Parameterization of dilute Ising model for iron-containing lanthanum gallate and aluminate solid solutions based on first-principles calculations

https://doi.org/10.1016/j.ssi.2020.115283Get rights and content

Highlights

  • Solid solution is represented as a set of configurations with estimated probabilities.

  • The dilute Ising model considers both magnetic and covalent contributions to energy.

  • The proposed model reproduces well the relative energies of different configurations.

  • Ga ↔ Fe and Al ↔ Fe interchange energies have different signs in the considered compounds.

  • Fe-clustering is more beneficial in Fe-doped LaGaO3 in comparison with LaAlO3.

Abstract

Calculations of the cubic models of two solid solutions LaGa0.5Fe0.5O3 and LaAl0.5Fe0.5O3 have been performed within the hybrid density functional theory. Multiple configurations have been considered for the solid solutions resulting from a different distribution of Fe atoms over the p-metal (Ga or Al) positions accounting for different spin orientation for Fe atoms themselves. It was demonstrated that 27 structures for both LaGa0.5Fe0.5O3 and LaAl0.5Fe0.5O3 should be treated to account for all the possible configurations in case of cubic 2 × 2 × 2 supercell. Optimized geometry, energy, and electron properties were calculated for all the obtained configurations. Statistical weights and probabilities were estimated for each symmetry non-equivalent configuration of the solid solutions within the canonical ensemble. A new parameterization for the dilute Ising model has been proposed. In this model, we account for non-magnetic contributions, which are absent in the simple Ising model, using the lattice approach based on the concept of interchange energy. Two model parameters (the magnetic coupling constant and interchange energy) were fitted to the calculated total energies of all considered configurations of both solid solutions. The dilute Ising model confirmed the benefit of Fe-clustering in doped lanthanum gallate against aluminate. Different signs of the estimated interchange energy enable us to explain the reasons for such differences.

Introduction

For the last decades and until today, the perovskite-type oxides are being actively explored due to a wide range of their applications. The slightest impurities in the perovskite structure can cause significant distinction in the electronic, optical, and magnetic properties. At the same time, the perovskite structure itself is tolerant for various substitutions in both A and B crystallographic orbits of ABO3, which opens large possibilities for doping with dia- and paramagnetic elements to obtain the necessary target products. That is the reason why complex oxides with perovskite structure became so popular.

A great number of works suggests some applications for the doped LaGaO3. For example, the increasing number of studies are focused on the photoluminescence properties of LaGaO3 doped with lanthanides (Sm3+, Tb3+, Bi3+, Eu3+, Tm3+, Yb3+) [[1], [2], [3]]. Besides this, the doped LaGaO3 oxides being mixed electronic-ionic conductors are frequently mentioned as promising candidates for solid oxide fuel cells (SOFC) [4,5]. Nevertheless, the practical applications of SOFCs are still limited due to the high costs of component materials: LaGaO3 is rather expensive material. In contrast, lanthanum aluminate, LaAlO3, is much cheaper and also studied profoundly [6]. However, a markedly smaller number of works relates to doped LaAlO3 systems.

Published experimental results [[7], [8], [9]] show that LaGaO3- and LaAlO3-solid solutions differ in properties. Thus, the studies of magnetic properties of LaMO3–LaAlO3 and LaMO3–LaGaO3 diluted solid solutions (M = Fe, Cr) [7,9] showed that the distribution of d-elements over B sites in perovskite structure is far from random. It was found that the fractions of dimer clusters in the diluted solutions exceeded the statistically probable fraction by a factor of about 2 in aluminates and 2.5 in gallates, whereas as was shown for lanthanum aluminate the fraction of clusters did not depend on the nature of 3d-element, but only on the composition of the host diamagnetic lattice. This means that in lanthanum gallate the clustering of paramagnetic M atoms is essentially greater than in the aluminate. A good example of such behavior is the systems LaFexAl1xO3 [8] and LaFexGa1xO3 [9]. The plots of paramagnetic components of magnetic susceptibility calculated per 1 mol of iron atoms are given in Fig. 1. Solid solutions of composition LaFexGa1xO3 and LaFexAl1xO3 (x = 0.01–0.10) were synthesized by ceramic method. The magnetic susceptibility χg was measured by Faraday method in the temperature range 77–400 K at 10 fixed values of magnetic field strength. The accuracy of measuring χg is 1%. The diamagnetic increments were introduced taking into account the susceptibility of diamagnetic matrix – LaGaO3 and LaAlO3 measured over the same temperature range. We are currently working on the paper with a more detailed discussion on the given experimental results. The slope of the isotherms shows that long order exchange interactions are antiferromagnetic in both systems and the paramagnetic component decreases with concentration to a greater extent in gallate. However, at low concentrations a large ferromagnetic component appears in the exchange according to magnetic susceptibility data for LaFexGa1xO3. This suggests that clustering of iron atoms in gallate is stronger than in aluminate. Nevertheless, these are model estimations based on experimental data, though carried out for a series of d-elements.

Antiferromagnetic-paramagnetic transition at TN ≃ 215 K was experimentally investigated by Komine et al. in the study of electronic conduction in LaGa0.5Fe0.5O3 [10]. The nature of the pre-peaks in the oxygen K-edge XANES (X-ray absorption near-edge structure) spectra of LaGaxFe1xO3 was discussed by Lafuerza et al. [11]. Their existence was interpreted in terms of a molecular-orbital bonding theory. The first pre-peak A is related to a 2p-3d bond (Osingle bondFe) and remains independent of Fe neighbors. The second pre-peak B is affected by the Ga substitution in LaFeO3, which changes both energy and intensity. This peak represents the states of σ symmetry, where 3d orbitals are strongly correlated with the coordination octahedra. The intensity of the peak B decreases with Ga substitution of one Fe atom in Fesingle bondOsingle bondFe triplet, and with the substitution of the two Fe atoms the peak is completely missing. In this respect, a more sensitive method is necessary for examination the orbital mixing at low concentrations of the 3d metal, i.e., the existence of Fe-clusters in the diluted solutions cannot be observed in XANES spectra.

Kuznetsov et al. [12] examined the magnetic properties of LaAl1xFexO3 solutions. Unexpectedly, the orthorhombic (Pbnm) → rhombohedral (R3¯c) phase transition was not observed over the whole range of concentrations, in contrast to other works [8,13]. The Mossbauer spectra of the LaFeO3, LaFe0.7Al0.3O3, and LaFe0.2Al0.8O3 confirm that the valence state of Fe atoms (3+) does not change with Al appearance. Ahmed et al. [14] recommended Fe-doped LaAlO3 as a prospective material for tunable capacitors and ferroelectric random access memory. The multiferroic LaAlyFe1yO3 (0 < y < 0.2) were prepared by citrate autocombustion method. It was shown [14] that the value of the Neel temperature in general decreases with the increasing of Al content (TN = 795 K for LaAl0.2Fe0.8O3).

In the theoretical study by Singh et al. [15], the photoluminescence properties of Fe-doped LaGaO3 were considered. First-principles density functional theory (DFT) calculations were carried out. The antiferromagnetic ordering of the solid solutions was established. The calculations showed that LaGaxFe1xO3 solutions have the direct band gap, which enables the observed photoluminescence. It is important that the calculations were carried out assuming the solid solution to be ordered and represented by a single structure. This approximation is the most commonly used in the first-principles calculations. Therefore, the interchange interactions between paramagnetic and diamagnetic atoms were ignored. However, it is these interatomic interactions that determine the magnetic and electrical performance of perovskites. The ions of Al3+ and Ga3+ have a similar valence electron structure and taking into account the disordered nature of the solid solutions is a key to the explanation of the observed differences. This means the discussed approximation becomes inappropriate for modeling of the solid solutions under consideration.

In this work, we do not assume that the solid solution is represented by a single ordered structure. We present an ab initio approach, which allows modeling of the magnetic properties of the solid solutions considering different distributions of the dopant atoms. To the best of our knowledge, the papers where the properties of LaFexGa1xO3 and LaFexAl1xO3 solid solutions were compared theoretically on ab initio level are absent.

Our paper is organized as follows: In Section 2, we discuss the choice of the computational scheme, the basics of the solid solutions calculations considering multiple configurations are presented in Section 3. In Section 4, we focus on the discussion of the results: Fe-clustering is analyzed in Section 4.1, 4.2 The dilute Ising model, 4.3 Stability investigation are devoted to the suggested parameterization of the dilute Ising model and stability investigation, respectively. Finally, the conclusions of the study are summarized.

Section snippets

Computational details

We used CRYSTAL14 computer code [16] to perform the quantum mechanical (QM) calculations within CO LCAO approximation (crystalline orbitals as a linear combination of atomic orbitals). Atomic orbitals are expanded into a set of localized atom-centered Gaussian-type orbitals (GTO's). To verify the computational scheme, we performed calculations of three phases for pure LaGaO3, LaAlO3 (Pbnm, R3¯c, Pm3¯m), and two phases for LaFeO3 (Pbnm [17], Pm3¯m). It was shown that hybrid functional PBE [18]

Modeling of solid solutions

Solid solutions do not form periodic systems. However, the periodicity of solid solutions is often assumed in the theoretical studies, when the proper supercell is created. CRYSTAL14 [16] computer code offers a more appropriate way to computational modeling of disordered systems. The solid solution is presented as an average over a set of configurations with distributions of different species over a given set of the atomic positions. The number of all the possible configurations of a solid

Fe-clustering and magnetic properties

The low energy structure for both compounds belong to the SIC 1 subSIC 4 (see Fig. 2) and has the maximum number, namely 8, of Fe(↑)single bondOsingle bondFe(↓) triplets with the antiferromagnetic (AFM) exchange. The energy of this AFM structure can be conditionally set as the zero-energy level. All the obtained 27 structures have different values of total energy Ek, which means they provide different contribution into the solid solution. This contribution should depend on the total energy Ek value of each structure

Conclusions

In this study, we performed first-principles hybrid DFT calculations considering all the possible distributions of Fe-atoms over the p-metal positions in the cubic model of LaGa0.5Fe0.5O3 and LaAl0.5Fe0.5O3 solid solutions. It is particularly important that the different spin distribution of the Fe atoms was investigated. The spin consideration is necessary for modeling systems with strong magnetic interactions, which affect the observed properties. The described approach can be applied to any

Prime novelty statement

We present the results of an ab initio study of the structure, stability, and magnetic properties of iron-containing solid solutions in two La-based perovskites. This work was initiated by the fact that the previous experimental studies indicate the existence of Fe-clusters in LaGaO3-LaFeO3 mixtures, whereas they were not observed in LaAlO3-LaFeO3 system.

The exact symmetry analysis was performed to obtain all the possible non-equivalent configurations of LaGa0.5Fe0.5O3 and LaAl0.5Fe0.5O3 cubic

Declaration of competing interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the Saint Petersburg State University (project number 110-8486). The authors gratefully acknowledge the University Computer Center for enabling to carry out high-performance computations.

References (44)

  • B.Y. Brach et al.

    Calculations of the distribution of the atoms of 3d elements in a solid solution and the parameters of the exchange interaction in dinuclear clusters

    J. Struct. Chem.

    (1982)
  • A.I. Fedortsov et al.

    State of atoms and interatomic interactions in complex perovskite-like oxides: XXXVI. 1. Magnetic susceptibility of LaFexGa1−xO3 and La1−0.2xSr0.2xFexMg0.2xGa1−1.2xO

    Russ. J. Gen. Chem.

    (2016)
  • S. Lafuerza et al.

    Origin of the pre-peak features in the oxygen K-edge x-ray absorption spectra of LaFeO3 and LaMnO3 studied by Ga substitution of the transition metal ion

    J. Phys. Condens. Matter.

    (2011)
  • M.V. Kuznetsov et al.

    Self-propagating high-temperature synthesis of aluminum substituted lanthanum ferrites LaFe1−xAlxO3(0 ≤ x ≤ 1.0)

    New J. Chem.

    (2015)
  • D. Kuščer et al.

    Defect structure and electrical properties of La1−ySryFe1−xAlxO3−δ

    J. Am. Ceram. Soc.

    (2001)
  • P. Singh et al.

    Fe doped LaGaO3: good white light emitters

    RSC Adv.

    (2016)
  • R. Dovesi et al.

    CRYSTAL14 User’s Manual

    (2014)
  • E. Heifets et al.

    Thermodynamic stability of stoichiometric LaFeO3 and BiFeO3: a hybrid DFT study

    Phys. Chem. Chem. Phys.

    (2017)
  • J.P. Perdew et al.

    Rationale for mixing exact exchange with density functional approximations

    J. Chem. Phys.

    (1996)
  • F. Corà

    The performance of hybrid density functionals in solid state chemistry: the case of BaTiO3

    Mol. Phys.

    (2005)
  • Energy-consistent pseudopotentials of the Stuttgart/Cologne Group

  • M. Dolg et al.

    Energy-adjusted pseudopotentials for the rare earth elements

    Theor. Chim. Acta

    (1989)
  • Cited by (0)

    View full text