DEFECTS, SURFACES, SOLID SOLUTIONS, AND REACTIVITY OF ADVANCED PEROVSKITES
E.A. Kotomin, Yu. Zhukovskii, S. Piskunov, R. Eglitis, and Yu. Mastrikov,
R.A. Evarestov (St. Petersburg University, Russia) ,
D.E. Ellis (Northwestern University, Evanston, Illinois, USA) ,
E. Heifets (California Institute of Technology, Pasadena, USA) ,
J. Felsteiner (Technion, Haifa, Israel) and A. Gordon (Haifa University, Tivon, Israel),
G. Borstel (Osnabruck University, Germany), J. Maier (MPI for Solid State Research, Stuttgart, Germany).

Understanding surface properties of perovskites is important for catalysis, growth of high T_c materials, micro- and optoelectronics. In collaboration with Northwestern University, Evanston, and California Institute of Technology, Pasadena, we have studied the atomic and electronic structures as well as thermodynamic stability of three double-layered (DL) SrTiO₃(001) surfaces: (i) SrO-terminated, (ii) TiO₂-terminated, and (iii) TiO₂-terminated perovskite with (2x1) substrate reconstruction. A thermodynamic stability diagram obtained from our first-principle calculations using hybrid B3PW exchange-correlation functional have shown that regular TiO₂- and SrO-terminated surfaces remain to be the energetically most stable. Application of hybrid functional allowed us to improve accuracy of our analysis with respect to pure DFT computations. We were able to obtain nearly perfect agreement of calculated and experimental formation energies and, therefore, to determine very well boundaries of stability of different crystals. The stability regions of (2x1) DL TiO₂- and DL SrO-terminated surfaces lie beyond the precipitation lines of SrO and TiO₂ compounds and, thus, are less stable with respect to regular SrTiO₃(001) surfaces. We expect that precipitation of strontium and titanium oxides will occur much earlier, than any of studied DL terminations can form. It allowed us also to suppose that the SrO or TiO₂ oxide films would grow preferably on SrTiO₃ perovskite through cluster formation rather than layer-by-layer deposition. Our simulations have also shown a substantial increase of Ti-O bond covalency on the DL surfaces. When oxygen gas partial pressure decreases at a constant temperature, either TiO₂ precipitation or Ti atom reduction occurs, which precipitate to metallic particles. Sr precipitation on SrTiO₃ surface will not occur. Increase of chemical potential of Sr will lead to precipitation of SrO.

In collaboration with Osnabruck University and California Institute of Technology, Pasadena, we have also calculated relaxation of BaTiO₃ and PbTiO₃ perovskite (001) surfaces and surface rumpling for two different terminations (BaO&PbO and TiO₂) as well as relaxation of BaTiO₃ and PbTiO₃(110) surfaces for three different terminations (Ba&Pb, TiO2, and O). The O-terminated A-type BaTiO₃(110) polar surface possesses a surface energy close to that for the (001) surface, which indicates that both (110) and (001) BaTiO₃ surfaces can exist simultaneously in perovskite ceramics. According to the results of our calculations, the energetically most unfavorable, and thereby most unstable are metal (Ba or Pb) terminated BaTiO₃ (3.24 eV) or PbTiO₃ (2.03 eV) (011) surfaces. It is interesting to notice, that the surface energies for the O terminated, A type BaTiO₃ and PbTiO₃ (011) surfaces coincide (1.72 eV).

Due to its antiferroelectric behavior, lead zirconate PbZrO₃ (PZ) is technologically important for many applications including actuators and high-energy storage devices. PbZrO₃ is also a parent compound of PbZr_1-xTi_xO₃ solid solutions, which are of high scientific and technological interest for their ferroelectricity and piezoelectricity, observed over a wide range of compositions. The structural and electronic properties of pure cubic and low-temperature orthorhombic PbZrO₃ (antiferroelectric phase), as well as cubic PZ containing single F centers (neutral oxygen vacancies) have been simulated in collaboration with Northwestern University, Evanston. The band gap obtained for PZ bulk (cubic phase) is in good agreement with the experimental data. The electronic charge redistribution calculated for a cubic bulk lead zirconate confirms a notable covalency of the Pb-O bond as proposed from analysis of X-ray powder diffraction data. This bond covalency is considerably increased in the orthorhombic phase. We have also began calculations on the atomic and electronic structure of PbZrO₃(001) surfaces for both phases. The band gaps calculated for ZrO₂- and PbO-terminated (001) surfaces are reduced as compared to that for PZ bulk. According to the results of our calculations, both ZrO₂- and PbO-terminated (001) PZ surfaces are stable, and since the surfaces energies for PbO and ZrO₂ (001) terminations are close (0.93 and 0.95 eV, respectively), they could coexist after cleavage of PZ bulk crystal. We have found a strong increase of the Zr-O bond covalency near the ZrO₂-terminated (001) surface as compared to the PZ bulk. Formation of an F center in cubic lead zirconate is accompanied by a substantial displacement (0.25 A) of the nearest lead atoms towards the vacancy. The F center traps 0.68 e and forms a defect level in the middle of the band gap (1.72 eV below the conduction band bottom) - unlike the shallow F level found in SrTiO₃ (0.5 eV). Thus, we predict that the presence of point defects affects the atomic polarization in lead zirconate and influences its ferroelectric properties.

Understanding and control of surface properties of pure and Sr-doped LaMnO₃ is important for applications in fuel cells, magnetoresistive devices, and spintronics. We have compared the atomic, electronic, and magnetic structures of LaMnO₃ bulk as well as the (001) and (110) surfaces calculated in collaboration with Max Planck Institute for Solid State Research, Stuttgart, and St. Petersburg University, using two ab initio approaches - hybrid B3PW functionals with optimized LCAO basis set (CRYSTAL-2003 code) and GGA-PW91 functional with plane wave basis set (VASP 4.6 package). We have demonstrated that a combination of non-local exchange and correlation used in hybrid functionals allows us to reproduce the experimental magnetic coupling constants Jab and Jc as well as the optical gap in much better agreement than other methods. Surface calculations performed by both methods using slab models show that the anti-ferromagnetic (AFM) and ferromagnetic (FM) (001) surfaces have lower surface energies than the FM (110) surface. Both the (001) and (110) surfaces reveal considerable atomic relaxations, up to the fourth plane from the surface, which reduce the surface energy by about a factor of two, being typically one order of magnitude larger than the energy difference between different magnetic structures. The calculated (Mulliken and Bader) effective atomic charges and the electron density maps indicate a considerable reduction of the Mn and O atom ionicity on the surface.

The fuel cell performance is defined by the O transport. To this end, we have performed DFT plane-wave calculations of the O atom adsorption and diffusion on the LaMnO₃ (001) surface. We have analyzed the effective charges and atomic displacements for non-stoichiometric symmetric (7 planes) and stoichiometric asymmetric (8 planes) slabs. Our calculations have shown a large difference between O adsorption on 7-plane and 8-plane slabs, which is probably caused by non-stoichiometric nature of the former. We predict that the energetically most favorable O adsorption position is over surface Mn ion, with the migration path along the (001) axis.


FIRST PRINCIPLES SIMULATION OF ELECTRONIC STRUCTURE FOR PERFECT AND DEFECTIVE BaF₂ AND CaF₂: BULK AND SURFACE
R. Eglitis,
H. Shi and G. Borstel (University of Osnabruck, Germany)

In collaboration with Osnabruck University, we have performed ab initio calculations on technologically important barium and calcium fluorides (perfect and defective bulk and densely-packed surfaces). BaF₂ is important as a candidate material for high-temperature batteries, fuel cells, chemical filters and sensors. CaF₂ has been identified as a prime candidate for windows operating at chemical laser wavelengths due to very low bulk absorption and exceptionally small thermal tensing coefficients. The hybrid B3PW method was used as implemented into the CRYSTAL-03 code which provides the best agreement with experiment for the band gap. When comparing the results of calculations on CaF₂ (111), (110), and (100) surfaces, we have confirmed that the CaF₂(111) surface is the most stable one, in agreement with the experiment, the same is true for the BaF₂(111) surface. The characterization of F centers in BaF₂ and CaF₂ is still a question of debate. The charge density map of the F center in CaF₂ shows that the charge is well localized inside the vacancy: the spin density on the F center has been found to be 0.716 e. The relaxation of atoms around the F center is rather small. CaF₂ experimentally exhibits optical absorption at 3.3 eV. Our results for defect levels suggest a possible mechanism for this absorption. The observed optical absorption may correspond to an electron transition from the F-center ground state, which lies by 6.75 eV above the top of the valence band, to the conduction band bottom. The F center defect band at G point is located 4.24 eV under the conduction band, which is qualitatively close to the experimentally observed absorption energy of 3.3 eV. For both BaF₂ and CaF₂ we also calculated the aggregation of two F centers (M center).

QUANTUM CHEMICAL INTERPRETATION OF X-RAY ABSORPTION SPECTRA IN ABO₃ COMPOUNDS
D. Bocharov,
A. Kuzmin (EXAFS Laboratory, ISSP, Riga, Latvia).

X-ray absorption spectroscopy is one of modern methods of experimental physics and materials science, which provides experimentalists with unique information on electronic, atomic and dynamic structure of materials. At the same time, quantum chemistry allows us to simulate reactivity, chemical properties, atomic and electronic structure of crystalline solids. In this work, the interpretation of experimental O K-edge X-ray absorption spectra (XAS) in perovskite-type WO₃ and AWO₃ compounds (A is the first group ion: H, Li, Na, K, Rb, Cs) has been performed when applying both DFT method as implemented in CRYSTAL-2003 code as well as conventional multiple-scattering approach realized in FEFF-8.2 code were used. Results of our calculations performed using both codes show qualitative agreement with available experimental data.
FEFF multiple-scattering calculations of XANES region allow us to obtain spectra qualitatively close to experimental ones. Calculated XAS have energy scale, which is non-linearly compressed. The position of the Fermi level was estimated with an accuracy of about ±2-4 eV. XAS is sensitive to the position of the Fermi level and its correction is required to achieve reasonable agreement with the experiment. A cluster size of about 6.5-7.5 Å around absorbing oxygen atom allows one to reproduce qualitatively experimental curves. The accuracy of CRYSTAL program to describe unoccupied electron states above the Fermi level was tested for AWO₃ systems. We found that several optimized basis sets give similar results for our compounds in the region up to 15 eV above the Fermi level, whereas strong deviation of calculated results for higher energies indicates CRYSTAL code limitations. CRYSTAL calculations allow us to estimate bonds covalency in AWO₃ compounds. For example, O-H bond is strongly covalent, W-O bond may be described in terms of both ionic and covalent contributions whereas O-A bonds are either pure ionic (A = K, Cs, Rb) or ionic with almost negligible covalency.

Back to Main Research Interests