The oxygen deficient Ruddlesden Popper La 3 Ni 2 O 7 d (d = 0.65) phase: Structure and properties

Transcription

1 Materials Research Bulletin 41 (2006) The oxygen deficient Ruddlesden Popper La 3 Ni 2 O 7 d (d = 0.65) phase: Structure and properties Viktor V. Poltavets a, Konstantin A. Lokshin b, Takeshi Egami b,c,d, Martha Greenblatt a, * a Department of Chemistry and Chemical Biology, Rutgers, The State University of New Jersey, 610 Taylor Road, Piscataway, NJ 08854, USA b Department of Materials Science and Engineering, University of Tennessee, Knoxville, TN 37996, USA c Department of Physics and Astronomy, University of Tennessee, Knoxville, TN 37996, USA d Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA Received 20 January 2006; accepted 23 January 2006 Available online 20 February 2006 Abstract La 3 Ni 2 O 7 d (d = 0.65) was synthesized by hydrogen reduction of the parent La 3 Ni 2 O 7 Ruddlesden Popper nickelate. The crystal structure of La 3 Ni 2 O 6.35 (space group: I4/mmm, a = (1) Å and c = (1) Å) has been determined from powder neutron diffraction data by the Rietveld method for the first time. The oxygen vacancies are located in the LaO x planes between two of the NiO 2 layers. Removal of these oxygen atoms from the parent phase results in a significant (0.4 Å) shrinkage of the perovskite block along c-direction and splitting of the Ni position. The major part of Ni cations is surrounded by five oxygen atoms forming square pyramids, while the rest are coordinated to six octahedrally arranged oxygen atoms. Over the K temperature range, the conductivity of La 3 Ni 2 O 6.35 follows Mott s variable range hopping model modified for a 2D case. # 2006 Elsevier Ltd. All rights reserved. Keywords: A. Oxides; B. Chemical synthesis; C. Neutron scattering; C. Thermogravimetric analysis; D. Crystal structure 1. Introduction There has been an interest in layered mixed valence nickelates due to their crystal and electronic structural similarity to superconducting cuprates. Especially interesting are the Ni + /Ni 2+ phases, with the same electronic configuration (3d 9 /3d 8 ) as the Cu 2+ /Cu 3+ complex oxides that show high temperature superconductivity. Phases, of the Ruddlesden Popper (RP) series, A (n+1) B n O 3n+1 d, containing Ni + /Ni 2+ cations have been investigated by several groups [1 3]. Several phases, for example LaSrNiO 3.1 [1],La 3 Ni 2 O 6.35 [2] and La 4 Ni 3 O 8 [3], are oxygen deficient and were prepared by a low temperature topotactic reduction of parent nickelates. Structural data, which are crucial for properties understanding, have been reported only for RP phases with n =1,La 2 NiO 4 s and LaSrNiO 3.1 [1], and n =3, La 4 Ni 3 O 8 [3]. For La 3 Ni 2 O 6.35, the most reduced phase of the n = 2 homologue known up to now, neither structural data nor a facile preparation method are reported. * Corresponding author. Tel.: ; fax: address: martha@rutchem.rutgers.edu (M. Greenblatt) /$ see front matter # 2006 Elsevier Ltd. All rights reserved. doi: /j.materresbull

2 956 V.V. Poltavets et al. / Materials Research Bulletin 41 (2006) A controlled reduction of La 3 Ni 2 O 7 in a thermogravimetric instrument was employed to isolate the La 3 Ni 2 O 6.35 phase [2,3]. It was found that the c unit cell parameter of the tetragonal reduced phase is approximately 0.46 Å smaller than that for the parent rhombohedral RP La 3 Ni 2 O 7 phase [2]. However, detailed structural information for the reduced phase, reflecting coordination of Ni cations, was missing. In this article, we present syntheses and a crystal structure study by powder neutron diffraction of the oxygen deficient RP phase, La 3 Ni 2 O Magnetic and electrical properties are also briefly reported. 2. Experimental La 3 Ni 2 O 7 was synthesized by the sol gel Pechini technique [4] with La 2 O 3 and Ni(OH) 2 as source of metals. La 2 O 3 was prepared by decomposition of La(NO 3 ) 3 6H 2 O (Alfa, 99.9%) at 750 8C for 2 h in air. The precise composition of Ni(OH) 2 (Acros, 99.9%) was determined by TGA. Stoichiometric amounts of La 2 O 3 and Ni(OH) 2 were dissolved in a solution of citric acid with several drops of concentrated nitric acid. When all the reactants had dissolved, an appropriate amount of ethylene glycol was added. The solution was stirred with moderate heating on a hot plate, resulting in the formation of a gel. The gel was dried at 300 8C and then heated at 600 8C overnight to remove organic matter. The resultant powder was pressed into pellets and annealed in oxygen flow at C for 100 h with two intermittent grindings. Finally, treatment at 500 8C for 5 h in oxygen flow was performed as a final step to achieve oxygen stoichiometry. La 3 Ni 2 O 6.35 was prepared by the reduction of the La 3 Ni 2 O 7 powder in flowing 10% H 2 in Ar at 400 8C for 12 h in a tube furnace. Powder neutron diffraction (PND) data were collected on 2.5 g sample at 300 K on the NPDF time-of-flight neutron diffractometer at the Lujan Neutron Science Center of the Los Alamos National Laboratory. Two-phase Rietveld refinement of the obtained data [5] was performed with GSAS [6] program with EXPGUI [7] interface. Vanadium was introduced as a second phase to the refinement, since small peaks from the vanadium sample container ware observed in the PND pattern. In the final runs, the scale factor, unit-cell parameters, peak profile parameters, background coefficients, absorption coefficient, atomic coordinates, isotropic atomic displacement parameters and fractional occupancy for oxygen atoms were simultaneously refined. The quality of agreement between the observed and calculated profiles was monitored by a set of conventional R factors, as described by Young [8]. The powder X-ray diffraction (PXD) patterns were recorded at room temperature over an angular range 58 2u 908 with a step of (2u) on a Bruker D8 Advance diffractometer (Bragg-Brentano geometry, Cu Ka radiation). Thermogravimetric analysis (TGA) was performed with a TA Instrument 2050 thermal analyzer. Samples were ramped at 2 8C/min to a final temperature of 900 8C. The dc magnetic susceptibility measurements were carried out on powder samples with a commercial SQUID magnetometer (Quantum Design, MPMS-XL). All measurements were performed by warming the samples in the applied field after cooling to 5 K in zero field (ZFC, zero field cooling) and by cooling the samples in the applied measuring field of 0.01, 0.1 or 1 T (FC, field cooling). Resistivity measurements were performed on the same machine with Keithley equipment and with a standard four-probe technique. 3. Results and discussion The purity of the parent La 3 Ni 2 O 7 phase was confirmed by the indexing of all of the observed reflections of the PXD pattern. The refined unit cell parameters of the orthorhombic La 3 Ni 2 O 7 phase, a = (1) Å, b = (1) Å and c = (3) Å, are in good agreement with previously reported data [2,9,10]. The oxygen stoichiometry of this phase was determined as La 3 Ni 2 O 7.04 by TGA in a 10% H 2 /Ar flow (Fig. 1). The reduction of La 3 Ni 2 O 7 during the TGA measurement occurs in two steps with a final formation of a mixture of La 2 O 3 and Ni (confirmed by PXD). A plateau in the TGA curve corresponds to the stabilization of oxygen deficient RP La 3 Ni 2 O 6.35 phase that was already noted in the literature [1,2]. However, attempts to prepare a large (several grams) sample of La 3 Ni 2 O 6.35 by the reduction of La 3 Ni 2 O 7 at 450 8C for 12 h in a 10% H 2 /Ar flow led to complete decomposition, unlike the controlled reduction in TGA [2]. A pure bulk 2.5 g powder sample of La 3 Ni 2 O 6.35 was prepared by annealing of the La 3 Ni 2 O 7 powder at 400 8C for 12 h in a tubular furnace in a 10% H 2 /Ar flow. It should be noted, that La 3 Ni 2 O 6.35 is metastable at the preparation condition. Namely, prolonged (60 h) annealing at the above specified conditions resulted in admixture of La 2 O 3 and Ni. The oxygen stoichiometry of the sample was derived as La 3 Ni 2 O 6.38(2) by TGA, which is in the range of the

3 V.V. Poltavets et al. / Materials Research Bulletin 41 (2006) Fig. 1. Thermogravimetric curves for La 3 Ni 2 O 7 and La 3 Ni 2 O 6.35 in 10% H 2 /Ar. previously reported values [2]. This sample, noted further in the text as La 3 Ni 2 O 6.35, was used for collection of powder neutron diffraction data, magnetic susceptibility and resistivity measurements. For the initial Rietveld refinement of the La 3 Ni 2 O 6.35 structure, atomic coordinates from an idealized tetragonal (space group: I4/mmm) n = 2 RP model were used. The refinement led to a reasonable fit of the PND data with reliability factors of R wp = 3.25%, R p = 2.25%. The obtained fractional occupancy of O1 position was 0.36(1), while those for O2 and O3 oxygen atom sites differed from unity for less than one standard deviation. An unusually large atomic displacement parameter for the Ni atom and its asymmetric coordination in the vicinity of oxygen vacancies suggested a splitting of the Ni position. The atomic position of Ni was split into two positions, Ni1 and Ni2, which correspond to Ni atoms near the filled and vacant O1 position, respectively. According to this model, the occupancy of Ni1 was equally constrained with the occupancy of O1 and the sum of Ni1 and Ni2 was fixed to 1. The refinement of the crystal structure with the split Ni atomic positions resulted in a significantly better profile match dropping the reliability factors to R wp = 2.95% and R p = 2.04%. The Rietveld refinement patterns and the refined structure of La 3 Ni 2 O 6.35 are shown in Figs. 2 and 3, respectively. The structural data and goodness of fit parameters for the final refinement are listed in Table 1. The major discrepancy of the data sets is observed around the {1 0 9} and {2 0 0} peaks. However, no evidence of symmetry lowering or superstructure formation for La 3 Ni 2 O 6.35 was found. We believe that this discrepancy is caused by staking fault defects, which are known to occur in reduced nickelates [11]. The oxygen composition derived from the Rietveld refinement is La 3 Ni 2 O 6.34(1), in good agreement with the TGA data. Comparison of the structural models of the parent RP La 3 Ni 2 O 7 and reduced La 3 Ni 2 O 6.35 phases shows that the reductive transformation is topotactic. Oxygen vacancies are found solely in the LaO 0.35 planes of the perovskite Fig. 2. Rietveld refinement profiles: observed (dots), calculated (solid lines), difference (bottom solid lines) and Bragg reflections (tick marks) for the fit to the PND pattern of the La 3 Ni 2 O 6.35 (upper tick marks, La 3 Ni 2 O 6.35 ; bottom tick marks, Vanadium).

4 958 V.V. Poltavets et al. / Materials Research Bulletin 41 (2006) Fig. 3. a) Idealized tetragonal (space group I4/mmm) structure of an n = 2 RP phase. Perovskite blocks of La 3 Ni 2 O 6.35, showing local coordination polyhedra around nickel: (b) corner-sharing six-coordinate octahedra for the Ni1 position, (c) five-coordinated square pyramid for Ni2 site. blocks of the structure. The coordination of a majority of the nickel atoms transformed from six (octahedral) to five (square pyramidal, see Fig. 3). The Ni atoms, located in the Ni2 position, lie in the planes of O3 oxygens. If a Ni1 cation, situated near the filled O1 position, would stay exactly in plane of the O3 oxygen atoms, then an impossibly short Ni1 O1 distance, equal to Å, and a much longer Ni1 O2 distance, Å, would exist. Therefore, the displacement of the Ni1 position toward the O2 oxygen occurs. Ni1 adopts a distorted octahedral coordination and does not lie in the plane of the O3 atoms. A difference in the coordination polyhedra for Ni1 and Ni2 might also reflect a nonuniform Ni + /Ni 2+ cation distribution over those positions. The Ni2 square pyramidal coordination is typical for Ni 2+ (d 8 ). The Ni2 O3 bond length, Å is very close to that of the corresponding Ni O, Å bond in La 2 NiO 4. Thus, this position is, most likely, exclusively occupied by Ni 2+, while the Ni1 position is shared between Ni 2+ and Ni + Table 1 Results of the Rietveld refinement of La 3 Ni 2 O 6.35 in the space group I4/mmm (no. 139) Atom Site x y z U iso (Å 2 ) Occupancy La1 2b (1) 1 La2 4e (1) (1) 1 Ni1 4e (4) (1) 0.34 (1) Ni2 4e (1) (1) 0.66 (1) O1 2a (1) 0.34 (1) O2 4e (3) (1) 1 O3 8g (1) (1) 1 Note. Lattice parameters (Å): a = b = (1), c = (1). R factors (%): R wp = 2.95, R p = 2.04.

5 V.V. Poltavets et al. / Materials Research Bulletin 41 (2006) Fig. 4. A plot of mass magnetization M vs. temperature for La 3 Ni 2 O 6.35 (ZFC, zero field cooling; FC, field cooling). in almost equal ratio. The larger atomic displacement parameter observed for Ni1 compared to that for Ni2 (Table 1)is attributed to the sharing of the Ni1 atomic site by both Ni + /Ni 2+. The temperature dependence of magnetization in Fig. 4 shows divergence between the ZFC and FC plots at low temperature. Such a behavior is typical for spin-glasses, and suggests mixed magnetic interactions in the system. Indeed, it is most likely that the superexchange interaction between Ni 2+ (S = 1) spins are negative, while the doped holes on oxygen, nominally resulting in Ni +, would introduce a positive exchange interaction [2,12]. Such a mixture of negative and positive exchange interactions can result in the spin-glass behavior. However, formation of Ni during reduction of various rare earth nickelates was previously observed [2,11,13]. It is possible that the magnetic response of a very small amount of Ni impurity is dominating the magnetic data, completely masking a weak response from the main phase. If we assume that is the case, the amount of Ni in our sample, which can be estimated in supposition that magnetization was saturated at 1 T, is about 0.25%. Further study of the magnetic behavior with the controlled Ni impurity is planned. The electrical resistivity measurements were carried out on polycrystalline pressed pellets in the temperature range of K. Below 170 K, the resistivity exceeded the values accessible by our apparatus. The electrical resistivity of La 3 Ni 2 O 6.35 can be reasonably fitted to Mott s law (a = 1/4) for variable range hopping (VRH) [14]; however, Mott s law modified for a 2D case (with a = 1/3) resulted in a better fit (Fig. 5). This model is also preferable due to the highly Fig. 5. The natural logarithm of resistivity, ln(r) vs. T ( 1/3) : observed data (dots) and fitting curve (solid line) according to Mott s variable-rangehopping model with a = 1/3.

6 960 V.V. Poltavets et al. / Materials Research Bulletin 41 (2006) two-dimensional nature of the crystal structure of the reduced phase (Fig. 3), i.e. higher conductivity could be expected in the NiO 2 planes, compared to that along the c axis. The limited temperature range prevents unambiguous identification of the conduction type. Nevertheless, the fit of the resistivity data to Mott s VRH model with localized states near the Fermi level [15]: r ¼ r 0 exp T0 T a ; T 0 ¼ b M k B g 0 j 2 ; a ¼ 1 3 yields T 0 = K and r 0 = V cm. In Eq. (1) T 0 is a measure of the degree of charge carrier localization, r 0 the pre-exponential factor, g 0 the density of states at the Fermi level, j the localization length and b M is a constant coefficient equal to 27/p for 2D systems. The fit of the resistivity data to Eq. (1) suggests Anderson localization due to electronic disorder in the compound. Anderson localization could be due to oxygen vacancy disorder and, associated with it, Ni + /Ni 2+ disorder in La 3 Ni 2 O Conclusions La 3 Ni 2 O 6.35, an oxygen deficient n = 2 RP phase has been synthesized by the reduction of the La 3 Ni 2 O 7 in flowing 10% H 2 in Ar at 400 8C. The Rietveld refinements of powder neutron diffraction data indicate that La 3 Ni 2 O 6.35 crystallizes in I4/mmm space group. Oxygen vacancies are located solely in the LaO x planes of the perovskites block of the structure, leading to a lowering of the coordination number from octahedral to square pyramidal for a majority of the nickel atoms. The temperature variation of resistivity follows Mott s law for variable-range hopping behavior. Acknowledgements This work was supported by the National Science Foundation through DMR at the Rutgers and DMR at the University of Tennessee. References (1) [1] M. Crespin, C. Landron, P. Odier, J.M. Bassat, P. Mouron, J. Choisnet, J. Solid State Chem. 100 (1992) 281. [2] Z. Zhang, M. Greenblatt, J.B. Goodenough, J. Solid State Chem. 108 (1994) 402. [3] Ph. Lacorre, J. Solid State Chem. 97 (1992) 495. [4] M. Pechini, Method of preparing lead and alkaline-earth titanates and niobates, US Patent 3,330,697, 11 July [5] H.M. Rietveld, J. Appl. Crystallogr. 2 (1969) 65. [6] A.C. Larson, R.B. Von Dreele, General Structure Analysis System (GSAS), Los Alamos National Laboratory Report LAUR , [7] B.H. Toby, J. Appl. Crystallogr. 34 (2001) 210. [8] R.A. Young, in: R.A. Young (Ed.), The Rietveld Method, Oxford University Press, Oxford, [9] V.I. Voronin, I.F. Berger, V.A. Cherepanov, L.Y. Gavrilova, A.N. Petrov, A.I. Ancharov, B.P. Tolochko, S.G. Nikitenko, Nucl. Instrum. Methods Phys. Res. Sect. A 470 (2001) 202. [10] C.D. Ling, D.N. Argyriou, G. Wu, J.J. Neumeier, J. Solid State Chem. 152 (2000) 517. [11] M.A. Hayward, M.J. Rosseinsky, Solid State Sci. 5 (2003) 839. [12] J.B. Goodenough, Struct. Bonding 98 (2001) 1. [13] M.A. Hayward, M.A. Green, M.J. Rosseinsky, J. Sloan, J. Am. Chem. Soc. 121 (1999) [14] N.F. Mott, J. Non-Cryst. Solids 1 (1968) 1. [15] N.F. Mott, E.A. Davis, Electronic Processes in Non-crystalline Materials, Clardenon Press, Oxford, 1979.