Investigations into the structure of La3Ni2−xFexO7±δ 51 D O I: 1 0. 15 82 6/ ch im te ch .2 01 9. 6. 2. 03 Kiselev E. A., Gaczynski P., Eckold G., Feldhoff A., Becker K.-D., Cherepanov V. A. Chimica Techno Acta. 2019. Vol. 6, No. 2. P. 51–71. ISSN 2409–5613 E. A. Kiseleva, P. Gaczynskib, G. Eckoldc, A. Feldhoffd, K.-D. Beckerb, V. A. Cherepanova a Department of Physical and Inorganic Chemistry, Institute of Natural Sciences and Mathematics, Ural Federal University, Kuybysheva St. 48, 620026, Yekaterinburg, Russia b Institute of Physical and Theoretical Chemistry, Technische Universität Braunschweig, Rebenring 56. 10, D-38106, Braunschweig, Germany c Institute of Physical Chemistry, Georg-August-Universität Göttingen, Tammannstr. 6, 37077, Göttingen, Germany d Institute of Physical Chemistry and Electrochemistry, Leibniz Universität Hannover, Callinstr. 3a, D-30179, Hannover, Germany eugene.kiselyov@urfu.ru Investigations into the structure of La3Ni2–xFexO7±δ The room-temperature (RT) 57Fe Mössbauer spectra of the La3Ni2–xFexO7±δ oxide solid solutions of Ruddlesden-Popper-type (x = 0.05, 0.10) reveal two doublets for Fe3+ ions in octahedral coordination by oxygen. The existence of two inequivalent sites for Fe at RT is at variance with the space groups Fmmm and Cmcm (Amam) which have been reported for La3Ni2O7±δ. This unexpected find- ing is discussed in connection with Patterson analyses and Rietveld refinements of powder XRD data for x = 0, 0.05, and 0.10. Alternative structural models have been proposed which can explain the spectroscopic findings and which are compatible with the results from X-ray diffraction. Keywords: La3Ni2O7; complex oxides; Ruddlesden-Popper phases; Mössbauer spectroscopy; crystal structure; Patterson analysis; Rietveld refinement. Received: 15.07.2019. Accepted: 01.08.2019. Published: 05.08.2019. © Kiselev E. A., Gaczynski P., Eckold G., Feldhoff A., Becker K.-D., Cherepanov V. A., 2019 Introduction Layered Ruddlesden-Popper (R-P) lanthanum nickelates of the general formula Lan+1NinO3n+1 have recently been studied for potential application as SOFC cathodes [1–5] for the members with n = 1, 2, 3. The structure is characterized by n perovskite-type layers, n·LaNiO3, stacked up in c-direction and separated by a single rock-salt-type layer of LaO, LaO(LaNiO3) n. Presently, however, La2NiO4+δ with n = 1 is the most studied within the series; data concerning La3Ni2O7±δ and La4Ni3O10–δ are still insufficient and sometimes contradic- tory. In the present paper, the focus will be on La3Ni2O7±δ (n = 2). La3Ni2O7±δ was first reported by Brisi et al. [6] as an intermediate phase in an at- tempt to synthesize La4Ni3O10–δ at 1100 °C. Drennan et al. [7] could directly obtain the  n = 2 compound by  annealing at 1150  °C in  air for 5  h with intermediate re-grindings. The unit-cell parameters were approximately determined as a = 5.42 Å, b = 5.47 Å, and c = 20.58 Å. However, elec- tron microscopy revealed intergrowth of all members of the series with n = 2, 3, and 4. Later, the  existence of  La3Ni2O7±δ was confirmed by Odier et al. [8] in the tem- perature range 900–1150  °C. However, small amounts of secondary phases were 52 still found present. Mohan Ram et al. [9] obtained La3Ni2O7±δ by decomposing stoi- chiometric amounts of  lanthanum and nickel nitrates and annealing at 1420 K for 10  h with frequent grindings. Although the X-ray diffraction patterns showed no evidence for the presence of other mem- bers of the series, HREM studies showed intergrowth of  small extent of  members with n = 3 and n = 5. Similar problems were also reported by Sreedhar et al. [10] who stated that intergrowth can be minimized by  annealing samples at  a  suitable tem- perature (1150  °C). According to  Zhang et al. [11], single phase La3Ni2O7±δ can be obtained by  solution synthesis fol- lowed by heat treatment and final anneal- ing of the pressed pellets for 4–5 days at 1150–1200 °C. These authors determined the oxygen content of the as-prepared n = 2 lanthanum nickelate by iodometric titra- tion and TGA to La3Ni2O6.92 [11]. Zhang et al. [11] were the  first to  re- fine (X-ray) diffraction data of La3Ni2O6.92 in the orthorhombic space group Fmmm which provides only a single site for nickel cations. This space group was also used by Carvalho et al. [12] in their refinement of  the  unit-cell parameters of  the  com- pound. These authors also showed that the oxygen content of as prepared samples depends on the  synthesis route: nitrate precursors usually were found to  yield oxygen-deficient materials, e. g. La3Ni2O6.93 which is close to results reported in Ref. [11]. On the other hand, slightly oxygen excess samples, e. g. La3Ni2O7.03, could be produced by the citrate technique. Gener- ally, least-square refinements of XRD data within the  Fmmm space group showed good agreement in the unit cell parameters [1, 11–13]. However, neutron diffraction data by Ling et al. [14] on La3Ni2O7.02 and La3Ni2O7.05 revealed a number of weak but clearly resolved peaks which could not be indexed by the Fmmm structural model, but could satisfactorily be refined within the Amam space group (a nonstandard set- ting of the Cmcm space group). The main features of this structure, which also pro- vides a  single unique site for nickel, are significantly distorted NiO6 octahedra and an elongation of Ni–O bonds in c-direction towards to the rock salt layers [14]. These re- sults were confirmed by Voronin et al. [15] using single-phase La3Ni2O7±δ. In this work, the standard Cmcm space group was used for refinement of neutron diffraction data and for EXAFS spectral analysis. The unit cell parameters and the  oxygen content of the La3Ni2O7±δ powder samples obtained by various authors [1, 6, 7, 9–17] are listed in Table 1 for comparison. The tempera- ture dependence of  the  oxygen content of the n = 2 R-P compound in air has been studied by Bannikov and Cherepanov [16] in the temperature range 900–1150 °C. It was shown that the stoichiometric param- eter δ of  La3Ni2O7–δ changes from about 0.057 at 900 °C to 0.072 at 1150 °C which is in good agreement with the results pre- sented in Refs. [1, 11, 12] for the samples prepared by nitrate route, Table 1. At  elevated temperatures, La3Ni2O7±δ has been found to  undergo a  structural phase transition. Sasaki et al. [13] observed a  discontinuous change in  the  tempera- ture dependence of the lattice parameter for the  c-axis of  the  unit cell at  about 550  K. The  transition has also been de- tected in lattice expansion [1], in magnetic [18], infrared [19], electrical conductivity [1, 18, 19], and in thermal measurements [18, 20]. From all these experiments, tran- sition temperatures between 550 and 600 K have been reported. More recently, Amow et al. [1] monitored changes in the high- temperature X-ray diffraction patterns 53 of La3Ni2O6.95 which led them to the con- clusion that at about 590 K a transforma- tion occurs from orthorhombic Fmmm to a higher symmetry tetragonal phase. Mössbauer spectroscopy has al- ready been applied to  the  neigh- boring Ruddlesden-Popper phases La2NixFe1–xO4+δ (n  =  1)  [21–24] and La4Ni3–xO10–δ (n  =  3)  [24–26] in  a  num- ber of studies. The 57Fe Mössbauer spectra of these compounds are composed of one and of two doublets, respectively. This ob- servation is in full agreement with the lay- ered structures of these compounds which provide one unique site and two non-equiv- alent sites for nickel in the R-P phases with n = 1 and n = 3, respectively. This has been considered strong evidence for the conjec- ture that Fe is incorporated on the tran- sition metal sites in  the  R-P lanthanum nickelates. Under this condition  — and provided that La3Ni2–xFexO7±δ crystallizes in space group Fmmm or Cmcm (Amam) — the Mössbauer spectrum of La3Ni2–xFexO7±δ is expected to consist of a single signal only. Among the layered Ruddlesden-Popper lanthanum nickelates, La3Ni2O7±δ is special in that the average oxidation state of nickel cations for the stoichiometric composition with δ = 0 is 2.5, indicating equal number of Ni2+ and Ni3+ ions. At low temperatures, this can open the possibility for charge or- dering of the nickel cations. Nevertheless, all structural models existing to date for La3Ni2O7±δ (Fmmm [1, 11–13], Amam [14], and Cmcm [15]) do not account for this special aspect in  providing only a  single site for Ni. The present Mössbauer study was undertaken in order to shed light in- to the  local structure of  La3Ni2–xFexO7±δ. In contrast to the neighboring R-P phases La2Ni1–xFexO4+δ and La4Ni3–xFexO10–δ which have already been well studied by Mössbau- er spectroscopy [21–26], the present work to our knowledge appears to be the first Mössbauer study of La3Ni2–xFexO7±δ. Table 1 Structural data of La3Ni2O7±δ obtained from X-ray and neutron diffraction, RT a (Å) b (Å) c (Å) Space group Oxygen content 7±δ Ref. 5.404(2) 5.452(2) 20.442(7) Fmmm 6.95 [1] 5.407(4) 5.454(4) 20.54(2) — — [6] 5.402(2) 5.453(7) 20.537(1) — — [7] 5.412 5.456 20.94 — — [9] 5.41 5.46 20.54 — — [10] 5.3961(6) 5.4498(5) 20.522(2) Fmmm 6.92 [11] 5.393(2) 5.451(2) 20.54(1) Fmmm 6.93 [12] nitrate route 5.400(2) 5.452(2) 20.52(1) Fmmm 7.03 [12] citrate route 5.3922(1) 5.4488(1) 20.5288(6) Fmmm 6.92 [13] 5.3928(1)* 5.4486(1)* 20.5185(5)* Amam* 7.02 [14] 5.3971(2)* 5.4501(2)* 20.507(1)* Amam* 7.05 [15] 20.502(1)* 5.4494(7)* 5.3981(7)* Cmcm* — [16] 5.409(2) 5.452(2) 20.537(3) — 6.72 [17] * neutron diffraction at 300 K. Space groups Amam and Cmcm differ only by the setting of the crystallographic axes. 54 Experimental In order to prepare the La3Ni2–xFexO7±δ (x  = 0, 0.05, 0.10) powder samples for our study, stoichiometric amounts of La(NO3)3 · 6H2O (chemical pure grade), Ni(CH3COO)2 · 4H2O (chemical pure grade), and of metallic iron were dissolved in an aqueous solution of nitric acid pre- pared from concentrated nitric acid (extra pure grade) and deionized water in the ra- tio 1:3 while heating and mixing on a hot plate. Metallic iron enriched in 57Fe was obtained by reduction of Fe2O3 (with iron enriched by 96.6 % in  the  isotope 57Fe) in  flowing H2 at 925  K for 6  hours and then quantitatively transferred into solu- tion. Subsequently, citric acid monohydrate C6H5O4(OH)3 · H2O (chemical pure grade) was added to  the  solution as  a  chelating and gelling agent. The thus-prepared pre- cursor solution was dried and decomposed forming a brown powder. The powder was calcined in  air at 1273  K for 30  minutes to  remove organic and carbon residuals and then cooled down to  RT. The  dark grey product was ground, pressed in- to pellets and annealed in  air at 1373  K for 24 h. In order to obtain single-phase La3Ni2–xFexO7±δ materials (x  = 0, 0.05, 0.10), 5–6 intermediate cycles of regrind- ing, pelletizing and annealing at 1373  K were required. Sample purity and morphol- ogy was checked by  REM/EDX analysis (Jeol JSM-6510, Bruker Nano XFlash 410). The measurements confirmed the absence of  any impurities and secondary phases in noticeable concentrations. Phase composition and crystal structure of the synthesized samples were examined by powder X-ray diffraction using an Equi- nox 3000 diffractometer (INEL, France) with Cu-Kα radiation (λ  = 1.54178  Å); data were collected in  the  asymmetric reflection mode. The  scattered radiation from the flat-plate samples was registered by  a  curved position-sensitive detector (CPS-590) within the angle interval 10–90° in 2θ. The  detector was calibrated using Na2Al2Ca3F14 as standard. The data acqui- sition time was 1 h for phase analysis and 17–21 h for structural analysis. All calculations including the Rietveld refinements and determinations of crys- tal structure of samples studied were per- formed using the  FullProf package [27]. The following structural parameters were varied in the refinement procedure: scale factor, unit cell parameters, zero shift, atomic positions, overall displacement pa- rameters (B-factors, Bov), background, peak shape, and width parameters. The  peak profiles were described by  the  Thomp- son-Cox-Hasting pseudo-Voigt function within the anisotropic strain broadening model. The instrumental resolution func- tion was determined by means of diffrac- tion experiments with a  CeO2 reference sample which had been annealed twice at 1473 K for 48 h with intermediate regrind- ing and then slowly cooled down to  RT. The background was estimated by linear interpolation between manually selected background points with refinable heights. Transmission 57Fe Mössbauer spectra were taken using a  standard Mössbauer system (Halder) in  the  sinusoidal driv- ing mode employing a 57Co/Rh γ-source with a  maximum activity of 1.91  GBq. The relatively low iron content and high electronic mass-absorption coefficient of La3Ni2–xFexO7±δ for the 14.4 keV Möss- bauer γ-radiation necessitated the  use of  samples enriched in 57Fe, see above. The least square fits of the spectra using lorentzian line shapes were performed by  means of  the  Recoil Mössbauer data evaluation software [28] yielding the fol- 55 lowing parameters: isomershift IS (mm·s‒1) relative to  α-Fe at  RT, quadrupole split- ting QS (mm·s‒1) of  doublets, full width Γ (mm·s‒1) at half maximum of lines, and (relative) spectral area A (%) of spectra. Mössbauer spectroscopy The  RT 57Fe Mössbauer spectra of La3Ni2–xFexO7±δ (x = 0.05, 0.10) are dis- played in Fig. 1. The highly symmetric spectra are com- posed of  two quadrupole-split doublets of almost identical intensity. This observa- tion gives evidence of the fact that iron cat- ions are incorporated on two inequivalent lattice sites which at present will be denot- ed by A and B for the site with the smaller and the larger quadrupole splitting (QS), respectively, QS(FeA) < QS(FeB). The pa- rameters obtained from the  least square fits of the spectra are presented in Table 2. The isomer shifts (IS) of 0.31 mm·s–1 of the two subspectra of La3Ni2–xFexO7±δ are identical in the two solid solutions studied (x = 0.05 and 0.10). It is also to be noted that these values are in close agreement with the 57Fe isomer shifts observed for the other octahedrally coordinated Lan+1NinO3n+1 R-P phases with n = 1 and n = 3 [21–26]. According to numerous Mössbauer stud- ies, e. g. Refs. [29, 30], such IS-values are characteristic for high-spin Fe3+ ions octa- hedrally coordinated by oxygen. The  RT quadrupole splittings (QS) of the two doublets in La3Ni2–xFexO7±δ are found to  differ significantly with values of about 0.43 mm·s–1 and 0.95 mm·s–1 for iron incorporated on sites of type A and B, respectively, Table 2. As also seen, Fig. 1, transmission can assume values as  low as  about 90 % in  parts of  the  spectra. Therefore, the absorbers used in the pre- sent study have to  be characterized as “thick”. This conclusion is also supported by the RT line widths of the two samples which take values ranging between 0.26 and 0.31  mm·s–1 and, thus, reveal sig- nificant broadening in  comparison with the theoretical width of a 57Fe Mössbau- er line of  about 0.20  mm·s–1 for “thin” Fig. 1. 57Fe Mössbauer spectra of La3Ni2−xFexO7±δ (x = 0.05 and 0.10) at RT Table 2 Parameters of the RT 57Fe Mössbauer spectra of La3Ni2−xFexO7±δ (x = 0.05 and 0.10): IS — isomer shift (vs α-Fe), QS — quadrupolar splitting, Г — full line width at half maximum (FWHM), A — experimental area fraction Compound IS (mm·s−1) QS (mm·s−1) Γ (mm·s−1) A (%) site La3Ni1.95Fe0.05O7±δ 0.31 0.41 0.30 50 A 0.31 0.97 0.26 50 B La3Ni1.90Fe0.10O7±δ 0.31 0.46 0.31 47 A 0.31 0.93 0.28 53 B 56 absorbers. In addition to line broadening, the thickness effect can also give rise to in- tensity distortion in the spectra as will be discussed below. Predominantly, low transmission and broadening of the experimental Mössbauer lines can be attributed to the substitution of nickel by iron strongly enriched in 57Fe (96.6 %) which causes large values for the total absorber thickness ta of the sam- ples. In the present case with 57Fe on two sites of the structure, ta it is defined as ta = (fA NA + fB NB)∙σ0 [31]. Here, fε represents the Debye-Waller factor of atoms of type ε (ε = A, B) and Nε their number per cm 2 of absorber; σ0 = 2.56·10 –18 cm2 is the ab- sorption cross section of  a  Mössbauer atom. In the present work, an estimate of 0.75 can be assumed for the  RT Debye- Waller factor. Using this approximate value, estimates of about ta = 4.0 and 8.0 are obtained for total absorber thicknesses of the La3Ni2–xFexO7±δ samples with x = 0.05 and 0.10, respectively, Fig. 1. For non-overlapping doublets of  lor- entzian shape, these ta-values would give rise to  theoretical line widths of  about 0.23 mm·s–1 and 0.25 mm·s–1 for x = 0.05 and 0.10, respectively, see e. g. Refs. [29, 31]. However, as seen, Fig.1, doublets are overlapping to  various degree. Notably the inner doublet A is almost fully over- lapped by  doublet B. In  the  limit of  full overlap, theoretical line widths amount to about 0.25 mm·s–1 and 0.30 mm·s–1 for x = 0.05 and x = 0.1, respectively [29, 31]. Indeed, in agreement with these expecta- tions, line widths observed for the inner doublet, A, are always larger than those of doublet B, Table 2. Remaining discrep- ancies between theoretical and experi- mental line widths can easily be attributed to the fact that quadrupolar interactions of the iron nuclei possess a distribution, see below, which provides an  additional source for line broadening. As  a  further consequence of  the  present large values of total absorber thickness ta, spectra un- dergo intensity distortion which always causes an underestimation of large spectral contributions on the basis of experimen- tal signal areas. As has been demonstrated by  Rancourt [31], such intensity correc- tions can be very large. In the present case, however, they are almost nonexistent (x = 0.05) or very moderate (~ 2 % for x = 0.10) due to the (almost) equal areas of the two experimental doublets. Thus, in the present case corrections to experimental signal in- tensities can well be neglected. Crystal structure As  discussed in  the  Introduction, to date the crystal structure of La3Ni2O7±δ has been assigned to  the  orthorhombic space groups Fmmm (№ 69) [1, 11–13] or Cmcm (№ 63) [14, 15]. According to these structural models, lanthanum cations occu- py two different crystallographic positions, but the nickel cations are located on a sin- gle site exclusively. This is in striking con- trast with the present results of the Möss- bauer measurements which clearly indicate the  existence of  two nonequivalent lat- tice sites for iron/nickel in  the  studied La3Ni2–xFexO7±δ solid solutions at  RT. In order to explain the unexpected obser- vations made in Mössbauer spectroscopy for the  n  =  2 member of  the  R-P series, in  the  following an  attempt will be un- dertaken to search for adequate structural models of La3Ni2–xFexO7±δ (x ≥ 0) account- ing for two inequivalent Ni/Fe sites. The XRD patterns of the  synthesized La3Ni2–xFexO7±δ (x  = 0, 0.05, 0.10) pow- der samples were indexed in orthorhom- 57 bic symmetry. Unit cell parameters a = 20.52298 Å, b = 5.45805 Å, c = 5.40101 Å, (α = β = γ = 90°) of La3Ni1.9Fe0.1O7±δ were obtained with the help of the DICVOL04 program [32] and the figures of merit pro- vided by this analysis amounted to M(20) = 56.2, F(20) = 60.7. Calculations by means of the ITO program [33] resulted in a pref- erable face-centered space group with al- most the same unit cell parameters of a = 5.4577Å, b = 20.5225Å, and c = 5.4020 (α = β = γ = 90°), and a figure of merit F(20) = 59.3. After preliminary profile-matching refinement within primitive orthorhom- bic Pmmm, probable space groups were searched for by means of the CheckGroup program interfaced by WinPlotr [34]. Thus, 48 possible space groups were obtained for consideration. Most of them are primitive with quite low values for figures of merit (<1.35). In  order to  reduce the  number of space groups, we focused on those which can be deduced from group-subgroup relations keeping both unit cell dimen- sions and orthorhombic crystal system of the parent group unchanged. Referring to  International Tables for Crystallogra- phy [35] only 6 subgroups remained: Ccca (No  68), Cmma (No  67), Cccm (No  66), Cmmm (No  65), Cmca (No  64), and Cmcm (No  63). Bearing in  mind the  re- sults of Mössbauer spectroscopy that reveal the existence of two nonequivalent sites for iron/nickel ions, the list of possible space groups shortens to three: Cmmm (No 65), Cmma (No 67), and Bmmb (non-standard setting of Cmcm space group) (No 63). The next step of our study was an at- tempt to determine the crystal structure of  La3Ni1.9Fe0.1O7±δ from XRD patterns by  constructing the  Patterson maps us- ing observed structure amplitudes and the peak search procedure of the GFourier program [34]. The  Patterson maps were constructed from the structural amplitudes extracted during the profile-matching fit according to the Le Bail algorithm for each chosen space group. Note that the  simplest crystal struc- ture solutions should be expected for both Fmmm and Cmmm models because these space groups correspond to exactly the same Patterson space groups Fmmm and Cmmm [35]. This means that one has exactly the  same symmetry both in  real space and Patterson space. As  far as  for X-ray diffraction, both nickel and iron atoms are indistinguishable and, thus, La3Ni1.9Fe0.1O7±δ contains three types of at- oms with substantially different numbers of electrons. Consequently, the observed peaks on the Patterson map might be re- solved except for those having the  same interatomic distances, for example the La1- La1 and Ni-Ni bond lengths in the perovs- kite layers of the Fmmm structure. Addi- tionally, if the heaviest atom (La) is placed into the origin of the unit cell with coor- dinates (0, 0, 0) — as we did in the present case — then the peaks observed in Patter- son space relate to interatomic vectors with coordinates (u = ±{x-0}, v = ±{y-0}, w = ±{z-0}) and coincide with atomic positions (±x, ±y, ±z) in real space. Accounting for these considerations, we were able to solve the  crystal structure for La3Ni1.9Fe0.1O7±δ in the framework of Fmmm and Cmmm space groups directly from the  peaks observed in  Patterson maps. The  results of  the  crystal structure solutions for La3Ni1.9Fe0.1O7±δ, which are summarized in Table 3, confirm our above mentioned expectations. As can be seen from Table 3, only a sin- gle crystallographic position is  available for nickel/iron cations within the Fmmm crystal structure of La3Ni1.9Fe0.1O7±δ. This is  in  agreement with the  earlier pro- 58 posed Fmmm space group for undoped La3Ni2O6.92 by Zhang et al. [11]. The main difference is that we place lanthanum into position (4a) at the origin of the unit cell (0, 0, 0). Zhang puts one of  the  oxygen anions in that position (4a) and the lan- thanum cations in position 4b with rela- tive coordinates (0, 0, 0.5). In  contrast, the Cmmm space group possesses two dif- ferent crystallographic sites (4k and 4l) for nickel cations. Thus, the Cmmm model can provide an explanation for the observed RT Mössbauer spectra of the La3Ni2–xFexO7±δ solid solutions (x = 0.05, 0.1). In  addition, it is  worth noting that the  space groups under consideration are subgroups of the space group Fmmm. Thus, if the  Wyckoff positions of  each atom in  Fmmm are known, we can eas- ily deduce the Wyckoff positions of each atom in  the  corresponding subgroup. The transformation of atomic coordinates from Fmmm (Table 3) to the correspond- ing subgroups was performed by the Pow- Table 3 Crystal structure solutions for La3Ni1.9Fe0.1O7±δ according to Fmmm and Cmmm models obtained from Patterson maps using peak search procedure of GFourier program [34] Peak No Relative atomic coordinates Occupation Peak height Wyckoff position Type of atomX Y Z Fmmm (69): a = 5.4020Å, b = 5.4570Å, c = 20.5170 Å general multiplicity 32 1 0 0 0 0.125 9872 4a La1 2 0 0 0.1837 0.25 6488 8i La2 3 0 0 0.4067 0.25 4024 8i Ni 4 0 0 0.5 0.125 1293 4b O1 5 0 0 0.3076 0.25 1238 8i O2 6 0.25 0.25 0.0922 0.5 1067 16j O3 Cmmm (65): a = 5.4014Å, b = 5.4576Å, c = 20.5182 Å general multiplicity 16 1 0 0 0 0.125 9892 2a La1 2 0 0.5 0.5 0.125 8427 2c La2 3 0 0.5 0.3158 0.25 6163 4l La4 4 0 0 0.1830 0.25 5778 4k La3 5 0 0.5 0.0907 0.25 3864 4l Ni2 6 0 0 0.4035 0.25 3855 4k Ni1 7 0 0.5 0.1990 0.25 1105 4l O4 8 0.25 0.25 0.0925 0.5 1028 8m O6 9 0.25 0.25 0.4030 0.5 992 8m O5 10 0 0.5 0 0.125 937 2b O1 11 0 0 0.4922* 0.25* 987 4l* O2 12 0 0 0.2992 0.125 881 4k O3 * here Z should be 0.5, the occupancy equal to 0.125 and the Wyckoff position should be 2d according to the chemical composition La3Ni1.9Fe0.1O7 59 derCell program [36] in  order to  reveal the possibility for the appearance of non- equivalent positions for nickel cations. As a result, of the six subgroups considered, only three would meet the latter criterion, namely Cmma (№ 67), Cmmm (№ 65), and Bmmb (№ 63) — a non-standard setting of the Cmcm space group. The Rietveld analysis of XRD patterns for La3Ni2–xFexO7±δ within the  suggested crystal structure models as  well as  thor- ough search for additional reflections with even tiny intensities which could definitely be attributed to a particular space group did not allow us to arrive at a final con- clusion. As an example, Fig. 2 illustrates the  refined Rietveld powder profiles of  La3Ni2–xFexO7±δ solid solutions within space groups Cmmm (A), Cmma (B), and Bmmb (C). Table 4 summarizes R-factors and chi-square values (χ2) for all refined patterns in the framework of the proposed structural models with two inequivalent sites for Ni/Fe  — space group Fmmm which provides only a single site for Ni/Fe is included for comparison. On the basis of these R-factor and χ2 data, the following observations can be made: (i) Fmmm ap- pears the least probable space group, (ii) fits within Cmmm and Cmma appear slightly superior to Bmmb, (iii) no significant dif- ferences can be observed between Cmmm and Cmma. Therefore, no valid conclu- sion can be drawn in respect to the true space group of La3Ni2–xFexO7±δ on the basis of the present X-ray diffraction data. Table  5 and Table  6 report the  re- sults of  the  Rietveld refinements for La3Ni2–xFexO7±δ in the two most probable space groups Cmma and Cmmm, respective- ly. The crystal structures of La3Ni2–xFexO7±δ (x  = 0, 0.05, 0.10) solid solutions con- fined to space groups Cmmm, Cmma, and Bmmb with two inequivalent sites (Ni1 and Ni2) for Ni/Fe are shown in  Fig.  3. For reason of comparison, the graph also shows the structural model for space group Fmmm. As can be seen from Fig. 3, in all cases the  lattice is  built up from double perovskite layers 2[La(Ni/Fe)O3] which are stacked between single rock-salt LaO layers along the c-axis. In Cmmm, Cmma, and Bmmb, the double perovskite layers consist of two different Ni/FeO6 octahedra (Ni1, Ni2) with slightly different distortions. Within the double perovskite layers the dif- ferently distorted octahedral Ni/Fe sites are arranged in different ways, Fig. 3B-D. Fig. 2. Rietveld refined XRD powder patterns of La3Ni2−xFexO7±δ solid solutions within space groups Cmmm for x = 0 (A), Cmma for x = 0.05 (B), and Bmmb for x = 0.10 (C) 10 20 30 40 50 60 70 80 90 0 2000 4000 6000 8000 10000 In te ns ity , a .u . 2θ, degree Observed Calculated Yobs – Ycal Bragg positions Observed Calculated Yobs – Ycal Bragg positions Observed Calculated Yobs – Ycal Bragg positions A 10 20 30 40 50 60 70 80 90 -5000 0 5000 10000 15000 20000 25000 30000 35000 In te ns ity , a .u . 2θ, degree B 10 20 30 40 50 60 70 80 90 0 10000 20000 30000 40000 50000 In te ns ity , a .u . 2θ, degree C 60 Finally, a- and b-parameters and the unit cell volume V of the La3Ni2–xFexO7±δ solid solutions are found to increase with Fe con- tent (x), (see Tables 5 and Table 6) which are in good agreement with size factor for 3d-metal cations: r(Fe3+) = 79 pm > r(Ni3+) = 74 pm [37]. Table 5 Rietveld refined atomic coordinates and bond lengths L(Ni-O) for La3Ni2−xFexO7 (x = 0, 0.05, 0.10) solid solutions within the Cmma model: 4g: La1 — (0,0.25,0.25), O3 — (0,0.25,0.75), La2,3/Ni1,2/O1,2 — (0,0.25,Z); 8l:O4,5 — (0.25,0,Z), RT Refined parameter x = 0 x = 0.05 x = 0.10 zero point 0.025(1) -0.0044(8) 0.0038(8) Bov, Å 2 1.66(4) 1.50(3) 1.58(3) a, Å 5.3895(1) 5.3970(1) 5.4019(1) b, Å 5.4462(1) 5.4521(1) 5.4574(1) c, Å 20.5250(3) 20.5280(2) 20.5192(2) V, Å3 602.46(1) 604.04(1) 604.91(1) Z(La2) 0.06994 0.06994 0.07007 Z(La3) 0.43006 0.43006 0.42993 Z(Ni1) 0.8476(2) 0.8472(1) 0.8469(1) Z(Ni2) 0.6524(2) 0.6528(1) 0.6531(1) Z(O1) 0.9526(6) 0.9454(7) 0.94918 Z(O2) 0.5474(6) 0.54719 0.55082 Z(O4) 0.6448(9) 0.6415(5) 0.6439(6) Table 4 Comparison of R-factors and χ2-values obtained for structural models of La3Ni2−xFexO7±δ solid solutions after Rietveld refinement Iron content Space group Rp Rwp Rexp χ 2 x = 0 Fmmm 4.65 6.02 4.56 1.74 Cmma 4.38 5.71 4.59 1.55 Cmmm 4.36 5.69 4.58 1.54 Bmmb 4.51 5.86 4.58 1.64 x = 0.05 Fmmm 2.85 3.89 2.19 3.16 Cmma 2.55 3.51 2.17 2.62 Cmmm 2.56 3.54 2.17 2.68 Bmmb 2.64 3.66 2.18 2.83 x = 0.1 Fmmm 2.95 4.17 1.90 4.82 Cmma 2.84 3.99 1.89 4.46 Cmmm 2.96 4.15 1.91 4.72 Bmmb 2.82 4.03 1.89 4.53 61 Table 6 Rietveld refined relative atomic coordinates and bond-lengths L(Ni-O) for La3Ni2−xFexO7 (x = 0, 0.05, 0.10) solid solutions within the Cmmm model: 2a: La1 — (0, 0, 0), 2b: O1 — (0, 0.5, 0) 2c: La2 — (0,0.5,0.5), 2d:O2 – (0, 0, 0.5), 4k: La3/Ni1/O3 — (0, 0, Z), 4l: La4/Ni2/O4 — (0, 0.5, Z), 8m: O5/O6 — (0.25, 0.25, Z), RT Refined parameter x = 0 x = 0.05 x = 0.10 zero point 0.025(1) -0.0043(8) 0.0038(8) Bov, Å 2 1.66(4) 1.47(3) 1.61(3) a, Å 5.3895(1) 5.3971(1) 5.4019(1) b, Å 5.4462(1) 5.4521(1) 5.4574(1) c, Å 20.5250(3) 20.5279(2) 20.5191(2) V, Å3 602.46(1) 604.044(8) 604.91(1) Z(La3) 0.18006 0.18016 0.17993 Z(La4) 0.31994 0.31984 0.32007 Z(Ni1) 0.4024(2) 0.4029 (1) 0.4031(1) Z(Ni2) 0.0976(2) 0.0971 (1) 0.0969(1) Z(O3) 0.298(1) 0.30079 0.2952(9) Z(O4) 0.20267 0.2049 (8) 0.19918 Z(O5) 0.3943(9) 0.4016 (3) 0.4021(4) Z(O6) 0.087(1) 0.0984 (3) 0.0979(4) L(Ni1/Ni2-O2/O1), Å 2.003(4) 1.994(2) 1.989(3) L(Ni1-O3), Å 2.15(3) 2.095(2) 2.21(2) 4L(Ni1/Ni2-O5/O6), Å 1.923(2) 1.9181(1) 1.9198(1) Average L(Ni1-O), Å 1.974(4) 1.9603(5) 1.980(3) L(Ni2-O4), Å 2.157(4) 2.21(2) 2.099(3) Average L(Ni2-O), Å 1.979(1) 1.980(3) 1.944(4) Сontinuation of table 5 Refined parameter x = 0 x = 0.05 x = 0.10 Z(O5) 0.838(1) 0.8370(7) 0.8389(8) L(Ni1-O3), Å 2.003(4) 1.995(2) 1.988(3) L(Ni1/Ni2-O1/O2), Å 2.16(1) 2.01(1) 2.099(3) 4L(Ni1-O5), Å 1.926(2) 1.929(2) 1.927(2) Average L(Ni1-O), Å 1.977(2) 1.954(3) 1.9657(8) L(Ni2-O3), Å 2.003(4) 2.169(2) 1.988(3) 4L(Ni2-O4), Å 1.922(2) 1.932(1) 1.929(2) Average L(Ni2-O), Å 1.974(2) 1.9818(7) 1.967(5) 62 Fig. 3. Crystal structures of La3Ni2−xFexO7 solid solutions within space groups Fmmm (A), Cmmm (B), Cmma (C), and Bmmb (D): grey balls — La3+, blue and dark grey octahedra represent oxygen environments of nickel/iron cations on Ni1- and Ni2-sites, respectively 63 Discussion T h e   RT M ö s s b a u e r s p e c t r a of La3Ni2–xFexO7±δ reveal that the sites for iron are characterized by two significantly different quadrupolar interactions, Table 2. In general, the 57Fe Mössbauer quadrupole splitting is given by [29] 21 11 2 3zz QS = eQV + η (1) where the so-called asymmetry parameter η is defined as η = (Vxx – Vyy)/Vzz (0 ≤ η ≤ 1) and where Vαα (α = x, y, z) represent the electric field gradients (EFGs) at the nu- cleus in the principal axis system with |Vzz| ≥ |Vyy| ≥ |Vxx|. Q denotes the quadrupole moment of the first excited nuclear state of 57Fe and e is the proton charge. In the case of  Fe3+ ions possessing the  d5 electronic configuration it is reasonable to assume that the experimental quadrupole splittings are dominated by the so-called lattice contri- bution to the electric field gradient which is determined by the positions and charges of the ions surrounding the Mössbauer at- oms in the lattice. According to Eq. (1), the quadrupole interaction is sensitive to lo- cal symmetry around the nuclear probes in that η = 0 for local tetragonal symmetry and that the interaction vanishes for local cubic symmetry, Vzz = 0. Thus, the splittings reflect an asymmetry of the charge distribu- tion in the crystal lattice around the nuclear probes and notably also the local distortion of their coordination polyhedra. The RT Mössbauer data clearly reveal that the two sites, A and B, occupied by iron differ significantly in respect to the quadru- polar interactions experienced by the probe nuclei, QS(FeA) < QS(FeB), Table 2. Under the  assumption that iron substitutes for nickel, this experimental finding is in con- flict with the structural assignments made to  date for La3Ni2O7±δ, i. e. space groups Fmmm [1, 11–13], Amam [14], and Cmcm [15], which provide only one unique site for Ni/Fe in  the  structure. The  assump- tion made here that iron adopts nickel sites is strongly supported by successful predic- tions made in respect to the neighboring R-P phases La2Ni1–xFexO4+δ [21–24] and La4Ni3–xFexO10–δ [24,26]. Their 57Fe Möss- bauer spectra are composed  — as  pre- dicted on the  basis of  their layered R-P structures — of one and of two doublets, respectively. Further on, this assumption is also supported by crystal chemical ar- guments for the transition metal cations which make high-spin Fe3+ ions with their intermediate ionic radius of 79 pm a  perfect substitute for both high-spin Ni3+ (74  pm) and Ni2+ (83  pm) cations [37]. Thus, provided that iron substitutes for nickel and that one of the aforemen- tioned space groups applies, only one sig- nal is expected in the Mössbauer spectrum of La3Ni2–xFexO7±δ (x > 0) which is in con- trast with experiment. Two possible explanations can be of- fered for this unexpected observation of  two nonequivalent sites for Fe/Ni. The  first would be the  conjecture that the  La3Ni2–xFexO7±δ solid solutions at  RT adopt a lower symmetry than hitherto as- sumed. This approach has been followed in Crystal structure section, where possi- ble structural models have been indicated. The second explanation, which will be dis- cussed below, involves a local reconstruc- tion of the R-P layer structure. Indeed, a low-symmetry phase at RT cannot be unexpected in view of the fact that the mixed-valent La3Ni2–xFexO7±δ con- tains Ni3+ and Ni2+ cations in similar con- centrations 64 [Ni2+] = 1  2δ and [Ni3+] = 1 ± 2δ – x (2) where square brackets denote the number of  species per formula unit. The  validity of Eq. (2) is based on the assumption that iron is always in the trivalent charge state as revealed by the present Mössbauer spectra (see Mössbauer spectroscopy section) and discussion below. The close agreement of [Ni2+] and [Ni3+] could give rise to charge ordering and to the formation of some kind of lower symmetry structure. Indeed, charge ordering in La3Ni2O7–δ at low temperatures has already been discussed by Taniguchi et al. [38] where it was proposed that ordering of electronic charge on nickel was induced by ordering of oxygen vacancies. In the same context, electrical conductivity, which is intimately related with electronic mate- rials properties, has given clear evidence of a transition from temperature-activated behavior to metal-type conduction at high temperatures with reported transition tem- peratures of 550 K [1, 18] and 600 K [19]. As  evidenced by  the  RT Mössbauer spectra, Fe3+ cations are almost evenly dis- tributed between the  two nonequivalent Ni1- and Ni2-sites for Ni/Fe in the lattice, Table 2. This can easily be explained as fol- lows. During their synthesis, samples have been exposed to high temperatures (up to 1373 K) where according to the above dis- cussion there is only one unique site for Ni/Fe and the Fe dopant will be randomly distributed on this unique site. Electron- ic charge ordering on the  nickel cations leading to  less symmetric structure oc- curs at  much lower temperatures where Fe3+ cations are not mobile anymore and, due to their random distribution on avail- able sites, show up in  almost equal pro- portions on the Ni1- and Ni2-sites which occur in equal numbers in the alternative structures previously discussed, Tables 4–6 and Fig. 3. Due to their fixed valence (see Mössbauer spectroscopy section) and discussion below, Fe3+ cations are not part of  the  electronic ordering process and, thus, serve as “spectators” of the structural changes induced by the evolving electronic charge order-disorder. In  an  ideally stoichiometric system, where x = 0 and δ = 0, Ni2+ and Ni3+ ions are present in exactly equal numbers, Eq. (2), and distributed over the two nonequivalent sites, Ni1 and Ni2, of the structure. Thus, two limiting cases of the cation distribu- tion are to be considered. In one case — cation distribution (I) — all sites of Ni1- type are occupied by (all) Ni2+ ions and, correspondingly, all Ni2-sites are occupied by (all) Ni3+ ions. In the case of cation dis- tribution (II), all Ni1-sites are occupied by Ni3+ and all Ni2-sites by Ni2+. In both situations, different, but single-valued electric field gradients (EFG) are expected at the two inequivalent sites of the struc- ture due to the fully ordered distribution of charges in the lattice. However, already the  introduction of  Fe3+ ions into sites of  the  Ni2+ sublattice will create charge disorder which will cause a  distribution of EFGs at both sites leading to line broad- ening. The variable oxygen content — defi- cit or excess — is associated with chang- ing concentrations of oxygen anions and changing amounts of Ni3+ and Ni2+ cations. This will also contribute to charge disorder in the lattice and, hence, contribute to line broadening in mixed crystals with x > 0 and |δ| > 0. Thus, charge disorder — in ad- dition to  absorber thickness  — explains the larger line widths for the sample with the higher dopant level, Table 2. In  an  alternative explanation for the  observation of  two sites for Fe/Ni in La3Ni2–xFexO7±δ, the possibility is to be 65 envisaged that a temperature-dependent reconstruction (intergrowth) of the Rud- dlesden-Popper structure may occur. Locally, three perovskite-type layers of iron/nickel atoms may be formed, like in La4Ni3–xFexO10–δ, which would provide two inequivalent sites for nickel (iron). With increasing temperatures, the  lo- cal three-layer intergrowth (n = 3) could change to  the  higher symmetric n = 2  two-layer structure of  La3Ni2O7. Such a  structural complexity could be a  con- sequence of  the  fact that for the  off-sto- ichiometric, iron-doped La3Ni2–xFexO7±δ (δ > 0, x > 0) the number of Ni3+ and Ni2+ cations no longer obeys the ideal 1:1 ratio but, instead, is given by [Ni3+]:[Ni2+] = 1  (x4δ)/(12δ). In this context, it is also interesting to recall Mössbauer results re- ported for La4Ni3–xFexO10–δ with x = 0.03 [25]. Here, the two quadrupole doublets at RT show splittings of 0.43 mm·s–1 and 0.95 mm·s–1 which is very close to the RT QS-values observed for La3Ni2–xFexO7±δ in  the  present study, Table  2. However, the  above scenario is  difficult to  accept in view of the fact that diffusional motion of all cations, La, Ni, and Fe, would be re- quired for changing the local intergrowth situation. This conclusion is also supported by experience gathered by many authors in the (sluggish) high-temperature synthe- sis of La3Ni2O7±δ, see e. g. Refs. [7, 8, 11]. The  discussion of  the  results from X-ray diffraction in Crystal structure sec- tion has led to  the  successful identifica- tion of several space groups which provide two nonequivalent sites for Ni/Fe in equal numbers. It must be admitted, however, that a final decision about the prevailing structure type could not be reached. There- fore, in the following another approach will be taken by considering the 57Fe quadru- polar interactions at the two inequivalent sites in La3Ni2O7. In this discussion it will be assumed that the  electric field gradi- ents at the iron nuclei, Eq. (1), arise from the so-called lattice contribution, Vαα(latt), i. e. from charges located on the ions sur- rounding the Mössbauer atom in the crys- tal lattice. Thus, the framework of the point charge model, the EFGs at the two sites for iron can be calculated in the principal axis system according to [29]: ( ) ( ) ( )2 2 5 0 1 (1 ) 3 4 i i i ii V V latt e z r r αα ∞ αα ∞ − = − γ = − γ = α − πε ∑ (3) Here, the  summation extends over the surrounding lattice ions of charge num- ber zi, and (1 – γ∞) and ε0 denote the Stern- heimer factor and the vacuum permittivity, respectively. The lattice EFGs were calcu- lated using the UNISOFT program pack- age [39] and quadrupole splittings were obtained by using a Sternheimer factor for Fe3+ ions of 1 – γ∞ = 10.42 [40] and a quad- rupole moment of Q = 0.209·10–28 m2 [41]. In  an  approximation to  the  complex situation in real La3Ni2–xFexO7±δ solid so- lutions, the EFGs at the Ni1- and Ni2-sites have been calculated for an  ideal rigid La3Ni2O7 lattice (x = 0, δ = 0) employing the crystallographic RT data of La3Ni2O7 obtained from the  Rietveld refinements partially reported in Tables 5 and 6. In all cases, the  assumption was made that the  formal charge number of  oxygen is –2. In the case of the Fmmm space group it was assumed that all Ni-sites are occu- pied by nickel cations possessing an aver- age charge of +2.5 — corresponding to no charge ordering. For the lower-symmetric space groups, the two above defined cation distributions, (I) and (II), have been con- sidered with charge numbers +2 for Ni2+ and +3 for Ni3+ and La3+. Table 7 reports the results of these calculations according 66 to Eqs. (1) and (3). It is to be noted that the  calculations neither account for Fe3+ cations introduced into the lattice by dop- ing nor for possible deviations from ex- act stoichiometry. Because of their small numbers and their random distribution, it can be expected that the  iron dopants and defects will not influence significantly the average electric field gradients at cation nickel sites, but will lead to a distribution of quadrupole interactions about the struc- ture-determined averages. As can be seen from Table 7 these ad- mittedly simplistic point charge calcula- tions, which also neglect lattice relaxation around the nuclear Fe probes and any effect of chemical bonding of the ferric cations, successfully yield 57Fe quadrupole splittings of the right order of magnitude, i. e. 0.1 < QS/mm·s–1 < 1. Calculated quadrupole splittings in the orthorhombic space groups Cmmm and Cmma, Table  7, are in  reasonable agreement with the experimental splittings of QS(FeB) ≈ 0.95 mm·s ‒1 and QS(FeA) ≈ 0.45 mm·s‒1, Table 2, for both types of cat- ion distributions. However, calculated QS-values are always found smaller than experimental ones by about 20 %, Table 7. In  view of  the  deficiencies of  the  point charge model in general and the simplify- ing assumptions made in the calculations, such deviations cannot be unexpected. In  contrast, for orthorhombic Bmmb no acceptable agreement is observed between experiment and calculated values. In addi- tion to consideration of the absolute values of quadrupolar interactions, special rele- vance is attributed to the ratio of splittings, RQ = QS(FeB)/QS(FeA), because of possi- ble error cancellation in the calculations. Table 7 Quadrupolar interactions in La3Ni2O7: 57Fe quadrupole splitting, QS, and splitting ratios RQ, at the nickel sites of La3Ni2O7 for space groups Fmmm, Cmmm, Cmma, and Bmmb and for the cation distributions, (I) and (II), which can be adopted by nickel in the lower symmetric structures. Calculations were performed in the framework of the point charge model according to Eqs. (1,3) with lattice parameters taken from respective Rietveld refinements Space group Cation distribution Site QS mm·s−1 RQ Fmmm — Ni 0.74 — Cmmm (I) Ni1 0.86 3.0 (I) Ni2 0.29 (II) Ni1 / A 0.36 2.2 (II) Ni2 / B 0.79 Cmma (I) Ni1 / B 0.81 2.3 (I) Ni2 / A 0.35 (II) Ni1 0.28 3.1 (II) Ni2 0.88 Bmmb (I) Ni1 0.83 1.1 (I) Ni2 0.75 (II) Ni1 0.92 1.4 (II) Ni2 0.66 67 At  RT, the  experimental ratio assumes values of  about RQ = 2.3 (x  = 0.05) and RQ = 2.0 (x = 0.10), Table 2. As can be seen from Table 7, such values are not consist- ent with those calculated for Bmmb where this ratio amounts to 1.1 or 1.4 for cation distributions (I) and (II), respectively. For Cmmm(I) as well as for Cmma(II) calcu- lated ratios are significantly larger than the experimental RQ-values. However, for Cmmm(II) and for Cmma(I) close agree- ment is  observed with calculated ratios of 2.2 and 2.3, respectively, indicated in “bold” in Table 7. Because we see no fur- ther means for differentiation between the two structures, we arrive at the conclu- sion that space group Cmmm with a cation distribution of  type (II) as  well as  space group Cmma with cation distribution (I) represent good structural models for La3Ni2–xFexO7±δ solid solutions at RT. In  the  present combined approach involving diffraction and spectroscopic work, A- and B-sites can be even be as- signed to the Ni1- and Ni2-sites in these space groups according to the quadrupole splittings calculated for the  respective types of cation distribution, see Table 7. This conclusion on the  crystal structure of La3Ni2–xFexO7±δ, which is in agreement with the space groups arrived at on the basis of general structural considerations and Ri- etveld refinements of the materials, narrows down considerably the number of possible combinations of space groups and cation distributions of La3Ni2–xFexO7±δ at RT. Conclusions At  RT, local probe 57Fe Mössbauer spectroscopy has provided indisputable evidence for the  existence of  two non- equivalent sites for Fe3+ in La3Ni2–xFexO7±δ with site populations closely correspond- ing to the 1:1 ratio. This unexpected re- sult is in contrast to the hitherto existing structure determinations of  La3Ni2O7±δ. The latter assumed space groups Fmmm or Cmcm (Amam) which both provide only a single unique site for nickel/iron. In view of the spectroscopic findings, the RT struc- ture of La3Ni2–xFexO7±δ has been reconsid- ered and X-ray diffraction patterns for samples with x = 0, 0.05, and 0.10 have been measured and refined by  the  Riet- veld method. The interest was in the iden- tification of alternative structural models which could replace the previously used space groups. From general considerations, we have selected the orthorhombic space groups Cmma, Cmmm, and Bmmb which fulfill the  spectroscopic requirements of providing two sites in equal numbers for Ni/F. According to Rietveld refinements, Cmma and Cmmm can be considered the preferred space groups. Calculations of the quadrupolar interactions at the iron nuclei in  La3Ni2O7 in  the  framework of the point charge model have confirmed this conclusion and provided additional information on the  specific distribution of Ni2+ and Ni3+ cations in these structures at RT. Acknowledgements K.-D.B. and P.G. would like to  thank the  Volkswagen-Foundation and the  State of Lower Saxony (Germany) for financial support of the present work. E.K. and V.C. ac- knowledge support of their work by the Ministry of Education and Science of the Russian Federation. A.F. is grateful to the Deutsche Forschungsgemeinschaft (DFG) for finan- cial support in the frame of grant FE928/7–1. Thanks are also due to Prof. Th. Gesing 68 (U Bremen, Germany) for chemical EDX analyses of samples. E.K and V. Ch. work was supported by Act 211 Government of the Russian Federation, agreement 02.A03.21.0006. References 1. Amow G, Davidson IJ, Skinner SJ. A comparative study of the Ruddlesden-Popper series, Lan+1NinO3n+1 (n = 1, 2 and 3), for solid-oxide fuel-cell cathode applications. Solid State Ionics. 2006;177(13–14):1205–10. DOI: 10.1016/j.ssi.2006.05.005. 2. Goettler R, Xing Z, Xue L, Hill M. Evaluation of Ruddlesden-Popper nickelate cath- odes for higher temperature SOFC. Ceramic Engineering and Science Proceedings. 2008;28(4):183–94. DOI: 10.1002/9780470339534.ch18. 3. Takahashi S, Nishimoto S, Matsuda M, Miyake M, Electrode properties of the Rud- dlesden–Popper series, Lan+1NinO3n+1 (n= 1, 2, and 3), as intermediate-temperature solid oxide fuel cells. J. Am. Ceram. Soc. 2010;93(8):2329–33. DOI: 10.1111/j.1551–2916.2010.03743.x. 4. Woolley RJ, Skinner SJ. Novel La2NiO4+δ and La4Ni3O10–δ composites for solid oxide fuel cell cathodes. J. Power Sources. 2013;243:790–95. DOI: 10.1016/j.jpowsour.2013.06.106. 5. Woolley RJ, Skinner SJ. Functionally graded composite La2NiO4+δ and La4Ni3O10–δ solid oxide fuel cell cathodes. Solid State Ionics. 2014;255:1–5. DOI: 10.1016/j.ssi.2013.11.041. 6. Brisi C., Vallino M, Abbattista F. Composition and structure of two hitherto uni- dentified phases in the system La2O3-NiO-O. J. Less Common Metals, 1981;79(2): 215–19. DOI: 10.1016/0022–5088(81)90070–9. 7. Drennan J, Tavares CP, Steele BCH. An electron microscope investigation of phases in the system La–Ni–O. Mat. Res. Bull. 1981;17(5):621–26. DOI: 10.1016/0025–5408(82)90044–7. 8. Odier P, Nigara Y, Coutures J, Sayer M. Phase relations in the La–Ni–O system: In- fluence of temperature and stoichiometry on the structure of La2NiO4. J. Solid State Chem. 1985;56(1):32–40. DOI: 10.1016/0022–4596(85)90249-X. 9. Ram RM, Ganapathi L, Ganguly P, Rao CNR. Evolution of three-dimensional char- acter across the Lan+1NinO3n+1 homologous series with increase in n. J. Solid State Chem. 1986;63(2):139–47. DOI: 10.1016/0022–4596(86)90163–5. 10. Sreedhar K, McElfresh M, Perry D, Kim D, Metcalf P, Honig JM. Low-temperature electronic properties of  the  Lan+1NinO3n+1 (n  = 2, 3, and∞) system: evidence for a crossover from fluctuating-valence to Fermi-liquid-like behavior. J. Solid State Chem. 1994;110(2):208–15. DOI: 10.1006/jssc.1994.1161. 69 11. Zhang Z, Greenblatt M, Goodenough JB. Synthesis, structure, and properties of the layered perovskite La3Ni2O7–δ. J. Solid State Chem. 1994;108(2):402–09. DOI: 10.1006/jssc.1994.1059. 12. Carvalho MD, Costa FMA, Pereira IS, Wattiaux A, Bassat JM, Grenier JC, Pouchard M. New preparation method of Lan+1NinO3n+1–δ (n= 2, 3). J. Mater. Chem. 1997;7(10):2107–11. DOI: 10.1039/A702424J. 13. Sasaki H, Harashina H, Taniguchi S, Kasai M, Kobayashi Y, Sato M, Sakata M. Struc- tural studies on the phase transition of La3Ni2O6.92 at about 550, K. J. Phys. Soc. Jpn. 1997;66(6):1693–97. DOI: 10.1143/jpsj.66.1693. 14. Ling CD, Argyriou DN, Wu G, Neumeier JJ. Neutron diffraction study of La3Ni2O7: Structural relationships among n = 1, 2, and 3 phases Lan+1NinO3n+1. J. Solid State Chem. 2000;152(2):517–25. DOI: 10.1006/jssc.2000.8721. 15. Voronin VI, Berger IF, Cherepanov VA, Gavrilova LY, Petrov AN, Ancharov AI, Nikitenko SG. Neutron diffraction, synchrotron radiation and EXAFS spectroscopy study of crystal structure peculiarities of the lanthanum nickelates Lan+1NinOy (n = 1, 2, 3). Nucl. Instrum. Methods Phys. Res., Sect. A. 2001;470(1–2):202–209. DOI: 10.1016/S0168–9002(01)01036–1. 16. Bannikov DO, Cherepanov VA. Thermodynamic properties of  complex oxides in the La–Ni–O system. J. Solid State Chem. 2006;179(8):2721–27. DOI: 10.1016/j.jssc.2006.05.026. 17. Savchenko VF, Ivashkevich LS,. Lubkina IYa, Russ. J. Inorg. Chem. 1988;33:30–33. 18. Kobayashi Y, Taniguchi S, Kasai M, Sato M, Nishioka, Kontani M. Transport and mag- netic properties of La3Ni2O7–δ and La4Ni3O10–δ. J. Phys. Soc. Jpn. 1996;65(12):3978–82. DOI: DOI.org/10.1143/jpsj.65.3978. 19. Gervais F, Odier P, Nigara Y. Plasmon behavior at the “semiconductor-metal” transi- tion in La2NiO4 and La3Ni2O7. Solid State Commun. 1985;56(4):371–74. DOI: 10.1016/0038–1098(85)90405–3. 20. Zinkevich M, Solak N, Nitsche H, Ahrens M, Aldinger F. Stability and thermody- namic functions of lanthanum nickelates. J. Alloys Compd. 2007; 438(1–2):92–99. DOI: 10.1016/j.jallcom.2006.08.047. 21. Fontcuberta J, Longworth G, Goodenough JB. Magnetic order or charge-density wave in La2NiO4 by Mössbauer spectroscopy. Phys. Rev. B. 1984;30(11):6320. DOI: 10.1103/PhysRevB.30.6320. 22. Tsipis EV, Naumovich EN, Patrakeev MV, Waerenborgh JC, Pivak YV, Gaczyński P, Kharton VV. Oxygen non-stoichiometry and defect thermodynamics in La2Ni0.9Fe0.1O4+δ. J. Phys. Chem. Solids. 2007;68(7):1443–55. DOI: 10.1016/j.jpcs.2007.04.006. 23. Klande T, Efimov K, Cusenza S, Becker KD, Feldhoff A. Effect of doping, micro- structure, and CO2 on La2NiO4+δ-based oxygen-transporting materials. J. Solid State Chem. 2011;184(12):3310–18. DOI: 10.1016/j.jssc.2011.10.019. 70 24. Klande T, Cusenza S, Gaczyński P, Becker KD, Dörrer L, Borchardt G, Feldhoff  A. In-situ Mössbauer studies of 57Fe-doped Ruddlesden–Popper type lanthanum nickel oxides. Solid State Ionics, 2012;222–223:8–15. DOI: 10.1016/j.ssi.2012.06.019. 25. Carvalho MD, Wattiaux A, Ferreira LP, Bassat JM. Mössbauer investigation of 57Fe doped La4Ni3O10±y phases. J. Solid State Chem. 2009;182(1):60–64. DOI: 10.1016/j.jssc.2008.10.006. 26. Tsipis EV, Patrakeev MV, Waerenborgh JC, Pivak YV, Markov AA, Gaczyński P, Kharton VV. Oxygen non-stoichiometry of Ln4Ni2.7Fe0.3O10–δ (Ln = La, Pr). J. Solid State Chem. 2007;180(6):1902–10. DOI: 10.1016/j.jssc.2007.04.025. 27. Rodriguez-Carvajal J. Commission on powder diffraction (IUCr). Newsletter, 2001;26:12. 28. Lagarec K, Rancourt DG. RECOIL, Mössbauer spectral analysis software for windows (version 1.02). Ottawa: Department of Physics, University of Ottawa;1998. 29. Gütlich P, Bill E, Trautwein AX. Mössbauer Spectroscopy and Transition Metal Chemistry. Berlin, Heidelberg: Springer; 2011. 568 p. DOI: 10.1007/978–3–540–88428–6. 30. Menil F. Systematic trends of the 57Fe Mössbauer isomer shifts in (FeOn) and (FeFn) polyhedra. Evidence of a new correlation between the isomer shift and the inductive effect of the competing bond TX (→ Fe)(where X is O or F and T any element with a formal positive charge). J. Phys. Chem. Solids. 1985;46(7):763–89. DOI: 10.1016/0022–3697(85)90001–0. 31. Rancourt DG. Accurate site populations from Mössbauer spectroscopy. Nucl. In- strum. Methods Phys. Res., Sect. B. 1989;44(2):199–210. DOI: 10.1016/0168–583X(89)90428-X. 32. Boultif A, Louër D. Powder pattern indexing with the dichotomy method. J. Appl. Crystallogr. 2004;37(5):724–31 DOI: 10.1107/S0021889804014876. 33. Visser JW. A fully automatic program for finding the unit cell from powder data. J. Appl. Crystallogr. 1969;2(3):89–95. DOI: 10.1107/S0021889869006649. 34. Roisnel T, Rodríquez-Carvajal J. WinPLOTR: a windows tool for powder diffraction pattern analysis. In Materials Science Forum (Vol. 378, No. 1, pp. 118–123). Transtec Publications, 1999. 35. Hahn Th. International Tables for Crystallography, Vol. A: Space-group symmetry. 5th ed. Dordrecht: Springer, 2005. 911 p. 36. Kraus W, Nolze G. POWDER CELL-a program for the representation and manipu- lation of crystal structures and calculation of the resulting X-ray powder patterns. J. Appl. Crystallogr. 1996;29(3):301–03. DOI: 10.1107/S0021889895014920. 37. Shannon RD. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Cryst. 1976; A32:751–67. DOI: 10.1107/S0567739476001551. 71 38. Taniguchi S, Nishikawa T, Yasui Y, Kobayashi Y, Takeda J, Shamoto SI, Sato  M. Transport, magnetic and thermal properties of  La3Ni2O7–δ. J. Phys. Soc. Jpn. 1995;64(5):1644–50. DOI: DOI.org/10.1143/jpsj.64.1644. 39. Eckold G. UNISOFT — A Program Package for Lattice-Dynamical Calculations: User Manual, 2nd rev. ed. Jülich: Jülich Forschungszentrum, 1992. 40. Stevens JG, Dunlap BD. Nuclear moments and moment ratios as  determined by Mössbauer spectroscopy. J. Phys. Chem. Ref. Data, 1976;5(4):1093–122. DOI: 10.1063/1.555541. 41. Liverts EZ, Zhetbaev AK. Sternheimer Quadrupole Factors in Ferric Compounds. Phys. Status Solidi B. 1982;111(2):469–475. DOI: 10.1002/pssb.2221110207.