Structural chemistry of SrMn, -,Fe,O, x z 0.3 Peter D. Battle,*" Courtenay M. Davison," Terence C. Gibb*' and Jaap F. Vente" ahorganic Chemistry Laboratory, South Parks Road, Oxford, UK OX1 3QR 'School of Chemistry, University of Leeds, Leeds, UK LS2 9JT A polycrystalline sample of nominal composition SrMn,~,Fe,~,O, was prepared by standard ceramic techniques and investigated at room temperature by iodometry, 57Fe Mossbauer spectroscopy and neutron powder diffraction. The former two techniques established 6 =0.11kO.01. Neutron diffraction showed that the sample wasobiphasic. The major component [98.3(5)% by mass] was a 10H perovskite [space group P~~/wwzc; a= 5.45035(5), c =22.3735(2) A], with the Sr-0 layers adopting a cccchcccch stacking sequence. A full structural refinement showed the composition to be SrMn,~,2Feo~2802~8,. The minor phase was a pseudo-cubic perovskite.The Mossbauer data suggest that the oxygen vacancies in the major phase are predominantly found in the coordination shell of Fe3+ cations. The sample did not show long-range magnetic order at room temperature. The crystal structure of all perovskites ABX, can be considered to consist of pseudo-close packed layers of stoichiometry AX, with the smaller cation, B, occupying octahedral holes between the layers. Successive AX, sheets can stack in either cubic close packed (ccp) or hexagonal close packed (hcp) arrange- ments, and several mixed stacking sequences are possible.' Consequently, the coordination polyhedra around B can link together by sharing either a common face or a common vertex.Only the former type of linkage is present in BaMnO,, and only the latter in CaTiO,. In other compounds, for example 6H BaTiO,, both types of linkage occur in a periodic manner. Many factors, including the electronic structure of the cation B and the size ratio, rA:rB, play a role in determining the structure adopted by a particular compound.2 As a result of this, compounds which might be expected to be structurally similar, for example SrMnO, and SrFeO,, can adopt very different crystal structures. At room temperature, the former3 has a 4H perovskite structure in which Mn209 dimers are linked together by vertex sharing, whereas the latter4 is a cubic perovskite containing only vertex-sharing Fe06 groups.The large number of phase transitions observed as a function of temperature and pressure in many perovskite system^^.^ sug-gests that the stability difference between the different stacking sequences is often small. Furthermore, the sensitivity of the structure to cation size and electron configuration suggests that a compound AB1 -xBx'03 containing two different trans- ition-metal cations might adopt a structure which, although still based on the stacking of AO, layers, is different from that of both ABO, and AB'O,. The magnetic and electrical proper- ties of perovskite oxides are controlled in part by the structures they adopt and, in view of the current interest in the magneto- resistance of Mn-containing perov~kites,~ we have begun a study of the compounds SrMn, -xFe,O,.We describe below a structural study, carried out using neutron diffraction and Mossbauer spectroscopy, of a composition close to SrMn,.,Fe, 303-6. The way in which we have chosen to write the formula of this compound hints at another important parameter in the structural chemistry of perovskites, i.e. the vacancy concentration in the anion sublattice. Experimental A black polycrystalline sample of overall composition SrMn, 7Fe0.303-6 was prepared by firing a well ground, pel- letized, stoichiometric mixture of SrCO,, MnO, and Fe,O, (Johnson Matthey 'Specpure' reagents) in a platinum cruc- ible at 1250 "C for 72 h in air, then cooling it at a rate of 100°C h-' to 500°C before removing it from the furnace.As the formation of the final product took place slowly, this heating cycle was repeated eight times, with intermediate regrinding and pelletizing, to ensure that the composition was homogeneous, X-Ray diffraction data collected in Bragg-Brentano geometry using Cu-Ka radiation were used to assess the purity of the product, and the oxygen content was deter- mined by iodometric titration. Neutron powder diffraction data were collected at room temperature on the diffractometer Dla at the Institut Laue Langevin, Grenoble. The sample was contained in a 16 mm diameter vanadium can whilst the bank of ten detectors was scanned through the angular range 6<28/<160 degrees an steps of 0.05'. The mean neutron wavelength was 1.956 A. Mossbauer data were collected in the temperature range 78 <T/K <290 using a 57Co/Rh source matrix held at room temperature; isomer shifts were determined relative to the spectrum of metallic iron.Results The initial characterisation of our sample by X-ray diffraction and chemical analysis suggested that it was a monophasic hexagonal material of composition SrMn,~,Feo~,02,90. The unit-cell dimensions were characteristic of a 10H perovskite structure. There are a number of possible stacking sequences' for close-packed SrO, layers which result in a 10H structure, and we analysed the observed neutron diffraction in order to determine which sequence is present in this particu- lar compound. The following scattering lengths were used: b,, =0.702, b,, = -0.373, b,, =0.954 and bo = 0.5805 x m.Preliminary refinements suggested that SrMn,~,Fe,,,O, is isostructural with Ba5W3Li,0151' and the data were consequently analysed using the space group P63/mmc, with three crystallographically distinct sites for each of Sr, Mn/Fe, and 0 (Fig. 1). The SrO, sublattice consists of layers stacked in a mixed cubic (c) and hexagonal (h)sequence which can be described as cccchcccch.' The resulting crystal structure can be seen to contain blocks of pseudo-cubic perovskite, three octahedra wide, separated by B209 dimers (B=Mn/Fe). During the early stages of refinement it became clear that the face-sharing octahedra, which form the dimers, are occupied only by Mn whereas the vertex-sharing octahedra in the perovskite blocks are occupied by a disordered distri- bution of Mn and Fe.However, our initial refinements, in which the Mn:Fe ratio was constrained to be 7:3, were unsatisfactory and we therefore allowed the composition of J. Muter. Chem., 1996, 6(7), 1187-1190 1187 0 0 0 0 0 0 0 0 1OH SrMno~,FeO,,O,, Fig. 1 10H crystal structure of SrMn, 72Feo 280287. Open circles represent Sr atoms. the hexagonal phase to vary. There was a marked improvement in the agreement indices as the phase became Mn rich. The difference between the observed and calculated diffrac- tion profiles at this stage consisted mainly of positive intensities with a 28 distribution which suggested that they were due to the presepce of a cubic second phase with a unit-cell parameter ~~3.85A.We hypothesised that this minority phase [ca. 1.7(5)%], which had not been detected during our preliminary X-ray study, might be near-stoichiometric SrFeO, consist-ent with the Fe deficiency in the hexagonal phase. All sub- sequent analysis of the neutron diffraction data was carried out by treating the sample as a two-phase mixture. The peak shape was constrained to be the same for both phases. In these final refinements we were able to establish that the oxygen vacancies in the primary phase occur only on the O(3)site in the centre of the blocks of pseudo-cubic perovskite (Fig. 1). The group of profile parameters being varied at this stage comprised two scale factors, a counter zeropoint, four peak- shape parameters (pseudo-Voigt function), five background parameters (shifted Chebyshev function), two unit-cell param- eters for the hexagonal phase and one for the cubic phase, and a preferred orientation correction for the hexagonal phase.The other parameters (all associated with the hexagonal phase) involved in the refinement were the nine variable atomic coordinates, the occupancy factor of O(3),the overall isotropic temperature factor, and the Mn:Fe ratios on the sites Mn/Fe( 1) and Mn/Fe(2). The latter sites were constrained to be fully occupied but no constraint was imposed on the Mn :Fe ratios, The second phase was finally treated as stoichiometric SrFeO, because the low concentration gave rise to very weak Bragg peaks which could not be used sensibly in a detailed structure refinement.In reality the phase is likely to be oxygen deficient, non-cubic and, perhaps, to contain some Mn. It could be postulated that it is the residue of a cubic 'high- temperature' phase with the same overall composition as the principal component, but this seems unlikely because the observed X-ray pattern improved substantially with successive anneals and the ultimate crystallinity of the hexagonal phase was excellent. Although we are unable to draw any meaningful conclusions about the second phase from these data, its inclusion in our analysis, albeit with a crude structural model, smoothed the refinement of the structure of the primary phase.The resulting agreement factors were as follows: R, =6.91YO, R, =5.19%, R,( 10H)=7.40%, x,d2 =1.004, DW -d =0.31 (lower limit of 90% confidence level=1.9013). The final observed and calculated diffraction profiles are shown in Fig. 2. The structural parameters of the 10H phase, the composition of which refined to be SrMn0.72Fe0.2802.87, are listed in Table 1 and the resulting bond lengths and bond angles are given in Table 2. The 57Fe Mossbauer spectrum of SrMno~7Feo~,0, observed-at room temperature is shown in Fig. 3, along with the spectrum calculated from a four-doublet fit. This fit is to some extent arbitrary, as the disordered nature of the crystal struc- ture will cause broadening of the resonances, and indeed the symmetrical Lorentzian line profile used here may be inappro- priate.However, a minimum of four doublets is required to represent the subtlety of the absorption profile, and the com- puted linewidths of all four components are similar. The parameters resulting from the fit are listed in Table 3. They are quoted to only two decimal places as they are undoubtedly average values representing a range of environments. The values of the isomer shifts are consistent with three components (1-3) from Fe3+ cations, while component 4has a substantially lower isomer shift and can be assigned to Fe4+ ions. The 1.7% by mass of the second phase will contribute <6% to the area of the spectrum, and with manganese substitution this value will decrease.There is no possibility of identifying this weak component separately. The asymmetry in the intensity of the spectrum (assuming that the recoilless fraction is the same at all sites) leads to an estimate of 24% of Fe4+ which is effectively independent of the other features of the model. If it is assumed that all the Mn is tetravalent, then this equates with an overall stoichiometry of SrMno.7Feo.302.89, in good agreement with that determined iodometrically. It also establishes that there must be ca. 20% of Fe4+ in the primary phase, although the precision of this estimate is limited by the low precision associated with our determination of the concentration of the second phase. In order to reconcile this Fe4+ concentration 20 40 60 80 100 120 140 160 2eldegrees Fig.2 Observed, calculated and difference neutron powder diffraction 87profiles for SrMn, 72Fe0 2802 at room temperature. Reflection positions are marked for both the primary and the secondary phases. Table 1 Crystallographic data for 10H SrMn, 72Fe0.2803-6u atom site X Y Z occupancy Sr(1) 2(d) 113 213 314 1 113 0.0503(2) 1Sr(2) 4~) 213 0.1544(1) 1Sr( 3) 4(e) 0 0 Mn/Fe(l) 2(a) 0 0 0 0.558/0.442( 5) Mn/Fe(2) 4(f) 113 213 0.6040(4) 0.531/0.469( 3) Mn(3) 4(f) 113 213 0.1936(3) 1 O(1) 6(h) 0.1801(3) 0.3602(7) 114 1 O(2) 12(k) 0.4985(2) -.0.0030( 5) 0.14816(8) 1 O(3) 12(k) 0.8323(3) 0.6647( 1) 0.0492( 1) 0.890(4) aspace group P6,lmmc. a=5.45035(5) A, c=22.3735(2) A, I/= 575.59( 1) A3.U,,,=0.0063(3) A'.mass frac: 98.3(5)%. 1188 J. Muter. Chem., 1996, 6(7), 1187-1190 Table 2 Bond lengths/A and selected bond angles/degrees in SrMno,72Feo~2802,87 Sr(1)-O( 1) Sr(1)-0(2) Sr(2)-O( 2) Sr(2)-O(3) Mn/Fe( 1)-0(3) Mn/Fe(2)-O(2) Mn/Fe(2)-0(3) Mn(3)-O( 1) Mn(3)-O( 2) Mn( 3)-Mn( 3) 2.728(4) (6 x) 2.777(2) (6 x) 2.688(3) (3 x) 2.722(3) (3 x) 1.928(3) (6 x) 1.871 (5) (3 x) 1.985(6) (3 x) 1.919(5) (3 x) 1.862(4) (3 x) 2.52( 1) Sr(2)-0(3) Sr(3)-O( 1) Sr(3)-O( 2) Sr(3)-O( 3) O(3)-Mn/Fe( 1)-O( 3) 0(2)-Mn/Fe (2)-0(2) 0(2)-Mn/Fe (2)-0(3) O(3)-Mn/Fe(2)-O(3) O( l)-Mn(3)-0( 1) O(l)-Mn(3)-0(2) 0(2)-Mn(3)-0(2) 2.725(3) (6 x) 2.733(4) (3 x) 2.729(3) (6 x) 2.835(4) (3 x) 90.6( 1) 94.6( 4) 89.53(8) 86.0( 3) 81.5(2) 92.49( 6) 92.9(2) Ba5W3Li201515were reported many years ago.The latter is isostructural with SrMn0.72Fe0.2802,87 whereas the former has a structure wherein Mn,O, dimers share vertices with trimers (Mn30,,) of face-sharing octahedra, thus leading to an hchchhchch stacking of BaO, layers, a sequence with a much greater proportion (60%) of hexagonal stacking than we have observed in SrMno,72Feo~2802~87. Cs4NiCdF,, has been shown’ to have a third 10H sequence, chhhcchhhc, which also contains 60% hexagonal stacking. The structure of SrMno.72Feo.2802.87 thus contains a relatively large component of cubic perovskite, a property which can be attributed to the relatively small size of the A-site cation.2 This can be seen in Fig.1, where it is possible to recognise not only segments of the SrMnO, struc- ture, i.e. Mn209 dimers sharing vertices with neighbouring octahedra, but also blocks of pseudo-cubic perovskite remi- niscent of SrFeO,. SrMn0,72Fe0.2802.87 is thus a compound of the type AB,-,B,’O,-, which adopts the structure of neither AB03-6 nor AB’03-6, but an intermediate structure contain- ing an intergrowth of elements of both end members. The occurrence of only Mn in the face-sharing octahedra is consist- ent with this descrip$on, The average Mn-0 bond distance in the +mers (1.891 A) is very similar to that found previysly (1.889 A) in SrMnO,,, as is the Mn-Mn distance (2.52 A, cf 2.50A). This supports the assumption that the Mn in the dimers is tetravalent, as does the value of 4.15 calculated for the bond valence of Mn(3).l6>l7 The location of the anion vacancies only on the site 0(3), i.e.a site in the centre of the Fe-rich region of the structure, is also consistent with the structural chemistry of SrFeO, -64 and SrMnO, . Throughout the above discussion we have described our sample using the chemical composition determined in the diffraction study. We believe that the large contrast between the scattering lengths of Mn and Fe makes neutron scattering a particularly sensitive analytical tool in this case. In addition to making it feasible to refine the cation stoichiometry, neutron diffraction was also the only technique which detected the presence of a pseudo-cubic impurity at 1.7% by mass.The disorder caused by the presence of anion vacancies and a near- random distribution of Mn and Fe over the vertex-sharing octahedral sites reduced the high resolution normally associ- ated with 57Fe Mossbauer spectroscopy, although it was still possible to deduce an accurate ratio for the Fe oxidation states. The Mossbauer spectra also show that the sample is paramag- netic at room temperature, an observation which is consistent with the absence of magnetic Bragg peaks in the neutron diffraction pattern. The isomer shifts and relative areas of the four doublets used to fit the spectra are consistent with the presence of Fe3+ and Fe4+ as distinct oxidation states in this compound; there is no evidence for the presence of an averaged or mixed valence state.The results described above suggest that the system SrMn, -xFe,03 -could be an abundant source of interesting structural chemistry. We are currently investigating the tem- perature dependence of both the structural and the electronic properties of these compounds as a function of composition, J. Muter. Chem., 1996,6(7), 1187-1190 1189 4-3-2-101 2 3 veiocity/mm s-l Fig. 3 Observed and fitted Mossbauer spectra of SrMno,7Feo,30,-, at room temperature. The maximum absorption is 3%. Table 3 Mossbauer parameters for SrMno,,Feo.,03-d at room temperature component 6/mm s-l d/mm s-l r/mm s-l area (%) 1 0.37 0.28 0.34 32 2 0.31 0.79 0.30 33 3 0.26 1.21 0.27 11 4 -0.06 0.29 0.27 24 with the stoichiometry (cu.7% Fe4+) determined by neutron diffraction, we are led to the conclusion that some reduction to Mn3+ occurs, presumably in the vicinity of the vacant anion sites. Alternatively, the precision associated with the occupancy factor of the O(3) site (Table 3) might have been overestimated. Assuming random placement of vacancies on O(3) sites and an occupancy of 0.89, then 63% of the Mn/Fe( 1)and Mn/Fe(2) sites have six 0 neighbours, and 37% have one or more vacancies. If the vacancies on O(3) are not random but are preferentially associated with a pair of Fe3+ cations, which is consistent with charge considerations, then a greater fraction of the iron will have a nearby vacancy. It is possible that components 2 and 3 in Table 3, which show a significant quadrupole splitting and represent 44% of the Fe, are associ- ated with a vacancy, while components 1 and 4 are fully coordinated to oxygen and show only a small quadrupole splitting. Thus there is some evidence in favour of a non-random distribution of anion vacancies, but this should not be regarded as proven.Discussion The structural chemistry of SrMno.72Feo.2802.87 shows a number of interesting features. The very adoption of the 10H structure is quite unusual. It does not occur in the BaMn, -xFex03-system,14 although 10H BaMnO, -x5 and and we hope to publish the results of these studies in the 8 J M Dance, J Darnet, A Tressaud and P Hagenmuller, Z Anorg Allg Chem ,1984,508,93near future 9 H M Rietveld, J Appl Crystallogr ,1969,2, 65 We are grateful to EPSRC for financial support 10 A C Larson and R B von-Dreele, General Structure Analysis System (GSAS), Los Alamos National Laboratones, Report References A L Patterson and J S Kasper, In International Tables for X-ray Crystallography, Kynoch Press, Birmingham, 1959, vol I1 J B Goodenough and J A Kafalas, J Solid State Chem, 1973, 6,493 P D Battle, T C Gibb and C W Jones, J Solid State Chem, 1988,74,60 J B MacChesney, R C Sherwood and J F Potter, J Chem Phys, 1965,43,1907 T Negas and R S Roth, J Solid State Chem ,1971,3,323 B L Chamberland, A W Sleight and J F Weiher, J Solid State Chem , 1970,1,506 M R Lees, J Barratt, G Balakrishnan and D M Paul, Phys Rev B, 1995,52, R14303 LAUR 86-748,1990 11 A J Jacobson, B M Collins and B E F Fender, Acta Crystallogr Sect B, 1974,30,816 12 B C Tofield, C Greaves and B E F Fender, Muter Res Bull, 1975,10,737 13 R J Hill and H D Flack, J Appl Crystallogr, 1987,20,356 14 V Caignaert, M Hervieu, B Domenges, N Nguyen, J Pannetier and B Raveau, J Solid State Chem ,1988,73,107 15 T Negas, R S Roth, H S Parker and W S Brower, J Solid State Chem, 1973,8,1 16 N E Brese and M O’Keeffe, Acta Crystallogr Sect B, 1991, 47,192 17 I D Brown and D Altermatt, Acta Crystallogr Sect B, 1985, 41,244 Paper 6/01020B, Received 12th February, 1996 1190 J Muter Chem, 1996, 6(7), 1187-1190