Non-stoichiometry, structural defects and properties of LaMnO3+d with high d values (0.11d0.29) J. A. Alonso,*a M. J. Martý�nez-Lope,a M. T. Casais,a J. L. MacManus-Driscoll,b P. S. I. P. N. de Silva,c L. F. Cohenc and M. T. Ferna�ndez-Dý�azd aInstituto de Ciencia deMateriales de Madrid, C.S.I.C. Cantoblanco, E-28049 Madrid, Spain bDepartment of Materials, Imperial College, L ondon, UK SW7 2BZ cBlackett L aboratory, Imperial College, L ondon, UK SW7 2BZ cInstitut L aue-L angevin, B.P. 156, F-38042 Grenoble Cedex 9, France Strongly oxygenated LaMnO3+d perovskites, with nominal MnIV contents up to 58%, have been prepared by thermal decomposition of metal citrates followed by annealings either in air or under high oxygen pressure (200 bar). A high-resolution neutron powder diVraction study of four representative samples with 0.11d0.29 reveals the presence of both La and Mn vacancies.Contrary to previous studies, it is found that there are a substantially higher proportion of Mn vacancies, depending rather sensitively on the oxidation conditions. The oxidation state for Mn calculated for the refined stoichiometry La1-xMn1-yO3 is in good agreement with the d values previously determined by thermal analysis.Further to this, it is also found that as d increases the MnMO bond lengths shorten, the MnMOMMn angles progressively increase and the perovskite structure becomes more regular, which is consistent with the incorporation of MnIV cations. The presence of Mn vacancies (as much as 13% in samples prepared under high oxygen pressure) perturbs the conduction paths for the transport of holes across MnMOMMn, weakening the double-exchange interaction.This structural disorder explains the observed decrease of the ferromagnetic Curie temperature (TC) as d increases. The structure and properties of the perovskite LaMnO3 and temperature) from orthorhombic to rhombohedral.10,11 In fact, the transition temperature from the orthorhombic (low-tem- related materials have been extensively studied in the past.1,2 More recently, the observation of large negative magnetoresist- perature form) to the rhombohedral (high-temperature form) was found to change10 from 600 °C for LaMnO3.00 to ca.ance eVects in mixed-valence manganites3 based on LaMnO3 has renewed the interest in the study of these systems.In -90 °C or LaMnO3.15. The crystal structure of the oxygen excess rhombohedral particular, the strongly correlated magnetic and electrical properties are being widely studied. The magnetoresistance is phases has been investigated by neutron diVraction.12,13 Tofield and Scott12 and, later on, Van Roosmalen et al.13,14 concluded associated with the insulator-to-metal transition which occurs in the vicinity of the Curie point, for ferromagnetic composi- that instead of incorporating oxygen interstitials, as directly suggested by the formula LaMnO3+d, this perovskite is defec- tions such as La0.7Ca0.3MnO3 which, thus, simultaneously exhibit metallic conductivity and ferromagnetism at low tem- tive in both La and Mn positions: for instance, a sample with a formal composition of LaMnO3.158 was found to exhibit an peratures.The presence of divalent alkaline earth cations on the La sites of the perovskite induces a MnIII–MnIV mixed- actual crystallographic formula of La0.95Mn0.95O3. Recently15 we have prepared LaMnO3+d phases with a wide valence state with the formation of holes which undergo fast hopping between the two oxidation states.The temperature of range of d values, 0.11d0.31, by annealing finely divided precursors at moderate temperatures, either in air or under both transitions depends on the MnIV content (for instance, in the La1-xCaxMnO3 system TC varies between 182 and 278 K high oxygen pressure. The samples were characterized by Xray diVraction and thermal analysis, allowing us to determine for MnIV contents between 14 and 38%, respectively.4) The parent perovskite LaMnO3.00 (where only MnIII is their oxygen contents (d) very reliably.This paper complements those previous studies giving details on the actual way of present) is antiferromagnetic below TN=140 K and shows a semiconducting behaviour over the whole temperature range.5 incorporation of the non-stoichiometry into the crystal structures, refined from high-resolution neutron powder diVraction.LaMnO3 crystallizes in the orthorhombic GdFeO3 structural type,6,7 with a=5.5392(6), b=5.6991(7), c=7.7175(9) A ° , in the Our main result is that the probability of creating equal numbers of La and Mn vacancies depends very strongly on Pbnm setting. A strong Jahn–Teller distortion of the oxide octahedra around the d4 MnIII cations has been identified from the preparative oxidation conditions.The influence of the observed defective structure on the magnetic properties is also a neutron diVraction study.7 A MnIII–MnIV mixed-valence state can also be induced in discussed. phases of nominal stoichiometry LaMnO3+d (with MnIV content of 2d per formula unit). These phases also show ferromag- Experimental netic behaviour for a suYciently high content of MnIV.For Four selected LaMnO3+d samples were prepared in polycrys- instance, LaMnO3.08 (with 16% of MnIV) was reported to be talline form by a citrate technique. Stoichiometric amounts of ferromagnetic with a Curie temperature TC=125 K).5 In fact, analytical grade La(NO3)3 6H2O and Mn(NO3)2 4H2O were GMR behaviour has been reported8,9 in samples with moderate dissolved in citric acid.The metal solution was slowly evapor- d values, showing overall compositions LaMnO3.17 and ated and the resultant resin was subsequently decomposed at LaMnO3.12. 600 °C for 12 h. A second treatment at 800 °C for 6 h enabled For LaMnO3+d the oxidation of some MnIII to MnIV ions the total elimination of the nitrates and organic materials.The reduces the driving force for Jahn–Teller distortion. Above d= 0.105 (i.e. 21% MnIV) there is a change in symmetry (at room finely divided precursor powders were annealed either in air J. Mater. Chem., 1997, 7(10), 2139–2144 2139(samples 3 and 4) or under 200 bar of oxygen pressure (samples 1 and 2), at temperatures ranging from 800 to 1100 °C.Then, the samples were slowly cooled to room temperature. The final materials were characterized by X-ray powder diVraction (XRD) with Cu-Ka radiation. The determination of d was performed by thermogravimetric (TG) analysis in a reducing H2–N2 flow, as indicated elsewhere.15 Neutron powder diVraction (NPD) data for LaMnO3+d were collected at room temperature in the high-resolution D2B diVractometer at the ILL, Grenoble, with a wavelength of 1.594 A ° , selected from a Ge monochromator.About 5 g of each sample were contained in a cylindrical vanadium can. The time consumed in each data collection was about 3 h. The thermal evolution of samples 2, 3 and 4 was studied in the multidetector DN5 diVractometer at the Siloe� reactor of the Centre d’Etudes Nucle�aires, Grenoble, with a wavelength of 2.488 A° in the temperature range 2–250 K.The Rietveld method16 was used to refine the crystal structures, using the FULLPROF program.17 The line shape of the diVraction peaks was generated by a pseudo-Voigt function, and the background refined to a fifth-degree polynomial. The coherent scattering lengths for La, Mn andO were, respectively, 8.24, -3.73 and 5.803 fm.In the final run the following parameters were refined: six background coeYcients, zeropoint, half-width, pseudo-Voigt and asymmetry parameters for Fig. 1 XRD patterns of LaMnO3+d , for samples 1 (above) to 4 (below) the peak shape; scale factors, positional, thermal isotropic factors and unit-cell parameters. The occupancy factors of La Table 2 Atomic parameters for rhombohedral LaMnO3+d (samples and Mn were also allowed to vary in the last stages of the 1–3) after the Rietveld refinement of NPD data at 295 K refinement.In the final cycle the shifts in the atomic parameters were zero up to the fourth decimal place. sample no. 1 2 3 Magnetization data were collected in an Oxford Instrument 3001 vibrating-sample magnetometer, in a remnant field of ca.atom La B/A 35(6) 1.38(6) 0.97(6) 30 Oe, between 10 and 300 K. focc 0.95(1) 0.97(1) 0.969(9) Mn B/A ° 2 0.3(1) 0.2(1) 0.26(10) Results focc 0.89(2) 0.87(2) 0.93(2) O x 0.4561(2) 0.4545(2) 0.4489(2) Structural features of LaMnO3+d B/A ° 2 1.48(3) 1.41(3) 1.22(4) focc 3.0 3.0 3.0 The XRD patterns of samples 1–4 were characteristic of aa/A ° 5.4905(1) 5.4951(1) 5.5222(1) monophase perovskites with rhombohedral (samples 1–3) or ba/A° 13.3077(1) 13.3030(1) 13.3317(1) orthorhombic (sample 4) symmetry, as shown in Fig. 1. V /A ° 3 347.42(1) 347.88(1) 352.08(1) discrepancy factors Observe that there is an increase of the rhombohedral splitting Rp (%) 4.13 4.12 4.12 of the peaks under less oxidizing conditions (i.e. high tempera- Rwp (%) 5.20 5.22 5.26 ture, air).Table 1 summarizes the preparation conditions, d Rexp (%) 3.96 4.19 4.26 values obtained by TG and unit-cell parameters determined x2 1.73 1.55 1.52 by XRD. For samples 1 to 3 both hexagonal (a, c) and RI (%) 7.34 7.73 6.89 rhombohedral (ar, ar) descriptions of the unit cell are given. aThe unit-cell parameters diVer slightly from those determined from The room-temperature NPD patterns of samples 1–3 were XRD (Table 1) owing to a small inaccuracy of the neutron wavelength.refined in the space group R39c, hexagonal description (Z=6), Note: space group R39c, Z=6. La atoms are at 6a positions, (0,0,1/4); taking as starting model that of the perovskite LaNiO3, which Mn at 6b, (0,0,0); O at 18e, (x,0,1/4). exhibits the same symmetry.18 La atoms are at 6a, (0,0,1/4) positions; Mn at 6b, (0,0,0); and O at 18e, (x,0,1/4).The final atomic parameters after the refinements are shown in Table 2. For sample 4, the profile refinement was performed in the orthorhombic space group Pbnm (Z=4), according to the The good matching of the fits is illustrated in Fig. 2(a) for sample 3. No extra lines or additional splitting of the peaks GdFeO3 structural model.La is placed at 4c, (x,y,1/4); Mn at 4b, (1/2,0,0); O(1) at 4c and O(2) at 8d, (x,y,z). Subsequently, was observed in any case. The refinement of the occupancy factor of La and Mn led to values significantly lower than 1. a detailed analysis of the profile made it necessary to consider Table 1 Preparation conditions, unit-cell parameters and volume per formula of LaMnO3+d , samples 1–4, determined from XRD data.Samples 1–3 are rhombohedral, space group R39c, Z=6; the main phase in sample 4 is orthorhombic, space group Pbnm, Z=4. sample no. 1 2 3 4 prep. conditions 800 (200 bar O2) 1000 °C (200 bar O2) 1000 °C (air) 1100 °C (air) d(TG) 0.29 0.26 0.15 0.11 a/A ° 5.4898(4) 5.4951(3) 5.5239(2) 5.54002(2) b/A° 5.4898(4) 5.4951(3) 5.5239(2) 5.4963(2) c/A ° 13.311(1) 13.3061(8) 13.3349(7) 7.7876(4) ar/A ° 5.453(1) 5.453(1) 5.470(1) — ar/° 60.450(5) 60.509(5) 60.64(5) — Vf/A° 3 57.90(1) 57.99(1) 58.73(1) 59.28(1) 2140 J.Mater. Chem., 1997, 7(10), 2139–2144Table 3 Atomic parameters for sample 4 [annealed at 1100 °C(air)] after the refinement of NPD data at 295 K. The sample consists of a mixture of a main orthorhombic phase (Pbnm, 64.4%) and a minor rhombohedral phase (R39c, 35.6%)a atom Pbnm R39c La x 0.9970(5) y 0.0171(4) B/A ° 2 0.83(5) 0.84(8) focc 0.978(2) 0.978(2) Mn B/A ° 2 0.41(7) 0.31(12) focc 0.946(4) 0.946(4) O(1) x 0.0645(5) 0.4466(4) y 0.4955(7) B/A° 2 1.09(6) 1.06(6) focc 1.0 3.0 O(2) x 0.7398(4) y 0.2705(4) z 0.0337(2) B/A ° 2 1.23(5) focc 2.0 a/A ° 5.5388(2) 5.5348(1) b/A° 5.4958(2) 5.5348(1) c/A ° 7.7843(2) 13.3438(3) RI 6.39 4.81 aDiscrepancy factors: Rp=6.32%, Rwp=4.18%, Rexp=2.38%, x2= 2.18.Note: for the orthorhomic phase: space group Pbnm, Z=4; La at 4c, (x,y,1/4); Mn at 4b, (1/2,0,0); O(1) at 4c and O(2) at 8d, (x,y,z). For the rhombohedral phase: space group R39c, Z=6; La at 6a (0,0,1/4); Mn at 6b (0,0,0); O at 18e, (x,0,1/4).additionally a minor rhombohedral LaMnO3+d phase (with space group R3 : c) to correctly fit the spectra. The strong overlapping between the reflections of both polymorphs had prevented the detection of the minor rhombohedral phase Fig. 2 Room temperature observed (crosses), calculated (solid line) prior to the Rietveld analysis of the neutron pattern. From the and diVerence (at the bottom) NPD profiles for (a) rhombohedral refined scale factors a composition of 63.6% (orthorhombic)+ LaMnO3.26, sample 2; and (b) orthorhombic LaMnO3.11, sample 4: 36.4% (rhombohedral) was determined for the mixture.The the second series of tick marks indicate the reflections of a minor La and Mn occupancy factors were constrained for both rhombohedral phase.For the sake of clarity only half of the experimenphases. Table 3 includes the results of the refinement. Fig. 2(b) tal points are represented. shows the observed and calculated NPD profiles for sample 4. Bond distances and angles are listed in Table 4. Observe that in the orthorhombic structure the MnO6 octahedra do not show any appreciable Jahn–Teller distortion, as expected for lengths progressively increase from 1.9494(6) A ° , for the perovskite prepared in the most oxidizing conditions, sample 1, the relatively high proportion of MnIV in the crystal (for d= 0.11, [Mn4+]=22%), which prevents the cooperative distor- to an average value of 1.976 A ° for the orthorhombic phase present in sample 4.tion of the octahedra observed7 in stoichiometric LaMnO3.00.As shown in Table 1, there is a net increase of the cell Some relevant parameters determined from the refinements are listed in Table 5. The crystallographic formulae of the volume per formula (Vf) from 57.90 to 59.28 A ° 3 as d decreases from 0.29 (sample 1) to 0.11 (sample 4), which is consistent LaMnO3+d phases, according to the refined occupancy factors for La and Mn, can be written as La1-xMn1-yO3, as indicated with the decreasing amount of MnIV cations in the structure.This is directly related to the regular variation of the MnMO in Table 5. Under more oxidizing conditions (from sample 4 to sample 1) the amount of Mn vacancies increases more distances, quoted in Table 4. Observe that MnMO bond Table 4 Selected bond distances (A ° ) and angles (degrees) for LaMnO3+d (samples 1–4) sample no 1 2 3 4 space group R39c R39c R39c R39c Pbnm MnMO(1) 1.9494(6) ×6 1.9514(7) ×6 1.9634(6) ×6 1.969(1) ×6 1.9788(6) ×2 MO(2) — — — — 1.975(2) ×2 MO(2) — — — — 1.975(2) ×2 LaMO(1) 2.504(1) ×3 2.497(1) ×3 2.479(1) ×3 2.472(1) ×3 2.891(5) MO(1) 2.7367(5) ×6 2.7377(5) ×6 2.7492(4) ×6 2.7543(5) ×6 2.656(5) MO(1) — — — — 2.432(4) MO(2) — — — — 2.639(3) ×2 MO(2) — — — — 2.500(3) ×2 MO(2) — — — — 2.811(3) ×2 LaMO 2.6591(2) 2.6576(2) 2.6591(2) 2.6602(3) 2.653(1) O(1)MMnMO(1) 90.84(6) 90.91(6) 91.13(5) 91.24(1) 180.0 O(2)MMnMO(2) — — — — 91.45(17) MnMO(1)MMn 165.782(7) 165.264(7) 163.477(7) 162.72(1) 159.15(5) MnMO(2)MMn — — — — 162.12(9) J.Mater. Chem., 1997, 7(10), 2139–2144 2141Table 5 Relevant parameters obtained from the structural data: crystallographic formulae, valence of Mn determined from the occupancy factors ( focc), thermogravimetric data (TG) and the bond valence theory ( b.v.); and tolerance factors, t Mn valence sample nominal actual no.composition stoichiometry focc TG b.v.a tb 1 LaMnO3.29 La0.95(1)Mn0.89(1)O3 3.54 3.58 3.65 0.965 2 LaMnO3.26 La0.97(1)Mn0.87(2)O3 3.55 3.52 3.61 0.963 3 LaMnO3.15 La0.97(1)Mn0.93(1)O3 3.33 3.30 3.50 0.958 4 LaMnO3.11 La0.978(2)Mn0.946(4)O3 3.24 3.22 3.26 0.949 aThe valence is the sum of the individual bond valences (si) for MnMO bonds within the MnO6 octahedra.Bond valences are calculated as si=exp[(r0-ri)/B]; B=0.37, r0=1.760 for the MnIIIMO2- pair, from ref. 22. Individual MnMO distances (ri) are taken from Table 3.bTolerance factors are calculated as t=LaMO /Ó2MnMO . Fig. 4 Magnetization vs. temperature plot for LaMnO3+d (H#30 Oe). Annealing temperature and atmosphere: (a) 1000 K, 200 bar O2; (b) 800 K, 200 bar O2; (c) 1000 K, air; (d) 1100 K air. Fig. 3 Variation of the La and Mn contents with d, determined from the neutron diVraction refinements. The full lines are guides for the eye.quickly than the La vacancies, as shown in Fig. 3. The very oxidizing conditions (under 200 bar O2) are more easily able to create Mn vacancies than La vacancies. The final valences for Mn calculated from the metal vacancy concentration agree quite well with those determined by thermal analysis, as shown in Table 5. The oxidative non-stoichiometry determined for these products is thus able to explain the macroscopic behav- Fig. 5 Thermal evolution of the NPD patterns for LaMnO3.11, sample iour, represented by the d values. 4. Below TC=165 K the magnetic contribution to the peaks of nuclear Comparing the observed MnMO distances to the ionic radii origin corresponds to the three-dimensional ferromagnetic ordering. sums19 for a random occupancy of MnIII/MnIV in the manganese positions, the observed values are systematically lower.In Fig. 6 the Curie temperatures are plotted vs. d, determined For instance, for sample 1, [MnMO]calc=1.983 A ° (observed from both magnetization data and low temperature NPD 1.949 A ° , Table 4); for sample 4, orthorhombic phase, experiments. Values of TC obtained by NPD are systematically [MnMO]calc=2.017 A ° (observed 1.976 A ° on average).This fact lower than those determined from the magnetization curves, can be understood as an eVect of the additional contraction as a consequence of a gradual transition from short- to longof the lattice due to the presence of metal vacancies. For the range magnetic order, the latter taking place at temperatures same reason, the calculation of the valence of Mn cations about 5–15 K lower.within the MnO6 coordination octahedra by means of Brown’s bond valence model20,21 systematically leads to values higher than observed (Table 5). Discussion In their pioneering work, Tofield and Scott12 showed that Comparison of low-temperature NPD and magnetization the defect chemistry of LaMnO3+d is better described with measurements randomly distributed La and Mn vacancies.Later on, Van Roosmalen et al.13,14 concluded from a neutron diVraction The magnetization and transport measurements have been studied for a more complete set of samples and will be experiment combined with density measurements that La and Mn vacancies are present in equal amounts in the solid, in described in detail elsewhere.22 Fig. 4 shows the magnetization vs.temperature curves for the samples discussed in the present such a way that the crystallographic formula should be written as La1-xMn1-xO3, with x=d/(3+d). Our neutron diVraction study. They show a ferromagnetic ordering of the Mn spins below a critical temperature. The ferromagnetic contribution study on samples with high d values, prepared at relatively low temperatures from finely divided precursors, confirms the to the neutron scattering is presented in Fig. 5, including the thermal evolution of the NPD patterns for sample 4. In all presence of significant amounts of La and Mn vacancies. Final Fourier synthesis did not yield any identifiable peaks which cases the magnetic structures can be described in terms of a single propagation vector, k=(0,0,0).23 could suggest the presence of additional oxygen atoms in 2142 J.Mater. Chem., 1997, 7(10), 2139–2144observation by To� pfer et al.,11 who described an orthorhombic perovskite-type structure for LaMnO3+d at d0.10, while for d>0.10 they identified a rhombohedral cell. Sample 4, with d=0.11, is at the boundary of the phase transformation, consistent with the observed phase separation.It is worth mentioning that, in spite of the change in crystal symmetry, the determined La and Mn vacancy concentration follows the trend shown in Fig. 3 for the three pure rhombohedral compositions. Some perovskites with oxidative non-stoichiometry have been reported to show ferromagnetic behaviour with Curie points in a comparable temperature range to those observed in the present work for the samples with lower d values.For instance, Ranno et al.24 described a TC of 125 K for LaMnO3.15, or (LaMn)0.95MnO3, whereas To� pfer et al.11 obtained TC= 171 K for LaMnO3.14; in both cases the composition is close to that of sample 3 (also d=0.15), with a refined stoichiometry La0.97Mn0.93O3, and TC (from magnetization data) of 155 K. Fig. 6 Variation of the Curie temperature with d. Full symbols from Fig. 6 shows that TC decreases as d increases, i.e. as the MnIV magnetization data, for samples prepared, from left to right, at 1100 K in air, 1000 K in air and 800 K in 200 bar O2. Open symbols from concentration increases. A priori, an opposite trend would be NPD data, for samples prepared, from left to right, at 1100 K in air, expected, since the introduction of MnIV cations (t2g3) into the 1000 K in air, 1000 K in 200 bar O2 (from ref. 23). MnIIIO6 array (high-spin MnIII: t2g3e2g1) of LaMnO3 by hole doping (or as a result of deviations from the stoichiometric composition) creates empty eg orbitals which favour the ferro- interstitial positions, even in the more oxygenated samples. Our previous suggestion15 that the structure of the samples magnetic double exchange.This would lead to increasing Curie temperatures with increasing d, as demonstrated in Ca-doped prepared under oxygen pressure would probably involve oxygen interstitials, is not confirmed. manganites La1-xCaxMnO3 up to the ‘optimum doping level’, of 33% MnIV. In addition, for a constant doping level, TC has The present NPD study shows definitively that the concentrations of both kinds of metal vacancies are not equal. been shown to increase as the tolerance factor of the perovskites is closer to 1, i.e.when the structure becomes more regular, Moreover, the concentration of Mn vacancies increases at a higher rate when the annealing conditions of the samples with MnMOMMn angles closer to 180° favouring the double-exchange interactions, as shown25 for the series become more oxidizing (i.e.lower temperatures and higher oxygen pressure), as shown in Fig. 3. The La and Mn vacancy (La0.7-xYx)Ca0.33MnO3. In the present case the observed trend in the variation of TC concentration only become closer for low d values, i.e. for samples annealed in air at higher temperatures.Note that the vs. d can be interpreted as a consequence of the presence of significant amounts of Mn vacancies in the structure. The final samples described by Tofield and Scott12 and Van Roosmalen et al.13,14 had undergone thermal treatments at even higher TC value and the strength of the ferromagnetic behaviour result from an interplay between defect density, MnIV content and temperatures of 1200 or 1300 °C, respectively. The soft-chemistry synthesis procedure described in the MnMOMMn bond angles. Even though the tolerance factors (Table 5) of the perovskite structures of samples 1–4 increase present work seems to favour the formation of highly defective perovskites, especially in the Mn sublattice.We show that the with d (which is related to the progressive shortening of the MnMO distances), the relatively high proportion of Mn vac- non-stoichiometry of LaMnO3 cannot be simply denoted with a single parameter, d or x, but requires the specification of two ancies (up to 13% in sample 2) perturbs the connecting paths for the holes to transport across MnMOMMn, and leads to parameters, x and y in La1-xMn1-yO3.The relative values of x and y depends dramatically on the preparation procedures weaker ferromagnetic coupling, via the double-exchange interaction.The presence of random vacancies may also have a of the samples, in such a way that ceramic synthesis at higher temperatures favours equal values of x and y. strong localising eVect on the available charge carriers, which reduces their mobility.24 Our results suggest that at moderate temperatures, in the range 800–1100 °C, the higher mobility of Mn cations allows them to predominantly migrate across the perovskite structure, Conclusions giving rise to a dtive Mn sublattice and consequently a higher oxidation state of the remaining Mn cations. The higher A room-temperature, high-resolution neutron diVraction study of four selected LaMnO3+d samples with high d values mobility of Mn is probably related to the smaller size of Mn vs.La cations. At higher temperatures the mobility of La (0.11d0.29), prepared by low-temperature treatments of citrate precursors, shows highly defective perovskite structures cations also becomes significant, giving rise to final products in which the amounts of both metal vacancies are closer.containing both La and Mn vacancies. The Mn vacancies are present in a substantially higher proportion with respect to In all cases the samples were prepared from stoichiometric mixtures of nitrates, with La5Mn=151 ratio. The observed the La vacancies in all the samples, although the diVerence in the number of both kinds of defects decreases under less deviation of the 151 stoichiometry in the final perovskite phases implies the presence of minor Mn-rich phases in the oxidizing conditions.(i.e. as d decreases). There is a good agreement between the Mn valences determined from thermal products. No diVraction peaks other than those corresponding to the perovskite oxides could be detected in either the XRD analysis and those estimated from the defect stoichiometry La1-xMn1-yO3.The cell volume, MnMO distances, or ND patterns. The segregated Mn-rich minor impurities must be present in a poorly crystallized or amorphous form, MnMOMMn angles and tolerance factors, t, of the perovskites vary regularly with d, i.e. with the oxidizing power of the probably as a thin layer covering the surface of the perovskite crystallites.materials. The high concentration of defects, mostly in the Mn sublattice, explains the anomalies observed in the magnetic The perovskites prepared at temperatures up to 1000 °C show rhombohedral symmetry, whereas that prepared at properties: the samples are ferromagnetic below TC which decreases when d increases, even though t increases. We 1100 °C (sample 4) could be identified as a mixture with a main orthorhombic phase.This fact is consistent with the conclude that the preparation by soft-chemistry methods of J. Mater. Chem., 1997, 7(10), 2139–2144 2143Ramakrisnan, R. Mahesh, N. Raganvittal and C. N. R. Rao, Phys. LaMnO3+d materials with high d contents starting from Rev. B, 1996, 53, 3348. La5Mn=151 mixtures leads to highly Mn defective phases 9 C.N. R. Rao and A. K. Cheetham, Science, 1996, 272, 369. with anomalous magnetic properties. The synthesis of undoped 10 A. Wold and R. J. Arnott, J. Phys. Chem. Solids, 1959, 9, 176. La–Mn–O materials with a perfect Mn sublattice would 11 J. To� pfer, J. P. Doumerc and J. C. Grenier, J. Mater. Chem., 1996, require starting mixtures with La5Mn<1 ratios, in such a way 6, 1511. 12 B. C. Tofield and W. R. Scott, J. Solid State Chem., 1974, 10, 183. that the selective creation of La vacancies allows Mn cations 13 J. A. M. Van Roosmalen, E. H. P. Cordfunke, R. B. Helmholdt and to reach the valence corresponding to the oxidation potential H. W. Zandbergen, J. Solid State Chem., 1994, 110, 100. given by the preparative conditions. 14 J. A. M. Van Roosmalen and E.H. P. Cordfunke, J. Solid State Chem., 1994, 110, 106. 15 J. A. Alonso, M. J. Martý�nez-Lope and M. T. Casais, Eur. J. Solid The authors acknowledge the financial support of the Spanish State Inorg. Chem., 1996, 33, 331. DGICyT to the project PB94–0046, and the Engineering and 16 H. M. Rietveld, J. Appl. Crystallogr., 1969, 2, 65. 17 J. Rodrý�guez-Carvajal, Physica B, 1993, 192, 55.Physical Science Research Council of the United Kingdom. 18 J. L. Garcý�a-Mun� oz, J. Rodrý�guez-Carvajal, P. Lacorre and J. B. Torrance, Phys. Rev. B, 1992, 46, 4414. 19 R. D. Shannon, Acta Crystallogr. Sect. A, 1976, 32, 751. References 20 I. D. Brown, in Structure and Bonding in Crystals, ed. M. O’Keefe and A. Navrotsky, New York, 1981, vol. 2, pp. 1–30. 1 G. H. Jonker and J. H. Van Santen, Physica, 1950, 16, 337. 21 N. E. Brese and M. O’Keefe, Acta Crystallogr. Sect. B, 1991, 47, 2 E. O.Wollan and W. C. Koeler, Phys. Rev., 1955, 100, 545. 192. 3 R. von Helmholt, J. Wecker, B. Holzapfel, L. Schultz and 22 P. S. I. P. N. de Silva, F. M. Richards, L. F. Cohen, J. A. Alonso, K. Samwer, Phys. Rev. L ett., 1993, 71, 2331. M. J. Martý�nez-Lope, M. T. Casais, T. Kodenkandath and 4 G. H. Rao, J. R. Sun, Y. Z. Sun, Y. L. Zhang and J. K. Liang, J. L. MacManus-Driscoll, J. Appl. Phys., submitted. J. Phys: Condens.Matter., 1996, 8, 5393. 23 J. A. Alonso, M. J. Martý�nez-Lope, M. T. Casais and A. Mun� oz, 5 B. C. Hauback, H. Fjellvag and N. Sakai, J. Solid State Chem., Solid State Commun., 1997, 102, 7. 1996, 124, 43. 24 L. Ranno, M. Viret, A. Mari, R. M. Thomas and J. M. D. Coey, 6 J. B. A. A. Elemans, B. Van Laar, K. R. Van der Veen and J. Phys: Condens. Matter, 1996, 8, L33. 25 J. Fontcuberta, B. Martý�nez, A. SeVar, S. Pin� ol, J. L. Garcý�a-Mun�oz B. O. Loopstra, J. Solid State Chem., 1971, 3, 238. and X. Obradors, Phys. Rev. L ett., 1996, 76, 1122. 7 P. Norby, I. G. K. Andersen, E. K. Andersen and N. H. Andersen, J. Solid State Chem. 1995, 119, 191. 8 R. Mahendiran, S. K. Tiwary, A. K. Raychadhuri, T. V. Paper 7/04088A; Received 11th June, 1997 2144 J. Mater. Chem., 19