5 ) : characterization of Fe ( III ) – Fe ( IV ) mixed valences *

D O I: 10 .1 58 26 /c hi m te ch .2 01 6. 3. 1. 00 4 I. Zvereva1*, T. Pavlova1, V. Pantchuk1, V. Semenov1, Y. Breard2, J. Choisnet2 1 Institute of Chemistry, St. Petersburg State University, 198904 Petrodvorets, Saint Petersburg, Russia 2 Laboratoire CRISMAT, UMR 6508 CNRS ENSICAEN and Caen University, 6 bd Maréchal Juin14050 Caen Cedex 4 France *Corresponding author: Tel.: +7 (904) 330-50-19 E-mail: irina.zvereva@spbu.ru


Introduction
Mixed valence states of 3d transition metals such as Ni, Co, Cu and Mn [1][2][3][4][5][6][7] can be stabilized in perovskite like oxides when working partial non-isovalent cationic substitutions.The fascinating electrical and magnetic properties of these oxide materials is strongly connected to the existence of these mixed valence states.Of importance to stabilize unusual oxidation degrees of 3d transition metals is the lowering of their site symmetry in the perovskite layers.In this respect the intergrowths of perovskite (P) layers and rocks-salt (RS) layers are favourable to get mixed valence states of the transition metal, whereas in the high symmetry field of the tridimensionnal perovskite structure such unusual oxidation degrees can disproportionate.In the manganites La 1+x Sr 2-x Mn 2 O 7 with the P 2 /RS intergrowth of a double perovskite block and one rock-salt layer, the existence of a colossal magnetoresistance (CMR) strongly depends on the mixed valence state of the manganese atoms [5,6].As to the oxygen content of these layered perovskite like phases, even if it has no concern to the oxygen stoichiometric LnSr 2 Mn 2 O 7 phases which contain Mn (III) and Mn (IV) [8], in most cases the oxygen deficiency properties are involved in the existence of the physical properties, as regularly checked in the cuprates and nickelates [1][2][3][4].
In the same way, systematic attention has been focused on the rich electrical and magnetic properties of perovskite like intergrowth structures of ferrites and their solid solutions [9][10][11][12][13][14].As an example the existence of two P 2 /RS type iron strontium mixed oxide is reported, namely the Fe(IV) one Sr 3 Fe 2 O 7 [12] and the Fe(III) one Sr 3 Fe 2 O 6 [9].Due to this and in the frame of our previous work on P/RS type chromium doped aluminates [15] and P 2 /RS chromium doped titanates [16], it was decided to look for compositions where it is possible to create mixed valence Fe(III) and Fe(IV) state of the iron atoms.
In this paper, we report on the partial substitution of iron atoms for titanium atoms in the P 2 /RS type strontium titanate Sr-3 Ti 2 O 7 [17] (Fig. 1) in terms of a structural analysis (XRPD) and a magnetic and Mössbauer characterization of iron low compositions of the solid solution Sr 3 Ti 2-x Fe x O 7-δ (x ≤ 0.5).The entire solid solution exists but up to now the reported results have a concern with iron richer compositions (2 ≥ x ≥ 0.5) [18,19].Even more iron diluted compositions (x ≤ 0.2) were never considered for crystal chemical and physical studies, as well.Consequently, the main goal of the present work consisted in clearing up the crystal chemical mechanism of charge compensation induced by the introduction of iron atoms in the matrix of Sr 3 Ti 2 O 7 : formation of the Fe (III) and Fe (IV) mixed valences together with the creation of oxygen vacancies.
According to the ceramics methods, the samples were pelletized and calcined in air at 1200 °C and then at 1350 °C for 40 h each.
The compositions x ≥ 0.2 were considered for X-ray structural analysis from cell parameters to structure calculations.The diluted compositions 0 < x ≤ 0.2 were retained for the magnetic and the Mössbauer study.In order to receive relevant information regarding the oxygen stoichiometry, different heating conditions were worked for some compositions asprepared in air: -an oxidizing treatment: 850 °C for 10 h and 150 bars oxygen pressure (x = 0.2 and x = 0.5) -a reducing treatment: a DTA Setaram device used with an hydrogen-argon atmosphere from room temperature up to 850 °C for 8 h.This was applied specifically to the richest iron composition x = 0.5.The iron content of the as-prepared samples was determined from atomic emission spectrometry.The maximum deviation between the theoretical and the experimental value of the iron content of a given sample does not exceed 5 %.
XRPD diffractograms were recorded with a Philips PW3020 diffractometer using the Cu Kα radiation in the 2q angular range 5-110 °, step size 0.04 ° and counting time 12s.Structure calculations were carried out with the FullProf code [20].
The magnetic susceptibility was measured according to the Faraday method in the temperature range 77-400 K.The precision is better than 2 %.Mössbauer spectra were recorded at room temperature by using spectrometer Wissel ( 57 Fe in a rhodium matrix), the isomeric shifts being calculated with respect to α Fe.In order to evaluate the part of paramagnetic species, the intensity of the signals was determined precisely up to the resonance factor.

XRPD results: cell constants and structure calculations
XRPD phase analysis ensured the existence of iron containing mixed oxide isotypic of Sr 3 Ti 2 O 7 (Fig. 1) which forms within the whole range of compositions (0 ≤ x ≤ 0.5).When the iron content of the Sr 3 Ti 2-x Fe x O 7-δ compositions does not exceed the value x = 0.3 no extra phase is observed.In the range of compositions 0.3 < x ≤ 0.5 some faint amount of a Sr 4 Ti 3 O 10 type phase i.e. a P 3 /RS intergrowth phase accompanies the major P 2 /RS phase.
The values of the tetragonal unit cell constants -a, c and volume V for x = 0.2 air prepared and after oxidation, x = 0.3 air prepared and x = 0.5 (air prepared, after oxidation and after reduction) are reported in Table 1.The corresponding variation versus x is shown in Fig. 2. In order to better understand the meaning of such a variation in terms of the cru- At first it should be stated that the nearly perfectly linear variation of V air the unit cell volume of the as-prepared compositions (Fig. 2a) brings evidence for the existence of a solid solution in the entire range of compositions 0 ≤ x ≤ 0.5.Moreover the variation of V strongly depends on the heating conditions: the oxidized compositions -Fe (IV) -exhibit a value of V ox smaller than the reduced one V red -Fe (III) whereas V air takes intermediate values.More precisely, the latter result is the combination of two different trends in the crystal chemical evolution of the solid solution Sr 3 Ti 2-x Fe x O 7-δ herein investigated: -in the oxidized compositions, the substitution of the smaller Fe 4+ cations (r CNVI = 0.585 Å) [21] for the Ti 4+ one (r CNVI = 0.605 Å) results in a decrease of V.
-in the reduced compositions, the creation of oxygen vacancies cancels the effect of the substitution of bigger Fe 3+ cations (r CNVI = 0.645 Å) for the Ti 4+ one, resulting in an overall decrease of V whose slope is weaker than in the oxidized compositions.
The precise contribution of a (Fig. 2b) and c (Fig. 2c) parameters to the variation of the unit cell volume is rather dif-ficult to ensure.At least one can assume the parameter a to be more sensitive to the decrease of size of the cations sitting in the octahedral sites.This result fully agrees with that is reported in the study of the compositions x ≥ 0.5 i.e. annealing at high oxygen pressure triggers a decrease  of the parameter a [19].On the contrary, the existence of a large amount of oxygen vacancies induces a pronounced lowering of the value of the parameter c.As a main result of the observed variation of the unit cell volume, it must be stated that the solid solution Sr 3 Ti 2-x Fe x O 7-δ which forms by heating in air contains a mixed valence state of the iron atoms.In order to learn about some modifications which are expected in the P 2 /RS intergrowth of the iron containing solid solution, it was decided to carry out a profile analysis of the XRPD difractograms of the compositions x ≥ 0.2.For x = 0.5 three cases were considered: as-prepared in air, oxidized and reduced samples and for x = 0.2 the as-prepared and the oxidized sample.The structure of Sr 3 Ti 2 O 7 [17] was retained: S.G.I4/mmm.Concerning the oxygen non-stoichiometry, XRPD is rather unsensitive to a small variation of the oxygen content.Consequently, only in a final step of the calculations of the air prepared compositions, a value of δ the oxygen deficiency arbitrarily fixed to the half of the maximum value corresponding to a full reduction was considered (δ = x/4).The results of the Mössbauer characterization here after reported ensured a value of δ close or lower than the half of a full reduction.As it was previously performed in the chromium containing solid solution Sr 3 Ti 2-x Cr x O 7-δ [16] we did an attempt to find the likely location of the oxygen vacancies in one of the three possible sites, namely the inner apical O 1 , the equatorial O 2 and the outer apical O 3 (Fig. 1b).In this respect, the main results to be received from the structural analysis are as follows: -the oxygen deficiency of the air prepared and reduced compositions occurs in the inner apical O 1 positions i.e. in the middle plane of the double perovskite block.This meets the different data obtained in the P 2 /RS cuprates [22] and the solid solution Sr 3 Ti 2-x Cr x O 7-δ [16].
-the equatorial M-O 2 distances (Table 2), within the precision of the calculation procedure are unsensitive to the substitution of iron atoms for titanium.
-as regularly observed in the intergrowth structures, there is an apical distorsion of the octahedra, as visible from the obtained values of the corresponding inner apical M-O 1 and outer apical M-O 3 distances (Table 2).The main data observed in Sr 3 Ti 2 O 7 i.e. the coupling of a longer inner M-O 1 distance with a smaller outer M-O 3 one is saved for the whole series of compositions.
Sr 3 Ti 2-x Fe x O 7-δ is shown in Fig. 3. Within the temperature range 77-400 K there is a monotonic decrease of χ versus T and for given temperature the observed value of χ systematically increases when x, the iron content, gets larger.The experimental data of the molar magnetic susceptibility have been used for calculating a paramagnetic value per one mole of iron, by subtracting the diamagnetic contribution of Sr 3 Ti 2 O 7 and iron atoms.The thermal variation of the paramagnetic susceptibility is described by a Curie-Weiss law χ = C/(T-q) over the whole temperature range under consideration.Curie constant C takes a value close to 4 emu.K, where as Weiss temperature q, slightly increases versus x (Table 3).The calculated effective magnetic moment m eff shows a complex dependence on both temperature and iron content, as visible in Fig. 4 for the air prepared compositions x = 0.02, 0.08; 0.12; 0.18.Such behaviour cannot be explained on the basis of one paramagnetic species and consequently, a mixed valence state of the iron cations which are introduced in the diamagnetic matrix of Sr 3 Ti 2 O 7 is likely to occur in the solid solution.Concerning the magnetic interactions, it can be reasonably assumed that they progressively change from an antiferromagnetic property to a ferromagnetic one, depending on an increasing of iron content.
The concentration dependence of m eff in the temperature range 298-400 K can be modelled in the following way: By extrapolating these equations at zero concentration of iron, the value of the effective magnetic moment m x→0 of a single iron cation in the solid solution above the room temperature (RT) can be estimated as nearly constant and equal to 5.45 MB.
The theoretical values of m eff of a single iron cation are 5.92 MB and 4.9 MB, for Fe 3+ (s = 5/2) and Fe 4+ (s = 2), respectively.Clearly, the observed value 5.45 MB gives evidence for the presence of iron cations with a number of unpaired electrons smaller than 5, very likely 4 as in the Fe 4+ species.If the exchange interactions between the paramagnetic iron species above RT are assumed to be weak enough not to induce a deviation of the effective magnetic moment with respect to the value of m eff for a single paramagnetic iron The observed effective magnetic moment in the solid solution with a zero concentration of iron can be modelled in terms on the only Fe 3+ and Fe 4+ cations, according to the following formula: x Fe Fe Fe Fe Fe Fe where m i and a i are the magnetic moment and the concentration of a given iron cation.Introducing into this equation the calculated value of m x→0 = 5.45 MB allows to calculate the concentration of Fe 4+ as equal to 0.48 (4).The presence of the two species Fe 3+ and Fe 4+ in the diluted solid solution looks unambiguous.
Considering the temperature dependence of the effective magnetic moment allows to point to the following statements: -for the lowest iron concentrations (x ≤ 0.08) the magnetic properties, up to a large extent, correspond to what is ex-pected from antiferromagnetic exchange interactions.
-for larger iron concentrations (x > 0.08) ferromagnetic exchange interactions take an increasing part depending on an increasing iron concentration (Fig. 4).
The Fe(III) -Fe(IV) mixed valence state of the iron atoms in the solid solution triggers three kinds of magnetic exchange interactions, namely Fe 3 -O-Fe 3+ , Fe 3+ -O-Fe 4+ et Fe 4 -O-Fe 4+ .Exchange interactions between Fe 3+ cations in the layered perovskite like phases are antiferromagnetic [23,24].When atoms with different electronic configuration are concerned, the exchange interactions are always ferromagnetic.As regards the Fe 4+ -O-Fe 4+ exchange interactions, they are either antiferromagnetic or ferromagnetic depending on the site symmetry of the iron atoms.In order to learn about the character of the exchange interactions in the Fe 4+ -O-Fe 4+ clusters, an analysis of the influence of the experimental heating conditions on the magnetic properties was carried out in the limiting composition x = 0.5 of the solid solution.The tempera- ture dependence of the effective magnetic moment observed in Sr 3 Ti 1.5 Fe 0.5 O 7-δ as prepared in air and heated in oxidizing or reducing conditions, is shown in Fig. 5.In the latter case, the sample contains only Fe 3+ species, as checked by thermal analysis, and the exchange interactions are antiferromagnetic.Above RT (Fig. 6) m eff in the reduced sample takes a value very similar to that of the Fe 3+ cation 5.92 MB.In the air prepared sample the value of m eff is intermediate between the reducing and the oxidizing cases, which result ensures the existence of two different exchange interactions.Finally, in the oxidized sample the Fe 4+ species are responsible of the strong ferromagnetic character of the exchange interactions, in agreement with the results reported for the ferrate Sr 3 Fe 2 O 7 [12].
At this stage, one result remains not immediately understandable in a simple way: the value of m eff even at temperatures higher than RT (Fig. 5) largely exceeds the value of single Fe 4+ cations: approximately 7 MB to be compared with 4.9 MB.One must take into account that for such iron concentration in the solid solution (25 %) the tendency of the paramagnetic species to aggregate will be important.It was previously observed and modelled in the P/ RS intergrowth structure of the solid solution YCaAl 1-x Cr x O 4 [25].Consequently, the actual value of the magnetic moment will be due not only to the single monomeric iron species but it will include the contribution of the various clusters, at least up to tetramers which likely have a concern to the observed magnetic moment.
As deduced from the temperature dependence of the effective magnetic moment, the ferromagnetism of the ex-change interactions undoubtedly increases versus the increasing amount of iron in the solid solution.More precisely, this data gives evidence for the increasing part of the Fe 4+ species.In order to receive another evidence for the presence of Fe 4+ and even more to calculate its concentration, the solid solution was studied by Mössbauer spectrometry.Fig. 6 shows Mössbauer spectra of three compositions x = 0.12; 0.16; 0.20 prepared in air.In any case there is the superposition of two signals with very different isomeric shifts d 1 = 0.431 mm/s and d 2 = -0.08 mm/scorresponding to the cations Fe 3+ and Fe 4+ , respectively [26][27][28].The observed value of the quadrupolar splitting for Fe 3+ DЕ 1 = 0.29 mm/s -is consistent with a lower site symmetry for Fe 3+ than for Fe 4+ DЕ 2 = 0.22 mm/s.On the basis of the Jahn-Teller effect of the 3d 4 Fe 4+ cations, a supplementary distortion of the corresponding (Fe 4+ O 6 ) octahedra is expected.In fact, the existence of oxygen vacancies in the inner apical positions of the double perovskite block, as ensured from the structure calculations, triggers a lowering of the site symmetry of the Fe 3+ cations.
For comparison, the solid solution Sr 3-x La x Ti 2-x Fe x O 7 was considered.In such case the fully charge compensated double substitution of the cationic couple x (La 3+ + Fe 3+ ) for x (Sr 2+ + Ti 4+ ) allows to maintain the oxygen stoichiometry i.e. there are no oxygen vacancies.The Mössbauer data observed for the composition x = 1 namely Sr 2.9 La 0.1 Ti 1.9 Fe 0.1 O 7 reveal the existence of one signal with an isomeric shift d = 0.32 mm/s which corresponds to an iron cation Fe 3+ in a high symmetry local field.Clearly, this is another proof that the Mössbauer spectrometry ensures the presence of the two species Fe 3+ and Fe 4+ in the air prepared solid solution Sr 3 Ti 2-x Fe x O 7-δ .
The analysis of the iron concentration dependence of the intensity of the two Mössbauer signals brings the opportunity to evaluate the respective parts of Fe 3+ and Fe 4+ .In Table 4 we report for the three compositions x = 0.12; 0.16; 0.20 the estimated values of the Fe 3+ and Fe 4+ concentration (%) and the corresponding values of y the (Fe 4+ ) composition and δ the oxygen deficiency.We receive the confirmation that the amount of Fe 4+ increases versus x the iron composition i.e. when the iron concentration in the solid solution is large enoughx ≥ 0.16 -the air prepared samples contain the Fe 4+ cations as main species.These results are in good agreement with the main information obtained from the magnetic properties.Finally an estimation of δ the oxygen deficiency in the solid solution Sr 3 Ti 2-x Fe x O 7-δ allows to ensure the oxygen non-stoichiometry property which is not large enough (δ < x/4) to be determined from XRPD calculations.

Conclusion
The solid solution Sr 3 Ti 2-x Fe x O 7-δ within its homogeneity range shows a remarkable ability to promote an oxidation of Fe(III) to Fe(IV) even annealed in air.The existence of a mixed valence state of the iron atoms with a major contribution of the Fe(IV) species is well established.In this respect, these new data well compare with those previously obtained in the case of chromium atoms in the solid solution Sr 3 Ti 2-x Cr x O 7-δ [16].In both cases the significant trend to get Fe(Cr)(IV) species № 1 | 2016 Chimica Techno Acta mainly results from the weak ability of these substituted titanates to tolerate the formation of oxygen vacancies in the middle plane of the double perovskite block.In this way, their crystal chemical properties are closer to that of the mangan-ites La 1+x Sr 2-x Mn 2 O 7 [8] than the cuprates La 2-x Sr(Ca) x Cu 2 O 6-x/2+δ [20].In the latter case, the middle plane of the double perovskite block is fully deprived of oxygen atoms.

Table 2
Metal oxygen distances (Å) in the (Ti, Fe)O 6 octahedra in the solid solution Sr 3 Ti 2-x Fe x O 7-δ

Table 3 Curie
constant C and Weiss temperature q in the air prepared) Sr 3 Ti 2-x Fe x O 7-δ x C, emu.