Effects of Doping and Oxygen Nonstoichiometry on the Thermodynamic Properties of Some Multiferroic Ceramics

3.1.1. Phase composition and crystalline structure

The room temperature XRD patterns (Fig. 2(a)) show perovskite single-phase, in the limit of XRD accuracy for all the investigated compositions after pre-sintering at 923 K/2 h followed by sintering at 1073 K/1 h and slow cooling. For all investigated ceramics, perovskite structure of rhombohedral R3c symmetry was identified, with a gradual attenuation of the rhombohedral distortion with the increase of BaTiO₃ content. This tendency to a gradual change towards a cubic symmetry with the BaTiO₃ addition is proved by the cancellation of the splitting of the XRD (110), (111), (120), (121), (220), (030) maxima specific to pure BiFeO₃ (2 ≈ 31.5^°, 39 ^°, 51^°, 57^°, 66^°, 70^°, 75^°), as observed in the detailed representation from Fig. 2(b). The evolution of the structural parameters provides an additional evidence for the influence of BaTiO₃ admixture in suppressing rhombohedral distortion (Fig. 3). Besides, the expansion of the lattice parameters induced by an increasing barium titanate content in (1−x)BiFeO₃ – xBaTiO₃ system was also pointed out (Prihor, 2009).

3.1.2. Microstructure

Surface SEM investigations were performed on both presintered and sintered samples. The SEM image of BiFeO₃ ceramic obtained after presintering at 923 K shows that the microstructure consists of intergranular pores and of grains of various size (the average grain size was estimated to be ~ 20 μm), with not well defined grain boundaries, indicating an incipient sintering stage (Fig. 4(a)). The SEM images of samples with x = 0.15 and x = 0.30 (Figs. 4(b) and Figs. 4(c) ) indicate that barium titanate addition influences drastically the microstructure. Thus, one can observe that BaTiO₃ used as additive has an inhibiting effect on the grain growth process and, consequently, a relative homogeneous microstructure, with a higher amount of intergranular porosity and grains of ~ one order of magnitude smaller than those ones of non-modified sample, were formed in both cases analyzed here.

BiFeO₃ pellet sintered at 1073 K/1h exhibits a heterogeneous microstructure with bimodal grain size distribution, consisting from large grains with equivalent average size of ~ 25 m and small grains of 3 - 4 m (Fig. 5(a)). The micrograph of the ceramic sample with x = 0.15 (Fig. 5(b)) shows that the dramatic influence of the BaTiO₃ on the microstructural features is maintained also after sintering. Thus, a significant grain size decrease was observed for sample with x = 0.15. Further increase of BaTiO₃ content to x = 0.30 (Fig. 5(c)) seems not to determine a further drop in the average grain size. Consequently, in both cases a rather monomodal grain size distribution and relative homogenous microstructures, consisting of finer (submicron) grains were observed (Ianculescu, 2008; Prihor, 2009). Irrespective of BaTiO₃ content, the amount of intergranular porosity is significantly reduced in comparison with the samples resulted after only one-step thermal treatment. This indicates that sintering strongly contributes to densification of the Bi_1-xBa_xFe_1-xTi_xO₃ ceramics.

3.1.3. Thermodynamic properties of Bi_1-xBa_xFe_1-xTi_xO₃

Of particular interest for us is to evidence how the appropriate substitutions could influence the stability of the Bi_1-xBa_xFe_1-xTi_xO₃ perovskite phases and then to correlate this effect with the charge compensation mechanism and the change in the oxygen nonstoichiometry of the samples.

In a previous work (Tanasescu, 2009), differential scanning calorimetric experiments were performed in the temperature range of 773-1173 K in order to evidence the ferro-para phase transitions by a non-electrical method. Particular attention is devoted to the high temperature thermodynamic data of these compounds for which the literature is rather scarce. Both the temperature and composition dependences of the specific heat capacity of the samples were determined and the variation of the Curie temperature with the composition was investigated. The effect of the BaTiO3 addition to BiFeO3 was seen as the decrease of the Curie transition temperature and of the corresponding enthalpy of transformation and heat capacity values (Tanasescu, 2009) (Fig. 6). A sharp decline in the T_C was pointed out for BiFeO₃ rich compositions (Fig. 6). In fact, the Cp of the rhombohedral phase (x = 0) is obviously larger than that of the Bi_1-xBa_xFe_1-xTi_xO₃ perovskite phases, whereas the Cp of each phase shows a weak composition dependence below the peak temperature. In particular, the value of Cp for x = 0.3 was found to be fairly low, which we did not show in the figure. The decreasing of the ferroelectric – paraelectric transition temperature with the increase of the BaTiO₃ amount in the composition of the solid solutions with x = 0 0.15 indicated by the DSC measurements is in agreement with the dielectric data reported by Buscaglia et al (Buscaglia, 2006).

Some reasons for this behaviour could be taken into account. First of all, these results confirm our observations that the solid solution system BiFeO₃ – BaTiO₃ undergoes structural transformations with increasing content of BaTiO₃. The decrease of the ferroelectric-paraelectric transition temperature Tc observed for the solid solution (1-x)BiFeO₃ – xBaTiO₃ may be ascribed to the decrease in unit cell volume caused by the BaTiO₃ addition. Addition of Ba²⁺ having empty p orbitals, reduces polarization of core electrons and also the structural distorsion. The low value obtained for Cp at x = 0.3 is in accordance with the previous result indicating that ferroelectricity disappears in samples above x ~ 0.3 (Kumar, 2000).

At the same time, the diffused phase transitions for compositions with x > 0.15 could be explained in terms of a large number of A and B sites occupied by two different, randomly distributed cationic specimens in the perovskite ABO₃ lattice. Previous reports on the substituted lanthanum manganites indicate that the mismatch at the A site creates strain on grain boundaries which affect the physical properties of an ABO₃ perovskite (Maignan, 2000). Besides, the role of charge ordering in explaining the magnetotransport properties of the variable valence transition metals perovskite was emphasized (Jonker, 1953). Investigating the influence of the dopants and of the oxygen nonstoichiometry on spin dynamics and thermodynamic properties of the magnetoresistive perovskites, Tanasescu et al (Tanasescu, 2008, 2009) pointed out that the remarkable behaviour of the substituted samples could be explained not only qualitatively by the structural changes upon doping, but also by the fact that the magneto-transport properties are extremely sensitive to the chemical defects in oxygen sites.

Though the effects of significant changes in the overall concentration of defects is not fully known in the present system of materials, extension of the results obtained on substituted manganites, may give some way for the correlation of the electrical, magnetic and thermodynamic properties with the defect structure. The partial replacement of Bi³⁺ with Ba²⁺ cations acting as acceptor centers could generate supplementary oxygen vacancies as compensating defects, whereas the Ti⁴⁺ solute on Fe³⁺ sites could induce cationic vacancies or polaronic defects by Fe³⁺ → Fe²⁺ transitions. The presence of the defects and the change of the Fe²⁺/ Fe³⁺ ratio is in turn a function not only of the composition but equally importantly of the thermal history of the phase. Consequently, an understanding of the high temperature defect chemistry of phases is vital, if an understanding of the low temperature electronic and magnetic properties is to be achieved. To further evaluate these considerations, and in order to discriminate against the above contributions, experimental insight into the effects of defect types and concentrations on phase transitions and thermodynamic data could give a valuable help.

For discussion was chosen the compound Bi_0.90Ba_0.10Fe_0.90Ti_0.10O₃ for which strong magnetoelectric coupling of intrinsic multiferroic origin was reported (Singh, 2008). The results obtained in the present study by using EMF and solid state state coulometric titration techniques are shown in the following.

The recorded EMF values obtained under the open circuit condition in the temperature range 923-1273 K are presented in Fig. 7. The thermodynamic data represented by the relative partial molar free energies, enthalpies and entropies of the oxygen dissolution in the perovskite phase, as well as the equilibrium partial pressures of oxygen have been calculated and the results are depicted in Figs. 8-11. A complex behavior which is dependent on the temperature range it was noticed, suggesting a change of the predominant defects concentration for the substituted compound.

As one can see in Fig. 7, at low temperatures, between 923 and ~1000 K, EMF has practically the same value E=0.475 V. Then, Fig. 7 distinctly shows a break in the EMF vs. temperature relation at about 1003 K, indicating a sudden change in the thermodynamic parameters. A strong increase of the partial molar free energy and of the partial pressure of oxygen was observed until 1050 K (Figs. 8 and 9) which can be due to structural transformation related to the charge compensation of the material system. Then, on a temperature interval of about 40 K the increasing of the energies values is smaller. After ~ 1090 K a new change of the slope in the ΔG¯O2and logpO2variation is registered on a temperature interval of about 130 degrees, followed again, after 1223 K, by a sudden change of the thermodynamic data, the higher ΔG¯O2value being obtained at about 1260 K.

The break point at about 1003 K is mainly due to first order phase transition in Bi_0.90Ba_0.10Fe_0.90Ti_0.10O₃ associated with the ferroelectric to the paraelectric transition T_C. The 10% BaTiO₃ substitution reduces the ferroelectric transition temperature of BiFeO₃ with about 100K_. This transition is also evident from calorimetric measurements (Tanasescu, 2009). The less abrupt first order transition at 1050 K is qualitatively in concordance with the transition to the polymorph which was previously identified in the literature for BiFeO₃ at 1198-1203K (Arnold, 2010; Palai, 2008; Selbach, 2009).

In Fig. 10 we represented the partial molar free energies of oxygen dissolution obtained in this study for both Bi_0.90Ba_0.10Fe_0.90Ti_0.10O₃ and BiFeO₃ at temperatures lower than their specific ferroelectric transition temperatures. We would like to specify that in the case of BiFeO₃, the EMF measurements were performed at temperatures not higher than 1073 K due to the instability of BiFeO₃ at higher temperatures. As one can see in Fig. 10, at 923 K, the partial molar free energies of oxygen dissolution in BiFeO₃ and Bi_0.90Ba_0.10Fe_0.90Ti_0.10O₃ samples are near each other. With increasing temperature, the highest ΔG¯O2values were obtained for BiFeO₃, suggesting an increased oxygen vacancies concentration in this compound. The result could be explained by the fact that at low temperatures the conduction is purely intrinsic and the anionic vacancies created are masked by impurity conduction. As the temperature increases, conduction becomes more extrinsic (Warren, 1996), and conduction due to the oxygen vacancies surface. This fact is also evident from the density measurements (Kumar, 2000), as well as electron paramagnetic resonance studies on perovskites (Warren, 2006). The increased concentration of oxygen vacancies in BiFeO₃ is consistent with the large leakage current reported for BiFeO₃ (Gu, 2010; Qi, 2005; Palkar, 2002; Wang, 2004). The electrical characteristics (Qi, 2005) indicated that the main conduction mechanism for pure BFO was space charge limited, and associated with free carriers trapped by oxygen vacancies. The coexistence of Fe³⁺ and Fe²⁺ causes electron hopping between Fe³⁺ and Fe²⁺ ions, oxygen vacancies acting as a bridge between them, which increases the leakage current. According to the defect chemistry theory, doping BiFeO₃ with aliovalent ions should change the oxidation state of iron and the concentration of oxygen vacancies. Qi and coworkers (Qi, 2005) have suggested as possible mechanisms to achieve the charge compensation in the 4+ cation-doped material: filling of oxygen vacancies, decrease of cation valence by formation of Fe²⁺, and creation of cation vacancies. Based on our results, the doping with Ti⁴⁺ is expected to eliminate oxygen vacancies causing the decreasing of ΔG¯O2and logpO2values.

Further clarification could be achieved by determining ΔH¯O2and ΔS¯O2values in particular temperature ranges in which the partial molar free energies are linear functions of temperature. Comparing the values obtained for Bi_0.90Ba_0.10Fe_0.90Ti_0.10O₃ in the temperature interval of 943-1003 K with the corresponding enthalpies and entropies values of BiFeO₃ in the 923-1073 K range (Fig. 11) one can observe that for the substituted compound, ΔH¯O2and ΔS¯O2values strongly increase (with ~450 kJ mol^-1 and ~480 J mol^-1 K^-1 respectively). This finding can be explained by the relative redox stability of the B-site ions which seems to modify both the mobility and the concentration of the oxygen vacancies. It is interesting to note that increasing temperature, after the first transition point, the enthalpies and entropies values strongly decrease (with ~ 468 kJ mol^-1 and ~1.4 kJ mol^-1 K^-1 respectively) up to more negative values. The negative values obtained for the relative partial entropies of oxygen dissolution at high temperature are indicative for a metal vacancy mechanism. Above 1053 K both the enthalpy and entropy increase again with increasing the temperature. The thermal reduction for transition metals tends to be easier with Ba doping. These may explain the reason for the different behaviors at higher temperature zone. Besides, oxygen vacancy order also show contribution to the observed phenomena, the increasing of the enthalpy and entropy values being an indication that the oxygen vacancies distribute randomly on the oxygen sublattice.

In order to further evaluate the previous results, the influence of the oxygen stoichiometry change on the thermodynamic properties has to be examined. The variation of the thermodynamic data of oxygen deficient Bi_0.90Ba_0.10Fe_0.90Ti_0.10O_3-δ samples was analyzed at the relative stoichiometry change Δδ = 0.01. In Figures 12 (a) and (b), two sets of data obtained before and after the isothermal titration experiments are plotted. Higher ΔS¯O2and ΔG¯O2values are obtained after titration at all temperatures until 1223 K; above 1223 K, the values after titration are lower than the corresponding values before titration.

Regarding the changes of logpO2and ΔH¯O2corresponding to the temperature range of 1093-1223 K (Fig. 13), one can observe that for Bi_0.90Ba_0.10Fe_0.90Ti_0.10O_3, both the variations of enthalpy and entropy decrease with the stoichiometry change suggesting the increase in the binding energy of oxygen and the change of order in the oxygen sublattice of the perovskite-type structure. The values of the relative partial entropies of oxygen dissolution are negative and this is indicative for a metal vacancy mechanism (Töfield, 1974). Due to the large decrease inΔS¯O2, it is considered that the oxygen vacancies would not randomly distribute on all of the oxygen sites but they would be distributed to some particular oxygen sites. It is also possible that the vacancy distribution is related to some crystallographic distortions or ordering of metal sites.

Presently, however, further details and measurements of the energy and entropy of oxygen incorporation into BiFeO₃-based materials at different values of nonstoichiometry are necessary in order to make clear the vacancy distribution with the stoiochiometry change.

3.2.1. Phase composition and crystalline structure

The room temperature X-ray diffraction pattern obtained for the presintered sample corresponding to the mixture 1 (Bi_0.9La_0.1FeO₃) shows a single phase composition, consisting of the well-crystallized perovskite phase (Fig. 14(a)). A small Mn addition (x 0.1) does not change the phase composition. The increase of the manganese amount to x = 0.2 determines the segregation of a small amount of Bi₃₆Fe₂O₅₇ secondary phase identified at the detection limit. For x 0.4 also small quantities of Bi₂Fe₄O₉ was detected as secondary phase, indicating the beginning of a decomposition process (Fig 14(a)).

From the structural point of view the XRD data pointed out that all the samples exhibit hexagonal R3c symmetry, similar to the structure of the paternal non-modified BiFeO₃ compound. Similar to Bi_1-xBa_xFe_1-xTi_xO₃ solid solutions, the increase of the manganese content does not determine the change of spatial group. However, certain distortions clearly emphasized by the cancellation of the splitting of some characteristic XRD peaks take place. Thus, Fig. 14(b) shows the evolution of the profile and position of the neighbouring (006) and (202) peaks specific to the Bi_0.9La_0.1O₃ composition when Mn is added in the system. One can observe that an amount of only 10% Mn replacing Fe³⁺ in the perovskite structure is enough to eliminate the (006) peak in the characteristic XRD pattern. A shift of the position of the main diffraction peaks toward higher 2 values was also pointed out (for exampe the (002) peak shifts from 2 = 39.5 ° for Bi_0.9La_0.1FeO₃ to 2 = 39.82° for Bi_0.9La_0.1Fe_0.5Mn_0.5O₃).

The increase of the manganese concentration determines the decrease of both a and c lattice parameters (Fig. 15(a)) and therefore a gradual contraction of the unit cell volume (Fig. 15(b)). This evolution suggests that most of the manganese ions are more probably incorporated on the B site of the perovskite network as Mn⁴⁺, causing the decrease of the network parameters because of the smaller ionic radius of Mn⁴⁺ (0.60 Å), comparing with that one corresponding to Fe³⁺ (0.64 Å). These results are in agreement with those ones reported by Palkar et al (Palkar, 2003).

3.2.2. Microstructure

As for Bi_1-xBa_xFe_1-xTi_xO₃ ceramics, SEM analyses were performed firstly on the pellets surface thermally treated at 923 K/2h. The surface SEM image of the Bi_0.9La_0.1FeO₃ sample indicates the obtaining of a non-uniform and porous microstructure, consisting from grains of variable sizes and a significant amount of intergranular porosity (fig. 16(a)). For the sample with x = 0.20, the presence of the manganese in the system induces the inhibition of the grain growth process and has a favourable effect on the densification (Fig. 16(b)).

The further increase of the Mn content to x = 0.50 enhances the densification, but seems to have not anymore a significant influence on the average grain size. Thus, the Bi_0.9La_0.1Fe_0.5Mn_0.5O₃ ceramic shows a dense, fine-grained (average grain size of ~ 2 µm) and homogeneous microstructure with a monomodal grain size distribution (Fig. 16(c)).

The same trend of the decrease of the average grain size with the addition of both La and Mn solutes was also observed in the case of the ceramics sintered at 1073 K for 1 hour. Thereby, unlike the non-homogeneous, rather coarse-grained BiFeO₃ sample (Fig. 5(a)), the average grain size in the La-modified Bi_0.9La_0.1FeO₃ ceramic is of only ~ 4 m (Fig. 17(a)). The increase of Mn addition causes a further decrease of the average grain size to ~ 2 m and a tendency to coalescence of the small grains in larger, well-sintered blocks (Fig. 17(b) and 17(c)).

3.3.3. Thermodynamic properties of Bi_1-xLa_xFe_1-yMn_yO₃ (x=0.1; y=0.2; 0.3)

Emphasizing the role of charge ordering in explaining the magnetotransport properties of the manganites, Jonker and van Santen considered that the local charge in the doped manganites is balanced by the conversion of Mn valence between Mn³⁺ and Mn⁴⁺ and the creation of oxygen vacancies, as well (Jonker, 1953). Investigating the influence of the dopants and of the nonstoichiometry on spin dynamics and thermodynamic properties of the magnetoresistive perovskites, Tanasescu et al (Tanasescu, 2008, 2009) demonstrated that the formation of oxygen vacancies and the change of the Mn³⁺/ Mn⁴⁺ ratio on the B-site play important roles to explain structural, magnetic and energetic properties of the substituted perovskite.

In BiFeO₃-BiMnO₃ system was already pointed out that, even though the Mn substitution does not alter the space group of BiFeO₃ for x ≤ 0.3, the possible variation of the valence state of Mn manganese together the oxygen hiperstoistoichiometry as a function of temperature and oxygen pressure could affect the crystallographic properties, electrical conductivity and phase stability of BiFe_1-xMn_xO_3+δ (Selbach, 2009). Excepting the communicated results on DSC investigation of Bi_1-xLa_xFe_1-yMn_yO₃ (x=0.1; y=0-0.5) (Tanasescu, 2010), no other work related to the thermodynamic behaviour of Bi_1-xLa_xFe_1-yMn_yO₃ were reported in the literature. In that study the evolution of heat of transformation and heat capacity in the temperature range of 573 – 1173 K was analyzed. The ferroelectric transition was shifted to a lower temperature for Bi_0.9La_0.1FeO₃ comparative to BiFeO₃, in agreement with literature data (Chen, 2008). However, a non-monotonous change of T_C, as well as of the thermochemical parameters is registered for the La and Mn co-doped compositions, depending on the Mn concentration, comparatively with undoped BiFeO₃. A sharp decline in the Tc was pointed out for x=0.3. One reason to explain this behavior is sustained by the structural results which were already pointed out in the previous section. The increase of the manganese concentration determines the decrease of both a and c lattice parameters and therefore a gradual contraction of the unit cell volume. This evolution suggests that most of the manganese ions are more probably incorporated on the B-site as Mn⁴⁺. Besides, as already was shown, in our samples the decreasing of the average grain size with the addition of both La and Mn solutes was observed in the case of presintered, as well as ceramics sintered at 1273 K (Ianculescu, 2009; Tanasescu, 2010). For finer particles where defect formation energies are likely to be reduced, the lattice defects, oxygen nonstoichiometry etc. appear to be sizable and significant changes in overall defect concentration are expected. So, an excess of Mn⁴⁺ ions and an increased oxygen nonstoichiometry are more likely. Due to the linear relationship between the Mn-O distortion and the Mn³⁺ content, one could expect to find in our samples a strong dependence of the energetic parameters on the Mn³⁺/ Mn⁴⁺ ratio and the oxygen nonstoichiometry.

In order to understand how the thermodynamic properties are related to the oxygen and manganese content in the substituted BiFeO₃, the thermodynamic properties represented by the relative partial molar free energies, enthalpies and entropies of oxygen dissolution in the perovskite phase, as well as the equilibrium partial pressures of oxygen have been obtained in a large temperature range (923-1123 K) by using solid electrolyte electrochemical cells method.

The obtained results are plotted in Figures 18 - 20. At low temperatures, between 923 and ~950 K for x=0.3 and between 923 and ~1000 K for x=0.2, theΔH¯O2values are increasing with temperature (Fig. 18(a)). The same trend is accounted for the ΔS¯O2variation (Fig. 18(b)). The break points at ~963 K and ~993 K obtained for x=0.3 and x=0.2, respectively (Fig. 18 (a)) are consistent with the Tc values of ferro-para transition in substituted samples. These values are near the ferroelectric Curie temperatures reported in the literature for BiFe_0.7Mn_0.3O₃ (926-957 K) and BiFe_0.8Mn_0.2O₃ (~990 K) with no La addition (Selbach, 2009; Sahu, 2007). However we have to notice that at the same Mn concentration x=0.3, the Tc value is slightly higher for the sample in which lanthanum is present, comparatively with the sample without La.

In Fig. 19 we represented the partial molar free energies of oxygen dissolution obtained in this study for Bi_0.9La_0.1Fe_0.8Mn_0.2O₃ and Bi_0.9La_0.1Fe_0.7Mn_0.3O₃ samples at temperatures lower than their specific ferroelectric transition temperatures, and for comparison, the ΔG¯O2values of BiFeO₃. As one can see in Fig. 19, the partial molar free energies of oxygen dissolution in the substituted samples are highest than thelogpO2values obtained for BiFeO₃, suggesting the increasing oxygen vacancies concentration with doping. Determining the ΔG¯O2and ΔG¯O2values in particular temperature ranges until T_C in which the partial molar free energies are linear functions of temperature (Figs. 19 and 20) one can observe that for the substituted compounds, the ΔH¯O2and ΔS¯O2values strongly increase in the doped compounds. Due to the relative redox stability of the B³⁺ ions (at the same A-site composition), both the mobility and the concentration of the oxygen vacancies will be modified. However it was noticed that the increase in B-site substitution from x = 0.20 to x = 0.30 at temperatures below 1000 K is followed by the decrease of ΔH¯O2and ΔS¯O2values with ~117.6 kJ mol^-1 and ~126 J mol^-1 K^-1 respectively (Fig. 20), suggesting the increase of the binding energy of oxygen and an increase of order in the oxygen sublattice of the perovskite-type structure with the Mn concentration.

After the ferroelectric transition temperatures and until 1043 K (for x=0.2), respectively until 1023 K (for x=0.3), a sharp decrease of theΔG¯O2values (Fig. 18(a)), together positive values of the enthalpies and entropies are revealed for both samples, indicating the decreasing of the thermodynamic driving force for oxygen vacancies formation and low ionic mobility. Taking into account the working conditions and the existing information as concerns the phase correlations of systems (Selbach, 2009, Carvalho, 2008), the dissociation reaction of the perovskite could be assumed to proceed in this temperature interval, Bi_0.9La_0.1Fe_1-xMn_xO₃ being in equilibrium with other two solid phases, namely sillenite and mullite type-phases (Bi₂₅Fe_1-yMn_yO₃₉ and Bi₂Fe_4-zMn_zO₉, respectively). Increasing the temperature until 1083 K (for x=0.2) and 1073 K (for x=0.3), the equilibrium will be driven back to the perovskite formation, the ΔH¯O2values of the sample with x=0.3 keeping higher than those of the sample with x=0.2 (Fig. 18(a)). At 1083 K (for x=0.2) and 1073 K (for x=0.3), the registered ΔS¯O2values have practically the same values as for the samples before decomposition. The next phase transition registered at 1093 K for Bi_0.9La_0.1Fe_0.7Mn_0.3O₃ is qualitatively in concordance with the transition to the polymorph which was previously identified in the literature for BiFeO₃ at 1198-1203K (Arnold, 2010; Palai, 2008; Selbach, 2009) and for BiFe_0.7Mn_0.3O₃ at 1145-1169 (Selbach, 2009). The temperature of transition to the γ phase corresponding to Bi_0.9La_0.1Fe_0.8Mn_0.2O₃ could be higher than 1123 K; it was not registered in our present experiment because the highest temperature of these measurements was 1123 K.

The obtained results evidenced the complex behavior of the partial molar thermodynamic data in substituted samples, suggesting a change of the predominant defects concentration as a function of temperature range and Mn concentration. Increasing the manganese content, the decreasing of the ferroelectric Curie temperature and of the transition temperature from paraelectric to γ phase it is noted.

To further evaluate the previous results, the influence of the oxygen stoichiometry change on the thermodynamic properties has been investigated. The variation of the partial molar thermodynamic data of Bi_0.9La_0.1Fe_0.8Mn_0.2O₃ (noted as BLFM0.2) and Bi_0.9La_0.1Fe_0.7Mn_0.3O₃ (noted as BLFM0.3) was examined before and after two successive titrations by the same relative oxygen stoichiometry change of Δδ = 0.02 in the oxygen excess region (Figures 21 and 22). Thus, the effect of the oxygen stoichiometry can be correlated with the influence of the substituent.

For x=0.2 higher values of the partial molar thermodynamic data are obtained after titration (Fig. 21). It is expected that the change in ΔG¯O2with δ in this case to be essentially determined by the change in S_O (config) and, therefore, the oxygen randomly distribute on the oxygen sites. Instead for x=0.3, both the variations of enthalpy and entropy decrease with the stoichiometry change (Fig. 22(b)), suggesting the increase in the binding energy of oxygen and change of order in the oxygen sublattice of the perovskite-type structure comparatively with the undoped compound. However it is interesting to note that the enthalpies obtained for x=0.3 after the first and the second titrations are near each other, suggesting a smaller dependence of the ΔH¯O2on the oxygen stoichiometry change at higher departure from stoichiometry. This result tends to agree with the assumption that metal vacancies prevail, since a value for enthalpy which is independent of nonstoichiometry is expected for randomly distributed and noninteracting metal vacancies (van Roosmalen, 1994; Tanasescu, 2005). The model based on excess oxygen compensated by cation vacancies and partial charge disproportionation of manganese ions was also proposed for other related systems, like LaMnO_3+δ (van Rosmallen, 1995; Töpfer, 1997), BiMnO_3+δ (Sundaresan, 2008), BiFe_0.7Mn_0.3O_3+δ (Selbach, 2009)_, La_0.5Bi_0.5Mn_0.5Fe_0.5O_3+δ (Kundu, 2008). However, the model could not explain the observed relationship in the entire oxygen-excess region. This statement was also discussed in the case of LaMnO_3+δ (Mizusaki, 2000; Nowotny, 1999; Tanasescu 2005) and could be subject for further discussion.

Considering the partial pressure of oxygen as a key parameter for the thermodynamic characterization of the materials, we investigated the variation of ΔS¯O2with the temperature, oxygen stoichiometry and the concentration of the B-site dopant (Fig. 23). Before titration, the ΔS¯O2values of the Bi_0.9La_0.1Fe_0.7Mn_0.3O₃ are higher than ΔH¯O2for Bi_0.9La_0.1Fe_0.8Mn_0.2O₃, excepting the value at 1123 K which is smaller for x=0.3 comparatively with the corresponding value for x=0.2, the result being correlated with the structural phase transformation noted for the composition with x=0.3 under 1123 K.

It is obtained that for both compounds, after titration, at the same deviation of the oxygen stoichiometry, logpO2shifted to higher values with increasing temperature. At the same temperature, the high deviation in the logpO2values with the stoichiometry change is obtained for the sample with x=0.2. Besides, at x=0.2, higher values of logpO2are obtained after the second titration, even though, increasing deviation from stoichiometry, a smaller increase of the partial pressure of oxygen was noted (Fig. 23). This could be explained by the fact that at high temperature, less excess oxygen is allowed. The sample with x=0.3 presents a smaller dependence of the logpO2on the oxygen nonstoichiometry. For small deviation from stoichiometry (Δδ=0.02), a small decrease in logpO2values is obtained, but after the second titration, the partial pressure increase again. At 1073 K, after the second titration, the same value as before titration is obtained. Increasing temperature to 1123 K, logpO2values increase again comparatively with the value before titration.

The obtained results could be correlated with some previously reported conductivity measurements. Singh et al (Singh, 2007) reported that small manganese doping in thin films of BiFeO₃ improved leakage current characteristic in the high electric field region, reducing the conductivity; others authors noted the increasing of the conductivity with increasing manganese content (Chung, 2006; Selbach, 2009, 2010). If the polaron hopping mechanism is supposed for the electrical conductivity at elevated temperatures, the electronic conductivity will increase in the samples with hiperstoichiometry. According to the evolution of the partial molar thermodynamic data of the oxygen dissolution, a decrease in the oxygen ionic conductivity (together the increasing of the electronic conductivity due to the electron-hole concentration increasing) will result in the sample with increased Mn content.

Even though there are disagreements between different works regarding the nature and the symmetry of the high temperature phases in the pure and substituted BiFeO₃, based on our data, we would like to point out that, in the condition of our experimental work, we may close to the stability limit of the Mn doped materials at temperatures around 1123 K. This is in accordance with theoretical consideration of the stability of ABO₃ compounds based on Goldschmidt tolerance factor relationship (Goldschmidt, 1926). The evolution with temperature and oxygen stoichiometry of the thermodynamic data suggest that excess oxygen causing an increase of the tolerance factor of the system will lead to the stabilization of the cubic phase at lower temperature with increasing the departure from stoichiometry.

At this point further studies are in progress, so that correlations could be established with the observed properties at different departures of oxygen stoichiometry, in both deficit and excess region for Bi_0.9La_0.1Fe_1-xMn_xO₃ materials.