The Effect of Mechanical Activation on the Synthesis and Properties of Multiferroic Lead Iron Niobate

Abstract

The present work studies the effect of high-energy mechanical activation using the planetary-centrifugal ball mill AGO-2 and subsequent annealing on the synthesis and magnetic properties of PbFe_0.5Nb_0.5O₃ (PFN). This technique enables one to perform the mechanically activated synthesis of PFN at much shorter time. The results of X-ray phase analysis, electron microscopic studies and transmission Mössbauer ⁵⁷Fe spectra measurement are presented and discussed. Mössbauer studies show that the temperature of magnetic phase transition in PbFe_0.5Nb_0.5O₃ powders can be changed by mechanical activation and subsequent annealing.

The present work studies the effect of high-energy mechanical activation using the planetary-centrifugal ball mill AGO-2 and subsequent annealing on the synthesis and magnetic properties of PbFe_0.5Nb_0.5O₃ (PFN). This technique enables one to perform the mechanically activated synthesis of PFN at much shorter time. The results of X-ray phase analysis, electron microscopic studies and transmission Mössbauer ⁵⁷Fe spectra measurement are presented and discussed. Mössbauer studies show that the temperature of magnetic phase transition in PbFe_0.5Nb_0.5O₃ powders can be changed by mechanical activation and subsequent annealing.

1 Introduction

Lead iron niobate PbFe_0.5Nb_0.5O₃ (PFN) is a ternary perovskite oxide possessing both ferroelectric (ferroelectric Curie temperature T _C  ≈ 380 K) and magnetic (temperature of antiferomagnetic phase transition, i.e. Neél temperature, T _N  ≈ 150 K) properties [1]. PFN is a promising basic material for multilayer capacitors, pyroelectric sensors, piezoelectric ceramics and electrostrictive actuators [2–6]. It is also a component of some novel room-temperature multiferroic magnetoelectrics [7].

The temperature of magnetic phase transition in Fe-containing oxides is believed to depend on the number of possible Fe–O–Fe linkages in a crystal lattice [8]. In ternary PbFe_0.5 B _0.5O₃ perovskites this number can be governed, e.g. by changing the degree S of B-cation ordering, as the ordering changes the number of magnetic ions in the neighboring unit cells and thus reduces the possible number of Fe–O–Fe linkages. However in contrast to ${\text{Pb}}B^{3 + } \,_{0.5} {\text{B}}^{5 + } \,_{0.5} {\text{O}}_{3}$ perovskites with $B^{ 3+ } = {\text{Sc}},{\text{Yb}};B^{ 5+ } = {\text{Nb}},{\text{Ta}}$, [9, 10], no superstuctural reflections on X-ray diffraction patterns due to B-cation ordering have been reported for PFN [8, 11]. At the same time, the experimental value of T _N , for PFN (≈150 K) is located approximately half-way between calculated values of this temperature for the fully ordered (T _N  = 0 K) and completely disordered (T _N  ≈ 400 K) states [8]. This fact is usually interpreted as an evidence of a partial ordering of Fe³⁺ and Nb⁵⁺ cations [8]. The possible reason of the absence of superstuctural reflections on X-ray diffraction patterns is a local character of ordering in PFN, i.e. mesoscopic domains of several nanometers in size with different values of S coexist in a crystal. Another evidence of the chemically inhomogeneous nature of PFN was obtained from the studies of magnetization [1, 11], ⁹³Nb and ¹⁷O NMR [12] and Mössbauer ⁵⁷Fe spectra [13], diffuse neutron scattering [11] as well as from the results of the first-principal calculations [13]. Long range AFM order seems to develop in Fe-rich-Nb-poor regions while, a short-range ordered spin-glass-like state can arise at low (T ≈ 10–20 K) temperatures from the Fe-poor-Nb-rich regions. AFM and spin-glass states can coexist in a wide temperature range [14]. Magnetic phase transition temperature in PFN is more than 100 K higher as compared to lead-free AFe_0.5Nb_0.5O₃ (A = Ca, Sr, Ba) perovskites [15, 16]. This dramatic difference is attributed to the possibility of magnetic super-exchange via an empty 6p state of Pb²⁺ ions [15, 16] as well as for the clastering of iron ions in the lattice [13]. Both the degree of B-cation compositional ordering and clastering of iron ions are known to be influenced by mechanical activation, which has been successfully used for the synthesis and preparation of ternary lead-based perovskite ferroelectrics [17, 18]. This method enables one to obtain nanodispersed powders with high sintering ability thus lowering substantially the sintering temperature of ceramic materials. We have found in the literature two works devoted to the application of the mechanical activation method for synthesis of PFN [19, 20]. However, the effect of mechanical activation on magnetic properties of PFN has not been studied yet. Moreover, the energy of mechanical activation, used in [19, 20], was rather modest and the activation time varied from 20 to 30 h. Recently the planetary-centrifugal ball mill AGO-2 enabling activation with much higher energy than in [19, 20] was designed [17]. The aim of the present work was to study the effect of high-energy mechanical activation using AGO-2 mill and subsequent annealing on the synthesis and magnetic properties of PFN.

2 Experimental

In order to compare the effect of high-energy milling on PFN synthesis, two batches were used: stoichiometric mixture of PbO, Fe₂O₃ and Nb₂O₅ oxide powders (I) and stoichiometric mixture of PbO and preliminary synthesized FeNbO₄ precursor (II). The starting oxides were Nb₂O₅ (the American Society for Testing Materials (ASTM) code 16-53, 37-1468), monoclinic modification, space group P2 (no. 3), OSCh reagent grade (high pure); Fe₂O₃ (ASTM code 85-599) hematite, rhombohedral modification, space group R-3C (no. 167), ChDA reagent grade (pure for analysis); PbO (ASTM code 72-93) massicot, orthorhombic modification, space group Pbcm (no. 57), ChDA reagent grade (pure for analysis), with Pb₃O₄ admixture (ASTM code 41-1493) minium, tetrahedral modification, space group P42/mbc (no. 135). The mechanical activation was carried out using the high-energy planetary-centrifugal ball mill AGO-2 under a ball acceleration of 40 g. A mixture of powdered reagents (10 g) was placed into steel cylinder together with 200 grams of steel balls 8 mm in diameter. Activation was carried out for 5, 10 and 15 min. After each 5 min of activation, the mill was stopped, the cylinders were opened, the powder was taken out and mixed then it was put back into the cylinders for further mechanical activation. The samples for subsequent annealing were pressed at 1000 kg/cm² without a plasticizer. Annealing of the samples, placed into a closed alumina crucible, was carried out in an electric oven at different temperatures for 2 h. Heating rate was 10 deg/min. The samples were cooled together with the furnace after it was switched off. X-ray phase analysis was performed using DRON-3 diffractometer and Cu–K_α radiation. Electron microscopic studies were carried out with a transmission electron microscope (TEM) JEM-2000FX II (JEOL) at the accelerating voltage of 200 kV. Transmission Mössbauer ⁵⁷Fe spectra were measured with the aid of MS-1104EM rapid spectrometer and analyzed using the original computer program UNIVEM. The sample holder was attached to the closed-cycle helium cryostat-refrigerator Janis Ccs-850 operating in the 12–320 К range.

3 Results and Discussion

The diffraction patterns of the samples of stoichiometric mixture of PbO, Fe₂O₃ and Nb₂O₅ oxide powders (I) before activation (a) and after activation for 5, 10, and 15 min (b)–(d), respectively, are shown in Fig. 2.1. Amorphization process indicated by the broadening of the diffraction reflexes and the initial stage of the formation of perovskite structure of PFN are well seen even after 5 min of activation. After 15 min of activation, only the lines corresponding to the perovskite structure are well seen in the diffraction pattern. It is worth noting that the diffraction pattern of the mixture I activated for 15 min using the high-energy planetary-centrifugal ball mill AGO-2 is very similar to the patterns reported in [19, 20] for the same mixtures activated in a shaker mill for 30 h or in a SPEX 8000D vibrating mill for 20 h, respectively.

Fig. 2.1

Diffraction patterns illustrating the effect of mechanical activation on the stoichiometric mixture of PbO, Fe₂O₃ and Nb₂O₅ (mixture I): a initial mixture, b after mechanical activation for 5 min, c after mechanical activation for 10 min, d after mechanical activation for 15 min; ▼—Nb₂O₅; +—PbO; ●—Fe₂O₃; *—PFN

One can see in Fig. 2.1d that besides the PFN reflexes, there are also two minor reflexes corresponding to Fe₂O₃ indicating that the synthesis of PFN is not complete. One can also expect that some amount of FeNbO₄ is present, because it appears in the activated mixture of Fe₂O₃ and Nb₂O₅ (Fig. 2.2b). However, due to a considerable widening of reflexes on the diffraction pattern, no PbO, Nb₂O₅, and FeNbO₄ reflexes could be unequivocally identified. The mean size of X-ray coherent scattering blocks, estimated using the Scherrer formula for the mixture I activated for 15 min, is about 13–20 nm. This value correlates well with the TEM images of the same powder. To have a deeper insight into the phase content of the mechanically activated mixture I we studied its room-temperature Mössbauer spectra before and after the activation (Fig. 2.3).

Fig. 2.2

X-ray diffraction patterns illustrating the stages of the FeNbO₄ synthesis: a initial mixture of Fe₂O₃ and Nb₂O₅, ●—Fe₂O₃, other reflections belong to Nb₂O₅; b after mechanical activation for 15 min, O—reflections of FeNbO₄; c after annealing at 1000 °C, ●—Fe₂O₃, other reflections belong to FeNbO₄−; d mixture of FeNbO₄ and PbO, +—basic reflections of PbO, †—basic reflections of Pb₃O₄, other reflections belong to FeNbO₄

Fig. 2.3

Room-temperature Mössbauer ⁵⁷Fe spectra: a initial mixture of PbO, Fe₂O₃ and Nb₂O₅ (mixture I); b Fe₂O₃ powder after mechanical activation for 15 min; c mixture of PbO, Fe₂O₃ and Nb₂O₅ after mechanical activation for 15 min

The spectrum of the initial mixture I is a sextet corresponding to Fe₂O₃ with a hyperfine field of 51.4 T. After mechanical activation of Fe₂O₃ powder for 15 min the mean size of X-ray coherent scattering blocks decreased from 35 to 14 nm and the sextet lines have become lower and broader (Fig. 2.3b). After mechanical activation of the initial mixture I for 15 min in the room-temperature Mössbauer spectra, besides sextet with a hyperfine field of 51.7 T, there it appears a doublet corresponding to PFN with admixture of FeNbO₄. The analysis of Mössbauer spectrum shows that 71.5 % of Fe remains in the Fe₂O₃ phase while 27.5 % occurs in the PFN or FeNbO₄ phase. Similar ratio of the Fe₂O₃ and PFN/FeNbO₄ phases was reported for the stoichiometric mixture of PbO, Fe₂O₃ and Nb₂O₅ activated in SPEX 8000D vibrating mill for 20 h [20].

The diffraction patterns of the disk samples pressed from the mixture I activated for 15 min and annealed for 2 h at different temperatures within 400–1000 °C range are shown in Fig. 2.4. After annealing at 400 °C, the diffraction patterns recorded from this sample exhibit almost no difference as compared to the patterns of unannealed mechanically activated sample (Fig. 2.1d). Only after annealing at 500 °C diffraction patterns start to differ from each other noticeably. One can see in Fig. 2.4a that the growth of Pb₂Fe₄Nb₄O₂₁ phase is prevailing. After annealing at 600 °C, the PFN phase starts to prevail (Fig. 2.4b), and after annealing at 800 °C its amount becomes much larger (Fig. 2.4c). However, after annealing at 900 °C again the growth of the amount of the Pb₂Fe₄Nb₄O₂₁ phase starts to dominate (Fig. 2.4d). After annealing at 1000 °C this phase is the major one (Fig. 2.4e) with small reflections of the PFN phase. Even after polishing (Fig. 2.4f) this is not pure PFN, but it contains noticeable amount of Pb₂Fe₄Nb₄O₂₁, that is, there is obviously a lack of lead oxide in the samples. Thus it seems that for obtaining the pure perovskite phase using the mechanical activation of the PbO, Fe₂O₃ and Nb₂O₅ oxides one needs to add an excess of PbO to the initial stoichiometric mixture.

Fig. 2.4

X-ray diffraction patterns of the samples pressed from the mixture of PbO, Fe₂O₃ and Nb₂O₅ (mixture I) activated for 15 min, after annealing for 2 h at different temperatures: a 500 °C, b 600 °C, c 800 °C, d 900 °C, e 1000 °C, f 1000 °C, polished sample; *—Pb(Fe_0.5Nb_0.5)O₃, ♦—Pb₂Fe₄Nb₄O₂₁

To synthesize FeNbO₄ precursor, necessary for obtaining the mixture II, a stoichiometric mixture of Fe₂O₃ and Nb₂O₅ corresponding to FeNbO₄ composition (Fig. 2.2a) was activated for 15 min. One can see in Fig. 2.2b that after such activation the reflexes on the diffraction pattern become substantially diffused and chemical interaction between initial components occurs, that is, reflexes corresponding to FeNbO₄ appear. The powder activated for 15 min was annealed at 1000 °C for 2 h. Iron niobate FeNbO₄ of monoclinic modification with space group P*/a (no. 13) was formed, but a small amount of unreacted initial Fe₂O₃ still remained (Fig. 2.2c).

Then sintered FeNbO₄ was crumbled and mixed with lead oxide β-PbO. It is known that the compositions of lead oxides may differ from the stoichiometry. One can see in the diffraction patterns (Fig. 2.1d) that a small amount of red tetragonal β-Pb₃O₄ is present in yellow rhombic lead oxide β-PbO. FeNbO₄ and PbO were averaged in a mortar, and then activated in the mill for 15 min. One can see in the diffraction patterns (Fig. 2.5a) that the product formed at once as a result of mechanical activation is single-phase cubic modification of PFN (ASTM code 32-522), space group Pm3 m (no. 221).

Fig. 2.5

X-ray diffraction patterns illustrating the effect of annealing on the mechanically activated stoichiometric mixture of PbO and FeNbO₄ (mixture II): a mixture of PbO and FeNbO₄ mechanically activated for 15 min, b after annealing at 700 °C, c after annealing at 800 °C (pure PFN), d after annealing at 1000 °C, e after annealing at 1000 °C and subsequent polishing the sample (pure PFN); ♦—Pb₂Fe₄Nb₄O₂₁, ¤—PbFe₁₂O₁₉, *—PFN

During annealing the mechanically activated mixture of PbO and FeNbO₄ within temperature range 400–700 °C, along with PFN, a small amount of the cubic phase Pb₂Fe₄Nb₄O₂₁ (ASTM code 50-445), space group F, (no. 0) is formed in the product (Fig. 2.5b). After annealing at 800 °C, the traces of this phase disappear from the diffraction pattern (Fig. 2.5c). The samples annealed at 1000 °C (Fig. 2.5d) exhibit the appearance of magnetoplumbite phase PbFe₁₂O₁₉ (ASTM code 84-2046) of hexagonal modification with space group P63/mmc (no. 194). The nucleation of this phase starts at a temperature about 900 °C and increases with an increase in annealing temperature. However, this phase is formed predominantly on the surface of the samples. After grinding off the surface layer of 0.1 mm thick, no traces of this phase are observed in the diffraction patterns (Fig. 2.5e).

After mechanical activation for 15 min, the both mixtures I and II have rather small mean size D of the X-ray coherent scattering blocks (10–20 nm) and their X-ray diffraction patterns are diffused. Annealing at 400 °C and 500 °C does not change substantially neither the D values nor the degree of X-ray diffraction reflexes’ diffusion. Annealing at higher temperatures leads to a fast growth of D values (Fig. 2.6) and sharpening of the X-ray diffraction reflexes with an increase of annealing temperature T _a . However above approximately 800 °C the D(T _a ) dependence saturates. At room temperature, Mössbauer ⁵⁷Fe spectra of all compositions studied appear to be doublets with the isomer shift of ≈0.4 mm/s (relative to the metallic iron). This value corresponds to the Fe³⁺ ions occupying the octahedral sites of the perovskite lattice. The quadrupole splitting in PFN can be caused by the asymmetric environment of the intraoctahedral Fe³⁺ ions resulting due to local heterogeneities in the compositional ordering of the Fe³⁺ and Nb⁵⁺ ions. This assumption is supported by the fact that, apart from a doublet, the Mössbauer spectrum of highly ordered ternary perovskite PbFe_0.5Sb_0.5O₃ contains also a singlet, which is attributed to compositionally ordered regions, whereas the doublet is related to the regions where a long-range ionic order is disturbed [21]. Quadrupole splitting Q of the Mössbauer ⁵⁷Fe doublet spectra of the mechanoactivated PFN powders with small (10–20 nm) mean size D of the X-ray coherent scattering blocks is substantially larger (Fig. 2.6) than that of the single crystal and ceramic samples, in which it is approximately the same (≈0.4 mm/s) [22]. As was already mentioned, annealing of the mechanoactivated PFN powders leads to the growth of D values. This growth is accompanied by gradual decrease of Q values with increasing the annealing temperature T _a (Fig. 2.6). At T _a  > 800 °C, the quadrupole splitting reaches the level typical of PFN single crystals and ceramics. It is worth noting that large values of quadrupole splitting were reported for the amorphous PFN and they also decreased as a result of annealing leading to the crystallization of the sample [23].

Fig. 2.6

Dependences of the mean size D of the X-ray coherent scattering blocks and the value of the quadrupole splitting Q of Mossbauer ⁵⁷Fe spectra on the annealing temperature T _a for the mixture of PbO and FeNbO₄ mechanically activated for 15 min

Now, we turn to the temperature dependence of the Mössbauer spectrum. Below T _N , at cooling, the Mossbauer spectrum transforms from the doublet to a sextet. This transformation is accompanied by a dramatic decrease of the doublet intensity normalized to its value measured at room temperature (Fig. 2.7). The position of the abrupt drop in the temperature dependence of the doublet intensity allows one to obtain the value of T _N from the Mössbauer experiment. This method was successfully used to determine values of T _N in several multiferroics and their solid solutions and the results obtained were very similar to the data got by traditional methods such as the magnetization or magnetic susceptibility measurements (see [22] and references therein). Thus, we have employed this opportunity to detect the temperature of magnetic phase transition in mechanically activated powders with the help of Mössbauer spectroscopy.

Fig. 2.7

Dependences of the doublet intensity in the Mössbauer ⁵⁷Fe spectrum normalized to its value measured at room temperature for the powders of the stoichiomentic mixture of PbO and FeNbO₄ mechanically activated for 15 min and subsequently annealed at different temperatures. Numbers near the curves represent the corresponding annealing temperature. The dashed line shows the data for PFN single crystal

Temperature dependencies of the normalized intensity of doublet in Mössbauer spectra for the powders of the stoichiomentic mixture of PbO and FeNbO₄ mechanically activated for 15 min and subsequently annealed at different temperatures are shown in Fig. 2.7. The steps on these dependencies correspond to the temperature of magnetic phase transition. For comparison, results for the PFN crystal that was chosen as a reference are shown. One can see that there are two steps on these dependencies for the powders annealed at 400 and 500 °C: one of the steps is at about 40–50 K, and the second one is at higher temperature which depends on annealing temperature T _a and shifts upward with an increase in T _a . The powders annealed at 700 °C and at higher temperatures exhibit only one step. The highest temperature of magnetic phase transition possesses the powder annealed at 700 °C. With an increase in T _a above 700 °C, T _N starts to decrease, that is, the effect of mechanical activation is annealing, so that for T _a  = 1000 and 1100 °C, the T _N value almost coincides with the one for the PFN crystal.

The low temperature of magnetic phase transition in the unannealed mechanically activated powder is likely to be connected with its «amorphization» . The magnetic state in this powder is, most probably, of the spin-glass type, which means that the magnetic order is only short-range. Actually, as this powder gives a normal diffraction pattern, though with broadened lines, this most likely means that the sizes of the regions of the perovskite phase are very small. Annealing is accompanied by an increase in the size of these regions, and the temperature of magnetic transition increases; the formation of long-range antiferromagnetic order is likely to occur. As we suppose, the T _N value depends also on the degree of iron clustering. Presently it seems that this degree varies comparatively slightly during annealing at temperatures below 700 °C, while at higher T _a the degree of iron clustering starts to decrease (Fig. 2.8).

Fig. 2.8

Dependences of the mean size D of the X-ray coherent scattering blocks and the Neél temperature T _N on the annealing temperature T _a for the mixture of PbO and FeNbO₄ mechanically activated for 15 min. The dashed line marks T _N value for PFN single crystal

Thus, the present hypothesis explaining the experiment is as follows: for T _a  < 700 °C the major factor determining the magnetic phase transition temperature is the coherent length, while for T _a  > 700 °C, it is the decrease in the degree of iron clustering with an increase in T _a .

In order to study if the changes of magnetic phase transition temperature observed for mechanically activated and subsequently annealed PFN powders are accompanied by the changes of the ferroelectric phase transition temperature, we studied the temperature dependence of the second harmonic generation intensity for the powder annealed at 700 °C and showing the largest increase of T _N as compared to PFN crystal. The experimental setup used was described in details elsewhere [24]. The results obtained are shown in Fig. 2.9. One can see that there is a distinct anomaly in the temperature dependence of the second harmonic generation intensity at about 110 °C. This temperature nicely correlates with the literature data for the ferroelectric-paraelectric phase transition temperature both in single crystals and ceramics of PFN [1, 7, 25]. The character of the dependence of the second harmonic generation intensity on temperature for the PFN powder studied shows some differences as compared to the similar dependencies for BaTiO₃ powders [24, 26]. However it is known that the shape of this curve may be changed crucially by changing the conditions of pressing the powders and their subsequent annealing [26].

Fig. 2.9

Temperature dependence of the normalized second harmonic generation signal for the powder of PbO and FeNbO₄ stoichiometric mixture mechanically activated for 15 min and subsequently annealed at 700 °C for 2 h

4 Summary

The use of the high-energy planetary-centrifugal ball mill AGO-2 enables one to perform the mechanically activated synthesis of PFN at much shorter time (15 min instead of 20–30 h).

Mössbauer studies have shown that the temperature of magnetic phase transition in PbFe_0.5Nb_0.5O₃ powders can be changed by mechanical activation and subsequent annealing. In particular, the increase by more than 50 K was achieved in the mechanically activated powder of stoichiometric PbO and FeNbO₄ mixture annealed at 700 °C. These results are in line with the model assuming that the temperature of magnetic phase transition in PbFe_0.5Nb_0.5O₃ depends substantially on the degree of Fe³⁺ and Nb⁵⁺ ions clustering [13]. At the same time the second harmonic generation studies did not reveal any substantial changes of the ferroelectric phase transition temperature in this powder.