Ferromagnetism evidence and size dependence in ferroelectric PZN-4.5PT nanoparticles

Among various everlasting desires of materials scientists, the most important is the research of materials with the best possible functionalities [1,2]. Transition metals have a very rich chemistry due to a variable oxidation state, which offers a multitude of interesting properties, if selected judicially. Numerous metal oxides have been investigated and are being studied for various technological applications [3–6]. To overcome the scarcity of single-phase multiferroics, various approaches have been used to artificially synthesize new multiferroic materials. With a steadily increasing interest to fabricate multiferroics, PbTiO₃ is being extensively looked into for composites of ferroelectric and ferromagnetic heterostructures, super-lattices composed of alternating ferroelectric and ferromagnetic layers and doping of suitable magnetic transition metals [7–10].

However, it has been reported in the literature that the reactivity and structural non-compatibility between the two materials or with the substrate generate difficulties in fabricating heterostructures and achieving coupling between the two order parameters.

Therefore, research has mainly focused on inducing magnetoelectric effects in the ferroelectric materials by suitable doping (Fe, Mn, Ni, etc.). Ferroelectric single crystals Pb(Zn_1/3Nb_2/3)O₃-PbTiO₃ (PZN-PT) may be promising full materials for such devices. Indeed, during the past few years, published results have pointed at high levels of piezoelectricity in Pb(Mg_1/3Nb_2/3)O₃-PbTiO₃ (PMN-PT) and PZN-PT single crystals [11–14], showing at least ferroelectric, ferroelastic and piezoelectric properties. However, the greatest difficulty in using such single crystals on electronic devices is to achieve them in thin layers form because of their incongruent melting property.

This paper describes ferroelectric 99.5Pb(Zn_1/3Nb_2/3)O₃-4.5PbTiO₃ (PZN-4.5PT) single crystals, synthesized by the flux method [12]. After having been ground into nanopowders, the undoped and Mn-doped PZN-4.5PT particles as well as bulk and oriented single crystals structural, ferroelectric and ferromagnetic properties were investigated.

Undoped and 1% Mn-doped PZN-4.5PT single crystals that had been synthesized by the flux method [12] were ground in a mortar to obtain a very fine nanopowder. To exclude contamination, grounding process was repeated in two different laboratories (in LVST, Paris 13, France and in LCPM, UASZ, Ziguinchor, Senegal, with different types of mortars fabricated with different chemical components (free-ferromagnetic elements: SiC (in LCPM, Senegal) and WC (in LVST, Paris, France). The as-synthesized powders were characterized by TEM analysis using a FEI CM10 microscope (Philips) by depositing a drop of nanoparticles suspension on carbon coated copper grids placed on a filter paper. Ferroelectric characterization was carried out on $1 \times 1 \times 0.1\ \text{cm}^{3}$ flat samples with two parallel faces sputtered with gold and oriented $\langle 110\rangle$. Poling was performed with an electric field of 12 kV/cm at room temperature. A modified Sawyer-Tower circuit at room temperature was used to measure the polarization vs. the DC field. From polarization-$E_{\mathrm{field}}$ curves, coercive field ($E_{\mathrm{c}}$), remanant polarization ($P_{\mathrm{r}}$) and spontaneous polarization ($P_{\mathrm{s}}$) were determined. The magnetic hysteresis loops were measured using a vibrating sample magnetometer (VSM) model Quantum Design Versalab for bulk, oriented bulk and nanoparticles for size effect. The results made possible to determine the saturation and remnant magnetization and coercivity.

Figure 1 presents TEM micrographs of the PZN-4.5PT (fig. 1(a) and fig. 1(b)) and PZN-4.5PT + 1% Mn (fig. 1(c) and fig. 1(d)) nanoparticles. According to these images, the undoped crystals consisted mainly of spherical nanoparticles with a diameter between 30 and 100 nm, whereas the doped ones were made up of both spherical nanoparticles with diameters ranging from 30 to 100 nm (fig. 1(c)) and nanosticks with lengths up to 200 nm and diameters smaller than 10 nm (fig. 1(d)).

Zoom In Zoom Out Reset image size

Figure 2 shows the ferroelectric and piezoelectric behavior of the PZN-4.5PT single crystals. Figure 2(a) and fig. 2(b) represent the P-E behavior hysteresis of, respectively, the undoped and Mn-doped PZN-4.5PT single crystals. $E_{\mathrm{c}}$ values are slightly the same around 430 V/mm. $P_{\mathrm{s}}$ and $P_{\mathrm{r}}$ decreased slightly (respectively from 0.31 to 0.28 C/m ²and from 0.25 to 0.22 C/m ²) with the doping. This effect can be explained by the fact that once the blocked dipoles were released by a relatively high field they struggled to return to their initial state when the field vanished.

Zoom In Zoom Out Reset image size

The acceptor dopants such as manganese are known to increase the values of coercive fields and stabilize the dipolar domain distribution within the material. The commonest explanation is linked to the existence of mobile oxygen vacancies that appear under stress and block the movement of domain walls making it more difficult to reverse the polarization and promoting the stability of domains induced by the polarization [15]. Figure 2(c) and fig. 2(d) show butterfly-shaped strain-electric field ($S\text{-}E$) curves. The initial loading cycle started at the point (0, 0). There was no strain response until the electric field reached the coercive field and the material switched to the polarized state. At point A, the material had positive remnant polarization and remnant strain. When a negative electric field was applied, the ions in the crystal structure developed a force that moved them toward a cubic state. When the coercive field was reached at point D, the field value was equal to $-E_{\mathrm{c}}$, the polarization direction began to switch. After the polarization had switched to the direction of the applied electric field, the latter acted to stretch the unit cell, giving a positive strain at point E.

When the electric field was subsequently reduced to zero, the remnant strain coincided with that at point A but the remnant polarization was negative. Thus, as a positive electric field was applied to the negatively polarized material, the ions in the crystal structure again developed a force that moved them toward a cubic state. The coercive field was reached at point B, and the polarization had completely switched direction again at point C. The electric field was reduced to zero and the material returned to point A with positive remnant polarization and remnant strain [16]. Mn doping decreased the relative strain maximum value from 0.25% ($2.5\cdot10^{-3}$) for the undoped crystals to a maximum of 0.20% ($2\cdot10^{-3}$) for their doped counterparts (fig. 2(e) and fig. 2(f)). These results confirmed the hardness effect of Mn-doping on the characteristics of PZN-PT single crystals.

Magnetic properties of bulks and nanopowders were investigated by measuring field dependence magnetization M(H) at room temperature.

Figure 3 shows the size dependence of PZN4.5PT nanpowders. It is clear from these curves that the magnetic behavior in PZN-4.5PT is related to the size of the crystals and would appear only at the nanoscale level (i.e., in form of powder) and could not be found in the bulk single crystals (which are paramagnetic, fig. 4(a)). There exists an optimized sizes range between 45 to 63 μm for which we have the highest value of the magnetization.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Indeed, due to a high surface-to-volume ratio of atoms in micro- or nanoparticles, a small amount of defects (in the form of cationic or anionic vacancies; formed due to a non stoichiometric ratios between anions and cations in binary semiconductors) is sufficient to show fascinating magnetic properties even when their bulk counterpart is non-magnetic [17] (fig. 4). One should recall that in all titanates, entropy requires that there are some oxygen vacancies. Ti⁴⁺ ions near such vacancies convert to Ti³⁺, which are magnetic. Therefore, titanates, especially in regions near domain walls (which trap oxygen vacancies) may exhibit unexpected magnetoelectric effects in substances such as SrTiO₃ (PbTiO₃ in our case) that nominally lack magnetic ions [18]. The mechanism that led to a ferromagnetic behavior being induced in the Ti ions would be T-center exchange (TCE) interactions of Ti³⁺-VO²⁻-Ti³⁺ groups (VO²⁻ is oxygen vacancy). Remnant magnetization ($M_{\mathrm{r}}$) and coercivity ($H_{\mathrm{c}}$) were measured at low field (fig. 4(b)) and were found equal respectively to $2 \times 10^{-4}\ \text{emu/g}$ and 63 Oe for undoped and, $7 \times 10^{-4}\ \text{emu/g}$ and 66 Oe for Mn-doped PZN-4.5PT indicating that the PZN-4.5PT nanoparticles possessed a weak ferromagnetic behavior. The values of the saturated magnetization ($M_{\mathrm s}$) were determined by extrapolating $1/H$ to zero field in the M vs. $1/H$ plot based on high field data (fig. 4(c)). $M_{\mathrm{s}}$ was found equal to $1.0 \times 10^{-3}$ and $7.5 \times 10^{-3}\ \text{emu}{}\cdot{}\text{g}^{-1}$ for the non-uniform undoped and Mn-doped powders. These values are similar to those found in BiFeO₃ ($\approx4 \times 10^{-2}\ \text{emu}{}\cdot{}\text{g}^{-1}$) [19]. The presence of an unsaturated hysteresis loop and small remnant magnetization for Mn-doped particles (fig. 4(b)) indicated the canting of the antiferromagnetic spin structure due to size mismatch and/or a ferromagnetic spin structure caused by a different magnetic nature of the substituted Mn [20] as compared to that of the Ti³⁺ ions. The increase in magnetization would be due to the presence of magnetic Mn²⁺ ions that could enhance ferromagnetic properties as well as oxygen vacancies. Mn doping and the reduction of the particle size provided an increased number of Ti³⁺-VO²⁻-Ti³⁺ groups (as explained in the previous paragraph), which improved the ferromagnetic behavior. Oxygen vacancies (VO²⁻) were easily created in the nanocrystals during the doping which significantly promoted the formation of the Ti³⁺-VO²⁻-Ti³⁺ groups and increased T-center exchange (TCE) interactions, leading to enhanced room-temperature ferromagnetism. These results are in agreement with Bahoosh et al. [21] who have shown that the observed ferromagnetic properties in BaTiO₃- or PbTiO₃-nanoparticles at room temperature could be originated due to the oxygen vacancies at the surface and to the appearance of a different valence state composed of Ti³⁺ or Ti²⁺. As a result one observes multiferroic properties of the NP. Whereas the polarization decreases with decreasing particle size, the magnetization increases below a critical particle size $N_{\mathrm{c}}$ that is of the order of 1–8 Å.

This critical particle size should be increased by an external magnetic field.

The field-cooled (FC) and zero-field–cooled (ZFC) magnetization measurements were performed from 50 to 295 K with 1000-Oe applied magnetic fields (fig. 4(d)) for the Mn-doped particles. It can be seen that at temperatures below 221 K, M(T) begins to deviate from linear dependence. The Néel temperature is determined to be $T_{\mathrm{N}}= 221\ \text{K}$. Such magnetic anomaly was found when antiferromagnetic correlations compete with ferromagnetic interactions. The temperature dependence of the difference magnetization $\Delta M = M_{\mathrm{FC}}-M_{\mathrm{ZFC}}$ is plotted in fig. 4(d). $\Delta M$ shows negative values above 121 K. This unusual feature was related to magnetostriction [22–24]. Our magnetization values were close to those found in BiFeO₃ multiferroic materials [19] and demonstrated that PZN-4.5PT nanoparticles can be considered as multiferroics with excellent properties taking into account their ferroelectric and ferroelastic characteristics.

We performed XRD measurements to reveal the phase composition (fig. 5). The XRD spectra show similar patterns for both samples with characteristic peaks of PZN-4.5PT.

Zoom In Zoom Out Reset image size

No peaks corresponding to other phases were observed in XRD spectra before and after magnetic measurements indicating pure PZN-4.5PT phase. More generally, it is well known from the literature that mechanical grounding and milling producing more defects in that way treated materials. It is suggested for some defect-rich regions, such as the surface, interface, and grain boundary (fig. 1). The importance of disordered interfaces and intergranular regions, vacancies, lattice or bond defects are suggested for driving high-temperature ferromagnetic-like behavior [25–27]. It has been also concluded that the proper combination of interpenetrating crystalline and amorphous phases is the most probable condition for the existence of ferromagnetism in pure ZnO [28]. Based on these results we can conclude that grain boundaries, size of the particles and the lattice or bond defects control the ferromagnetic-like properties of the PZN-4.5PT nanoparticles.

Investigations were conducted to determine the different Ti ions present in the crystals. The energies of the Ti³⁺ and nominally Ti⁴⁺ samples are compared by Waychunas [29]. The mean ls to 3d energy for the two Ti³⁺ samples is 4967.5(l) eV, and for the Ti⁴⁺ samples is 4969.1(l) eV, a shift of 1.6 eV. In fig. 6, the Ti K-edge Xanes absorption of $\langle001\rangle$ and $\langle 110\rangle$ oriented PZN-4.5PT single crystals using a synchrotron lightsource indicates the presence of mixed Ti³⁺ and Ti⁴⁺ ions at the pre-edge position (from 4963–4973 eV). The presence of Mn²⁺ and Mn³⁺ ions in the doped crystals may increase the oxygen vacancies. The transformation of Ti⁴⁺ ions to magnetic Ti³⁺ explain why doping could increase the ferromagnetism of the nanoparticles. It is only at 4967.05 eV that the value of Ti ions for Mn-doped crystals is higher than the other crystals (fig. 6). This indicates the presence of Ti³⁺ ions at this position as suggested by [27]. The presence of more Ti³⁺ ions in Mn-doped crystals confirms the hypothesis that oxygen vacancies (VO) would facilitate the transformation of Ti⁴⁺ to Ti³⁺ ions.

Zoom In Zoom Out Reset image size

In this study we have shown that PZN-4.5PT shows excellent magnetic properties when reduced into nanopowder form. The magnetic behavior of undoped and Mn-doped PZN-4.5PT is related to sizes and would appear only at the nanoscale level. Remnant magnetization ($M_{\mathrm{r}}$) and coercivity ($H_{\mathrm{c}}$) were measured at low field indicating that the PZN-4.5PT nanoparticles possessed a weak ferromagnetic behavior.

The magnetization values demonstrated that PZN-4.5PT nanoparticles could be considered as multiferroic materials with excellent properties taking into account their ferroelectric and ferroelastic characteristics. The use of the PZN-4.5PT as nanoparticles thin layers allows potential applications in electronic devices.

This work is supported by Agence Universitaire de la Francophonie Bureau Afrique de l'Ouest (AUF-BAO) and LAAAMP joint Project of AfLS, ICTP and IUCr.