Background

BiMnO3 has received huge interest due to the possibility of coexistence of ferroelectricity and ferromagnetism [1-11]. BiMnO3 has monoclinic symmetry with lattice parameters a = 9.533 Å, b = 5.606 Å, c = 9.854 Å, and β = 110.667°. The ferromagnetism has been explained in terms of orbital ordering of Mn4+ ions, while the Bi-6s lone pair was expected to result in ferroelectricity. The existence of ferromagnetism has been confirmed for both stoichiometric and ceramic BiMnO3. Most studies of ceramic BiMnO3 showed around the same ferromagnetic transition temperature of approximately 105 K, with a saturated magnetic moment of 3.6 μB/Mn, consistent with the high spin configuration of the Mn4+ ion. However, ferroelectricity of ceramic samples has not been measured. A sizable single crystal has not been obtained, and most measurements have been performed on polycrystalline samples requiring high-pressure synthesis.

It has been very difficult to obtain good stoichiometry in thin film BiMnO3, and the physical properties measured for thin films of BiMnO3 having unsatisfactory stoichiometry are often widely distributed. Thin film studies on BiMnO3 have not detected ferroelectricity very clearly, while other physical properties measured for the films varied across a wide spectrum. The films also exhibited non-optimum magnetic properties. Ferromagnetic transition temperatures and saturated magnetic moments were smaller than those reported for stoichiometric ceramic BiMnO3; thus, multiferroicity has not yet been accurately ascertained for stoichiometric BiMnO3.

The first thin film of BiMnO3 on SrTiO3 (001) substrate had Tc ~ 105 K, and an x-ray diffraction rocking curve peak had full width at half maximum of approximately 1.1° [12]. In that study, no measurement was made of saturated magnetic moment, M sat, and the existence of ferroelectricity was not confirmed. Son et al. reported writing polarization bits on BiMnO3 thin films with a low Tc of approximately 50 K and full width at half maximum of approximately 0.4° [13]. Pt/SrTiO3/BiMnO3/SrTiO3/Pt and SrRuO3/SrTiO3/BiMnO3/SrTiO3/SrRuO3 capacitors were also reported to show good ferroelectric properties with a remnant polarization of around 9 to 16 μC/cm2 but in combination with a very small saturated magnetic moment, M sat < 1.0 μB/Mn [14]. It is notable that ferroelectricity was reported to arise from SrTiO3 itself [15-17]. A clear polarization electric field hysteresis curve was observed for a BiMnO3/SrTiO3 (001) structure grown using pulsed laser deposition with a high Bi-rich target of Bi2.4MnO3 [18]. However, the ferromagnetic properties of the film were not optimum: Tc ~ 85 K, M sat ~ 1 μB/Mn. Ferromagnetic properties measured for BiMnO3 films made using chemical solution deposition or rf-magnetron sputtering were less favorable compared to those of stoichiometric ceramic BiMnO3 [19,20]. The depression in Curie temperature can be attributed to a non-stoichiometric composition, to strain, or to size effects [1].

Overall, the growth of BiMnO3 thin films with correct stoichiometry, free from vacancies, and with ferromagnetic properties similar to those measured in bulk samples has not yet been reported. Thin film growth of BiMnO3 suffers from high Bi volatility. To study the pertinent problem of multiferroicity in BiMnO3, we fabricated thin films of BiMnO3 with magnetic properties and stoichiometry matching those reported for high-pressure fabricated stoichiometric BiMnO3. Using these films, we investigated the existence of ferroelectricity in stoichiometric BiMnO3.

Methods

We fabricated BiMnO3 thin films on a SrTiO3 (001) substrate using a pulsed laser deposition method [21-24]. A KrF excimer laser with repetition rate of 4 Hz was used, and the optimum growth temperature was found to be very narrow: around approximately 500°C with oxygen partial pressure of approximately 10 mTorr. We used a freshly ground surface of Bi1.2MnO3 as the target. Note that the Bi overstoichiometry is rather small; together with the precise growth conditions, these characteristics of the target are one reason for the wide spectrum of physical properties reported in films. The number of pulses required per monolayer of BiMnO3 was about 13.6. The thickness was estimated to be around t ~ 88 nm, using field emission scanning electron microscope. We performed a detailed x-ray diffraction study of the epitaxial structure of the BiMnO3 films using high-resolution x-ray diffraction. For electrical transport studies, we used a physical property measurement system (Quantum Design, PPMS, San Diego, USA). Magnetic properties were determined using a superconducting quantum interference device (Quantum Design, MPMS, San Diego, USA). Ferroelectric characterization measurement with capacitance geometry was done on Nb-doped SrTiO3 substrate\BiMnO3\Au sample using a cryogenic probe station (Lake Shore Cryotronics, Inc., Westerville, USA) and semiconductor parameter analyzer (Agilent Technologies, Santa Clara, USA). The area of Au top electrode was approximately 100 μm × 100 μm.

Results and discussion

Figure 1a shows the θ − 2θ patterns of the BiMnO3/SrTiO3 (001) structure. The (010) and (020) BiMnO3 reflection peaks are clearly visible to the left of the SrTiO3 substrate peaks. The calculated out-of-plane lattice constant for BiMnO3 film peaks was 3.985 Å. No other Bragg diffraction peaks were observed for the films. The x-ray rocking curve of the (010) BiMnO3 peak revealed a full width at half maximum as small as approximately 0.067°, lower than the 0.4° and 1.1° reported in previous studies [12,13].

Figure 1
figure 1

The θ  − 2 θ patterns of the BiMnO 3 /SrTiO 3 (001) structure and the reciprocal space maps. (a) XRD θ − 2θ patterns for the BiMnO3/SrTiO3 (001) heterostructure. Inset shows the rocking curve for a (010) BiMnO3 peak. The x-ray rocking curve of the (010) BiMnO3 peak revealed a full width at half maximum as small as approximately 0.067°. (b) X-ray reciprocal space mapping around the SrTiO3 (114) plane shows well-developed peaks for BiMnO3 in the lower region and two strong substrate peaks in the upper region.

Full size image

The reciprocal space maps shown in Figure 1b confirm nearly coherent growth of a BiMnO3 film on an SrTiO3 (001) substrate with an in-plane lattice constant of 3.909 Å. The calculated volume of one unit cell of BiMnO3 (V film = 60.89 Å3) in the film was 98.9% that of bulk monoclinic BiMnO3 (V bulk = 61.58 Å3). The roughly 1% volume reduction is mostly due to compressive strain from the SrTiO3 substrate. The slightly smaller unit cell volume measured for the film demonstrates that our stoichiometric BiMnO3 has negligible cation or oxygen vacancies.

The ferromagnetic properties for our stoichiometric BiMnO3 films were investigated, to compare their performance with that reported for stoichiometric ceramic bulk BiMnO3. Figure 2a shows the temperature dependence of magnetization at 1 T after 7 T field cooling. We measured a Tc of 105 K, close to that of the stoichiometric BiMnO3 bulk sample. Figure 2b shows magnetic hysteresis (M-H) curves at 5 K. The saturated magnetic moment is as high as 3.8 μB/Mn, close to that reported for stoichiometric BiMnO3 bulk. The magnetic coercive field was approximately 700 Oe, which is slightly larger than that measured for other manganese perovskite oxides such as (La,Ca,Sr)MnO3, and about two orders of magnitude smaller than that reported for SrRuO3 [24,25]. For ferromagnetic perovskite oxides, a larger epitaxial strain usually results in enhancement of magnetic coercive field and the lattice mismatch of BiMnO3 with respect to the SrTiO3 substrate is larger than the lattice mismatch of (La,Ca,Sr)MnO3 with respect to the commonly used of SrTiO3 substrate.

Figure 2
figure 2

Magnetization curves and magnetic hysteresis curves for BiMnO 3 /SrTiO 3 (001) structure. (a) Magnetization curves for BiMnO3/SrTiO3 (001) structure at 1 T after 7-T high field cooling. (b) Magnetic hysteresis curves for BiMnO3/SrTiO3 (001) structure at 5 K.

Full size image

After confirming good stoichiometry and ferromagnetic properties, comparable to those of stoichiometric bulk BiMnO3, we investigated the existence of multiferroicity in our BiMnO3 films. First, in-plane resistivity was measured, as shown in Figure 3a. The resistivity of the film shows semiconducting behavior with ρ(T = 300 K) ~ 4 × 104 Ω⋅cm and ρ(T = 100 K) ~ 1011 Ω⋅cm. This room temperature resistivity is larger than the room temperature values of 2 × 104 Ω⋅cm measured for polycrystalline ceramics [5] and 1.8 × 102 Ω⋅cm reported for epitaxial films [26]. However, this value is much smaller than the 5 × 107 Ω⋅cm measured for a ‘highly resistive film’ [27]. It is notable that the out-of-plane lattice parameter of 4.004 Å reported for the ‘highly resistive film’ is significantly larger than our value of 3.985 Å, and that the saturated magnetic moment of 2.0 μB/Mn for the ‘highly resistive film’ is much smaller than our value of 3.8 μB/Mn. Usually, a larger unit cell volume in perovskite-based metal oxides arises from cation or oxygen vacancies, which dramatically change transport properties more than magnetic properties [28]. Gajek et al. demonstrated spin filtering in the BiMnO3 junction [26] and observed that significant changes of unit cell volume measured in films arise from Bi vacancies that locally disturb the complex orbital ordering essential for long-range ferromagnetic order in BiMnO3. A change of unit cell volume was accompanied by small room temperature resistivity values and lower saturated magnetic moment in the magnetic hysteresis curve [26].

Figure 3
figure 3

Temperature dependence of resistivity and charge-electric field (Q-E) curve for BiMnO 3 /SrTiO 3 (001) structure. (a) Temperature dependence of resistivity for BiMnO3/SrTiO3 (001) structure. (b) Charge-electric field (Q-E) curves for BiMnO3/SrTiO3 (001) structure at 5 K.

Full size image

Finally, we measured the charge-electric field (Q-E) curve to obtain evidence of ferroelectricity in the stoichiometric BiMnO3 film. Figure 3b shows Q-E curves at 5 K, where leakage problems, as shown in Figure 3a, do not occur. The Q-E curve measured at 5 K was fairly linear and free from hysteresis, and no ferroelectric order was observed. The absence of ferroelectric order in our stoichiometric film seems to be strengthened by the observation that the unit cell volume supports stoichiometry, that crystallinity is excellent, and ferromagnetic properties are as good as those reported for stoichiometric BiMnO3 samples.

Recently, there has been doubt about observations of ferroelectricity in some BiMnO3 films. It was reported theoretically that the ground state for BiMnO3 either with or without strain should be a centrosymmetric structure [29,30]. Rigorous structural studies on ceramic samples using transmission electron microscope and neutron diffraction data showed that BiMnO3 crystallizes in the centrosymmetric space group C2/c at 300 K [31]. It was suggested that the weak ferroelectric polarizations measured on BiMnO3 samples originated from an ordered oxygen deficiency [32].

Conclusions

In summary, we investigated the existence of ferroelectricity in stoichiometric BiMnO3. We produced high-quality thin films with good stoichiometry and with magnetic properties - such as Tc and saturated magnetic moment - comparable to those reported for bulk stoichiometric BiMnO3. The structural quality was evidenced by narrow full width at half maximum for XRD peaks and good reciprocal space mapping data. Since vacancies in perovskite oxide film affect transport properties more than ferromagnetic properties, we believe that our stoichiometric BiMnO3 films should have sufficient quality for ascertaining the existence of ferroelectricity in stoichiometric BiMnO3. We found that the resistivity of the film demonstrates semiconducting behavior, with ρ(T = 300 K) ~ 4 × 104 Ω⋅cm. The Q-E curve measured at 5 K was fairly linear and free from hysteresis, and no ferroelectric order was observed. This finding is consistent with the centrosymmetric crystal structure recently suggested by theoretical calculations and structural studies on ceramic samples of stoichiometric BiMnO3. If ferroelectricity does exist in both stoichiometric BiMnO3 and non-stoichiometric BiMnO3, then Bi-6s lone pair scenario should be the best answer for the origin. Summarizing our work and other works, the existence of ferroelectricity seems to depend on the stoichiometry very sensitively. Then, other origin should be considered at least together since Bi-6s lone pair exists both for stoichiometric BiMnO3 without showing FE and non-stoichiomeric BiMnO3 showing FE.