In recent times, BiFeO3 has been considered as an important material for the development of multifunctional devices because of its distinctive ferroelectric, magnetic, piezoelectric, and optical properties. These include a high Currie temperature of ferroelectricity (TC1100 K), (Venevtsev et al., 1960), high Néel temperature of antiferromagnetism (TN650 K), (Kiselev et al., 1963)lead-free piezoelectricity, and large flexibility in the wavelength of visible light region. These features make BiFeO3 particularly applicable in the fields of ferroelectrics, magnetics, piezoelectrics, and optics; in addition, the cross correlation of these properties can be expected above room temperature (RT). [Fig. 1] BiFeO3 has a perovskite-type crystal structure that is rhombohedrally distorted in the  direction and crystallized in the space group R3c [Fig. 2].(Kubel et al., 1990) The ferroelectric performance of BiFeO3 is comparable to that of conventional ferroelectric materials such as Pr(Zr,Ti)O3 (PZT) because BiFeO3 exhibits excellent spontaneous polarization at RT. Theoretically, spontaneous polarization corresponds to crystal symmetry, wherein the rhombohedral and tetragonal BiFeO3 structures are expected to show spontaneous polarizations of ≈100 C/cm2 in the  direction and ≈150 C/cm2 in the  direction, respectively. [Fig. 3(a)](Edereret al., 2005) In fact, these theoretically predicted large spontaneous polarizations in BiFeO3 are almost consistent with the experimental results [Figs. 3(b) andFigs. 3(c) ] (Li et al., 2004, Yunet al., 2004), stating that BiFeO3 is favorable for use in ferroelectric random access memory (FeRAM)applications. However, the practical application of BiFeO3 thin films has been limited by their large leakage current density and large coercive field at RT, (Naganumaet al., 2007, Pabst et al., 2007) as a result, BiFeO3 thin films easily undergo electrical breakdown when a large leakage current passes through them before the polarization is switched. Therefore, in order for BiFeO3 films to find practical future application, the leakage current and/or coercive field of these films must be reduced. In term of magnetic properties, BiFeO3 is antiferromagnetic with aG-type spin configuration; (Kubelet al., 1990, Edereret al., 2005) that is, nearestneighborFe moments are aligned antiparallel to each other, and there is a sixfolddegeneracy, resulting in an effective“easy magnetization plane”forthe orientation of the magnetic moments within the (111) plane. It should be noted that the antiferromagnetic (111) plane is orthogonal to the ferroelectric polarization direction  in the rhombohedral structure. [Fig. 4(a)] The orientation of theantiferromagnetic sublattice magnetization is coupled through ferroelastic strain due to crystal symmetry, and it shouldalways be perpendicular to the ferroelectric polarization  direction. Therefore, a polarization switch to either 71 or 109 should change theorientation of the antiferromagnetic plane. This change in orientation would, however, not occur in the case of 180 to 180 ferroelectric polarization switching. [Fig. 4(b)-4(d)] In fact, experimentally, (Zhao et al., 2006) the ferroelectric domain and antiferromagnetic domain in BiFeO3 (100) epitaxial films are strongly coupled (magneto-electric (ME) coupling) in the orthogonal configuration, and the orientation of the antiferromagnetic plane is switched by a 109 switch in the ferroelectric domain. In effect, the magnetization configuration can be controlled by the application of an electric field through ferroelectric domain switching; by means of this mechanism, it is possible to realize voltage control of magnetic random access memory (V-MRAM). The use of V-MRAM can drastically reduce electrical consumption when compared with spin-MRAM which is operated by spin-polarized current. In terms of “how to detect the change of magnetization induced by ferroelectric domain switching”, it can be seen that owing to Dzyaloshinskii-Moriya (DM) interaction, (Dzyaloshinskii, 1957, Moriya, 1960) the symmetry permits a canting of the antiferromagnetic sublattices, resulting in a local weak spontaneous magnetization. This magnetization is macroscopically canceled by a spiral spin structure caused by the rotation of the antiferromagnetic axis through the crystal with an incommensurately long-wavelength period of 62 nm. (Sosnowska et al., 1982) This spiral spin structure might be suppressed in the film form of BiFeO3 and the resulting magnetic moment is caused by a weak ferromagnetism of 0.1 B/Fe atom. However, this small magnetic moment is not suitable for application in devices such as spintronics because of the difficulty associated with the direction of a weak magnetic moment using a magnetic sensor, as well as the spin-filter effect. Therefore, in order to detect the change of magnetic states driven by ferroelectric domain switching (i) introduction of ferrimagnetic spin order into BiFeO3 having a rombohedral structure or (ii) detection of the ferromagnetization change through exchange coupling with BiFeO3 (Chu et al., 2008) are the predominant candidates. In addition, the BiFeO3 films show distinctive optical properties, as previously mentioned. BiFeO3 has the highest flexibility among the oxide materials: a flexibility of 3.22 at a wavelength of 600 nm; this flexibility is expected to cross-correlate with the other physical properties. (Shima et al., 2009) The details of the optical properties of BiFeO3 films have been relegated to a subsequent
discussion. Based on these background considerations, the main focus of this chapter is the structure, ferroelectricity, and magnetism of BiFeO3 films. In addition, we discuss the effect of the substitution of various 3d transition metals into the B-site of BiFeO3 films on the films’ structural, ferroelectric, and magnetic properties. We also propose a new multiferroic material having a ferrimagnetic spin order and large spontaneous polarization at RT. This chapter includes three sections. Section 1 presents the fundamental characteristics of pure BiFeO3 films. Section 2 describes substitution of various 3d transition metals for Fe in BiFeO3 films, in small amount (5 at.%), in order to achieve ferrimagnetic spin ordering together with low leakage current density. Finally, in Section 3, we demonstrate the substitution of a large amount of Co (58 at.%).
2. Structural, electronic, and magnetic properties of BiFeO3 films
In order to elucidate the fundamental properties of BiFeO3 films, here, we begin by systematically investigatingthe effect of annealing temperature on the structural, electrical, and magnetic properties of BiFeO3 polycrystalline film. The results described in this section is based on our recent work as follows; Naganuma et al. TMRSJ, 2007,Naganuma et al. MT, 2007, Naganuma et al. IF, 2007, Naganuma et al. TUFFC, 2008, Naganuma et al. JJAP, 2008, Naganuma et al. APEX, 2008, Naganuma et al.JCSJ, 2010, Naganuma et al. JAP, 2011.
Polycrystalline BiFeO3 films were prepared by a chemical solution deposition (CSD) method. The preparation processes are shown in Fig. 5 and can be summarized as follows: an enhanced-metal-organic-decomposition (E-MOD) solution (with stoichiometric composition Bi:Fe=1:1, 0.2 mol/l) was used as the precursor solution. The precursor solution was spin-coated onto Pt (150 nm)/Ti (5 nm)/SiO2/Si(100) substratesat a rotation speed between4000 and 6000 rpm for 50 s. Pt and Ti were deposited by r.f. magnetron sputtering using Ar gas at RT. The spin-coatedfilms were dried at 150 °C for 1 min andcalcined at 350 °C for 5 min. The spin coating andcalcination processes were repeated 4-5 times, after which the filmswere sintered in air at 400 - 800 °C for 10 min by rapid thermal annealing (RTA; ULVAC mila-5000). The films had a thickness of approximately 200 nm. The surface morphology of the films was observed by atomic force microscopy (AFM; SII SPI3800N) and scanning electron microscopy (SEM; JEOL JSM-6380). The crystal structure and orientation were confirmed by X-ray diffraction (XRD; PANalytical X’Pert MRD) with Cu-K radiation and transmission electron microscopy (TEM; JEOL JEM-2100F, JEOL JEM-3000F, LEO-922) working at 200 and 300 kV. The leakage current density was measured using a pico-ampere meter (HP 4140B). The ferroelectric hysteresis (P-E) loop of the films was measured using a ferroelectric tester (aixACCT TF-2000, TOYO FCE-1A) with a single triangular pulse. Positive-Up-Negative-Down (PUND) measurement was also used for evaluating electrical properties. The magnetic properties were measured with a vibrating sample magnetometer (VSM; Tamakawa) at RT, and a superconducting quantum interference device (SQUID; Quantum design MPMS) magnetometer was used for the in-plane direction.
Figure 6 shows the XRD profiles of BiFeO3 films deposited on Pt/Ti/SiO2/Si(100) substrates annealed at various temperatures. For each annealing temperature, astrongdiffraction peak at 2=39.8was observed due to the Pt(111) plane. At an annealing temperature of 400 °C, a weak diffraction peak was observed in the region of 2=46; however,it was not clear whether this diffraction peak originated from BiFeO3(024) at 2=45.74or Pt(200) at 2=46.24. Therefore, in order to clarify the origin of the diffraction peak in the region of 2=46, XRD analyses were undertaken for the Pt/Ti/SiO2/Si(100) substrate only.On the basis of this experiment, the diffraction peak in the region of 2=46 was identified as Pt (200) for the sample annealed at 400 °C. Therefore, the structure of the film annealed at 400 °C was amorphous or nanocrystalline. The formation of the polycrystalline BiFeO3 films for annealing temperature above 450 °C was indicated by the appearance of numerous diffraction peaks attributed to the BiFeO3 structure above the stated temperature. The strong  orientation of the bottom Pt layer did not affect BiFeO3 crystal growth. Atannealing temperature above 700 °C, secondary phasessuch as-Fe2O3 and BiPt were formed.The observation of these phases indicates that the excess Fe formed -Fe2O3 due to diffusion of Bi into the Pt electrode at high annealing temperatures. The indication is that the single phase ofrandomly oriented polycrystalline BiFeO3 film was formed at annealing temperatures between450 and 650 °C.
Cross-sectional TEM observation was used to investigate the quality of the polycrystalline BiFeO3 films annealed at 550 °C. The observation, presented in Fig 7(a) indicates that the bottom Pt layer shows the (111) texture structure, which is consistent with the XRD profiles. High-resolution (HR) TEM was used to investigate the grain boundary phases in thepolycrystalline structure within the square area outlined in Fig. 7(a). The HRTEM image of the grain boundary [Fig. 7(b)] and fast Fourier transform (FFT) pattern from two grains [Fig. 7(c), 7(d)] are shown. These investigations show that high quality polycrystalline BiFeO3 films were successfully fabricated by means of the CSD method
In the case of BiFeO3, crystal symmetry exerts a strong influence on the ferroelectric polarization; (Edereret al., 2005) therefore, the crystal symmetry of BiFeO3 was determined by simulation of the HRTEM images and nanobeam diffraction (NBD) patterns. Figure 8(a) shows the HRTEM image (a), corresponding FFT pattern (b), NBD pattern (c), simulated electron diffractionpattern (d), simulated lattice fringe image embedded in the observed HRTEM image (e), atom position (f), and distance between atoms (g). The HRTEM image containsperiodic lattice fringes along the  direction with spacings of approximately 0.396 nm which is in goodagreement with Kubel’s report. (Kubelet al., 1990) The electron diffractionpattern was simulated using the MacTEMPAS computerprogram by applying the multislice method (Kirklandet al., 1998) and using the lattice parameters of the rhombohedral R3c and the tetragonal Pbmmlattices.( Kubelet al., 1990, Wanget al., 2003, Yunet al., 2004) A comparison of the simulated electrondiffraction pattern with the NBD and FFT patterns shows that theBiFeO3 layer has a rhombohedral R3c structure. The simulatedlattice fringe image of R3c corresponded exactly to the HRTEM image. The position of the atoms in the HRTEM image and the periodicity of the atoms based on R3c symmetry are indicated in Fig. 8(d) and 8(e).
Figure 9 shows the AFM and SEM images of the BiFeO3 films as a function of annealing temperature. At temperatures of 400 and 450 °C [Fig. 9(a) and 9(b)], a homogeneous surface was formed and no obvious grains were detected for the sample annealed at 400 °C. The appearance of grains was observed in samples annealed at temperatures above 450 °C. In the wide area AFM images, very little variation in the grain size was seen with an increase in the annealing temperature between 450 and 750 °C. In contrast, the expanded AFM images show grains with sizes of several tens of nanometers, indicating that the micron sized grains consisted of anagglomeration of small grains, several tens of nanometers in diameter.Fig. 9(c) The size of the smaller grains increased as the annealing temperature increased. In particularly, there was a drastic increase in the size of the smaller grains above 700 °C. The sample annealed at 800 °C could not be analyzed using AFM because the specimen was easily stripped away from the substrate. Therefore, the surface morphology of this specimen was observed using SEM. [Fig. 9(i)] Square-shaped grains were observed after annealing at 800C; this can be identified as the secondary phases of-Fe2O3 and BiPt. These observations indicate that the microstructure of BiFeO3 films is drastically influenced by the annealing temperature.
There are many reports that focus on leakage current density; however, only a few of these have discussedthe mechanism underlying the leakage current.Hence, the topic of leakage current density is still open to discussion and can be considered an important issue from the viewpoint ofmemory applications. Herein, the leakage currentmechanism operating in the BiFeO3 film is discussed as a function of the annealing temperature.Figure 10shows (a) the leakage current density (J) v.s.electric field (E),(b) Schottky emission plot (log J v.s.E1/2),(c)Ohmic plot (double logarithm plots),(d) Fowler-Nordheim plot (log(J/E2) vs 1/E),(e)Poole-Frenkel plot (log(J/E) v.s.E1/2) plots, and (f) space-charge-limited current (log(J/E) v.s. log V) for theBiFeO3 film annealed at various temperatures. (Naganuma & Okamura, JAP 2007, Naganuma et al., IF 2007) The measurement was carried out at RT. As shown in Fig. 10(a), the leakage current density of BiFeO3 filmstended toincrease with increasing annealingtemperature. Careful observation of the slopes of the curves shows three steps corresponding to three kinds of leakage current mechanisms. First, wediscuss the interfacial limited leakagecurrent mechanism, taking into consideration Schottky emission. Using Schottky emission, the relative permittivityand barrier height were estimated tobe 0.4 and 0.6 eV, respectively [Fig. 10(b)]. On the other hand, the inclination at alow electric field in the double logarithm plots, as shown in Fig. 10(c) was around 1.1–1.2. Compared to the Schottoky-emission conduction, Ohmic conduction seems to be adaptable at low electric field. The inclinationat a high electric field in the double logarithm plots in Fig. 10(b) was around 2, indicating that the leakage currentbehavior at high electric field was dominated
by space-charge-limited current (SCLC). Next, we discuss the leakagecurrent mechanism before the start of the SCLC. The barrierheight deduced from the Fowler-Nordheim equation wasaround 0.019 eV [Fig. 10(d)]. This barrier height is quitesmall for Fowler-Nordheim tunneling conduction.The relative permittivity calculated using the Poole-Frenkel equation was around 0.1–0.2 [Fig. 10(e)]. Here, it is though that the leakage current mechanism changed as follows:Ohmic conduction occurred at a low electric field; Poole-Frenkeltrap limited conduction appeared as the electric field increased; and SCLC was activated at a high electric field.
Figure 11 shows the P-E hysteresis loops of the BiFeO3 films annealed at various temperatures. The P-E hysteresis loop was measured at RT using a frequency of 2 kHz
(aixACCT TF-2000).The P-E hysteresis loop has an unsaturated, loose shape even in the case of the BiFeO3 film annealed at 450 °C, which is indicative of a low leakage current density.
Two methods were employed in order to reduce the influence of leakage current density on theP-E hysteresis measurements: (i) a high frequency system was used and (ii) measurements were taken at low temperatures. Figure 12(a) shows the P-E hysteresis loops for BiFeO3 film annealed at 450 °C; the loops were obtained using the 100 kHz high frequency system. The solid line indicates the use of phase-delay compensation, and the dotted line indicates the case inwhich there was no compensation.
The compensation of phase delay from the high voltage amplifier and circuit cable was estimated by using a standard capacitor of 100 pF. The changes inthe electric field and polarization against time are shown in Fig. 12(b).Changes in polarization appear to be delayed with respect tothe changes in the applied electric field by approximately 500 nsec. When a phase delay compensation of 500 nsec was applied, the shape of the P-Ehysteresis loop was improved to high squareness when compared with the low frequency measurement system, as shown in Fig. 11. The details of the compensation method are addressed elsewhere (Naganuma et al., JCSJ 2010).
Figure 13 show the P-E hysteresis loops for BiFeO3 films annealed at various temperatures, measured at RT using a high frequency 100 kHz system. For the BiFeO3 films annealed at 400 °C, the paraelectrics due to amorphous structure was observed. The P-E hysteresis begins to be observed from 450 °C and a relatively high remanent polarization of 80-90 C/cm2 was obtained for the BiFeO3 film annealed at 500C. However, it seems that leakage current still influences the shape of the hysteresis loops for films annealed at high temperatures. Above 600C, the P-E loops assume an unsaturated, loose shapeand spontaneous polarization cannot be estimated. Figure 13(h) shows the electric field (E) dependence of remanent polarization (Pr) estimated from the P-E hysteresis loops. The Pr increased with increasing electric field and there was no clear tendency to saturate. It can thus be seen that the influence of leakage current on the P-E hysteresis loops was clearly reduced by increasing the measurement frequency.
Figure 14 shows the P-E hysteresis loops for BiFeO3 films annealed at various temperatures, measured at -183 °C using a measurement frequency of 1 kHz. At -183 °C, the leakage current density was significantly decreased to below 1.0 10-8 A/cm2 at 0.1 MV/cm; therefore, the influence of leakage current density on the ferroelectric measurement could be excluded. By decreasingthemeasurement temperature, the P-E hysteresis loops could be observed forthe samplesannealed in the temperature range between 450 and 750 °C. No ferroelectricity was observed at the annealing temperature of 800 °C due to the disappearance of the BiFeO3 phase. The shape of the P-E hysteresis loops varied for each annealingtemperature. Double P-E hysteresis loops were observed for the samples annealed at 450 °C,whereas the shape of theP-E hysteresis loop for the sample annealed at 500 °C was insufficiently saturated. This is because the electric coercive field was high for the sample annealed at the lower temperature, and it was not enough to saturate the polarization by the electric field at around 1.3 MV/cm. The remanent polarization and electric coercive field as a function of the annealing temperature are summarized in Fig. 14(g). The remanent polarization of the BiFeO3 films increased linearly with increasing annealing temperature up to 650 °C and decreased above the annealing temperature of 700 °C. The electric coercivity fieldof the BiFeO3 films decreased as the annealing temperature increased. The highestremanent
polarization, as well as the lowest electric coercive field of the BiFeO3 film appeared at the annealing temperature of 650 °C. The remanent polarization and the electric coercive field were 89 C/cm2 and 0.31 MV/cm, respectively, which are be comparable to the recent reports of high remanent polarization. These results reveal that the ferroelectric properties such as remanent polarization and the electric coercive field of the BiFeO3 films are strongly affected by the annealing temperature of the CSD processes even though single phase, the polycrystalline BiFeO3 films were formed.
Figure 15 shows the magnetization (M-H) curve at RT for the BiFeO3 film annealed at 650 °C. The magnetization increased linearly at high magnetic field due to the anitiferromagnetic spin structure. In the zero fields region, nonlinear hysteresis with a very small remanent magnetization was observed, which might be considered to be the weak ferromagnetism due to DM interaction (Dzyaloshinskii, 1957, Moriya, 1960) or strain induced magnetization. Interestingly, non-linearity is often reported near the zero fields for the film form of BiFeO3; (Kiselevet al., 1963, Naganuma et al., TUFFC 2008, Yunet al., 2004, Bai et al., 2005) however, non-linearity has not been observed in the case of bulk BiFeO3. (Bai et al., 2005, Lebeugleet al., 2007) This means that the non-linearity of the M-H curves is mainly affected by strain-induced changes of the spiral structure in the film. There are several reports which discuss the strain induced changes of the spiral structure in BiFeO3 films. However, the details of the process are still debatable.
3. Effect of B-site substitution of Cr, Mn, Co, Ni, and Cu for Fe in BiFeO3 on structural, electrical and magnetic properties
In the second section, single phase of polycrystalline BiFeO3 films were successfully fabricated on Pt/Ti/SiO2/Si(100) substrates, and a high polarization of 89 C/cm2 with a switching field of 0.31 MV/cm was obtained at -183 °C for films annealed at 650 °C. However, the large leakage current, relatively large switching field of polarization, and antiferromagnetic spin configuration of BiFeO3 films make it difficult to use these films in novel electrical applications such as spintronics devices. In this section, engineering of these physical properties is investigated by substitution of Fe in BiFeO3 with various 3d transition metals. (Naganuma et al., APL 2008, Naganuma et al., JAP 2008, Naganuma et al., JE 2009, Naganuma et al., JMSJ 2009)
Cr, Mn, Co, Ni, and Cu substituted BiFeO3 films (200 nm in thickness) were fabricated by the CSD method onto Pt/Ti/SiO2/Si (100) substrates followedby post annealing in air at 650 °C for 10 min. The composition of the E-MOD was adjusted as follows: Bi(Fe0.95M0.05)O3 where M =Cr, Mn, Co, Ni, and Cu. Thefilm structure was confirmed by the/2XRD pattern. The ferroelectric properties weremeasured using ferroelectric testers (TOYO Corporation:FCE-1A for RT and aixACCT: TF-2000 for -183 °C).The leakage current wasmeasured using a picoampere meter and the pulse response forms of the PUND measurement. The details of the estimation method are discussed elsewhere (Naganuma et al., APEX 2008). The magneticproperties were measured using aVSM at RT for the in-plane direction.
Figure 16(a) shows the XRD profiles of Cr, Mn, Co, Ni,and Cu of 5 at. % substituted BiFeO3 films:[Bi(M0.05Fe0.95)O3, M= Cr, Mn, Co, Ni,and Cu]. Diffraction peaks caused by the BiFeO3 structure were observed,indicating the formation of a polycrystalline structure.The (012) diffraction peak of the Co-substituted BiFeO3 filmwas stronger than those of the other substitutive metals; this implies theformation of a 012-textured structure. In the case of Cr-substituted BiFeO3, a secondary phase of Bi7CrO12.5 was formed in addition to the BiFeO3 phase.
Figure 17 shows the leakage current density ofBi(M0.05Fe0.95)O3 films measured with the picoampere meter at RT.The leakage current density of the Ni-substituted BiFeO3 film couldnot be preciselyevaluated because of a considerably highleakage current. The leakagecurrent density of the pure BiFeO3 film increased more rapidlythan those of the films with substitutions in response to increases in the electric field. However, even for thetransmission metal (TM) substituted films, it was difficult to measure the leakage currentdensity above 0.2 MV/cm using the picoampere meter because of dielectric breakdown. In order to evaluate theleakage current density at higher electric fields, the leakagecurrent density was estimated from the pulse response formsof the PUND measurements. In this way, the leakage currentat high electric field can be measured by the reduction in theJoule heat damage.(Naganuma et al., APEX 2008) An electric field more than 0.36 MV/cm could be applied, which is higher than thatmeasured by the picoampere meter. Figure 17(b) shows theleakage current density estimated from the response forms ofthe up pulse. The leakage current density of the Ni-substituted BiFeO3 film could also be measured by this method, and it wasfound to be considerably higher than that of the pure BiFeO3film. This indicates that the PUND method can be used formaterials with a high leakage current density. The substitutions ofMn, Co, and Cu to the BiFeO3 films effectively reduced theleakage current density in the high electric field region.
Figure 18 shows the ferroelectric hysteresis loops of theBi(M0.05Fe0.95)O3 films measured with a 100 kHz driving system at RT usingthe ferroelectric tester, and those measured at -183 °C using a 2 kHz driving system. Ferroelectric hysteresis loopscould not be observed for the Ni-substituted BiFeO3 film. Thepure BiFeO3 film showed an expanded hysteresis loop at RT [Fig. 18(a)], which could be attributed to the leakagecurrent component. The squareness of the ferroelectric hysteresisloops was clearly improved by the substitution of Mn, Co,and Cu to the BiFeO3 films. This squareness is attributed tothe reduction in the leakage current density in the high electricfield region. Although the leakage current density is reducedby the substitution of Mn, Co, and Cu, it is still difficult toapply a high electric field at RT. Therefore,the ferroelectric hysteresis loops were measured at -183 °C usingthe 2 kHz driving system [Fig. 18(b)]. At -183 °C, the leakage currentdensity was considerably lower than the inversion currentdue to domain switching, as inferred from the current responseof the PUND measurements. In fact, the ferroelectrichysteresis loops did not expand and showed high squarenessat -183 °C. Ecversus E plots [Fig. 18(c)] show that the Ecwas reduced by the substitution of Co and Cu. In contrast, the substitution of Mn and Cr to the BiFeO3 films produced a higherEc compared to the pure BiFeO3 film. Inthe Co- and Cu-substituted BiFeO3 films, the Prversus E plots [Fig. 18(d)] almost overlappedup to 1.3
MV/cm. Thus, Co and Cu substitutionreduced theEc of polycrystalline BiFeO3filmswithout reducing Pr, which is suitablefor memory and/or piezoelectric devices.
Figure 19 shows the magnetization curves of theBi(M0.04Fe0.96)O3 films measured at RT. As mentioned in the previsou section, thepure BiFeO3 films showed small magnetization. However, thesubstitution of Co, Ni,and Cu caused an increase in the magnetization, indicating substitution of these TM into the B sites ofFe, although it was not clear whether all the TMs weresubstituted into the B-sites. In the case of Co-substituted BiFeO3, there was an increase in magnetization accompanied by the appearance ofspontaneous magnetization and the coercive field of 2 kOe was observed at RT.In addition, according to other report, (Zhang et al., 2010) clear observation of the magnetic domain structure using magnetic force microscopy (MFM) at RT was observed in 4 at.% Co-substituted BiFeO3 has been reported. Based on these results, the increased magnetization in Co-substituted BiFeO3 was confirmed by both macroscopic and local measurement methods.
Cross-sectional TEM observation was carried out in order to clarify the influence of magnetic impurities on spontaneous magnetization in Co-substituted BiFeO3 films. (Naganuma et al., JMSJ 2009) Co-substituted BiFeO3 film was deposited on a Pt/Ti/SiO2/Si (100) substrate having a relatively flat surface. Grains of approximately hundreds of nm in size were formed. [Fig. 20(a)] Obvious secondary phases could not be observed in the wide area images. Figure 20(b) shows the NBD patterns for the [-1 3 -2] direction of the Co-substituted BiFeO3 layer. Analysis of the NBD pattern shows that the crystal symmetry is rombohedral with a R3c space group, and the lattice parameters are a=0.55 nm, c=1.39 nm. The high-resolution TEM image around the grain boundary is shown in Fig. 20(c). Grain boundary formation is evident but the grain boundary phases could not be observed in this film. Therefore, it can be inferred that Co was substituted for Fe in BiFeO3, and the magnetization enhancement might not be attributed to magnetic impurity phases. It was concluded that the substitution of small of Cointothe B-sites of BiFeO3 could improve the leakage current property, reducethe electric coercive field without degrading the remanentpolarization, and induce spontaneous magnetization at RT.
4. Multifunctional characteristics of BiCoO3-BiFeO3 solid solution epitaxial films
As clarified in the third section, the 4 or 5 at.%-Co-substituted BiFeO3 polycrystalline films exhibited excellent electrical and magnetic characteristics. Substitution with larger amounts of Co was expected to result in further enhancement of the electrical and magnetic properties. It should be noted that high-pressure behavior becomes dominant in the highly Co-substituted BiFeO3 films due to the high-pressure phase of BiCoO3. In fact, a maximum of approximately 8 at.% Co can be substituted forFe in the case of polycrystalline films while maintaining a single phase, whereas secondary phases of BiOx are formed at Co concentrations above 8 at.%. (Naganuma et al., JAP 2008) Because the character of BiCoO3 is strongly influenced at high Co-substitution, hereafter, we refer to highlyCo-substituted BiFeO3 films as BiCoO3-BiFeO3. In one of our studies,(Naganuma et al., JAP 2009) the high-pressure phase of BiMnO3 was successfully stabilized in a thin-film form by using epitaxial strain.In accordance with this study,solid solution films of BiCoO3-BiFeO3 having a high BiCoO3 content could also be stabilized on SrTiO3 (100) single crystal substrates by epitaxial strain. In this section, the structural, (Yasuiet al., JJAP 2007) ferroelectric, (Yasuiet al., JJAP 2008, Yasui et al., JAP 2009) and magnetic properties (Naganuma et al., JAP 2011) of epitaxial BiCoO3-BiFeO3films grown on SrTiO3 substrates up to a BiCoO3 concentration of 58 at.% are systematically investigated.
The BiFeO3–BiCoO3 solid solution films were grown on SrTiO3 (100)substrates at 700 °C by metalorganic chemical vapor deposition (MOCVD)established in Funakubo laboratory, andBi[(CH3)2-(2–(CH3)2NCH2C6H4)], Fe(C2H5C5H4)2, Co(CH3C5H4)2andoxygen gas was used as the source materials.A vertical glasstypereactor maintained at a pressure of 530 Pa was used forthe film preparation. The films were deposited by MOCVDusing pulse introduction of the mixture gases with Bi, Fe,and Co sources (pulse-MOCVD). The thickness of thesefilms wasapproximately 200 nm.( Yasuiet al., JJAP 2007)The crystal structure of the deposited films was characterizedby high-resolution XRD (HRXRD)analysis using a four-axis diffractometer (Philips X’-pertMRD). HRXRD reciprocal space mapping (RSM) aroundSrTiO3 004 and 204 was employed for a detailedanalysis of crystal symmetry. The cross-sectional TEM (Hitachi HF-2000) observation working at 200 kV was used for microstructural analysis. The crystal symmetry was also identified using Ramanspectroscopy by K. Nishida.( Yasuiet al., JJAP 2007) Raman spectra were measured using asubtractive single spectrometer (Renishaw SYSTEM1000)with a backward scattering configuration. A laser beam wasfocused on the film surface, and the beam spot wasapproximately 1 m. The measurement time was fixed at100 s. The leakage current v.s. electrical field and P-E loops were measured with asemiconductor parameter analyzer (HP4155B, Hewlett-Packard) and ferroelectric tester (TOYO Corporation, FCE-1A). The magnetic properties were measured in the in-plane direction using SQUID.
Figure 21 shows the typical /2 and pole-figure HRXRD profiles of BiFeO3–BiCoO3 solid solution films (BiCoO3 concentrations of 0, 16, 21, and 33 at.%) grown on SrTiO3 (100)substrates. Although Bi2O3 of secondary phase has a tendency to be formed at a high BiCoO3 concentration, the single phase of BiFeO3–BiCoO3 was successfully obtained by optimizing preparation conditions. The pole-figure HRXRD profiles indicate that all the films were epitaxially grown on SrTiO3 (100)substrates. The magnified/2 XRD profiles around BiFeO3–BiCoO3 002 indicate that the 002 peak shifted to high angle upon increasing the BiCoO3concentration, which indicates that the lattice constant for the out-of-plane direction approximated that of the SrTiO3 substrates at high BiCoO3 concentration.
The structures of the bulk forms of BiFeO3 and BiCoO3 are rombohedral and tetragonal, respectively. Conventional /2 XRD measurement cannot be used to identify whether the crystal symmetry is rombohedral or tetragonal in the case of the BiFeO3–BiCoO3 solid solution films. Therefore, HRXRD-RSM measurements around SrTiO3004 and 204 were employed in the investigation of the crystal symmetry of the films. [Fig. 22]The pure BiFeO3 film exhibited rhombohedral/monoclinic symmetry, as indicated by the existence of two asymmetric 204 spots in Fig. 22(b) and only one center spot of 004 in Fig. 22(a).This result is in agreement with that reported by Saito et al.for epitaxial BiFeO3 films grown on SrRuO3 (100)/SrTiO3(100) substrates. (Saitoet al., JJAP 2006) On the other hand, only single 204 and 004 spots were found for the film with aBiCoO3 of 33 at.%[Figs. 22(g) and Figs. 22(h) ], which indicates tetragonal crystal symmetry.Figures 22(c),Figs. 22(d) , Figs. 22(e) , and Figs. 22(f) show theHRXRD-RSM profilesfor the films with 16 and 21 at.% BiCoO3,respectively. Three peaks including one parallel spot andtwo tilting spots with a SrTiO3 orientation for 204, inboth films, represented the existence of a mixture of(rhombohedral/monoclinic) and tetragonal symmetries.
Raman spectroscopy was carried out in order to precisely check the change in crystal symmetry by K Nishida. (Yasuiet al., JJAP 2007)Raman spectra of the BiFeO3-BiCoO3 films and that of the SrTiO3 substrate are shown in Fig. 23. The SrTiO3 substrate shows a peak at 81 cm-1, which is shifted to a value within 75-78 cm-1 for the films with 0-33 at.%BiCoO3. It was confirmed that the peak observedfor the films does not originate from the SrTiO3 substrate.The decrease in the intensity of the SrTiO3 peak with increasingfilm thickness forpure BiFeO3 and the disappearance of the peak at600 cm-1, as shown in Fig. 23, are also in agreement with theabove results. The typical rhombohedral symmetry observed for bulk BiFeO3 was indicated for the pure BiFeO3 film and 16 at.% BiCoO3film. Different patterns with rhombohedral symmetry wereobserved for the film with 33 at.%BiCoO3, which was shown to have tetragonalsymmetry from the analysis of the HRXRD-RSM data. Furthermore, thispeak of film was very similar to that of BiCoO3 powder which has been confirmedto have tetragonal symmetry. For thefilms with 21 at.% BiCoO3, it was ascertained from Fig. 23 thatthe tetragonal and rhombohedral symmetries coexisted, which is almost consistent with the findings of the HRXRD-RSM experiment. It was revealed that the phase transition in BiFeO3–BiCoO3 from(rhombohedral/monoclinic) symmetry to tetragonal symmetryis similar to the morphotropic phase boundary (MPB) in Pb(ZrxTi1-x)O3.
Figure 24 shows the leakage current v.s. electrical field measurements taken at RT and P-E hysteresis loops measured at -193 °C for theBiFeO3-BiCoO3 films. The leakage current
density at RT was very large for the BiFeO3-BiCoO3 films with high BiCoO3 concentration, andtheleakage current density increased with increasing BiCoO3 concentration. Because of the magnitude of the leakage current at a BiCoO3concentration of 33 at.%, a leakage current measurement could not be evaluated for this film at RT using the semiconductor parameter analyzer. Although the previous discussions indicated that a small amount of Co-substitution can effectively reduce the leakage current, it can be seen from these that a large amount of Co-substitution degraded the leakage current property. In order to reduce the influence of leakage current density on the P-E hysteresis measurement for samples having a high BiCoO3 concentration, the P-E loops were measured at a low temperature of -193C. The P-E loops observed at -193 °C were of relatively high squareness and the influence of leakage current density on the P-E loops could be successfullyexcluded at this temperature, except for the BiCoO3 concentration of 33 at.%. At -193C, spontaneous polarization decreased, and the coercive field of BiFeO3-BiCoO3 films increased with increasing BiCoO3concentration.
In the case of filmswith weak ferromagnetism such as BiFeO3films on substrates, eliminating the magnetization of the substrates from the films is important for acurrate evaluation of the magnetic properties of the films.Therefore, here, the magnetic properties of SrTiO3 substrates were carefully evaluated. Figure 25(a) shows theM-H curves for two different weights of SrTiO3 substrates. The SrTiO3 substrates show a negative slope due to dimagnetism. The magnetization at 50 kOe (M50kOe) for various weights of the SrTiO3 substrates is plotted in Fig.25(b). The absolute value of magnetization decreases with a decrease in thesubstrate weight, but some of the magnetization is retained even at zero weight. This retained magnetization is considered to be the background caused by the straw of the sample holder. In this study, standard straws produced by Quantum Design Inc. were used. Figure 25(c) shows the M-H curves ofthe SrTiO3 substrate (weight = 0.0471 g) at 10 and 300 K. The hysteresis was not observed near the zero-field even at 10 K, indicating low magnetic impurity inthe SrTiO3 substrates and sample holder. The temperature dependence of M50kOeis shown in Fig.25(d). The diamagnetism slope decreased slightly with the temperature, however, it was not strongly influenced by the temperature. In this study, the magnetic properites of the films were carefullyevaluated by eliminatingSrTiO3 substratemagnetization, and the same sample holder was used in all the magnetic measurements to exculde the effect of differences among straws.
Figure 26 shows the M-H curves measured at 300 K and the corresponding magnetic parameters that were estimated from the M-H curves. For pure BiFeO3, the magnetization increased linearly ata high magnetic field. [Fig. 26(a)] Small hysteresis was observed near the zero fields, which is relatively obvious compared with that of polycrystalline BiFeO3 films. [Fig. 15]For BiCoO3 concentrations of 18–25 at.%, magnetization was clearly enhanced, and Hcwas observed. [Figs. 26(b) and 26(c)] ForaBiCoO3 concentration of 58 at.%,the M-H curve was almost identical to that of pure BiFeO3 films. There is an apparent linear increase in the magnetization at high-magnetic field for all the specimens. It was reported that by substitutingA-site Bi ions in bulk BiFeO3 with Gd or Nd, spontaneous magnetization was observed, and the magnetization increased linearly in the high-magnetic field region, which is in agreement with our results. Although it is difficult to accurately evaluate the slope at a high field due to film form, it can be considered that the antiferromagnetic spin structure still remained after substitution at the A- or B-site.The magnetic parameters M50kOe, remanent magnetization (Mr), and coercive field (Hc),estimated from the M-H curves are shown in Figs. 26(d) - 26(g). M10kOe forpolycrystallineBiCoO3-BiFeO3 films is also plotted in Fig. 26(e). The acronyms M50kOe and M10kOe indicate the magnetization at 50 kOe and 10 kOe, respectively. It was revealed that theM50kOe, Mr, and Hcvalues increased with the BiCoO3 concentration in the rhombohedral structure. This indicates theformation of ferro-like magnetic ordering. M50kOe,Mr, and Hcwere maximally enhanced at MPB composition. For a BiCoO3 concentration above 30 at.%, corresponding to a tetragonal structure,M50kOe,Mr, and Hc showed a tendency to decrease. These results indicate that the enhancement of the magnetic orderingin the MPB cannot be explained simply byferrimagnetism ina double-perovskite structure, because maximum magnetization does not take place at the half-composition. In addition, the clear relationship between the change in the magnetization and the phase transition shows that the enhancement of magnetization was not attributable to magnetic impurities.
Figure 27(a) and 27(b) show the M-H curves for 300 and 10 K for the BiFeO3–BiCoO3 film with 15 at.% of BiCoO3 concentration. Interestingly, the slope at high magnetic field became larger when decreased the temperature to 10 K. Figure 27(c) shows the temperature dependence M50kOe, Mr, and Hc. M50kOeand Mr increased with decreasing temperature; however, thesewere not show strong temperature dependence. In contrast, Hc clearly increased with decreasing temperature.
Because BiFeO3and BiCoO3 are synthesized under atmospheric pressure and a very high pressure phase, respectively, it is possible that the formation of magnetic impurities such as Co, CoFe2O4, and Fe3O4etc.,may adversely affect the magnetic properties at high concentrations of BiCoO3. In our previous studies, apparentmagneticimpurities were not observed in the XRD measurement; however, nanosized magnetic particles are difficult to detect by XRD measurements. The superparamagnetic limit is a few nanometers in diameter for Co, CoFe2O4, and Fe3O4etc. Particles with such small sizes can be detected by TEM. Therefore,the microstructure of the film was confirmed by a cross-sectional TEM observation for a BiCoO3 concentration of 17 at.%. [Fig.28] No obvious magneticimpurities were observed in the TEM image, [Fig. 28(a)] and there was no diffraction spot attributed to magneticimpurities in the NBD pattern. [Fig. 28(b)] Our previous studies on nanoparticles suggest that particles that are a few nanometers in size can be confirmed by NBD, indicating that the influence of magnetic impuritiesmight be ignored in our discussion. Although a further detailed investigation of the microstructure by high-resolution TEM observation is necessary, the enhancement of the magnetic properties might be attributable to ferro-like magnetic ordering.
Here, we briefly discuss about possibility of the magneto-electric (ME) effect at RTin BiCoO3–BiFeO3 solid solution. As mentioned above, the magnetization of BiCoO3–BiFeO3 films was enhanced with rombohedral structure at a BiCoO3 concentration below 18 at.%. In previous report, (Chu et al., 2008) for BiFeO3 with rombohedral structure, a strong coupling was reported between the ferroelectric domains of rhombohedral 71 and 109and the antiferromagnetic domains, and the anitiferromagnetic domains were reversed by ferroelectric switching at RT. In accordance withthisBiFeO3 regime,the BiCoO3–BiFeO3films below BiCoO3 concentration 18 at.% can potentially exhibit the ME effect with a macroscopic magnetization change because the rhombohedral domains exist at BiCoO3 concentration below 18 at.%. The macroscopic magnetization changes operated at RT are useful in spintronics applications such as multi-valued memory using a spin-filter device etc. To confirm the ME effect, we will clarify the role of the substitution of Fe atom for Co atom and the origin of the enhanced magnetization in BiCoO3–BiFeO3 films. In addition, we expect to observe the magnetization changes driven by the electric field as well as external pressure in the MPB (BiCoO3concentration of 20 – 25 at.%) because the MPB phase shows a large displacement due to a large piezoelectric effect.
High quality single phase BiFeO3 polycrystalline films with a space group of R3c were fabricated on Pt/Ti/SiO2/Si (100) substrates. The leakage current density of the films at RT was large and strongly affected the ferroelectric measurement. The ferroelectric measurement was carried out at low temperature to reduce the leakage current, and a large polarization of 89 C/cm2 and a coercive fieldof 0.31 MV/cm were observed. The magnetic properties at RT were primarily due to antiferromagnetism. The magnetic properties at RT were drastically enhanced by substitution of Fe in BiFeO3 with 4 at% Co, which implies the induction of ferro-like magnetic ordering. The large leakage current and coercive field were simultaneously successfully reduced by substitution of Fe with 5 at.% Co. Epitaxial strain was employed in the preparation of films with high levels of Co substitution for Fe in BiFeO3 because, under these conditions, the high-pressure phase of BiCoO3 dominates stability. (hereafter, we refer to hese highly Co-substituted. BiFeO3 films as BiCoO3-BiFeO3). The magnetization of the BiCoO3-BiFeO3 films increased drastically with an increase in the BiCoO3 concentration, and the maximum magnetization was observed at 20-25 at.% substitution.Above aBiCoO3 concentration of 25 at.%, there is a decrease in magnetization,which corresponds to the change from rhombohedral to tetragonal structural composition. Interestingly, the magnetization was maximally enhanced at the MPB of the rombohedral structure of BiFeO3 and the tetragonal structure of BiCoO3. It is well known that large piezoelectricity can be expected in the MPB; therefore,the cross-correlation between piezoelectricity and magnetism can be expected in theMPB. Furthermore, this materialhas the capacity possibility to show wide cross-correlationamong magnetism, ferroelectricity, piezoelectricity, and optical properties. Epitaxial BiCoO3-BiFeO3 solid solutionscan open up an avenue for the development of new multifunctional materials, may have potential application in devices such as multivalued memories, spin-filter devices, V-MRAM, magnetic/electric field tunability or flexibility, and piezoelectric materials with MPB etc.
The author extends appreciation for the collaborativecontribution of Prof. S. Okamura, Tokyo University of Science, to the entire study. Prof. H. Funakubo and Ph.D. student S. Yasui, Tokyo Institute of Technology, Prof. K. Nishida, National Defense Academy of Japan, Dr. T. Iijima, National Institute of Advanced Industrial Science and Technology collaborated strongly in the preparation and characterizations of BiFeO3-BiCoO3 epitaxial films presented in Section 3.The TEM observations were carried out by Dr. Andras Kovacs of Oxford University and Dr. Bae In-Tae, Binghamton University, State University of New York. The author also expresses gratitude to Prof. Y. Ando for the opportunity to write this chapter.This study was partly supported by the Grant-in-Aid forYoung Scientist Start-up program (Grant No. 18860070), Young Scientists B (No. 20760474), Young Scientists A (No. 22686001), Grand-in-Aid for Scientific Research and the Elements Science and Technology Projectfrom the Ministry of Education, Culture, Sports, Science andTechnology of Japan, by the Sasakawa Scientific ResearchGrant from The Japan Society (Grant No. 19-216), by Tohoku University Exploratory Research Program for Young Scientists (TU-ERYS), and by TANAKA Co, Ltd research found (No. J090809317).