Multiferroic materials have recently attracted extensive attention due to their attractive physical properties, such as ferromagnetism, ferroelectricity and ferroelasticity. Moreover, these coexisting properties can also result in coupling effects, such as the magnetoelectric (ME) effect, the cross-coupling between the ferromagnetic and ferroelectric properties. Based on the modified requirements: multifunctional, miniaturized and high-performance circuit, the storage devices are seeking for other possibilities except the traditional semiconductor technics^[1,2,3], like magnetoelectric random access memory (MERAMs) worked as electrically written and magnetically read^[4] and multiple-state magnetoelectric memory^[5].

As a member of multiferroic family, GaFeO₃ (GFO) is one of the most promising materials because of its good piezoelectricity at room temperature^[6,7,8], near room- temperature ferrimagnetism^[9,10], and pronounced low- temperature ME effects^[11]. Although GFO owns such attractive features, its large electrical leakage and lower magnetic transition temperature limit its application as storage devices. Therefore, its leakage current must be reduced and its magnetic transition temperature (T_C) and magnetic properties should be improved^{[12,13,14,15]}.

The conduction mechanism of GFO ceramics is studied through annealing at different atmosphere, and it was found that the leaky behavior of GFO mainly results from the oxygen vacancies and Fe valance fluctuation^[16,17,18]. Several groups^{[17,18,19,20,21,22]} explored the effects of doping in order to reduce the leakage currents and discovered that the incorporation of some bivalent cation ions can effectively decrease the leakage current by weakening the hopping processes between Fe²⁺ and Fe³⁺ states. Han, et al^[23] demonstrated that both magnetization and T_C of GFO nanoparticles are enhanced with the presence of Co contents. Moreover, the reduction of the leakage current is then expected in bivalent Co ions substituted bulk GFO. In this work, we studied the effect of Co incorporation on the structure, conductivity and magnetism of GFO ceramics. The results manifested that the multiferroic properties of GFO were well improved with proper Co substitution. The reasons for the leakage reduction and improved multiferroic properties were discussed based on components and lattice distortion.

A series of GaFe_1-xCo_xO₃ (x= 0, 0.02, 0.05, 0.07, 0.10) ceramics have been fabricated with conventional solid- state sintering technology. The starting powders including gallium oxide [Ga₂O₃, (99.99%, aladdin)], ferric oxide [Fe₂O₃ (99.99%, Rare Metallic Co Ltd)] and cobalt monoxide [CoO (AR, aladdin)] were carefully weighed in stoichiometric ratio and thoroughly balled-milled with absolute ethanol and agate balls for 12 h. After that, the mixed powder was dried and passed through a 100-mesh (150 μm) sieve. The dried mixture was precalcined at 900 ℃ in air in a muffle furnace for 6 h. After it cooled down, the mixture was ground into powder again and pressed into rods by the cold isostatic press, then the rods were calcined at 1350 ℃ in air for 10 h afterwards. Samples with different cobalt contents, 0, 2at%, 5at%, 7at%, and 10at% were denoted as GFCO-0, GFCO-1, GFCO-2, GFCO-3, and GFCO-4, respectively. The as-prepared samples were sliced into flat pieces or ground into powder for further measurement.

The powder X-ray diffraction patterns (XRD) for the crushed GFCO ceramics were obtained at room temperature using an Ultima-IV (l=0.15406 nm) with CuKa radiation (l=0.15406 nm), and the step scan mode was chosen with 0.01° per step and 2 s collection time. Refinement of the lattice constants was achieved with the Fullprof package. The morphology of the GFCO-x ceramics was observed with scanning electron microscopy (SEM, S-3400N TypeI), and the homogeneity of the components was determined with energy-dispersive X-ray spectroscopy (EDX, S-4800). A physical properties measurement system (PPMS-9(dxl)) was used to investigate the magnetization of the samples. The M-H loops were carried out at 300 K with a field of 1.58´10⁵ A/m, while the magnetization versus temperature curves were obtained with a field of 7.9´10⁴ A/m and between 50 and 390 K. For the measurement of the ferroelectric properties, electrodes were prepared on the samples via gold electroplating. I-V curves and P-E loops at room temperature were measured using a Keithley 6517B electrometer and ferroelectric analyzer (TF2000E).

The XRD patterns of the GFCO-x ceramics at room temperature are shown in Fig. 1. The XRD spectra confirmed that cobalt was doped in the gallium ferrite without original phases, such as CoO, Fe₂O₃, Ga₂O₃. There were two phases observed, one was GaFe_1-xCo_xO₃ possessed an orthorhombic structure with Pc2₁n space group, the other was secondary phase with a cubic structure and the Fd3¯m space group. The FullProf package was used to obtain the Rietveld refinement results, which were exhibited in Fig. 2. The refinement results also obtained the ratio of both phases which were listed in Table 1. The refined lattice parameters of the undoped GFO, namely, a=0.873659 nm, b=0.938267 nm, and c=0.50766 nm, are consistent with those in previous reports^[11,24].

Fig. 3 depicted the variation in the cell parameters of GFCO-x, and the cell unit volume and lattice parameters decreased with the incorporation of Co, which correlated with the ionic radius of Co²⁺ (0.074 nm) and Co³⁺ (0.061 nm) being smaller than that of Fe²⁺ (0.078 nm) and Fe³⁺ (0.064 nm). The orthorhombic distortion values (c/a and $\sqrt{2}$ c/b) were 0.581074 and 0.765176 for GFO, respectively, while the values of c/a and $\sqrt{2}$ c/b for GFCO-x (0.58116 and 0.765325, 0.51221 and 0.765472, and 0.581314 and 0.765560, respectively) were larger than those of GFO, which implied an increase in the lattice distortion. The numerical criteria R_WP and χ² were 15-18 and 1-2, proving reliability for this refinement.

Fig. 4 illustrates the morphologies of the GFCO-x samples. The particle size ranged from 2-14 μm, and their average sizes were 6.7, 4.4, 5.4, and 5.1 μm, respectively. Hence, the proportion of grain boundaries increased due to the incorporation of Co. Abnormally small grains denoted as the secondary phase were observed because of the doping of Co. In addition, the EDX results of GFCO-1, as shown in Fig. 4(e), confirmed the existence of secondary phase and suggested that the chemical compositions of the samples comprised GaFe_0.99Co_0.01O₃ for the main phase and Ga_0.9Fe_1.4Co_0.7O₄ for the secondary phase. Moreover, the microstructure of the secondary phase is identical to Ga doped CoFe₂O₄^[25].

The curves for the current density versus applied field for the GFCO-x (x=0, 1, 2) samples measured by a ferroelectric analyzer (TF2000E) are shown in Fig. 5(a). The current density decreased greatly due to the presence of Co, especially when doping a slight amount, while the insets of Fig. 5(a) showed that the polarization value decreased drastically. However, the current density increased so rapidly when the Co content reached 7% that we could not measure the current with the same tester. Instead, the current was measured with a Keithley 6517B electrometer, as shown in Fig. 5(b), while it’s obvious that the incorporation of 2at% cobalt decreased leakage current density sharply.

The variation in the current density versus Co content as depicted in Fig. 5(b), is similar to that for GFO-doped Ni^[17]. The charge conduction decreased sharply and then increased with Co content increasing. This is commonly attributed to the substitution of Fe²⁺ with other bivalent ions. The possible process can be described using Kröger- Vink notation as follows.

Oxygen vacancies produced inevitably during the preparation of GFCO ceramics can be written as:

where $\text{O}_{\text{O}}^{\times }$ is the loss of lattice oxygen, and ${{\text{{V}''}}_{\text{O}}}$ is a vacancy in the oxygen site with double positive charge. The released charge may be captured by Fe³⁺ ions to satisfy the charge balance, and then Fe³⁺ is reduced to Fe²⁺:

The coexistence of Fe³⁺ and Fe²⁺ allows electrical conduction via the hopping mechanism. However, the substitution of Fe with Co creates holes according to:

Where $\text{C{o}'}_{\text{Fe}}^{{}}$ indicates that Co is in the Fe site and has an extra negative charge, and ${{h}^{\centerdot }}$ stands for a hole. These holes are recombined with electrons as follows:

It is believed that the charge carriers are responsible for the leakage current of GFCO, and the recombination of holes and electrons weakens this electrical conduction.

Another reason for the decreased electrical conduction is the incorporation of elements that cause a large lattice distortion and increase the scattering of the conduction electrons. As mentioned earlier, the values of c/a and $\sqrt{2}$c/b for the GFCO-x samples are larger than those for the GFO, which indicates a larger lattice distortion. Fig. 5(b) shows that the leakage current density decreased by almost seven orders of magnitude; nonetheless, referring to published reports, the incorporation of elements reduced the leakage current density by up to four orders of magnitude at most^[17,19-20]. Thus, there are probably other reasons to explain the decrease in the electrical conduction, such as the introduction of a secondary phase. SEM analysis of the GFCO-x microstructures depicted the presence of a very small secondary phase that was distributed uniformly in the samples, which increased the proportion of crystal boundaries and hindered the electron scattering between the grains, thus, electrical conduction further decreased. However, when the contents of substituted Co are too high compared to the number of oxygen vacancies, Co plays as a p-type acceptor, and the material becomes a p-type conductor. Moreover, the proportion and size of secondary phase increased correspondingly, so the leakage current of 7% Co-doped samples increased.

Fig. 6(a) shows the M(T) curves for the GFCO-x (x= 0, 1, 2, 3, 4) samples, which were measured in zero-field- cooled (ZFC) mode. The hysteresis loops of all GFCO samples at room temperature with an applied field of 20000 Oe were displayed in Fig. 6(b). Fig. 6 indicated that the Co doping GFO ceramics owned enhanced magnetism which is in accordance with previous results^{[21-22,26-27]}. The magnetism of GaFeO₃ can be explained by the presence of unequal and antiparallel Fe moments and Co moments in Fe1 and Fe2 sites along the c axis^[11,28]. With the substituted Co on Fe-site whose magnetic moment isn’t equal to Fe’s, the net magnetic moment increased along the c axis. GFCO ceramics still possessed magnetism after the main phase transformed into a paramagnet at elevated temperatures. Hence, it suggested that there was two T_C in GFCO ceramics, one is near 200 K and the other is higher than 400 K, and that the secondary phase is responsible for the higher T_C. According to published papers^[25], CoGa_xFe_2-xO₄ (x= 0-0.8) possessed varied T_C from 773 to 523 K. While both CoGa_xFe_2-xO₄ and the secondary Ga_0.9Fe_1.4Co_0.7O₄ phase had cubic microstructures with a Fd¯3m space group and similar elemental proportions. Thus, we inferred that the as-prepared GFCO-x ceramics had two magnetic transition temperatures, and the higher T_C was attributed to the secondary phase.

In conclusion, GaFe_1-xCo_xO₃ (x=0, 0.02, 0.05, 0.07, 0.10) ceramics were prepared by the conventional solid-state method. X-ray characterization and Rietveld refinement results revealed that the as-prepared GFCO-x samples possessed an orthorhombic crystal structure with a Pc2₁n space group for the GFO and a cubic crystal structure with a Fd¯3m space group for the secondary phase. Moreover, the more Co content was doped, the higher proportion of the secondary phase there was, so did the degree of crystal distortion. The surface morphology of the GFCO samples was analyzed using SEM, which indicated that the mean grain size diminished and the proportion of grain boundary increased because of the presence of Co. The Co content doping and secondary phase decreased the leakage current effectively, especially the leakage current of the 2at% Co-doped GFO ceramics decreased by approximately 7 orders of magnitude compared to that of the undoped GFO ceramics. Moreover, doping Co improved the magnetic property because of the secondary phase and enhanced lattice distortion. This work shows that Co has a positive influence on magnetism and trace amounts of Co incorporation in GaFeO₃ can decrease leakage current sharply, while excess Co leads to the drastic rebound in leakage current.