Dimensionality tuning of the electronic structure in Fe3Ga4 magnetic materials

This work reports on the dimensionality effects on the magnetic behavior of Fe3Ga4 compounds by means of magnetic susceptibility, electrical resistivity, and specific heat measurements. Our results show that reducing the Fe3Ga4 dimensionality, via nanowire shape, intriguingly modifies its electronic structure. In particular, the bulk system exhibits two transitions, a ferromagnetic (FM) transition temperature at T1 = 50 K and an antiferromagnetic (AFM) one at T2 = 390 K. On the other hand, nanowires shift these transition temperatures, towards higher and lower temperature for T1 and T2, respectively. Moreover, the dimensionality reduction seems to also modify the microscopic nature of the T1 transition. Instead of a FM to AFM transition, as observed in the 3D system, a transition from FM to ferrimagnetic (FERRI) or to coexistence of FM and AFM phases is found for the nanowires. Our results allowed us to propose the magnetic field-temperature phase diagram for Fe3Ga4 in both bulk and nanostructured forms. The interesting microscopic tuning of the magnetic interactions induced by dimensionality in Fe3Ga4 opens a new route to optimize the use of such materials in nanostructured devices.

Scientific RepoRts | 6:28364 | DOI: 10.1038/srep28364 antiferromagnetism (AFM) takes place before vanishing at a Néel temperature T 2 = 390 K 13 . This behavior has been explained by Moriya and Usami's theory, which predicts coexistence of FM and AFM states in itinerant electron systems 13,14 . To investigate these peculiar magnetic and structural behaviors, several chemical substitution studies have been performed on both Fe and Ga sublattices [15][16][17][18] . In particular, it has been recently reported that when grown by the alternative deposition of Fe 3 GaAs and GaAs on a GaAs (001) substrate, the Fe 3 Ga 4 compound presents a distinctive photo-enhanced magnetization at room temperature 19 . On the other hand, Fe-Ga nanowires, with different composition from that studied in this work, have been successfully fabricated by electrochemical deposition and have been widely investigated for potential use as sensing elements in a variety of microelectromechanical and nanoelectromechanical-based biomimetic devices 20,21 . Therefore, it is very instructive to study the dimensionality effects on Fe 3 Ga 4 interesting properties. To the best of our knowledge, no such studies about Fe 3 Ga 4 compound have not yet been reported in the literature.
In this work, we report the dimensionality effects on the Fe 3 Ga 4 magnetic field-temperature phase diagram (H-T), which was constructed for both bulk and nanowire systems using magnetization, specific heat and electrical resistivity measurements. The results are discussed within the framework of the Moriya and Usami's theory on magnetic phase transitions in itinerant electron systems. As such, this work reveals unambiguous evidence for a dimensionality tuning of the Fe 3 Ga 4 electronic structure that modifies the microscopic exchange parameters considered in Moriya and Usami's theory. More generally speaking, we strongly believe that the possibility of growing intermetallic compounds in nanowire form will open an interesting branch in the understanding of fundamental properties, as well as permitting to control the nanowire characteristics for desired applications.

Results and Discussion
Morphology and Composition. Scanning electron microscopy (SEM) was first performed to verify the presence, dimensions and composition of the Fe 3 Ga 4 nanowires. As shown in Fig. 1(a), most of the observed isolated nanowires exhibit a diameter of ~ 250 nm and a length of ~ 25 μm, giving a length/diameter aspect ratio of around 100. Figure 1(b) displays a magnified view of two nanowires showing their surface quality by presenting a very low roughness and no apparent defects. Its energy dispersive X-ray spectrometry (EDS) mapping for Fe Kα and Ga Lα energies, shown respectively in Fig. 1(c-d), clearly states that both Fe and Ga elements are present in the nanowires. However, the quantitative chemical composition given by EDS was not used since the analyzed surface was not polished and that Fe has a weak detection power compared to Ga, due to its low atomic number.
Structural Characterization. X-ray diffraction (XRD) data were used to identify the obtained bulk crystalline phase. The measurements were performed using Cu Kα radiation and θ-2θ scans were recorded in the 25°-55° range. From the study of the X-ray diffraction pattern we can affirm that the Fe 3 Ga 4 phase was formed in the expected C2/m (#12) space group. The obtained crystals present a higher crystalline quality compared to those fabricated via arc melting 13 . Furthermore, no formation of additional phases was observed, such as FeGa 3 , which is commonly observed during Fe 3 Ga 4 growth 22 .
On the other hand, the nanowires structural characterization required synchrotron X-ray diffraction and absorption measurements, both performed at the Brazilian Synchrotron Light Laboratory (LNLS). The main difficulty encountered arises from the alumina membrane present around the nanowires, the latter only forming a small percentage of the measured system.
In a first step, we used the X-ray Diffraction and Spectroscopy (XDS) beamline to acquire the XRD spectra of the membrane with embedded nanowires, as well as the untreated empty porous alumina membrane. The data were collected in transmission mode at room temperature and atmospheric pressure with a wavelength of 0.6199 Å. The spot size was 2.0 × 0.2 mm 2 . As we can see in Fig. 2, the diffraction profile of the empty membrane could be reasonably well modeled with the program suite GSAS+ EXPGUI 23,24 by an admixture of ηand θ-Al 2 O 3 phases (space groups Fd-3m and C2/m, respectively) 25  where the respective compositions of the membrane phases were estimated from the unit cell volumes through the Vegard's law. Extra Bragg peaks due to additional polycrystalline phases were also observed in the membrane with embedded nanowires profile, with intensities of ~ 10% or lower with respect to the strongest peak of the main monoclinic Al 1.4 Ga 0.6 O 3 membrane phase. These extra peaks could not be related to any known phase of the Fe-Ga binary system, and most likely arise from additional phases of the Al 2−x Ga x O 3 membrane. Efforts towards an unambiguous identification of the minor membrane phases were unsuccessful, which is justified considering the possibilities of compositional fluctuations that would shift the Bragg peaks with respect to the expected angles, preferred orientation of membrane phases and the large variety of possible Al 2−x Ga x O 3 phases. In any case, neither Fe 3 Ga 4 nor any other known Fe-Ga binary compound such as FeGa 3 29 were unambiguously identified in the membrane with nanowires diffractogram (see Fig. 2 inset), meaning that the weight fraction of any such metallic phases within the membrane must be below 2%. This conclusion is consistent with our SEM analysis, which also indicated a low filling factor of the pores. We should mention that, while the Al ions in the membrane are partly replaced by Ga during the Fe 3 Ga 4 crystal growth procedure, we cannot discard the possibility that, conversely, some Ga ions in the crystals are partly replaced by Al 18 .
In order to obtain structural information of the phases containing Fe, an element-specific technique became necessary. X-ray absorption measurements were therefore performed at the Fe K-edge in fluorescence mode at the XAFS2 beamline on the embedded nanowires, while bulk Fe 3 Ga 4 and hematite (α-Fe 2 O 3 ) were taken as standard samples. The near-edge region of the spectrum (XANES) is shown in Fig. 3(a), whereas a more extended region up to 7350 eV, covering the first extended X-ray absorption fine structure (EXAFS) oscillations, is displayed in Fig. 3(b). The prominent XANES peak at 7130 eV found for the embedded nanowires system indicates the presence of oxidized Fe 3+ ions, while the strong pre-edge feature at 7112 eV is consistent with the presence of a significant fraction of non-oxidized Fe. The observed spectrum could be fitted by a combination of the spectra from Fe 3 Ga 4 (62(1)%) and α-Fe 2 O 3 (38(1)%), with good agreement both in near-edge region ( Fig. 3(a)) and for first EXAFS oscillations above 7170 eV ( Fig. 3(b)). Based on this analysis, we attribute the presence of oxidized Fe 3+ in our membrane with nanowires sample to a small degree of substitution of Fe atoms into the Al 2−x Ga x O 3 membrane, while the major metallic Fe component found in our XAS spectrum is attributed to the Fe 3 Ga 4 nanowires. Despite the good overall agreement provided by our simple analysis using Fe 2 O 3 and Fe 3 Ga 4 standards, a clear discrepancy between the observed spectrum for the embedded nanowires sample and the mixture of standards can be seen at ~ 7160 eV. This may be attributed to a distinct non-local electronic structure of α-Fe 2 O 3 with respect to Fe 3+ ions impurity into the Al 2−x Ga x O 3 membrane, which may lead to large energy shifts of high-energy XANES excitations involving charge transfer from Fe to neighboring ions.

Magnetic Characterization.
In order to investigate the magnetic properties and build the magnetic phase diagram of the Fe 3 Ga 4 bulk and nanowire array, magnetization measurements were acquired both as a function of the temperature T (2-400 K) and the applied magnetic field H (± 20 kOe). For the nanowire array characterization, the magnetic field was applied parallel to the nanowires axis.
From the magnetization as a function of temperature curves, the two expected magnetic phase transitions are distinguished in terms of a magnetization drop around 50 K (T 1 ) and a peak around 390 K (T 2 ). In the case of nanowires, the reduced dimensionality affects T 1 and T 2 values in opposite ways. As we can see in Fig. 4, T 1 increases while T 2 decreases. Moreover, the relative magnetization drop at the first temperature transition T 1 is smaller for the nanowire array compared to the bulk one. This could suggest that we are dealing with a phase transition from a FM to another magnetic phase order with net macroscopic magnetization in the nanowire case (FERRI or coexistence of FM and AFM). As discussed later, this phase coexistence is predictable from Moriya's theory. This is the first evidence that the nature of this transition differs in the nanowires and in the bulk samples. We also observe that the temperature-dependent magnetization curves exhibit appreciable thermal hysteresis around T 1 , but only in the Fe 3 Ga 4 nanowires case (not shown here), which decreases when we increase the applied magnetic field. These effects persist for all magnetic field values used in the magnetization measurements, and are characteristic of first order phase transition. In the case of bulk samples, no thermal hysteresis is noticed. Furthermore, we can see that the magnetic signal from Fe 3 Ga 4 phase (in emu/g) is about 500 times smaller that for the bulk crystal, which is consistent with our estimative by XRD that the weight fraction of the Fe 3 Ga 4 phase in the membrane is below 2%.
Between 50 and 300 K, we observe a field-induced transition on the field-dependent magnetization curves for the bulk compound (Fig. 5). On the other hand, no such transition is detected for the nanowire system, but rather a large magnetic susceptibility and the presence of a magnetic hysteresis between 2 and 300 K, also supporting the FERRI or coexistence FM and AFM behavior suggested before. The magnetic results for bulk Fe 3 Ga 4 are in good agreement with those obtained by Kawamiya and Adach 13 and interpreted based on the Moriya and Usami's theory for magnetic phase transition in itinerant electron systems. The theory takes into account the coexistence of uniform and staggered magnetization using a Landau free energy expression, which contains both magnetization terms (up to the fourth power) and the coupling terms between both magnetization components 14 . In the absence of magnetic anisotropy, the relative magnitudes of the free energy coefficients give rise to four different magnetic phase diagrams, one of which corresponding to the experimentally obtained behavior for bulk Fe 3 Ga 4 .
Specific Heat Characterization. Specific heat (C) measurements were performed in the 2-100 K and 2-20 kOe ranges of temperature and magnetic field, respectively. As shown in Fig. 6(a,b), the linear fits of the C/T  versus T 2 curves at low temperatures give similar Debye temperatures (θ B = 230(2) K and θ N = 253(15) K for bulk and nanowires, respectively). Since θ is related to the phonon spectrum, which in turn is related to the material crystalline structure, this result demonstrates that the nanowire array exhibits the same crystalline structure than that of Fe 3 Ga 4 bulk phase. Moreover, we observe a reduction in the contribution of the conduction electrons to the specific heat when the system dimensionality is reduced. Remarkably, the electronic coefficient (γ) drops from 25.60 mJ/mol.K 2 to almost zero (within experimental error), indicating that the Fe 3 Ga 4 compound tends to become insulating in the nanowire morphology. These data are consistent with the magnetization results, since that a FERRI or FM behavior is most likely to be expected in localized electron magnetism. In addition, field-dependent specific heat measurement performed at 80 K shows a maximum at about 4 kOe for bulk Fe 3 Ga 4 (see inset Fig. 6(a)), which corresponds to the field transition exhibited in magnetization hysteresis curves. Similar behavior is observed for other temperature values. However, no such maximum is detected the 0-20 kOe range for the nanowire system (see inset Fig. 6(b)), indicating the absence of field-induced metamagnetic transition.
Electrical Characterization. Bulk electrical resistivity measurements were performed using a standard four-probe technique in the 2-400 K temperature range and under magnetic fields of 0 and 7 kOe. Without applied magnetic field, temperature-dependent electrical resistivity curve exhibits a slope change at approximately 65 K (Fig. 7, see upper left inset). This temperature roughly corresponds to the bulk T 1 value in temperature-dependent magnetization curve done under 500 Oe applied magnetic field. The linear dependence of resistivity as a function of squared temperature (Fig. 7, lower right inset) indicates that spin fluctuations play an important role at low  temperature 30,31 . These results reveal that the resistivity of metamagnetic compounds is dominated by spin fluctuations, an important aspect that justifies the use of Moriya's theory to explain the Fe 3 Ga 4 compound magnetic behavior. With a magnetic field of 7 kOe, the temperature where the resistivity slope is changing, which happens to be 70 K, increases, in agreement with the magnetization results (not shown here).
Magnetic Phase Diagram. Experimental magnetic, specific heat and electrical results yield to the built magnetic phase diagrams of the Fe 3 Ga 4 bulk and for the system at low dimensionality (Fig. 8). Based on our observations and Moriya and Usami's theory, we have strong evidences that the effects of reducing dimensionality can be explained as resulting from modifications of the system free energy. Those modifications originate from electronic structure changes, and end up affecting the compound magnetic phase diagram. Thus, Fe 3 Ga 4 bulk compound would exhibit at T 1 either a ferromagnetic to antiferromagnetic transition (in agreement with the literature for bulk), or a ferromagnetic to ferrimagnetic or coexistence of ferromagnetic and antiferromagnetic transition, if under nanowire form. The specific heat data are also consistent with these conclusions, since a FERRI or FM behavior is most likely to be expected in localized electron magnetism. Finally, at T 2 , the applied field decreases the temperature of maximum susceptibility similarly for both systems.

Conclusions
We have successfully grown both Fe 3 Ga 4 bulk and nanowires by the new technique of metallic-flux nanonucleation. All the results suggest that, based on the Moriya's theory, there is a transition from ferromagnetic to antiferromagnetic state for Fe 3 Ga 4 bulk, while that for Fe 3 Ga 4 nanowire array, there is a transition from ferromagnetic to ferrimagnetic or a coexistence of ferromagnetic and antiferromagnetic state. These changes in the nature and aspects of the magnetic transitions are supported to arise from electronic structure modifications, which can be tuned by controlling the system dimensionality. These results open the possibility for studying fundamental aspects of magnetic behavior of any intermetallic compound that can be fabricated by the metallic flux method, as well as could allow the design of new nanostructures used in nanometric devices with desired physical features.

Methods
Bulk single crystals as well as nanowires specimens of Fe 3 Ga 4 compound were synthesized through the innovating method of metallic-flux nanonucleation (MFNN). The MFNN technique is based on the conventional flux-growth technique 32 performed in a nanoporous alumina template that mediates the preferential nucleation of the single crystals in the desired geometry. First, alumina templates were prepared via a hard anodization process 33 . High purity (99.999%) aluminum foils were degreased, cleaned, dried, thermally treated at 400 °C for 3 hours and finally electropolished. The anodization procedure was performed in a 0.3 M oxalic acid bath at 1 °C. After a first anodization performed at 40 V for 5 min, the anodization voltage was gradually increased until reaching 120 V at a 0.8 V/s rate, where it was kept for 2 hours. The obtained templates are 100 μm thick, with 250 nm diameter pores arranged in a hexagonal array with 270 nm interpore distance. Both Fe 3 Ga 4 compounds in bulk and nanowire shape were simultaneously obtained in the same bath using the Ga self-flux technique. For this procedure, appropriated quantities of high purity (99.995%) Fe and (99.999%) Ga precursors were deposited in an alumina crucible, together with the nanoporous alumina template properly fixed at the crucible bottom and sealed in a quartz tube under vaccum. In order to form Fe 3 Ga 4 crystals via the flux-growth technique, the sealed tube was heated up to 1100 °C for 12 h, cooled down to 700 °C at 3 °C/h and finally cooled down to 400 °C at 8 °C/h. The filled pores embedded in the nanoporous alumina template were subsequently carefully separated from the bulk crystals.
Since the MFNN method yields a simultaneous synthesis of crystalline bulk and nanowires, any intermetallic compound that can be prepared by the flux-growth method can also be, in principle, synthesized in nanowire form. This represents an enormous advantage for the MFNN method.