betaFeedback

WorksPhysics and Astronomy

Enhancing the magnetism and giant anomalous Hall effect in thin Fe-Al films via B2 nanophase growth

Abstract

The properties of alloys that undergo to chemical order-disorder transformations depend heavily on the degree of ordering in the crystal lattice. In the literature, it is well established that the ordering in a magnetic alloy such as Fe-rich Fe_xAl_1-x (x>0.5) leads to reducing its magnetization and even to a transition from the ferromagnetic (FM) to paramagnetic (PM) state at x<0.7. Studying the ordering kinetics in thin (50 nm) Fe_xAl_1-x films with a non-stoichiometric composition (0.5<x<0.7), we demonstrate the opposite behavior: When the alloy is aged at a high temperature Ta>600 °C, the ordering process is accompanied by an increase in magnetization and related properties. For example, we find the further enhancement of the giant anomalous Hall (AH) effect found recently in Fe_xAl_1-x alloys. Based on both experimental data and theoretical modeling, we argue that these properties are enhanced due to the nucleation and growth of the B2-Fe_0.5Al_0.5 phase. Growing B2 nanocrystals enable segregation and clustering of excess Fe in the alloy. It has been revealed that the PM phase, which is formed in the aged samples and contains Fe-enriched superparamagnetic clusters, contributes to the AH resistivity even more than the FM phase in the as-grown sample. Our findings open a route for improving the properties of functional alloys.

Introduction

The study of the ordered and disordered states in Fe(Co)-rich Fe(Co)x_xAl1−x_{1-x} alloys and their derivatives, such as Fe2_2(Co2_2)XAl(Si) Heusler alloys with X = Co, Fe, Mn, Cr, V,..., is currently of great interest due to their half-metallic behavior [1, 2] and sharp magnetic transitions [3, 4], which have thermoelectric [5, 6], spintronic [7, 8, 9, 10, 11, 12, 13], magnetocaloric [14, 15, 16], and microbiological [17] applications. The order-disorder transformations in binary Fex_xAl1−x_{1-x} alloys, which crystallize into the cubic lattice of the CsCl type, have been studied for decades, since the first observations of ordering and superstructure formation in these systems in 1932 [18]. In a perfectly chemically ordered stoichiometric alloy, atoms of all species occupy regular positions in the atomic lattice, and so, the long-range order (LRO) in the alloy becomes equal to unity. The vast majority of alloys exhibit a strong dependence on their physical properties on the atomic arrangement and LRO. In particular, studies on the magnetic and structural properties of the Fex_xAl1−x_{1-x} system have shown [19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29] that magnetization in the alloy only decreases with increasing LRO. In Figure 1a, this magnetization behavior is shown schematically as a function of aging time. In the disordered A2 state, the alloy is ferromagnetic (FM) at room temperatures, but it becomes paramagnetic (PM) when it transitions to the ordered B2 phase. This transformation occurs in the alloys with a Fe content ranging between x=0.5x=0.5 and x=0.7x=0.7 at aging temperatures TaT_a below the critical temperature TcT_c for the B2↔\leftrightarrowA2 phase transition, as illustrated in the structural (Figure 1b [30]) and magnetic (Figure 1c [19]) phase diagrams of this system. The established view in the literature [19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29] is that the ordering transformation in Fex_xAl1−x_{1-x} is a single-phase process, where magnetism is progressively suppressed due to continuous atomic ordering, as schematically illustrated in Figure 1d for Fe0.6_{0.6}Al0.4_{0.4}.

Schematic plot of magnetization and structural and magnetic phase diagrams for Fe-Al alloys

Figure 1. (a) Schematic plot of magnetization as a function of aging time in Fex_xAl1−x_{1-x} (0.5<x<0.70.5<x<0.7) at T<TcT<T_c. The insets show the atomic structure of the alloy in its two structural states. In the disordered A2 state, the alloy is ferromagnetic (FM), while it becomes paramagnetic (PM) in the ordered B2-Fe0.5_{0.5}Al0.5_{0.5} state. (b-c) Structural (b) and magnetic (c) phase diagrams, which are adapted from Ref. [30] and Ref. [19], respectively. In the structural phase diagram, the A2↔B2 and solid↔liquid transitions are indicated by black and red lines, respectively. The magnetic phase diagram is plotted for the ordered B2-Fex_xAl1−x_{1-x} alloy, assuming continuous ordering of the alloy, as illustrated in (d). In the diagram, the PM↔FM and PM↔mictomagnet [4] transitions are indicated. (d-e) Schematic illustrations for two possible scenarios of ordering in a nonstoichiometric binary alloy. One of them (d) is continuous ordering at which excess atoms occupy sites randomly on the sublattice of another component. Another one (e) takes place through nucleation and growth of an ordered stoichiometric phase, with segregation and clustering of excess atoms.

In this model, excess Fe atoms randomly occupy the Al sublattice. This ordering mechanism is thermodynamically permissible, as it aligns with the Lifshitz criterion for second-order transformations in the CsCl-type lattice [31].

However, an alternative, first-order transformation pathway involving the nucleation and growth of a stoichiometric ordered phase is also plausible, particularly within a miscibility gap [32]. Such a heterogeneous process (Figure 1e) is often catalyzed by lattice defects [33, 34, 35, 36] and, crucially, would lead to a fundamentally different microstructure. Indeed, growing evidence from studies on mechanical strengthening in Fe-Al-based alloys confirms that B2-phase precipitation within a disordered matrix is a viable mechanism [37, 38, 39, 40, 41]. Yet, the impact of this alternative ordering scenario on the system's magnetic and transport properties remains largely unexplored, creating a critical knowledge gap. Our work addresses this gap by directly investigating the high-temperature ordering kinetics and its consequences for magnetism. In our work, we have investigated the kinetics of chemical ordering in thin (50 nm) films of nonstoichiometric Fex_xAl1−x_{1-x} alloys (0.5<x<0.70.5<x<0.7) using methods of both structural and magnetic characterization. In contrast to previous studies [19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29], we found that, at sufficiently high Ta>600T_a>600 °C but still lower than TcT_c, chemical ordering is accompanied by not a decrease in magnetization but rather by its increase. This improvement of the magnetism is also reflected in other properties such as the intensity of the magneto-optic Kerr effect (MOKE) and the giant anomalous Hall (AH) resistivity ρAH\rho_{\mathrm{AH}} found recently in Fe-Al alloys [5, 6, 8, 10, 29]. We have revealed, however, that high-temperature aging results in an even higher ρAH\rho_{\mathrm{AH}} than that in the as-grown sample, which has a higher magnetization. To understand these behaviors, we analyzed our samples using transmission electron microscopy (TEM), including the Electron Spectroscopy Image (ESI) and selective area electron diffraction (SAED) modes. In these studies, we found that the ordering occurs through classical nucleation and growth of nuclei of the ordered B2-Fe0.5Al0.5\mathrm{Fe}_{0.5}\mathrm{Al}_{0.5} (B2) phase, so that the growing nuclei force excess Fe atoms to segregate into the surrounding matrix. This scenario for chemical ordering is schematically illustrated in Figure 1e. In fact, the enhancement of magnetization – we observe in our films at high TaT_a – can reflect the clustering of excess Fe in the alloy. The clustering tendency is indicated in the ρAH\rho_{\mathrm{AH}} data we have obtained. Our experimental observations are supported by molecular dynamics (MD) simulations. Thus, we demonstrate that the long-standing belief that chemical ordering invariably suppresses magnetism is not universal. With the prototypical FexAl1−x\mathrm{Fe}_x\mathrm{Al}_{1-x} system, we establish that high-temperature aging triggers an ordering mechanism via B2 discontinuous nucleation, enabling the enhancement of magnetization and related properties due to the formation of Fe-rich magnetic clusters.

Methods

Sample preparation

50 nm thick FexAl1−x\mathrm{Fe}_x\mathrm{Al}_{1-x} films were prepared with dc magnetron co-sputtering from individual high purity iron (Fe – 99.99 %) and aluminum (Al – 99.999 %) targets at room temperature. As substrates, we used 100 Si wafers coated with 100 nm of SiO2 and 50 nm of Si3N4 (Si/SiO2/Si3N4), fused quartz with a thickness of 0.8 mm, and commercially available 50 nm thick Si3N4 membranes. For film sputtering, an AJA 2200 multichamber system was employed at a basic pressure down to ∼6×10−8\sim6 \times10^{-8} mbar, with introducing argon gas of 99.999% purity into the chamber using gas mass flow controller attached to the system. The base pressure in the system during sputtering was maintained in the range of 4.0 – 5.0 x 10−310^{-3} mbar. Film composition was varied by changing magnetron powers and thus sputtering rates of Fe and Al and checked by measuring glancing incidence X-ray reflectivity (XRR) with a Bruker D8 Discover x-ray diffractometer. To determine the composition, XRR curves were modeled and fitted to the experiment with employing the DIFFRAC.Leptos 7.04 (Bruker AXS) software; see Supplemental Information for greater detail. Thermal treatments at temperatures up to 800 °C were carried out in a high-vacuum chamber, with residual gas pressure of approximately 10−410^{-4} mbar. In addition, a rapid thermal annealing (RTA) system was used for short-term (down to ∼10\sim10 s) heat treatments at TaT_a up to 1000 °C. The RTA process involved the use of He or Ar gases that flowed through a heated sample at a rate of 10 l/min and a pressure of 1.5 kg/cm².

Sample characterization

The phase transformations that occur in our films were investigated by examining both their structural and magnetic properties. We studied the crystalline structure of the films using TEM, which was operated at a LIBRA 200 MC apparatus (Carl Zeiss, Jena), operating at 200 kV. The apparatus was equipped with an OMEGA electron energy loss spectrometer, which allowed us to map the distribution of Fe and Al over the film surface in the ESI mode. In the TEM experiments, we determined the distribution in grain sizes as well as the LRO by analyzing SAED patterns. In the as-grown films, the average crystallite diameter was evaluated as 10.5±1.0 nm [26, 27]. The order degree was determined as LRO = (Is/If)exp/(Is/If)id\sqrt{(I_s/I_f)_{\mathrm{exp}}/(I_s/I_f)_{\mathrm{id}}}, [42] where (Is/If)exp(id)(I_s/I_f)_{\mathrm{exp(id)}} are ratios between intensities of a superstructure (Is)(I_s) and fundamental (If)(I_f) diffraction peaks, which are found experimentally [(Is/If)exp][(I_s/I_f)_{\mathrm{exp}}] and taken from the calculations for the B2 structure [(Is/If)id][(I_s/I_f)_{\mathrm{id}}].

The magnetic properties were studied at room temperature with ferromagnetic resonance (FMR), MOKE magnetometry, and by measuring the AH resistivity. FMR data were collected with a Bruker EMX 10/12 spectrometer of electron paramagnetic resonance at a fixed frequency of 9.5 GHz. The detection of the resonance lines from the uniform precession modes allows for determining the saturation magnetization 4πM4\pi M from the Kittel resonance equation. The used MOKE setup was home-built and based on a Faraday modulator technique. As a light source, a He-Ne laser (λ\lambda=633 nm, 5 mW, Thorlabs HRS015B) was employed. The MOKE intensity was measured as a function of magnetic field HH applied in the film plane to generate a MOKE hysteresis loop. The AH resistivity was measured using the standard four-probe method by applying a dc current of II=10 mA. Original data obtained by taking FMR spectroscopy and MOKE magnetometry are presented in Supplemental Information.

Modeling

MD simulations were performed using the LAMMPS [43] package to investigate the relaxation of a Fex_xAl1−x_{1-x} (x=0.6x=0.6) crystal to the thermodynamic equilibrium at elevated temperatures. The initial atomic configuration was constructed based on the crystallographic data for the disordered A2 state, obtained from the CIF file (607483-ICSD) for the Fe0.75_{0.75}Al0.25_{0.25} alloy. In this structure, a part of Fe atoms was substituted with Al in a random manner, in order to change the composition from x=0.75x=0.75 to the desired one, x=0.6x=0.6.

The crystal was generated by replicating the unit cell in three dimensions to form a 5×5×55 \times5 \times5 nm3^3 monocrystalline supercell consisting of different random realizations of the disordered structure from one unit cell to another one. Periodic boundary conditions were applied along all three Cartesian directions. Interatomic interactions were described using an Embedded-Atom Method (EAM) potential [44]. The resulting atomic configurations were visualized and analyzed using the OVITO software package [45].

Results

Enhancing the magnetism and AH resistivity

In our Fex_xAl1−x_{1-x} films, we have investigated how the magnetization 4πM4\pi M determined using FMR as well as some related properties such as the MOKE intensity and ρAH\rho_{\mathrm{AH}}, behave with increasing TaT_a. At sufficiently low Ta≤600T_a \le600 °C, the samples studied exhibit a decrease in 4πM4\pi M, down to vanishing room-temperature FMR responses in the samples with a sufficiently low Fe content, x<0.6x<0.6. Basically, such a behavior is consistent with the previous data reported [19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29]. However, we observe the opposite trend at a higher TaT_a.

Figure 2 (a) shows 4πM4\pi M versus aging time tt for the samples with different compositions – x=0.56x=0.56, x=0.65x=0.65, and x=0.68x=0.68. The films were aged at 600 °C for 0.5, 5, 50, and 500 min. Also, the samples aged at 600 °C for 500 min were subsequently aged at 900 °C for 0.5 min in the RTA system. We see that, after RTA aging at 900 °C, the samples with such compositions exhibit a significant increase in magnetization at room temperature. Quantitatively, the same trend was observed in the specimens aged at 800 °C for 1 hour in a high-vacuum chamber. This behavior is in contrast to what we observe at a lower Ta≤600T_a \le600 °C. We note that the Fe content of x=0.56x=0.56 was lowest, at which we detected the reentrance of FMR response after elevation of TaT_a up to 900 °C. The original FMR data for all compositions studied are presented in Supplemental Information.

Four-panel plots showing magnetization and anomalous Hall resistivity

Figure 2. (a) Time evolution of magnetization 4πM4\pi M in the films with different Fe content, which were aged at 600 °C and subsequently at 900 °C. The quantity of 4πM4\pi M was measured with FMR by using Kittel resonance equations. (b-d) Room temperature AH resistivity ρAH/ρ0\rho_{\mathrm{AH}}/\rho_0 (ρ0\rho_0 is this quantity for pure Fe) versus applied magnetic field HH for the samples with x=0.56x=0.56 (b), x=0.65x=0.65 (c), x=0.68x=0.68 (d) in the as-grown films as well as after film aging at 600 °C and 900 °C. There are fast (FM) and slow (PM) stages in the ρAH\rho_{\mathrm{AH}} evolution. The increase in the PM susceptibility after aging at 900 °C, results from a contribution of superparamagnetic clusters, which form owing to squeezing out the excess Fe from regions of the B2 phase.

Here, we also emphasize that we do not observe the oxidation effects [46] on magnetization of the samples aged at high TaT_{\mathrm{a}} up to 900 °C (at least, at short annealing times ∼100\sim100 s), We find, for example, that RTA under He flow at the air atmosphere and high-vacuum annealing induce the same changes in magnetization. The MOKE intensity measured for the as-grown samples and after RTA and high-vacuum thermal annealing is illustrated in Table 1. The original MOKE data obtained (MOKE hysteresis loops) are presented in Supplemental Information for all compositions studied.

sample2θc2\theta_{\mathrm{c}}, degr.xas-grown600 °C 100 s / 1 hour800 °C 100 s / 1 hour900 °C 30 s
Fe(110)-Al(150)0.670.681.40.25 / 0.20.5 / 0.40.7
Fe(100)-Al(150)0.660.650.80.1 / 0.050.15 / 0.10.3
Fe(85)-Al(150)0.640.560.2 (1.0)<0.05<0.05<0.05<0.05<0.05<0.05 (0.05)

Table 1. Comparison of RTA and longtime (1 hour) high-vacuum thermal annealing on MOKE; see Supplemental Information for the original data. The measured MOKE intensity is a difference in Kerr rotation in saturated positive and negative applied magnetic fields, HH. In brackets (sample with x=0.56), the MOKE intensity is indicated, which was measured by taking a polar MOKE at H=±5H=\pm5 kOe. The film composition was changed by varying the magnetron powers indicated in the first column by numbers in units of W. 2θc2\theta_{\mathrm{c}} is the double total external reflection edge, which is related to the average dielectric susceptibility ⟨ε⟩\langle\varepsilon\rangle across the film as cos⁡2θc=Re⁡⟨ε⟩\cos^2\theta_{\mathrm{c}}=\operatorname{Re}\langle\varepsilon\rangle [47].

It is of special interest to study the AH resistivity of FexAl1−x\mathrm{Fe}_{x}\mathrm{Al}_{1-x} [5, 6, 8, 10, 29]. In such Fe-based alloys, due to their half-metallicity and thus high spin polarization [1, 2], the AH effect can be called as the giant AH effect, which can be even much larger than that in pure Fe (ρ0=0.18 μΩ cm\rho_0=0.18\ \mu\Omega\ \mathrm{cm} [10]). The empirical relationship between the AH resistivity and magnetization can be expressed as ρAH=Rs4πMz\rho_{\mathrm{AH}}=R_{\mathrm{s}}4\pi M_z, where RsR_{\mathrm{s}} is the conventional notation for the AH coefficient and 4πMz4\pi M_z is the out-of-plane component of magnetization. In our case, the studies of ρAH\rho_{\mathrm{AH}} enable a new kind of information about the ordering tendencies in the samples under study. Figure 2 (b-Figure 2d) shows our results of measuring ρAH=VHL/I\rho_{\mathrm{AH}}=V_{\mathrm{H}}L/I ρAH=VHL/I\rho_{\mathrm{AH}}=V_{\mathrm{H}}L/I (VHV_{\mathrm{H}} is the measured Hall voltage, LL is the film thickness, and II is the current passed through the film) as a function of applied magnetic field HH. In the plots, the quantity of ρAH\rho_{\mathrm{AH}} is normalized by ρ0\rho_0. The dependencies obtained are shown for the samples in their as-prepared state as well as after aging at 600 °C and 900 °C. We find that both as-grown and aged films exhibit the two-stage behavior of the ρAH\rho_{\mathrm{AH}} evolution with increasing HH. The occurrence of fast and slow stages in the ρAH\rho_{\mathrm{AH}}–versus–HH dependence can be associated with the FM and PM phases, respectively. It is striking that there is a significant enhancement of the PM susceptibility after aging at 900 °C by comparison that at 600 °C, as indicated by tilted thin lines. We see that, at sufficiently high H>20H>20 kOe, the quantity of ρAH\rho_{\mathrm{AH}} in the sample aged at 900 °C can even exceed that for the as-grown film, whose magnetization remains much higher.

LRO determination

To elucidate the relationship between the magnetism and LRO in our films, we determined the intensity of diffraction rings in the SAED patterns as functions of the aging time (tt) at the same aging temperatures, 600 °C and 900 °C. It is found that the increase in magnetization and related properties we observe is accompanied by a further increase in diffraction intensity from the B2 phase, which is present even in the as-grown films [26, 27]. Generally, chemical ordering in a magnetic alloy has to suppress the ferromagnetism because of reducing the number of nearest Fe neighbors in the atomic lattice [48].

Figure 3a shows the time evolution of the ratio between intensities of the strongest B2 (100) and fundamental (110) diffraction peaks, I100/I110I_{100}/I_{110}, the square root of which is proportional to LRO [42]. A representative SAED pattern [26, 27] of the aged samples, with indication of the 100 (highlighted in red) and 110 diffraction rings, is presented in Figure 3b. In greater detail, Figure 3 (c-Figure 3d) depicts the SAED intensity versus the scattering vector q=2sin⁡θ/λq=2\sin\theta/\lambda with θ\theta and λ\lambda being the electron diffraction angle and electron wavelength, respectively. As seen from Figure 3c, up to three B2 peaks – 100, 111, 210 – are detectable in the used SAED mode. Positions of the diffraction maxima are at qhkl=h2+l2+k2/aq_{hkl}=\sqrt{h^{2}+l^{2}+k^{2}}/a, where hh, kk, ll are Miller indices and aa is the interplane distance. For the B2 phase, the latter quantity is 0.287 nm, 0.291 nm and 0.288 nm in the as-grown sample, after aging at 600 °C, and after aging at 900 °C, respectively. We also note that, as shown in Figure 3d, a new diffraction ring is detectable at q=4.2q=4.2 nm−1^{-1} after RTA aging at 900 °C, which can be attributed to an oxide phase such as Fe2_2O3_3 or Al2_2O3_3. As there is no its effect on the magnetization (Table 1), one may suppose that this peak occurs from the oxide passivation layer, which forms on the sample surface under a heat treatment.

SAED diffraction patterns and intensity plots

Figure 3. (a) Time evolution of the quantity of I100/I110∝LRO2I_{100}/I_{110} \propto\mathrm{LRO}^{2}, where I100I_{100} and I110I_{110} are intensities of the strongest B2 (100) and fundamental (110) diffraction peaks. The dependences obtained by determining the intensities of diffraction rings in the SAED patterns. The horizontal dashed line (theoretical limit) is the calculated (I100/I110I_{100}/I_{110}) for the B2 structure [49]. (b) A representative SAED pattern, which indicates the strongest B2 (100) and fundamental (110) diffraction peaks. (c) SAED intensity versus scattering vector qq in the as-grown sample with x=0.56x=0.56 and after its aging at elevated TaT_{\mathrm{a}}, with indication up to three B2 diffraction peaks – 100, 111, and 210. (d) The SAED intensity in the logarithmic scale, with indication the strongest B2 diffraction peak (100) as well as the diffraction peak from an oxide layer on the sample surface.

From the plotting in Figure 3, we see that after aging at 900 °C the superstructure peak intensity becomes larger than that at 600°C, at which the magnetization remains still small after heat treatment (Figure 2a). Moreover, it is even can be close to its theoretical limit [49] indicated in Figure 3a (dashed horizontal line). Thus, we find that, in our Fex_xAl1−x_{1-x} sample aged at high TaT_{\mathrm{a}} (but still lower TcT_{\mathrm{c}}), there is an inconsistence between the evolutions of magnetization and LRO.

B2 nucleation: Experiment and modeling

To understand a reason(s) for the observed (1) inconsistency between the magnetism and LRO and (2) enhancement of magnetism at high-temperature aging, the film crystalline structure was examined with TEM, including the ESI mode. Figure 4a and Figure 4b depict TEM bright field overview micrographs of the sample with x=0.56x=0.56 after aging at 600 °C and subsequently at 900 °C. We note that the SAED pattern corresponding to the image for the sample after aging at 900 °C shows a high B2 diffraction intensity, as illustrated in Figure 3a. Both TEM overview micrographs indicate that the film microstructure is inhomogeneous to contain bright regions embedded in the crystal structure of the alloy. As TaT_{\mathrm{a}} increases from 600 °C to 900 °C, we observe enlargement of the bright regions up to submicron dimensions, as shown in Figure 4b.

TEM bright field overview micrographs and HRTEM micrograph of the sample with $x=0.56$

Figure 4. (a-b) TEM bright field overview micrographs of the sample with x=0.56x=0.56 aged at 600 °C (a) and 900 °C (b). The bright regions in the images consist of assemblies of B2 nuclei, which are highlighted with red contours. (c) HRTEM micrograph of the morphological structure, where a few B2 nuclei are identifiable. The inset is the Fast Fourier Transform of the HRTEM micrograph, which indicates the formation of the B2 structure through the appearance of the 100 reflection.

Examining the bright regions, like that seen in top of Figure 4b, has revealed that they contain the assemblies of well-shaped crystals with diameters of 10-20 nm (highlighted with red contours). Figure 4c shows a high-resolution TEM image and its Fast Fourier Transformation (FFT) in the inset, where a few such nanocrystals can be identified (indicated by arrows) through the effect of the periodic atomic arrangement in the alloy. The FFT we show allowed us to reveal the 100 reflection from the B2 structure in addition to the 110 fundamental one.

However, the question about identifying the ordering scenario (Figure 1d, e) still remains. To find an answer to that question, ESI maps were collected inside of a one of the bright regions seen in Figure 4a, which contain a higher concentration of B2 nuclei than the environment does. In Figure 5a, we show a TEM micrograph of a film location, which contains this region. Figure 5b and Figure 5c show the distributions of Fe and Al concentrations over the scanned location, respectively. The elemental maps were obtained near the Fe L and Al K absorption edges using a 10-eV spectrometer slit width during 20-minute exposure for each signal and background image. The one-window contrast evaluation method was used to estimate the background signal, which allowed for a qualitative evaluation of the chemical distribution of elements in the alloy.

TEM micrograph and ESI maps of Fe and Al concentration distributions

Figure 5. (a) Zero loss bright field TEM micrograph of a film location, which contains a zone with a high concentration of B2 nuclei (bright region in the center). (b-c) ESI maps of Fe and Al concentration distribution near the Fe L (b) and Al K (c) absorption edges, which are taken inside the bright region. The insets show the corresponding electron energy loss spectra averaged through the scanned area.

We have found that the Fe concentration within the bright zone is lower than that of the surrounding matrix (Figure 5b), while the Al concentration remains uniform throughout the film area (Figure 5c). Thus, we conclude that achieving a higher LRO corresponds to a higher concentration of B2 nuclei in the aged alloy, which is accompanied by squeezing out the excess Fe into the surrounding matrix.

Figure 6 shows the atomic structures in the Fe0.6_{0.6}Al0.4_{0.4} crystal, which were simulated with the MD method at 1100 °C with different aging times – 10 ns, 40 ns, and 60 ns. During the early stages of ordering occurring through thermal vacancies in the crystal lattice, the B2 nucleation was observed. For example, we see (1) the formation of a B2 nucleus in the right upper corner of the “40 ns” pattern and (2) segregation and clustering of Fe (blue tones), as pointed out by arrows. It is interesting to note that, in such a temperature regime, the ordering process is activated within the nanosecond timescale. Indeed, there is a pronounced difference in atomic arrangements of the “10 ns” and “40 ns” patterns, while the “40 ns” and “60 ns” patterns are already close each to other. Such an ordering rate is compatible with analytical simulations performed earlier [50] by combining the Landau theory for second-order phase transitions [51] and the Metiu, Kitahara, and Ross equation for LRO relaxation [52].

MD simulations of the B2 nucleation in the Fe0.6Al0.4 alloy aged for different times at 1100 °C

Figure 6. MD simulations of the B2 nucleation in the Fe0.6_{0.6}Al0.4_{0.4} alloy aged for different times at 1100 °C. Blue and yellow tones are Fe and Al atoms, respectively. The formation of the B2 nucleus, indicated by the dashed contour, occurs after aging for 40 ns. As a result of this process, excess Fe atoms are expelled to other locations, as indicated by arrows.

Discussion

Based on the FMR/MOKE, TEM/SAED/ESI, ρAH\rho_{\mathrm{AH}}-versus-HH data, and MD simulations we find that, after aging at 900 °C, the alloy appears to be compositionally nonuniform, consisting of two regions. One of them in this dual-phase hybrid is a nearly pure B2 phase. Indeed, the 100 diffraction ring intensity becomes close (Figure 3a) to that calculated for the B2 structure, I100/I110=0.19I_{100}/I_{110}=0.19 [49].

Another phase is Fe-enriched precipitates embedded in the B2 matrix. It is highly likely that the clustering of excess iron is responsible for the observed improvement in the magnetism (Figure 2a). By contrast, we observe mostly a decrease in magnetization and related properties at 600 °C, and so, segregation of Fe and its precipitation are not effective in such a heat treatment regime.

We should also note that the ordering scenario through nucleation and growth of the B2-Fe0.5_{0.5}Al0.5_{0.5} phase (Figure 1e) in the nonstoichiometric alloy is incompatible with the magnetic phase diagram adopted for Fex_xAl1−x_{1-x} (Figure 1c). Although this diagram is derived from experimental data, it can be interpreted as showing magnetic phases and their transitions in a single-phase B2-Fex_xAl1−x_{1-x} alloy (Figure 1d). However, Fe precipitates in the ordered alloy have to increase the critical temperatures for magnetic transitions, which are displayed in the current magnetic diagram to be not higher than a few tens of Kelvin.

Finally, we would like to emphasize that high-temperature aging (900 °C) causes the further ordering in the alloy with massive B2 formation, in which magnetization is very small (4πM<0.14\pi M<0.1 kG) [19, 22, 24]. Therefore, the explanation of the increase in ρAH\rho_{\mathrm{AH}} we observe (Figure 2b-Figure 2d) can be found by considering the contributions from superparamagnetic clusters [27], which form owing to squeezing out the excess iron by growing B2 nuclei. Sizes of these clusters can be roughly evaluated from the following relationship between magnetization MM and applied magnetic field HH in a thin PM film:

M=CJT+(4π/3)CJH(1)M = \frac{C_J}{T+(4\pi/3)C_J}H \tag*{(1)}

where TT is the ambient temperature, CJ=Nμ2/3kBC_J=N\mu^2/3k_B is the Curie constant, NN is the concentration of superparamagnetic clusters, μ\mu is the total magnetic moment of a cluster, and kBk_B is the Boltzmann constant. Equation (1) is valid if μH≪kBT\mu H\ll k_BT. It follows from this condition that the cluster can contain at most ∼10\sim10 magnetic atoms at TT close to room temperatures, at which ρAH\rho_{\mathrm{AH}} was measured. It here is interesting that, at sufficiently high H>20H>20 kOe, the contribution of the clusters into ρAH\rho_{\mathrm{AH}} can even exceed that in the as-prepared alloy, whose magnetization is higher.

Conclusion

In summary, we have investigated the kinetics of chemical ordering in nonstoichiometric Fex_xAl1−x_{1-x} (0.5<x<0.70.5<x<0.7) thin-film alloys and uncovered a non-monotonic relationship between the magnetism and aging temperature. While annealing at lower temperatures (Ta≤600T_a\leq600 °C) leads to the well-established decrease in magnetization due to continuous atomic ordering, high-temperature annealing (Ta=900T_a=900 °C) triggers an opposite trend, enhancing the magnetization and related properties such as the anomalous Hall resistivity.

The key to understanding this behavior lies in identifying a change in the fundamental ordering mechanism, which is the nucleation and growth of the stoichiometric ordered (B2) phase, a first-order transformation pathway. Crucially, this process forces the segregation of excess Fe into the surrounding matrix, as directly evidenced by our TEM/ESI analysis. The observed magnetization enhancement is a direct consequence of the formation of Fe-enriched clusters within the ordered B2 matrix. This qualitative model successfully explains the distinctive field evolution of the anomalous Hall resistivity. Our conclusions are further supported by molecular dynamics simulations of the Fe0.6_{0.6}Al0.4_{0.4} alloy.

Thus, beyond the previously documented continuous ordering, Fe-Al alloys can access a competing ordering pathway via phase separation. This work clarifies the complex interplay between structural order and magnetism in this prototypical system and provides a revised framework for controlling functional properties in ordering alloys. We finally note that after their high-temperature treatments such alloys demonstrate excellent magnetic and electronic properties even with no sintering oxygen-containing solid solutions [46].

Acknowledgements

The authors express their gratitude to Dr. M. Bohra (Mahindra University, Hyderabad, India) for fruitful discussions. This work was supported by Ministry of Science and Higher Education of the Russian Federation (State Contract FFUF-2024-0021). MD modeling was performed using the resources of the “Lobachevsky” computing cluster of the Lobachevsky State University.

Supplemental Information Raw Data: X-ray reflectivity (XRR) analysis, Ferromagnetic resonance (FMR) spectroscopy, and Magneto-optical Kerr effect (MOKE) magnetometry.

Supplemental Information: Original data

XRR curves for determining the composition

Figure S1. X-ray reflectivity (XRR) at the Cu Kα_{\alpha} edge (λ=0.154\lambda=0.154 nm) versus the double angle of grazing incidence for three as-prepared films of the Fex_xAl1−x_{1-x} alloy sputtered on Si(500 μ\mum)/SiO2_2(100 nm)/Si3_3N4_4(50 nm) substrates. The samples have different magnetizations measured by FMR and MOKE magnetometry. The composition, xx, was determined using two methods. One way was to measure the edge of total external reflection as the maximum of the angle derivative of XRR. Another way was modeling XRR curves and their fitting to the experimental ones shown in this plot using DIFFRAC.Leptos 7.04 (Bruker AXS) software [A.Ulyanenkov, LEPTOS: A universal software for X-ray reflectivity and diffraction, Advances in Computational Methods for X-Ray and Neutron Optics, 5536 (2004) 1-15].

Fitting of interference fringes in XRR curves allowed for determining precisely the atomic density of a sample, ρs\rho_s. The composition value was found as x=(ρs−ρAl)/(ρFe−ρAl)x = (\rho_s - \rho_{\mathrm{Al}})/(\rho_{\mathrm{Fe}} - \rho_{\mathrm{Al}}), where ρAl=2.7\rho_{\mathrm{Al}}=2.7 g/cm3^3 and ρFe=7.9\rho_{\mathrm{Fe}}=7.9 g/cm3^3 are the atomic densities of Al and Fe bulks. The results of the measurements are summarized are summarized as follows:

Table S1: 2θc2\theta_c is the double total external reflection edge and ρs\rho_s are measured atomic densities of the surface oxide, Fe-Al films, and SiO2\mathrm{SiO_2} substrate. The thicknesses and surface roughness determined by fitting the modeled layered system to the experimental XRR curves are given as well.

  • #567 (2θc=0.64∘2\theta_c=0.64^\circ)

layerThickness, nmRoughness, nmρs\rho_s, g/cm3^3
oxide5.741.952.73
Fe-Al56.22.415.58
Substrate SiO2\mathrm{SiO_2}1.042.33

Table S1-567.

  • #566 (2θc=0.66∘2\theta_c=0.66^\circ)

layerThickness, nmRoughness, nmρs\rho_s, g/cm3^3
oxide7.24.75.48
Fe-Al43.61.55.97
Substrate SiO2\mathrm{SiO_2}1.222.33

Table S1-566.

  • 568 (2θc=0.67∘2\theta_c=0.67^\circ)

layerThickness, nmRoughness, nmρs\rho_s, g/cm3^3
oxide5.52.03.0
Fe-Al53.62.26.19
Substrate SiO2\mathrm{SiO_2}1.12.33

Table S1-568.

FMR spectroscopy

  • 56 at. % Fe (#567)

FMR spectra for the as-grown sample and after aging at 600 °C and 900 °C

Figure S2. FMR spectra obtained at room temperature for the as-grown sample with x=0.56x=0.56 and after its aging at elevated TaT_a. The spectra are shown for two geometries of an applied static field HH: (1) the field is parallel to the film plane (90∘90^\circ) and (2) the field is normal to the film plane (0∘0^\circ). The reentrance of the FMR response is observable after high-temperature aging at 900 °C (indicated by arrows).

• 65 at. % Fe (#566)

Figure S3. FMR spectra obtained at room temperature for the as-prepared, and aged sample with a composition of x=0.65x=0.65. The spectra are shown for two geometries of an applied static field HH: (1) the field is parallel to the film plane (noted as “90”) and (2) the field is normal to the field plane (noted as “0”). The as-prepared film is not uniform across its thickness.

• 68at. % Fe (#568)

Figure S4. FMR spectra obtained at room temperature for the as-prepared, and aged sample with a composition of x=0.68x=0.68. The spectra are shown for two geometries of an applied static field HH: (1) the field is parallel (HpH_p) to the film plane and (2) the field is normal (HnH_n) to the field plane. The resonance fields are given in units of kOe and kG, respectively.

MOKE magnetometry

  • 68 at. % Fe (#568)

Figure S5. Longitudinal/polar MOKE magnetometry of the as-prepared, aged, and quenched samples with 68, 65, and 56 atomic percent of Fe. The MOKE intensity is indicated in the plots.

  • 65 at. % Fe (#566)

Figure 1 (PDF p. 18).

  • 56 at. % Fe (#567)

Figure 7 (PDF p. 19).

References

  1. [1]Elphick K, Frost W, Samiepour M, T. Kubota T, Takanashi K, Hiroaki S, Mitani S, Hirohata A (2021) Heusler alloys for spintronic devices: review on recent development and future perspectives. Sci Technol Adv Mater 22 (1): 235–271.DOI
  2. [2]Manna K, Sun Y, Muechler L, Küebler J, Felser C (2018) Heusler, Weyl and Berry. Nat Rev Mater 3: 244–256.DOI
  3. [3]Arrott A, Sato H (1959) Transitions from ferromagnetism to antiferromagnetism in Iron-Aluminum alloys. Experimental Results. Phys Rev 114 (6): 1420-1426.DOI
  4. [4]Shull RD, Okamoto H, Beck PA (1976) Transition from ferromagnetism to micromagnetism in Fe-Al alloys. Solid State Com 20: 863-868.
  5. [5]Sakai A, Minami S., Koretsune T, Chen T, Higo T, Wang Y, Nomoto T, Hirayama M, Miwa S, Nishio-Hamane D, Ishii F, Arita R, Nakatsuji S (2020) Iron-based binary ferromagnets for transverse thermoelectric conversion. Nature 581: 53–57.DOI
  6. [6]Zhou W, Sakuraba Y (2020) Heat flux sensing by anomalous Nernst effect in Fe–Al thin films on a flexible substrate. Appl Phys Express 13 (4): 043001.
  7. [7]Chen YJ, Basiaga D, O’Brien JR, Heiman D (2004) Anomalous magnetic properties and Hall effect in ferromagnetic Co2MnAl epilayers. Appl Phys Lett 84 (21): 4301–4303.DOI
  8. [8]Lv B, Liu P, Wang Y, Cao C, Si M (2022) A large anomalous Hall conductivity induced by Weyl nodal lines in Fe70Al30. Appl Phys Lett 121: 072405.DOI
  9. [9]Vilanova Vidal E, Stryganyuk G, Schneider H, Felser C, Jakob G (2021)Exploring Co2MnAl Heusler compound for anomalous Hall effect sensors. Appl Phys Lett 99: 132509.
  10. [10]Nakatani T, Kulkarni PD, Suto H, Masuda K, Iwasaki H, Sakuraba Y (2024) Perspective on nanoscale magnetic sensors using giant anomalous Hall effect in topological magnetic materials for read head application in magnetic recording. Appl Phys Lett 124(7): 070501.
  11. [11]Li P, Koo J, Ning W, Li J, Miao L, Min L, Zhu Y, Wang Y, Alem N, Liu C-X, Mao Z, Yan B (2020) Giant room temperature anomalous Hall effect and tunable topology in a ferromagnetic topological semimetal Co2MnAl. Nat Commun 11: 3476.DOI
  12. [12]Gupta R, Husain S, Kumar A, Brucas R, Rydberg A, Svedlindh P (2021) Co2FeAl full Heusler compound based spintronic terahertz emitter. Adv Optical Mater 9: 2001987.DOI
  13. [13]Bierhance G, Markou A, Gueckstock O, Rouzegar R, Behovits Y, Chekhova AL, Wolf M, Seifert T, Felser C, Kampfrath T (2022) Spin-voltage-driven efficient terahertz spin currents from the magnetic Weyl semimetals Co2MnGa and Co2MnAl. Appl Phys Lett 120: 082401DOI
  14. [14]Sharma V, Chaudhary V, Ramanujan RV (2018) Near-room-temperature magnetocaloric properties in Fe75−xMnxAl25 alloys. IEEE Trans Magn 54: 2501105.DOI
  15. [15]M. Bohra, S. Pardeshi, A. Annadi, Singh V, López-Martín R, Toulkeridou E, Grammatikopoulos P (2025) Comprehensive modeling of magnetocaloric effects in Fe-Al alloys across magnetic transitions. Appl Phys A 131: 563.DOI
  16. [16]Ravisankar NK, Rajasabai SP (2023) Large magnetocaloric effect in DO3-type Fe3Al alloys. Physica B: Condens Matter 651: 414585.DOI
  17. [17]Vorobjova AI, Tishkevich DI, Outkina EA, Yao Y, Razanau IU, Zubar TI, Rotkovich AA, Bondaruk AA, Sayyed MI, Trukhanov SV, Kubasov IV, Fedosyuk VM, Trukhanov AV (2024) Fabrication of composite nanostructures for impedance biosensors using anodic aluminum oxide templates and carbon nanotubes.Ceram Int 50: 45703-45712.DOI
  18. [18]Bradley AJ, Jay AH (1932) The formation of superlattices in alloys of iron and aluminium. Proc Royal Soc A (London) 136 (829): 210-232.DOI
  19. [19]Kuentzler R (1983) Ordering effects in the Fe-Al system. J Phys 44: 1167-1178.DOI
  20. [20]Hernando A, Amils X, Nogues J, Surinach S, Baro MD Ibarra MR, (1998) Influence of magnetization on the reordering of nanostructured ball-milled Fe-40 at. % Al powders. Phys Rev B 58:R11864.DOI
  21. [21]Kudryavtsev YV, Nemoshkalenko VV, Lee YP, Kim KW (1997) Magneto-optical and optical properties of ordered and disordered Fe–Al alloy films. J Appl Phys, 82 (10): 5043–5049.DOI
  22. [22]Yao X, Wiedwald U, Trautwetter M, Ziemann P (2014) Growth modes and epitaxy of FeAl thin films on a-cut sapphire prepared by pulsed laser and ion beam assisted deposition. J Appl Phys 115: 023507.DOI
  23. [23]M. O. Liedke, W. Anwand, R. Bali, Cornelius S, Butterling M, Trinh TT, Wagner A, Salamon S, Walecki D, Smekhova A, Wende H, Potzger K (2015) Open volume defects and magnetic phase transition in Fe60Al40 transition metal aluminide, J Appl Phys 117: 163908.DOI
  24. [24]Ehrler J, Sanyal B, Grenzer J, Zhou S, Böttger R, Eggert B, Wende H, Lindner J, Fassbender J, Christoph Leyens C (2020) Magneto-structural correlations in a systematically disordered B2 lattice. New J Phys 22: 073004.DOI
  25. [25]Toyoki K, Kitaguchi D, Shiratsuchi Y, Nakatani R (2024) Influence of long- and short-range chemical order on spontaneous magnetization in single-crystalline Fe₀.₆Al₀.₄ compound thin films. J. Phys: Condens Matter 36:135805.DOI
  26. [26]Pashenkin IY, Tatarsky DA, Churin SA, Nechay AN, Drozdov MN, Sapozhnikov MV, Polushkin NI (2024) Magneto-optical control of ordering kinetics and vacancy behavior in Fe–Al thin films quenched by laser, Adv Eng Mater 26 (24): 2401044.arxiv.org/abs/2407.16423
  27. [27]Pashenkin IY, Gorev RV, Kuznetsov MA, Tatarsky DA, Churin SA, Yunin PA, Drozdov MN, Krivulin DO, Sapozhnikov MV, Demidov ES, Skuratov VA, Kudyukov EV, Kurylandskaya GV, Fraerman AA, Polushkin NI (2024) Magnetic and magnetocaloric modifications near room temperature in Fe₀.₆Al₀.₄ nanoalloys under irradiation by swift heavy ions. J Phys Chem C, 128(21): 8853–8860.DOI
  28. [28]Andrade LHF, Campos LMB, Pedrosa TA, Krohling A, Ardisson JD, Fernandez-Outon LE (2025) Magnetic properties and structural characterization of CoₓFe₍₁₋ₓ₎Al Heusler alloys synthesized from Co/Fe/Al multilayers. AIP Advances 15:035105.DOI
  29. [29]Kurdnovský J, Drchal V, Máca F, Turek I., Khemelevsky S (2020) Large anomalous Hall angle in the Fe₆₀Al₄₀ alloy induced by substitutional atomic disorder. Phys Rev B 101: 054437.DOI
  30. [30]Stein F, Palm M (2007) Re-determination of transition temperatures in the Fe–Al system by differential thermal analysis. Int Mat Res (formerly Z Metallkd) 98(7): 580-588.DOI
  31. [31]Lifshitz EM (1942) On the theory of second-order phase transitions. II. Second-order phase transitions in alloys. J Phys USSR 7(61): 251; De Fontaine D. (1975) k-Space symmetry rules for order-disorder reactions. Acta Met 23(5): 553-571.
  32. [32]Allen SM, Cahn JW (1976) Mechanisms of phase transformations within the miscibility gap in Fe-rich Fe-Al alloys. Acta Met 24: 426-437.
  33. [33]Polak M, Rubinovich L (2000) The interplay of surface segregation and atomic order in alloys. Surf Sci Rep 38: 127-194.DOI
  34. [34]Cerháti C, Bakker H, Beke DL (1993) Kinetics of surface segregation in alloys. Surf Sci 290: 345-361.
  35. [35]Tucker V, Wen D, Cooley V, Park J-S, Titus MS (2025) Theoretical and experimental quantification of Suzuki segregation enthalpy and strengthening mechanisms in a binary alloy. Communications Materials 6:142.DOI
  36. [36]Sokolovskiy VV, Gruner ME, Entel P, Acet M, Çakır A, Baigutlin DR, Buchelnikov VD (2019) Segregation tendency of Heusler alloys. Phys Rev Materials 3: 084413.DOI
  37. [37]Jiang S, Wang H, Wu Y, Liu XJ, Chen HH, Yao ML, Gault B, Pomge D, Raabe D, Hirata A, Chen MW, Wang YD, Lu ZP (2017) Ultrastrong steel via minimal lattice misfit and high density nanoprecipitation. Nature 544: 460–465.DOI
  38. [38]Zhou Y, Zhou D, Jin X, Zhang L, Du X, Li B (2018) Design of non-equiatomic medium-entropy alloys. Sci Rep 8: 1236.DOI
  39. [39]Wang Z, Wang Q, Niu B, Dong C, Zhang Hongwei, Zhang Haifeng, Liaw PK (2021) Coherent precipitation and stability of cuboidal B2 nanoparticles in a ferritic Fe–Cr–Ni–Al superalloy. Mater Res Lett 9 (11): 458-466.
  40. [40]Kannan R, Pierce D, Lee Y, Poplawsky J, Unocic K, Fancher C, Hoffmann W, Lienert TJ, Nandwana P (2024) Controlling microstructure and B2 ordering kinetics in Fe–Al system through additive manufacturing. J Mater Res Technol 33: 1692–1703.DOI
  41. [41]Almessiere MA, Algarou NA, Slimani YKlylgach DS, Baykal A, Korkmaz AD, Zubar TI, Silibin MV, Vakhitov MG, UI-Hamid A, Trukhanov SV, Trukhanov AV 2024 Structure, magnetic and electrodyamic properties of SrInxFe12−xO4/Ni0.5Zn0.5Fe2O4 composites at low temperature. J. Amer Ceram Soc 107: 4936-4948.
  42. [42]Cullity BD, Elements of X-ray Diffraction ((Addison-Wesley, Reading, MA, USA, 1956).
  43. [43]Hirel P (2015) Atomsk: A tool for manipulating and converting atomic data files. Comput Phys Commun 197:212-219.
  44. [44]Mendelev MI, Srolovitz DJ, Ackland GJ, Han S (2011) Effect of Fe segregation on the migration of a non-symmetric Σ5 tilt grain boundary in Al. J Mater Res 20: 208-218.
  45. [45]Stukowski A (2010) Visualization and analysis of atomistic simulation data with OVITO–the open visualization tool. Modelling Simul Mater Sci Eng 18:015012.DOI
  46. [46]Trukhanov AV, Turchenko VO, Bobrikov IA, Trukhanov SV, Kazakevich IS, Balagurov AM (2015) Crystal structure and magnetic properties of the BaFe12−xAlxO19 (x=0.1-1.2) solid solutions, J Magn Magn Mater 393: 253-259.
  47. [47]Akhsakh al yan AD, Fraerman AA, Polushkin NI, Platonov YY, Salashchenko NN (1991) Determination of layered synthetic microstructure parameters. Thin Solid Films 203(2): 317-326.DOI
  48. [48]Beck PA (1971) Some recent results on magnetism in alloys. Metall Trans 2: 2015–2024.DOI
  49. [49]https://www.icdd.com/; pdf entry #01-078-3470 (uploaded Mar 28, 2024).
  50. [50]Polushkin NI (2021) Chemical order relaxation in a substitutional solid alloy around the critical temperature. Phys Rev B 103(10): 104207.DOI
  51. [51]Landau LD, Lifshitz EM, Statistical Physics, 3rd edn. Part 1 (Elsevier 2013), pp. 446-516 (Ch. XIV).
  52. [52]Metiu H, Kitahara K, Ross J (1976) Stochastic theory of the kinetics of phase transitions, Chem Phys 64: 292-299.DOI

Paper details

vlnc:W_jz6y84ft
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents