Magnetic transition in the Heusler compounds Fe_3-xMn_xSi

Fe_3-xMn_xSi is a Heusler compound with L2₁ structure. Heusler compounds have attracted significant attention owing to their potential to be used as new functional materials. In particular, Heusler compounds that have properties expected of half-metals have been discovered in recent studies;^1,2 Fe_3-xMn_xSi was theoretically predicted to be a half-metal for a wide range of x.^3–5 Magnetic properties of Fe_3-xMn_xSi have been studied for a wide x range. It is well-known that Fe₃Si is ferromagnetic. The ferromagnetic phase below Curie temperature T_C is defined as the F phase in this paper. Although T_C decreases with increasing x, for x > 0.75, an antiferromagnetic phase with a ferromagnetic component appears at T_A below T_C, which has been confirmed by neutron diffraction studies.^6,7 This implies that the ground state is not half-metallic. The phase below T_A is defined here as the AF phase, and the transition is defined as the AF transition. With increasing x, T_C continues to decrease and approaches T_A until x ∼1.7.⁷ Previously, for x ≥ 1.65, a meta-magnetic-like transition was observed in the magnetization vs. magnetic field, M(B), at temperatures lower than T_A.^7–9 In our recent study of the temperature dependence of magnetization M(T) for x = 1.7, we discovered another transition at a temperature lower than T_A. This transition is characterized by a rapid decrease in M with decreasing T, and is regarded as another antiferromagnetic transition; the phase and the transition temperature will be referred to as the AF2 phase and T_A2, respectively.¹⁰ It was shown that this AF2-AF transition corresponds to the meta-magnetic-like transition. We observed hysteresis in M(T) between for a zero-field cooling process (ZFC) and a field cooling process (FC) at high fields below T_A2. It was found that T_A2 depends on magnetic field, in contrast to T_A, which shows a weak dependence on B.¹¹ The AF2 phase abruptly appears around x∼1.7 when varying x.¹⁰ 

The relationship of these transitions and detailed information regarding the phases is still unknown. Although we believe that the AF2 phase is a typical antiferromagnetic phase without ferromagnetic components, we do not understand anomalous hysteresis related to this phase at rather high fields. Therefore, we measured the magnetization of Fe_3-xMn_xSi for 1.75 ≤ x ≤ 1.85 and attempted to discover the properties of these phases and transitions.

Polycrystalline samples of Fe_3-xMn_xSi with 1.65 ≤ x ≤ 1.85 were prepared by arc-melting high-purity constituent elements in a high-purity argon atmosphere. The buttons of the samples were flipped and re-melted several times to insure homogeneity. Fe_1.3Mn_1.7Si was annealed at 800 °C for 2 days. Fe_1.2Mn_1.8Si were annealed at 600 °C for 7 days. We found that annealing improved the sharpness of the transitions but did not influence transition temperatures. The crystal structures were examined using X-ray powder diffraction measurements with Cu Kα radiation. The results showed that the prepared samples were single-phase, and the crystal structure was L2₁. Magnetization was measured using superconducting quantum interference device (SQUID) magnetometers (MPMS: Quantum Design) with magnetic field range of 0–7 T and temperature range of 2–300 K.

Figure 1(a) shows M(T) for x = 1.65 for different values of the magnetic field in the range 0 < B ≤ 3 T. The behavior observed is the same as that for x ≤ 1.6,^10,12,13 which implies that the transitions at T_C = 105 K and T_A = 68 K were present, and the transition at T_A2 did not occur. The M(T) curve for x = 1.65 shows the noted hysteresis below T_A between ZFC and FC at a low field of 0.01 T, as was already known. However, even at 1 T, hysteresis between the curves for ZFC and FC is observed. The origin is presumed to be different from that at 0.01 T. This resembles the behavior of M(T) for x = 1.7 for B ≥2 T below T_A2. This may suggest that the AF2-AF transition is already present for x = 1.65.

Figures 1(b)–(d) show (b) M(T) for x = 1.75, (c) M(T) for x = 1.8, and (d) the details of x =1.8 at low fields. Magnetization decreases with increasing x. The M(T) curves for x = 1.75, 1.8, and 1.85 are similar to each other, and all curves appear to have the AF transition, the temperature of which is tentatively considered to be T_A. A rather rapid decrease in M(T) with decreasing T from a temperature below T_A is observed at rather high fields above ∼2 T, which shows the AF2-AF transition (T_A2). Even though for x =1.65 in B = 0.01 T, a rapid change of M(T) indicating the ferromagnetic transition was observed at T_C, these curves does not show such a change even at low fields. Considering together this fact and the Arrott plot as described later, we can conclude that for 1.75 ≤ x ≤ 1.85 of Fe_3-xMn_xSi, the ferromagnetic transition does not occur. For 1.75 ≤ x ≤ 1.85, a transition, which might be called T_A2, is not recognized at a temperature lower than T_A for B < ∼2 T. Although T_A does not change with B toward high fields for all x, T_A2 is dependent on B. The magnetization M(T) for x = 1.8 shows hysteresis below T_A2 between ZFC and FC at 5 T, whereas at a lower field of 3 T, it disappears. This behavior is similar to the hysteresis in M(T) observed below T_A2 for x = 1.7, and the hysteresis observed for x = 1.65 at B = 1 T is reminiscent of this.¹⁰ 

Figure 2(a) shows the results of M(B) for x = 1.8 measured at different temperatures. A meta-magnetic-like transitions is observed at T ≤ 55 K, and hysteresis at the transition between increasing and decreasing B is observed at T ≤ 30 K. At low temperatures in the low field region where the AF2 phase exists, M increases with linearly with B, as in a typical antiferromagnet. This basically agrees with the results of the previous study for x=1.7.¹⁰ However, in Fig. 2(a), it is difficult for us to recognize that the ferromagnetic phase is present as for x = 1.7. We therefore created Arrott plots, M² vs. B/M, as shown in Figs. 2(b) for x = 1.7 and (c) for x = 1.8. Consequently, it can be seen that spontaneous magnetization exists for x =1.7, because there are positive M²-intercepts, whereas it does not exist for x =1.8. The Arrott plots show the absence of spontaneous magnetization for 1.75 ≤ x ≤ 1.85. Considering this result together with the M(T) data, we can conclude that for x ≥ 1.75, the ferromagnetic transition does not exist at low fields, and spontaneous magnetization, which the AF phase should show, is not present for x ≥ 1.75.

The B-T magnetic phase diagram relating to the magnetic transitions given by the above results and the previous study is shown in Fig. 3. The transition temperature T_C is 85 K for x = 1.7 but disappears for x ≥ 1.75. The transition temperature T_A is almost independent of x and B, and therefore, it is shown by a single broken line to avoid confusion in Fig. 3. On the contrary, T_A2 has both x and B dependences. The transition fields for T_A2 (the AF2-AF transition) shift to large B with increasing x, whereas the AF2-AF transition lines approach the T_A line at lower fields irrespective of x for x ≥ 1.75. The AF2-AF transition lines appear to merge with the T_A line at lower fields. Below ∼2 T, the existence of the AF2-AF transition is unclear. Considering these results together, we can conclude that for x ≥ 1.75 in near zero fields, the F phase disappears and the transition from the paramagnetic phase directly to the AF2 phase occurs at T_A, even though this T_A does not necessarily have the same meaning originally defined. Above ∼2 T, the AF2-AF transition appears, and the transition line shifts to higher fields with increasing x. In addition, hysteresis, which is characteristic of the AF2-AF transition at low T and high B, was also observed, even though it becomes less remarkable with increasing x. However, these transitions in near zero field are close, and distinction is subtle. To obtain a firm conclusion, more studies are needed.

In Heusler compounds, Mn atoms often induce antiferromagnetism. For instance, it was shown that surplus Mn atoms are antiferromagnetically coupled in Mn-doped Ni₂MnGa, which is known as a ferromagnetic shape-memory Heusler compound.¹⁴ Phenomenologically also in Fe_3-xMn_xSi extra Mn atoms are considered to decrease ferromagnetic interactions and to induce the AF phase. Furthermore, the existence of the AF2 phase suggests that for x ∼ 1.7 where ferromagnetic and antiferromagnetic interactions compete and ferromagnetism disappears, a complex magnetic phase diagram in magnetic field might arise, as was reported in doped MnSi.¹⁵ 

Information about the F, AF, and AF2 phases and their transitions for Fe_3-xMn_xSi with 1.65 ≤ x ≤ 1.85 was obtained from magnetic measurements. For x = 1.65, transitions at T_C and T_A were observed as is known for x ≤ 1.6. For 1.75 ≤ x ≤ 1.85 a rather rapid decrease in M(T) with decreasing temperature was found at a temperature lower than T_A, which indicates an AF2-AF transition at T_A2. The magnetization M(B) showed the meta-magnetic-like transition with hysteresis, which is found to correspond to the transition at T_A2 observed in M(T). These results agree with those of the AF2-AF transition for x = 1.7. The AF2-AF transition line in the B-T phase diagram shifts toward high field with increasing x, whereas the transition temperature at low fields does not depend strongly on x.

As opposed to the case for x = 1.7, we conclude from the M(T) and M(B) results that the ferromagnetic transition and spontaneous magnetization disappear for x ≥ 1.75. Consequently, at B lower than ∼2 T, a transition from the paramagnetic phase directly to the AF2 phase occurs. The AF2-AF transition line in the B-T phase diagram seems to terminate at the AF transition line below B = 2 T. At higher fields a transition from the paramagnetic to the AF phase and, subsequently, to the AF2 phase occurs.

We thank G. Adachi, for cooperation. This study was carried out as a joint research in the Institute for Solid State Physics, the University of Tokyo (ISSP). We are grateful to T. Yamauchi at the Materials Design and Characterization Laboratory, ISSP, for advice and help. Authors acknowledge support from JSPS KAKENHI Grant Number JP17K06774.