The growth of MnSi films have been well described in previous reports with various methods.2,9,21-25 However, most techniques to grow MnSi required specific facilities with an ultrahigh vacuum environment, while development for conventional magnetron sputtering with relatively low base pressure is not introduced yet. Since the lattice mismatch between Si (001) substrate and cubic MnSi structure is estimated to be around 19%, we have tested to find optimal growth conditions of the MnSi films on Si (001) substrates. Co-sputtering method with Mn and Si targets were employed, and growth conditions such as RF power, growth temperature, and annealing treatments were minutely controlled to grow the MnSi films (Table S1 of the supplementary information). Aguf et al. reported that as-deposited MnSi films were amorphous unless they were crystallized by annealing treatment.23 Indeed, we found that initially deposited amorphous MnSi turned into crystallized MnSi phase after annealing treatment (Fig. S1 of the supplementary information). Most results using Si (001) substrate, however, showed that mixed phases of MnSi and Mn5Si3 were observed by XRD measurements. For this reason, Si (001) substrates were replaced by Al2O3 substrates having a low lattice mismatch (~ 4.2%).
Fig. 1 presents the XRD patterns of the MnSi films grown on Si (black solid line) and Al2O3 (blue and red solid lines) substrates, where the MnSi films on Si (001) and on Al2O3 #1 were deposited with same growth conditions (15 W for Mn power, 100 W for Si, 590 °C annealing treatment). Note that the substrate peaks were not displayed for all samples because grazing incident X-ray diffraction technique was used. The asterisk in the Figure indicates Mn5Si3 (ICSD card no. 04-003-4114) phase. For the MnSi film on Si (001), MnSi peaks were mainly observed, in addition five peaks matched with Mn5Si3 phase and several unknown impurity peaks were detected. However, we found out that the peaks related with Mn5Si3 phase were suppressed and the unknown peaks disappeared for the MnSi on Al2O3 #1. Furthermore, MnSi on Al2O3 #2 sample, for which Mn power and annealing temperature decreased to 10 W and 550 °C, respectively, showed only MnSi (ICSD card no. 04-004-7568) peaks.
Although as-grown MnSi on Al2O3 #2 showed somewhat defective surface, highly uniform and low uneven surface was observed, as shown in SEM image of Fig. 2a and AFM topographic image of Fig. 2b. In the 15×15 μm scale of AFM image, the root mean squared (RMS) roughness was measured to be under 1 nm. To characterize detailed structure and chemical composition, the cross-sectional TEM analyses of as-grown MnSi on Al2O3 #2 were carried out. Fig. 2c shows the representative cross-sectional TEM image of MnSi on Al2O3 #2 at the interfacial region. Note that no stacking faults nor significant defects were observed. When one grows MnSi films grown by conventional sputtering in a relatively low vacuum chamber, it is hard to expect that MnSi grows epitaxially to the preferred direction of the surface of substrates, considering the structural parameters such as lattice mismatch and chemical bonding. Our MnSi films grown on Al2O3 have polycrystalline nature, confirmed by XRD patterns (Fig. 1) and fast Fourier transform (FFT) of TEM image [inset of Fig. 2c]. We have examined chemical composition of as-grown MnSi films. As seen in TEM-EDS mapping of Fig. 2d, the presence of only Mn and Si elements was detected at several different regions, and the atomic ratio of Mn : Si = 1 : 1.1 is estimated. We tested the growth rate of MnSi films by controlling growth time. The thickness of as-grown MnSi films showed a linear behavior for the growth time (Fig. S2 of the supplementary information).
Fig. 3a shows the temperature dependence of magnetization for MnSi on Al2O3 (thickness; 25 nm) measured in an out-of-plane magnetic field of 1 kOe. The magnetization dropped significantly at temperatures above 25 K, indicating the ferromagnetic transition temperature (TC), similar to the bulk MnSi.26,27 The resistivity depending on the temperature exhibited metallic behavior above TC, as shown in Fig. 3b. Below TC, the resistivity tended to decrease with T2 dependence as decreasing temperatures, owing to the coupling of charge carriers to spin fluctuations in helimagnetic phase.28 As seen in the inset of Fig. 3b, the derivative of resistivity versus temperature highlighted TC of MnSi films, around 25 K. The polycrystal and defects on the surface give rise to the low residual resistivity ratio, [ρ(300 K) / ρ(5 K)] ~ 1.7.
Fig. 3c shows the magnetoresistance for the magnetic fields perpendicular to the film plane at different temperatures of 2 K, 25 K, and 50 K. As we discussed above, since as-grown MnSi films had polycrystalline nature, the magnetic phase transition from the magnetoresistance was not clearly observed. In low magnetic fields, however, the temperature dependence of the magnetoresistance exhibited distinguishable features. As the temperature increased, the shape of the magnetoresistance in the vicinity of zero magnetic field changed from flat (2 K) to sharp (25 K) and broad (50 K) peaks.
Spin-chirality-driven Hall effect, THE can be induced by DMI arising from strong spin-orbit coupling and non-centrosymmetric B20 crystal structure,29 considered as a hallmark for the existence of skyrmion phase. We have performed Hall resistivity measurement to observe abnormal resistivity related with THE. The total Hall resistivity can be expressed as a combination of three components:
where ρnormal, ρAHE, and ρTHE are the normal, anomalous, and topological Hall resistivities, respectively. R0 is the normal Hall coefficient, and α, β, and b are the constants corresponding to the skew scattering, side jump, and intrinsic contributions to the anomalous Hall resistivity. Also, nSkx is the relative skyrmion density, P is the polarization of the conduction electrons, RTH is the topological Hall coefficient, and Beff is the effective magnetic field derived from the real-space Berry phase.20,30 Topological Hall contribution can be extracted by subtracting the normal and anomalous Hall resistivity terms from the measured total Hall resistivity.
Fig. 4a shows deconvoluted Hall data to extract the THE signal at 10 K as the blue curve, including normal (green line) and anomalous (red curve) Hall resistivities. Note that the positive slope of ρnormal indicates p-type majority carriers, and ρAHE is negative, consistent with those of MnSi bulk,31 thin film,9 and nanowire.20 The ρnormal is obtained from the linear fit at high magnetic fields, and ρAHE is directly taken from the magnetization data. The ρTHE depending on the temperature is displayed in Fig. 4b. Interestingly, the sign of ρTHE flipped at the border of 25 K, where the magnetic transition was expected. The sign of ρTHE is very sensitive to the spin polarization of charge carriers. In the band structure of MnSi, the localized electrons in d band affect the density of states near Fermi level, while itinerant electrons in s band are contributed meagerly in band structure,31 allowing the spin polarization to be delicate. In addition, since the spin polarization can be changed by external factors such as tensile strain and crystal purity with temperature,9 the flipped sign of ρTHE in our polycrystalline MnSi sample is reasonable. Fig. 4c presents the contour mapping of ρTHE as a function of magnetic field and temperature. While skyrmion phase in bulk MnSi was observed in narrow temperature range close to the magnetic transition temperature, non-zero ρTHE was collected from 2 K to 40 K regardless the sign. The absolute value of ρTHE had a maximum of 36 nΩ.cm at 10 K and 4 kOe, larger than thin films grown by MBE (10 nΩ.cm),9 bulk (4.5 nΩ.cm),32 and nanowire (15 nΩ.cm)20 but similar to the thin films grown by off-axis magnetron sputtering with ultrahigh vacuum chamber.25
ρAHE consists of three components; skew scattering, side jump, and intrinsic contribution. Implication in the scaling of anomalous Hall contribution is that ρAHE is proportional to the intrinsic contribution, , associated with momentum-space Berry phase.33 In Fig. 4d, we plot the ρAHE against at 20 kOe, showing obvious deviation from linear dependence. The breakdown of the scaling suggests that anomalous Hall effect is relevant to extrinsic skew scattering and side jump contributions caused by impurities and defects in our polycrystalline MnSi sample, remaining the stabilization of the skyrmion phase in a broader range of temperatures and magnetic fields.