Engineering helimagnetism in MnSi thin films

Magnetic skyrmion materials have the great advantage of a robust topological magnetic structure, which makes them stable against the superparamagnetic ef-fect and therefore a candidate for the next-generation of spintronic memory de-vices. Bulk MnSi, with an ordering temperature of 29.5 K, is a typical skyrmion system with a propagation vector periodicity of ∼ 18 nm. One crucial prerequi-site for any kind of application, however, is the observation and precise control of skyrmions in thin films at room-temperature. Strain in epitaxial MnSi thin films is known to raise the transition temperature to 43 K. Here we show, using magnetometry and x-ray spectroscopy, that the transition temperature can be raised further through proximity coupling to a ferromagnetic layer. Similarly, the external field required to stabilize the helimagnetic phase is lowered. Transmission electron microscopy with element-sensitive detection is used to explore the structural origin of ferromagnetism in these Mn-doped substrates. Our work suggests that an artificial pinning layer, not limited to the MnSi / Si system, may enable room temperature, zero-field skyrmion thin-film systems, thereby opening the door to device applications. C 2016 Author(s). All article content, except where other-wise noted, is licensed under a Creative Commons


I. INTRODUCTION
Magnetic skyrmions 1 are topologically stable, vortex-like magnetization states that form periodic, six-fold symmetric lattices (with periodicities of approximately 3-30 nm). 2,3 Since each skyrmion can be seen as an ultra-stable carrier of information, the crystal itself can be regarded as a high-density, non-volatile information matrix, overcoming the limitations set by the superparamagnetic limit. 4 Owing to the ultra-low current densities needed to manipulate them, 5 skyrmions have become the focus of novel magnetic memory and logic concepts. 6,7 The broken inversion symmetry of the spin-orbit coupling is the key condition for the appearance of the Dzyaloshinskii-Moriya interaction (DMI), which is the essential ingredient for the existence of skyrmions. Currently, there are two major types of materials that support magnetic skyrmions. First, skyrmions are found in non-centrosymmetric crystals which are governed by the DMI such as the cubic B20 helimagnetic transition metal silicides and germanides, MnSi, 2 FeGe, 8 and MnGe, 9 as well as new space group P4 1 32 materials. 10 Second, skyrmions are observed in thin films 4,11 or heterostructures in which the DMI is induced by the specific structural boundary conditions. [12][13][14][15] Though the structural DMI systems support room-temperature skyrmion bubbles, only crystalline DMI systems support extremely long-range ordered skyrmion lattices, which unlock novel physics, such as emergent electromagnetism, magnetic monopoles and a rich variety of phase transitions, which are promising candidate for advanced applications. Therefore, increasing the magnetic ordering temperature and decreasing the skyrmion-stabilizing magnetic field is very important for the future applications of these materials.
MnSi is a weak itinerant, helically ordered ferromagnet with a magnetic bulk moment of µ bulk = 0.39 µ B /Mn, 16 naturally occurring in interplanetary dust particles. 17 The bulk Curie temperature, T C , is 29.5 K. 16 For skyrmion-based spintronic applications, thin film materials are pivotal. 18 High-quality thin film MnSi has been grown by molecular beam epitaxy (MBE) on Si(111). 19 Owing to the difference in lattice constants between MnSi and Si, the film is tensily strained in the plane, with the strain monotonically decreasing with film thickness. 20 The out-of-plane strain, on the other hand, is compressive and a nonmonotic function of film thickness. 20 As a result of strain in the film, T C becomes a function of film thickness and increases over 10 nm up to 43 K. 20 Currently, there are two fundamental issues that hamper the application of skyrmion-carrying materials systems. Firstly, the candidate materials should have an ordering temperature above room temperature. Secondly, the skyrmion phase has to be stabilized in zero applied field. However, these issues remain challenging as there are no reported B20-type helimagnetic materials with a T C above room temperature, and an external field is always required to manipulate the helimagnetism and stabilize the skyrmion phase. 2 Utilizing coupling mechanisms in artificial magnetic heterostructures provides a flexible playground for engineering B20 systems, as well as manipulating their helimagnetism. Here, we investigate the magnetic and structural properties of MBE-grown MnSi thin films using magnetometry, x-ray spectroscopy, electron microscopy, and x-ray diffraction. We find ferromagnetism in the seed layer, which persists up to room temperature, and which results in proximity coupling with the MnSi layer above.

II. THIN FILM GROWTH
Epitaxial MnSi(111) thin films were grown on Si(111) substrates measuring 10 × 12 mm 2 using MBE. The Balzers MBE system has a base pressure of 5 × 10 −10 mbar and is equipped with electron beam evaporators and effusion cells. Flux control is achieved via cross-beam mass spectrometry. Prior to loading, the Si wafers were first degreased, followed by etching in hydrofluoric acid and oxidation by H 2 O 2 . Annealing at 990 • C, and growing a Si buffer layer, leads to the 7 × 7 reconstruction, as confirmed by reflection high energy electron diffraction (RHEED). The sample is then cooled down to room temperature and ∼3 monolayers of Mn were deposited before they are reacted with the Si surface at an elevated temperature of ∼400 • C. One monolayer (ML) corresponds to 7.82 × 10 14 atoms/cm 2 . This leads to the reactive formation of an epitaxial MnSi seed layer. 21 For thicker Mn films, the supply of Si from the bulk is becoming the limiting factor, leading to deep holes. 21 The MnSi layer has a ( √ 30 × √ 30)R30 • structure, as determined by RHEED. Growth of the MnSi layer proceeds by stoichiometric deposition of Mn and Si, with the substrate at 250 • C, up to a thickness of ∼500 Å without any signs of the formation of a secondary phase in x-ray diffraction (XRD). The thicknesses obtained by x-ray reflectometry and in-situ film thickness monitoring agree well. 22

A. Superconducting quantum interference device (SQUID) studies
First, the temperature-magnetic field (T-B) phase diagram (PD) is studied as a function of MnSi thickness using SQUID magnetometry. 23 A one-to-one correspondence of the magnetic PD has been established between, on the one hand, studies of bulk samples using small-angle neutron scattering 2 and Lorentz transmission electron microscopy (LTEM) 24 and, on the other hand, ac and dc magnetometry. The details of the magnetic structures existing inside of the PD boundaries for MnSi thin films are still controversial in the literature: whereas some authors report the existence of a disordered skyrmion phase, 18 others argue that the observed data can be explained by a uniform conical phase. 25 However, the details of the magnetic structures inside the PD are not the focus of this work, as they require a different characterization approach. Here, we focus on the film-thickness dependence of the PD boundaries only, since these variations suggest similar changes inside of the PD boundaries, as has been observed in other B20 thin film systems. 8 The upper boundary, H c2 , which defines the transition from conical to field-polarized (FP) state, can be defined as the field at which the magnetic susceptibility is zero and is obtained from magnetization-field (M-H) curves. A linear background signal (containing, among others, the diamagnetic contribution of the substrate) was subtracted before performing this analysis. The transition temperature, T C , is obtained from M-T profiles measured in different applied fields. The samples were field cooled to 5 K from room temperature in a field of 2 T. M-T curves were measured during the warm-up, at fixed field values. For the helimagnetic-to-paramagnetic transition, T C is defined by the temperature point where d 2 M/dT 2 = 0.
The amalgamation of H c2 and T C defines the general shape of the magnetic PD, and the in-plane PD is shown in Fig. 1. The films show easy-plane anisotropy, and T C is between 38 K and 47 K, depending on the film thickness. The saturation magnetization M S is (0.41 ± 0.03) µ B /Mn, assuming that the film is phase-pure in the B20 phase. The phase boundaries have a similar shape, however, the boundary values vary systematically with film thickness. Note the out-of-plane PD exhibit the same general trends with film thickness. First, T C decreases as the thickness increases. This is commonly recognized as a strain effect since as the film becomes thicker, it should release the overall strain and eventually approach the bulk value of 29.5 K. Second, H c2 increases with film thickness for both in-plane and out-of-plane applied fields. This is an unusual behavior and cannot be simply explained by magnetic anisotropy effects, which, for thinner films with their stronger easy-plane anisotropy, would favor a lower in-plane and a higher out-of-plane H c2 . This means that the thinner films are easier to unwind from the conical state to the FP state, possibly pointing towards the existence of an 'internal' pinning field. This pinning effect has the same strength regardless of the sample thickness. We interpret this phenomenon as a magnetic proximity effect between the B20 helimagnetic thin film and the underlying interfacial volume, consisting of the MnSi seed layer and the upper layers of the Si substrate, as will be shown in the following.  made in total-electron-yield (TEY) and fluorescence-yield (FY) mode. Element-specific XMCD is used to probe the local electronic character of the magnetic ground state. 26 This technique allows for an unambiguous determination of the electronic and magnetic state of transition metals. XMCD is obtained from the difference between two XAS spectra recorded with the x-ray helicity vector and applied magnetic field parallel and antiparallel, respectively. The magnetic field is parallel to the x-ray beam which impinges at normal incidence. The XMCD is obtained by reversing the polarization of the incident x-rays to avoid having to change the magnetic field of the superconducting magnet. At 40 K, i.e., a few degrees below the T C of the 50-nm-thick MnSi film, the XMCD signal obtained in an applied field of 10 T shows ferromagnetic ordering both in surface-sensitive TEY [3-5 nm probing depth, see Fig. 2(a)] and bulk-sensitive FY mode [ Fig. 2(b)]. The XMCD TEY spectrum shows the signature of Mn 2+ . 27 The XMCD FY spectrum, on the other hand, suffers from strong saturation effects. 28 At 300 K, the surface-sensitive TEY signal has vanished (not shown), whereas the bulk-sensitive FY signal is still persistent [Fig. 2(c)]. An open-loop hysteresis curve is also seen in SQUID magnetization measurements [ Fig. 2(d)]. This behavior clearly points towards room temperature ferromagnetic ordering in the seed layer region.  The magnetization curves of a pure, reactively grown seed layer, shown in Fig. 2(e) with the field applied in-plane and out-of-plane, exhibit hysteretic behavior with a remanent magnetization [see inset in Fig. 2(e)], indicative of ferromagnetic ordering. Also, in-plane and out-of-plane magnetization curves are different, meaning that the observed ferromagnetism is not due to random magnetic clusters, but instead due to some long-range-ordered ferromagnetic phase showing magnetic anisotropy. In the following structural study, we try to shed light in the structural properties of the MnSi/Si interface region.

A. High-resolution TEM of interface
High-resolution TEM (HRTEM, phase contrast) and scanning TEM (STEM) in high-angle annular dark-field mode (STEM-HAADF) was carried out on a JEOL JEM 2100F to study the structural properties of the interfacial region in greater detail. Selected area electron diffraction (SAED) was recorded using a Gatan Ultrascan 2k × 2k CCD. Electron energy-loss spectroscopy (EELS) was performed in energy-filtered TEM mode making use of a post-column Gatan Quan-tumSE GIF. The sample for TEM analysis was prepared by mechanical polishing and subsequent low energy Ar-ion milling. Figure 3(a) shows a cross-sectional, bright-field TEM image of the MnSi/Si interface with a high-resolution (phase contrast) image as inset. The investigated film exhibits a low defect density. The SAED pattern of the film [ Fig. 3(b)] is indexed with the cubic B20 MnSi phase (space group  is indexed with the diamond space group Fd3m and the zone axis is identified as [110]. The SAED patterns of the interface also indicate that the (111) planes of MnSi and the (111) planes of the Si substrate are parallel. From these observations, it can be concluded that the epitaxial relationship between the MnSi film and the Si substrate is MnSi{111}<112 >∥ Si{111}<110>, as expected. 20 To investigate the Mn diffusion into the Si(111) substrate, we used EELS spectra obtained in STEM mode with a probe size of about 1 nm in diameter. The STEM-HAADF Z-contrast image [ Fig. 3(d)] clearly shows the uniform MnSi film, as well as a zone of local interface reaction with the Si(111) substrate. Point analysis was performed at several locations and the corresponding EELS spectra are shown on the right-hand side of Fig. 3(d). The spectra from the MnSi film (III) and the substrate close to the interface (II) show a signal at 640 eV, corresponding to Mn L edge, whereas Si far away from the interface (I) does not show a Mn signal. This measurement confirms that Mn diffused into the Si substrate.

B. Investigation of the inclusions by HRTEM and GIXRD
To analyze the Mn reaction in greater detail we performed HRTEM and grazing-incidence x-ray diffraction (GIXRD) analysis. A cross-sectional HRTEM image of the MnSi/Si(111) interface region is shown in Fig. 4(a). It should be noted that only a minority of the Si substrate is affected by this strong Mn diffusion, leading to an inclusion measuring ∼12 × 22 nm 2 . The x-ray reflectometry data can be fitted by only taking the B20 MnSi, a homogeneous MnSi seed layer, and the substrate into account. 22 To identify the phase formed at the interface, different reflections were separately masked and the corresponding image reconstructed by an inverse Fourier transform (not shown). The extracted d-spacing of 3.81 Å is not present in either MnSi or Si, and points towards the presence of Si-rich manganese silicides, such as Mn 4 Si 7 .
The GIXRD patterns were recorded on a Rigaku Smartlab diffractometer using Cu K α radiation (λ = 1.54 Å) and a scan rate of 0.5 • min −1 , in a 2θ range of 10 • -110 • at incidence angles between 0.25 • and 2.75 • , aligning with an arbitrary direction avoiding major reflections from both Si and MnSi. GIXRD reveals multiple peaks which are consistent with the formation of polycrystalline Mn 4 Si 7 [cf. Fig. 4(b), incident angle 0.5 • ], however, we cannot rule out the possibility of more than one Mn x Si y phase within the seed layer. Nevertheless, GIXRD confirms the existence of additional phases below the ordered B20 MnSi layer, consistent with results of the magnetic and structural studies reported here.

V. DISCUSSION
In order to identify the source of the magnetic properties at the interface, one has to look at the magnetic properties of the structural entities present. Single crystals of Mn 4 Si 7 show a T C ≈ 40 K, 29 however, nanoinclusions in Si exhibit size-dependent ferromagnetic behavior up to 320 K (Ref. 30) due to stoichiometry, strain, charge accumulation, and interfacial magnetism. 31 The intrinsic mechanism leading to high-temperature ferromagnetic order in Mn 4 Si 7 is exchange coupling between unbound 3d orbitals of Mn defects, mediated by spin fluctuations which exist at temperatures far above the bulk T C . 32 Another possibility is the ferromagnetic ordering at room temperature associated with slightly non-stoichiometric Si 1−x Mn x (x ≈ 0.51-0.55). 33 Although we find no hints for this behavior either in magnetometry or TEM, it cannot be fully excluded. In fact, deviations from the ideal stoichiometries could also be responsible for above room temperature ferromagnetism at interfaces between Mn and Si films 34 and around inclusions. 31 Previous studies on Mn-implanted Si have found above-room temperature ferromagnetism, 35 without the formation of Mn clusters, 36 and a tendency for Mn to diffuse to the surface. 37 X-ray-absorption studies show that Mn is neither incorporated substitutionally nor interstitially in the Si lattice, and the detailed doping scenario depends on the preparation conditions. 38 Independent of the precise cause of the ferromagnetism, it is clear from the magnetometry data in Fig. 2(e) that the interface is ferromagnetically ordered, has an easy-plane anisotropy, and saturates at lower fields than the B20 MnSi. As the ferromagnetic layer is in contact with the helimagnetic layer, proximity coupling seems to exist, as illustrated in the inset to Fig. 1. For thinner films, the pinning is dominant, whereas for thicker films, the intrinsic properties of bulk B20 MnSi start taking over. Therefore, the interface can be regarded as a magnetic pinning layer that provides an effective 'internal field'. This internal field assists the unwinding of the conical state, transitioning into the FP state as observed in Fig. 1. From the presented data it is clear that this effect can be used to optimize and engineer the skyrmion phase, potentially allowing for its stabilization at zero applied field and higher T C .

VI. SUMMARY AND CONCLUSIONS
In conclusion, we have shown that MnSi films grown on Si(111) experience proximity coupling to a ferromagnetic layer located at the interface between MnSi and Si using XMCD, magnetometry, and TEM. We have demonstrated an effective method to modify the phase diagram of MnSi thin films, utilizing the interlayer exchange interaction. In this case, one requires less external field to stabilize the helimagnetic phase and T C can be significantly increased. We would like to emphasize that this method is not limited to MnSi/Si, but also applies to other B20 helimagnetic systems. One promising application for this approach is FeGe that has a higher, near-room temperature T C . By fabricating an artificial pinning layer, such as NiFe, one may eventually realize room temperature, zero-field skyrmion thin film systems.