betaFeedback

WorksPhysics and Astronomy

Topological magnons driven by the Dzyaloshinskii-Moriya interaction in the centrosymmetric ferromagnet Mn5_5Ge3_3

Abstract

The phase of the quantum-mechanical wave function can encode a topological structure with wide-ranging physical consequences, such as anomalous transport effects and the existence of edge states robust against perturbations. While this has been exhaustively demonstrated for electrons, properties associated with the elementary quasiparticles in magnetic materials are still underexplored. Here, we show theoretically and via inelastic neutron scattering experiments that the bulk ferromagnet Mn5_5Ge3_3 hosts gapped topological Dirac magnons. Although inversion symmetry prohibits a net Dzyaloshinskii-Moriya interaction in the unit cell, it is locally allowed and is responsible for the gap opening in the magnon spectrum. This gap is predicted and experimentally verified to close by rotating the magnetization away from the cc-axis with an applied magnetic field. Hence, Mn5_5Ge3_3 realizes a gapped Dirac magnon material in three dimensions. Its tunability by chemical doping or by thin film nanostructuring defines an exciting new platform to explore and design topological magnons. More generally, our experimental route to verify and control the topological character of the magnons is applicable to bulk centrosymmetric hexagonal materials, which calls for systematic investigation.

Introduction

Recent breakthroughs in the physics of electrons in solids resulted from the application of topological concepts to the quantum-mechanical wave function, highlighting the role of the Berry phase [1]. For instance, the modern understanding of the integer quantum Hall effect in a two-dimensional (2D) system is that of a gapped bulk with non-zero Chern numbers which imply the existence of chiral edge states responsible for the quantized conduction [2]. Three-dimensional (3D) topological insulators are gapped by the spin-orbit interaction, leading to Dirac-like surface states with a linear dispersion and spin-momentum locking that underpin the quantum spin Hall effect [3]. There is now a systematic classification of the possible topological phases in electronic systems, encompassing also gapless systems such as Weyl semimetals, where the dimensionality but also spatial and magnetic symmetries are prominent [4, 5, 6].

Topology is agnostic to whether the (quasi)particles are fermions or bosons, so that magnons can also be responsible for novel physical effects [7]. Perhaps the first example is the Hall effect experienced by a thermally-induced magnon current in ferromagnetic (FM) Lu2_2V2_2O7_7, with the Dzyaloshinskii-Moriya interaction (DMI) playing a similar role to the one of spin-orbit coupling (SOC) for electronic systems [8, 9, 10]. Magnetic materials with an hexagonal lattice generally exhibit a Dirac-like magnon dispersion at the K-point in the Brillouin zone, which if gapped signals the existence of non-trivial topology [11, 12, 13, 14, 15]. Such topological magnon insulators were experimentally identified in 2D FM materials [16, 17, 18], while gapless Dirac magnons were characterized in 3D magnetic materials such as antiferromagnetic (AFM) Cu3_3TeO6_6 [19] and CoTiO3_3 [20, 21], and the elemental FM Gd [22]. While a symmetry-based approach has been proposed to predict materials hosting topological magnons [23], experimentally confirming the topological character is challenging, as the magnon Hall effect is difficult to measure and other signatures such as a characteristic winding of the scattering intensity have only recently been detected [21].

In this article, we study the 3D centrosymmetric ferromagnet Mn5_5Ge3_3 [24, 25]. This material exhibits significant anomalous Hall [26] and Nernst [27] effects which are signatures of large electronic Berry phases. Its Curie temperature (TCT_{\mathrm{C}}) can be enhanced by carbon doping [28] as successfully described by a previous theoretical work [29], but its magnonic properties remain unexplored. Here, we theoretically predict and experimentally confirm the existence of gapped Dirac magnons at the K-point due to the DMI. This gap can be closed by rotating the magnetization direction with an external magnetic field, thus validating the proposed gap mechanism and confirming the topological character of the magnons at the K-point. Our experimental route to verify and control the topological character of the magnons is not limited to Mn5_5Ge3_3 and should also be applicable to other bulk centrosymmetric hexagonal materials.

Results

Basic properties of Mn5_5Ge3_3

A high-quality Mn5_5Ge3_3 single crystal of about 10 g has been grown using the Czochralski method. The space group is P63/mcmP6_3/mcm and the unit cell contains 10 Mn atoms (and 6 Ge atoms), with Mn1 and Mn2 occupying the Wyckoff positions 4d4d and 6g6g, respectively [24]. In the abab-plane, Mn1 adopts a honeycomb lattice while Mn2 adopts an hexagonal arrangement, Figure 1a. Along the cc-axis, Mn1 forms chains and Mn2 columns of face-sharing octahedra, Figure 1b. The Curie temperature (TCT_\mathrm{C}) is around 300 K and the magnetic moments of the Mn1 and Mn2 atoms are 1.96(3) μB\mu_\mathrm{B} and 3.23(2) μB\mu_\mathrm{B}, respectively, aligned along the cc-axis [24, 25].

Figure 1. Crystal structure and theoretical magnon bands of Mn5_5Ge3_3. a Top view of the crystal structure, showing bonds with length up to 4.3 Å. The unit cell is indicated by the thin black lines. b Perspective view of the crystal structure, omitting the Ge atoms. The first few magnetic exchange interactions are marked with the respective non-vanishing Dzyaloshinskii-Moriya interaction (DMI) vectors located in the bond centers. The values are collected in Table I. c Computed magnon bands using the magnetic exchange interactions extracted from DFT calculations performed for the DFT optimized structure (blue) and for the experimental structure reported in Ref. [24] (red). The solid and dashed lines show the magnon bands obtained with and without the DMI, respectively. With DMI, a gap opens at the K-point. The magnetization is set along the cc-axis, which is the easy-axis of the material. The inset indicates the location of the high-symmetry points in the hexagonal Brillouin zone.

Magnetic properties from first principles

Density functional theory (DFT) calculations were performed prior to the inelastic neutron scattering measurements in order to provide an initial picture of the expected magnon excitations and to identify interesting regions in (Q\mathbf{Q},EE) space. The theoretical magnetic moments from juKKR (2.11 μB\mu_\mathrm{B} and 3.14 μB\mu_\mathrm{B} for Mn1 and Mn2, respectively) and the computed Heisenberg exchange interactions are comparable to the ones previously reported [29], as seen in Table I. Spin-orbit coupling leads to significant DMI (c.f. Table I), much weaker symmetric anisotropic exchange (not included in Eq. (1)), and a small uniaxial magnetic anisotropy energy (K≈−0.1K \approx-0.1 meV), so that the relevant spin Hamiltonian (with ∣Si∣=1|\mathbf{S}_i| = 1) reads:

H=K∑i(Siz)2−∑i,jJijSi⋅Sj−∑i,jDij⋅(Si×Sj).(1)\mathcal{H} = K \sum_i (S_i^z)^2 - \sum_{i,j} J_{ij} \mathbf{S}_i \cdot\mathbf{S}_j - \sum_{i,j} \mathbf{D}_{ij} \cdot(\mathbf{S}_i \times\mathbf{S}_j). \tag*{(1)}

Here JijJ_{ij} are the Heisenberg exchange interactions and Dij\mathbf{D}_{ij} are the DMI vectors which mostly align along the c-axis, with the strongest shown in Figure 1b. We discovered that some of the magnetic interactions, namely the AFM ones, are quite sensitive to small changes in the unit cell parameter and the atomic positions. The impact of this can be seen in the theoretical magnon dispersions shown in Figure 1c, where we compare the results obtained with the experimental crystal structure parameters and with the theoretically optimized ones. In both cases there are two magnon bands in the energy range of experimental interest, with an energy gap at the K-point where otherwise a Dirac-like crossing of the bands would be expected by symmetry. This is straightforwardly verified to arise from the DMI, as omitting it from the magnon calculation leads to the closing of the gap.

TypeRef. [29] Distance (Å)Ref. [29] Value (meV)Experimental structure [24] Distance (Å)Experimental structure [24] Value (meV)DFT structure Distance (Å)DFT structure Value (meV)
J1J_1Mn1-Mn12.52629.12.52730.872.48531.59
J2J_2Mn1-Mn23.0688.03.0658.653.0217.82
J3J_3Mn2-Mn22.974-2.02.983-1.273.013-0.21
J4J_4Mn2-Mn23.0556.93.0586.843.0336.10
J5J_5Mn1-Mn14.148-1.44.148-3.864.112-2.52
J6J_6Mn2-Mn24.2639.44.2719.974.2769.86
∣D2∣\lvert D_2\rvertMn1-Mn2——3.0650.573.0210.59
∣D3∣\lvert D_3\rvertMn2-Mn2——2.9830.503.0130.45

Table 1. TABLE I. Magnetic exchange interactions in Mn5_5Ge3_3 from DFT. We compare our results using the experimental crystal structure [24] with those of Ref. [29], and to our results using the theoretically optimized crystal structure. Positive (negative) values characterize FM (AFM) coupling. The corresponding pairs are indicated in Figure 1b. The calculated magnetic interactions are long-ranged and only the first few values are given in the table. The listed values are significant to the displayed decimal precision.

Dzyaloshinskii-Moriya interaction in centrosymmetric systems

The DMI is the key magnetic interaction for the subsequent interpretation of our experimental findings, so before we continue we wish to clarify how it can be present and have an effect in a centrosymmetric material. In his seminal paper [30], Moriya established the symmetry rules that the interaction named after Dzyaloshinskii and himself must obey. The most famous of these rules is that if two spins are connected by an inversion center then the respective DMI must identically vanish. This pointedly explains why we find finite DMI in our calculations for Mn5_5Ge3_3: it is finite for those spin pairs that do not contain an inversion center in the midpoint of the corresponding bond, such as those illustrated in Figure 1b.

Centrosymmetry does ensure that the net DMI of the unit cell is zero, which is also verified in our simulations. This means that the ferromagnetic domain walls are not chiral and that magnetic skyrmions cannot form, in agreement with Neumann’s principle. In contrast, magnons can still be influenced by the local DMI. In a semiclassical picture, the spins at different sites precess with different phases and/or amplitudes, so that certain pairs of spins are noncollinear and can be affected by the torque arising from the DMI. This is a strong effect at the K-point, where two magnon modes with opposite chirality cross and the degeneracy is lifted in a non-perturbative way by the DMI. We now report the experimental observation of this effect and its implications.

Magnons from inelastic neutron scattering

Inspired by these theoretical predictions, the experimental magnon spectrum of Mn5_5Ge3_3 was investigated by INS. Several constant-Q\mathbf{Q} and constant-EE scans have been performed at T=10T = 10 K along the reciprocal space directions (h,0,0)(h, 0, 0), (h,0,2)(h, 0, 2), (h,h,0)(h, h, 0), and (0,0,l)(0, 0, l), with representative examples shown in Figure 2a. The measured scattering intensity (circles) was fitted with Gaussian line shapes on top of a sloping background (lines).

Figure 2. Magnon bands of Mn5_5Ge3_3 determined by inelastic neutron scattering. a Representative measurements obtained at IN22 with constant-Q\mathbf{Q} scans (symbols) at T=10 KT = 10\,\mathrm{K} with ∣kf∣=2.662 A˚−1|k_\mathrm{f}| = 2.662\,\mathrm{\mathring{A}}^{-1} and fits used to determine the excitation energies (lines). The purple arrows and rectangles indicate the peak positions in the corresponding (q,E)(\mathbf{q},E) region of the dispersion. The dashed lines show the individual contributions of the various Gaussian peaks and of the linear background. The error bars indicate one standard deviation (square root of the neutron counts). b Experimentally determined magnon bands (symbols) and fit to a simplified model described in the text (lines). The solid and the dashed lines show the magnon bands obtained with and without the Dzyaloshinskii-Moriya interaction (DMI), respectively. The error bars indicate the uncertainty in the peak location from least-squares fitting.

The magnetic nature of the excitations has been confirmed through their temperature dependence (see Supplementary Figure 4–8 [31] and Figure 3a below), and the obtained qq and EE position is given as red symbols in Figure 2b. For the in-plane directions (Γ−M−K−Γ)(\Gamma- \mathrm{M} - \mathrm{K} - \Gamma) one can distinguish the presence of three modes: two acoustic-like modes dispersing upwards in energy away from Γ\Gamma and additionally a third higher-energy mode with a steep dispersion along Γ−M\Gamma- \mathrm{M} and weakly dispersive along K−M\mathrm{K} - \mathrm{M}. In contrast, in the same investigated EE range a single stiff spin-wave mode is observed for the out-of-plane direction (Γ−A)(\Gamma- \mathrm{A}).

Figure 3. Temperature and magnetic field dependence of the magnon peaks at the K-point determined by inelastic neutron scattering. a Temperature dependence. At T=10 KT = 10\,\mathrm{K} three peaks are seen between 8 meV8\,\mathrm{meV} and 22 meV22\,\mathrm{meV}. At T=398 KT = 398\,\mathrm{K} (above TCT_\mathrm{C}) only a broad quasielastic signal is observed. Neutron intensity for the data at 10 K10\,\mathrm{K} and 398 K398\,\mathrm{K} is given on the right and left axis, respectively. b Dependence on magnetic field applied along the cc-axis at T=10 KT = 10\,\mathrm{K}. The field has almost no effect on the location of the two peaks. c Dependence on magnetic field applied along the a∗a^*-axis at T=10 KT = 10\,\mathrm{K}. In zero field the magnetization is along the cc-axis and two peaks are visible. The application of a transverse magnetic field saturates the magnetization along the aa-axis and merges the two peaks into one, demonstrating the closure of the magnon energy gap. The data shown in a were obtained at IN22 (∣kf∣=2.662 A˚−1\lvert\mathbf{k}_f\rvert= 2.662\,\mathrm{\AA}^{-1}) and the data in b-c at IN12 (∣kf∣=1.971 A˚−1\lvert\mathbf{k}_f\rvert= 1.971\,\mathrm{\AA}^{-1}). The error bars indicate one standard deviation (square root of the neutron counts).

Figure 4. Magnon surface states in a (011ˉ0)(01\bar{1}0)-oriented slab of Mn5_5Ge3_3. a Relation between the primitive cell and the rectangular cell used to define the slab, viewed along the cc-axis, using the simplified model that fits the experimental data in Figure 2b. The Mn1 chains are indicated by the two different shades of red, while the effective spin taken to represent the Mn2 sites is shown in blue. b Path in reciprocal space used for the magnon calculation. c Magnon bands of a 20-unit-cell-wide slab computed with periodic boundary conditions (PBC). d Magnon bands of a 20-unit-cell-wide slab computed by changing periodic to open boundary conditions (OBC) along the (011ˉ0)(01\bar{1}0) orientation. Results obtained excluding/including the Dzyaloshinskii-Moriya interaction (DMI) are shown by dashed/solid lines. The green arrow in (d) indicates the DMI-enabled magnon surface state.

We now turn to the theoretical interpretation of these measurements. Comparing the experimental results of Figure 2b with Figure 1c, we find that the spin model derived from the DFT calculations qualitatively reproduces several features. The absence of a clearly visible gap in the spin-wave spectrum near zero energy transfer (Γ\Gamma-point) agrees with the expected weakness of the uniaxial magnetic anisotropy [25], as also computed from DFT. The theoretical dispersion along Γ−A\Gamma- \mathrm{A} is slightly stiffer than the experimental one, and a simulation of the INS intensity reveals that the second mode which is higher in energy should be invisible (see Supplementary Fig. 14 [31]). We find rather poor quantitative agreement between theory and experiment in the Γ−K−M\Gamma- \mathrm{K} - \mathrm{M} plane, which is probably related to the already-identified strong dependence of the magnetic interactions computed from DFT on small changes of the structural parameters. We verified that this dependence is systematic. by considering various deformations of the unit cell (see Supplementary Figs. 11 and 12 [31]). However, the most crucial feature is observed both in theory and in experiment, that is the existence of a gap at the K-point where two magnon bands should otherwise cross.

In order to provide a more quantitative description of the experimentally obtained magnon bands for the in-plane directions, we constructed a simplified effective spin model (additional details are given in the Supplementary Information, Section II.E [31]). We replace each Mn2 triangle at a constant height in the unit cell with a single effective spin S=9/2S = 9/2 (S=1S = 1 for Mn1), so that the column of Mn2 octahedra is replaced by a spin chain. Seen from the cc-axis, the unit cell for this model thus contains two Mn1 chains and one effective Mn2 chain, and we determine the model parameters using the measured magnon energies at the high-symmetry points. The lines in Figure 2b show that the results of this model approach indeed provide a realistic band dispersion, and confirm once more that the gap at the KK-point is a consequence of a finite DMI. The model results also highlight a peculiarity of the measured magnon energies in the vicinity of Γ\Gamma, to which we shall briefly return in the Discussion.

Closing the topological magnon gap

Although there are strong theoretical arguments in favor of the topological character of the magnon gap at the KK-point, a convincing experimental demonstration is in order. The gap is expected to arise from the DMI, but such a microscopic interaction cannot easy be manipulated experimentally. However, it is known that the impact of the DMI on the FM magnon spectrum depends on the relative alignment between the vectors that characterize this interaction and the FM magnetization [32, 33, 34]. Adapting these arguments to the hexagonal symmetry of Mn5_5Ge3_3 leads to the prediction that the magnitude of the gap should depend on the relative alignment of the FM magnetization and the crystalline cc-axis, and in particular should vanish if the two are perpendicular. We have verified that the magnon gap closes both in the DFT-based spin model and in the one fitted to the experimental measurements when the magnetisation is perpendicular to the cc-axis (as it does when disregarding the DMI). The magnon dispersions computed with DMI and setting the magnetization along the a∗a^*-axis are identical to the dashed lines shown in Figure 1c and Figure 2b, which were obtained by excluding the DMI from the calculations.

This hypothesis can be experimentally tested by applying an external magnetic field. First we rule out the possibility of a phonon contribution to the inelastic peaks at the KK-point, by heating the sample above its TCT_{\mathrm{C}}, as seen in Figure 3a, and verifying that the peaks disappear conforming their magnetic origin. The measurements reported so far in this work were performed in zero field, for which the magnetization is parallel to the cc-axis due to the uniaxial magnetic anisotropy. Applying a magnetic field along the cc-axis should lead to a very small Zeeman shift of the magnon energies, and this is indeed what we observed, as seen in Figure 3b. If a magnetic field of similar magnitude is applied along the a∗a^*-axis it overcomes the magnetic anisotropy energy and saturates the magnetization [25]. The results

obtained in this way are shown in Figure 3c. Now the effect of the field cannot be explained by a simple Zeeman shift, and instead we find the anticipated closing of the magnon gap. The two peaks observed in zero field coalesce into a single one with an integrated intensity approximately matching the sum of intensities for the two peaks in zero field (see also Supplementary Fig. 10 [31]), although shifted to slightly higher energy than anticipated. The distinct response of the magnon excitation to a magnetic field applied to orthogonal crystal directions is consistent with the DMI mechanism, and so the gapped Dirac magnons at the K-point should consequently have a topological nature.

Ruling out alternative explanations for a gap at the K-point

Next we rule out potential alternative mechanisms to the DMI that could lead to a magnon gap opening at the K-point, such as dipolar, Kitaev and magnon-phonon interactions:

(i) Dipolar interactions are long-ranged but much weaker than the magnetic exchange interactions, so their effect is usually seen for rather small wave vectors in the vicinity of the Γ\Gamma-point. Even if they did lift the magnon degeneracy at the K-point, their intrinsic weakness could not account for the observed magnitude of the gap.

(ii) Kitaev interactions were proposed for instance in Ref. [35] to explain measurements on CrI3\mathrm{CrI}_{3} but are ruled out for Mn5Ge3\mathrm{Mn}_{5}\mathrm{Ge}_{3} both by our simulations and by general considerations. The interactions extracted from our DFT calculations include both the Heisenberg exchange, the DMI and the symmetric anisotropic exchange (SAE), which includes the Kitaev interaction. The SAE was found to be rather weak and unable to open the observed magnon gap at the K-point. The weakness of the SAE (and so of potential Kitaev interactions) could be anticipated from the weak magnetic anisotropy measured for this system. This reflects the lack of heavy elements in the material that could supply a strong spin-orbit interaction, which is a key ingredient for obtaining a sizeable Kitaev interaction. We are also not aware of any Kitaev material candidates containing only elements from the first four rows of the periodic table (i.e. Z<36Z < 36), likely due to the preceding reason. To make this argument more quantitative, we employ the theory of magnetic exchange interactions for systems with weak SOC presented by Moriya [30], Eqs. 2.3 and 2.4. The DMI is first-order in the weak SOC, while the Kitaev interaction is part of the SAE which is second-order in SOC and so is much weaker than the DMI. The Kitaev interaction, if present, would contribute to the magnetic anisotropy energy, which is about 1 meV/f.u. for Mn5Ge3\mathrm{Mn}_{5}\mathrm{Ge}_{3} (the DMI does not contribute to the magnetic anisotropy energy of the ferromagnetic state). To give an estimate of the potential magnitude of the Kitaev interaction using only experimental input, we distribute the magnetic anisotropy energy on one of the set of bonds for which we identified the DMI, bond #2 indicated in Figure 1b. This set of bonds occurs four times in the unit cell, as it connects each Mn1 site to its six Mn2 neighbours, and so could have 1 meV/24 = 0.04 meV maximum Kitaev strength. The SAE obtained directly from the DFT calculations is about 0.02 meV in magnitude, which is in line with this estimate, and is one order of magnitude smaller than the values found for the DMI (0.57 meV for the set of bonds #2), as expected from the theory of the magnetic exchange interactions for systems with weak SOC.

(iii) Magnon-phonon interactions can result in gaps at the crossing points between the magnon and phonon branches. However, our INS data ruled out this possibility. The measured excitations at different Q\mathbf{Q} vectors around and at the K-point are solely of magnetic origin. This has been verified through their temperature dependence. All the peaks observed in the energy range from 10 to 20 meV at T=10T = 10 K are replaced by a broad quasi-elastic signal (centered at 0 meV) above the ordering temperature as shown in Figure 3a. Hence no phonon modes were detected in the vicinity of the K-point with an energy compatible with that of the magnons, which is a requirement for the gap opening mechanism through magnon-phonon interactions.

Therefore, we can assert that the only reasonable mechanism for the gap opening at the K-point is the DMI.

Simulation of magnon surface states

Here we explore the expectation that if the bulk magnon band structure has some topologically non-trivial character it should be accompanied by magnon surface states. To do so, we compute the magnon band structure of slabs which are finite in one direction and periodic (i.e., infinite) in the other two directions. Comparing the simulations performed for the same slab with periodic and open boundary conditions along the chosen surface normal enables us to identify the energy range corresponding to the surface projection of the bulk magnon bands. Surface magnons are then expected to appear in the regions of (E,q)(E,\mathbf{q}) where bulk magnon bands are absent.

To illustrate this point, we performed simulations using the simplified spin model depicted in Figure 4a with parameters fitted to the experimental measurements in Figure 2b. We created a rectangular supercell and extended it by 20 unit cells along the (011ˉ0)(01\bar{1}0) direction of the original hexagonal lattice. The chosen path in reciprocal space is perpendicular to the surface normal and shown in Figure 4b. Figure 4c shows that indeed there is a pocket in the K–M–K path and with energies between 10 and 12 meV from which bulk magnons are absent, with or without DMI. Figure 4d shows the modified magnon band structure upon truncating the crystal along the (011ˉ0)(01\bar{1}0) direction, i.e., making a horizontal cut through the lattice shown in Figure 4a. We indeed find that magnon surface states do appear in the identified region where bulk magnon bands were absent. Without the DMI, these surface magnons are disconnected and gapped from each other. The DMI restructures the band connectivity and leads to a crossing that resembles a distorted Weyl-like crossing. The crossing is not located at the M-point as the slab loses the acac-mirror plane, retaining instead a twofold rotation around the cc-axis. There are other surface magnons at around 3 meV and 18 meV that are only weakly affected by the DMI, and result from the reduced coordination number introduced by creating the surface. Lastly, the identified surface magnons were found to extend only a couple of unit cells towards the interior of the slab, confirming their localization at the surface.

The experimental detection of the predicted surface magnons is quite challenging, as the scattering volume is too small for a straightforward detection using INS. Other techniques such as Brillouin light scattering (for magnons near the Γ\Gamma-point) or electron energy loss spectroscopy could be considered for this purpose [36].

Discussion

We now briefly discuss some outstanding points. Firstly, we return to the quantitative disagreement between theory and experiment concerning the magnon bands. The main issue seems to be an overestimation of the magnitude of the magnetic interactions in the DFT calculations, which has also been found in other studies. A recent example from the literature is Co3Sn2S2\mathrm{Co}_{3}\mathrm{Sn}_{2}\mathrm{S}_{2} [37], where two different DFT approaches are compared with experiment, with disagreements also in the Γ−M−K−Γ\Gamma-\mathrm{M}-\mathrm{K}-\Gamma plane but not in the Γ−A\Gamma-\mathrm{A} direction. Another issue is that the simplified spin model does not adequately capture the relative intensities of the two peaks around the K-point, despite providing a reasonable description of the experimental magnon energies. This is likely due to the model assumptions, namely treating the Mn2\mathrm{Mn}_{2} sites as a single effective spin and neglecting the atomic magnetic form factor, as well as employing a simple Lorentzian broadening for the peaks. This shows the need for further developments on the theoretical side. We also noticed that the ratio of the intensities of the two peaks varies slightly in different experimental setups, see Figure 3b-Figure 3c. This might be due to changes in the domain state of the sample arising from the measurement history.

Our INS measurements for Mn5Ge3\mathrm{Mn}_{5}\mathrm{Ge}_{3} revealed a peculiar dispersion for the second mode in the Γ−K\Gamma-\mathrm{K} direction. Such a steep linear-like dispersion close to the Γ\Gamma-point resembles an AFM magnon or a phonon mode. We restate that the material is FM and the measured excitations are of magnetic nature as verified by measurements above TCT_{\mathrm{C}} (Figure 3a and Supplementary Figs. 4-8 [31]), so that both simple explanations are ruled out. However, it is possible to have formation of hybrid collective modes. INS can identify the magnetic and phononic component of such excitations, which are referred as magnon-polarons (magnetoelastic modes), and are reported in several magnetic materials [38, 39]. Avoided crossings are the common signature of magnetoelastic interactions, but they also underpin the magnetovibrational scattering term in the INS scattering cross section [40, 41]. Additional discussion is given in the Supplementary Information, Section I.C [31]. We propose that Mn5_5Ge3_3 is also an interesting 3D FM candidate material for detailed investigation of magnetoelastic effects [42, 43].

The key interaction responsible for opening the magnon gap at the K-point and thus endowing the Dirac magnons with a topological character is the DMI. There are several possible routes by which this interaction could be engineered, so that the magnon properties can be tuned for specific purposes for magnonic devices operating in a broad temperature range. Firstly, chemical substitution of Ge by Si has been explored in the literature to connect to the multifunctional AFM Mn5_5Si3_3 [44, 45, 46, 47]. Substituting Ge by Si reduces TCT_{\mathrm{C}} [48], but the impact on the magnons and on the DMI is unknown. On the other hand, carbon implantation is demonstrated to enhance TCT_{\mathrm{C}} [27, 49] for which the imprint on the Heisenberg exchange interactions has been theoretically established [29], but once again the effect on the magnons and the DMI remains unexplored. Lastly, Mn5_5Ge3_3 can also be grown in thin film form [27, 49]. This could modify the DMI by epitaxial strain, which would also be interesting in connection to potential magnetoelastic effects, by interfaces to other materials, or by quantum confinement effects if the thickness is just a few nanometers.

In conclusion, we presented a combined theoretical and experimental study of the magnons in the centrosymmetric 3D FM Mn5_5Ge3_3. Despite the inversion symmetry, significant DMI has been theoretically identified on Mn-Mn bonds which do not contain an inversion center. This DMI is responsible for opening a gap in the magnon spectrum at the K-point, where otherwise symmetry would enforce a Dirac-like crossing of the magnon bands. INS measurements of the magnon spectrum show qualitative agreement with the main points predicted by theory, and confirm the expected gap at the K-point. We experimentally observe the closing of the gap by rotating the magnetization from the cc-axis to the a∗a^{*}-axis with a magnetic field. This both validates the gap generation mechanism and the topological nature of the magnons at the K-point, thus establishing Mn5Ge3\mathrm{Mn}_{5}\mathrm{Ge}_{3} as a realization of a gapped Dirac magnon material in three dimensions. The ability to control the gap at the K-point with an external magnetic field will also impact topological magnon surface states, and deserves further study. As the macroscopic magnetic properties of Mn5Ge3\mathrm{Mn}_{5}\mathrm{Ge}_{3} can be tuned by chemical substitution of Ge with Si or by carbon implantation, and it can also be grown as thin films in spintronics heterostructures, we foresee that the features of the newly-discovered topological magnons can be engineered and subsequently integrated in novel device concepts for magnonic applications. Looking beyond Mn5Ge3\mathrm{Mn}_{5}\mathrm{Ge}_{3}, the physical mechanism leading to the formation of topological magnons at the K-point should be present in many other bulk centrosymmetric hexagonal materials, which opens an exciting avenue for future investigations.

Methods

Experimental methods

Inelastic neutron scattering (INS) experiments have been carried out on a cold (IN12) and a thermal (IN22) triple axis spectrometer at the Institut Laue-Langevin, in Grenoble, France. We use the hexagonal coordinate system and the scattering vector Q\mathbf{Q} is given in reciprocal lattice units (r.l.u.). The wave-vector q\mathbf{q} is related to Q\mathbf{Q} through Q=q+G\mathbf{Q} = \mathbf{q} + \mathbf{G}, where G\mathbf{G} is a Brillouin zone center. Inelastic scans were performed with constant ∣kf∣\lvert\mathbf{k}_{f}\rvert, where kf\mathbf{k}_{f} is the wave-vector of the scattered neutron beam. Data were collected holding either the energy (constant-EE) or the scattering vector (constant-Q\mathbf{Q}) fixed. Further details on the experimental procedures and additional measurements can be found in the Supplementary Information, Section I [31].

Theoretical methods

The theoretical results were obtained with DFT calculations and the extracted spin Hamiltonian. The unit cell parameters and the atomic positions were optimized with the DFT code Quantum Espresso [50]. The magnetic parameters were computed with the DFT code juKKR [51, 52], which are then used to solve a spin Hamiltonian in the linear spin wave approximation [53]. Further details on all these aspects can be found in the Supplementary Information, Section II [31].

Data availability

The authors declare that the data supporting the findings of this study are available within the paper, its supplementary information file and in the following repositories [54, 55, 56, 57, 58].

Code availability

The DFT simulation packages Quantum Espresso and juKKR are publicly available (see Methods). The code for the solution of the linear spin wave problem is available from the corresponding authors upon request.

References

Acknowledgements

We thank S. Raymond for discussions and comments. The work of M.d.S.D. made use of computational support by CoSeC, the Computational Science Centre for Research Communities, through CCP9. N.B. acknowledges the support of JCNS through the Tasso Springer fellowship. F.J.d.S. acknowledges support of the European H2020 Intersect project (Grant No. 814487), and N.M. of the Swiss National Science Foundation (SNSF) through its National Centre of Competence in Research (NCCR) MARVEL. This work was also supported by the Brazilian funding agency CAPES under Project No. 13703/13-7 and the Deutsche Forschungsgemeinschaft (DFG) through SPP 2137 “Skyrmionics” (Project LO 1659/8-1). We gratefully acknowledge the computing time granted through JARA on the supercomputer JURECA [74] at Forschungszentrum Jülich GmbH and by RWTH Aachen University. The neutron data collected at the Institut Laue-Langevin are available at Refs. [54, 55, 56, 57].

Author contributions

MdSD, NB and FJdS contributed equally to this work. MdSD and NB conceived the project together with TB and SL. MdSD performed most DFT calculations and the spin-wave modelling, with additional calculations performed by FJdS. JP grew the Mn5_5Ge3_3 single crystal. NB performed the experimental measurements and the corresponding data analysis. KS and FB were local contacts of IN12 and IN22 and provided instrument support. The theoretical aspects of the work were discussed with NM, SB and SL. All authors participated in the discussion of the results. MdSD, NB and FJdS wrote the manuscript with input from all authors.

Competing interests

The authors declare no competing interests.

References

  1. [1]M. V. Berry, “Quantum phase factors accompanying adiabatic changes,” Proc. R. Soc. Lond. A 392, 45–57 (1984).
  2. [2]D. J. Thouless, M. Kohmoto, M. P. Nightingale, and M. den Nijs, “Quantized Hall Conductance in a Two-Dimensional Periodic Potential,” Phys. Rev. Lett. 49, 405–408 (1982).
  3. [3]M. Z. Hasan and C. L. Kane, “Colloquium: Topological insulators,” Rev. Mod. Phys. 82, 3045–3067 (2010).
  4. [4]Barry Bradlyn, L. Elcoro, Jennifer Cano, M. G. Vergniory, Zhijun Wang, C. Felser, M. I. Ar oyo, and B. Andrei Bernevig, “Topological quantum chemistry,” Nature 547, 298–305 (2017).DOI
  5. [5]Hoi Chun Po, Ashvin Vishwanath, and Haruki Watanabe, “Symmetry-based indicators of band topology in the 230 space groups,” Nat. Commun. 8, 50 (2017).DOI
  6. [6]Jorrit Kruthoff, Jan de Boer, Jasper van Wezel, Charles L. Kane, and Robert-Jan Slager, “Topological Classification of Crystalline Insulators through Band Structure Combinatorics,” Phys. Rev. X 7, 041069 (2017).DOI
  7. [7]Paul A. McClarty, “Topological Magnons: A Review,” Annu Rev Conden Ma P 13, 171–190 (2022).DOI
  8. [8]Y. Onose, T. Ideue, H. Katsura, Y. Shiomi, N. Nagaosa, and Y. Tokura, “Observation of the Magnon Hall Effect,” Science 329, 297–299 (2010).DOI
  9. [9]Ryo Matsumoto and Shuichi Murakami, “Theoretical Prediction of a Rotating Magnon Wave Packet in Ferromagnets,” Phys. Rev. Lett. 106, 197202 (2011).pubmed.ncbi.nlm.nih.gov/21668195
  10. [10]M. Mena, R. S. Perry, T. G. Perring, M. D. Le, S. Guerrero, M. Storni, D. T. Adroja, Ch. Rüegg, and D. F. McMorrow, “Spin-Wave Spectrum of the Quantum Ferromagnet on the Pyrochlore Lattice Lu2V2O7,” Phys. Rev. Lett. 113, 047202 (2014).DOI
  11. [11]Lifa Zhang, Jie Ren, Jian-Sheng Wang, and Baowen Li, “Topological magnon insulator in insulating ferromagnet,” Phys. Rev. B 87, 144101 (2013).
  12. [12]Alexander Mook, Jürgen Henk, and Ingrid Mertig, “Edge states in topological magnon insulators,” Phys. Rev. B 90, 024412 (2014).DOI
  13. [13]R. Chisnell, J. S. Helton, D. E. Freedman, D. K. Singh, R. I. Bewley, D. G. Nocera, and Y. S. Lee, “Topological Magnon Bands in a Kagome Lattice Ferromagnet,” Phys. Rev. Lett. 115, 147201 (2015).DOI
  14. [14]J. Fransson, A. M. Black-Schaffer, and A. V. Balatsky, “Magnon Dirac materials,” Phys. Rev. B 94, 075401 (2016).DOI
  15. [15]Sergey S. Pershoguba, Saikat Banerjee, J. C. Lashley, Jihwey Park, Hans Ågren, Gabriel Aeppli, and Alexander V. Balatsky, “Dirac Magnons in Honeycomb Ferromagnets,” Phys. Rev. X 8, 011010 (2018).DOI
  16. [16]Lebing Chen, Jae-Ho Chung, Bin Gao, Tong Chen, Matthew B. Stone, Alexander I. Kolesnikov, Qingzhen Huang, and Pengcheng Dai, “Topological Spin Excitations in Honeycomb Ferromagnet CrI3,” Phys. Rev. X 8, 041028 (2018).DOI
  17. [17]Lebing Chen, Jae-Ho Chung, Matthew B. Stone, Alexander I. Kolesnikov, Barry Winn, V. Ovidiu Garlea, Douglas L. Abernathy, Bin Gao, Mathias Augustin, Elton J. G. Santos, and Pengcheng Dai, “Magnetic Field Effect on Topological Spin Excitations in CrI₃,” Phys. Rev. X 11, 031047 (2021).
  18. [18]Fengfeng Zhu, Lichuan Zhang, Xiao Wang, Flaviano José dos Santos, Junda Song, Thomas Mueller, Karin Schmalz, Wolfgang F. Schmidt, Alexandre Ivanov, Jitae T. Park, Jianhui Xu, Jie Ma, Samir Lounis, Stefan Blügel, Yuriy Mokrousov, Yixi Su, and Thomas Brückel, “Topological magnon insulators in two-dimensional van der Waals ferromagnets CrSiTe₃ and CrGeTe₃: Toward intrinsic gap-tunability,” Sci. Adv. 7, eabi7532 (2021).DOI
  19. [19]Weiliang Yao, Chenyuan Li, Lichen Wang, Shangjie Xue, Yang Dan, Kazuki Iida, Kazuya Kamazawa, Kangkang Li, Chen Fang, and Yuan Li, “Topological spin excitations in a three-dimensional antiferromagnet,” Nat Phys 14, 1011–1015 (2018).DOI
  20. [20]Bo Yuan, Ilia Khait, Guo-Jiun Shu, F. C. Chou, M. B. Stone, J. P. Clancy, Arun Paramekanti, and Young-June Kim, “Dirac Magnons in a Honeycomb Lattice Quantum XY Magnet CoTiO₃,” Phys. Rev. X 10, 011062 (2020).
  21. [21]M. Elliot, P. A. McClarty, D. Prabhakaran, R. D. Johnson, H. C. Walker, P. Manuel, and R. Coldea, “Order-by-disorder from bond-dependent exchange and intensity signature of nodal quasiparticles in a honeycomb cobaltate,” Nat Commun 12, 3936 (2021).
  22. [22]A. Scheie, Pontus Laurell, P. A. McClarty, G. E. Granroth, M. B. Stone, R. Moessner, and S. E. Nagler, “Dirac Magnons, Nodal Lines, and Nodal Plane in Elemental Gadolinium,” Phys. Rev. Lett. 128, 097201 (2022).DOI
  23. [23]Mohammed J. Karaki, Xu Yang, Archibald J. Williams, Mohamed Nawwar, Vicky Doan-Nguyen, Joshua E. Goldberger, and Yuan-Ming Lu, “An efficient material search for room-temperature topological magnons,” Science Advances 9, eade7731 (2023).DOI
  24. [24]J B Forsyth and P J Brown, “The spatial distribution of magnetisation density in Mn₅Ge₃,” J. Phys.: Condens. Matter 2, 2713–2720 (1990).
  25. [25]N. Maraytta, J. Voigt, C. Salazar Mejía, K. Friese, Y. Skourski, J. Perßon, S. M. Salman, and Th. Brückel, “Anisotropy of the magnetocaloric effect: Example of Mn₅Ge₃,” J. Appl. Phys. 128, 103903 (2020).DOI
  26. [26]Changgan Zeng, Yugui Yao, Qian Niu, and Hanno H. Weitering, “Linear Magnetization Dependence of the Intrinsic Anomalous Hall Effect,” Phys. Rev. Lett. 96, 037204 (2006).pubmed.ncbi.nlm.nih.gov/16486762
  27. [27]R. Kraft, S. Srichandan, G. Fischer, and C. Sürgers, “Anomalous Nernst effect in ferromagnetic Mn₅Ge₃Cₓ thin films on insulating sapphire,” J Appl Phys 128, 033905 (2020).DOI
  28. [28]M Gajdzik, C Sürgers, M.T Kelemen, and H.v Löhneysen, “Strongly enhanced Curie temperature in carbon-doped Mn₅Ge₃ films,” J. Magn. Magn. Mater. 221, 248–254 (2000).DOI
  29. [29]I. Slipukhina, E. Arras, Ph. Mavropoulos, and P. Pochet, “Simulation of the enhanced Curie temperature in Mn₅Ge₃Cₓ compounds,” Appl Phys Lett 94, 192505 (2009).DOI
  30. [30]Toru Moriya, “Anisotropic Superexchange Interaction and Weak Ferromagnetism,” Phys. Rev. 120, 91–98 (1960).DOI
  31. [31]See Supplemental Material for supporting experimental and theoretical information, which includes Refs. [59–73].
  32. [32]L Udvardi and L Szunyogh, “Chiral Asymmetry of the Spin-Wave Spectra in Ultrathin Magnetic Films,” Phys. Rev. Lett. 102, 207204 (2009).pubmed.ncbi.nlm.nih.gov/19519069
  33. [33]Kh. Zakeri, Y. Zhang, J. Prokop, T.-H. Chuang, N. Sakr, W. X. Tang, and J. Kirschner, “Asymmetric Spin-Wave Dispersion on Fe(110): Direct Evidence of the Dzyaloshinskii-Moriya Interaction,” Phys. Rev. Lett. 104, 137203 (2010).pubmed.ncbi.nlm.nih.gov/20481909
  34. [34]Flaviano José dos Santos, Manuel dos Santos Dias, and Samir Lounis, “Nonreciprocity of spin waves in noncollinear magnets due to the Dzyaloshinskii-Moriya interaction,” Phys. Rev. B 102, 104401 (2020).DOI
  35. [35]Inhee Lee, Franz G. Utermohlen, Daniel Weber, Kyusung Hwang, Chi Zhang, Johan van Tol, Joshua E. Goldberger, Nandini Trivedi, and P. Chris Hammel, “Fundamental Spin Interactions Underlying the Magnetic Anisotropy in the Kitaev Ferromagnet CrI₃,” Phys. Rev. Lett. 124, 017201 (2020).DOI
  36. [36]Khalil Zakeri, “Probing of the interfacial Heisenberg and Dzyaloshinskii–Moriya exchange interaction by magnon spectroscopy,” J. Phys.: Condens. Matter 29, 013001 (2016).DOI
  37. [37]Qiang Zhang, Satoshi Okamoto, German D. Samolyuk, Matthew B. Stone, Alexander I. Kolesnikov, Rui Xue, Jiaqiang Yan, Michael A. McGuire, David Mandrus, and D. Alan Tennant, “Unusual Exchange Couplings and Intermediate Temperature Weyl State in Co₃Sn₂S₂,” Phys. Rev. Lett. 127, 117201 (2021).DOI
  38. [38]S. Petit, F. Moussa, M. Hennion, S. Pailhès, L. Pinsard-Gaudart, and A. Ivanov, “Spin Phonon Coupling in Hexagonal Multiferroic YMnO₃,” Phys. Rev. Lett. 99, 266604 (2007).pubmed.ncbi.nlm.nih.gov/18233597
  39. [39]A. S. Sukhanov, M. S. Pavlovskii, Ph. Bourges, H. C. Walker, K. Manna, C. Felser, and D. S. Inosov, “Magnon-polaron excitations in the noncollinear antiferromagnet Mn₃Ge,” Phys. Rev. B 99, 214445 (2019).DOI
  40. [40]G. L. Squires, Introduction to the Theory of Thermal Neutron Scattering (Cambridge Univer- sity Press, 2012).
  41. [41]S. Raymond, N. Biniskos, K. Schmalzl, J. Persson, and T. Brückel, “Total interference between nuclear and magnetovibrational one-phonon scattering cross sections,” J. Phys.: Conf. Ser. 1316, 012018 (2019).DOI
  42. [42]Xiaoou Zhang, Yinhan Zhang, Satoshi Okamoto, and Di Xiao, “Thermal Hall Effect Induced by Magnon-Phonon Interactions,” Phys. Rev. Lett. 123, 167202 (2019).DOI
  43. [43]Gyungchoon Go, Se Kwon Kim, and Kyung-Jin Lee, “Topological Magnon-Phonon Hybrid Excitations in Two-Dimensional Ferromagnets with Tunable Chern Numbers,” Phys. Rev. Lett. 123, 237207 (2019).DOI
  44. [44]Christoph Sürgers, Gerda Fischer, Patrick Winkel, and Hilbert v Löhneysen, “Large topological Hall effect in the non-collinear phase of an antiferromagnet,” Nat. Commun. 5, 3400 (2014).DOI
  45. [45]N. Biniskos, K. Schmalzl, S. Raymond, S. Petit, P. Steffens, J. Persson, and T. Brückel, “Spin Fluctuations Drive the Inverse Magnetocaloric Effect in Mn₅Si₃,” Phys. Rev. Lett. 120, 257205 (2018).DOI
  46. [46]F. J. dos Santos, N. Biniskos, S. Raymond, K. Schmalzl, M. dos Santos Dias, P. Steffens, J. Persson, S. Blügel, S. Lounis, and T. Brückel, “Spin waves in the collinear antiferromagnetic phase of Mn₅Si₃,” Phys. Rev. B 103, 024407 (2021).
  47. [47]N. Biniskos, F. J. dos Santos, K. Schmalzl, S. Raymond, M. dos Santos Dias, J. Persson, N. Marzari, S. Blügel, S. Lounis, and T. Brückel, “Complex magnetic structure and spin waves of the noncollinear antiferromagnet Mn₅Si₃,” Phys. Rev. B 105, 104404 (2022).DOI
  48. [48]F.Q. Zhao, W. Dagula, O. Tegus, and K.H.J. Buschow, “Magnetic-entropy change in Mn₅Ge₃−ₓSiₓ alloys,” Journal of Alloys and Compounds 416, 43–45 (2006).DOI
  49. [49]L.-A. Michez, M. Petit, V. Heresanu, V. Le Thanh, E. Prestat, F. d’Acapito, Q. Ramasse, F. Boscherini, P. Pochet, and M. Jamet, “Unveiling the atomic position of C in Mn₅Ge₃Cₓ thin films,” Phys. Rev. Mater. 6, 074404 (2022).DOI
  50. [50]Paolo Giannozzi, Stefano Baroni, Nicola Bonini, Matteo Calandra, Roberto Car, Carlo Cavazzoni, Davide Ceresoli, Guido L. Chiarotti, Matteo Cococcioni, Ismaila Dabo, Andrea Dal Corso, Stefano de Gironcoli, Stefano Fabris, Guido Fratesi, Ralph Gebauer, Uwe Gerstmann, Christos Gougoussis, Anton Kokalj, Michele Lazzeri, Layla Martin-Samos, Nicola Marzari, Francesco Mauri, Riccardo Mazzarello, Stefano Paolini, Alfredo Pasquarello, Lorenzo Paulatto, Carlo Sbraccia, Sandro Scandolo, Gabriele Sclauzero, Ari P. Seitsonen, Alexander Smogunov, Paolo Umari, and Renata M. Wentzcovitch, “QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials,” J. Phys.: Condens. Matter 21, 395502 (2009).DOI
  51. [51]David Siegfried Georg Bauer, Development of a relativistic full-potential first-principles multiple scattering Green function method applied to complex magnetic textures of nano structures at surfaces, Ph.D. thesis, RWTH Aachen (2014).
  52. [52]https://jukkr.fz-juelich.de/.
  53. [53]Flaviano José dos Santos, Manuel dos Santos Dias, Filipe Souza Mendes Guimarães, Juba Bouaziz, and Samir Lounis, “Spin-resolved inelastic electron scattering by spin waves in noncollinear magnets,” Phys. Rev. B 97, 024431 (2018).DOI
  54. [54]https://doi.ill.fr/10.5291/ILL-DATA.4-01-1682 (2021).
  55. [55]https://doi.ill.fr/10.5291/ILL-DATA.CRG-2853 (2021).
  56. [56]https://doi.ill.fr/10.5291/ILL-DATA.INTER-547 (2021).
  57. [57]https://doi.ill.fr/10.5291/ILL-DATA.CRG-2755 (2021).
  58. [58]M. dos Santos Dias, N. Biniskos, F. J. dos Santos, K. Schmalzl, J. Persson, F. Bourdarot, N. Marzari, S. Blügel, and S. Lounis, “Topological magnons driven by the Dzyaloshinskii-Moriya interaction in the centrosymmetric ferromagnet Mn₅Ge₃,” Materials Cloud Archive 2023.12 (2023), 10.24435/materialscloud:xq-5d.DOI
  59. [74]Jülich Supercomputing Centre, “JURECA: Data Centric and Booster Modules implementing the Modular Supercomputing Architecture at Jülich Supercomputing Centre,” Journal of large-scale research facilities 7, A182 (2021).DOI

Paper details

vlnc:W_c3nhf2f7
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