betaFeedback

WorksPhysics and Astronomy

Absence of magnetic order in RuO2_2: insights from μμSR spectroscopy and neutron diffraction

Abstract

Altermagnets are a novel class of magnetic materials besides ferro- and antiferromagnets, where the interplay of lattice and spin symmetries produces a magnetic order that is staggered both in coordinate as well as momentum space. The metallic rutile oxide RuO2_2, long believed to be a textbook Pauli paramagnet, recently emerged as a workhorse altermagnet when resonant X-ray and neutron scattering studies reported nonzero magnetic moments and long-range collinear order. While experiments on thin films seem consistent with altermagnetic behavior, the origin and size of magnetic moments in RuO2_2 still remain controversial. Here we show that RuO2_2 is nonmagnetic, regardless if as bulk or thin film. Employing muon spin spectroscopy as a highly sensitive probe of local magnetic moments complemented by density functional theory, we find at most 1.4×10−41.4 \times 10^{-4} μBμ_B/Ru in bulk RuO2_2 and at most 7.5×10−47.5 \times 10^{-4} μBμ_B/Ru in epitaxial films. In their essence, these moments reflect the detection limit of our spectrometers and are orders of magnitude smaller than previously reported neutron results, i.e., the moments previously assumed to rationalize altermagnetic behavior. Our own neutron diffraction measurements on RuO2_2 single crystals identify multiple scattering as a likely source for this discrepancy.

Introduction

Altermagnetism, a magnetic state of matter that shares properties with both conventional ferro- and antiferromagnets [1, 2], has recently garnered considerable attention. While altermagnets exhibit the typical ferromagnetic signature of spin split energy bands, they still retain the zero net magnetization that is characteristic of an antiferromagnet [3]. Since one of the first articles on the subject [4], a number of theoretical and experimental studies aimed at exploring fingerprints of altermagnetism have been conducted [5, 6, 7, 8, 9, 10, 11, 12, 13], but a definitive reproducible experimental detection of the altermagnetic state remains elusive. A plurality of experiments, in particular, have been performed on the metallic rutile oxide RuO2_2 [6, 8, 13, 14, 15, 16, 17, 18, 19], where both resonant X-ray scattering [20] and neutron diffraction experiments [21] reported a q=0q=0 antiferromagnetic pattern that is consistent with altermagnetic symmetry constraints. Nonetheless, in the earlier neutron report [21], two caveats were noted: firstly, polarized neutron diffraction refinement resulted in an ordered moment of approximately ∼0.05 μB\sim0.05\,\mu_{\mathrm{B}}/Ru. This value is at least one order of magnitude smaller than the numbers used to interpret subsequent experiments [6, 8], but still two orders of magnitude larger than what was found recently by muon spin relaxation/rotation (μ\muSR) technique on RuO2_2 single crystals [18].

Secondly, electronic structure calculations in the framework of density functional theory (DFT) converge, without additional nudging, to a nonmagnetic solution [15, 21, 22]. Only the addition of a sizable Hubbard UU – somewhat uncommon for good 4d metals – can generate ordered magnetic moments. In this context, it has been suggested that the presence of Ru vacancies may lead to magnetism at a lower and more realistic Hubbard UU [22] than that used in Refs. [4, 23].

In view of the popularity and conceptual importance of RuO2_2 as a prototypical altermagnet, it is essential to verify the existence and strength of the magnetic order in this compound. We thus have revisited the question about ordered moments in RuO2_2 and performed careful neutron diffraction experiments employing state-of-the art time-of-flight detection. Surprisingly, we found that the previously reported and alleged magnetic (1,0,0)(1,0,0) reflection [21] is likely related to a multiple scattering artifact rather than to a magnetic origin. Further, we employed the highly sensitive μ\muSR technique, complemented by DFT calculations of the muon stopping sites and hyperfine fields, and determined an upper bound on magnetic moments of 1.4×10−4 μB1.4 \times10^{-4}\,\mu_{\mathrm{B}}/Ru in the bulk and 7.5×10−4 μB7.5 \times10^{-4}\,\mu_{\mathrm{B}} in epitaxial thin films of RuO2_2, consistent with a recent bulk value obtained by an independent group [18].

These moments are two orders of magnitude smaller than what was reported by the previous neutron study [21], and essentially reflect the detection limit of our spectrometer. Hyperfine field calculations in stoichiometric and Ru-vacant RuO2_2 further exclude any fortuitous cancellation of hyperfine fields at the muon stopping site. We thus conclude that RuO2_2 exhibits no long-range magnetic order, neither in the bulk nor in epitaxial thin films, and the altermagnetic signatures reported in multiple earlier publications [6, 8, 11, 13, 16, 17] likely have extrinsic origin.

Experimental Results

RuO2_2 Samples

To shed light on the magnetic properties of RuO2_2, we investigated three different types of samples: (i) Millimeter sized RuO2_2 crystals that were grown by chemical vapor transport at the Crystal Growth Facility of the EPFL in Lausanne, Switzerland, following a recipe described in Ref. [24]; (ii) 11 nm thick RuO2_2 epitaxial thin films grown on TiO2_2(110) by pulsed laser deposition at the Institute of Physics, Würzburg, Germany, following our recipe developed in Ref. [8]; and (iii) 99.9% pure (trace metals basis) RuO2_2 powder, purchased from Sigma-Aldrich that was measured without further preparation steps. The RuO2_2 crystalline structure was ensured for all three sample types by x-ray diffraction techniques before the μ\muSR and/or neutron diffraction experiments.

Neutron diffraction

Reassessing the results of the earlier study [21], we first performed single crystal neutron diffraction on the RuO2_2 crystals. Hereby, we employed the large pixelated detector array of the state-of-the-art time of flight diffractometer WISH at the ISIS Neutron and Muon Source of the STFC Rutherford Appleton Laboratory [25]. Measurements at 1.5 K on four RuO2_2 crystals, three of them identified as multi grain, one of them identified as single grain, all led to consistent conclusions. The crystals were aligned in the (h,0,l)(h,0,l) horizontal scattering plane, a large portion of which can be measured in a single shot using the time-of-flight dimension, with the possibility to rotate the crystals about the vertical b∗b^{*} direction.

As discussed above, altermagnetism in RuO2_2 implies a q=0\mathbf{q}=0 propagation vector and antiparallel magnetic coupling of the two Ru sites in the primitive unit cell, with moments polarized along the crystal c-axis. The structure is expected to result in a strong magnetic scattering to the structurally forbidden (1,0,0)(1,0,0) reflection [21]. This is due to the fact that the two magnetic sites sum over in the corresponding structure factor and the moments are perpendicular to the scattering vector. In addition, among all

possible magnetic reflections with q=0\mathbf{q}=0, the (1,0,0)(1,0,0) one has the largest dd-spacing and therefore the largest value of the magnetic form factor. We thus focused our attention at the (1,0,0)(1,0,0) reflection and measured it at different scattering angles 2Θ2\Theta, thereby using different neutron wavelengths λ\lambda, and found a strong dependence on the scattering geometry. At some scattering angles, the reflection was not observed even after ten hours of statistics, in spite of a good optimisation with respect of the neutron flux, whereas at some other angles (mainly low angles) it was detectable within a few minutes with a poorer flux optimisation (Figure 1). This behaviour is not typical for a real Bragg peak and is known as Renninger effect [26, 27, 28], caused by multiple scattering processes (often referred to as double scattering).

Neutron diffraction data and reciprocal-lattice illustration

Figure 1. Neutron diffraction data collected on RuO2_2 single crystal at T=1.5T=1.5 K with the (h,0,0)(h,0,0) reflections at the scattering angle 2Θ=71∘2\Theta=71^\circ (a) and 2Θ=32∘2\Theta=32^\circ (b). The counting time was 10 hours and 15 minutes, respectively. The data demonstrate 1D cuts of the reciprocal lattice along the (h,0,0)(h,0,0) direction. The inset of panel (a) shows a zoomed region at a vicinity of the (1,0,0)(1,0,0) reflection. The inset of panel (b) shows a simple two-dimensional illustration of the double diffraction, taking place when the Ewald sphere intersects both the K1=(1,0,0)\mathbf{K}_1=(1,0,0) and the K2=(2,1,0)\mathbf{K}_2=(2,1,0) reciprocal lattice points. The beam, first scattered by the (2,1,0)(2,1,0) planes in the direction k2\mathbf{k}_2, is then scattered again by the (−1,−1,0)(-1,-1,0) planes in the direction k1\mathbf{k}_1, where the (1,0,0)(1,0,0) scattering is expected. The intersection of the (h,k,0)(h,k,0) reciprocal plane and the Ewald sphere is shown by the red circle for the first scattering process (from the (2,1,0)(2,1,0) planes) and by the blue circle for the second scattering process (by the (−1,−1,0)(-1,-1,0) planes).

This can happen when two reciprocal lattice nodes, K1K_1 and K2K_2, are present simultaneously on the surface of the Ewald sphere. Let us denote the Miller indices of the corresponding families of lattice planes as (h1,k1,l1)(h_1,k_1,l_1) and (h2,k2,l2)(h_2,k_2,l_2), then the corresponding Laue conditions are:

k1−k0λ=K1andk2−k0λ=K2,\frac{\boldsymbol{k}_1-\boldsymbol{k}_0}{\lambda}=\boldsymbol{K}_1 \quad\text{and} \quad\frac{\boldsymbol{k}_2-\boldsymbol{k}_0}{\lambda}=\boldsymbol{K}_2,

where k0\boldsymbol{k}_0 is the unit vector along the incident beam and k1\boldsymbol{k}_1, k2\boldsymbol{k}_2 are directions of the diffracted beam. By subtracting the second equation from the first, one obtains

k1−k2λ=K1−K2.\frac{\boldsymbol{k}_1-\boldsymbol{k}_2}{\lambda}=\boldsymbol{K}_1-\boldsymbol{K}_2.

This indicates that the beam diffracted by the (h1,k1,l1)(h_1,k_1,l_1) in the direction k1\boldsymbol{k}_1 overlaps with a double-scattering beam, first by the plane (h2,k2,l2)(h_2,k_2,l_2) in the direction k2\boldsymbol{k}_2 and then by the plane (h1−h2,k1−k2,l1−l2)(h_1-h_2,k_1-k_2,l_1-l_2) in the direction k1\boldsymbol{k}_1 (see inset of Figure 1(b) for an illustration). If (h1,k1,l1)(h_1,k_1,l_1) is a systematically absent structural reflection, then the Renninger effect may result in an apparent violation of the reflection conditions [29] or can be interpreted as magnetic scattering.

As the radius of the Ewald sphere is defined by the wavelength λ\lambda of the incident and scattered neutrons, the double scattering process is expected to be strongly λ\lambda-dependent and should not take place if the necessary conditions are not satisfied. Observation of the doubly scattered beam will thus only exist in a finite angular range due to beam divergence and finite extension of the Bragg spots in reciprocal space, controlled by sample quality and instrument resolution. At low scattering angles, where the Bragg law selects short wavelengths, the radius of the Ewald sphere is big, and the sphere has a large surface in reciprocal space. Further, instrumental resolution also quickly degrades with decreasing scattering angles. This strongly increases the probability that two (or more) reciprocal spots hit the sphere producing the multiple scattering. Consequently, the Renninger effect is the most natural explanation of the presence of the (1,0,0)(1,0,0) reflection in RuO2_2 and its strong dependence on the scattering angle (or equivalently the wavelength of neutrons). In the context of the present study, the key experimental observation is the lack of the (1,0,0)(1,0,0) reflection in the well-optimised (in respect of the neutron flux) scattering geometry. After ten hours of measurements, the statistic of the data allows us to rule out the q=0\boldsymbol{q}=\boldsymbol{0} magnetic ordering with moment size bigger than 0.011μB0.011\mu_{\mathrm{B}}, as previously suggested in Ref. [21]. Moreover, we explored a large portion of the reciprocal space available in the time-of-flight scattering experiment and have not found any evidence of magnetic ordering either with q=0\boldsymbol{q}=\boldsymbol{0} or with some other propagation vector. This strongly supports the non-magnetic nature of the RuO2_2 crystals.

Muon spin resonance (μSR)

For further insights on their size, we employed the μ\muSR technique as a highly sensitive probe of static local magnetic moments. The μ\muSR experiments on the powder and single crystalline samples were conducted at the General Purpose Surface-Muon (GPS) Instrument [30] at the Swiss Muon Source piM3.2 beamline of the Paul Scherrer Institute (PSI). A 4^4He continuous-flow cryostat was used to perform μ\muSR measurements between 5 K and 290 K. The thin films were studied in the Low-Energy Muons Facility (LEMu) at the μ\muE4 beamline at PSI [31, 32]. For these measurements, the thin-film samples were glued onto the cold finger of a 4^4He flow cryostat using silver paint. The data analysis for all the μ\muSR spectra was conducted using the musrfit package [33].

In a μ\muSR measurement, fully polarized muons are implanted into the sample. Muons decay with a lifetime of 2.2 μ\mus, emitting a positron preferentially in the direction of the spin polarization at the time of decay. The spatial distribution of these positrons is monitored as a function of time using plastic scintillating detectors placed around the studied sample. Therefore, the asymmetry in the decay of positrons on opposite sides of the sample, A(t)A(t), is proportional to the polarization along the corresponding axis. In practice, A(t)A(t) is calculated via

A(t)=NU−αUDNDNU+αUDND,(1)A(t)=\frac{N_{\mathrm{U}}-\alpha_{\mathrm{UD}}N_{\mathrm{D}}}{N_{\mathrm{U}}+\alpha_{\mathrm{UD}}N_{\mathrm{D}}}, \tag*{(1)}

where NUN_{\mathrm{U}} and NDN_{\mathrm{D}} are the number of positrons detected as a function of time in detectors placed above (U) and below (D) the sample, respectively. αUD\alpha_{\mathrm{UD}} is a geometric correction factor to account for the different efficiencies of the U and D detectors.

In the bulk μ\muSR measurements (GPS), we performed primarily measurements on the powder sample in zero applied external field (ZF) and in weak transverse field (TF - transverse to the initial muon spin polarization). A few measurements obtained from the single crystal gave results identical to the powder measurements. Therefore, we hereafter focus only on the powder measurements. In Figure 2(a), we show typical asymmetry curves measured at three different temperatures. We observe no significant relaxation or depolarization at any temperature and no missing asymmetry, confirming that the local magnetic fields sensed by the implanted muons are extremely small. Only a very small increase in the relaxation rate around ∼150\sim150 K is observed. Similar results were obtained from μ\muSR measurements while applying a TF of 5 mT. A few representative asymmetry spectra are depicted in Figure 2(b). The applied field in this case leads to a Lamor precession of the muons’ spin polarization, and therefore, an oscillation in the asymmetry spectra. The initial amplitude of these oscillations is proportional to the paramagnetic volume fraction of the sample.

Asymmetry measured in zero and transverse magnetic fields at various temperatures

Figure 2. The asymmetry measured in (a) zero and (b) transverse magnetic field at various temperatures in bulk RuO2_2.

Just as in the case of the ZF measurements, no significant temperature dependence was observed ruling out any static magnetic order in RuO2_2 (neither short nor long range). To analyse the data, we fit the asymmetries using an exponentially damped oscillation, A(t)=A0exp⁡(−λμt)cos⁡(ωt+φ)A(t)=A_0\exp(-\lambda_\mu t)\cos(\omega t+\varphi). Both ZF and TF data were fitted using a temperature independent initial asymmetry A0A_0 and phase φ\varphi, while the relaxation rate λμ\lambda_\mu and frequency ω\omega were temperature dependent. For the ZF measurements, we use ω=0\omega=0 and φ=0\varphi=0. The relaxation rates obtained from these fits are shown in Figure 3. We note a very small temperature dependence in λμ\lambda_\mu measured in both ZF and TF. In particular, in the ZF measurements we observe a peak around ∼150\sim150 K, which corresponds to the small enhancement in relaxation seen in the raw spectra of Figure 2(a). In contrast, in the TF measurements, λμ\lambda_\mu increases slightly below ∼200\sim200 K but saturates below ∼150\sim150 K. This small temperature dependence and extremely small relaxation rate, however, cannot be attributed to static magnetic order in RuO2_2, which would generally result in much larger values of λμ\lambda_\mu.

Temperature dependence of the muon relaxation rate

Figure 3. Temperature dependence of the muon relaxation rate in bulk RuO2_2 for zero-field and transverse-field geometry.

In order to probe a possible magnetic order (or static magnetic moments) near grain boundaries, interface, or other defects, we performed low energy μ\muSR (LE-μ\muSR) measurements on our 11 nm thick epitaxial RuO2_2(110) thin films deposited on top of a TiO2_2 rutile substrate.

the implantation energy of the muons was varied between 1 keV and 12 keV, resulting in an implantation depth (probing depth) between ∼5\sim5 nm and ∼100\sim100 nm (see Figure 4). We performed LE-μ\muSR measurements in TF of 10 mT as

Calculated muon stopping profile as a function of depth

Figure 4. Calculated muon stopping profile as a function of depth for the different implantation energies in an 11 nm thick RuO2_2 on TiO2_2 substrate.

a function of temperature at two implantation energies: E=1.5E=1.5 keV, where most of the muons stop in the RuO2_2 layer; and E=12E=12 keV, where practically all the muons stop in the TiO2_2 substrate. These energies were selected based on TRIM.SP calculations [34, 35] (Figure 4). Like the bulk data, the obtained spectra were fitted to an exponentially damped oscillation. The resulting damping rate, λμ\lambda_\mu, is plotted as a function of temperature for both implantation energies in Figure 5.

Temperature dependence of the muon relaxation rate in a film of 11 nm RuO₂ on TiO₂

Figure 5. Temperature dependence of the muon relaxation rate in a film of 11 nm RuO₂ on TiO₂ measured in 10 mT transverse-field. At E=1.5E = 1.5 keV most of the muons stop in the RuO₂ layer while at E=12E = 12 keV they stop in the substrate.

We find that λμ\lambda_\mu in the RuO₂ layer, though generally higher than what we found in the bulk, remains quite small. This rules out the presence of large magnetic fields in these films. We also note that λμ\lambda_\mu in RuO₂ is much larger than that measured in the TiO₂ substrate. Therefore, the enhanced relaxation rate in the RuO₂ layer cannot be attributed to the fraction of muons stopping in the substrate. Moreover, the gradual increase in λμ\lambda_\mu with decreasing temperature rules out background contribution due to back-scattered muons [36]. Instead, we believe that the observed increase in relaxation compared to the bulk is due to the presence of somewhat larger magnetic moments in films of RuO₂. The origin of these magnetic moments may be defects or vacancies, which might be more abundant in the films.

Discussion

In order to better interpret the results of the μ\muSR measurements, we performed first-principles calculations based on spin-polarized density functional theory (DFT), and enforcing a magnetic moment on the Ru atoms by adding a large effective Hubbard UU [22], as also done in other computational studies [4, 23]. To emulate the muon stopping site, we used the standard protocol described in Ref. [37]. As muon and proton are identical from the point of view of adiabatic DFT calculations, we used a proton in place of the muon, thinking of the muon as a light proton [38]. The total number of electrons was kept unchanged. This proton was implanted into a sufficiently large (3×3×3)(3 \times3 \times3) supercell to avoid interaction with its replicas in the neighbouring cells [37]. Otherwise, we performed DFT calculations

with the projector augmented-wave (PAW) method [39, 40], as implemented in VASP [41, 42, 43, 44, 45, 46]. As exchange-correlation functional we considered the generalized gradient approximation (GGA) [47] and added an effective Hubbard Ueff=U−J=1.4U_{\mathrm{eff}} = U - J = 1.4 eV using the rotational invariant GGA+U introduced in Ref. [48] to ensure a sizable magnetic moment on the Ru atoms and a non-zero hyperfine field. The cutoff energy for the plane-wave basis was set to 300 eV.

RuO₂ crystallizes in the rutile structure, space group P4₂/mnm, as depicted in Figure S1. This structure, which we here call “pristine”, serves as the basis for DFT calculations involving a muon. The calculated magnetic moment on the Ru atoms, for Ueff=1.4U_{\mathrm{eff}} = 1.4 eV, was 0.866 μB\mu_{\mathrm{B}}/Ru. This magnetic moment was used in the interpretation of the experiments done in Refs. [5, 6, 8, 11, 13, 20, 21, 49]. In addition to the pristine structure, we emulated the effect of potential Ru vacancies by removing one Ru from this supercell (i.e., 1.8 % of Ru vacancies as reported in at least one case [50]), which enhances the tendency to magnetism [22]. We also checked the stability of the results against adding spin-orbit coupling (SOC), and found that both in the pristine case and in the presence of a Ru vacancy, SOC introduces negligible effects, in agreement with the previous literature [23, 51]. To optimize computing time we thus performed our calculations without SOC.

Crystal structure of RuO2 and Coulomb-potential isosurfaces

Figure S1. (a): Crystal structure of RuO2: Ru atoms are represented in gray while O atoms are in red. The projection on the abab plane of the rotated RuO6 octahedra is shown in shaded gray (note a 90° rotation of the octahedra around two sublattice Ru). The square shows the unit cell used in calculations without a muon. (b): Top view of the isosurface of the Coulomb potential with value −11.2 eV calculated with Ueff=1.4U_{\mathrm{eff}} = 1.4 eV, indicating the initial approximation for muon stopping sites. (c): Same, a side view

First, we identified probable stopping sites for a muon by looking at the electrostatic potential in the pristine unit cell, as shown in Figure 6. This electrostatic potential is defined as the sum of the ionic potential and the Hartree potential. As Figure 6 shows, and Figure 7 further highlights, the electrostatic potential suggests the presence of channels along the cc direction, where we expect the muon to be initially attracted to. Within a channel, the electrostatic potential minima are the points farthest away from the Ru ions. They are indicated in Figure 7 by little orange balls.

Crystal structure and electrostatic potential maps for different cuts along c

Figure 6. Crystal structure and electrostatic potential maps for different cuts along cc in units of the lattice parameter. The electrostatic potential is represented by the color map from red to blue. The purple shade indicates regions where the (negative) electrostatic potential is the largest, i.e., the likely muon stopping sites. The color map is limited to the unit cell.

Initial approximation of the stopping sites of the muon

Figure 7. Initial approximation of the stopping sites of the muon, represented by orange balls, where the electrostatic potential is highest. The color map indicates that, the maxima of the potential occurs at the furthest distance from Ru atoms. These positions are the most likely stopping sites along the channel, as seen in the minima of the energy in Figure 9.

Taking a closer look at Figure 7, and focusing on one color, i.e., one value of the electrostatic potential, we can construct an equipotential surface that is oriented along the cc axis (see Figure S1 for an example). The tube-like shape indicates the above-mentioned channel to which a muon will be immediately attracted. Full optimization of the muon, Ru and O positions further shows that the muons displace away from the axis of the tube to form a bond with one of the nearby oxygens (see Figure 8), as is typical in ionic crystals [37]. Although along the cc-axis the radii of the tubes diminish in the region between two Ru layers, there is a possibility of a thermally-activated muon diffusion along these channels, which can affect the muon depolarization by averaging the local magnetic fields. In order to check this, we performed constrained optimization, fixing the zz coordinate of the muon and relaxing all other internal parameters. Since this is a rather time-consuming procedure, we performed all calculations on an 3×3×43 \times3 \times4 kk-mesh, starting from a low kk-mesh density and increasing it as the calculations converged. The result is shown in Figure 9. We find the minima labeled 1-3 in Figure 7 to be at least 600 meV deep. This barrier height ensures that muons do not diffuse between these stopping sites under our experimental conditions.

Final stopping sites of the muon

Figure 8. Final stopping sites of the muon (represented in orange or blue) for different positions along the cc axis. In blue we specify those positions with the lowest energy, i.e. the most likely stopping sites.

Energy calculated for a diffusion path of a muon along the c direction

Figure 9. Energy calculated for a diffusion path of a muon along the cc direction from one Ru spin-equivalent plane to the next, in fractional coordinates with Ueff=1.4U_{eff}=1.4 eV. The muon’s position along aa and bb was relaxed. The most likely stopping sites are in the planes of the Ru atoms, as labeled in Figure 7. The line is a guide to the eye. Please note that the energy landscape should be periodic, and the observed deviations on the order of ±50\pm50 meV serve as an independent gauge of reliability of our calculations.

Beyond the pristine system, long range static magnetic order in RuO2_2 was suggested to be induced by hole doping due to Ru vacancies [22]. To account for this possibility, we conducted two sets of calculations: with and without a Ru vacancy in the 3×3×33 \times3 \times3 supercell. In the former case, we considered both the same stopping site as in a pristine crystal, again with full re-optimization, and a second case where the muon was put near a vacancy, followed by additional atomic relaxation (see Supplementary Information). As seen in Figure S3, the electrostatic potential increases at the Ru vacancy (in absolute values). When placing a muon nearby the vacancy, however, it converged to a position further away from the Ru atoms than in the pristine case, revealing no new energetically favorable stopping sites.

Top view of a color map of the electrostatic potential of a pristine crystal with a Ru vacancy

Figure S3. Top view of a color map of the electrostatic potential of a pristine crystal with a Ru vacancy. Here, the UeffU_{\mathrm{eff}} is 1.4 eV. A muon, represented in orange, was placed nearby the RuRu vacancy, shown in white, to check for possible new stopping sites. The arrow points to the muon, which is shown in the final stopping site, where it sits near an OO atom and creates a bond with it.

The hyperfine field acting on the muon was calculated for both the structures with and without Ru vacancies. For different UeffU_{\mathrm{eff}} in the pristine case, the values of the hyperfine field and the magnetic moment of the Ru atoms are shown in Figure 10. As expected in a good metal, we find that the hyperfine field is dominated by the Fermi-contact term due to the finite polarization of the electronic cloud, and the dipole field is much smaller. The hyperfine field shows a non-monotonic dependence on the applied Hubbard UeffU_{\mathrm{eff}} and the resulting moment, but never drops below ≈1\approx1 T.

Figure 10. Hyperfine field and magnetic moment on Ru as a function of Hubbard UeffU_{eff}. The Fermi-contact term provides the dominant contribution to the hyperfine field, being the dipole field contribution negligible. For UeffU_{eff} values smaller than the plotted ones, the Ru atoms have no magnetic moment. The lines are guides to the eye.

The calculations with a vacancy did not generate a new stopping site, and the calculated hyperfine field was similar to that for a pristine crystal. That is to say, when the calculations are artificially nudged into a magnetic solution by using a large UeffU_{\mathrm{eff}}, the resulting Fermi-contact hyperfine field is as few orders of magnitude stronger than admissible by the experimental constraints. Therefore, we conclude that the nonmagnetic calculations without UU (as typical for 4d metals) describe our samples better. Note that in the latter case (and in fact for any Ueff≲0.6U_{eff}\lesssim0.6 eV) the calculations converge to a fully nonmagnetic state, where not only the net magnetic moment around Ru is zero, but also the local spin density is zero everywhere.

Given the large Fermi-contact hyperfine field, we can confidently exclude the presence of anti(alter-)magnetic order. However, even if we exclude metallic electrons at the Fermi level from consideration, the dipole field induced by magnetic moments on Ru already allows us to exclude any static magnetic order. Indeed, the knowledge of the muon stopping site can be used to estimate the relationship between the muon relaxation rate λμ\lambda_\mu and the magnitude of magnetic moments mm in RuO2_2. Due to the small magnitude of λμ\lambda_\mu and exponential relaxation of the asymmetry, we assume a dilute arrangement of randomly oriented static moments in our model. Even if the large Fermi-contact field was averaged out, these random moments (located on Ru sites or on magnetic impurities) would still produce a dipolar magnetic field distribution that leads to a relaxation/damping of the polarization of implanted muons.

Considering the exponential relaxation observed experimentally, we assume that the muons in RuO2_2 experience a distribution of static magnetic fields f(∣B∣)f(|B|) that can be described by a Lorentzian function,

f(∣B∣)=1π2Λ(Λ2+B2)24πB2,(2)f(|B|)=\frac{1}{\pi^{2}}\frac{\Lambda}{(\Lambda^{2}+B^{2})^{2}}4\pi B^{2}, \tag*{(2)}

where Λ\Lambda is the half-width at half-maximum. In a physical context, Λ\Lambda can be regarded as the magnitude of the local field BlocB_{\mathrm{loc}}, which can be written in terms of the relaxation rate λμ\lambda_\mu in ZF,

Bloc≈Λ≈34λμγμ,(3)B_{\mathrm{loc}}\approx\Lambda\approx\frac{3}{4}\frac{\lambda_\mu}{\gamma_\mu}, \tag*{(3)}

where γμ=851.615\gamma_\mu=851.615 MHz/T is the muon’s gyromagnetic ratio. Therefore, we can put an upper limit on the local static field experienced by muons of 1.3×10−51.3\times10^{-5} T in the bulk (λμ∼0.015\lambda_\mu\sim0.015 μ\mus−1^{-1}) and 7.0×10−57.0\times10^{-5} T in the films (λμ∼0.08\lambda_\mu\sim0.08 μ\mus−1^{-1}). Assuming that these fields stem from a dipolar interaction between the muon and a neighbouring Ru atom, they can be translated into upper limits on the magnetic moment of 1.4×10−4μB1.4\times10^{-4}\mu_B in bulk and 7.5×10−4μB7.5\times10^{-4}\mu_B in films.

Conclusions

Both neutron scattering experiments on single crystals and μ\muSR experiments on single crystals, powder and thin films conclusively show that the magnetic moments detected in RuO2_2 are several orders of magnitude smaller than previously suggested by polarized neutron scattering (0.05μB0.05\mu_B) [21], and essentially comparable to the detection limit of the individual spectrometer facility. In particular, both the absence of a magnetic Bragg peak in neutron scattering and the measured damping rates of muon spin polarization in our μ\muSR experiments, present a clear evidence of the non-magnetic ground state of RuO2_2. This is in agreement with the DFT+UU results calculated for Hubbard UU values that are realistic for the RuO2_2 compound [22]. Our findings strongly suggest that the RuO2_2 altermagnetic signatures reported in multiple previous studies likely have extrinsic origin.

Acknowledgments

The authors thank Helena Reichlova, Dominik Kriegner, Libor Šmejkal, Thomas Jungwirth and Huibo Cao for helpful discussions. Funding support came from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy through the Würzburg-Dresden Cluster of Excellence on Complexity and Topology in Quantum Matter ct.qmat (EXC 2147, Project ID 390858490) and through the Collaborative Research Center SFB 1170 ToCoTronics (Project ID 258499086). L.G.G. and R.V. further gratefully acknowledge support by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) for funding through project TRR 288 — 422213477 (project B05). This work is partially based on experiments performed at the Swiss Muon Source (Sμ\muS), Paul Scherrer Institute, Villigen, Switzerland. I.I.M. was supported by the Army Research Office under Cooperative Agreement Number W911NF-22-2-0173. He also acknowledges Heraeus Foundation for supporting his visits to the University of Frankfurt.

Author contribution

P.K. and L.G.G. contributed equally. I.I.M., R.V., S.M. conceived the project. P.K. and S.M. synthesized and characterized the RuO2_2 thin films. D.K. and P.M. performed the neutron scattering experiments and analysis. P.K., S.M., Z.S., A.S. and T.P. performed the μ\muSR experiments and analysis and L.G.G. and I.I.M. performed the DFT+U+μ\mu calculations. All authors contributed to the discussions and the writing of the paper.

References

Supplementary material

RuO2_2 crystallizes in the rutile crystal structure, space group P42/mnmP4_2/mnm. The unit cell of the crystal is shown in the left of Figure S1, such that the unit cell is formed by two Ru atoms, to account for the spin polarized component, and four oxygen atoms. The octahedra formed by the oxygens around the Ru atoms are shown in light gray color, exposing the four-fold rotation from one Ru sub-lattice to the other. We used this structure to study the behavior of the local potential in the pristine case. The relaxed lattice parameters used are a=b=4.533132a=b=4.533132 Å and c=3.124167c=3.124167 Å.

The middle and right panels of Figure S1 show different perspectives of an isosurface of the electrostatic potential within the unit cell. The top view of the middle panel shows the maxima of the potential to occur in between the Ru rows, outside of the octahedra shown in the left panel. Moreover, there is a slight tilt in the isosurface due to the coordination of the nearby oxygens. The right panel shows the tube-like shape of the isosurface along the crystal cc-axis, as well as the supercell used in the calculations. The muon decays somewhere along this tube. The reason to use a supercell is such that the interaction between a muon and its replica in the neighbouring cell is negligible [37].

Figure S2 shows the energy of the system, compared to the ground state energy, as well as the charge captured by the muon for different positions along the zz axis, in fractionalized coordinates. The minima of the energy occur when closest to the oxygens. Multiple calculations were performed in order to avoid finding a local minimum of the potential energy surface. On the other hand, the muon capturing charge from the environment is a possible occurrence [37].

Figure S2. Energy of the system (in meV) -left axis- and charge captured by the muon (in electron units) -right axis- for different positions of the muon along the cc direction (the positions are given in fractionalized units of the cc lattice parameter). The charge the muon is captured from the bond it creates with the nearest O atom, as shown in Figure 8. The closer to the oxygen atom the muon is, the more screening the muon has and therefore the captured charge is lower, which occurs when the energy of the system is the lowest as well.

The evolution of the charge captured by the muon shows that, when it is closest to the oxygens (see Figure 8), the electron screening is higher and hence there is a lower charge captured by the muon.

Figure S3 shows a supercell of 3×3×33 \times3 \times3 with one Ru vacancy, represented in white centered in the unit cell. This vacancy accounts for ≈2%\approx2\% of vacancies, as reported in Ref. [50]. A color map is used to represent the electrostatic potential, with the previously seen absolute maxima in between the octahedra. Moreover, the new electrostatic potential due to the Ru vacancy is shown. Note the breaking of C4_4 symmetry from the oxygens that would have been bonded to the missing Ru. Those oxygens are the ones with the largest displacement, 0.0953 Å, towards the vacancy. Furthermore, around the vacancy, the electrostatic potential indicates a possible new stopping site. Hence, a muon was placed nearby the vacancy, more specifically in the (001) plane with the vacancy, to study the energy of the system and compare it to the previously found stopping sites to determine if it was likely the muon ended there. The muon was placed in this plane given the knowledge of its preference to bond with the oxygen and the previously minima energy found in the pristine case. Note that the final stopping site of the muon was not at the absolute maxima of the electrostatic potential, due to this tendency to bond with an OO atom. However, no new possible stopping sites that were energetically favorable were found. Therefore, the distance between the most likely stopping sites of the muon found in Figure 8 and the closest Ru atom was used to analyze the data of the μ\muSR experiments.

References

  1. [1]Mazin, I. Editorial: Altermagnetism - a new punch line of fundamental magnetism. Phys. Rev. X 12, 040002 (2022).DOI
  2. [2]Šmejkal, L., Sinova, J. & Jungwirth, T. Emerging research landscape of altermagnetism. Phys. Rev. X 12, 040501 (2022).arxiv.org/abs/2204.10844
  3. [3]Cuono, G., Sattigeri, R. M., Skolimowski, J. & Autieri, C. Orbital-selective altermagnetism and correlation-enhanced spin-splitting in strongly-correlated transition metal oxides. Journal of Magnetism and Magnetic Materials 586, 171163 (2023).DOI
  4. [4]Šmejkal, L., González-Hernández, R., Jungwirth, T. & Sinova, J. Crystal time-reversal symmetry breaking and spontaneous hall effect in collinear antiferromagnets. Science Advances 6, eaaz8809 (2020).DOI
  5. [5]Samanta, K. et al. Crystal hall and crystal magneto-optical effect in thin films of SrRuO₃. Journal of applied physics 127, 21 (2020).DOI
  6. [6]Feng, Z. et al. An anomalous hall effect in altermagnetic ruthenium dioxide. Nature Electronics 5, 735–743 (2022).DOI
  7. [7]Gonzalez Betancourt, R. D. et al. Spontaneous Anomalous Hall Effect Arising from an Unconventional Compensated Magnetic Phase in a Semiconductor. Physical Review Letters 130, 036702 (2023). 2112.06805.
  8. [8]Tschirner, T. et al. Saturation of the anomalous Hall effect at high magnetic fields in altermagnetic RuO₂. APL Materials 11 (2023). 2309.00568.
  9. [9]Liu, Q. et al. Inverse Magnetocaloric Effect in Altermagnetic 2D Non-van der Waals FeX (X = S and Se) Semiconductors. Advanced Functional Materials 2402080, 1–10 (2024).DOI
  10. [10]Reimers, S. et al. Direct observation of altermagnetic band splitting in CrSb thin films. Nature Communications 15, 2116 (2024). 2310.17280.
  11. [11]Zhou, X. et al. Crystal Thermal Transport in Altermagnetic RuO₂. Physical Review Letters 132, 056701 (2024). 2305.01410.DOI
  12. [12]Hariki, A. et al. X-Ray Magnetic Circular Dichroism in Altermagnetic α-MnTe. Physical Review Letters 132, 176701 (2024). 2305.03588.DOI
  13. [13]Guo, Y. et al. Direct and Inverse Spin Splitting Effects in Altermagnetic RuO₂. Advanced Science 2400967, 1–9 (2024).DOI
  14. [14]Occhialini, C. A. et al. Local electronic structure of rutile RuO₂. Physical Review Research 3, 033214 (2021). 2108.06256.arxiv.org/abs/2108.06256
  15. [15]Liang, Q., Bieberle-Hütter, A. & Brock, G. Antiferromagnetic RuO₂: a stable and robust oer catalyst over a large range of surface terminations. The Journal of Physical Chemistry C 126, 1337–1345 (2022).DOI
  16. [16]Karube, S. et al. Observation of Spin-Splitter Torque in Collinear Antiferromagnetic RuO₂. Physical Review Letters 129, 137201 (2022). 2111.07487.DOI
  17. [17]Fedchenko, O. et al. Observation of time-reversal symmetry breaking in the band structure of altermagnetic RuO₂. Science Advances 10, eadj4883 (2024).DOI
  18. [18]Hiraishi, M. et al. Nonmagnetic Ground State in RuO₂ Revealed by Muon Spin Rotation. Physical Review Letters 132, 166702 (2024). 2403.10028.DOI
  19. [19]Prateek, K. et al. Fabrication and properties of lateral Josephson junctions with a RuO₂ weak link. Superconductor Science and Technology 37, 035020 (2024).DOI
  20. [20]Zhu, Z. H. et al. Anomalous antiferromagnetism in metallic RuO₂ determined by resonant x-ray scattering. Physical review letters 122, 017202 (2019).
  21. [21]Berlijn, T., Snijders, P. C., Delaire, O., Zhou, T. A., H. D.and Maier & Cao, H. B. Itinerant antiferromagnetism in RuO₂. Physical review letters 118, 077201 (2017).
  22. [22]Smolyanyuk, A., Mazin, I. I., Garcia-Gassull, L. & Valentí, R. Fragility of the magnetic order in the prototypical altermagnet RuO₂. Physical Review B 109, 134424 (2024).
  23. [23]Ahn, K.-H., Hariki, A., Lee, K.-W. & Kuneš, J. Antiferromagnetism in RuO₂ as d-wave pomeranchuk instability. Physical Review B 99, 184432 (2019).DOI
  24. [24]Schäfer, H., Schneidereit, G. & Gerhardt, W. Zur Chemie der Platinmetalle. RuO₂ Chemischer Transport, Eigenschaften, thermischer Zerfall. Zeitschrift für anorganische und allgemeine Chemie 319, 327–336 (1963). [25] Chapon, L. C. et al. Wish: The new powder and single crystal magnetic diffractometer on the second target station (2011).
  25. [26]Renninger, M. Umwegnanegung, eine bisher unbeachtete Wechselwirkungserscheinung bei Raumgitterinterferenzen. Zeitschrift für Physik 106, 141–176 (1937).DOI
  26. [27]Speakman, J. C. The renninger effect – an example and its implication. Acta Crystallographica 18, 570–571 (1965).DOI
  27. [28]Giacovazzo, C. et al. Fundamentals of Crystallography (Oxford University Press, 2011).
  28. [29]Sazonov, A. et al. Origin of forbidden reflections in multiferroic Ba₂CoGe₂O₇ by neutron diffraction: symmetry lowering or renninger effect? Journal of Applied Crystallography 49, 556–560 (2016).DOI
  29. [30]Amato, A. et al. The new versatile general purpose surface-muon instrument (GPS) based on silicon photomultipliers for μSR measurements on a continuous-wave beam. Review of Scientific Instruments 88, 093301 (2017).DOI
  30. [31]Morenzoni, E. et al. Generation of very slow polarized positive muons. Physical Review Letters 72, 2793 (1994).pubmed.ncbi.nlm.nih.gov/10055978
  31. [32]Prokscha, T. et al. The New μE4 Beam at PSI: A Hybrid-Type Large Acceptance Channel for the Generation of a High Intensity Surface-Muon Beam. Nuclear Instruments and Methods in Physics Research Section A 595, 317–331 (2008).DOI
  32. [33]Suter, A. & Wojek, B. Musrfit: A Free Platform-Independent Framework for μSR Data Analysis. Physics Procedia 30, 69–73 (2012).DOI
  33. [34]Morenzoni, E. et al. Implantation studies of keV positive muons in thin metallic layers. Nuclear Instruments and Methods in Physics Research Section B 192, 254–266 (2002).DOI
  34. [35]Eckstein, W. Computer Simulation of Ion-Solid Interactions, vol. 10 of Springer Series in Materials Science (Springer, 1991).DOI
  35. [36]Suter, A., Martins, M. M., Ni, X., Prokscha, T. & Salman, Z. Low Energy Measurements in Low-Energy μSR. Journal of Physics: Conference Series 2462, 012011 (2023).
  36. [37]Blundell, S., De Renzi, R., Lancaster, T. & Pratt, F. L. Muon Spectroscopy: An Introduction (Oxford University Press, 2022).
  37. [38]Blundell, S. J. Spin-polarized muons in condensed matter physics. Contemporary Physics 40, 175–192 (1999).DOI
  38. [39]Kresse, G. & Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Physical Review B 59, 1758 (1999).DOI
  39. [40]Blöchl, P. E. Projector augmented-wave method. Physical Review B 50, 17953 (1994).pubmed.ncbi.nlm.nih.gov/9976227
  40. [41]Kresse, G. & Hafner, J. Ab initio molecular dynamics for liquid metals. Phys. Rev. B 47, 558–561 (1993).
  41. [42]Kresse, G. & Hafner, J. Ab initio molecular-dynamics simulation of the liquid-metal–amorphous-semiconductor transition in germanium. Phys. Rev. B 49, 14251–14269 (1994).pubmed.ncbi.nlm.nih.gov/10010505
  42. [43]Kresse, G. & Furthmüller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Computational Materials Science 6, 15–50 (1996).DOI
  43. [44][44*] Kresse, G. & Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996). Publisher:pubmed.ncbi.nlm.nih.gov/9984901
  44. [45]Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized Gradient Approximation Made Simple. Physical Review Letters 77, 3865–3868 (1996). Publisher: American Physical Society.pubmed.ncbi.nlm.nih.gov/10062328
  45. [46]Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized Gradient Approximation Made Simple [Phys. Rev. Lett. 77, 3865 (1996)]. Physical Review Letters 78, 1396–1396 (1997). Publisher: American Physical Society.pubmed.ncbi.nlm.nih.gov/10062328
  46. [47]Ernzerhof, M., Burke, K. & Perdew, J. P. Density functional theory, the exchange hole, and the molecular bond. Theoretical and Computational Chemistry 4, 207–238 (1996).DOI
  47. [48]Liechtenstein, A. I., Anisimov, V. I. & Zaanen, J. Density-functional theory and strong interactions: Orbital ordering in mott-hubbard insulators. Physical Review B 52, R5467 (1995).pubmed.ncbi.nlm.nih.gov/9981806
  48. [49]Adamantopoulos, T. et al. Spin and orbital magnetism by light in rutile altermagnets (2024).
  49. [50]Diulus, J. T., Tobler, B., Osterwalder, J. & Novotny, Z. Thermal oxidation of Ru (0001) to RuO₂ (110) studied with ambient pressure x-ray photoelectron spectroscopy. Journal of Physics D: Applied Physics 54, 244001 (2021).DOI
  50. [51]Jovic, V. et al. Direct nodal lines and flat-band surface state in the functional oxide RuO₂. Physical Review B 98, 24 (2018).DOI

Paper details

vlnc:W_rf69y6m3
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