Abstract
Spins of single atoms adsorbed on substrates are promising building blocks for spintronics and quantum computation schemes. To process spin information and for increased magnetic stability, these spins have to be coupled to arrays. For a single atom, a high symmetry of the environment increases its spin stability. However, little is known about the role of the symmetry of the magnetic couplings in the arrays. Here, we study arrays of atomic spins coupled via Ruderman−Kittel−Kasuya−Yosida interaction, focusing on Dzyaloshinskii−Moriya and symmetric anisotropic exchange. We show that the high spin stability of a trimer can be remotely detected by a nearby atom, and how the Dzyaloshinskii−Moriya interaction leads to its destabilization. Adding more nearby atoms further destabilizes the trimer, due to a nonlocal effective transverse anisotropy originating in the symmetric anisotropic exchange. This transverse anisotropy can be quenched for highly symmetric structures, where the spin lifetime of the array increases drastically.
Introduction
In magnetic systems, the enhancement, preservation or breaking of symmetries have strong implications for the groundstate and the dynamical properties. For instance, twodimensional spin lattices with C_{4V} symmetry and antiferromagnetic nearestneighbor interactions form a stable diagonal rowwise antiferromagnet, while a C_{6V} symmetry is predicted to have a spin liquid ground state^{1,2,3}. The breaking of inversion symmetry is responsible for the presence of the Dzyaloshinskii−Moriya (DM) interaction between the spins in materials with large spin−orbit interaction^{4,5}, which promotes the formation of magnetic skyrmions^{6,7,8,9}. Another important example is the symmetry of the Coulomb field of the atoms around the spin, the socalled crystal field, which induces a magnetic anisotropy energy that leads to the stabilization of the spin along particular directions^{10}. It has recently been shown that, when the crystal field of a single absorbed atomic spin has high symmetry^{11}, and the spin is decoupled from the conduction electrons of the substrate, the spin state lifetime can be enormously enhanced, up to hours^{12}. This brings the use of such atomic spins as bits of information technology into reach.
In order to read, write and process the information stored in the state of such atomic spins, they have to be coupled to neighboring spins. The transfer of information can be realized via dipolar interaction^{13}, direct or superexchange interaction^{14}, or the conductionelectronmediated Ruderman−Kittel−Kasuya−Yosida (RKKY) interaction between spins on metallic surfaces^{15,16,17}. The latter is unique in its flexibility to tailor the properties of the spinsystem in terms of coupling strength and noncollinearity^{18}. The general capability of RKKYcoupled networks to transfer and process information at the nanoscale has been shown by realizing fewatom spinbased logic gates^{19}. One might naively expect that enhancing the number of atoms in such a network will drive the system into the regime of a classical magnet, where spinfluctuations are less important. Indeed, exchangecoupling a few spins to larger ensembles via strong isotropic ferromagnetic (FM)^{20} or antiferromagnetic (AFM)^{21,22} Heisenberg exchange interactions has been shown to enhance the spin stability with respect to that of the constituents^{23}. However, the coupling more generally has other contributions, such as the DM and the symmetric anisotropic Heisenberg exchange^{24}. The effects of such contributions on the spin fluctuations and whether they are governed by symmetries remain to be explored. An experimental and theoretical approach towards this end for RKKYcoupled spins is given in the following, where the spinfluctuations of atombyatom engineered complexes consisting of one cluster and a different number of RKKYcoupled satellite atoms are investigated. The results of timeresolved spinpolarized scanning tunneling microscopy experiments are analyzed and interpreted taking into account the outcome of density functional theory (DFT) calculations and dynamic simulations.
Results
Trimer–satellite atom complexes
A prototypical system for these studies consists of iron (Fe) atoms on a Pt(111) surface (Fig. 1a, see Methods), which adsorb either on the facecentered cubic (fcc) or on the hexagonal closepacked (hcp) sites. In both cases, the atoms can be approximately described by an effective spin with a quantum number of S_{a} = 5/2, and a preferred spin orientation along the surface normal (fcc, easyaxis anisotropy) or in the surface plane (hcp, easyplane anisotropy)^{25}. Up to distances of several lattice spacings, there is a sizeable RKKY interaction between the Fe atoms with significant Heisenberg as well as DM contributions. Varying the separation of the spins gives control over the magnitudes and signs of the Heisenberg and DM parts, in order to tailor noncollinear spin states^{18,26}. Contrary to what one may expect intuitively, the strong interaction of Fe atoms with the Pt(111) substrate does not prevent ground state spin lifetimes of hours, when at least three iron atoms are manipulated into a closepacked fcctopstacked symmetrical trimer (Fig. 1a) with a spin of S_{t} = 11/2^{27}. Since the spin state of these trimers (up or down, perpendicular to the surface) can be easily set by a short pulse of spinpolarized electrons with sufficient energy, they present ideal inputs/switches for RKKYcoupled networks of atoms^{19}. However, it is a priori unclear whether the RKKY interaction, e.g. between such a trimer and a closeby satellite atom (Fig. 1a), further stabilizes or destabilizes the spin state of the coupled network. In order to investigate this question, we first place a single Fe atom on an fcc site at a distance of d = 0.97 nm to a wellcharacterized fcc top trimer^{27}, via lateral atom manipulation (Fig. 1a).
Two spinstate telegraph noise
The strong impact of the RKKY coupling on the dynamics of the trimer−satellite atom system becomes apparent when a magnetic tip (sensitive to the outofplane component of the sample spinpolarization) is positioned on either of the two constituents of this complex (Fig. 1b). For both tip positions, we observe a telegraph noise between two spin states, 0 and 1, with corresponding lifetimes of τ_{0} and τ_{1} and mean lifetime \(\tau = (\tau _0^{  1} + \tau _1^{  1})^{  1}\) on the order of 10 ms to 10 s. Similar spinstate lifetimes have been observed for isolated uncoupled trimers and are associated with the tunneling electroninduced dynamics between two degenerate outofplane oriented ground states of the trimer^{27}. However, the observation of a lifetime on this order of magnitude for the spin state of the satellite atom is surprising. For isolated uncoupled atoms, the spin ground state lifetimes have been estimated to be on the order of 1 ms^{28} prohibiting the observation of the corresponding telegraph noise due to the limited time resolution of the method. Therefore, the telegraph noise detected on the satellite atom (Fig. 1b bottom trace) must be a consequence of its RKKY interaction with the slowly switching trimer. We propose the mechanism illustrated in Fig. 1c, which has been confirmed by our simulations (see below and Supplementary Notes 6 and 7). The RKKY interaction with the trimer induces an asymmetry \(\frac{{\tau _{\mathrm{1}}^{{\mathrm{a}},{\mathrm{short}}}  \tau _0^{{\mathrm{a}},{\mathrm{short}}}}}{{\tau _1^{{\mathrm{a}},{\mathrm{short}}} + \tau _0^{{\mathrm{a}},{\mathrm{short}}}}}\) in the short spingroundstate lifetimes of the satellite atom (\(\tau _{0/1}^{{\mathrm{a}},{\mathrm{short}}}\)). Although the SPSTM measurement averages out this fast telegraph noise, a changing asymmetry induced by switches of the trimer spin leads to a changing apparent height z of the satellite atom, which is detectable via the longer lifetimes \(\tau _0^{\mathrm{a}}\) and \(\tau _1^{\mathrm{a}}\). Thereby, the telegraph noise which is measured on the satellite atom reflects only the slow dynamics of the coupled system and essentially enables a nonlocal readout of the trimer spin similar to ref. ^{21}. This interpretation is further corroborated by the correlation between the telegraph noises measured on the trimer and satellite atom (Fig. 1d, e). In consecutively recorded line scans along the dashed line in Fig. 1d, which are stacked on top of each other in Fig. 1e, an anticorrelated switching of the trimer and satellite atom is found. Whenever the trimer is in the high state, the apparent height of the atom is low, and vice versa. Such anticorrelations or correlations of the two telegraph signals furthermore enable to conclude on a dominating AFM or FM orientation between the trimer and the satellite atom, since atom^{29} and trimer possess the same sign of the vacuum spinpolarization (Supplementary Note 4). The anticorrelation observed in Fig. 1e therefore indicates an AFM spin orientation in this particular complex.
Dependence of lifetime on trimer to satelliteatom distance
Although the telegraph noises measured in the two situations depicted in the insets in Fig. 1b, i.e. directly on the trimer (black) or nonlocally on the satellite atom (red), reflect the dynamics of the coupled system, they strongly differ for the two cases. Consequently, these differences must originate from different probabilities to switch the trimer–satellite atom complex for the two tunneling paths. Interestingly, for the particular complex in Fig. 1, the tunnel current is more likely to induce switching via the atom than via the trimer (\(\tau _{\mathrm{a}} < \tau _{\mathrm{t}}\)), and both lifetimes are smaller than the lifetime measured on the same trimer in the absence of the satellite atom^{27}. Counterintuitively, the trimer spin is found to be destabilized by the nearby atom.
In order to systematically study the influence of the satellite atom on the magnetization dynamics of the complex, various structures are constructed and characterized in terms of dynamics and coupling behavior. To this end, the satellite atom is moved to adsorption sites at various distances to the trimer via lateral atom manipulation (Fig. 2a). In all these complexes, a finite coupling was revealed by a telegraph noise on the satellite atom analogous to the observations for position a in Fig. 1. In dependence of the actual position of the satellite atom, the telegraph noises of trimer and atom are correlated or anticorrelated, indicating a complex oscillation between FM and AFM orientation illustrated in Fig. 2a by the corresponding color. This oscillatory behavior is reminiscent of an RKKY interaction. Indeed, the measured orientations are consistent with the configurations that have been obtained from the minimization of a classical Heisenberg model using density DFT parameters as input (Fig. 2b, see Methods). The DFT calculations were performed for all fcc positions of the satellite atoms under the assumption of a collinear internal spin configuration in the outofplane direction for the trimer, which is a valid approximation for fcc trimers due to only tiny noncollinear deviations^{27}. The calculations reveal that the magnetic moment of the satellite atom is almost collinear to that of the trimer (along the zaxis, which is the easy axis for fcc atoms). Only for the farthest separation between trimer and satellite atom are the experimental and theoretical results in contradiction, which might be due to inaccuracies in the calculation associated with the small size of the interaction. The FMAFM oscillation results from a complex interplay of different contributions to the RKKY interaction, as revealed by mapping the DFT calculated energies to a generalized classical Heisenberg model which then results in the quantum Heisenberg Hamiltonian
with the vector spin operators of the satellite atom \({\hat{\mathbf{S}}}_{\mathrm{a}}\) and trimer \({\hat{\mathbf{S}}}_{\mathrm{t}}\) and their respective zcomponents \(\hat S_{\mathrm{a}}^z\) and \(\hat S_{\mathrm{t}}^z\). Besides the magnetocrystalline anisotropies (MAE) of atom (\({\cal{D}}_{\mathrm{a}}\)) and trimer (\({\cal{D}}_{\mathrm{t}}\)), it includes the most general form of Heisenberg exchange including the isotropic (J_{iso}) and the symmetric anisotropic (J_{aniso}) parts, as well as the DM interaction (D) (see Supplementary Notes 5, 6 for the definition of the operators and the relation between classical and quantum Heisenberg Hamiltonian). The symmetric anisotropic part, J_{aniso}, is also called the pseudodipolar interaction or a twoion anisotropy term, which contrary to the isotropic part, J_{iso}, favors the orientation of the moments along specific directions similarly to MAE, leading to nonlocal effective magnetic anisotropies.
Figure 2c shows the distancedependent strengths of J_{iso}, J_{aniso} and D, whose components all display oscillations in the sign typical for RKKY interaction (see Fig. 2d). These sign oscillations, together with a rapid decrease in the strength of the RKKY interaction (Fig. 2c) and the relatively strong outofplane MAE of the fcc satellite atoms \({\cal{D}}_{\mathrm{a}} =  0.19\,{\mathrm{meV}}\)^{25} and trimer \({\cal{D}}_{\mathrm{t}} =  0.09\,{\mathrm{meV}}\)^{27} lead to an almost collinear spin orientation between trimer and satellite atom along the surface normal, which alternates between FM and AFM as their separation is varied. For complexes including hcp satellites, the experimental results (Supplementary Note 1) point towards a deviation of the atom’s spin orientation from the surface normal due to the inplane MAE.
In order to investigate the influence of these RKKY couplings on the dynamics of the complexes, time traces like those in Fig. 1b were recorded for each built configuration. The lifetimes \(\tau _{\mathrm{a}}\) and \(\tau _{\mathrm{t}}\) measured with the tip on the satellite atom, respectively the trimer, are plotted in Fig. 3a against the atomtrimer separation. Obviously, there are two qualitatively different regimes. For smaller distances (\(d < 1.25\,{\mathrm{nm}}\), see vertical dashed line), there is a destabilization of the trimer by the satellite atom in comparison to the freestanding case (horizontal dashed line). Independently of whether the complex lifetime is measured via the atom (red) or directly on the trimer (black), the values are about one order of magnitude smaller. In contrast, for larger distances (\(d > 1.25\,{\mathrm{nm}}\)), the lifetime of the complex strongly differs between the two measurement modes: When the current runs through the trimer (\(\tau _{\mathrm{t}}\)) the lifetime is comparable to that of an isolated trimer, which indicates a negligible influence of the satellite atom on the trimer dynamics. However, the twostate telegraph noise can still be observed if the current runs through the satellite atom (\(\tau _{\mathrm{a}}\)), proving a finite RKKY coupling between the two constituents. The accordinglymeasured values of \(\tau _{\mathrm{a}}\) are two orders of magnitude larger than \(\tau _{\mathrm{t}}\), indicating that the influence of the measurement process on the trimer dynamics is drastically reduced. No clear correlation is found between the lifetime and the type of the adsorption site of the satellite atom nor the type of coupling, AFM or FM. Also, the destabilization effect in the small distance regime is independent of the used bias voltage, as it becomes obvious by comparison of the data from the freestanding trimer and the complex recorded over a large range of bias voltages (Fig. 4l).
To unravel the physics behind the sharp crossover between the two regimes, we performed simulations within a master equation approach, in conjunction with the effective quantum spin Hamiltonian given above^{20,27} using the exchange parameters extracted from the DFT calculations (see Methods). In both regimes, the simulations reproduce the experimental data quantitatively (Fig. 3a). Only for larger distances and when the current flows through the satellite atom, does the model overestimate the lifetimes, which might be explained by quantum fluctuations due to additional transversal magnetic anisotropies^{28} or zeropoint spin fluctuations^{30} which go beyond those included in the master equation approach. Note, that an effective tipmagnetic field^{27,28} may have residual effects on the lifetime difference between the measurements with the tip above of the trimer or above the satellite atom, which are not considered in the simulations. However, the effective field of the particular tip used for the measurements of Fig. 3a has been estimated to be smaller than 0.17 T which results in negligible changes of the lifetimes by at most a factor of 2 (Supplementary Notes 2 and 3). Moreover, an analysis using the magneticfieldindependent geometric means for the determination of the lifetimes instead of the harmonic means used for Fig. 3a, and additional data measured on another set of manipulated complexes using another tip, both show quantitatively similar results (Supplementary Note 3), ruling out a major role of the tipeffective magnetic field for the observed distance dependence of the lifetime.
The simulations enable to identify the origin of the steep increase of the lifetimes around 1.25 nm by independently sweeping D_{x} and J_{iso} (Fig. 3c). In the experimentally relevant parameter regime of atom−trimer separations of around 1.25 nm, there is a strong drop of the lifetime when the DM interaction strength exceeds ≈6 × 10^{−3} meV. The effect of the isotropic exchange interaction on the lifetime is much less pronounced. This strongly points towards a dominant destabilization of the trimer spin by the DM interaction. Further studies separating the dependence of the lifetime on the three different exchange contributions reinforce this finding (Fig. 3b). Contrary to the DM interaction, the symmetric anisotropic exchange interaction can affect the lifetimes in two ways: It can either stabilize the complex by inducing an effective easyaxis anisotropy (e.g. via a ferromagnetic zzcomponent) or destabilize it, similarly to the DM interaction, by inducing an effective transversal anisotropy. Figure 3b shows that the destabilization of the lifetime for an atom−trimer separation of around 1.25 nm is mainly driven by the DM interaction, whereas in the close distance regime the symmetric anisotropic part of the exchange interaction dominates the destabilization.
Dependence of the dynamics on symmetry
The observed destabilization of the spinstructure by coupling the trimer to a satellite atom is rather counterintuitive, as generally systems tend to stabilize with an increasing number of spins^{20,22}. Importantly, the addition of the satellite atom not only increases the number of spins, but also lowers the symmetry of the system. These two effects on the lifetime, i.e. number of spins and symmetry, are investigated in the following by adding different numbers of satellite atoms in different configurations to the complex (Fig. 4a–l). The lifetime depends in a rather intricate manner on the number of satellite atoms. It first increases for two satellite atoms (Fig. 4c, l), but then stays almost the same for three (Fig. 4e) and four (Fig. 4f) satellite atoms, or is even drastically reduced for a different position of the four satellite atoms (Fig. 4g). Note, that from (Fig. 4b) to (Fig. 4g) all satellite atoms have a distance d < 1.25 nm to the trimer, i.e. within the regime of strong coupling (cf. Fig. 3a). In most of these complexes the lifetime is smaller than for the symmetrical case of the trimer without any satellite atoms (Fig. 4a). Only if the satellite atoms are added in a way that conserves the C_{3V} symmetry of the trimer/substrate (Fig. 4d), can the lifetime be further increased by an order of magnitude above the one of the isolated trimer.
This suggests that keeping a high symmetry of the RKKY interactions, and in particular of the symmetric anisotropic contribution which dominates the destabilization in the complex with one satellite atom in the close distance regime, restores the spinstability. The effect of the symmetry of the RKKY interactions on the lifetime can be rationalized by the DFT calculation of the effective magnetic anisotropy energy of all trimer–satellite atom complexes when the spins of all constituents are collinearly rotated in the surface plane (Fig. 4m). Interestingly, the RKKY interaction induces an effective nonlocal transversal magnetic anisotropy in the complex spin system, whenever the C_{3V} symmetry is broken resulting in C_{S} or no symmetry. The crucial effect of the symmetry on this nonlocal transversal magnetic anisotropy and thereby on the lifetime of the complex is further proven by simulations of slight distortions of the symmetric complex of Fig. 4d (Supplementary Note 7 and Supplementary Fig. 10). Already a slight misalignment of one of the satellite atoms from the symmetric position decreases the lifetime by an order of magnitude, although the anisotropies and exchange couplings are kept constant. Onsite transversal magnetic anisotropy in singlespin effective Hamiltonians is well known to induce instability of the lifetime via mixing of the spin states^{31,32}. However, the related effect in a multispin complex originating from the nonlocal RKKY interactions between the spins, which we observe here, has not been studied so far, and cannot be reduced to the singlespin effective Hamiltonian picture.
We consequently expect that it might be possible to further stabilize the complex by adding a larger number of satellite atoms in symmetrical positions. This idea is experimentally verified by the formation of a symmetrical complex with six satellites shown in Fig. 4h. Successively adding three more atoms on symmetrical positions to the complex in (Fig. 4d), the lifetime is further increased by more than an order of magnitude (see the intermediate steps in (Fig. 4i, j) and the corresponding lifetimes in Fig. 4l).
Discussion
In conclusion, we revealed the dynamic behavior of RKKYcoupled complexes of two and more constituents. There are three regimes depending on the coupling strength between the trimer and a single satellite atom: In the weak coupling regime, the nearly intrinsic dynamics of the trimer could be observed while probing the satellite atom. This enables to nonlocally read and write the trimer spin via an RKKYcoupled network of atoms. In the intermediate coupling regime, the DM interaction leads to a significant destabilization when its magnitude exceeds a certain threshold. In the strong coupling regime, the symmetric anisotropic exchange interaction leads to a nonlocal transverse effective magnetic anisotropy which generally destabilizes the system. However, in the latter case, by engineering the symmetry of the complex, the magnetic stability can be restored and even enhanced by an order of magnitude. This demonstrates an interesting direction for the tailoring of RKKYcoupled networks of atoms towards usage in spintronics elements.
Methods
Experimental procedures
All measurements have been performed under ultrahigh vacuum conditions in a homebuilt scanning tunneling microscope facility at a base temperature of T = 300 mK. A superconducting magnet enabled to apply a magnetic field B perpendicular to the sample surface^{33}. The Pt(111) single crystal was cleaned in situ by argon ion sputtering and annealing cycles and a final flash, and single Fe atoms were subsequently deposited onto the cold substrate, as described in ref. ^{25}. The experiments described here have been conducted using the same preparation as used for the measurements presented in refs ^{26,27,28}.
The trimer was constructed by lateral atom manipulation^{34} in the pulling mode. The construction and identification is described in detail in ref. ^{27}. With the same technique, the arrangement of satellite atoms in the complexes is changed. During such a rearrangement of the complex, residual vertical or lateral drift is compensated. This enables to exclude tip changes by comparing the absolute substrate height before and after each manipulation. The stacking of the new position of the satellite atom is revealed by inelastic scanning tunneling spectroscopy^{25}. The trimer is not moved within the presented experiments.
All measurements presented here are performed using a spinpolarized tip with a magnetically stable sensitivity to the outofplane component of the sample magnetization. To this end, the spin sensitivity of a Crcoated tungsten tip was optimized by picking up single iron atoms until a high twostate telegraph noise is observed on the trimer.
Constantcurrent images were taken using a tunneling current I with a bias voltage V applied to the sample. For the dynamic measurements, the tip is placed stationary above the atom or the trimer and the apparent height is recorded with running feedback loop stabilizing the current at I.
Ab initio calculations
For the ab initio DFT calculations, we employed the Korringa−Kohn−Rostoker Green function method in full potential with spin−orbit coupling added to the scalarrelativistic approximation^{35}. The exchange and correlation effects are treated in the local spin density approximation as parametrized by Vosko et al.^{36}. The pristine Pt(111) surface is modeled by 40 Pt layers augmented by two vacuum regions. The magnetic Fe atoms are placed on the surface in the fccstacking position, using an embedding method. The magnetic anisotropy energies and the exchange interactions are obtained from band energy differences using the magnetic force theorem^{37} and the infinitesimal rotation method^{37,38}, respectively. Both are used in an extended classical Heisenberg model to obtain the ground state spin configuration (see Supplementary Note 5). The DFT results for the spin moments of the iron atoms and their magnetic anisotropies and magnetic pair interactions were used to parametrize the master equation model.
Master equation model
We use a master equation model to simulate the lifetime in the telegraph noise experiment. The magnetic clusters are described within the quantum Heisenberg model (Eq. 1) following refs ^{20,27}. The interaction of the tunneling electrons with the spins is described by an Applebaum Hamiltonian^{39}. The ratio between the coupling of the spins to the tip and to the surface at a bias voltage of 5 V is set to \(\upsilon ^{\mathrm{T}}\)/\(\upsilon ^{\mathrm{S}} = 0.055\), as determined in a previous experiment^{27}. The lifetimes are determined using a master equation approach combined with Fermi’s Golden rule^{39}.
Data availability
The authors declare that the main data supporting the findings of this study are available within the article and its Supplementary Information files. Extra data are available from the corresponding author upon reasonable request.
References
Anderson, P. W. Resonating valence bonds: a new kind of insulator? Mater. Res. Bull. 8, 153–160 (1973).
Normand, B. Frontiers in frustrated magnetism. Contemp. Phys. 50, 533–552 (2009).
Balents, L. Spin liquids in frustrated magnets. Nature 464, 199 (2010).
Dzyaloshinsky, I. A thermodynamic theory of “weak” ferromagnetism of antiferromagnetics. J. Phys. Chem. Solids 4, 241–255 (1958).
Moriya, T. Anisotropic superexchange interaction and weak ferromagnetism. Phys. Rev. 120, 91–98 (1960).
von Bergmann, K. et al. Observation of a complex nanoscale magnetic structure in a hexagonal Fe monolayer. Phys. Rev. Lett. 96, 167203 (2006).
Bode, M. et al. Chiral magnetic order at surfaces driven by inversion asymmetry. Nature 447, 190 (2007).
Heide, M., Bihlmayer, G. & Blügel, S. Dzyaloshinskii−Moriya interaction accounting for the orientation of magnetic domains in ultrathin films: Fe/w(110). Phys. Rev. B 78, 140403 (2008).
Heinze, S. et al. Spontaneous atomicscale magnetic skyrmion lattice in two dimensions. Nat. Phys. 7, 713 (2011).
Bloch, F. & Gentile, G. Zur Anisotropie der Magnetisierung ferromagnetischer Einkristalle. Z. f.ür. Phys. 70, 395–408 (1931).
Rau, I. G. et al. Reaching the magnetic anisotropy limit of a 3d metal atom. Science 344, 988–992 (2014).
Donati, F. et al. Magnetic remanence in single atoms. Science 352, 318–321 (2016).
Natterer, F. D. et al. Reading and writing singleatom magnets. Nature 543, 226 (2017).
Lin, C.Y. et al. Magnetic interaction between surfaceengineered rareearth atomic spins. Phys. Rev. X 2, 021012 (2012).
Ruderman, M. A. & Kittel, C. Indirect exchange coupling of nuclear magnetic moments by conduction electrons. Phys. Rev. 96, 99–102 (1954).
Kasuya, T. A theory of metallic ferro and antiferromagnetism on Zener’s model. Prog. Theor. Phys. 16, 45–57 (1956).
Yosida, K. Magnetic properties of cumn alloys. Phys. Rev. 106, 893–898 (1957).
Steinbrecher, M. et al. Noncollinear spin states in bottomup fabricated atomic chains. Nat. Commun. 9, 2853 (2018).
Khajetoorians, A. A., Wiebe, J., Chilian, B. & Wiesendanger, R. Realizing allspin–based logic operations atom by atom. Science 332, 1062–1064 (2011).
Khajetoorians, A. A. et al. Currentdriven spin dynamics of artificially constructed quantum magnets. Science 339, 55–59 (2013).
Yan, S. et al. Nonlocally sensing the magnetic states of nanoscale antiferromagnets with an atomic spin sensor. Sci. Adv. 3, e1603137 (2017).
Loth, S., Baumann, S., Lutz, C. P., Eigler, D. M. & Heinrich, A. J. Bistability in atomicscale antiferromagnets. Science 335, 196–199 (2012).
Marciani, M., Hübner, C. & Baxevanis, B. General scheme for stable single and multiatom nanomagnets according to symmetry selection rules. Phys. Rev. B 95, 125433 (2017).
Juba, B. et al. Chiral magnetism of magnetic adatoms generated by rashba electrons. New J. Phys. 19, 023010 (2017).
Khajetoorians, A. A. et al. Spin excitations of individual Fe atoms on Pt(111): Impact of the sitedependent giant substrate polarization. Phys. Rev. Lett. 111, 157204 (2013).
Khajetoorians, A. A. et al. Tailoring the chiral magnetic interaction between two individual atoms. Nat. Commun. 7, 10620 (2016).
Hermenau, J. et al. A gateway towards noncollinear spin processing using threeatom magnets with strong substrate coupling. Nat. Commun. 8, 642 (2017).
Hermenau, J., Ternes, M., Steinbrecher, M., Wiesendanger, R. & Wiebe, J. Long spinrelaxation times in a transitionmetal atom in direct contact to a metal substrate. Nano Lett. 18, 1978–1983 (2018).
Steinbrecher, M. et al. Absence of a spinsignature from a single Ho adatom as probed by spinsensitive tunneling. Nat. Commun. 7, 10454 (2016).
IbañezAzpiroz, J., dos Santos Dias, M., Blügel, S. & Lounis, S. Zeropoint spinfluctuations of single adatoms. Nano Lett. 16, 4305–4311 (2016).
Delgado, F. & FernándezRossier, J. Spin decoherence of magnetic atoms on surfaces. Prog. Surf. Sci. 92, 40–82 (2017).
Hübner, C., Baxevanis, B., Khajetoorians, A. A. & Pfannkuche, D. Symmetry effects on the spin switching of adatoms. Phys. Rev. B 90, 155134 (2014).
Wiebe, J. et al. A 300 mk ultrahigh vacuum scanning tunneling microscope for spinresolved spectroscopy at high energy resolution. Rev. Sci. Instrum. 75, 4871–4879 (2004).
Eigler, D. M. & Schweizer, E. K. Positioning single atoms with a scanning tunnelling microscope. Nature 344, 524–526 (1990).
Bauer, D. S. G. Development of a Relativistic FullPotential FirstPrinciples Multiple Scattering Green Function Method Applied to Complex Magnetic Textures of Nano Structures at Surfaces. PhD Thesis, RWTH Aachen University (2013).
Vosko, S. H. L., Wilk, L. H. & Nusair, M. Accurate spindependent electron liquid correlation energies for local spin density calculations: a critical analysis. Can. J. Phys. 58, 1200–1211 (1980).
Liechtenstein, A. I., Katsnelson, M. I., Antropov, V. P. & Gubanov, V. A. Local spin density functional approach to the theory of exchange interactions in ferromagnetic metals and alloys. J. Magn. Magn. Mater. 67, 65–74 (1987).
Mankovsky, S. et al. Effects of spinorbit coupling on the spin structure of deposited transitionmetal clusters. Phys. Rev. B 80, 014422 (2009).
Delgado, F. & FernándezRossier, J. Spin dynamics of currentdriven single magnetic adatoms and molecules. Phys. Rev. B 82, 134414 (2010).
Acknowledgements
J.H., M.S., R.W. and J.W. acknowledge funding by the SFB668 and by the Cluster of Excellence ‘Advanced Imaging of Matter’ (EXC 2056—project ID 390715994) of the Deutsche Forschungsgemeinschaft (DFG). S.B., M.d.S.D. and S.L. acknowledge funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (ERCconsolidator grant 681405—DYNASORE) and the computing time granted by the JARAHPC Vergabegremium and VSR commission on the supercomputer JURECA at Forschungszentrum Jülich. M.M. was supported by the Netherlands Organization for Scientific Research (NWO/OCW), an ERC Synergy Grant and the French Agence Nationale de la Recherche (ANR) under Grant TopoDyn (ANR14ACHN0031).
Author information
Authors and Affiliations
Contributions
J.H. and J.W. designed the experiments. J.H. carried out the measurements together with M.S. and did the analysis of the experimental data. S.B. performed the DFT calculations. M.M., S.B. and M.d.S.D. extended the master equation model, which S.B. used to simulate the data. J.H. and J.W. wrote the manuscript, to which S.B., M.M., M.S., M.d.S.D., R.W. and S.L. contributed via discussions and corrections.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Journal peer review information: Nature Communications thanks Cyrus Hirjibehedin, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Hermenau, J., Brinker, S., Marciani, M. et al. Stabilizing spin systems via symmetrically tailored RKKY interactions. Nat Commun 10, 2565 (2019). https://doi.org/10.1038/s41467019105162
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467019105162
This article is cited by

Remote detection and recording of atomicscale spin dynamics
Communications Physics (2020)

Longrange focusing of magnetic bound states in superconducting lanthanum
Nature Communications (2020)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.