Spatially Resolved Magnetic Anisotropy of Cobalt Nanostructures on

Letter pubs.acs.org/NanoLett

Spatially Resolved Magnetic Anisotropy of Cobalt Nanostructures on the Au(111) Surface Puneet Mishra,*,† Zhi Kun Qi,† Hirofumi Oka,† Kohji Nakamura,‡ and Tadahiro Komeda*,† †

Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, Katahira 2-1-1, Aoba-ku, Sendai 980-8577, Japan Department of Physics Engineering, Mie University, Tsu, Mie 514-8507, Japan

‡

S Supporting Information *

ABSTRACT: Understanding the origin of perpendicular magnetic anisotropy in surface-supported nanoclusters is crucial for fundamental research as well as data storage applications. Here, we investigate the perpendicular magnetic anisotropy energy (MAE) of bilayer cobalt islands on Au(111) substrate using spin-polarized scanning tunneling microscopy at 4.6 K and first-principles theoretical calculations. Au(111) substrate serves as an excellent model system to study the effect of nucleation site and stacking sequence on MAE. Our measurements reveal that the MAE of bilayer islands depends strongly on the crystallographic stacking of the two Co layers and nucleation of the third layer. Moreover, the MAE of Co atoms on Au(111) is enhanced by a factor of 1.75 as compared to that reported on Cu(111). Our first-principles calculations attribute this enhancement to the large spin−orbit coupling of the Au atoms. Our results highlight the strong impact of nanometer-scale structural changes in Co islands on MAE and emphasize the importance of spatially resolved measurements for the magnetic characterization of surface-supported nanostructures. KEYWORDS: Magnetic anisotropy, magnetization switching, nanoclusters, scanning tunneling microscopy, first-principles calculations

T

This technique allows us to investigate the surface magnetism with atomic-scale resolution.12−14 Au(111) substrate produces rich structural variations in Co islands due to its herringbone reconstruction creating fcc/hcp surface domains. We chose Au(111) as a substrate because it serves as an excellent model system to study the effect of nucleation site and stacking sequence on the MAE. Our measurements reveal that MAE is significantly enhanced for a specific crystallographic stacking of Co islands in each of the two surface domains. Interestingly, islands in less-preferred stacking show larger anisotropy. Additionally, we observe an enhancement in the stackingaveraged MAE of cobalt on Au(111) by a factor of 1.75 as compared to that reported on Cu(111).15 Our first-principles calculations attribute this enhancement to the large SOC of Au atoms. Furthermore, we observe a strong reduction of MAE due to the nucleation of the third layer, consistent with the trend of spin-reorientation transition reported in Co on Au(111). Our results emphasize the importance of interface effects in determining the MAE of magnetic nanostructures. Figure 1a shows a 60 nm × 60 nm STM image of Co islands on Au(111) surface. Cobalt nucleates at the elbows of the Au(111) reconstruction,16,17 resulting in the formation of regular arrays of bilayer islands with 3-fold symmetry. Au(111) surface reconstruction leads to two different kinds of nucleation sites, viz., fcc and hcp, at the herringbone elbows18 (Supporting Information, Section S1, and Figure S1). As inferred from the direction of the triangular shape of Co islands, two orientations

he ongoing miniaturization of data storage devices has led to a drastic reduction in the size of the magnetic bits. As the bit size reduces, the random reorientation of its magnetization direction occurs due to thermal fluctuations causing loss of information stored in the magnetic bit. One of the ways to overcome this problem is by developing materials with strong magnetic anisotropy. Understanding the origin of magnetic anisotropy, especially in surface-supported nanostructures, is important to fully exploit such materials for applications. In surface-supported nanostructures, the magnetic anisotropy energy (MAE) can be modified in a controlled manner by varying the particle size, shape, and choosing appropriate substrate for inducing interface-driven effects.1,2 The interfaces, in particular, combining a 3d ferromagnet with its large spin moment and 4d or 5d metal with its large spin− orbit coupling (SOC) offer a fertile platform to explore the mechanism of enhancements of MAE.3−6 Accounting for the precise structure-magnetic property correlation is essential to tune the MAE in such systems. Spatially averaging techniques, namely, magneto-optical Kerr effect (MOKE), and X-ray magnetic circular dichroism (XMCD), have been remarkably successful in revealing the magnetic properties such as magnetic moments and MAE in a monodisperse ensemble of surfacesupported nanostructures.1,7,8 However, for ensembles containing several species, unveiling the structure−magnetic property correlation, including the interface effects, demands a spatially resolved magnetic characterization technique.2,9−11 Here we report spatially resolved measurements of MAE of bilayer Co islands on Au(111) using a low-temperature spinpolarized scanning tunneling microscopy (SP-STM) technique. © XXXX American Chemical Society

Received: July 21, 2017

A

DOI: 10.1021/acs.nanolett.7b03114 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

anisotropy and small Co coverages studied here (0.5−1.1 ML).19 Switching fields measured for bilayer cobalt islands of varying sizes are summarized in Figure 2a. The island size in the

Figure 1. Spin-polarized differential conductivity of Co islands. (a) 60 nm × 60 nm (b) 20 nm × 20 nm STM image of Co islands on Au(111) surface. dI/dV images (Vbias = −0.6 V, I = 200 pA) taken at (c) −1.5 T, (d) −2.5 T, and (e) −3 T. (f) Hysteresis curve of the dI/ dV signal measured over islands marked 2.

Figure 2. Energy barrier for magnetization reversal of Co islands. (a) Switching field of 137 bilayer Co islands as a function of the island size. (b) Variation of the energy barrier of the Co islands with size. The red line is a fit to the data as discussed in the text. Classification of the energy barrier for islands nucleated at the (c) fcc domain and (d) hcp domain.

rotated by 180° are found on each of the two surface domains; one orientation is preferred: for fcc (hcp) domain, the major orientation is indicated by red (blue) arrows. Our analysis reveals that the normalized ratio of islands of major orientation versus minor orientation are 0.75:0.25 and 0.65:0.35 for the fcc and the hcp domains, respectively. The observation of two orientations is attributed to the two different crystallographic stacking sequences of the Co layers as confirmed by our firstprinciples calculations (Supporting Information, Figure S2). The magnetic properties of Co islands were probed using a Pt/Ir tip after in situ capture of a Co cluster from the surface (Supporting Information, Figure S5). Figure 1c−e shows dI/dV images at Vbias = −0.6 V taken over the same region as in Figure 1b at different external magnetic fields (Supporting Information, Figure S6). Co islands exhibit spontaneous out-of-plane magnetization which is manifested in a two-level contrast in the dI/dV images. The two levels of contrast correspond to the parallel or antiparallel alignment of the island magnetization with respect to the out-of-plane component of the tip magnetization. An external magnetic field applied perpendicular to the sample surface changes the magnetization state of the Co islands. As the applied field is increased to −2.5 T and subsequently −3.0 T, the dI/dV signal contrast over islands marked 4 and 2 changes from bright to dark, respectively, indicating the switching of their magnetizations parallel to that of the tip. The switching fields, Hsw, of individual islands are obtained from measurements of the dI/dV hysteresis loop, where dI/dV signal over the island is plotted as a function of the magnetic field. Figure 1f shows a representative hysteresis loop obtained for the island marked 2 from dI/dV images acquired at Vbias = −0.6 V. The symmetric nature of the loop with a butterfly shape can be ascribed to the magnetization reversal of both the tip and the Co island during the field sweep. This also indicates the robustness of the spin contrast obtained in our measurements. As inferred from Figure 1f, Hsw of the island marked 2 is 3.0 T (Supporting Information, Section S5). Measured Hsw for an individual island was confirmed to be within ±0.05 T for repeated measurements for a few selected Co islands. This behavior is expected due to the negligible dipolar coupling between the islands, owing to the large out-of-plane magnetic

number of atoms N = 2Aρ, where A is the area of the island estimated from the topographic STM images and ρ is the areal density of Co atoms. Assuming pseudomorphic growth of cobalt islands on Au(111), ρ = 1.387 × 1015 atoms/cm2.20 Hsw was found to increase for island sizes ranging from N = 130 to N = 3500 atoms. This behavior is, in parts, reminiscent of the observation of increasing Hsw for Co islands on Cu(111) surface for 700 ≤ N ≤ 7500 atoms.15 There are, however, some illustrative differences: (i) the Hsw values are significantly higher than those observed for similarly sized islands on Cu(111) surface, and (ii) the Hsw values exhibit significantly large scattering on Au(111). For some islands, the switching field exceeds 5 T which is twice as large as the maximum Hsw observed on Cu(111) substrate15 (Supporting Information, Section S5). This observation clearly indicates a higher energy barrier, separating the two magnetization orientations, for Co islands on Au(111) surface. The measured Hsw allows us to extract the energy barrier for magnetization reversal.15 At finite temperature T, the energy barrier ΔE is related to Hsw as Hsw =

⎡ ⎛k T t ⎞1/2 ⎤ 2ΔE ⎢ 1 − ⎜ B ln meas ⎟ ⎥ Nμ ⎢⎣ τ0 ⎠ ⎥⎦ ⎝ ΔE −10

(1)

where μ = 2.1 μB/atom, τ0 = 10 s, tmeas = 100 s, and T = 4.6 K is the temperature at which the experiments were performed. The estimated energy barrier ΔE, as depicted in Figure 2b, depends linearly on N. This observation agrees well with the Neel−Brown model of thermally assisted magnetization reversal over a simple potential barrier,22,24 implying a coherent rotation of all the constituent spins in an island. The energy barrier then corresponds to a total sum of the effective magnetic anisotropy energies of all the atoms in an island, ΔE = KN, as proposed in the Stoner−Wolfharth model.25 Here, K is the effective MAE of a cobalt atom, and N is the size of the island in number of atoms. Fitting the experimental data with 21

B

2,22,23

15

DOI: 10.1021/acs.nanolett.7b03114 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters ΔE(N) = K(N − N0) gives K = 0.258 ± 0.006 meV/atom and N0 = 80 ± 21 atoms. A small value of N0 indicates that unlike the cobalt islands on Cu(111), which show exchange-spring behavior,15,26 here nearly all cobalt atoms contribute to the MAE. More importantly, the MAE of cobalt atoms is enhanced by a factor of 1.75 when Au(111) is used as a substrate, as compared to 0.148 meV/atom reported for Co islands on Cu(111).15 Figure 2c and d shows the energy barrier classified on the basis of Au(111) surface domain and crystallographic stacking. As stated earlier, Co islands nucleate either on fcc or hcp domains and have two orientations determined by the stacking sequence of the two Co layers. Fitting the data, using the Stoner−Wolfharth model,25 for islands on fcc domain yields Kfcc‑major = 0.256 ± 0.007 meV and Kfcc‑minor = 0.337 ± 0.014 meV. A similar analysis for islands at the hcp region gives Khcp‑major = 0.255 ± 0.008 meV and Khcp‑minor = 0.312 ± 0.008 meV. Scattering observed in the Hsw (Figure 2a) and ΔE (Figure 2b) values with the island size N, as noted earlier, is a consequence of this large variation in the stacking dependent MAE for different Co islands. While the effective MAE values are independent of the fcc/hcp domain, a large influence of stacking sequence of the two Co layers is observed. The effective MAE of Co atoms for islands in minor orientation is more than twice as large as 0.148 meV/atom reported earlier for similar islands on Cu(111).15 To understand the origin of the enhanced MAE of Co atoms on Au(111) compared to that on Cu(111), the spin- and orbital-resolved band structure, and the magnetocrystalline anisotropy (MCA) energy contribution are calculated (Supporting Information, Figures S2−S4). Our first-principles calculations predict the MCA energy, Kmca, to be 0.17 meV/ atom for a Co bilayer on Cu(111). Changing the substrate to fcc-Au(111) yields a significantly enhanced Kmca of 0.27 meV/ atom for the lowest-energy stacking (ABCAC) corresponding to the fcc-major islands (Supporting Information, Figure S2 and Table S1). Here, in the stacking-sequence ABCAC, the first three letters from left correspond to the Au layers, while the last two letters designate Co layers. The calculated value of Kmca is consistent with our experimental findings. If the SOC of Au atoms is switched off in the former calculation, then the Kmca is reduced to 0.18 meV/atom for Co on Au(111). This is almost similar to the value obtained on Cu(111) where substrate induced SOC is very weak. Since the Kmca for a free-standing Co bilayer is obtained to be 0.12 meV/atom, the interface formation with Au(111) or Cu(111) itself does not enhance the Kmca of Co bilayer so much, but the most important contribution toward the observed enhancement of MAE is derived from the large SOC of the Au atoms. A clear enhancement of MAE for islands in minor orientation for both the surface domains highlights the high sensitivity of our measurements. Differences in the Kmca obtained from our calculations for the two lowest-energy stacking sequences at each domain are nearly equal to the energy resolution of the calculations (60 μeV) (Supporting Information, Figure S2 and Table S1). Additionally, due to the herringbone reconstruction, each Co island is distributed over both the fcc domain as well as the hcp domain, irrespective of its nucleation site. This situation necessitates taking into account the herringbone reconstruction of Au(111) to obtain satisfactory matching with the experiments. Furthermore, earlier reports of calculations of MCA of zero-, one-, and two-dimensional transition metal systems on noble metal support have shown deviations up to tens of

percent with respect to the experimental estimates.7,9 This implies the possibility of further refinements in the theoretical calculations to incorporate small interfacial defects and features which can cause appreciable changes in the MCA energy. For this purpose, spatially resolved experimental results on model systems, as presented here, can provide valuable inputs. The effect of local structural changes on the MAE is further investigated using SP-STM. Cobalt growth on Au(111) is three-dimensional in nature. As the amount of deposited Co increases, several islands exhibit partial third layer coverage as can be seen for islands 3 and 4 in Figure 1b. Hsw obtained for Co islands with different third layer coverages are shown as solid symbols in Figure 3a as a function of size in the total

Figure 3. Effect of third layer on the magnetic anisotropoy energy. (a) Switching field of 54 Co islands with third layer coverage as a function of the island size. Data points for the bilayer islands are also shown for comparison. (b) Variation of the energy barrier for Co islands with third layer coverage as a function of their sizes. (c) Variation of the effective MAE with thickness.

number of atoms. Here, the third layer coverage, θ, is defined as the ratio of a third layer area with respect to that of the bilayer of the individual islands. Comparing with the results of bilayer islands (open circles), the Hsw is apparently reduced due to the presence of third layer, and this effect becomes more pronounced with increasing coverage. The energy barrier is extracted for Co islands with the third layer in the same manner as for bilayer islands described above, and the result is presented in Figure 3b. Data points of each coverage range vary linearly with the size. However, the slope decreases with increasing third layer coverage, implying that the MAE strongly depends on the coverage, in other words, the thickness of Co islands on Au(111). C

DOI: 10.1021/acs.nanolett.7b03114 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters For a more quantitative description, the effective MAE, K, needs to be extracted for individual islands investigated in Figure 3a and b. K can be extracted from the knowledge of magnetization reversal mechanism operative for these Co islands. Here we consider two possible magnetization reversal mechanisms, coherent rotation and domain wall formation (Supporting Information, Section S6 and Figure S8). Recent studies have demonstrated a crossover of the magnetization reversal mechanism, in Co nanostructures, from (quasi)coherent rotation of macrospin to domain wall formation at a critical size.8,15,19 While the energy barrier can be described as ΔEcr = KN for the coherent rotation, it is related to K as ΔEdw = 4σ AK for the domain wall formation. Here σ is the area of the domain wall and A the exchange constant. The competition between the two energies, ΔEcr and ΔEdw, determines the magnetization reversal mechanism, and the system goes to an energetically more favorable state. Considering A = 27.1 meV/atom,15,27 and K = 0.337 meV/ atom as the upper limit of the MAE, our analysis predicts a transition from coherent rotation of macrospin to domain wall formation for bilayer islands beyond a critical size of 4400 atoms. This prediction is consistent with our experimental findings for bilayer islands presented in Figure 2. Furthermore, the critical size for 3 ML Co islands is estimated to be 6700 atoms. Thus, the Co islands with the third layer in Figure 3a and b are in the regime of the coherent rotation magnetization reversal and their individual magnetic anisotropy energies can be extracted with ΔEcr = KN. Figure 3c summarizes island-specific MAE values of the Co islands as a function of thickness. In spite of the large scattering observed in the data points, a definite trend can be seen. The MAE decreases from 0.26 to 0.08 meV/atom as the thickness changes from 2.0 to 2.8 ML. The limiting values observed for 2.0 and 2.8 ML are remarkably close to 0.258 meV/atom estimated for purely bilayer islands above (dashed blue line in Figure 3c) and 0.074 meV/atom theoretically computed for 3 ML Co layer on Au(111)28 (dotted red line in Figure 3c), respectively. The data points in Figure 3c are classified into categories depending on the island size. Since the trend is observed for all the categories, it does not originate from differences in the size. Furthermore, a linear extrapolation of the data predicts that the MAE of the Co atoms changes from positive to negative at a thickness of 3.7 ± 0.3 ML, indicating spin-reorientation transition (SRT). This prediction is consistent with the earlier experimental29,30 and theoretical28 reports of SRT in Co/Au(111) at 4.2 ML. Our spatially resolved measurements demonstrate that the presence of even fractional amount of third layer significantly reduces the MAE of cobalt atoms as compared to that for a purely bilayer island. Observation of magnetization reversal by coherent rotation of macrospin for bilayer islands up to N = 3500 atoms contradicts an earlier report, using spatially averaging technique, MOKE, where deviation from coherent rotation occurs beyond N = 750 atoms.19 Although a direct comparison between ref 19 and our results is not straightforward since the magnetization reversal was thermally activated rather than driven by magnetic field, this discrepancy can be resolved, qualitatively, by taking into account the effect of third Co layer. The conclusion in ref 19 is based on the assumption that the presence of third layer has negligible effect on the MAE of Co islands. On the other hand, several islands have varying fraction of third layer coverage even at island sizes N ≥ 750 atoms and the magnetic anisotopy of Co decreases with the increasing

third layer coverage as discussed above. This observation is valid for islands even with similar sizes. This can be clearly seen for data points for 2000−2500 atoms (solid symbol) in Figure 3c. A strong reduction of the MAE is expected for measurements using spatially averaging techniques over Co islands with different third layer coverages. Therefore, our spatially resolved measurements on purely bilayer islands corroborate that the magnetization reversal proceeds by the coherent rotation of macrospin for islands at least up to N = 3500 atoms. In conclusion, the evolution of MAE in Co islands on Au(111) was investigated by means of spatially resolved technique, SP-STM, and first-principles calculations. Our quantitative analysis of experimental data revealed that the MAE of bilayer islands strongly depends on the crystallographic stacking of the two Co layers and the formation of third layer. The MAE of bilayer islands on Au(111) is enhanced by a factor of 1.7 and more than 2 for islands in a major and in a minor stacking, respectively, as compared to that reported on Cu(111). Our first-principles calculations attributed this enhancement to the large spin−orbit coupling of Au atoms. Our results highlight the strong impact of atomic-scale structural changes in Co islands on the MAE and emphasize the importance of spatially resolved techniques to characterize magnetic nanostructures. Methods. Experiments were performed at 4.6 K in a commercial low-temperature STM from Unisoku with a base pressure of 10−8 Pa. Magnetic fields up to B = ±5 T were applied normal to the sample surface, which is the easy magnetization direction of bilayer Co islands on Au(111). The differential conductance (dI/dV) images at a fixed sample bias voltage (Vbias) were acquired simultaneously with the topographic images using conventional lock-in detection with the feedback loop closed. Au(111) substrate was prepared by repeated cycles of Ar ion sputtering followed by annealing. Submonolayer Co was deposited by electron bombardment heating of a Co rod. Samples were then quickly transferred to the STM precooled at 4.6 K. Calculations were performed based on generalized gradient approximation 31 by using the full potential linearized augmented plane-wave (FLAPW) method with a single slab geometry32−34 (Supporting Information, Section S2). MCA energy is calculated with the second variational SOC treatment based on the force theorem.34

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.7b03114. Co growth on Au(111), details and results of the theoretical calculations, determination of switching field of Co islands, and determination of the switching mechanism of Co islands (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail (P.M.): [email protected]. *E-mail (T.K.): [email protected]. ORCID

Puneet Mishra: 0000-0003-2890-432X D

DOI: 10.1021/acs.nanolett.7b03114 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters Present Address

(19) Rohart, S.; Campiglio, P.; Repain, V.; Nahas, Y.; Chacon, C.; Girard, Y.; Lagoute, J.; Thiaville, A.; Rousset, S. Phys. Rev. Lett. 2010, 104, 137202. (20) Krupski, K.; Moors, M.; Józw ́ ik, P.; Kobiela, T.; Krupski, A. Materials 2015, 8, 2935−2952. (21) Šipr, O.; Bornemann, S.; Minár, J.; Polesya, S.; Popescu, V.; Šimunek, A.; Ebert, H. J. Phys.: Condens. Matter 2007, 19, 096203. (22) Brown, W. F. Phys. Rev. 1963, 130, 1677−1686. (23) Brown, W. F.; Hanton, J. P.; Morrish, A. H. J. Appl. Phys. 1960, 31, S214−S215. (24) Wernsdorfer, W.; Orozco, E. B.; Hasselbach, K.; Benoit, A.; Barbara, B.; Demoncy, N.; Loiseau, A.; Pascard, H.; Mailly, D. Phys. Rev. Lett. 1997, 78, 1791−1794. (25) Stoner, E. C.; Wohlfarth, E. P. Philos. Trans. R. Soc., A 1948, 240, 599−642. (26) Rohart, S.; Repain, V.; Thiaville, A.; Rousset, S. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 76, 104401. (27) Speckmann, M.; Oepen, H. P.; Ibach, H. Phys. Rev. Lett. 1995, 75, 2035−2038. (28) Campiglio, P.; Breitwieser, R.; Repain, V.; Guitteny, S.; Chacon, C.; Bellec, A.; Lagoute, J.; Girard, Y.; Rousset, S.; Sassella, A.; Imam, M.; Narasimhan, S. New J. Phys. 2015, 17, 063022. (29) Allenspach, R.; Stampanoni, M.; Bischof, A. Phys. Rev. Lett. 1990, 65, 3344−3347. (30) Rodary, G.; Repain, V.; Stamps, R. L.; Girard, Y.; Rohart, S.; Tejeda, A.; Rousset, S. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 75, 184415. (31) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865−3868. (32) Wimmer, E.; Krakauer, H.; Weinert, M.; Freeman, A. J. Phys. Rev. B: Condens. Matter Mater. Phys. 1981, 24, 864−875. (33) Weinert, M.; Wimmer, E.; Freeman, A. J. Phys. Rev. B: Condens. Matter Mater. Phys. 1982, 26, 4571−4578. (34) Nakamura, K.; Ito, T.; Freeman, A. J.; Zhong, L.; Fernandez-de Castro, J. Phys. Rev. B: Condens. Matter Mater. Phys. 2003, 67, 014420.

(P.M.) Department of Physics, Indian Institute of Technology Patna, Bihta, Patna-801103, India. (H.O.) Advanced Institute for Materials Research (WPI-AIMR), Tohoku University, Sendai 980−8577, Japan Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The following financial supports are acknowledged: Grant-inAid for Scientific Research on Innovative Areas (MEXT KAKENHI grant no. JP25110005) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) (T.K., P.M., H.O.); JSPS KAKENHI grant no. JP16H03863 and JP17K19047 (T.K.); JSPS Kakenhi grant no. 17H02779 and 15K20878 (H.O.). Computations were performed at Research Institute for Information Technology, Kyushu University. P.M. thanks CSIR for SRA during the preparation of this manuscript. Authors thank Prof. Osamu Kitakami for fruitful discussion.

■

REFERENCES

(1) Gambardella, P.; Dallmeyer, A.; Maiti, K.; Malagoli, M. C.; Eberhardt, W.; Kern, K.; Carbone, C. Nature 2002, 416, 301−304. (2) Krause, S.; Herzog, G.; Stapelfeldt, T.; Berbil-Bautista, L.; Bode, M.; Vedmedenko, E. Y.; Wiesendanger, R. Phys. Rev. Lett. 2009, 103, 127202. (3) Weller, D.; Wu, Y.; Stöhr, J.; Samant, M. G.; Hermsmeier, B. D.; Chappert, C. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 49, 12888−12896. (4) Nakajima, N.; Koide, T.; Shidara, T.; Miyauchi, H.; Fukutani, H.; Fujimori, A.; Iio, K.; Katayama, T.; Nývlt, M.; Suzuki, Y. Phys. Rev. Lett. 1998, 81, 5229−5232. (5) Bornemann, S.; Šipr, O.; Mankovsky, S.; Polesya, S.; Staunton, J. B.; Wurth, W.; Ebert, H.; Minár, J. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 86, 104436. (6) Ouazi, S.; Vlaic, S.; Rusponi, S.; Moulas, G.; Buluschek, P.; Halleux, K.; Bornemann, S.; Mankovsky, S.; Minár, J.; Staunton, J. B.; Ebert, H.; Brune, H. Nat. Commun. 2012, 3, 1313. (7) Gambardella, P.; Rusponi, S.; Veronese, M.; Dhesi, S. S.; Grazioli, C.; Dallmeyer, A.; Cabria, I.; Zeller, R.; Dederichs, P. H.; Kern, K.; Carbone, C.; Brune, H. Science 2003, 300, 1130−1133. (8) Cavallin, A.; Natterer, F. D.; Ouazi, S.; Moulas, G.; Lehnert, A.; Rusponi, S.; Brune, H. Phys. Rev. B: Condens. Matter Mater. Phys. 2014, 90, 144427. (9) Balashov, T.; Schuh, T.; Takács, A. F.; Ernst, A.; Ostanin, S.; Henk, J.; Mertig, I.; Bruno, P.; Miyamachi, T.; Suga, S.; Wulfhekel, W. Phys. Rev. Lett. 2009, 102, 257203. (10) Loth, S.; Baumann, S.; Lutz, C. P.; Eigler, D. M.; Heinrich, A. J. Science 2012, 335, 196−199. (11) Khajetoorians, A.; Wiebe, J.; Chilian, B.; Lounis, S.; Blugel, S.; Wiesendanger, R. Nat. Phys. 2012, 8, 497−503. (12) Bode, M. Rep. Prog. Phys. 2003, 66, 523. (13) Wiesendanger, R. Rev. Mod. Phys. 2009, 81, 1495−1550. (14) Oka, H.; Brovko, O. O.; Corbetta, M.; Stepanyuk, V. S.; Sander, D.; Kirschner, J. Rev. Mod. Phys. 2014, 86, 1127−1168. (15) Ouazi, S.; Wedekind, S.; Rodary, G.; Oka, H.; Sander, D.; Kirschner, J. Phys. Rev. Lett. 2012, 108, 107206. (16) Voigtländer, B.; Meyer, G.; Amer, N. M. Phys. Rev. B: Condens. Matter Mater. Phys. 1991, 44, 10354−10357. (17) Meyer, J.; Baikie, I.; Kopatzki, E.; Behm, R. Surf. Sci. 1996, 365, L647−L651. (18) Morgenstern, K.; Kibsgaard, J.; Lauritsen, J. V.; Lægsgaard, E.; Besenbacher, F. Surf. Sci. 2007, 601, 1967−1972. E

DOI: 10.1021/acs.nanolett.7b03114 Nano Lett. XXXX, XXX, XXX−XXX