Electronic and Spin States of SrRuO Thin Films: an X-ray Magnetic Circular 3 Dichroism Study S. Agrestini,1 Z. Hu,1 C.-Y. Kuo,1 M. W. Haverkort,1 K.-T. Ko,1 N. Hollmann,1 Q. Liu,1 E. Pellegrin,2 M. Valvidares,2 J. Herrero-Martin,2 P. Gargiani,2 P. Gegenwart,3 M. Schneider,4 S. Esser,3 A. Tanaka,5 A. C. Komarek,1 and L. H. Tjeng1 1Max Planck Institute for Chemical Physics of Solids, No¨thnitzerstr. 40, 01187 Dresden, Germany 2ALBA Synchrotron Light Source, E-08290 Cerdanyola del Vall`es, Barcelona, Spain 3Experimental Physics VI, Center for Electronic Correlations and Magnetism, University of Augsburg, 86159 Augsburg, Germany 5 4I. Physikalisches Institut, Georg-August-Universita¨t G¨ottingen, D-37077 G¨ottingen, Germany 1 5Department of Quantum Matter, ADSM, Hiroshima University, Higashi-Hiroshima 739-8530, Japan 0 2 (Dated: January 28, 2015) n We report a study of the local magnetism in thin films of SrRuO3 grown on (111) and (001) a oriented SrTiO3 substrates using x-ray magnetic circular dichroism spectroscopy (XMCD) at the J Ru-L2,3edges. TheapplicationofthesumrulestotheXMCDdatagivesanalmostquenchedorbital 7 momentandaspinmomentclosetothevalueexpectedforthelowspinstateS =1. Full-multiplet 2 cluster calculations indicate that the low spin state is quitestable and suggest that the occurrence of a transition to the high spin state S =2 in strained thin films of SrRuO3 is unlikely as it would ] betoo expensivein energy. l e - PACSnumbers: 75.70.Ak,75.47.Lx,78.70.Dm,72.80.Ga r t s t. Despite being investigated already for about five urated moment of 3.4 µB/Ru ion17,18, a value that is a decades the physical properties of SrRuO keeps fasci- muchhigher than that observedin bulk SrRuO and ex- m 3 3 nating the scientific community. SrRuO is one of the ceeds the atomic moment of 2 µ /Ru ion expected for 3 B - few known 4d transition metal oxide ferromagnets with a S = 1 spin state. In order to explain the SQUID re- d n Tc ashighas160K1,2. Itsnon-integermagneticmoment sultsithasbeenproposed18thatthetrigonalcompressive o hasbeeninterpretedintermsofasurprisingrareexample strain induced by the (111)STO substrate onto the film c of itinerant ferromagnetism in oxides3,4. More recently, would stabilize the high spin state (HS) S = 2, which [ the possibility ofemployingthin films ofSrRuO ascon- is very surprising as a HS state is unusual in 4d oxides. 3 1 ducting layer in epitaxial heterostructures of functional Even more intriguing, an unquenched orbital moment of v oxideshasarousedawideattentionfromthe appliedsci- about 0.32µB has been reportedfor these strainedfilms 9 ence community5. on the basis of XMCD measurements at the Ru M 2,3 9 edges18. However,theoreticalstudies,whichinvestigated 8 SrRuO3 isaperovskitecompoundwithanorthorhom- the effect of substrate-inducedcompressive strain on the 6 bic GdFeO3 type structure6,7. The orthorhombic dis- physical properties of SrRuO could not find evidence 0 tortion arises from the zig-zag tilting, along the c-axis, 3 in support of the alleged stabilization of a HS state or . and rotation, around the b-axis, of the corner-sharing 1 even suggested the reduction of the magnetic moment 0 RuO6 octahedra. Despite this distortion the RuO6 oc- from bulk values13,19. Further, a very recent study20 on 5 tahedra remain nearly regular7–9. In a localized picture, SRO/(111)STO has reported magnetization values sug- 1 the strong crystal field at the octahedral site splits the gestingaLSstate,incontradictionwiththe resultspub- v: Ru 4d bands of the Ru4+ ions into eg and t2g levels, lished earlier17,18. Understanding the stability of the Xi Tleahdeoinrgetticoalaclaolwcuslaptiinon(Ls1S0)–1t342gancdonafiXgu-rraaytimonagwnietthicSci=rcu1-. magneticgroundstateofSrRuO3 isobviouslyaveryim- portantaspect for controlling the magnetic properties of r lar dichroism (XMCD) study14 suggest that the orbital a heterostructures involving SrRuO as conducting layer. 3 moment in SrRuO should be quenched. High magnetic 3 In this work we address two questions: 1) whether field measurements on a bulk single crystal give a satu- compressive strain can induce a spin state transition in rated magnetization of 1.6 µ /Ru ion15, a value similar B SRO/(111)STO and 2) whether the orbital moment is to the orderedmagnetic moment determined by neutron quenched. To this end, we have performed an inves- diffraction experiments2,9. tigation of the XMCD signal at high magnetic field at While the technology for growing high quality the Ru L edges of SrRuO films under different com- 2,3 3 SrRuO thin films on (001) oriented SrTiO substrates, pressivestrains (trigonalstrainfor the case of(111)STO 3 3 SRO/(001)STO, was developed long time ago and is substrate and tetragonalstrainfor the case of (001)STO well known5, the systematic growth of thin films on substrate) compared with the case of a SrRuO single 3 (111) oriented SrTiO substrates, SRO/(111)STO, is crystal. XMCD is a well-established technique to study 3 quiterecent16,17. VerysurprisinglythefirstSQUIDmea- local magnetic properties. The XMCD signal can be an- surements of SRO/(111)STO films have provided a sat- alyzed by means of sum rules21,22, allowing for a direct 2 angle X-ray scattering. Details of their preparation and s) 4 a) XAS E surface structure characterizationare reportedin Ref. 24. Large nit E surface single crystals of SrRuO were grown by floating zone u 3 b. 3 technique. The purity and quality of the crystal were ar checked by x-ray diffraction. Susceptibility measure- ( y 2 ments using a MPMS squid magnetometer show a bulk sit Film SRO(27nm)/(111)STO ferromagnetic transition at T = 160 K for the single n c nte 1 crystal, and between 154 K and 147 K for the films de- I pendingonthefilmthickness. Thex-raylineardichroism 0 s) (XLD) and x-ray magnetic circular dichroism (XMCD) b. unit 0.2 b) XLD E surface - E surface epxeprfeorrimmeedntsatatthteheBLR2u9-LB2,o3reeadsgebsea(m28l0in0e-3a0t00theeVA) LwBerAe ar 0.1 Film SRO(27nm)/(111)STO synchrotron radiation facility in Barcelona. The energy y ( 0.0 resolutionwas1.4eVandthedegreeofcircularpolariza- sit-0.1 n tion delivered by the Apple II-type elliptical undulator e-0.2 Int 2800 2850 2900 2950 3000 was adjusted to 70% as balanced trade off between de- greeofpolarizationandphotonfluxrequiredwhenwork- Photon Energy (eV) ing at high photon energies and high undulator harmon- FIG. 1: Ru-L2,3 XAS spectra of a 27 nm SRO/(111)STO ics. The degree of linear polarization for XLD is close film for linearly polarized light coming in with the electric to 100%. The XMCD signal was measured using a mag- field vector E normal [dark (blue) lines] and parallel [light neticfieldof6Teslawiththesampleatatemperatureof (red) lines] to thefilm surface. 50 K.The spectra were recordedusing the total electron yieldmethod(bymeasuringthesampledraincurrent)in achamberwithavacuumbasepressureof2x10−10mbar. experimentaldeterminationofthedesiredquantumnum- Thesinglecrystallinesamplewascleavedinsitutoobtain bers L and S . The energy separation between Ru L z z 3 acleansamplesurfacenormaltothe(110)direction. The and L edges of about 150 eV is much larger than the 2 XASspectrawerecollectedinbothB =6Tand-6Tap- multiplet effects (a few eV), and therefore the spectra areverysuitablealsoforspinsumruleanalysis22. Inad- plied fields and ingroups offour or quartet(σ+σ−σ−σ+ or σ−σ+σ+σ−, where σ+ and σ− indicate photon spin dition, the signal-to-background ratio at the L edges 2,3 parallel or antiparallel to the applied field, respectively) is higher thanatthe M edges. We wouldlike to stress 2,3 in order to minimize the effect of any time dependence that obtaining a reasonable degree of circular polarized in the X-ray beam on the measured spectra. lightatthephotonenergiesoftheLedgesof4delements is challenging23 and only thanks to the development of In Fig. 1 we report the Ru-L2,3 XAS measured on a the new BOREAS beamline this XMCD investigationof 27 nm SRO/(111)STO film for linearly polarized light the Ru L edges has been possible. In addition, a com- coming in with the electric field vector E normal [dark 2,3 parison of the line shape to full-multiplet theory can be (blue) lines] and parallel [light (red) lines] to the film madetounraveldetailsofthewavefunctionsformingthe surface. The Ru 2p core-hole spin-orbit coupling splits ground state. the spectrum roughly in two parts, namely the L3 (at hν ≈ 2840 eV) and L (at hν ≈ 2970eV) white lines re- Single crystalline thin films of SrRuO were grown on 2 3 gions. A clear linear dichroism (XLD) can be observed, SrTiO substrateswithdifferentorientationsbymetalor- 3 whichisanindicationthatthefilmisunderin-planecom- ganic aerosol deposition. Thin SrRuO films grown on 3 pressivestrain. Infact,in-planecompressivestrainleads (001) and (111) oriented substrates were determined by to a trigonal elongation of the RuO octahedron along X-ray diffraction (XRD) to have (100)c and (111) orien- 6 the (111) axis. As consequence, the t orbitals are split tation, respectively. (In this report, we use pseudocubic 2g in a and eπ orbitals, with the a orbital lying higher notation for SRO films. (110)orthorhombicand (101)or- 1g g 1g in energy and, hence, having more holes. The experi- thorhombicisequivalentto(100)cand(111)inthepseu- mentallyobservedlargerspectralweightforE normalto docubic notation). The XRD results show that the films thefilmsurfaceisaresultoftheunevenholedistribution grownon(111)orientedsubstratesexhibitanelongation among the t orbitals induced by the strain. oftheout-of-planelatticeconstant(3.946(1)and3.950(1) 2g ˚A for the 80 and 27 nm thick films, respectively) com- The top panel of Fig. 2 shows the Ru-L XAS 2,3 pared to bulk SrRuO (≃ 3.93 ˚A18). This systematic measured on SRO/(111)STO and SRO/(001)STO films 3 evolution of the out-of-plane constant with film thick- and, for comparison, on a SrRuO single crystal. The 3 ness (the thinner the film, the larger the out-of-plane XAS spectra were taken using circular polarized light constant)isaneffectofthestrain: undercompressivein- with the photon spin parallel (σ+, red curves) and an- − plane strain the in-plane lattice constant shrinks, while tiparallel (σ , blue curves) aligned to the magnetic the out-of-plane lattice constant becomes elongated, in field. The difference spectrum (σ− − σ+), i.e., the order to roughly preserve the unit cell volume20. The XMCD spectrum, is reported in the bottom panel of thickness of the SrRuO films was determined by small- Fig. 2. The spectra were collected with the beam 3 3 x 22 L3 XAS a) zB B 20o 20 Film SRO(27nm)/(111)STO L2 z grazing incidence B surface x 18 Film SRO(27nm)/(111)STO SrRuO3film s)16 normal incidence B surface SrTiO3 substrate nit b. u14 Fnoilmrm SaRl iOnc(i8d0ennmce)/(111)STO B surface Normal incidence Grazing incidence r (a12 FIG.3: Setupof theXMCD experiments: themagnetic field y Film SRO(45nm)/(001)STO B is applied parallel to the Poynting vector of the circularly sit10 grazing incidence B surface polarized photons and forms an angle of 90o (20o) in normal n e (grazing) incidence with the sample surface. nt 8 Film SRO(45nm)/(001)STO I normal incidence B surface 6 of an out-of-plane easy axis for SrRuO3 films grown on single crystal STO as reported in literature17,20. The reduced mag- 4 grazing incidence B surface netic anisotropy shown by our XMCD measurements in the case of SRO/(111)STO films with respect to 2 single crystal SRO/(001)STO film is in fair agreement with previous normal incidence //(110) B surface SQUID measurements20. Both XAS and XMCD spec- 0 2.2 trameasuredontheSRO/(111)STOandSRO/(001)STO L2 b) films appear fairly identical to those measured on the 2.0 L3 Film SRO(27nm)/(111)STO bulk single crystal, without clear evidence of changes in 1.8 grazing incidence B surface thespectrallineshapeandinthesizeoftheXMCDsignal that otherwise could suggest a different spin state. Film SRO(27nm)/(111)STO 1.6 normal incidence B surface 1s).4 Sample incidence Lz/2Sz Lz 2Sz 011nsity (arb. unit...802 FgFnFnrooiiilllammmrrzmm inSSSaagRRRll iiiOOOnnnccc(((ii448iddd550eeennnnnnmmmcccee)))e/// (((001001111)))SSSTTTOOO BB ssuurrffaaccee SSSSRRRROOOO((((82440755CCnnnnmmmmrryy))))ss////tt((((aa1100ll11001111))))SSSSTTTTOOOO BBBBBB//⊥⊥⊥⊥//ssssssuuuuuurrrrrrffffffaaaaaacccccceeeeee 000000......000000110111 000000......000000120312 112111......970995 0e.6 B surface SRO(27nm)/(111)STO B//surface 0.01 0.02 1.8 nt I single crystal 0.4 grazing incidence B surface TABLE I: L/2S ratio, orbital and spin moment as estimated 0.2 using sum rules. The values were divided by a factor 0.7 single crystal normal incidence //(110) to take into account that the beam was only 70% circular B surface 0.0 polarized. -0.2 XMCD - The material metallicity and life time broadening (∼2 -0.4 eV)maylimittheinformationthatcanbeobtainedfrom 2850 2900 2950 3000 the lineshape about the Ru ground state. However, it is Photon energy (eV) possible to use the sum rules for XMCD developed by FIG. 2: Ru-L2,3 XAS spectra a) and XMCD spectra b) of Thole and Carra et al.21,22 to extract from our XMCD SrRuO3 films and bulk single crystal measured at T = 50 K data the orbital (Lz) and spin (2Sz) moments: and H = 6 T. The spectra are vertically shifted for clarity. sThhoewspaecsmtraaltlaakneinsoattronpoyr.maTlh(e90doa)sahneddgbrlaazcikngcu(r2v0eo)rienpcriedseennctes L = 4 · RL2,3(σ+−σ−)dE ·N , (1) theedge jump. z 3 RL2,3(σ++σ−)dE h R (σ+−σ−)dE−2R (σ+−σ−)dE in grazing (B//surface) and in normal (B⊥surface) in- 2S +7T =2· L3 L2 ·N , cidence, see Fig. 3 for experimental geometry. The z z R (σ++σ−)dE h L2,3 XMCDsignalislargerforB⊥surfacethanforB//surface (2) by about 30% for the SRO/(001)STO film, and by For ions in octahedral symmetry the magnetic dipole about 5% for the 27 nm SRO/(111)STO film. The momentT isasmallnumberandcanbe neglectedcom- z anisotropy of the XMCD signal agrees with the picture pared to S 25. The number of holes in the 4d shell was z 4 4 10Dqeff (eV) s) Film SRO(27nm)/(111)STO XAS -10 2.0 2.2 2.4 2.6 2.8 nit L3 B surface XMCD u rb. L2 -11 2y (a V)-12 nsit gy (e-13 nte ner 3.0 SrRuO3 I E-14 V) 0 q (eeff2.5 -15 0D2.0 S = 0 2830 2835 2840 2845 2960 2965 2970 2975 2980 1 S = 1 0.0 0.5 1.0 1.5 Photon energy (eV) 10Dqionic(eV) S = 2 -16 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 aFI2G7.n4m: ZtohoinmfiolfmRouf-LSr2R,3uisOo3trgorpoiwcnXAonS(a1n1d1)XoMriCenDtesdpeScrtTraiOo3f 10Dqionic(eV) substrate. Theverticaldashedlinescorrespondstotheenergy FIG.5: EnergyleveldiagramoftheRu4+ ionasafunctionof position of themaxima in the XASand XMCD spectra. theionic crystal field 10Dq in a cubiclocal symmetry. Black dotted, blue solid and red dashed lines correspond to levels with S = 0, 1and 2 spin state, respectively. The inset shows estimated to be about N = 5.2 by our cluster calcula- the evolution of the effective crystal field versus ionic crystal h tions, in agreement with previous estimates26, reflecting field. Theverticalmagentasolidlineindicatesthe10Dqeff of the highly mixed p − d covalency of the ground state SrRuO3asobtainedbythesimulationoftheXASandXMCD spectra. in SrRuO . In estimating the XAS intensities, the edge 3 jumpbackground,describedasarctanfunction,hasbeen subtracted from the XAS spectra (dashed curve in Fig. parameters. In the simulations we considered a RuO 6 2). The results of the application of the sum rules are cluster with a cubic symmetry as the octahedra in bulk reported in table 1. The orbital moment Lz is found SrRuO3 are fairly regular7–9. The calculations were per- to be almost quenched for all samples, including the formed using the XTLS 8.3 code30 with the parameters SRO/(111)STO. The spin contribution to the magnetic given in Ref. 31. We applied to the spectra additional momentinthe80and27nmSRO/(111)STOfilmsinnor- Gaussian (1.4 eV) and Lorentzian (2 eV) broadening in mal incidence is found to be close to the value expected order to take into account experimental resolution and for a S = 1 spin state and very similar to that found lifetime effects, respectively. for the bulk single crystal27. These results are in clear In Fig. 5 we report the energy level diagram of the contradiction with the much larger saturated moment Ru4+ ion as a function of the ionic crystal electric field values reported earlier18 from SQUID measurements on 10Dq in a cubic local symmetry. The energy dif- ionic SRO/(111)STO films. ference between the two S = 2 levels with orbital oc- Asmentionedbefore,thespectraseemtoberatherfea- cupation t32ge1g and t22ge2g (bottom and top red dashed tureless, but a closer look reveals that for both L and line in Fig. 5, respectively) can be taken as a measure 3 L2 edges the maximum in intensity of the XAS spec- of the effective crystal electric field 10Dqeff, i.e. the trum lies 1.5 eV higher in energy position than that splitting between t2g and eg levels including the effect of of the XMCD spectrum (see Fig. 4). We see exactly the hybridization with the oxygens. The diagram shows the same difference in energy position in the spectra of that for 10Dqeff > 2.15 eV (10Dqionic > 0.41 eV) the all samples. A similar energy shift of the XMCD peak level with configuration t42ge0g is the lowest energy level with respect to the XAS peak was previously observed (bottom solid blue line) and the ground state of Ru4+ for the Ru-M edges14 and can be understood consid- ion has S = 1 spin state. As the crystal field is re- 2,3 ering that only the t orbitals contribute to the XMCD duced across the critical value of 10Dq ≤ 2.15 eV 2g eff signal, while both t and e orbitals contribute to the (10Dq ≤ 0.41 eV) the t3 e1 level (red dashed line) 2g g ionic 2g g XAS spectrum with the XAS maximum corresponding becomesthelowestenergylevelandtheHSstateisstabi- to the signal from the unoccupied e levels. Therefore, lized. The non-magneticS =0state (black dottedlines) g this energypositiondifference providesa veryimportant liesalwaysmuchhigherinenergyandneverbecomesthe information as it reflects the crystal field splitting 10Dq ground state for any value of the cubic crystal field. between the t and e orbitals. In order to determine In Fig. 6 we show the comparison of the simulated 2g g quantitatively 10Dq we have performed simulations of XAS and XMCD spectra at the L edges with the 2,3 the XAS and XMCD spectra using the well-proven full- experimental spectra measured on SRO/(111)STO film. multiplet configuration-interaction approach28,29. It ac- The simulated XAS and XMCD spectra were calculated countsforthe intra-atomic4d−4dand2p−4dCoulomb for different values of the ionic and effective crystal field interactions, the atomic 2p and 4d spin-orbit couplings, splitting. Forthesakeofclarity,thespectrawerenormal- the oxygen 2p−4d hybridization, and local crystal field izedtotheheightofthepeakandtheXMCDsignalatthe 5 state becomes stable only for smaller crystal field split- Film SRO(27nm)/(111)STO a) b) ting, 10Dq < 2.15 eV. In the hypothesis of a HS spin B surface eff XAS state as a ground state the simulated XMCD spectrum experiment L3 L2 XMCD looksverydifferent fromthe experimentalone: 1) atthe L and L edge the XMCD lineshape is not anymore 3 2 s) XAS asymmetric; 2) at the L2 edge the XMCD maximum oc- nit calculation XMCD curs at higher photon energy than the XAS maximum, u b. 10Dqeff=2.84 eV which is opposite to what has been experimentally ob- ar =2.77 served. As it can be seen in the energy level diagram y ( =2.70 reported in Fig. 5 SrRuO is located very far from the sit b e s t =2.62 3 n =2.55 stability region for the HS state. e nt =2.48 I =2.41 To summarize, we have used XMCD spectroscopy to =2.34 =2.26 investigate the local magnetism in thin films of SrRuO3 L S =2.19 grown on (111) and (001) oriented SrTiO3 substrates. H S =2.12 We have found that the orbital moment is almost =2.05 =1.98 quenched and the spin is close to the value expected for =1.90 a S = 1 spin state. From a comparison of the exper- 2835 2840 2845 2965 2970 2975 imental with simulated spectra we could determine the Photon Energy (eV) effective crystal field. The hypothesis of a compressive FIG. 6: Experimental (top) and simulated (below) XMCD strain-inducedspin state transition,as proposedin liter- andXASspectrafordifferent10Dqeff valuesattheL3(panel atureonthebasisofSQUIDmeasurements,canberuled a)andL2 (panelb)edges. Forabettercomparison thespec- outas the stabilizationofthe high spin state with S =2 trawerenormalizedtothepeakheightandtheRuL3XMCD would be too costly in energy. spectrawasreversed. Theverticaldottedlinescorrespond to the energy position of the maxima in the XAS and XMCD The XMCD experiments were performed at the spectra. BOREAS beamline of ALBA Synchrotron with the col- laboration of ALBA staff. The research leading to these L edge was reversed. The calculated spectra show that resultshas receivedfunding fromthe EuropeanCommu- 3 the peak position depends on the value of crystal field nity’s Seventh Framework Programme (FP7/2007-2013) splitting. The experimental energy separation between under grantagreementn.o312284. Q.Liureceivedfinan- the maxima of the XMCD and XAS spectra can be cor- cial support from the European Union through the ITN rectlysimulatedfor10Dq =2.62eVandthelineshape Soprano Network (Grant No. PITN-GA-2008-214040). eff of the calculated spectra is fairly similar to that of the K.-T. Ko acknowledges support from the Max Planck- experimental spectra. For such a value of 10Dq the POSTECH Center for Complex Phase Materials (No. eff Ru4+ ions are in a LS S =1 ground state. The HS spin KR2011-0031558). 1 A. Callaghan, C. W. Moeller, and R. Ward, Inorg. Chem. 10 M.S.LaadandE.Mu¨ller-Hartmann,Phys.Rev.Lett.87, 5, 1572 (1966). 246402 (2001). 2 J. M. Longo, P. M. Raccah, and J. B. Goodenough, J. 11 Horng-Tay Jeng, Shi-Hsin Lin, and Chen-Shiung Hsue, Appl.Phys. 39, 1327 (1968). Phys. Rev.Lett. 97, 067002 (2006). 3 D. J. Singh, J. Appl.Phys. 79, 4818 (1996). 12 James M. Rondinelli, Nuala M. Caffrey, Stefano Sanvito, 4 P.B.Allen, H.Berger, O.Chauvet,L.Forro, T.Jarlborg, and Nicola A. Spaldin,Phys. Rev.B 78, 155107 (2008). A. Junod, B. Revaz, and G. Santi, Phys.Rev. B 53, 4393 13 Bongjae Kim and B. I. Min, Phys. Rev. B 89, 195411 (1996). (2014). 5 seeareviewinGertjanKoster,LiorKlein,WolterSiemons, 14 J.Okamoto,T.Okane,Y.Saitoh,K.Terai, S.-I.Fujimori, GuusRijnders,J.StevenDodge,Chang-Beom Eom, Dave Y. Muramatsu, K. Yoshii, K. Mamiya, T. Koide, A. Fuji- H. A. Blank, and Malcolm R. Beasley, Rev. Mod. Phys. mori, Z. Fang, Y. Takeda, and M. Takano, Phys. Rev. B 84, 253 (2012). 76, 184441 (2007). 6 R.J.BouchardandJ.L.Gillson, Mater.Res.Bull.7,873 15 G.Cao,S.McCall,M.Shepard,J.E.Crow,R.P.Guertin, (1972). Phys. Rev.B 56, 321 (1997). 7 C.W.Jones,P.D.Battle,P.Lightfoot,andW.T.A.Har- 16 J.Chang,Y.-S.Park,J.-W.Lee,andS.-K.Kim,J.Cryst. rison, ActaCrystallogr. Sect. C 45, 365 (1989). Growth 311, 3771 (2009). 8 S. Gardner, G. Balakrishnan, and D. Mck. Paul, Physica 17 A.Grutter, F. Wong, E. Arenholz,M. Liberati, A. Vailio- C 252, 303 (1995). nis, and Y.Suzuki,Appl.Phys. Lett. 96, 082509 (2010). 9 S. N. Bushmeleva, V. Y. Pomjakushin, E. V. Pom- 18 A. J. Grutter, F. J. Wong, E. Arenholz, A. Vailionis, and jakushina, D. V. Sheptyakov, and A. M. Balagurov, J. Y. Suzuki,Phys. Rev.B 85, 134429 (2012). Magn. Magn. Mater. 305, 491 (2006). 19 A. T. Zayak, X. Huang, J. B. Neaton, and K. M. Rabe, 6 Phys. Rev.B 77, 214410 (2008). Ru-M2,3 XMCDmeasurementsisonly∼0.6µB.However, 20 B. Lee, O.-U. Kwon, R. H. Shin, W. Jo, and C. U. Jung, astheauthorsadmitted,theelectron-core-holeinteraction, Nanoscale Res. Lett. 9, 8 (2014). which in the case of the M2,3 edges mixes p3/2 and p1/2 21 B.T.Thole,P.Carra,F.Sette,andG.vanderLaan,Phys. components, the use of scraped surfaces of polycrystalline Rev.Lett. 68, 1943 (1992). samples and the relative small magnetic field (2 T) might 22 P. Carra, B. T. Thole, M. Altarelli, and X. Wang, Phys. have caused an underestimation of the moment reported Rev.Lett. 70, 694 (1993). in thelatter work. 23 M.A.Tomaz,TaoLin,G.R.Harp,E.Hallin,T.K.Sham, 28 F. de Groot, J. Electron Spectrosc. Relat. Phenom. 67, andW.L.O’Brien,J.Vac.Sci.Technol.A16,1359(1998). 529 (1994). 24 M. Schneider, V. Moshnyaga, P. Gegenwart, Journal of 29 See the ”Theo Thole Memorial Issue”, J. Electron. Spec- Physics: Conference Series 200, 012178 (2010). trosc. Relat. Phenom. 86, 1 (1997). 25 Y.Teramura,A.Tanaka,andT.Jo,J.Phys.Soc.Jpn.65, 30 A. Tanakaand T. Jo, J. Phys. Soc. Jpn. 63, 2788 (1994). 1053 (1996). 31 U =3.0eV,U =4.0eV,pdσ=2.19eV,pdπ=-1.03eV dd pd 26 E. B. Guedes, M. Abbate, K. Ishigami, A. Fujimori, K. and ∆ = 1.0 eV. Slater integrals were reduced to 70 % of Yoshimatsu, H. Kumigashira, M. Oshima, F. C. Vicentin, the Hartree-Fock values. The Ru 4d spin-orbit parameter P. T. Fonseca, and R. J. O. Mossanek, Phys. Rev. B 86, was set to zero to avoid the J = 0 (singlet) state to form 235127 (2012). thegroundstate,andinstead,toallowtheS =1orS =2 27 The spin contribution obtained by Okamoto et al.14 from states to become stable.