Conference-ISSS-7-Magnetic Properties of Iron Ultrathin Films Intercalated in Graphene / Ni ( 111 )

We measured element specific magnetization curves for graphene/Fe/Ni(111) with a intercalated monolayer of Fe, obtained from external magnetic field dependence of X-ray absorption at Fe and Ni L3 edges. In-plane magnetic anisotropy was clearly confirmed in both Fe and Ni magnetic sites and the strong ferromagnetic coupling of the Fe spins with the Ni spins were found at the interface. A detailed analysis of X-ray magnetic circular dichroism (XMCD) spectra leads to significant values of in-plane and out-of-plane orbital magnetic moment at the Fe site, whose anisotropy is vanishingly small. [DOI: 10.1380/ejssnt.2015.312]


I. INTRODUCTION
Graphene is one of prospective candidates for the base material for developing spintronics devices [1,2], because ballistic electronic transport is realized in the graphene layer where spin-conserving current can be utilized if spin polarized electrons are injected effectively [3,4].Since the spin polarized electrons should be provided from a ferromagnetic electrode to the graphene layer through their interface, information on the magnetic properties and the spin polarization of valence electrons at the interface are essentially important to establish the spin injection technique practically.
Fabrication techniques of monatomic layered structures provide high-quality graphene on a Ni(111) single crystal and an iron-intercalated structure of graphene/Fe/Ni(111) in ultra high vacuum (UHV) environment [5,6].Post-annealing procedure after iron deposition onto graphene/Ni(111) promotes the intercalation of Fe underneath graphene, leading to inert effect of the Fe ultrathin film which is protected by graphene as passivation layer [6].It is expected that the artificial monatomic structure of graphene/Fe/Ni(111) can be utilized as source electrode of spin injection devices, because chemical and ferromagnetic state is maintained by the inert effect against atmospheric exposure.
Although investigations of macroscopic magnetic properties and their microscopic origin in graphene/Fe/Ni(111) are essential to examine a potential of spin injection, few magnetic studies have been so far carried out.X-ray magnetic circular dichroism (XMCD) spectra at the carbon K-edge have revealed an induced spin magnetic moment at the carbon site in the graphene layer, which is positive evidence for the spin injection [7].Additional XMCD analysis for X-ray absorption spectra at Fe L 2,3 edges have resulted in a certain value of spin magnetic moment in the Fe monolayer contacted with the graphene layer.However, the XMCD study has not reported the orbital magnetic moment and the magnetic dipole moment, whose axial components also have not been revealed along parallel and perpendicular direction to the sample surface.This analysis is inadequate to comprehend the electronic state of 3d orbitals, especially, anisotropy in the orbital magnetic moment that can cause magnetic crystalline anisotropy [8,9].Macroscopic magnetic properties in the ferromagnetic Fe layer are also important to design the spin devices.Magnetic anisotropy is crucial because it determines whether the direction of spins injected into the graphene layer is in-plane or out-of-plane.In the case of graphene/Ni(111), electron spins at the interface are aligned normal to the surface in spite of in-plane magnetization inside the substrate of Ni [10].For the graphene/Fe/Ni(111) structure, no information on the magnetization behavior has been provided including its magnetic anisotropy.
In this study, we have fully investigated magnetic anisotropy and local magnetic moment of Fe in graphene/Fe/Ni(111) in the ultrathin film limit, as well as Fe/graphene/Ni(111) structure, by means of XMCD in soft X-ray region.Element specific magnetization curves indicate Fe spins coupling with Ni spins in the substrate strongly, exhibiting in-plane easy axis of magnetization.Quantitative analysis of the XMCD spectra has derived all components of local magnetic moment including orbital magnetic moment, whose values have found to be fairly large without significant anisotropy.

II. EXPERIMENTAL
All the sample preparations and the XMCD experiments were performed in situ at HiSOR-BL14 in Hiroshima Synchrotron Radiation Center.The Ni(111) single crystal was cleaned by cycles of ion sputtering (Ar + , 1.0 kV) and annealing (600 • C, 15 min) until no contamination was detectable by Auger electron spectroscopy (AES).Graphene was grown on the clean Ni(111) surface by cracking propylene (C 3 H 6 ) at a temperature of 500 • C with exposing the substrate to an atmosphere of 1×10 −4 Pa of C 3 H 6 for 5 min.Iron were deposited onto graphene/Ni(111) at room temperature by e-beam evaporation.After the iron deposition onto graphene/Ni(111), post-annealing procedure at 330 • C for 5 min completes intercalation of the Fe atoms underneath the graphene layer.
X-ray absorption measurements were carried out at room temperature using total electron yield method.Monochromated synchrotron radiation was made incident on the sample surface, whose resolution and circular polarization were set to ∆E = 0.4 eV and P c = 70%, respectively.Magnetization curves (M−H curves) were ob-tained from magnetic field dependence of the absorption intensities at Fe and Ni L 3 edges.In the M−H curve measurements, normal (incident angle θ = 0 • ) and grazing (θ = 60 • ) incidence geometry were applied to detect the out-of-plane and in-plane magnetization, respectively.Periodically-amplified magnetic field was applied along the photon helicity vector (i.e.light incident direction), whose amplitude was ±0.3 T. The method of M−H curve analysis is based on a previous literature [11].
Spectra of XMCD were measured at the Fe L 2,3 absorption edges with a fixed photon helicity.Both normal (θ = 0 • ) and grazing (θ = 54.7 • ) incidence geometry were applied in the XMCD spectroscopy.For the normal incidence measurements, external magnetic field of 1.1 T was applied perpendicular to the surface.For the grazing incidence case, magnetic field of 0.3 T was applied parallel to the surface (i.e.in-plane direction).The direction of magnetic field was switched parallel and antiparallel to the photon helicity at each energy step during a phonon energy scan of spectral measurement.The absorption spectra for the direction of magnetization parallel (µ + ) and antiparallel (µ − ) to the photon helicity were recorded independently.The corresponding XMCD spectrum was obtained from µ + −µ − as difference spectrum.

III. RESULTS AND DISCUSSION
Figure 1 shows differential spectra of Auger electrons excited with 3 keV electron beam, for Fe/graphene/Ni(111) before intercalation and graphene/Fe/Ni(111) where the thickness of Fe is 1.1 ML.The spectra are composed of Auger electron signals of C KLL, Fe LMM and Ni LMM, located around 275, 650 and 800 eV, respectively.Blue and red spectral curves correspond to the samples before and after the intercalation treatment, respectively, and these spectra are normalized to the amplitude of Fe LMM signal at 654 eV.After annealing for the intercalation, amplitude ratio of C KLL to Fe LMM becomes larger than before the process.Since escape depth of the Auger electron with characteristic energy of several hundred electron volts is order of an angstrom, Auger electron intensities are sensitive to overlayer thickness and enhanced by reaching the source point to a surface of specimen.The intercalation treatment makes graphene rise to surface of the layered structure, in contrast to Fe atoms moving Red and blue curves are XAS spectra measured with the circularly polarized light for two opposite magnetization direction µ+ (parallel to the photon helicity vector) and µ− (antiparallel to one), respectively.Green spectrum is generated from (µ++µ−), whose background component is notated as dashed line.The XMCD spectrum is directly obtained form (µ+−µ−), shown as black thick curves.XMCD spectra for the as-deposited sample are also plotted as black thin curves.
inside.The increase of the amplitude ratio of carbon to iron signal is consistent with the structural modification by the intercalation.Figure 3 shows element specific magnetization curves (M−H curves) of graphene/Fe/Ni(111) with the Fe thick-ness of 1.2 ML, derived from external magnetic field dependence of X-ray absorption intensities at the L 3 edges of Fe and Ni.For each measurement of Fe and Ni, a set of M−H curves for in-plane (notated as red curve) and out-of-plane (blue one) magnetization is obtained.For the measurement at Fe L 3 edge, M−H curves of Fe/graphene/Ni(111) before the intercalation treatment are also shown by thin curves.The in-plane and out-ofplane magnetization curves are originated from XMCD signals in grazing and normal incidence geometry, respectively.In the grazing incidence, the amplitude is proportional to a projection of in-plane spin magnetic moment along the incidence axis.In the normal incidence, a spin component purely perpendicular to the sample surface is detected.The amplitudes of M−H curves are normalized to height of the absorption peak at L 3 edges from the baseline in the pre-edge region.Firstly, it is noticed that the magnetic field dependent behavior of Fe and Ni spin is qualitatively same in both magnetization directions of in-plane and out-of-plane.Particularly the M−H loops for in-plane magnetization are found to be similar between Fe and Ni, and exhibit close values in coercive field and saturation magnetic field.This indicates a conformity behavior of the local spins toward external magnetic field, caused by a strong ferromagnetic coupling between Fe and Ni spins at the interface.Second noteworthy point is in-plane magnetic anisotropy in the intercalated Fe layer, evidenced by existence of a loop feature only in the in-plane magnetization curve.The situation that in-plane ferromagnetic order emerges at room temperature, is quite different from the case of Fe/graphene/Ni(111) without intercalation.The Fe simply deposited on graphene/Ni(111) does not show remanent magnetization any longer.The deposited Fe is not epitaxially grown on the graphene, and probably forms grain or island structures microscopically in contrast to the uniform layer structure of intercalated Fe.The disappearance of ferromagnetic order is assumed to be due to disconnection of the three dimensional Fe islands.Additionally, from the M−H curves of Fe/graphene/Ni(111), we note that applied magnetic field yields out-of-plane magnetization a little more easily than in-plane one.This is because the three dimensional growth weaken in-plane shape magnetic anisotropy.In contrast, the intercalated Fe layer shows in-plane magnetic anisotropy definitely, which is explained by ferromagnetic coupling between Fe and Ni spins and magnetic shape anisotropy in the Ni substrate contributing to in-plane magnetic easy axis.
In order to complete XMCD analysis of graphene/Fe/Ni(111), we have observed XMCD spectra at the Fe L 2,3 absorption edges for both samples fully magnetized in-plane and out-of-plane.Saturation magnetic field along the sample surface is confirmed to be smaller than 0.3 T from the M−H curve measurement.Applied magnetic field of 1.1 T perpendicular to the surface is adequate to saturate the magnetic moment against an anisotropy magnetic field that is estimated at 0.6 T approximately.According to optical sum-rules [12,13]  cidence.Figure 4 shows XAS and XMCD spectra for graphene/Fe/Ni(111) with the Fe thickness of 1.8 ML, measured at Fe L 2,3 absorption edges in normal and grazing incidence geometry with external magnetic field of 1.1 T. In the XAS spectra, two white line peaks are observed are observed at 708 and 721 eV, corresponding to the Fe L 3 and L 2 absorption edges, respectively.In the sum rule analysis, the amplitude of XMCD spectra is normalized by the integration intensity of µ + +µ − spectrum whose background is subtracted.Integration intensities of the normalized XMCD spectrum for normal incidence are evaluated at both regions of L 3 and L 2 edges, resulting in the perpendicular orbital magnetic moment (m ⊥ orb ) and the effective spin magnetic moment, which is a liner combination of the pure spin magnetic moment (m spin ) and the perpendicular dipole magnetic moment (m ⊥ T ).From the other measurement with the incident angle of 54.7 • , m spin and a liner combination of m ⊥ orb and m ∥ orb (in-plane orbital magnetic moment) are derived by a magic angle relation [14].In Table I, the complete component sets of local magnetic moment at the Fe atomic site are summarized for Fe thicknesses in the ultrathin limit, obtained as a solution of the simultaneous equations from the sum rule formula.The value of m spin is consistent with the previously reported [7], and found to be slightly larger than that of bulk Fe [15].Fairly large values of the orbital magnetic moments are also revealed.The values deviated from the bulk case are probably due to low dimensionality of the ultrathin layer structure of the intercalated Fe.An important finding here is that no significant difference between m ⊥ orb and m ∥ orb is observed.In other words, the anisotropy in the 3d orbitals of Fe is negligible, in spite of an anisotropic circumstance of the intercalated layer sandwiched with the other element layers.The isotropic situation in 3d orbitals is different from the case of ultrathin films of Co(111) with perpendicular magnetic anisotropy, which are grown on graphene or fcc(111) surfaces of metals [16][17][18].The difference in orbital anisotropy between magnetic elements is attributed to a variety of their valence band structures, because the orbital anisotropy and magnetocrystalline anisotropy are determined through a second-order perturbation of angular momentum between 3d electronic states [8].The perturbation process between occupied states of upper majority spin bands (derived from e g states) and unoccupied states of lower minority spin bands (derived from t 2g states) is essential for the perpendicular magnetic anisotropy in the Co(111) films, and a proximity of these key states to the Fermi level is crucial to enhanced the orbital anisotropy [19,20].In the case of Fe, the essential states are shifted to higher energy apart from the Fermi level because of smaller number of 3d electrons.This is reason why the orbital anisotropy of Fe is suppressed in our result.In order to examine an effect of the intercalation on the orbital anisotropy, we measured XMCD spectra of Fe/graphene/Ni(111) without intercalation at Fe L 2,3 edges, whose normalized XMCD curves are shown for both normal and grazing incidence in Fig. 4. Smaller amplitudes of the XMCD signals were observed than the intercalated sample, because of unsaturated magnetization that is confirmed in the M−H curves shown in Fig. 3.In unsaturation case, we can only discuss on the ratio of orbital magnetic moment to spin one from the sum rule analysis, instead of the complete set of local magnetic moment of Fe.The values of ratio are listed in Table II, for both as-deposited and intercalated samples.We note that the anisotropy of orbital magnetic moment is negligible in both cases although slight modification is found.Consequently, contact and ferromagnetic interaction with the local spin site of Ni substrate are crucial in the in-plane ferromagnetic order of Fe, and the effect of magnetocrystalline anisotropy is minor in the graphene/Fe/Ni(111) structure.This fact clarified from a microscopic viewpoint supports the macroscopic magnetic behavior of magnetic anisotropy similar to the original Ni substrate.

IV. CONCLUSION
We have fully investigated magnetic anisotropy and local magnetic moment in the graphene/Fe/Ni(111) system by means of XMCD spectroscopy.In the M−H curves, in-plane magnetic anisotropy of the Fe film was clearly observed as well as shape magnetic anisotropy of the Ni substrate with in-plane easy axis.It is revealed that the spin magnetic moments of Fe and Ni are ferromagnetically coupled and saturated in relatively small external magnetic field.We have completely determined the component set of Fe 3d local magnetic moment in the inter- calated layer.Although enhancement of spin and orbital magnetic moment is found as typical phenomena in ul-

FIG. 1 .
FIG.1.Differential spectra of Auger electrons for samples before and after the intercalation treatment.Blue and red curves correspond to spectra for Fe/graphene/Ni(111) and graphene/Fe/Ni(111), whose thickness of Fe is 1.1 ML.

FIG. 2 .
FIG. 2. Images of LEED for (a) graphene/Ni(111) and (b) graphene/Fe/Ni(111) where the thickness of Fe is 1.1 ML.Both images were collected at an electron energy of 128eV.

FIG. 4 .
FIG. 3. M−H curves for (a) Fe and (b) Ni elements of graphene/Fe/Ni(111) with the Fe thickness of 1.2 ML, are shown by thick curves.In the figure (a), M−H curves for the as-deposited sample are also shown by thin curves, whose amount of Fe is 1.6ML.The red and blue curves are corresponding to the measurements for in-plane and out-of-plane magnetization, respectively.

Figure 2
shows images of low energy electron diffraction (LEED) for graphene/Ni(111) and graphene/Fe/Ni(111) properly fabricated in our experiment.In both of the LEED images, sharp three-fold symmetry spots are found, coming from fcc(111) structure of the Ni substrate.The excellent LEED spots after the intercalation treatment mean good crystallinity after formation of the graphene/Fe/Ni(111) structure by the post-annealing.All the graphene/Fe/Ni(111) samples for our XMCD studies were fabricated in the same manner, whose signals of composition and no contamination were verified by AES.
, a complete component set of local magnetic moment can be analyzed, if the XMCD spectra of the identical sample are measured with full perpendicular magnetization in normal incidence as well as measured with saturated in-plane magnetization in grazing inhttp://www.sssj.org/ejssnt(J-Stage: http://www.jstage.jst.go.jp/browse/ejssnt/) e-Journal of Surface Science and Nanotechnology

TABLE I .
The complete component sets of local magnetic moment at the Fe atomic site are summarized, obtained from optical sum-rules.