Magnetooptic enhancement and magnetic properties in Fe antidot films with hexagonal symmetry
E. Th. Papaioannou, V. Kapaklis, P. Patoka, M. Giersig, P. Fumagalli, A. Garcia-Martin, E. Ferreiro-Vila, G. Ctistis
MMagnetooptic enhancement and magnetic properties in Fe antidot films withhexagonal symmetry
E. Th. Papaioannou ∗ and V. Kapaklis Department of Physics and Materials Science, Uppsala University, 75121 Uppsala, Sweden
P. Patoka
Helmholtz-Zentrum Berlin f¨ur Materialien und Energie GmbH, 14109 Berlin, Germany
M. Giersig and P. Fumagalli
Institut f¨ur Experimentalphysik, Freie Universit¨at Berlin, 14195 Berlin, Germany
A. Garcia-Martin and E. Ferreiro-Vila
Instituto de Microelectr`onica de Madrid (IMM-CNM-CSIC), 28760 Madrid, Spain
G. Ctistis † FOM Institute for Atomic and Molecular Physics (AMOLF),Center for Nanophotonics, 1098 XG Amsterdam, The Netherlands andComplex Photonic Systems (COPS), MESA+ Institute for Nanotechnology, University of Twente, The Netherlands (Dated: October 31, 2018)The magnetooptic and magnetic properties of hexagonal arrays of holes in optically thin ironfilms are presented. We analyze their dependence on the hole radius and compare the results to acontinuous iron film of same thickness. We observe a large enhancement of the magnetooptic Kerrrotation with respect to that of the continuous film, at frequecies where surface plasmon excitationsare expected. The spectral position of the Kerr maxima can be tuned by the size and the distancebetween the holes. Additional simulations are in very good agreement with the experiment andthus confirm the effect of the surface plasmons on the Kerr rotation. The altering of the magneticproperties by the hole array is also visible in the hysteretic behavior of the sample where a significanthardening is observed.
PACS numbers: 81.16.Be, 75.70.Ak, 75.75.+a, 78.20.Ls, 75.70.Kw, 75.60.-d
I. INTRODUCTION
Patterning holes into ferromagnetic thin films (anti-dots) is an effective way to engineer their magnetic prop-erties. The presence of antidots alters the demagneti-zation fields of the structures. At the same time theholes are pinning sites for domain walls. These twoparameters have been shown to influence the coerciv-ities and remanences , anisotropies , and switchingcharacteristics . In parallel, the antidot structures withnoble metals have been studied for their optical proper-ties after the pioneering work of Ebbesen et al. and thediscovery of extraordinary optical transmission of lightthrough these subwavelength structures at certain reso-nant frequencies or angle of incidence. The investigationshave not included the presence of a magnetic field sinceit is known that for plasmonic metals the influence ofa magnetic field is very low and thus the use of veryhigh magnetic fields would be necessary to observe inter-actions. However, by using ferromangetic materials theinteractions with a magnetic field is much stronger thanfor the noble metals that would in principle lead to mea-surable magneto-plasmonic interactions, even in the pres-ence of the absorptive losses for the plasmons. Indeed,the interaction of light with the ferromagnetic nanoscalearrays of holes in an applied magnetic field has shown ex- citing optical and magnetooptic properties. In particular,magneto-plasmonic interactions were observed by Ctistis et al. , where an extraordinary magnetooptic responseof antidot hexagonal arrays of Co was revealed, and byGonzalez-Diaz et al. who reported an enhancement ofthe magnetooptic response for Ni nanowires embedded inan alumina matrix due to the propagation of plasmonsin the nanowires. Furthermore recently, predictions ofa signicant enhancement of the magneto-optic transverseKerr effect were reported for the case of a magnetic/noblemetal film perforated with subwavelength slit arrays .In this work, we have succesfully fabricated subwave-length hole arrays in optically thin Fe films with differenthole sizes. We present a detailed study on the size de-pendencies of the magnetic and magnetooptic properties. II. EXPERIMENTAL DETAILS
The Fe antidot arrays were prepared on a Si(111) sub-strate using self-assembly nanosphere lithography withpolystyrene (PS) spheres. For a more detailed descrip-tion of the process we refer to Ref. and will give only ashort summary here. We used monodisperse PS sphereswith a diameter of 470 nm as a mask template. Af-ter the formation of a hexagonal closed-packed structure a r X i v : . [ c ond - m a t . m e s - h a ll ] D ec with the PS-spheres, the diameter of the spheres has beenshrunk by means of reactive ion-etching (RIE). The con-trol of the etching parameters enables us to control thediameter of the PS-spheres, here 248 and 297 nm , re-spectively. The thus obtained structure served as a maskfor the evaporation of the 100 nm thick iron films bymeans of molecular-beam epitaxy and a base pressure of10 − mbar . The used evaporation conditions lead to apolycrystalline structure of the iron films. In order toobtain a smooth and continuous Fe film, a seed layer of2 nm Ti was deposited prior the Fe evaporation. Toprevent subsequent oxidation of the surface a cappinglayer of 2 nm gold was deposited on top. Thereafter,a chemical treatment dissolved the PS-spheres leavingthe metal film with a hexagonal array of holes behind[see Fig. 1(a)]. Due to this treatment and the measure-ments under ambient conditions, a final oxidation of theFe films can not be excluded. To characterize the to-pography as well as the micromagnetic behavior of thesamples, an atomic force microscope (Nanosurf MobileS) operating in tapping mode was used. By measuringthe phase shift of the cantilever oscillation we were ableto observe the magnetic interaction of the cobalt coatedcantilever tip with the underlying sample (oscillating fre-quency 75 kHz , spring constant 2 . N/m ). An AFM im-age of the 3-dimensional reconstruction of the patternedFe film with holes diameter 248 nm is presented in theinset of Fig. 1(a). The MFM measurements were per-formed at room temperature without applied externalfield implying that the remanent magnetic state of thesample is studied. Additionally, we performed micromag-netic simulations using an object-oriented micromagneticframework (oommf) from NIST to explain micromag-netic distributions to the spin orientation imaged withthe magnetic-force microscope.Furthermore, the macroscopic optical and magneticproperties of the films were investigated using a magne-tooptic Kerr spectrometer in the longitudinal and polarconfiguration under ambient conditions. A schematic ofthe longitudinal Kerr setup is shown in Fig. 1(b). Thelongitudinal Kerr magnetometer, based on the use of aphotoelastic modulator (PEM) operating at 50 kHz , al-lowed the simultaneuous meauserement of both Kerr ro-tation and ellipticity. The incident polarization set bythe first polarizer, corresponds to s-polarized light. Afterthe sample, the beam passes through the modulator andan analyser. The PEM retardation axis was parallel tothe plane of incidence. The analyzer is oriented at 45 ◦ with respect to the PEM retardation axis. The measure-ments were performed at a photon energy of 1 . eV atan angle of incidence of 24 ◦ with respect to the samplesurface normal. The polar Kerr magnetometer, based onthe use of a Faraday modulator , was used to recordpolar Kerr hysteresis loops at selected energies at themaximum magnetic field of 1 . T . Furthermore, polar-MOKE spectroscopy at room temperature (RT) was per-formed with an applied magnetic field of 1 . T , at anangle of incidence of 5 ◦ with respect to the sample surface x-Direction ( μ m)0 20020 y - D i r e c t i on ( μ m ) (a) x y (b) FIG. 1: (color online) (a) Atomic force micrograph of a100 nm thick Fe film with hexagonal arrays of holes. Thepitch size of the array is a = 470 nm and the hole size is d = 248 nm . A large defect-free area is visible. The insetshows a 3D representation of the atomic force micrograph.(b) Schematic of the experimental setup for the longitudinalKerr measurements. The angle of incidence of the light is 24 ◦ . normal and at photon energies between 1 . . eV . III. RESULTS AND DISCUSSION
Typical reflectivity spectra are plotted together withpolar magnetooptic spectra in Fig. 2. The reflectivityspectra shown in Fig. 2 (top graph) are measured for twodifferent hole arrays with 248 nm (circles) and 297 nm (squares) hole diameter, respectively. The spectra showsimilar spectral behavior. The main features are the min-ima in reflectivity at ∼ . eV and at ∼ . eV . Theseminima are a result of the resonant coupling of light tosurface plasmon (SP) excitations of both interfaces of the R e l a t i v e R e f l e c t i v i t y Wavelength (nm)-0.6-0.5-0.4-0.3-0.2 K e rr R o t a t i on ( deg ) Photon Energy (eV)
248 nm holes 297 nm holes 100 nm Fe film
FIG. 2: (color online) Relative reflectivity (top) and mag-netooptic polar Kerr (bottom) measurements for two differ-ent Fe hole diameters with holes diameter of 248 nm (opencircles) and 297 nm (open squares), respectively. The Kerrspectrum of a continuous Fe film (filled circles) is shown asreference. The Kerr spectra are recorded at the saturationstate of the samples at magnetic field of 1 . T . The mag-netooptic response of the antidot samples is strongly affectedand enhanced at ∼ . eV , and at ∼ . eV as a direct re-sponse to surface plasmon excitation. Fe film perforated with a hexagonal array of holes witha lattice constant of a = 470 nm . Figure 2 (bottomgraph) shows the magnetooptic spectra of the two holearray samples. Additionally, the spectrum of a Fe filmof same thickness as for the hole arrays is shown. In thelow energy regime, the Kerr rotation is smaller than thecontinuous film, which is expected, if we consider that wehave a smaller amount of magnetooptically active mate-rial in the hole arrays. Nevertheless, above 2 . eV a verystrong enhancement of the Kerr rotation is observed forboth hole arrays with two maxima around 2 . eV and3 . eV . The maximum Kerr rotation values at these en-ergies are nearly two and three times bigger than the val-ues of the continuous film. It is worth to notice that thespectacular enhancement is visible at the same energiesas the features in the reflectivity spectra, mirroring thefact that surface-plasmon excitation of the array influ-ences the magnetooptic properties of the Fe film. Above4 eV , Kerr rotation is decreasing, however it is still much R e l a t i v e R e f l e c t i v i t y continuousd = 150nmd = 200nmd = 248nmd = 297nm Photon Energy (eV) -0.8-0.6-0.4-0.20 K e rr R o t a t i on ( deg ) FIG. 3: (color online) Calculated reflectivity and polar Kerrrotation versus photon energy. The array parameters for thehole arrays are as follows: a = 470 nm as pitch size for allarrays, d = 150 nm , d = 200 nm , d = 248 nm and d =297 nm . The two bigger hole diameter correspond to theexperimental spectra shown in Fig. 2. The spectrum for theFe film is also shown. A continuous enhancement of the polarKerr rotation as the reflectivity decreases is observed as thehole diameter increases. bigger than the one of the continuous film. The resultsare similar to earlier measurements on Co hole arrays .In both cases we have a strong enhancement of the po-lar Kerr values which is supported by surface-plasmonresonances at specific energies.For a more thorough understanding of the underly-ing effects, a theoretical approach has been performed.The calculations employed a scattering matrix methodspecifically adapted to consider magnetooptic effects .In Fig. 3 calculated spectra for the reflectivity (topgraph) and the polar Kerr rotation (bottom graph) ofthe corresponding two hole arrays of Fe and the continu-ous film are presented. Furthermore, in the same graph,we have simulated two more samples with hole diame-ters of 150 nm and 200 nm in order to reveal how thediameter d rules the behavior of such arrays.The simulations are in good qualitative agreement withthe experimental results and provide guidelines to arti-ficially control the optical and magnetooptic propertiesof the antidots. The calculated reflectivity curves ex-hibit similar features for all samples. The first minimumis almost constant at 2 . eV indicative of the fact thatthe optical response depends on the interhole separation a = 470 nm , that is the same for all the samples. Theinteraction of light with the hexagonal periodicity of holearrays leads to enhancement of plasma oscillations of theelectrons (minimum in reflectivity) at resonant frequen-cies. The second minimum is much broader and shiftsslightly towards lower energies as the diameter of theholes increases. The features of the calculated curvesagree quite well with the experimental reflectivity of thetwo samples at the positions of the two minima at 2 . eV ,and 3 . eV .The enhancement of plasmon oscillations at specificenergies is coupled to the magnetooptic response. Thesimulations of the Kerr spectra reproduce the experi-mental enhancement of the magnetooptic activity at theend of the visible region and in the UV. The energy po-sitions of the maxima agree well with the experiment.Even more, the simulations reveal clearly the behaviorof the antidots with different diameters. The Kerr re-sponse is enhanced as we go to higher diameters and thereflectivity decreases. The presence of bigger holes bringsthe nanoholes closer and enables the excitation of surfaceplasmons to interact more sufficiently with the adjacentnanoholes. As a result the enhancement of the Kerr effectis maximized at the resonant frequency of 2 . eV for thesample with d = 297 nm . At the same time, by increas-ing the diameter a shift to lower energies for the secondKerr maximum at energies above 3 . eV is observed.The calculated absolute rotation values are differentthan the experimentally measured. The peak at 2 . eV is more pronounced in the simulation than the one at3 . eV . The difference could be attributed to an insuffi-cient description of the optical constants of the materials,especially inside the holes. It is expected that the pres-ence of an oxide Fe layer at the side walls of the holeswill modify the refractive index n of the material andconsequently changing the plasmonic characteristics re-sponsible for the magnetooptic enhancement. Recently,Gonzalez-Diaz et al. reveal the role of the increase ofthe refractive index of a material that fills the pores inthe magnetooptic response by observing a red-shift in theKerr rotation maxima.To investigate the magnetic behavior closer, hysteresisloops both in longitudinal and polar configuration weremeasured. Figure 4 shows the results for a Fe film (toprow), a 248 nm hole array (middle row), and a 297 nm hole array (bottom row), respectively. The hysteresiscurve for the continuous film, taken in longitudinal con-figuration[Fig. 4(a)], confirms that the easy axis of themagnetization lies in the film plane due to the shapeanisotropy of the Fe film. Keeping the applied magneticfilm in-plane but in different directions of the 2D unitcell, covering thereby the full 360 ◦ range, enables us todetermine that the sample does not show any in-plane -1.00.01.0 K e rr R o t a t i on ( a r b . un i t s ) -1.00.01.0-1.00.01.0 -100 0 100 Magnetic Field (mT) -1.00.01.0-1.00.01.0-1.00.01.0 K e rr R o t a t i on ( a r b . un i t s ) -1000 0 1000 Fe -film248 nm holes297 nm holes (a) (b)
FIG. 4: (color online) Longitudinal (a) and polar Kerr (b)hysteresis loops for a Fe thin film (closed circles) and fortwo antidot Fe samples with d = 248 nm (open circles) and297 nm (open squares) hole diameter. The lattice constant ofthe antidot array is a = 470 nm for both samples. The y-axisis normalized to unity for a better comparison. The mag-netic behavior is dominated by the size of the holes. Highercoercivity and the appearance of out-of-plane magnetizationcomponents with increasing hole size is visible. anisotropy, as expected for a polycrystalline film. Sincethe material itself does not possess an intrinsic magneticanisotropy, the different hysteresis loops that we observefor the antidot films indicate the dominant role of the sizeand the arrangement of the holes in the reversal magneticbehavior. The two patterned samples exhibit a larger co-ercivity ( H C ≈ mT ), which is twice as large comparedto the continuous film. The trend of magnetic hardeningfor the hole arrays can be attributed to the presence ofthe holes.In particular, the holes introduce large areas of air-metal boundaries, only around 70% of the surface is cov-ered with material. As a consequence these boundariesmodify the demagnetizing field distribution in the film.At the same time they serve as domain wall pinning sites.Even more, the shape of the hysteresis curves is markedlymodified by the presence of the nonmagnetic holes, theyare more squared in shape.Although the hexagonal arrangement of the holesshould introduce a threefold in-plane anisotropy as ob-served in micromagnetic simulations , measurements ofhysteresis loops in various in-plane magnetic field direc-tions do not show any changes, neither in coercive field( H C ) nor in saturation field ( H sat ). The absence of ahard and easy in-plane axis can be explained by the factthat the illumination spot during the experiment is notfocussed and covers an area ( ∼ mm ), which probesmore than one structural domain. Therefore, we aver- y D i r e c t i on ( µ m ) x Direction (µm) (a) (b) FIG. 5: (a) Magnetic-force micrograph of a nanohole array( d = 248 nm ) with a 100 nm thick Fe film without appliedexternal magnetic field. The color scale describes differentmagnetization orientations. (b) Micromagnetic simulation ofthe same structure as in (a) in remanence. The spin orien-tation is denoted by the arrows and the color scale, showingmainly an in-plane orientation affected by the holes. age over all possible orientations breaking thereby thein-plane shape anisotropy.Figure 4(b) shows hysteresis measurements in the po-lar configuration. As can be seen from the graphs, allsamples have their hard magnetization axis out of plane.For the reference sample, the hysteresis loop shows a typ-ical hard-axis behavior with no remanence and a satura-tion field that we have hardly reached with our setupof B = 1 . T . The hole arrays, however, have mod-ified the hysteresis loops. By increasing the hole sizethe saturation field decreases strongly and reaches thevalue of H sat = 1 T for the sample with 297 nm holediameter. Simultaneously, a small hysteresis and rema-nence appear. The local dipolar fields introduced by thehole edges give rise to the out of plane magnetizationcomponents in competition with the intrinsic in planeanisotropy of the samples.Magnetic force microscopy was used to visualize theformation of magnetic domains in such systems in theremanent state. Figure 5(a) shows a magnetic force mi-crograph of the sample with the 248 nm holes. Brightand dark regions are distributed among the holes withoutany correlation to the simultaneously recorded topogra-phy (not shown here). The MFM technique is based onthe interactions of the tip with the magnetic charges inthe sample. This interaction gives information about thestray field of the sample. The stray field of the pat-terned sample is strongly influenced by the presence ofthe holes since the magnetic field lines can close throughthe holes. The strong contrast can be attributed to smallout-of-plane components (see Fig. 4(b)), creating mag-netic poles that give rise to dark or bright regions. Sim-ilar domain configuration was observed in a Ni film of55 nm thickness but for much smaller hole diameter of50 nm . Even though the contrast is not so large inbetween neighboring holes, one can recognize upon sixdarker regions (different domain configurations) aroundall the holes separated by lighter areas. Figure 5(b) shows micromagnetic simulations per-formed in remanence. The boundary conditions of thesimulation are open and the calculated cell is displayedin Figure 5(b). The dimensions in the simulation areidentical with that of the real structure and are given tobe 1 . µm × . µm and the cell size of 10 nm was usedin order to reduce computation time. We also calculatedthe magnetic configuration for a larger structure withmore than the displayed structural unit cell but foundno differences between the calculations. The magnetichistory of the sample in the simulation is set to be thesame as in the experiment during the MFM measure-ments and is as follows: Starting with a random mag-netization we first magnetize the sample in the negativex-direction with an applied field of B = 1 T and thenturn off the field and leave the magnetization relaxing.The parameters used for the calculations are: saturationmagnetization M S = 1 . × Am − , exchange constant A = 21 × − Jm − , and a cubic anisotropy with ananisotropy constant of K = 48 × Jm − . As conver-gence criterion the misalignment between magnetizationand effective field was used and set to be lower than 10 − in each computation cell. The film thickness was 100 nm .The arrows in the picture denote the in-plane spin orien-tation of the computation cells. It is visible that the holesaffect the spin orientation. In particular, the remanentspin configuration can generally be divided in differentarrangements. One group is along the x-direction (0 ◦ with the x-axis). Domains are pinned along this direc-tion and they are placed in the central region among ad-jacent holes. Different groups of spin configurations areformed having an angle ± ◦ , ± ◦ , ± ◦ to the x-axis.These configurations smoothly circle around the holes asa result of minimizing the total energy between two com-peting terms: the magnetostatic and the exchange energyacross the domain walls. We can see that the simulationis in qualitative agreement with the experiment. The re-vealed correlation between the formation of domains andthe periodic structure results in pinning effects, whichjustify the magnetic hardening observed in the MOKEloops in Fig. 4. IV. CONCLUSIONS
The magnetic and magnetooptic properties of hexago-nal arrays of holes in thin Fe films were presented. We an-alyzed the dependence of the magnetic and magnetoopticproperties on the hole size and we compared them with asimilar continuous Fe film. Extraordinary enhancementof the magnetooptic Kerr rotation is observed, which isrelated to the surface plasmon resonances and the holediameter. The very good agreement with theoretical sim-ulations gives us the ability to fully control of the prop-erties and apply the structures for technological appli-cations. Keeping constant the interhole distance a themaximum magnetooptic enhancement can be tuned byincreasing the hole diameter. The magnetic characteri-zation revealed the magnetic hardening, and the presenceof out of plane magnetization components that give riseto different domain configurations around the holes. Acknowledgments
E. Th. P. acknowledges the financial support fromthe Icelandic Science Foundation and the Swedish Foun-dation for International Cooperation in Research and Higher Education (STINT). M. G. thanks the Helmholtz-Zentrum Berlin for financial support. A.G.-M. and E.F.-V. acknowledge financial support from the EU ProjectNMP3-SL-2008-214107-Nanomagma and from the Span-ish MICINN (Consolider 2010 ref. CSD2008-00023-Funcoat and MAT2008-06765-C02-01/NAN). E.F.-Valso acknowledges financial support from the CSIC viathe JAE-Pre program. The authors acknowledge alsothe Knut and Alice Wallenberg Foundation. ∗ Electronic address: [email protected] † Electronic address: [email protected] G. Ctistis, E. Papaioannou, P. Patoka, J. Gutek, P. Fuma-galli, and M. Giersig, Nano Lett. , 1 (2009). A. O. Adeyeye, J. A. C. Bland, and C. Daboo, Appl. Phys.Lett. , 3164 (1997). C. T. Yu, H. Jiang, L. Shen, P. J. Flanders, and G. J.Mankey, J. Appl. Phys. , 6322 (2000). P. Vavassori, G. Gubbiotti, G. Zangari, C. T. Yu, H. Yin,H. Jiang, and G. J. Mankey, J. Appl. Phys. , 7992(2002). C. C. Wang, A. O. Adeyeye, and Y. H. Wu, J. Appl. Phys. , 6644 (2003). L. J. Heyderman, F. Nolting, D. Backes, S. Czekaj,L. Lopez-Diaz, M. Klaui, U. Rudiger, C. A. F. Vaz, J. A. C.Bland, R. J. Matelon, et al., Phys. Rev. B , 214429(2006). T. W. Ebbesen, H. J. Lezec, H. F. Ghaemi, T. Thio, andP. A. Wolff, Nature , 667 (1998). J. B. Gonzalez-Diaz, A. Garcia-Martin, G. Armelles,D. Navas, M. Vazquez, K. Nielsch, R. B. Wehrspohn, andU. Gsele, Adv. Mater. , 2643 (2007). V. I. Belotelov, D. A. Bykov, L. L. Doskolovich, A. N.Kalish, and A. K. Zvezdin, J. Opt. Soc. Am. B , 1594 (2009). A. Kosiorek, W. Kandulski, P. Chudzinski, K. Kempa, andM. Giersig, Nano Lett. , 1359 (2004). M. J. Donahue and D. G. Porter, Oommf User’s guideVersion 1, National Institute of Standard and Technologyand Gaithersburg MD (1999), URL http://math.nist.gov/oommf . E. T. Papaioannou, M. Angelakeris, N. K. Flevaris, P. Fu-magalli, C. Mueller, A. Troupis, A. Spanou, V. Karoutsos,P. Poulopoulos, V. Kapaklis, et al., J. Appl. Phys. ,023913 (2007). A. Garcia-Martin, G. Armelles, and S. Pereira, Phys. Rev.B , 205116 (2005). J. B. Gonzalez-Diaz, J. M. Garcia-Martin, A. Garcia-Martin, D. Navas, A. Asenjo, M. Vazquez, M. Hernandez-Velez, and G. Armelles, Appl. Phys. Lett. , 263101(2009). C. C. Wang, A. O. Adeyeye, and N. Singh, Nanotechnology , 1629 (2006). M. Jaafar, D. Navas, A. Asenjo, M. Vazquez,M. Hernandez-Velez, and J. M. Garcia-Martin, J. Appl.Phys.101