Magnetoplasmonic properties of perpendicularly magnetized [ Co/Pt ] N nanodots
Francisco Freire-Fernández, Rhodri Mansell, Sebastiaan van Dijken
MMagnetoplasmonic properties of perpendicularly magnetized [Co / Pt] N nanodots Francisco Freire-Fern´andez, ∗ Rhodri Mansell, and Sebastiaan van Dijken † NanoSpin, Department of Applied Physics, Aalto University School of Science, P.O. Box 15100, FI-00076 Aalto, Finland (Dated: November 18, 2019)We demonstrate a ten-fold resonant enhancement of magneto-optical effects in perpendicularlymagnetized [Co / Pt] N nanodots mediated by the excitation of optimized plasmon modes. Twomagnetoplasmonic systems are considered; square arrays of [Co / Pt] N nanodots on glass and identicalarrays on a Au/SiO bilayer. On glass, the optical and magneto-optical spectra of the nanodotarrays are dominated by the excitation of a surface lattice resonance (SLR), whereas on Au/SiO ,a narrow surface plasmon polariton (SPP) resonance tailors the spectra further. Both the SLR andSPP modes are magneto-optically active leading to an enhancement of the Kerr angle. We detailthe dependence of optical and magneto-optical spectra on the number of Co/Pt bilayer repetitions,the nanodot diameter, and the array period, offering design rules on how to maximize and spectrallytune the magneto-optical response of perpendicularly magnetized [Co / Pt] N nanodots. I. INTRODUCTION
Perpendicularly magnetized Co/Pt bilayers and mul-tilayers are widely investigated in the fields of nano-magnetism and spintronics. While research was mo-tivated initially by their potential use in magneticdata storage devices , Co/Pt and other similar struc-tures have been instrumental also in studies on do-main wall dynamics , current-induced magnetizationswitching , current-driven motion of chiral domainwalls , ionic control of magnetism , and magneticskyrmions . The attractive properties of Co/Ptarise from the interface nature of its perpendicularmagnetic anisotropy (PMA) , allowing it to be tai-lored by variation of the Co layer thickness or interfacechemistry and, in combination with the Dzyaloshinskii-Moriya interaction (DMI), it can be used to stabilizechiral spin textures. Moreover, the demonstration ofall-optical switching of perpendicular magnetization inCo/Pt multilayers has triggered a lively debate on theorigin of AOS in thin ferromagnetic films.The magneto-optical properties of Co/Pt multilayersare used mainly for magnetic characterization, often inthe polar Kerr effect configuration . The complex re-fractive index and magneto-optic Voigt parameter deter-mine the optical and magneto-optical response of contin-uous Co/Pt films. The gradual variation of both parame-ters with wavelength produces a rather smooth magneto-optical Kerr effect (MOKE) spectrum . In patternednanostructures, this no longer necessarily holds true. IfCo/Pt multilayers are patterned into dots wherein theresonance condition of the free electrons matches thewavelength of incident light, a localized surface plasmonresonance (LSPR) is excited. For other magnetic nanos-tructures, it has been demonstrated already that the op-tical near-fields of LSPR modes can resonantly enhanceand change the polarity of magneto-optical signals .This phenomenon is explained by the excitation of twoelectric dipoles (LSPRs) within the nanodots. The firstdipole is excited parallel to the incident electric field.The second dipole is excited orthogonally to both the first dipole and the direction of the magnetization of thenanodot, and is induced by spin-orbit coupling . Be-cause the phase and amplitude relations of the two elec-tric dipoles determine the magneto-optical response, it nolonger depends solely on intrinsic material parameters.Tailoring of optical near-fields by variation of the nan-odot size, shape, or their ordering in periodic arrays, asroutinely exploited in plasmonics, can therefore be usedto design magneto-optical spectra .Compared to metallic plasmonic systems comprisingAg, Au, or Al, magnetic metals suffer from larger ohmiclosses because of their higher electrical resistivity. Con-sequently, plasmonic resonances of magnetic nanostruc-tures are broader and less intense. To mitigate losses,hybrid magnetoplasmonic materials combining noble andmagnetic metals have been explored. Examples include,Au/Co/Au trilayers , nanosandwiches , or nanorods ,and core-shell Co/Ag or Co/Au nanoparticles . TheCo/Pt multilayers, considered in this paper would not,a priori, circumvent losses in a similar fashion becausethe electrical resistivity of Pt is higher than that of Co.Consequently, it remains an open question to what ex-tent plasmon resonances could be exploited to tailor themagneto-optical response of this attractive PMA system.The ability to drastically enhance optical near-fields inCo/Pt nanostructures using plasmonics would providepromising links to ongoing research on spintronics andAOS of magnetic materials.Here, we study the optical and magneto-optical prop-erties of [Co/Pt] N nanodots. We consider two magneto-plasmonic systems; (i) periodic [Co/Pt] N nanodot arrayson glass substrates and (ii) identical nanodot arrays onAu/SiO bilayers. We demonstrate that both plasmonicnanostructures allow for the design of strong magneto-optical responses through the excitation of collective sur-face lattice resonances (SLRs, system (i) and (ii)) or theexcitation of surface plasmon polaritons (SPPs, only sys-tem (ii)). Design rules for strong magnetoplasmonic ef-fects in this PMA system are derived by characterizinga large number of samples with varying numbers of bi-layer repetitions ( N ), nanodot diameters ( D ), and arrayperiods ( P ). a r X i v : . [ phy s i c s . op ti c s ] N ov (a) (c) (b) -1000 -500 0 500 1000-1.00-0.75-0.50-0.250.000.250.500.751.00 P o l a r M O KE i n t en s i t y ( a . u . ) µ
0H (mT) -1000 -500 0 500 1000-1.00-0.75-0.50-0.250.000.250.500.751.00 P o l a r M O KE i n t en s i t y ( a . u . ) µ
0H (mT) -1000 -500 0 500 1000-1.5-1.0-0.50.00.51.01.5 P o l a r M O KE i n t en s i t y ( a . u . ) N=1N=3N=5N=10N=20N=30 µ
0H (mT)
N=1N=3N=5N=10N=20N=30N=1N=3N=5N=10N=20N=30
FIG. 1. Polar MOKE hysteresis curves of square arrays of [Co / Pt] N nanodots with a diameter of (a) 200 nm, (b) 150 nm,and (c) 100 nm. The array period ( P ) is 400 nm and the number of bilayers repetitions ( N ) varies from 1 to 30. The insetin panel (b) shows an atomic force microscopy image of a [Co / Pt] nanodot with a diameter of 150 nm. The white scale barcorresponds to 200 nm. II. EXPERIMENT
The periodic arrays in this study consist ofTa(2) / Pt(4) / [Co(1) / Pt(1)] N / Pt(2) multilayer nanodots,where the numbers in brackets indicate the layer thick-ness in nanometers. Hereafter, we will simply refer tothe multilayer structure as [Co / Pt] N . The nanodot ar-rays are fabricated on glass substrates and Si substratescovered by a Au(150)/SiO (20) bilayer using electronbeam lithography and lift-off. The [Co / Pt] N multilayersare grown by magnetron sputtering, whereas SiO andAu are deposited by atomic layer deposition and elec-tron beam evaporation, respectively. Using this nanofab-rication process, square arrays of [Co / Pt] N nanodotswith the following parameters are patterned on glass andSi/Au/SiO : The number of bilayer repetitions is variedas N = 1, 3, 5, 10, 20, 30. For each N , arrays with periods P = 350 nm, 400 nm, 450 nm, 500 nm are patterned. Fi-nally, samples with nanodot diameters D = 100 nm, 150nm, and 200 nm are fabricated for each combination of N and P .The optical and magneto-optical properties of the sam-ples are characterized in a magneto-optical spectrometerthat can be configured for transmission (Faraday effect)and reflection (Kerr effect) measurements. The setupconsists of a NKT SuperK EXW-12 supercontinuum laserwith an acousto-optical filter, polarizing and focusing op-tics, a Hinds Instruments I/FS50 photoelastic modulator,and a photodetector. The laser wavelength is varied be-tween 475 nm and 1050 nm. We use linear polarizedlight at normal incidence with the electric field alignedalong one the x -axis of the square nanodot arrays. Dur-ing measurements, a ± / Pt] N nanodots be-tween two perpendicular directions. The magneto-opticalKerr rotation ( θ ) and Kerr ellipticity ( (cid:15) ) are simultane-ously recorded by lock-in amplification of the modulated signal at 50 kHz and 100 kHz. From these data the Kerrangle (Φ) is calculated as Φ = √ θ + (cid:15) . All measure-ments are performed with the [Co / Pt] N nanodots im-mersed in oil. The refractive index of the oil matchesthat of the glass substrate ( n = 1.52). The resulting uni-form dielectric environment facilitates the excitation ofintense plasmon resonances. III. RESULTS AND DISCUSSIONA. Magnetic characterization
The magnetic properties of the nanodots are stronglyinfluenced by the number of Co/Pt bilayer repetitions.Figure 1 shows polar MOKE hysteresis curves as a func-tion of N and D . As is expected, the Co/Pt multi-layers display an out-of-plane magnetic easy axis dueto hybridization of the electron orbitals at the Co/Ptinterfaces . The hysteresis curves in Figs. 1(a)-(c)exhibit fully remanent magnetization for most values of N , with only a slight decrease for the thickest multilayers.Regardless of the nanodot diameter, the coercivity of thearrays displays a similar trend as a function of N . Thecoercivity increases initially and reaches a maximum forfive bilayer repetitions. Beyond this, it slowly decreasesagain in the thicker nanodots. The Co layers are ferro-magnetically exchanged coupled to each other throughthe 1 nm thick Pt layers , which leads to increasingcoercivity with increasing N . For thicker multilayers,larger dipolar fields allow nucleation of reverse domainsat lower fields, leading to a reduction in coercivity andremanence . (a) (c)(d) (e) (b) (f)
600 800 10000.800.850.900.951.00 T r an s m i ss i on Wavelength (nm)
350 nm400 nm450 nm500 nm
600 800 10000.50.60.70.80.91.0 T r an s m i ss i on Wavelength (nm)
350 nm400 nm450 nm500 nm
600 800 10000.20.30.40.50.60.70.80.91.0 T r an s m i ss i on Wavelength (nm)
350 nm400 nm450 nm500 nm
600 800 10000.800.850.900.951.00 T r an s m i ss i on Wavelength (nm)
600 800 10000.50.60.70.80.91.0 T r an s m i ss i on Wavelength (nm)
600 800 10000.20.30.40.50.60.70.80.91.0 T r an s m i ss i on Wavelength (nm)
N=1N=3N=5N=10N=20N=30Random Random RandomN=1N=3N=5N=10N=20N=30 N=1N=3N=5N=10N=20N=30
FIG. 2. Optical transmission spectra of square arrays of [Co / Pt] N nanodots as a function of N (a-c) and P (d-f). Panels (a-c)show data for arrays with P = 400 nm and D = 200 nm (a), 150 nm (b), and 100 nm (c). The black lines represent spectra ofrandomly distributed [Co / Pt] nanodots with the same packing density. The vertical dashed lines denote the position of the(0, ±
1) DOs, the downward pointing arrows indicate the single-particle LSPRs, and the upward pointing arrows mark the SLRwavelengths. Panels (d-f) depict data for arrays with N = 20 nm and D = 200 nm (d), 150 nm (e), and 100 nm (f). B. Optical and magneto-optical properties of [Co / Pt] N nanodot arrays on glass Firstly, we discuss the optical response of [Co / Pt] N nanodot arrays on glass. Figures 2(a)-(c) show opticaltransmission spectra of [Co / Pt] N arrays with P = 400nm and D = 200 nm (a), 150 nm (b), and 100 nm (c).Nanodots of these sizes support the excitation of LSPRsat visible wavelengths, as the solid lines recorded on ran-domly distributed [Co / Pt] nanodots demonstrate. Thedownward pointing arrows mark the LSPR wavelengths.The LSPRs are broad because of large ohmic losses inCo and Pt. In periodic nanodot arrays, coupling betweenthe single-particle LSPRs and the diffracted orders (DOs)of the array produces asymmetric Fano-like excitations,known as surface lattice resonances (SLRs) . In uni- form dielectric environments (glass/index matching oil inour experiment, n = 1.52), the DO wavelengths are givenby λ p,q = nP (cid:112) p + q , (1)where p and q indicate the order of diffraction alongthe x - and y -axis of the nanodot array. In opticaltransmission spectra, the DOs appear as sharp peaks.For instance, the vertical dashed lines in Figs. 2(a)-(c) mark the (0, ±
1) DOs of arrays with P = 400 nm( λ , ± = 1 . ×
400 = 608 nm).Collective SLR modes absorb the incident light effi-ciently and, hence, they produce a minimum in opti-cal transmission spectra. The upward pointing arrowsin Figs. 2(a) and 2(b) indicate the SLRs arising from hy-bridization between the single-particle LSPRs and the(0, ±
1) DOs of the array. Because larger and thickernanodots absorb more light, the transmission signal atthe SLR wavelength decreases with increasing D and N . The SLRs are more narrow than the LSPRs be-cause diffraction in the array plane produces scatteredfields that counter damping of the single-particle plas-monic response . In our [Co / Pt] N nanodot arrays, theSLR linewidth decreases as the number of bilayer repeti-tions increases. Moreover, changes in the size and aspectratio of the nanodots translate into a spectral shift ofthe LSPR and, thereby, the SLR mode. For instance,Figs. 2(a) and 2(b) show how an increase of N and adecrease of D blue-shift the SLR transmission minimum.For [Co / Pt] N nanodots with D = 100 nm, the LSPRis blue-shifted well below the DO wavelength. Togetherwith the reduced polarizability of this small-volume par-ticle, it prevents the excitation of a pronounced SLRmode.Figures 2(d)-(f) summarize the dependence of opticaltransmission spectra on the array period. An increaseof P red-shifts the DOs, as described by Eq. 1. Conse-quently, the SLR wavelength moves up too. For nanodotarrays with P = 450 nm and 500 nm, sharp transmissionpeaks at the spectral position of ( ± ±
1) DOs are visiblebelow 600 nm also. The SLR linewidth depends on thespectral overlap between the LSPR and DO modes .In our [Co / Pt] samples with D = 200 nm and 150 nm,a red-shift of the DO towards the single-particle LSPRreduces the SLR linewidth. The results of Fig. 2 demon-strate that [Co / Pt] N nanodot arrays support the excita-tion of intense SLRs if D ≥
150 nm and N ≥
5. The min-imal SLR linewidth is about 100 nm, which is comparableto that of previously studied magnetoplasmonic arraysmade of Ni nanodisks or Au/SiO /Ni dimers .The optical reflectivity spectra of Figs. 3(a)-(c) vali-date the excitation of SLRs in the [Co / Pt] N nanodot ar-rays. In this measurement geometry, scattering from theSLR mode into the far field enhances the reflectivity ( R )at the SLR wavelength. We note that the optical reflec-tivity of a nanodot array does not exactly correspond to1 - T , because R depends on scattering by the nanodots,while the optical transmission T is affected by the scat-tering and absorption of light. SLRs do not only deter-mine the optical response of the [Co(1) / Pt(1)] N nanodotarrays, but also their magneto-optical activity. In ourexperiments, we irradiate the nanodots with linear po-larized light at normal incidence. In this geometry, theincident electric field ( E x ) induces an oscillating electricdipole ( p x ) in the metal nanodots, corresponding to theLSPR mode. The electric dipole is given by p x = α xx E x ,where α xx is a diagonal component of the nanodot polar-izability tensor. In the presence of perpendicular mag-netization, spin-orbit coupling produces a second weakerelectric dipole orthogonal to the optically excited dipoleand the direction of magnetization ( p y = α xy E x ). Forsingle or randomly distributed magnetic nanodots, the real and imaginary part of the p y /p x ratio determine toKerr rotation and Kerr ellipticity, respectively .The optical reflectivity of a periodic plasmonic ar-ray is proportional to the effective polarizability squared( R ∝ | α eff | ) . The effective polarizability of nanodotsin an array accounts for polarizing effects caused by theincident radiation and the electric fields from other nan-odots in the array. For normal incident light with linearpolarization along the x -axis ( E x ), the relevant diagonalcomponent of the effective polarizability tensor ( α eff , xx )can be written as α eff , xx = 11 /α xx − S x (2)In Eq. 2, α xx is the polarizability of a single metal nan-odot and S x is the geometrical lattice factor for an in-coming electric field along x . The effective polarizabilityof a plasmonic array is thus resonantly enhanced whenthe real part of the denominator in Eq. 2 (Re(1 /α xx )- Re( S x ), becomes zero. This condition corresponds tothe SLR wavelength. The SLR linewidth depends onthe imaginary value of the denominator in Eq. 2. Asthe polarizability of magnetic nanodots is small com-pared to that of noble metals, i.e. 1 /α xx is large, theSLRs of magnetoplasmonic arrays tend to be substan-tially broader than their noble metal counterpart. Inmagnetic nanodot arrays with perpendicular magnetiza-tion, E x induces a second orthogonal SLR mode throughspin-orbit coupling. This mode can be thought of as aris-ing from diffractive coupling of the local electric dipolesalong y ( p y ). The effective polarizability of the orthogo-nal SLR mode is given by α eff , xy = α xy α xx α yy (1 /α yy − S x )(1 /α xx − S y ) , (3)where α eff , yy is the second diagonal component of theeffective polarizability tensor of the magnetic nanodotarray, α eff , xy is the off-diagonal component, and S y is thelattice factor for radiation along y . The magneto-opticalKerr angle of the array (Φ) is then given byΦ = (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) α eff , xy α eff , xx (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) . (4)For square arrays of circular magnetic nanodots, α xx = α yy and S x = S y . From Eqs. 2 - 4 it follows that the spec-tral positions of the DOs and SLRs in optical reflectivityand MOKE spectra are similar in this case. At the SLRwavelength, the Kerr rotation is maximized and the Kerrellipticity crosses zero . As a result, the maximum Kerrangle (Φ = √ θ + (cid:15) ) may be slightly shifted away fromthe SLR wavelength. If the symmetry of the nanodots orthe array is broken, the reflectivity and MOKE spectracan be vastly different, as demonstrated experimentally (d) (f)(e) N=1N=3N=5N=10
N=20N=30
600 800 10000246810121416 K e rr ang l e ( m r ad ) Wavelength (nm)
200 nm150 nm100 nm
600 800 100002468101214161820 K e rr ang l e ( m r ad )
350 nm400 nm450 nm500 nm
Wavelength (nm)Wavelength (nm)
600 800 10000246810121416 K e rr ang l e ( m r ad ) (b)
600 800 10000.050.100.150.200.250.300.35 R e f l e c t i v i t y Wavelength (nm)
350 nm400 nm450 nm500 nm (c)
600 800 10000.050.100.150.200.25 R e f l e c t i v i t y Wavelength (nm)
200 nm150 nm100 nm (a)
600 800 10000.050.100.150.200.250.300.35 R e f l e c t i v i t y Wavelength (nm)
N=1N=3N=5N=10N=20N=30
FIG. 3. Optical reflectivity spectra (a-c) and polar MOKE spectra (d-f) of square arrays of [Co / Pt] N nanodots as a functionof D (a,d), P (b,e), and N (c,f). In (a,d) N = 10 and P = 400 nm, in (b,e) N = 10 and D = 200 nm, and in (c,f) P = 400nm and D = 200 nm. for circular Ni nanoparticles in rectangular arrays andelliptical Ni nanodots in square arrays .If we compare the optical reflectivity and MOKE spec-tra of our square [Co / Pt] N nanodot arrays (Fig. 3), weindeed observe the expected resemblance. Excitation ofthe SLR mode produces a resonant enhancement of themagneto-optical Kerr angle. More intense SLR modesresult in larger Kerr signals for most samples, as demon-strated by the dependence on nanodot diameter (Figs.3(a),(d)) and array period (Figs. 3(b),(e)). However, theKerr angle varies non-monotonically with the number ofbilayer repetitions, with N = 10 displaying the strongestMOKE signal (Fig. 3(f)). To distinguish between plas-monic or magnetic effects that could be responsible forthis, it is instructive to compare the MOKE spectra ofthe nanodot arrays to those of continuous [Co / Pt] N mul-tilayers. Figure 4 shows that the MOKE signal decreasesalso with N in continuous multilayers. From this we con-clude that the upper Co layers in thick films exhibit areduced magnetic moment, most likely caused by larger interface roughness. In the continuous multilayers, theKerr angle already reduces for N = 10. In the nanodotarrays, this intrinsic magnetic effect is initially compen-sated by the excitation of a more intense SLR mode for N = 10 than N = 5, producing a larger MOKE signal for N = 10. To further quantify the resonant enhancementof the Kerr angle in [Co / Pt] N nanodot arrays, we com-pare the MOKE spectrum of D = 200 nm, P = 400 nm,and N = 10 (black data in Fig. 3(d)) to that of a con-tinuous multilayer with N = 10 (green data in Fig. 4).In these two cases, the maximum Kerr angle is similar,15.5 mrad versus 13 mrad. However, the packing densityof the [Co / Pt] nanodots is only 20%. This suggest aSLR-induced resonant enhancement of the MOKE signalby a factor ∼
6. Another way of assessing the effect ofthe SLR on the magneto-optical response of the nanodotarrays directly compares the Kerr angle measured on-and off-resonance. Using data of the same nanodot arraywe find that a small off-resonance signal of 1.6 mrad at500 nm is resonantly enhanced by the SLR mode to 15.5
500 600 700 800 900 10000246810121416 K e rr ang l e ( m r ad ) Wavelength (nm)
N = 1N = 3N = 5N = 10N = 20N = 30
FIG. 4. Polar MOKE spectra of continuous [Co / Pt] N multi-layers as a function of N . mrad at 800 nm (black curve in Fig. 3(d)). C. Optical and magneto-optical properties of [Co / Pt] N nanodot arrays on Si/Au/Si O While [Co / Pt] N nanodot arrays on glass enable res-onant enhancements of their magneto-optical activityvia the excitation of SLRs, the magneto-optical reso-nances are still relatively broad because of the stronglydamped plasmonic component of SLR modes. Propagat-ing surface plasmon polaritons (SPPs) excited at a no-ble metal/dielectric interface exhibit considerably lowerdamping. Free space photons cannot excite SPPs as theirdispersion relation lies under the light line . To over-come this momentum mismatch, SPPs are often excitedusing a prism or nanostructure array that acts as grat-ing coupler. In the experiments discussed in this section,we explore the optical and magneto-optical properties of[Co / Pt] N nanodot arrays on top of Au/SiO bilayers,wherein the Au and SiO layers are 150 nm and 20 nmthick, respectively. The rational behind this hybrid mag-netoplasmonic structure is that the [Co / Pt] N array fa-cilitates the excitation of low-loss SPPs at the Au/SiO interface. In turn, the slowly decaying near-fields of theSPP modes on the dielectric side of the Au/SiO inter-face induce a narrow-linewidth magneto-optical responseon the [Co / Pt] N nanodots. Recently, this low-loss mag-netoplasmonic excitation mechanism was demonstratedfor the first time using Ni nanodisks .The experiments on [Co / Pt] N nanodot arrays on topof Au/SiO bilayers are performed in reflection (the Aufilm blocks transmission). Absorption of light by the ex-citation of plasmon modes suppresses the optical reflec-tivity in this measurement geometry. The momentummismatch between free space photons and SPPs is over-come by the extra momentum provided by the nanodot array k SP P = k + p G x + q G y . (5)Here, k SP P is wave vector of the SPP mode, k =( ω/c ) sin( θ ) is the wave vector of free space photons withangular frequency ω traveling at the speed of light c and θ is the angle of incidence, G x,y = πP x,y e x,y are the re-ciprocal lattice vectors for array periods P x,y along thedirection of unit vectors e x,y , and p, q indicate the or-der of diffraction along the x - and y -axis of the nanodotarray. For a square array with period P acting as grat-ing coupler, the free space wavelength of the SPP modecorresponds to λ (cid:48) p,q = P (cid:112) p + q · (cid:114) (cid:15) d (cid:15) m (cid:15) d + (cid:15) m , (6)where (cid:15) m and (cid:15) d are the dielectric constants of themetal film and the dielectric layer, respectively. Inmetal/dielectric bilayers with a periodic nanodot arrayon top, the SPP wavelength is greater than that of theDO.Figures 5(a)-(c) show optical reflectivity spectra of[Co / Pt] N nanodot arrays on a Au/SiO bilayer. Thearray period is 400 nm in panels (a)-(c) and D = 200nm (a), 150 nm (b), and 100 nm (c). Two clear reflec-tivity minima are measured. The first narrow resonancemarked by an upward pointing arrow occurs just abovethe wavelength of the (0, ±
1) DOs of the array (verti-cal dashed line in Figs. 5(a),(b)). This resonance corre-sponds to a SPP mode. Based on Eq. 6 and the dielectricconstants of Au and SiO determined by ellipsometry, weestimate an SPP excitation wavelength of 638 nm for P = 400 nm, in excellent agreement with the experimentaldata of Figs. 5(a),(b). The second reflectivity minimumoccurring at larger wavelength is considerably broader(labeled by a downward pointing arrow). This resonancecorresponds to the SLR mode in the [Co / Pt] N nanodotarray. Compared to the same array on glass, the SLRmode on Au/SiO is red-shifted (compare Figs. 5(a),(b)and Figs. 2(a),(b)). This effect is explained by the for-mation of image dipoles in the Au film when the SiO layer is thin, as discussed previously for noble metal plas-monic systems . Energy absorption by the SPP andSLR modes increases with N and D . Figures 5(d)-(f)summarize tuning of the optical reflectivity spectra bythe array period. In these measurements, N = 20. Sincethe SPP and SLR modes both depend on the spectral po-sition of the DOs, the two reflectivity minima red-shiftwith increasing P .We now focus our attention to the magneto-optical re-sponse of the [Co / Pt] N nanodot arrays on Si/Au/SiO .Figure 6 shows the Kerr angle dependence on the numberof Co/Pt bilayer repetitions and the array period. Ob-viously, the SPP and SLR modes both produce a strongresonant enhancement of the MOKE signal. The SLR-induced Kerr angle and its dependence on N and P is (a) (c)(d) (e) (b) (f)
600 800 10000.750.800.850.900.951.00
Wavelength (nm)
600 800 10000.40.60.81.0
Wavelength (nm)
600 800 10000.00.20.40.60.81.0
Wavelength (nm)
600 800 10000.20.30.40.50.60.70.80.91.0
Wavelength (nm)
350 nm400 nm450 nm500 nm
600 800 10000.750.800.850.900.951.00
Wavelength (nm)
350 nm400 nm450 nm500 nm
600 800 10000.00.10.20.30.40.50.60.70.80.9
Wavelength (nm)
350 nm400 nm450 nm500 nmN=3N=5N=10N=20 N=3N=5N=10N=20 N=3N=5N=10N=20 R e f l e c t i v i t y R e f l e c t i v i t y R e f l e c t i v i t y R e f l e c t i v i t y R e f l e c t i v i t y R e f l e c t i v i t y FIG. 5. Optical reflectivity spectra of square arrays of [Co / Pt] N nanodots on a Au/SiO bilayer as a function of N (a-c) and P (d-f). Panels (a-c) show data for arrays with P = 400 nm and D = 200 nm (a), 150 nm (b), and 100 nm (c). The verticaldashed lines denote the position of the (0, ±
1) DOs, the upward pointing arrows mark the SPP mode, and the downwardpointing arrows indicate the SLR. Panels (d-f) depict data for arrays with N = 20 nm and D = 200 nm (d), 150 nm (e), and100 nm (f). similar to that observed in [Co / Pt] N nanodot arrays onglass, except for a red-shift of the resonances caused bythe formation of image dipoles in the Au film. Section Bdescribes how SLR modes produce a strong MOKE sig-nal in magnetic nanodot arrays. Here, we focus on themagneto-optical activity of the SPP mode. The linewidthof the SPP-induced MOKE resonance is small, with thefull width at half maximum (FWHM) ranging from 54nm ( D = 200 nm, P = 350 nm) to 18 nm ( D = 200 nm, P = 500), which is similar to the SPP resonances linewidthin the optical reflectivity spectra of Fig. 5. While theSPP mode is excited at the Au/SiO interface, its near-field only decays slowly within the dielectric film. Typi-cally, the decay length of the SPP electric field is abouthalf the wavelength of light involved . Because the SiO layer in our magnetoplasmonic structures is much thinner than this decay length, the [Co / Pt] N nanodots patternedon top of SiO are driven into resonance. Consequently,the SPP mode induces an electric dipole ( p x ) in the mag-netic nanodots parallel to the incident electric field ( E x ).Via spin-orbit coupling in [Co / Pt] N this again producesan orthogonal electric dipole ( p y ), rendering a magneto-optical signal at the SPP wavelength. Since the SPP reso-nances forces the free electrons of the [Co / Pt] N nanodotsinto oscillation, the induced magneto-optical response isnot strongly affected by damping in the nanodots. Thispoint, illustrating a powerful loss mitigation strategy formetallic magnetoplasmonics, was recently substantiatedby the demonstration of similar SPP linewidths for Niand Au nanodot arrays on Au/SiO bilayers .The magnitude of the Kerr angle at the SPP wave-length increases initially with the number of Co/Pt bi- (a) (b)
600 800 1000024681012141618202224 K e rr ang l e ( m r ad ) Wavelength (nm)
350 nm400 nm450 nm500 nm
600 800 1000024681012141618202224 K e rr ang l e ( m r ad ) Wavelength (nm)
N = 3N = 5N = 10N = 20
FIG. 6. Polar MOKE spectra of square arrays of [Co / Pt] N nanodots on a Au/SiO bilayer as a function of N (a) and P (b). In (a) D = 200 nm and P = 400 nm. In (b) D = 200nm and N = 10. layer repetitions before it saturates at N = 10 - 20 (Fig.6(a)). This effect, which is similar to that observed forthe SLR mode, again correlates with a reduction of themagnetic moment in the upper Co layers of thick mul-tilayers (Fig. 4). The maximum Kerr angle measuredat the SPP wavelength is comparable to that producedby the SLR ( ∼
20 mrad for N = 10, D = 200 nm, and P = 350 nm). Variation of the array period offers anattractive means of tuning the wavelength of the narrowSPP-induced MOKE resonance. The spectral position ofthe Kerr angle maximum is given by Eq. 6, which can beused as a design tool in magnetoplasmonic applications.The decrease of the Kerr angle with array period (Fig.6(b)) is caused predominantly by a reduction of the nan-odot packing density from 26% for P = 350 nm to 13%for P = 500 nm. IV. SUMMARY
In this article, we report on the magnetoplasmonicproperties of perpendicularly magnetized [Co / Pt] N nan-odot arrays on glass and Si/Au/SiO . On glass, the op-tical and magneto-optical responses are dominated bythe excitation of a collective SLR mode, arising fromdiffractive coupling between single-particle LSPRs. OnSi/Au/SiO , a red-shifted SLR mode is also measured. In addition, a spectrally more narrow resonance appearsin optical reflectivity and MOKE spectra. This feature isinduced by the excitation of a SPP mode at the Au/SiO interface. Because the [Co / Pt] N nanodots are placedwithin the SPP near-field, plasmon resonances are ex-cited in the nanodots at the SPP wavelength. In bothcases, whether plasmon resonances in the nanodots aredirectly excited by the incident electric field or via a SPP,spin-orbit coupling produces a second plasmon mode inthe orthogonal direction. This effect causes linear polar-ized light to undergo a rotation and to become ellipticalupon reflection from the [Co / Pt] N nanodot samples.Compared to continuous [Co / Pt] N films or off-resonance measurement conditions, optical near-field en-hancements at the SPP and SLR wavelength can in-crease local magneto-optical effects by up to one orderof magnitude. Plasmon-enhanced MOKE signals requirea [Co / Pt] N nanodot diameter of more than 100 nm and10 - 20 Co/Pt bilayer repetitions maximize the Kerr an-gle. Because the SLR mode depends on single-particleLSPRs, its spectral position is tuned by variation of themultilayer thickness and the nanodot diameter. Both theSLR and SPP modes depend on the DOs of the arrayand, consequently, their spectral position depends on thearray period.Plasmon-enhanced magneto-optical effects in[Co / Pt] N nanodots and the flexibility to tailor stronglight-matter interactions in this PMA material systemmay be utilized in AOS experiments or as an efficientinterface linking photonic and spintronic devices . Ourfindings could also be exploited in plasmonic lasing.Recently, lasing was demonstrated for the first time in amagnetic system . In this study, the SLR mode of Ninanodot arrays on glass acted as a feedback mechanismfor lasing from an organic gain medium. We expectthat the much narrower SPP resonances in [Co / Pt] N nanodot arrays on Au/SiO would provide more efficientfeedback, with the potential of realizing magnetic fieldcontrol of plasmonic lasing. V. ACKNOWLEDGEMENTS
This work was supported by the Academy of Finland(Grant Nos. 316857, 295269, and 306978). F.F-F. ac-knowledges financial support from the Finnish Academyof Science and Letters (Vilho, Yrj and Kalle Visln Fund).Lithography was performed at the Micronova Nanofabri-cation Centre, supported by Aalto University. ∗ francisco.freirefernandez@aalto.fi † sebastiaan.van.dijken@aalto.fi M. T. Johnson, P. J. H. Bloemen, F. J. A. den Broeder,and J. J. de Vries, Rep. Prog. Phys. , 1409 (1996). B. D. Terris and T. Thomson, J. Phys. D Appl. Phys. ,R199 (2005). S. Lemerle, J. Ferr´e, C. Chappert, V. Mathet, T. Gia-marchi, and P. Le Doussal, Phys. Rev. Lett. , 849 (1998). P. J. Metaxas, J. P. Jamet, A. Mougin, M. Cormier,J. Ferr´e, V. Baltz, B. Rodmacq, B. Dieny, and R. L.Stamps, Phys. Rev. Lett. , 217208 (2007). S. Mangin, D. Ravelosona, J. A. Katine, M. J. Carey, B. D.Terris, and E. E. Fullerton, Nat. Mater. , 210 (2006). I. M. Miron, K. Garello, G. Gaudin, P.-J. Zermatten, M. V.Costache, S. Auffret, S. Bandiera, B. Rodmacq, A. Schuhl,and P. Gambardella, Nature , 189 (2011). L. Liu, O. J. Lee, T. J. Gudmundsen, D. C. Ralph, andR. A. Buhrman, Phys. Rev. Lett. , 096602 (2012). P. P. J. Haazen, E. Mur`e, J. H. Franken, R. Lavrijsen,H. J. M. Swagten, and B. Koopmans, Nat. Mater. , 299(2013). S. Emori, U. Bauer, S.-M. Ahn, E. Martinez, and G. S. D.Beach, Nat. Mater. , 611 (2013). K.-S. Ryu, L. Thomas, S.-H. Yang, and S. Parkin, Nat.Nanotechnol. , 527 (2013). U. Bauer, L. Yao, A. J. Tan, P. Agrawal, S. Emori, H. L.Tuller, S. van Dijken, and G. S. D. Beach, Nat. Mater. , 174 (2015). S. Woo, K. Litzius, B. Kr¨uger, M.-Y. Im, L. Caretta,K. Richter, M. Mann, A. Krone, R. M. Reeve, M. Weigand,P. Agrawal, I. Lemesh, M.-A. Mawass, P. Fischer,M. Kl¨aui, and G. S. D. Beach, Nat. Mater. , 501 (2016). O. Boulle, J. Vogel, H. Yang, S. Pizzini, D. de SouzaChaves, A. Locatelli, T. O. Mente¸s, A. Sala, L. D.Buda-Prejbeanu, O. Klein, M. Belmeguenai, Y. Rous-sign´e, A. Stashkevich, S. M. Ch´erif, L. Aballe, M. Foerster,M. Chshiev, S. Auffret, I. M. Miron, and G. Gaudin, Nat.Nanotechnol. , 449 (2016). C. Moreau-Luchaire, C. Moutafis, N. Reyren, J. Sampaio,C. A. F. Vaz, N. van Horne, K. Bouzehouane, K. Garcia,C. Deranlot, P. Warnicke, P. Wohlh¨uter, J.-M. George,M. Weigand, J. Raabe, V. Cros, and A. Fert, Nat. Nan-otechnol. , 444 (2016). N. Nakajima, T. Koide, T. Shidara, H. Miyauchi, H. Fuku-tani, A. Fujimori, K. Iio, T. Katayama, M. N´yvlt, andY. Suzuki, Phys. Rev. Lett. , 5229 (1998). F. Hellman, A. Hoffmann, Y. Tserkovnyak, G. S. D. Beach,E. E. Fullerton, C. Leighton, A. H. MacDonald, D. C.Ralph, D. A. Arena, H. A. D¨urr, P. Fischer, J. Grollier,J. P. Heremans, T. Jungwirth, A. V. Kimel, B. Koop-mans, I. N. Krivorotov, S. J. May, A. K. Petford-Long,J. M. Rondinelli, N. Samarth, I. K. Schuller, A. N. Slavin,M. D. Stiles, O. Tchernyshyov, A. Thiaville, and B. L.Zink, Rev. Mod. Phys. , 025006 (2017). C.-H. Lambert, S. Mangin, B. S. D. C. S. Varaprasad,Y. K. Takahashi, M. Hehn, M. Cinchetti, G. Malinowski,K. Hono, Y. Fainman, M. Aeschlimann, and E. E. Fuller-ton, Science , 1337 (2014). W. B. Zeper, F. J. A. M. Greidanus, P. F. Carcia, andC. R. Fincher, J. Appl. Phys. , 4971 (1989). R. Atkinson, S. Pahirathan, I. W. Salter, P. J. Grundy,C. J. Tatnall, J. C. Lodder, and Q. Meng, J. Magn. Magn.Mater. , 131 (1996). J. Chen, P. Albella, Z. Pirzadeh, P. Alonso-Gonz´alez,F. Huth, S. Bonetti, V. Bonanni, J. ˚Akerman, J. Nogu´es,P. Vavassori, A. Dmitriev, J. Aizpurua, and R. Hillen-brand, Small , 2341 (2011). V. Bonanni, S. Bonetti, T. Pakizeh, Z. Pirzadeh, J. Chen,J. Nogu´es, P. Vavassori, R. Hillenbrand, J. ˚Akerman, and A. Dmitriev, Nano Lett. , 5333 (2011). G. Armelles, G. Cebollada, A. Garc´ıa-Mart´ın, and M. U.Gonz´alez, Adv. Opt. Mater. (2013). N. Maccaferri, A. Berger, S. Bonetti, V. Bonanni,M. Kataja, Q. H. Qin, S. van Dijken, Z. Pirzadeh,A. Dmitriev, J. Nogu´es, J. ˚Akerman, and P. Vavassori,Phys. Rev. Lett. , 167401 (2013). M. Kataja, T. K. Hakala, A. Julku, M. J. Huttunen, S. vanDijken, and P. T¨orm¨a, Nat. Commun. , 7072 (2015). N. Maccaferri, L. Bergamini, M. Pancaldi, M. K. Schmidt,M. Kataja, S. v. Dijken, N. Zabala, J. Aizpurua, andP. Vavassori, Nano Lett. , 2533 (2016). V. V. Temnov, G. Armelles, U. Woggon, D. Guzatov,A. Cebollada, A. Garcia-Martin, J.-M. Garcia-Martin,T. Thomay, A. Leitenstorfer, and R. Bratschitsch, Nat.Photon. , 107 (2010). J. B. Gonz´alez-D´ıaz, A. Garc´ıa-Mart´ın, J. M. Garc´ıa-Mart´ın, A. Cebollada, G. Armelles, B. Sep´ulveda,Y. Alaverdyan, and M. K¨all, Small (2008). G. Armelles, A. Cebollada, F. Garc´ıa, A. Garc´ıa-Mart´ın,and N. de Sousa, ACS Photon. , 2427 (2016). L. Wang, C. Clavero, Z. Huba, K. J. Carroll, E. E. Car-penter, D. Gu, and R. A. Lukaszew, Nano Lett. , 1237(2011). Y. Song, J. Ding, and Y. Wang, J. Phys. Chem. C ,11343 (2012). J. W. Knepper and F. Y. Yang, Phys. Rev. B , 224403(2005). R. Sbiaa, Z. Bilin, M. Ranjbar, H. K. Tan, S. J. Wong,S. N. Piramanayagam, and T. C. Chong, J. Appl. Phys. , 103901 (2010). S. Zou, N. Janel, and G. C. Schatz, J. Chem. Phys. ,10871 (2004). V. G. Kravets, F. Schedin, and A. N. Grigorenko, Phys.Rev. Lett. , 087403 (2008). B. Augui´e and W. L. Barnes, Phys. Rev. Lett. , 143902(2008). V. G. Kravets, A. V. Kabashin, W. L. Barnes, and A. N.Grigorenko, Chem. Rev. , 5912 (2018). W. L. Barnes, Am. J. Phys. , 593 (2016). S. Pourjamal, M. Kataja, N. Maccaferri, P. Vavassori, andS. van Dijken, Nanophotonics , 905 (2018). C. Bohren and D. Huffman,
Absorption and Scattering ofLight by Small Particles (John Wiley and Sons, 1983). S. Pourjamal, M. Kataja, N. Maccaferri, P. Vavassori, andS. van Dijken, Sci. Rep. , 9907 (2019). W. L. Barnes, A. Dereux, and T. W. Ebbesen, Nature , 824 (2003). F. Freire-Fern´andez, M. Kajava, and S. van Dijken,arXiv:1909.07002. T. W. Ebbesen, H. J. Lezec, H. F. Ghaemi, T. Thio, andP. A. Wolff, Nature , 667 (1998). N. Papanikolaou, Phys. Rev. B , 235426 (2007). W. R. Holland and D. G. Hall, Phys. Rev. Lett. , 1041(1984). P. Nordlander and E. Prodan, Nano Lett. , 2209 (2004). M. L. M. Lalieu, R. Lavrijsen, and B. Koopmans, Nat.Commun. , 110 (2019). S. Pourjamal, T. K. Hakala, M. Ne˘cada, F. Freire-Fern´andez, M. Kataja, H. Rekola, J.-P. Martikainen,P. T¨orm¨a, and S. van Dijken, ACS Nano13