Magneto-Optical Enhancement by Plasmon Excitations inNanoparticle/Metal StructuresM. Rubio-Roy,† O. Vlasin,† O. Pascu,† J. M. Caicedo,† M. Schmidt,† A. R. Goni,†,‡ N. G. Tognalli,§

A. Fainstein,§ A. Roig,† and G. Herranz*,†

†Institut de Ciencia de Materials de Barcelona ICMAB-CSIC, Campus de la UAB, 08193 Bellaterra, Catalonia, Spain‡ICREA, Passeig Lluís Companys 23, 08010 Barcelona, Catalonia, Spain§Centro Atomico Bariloche, Instituto Balseiro, Comision Nacional de Energía Atomica, 8400 S. C. de Bariloche, Río Negro, Argentina

*S Supporting Information

ABSTRACT: Coupling magnetic materials to plasmonic structures provides a pathway to dramatically increase the magneto-optical response of the resulting composite architecture. Although such optical enhancement has been demonstrated in a varietyof systems, some basic aspects are scarcely known. In particular, reflectance/transmission modulations and electromagnetic fieldintensification, both triggered by plasmon excitations, can contribute to the magneto-optical enhancement. However, aquantitative evaluation of the impact of both factors on the magneto-optical response is lacking. To address this issue, we havemeasured magneto-optical Kerr spectra on corrugated gold/dielectric interfaces with magnetic (nickel and iron oxide)nanoparticles. We find that the magneto-optical activity is enhanced by up to an order of magnitude for wavelengths that arecorrelated to the excitation of propagating or localized surface plasmons. Our work sheds light on the fundamental principles forthe observed optical response and demonstrates that the outstanding magneto-optical performance is originated by the increaseof the polarization conversion efficiency, whereas the contribution of reflectance modulations is negligible.

■ INTRODUCTIONStructuring of matter is a versatile way to manipulate lightpropagation in media and to obtain engineered electromagneticresponses. Examples of tailoring light−matter interactions arefound in different fields of photonics, from metamaterials1 tophotonic crystals2 and surface plasmons in corrugated metal/dielectric interfaces.3 For instance, photonic crystals areperiodically arranged structures engineered at scales compara-ble to the electromagnetic wavelength that are designed to trapand guide light.4 For a particular range of wavelengths, thesesystems exhibit a bandgap for which the electromagnetic wavescannot propagate in the crystal and also exhibit opticalnonlinear effects at photonic band edges.5 On the otherhand, light can be also coupled to surface plasmon excitations atmetal/dielectric interfaces, that is, coherent electron oscillationsthat are coupled to electromagnetic radiation.3 The electro-magnetic energy of surface plasmons is strongly confined to theinterface, and huge values of the electric field are attained there,so that the electric dependent response may be enormously

enhanced, as it is the case of the surface enhanced Ramanspectroscopy (SERS).6,7

The interest of magneto-optical materials for applications indata storage and optical communications has spurred theresearch on new materials exhibiting large magneto-opticalresponses at the operating wavelengths. This has motivated theresearch on photonic/plasmonic devices coupled to magneticmaterials. In this line, magnetophotonic crystals have beenintensively investigated, in which nonreciprocal optical effects8

and large magneto-optical responses at stop-band edgefrequencies have been reported.9−20 The interest in magneticmaterials has also been extended to plasmonics. Beyond theconventional approach based on noble metals such as Ag or

Special Issue: Colloidal Nanoplasmonics

Received: January 13, 2012Revised: May 15, 2012Published: May 17, 2012

Article

pubs.acs.org/Langmuir

© 2012 American Chemical Society 9010 dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−9020

Au,21 recently, a series of studies on the plasmonic properties ofstructures containing ferromagnetic materials has been carriedout. These studies include materials such as Au/Co/Aumultilayers22−24 periodically patterned ferromagneticfilms25−29 and noble metal/ferromagnet structures,30 magneticnanowires,31 or nanodisk arrays.32,33 In such a large variety ofsystems, remarkable magneto-optical enhancements have beenobserved at wavelengths where surface plasmons are excited. Amore detailed knowledge about the fundamental mechanismsleading to such improved optical performance is crucial todesign better magneto-optical devices. In particular, the twobasic mechanisms that may give rise to such increased activityare the suppression of reflectivity at the plasmon resonance andthe increase of polarization conversion driven by the dramaticintensification of the electromagnetic field.29,34 Although inprevious work evidence has been obtained that the degree ofmagneto-optical enhancement correlates with the electric fieldintensification,35 a direct quantitative measurement of bothcontributions is missing.

We have addressed this issue in magneto-plasmonic systemsconsisting of two structures, namely, (i) an hexagonal array ofspherical truncated gold voids containing a small amount of Ninanoparticles and (ii) three gold nanodisk arrays with differentdiameters on an amorphous-SrTiO3/Si(100) substrate, coveredwith iron oxide nanoparticles. As expected, such materialsexhibit a large magneto-optical response at wavelengths whereplasmons are excited, by up to an order of magnitude largerthan the reference magneto-optical signal. Interestingly, wedemonstrate that whereas structures of type (i) sustain mainlypropagating surface plasmons, in systems of type (ii) onlylocalized plasmons are excited. This has allowed us to ascertainwhether the different nature of these excitations has an impacton the character of the observed magneto-optical enhancement.Our results clearly demonstrate that for both kinds of magneto-plasmonic systems the intrinsic polarization conversionefficiency is spectacularly increased, being at least 1 order ofmagnitude larger than that from reflectance modulations.

Figure 1. SEM images of the Au void surface infiltrated with nanoparticles. (a) Au void array. Inset: detail of one of the voids, where nickelnanoparticles are observed. (b) non-structured area (the white circle emphasizes a group of nanoparticles). Panels (c−e) show the SEM images ofthe nanodisk arrays with diameters d ≈ 57, 79, and 95 nm, respectively.

Langmuir Article

dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−90209011

■ MATERIALS AND METHODSGold Nanovoid Arrays. The void array was produced via a

template self-assembly of polystyrene latex spheres on a gold film onglass, followed by an electrochemical deposition of gold and a finalchemical etching of the latex sphere template.36,37 In this process, asolution of polymer spheres fills a thin fluid cell, which is made of agold-coated glass slide, a clean glass slide, and sidewalls made from a300 μm thick spacer of Parafilm. The gold slide is coated withcysteamine to reduce the contact angle of the aqueous liquid placed onthe surface. This forms a sweeping meniscus tail as the fluid dries,which pulls spheres to the evaporation line where they form a close-packed monolayer domain. After template formation the sample isplaced in an electrochemical plating bath. The degree of truncation canbe finely controlled by the deposition time. Therefore, very clean,reproducible, and well-defined nanostructured surfaces are obtainedover areas on the order of 1 cm2. The sample analyzed here wasretracted from the solution during the metal deposition in order toachieve a degree of truncation of d/2R ≈ 0.4, where d ≈ 240 nm is theAu film thickness while 2R ≈ 600 nm is the diameter of the spheres.The resulting structure exhibits a period of ≈650 nm, being thestructural order of the Au nanovoids preserved on large scales up tohundreds of micrometers. Due to the truncated sphere geometry, a topview image of the structure after the chemical deposition (as seen inFigure 1a) gives a void diameter of ≈400 nm, that is, significantlysmaller than the sphere diameter in the template. Nickel nanoparticleswith diameters below 10 nm, synthesized by a ligand stabilizedsolution-phase synthesis (see ref 38 and the description below), weredeposited on the Au void surface by vertical immersion on a colloidaldispersion in hexane. The fabrication process also yields a non-structured Au surface outside the area where the spherical void array isdefined (Figure 1b). Optical and magneto-optical characterizations ofnanoparticles deposited on this Au flat area were used as reference toanalyze the effect of plasmons on the magneto-optical properties of thenanovoid arrays.Gold Nanodisk Arrays. Gold nanodisk arrays with variable

diameter were fabricated by electron-beam lithography on anamorphous SrTiO3 dielectric layer with thickness ≈60 nm depositedby pulsed laser deposition at room temperature on p-type dopedSi(100) substrates. Three different nanodisk arrays were defined, withdiameters d ≈ 57, 79 and 95 nm, respectively and period ≈200 nm(see Figure 1c−e). Iron oxide magnetic nanoparticles with diametersbelow 10 nm were deposited on the prepared substrates by immersingthe nanodisk array templates in a colloidal solution containing thenanoparticles dispersed in hexane. Details on nanoparticle character-ization and chemical composition are given below. The fabricationprocess defines large areas of exposed flat amorphous SrTiO3 surfaces;on these non-structured regions, free of any Au nanodisk, iron oxidenanoparticles were deposited and their optical and magneto-opticalcharacterizations were used as a reference to analyze the effect ofplasmons on the magneto-optical properties of the nanodisk arrays.Nanoparticle Synthesis and Characterization. Nickel nano-

particles were synthesized by high-temperature organometallicdecomposition route. The reaction mixture containing 2 mmol(0.514 g) of Ni(acac)2 precursor, 2 mmol (0.63 mL) oleic acid,4 mmol (1.8 mL) trioctylphosphine and 14 mL oleylamine solvent wasslowly heated up to 130 °C under a flow of high purity argon andmagnetic stirring. The mixture was kept at 130 °C for 20 min followedby further heating up to the reflux point (250 °C) and maintained atthis temperature for 30 additional minutes. The color of the solutionchanged from green to dark green and finally to black. After cooling toroom temperature, the nanoparticles were precipitated by addingethanol, followed by centrifugation. The precipitate was dried in anoven (at 70 °C) overnight and weighted. Finally, the as-synthesizednickel nanoparticles were kept in a hexane dispersion of knownconcentration. Transmission electron microscopy (TEM) of the Ninanoparticles shows that they are spherical with an average size of8 ± 1 nm (Figure 2a). Lattice spacings from the selected area electrondiffraction (SAED) patterns (Figure 2b) are in good agreement withthose of metallic fcc-Ni.

For the synthesis of iron oxide nanoparticles, a microwave-assistedsol−gel route was used. Briefly, in a typical reaction, 0.35 mmol ofFe(acac)3 and 1.05 mmol of oleic acid were dissolved in 1.5 mLanhydrous benzyl alcohol at 60 °C for 5 min under magnetic stirringfor complete dissolution of the precursors. The mixture had atransparent dark-red color. The precursor solution was further heatedin the same microwave reactor to 160 °C, and this temperature waskept stable for 5 min. Then, the solution was automatically cooleddown to 50 °C by compressed N2 in approximately 3 min. The finalsuspension was black. The nanoparticles were separated by addingethanol (40 mL), followed by double centrifugation at 4000 rpmduring 20 min. Finally, the black precipitate was redispersed in 2 mLhexane containing 10 μL oleic acid and used for further character-izations and deposition onto the Au dots. By this procedure, roughlyspherical iron oxide nanoparticles with a mean size of 8.3 ± 2.5 nm(see TEM micrographs in Figure 2c) were produced. Titrationanalyses revealed the presence of only 5% of Fe2+ relative to the totalFe in comparison with an expected 33% for pure magnetite. We thusconclude that the material consists of a mixture of the two phases(magnetite and maghemite) and hereafter will be referred as iron oxidenanoparticles.

Optical Reflectivity Spectroscopy. For the reflectance measure-ments, a self-made optical setup consisting on a halogen lamp, a 50×microscope objective, an optical Fourier transform doublet, and amotorized CCD spectrometer allowed us to perform measurementswith incident angles from about −40° to 40° and wavelengths from≈400 to 800 nm (see the sketch in Figure 3a). The light spot can befocused to ≈100 μm in this setup.The measurements were done withs- and p-polarized incident light, and all of them were referred to thereflectance of a silver mirror to obtain the reflectivity of the samples.The final output generates contour plots that permit the analysis of thedispersion relationship (i.e., frequency versus wavevector or,equivalently, wavelength versus angle of incidence) of lightpropagating in media.

Magneto-Optical Spectroscopy. Kerr rotation (θ) and ellipticity(ε) were obtained at room temperature from the analysis of thepolarization of light reflected on samples. We probed the magneto-optical signals in polar configuration in the range of wavelengths≈400−850 nm, with s- and p-polarized light incident at angles close to

Figure 2. TEM micrographs of nanoparticles. (a) Nickel nanoparticles.Inset: detail of one nickel nanoparticle. (b) SAED pattern with theindication of the Miller indices. (c) Iron oxide nanoparticles. (d)SAED pattern with the indication of the Miller indices.

Langmuir Article

dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−90209012

the normal to the surface. In the setup, light from a 150 W Xe arc lamp(ZolixTechnology) is dispersed by a monochromator (Zolix l-150),collimated, and then linearly polarized by the action of a Glan-Thompson prism, which is rotated 45° with respect to the modulatoraxis of a photoelastic modulator (PEM), so that the phases of the s-and p-components of the polarized light are modulated periodically ata frequency of ≈50 kHz (see Figure 3b for a sketch of theexperimental setup). After the PEM, the light is reflected from thesamples (with a typical light beam diameter of about ≈1 mm) anddirected toward a detector through a polarizing analyzer. The positionof the optical axis of the analyzer can be set at two different angles withrespect to the PEM axis, namely, 0° and 90°, to record the magneto-optical signals of s- and p-polarized light, respectively. Finally, thesignal collected from the detector is brought to a lock-in amplifiersynchronized to the frequency of the PEM retardation angle φ. In thisoptical arrangement, the ellipticity and rotation are given by theexpressions

θφ

θφ

εφ

θφ

= = −

= = −

ω ω

ω ω

cJ

I

I cJ

I

I

cJ

I

I cJ

I

I

14 ( )

,1

4 ( )

14 ( )

,1

4 ( )

s2 0

2 ,s

0p

2 0

2 ,p

0

s1 0

1 ,s

0p

1 0

1 ,p

0 (1)

where Ji(φ0) are the Bessel functions of the first kind, the PEMretardation angle is set to φ0 ≈ 137.8°, I0 is the dc-component of thedetected light, and (Iω,s, Iω,p) and (I2ω,s, I2ω,p) are the first and secondharmonics of the detected light measured by the lock-inamplifier.19,20,38 The calibration constant c is determined experimen-tally. Thus, as shown in eq 1, the rotation and ellipticity are measuredby inspection of the second and first harmonic of the detector signal,respectively.

■ RESULTS AND DISCUSSIONSurface Morphology. After vertical immersion of the Au

void and nanodisk arrays in colloidal dispersions of magneticnanoparticles in hexane, we carried out scanning electron

microscopy (SEM) to characterize the surface morphology.Since the individual nanoparticles were detectable by directvisual inspection of the SEM images (see Figure 1), weexploited the latter to assess the degree of nanoparticle coatingon these structures. For instance, Figure 1a and b revealed thepresence of individual Ni nanoparticles on the Au void array, asappreciated in the inset of Figure 1a (displaying a zoom of asingle Au cavity) as well as in Figure 1b, where one group ofnanoparticles is emphasized by a dashed circle. The direct visualinspection of these images allowed us to estimate that insidethe void cavities the surface nanoparticle density was around10−3 nanoparticles/nm2, equivalent to a surface coverage of≈6%. The distribution of the nanoparticles was found to bequite uniform (less than 20% variation from cavity to cavity).Outside the cavities of the Au void array, on the non-structuredarea (see Figure 1b), the nanoparticle surface density appearedto be slightly higher, around 2 × 10−3 nanoparticles/nm2,equivalent to a surface coverage ≈12%. SEM images of the Aunanodisk arrays also allowed identifying the presence of theindividual nanoparticles (Figure 1c−e). An inspection of thesefigures immediately reveals a significantly larger and uniformnanoparticle surface density. Indeed, with independence of thenanodisk size, the iron oxide nanoparticle surface density wasestimated to be around 8 × 10−3 nanoparticles/nm2 and surfacecoverage ≈50%, significantly larger than the values found in theAu voids.

Optical Reflectance Spectroscopy. Figures 4 and 5display the contour plots of the reflectance spectra of the Auvoid and nanodisk arrays, respectively, measured in the range ofwavelengths λ ≈ 400−800 nm for both s- and p-linearly polarized light. From these experiments, we obtainedthe absolute values of the diagonal Fresnel reflectionscoefficients |rss| and |rpp| that give the relative intensities of s-and p-polarized lights between incident and reflected waves.

Au Void Array. The reflectance spectra of the Au voidsurface coated with nickel nanoparticles reveal the presence ofseveral dispersive absorption lines that are dependent on boththe light polarization and the relative orientation of the plane ofincidence with respect to the void geometry (see Figure 4 andthe sketch therein). Along these dispersive bands, theabsorption peak is located at wavelengths that depend stronglyon the angle of incidence α (and wavevector), indicating theexcitation of delocalized Bragg surface plasmon modes.39−41

We note that although for d/2R ≈ 0.4, a Mie mode (1P+) isexpected to be excited at wavelength λ ≈ 650 nm, its resonanceis significantly weaker than that of Bragg modes,41 and this maybe the reason for which its excitation is not apparent in thespectra (Figure 4). Another possible explanation is that arandom distribution of nanoparticles inside the Au voids caninduce fluctuations of the Mie resonances from cavity to cavity,smearing out the localized resonance when probed with a largelight spot (≈100 μm).Our measurements on the Au void surface indicate that

Bragg modes are observed for both s- and p-polarized incidentlight, in agreement with other reports.39 Interestingly, thedelocalized character of these collective excitations is furtherconfirmed by inspection of the reflectance contour mapsrecorded at different azimuth angles. In particular, thedispersion relation is slightly modified as the plane of incidenceof light is rotated around an axis normal to the sample surface,being the spectra periodic with azimuth period Δϕ = 60°, asexpected from the void hexagonal packing (Figure S1 in theSupporting Information). We note also that a global inspection

Figure 3. Sketches of the experimental setups for (a) opticalreflectivity spectroscopy and (b) magneto-optical spectroscopy.

Langmuir Article

dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−90209013

of all the spectra measured at different azimuth angles showsthat similar absorption bands are excited either by incident s- orp-polarized light, although for each particular angle the spectramay differ significantly. Thus, the contour plots measured withp-polarized light at azimuth angles ϕ = 20°−40° are similar tothose obtained for s-polarized light with azimuth angles ϕ = 0°or ϕ = 60°, and vice versa plots measured with s-polarized lightat azimuth angles ϕ = 20°−40° are somewhat similar to thoseobtained for p-polarized at angles ϕ = 0° or ϕ = 60°(Supporting Information). This feature can be understood interms of the orientation of light polarization with respectto the void geometry, making the coupling to plasmons ofs-(p-)polarized vector fields for one particular direction of theplane of incidence similar to the coupling to plasmons ofp-(s-)polarized vector fields with a plane of incidence rotatedΔϕ = 30° (see the sketch in Figure S1 in the SupportingInformation and Figure 4).Bragg plasmons propagating in a close-packed two-dimen-

sional structure of Au void arrays have been modeled inprevious experimental and theoretical studies. Assuming thatlight is scattered by planes spaced by [3/4(ma + nb )]1/2, where(m, n) are integers and (a , b ) are the two lattice vectors, theBragg plasmon energies can be obtained as a function of theincident angle (see Figure 3 in ref 39). We used these data toidentify the dispersive absorption bands of Figure 4 with thedifferent Bragg mode wavevectors qmn.

39 Figure 4a and b showsthe spectra recorded with the plane of incidence oriented alongthe direction with azimuth ϕ = 0° (see the sketch in this

figure), for s- and p-polarized light, respectively. For p-polarizedlight, we associate the lower energy dispersive band (longerwavelengths) to the q11 mode, while we relate the highestobserved band to the q10 and q01 modes (Figure 4b).Propagating q10, q01 and q11 modes may still be present, albeittheir presence can be damped due to the strong absorption forwavelengths below ca. λ ≈ 500 nm that is related to interbandtransitions of d-electrons in gold.42,43 For s-polarized light, thestrong absorption band is assigned to the q11 ,q01 modes, whileother modes (q10, q10 and q11, q01) may be also damped atshorter wavelengths (Figure 4a). The spectra recorded with theplane of incidence oriented along ϕ = 30° exhibit similarplasmonic excitations, although the assigned modes for s- andp- polarization are now interchanged with respect to ϕ = 0°, inagreement with the cross-correlation of s- and p-spectra withperiodicity Δϕ = 30° discussed above.

Au Nanodisk Arrays. The reflectance contour plots recordedfor s-polarized light on the Au nanodisk arrays are shown inFigure 5. We see that non-dispersive reflection maxima arelocated at wavelengths λ ≈ 660 nm (Figure 5a, d ≈ 57 nm) andλ ≈ 710−720 nm (Figure 5b, c, d ≈ 79, 95 nm), whichcorrespond to resonantly enhanced scattering cross section forwavelengths at which localized plasmons are excited.3 Note thatthe observed wavelength resonances are red-shifted withincreasing diameter, as expected.3 The localized nature ofplasmons excited in the nanodisks is emphasized by the factthat, contrary to the behavior observed in the Au void template,the displayed reflection maxima are nondispersive (Figure 4)and the reflection spectra are essentially insensitive to light

Figure 4. Optical reflectance contour maps (|rss|2 and |rpp|

2) measuredin the Au void surface infiltrated with nickel nanoparticles, for s-polarization (a, c) and for p-polarization (b, d). Data was acquired forthe plane of incidence along the azimuth angles ϕ = 0° (a, b) and 30°(c, d) (see the sketch for the definition of the geometry of theexperiments). The different assigned Bragg modes are indicated bydashed lines.

Figure 5. Optical reflectance contour maps measured for s-polarizationin the Au nanodisk arrays with diameters (a) d ≈ 57 nm, (b) d ≈79 nm, and (c) d ≈ 95 nm, giving the values of |rss|.

Langmuir Article

dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−90209014

polarization, either s or p (not shown). We note also that thedistance between nanodisk centers, being ≈200 nm, is probablytoo large for the interaction of the localized plasmons, inagreement with the expected separation needed for plasmonhybridization which should in the order of a few tens ofnanometers.44−46

Magneto-Optical Spectroscopy. The effects of anarbitrary optical system on the polarization of light reflectedfrom its surface are governed by the diagonal Fresnel reflectioncoefficients rss and rpp that describe the changes of amplitudeand phase for s- and p-polarized light, respectively, as well asthe off-diagonal rsp and rps coefficients that reflect theconversion between s- and p-polarizations. After reflection ona magnetic surface, an orthogonal polarization appears as aconsequence of the magneto-optical activity. For instance, forincident p-polarized light, the reflected wave acquires anadditional small s-polarization and, conversely, a small p-polarization comes up from reflection of an incident s-polarizedlight. Thus, the relative intensity and phase of the polarizationsof the incident and reflected waves gives way to a magneto-optical activity described in terms of rotation

θ θ= =⎡⎣⎢

⎤⎦⎥

⎡⎣⎢⎢

⎤⎦⎥⎥

r

r

r

r;s

ps

ssp

sp

pp (2)

and ellipticity

ε ε= =⎡⎣⎢

⎤⎦⎥

⎡⎣⎢⎢

⎤⎦⎥⎥

r

r

r

r;s

ps

ssp

sp

pp (3)

where the subindices s and p indicate the polarization of theincident light and ℛ and are the real and imaginarycomponents of the reflection coefficient ratios, respectively.The understanding of the microscopic mechanisms that

govern these magneto-optical effects are based on theknowledge of the electronic structure and electron wavefunctions in solids, and the transitions of electrons between thedifferent bands induced by electromagnetic waves.47,48 For theparticular case of the polar Kerr magneto-optical configurationused in our experiments, the complex Kerr angle is shown to berelated to the diagonal (εxx) and off-diagonal (εxy) permittivitytensor components through47,48 η = θ + iε ≈ (−εxy/(εxx −1)(εxx)

1/2). Thus, we see that the appearance of off-diagonalpermittivity components is an essential ingredient for theemergence of magneto-optical activity. In the dipolarapproximation, it can be shown that47,48

∑ε ρ ω

ω ω ω ωω ω ω

∝ | + | − | − |

− − Γ− + Γ + Γ

≠ d id d id{ }

( )( ) 4

ij i jab

a ab abi

abj

abi

abj

ab ab ab

ab ab ab

( )2 2

2 2 2

2 2 2 2 2(4)

where the sum is taken over ground (a) and excited (b) states,ρa is the probability for one electron to reside in the a-state, anddabi , dab

j are matrix elements of the ith or jth component of thedipole moment connecting a- and b-states that are separated byan energy ℏωab, being Γab the half-width of the a to b transition.Note that dipole matrix elements are defined as dab

i = ⟨a| Ei(xi)xi|b⟩, Thus, an electric field intensification and the ensuingincrease of density of optical modes49 occurring in the vicinityof surfaces where plasmons are excited might provide a natural

Figure 6. Hysteresis loops for s- and p-polarization of the ellipticity (a, g) and rotation (d, j) of nickel nanoparticles deposited inside the Au voids(red curves) and on the non-patterned area outside the voids (black curves). The figure also includes the contour plots obtained for s-polarization ofthe ellipticity (b, c) and rotation (e, f), and for p-polarization of the ellipticity (h, i) and rotation (k, l).

Langmuir Article

dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−90209015

mechanism to amplify the magneto-optical activity. One way todo this is to place the magneto-optical material at small regionswhere plasmons are excited, as is the case of the patternedmetal surfaces discussed here, where the electromagnetic fieldsare strongly confined around the metal/dielectric interfaces andincreasing hugely the electromagnetic density of states.3

Alternatively, according to eqs 2 and 3, the intensification ofthe magneto-optical activity may be also originated by adecrease of the diagonal coefficients rss and rpp, providing anindependent pathway toward magneto-optical enhance-ment.23,50 As described in the following, the combination ofoptical and magneto-optical spectroscopy has allowed us toquantify the contribution of both factors. We find that theenhancement of the magneto-optical activity in the plasmonicstructures here presented is essentially due to an intrinsicincrease of the off-diagonal coefficients terms rsp and rps. Beforeproceeding, we note that the Au void and nanodisk arrays didnot exhibit magneto-optical activity before the impregnationwith the magnetic nanoparticles.Magneto-Optical Spectra of the Au Void Surface

Infiltrated with Nickel Nanoparticles. The rotation (θ) andellipticity (ε) hysteresis loops where recorded as a function ofthe light wavelength within λ ≈ 390−850 nm with a stepresolution of Δλ = 1 nm for magnetic fields |H| ≤ 15 kOe. Thedata were measured in polar Kerr configuration with lightincident at α ≈ 9.5° from the normal to the sample surface andthe plane of incidence of the light was along the azimuth angleϕ = 30°. Figure 6a and d shows the ellipticity and rotationloops measured with incident s-polarized light at the selectedwavelengths indicated in the figure, whereas Figure 6g and jdisplays the magneto-optical data for p-polarization. All thesefigures show the curves recorded with the beam light focusedinside the structured area (Au voids, dashed lines) as well asoutside (non-structured Au surface, full lines). The S-like shapeof the loops and the negligible coercive field are typical ofsuperparamagnetic behavior, representative of aggregates ofmagnetic nanoparticles, with saturation field H ≈ 10 kOe.Remarkably, at the selected wavelengths, the magnitude of themagneto-optical hysteresis loops recorded inside the voids areof the order of a few mrad, in contrast to the values measuredin the non-structured area (∼0.1 mrad); see Figure 6a, d, g, andj. This large magneto-optical enhancement, of more than 1order of magnitude for some wavelengths, comes up in spite ofthe nanoparticle coverage of the non-structured area beingroughly two times larger than that of the Au voids. This clearlyindicates that the magneto-optical response of nanoparticlesdeposited on the structured Au surface is strongly modifiedwith respect to the response of the same nanoparticles on thenon-structured area. This conclusion is further confirmed byinspecting the magneto-optical contour plots obtained bymapping the rotation and ellipticity as a function of wavelengthand magnetic field (|H| ≤ 15 kOe). The contour plots weremeasured with light incident inside the Au void array for s-polarized light (ellipticity and rotation in Figure 6b and e,respectively) and for p-polarized light (ellipticity and rotation inFigure 6h and k, respectively). Similar plots were also obtainedfor the non-structured surface (Figure 6c, f, i, l). We observethat while the plots corresponding to the non-structured areasappear relatively featureless on the plotted scale, with rotation/ellipticity values below ≈0.5 mrad, the plots corresponding tothe voids display relatively narrow spectral regions where themagneto-optical response is remarkably intensified, with valuesup to ≈3 mrad. For instance, for s-polarized incident light the

ellipticity is strongly peaked at around λ ≈ 400, 550, and590 nm (Figure 6b), whereas the rotation is especially intensiveat λ ≈ 540 nm. On the other hand, for p-polarized light, theellipticity is faintly intensified around λ ≈ 400 nm, whereas therotation is significantly enhanced at λ ≈ 540 and 625 nm.The magneto-optical responses are then enhanced around

spectral regions that are close to the wavelengths whereplasmons are excited for light incident at α ≈ 9.5° (Figure 4).To further investigate the connection between magneto-opticsand plasmonics, we extracted from the measurements of therotation and ellipticity the absolute value of the Kerr angle |ηs,p|= (θs,p

2 + εs,p2)1/2 for s- and p-polarized light, shown in Figure

7a and d, respectively. Since the values of the diagonal

reflection coefficients |rss| and |rpp| are directly drawnfrom the reflectance spectra of Figure 4, we were in positionto obtain the off-diagonal coefficients |rsp| and |rps| through eqs2−3, as |rsp,ps| = |ηs,p| × |rss,pp|. The values of |ηs|, |rsp|, and |rss| fors-polarized light are shown in Figure 7a−c, whereas |ηp|, |rps|,and |rpp| for p-polarized light are displayed in Figure 7d−f. Alldata in Figure 7 are plotted for light focused both inside andoutside the structured area. The different Bragg modewavelengths are inferred by the appearance of dips in thereflectance spectra, and are indicated by arrows in the curves of|rss| (Figure 7c) and |rpp| (Figure 7f), which are obtained fromcuts made at an angle of incidence of α = 9.5° of the reflectancespectra of Figure 4c and d, respectively. We observe that theKerr angles |ηs|, |ηp| and the off-diagonal coefficients |rsp|, |rps| areparticularly enhanced in the void areas with respect to the non-structured regions for a certain range of wavelengths. Thus, fors-polarized light, two main regions are identified for which the

Figure 7. Plots of the Kerr angles (a) |ηs| and (d) |ηp|, the off-diagonalreflectance coefficients (b) |rsp| and (e) |rps| and the diagonalreflectance coefficients (c) |rss| and (f) |rpp| measured in the Auvoids. The values of |ηs|, |ηp|, |rsp|, and |rps| were obtained at field H =10 kOe, where the magnetization of the nickel nanoparticles issaturated. The values of |rss| and |rpp| were obtained from cuts atincidence angle 9.5° of the spectra of Figure 3a and b, respectively.

Langmuir Article

dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−90209016

patterned area exhibits |ηs| and |rsp| that are up to an order ofmagnitude larger than in the nonpatterned surface. One islocated around a narrow region around λ ≈ 540 nm, for whichthe Bragg modes q10,q01 are excited. These modes, althoughpredicted by the theory,39 are hardly identified in thereflectance spectra (Figures 4c and 7c), but Figure 7a and bdemonstrates that they have a conspicuous effect on themagneto-optical properties. A second region is found within abroad region λ ≈ 600−800 nm, centered at λ ≈ 700 nm, whichis relatively close to the wavelengths at which the q11 Braggmodes are excited. We note that q10,q01 modes seem to have aless relevant influence on the magneto-optical activity.Similarly, for p-polarized light, |ηp| and |rps| inside the voidsurface are magnified by around of order of magnitude withrespect to the non-structured region for wavelengths centeredon a narrow region λ ≈ 540 nm. A second broader region isidentified around 630 nm. Although it might be associated toBragg modes q11, q01 and q1 0, q10, one cannot exclude thatlocalized Mie-like resonances play a role on the magneto-optical enhancement around these broader spectral regions. Wenote that these localized resonances might be smeared outwhen observed using a large light spot, but they may be presentat each individual cavity, with resonance frequencies fluctuatingbecause of random distributions of nanoparticles inside thevoids.Magneto-Optical Spectra of the Au Nanodisk Arrays

Infiltrated with Iron Oxide Nanoparticles. An analogousmagneto-optical spectral study was carried out on the Aunanodisk arrays infiltrated with iron oxide nanoparticles. Figure8a and f shows the ellipticity and rotation hysteresis loops,respectively, measured for s-polarized light focused on the threedifferent nanodisk patterns (d ≈ 57, 79, 95 nm) as well as onnanoparticles deposited on the non-structured surface. Weobserve that, as before, the shape of the hysteresis loopsindicates a superparamagnetic behavior of the infiltratednanoparticles. The downturn in the magnitude of the hysteresisloops for fields |H| ≥ 4 kOe is related to a diamagneticmagneto-optical contribution coming from the SrTiO3substrate underneath the nanoparticles, that at high enoughfields, above the magnetization saturation, can overcome thesignal from nanoparticles. This is the expected behavior fromthe diamagnetic response of SrTiO3 (see Figure S2 in theSupporting Information). This contribution can be removed toobtain the contributions of nanoparticles alone. Interestingly,the magneto-optical activity of the nanoparticles deposited onthe patterned areas reaches values of up to ∼6 mrad, whereasnanoparticles on the nonpatterned area exhibit much lowerellipticity/rotation, in the order of ∼0.8 mrad. As before, this isa clear indication that the magneto-optical spectra aresubstantially modified when nanoparticles are deposited onthe nanostructured metallic surfaces.Contour plots for the ellipticity (Figure 8b−e) and rotation

(Figure 8g−j) were obtained by mapping the magneto-opticalactivity as a function of wavelength (λ ≈ 390−850 nm) andmagnetic field (|H| ≤ 7 kOe). Again, this allowed us to get anoverall picture of the effect on the magneto-optical spectra ofnanostructuring the metal/dielectric surface. In particular, weagain observed that the magneto-optical activity was signifi-cantly enhanced around relatively narrow spectral regionscentered at wavelengths in the vicinity of the localized plasmonresonances. In contrast, for light incident on the non-structuredarea, the spectra were quite featureless when plotted on thesame scale. For instance, for nanodisks with diameter d ≈

57 nm, the ellipticity was especially intensified around λ ≈ 670nm (Figure 8b), while the rotation was peaked at differentwavelengths around λ ≈ 400, 470, 610, and 725 nm (Figure8g), with values that reach up to ≈6 mrad. For d ≈ 79 nm, theellipticity appears particularly intensified around λ ≈ 675 and725 nm (Figure 8c), whereas the rotation is faintly magnifiedaround λ ≈ 725 nm (Figure 8h). Finally, for d ≈ 95 nm, theellipticity appears to be increased at λ ≈ 520, 650, and 725 nm(Figure 8d), with the rotation slightly enhanced at λ ≈ 390 and590 nm (Figure 8i).As discussed previously for the Au void array, the relationship

between the magneto-optical spectra and plasmonics can be bet-

Figure 8. Hysteresis loops for s- polarization of the ellipticity (a) androtation (d) of iron oxide nanoparticles deposited on the nanodiskarrays (red, magenta, and green lines) and on the nonpatterned areaoutside the disks (black curves). The figure also includes the contourplots obtained for s-polarization of the ellipticity (b−e) and rotation(g−j).

Langmuir Article

dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−90209017

ter appreciated by plotting |ηs,p| = (θs,p2 + εs,p

2)1/2, |rsp| and |rss|(see Figure 9). We observe that the Kerr angle is stronglypeaked at wavelengths λ ≈ 720 nm for disks d ≈ 79 and 95 nm,and λ ≈ 670 nm for d ≈ 57 nm, with maximum values thatreach up to |ηs| ≈ 6.7 mrad (Figure 9a). Note that in the non-structured region |ηs| displays significantly lower values all overthe spectrum, and it does not exhibit any peak within λ ≈ 600−800 nm, with values |ηs| ≤ 0.8 mrad over that spectral region.The behavior of |rsp| is quite similar that of to |ηs|, with peakvalues within λ ≈ 600−800 nm for the nanodisk area, and beingquite flat over that region for the non-structured surface (Figure9b). When compared to the reflectance |rss| spectra, we observethat the maxima of |ηs| and |rsp| appear to be in the vicinity ofmaxima for |rss| (Figure 9c), in agreement with the resonantlyenhanced light scattering cross section for localized plasmons.Thus, we demonstrate that the coupling of the magneto-opticalactivity of magnetic nanoparticles with localized plasmonexcitations of the Au nanodisk arrays gives way to a significantmodification of the magneto-optical spectral response and to anenhancement of the magneto-optical activity.About the Origin of the Magneto-Optical Enhance-

ment. We recall from eqs 2 and 3 that the absolute values ofthe Kerr angle are given by |ηs| = (|rsp|/|rss|), |ηp| = (|rps|/|rpp|),

for s- and p-polarized light, respectively. Thus, with thecorrelation between magneto-optical properties and plasmonexcitationsbeing established, a question arises to which degreethe observed magneto-optical enhancements of |ηs|and |ηp| aredue to an intensification of the off-diagonal reflectance |rsp|, |rps|coefficients, associated to the plasmon-assisted electric fieldenhancement, or to a modification of the spectral response ofthe diagonal |rss| |rpp| terms. To answer this question, we havecalculated from the experimental values the ratios (|ηs|pat/|ηs|nonpat) = (|rsp|pat/|rsp|nonpat)(|rss|nonpat/|rss|pat) and (|ηp|pat/|ηp|nonpat) = (|rps|pat/|rps|nonpat)(|rpp|nonpat/|rpp|pat), where thesubindices pat and nonpat correspond to the spectra recordedin structured (void or nanodisk arrays) and flat non-structuredsurfaces, respectively. Thus, we are in a position to find out theseparate contributions of the diagonal (|rpp|nonpat/|rpp|pat) andoff-diagonal (|rsp|pat/|rsp|nonpat) terms to the magneto-opticalenhancements observed in Figure 7a,d and 9a. In Figure 10, we

have plotted the ratios of the off-diagonal and diagonalcoefficients for the Au void array, measured for s- (Figure10a) and p- (Figure 10b) polarizations, and for the threenanodisk arrays (Figure 10c). The regions with enhancedmagneto-optical activity are identified in these figures byarrows. An inspection of these data reveals immediately that themagneto-optical enhancement is determined almost entirely by

Figure 9. Plots of the Kerr angle |ηs| (a), the off-diagonal reflectancecoefficient |rsp| (b), and the diagonal reflectance coefficient |rss| (c),measured in the Au nanodisk arrays. The values of |ηs| and |rsp| wereobtained at field H = 4 kOe, where the magnetization of the iron oxidenanoparticles is saturated. The values of |rss| where obtained from cutsat normal incidence of the spectra of Figure 4.

Figure 10. Plots of the ratios of the off-diagonal (|rsp|pat/|rsp|nonpat,|rps|pat/|rps|nonpat) and diagonal (|rss|nonpat/|rss|pat, |rpp|nonpat/|rpp|pat)reflectance coefficients obtained in the Au voids for s-polarization(a) and p-polarization (b), and in the nanodisk arrays for s-polarization(c). Based on the spectra of Figures 3 and 4, the wavelengths at whichplasmons are excited are indicated by arrows.

Langmuir Article

dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−90209018

an extraordinary increase of the off-diagonal reflectancecoefficients, whereas the diagonal terms play a minor role.For instance, for the Au voids, the coefficients |rsp| and |rps| areenhanced by up to a factor ≈50 in the structured regions,whereas the diagonal coefficients |rss| and |rpp| are increased by afactor ≈1−5 in the relevant spectral regions; see Figure 10a andb. The increase of |rsp| is still more spectacular in the nanodiskarrays, where for the relevant region enhancement factors ofabout 3 orders of magnitude are found, whereas the ratio of |rss|remains close to the unity or is even suppressed in the regionwhere plasmons are excited. We can conclude, then, that themagneto-optical enhancement of nanoparticles deposited onnanostructured metal/dielectric surfaces is almost entirely dueto an outstanding increase of the conversion rate betweenorthogonal polarizations, rather than by a modification of thereflectance.

■ CONCLUSIONS

We found that the magneto-optical spectra of magneticnanoparticles inside nanostructured metal/dielectric surfacesare dramatically modified compared to the spectra ofnanoparticles randomly distributed on non-structured flatsurfaces. In particular, the magneto-optical activity is sub-stantially increased for wavelengths at which surface plasmons,either localized resonances or extended Bragg modes, areexcited on the nanostructured metal/dielectric interfaces. Wehave achieved a quantitative evaluation of the contributions ofboth intrinsic polarization conversion efficiency and reflectancemodulations to the observed magneto-optical enhancement.Interestingly, we have demonstrated that plasmon-induced fieldenhancement is responsible for the spectacular increase of theintrinsic polarization conversion efficiency, which is at least 1order of magnitude larger than that from reflectancemodulations, thus shedding light on the fundamental principlesfor the observed optical response.Our results prove the potential of surface plasmons to

generate large magneto-optical signal enhancements at specificwavelengths and to design promising strategies to modify thespectral optical responses of magneto-optical materials. Hybridsystems composed of magnetic nanoparticles with corrugatedmetal/dielectric surfaces offer a promising strategy for newapplications, where very small changes of polarization of lightassociated to magneto-optical effects can be exploited forsensing applications. This approach offers a high potential forsensing, especially for wavelengths that match the plasmonresonances with targeted patterned surfaces, giving highsensitivity even for low coverage of nanoparticles.

■ ASSOCIATED CONTENT

*S Supporting InformationDependence on the azimuth angle of the optical reflectancespectra of Au voids infiltrated with nanoparticles. This materialis free of charge via the Internet http://pubs.acs.org.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: gherranz@icmab.es.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

We acknowledge financial support from the Spanish Govern-ment (CONSOLIDER Nanoselect-CSD2007-00041,MAT2011-29269-C03-01, MAT2009-08024, MAT2009-06885-E, FPI grant of J.M.C. and CSIC JAE-predoc grants ofO.V. and M.S.), the Generalitat de Catalunya (2009SGR-376,2009SGR-203 and FI grant of O.P.) and from ANPCyT(Argentina, PICT08-1617). We acknowledge Dr. F. Sanchezfor his support and scientific advising in pulsed laser depositiongrowth.

■ REFERENCES(1) Soukoulis, C. M.; Linden, S.; Wegener, M. Negative RefractiveIndex at Optical Wavelengths. Science 2007, 315, 47−49.(2) Galisteo-Lopez, J. F.; Ibisate, M.; Sapienza, R.; Froufe, L.; Blanco,A; Lopez, C. Self-Assembled Photonic Structures. Adv. Mater. 2011,23, 30−69.(3) Maier, S. A. Plasmonics: Fundamentals and Applications; SpringerScience + Business Media LLC: New York, 2007.(4) Yablonovitch, E. Inhibited Spontaneous Emission in Solid-StatePhysics and Electronics. Phys. Rev. Lett. 1987, 58, 2059−2062.(5) Yablonovitch, E. Photonics: One-way road for light. Nature 2009,461, 744−745.(6) Tognalli, N. G.; Cortes, E.; Hernandez-Nieves, A. D.; Carro, P.;Usaj, G.; Balseiro, C. A.; Vela, M. E.; Salvarezza, R. C.; Fainstein, A.From Single to Multiple Ag-Layer Modification of Au NanocavitySubstrates: A Tunable Probe of the Chemical Surface-EnhancedRaman Scattering Mechanism. ACS Nano 2011, 5, 5433−5443.(7) Le Ru, E. C.; Meyer, M.; Etchegoin, P. J. Proof of Single-Molecule Sensitivity in Surface Enhanced Raman Scattering (SERS) byMeans of a Two-Analyte Technique. J. Phys. Chem. B 2006, 110,1944−1948.(8) Wang, Z.; Chong, Y.; Joannopoulos, J. D.; Soljacic, M.Observation of unidirectional backscattering-immune topologicalelectromagnetic states. Nature 2009, 461, 772−775.(9) Takahashi, K.; Kawanishi, F.; Mito, S.; Takagi, H.; Shin, K. H.;Kim, J.; Lim, P. B.; Uchida, H.; Inoue, M. Study on magnetophotoniccrystals for use in reflection-type magneto-optical spatial lightmodulators. J. Appl. Phys. 2008, 103, 07B331.(10) Boriskina, J. V.; Erokhin, S. G.; Granovsky, A. B.; Vinogradov, A.P.; Inoue, M. Enhancement of the magnetorefractive effect inmagnetophotonic crystals. Phys. Solid State 2006, 48, 717−721.(11) Fedyanin, A. A.; Aktsipetrov, O. A.; Kobayashi, D.; Nishimura,K.; Uchida, H.; Inoue, M. Enhanced Faraday and nonlinear magneto-optical Kerr effects in magnetophotonic crystals. J. Magn. Magn. Mater.2004, 282, 256−259.(12) Inoue, M.; Fujikawa, R.; Baryshev, A.; Khanikaev, A.; Lim, P. B.;Uchida, H.; Aktsipetrov, O.; Fedyanin, A.; Murzina, T.; Granovsky, A.Magnetophotonic crystals. J. Phys. D: Appl. Phys. 2006, 39, R151.(13) Zvezdin, A. K.; Belotelov, V. I. Magnetooptical properties of twodimensional photonic crystals. Eur. Phys. J. B 2004, 37, 479−487.(14) Liu, N.; Guo, H.; Fu, L.; Kaiser, S.; Schweizer, H.; Giessen, H.Three-dimensional photonic metamaterials at optical frequencies. Nat.Mater. 2008, 7, 31−37.(15) Fang, M.; Volotinen, T. T.; Kulkarni, S. K.; Belova, L.; V. Rao,K. Effect of embedding Fe3O4 nanoparticles in silica spheres on theoptical transmission properties of three-dimensional magneticphotonic crystals. J. Appl. Phys. 2010, 108, 103501.(16) Simkiene, I.; Reza, A.; Kindurys, A.; Bukauskas, V.; Babonas, J.;Szymczak, R.; Aleshkevych, P.; Franckevicius, M.; Vaisnoras, R.Magnetooptics of opal crystals modified by cobalt nanoparticles. Lith.J. Phys. 2010, 50, 7−15.(17) Pavlov, V. V.; Usachev, P. A.; Pisarev, R. V.; Kurdyukov, D. A.;Kaplan, S. F.; Kimel, A. V.; Kirilyuk, A.; Rasing, Th. Optical study ofthree-dimensional magnetic photonic crystals opal/Fe3O4. J. Magn.Magn. Mater. 2009, 321, 840−842.

Langmuir Article

dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−90209019

(18) Caicedo, J. M.; Taboada, E.; Hrabovsky, D.; Lopez-García, M.;Herranz, G.; Roig, A.; Blanco, A.; Lopez, C.; Fontcuberta, J. Facileroute to magnetophotonic crystals by infiltration of 3D inverse opalswith magnetic nanoparticles. J. Magn. Magn. Mater. 2010, 322, 1494−1496.(19) Caicedo, J. M.; Pascu, O.; Lopez-García, M.; Canalejas, V.;Blanco, A.; Lo pez, C.; Fontcuberta, J.; Roig, A.; Herranz, G.Magnetophotonic Response of Three-Dimensional Opals. ACS Nano2011, 5, 2957−2963.(20) Pascu, O.; Caicedo, J. M.; Lopez-García, M.; Canalejas, V.;Blanco, A.; Lopez, C.; Arbiol, J.; Fontcuberta, J.; Roig, A.; Herranz, G.Ultrathin conformal coating for complex magneto-photonic structures.Nanoscale 2011, 3, 4811−4816.(21) Raether, H. Surface Plasmons; Springer-Verlag: Berlin,Heidelberg, 1988.(22) Hermann, C.; Kosobukin, V. A.; Lampel, G.; Peretti, J.; Safarov,V. I.; Bertrand, P. Surface-enhanced magneto-optics in metallicmultilayer films. Phys. Rev. B: Condens. Matter Mater. Phys. 2001, 64,235422.(23) Gonzalez-Díaz, J. B.; García-Martín, A.; Armelles, G.; García-Martín, J. M.; Clavero, C.; Cebollada, A.; Lukaszew, R. A.; Skuza, J. R.;Kumah, D. P.; Clarke, R. Surface-magnetoplasmon nonreciprocityeffects in noble-metal/ferromagnetic heterostructures. Phys. Rev. B:Condens. Matter Mater. Phys. 2007, 76, 153402.(24) Temnov, V. V.; Armelles, G.; Woggon, U.; Guzatov, D.;Cebollada, A.; Garcia-Martin, A.; Garcia-Martin, J. M.; Thomay, T.;Leitenstorfer, A.; Bratschitsch, R. Active magneto-plasmonics in hybridmetal−ferromagnet structures. Nat. Photonics 2010, 4, 107−111.(25) Ctistis, G.; Papaioannou, E.; Patoka, P.; Gutek, J.; Fumagalli, P.;Giersig, M. Optical and Magnetic Properties of Hexagonal Arrays ofSubwavelength Holes in Optically Thin Cobalt Films. Nano Lett. 2009,9, 1−6.(26) Papaioannou, E. T.; Kapaklis, V.; Patoka, P.; Giersig, M.;Fumagalli, P.; Garcia-Martin, A.; Ferreiro-Vila, E.; Ctistis, G. Magneto-optic enhancement and magnetic properties in Fe antidot films withhexagonal symmetry. Phys. Rev. B: Condens. Matter Mater. Phys. 2010,81, 054424.(27) Torrado, J. F.; Papaioannou, E. T.; Ctistis, G.; Patoka, P.;Giersig, M.; Armelles, G.; Garcia-Martin, A. Plasmon inducedmodification of the transverse magneto-optical response in Fe antidotarrays. Phys. Status Solidi RRL 2010, 4, 271−273.(28) Torrado, J. F.; Gonzalez-Díaz, J. B.; Armelles, G.; García-Martín,A.; Altube, A.; Lopez-García, M.; Galisteo-Lopez, J. F.; Blanco, A.;Lopez, C. Tunable magneto-photonic response of nickel nanostruc-tures. Appl. Phys. Lett. 2011, 99, 193109.(29) Papaioannou, E. T.; Kapaklis, V.; Melander, E.; Hjorvarsson, B.;Pappas, S. D.; Patoka, P.; Giersig, M.; Fumagalli, P.; Garcia-Martin, A.;Ctistis, G. Surface plasmons and magneto-optic activity in hexagonalNi anti-dot arrays. Opt. Express 2011, 19, 23867−23877.(30) Belotelov, V. I.; Akimov, I. A.; Pohl, M.; Kotov, V. A.; Kasture,S.; Vengurlekar, A. S; Gopal, A. V.; Yakovlev, D. R.; Zvezdin, A. K.;Bayer, M. Enhanced magneto-optical effects in magnetoplasmoniccrystals. Nat. Nanotechnol. 2011, 6, 370−376.(31) Gonzalez-Díaz, J. B.; García-Martín, A.; Armelles, G.; Navas, D.;Vazquez, M.; Nielsch, K.; Wehrspohn, R. B.; Gosele, U. EnhancedMagneto-Optics and Size Effects in Ferromagnetic Nanowire Arrays.Adv. Mater. 2007, 19, 2643−2647.(32) Du, G. X.; Mori, T.; Suzuki, M.; Saito, S.; Fukuda, H.;Takahashi, M. Evidence of localized surface plasmon enhancedmagneto-optical effect in nanodisk array. Appl. Phys. Lett. 2010, 96,081915.(33) Chen, J.; Albella, P.; Pirzadeh, Z.; Alonso-Gonzalez, P.; Huth, F.;Bonetti, S.; Bonanni, V.; Åkerman, J.; Nogues, J.; Vavassori, P.;Dmitriev, A.; Aizpurua, J.; Hillenbrand, R. Plasmonic NickelNanoantennas. Small 2011, 7, 2341−2347.(34) Newman, D. M.; Wears, M. L.; Matelon, R. J.; McHugh, D.Non-linear optics and magneto-optics on nano-structured interfaces.Appl. Phys. B: Lasers Opt. 2002, 74, 719−722.

(35) Newman, D. M.; Wears, M. L.; Matelon, R. J.; Hooper, I. R.Magneto-optic behaviour in the presence of surface plasmons. J. Phys.:Condens. Matter 2008, 20, 345230.(36) Bartlett, P. N.; Baumberg, J. J.; Birkin, P. R.; Ghanem, M. A.;Netti, M. C. Highly Ordered Macroporous Gold and Platinum FilmsFormed by Electrochemical Deposition through Templates Assembledfrom Submicron Diameter Monodisperse Polystyrene Spheres. Chem.Mater. 2002, 14, 2199−2208.(37) Lacharmoise, P. D.; Tognalli, N. G.; Goni, A. R.; Alonso, M. I.;Fainstein, A.; Cole, R. M.; Baumberg, J. J.; García de Abajo, J.; Bartlett,P. N. Imaging optical near fields at metallic nanoscale voids. Phys. Rev.B: Condens. Matter Mater. Phys. 2008, 78, 125410.(38) Pascu, O.; Caicedo, J. M.; Fontcuberta, J.; Herranz, G.; Roig, A.Magneto-Optical Characterization of Colloidal Dispersions. Applica-tion to Nickel Nanoparticles. Langmuir 2010, 26, 12548−12552.(39) Kelf, T. A.; Sugawara, Y.; Cole, R. M.; Baumberg, J. J.;Abdelsalam, M. E.; Cintra, S.; Mahajan, S.; Russell, A. E.; Bartlett, P. N.Localized and delocalized plasmons in metallic nanovoids. Phys. Rev. B:Condens. Matter Mater. Phys. 2006, 74, 245415.(40) Cole, R. M.; Baumberg, J. J.; Garcia de Abajo, F. J.; Mahajan, S.;Abdelsalam, M.; Bartlett, P. N. Understanding Plasmons in NanoscaleVoids. Nano Lett. 2007, 7, 2094−2100.(41) Tognalli, N. G.; Fainstein, A.; Calvo, E. J.; Abdelsalam, M.;Bartlett, P. N. Incident Wavelength Resolved Resonant SERS on AuSphere Segment Void (SSV) Arrays. J. Phys. Chem. C 2012, 116,3414−3420.(42) Fox, M. Optical Properties of Solids; Oxford University Press:Oxford, 2001.(43) Lide, D. R. CRC handbook of chemistry and physics; CRC Press:Boca Raton, FL, 1996.(44) Tognalli, N.; Fainstein, A.; Calvo, E.; Bonazzola, C.; Pietrasanta,L.; Campoy-Quiles, M.; Etchegoin, P. SERS in PAH-Os and goldnanoparticle self-assembled multilayers. J. Chem. Phys. 2005, 123,044707.(45) Nordlander, P.; Oubre, C.; Prodan, E.; Li, K.; Stockman, M. I.Plasmon Hybridization in Nanoparticle Dimers. Nano Lett. 2004, 4,899−903.(46) Jung, J.; Pedersen, T. G.; Søndergaard, T.; Pedersen, K.;Nylandsted Larsen, A.; Nielsen, B. B. Electrostatic plasmon resonancesof metal nanospheres in layered geometries. Phys. Rev. B: Condens.Matter Mater. Phys. 2010, 81, 125413.(47) Zvezdin, A.; Kotov, V. Modern Magnetooptics and MagnetoopticalMaterials; Taylor & Francis: London, 1997.(48) Antonov, V.; Harmon, B.; Yaresko, A. Electronic Structure andMagneto-Optical Properties of Solids; Springer: New York, 2004.(49) Snoeks, E.; Lagendijk, A.; Polman, V. Measuring and Modifyingthe Spontaneous Emission Rate of Erbium near an Interface. Phys. Rev.Lett. 1995, 74, 2459−2462.(50) Gonzalez-Díaz, J. B.; García-Martín, J. M.; García-Martín, A.;Navas, D.; Asenjo, A.; Vazquez, M.; Hernandez-Velez, M.; Armelles, V.Plasmon-enhanced magneto-optical activity in ferromagnetic mem-branes. Appl. Phys. Lett. 2009, 94, 263101.

Langmuir Article

dx.doi.org/10.1021/la301239x | Langmuir 2012, 28, 9010−90209020