100 CHAPTER 6. COMPOSITE MEDIA BASED ON AU/LIF ARRAYSThepatternisparticularlyemphasizedbythecorrespondingFourierspectraldensity(inset<strong>of</strong> fig. 6.2(d)), which is characterized by two sharp peaks confined along the [1¯10] <strong>and</strong> the[001] directions, indicating the presence <strong>of</strong> periodic features both along <strong>and</strong> across theripples. These observations suggest that, even if the Fe 3 O 4 /OA NPs do not follow the AuNPs arrangement with a one-to-one correspondence, still the Au NPs provide preferentialsites for the adsorption <strong>of</strong> the magnetic NPs, so that their arrangement can be effectivelyreproduced by groups <strong>of</strong> few Fe 3 O 4 /OA NPs.Similarly to the previous cases, the optical response <strong>of</strong> the sample was optically characterizedby means <strong>of</strong> s- <strong>and</strong> p-polarized light reflectivity, with the plane <strong>of</strong> incidenceeither parallel or perpendicular to the LiF ripples. The R S <strong>and</strong> R P spectra acquired atan incidence angle <strong>of</strong> θ = 50 ◦ , before <strong>and</strong> after the Fe 3 O 4 /OA deposition, are reported infig. 6.3.The presence <strong>of</strong> the magnetite layer affects the LSPs by systematically redshiftingthe resonances <strong>and</strong> increasing their width; the overall intensity <strong>of</strong> the reflectivity is alsopspsbefore depositionafter deposition0.04R P||0.30R S┴R P0.02R S0.200.100.004006008001000a. [nm]b.400600 800 [nm]10000.24R S||0.05R p┴R S0.16R P0.030.080.01400600 800 [nm]1000c. d.400600 800 [nm]1000Figure 6.3: Reflectivities, at θ = 50 ◦ , measured before (red curves) <strong>and</strong> after (blackcurves) the deposition <strong>of</strong> ≈ 1 monolayer <strong>of</strong> Fe 3 O 4 OA NPs on 2D <strong>arrays</strong> <strong>of</strong> gold NPs (seefig. 6.2). Plane <strong>of</strong> incidence either parallel (left panels) <strong>and</strong> perpendicular (right panels) tothe ripples. EM excitation along (panels (a), (b)) <strong>and</strong> across (panels (c), (d)) the ripples.
101increased. Before the deposition (red curves in fig. 6.3), the longitudinal LSP mode isfound at λ L = 620 nm, with a width <strong>of</strong> Γ L = 170 nm; the transverse mode is insteadlocated at λ T = 570 nm <strong>and</strong> has a width <strong>of</strong> Γ T = 125 nm. Following the deposition (blackcurves in fig. 6.3), both modes redshift by few tens <strong>of</strong> nm, the L mode at λ ∗ L = 645 nm<strong>and</strong> the T mode at λ ∗ T = 585 nm, while the corresponding widths become Γ∗ L = 215 nm<strong>and</strong> Γ ∗ T = 145 nm, respectively.This trend is in accordance with the qualitative considerations in §1.3.3, where wediscussed the dependence <strong>of</strong> the dielectric environment on the LSP resonances. In particular,the simultaneous redshift <strong>of</strong> the resonances can be explained in terms <strong>of</strong> the Fröhlichcondition Re [ ε L,Tm (λ L,T ) ] = −2ε h (see eq. (1.45)), where ε m , ε L,Th<strong>and</strong> λ L,T are thedielectric constant <strong>of</strong> the individual Au NPs <strong>and</strong> <strong>of</strong> the host <strong>and</strong> the position <strong>of</strong> the resonances,respectively. ε m is shown in fig. 5.6(b), <strong>and</strong> its real part has a negative slopefor increasing wavelengths; in fig. 6.4 we report instead the dielectric constant <strong>of</strong> bulkmagnetite, extracted from ref. [220]. In analogy with the model <strong>of</strong> the previous chapter,in first approximation the dielectric constant ε h can be expressed as a linear combination<strong>of</strong> the dielectric constants <strong>of</strong> the materials surrounding the Au NPs; therefore, followingthe deposition <strong>of</strong> the Fe 3 O 4 /OA NPs, ε h is expected to increase, because the real part<strong>of</strong> the dielectric constant <strong>of</strong> magnetite reads about ε 1 = 5 in correspondence <strong>of</strong> the goldnanoparticles LSP resonances. Then, according to the Fröhlich condition, if ε h increases<strong>and</strong> ε m has a negative slope as a function <strong>of</strong> the wavelength, the LSP resonances <strong>of</strong> thegold NPs must red shift when embedded in the Fe 3 O 4 /OA layer.64.053.5Re []43.0Im []32.522.040060080010001200 [nm]Figure 6.4: Real (red line) <strong>and</strong> imaginary (black line) parts <strong>of</strong> the dielectric constant <strong>of</strong>bulk magnetite in the visible <strong>and</strong> near-IR range, from ref. [220].In order to quantitatively evaluate the effects <strong>of</strong> the Fe 3 O 4 /OA NPs on the <strong>plasmonic</strong>response <strong>of</strong> the Au NPs array, we can apply the theoretical framework developed in theprevious chapter to the current system. In fig. 6.5(a) we report the computed R S spectrafor the “bare” Au/LiF nanostructures, compared to the corresponding experimentalvalues, for the L <strong>and</strong> T LSP modes, at an angle <strong>of</strong> incidence <strong>of</strong> θ = 50 ◦ . In analogywith before, the morphological parameters employed for the calculations were obtainedfrom the analysis <strong>of</strong> the AFM data: for the NPs spacings across <strong>and</strong> along the ripples weused d x = d y = 33 nm, while the values <strong>of</strong> the ellipsoids semiaxes were a x = 11.2 nm,a y = 14.0 nm <strong>and</strong> a z = 6.4 nm.The corresponding spectra in presence <strong>of</strong> the Fe 3 O 4 /OA NPs were computed keepingthe same input parameters, except two specific modifications. Considering a single