Figure 1 shows a conceptual sketch of the device. One may see the array of the nanoparticles that are made up of two layers (Au on top of a Si) that are being illuminated by two light beams. On the upper left corner we present the polarization of the electric field relative to a single nanoparticle. Note that we refer to longitudinal oscillation only where the E-field is parallel to the long axis of the nanoparticle. The transverse case does not depend on the aspect ratio of the nanoparticle and therefore no tunable effect can be obtained. We note that the device presented in Figure 1 uses an array of nanoparticles in order to increase the accumulated effect.

The structure is continuously illuminated by a uniformly distributed beam at the wavelength’s range of interest. When the pump illumination is absent we have light excitation at a wavelength that corresponds to the geometry of the device [9]. By using a high intensity pump illumination which is absorbed by the silicon nanorod at the bottom, the free charge carrier density is changed and thus the refractive index of the device. When the carrier density inside the silicon nanorod is high enough we essentially obtain a metallic like nanorod attached to the gold nanoparticle which results in a different aspect ratio. Using this mechanism we can control the aspect ratio of the structure by means of external illumination, thus changing the wavelength of the peak power excitation.

The mechanism of change in the refractive index in silicon as a function of the carrier concentration is known as the plasma dispersion effect. This change is given by (1) [10]: (1) Δ n = − e 2 λ 0 2 8 π 2 c 2 ɛ 0 n ( Δ N e m e * + Δ N h m h * )

For the real part of the refractive index, the change in the absorption coefficient equals: (2) Δ α = e 3 λ 0 2 4 π 2 c 3 ɛ 0 n ( Δ N e μ e ( m e * ) 2 + Δ N h μ h ( m h * ) 2 )where e is the electron charge, λ₀ is wavelength of light in free space, c is the speed of light, ɛ₀ is the vacuum permittivity, n is the refractive index, ΔN_e and ΔN_h are the change in the carrier concentration of the electrons and holes, respectively. μ_e and μ_h are the mobility of the electron and holes. m^*_e and m^*_h are the effective electron and hole masses, respectively. The change of the free carrier concentration is: (3) Δ N e = Δ N h = η × P h × νwhere η is the quantum efficiency, P is the intensity of the pump illumination and hυ is the energy of each photon (h is Plank’s constant and υ is the optical frequency of the photon). From equations (1) and (2) we can see that the pump illumination causes the real part of the refractive index to decrease, while increasing the imaginary part of the refractive index, thus obtaining values close to that of a metal.

The Discrete Dipole Approximation (DDA) is a flexible method for computing the scattering and absorption by particles with arbitrary geometry. In this method, the Maxwell’s equations are calculated by dividing the geometry of a single nanoparticle into a target array of N dipoles (j = 1…N) as presented in Figure 2. For each dipole the solution for the oscillating dipole moment can be found and the scattering and absorption cross sections are then calculated.

The electric field at location r_j can be written as the sum of incident electric field and the effect of the other N-1 dipoles: (4) E j = E inc , j − ∑ k ≠ j A jk × P kwhere E_inc,j is the incident electric field, A_jk is an interaction matrix between j and k dipoles and P_k is the dipole moment at the k'th element. A is a N × N interaction matrix whose elements are 3 × 3 tensors which can be defined by: (5) A jk = exp ( iKr jk ) r j k × [ k 2 ( r ^ jk r ^ jk − 1 3 ) + iKr jk − 1 r jk 2 ( 3 r ^ jk r ^ jk − 1 3 ) ] ,   j ≠ k A jj = α − 1where K is the wave vector, r_jk is the distance between j and k dipoles, r̂_jk is a unit vector and α is the polarizability which is geometry dependent and takes into account the dielectric constant of the nanoparticles and the surrounding medium. Using these relations, a system of equations can be solved in order to find the dipole moments: (6) A × P = E inc , j

The extinction and absorption cross sections can be evaluated by: (7) C ext = 4 π K | E 0 | 2 ∑ j = 1 N Im ( E inc , j * × P j )and: (8) C abs = 4 π K | E 0 | 2 ∑ j = 1 N { Im ( P j × ( α − 1 ) * × P j * ) − 2 3 K 3 | P j | 2 }where the scattering cross section is C_sca = C_ext − C_abs and * denotes the complex conjugate. The numerical analysis of the device was carried out using the freely available DDSCAT 7.0 [11] software which uses DDA method to calculate the scattering and absorption of nanoparticles with arbitrary shape and complex refractive index.

The resonance wavelength excitation was tested at the range of 300 nm–700 nm, while the wavelength step was chosen to be 2 nm at the critical area (at the peak region) and 10 nm outside the peak region. The distance between dipoles was chosen to be 1 nm, in order to obtain accurate results while not increasing the computational cost significantly (the inter dipole distance affects the accuracy of the numerical solution while a denser inter dipole matrix will improve the accuracy of the solution, while increasing the computational cost). The polarization of the incident light was parallel to the long axis of the nanoparticle. The propagation direction was as indicated in Figure 1. The cross section of each nanoparticle was 10 nm × 10 nm while a comparison between nanoparticles with lengths of 40 nm and 50 nm (i.e., different aspect ratio) was performed. For the 40 nm nanoparticles we also investigated the behavior of a single elongated Si nanoparticle which has the overall size equal to the original size of the Si and Au nanoparticles when placed one on top the other.

The simulation results can be seen in Figure 3. Figure 3(a) shows the absorption cross section calculated for a 40 nm long structure with different pump illumination intensities. One may see that as the intensity of pump illumination increases the absorption peak and its wavelength shift from 562 nm with a pump of 19.65 mW/μm² towards the blue wavelength of 476 nm with a pump of 37.33 mW/μm². While the 50 nm particle exhibits the same behavior, but the entire image is red shifted, see Figure 3(b). These results can be explained by the fact that the higher the aspect ratio is, the higher the peak wavelength is. Illuminating the silicon with the pump increases its free carrier concentration towards that of the gold nanoparticle to a point where it becomes a single and effectively thicker gold nanoparticle and thus has a smaller aspect ratio.

The results for the Si single nanoparticle, shown in Figure 3(c), indicate that the single silicon particle can have a broader range of peak absorption wavelengths (range of 484 nm to 656 nm in comparison to a range of 476 nm to 562 nm for the case of gold-silicon structure). However, looking at the width of the peaks we can see that silicon has a full width half maximum (FWHM) which is 10% larger than that of the silicon-gold structure. Note, that for high pump illumination the results for a single silicon nanoparticle and a Au-Si structure show the same location for the peak, which indicates that for that intensity silicon becomes as conductive as gold.

The dependence of the peak location on the pump illumination intensity is depicted in Figure 3(d). According to the results, the dependency between the peak wavelength and the pump intensity is linear, thus we used a linear fit to obtain the equation of the peak wavelength λ_max[μm] = −4.777×10⁻³P+649×10⁻³ for the 40 nm long nanoparticle. Since the linear relation between the absorption peak wavelength and the aspect ratio is also known [12], we can design the device to have the exact desirable peak wavelengths.

We examined the absorption of Au nanospheres in water using Shimadzu UV-1650pc spectra-photometer. The results can be seen in Figure 4.

The purpose of this experiment was to show that indeed as previously claimed, changing the geometrical properties or the conductivity of nanoparticles (which can be controlled by external illumination) affect their spectral behavior. The size of the nanospheres was 5, 30 and 80 nm and the absorption spectra resulted peaks were at wavelengths of 517, 526 and 549 nm, respectively.

We can see that by enlarging the nanospheres we can obtain a red shift of the spectra (the colored images as captured by the spectrophotometer are seen in the right upper corner of the figure). We can also see that while the 5 nm particles have no distinct peak, the 80 nm particles have a distinct peak which is wider in its area than that of obtained for the 30 nm particles. The experimental results presented in Figure 4 fit well with the previously presented numerical/mathematical modeling.

The demonstrated concept can be used for the realization of sensing as well as reconfigurable all-optical devices. In case of a sensor application the output of the device is a photonic output which provides photonic readout according to the external intensity of the pump illumination that is being illuminated over the device shown in Figure 1. In addition, a configuration such as the one shown in Figure 1 can be used as a multiplexer where the pump illumination controls the output of the device by selecting the proper input wavelength channel. Furthermore, the device can be used as an optical switch as well as a tunable filter which is polarization dependent in optical communications links. Figure 5 depicts an example of a nanoparticle based all-optical modulator. Here we have a waveguide with a nanoparticle inside an air gap. The particle is illuminated from above with a white light. By designing the waveguide in such a way that it supports only guided modes for wavelengths that are smaller than the zero pump peak wavelength of the particle, we obtain radiative modes when the pump illumination is absent and therefore the optical power is not guided towards the output. When the pump is present we obtain guided modes and then the power is propagated towards the output.

Figure 6 shows a top view Scanning Electron Microscope (SEM) image of the device reported in Figure 1. The fabrication itself included several stages. First we used the Bestec system to evaporate 10 nm of silicon on top of a silica wafer. Right after, 10 nm of Au was evaporated again on top of the silicon layer. The next stage was to etch the surface, using the Focused Ion Beam (FIB) in such a way that an array of nanorods that are made out of two layers (i.e., Au on top of a silicon layer) was generated.