Aluminum for nonlinear plasmonics: Methods Section Marta Castro-Lopez, Daan Brinks, Riccardo Sapienza, and Niek F. van Hulst, ICFO - Institut de Ciencies Fotoniques, and ICREA - Institució Catalana de Recerca i Estudis Avançats E-mail: Niek.vanHulst@ICFO.es Fabrication All antennas presented in this study were nanorod metallic structures made by e-beam lithography using PMMA as positive photo-resist and depositing the metals by thermal evaporation. The height of these nanorods was fixed at 40 nm for all the experiments described in this paper. In the first experiment the width was kept constant at 40 nm while the length was varied from 50 nm to 540 nm in steps of 10 nm. Rods with increasing length were arranged on a single line with a separation of 2 µm between consecutive rods. Four copies of this line were placed on the same matrix to check the reproducibility of the lithography process. Scanning electron microscope (SEM) images were taken to check surface and shape features of the rods. In following experiments the length and the width of the rods were both varied in steps of 10 nm from 50 nm to 300 nm. The rods were arranged in a matrix where the length increased in To whom correspondence should be addressed ICFO - Institut de Ciencies Fotoniques, Mediterranean Technology Park, 08860 Castelldefels (Barcelona), Spain ICREA - Institució Catalana de Recerca i Estudis Avançats, 08010 Barcelona, Spain 1
Figure 1: SEM and 3D representation of the matrix used for the 2D polarization map and polarizability comparison Matrix of nanorods where the length increased in the horizontal axis in steps of 10 nm while the width increased in the vertical axis also in steps of 10 nm (from 50 nm to 300 nm). Note that the diagonal of the matrix consist of squares with increasing side length. the horizontal axis while the width increased in the vertical axis (Figure 1). Using this type of arrangement, the diagonal of this matrix consisted of squares with increasing side length also in 2
Figure 2: Simulations of field enhancement at 600 nm wavelength. Integrated E 4 around antennas of increasing length from 50 nm to 150 nm. The resonance lengths for Au, Ag and Al are around 65 nm, 85 nm and 135 nm respectively. steps of 10 nm. Setup The experiment was done by scanning the sample through the laser focus using an X-Y piezo scanning unit. At each step the light generated at the sample is collected in a confocal configuration, therefore creating images of 1024 by 1024 pixels. The pixels are spaced 58 nm in x and y. An intensity image is created by binning all detected photons per pixel for 1 ms. The images are colorcoded for polarization, with red being emission polarized along the antenna axis and the excitation polarization, and green being emission polarized perpendicular to the antenna axis. Two photon absorption in the antennas was induced by illumination with broadband laser pulses (center wavelength 780 nm, bandwidth 120 nm), compressed to 15 fs with a pulse shaper based on a spatial light modulator. 1 A 200 nm band of the resulting luminescence from the metals (450-650 nm) was collected. The signals were separated from the excitation light by suitable dichroic mirrors and band pass filters and detected on a pair of polarization split APDs, allowing us to determine the 3
linear polarization components of the emitted light. Using a 1.3. NA oil immersion objective, for both illumination and collection, we achieved the necessary peak powers for detectable luminescence while keeping the CW fluency between 1 and 3 GW/cm 2 (CW power 30µW, repetition rate 85 MHz, excitation spot size 400 nm). The polarization of the excitation beam was chosen linear and parallel to the antenna axis for all experiments described in this paper; for the experiment in figure 1, the polarization was horizontal. The Degree of Linear Polarization (DoLP) defined as (I I )/(I +I ) was measured splitting parallel and perpendicular polarization component of the luminescence via a polarization beam splitter and measuring both components using two equal Avalanche Photo Diodes (APDs). Simulations Finite Difference Time Domain (FDTD) simulations were performed on a system composed of a glass substrate, a thin film (10 nm thickness) of Indium Tin Oxide (ITO) and a nanorod with rounded corners which is excited under the same conditions as in the experiment. The nanorod dimensions were set to 40 nm in height, 40 nm in width and various lengths. We used the Johnson and Christy 2 dielectric constants for characterizing gold and silver and the CRC (Handbook of Chemistry and Physics) 3 constants for aluminum. This simulation was repeated for each metal and for all the rod lengths. For each rod length we interpolate the corresponding energy distribution into a uniform cell grid and calculate the intensity square E 4 (which is proportional to the two photon absorption probability) for each cell. Integrating this intensity square over a small volume around the rod (30 nm bigger than the rod on each direction) we were able to plot the excitation intensity as a function of rod length. As the surface or volume nature of TPPL is still not clear, we performed an alternative analysis of the simulations integrating E 4 only along the surface of the rod (from 2 nm outside to 3 nm inside the rod) and not through the whole volume. Doing so, the difference between metals decreases by half its value due to the smaller skin depth and penetration into the surrounding medium of aluminum at the detection wavelengths. 4 Still, there 4
Figure 3: Emission spectrum of gold and aluminum nanorods with lengths around their respective resonance points, excited at 796 nm wavelength. (a) Gold nanorods of length 130 nm and width 100 nm (red line) and length 140 nm and width 90 nm (black dotted line); (b) Aluminum nanorods of length 160 nm and width 110 nm (red line) and length 170 nm and width 130 nm (black dotted line). The intensity of the TPPL signal is similar for both metals while the SHG signal is much higher in the case of aluminum. was two orders of magnitude difference between the simulated E 4 of aluminum and silver and more than one order ( 25 times) between aluminum and gold. We also performed the same simulations fixing the excitation wavelength at 600 nm which corresponds to the maximum of the luminescence emission for the three metals. Figure 2 shows the resulting E 4 integrated along the rod volume as a function of the length of the rod. The resonance length at the luminescence wavelengths sits around 65 nm for gold, 85 nm for silver and 135 nm for aluminum. 2nd and 3rd order Susceptibilities In previous studies of nonlinearities in metals the measured second order nonlinear optical coefficient χ (2) for aluminum thin films was found to be one order of magnitude higher than that for silver and gold (1 10 16 m/v vs 3.2 10 17 m/v and 3.2 10 17 m/v respectively 5 ). However, 5
as discussed in the text, it is very difficult to extrapolate this results to nanoparticles. 6 In Figure 3 we compared representative spectra from resonant aluminum and gold antennas. For this experiment we used a more narrow band laser in order to separate the SHG peak and the TPPL better spectrally (Coherent Mira900 laser, central wavelength 796 nm, bandwidth 6 nm). We compare the spectrum of gold and aluminum rods of different widths around the resonance length of each metal. The SHG peak from aluminum rods is between 2 and 10 times higher than that of gold rods, while the TPPL has nearly the same intensity for the same excitation conditions. In combination with the expected 50 times lower field around the aluminum particles, this tentatively puts the χ (2) of aluminum particles about 100 to 500 times higher than that of gold. Assuming similar quantum efficiencies (η) for TPPL in both materials, the χ (3) would tentatively about 70 times higher. References [1] R. Hildner, D. Brinks, F. D. Stefani and N. F. van Hulst, Chem. Phys. Phys. Chem. 2010, 10.1039. [2] P. B. Johnson and R. W. Christy, Phys. Rev. B (1972), 6, 4370-4379. [3] D. R. Lide, "Handbook of Chemestry and Physics", National Institute of Standards and Technology. [4] F. J. Garcia de Abajo, Rev. Mod. Phys. 2007, 79, 4. [5] D. Krause, C. W. Teplin and C. T. Rogers, J. App. Phys. 2004, 96, 7. [6] M. Zdanowicz, S. Kujala, H. Husu and M. Kauranen, New J. of Phys. 2011, 13, 023025. 6