Third harmonic upconversion enhancement from a single. semiconductor nanoparticle coupled to a plasmonic antenna

Transcription

1 Third harmonic upconversion enhancement from a single semiconductor nanoparticle coupled to a plasmonic antenna Heykel Aouani, Mohsen Rahmani, Miguel Navarro-Cía and Stefan A. Maier This document includes the following supplementary information: 1- Fabrication of the ITO nanoparticles decorated with plasmonic dimers 2- Characterization of structures with Fourier transform infrared spectroscopy 3- Experimental setup and measurements 4- THG measurements from hybrid plasmonic antennas 5- Use of third harmonic signal for probing plasmonic hot spots 6- Effective third order susceptibilities and conversion efficiencies of the nonlinear upconversion nanosystems 7- Current factors limiting the nonlinear upconversion efficiency rates 8- Additional linear and nonlinear simulations 1

2 1- Fabrication of the ITO nanoparticles decorated with plasmonic dimers The nonlinear upconversion systems investigated in this work were fabricated via combination of etch-down and lift-off approaches accompanied with precise alignments. Unlike conventional hybrid structures which can be fabricated with dual lift-off methods, the fabrication procedure developed in this work allows employing non-directional deposition methods such as sputtering for fabricating objects with characteristic dimensions down to 25 nm. Indeed, a pure lift-off approach cannot be employed here as the non-directional plasma blocks define nanoscale windows on the resist before the material reaches the substrate. These limitations hinder sputtered nanomaterials to be fabricated via lift-off, which prevent the use of many promising materials in nanophotonics. Here, we tackle this issue by combining etchdown and lift-off approaches. In a first step, a quartz substrate was covered with a 40 nm ITO film using sputtering. Subsequently, high resolution ITO nanodots and alignment markers were defined in negative resist (HSQ) by electron-beam lithography. Ion beam etching (Ar ions) of the ITO layer was then performed to generate the ITO nanoparticles. Next, the substrate was coated with PMMA resist and the antennas shapes were defined in PMMA around the ITO nanodots with a nanometric alignment precision. Hereafter, the sample was covered with a 2 nm Cr adhesion layer and a 40 nm Au film by thermal evaporation. A final lift-off step enables to obtain the nonlinear upconversion nanosystem made by a single 25 nm ITO dot at the gap of a gold plasmonic dimer. We emphasize that this procedure leads to a precise control of the distance between the ITO nanoparticle and the gold structure. A schematic description of the fabrication process is presented in Supplementary Fig. 1. Let us point out that ITO has been chosen for our nonlinear investigations given its high third harmonic susceptibility. 2

3 Supplementary Fig. 2: Schematic representation of the fabrication process developed to manufacture the nonlinear upconversion system made by a single ITO nanoparticle decorated with a plasmonic gold dimer. 2- Characterization of structures with Fourier transform infrared spectroscopy The extinction spectra of the fabricated structures (defined as 1 transmission) were measured by Fourier transform infrared spectroscopy (FTIR, Bruker Hyperion 2000) through an array of plasmonic antennas (approximately 120 structures, with a pitch of 2 μm) with and without an ITO nanoparticle at their gap at normal incidence under linear polarization. The 2 μm pitch ensured insignificant coupling between neighbouring nanostructures, as confirmed by numerical simulations. For measurements in the near-infrared regime, we used an InGaAs detector cooled with liquid nitrogen. As the extinction spectra presented in Fig. 1f were measured for structures under parallel polarized excitation, i.e., electric field parallel to the dimer axis, the extinction spectra for a 35 nm gap nanorod dimer with / without an ITO nanoparticle under perpendicular polarized excitation are presented in Supplementary Fig. 2. As expected, the extinction spectra do not exhibit plasmonic resonances in this spectral window when the structures are excited under perpendicular polarization. 3

4 Supplementary Fig. 2: Experimental extinction cross section of the 35 nm gap nanorod dimers with (red dots) and without (black dots) a 25 nm ITO nanoparticle at their gaps under perpendicular polarized excitation. 3- Experimental setup and measurements The inverted microscope developed for this work uses a Yb:KGW femtosecond PHAROS laser system as a pump of a collinear optical parametric amplifier ORPHEUS with a LYRA wavelength extension option (Light Conversion Ltd, Lithuania, pulse duration 140 fs, repetition rate 100 khz). The excitation beam is reflected by a shortpass dichroic mirror (Thorlbas DMSP805 and DMSP1000) and focused on the sample plane by a dry microscope objective (Nikon S Plan Fluor x40, 0.6 NA). For nonlinear experiments, the fundamental incident wavelength is set at 1500 nm, and the backward-emitted third harmonic generation at 500 nm is collected via the same objective. The third harmonic signal is then directed to a 70:30 cube beamsplitter that separates the beam towards a spectrograph (PI Acton SP2300 by Princeton Instruments) for spectral measurements and towards an avalanche photodiode (MPD PDM Series by Picoquant) for nonlinear imaging. Accurate positioning of the sample at the laser focus spot is ensured by a multi-axis piezoelectric stage (Nano-Drive, Mad City Labs). For all experimental measurements, the excitation power was set below 50 μw (peak 4

5 intensities of 45.7 GW/cm 2 ) in order to prevent sample damage. A schematic representation of the nonlinear microscope developed in this work is presented in Supplementary Fig. 3. For measurements on ITO nanoparticles, the third harmonic generation spectra at λ= 500 nm were obtained by integrating the nonlinear signal for 1800 s with an average incident power set at 50 μw thought an array of nanoparticles (pitch of 200 nm). As the signal to noise ratio is dramatically improved in presence of a plasmonic dimer, the third harmonic generation spectra of the nonlinear upconversion nanosystems were integrated for 5 s with an average excitation power set at 50 μw. All the experimental spectra presented in the main text have been normalized by the integration time and by the area of the ITO nanoparticle in order to present a third harmonic intensity expressed in photon.s -1.μm 2. Supplementary Fig. 3: Schematic description of the nonlinear microscope used for our investigations. 5

6 4- THG measurements from hybrid plasmonic antennas The third harmonic generation signal from the plasmonic dimers without ITO nanoparticle at their gap has been carefully evaluated in order to quantify the intrinsic nonlinear gold background. Typical raw spectra determined for a 35 nm gap nanorod dimer with and without ITO nanoparticle at its gap and excited under parallel and perpendicular polarizations are presented in Supplementary Fig. 4 (average excitation power of 50 μw). As can be seen, a strong polarization dependence of the THG signal is highlighted in both cases. Under parallel polarization, the THG response of the ITO nanoparticle + plasmonic dimer combined system is more than 16 times higher than the intrinsic THG background of the plasmonic dimer itself. Under perpendicular polarization, the THG response of the ITO nanoparticle + plasmonic dimer is similar (the nonlinear signal from the ITO nanoparticle is not enhanced under perpendicular polarization), thus demonstrating that the THG signal from the ITO nanoparticle + plasmonic dimer under parallel polarized excitation mainly comes from the ITO nanoparticle. This is further corroborated with the nonlinear simulations in Supplementary Section S9. For each nonlinear upconversion system investigated, we have taken into account the intrinsic nonlinear third harmonic from the gold dimer without ITO nanoparticle at its gap by considering it as a background, which was subtracted from the spectra of hybrid upconversion nanosystem responses presented in the main body of the paper. Taking into account the dimensions of the 25 nm ITO particle and the dimensions of the 35 nm gap nanorod dimer (2 nm 100 nm 280 nm), and using the data presented in Supplementary Fig. 4, we determined a third harmonic generation upconversion efficiency 3 orders of magnitude higher for the ITO nanoparticle coupled into the dimer s gap compared to the dimer itself (third harmonic upconversion of for the ITO nanoparticle at the gap of the nanorod dimer and for the nanorod dimer itself). 6

7 Supplementary Fig. 4: a, Raw spectra of the third harmonic generation from a 35 nm gap nanorod gold dimer excited under parallel polarization with (red) and without (black) an ITO nanoparticle at its gap. b, Raw spectra of the third harmonic generation from a 35 nm gap nanorod gold dimer excited under perpendicular polarization with (red) and without (black) an ITO nanoparticle at its gap. As the introduction of a dielectric nanoparticle at the gap of a metallic nanodimer shifts its plasmonic resonance, a special attention has been devoted to characterize the third harmonic responses from plasmonic dimers with and without an ITO nanoparticle at their gaps in a much larger spectral window. In order to avoid any ambiguity regarding the origin of the nonlinear signal, the fundamental excitation wavelength of the incident pulsed laser was tuned from 1480 to 1630 nm, which correspond to the resonance peak in the extinction spectra of the 35 nm gap nanorod dimer without and with ITO respectively, see Fig. 1f. The evolution of the third harmonic generation intensity from the structures under investigation is presented in Supplementary Fig. 5 for various fundamental excitation wavelengths (average excitation power set at 20 μw). Please note that these experimental data have been processed by taking into account the transmission / reflection coefficients of the different optical elements and the quantum efficiency of the CCD detector. For each of these multi-wavelength measurements, the THG intensity from the plasmonic dimer with an ITO nanoparticle at its gap is 1 order of magnitude higher than the THG intensity from the plasmonic dimer without ITO, thus confirming that the nonlinear signal originates from the ITO nanoparticle, and is not due to a shift of the plasmonic resonance position. 7

8 Supplementary Fig. 5: a, Evolution of the third harmonic generation from a 35 nm gap nanorod gold dimer excited under parallel polarization without an ITO nanoparticle at its gap for various fundamental wavelengths. b, Evolution of the third harmonic generation from a 35 nm gap nanorod gold dimer excited under parallel polarization with an ITO nanoparticle at its gap for various fundamental wavelengths. Lastly, we have investigated the case when the relative position ITO nanoparticle is shifted inside the gap of the 35 nm gap nanorod dimer. The SEM image of such configuration and its corresponding third harmonic generation spectrum are presented in Supplementary Fig. 6. As can be seen, the THG signal decreases when the ITO nanoparticle is shifted from the center to the bottom edge of the gap. From these data, we computed an intensity enhancement inside the ITO nanoparticle of about 76, in good agreement with the value of about 82 determined by numerical simulations (see Section S8, Supplementary Fig. 9d for more details about simulations). Supplementary Fig. 6: Evolution of the third harmonic generation from an ITO nanoparticle coupled to a 35 nm gap nanorod gold dimer excited under parallel polarization when the relative position of the ITO nanoparticle is shifted inside the gap. 8

9 5- Use of third harmonic signal for probing plasmonic hot spots As mentioned in the main body of this paper, indium tin oxide was chosen as the nonlinear material for our investigations. Under excitation at frequency ω, its third order susceptibility χ induces a nonlinear polarization: P = ε 0 χ E ω E ω E ω, (Eq. 1) which in turn generates a radiation of intensity I α P 2. The third order polarization P H of an ITO nanoparticle localized at the gap of a plasmonic dimer can be expressed as: P H = ε 0 χ E ω H E ω H E ω H, (Eq. 2) ω where E H is the field inside the ITO nanoparticle at the gap of the plasmonic dimer investigated. By introducing the excitation intensity enhancement factor η at the gap of plasmonic dimers defined by η= (E ω H / E ω ) 2, Eq. 2 can be rewritten: P H = ε 0 χ η 3/2 E ω E ω E ω = η 3/2 P, (Eq.3) which implies: η= (I H / I ) 1/3, (Eq.4) where I and I H are respectively the third harmonic intensity from the Ø 25 nm ITO nanoparticle at the gap of a plasmonic dimer or isolated. The Eq. 4 and the experimental data presented in the Fig. 2a and Fig. 3c,e enable the immediate probing of the hot spot intensity inside the ITO nanoparticle at the gap of the plasmonic dimers investigated and presented in Fig. 3d,f. 6- Effective third order susceptibilities and conversion efficiencies of the nonlinear upconversion nanosystems In order to investigate the nonlinear performances of the nonlinear upconversion systems presented in this work, we started by determining the third harmonic conversion efficiency 9

10 defined as the ratio of third harmonic intensity I relative to the fundamental average incident power. The experimental values determined for I are summarized in Supplementary Table 1. Supplementary Table 1: Experimental third harmonic radiated intensity I by the various nonlinear upconversion nanosystems investigated. Using Eq. 3, we can introduce an effective third order susceptibility χ Η for the nonlinear upconversion nanosystems defined by χ Η = (χ ) 3/2. In order to investigate the potential of the hybrid systems as ultrabright nanosources of upconverted third harmonic light, we plotted in Supplementary Fig. 7 the THG conversion efficiency and effective susceptibility for a 25 nm ITO nanoparticle decorated with the various plasmonic dimers. Upconversion efficiencies between 0.4 and % were experimentally achieved, which is 1000 folds greater than the highest values reported for intrinsic second order processes from plasmonic antennas 1. Third order effective susceptibilities ranging from 249 to 3543 nm 2 /V 2 were here computed for the ITO-plasmonic hybrid systems, much larger than the 0.2 nm 2 /V 2 recently reached with gold nanostructures alone 2. Altogether, the high third harmonic conversion efficiency and effective susceptibility of ITO-plasmonic hybrid systems make them suitable candidates for nanoscale nonlinear upconversion of light. 10

11 Supplementary Fig. 7: Third order upconversion efficiency (solid bars) and effective susceptibility (grid bars) corresponding to an ITO nanoparticle decorated with the various plasmonic motifs under investigation. 7- Current factors limiting the nonlinear upconversion efficiency rates Although, in theory, the nonlinear upconversion efficiency from the hybrid nanosystem can be increased until converting all the fundamental red photons (λ= 1500 nm) into third harmonic green photons (λ= 500 nm), the damage thresold prevents achieving upconversion efficiencies greater than %. As we can see in Supplementary Fig. 8, an average excitation power of 100 μw modifies the shape of the plasmonic dimer and destroys the ITO nanoparticle at its gap. To overcome these limitations, and as a future work, we are planning to embed our samples in a low refractive index superstrate in order to make the nonlinear upconversion nanosystems more resistive to high incident powers. 11

12 Supplementary Fig. 8: SEM picture of a 25 nm ITO particle at the gap of a nanocylinder dimer after excitation under low (left-side image, after average excitation power of 50 μw) or high laser power (right-side image, after average excitation power of 100 μw). Scale bar, 200 nm. In the right-side SEM image, the ITO nanoparticle is destroyed and one of the dimer s arms is rotated because of the heat induced in the Cr adhesion layer between the gold film and the quartz substrate. 8- Additional linear and nonlinear simulations As can be seen in Supplementary Fig. 9, the nanorod dimer without anything at the gap has its maximum intensity at the interfaces metal-gap air. Because of the small gap, the peaks at both sides of the gap are strongly coupled leading to a plateau of ~255 intensity enhancement within the gap for the x-cut denoted by the white dotted line. However, when the ITO is placed at the gap, the field distribution is changed in the surrounding of the ITO nanoparticle and additional intensity peaks emerge at the interface ITO-gap air. Now, within the ITO nanoparticle, there is an almost flat intensity enhancement of ~110. If the ITO nanoparticle is shifted to the bottom edge of the gap, there is a significant reduction of the intensity enhancement within the plateau generated by the ITO nanoparticle. In this case, the intensity enhancement drops up to ~70. The perturbation on the field distribution, and thus, the change on the intensity enhancement at the gap, depends on the geometry and optical dielectric function of the probing element. Notice that quantum dots have high refractive index, and thus, show even higher invasiveness 3 than the here proposed ITO-based technique. Probing elements with low refractive index would be necessary to pave the way for quantification of the hot spot on gap-free plasmonic dimers. However, the technique proposed in the main manuscript is highly relevant for plasmonic applications, i.e. to probe the excitation intensity 12

13 enhancement surrounding / inside a single emitter when coupled to a plasmonic antenna. Indeed, the parameter of interest for plasmonic applications (sensing, photovoltaic to cite few of them) is the intensity enhancement in presence of a localized nanoemitter (molecule, quantum dot, nanoparticle...) and not the enhancement provided by an isolated plasmonic antenna. The conclusions of the manuscript are built on the assumption that any effect occurring at the third harmonic (e.g. losses) has a negligible influence on the overall non-linear detection, hence Eq. 4. To sustain the assumption, 3D FDTD nonlinear simulations were performed using FDTD Solution v8.6. The same simulation setup as in the linear analysis (described in Methods) is used unless anything else is stated next. Supplementary Fig. 9: a, Intensity enhancement along the white dotted lines shown in the twodimensional colour maps of panel b, c and d. b and c, FDTD intensity enhancement maps at the middle cross-sectional plane of the hybrid and nanodipole-only configuration, respectively. d, FDTD intensity enhancement maps at the middle cross-sectional plane of the hybrid configuration when the ITO nanoparticle is shifted to the bottom edge of the gap. 13

14 Given the complexity of the analysis, the computation effort is reduced by considering: (i) the nanodipole as two ideal rectangles, and (ii) the perfect ITO cylinder to be placed at the centre of the gap. This allows us to use a twofold symmetry. See Supplementary Fig. 30 for the sketch of the 35 nm gap dimer considered along with its dimensions. The optical dielectric functions of Au, Cr, SiO 2 are taken from tabulated data 4 as in the linear analysis of the main body of this manuscript. However, the dielectric dispersion of the materials is fitted in the spectrum range from 400 nm to 1600 nm by a 6-coefficient model, allowing a tolerance of 0.1 and enforcing passivity. In addition, the nonlinear response of Au is considered with its intrinsic χ Au = m 2 /V 2. Meanwhile, ITO is characterized with a constant ε r = 2.89 and χ ITO = m 2 /V 2 according to the literature 5. We investigate three limiting cases: when either χ Au or χ ITO is set to zero along with the situation where χ Au and χ ITO are considered simultaneously. The plane-wave excitation has an amplitude of V/m, which has been chosen to fulfil the condition χ E(t) 2 << ε r, where E(t) is the temporal induced electric field within the ITO particle. This excitation is a realistic narrowband temporal pulse with central wavelength of 1500 nm and spectral width ~23.65 nm (i.e. pulse length of 140 fs). 35 nm 280 nm 100 nm ITO: Ø20 nm 40 nm 45 nm Supplementary Fig. 40: Sketch of the nano-dipole and nano-discs along with the geometrical parameters. The Cr layer shown in grey between the Au dimer and the SiO 2 substrate has a thickness of 2 nm. 14

15 The TFSF volume default grid is 7 nm 7 nm 3 nm. For the gap region, the volume with dimensions 120 nm 140 nm 50 nm was discretized with a cubic grid of 2.5 nm 2.5 nm 1.5 nm. An even smaller cubic grid of 0.3 nm 0.3 nm 1 nm is overridden in the region enclosing the ITO cylinder. The maximum simulation time is set to 1700 fs. The time stepping stability factor was set to 0.95, which corresponds to a time step of δt = fs. I H and I are calculated as the total power flowing outward through a volume enclosing the TFSF source when the Ø20 nm ITO nanoparticle is at the gap of a plasmonic dimer or isolated, respectively. The results of this nonlinear analysis are shown in in Supplementary Table 2, 3 and 4. As it can be seen in Supplementary Fig. 9, high intensity is displayed at the circular interface between the ITO and air. Given the unavoidable numerical error induced by the hexahedral mesh failing to map perfectly circular geometries, we provide the average intensity computed from the stepwise approximated ITO cylinder and from the ITO-enclosing cuboid. These two cases define the worst and best case scenarios, respectively, i.e. it is expected to have ( I / I ) 1/3 between these two cases if our assumption is correct. Indeed, ( I / I H H ) 1/3 for the case with χ Au and χ ITO falls within this range, which supports the assumption (that the effects at 500 nm can be neglected) and analysis carried out in the main body of this manuscript. Arguably, this is expected since 500 nm is below the localized surface plasmon resonance in Au metallic nanoparticles. Therefore, the nanoantenna is not activated, and thus, its interaction with the THG is reduced. When the nanoantenna operates at its resonance, its effect cannot be ignored. It is also relevant to point out that these nonlinear simulations confirm that the main contributor to the THG is the ITO since the results for χ Au & χ ITO (Supplementary Table 2) are comparable to the case with χ Au set to zero (Supplementary Table 3), whereas the case with χ ITO set to zero has a three orders of magnitude lower THG (Supplementary Table 4). 15

16 Nano-dipole (χ Au & χ ITO ) Average intensity within Simulated structure the cuboid the ITO cylinder enclosing the ITO Scattered power at at λ = 1500 nm cylinder λ = 500 nm [a.u.] [a.u.] at λ = 1500 nm [a.u.] ITO cylinder Dimer + ITO cylinder Intensity enhancement η = η = / 3 I H = I Supplementary Table 2: Numerical results determined from nonlinear analysis (χ Au & χ ITO ). Nano-dipole (χ Au = 0 & χ ITO ) Average intensity within Simulated structure the cuboid the ITO cylinder enclosing the ITO Scattered power at at λ = 1500 nm cylinder λ = 500 nm [a.u.] [a.u.] at λ = 1500 nm [a.u.] ITO cylinder Dimer + ITO cylinder Intensity enhancement η = η = / 3 I H = I Supplementary Table 3: Numerical results determined from nonlinear analysis (χ Au = 0 & χ ITO ). 16

17 Nano-dipole (χ Au & χ ITO = 0) Average intensity within the cuboid Simulated structure the ITO cylinder enclosing the ITO at λ = 1500 nm cylinder [a.u.] at λ = 1500 nm [a.u.] ITO cylinder Dimer + ITO cylinder Intensity enhancement Scattered power at λ = 500 nm [a.u.] η = η = Supplementary Table 4: Numerical results determined from nonlinear analysis (χ Au & χ ITO = 0). References: 1. Aouani, H. et al. Multiresonant broadband optical antennas as efficient tunable nanosources of second harmonic light. Nano Lett. 12, (2012). 2. Renger, J., Quidant, R., Van Hulst, N. & Novotny, L. Surface-enhanced nonlinear four-wave mixing. Phys. Rev. Lett. 104, (2010). 3. Bermúdez Ureña, E. et al. Excitation enhancement of a quantum dot coupled to a plasmonic antenna. Adv. Mater. 24, OP314 OP320 (2012). 4. Palik, E.D. Handbook of optical constants of solids (Academic Press, 1985). 5. Humphrey, J.L. & Kuciauskas, D. Optical susceptibilities of supported indium tin oxide thin films. J. Appl. Phys. 100, (2006). 17