1. Introduction

There is a considerable interest in the study of nanoscale optical antennas, largely due to their ability to produce giant and highly localized electromagnetic fields [1,2]. Important applications include near field microscopy [3], high resolution lithography [4–7], spectroscopy [8,9], novel nanophotonic devices [10–12], optical tweezers for nanoparticle trapping [13] and high order harmonic generation [14], etc. Numerical simulations and experimental measurements show that a high intensity hot spot could be achieved in the gap of the metallic nanoantennas in the visible [1–4], infrared (IR) [11, 12] and THz domains [15]. Such a high-intensity localized spot could be very useful for future on-chip optical applications [16].

The UV region of the electromagnetic spectrum has received increased attention because of numerous civil and military industrial applications [17]. Additionally, there is a great interest in solid-state UV light sources for chemical and biological agent detection and efficient solid-state lighting. The availability of chip-scale UV light sources may also open up new applications in medical research for early disease detection [18]. Consequently, achieving similar high intensity and strongly localized UV spots for future UV on-chip applications is an active area of research.

In recent years, some groups reported measurements or numerical modeling results employing UV light to excite UV localized surface plasmons (LSP) for near-field nanolithography [5, 7, 19, 20]. They presented lithographic results on photo-resist layers as an indirect demonstration of localized UV optical field. The near-field characteristics are intimately linked to the properties of any device in SPP circuits, especially in the subwavelength regime.[21] It is therefore valuable to provide direct access to the near-field of the modes. In order to develop novel nano-structured components that control the distribution of the UV LSP modes, it is very important to directly image the near-field distribution of the hot spot and investigate their optical properties. In a recent publication, we have demonstrated our capability to study the optical properties of UV surface plasmon standing waves on an Al/Al₂O₃ film surface using a UV-compatible near-field scanning optical microscope (NSOM) system [22]. So far, no direct near-field imaging of optical nanoantennas operating in the UV domain has been reported. In this work, we employ our UV-NSOM system to study the optical field distribution on a metallic bowtie nanoantenna structure in the UV domain. We believe that direct mapping of UV LSP modes supported by the nanoantenna structure is an important step towards developing a novel optical scanning probe for high-resolution imaging, spectroscopy and lithography techniques at UV wavelengths.

2. Sample preparation and characterization

Since the experiments are designed for operation in the UV spectral range, aluminum was selected as the material. In this experiment, the Al bowtie nanoantenna was fabricated on a fused silica substrate. A 200nm-thick Al layer was deposited onto the substrate plate by a magnetically controlled sputtering process. The desired structure was then created utilizing a high-precision focused ion-beam milling technique (FEI DB-235). Figure 1(a) shows a scanning electron microscope (SEM) image of one of the nanoantennas that were fabricated. The geometric parameters of the nanoantenna are annotated in Fig. 1(b). The length (L) and width (d) of the antenna structure in Fig. 1(a) was measured to be approximately 200 nm and 185 nm, respectively. The antenna gap is about 50nm.

 

Fig. 1. (a) The SEM image of a bowtie nanoantenna. (b) Illustration of the bow-tie nanoantenna geometry. (c) A sketch of the UV NSOM operating in the collection mode.

Download Full Size | PDF

A modified NSOM was employed to measure the optical field distribution on the surface of the nanostructures [Fig. 1(c)]. The optics of the NSOM system was modified for operation with UV light, permitting operation down to the Deep UV wavelength range. In the measurement, the UV NSOM is configured to operate in the collection mode. A UV laser at a wavelength of 364 nm is employed to illuminate the bowtie nanoantenna through the substrate [Fig. 1(c)]. A linear polarizer was used to control the polarization direction of the incident light. The NSOM probe, which had an aperture size of 50nm, was placed about 10 nm above the sample to pick up the optical signals. This is achieved by first bringing the NSOM tip in contact with the top surface of our bowtie nano-antenna structure, and finding the structures from the surface topography scan. Then we move the tip up to a fixed height of 10 nm. During this process we closely observe the beam deflection. It could be seen that the deflected beam shifts did not change during the measurement, indicating that the tip height was fixed. Under this condition, we observed obvious localized field distribution. The UV optical field distribution in the vicinity of the nanoantenna structure is shown in Fig. 2(a). The data clearly demonstrates that the bowtie antenna produced a hot spot in the gap area between the two triangular antenna arms that is greatly enhanced relative to areas away from the gap. We believe that this strong field enhancement arises from a combination of effects, including the LSP resonance, the sharp metallic tip with thin wedges, and electromagnetic field localization in the antenna gap nano-capacitor [11]. The spatial profile of the LSP mode is expected to be confined to the immediate vicinity of the 50 nm nanoantenna gap. However, the measured optical spot size along the antenna axis [the x direction in Fig. 1 (b)] was larger than the 50 nm gap. This discrepancy is mainly due to the finite-size of the NSOM probe aperture, which is comparable in size to the gap. The observed profile should be a convolution of the actual optical field and the collection function of the NSOM probe. Here we employed three-dimensional (3D) finite-difference time-domain (FDTD) modeling [23] to calculate the expected optical profile. The entire volume of the 3D simulation is treated using a grid size of 5 nm. The excitation source used in our simulation is a 364 nm UV plane wave polarized along the x-axis. The plane wave is launched normal to the bowtie antenna through the substrate. Aluminum and SiO₂ are used as bowtie and substrate materials respectively. The permittivity of Al at this fixed wavelength is approximately -19.459+i3.606 [24]. In this simulation, all the geometric parameters of the nanoantenna are the same as those of the real device shown in Fig. 1(b). Figure 2(b) presents the calculated electric field distribution 10nm above the bowtie antenna in the x-y plane. These results clearly show that the electric field is strongly confined within the antenna gap. The fine structure of the electric field confinement was not totally resolved in NSOM imaging because of the aforementioned finite NSOM probe aperture. To compare our simulation data to the measured data in Fig. 2(a), we performed a convolution procedure on the FDTD simulation as shown in Fig. 2(c). The 2-D convolution was done by replacing each data point from the FDTD simulation with the average value of data points within a numerical circular aperture with a diameter of 50 nm. Since the data presented in Fig. 2(b) is a collection of discrete pixels, this 2-D convolution actually outputs a weighted average of each pixel’s neighborhood [25]. A comparison of the optical field localization along the x direction is shown in Fig. 2(d). It can be clearly seen that after the convolution, the processed image agrees reasonably with the experimental data.

 

Fig. 2. (a) The optical field distribution around the nanoantenna structure measured in the near field. (b) FDTD simulation of the optical field intensity of x-y plane on the surface. (c) Convolved optical intensity distribution from (b). (d) Comparison the intensity cross-section taken from (c) with the experiment data.

Download Full Size | PDF

3. Optimization of the bowtie nanoantenna dimensions

The field enhancement in the antenna gap can be optimized when the antenna is illuminated at the resonant wavelength, which is mainly determined by the geometric shape. This behavior was discussed by H. Fischer et. al. [26] using Green’s tensor technique in the visible spectral range. In this report we performed FDTD simulations to find the resonance condition for bowtie nanoantennas operating at UV wavelengths. We simulated a set of bowtie antennas with a fixed gap width but various lengths L and tip angles θ. In these simulations, the antenna thickness t and the antenna gap g are fixed at 200 nm and 50 nm, respectively. The electric field amplitude 10 nm away from the bowtie antenna gap is recorded and normalized to the input electric field. As shown in Fig. 3(a), there exist several peaks of localized optical field amplitude or various values of length, L. We also plotted the length L associated with each resonance peak as a function of the order of resonance [Fig. 3 (b)]. Using a linear least square fit, the resonance periodicity was found to be 205±5 nm regardless of the value of the tip angles, indicating that the length L predominantly determines the resonances of the nanoantennas. Keeping this in mind, we simplified our experiments and only varied the length of the antenna. The results are discussed in the following.

 

Fig. 3. (a) Electric field amplitude vs. length L for the bowtie antenna different a function of tip angle. The field is normalized to the amplitude of incident electric field. (b) Resonance peak position as a function of the order of resonance. Dashed lines are linear least square fit of the data with different tip angle.

Download Full Size | PDF

In order to experimentally examine this resonance behavior, we fabricated a group of bowtie antennas with different arm lengths that are positioned next to each other so that they can be imaged under identical illumination conditions [see Fig. 4 (a)]. The arm length L from the left to the right ranges from 300 nm to 750 nm with an increment of 50 nm. The NSOM imaging results are shown in Fig. 4(b). The experimental configuration is the same as the one used for measurement in Fig. 2(a). Since these antennae are imaged in a single scan, the optical field intensities around each of the elements are comparable. One can see a clear variation of the LSP intensities. Stronger intensities can be observed in samples 1, 6, 7 and 10. To quantitatively compare with the simulation, we determined the total intensity transmitted through each antenna by performing a 2D integration centered at the LSP point. The comparison shown in Fig. 4(c) demonstrates that the measured resonance behavior agrees well with simulations. The predicted 2nd, 3rd and 4th order resonances supported by the nanoantenna structures with various arm lengths are observed in this measurement.

 

Fig. 4. (a) SEM image of a bowtie antenna array. (b) NSOM image of the bowtie antenna array in (a). (c) Comparison of experimental data with computed field amplitude.

Download Full Size | PDF

4. Summary

Using a UV-NSOM operating in the collection mode, we successfully performed a direct measurement of the near-field intensity profile of UV LSP near the gap of bow-tie nanoantennas. Highly confined UV hot spots are observed in the near-field of these nanostructures. FDTD simulations show good agreement with the experiments as long as the limited NSOM resolution is taken into account. The phenomena of pronounced resonances in the LSP intensity as a function of bowtie geometry (e.g.: arm length) is discussed in the context of optimizing the nanoantennas at UV wavelengths. We find good agreement between simulation and experiment. We believe that direct mapping of the UV light enhancement from a metallic bowtie nanoantenna has potential impact on novel photonic applications in the UV domain, such as photolithography, single-molecule imaging, tip-enhanced Raman spectroscopy, and nanoparticle trapping. In addition, direct measurement of the optical field near the nanoantenna contributes to a better understanding of the underlying physics of these nanoscale optical antennas, which is crucial to the studies of sub-wavelength optics on a chip.