1. Introduction

Optical antennas are nanostructures that convert propagating radiation into strongly localized and resonantly enhanced near-fields [1–3]. The enhanced near-fields of an optical antenna can be used for enhancing nonlinear effects (e.g. two-photon excited photoluminescence (TPPL) [1, 4]), for enhancing spectroscopy on molecular species using enhanced fluorescence [5,6], surface enhanced Raman scattering (SERS) [7] or surface enhanced IR absorption (SEIRA) [8, 9]. Furthermore, optical antennas are used as highly efficient probes for high-resolution near-field microscopy [5, 10–14]. Among the various shapes and configurations of optical antennas [15], the thin wire antenna or λ/2-dipole antenna is one of simplest and most abundantly used concepts.

Optical antennas are commonly fabricated by means of electron-beam lithography in planar arrays on a substrate and characterized subsequently with far-field [9] and near-field methods [16, 17]. For investigating single antennas (e.g. for SEIRA experiments [8]), metallic nanowires of a well-controlled diameter and length can be obtained by chemical growth of metallic nanowires in an ion-track etched membrane [18] and subsequent dissolving of the membrane. These antennas are then characterized when they are lying on a substrate. For the fabrication of standing antennas with a high aspect-ratio, the use of porous anodized alumina membranes as templates for a metallization can result in large arrays of nanowires. For sensing applications in liquid environment, standing antennas are beneficial over lying antennas because they allow the analyte to flow freely between the antennas, and the “hot spots” which exhibit highest high field enhancement are directly accessible for the molecule to be probed. Such arrays of standing metallic antennas have been employed as metamaterials for sensing [19–21] and for super-resolution imaging [22]. However, with the aforementioned fabrication scheme, the spatial arrangement of each individual upright thin wire antenna can only be poorly controlled. Here, we present a simple and well-controlled laser-based fabrication approach that leads to large arrays of upright standing gold nanoantennas with a high aspect ratio and resonances in the mid-infrared spectral range.

2. Fabrication of standing nanoantennas

In this study, the controlled fabrication of arrays of identical, upright standing gold nanoantennas (Fig. 1) is carried out by irradiating thin gold films with single pulses at a central wavelength of 800 nm of a Ti:Sapphire fs-laser system (Coherent, Libra). As every antenna is generated by a single pulse of laser radiation the pulse to pulse stability of less than 0.2 percent is important to ensure an energetically identical initialization of the formation dynamics for every single antenna. In the following, three different microscope objectives with numerical apertures (NA) of 0.4 (Olympus, MSPlan 20x), 0.25 (Olympus, MPlan N 10x) and 0.15 (Zeiss, Epiplan Neofluar 5x) are applied to focus the Gaussian beam profile of the single pulse laser radiation onto the surface of the gold thin film resulting in focus diameters from 4 µm to 1.5 µm. Under these focusing conditions spiky gold nanoantennas (Fig. 1, inset) with aspect ratios greater than 1:10 are generated at pulse energies (E_p) of approximately 57 nJ and 12 nJ. The presented nanoantennas are generated at pulse durations of 500 fs, whereas it has been shown previously that for pulse durations up to eight picoseconds no significant differences in the generated structures are observable [23]. Considering the underlying ultrafast melting process [23–26] a negligible influence of the pulse duration on the formation process is reasonable as long as the pulse duration is comparably small to the electron-phonon relaxation time in gold of approximately ten picoseconds [25]. Beside the parameters of the laser radiation like the pulse energy or the focusing conditions, the formation dynamics is very sensitive to imperfections and slight changes of the gold filmthickness. Therefore, the applied sample system consists of a 50 nm thick gold film deposited on a fused silica substrate by means of magnetron sputtering. The deposited gold films exhibit a root mean square roughness of less than 2.5 nm. Omitting an additional adhesive layer decreases the interaction forces between substrate and gold thin film, which is essential for the formation dynamics. A linear x,y-piezostage moves the sample relative to the pulsed laser radiation, separating adjacent antennas. Hence, the distance between two antennas within a single row of the meander-shaped scan strategy is determined by the feed rate of the piezostage of 2 mm/s and the repetition rate of 80 Hz of the pulsed laser radiation. To suppress the influence of adjacent nanoantennas during the formation process the distance of neighboring structures should be greater than twice the focus diameter. As long as this distance is taken into account, neither the scan strategy nor the scan speed or the repetition rate has any impact on the fabricated structures. SEM images captured under an angle of 45° reveal a comprehensive estimation of the relevant values: reproducibility, shape and length of the generated nanoantennas. Within an antenna field of 300 antennas the latter varies approximately by 10 percent.

 

Fig. 1 The large area SEM image shows the reproducible generation of approximately 40 nanoantennas generated with single pulses of fs-laser radiation at pulse energies of 21.2 nJ, focused onto the surface of a 50 nm thick gold thin film with a NA of 0.25. The inset shows a representative upright standing nanoantenna.

Download Full Size | PDF

3. Infrared absorption measurement and modeling of the resonance wavelength

3.1 Fourier-Transformed-Infrared-Spectroscopy (FTIR)

Owing to the orientation of the nanoantennas normal to the surface we performed spectroscopic measurement in grazing incidence geometry with p-polarized light. In this configuration, the incoming electric field vector almost matches the long axis of the nanoantenna, leading to an optimized optical extinction spectrum essential to determine the experimental resonance wavelength λ₀ of the antennas. These measurements are performed by the use of a FTIR spectrometer (Bruker, Vertex 70 spectrometer combined with a Hyperion 2000 microscope). A grazing incidence objective (Bruker) routes the infrared radiation under a central angle of 83° in reflecting geometry two times across the sample. All presented spectra are calculated on the basis of 500 scans collecting the reflected signal of a 120 µm x 120 µm large sample area, which corresponds to a field of approximately 300 nanoantennas. Afterwards, the spectra are normalized to the reflection spectrum of the plain gold thin film.

3.2 λ₀/4-model

A simple approach using a monopole model [10, 20] is applied to describe the correlation of the antenna length l with the corresponding resonance wavelength λ₀. Therefore, the length of the antennas is defined as the vertical extend of the antenna from the top of the bump-like structure to the tip of the antenna (Figs. 2(a) and 2(b)). By including the gold thin film beneath the antenna and irradiating the antenna under a grazing incidence angle, a separation of charges with opposite sign occurs as shown in Fig. 2(c). Due to the metallic properties of the gold film the charge distribution in the conductive layer is assumed as homogenous (Fig. 2(c)). An alternative approach to describe the charge distribution and the resulting electric field is based on the replacement of the present sample system depicted in Fig. 2(b) by a thin wire standing on a semi-infinite, conductive substrate. For this case the electric charge distribution can be described by introducing a mirror plane and an image charge as shown in Fig. 2(d). The resulting effective dipole antenna has a lateral extension which is twice as long compared to the originally defined antenna length in Fig. 2(b). In consequence, the theoretically expected resonance wavelength should obey the linear correlation of

 

Fig. 2 (a) SEM image of a spiky nanoantenna. (b) Schematic of a nanoantenna defining the length of the antenna. (c) Induced charge distribution of a nanoantenna on a conductive gold thin film as a result of an incident electromagnetic wave. (d) The charge distribution shown in (c) can be described by introducing a mirror plane and the corresponding image charge.

Download Full Size | PDF

4. Experimental results

The gold nanoantenna fabrication technique based on the irradiation with single, ultrashort pulsed laser radiation exhibits an exceptional simple access to a controlled manipulation of the intrinsic resonance wavelength by an adjustment of the length of the nanoantennas. In the following, we will present three different pathways to adjust the antenna length and the corresponding resonance wavelength in a graduated manner, ranging from a coarse tuning step with tunabilities of the length of the nanoantennas in the regime of several micrometers over a fine tuning in the regime of several hundreds of nanometers down to an ultrafine tuning offering precise tuning in the range of tens of nanometers.

4.1 Coarse tuning

A coarse tuning of the length of the nanoantennas is realized by varying the focusing conditions in terms of choosing different numerical apertures. Increasing the NA reduces the focus diameter, the lateral extent of the melting pool and the volume of ejected material forming the nanoantenna. Upscaling of the nanoantenna length is only limited by the stability of the formation process. In preliminary experiments we determined a maximum length of three micrometers for stable nanoantennas. For longer antennas thin film properties like morphology, crystallinity and texture emerge a profound impact on the formation process. To prevent an undesirable broadening of the resonance wavelength and to achieve the highest grade of reproducibility of generated nanoantennas, we restrict the experiments to antenna lengths up to two micrometers by generating nanoantennas with NAs of 0.4 and 0.15.

The normalized FTIR spectra in Fig. 3 show the wavelength dependent reflectance of the thin gold film modified by the characteristic dip owing the enlarged extinction of the corresponding nanoantennas at the resonance wavelength λ₀. Using focusing optics with NAs of 0.4 (blue circle) and 0.15 (red square) at pulse energies of 12.3 nJ and 57 nJ respectively, a variation of the antenna length from 0.42 µm up to 1.61 µm is realized resulting in a corresponding shift of the characteristic resonance wavelength λ₀ of the nanoantennas from 2.46 µm to 6.51 µm. Both spectra are collected by irradiating a surface area of 120 µm x 120 µm. This technique restricts the detection of a reflected signal of a distinct number of approximately 300 nanoantennas and ensures the measurement of a selected area which previously has been checked for imperfections via SEM. Furthermore, SEM images validate the spiky shape of the nanoantennas and serve as a rough estimate of the length of the generated nanoantenna.

 

Fig. 3 FTIR spectra of two antenna fields of 120 µm x 120 µm in size including approximately 300 nanoantennas. A coarse tuning of the characteristic resonance wavelength is achieved by varying the NA of the focusing optics. SEM images of representative nanoantennas validate the correlation between resonance wavelength λ₀ and the geometrical length l of the nanoantennas by a factor of four. (blue curve: NA = 0.4, E_p = 12.3 nJ, λ₀ = 2.46 µm, l = 0.42 µm; red curve: NA = 0.15, E_p = 57 nJ, λ₀ = 6.51 µm, l = 1.61 µm)

Download Full Size | PDF

4.2 Fine tuning

Depending on the applied pulse energies under constant focusing conditions (NA = 0.40), a fine tuning of the antenna length in the regime of several hundreds of nanometers is realized (Fig. 4(a)). Increasing the pulse energy until nanoantennas emerge, the highest spectroscopically investigated nanoantenna is generated at pulse energies of 11 nJ resulting in an antenna length of l = 0.73 µm and a corresponding resonance wavelength of 3.33 µm (Fig. 4). A further increase of the deposited pulse energy leads to the generation of antennas of decreasing length down to 0.42 µm. One possible explanation for this effect is the enhanced melt ejection [24] at higher pulse energies reducing the effective mass of gold remaining in the melt pool [25] to form the nanoantenna. A second supposable scenario is the temporal delayed solidification of the molten material due to the additional energy input. Hence, shrinkage of the molten material ejected in form of the antenna seems plausible in the advanced state of the formation dynamics. In accordance with the results of the coarse tuning (Fig. 3), a direct comparison of the estimated resonance wavelength with the length of the nanoantennas determined by SEM, reveals an appropriate description of these correlated values by the λ₀/4-model (Fig. 4(b)). The resonance wavelengths of the antennas tend to be red-shifted. However, considering the simplicity of the assumed theoretical model and the rough estimation of the nanoantenna length from a 45° tilted SEM image, the accordance of the experimental values within an error of approximately 15 percent is reasonable.

 

Fig. 4 (a) FTIR spectra of a gold thin film covered fs-laser radiation induced nanoantennas. The shift in the central resonance wavelength λ₀ is achieved by systematic variation of the pulse energy applied to generate the characteristic length of the corresponding nanoantenna. SEM images illustrate the possibility of tuning the length of the nanoantenna by varying the pulse energy. The presented spectra originate of six fields of nanoantennas generated at pulse energies of 12.3 nJ down to 11 nJ in steps of 0.3 nJ. (blue curve: NA = 0.4, E_p = 12.3 nJ, λ₀ = 2.46 µm, l = 0.42 µm; red curve: NA = 0.4, E_p = 11 nJ, λ₀ = 3.33 µm, l = 0.73 µm) (b) A comparison of the resonance wavelength observed for nanoantennas generated with three different numerical apertures with the length of the nanoantennas extracted from the SEM images yields good agreement of the assumed λ₀/4-model (red, solid line).

Download Full Size | PDF

4.3 Ultrafine tuning

The ultrafine tuning of the resonance wavelength consists of a post treatment step permitting wavelength shifts with accuracy in the regime of tens of nanometers. Therefore, subsequent to the laser induced generation process of nanoantennas, an additional gold film is deposited by means of thermal evaporation. The thickness of the additional gold thin film is measured by the quartz crystal of the thermal evaporator (Leica, MED 020) delivering reliable values of the film thickness with nanometer precision. The additional evaporated thin film thickness and the corresponding shift of the resonance wavelength Δλ₀ is presented in Fig. 5. In particular, the presented data are based on an alternating procedure of evaporation steps and the subsequent measurement of the extinction spectrum. A slope of 3.9 of the linear correlation between the additional gold thin film thickness and the shift of the resonance wavelength obeys the theoretical prediction of the assumed λ₀/4-model. However, the shift of the resonance wavelength caused by the deposition of an additional gold film contradicts the assumption of a uniform thin film growth on the antenna itself and the surrounding gold film. Therefore, the origin of the shift of the resonance wavelength is content of our current research.

 

Fig. 5 In post treatment steps additional gold thin films of different thicknesses are thermally evaporated covering the laser generated nanoantennas. The corresponding shift of the observed resonance wavelength in FTIR measurements is shown as a function of the additional gold film thickness. The slope of 3.9 of the assumed linear correlation of the additional gold thin film thickness and the corresponding shift of the resonance wavelength suggests a dependency according to the predicted λ₀/4-model.

Download Full Size | PDF

5. Conclusion and outlook

In conclusion, we fabricated laser generated nanoantennas applying single pulses of fs-laser radiation to induce the melt based formation process. Apparently, the presented nanoantennas exhibit great potential for near-field enhanced measurements due to their geometrical extent and their high aspect ratio. Thus, the control over the nanoantenna length and therefore the corresponding resonant extinction of these antennas is of great importance for applications like enhanced vibrational spectroscopy [8, 27] and antenna-enhanced near-field infrared spectroscopy [14, 28]. Therefore, three different tuning steps are presented to adjust the resonance wavelength of the fabricated antennas in a desirable wavelength spectrum from 2 µm up to 8 µm. Each tuning step possesses a characteristic precision enabling the graduated adjustment of the resonance wavelength in steps of the several micrometers, hundreds and tens of nanometers, respectively.

Beside the experimental results a monopole model is applied predicting a dependency of the resonant wavelength and the antenna length by a factor of four. This so-called λ₀/4-model is consistent with the fundamental electrostatic model applicable for metallic surfaces using image charges to elucidate the electrostatic field distribution of an electric charge in the vicinity of a metallic surface. Within the assumed errors this simple model is consistent with the experimentally observed resonance wavelengths. Addressing promising applications, near-field enhanced optical measurements using upright standing fs-laser induced nanoantennas are content of our current research.