The field of nanoplasmonics promises manipulation of electromagnetic wave propagation almost at will.1, 2 Tailored nanostructures, usually metallic, are the centerpiece to consciously control light-matter interaction, whereby fascinating optical effects like cloaking,3 superlensing4 or negative refraction5 become reality.
Chiral nanostructures cannot be superimposed on their mirror image, defining chirality as a geometrical property. Due to this fact the optical response of chiral nanomaterial differs when left circular polarized (LCP) or right circular polarized (RCP) light is incident. For example, circular dichroism has been observed when a chiral nanomaterial sample was illuminated by LCP and RCP light.6–9 When illuminated with linearly polarized light, large polarization rotation is also possible.10, 11 Furthermore, chiral nanomaterials are envisioned to be advantageous to realize negative refraction12–14 or to exhibit a repulsive Casimir effect.15 It has recently been demonstrated that the chiral optical response is much stronger if configurational rather than constitutional chirality is involved.6
One of the most prevalent challenges for the realization of chiral shaped nanostructures is their extension into all three spatial directions allowing for a pronounced and well measurable chiral optical response. Existing approaches for chiral nanomaterial fabrication are mainly separable into bottom-up and top-down approaches. The former involves chemical approaches wherein growth and self-assembly of metallic nanocrystals into chiral shapes or assemblies are of common use.16–18 The latter comprises deterministic, for example, lithography-based nanofabrication routes recently summarized by a comprehensive review article from Lindquist et al.19 So far, the majority of top-down approaches have mainly focused on subsequent stacking of inherently nonchiral nanostructures. As a result, this yielded either a single electrically connected chiral shape20, 21 or a union of electrically separated and twisted nanostructures comprising an overall three-dimensional and chiral assembly.10, 22–25 Other notable approaches to achieve chirality have been performed by combining direct laser writing and electroplating7 and via electron beam induced writing of gold nanostructures.26 To date, all these methods hit one of the following limits: either the method is time-efficient but lacks of flexibility and reproducibility, or vice versa.
In this work, we introduce a technique termed on-edge lithography (OEL). We show how this method is capable of efficient, accurate and reproducible chiral nanomaterial fabrication by simultaneously reaching small feature sizes. The key idea of OEL is to process electron-beam resist spin-coated on a prestructured surface, further referred to as the template, by electron beam lithography (EBL). The process is schematically depicted in Figure 1 with further details given in the Experimental Section. The use of OEL reduces the challenge of three-dimensional nanostructuring to the preparation of a conveniently designed template. Although the herein described template was prepared by EBL as well, this step can in principle be achieved on large scale by a huge variety of fast and cost-effective fabrication tools. When EBL-generated nanostructures are attached to the edges of such a template, they will finally unfold into three-dimensional nanostructures, giving rise to the nomination “OEL”. Accordingly, OEL holds solid promise to reduce the effort to achieve three-dimensional nanostructuring to one lithographic step.
Without loss of generality we demonstrate a chiral nanostructure as an example of OEL. Our developed chiral nanostructure design (golden shape in Figure 1e) consists of a thin slab bent to a quarter of a ring (further referred to as the “foot”). In addition, it features an extension perpendicular to the ring plane on one end of the foot (further referred to as the “arm”), whereby the design fulfills the condition of chirality. Since this geometry exhibits a similarity to an L-shape it will further be denoted “3D-L particle”. Figure 2a depicts a tilted view scanning electron microscopy image of the final structure. The arms of the chiral shape (length 170 nm and width 80 nm) located right on top of the grating bar are clearly visible, whereas the foot lies in the grating groove (groove radius 210 nm, foot width 80 nm). It is worthwhile to note that the position accuracy of the EBL system is specified as below 20 nm under appropriate conditions.
We performed rigorous electromagnetic simulations of the 3D-L particle by means of the commercial software package Fullwave (Rsoft Design Group Inc). The optical responses for incident RCP and LCP light were investigated while periodic boundary conditions in the plane were assigned to the unit cell which contained two 3D-L particles facing each other. The dielectric function of gold was described using the model proposed by Rakić et al.27 For the fused silica template we adopted the dispersion published by Malitson.28 Figure 3a depicts the transmittance spectra obtained after simulation. For both incident polarizations two distinct resonant features are observed as dips in the transmittances at wavelengths of 990 and 1770 nm for RCP (solid line) and 800 and 1770 nm for LCP (dashed line) light, respectively. Over the whole spectral range, notable differences between the transmitted intensity can be observed as footprints of circular dichroism stemming from the 3D-L particles (black solid line). This quantity peaks resonantly at the three specific center wavelengths 800, 990, and 1770 nm. The electron current density within a single 3D-L particle is monitored in Figure 3b–e to deduce the nature of the respective plasmonic modes. At the fundamental mode at longer wavelengths (1770 nm for both LCP and RCP), we observe a maximum electron current in the center of the 3D-L particle concentrated close to the junction between the foot and arm. In contrast, the first higher harmonic mode at shorter wavelengths (800 nm for LCP and 990 nm for RCP) features a node of the electron current at this specific spot, while the maximum currents are observed at the extremities of the particle.
The measured transmittance spectra of the fabricated 3D-L particle array are displayed in Figure 4. The blue solid line representing RCP light features mainly one transmittance dip in the full spectral range under investigation. The measured data indicate a minimum value at 990 nm wavelength where the transmittance drops down to 20%. Additionally, the graph progression exhibits a shoulder at wavelengths around 1200 nm broadening the overall transmittance dip width. In contrast, the blue dashed line (LCP) indicates only a slight decrease in transmittance starting at wavelengths below 1300 nm. The transmittance difference of the two spectra (black solid line) reveals a notable and broadband circular dichroism of over 40% between 950 nm and 1280 nm wavelength, while the peak value exceeds 45% at 990 nm wavelength. This proves that OEL is capable of chiral nanomaterial fabrication. Furthermore, we figured out two specific optical properties of the investigated nanostructure array. First, RCP light is more effectively blocked than LCP light in the measured spectral range. Second, the handedness of the 3D-L particles and the handedness of the favorably blocked polarization state are of opposing case.
Comparing the simulations based on an ideally shaped 3D-L particle to the measurement results we encountered some discrepancy. Although the low-wavelength resonance position at 990 nm wavelength is well confirmed, the fundamental mode appears far more red-shifted in the simulation data. This discrepancy can be explained by a closer look at the structural details of the 3D-L particles stemming from peculiarities of OEL. In Figure 2c, a discontinuous gold film at the junction between arm and foot becomes apparent. This specific feature was not included in the simulation so far. Furthermore, one would expect a template-dependent gold film thickness gradient along the groove profile due to fabrication relevant aspects. Thus, implementing a gap and a thickness gradient affects the optical response as depicted in Figure 5. The resulting transmittance spectra for incident RCP and LCP light are displayed as a green solid and dashed line, respectively. We note that the fundamental mode resonance of RCP light is blue-shifted and for a gap of 50 nm and a cosinusoidal gold film thickness gradient, a convincing agreement between simulated and measured RCP light data exists. For LCP light the simulated data still features two transmittance dips, which were not observed in the measurements. In Figure 5b–e, a snapshot of the electron current density is displayed. Due to the gap, electron oscillations from LCP and RCP light at a wavelength of 1280 nm are mainly excited in the foot, while the excitation in the arm is negligible. At a shorter wavelength the excited currents in the arm and foot are comparable, but we observe the oscillations in the foot lagging behind the oscillations in the arm by around π/4.
Finally, we investigate the influence of the existent gap in the fabricated 3D-L particles on their resulting optical response. Intuitively, this gap has a larger impact on the fundamental plasmonic mode at longer wavelengths, since the electron current distribution as shown in Figure 5b,d cannot be sustained. On the other hand, the first-order harmonic mode at shorter wavelengths can effectively be excited even without a continuous connection between the arm and foot of the 3D-L particle, since the electron currents show a node at precisely this spot (Figure 5c,e). These assumptions have been sustained by further rigorous simulations. It was revealed that the blue-shift of the resonance at longer wavelengths can be triggered by the size of the gap, while neither the spectral position nor the strength of the resonance at shorter wavelengths is notably affected. With a gap size of 50 nm corresponding to the fabricated sample, both plasmonic resonances overlap and form a broader operation band. Accordingly, the gap size in 3D-L particles can be used to tailor circular dichroism either for broadband operation or in spectrally separated resonant domains. Technically, in OEL fabrication technology, the gap size is a direct measure of the sharpness of the edges defined on the prestructured template. Depending on the route how this template is provided, a multitude of 3D nanostructures beyond 3D-L particles can be created for various nanometer-scaled polarization optics.
In conclusion, we succeeded to apply EBL on nonplanar surfaces. The presented OEL technique offers an accurate and reproducible method to generate three-dimensional and/or chiral nanomaterials within one lithographic step and allows for nanometer-scale feature sizes. As an example of the OEL technique, a chiral nanostructure termed 3D-L particle was demonstrated. The optical investigation confirms notable and broadband circular dichroism at near-infrared wavelengths. The spectral width of the operation band is tunable by tailoring topographic parameters of the nanostructures. The presented nanostructure is usable as a plasmonic polarization filter and/or for biosensing. Additionally, the increased fabrication efficiency allows for voluminous chiral nanomaterial generation.