1. Introduction

Subwavelength metal helix structures have been developed to perform extraordinary optical properties including broadband circular polarizer [1], circular dichroism [2] and negative index of refraction [3]. Such three-dimensional plasmonic nanostructures have also been proposed to be potentially applied in biosensing [4]. A regular and periodic helix array was developed using direct laser writing technique and functions as a polarizer over wavelengths from $3.5 \text{μm}$ to $7.5 \text{μm}$ [1]. Those periodically distributed nanohelices with feature sizes including pitch length of 2 $\text{μm}$ and arm width of 1$\text{μm}$ cause the circular polarization dependent transmission and reflection occurring at terahertz frequencies. Such broadband circular polarizer is desired to act at visible wavelengths. Therefore, it is desired for scientists and engineers to fabricate smaller sized metal helices and observe the associated optical properties at shorter wavelengths. Recently, glancing angle deposition (GLAD) [5] was applied to fabricate a metal nanohelix array on a seeded surface that was cooled down to a temperature below −100°C [6]. In a previous work, the period of the seeds on substrate surface is 76 nm and gold helices with pitch length of 100 nm and arm width of 25 nm were grown up to be 2-turn gold nanohelix array. However a patterned substrate with seeds regularly distributed on its surface requires complicated and expensive lithography procedure. In order to reach one-step deposition, shadowing effect has to be introduced in the initial stage of film growth successfully [7]. In this work, gold nanohelix arrays were fabricated on nonseeding glass substrates with GLAD. During deposition, the substrate is tilted an angle with respect to the incident flux direction and spins with a spin rate. By controlling the speed of substrate rotation with respect to the deposition rate, a spiral-like NHAs were then developed. Within the range of spin rates that lead to spiral-like nanohelices, a slower spin rate would lead to larger helices whose radius of curvature is larger. Three different sized NHAs were fabricated and their optical properties were measured and compared. As an important application of noble metal nanostructures the Surface-enhanced Raman Scattering (SERS) from the three gold NHAs were investigated. The strong SERS comes from localized surface plasmon resonances of metallic nanostructures that are capable of confining light into subwavelength volumes, generating intense local electromagnetic fields in their proximities (called hot spots). Recently, nanostructures have been developed from two dimensional to three dimensional morphologies to support stronger SERS [8] because compared with two dimensional structure, the three dimensional structure owns larger surface area enabling the extinctance and detection of more target molecules to exhibit a stronger SERS response. Therefore the gold NHA has the potential to function as a highly sensitive SERS substrate. The SERS from Au NHAs with different morphologies are measured and analyzed with near field simulation in this work.

2. Fabrication and morphology

Electron beam evaporation was applied to grow Au NHAs firstly. In this work, the substrate normal was tilted at an angle of 89° from the direction of incidence of the vapor during the deposition process. The center of the substrate and the evaporation source separated vertically by a distance of 350 mm. The substrate cooling was achieved to −140 °C by introducing liquid nitrogen passing through a stainless steel pipe underneath the substrate holder. The holder was made of cooper and 10 mm thick. A K-type thermocouple was directly mounted next to the substrate on the holder surface.Pumping yielded a background pressure of $4\times {10}^{-6}$ torr before evaporation. The deposition rate was controlled using a quartz oscillator adjacent to the substrate holder. The deposition rate of Au was maintained at 0.3 nm/s. The film thickness was monitored by a quartz oscillator. Sample1,sample2, and sample3 were grown at this oblique angle by depositing a desired turns of gold with a deposition thickness of 15000 Å, 11250 Å, and 9000 Å respectively.Three spin rates $\text{ω}$ of the substrate ($\text{ω}=0.018 \text{rpm}, \text{ω}=0.024 \text{rpm}, \text{and ω}=0.030 \text{rpm}$) were chosen with respect to the deposition rate to grow different sized Au nanohelices. Figure 1 (a) and (b) present the cross-sectional and top-view scanning electron microscopic (SEM) images of the 1.5-turn Au NHA deposited at $\text{ ω}=0.018 \text{rpm}$ (sample-1). From Fig. 1(a), the average diameter of the arms is 46 nm. The pitch length and the radius of curvature of the nanohelices are 219 nm and 152 nm, respectively. Due to the competition effect [9] during the helix growth, some incomplete grown helices (for example 0.5-turn or 1-turn helices) randomly distributed on the surface. The competitive growth mode during GLAD from the oblique deposition angle that results in strong atomic shadowing, and favors the growth of longer nanorods at the expense of shorter ones that die out. From Fig. 1(b), the average distance between two adjacent 1.5-turn helixes is 96 nm. The mean width of rods near the top of NHA become broader due to the fan-out phenomenon. Figures 1(c) and 1(d) show the cross-section and top-view SEM images of the 1.5-turn Au NHA grown at a spin rate of 0.024$\text{rpm}$ (sample-2). From Fig. 1(c), the average diameter of the arms is 50 nm. The pitch length and the radius of curvature of the helices are 159 nm and 105 nm, respectively. From Fig. 1(c), The average distance between two adjacent helixes is 125 nm. Figures 1(e) and 1(f) show the cross-section and top-view SEM images of the 1.5-turn Au NHA grown at a spin rate of $\text{ω}=$0.030 $\text{rpm}$ (sample-3). From Fig. 1(e), the average diameter of the arms is 48 nm. The pitch length and the radius of curvature of the helices are 130 nm and 75 nm, respectively. From Fig. 1(f), The average distance between two adjacent helixes is 162 nm. It is shown that the radius of curvature becomes less for a higher spin rate. The pitch angle defined as the angle between the growth direction of rods and the substrate surface are 28°, 34° and 40° for sample-1, sample-2 and sample-3, respectively. The aforementioned quantities including spin rates, radius of curvature, diameter of the arms, nanohelix spacing, pitch length, and pitch angle of three samples are listed in Table 1. and Fig. 2.

 

Fig. 1 Top-view and cross-section SEM images of 1.5-turn Au NHAs deposited at ω = 0.018 rpm (a, b), ω = 0.024 rpm (c, d) and $\text{ω}= 0.030 \text{rpm}$ (e, f).

Download Full Size | PDF

Table 1. spin rates, radius of curvature, diameter of the arms, nanohelix spacing, pitch length, and pitch angle of Au NHAs

View Table

 

Fig. 2 Schematic diagram of a nanohelix.

Download Full Size | PDF

3. Optical property

The transmittance and reflectance spectra of the three samples were measured with left-handed circular polarization (LCP) incident light or right-handed circular polarization (RCP) incident light [10], as shown in Fig. 3. Both transmittance and reflectance spectra of the three samples have similar trends. For transmittance spectra, both RCP and LCP transmittance have local maxima at wavelengths around 490 nm. For wavelengths over 585 nm, a transmittance valley occurs for each sample. All reflectance spectra have a minimum around $\text{λ}=500 \text{nm}$. For sample-1, the transmittance is less than 8.32% over wavelengths from 400 nm to 870 nm. The transmittance difference $\Delta T=T_{RCP}-T_{LCP }$ decreases from 1.79% at $\text{λ}=625 \text{nm}$ to zero at $\text{λ}=796 \text{nm}$ and then decrease to be –1.32% at $\text{λ}=870 \text{nm}$. The RCP and LCP reflectance are very similar with reflectance difference $\Delta R=R_{RCP}-R_{LCP }$ less than 0.84% and both minima are 1.55% at $\text{λ}=445 \text{nm}$ The maximal reflectance are 13% for RCP and 12.22% for LCP at $\text{λ}=870 \text{nm}$ The extinctance of sample-1 is larger than other two samples. The extinctance E defined as the sum of scatterance and absorptance was derived using the simple relationship E = 1-R-T. For sample-2, the local transmittance maximum is around 7.77% for both LCP and RCP spectra. The transmittance difference $\Delta T=T_{RCP}-T_{LCP }$ decreases from 0.93% at $\text{λ}=500 \text{nm}$ to zero at $\text{λ}=671 \text{nm}$ and then decrease to be –2.9% at $\text{λ}=870 \text{nm}$. The RCP and LCP reflectance are very similar with reflectance difference $\Delta R$ less than 0.3% for wavelengths less than 500 nm. The reflectance difference $\Delta R$ increases from 1.01% at λ = 600 nm to 2.63% at λ = 870 nm. For sample-3, the local transmittance maximum is around 9% for both LCP and RCP spectra. The transmittance difference $\Delta T=T_{RCP}-T_{LCP }$ decreases from 0.63% at $\text{λ}=500 \text{nm}$ to zero at $\text{λ}=675 \text{nm}$ and then decrease to be –5.59% at $\text{λ}=870 \text{nm}$. The RCP and LCP reflectance are very similar with reflectance difference $\Delta R$ less than 0.6% for wavelengths less than 500 nm. The reflectance difference $\Delta R$ is less than 1.04% for wavelengths from $\text{λ}=400 \text{nm}$ to $\text{λ}=870 \text{nm}$. The circular dichroism of the NHA is described with g-factor defined as the equation g = ∆E/E, where ∆E is difference of extinctance between left-handed and right-handed polarized illumination and E is the average extinctance. Figure 4 shows the g-factor spectra of the three samples. The g-factor of sample-1 is small and within the range between 0.004 and −0.006 over wavelengths from 400 nm to 870 nm. The g-factor of sample-2 is smoothly increasing from 0.001at λ = 400 nm to 0.035 at λ = 670 nm, and then decays to −0.003 at λ = 870 nm. The g-factor of sample-3 is also smoothly increasing from −0.004 at λ = 400 nm to 0.011 at λ = 616 nm, and then increase to −0.071 at λ = 870 nm.

 

Fig. 3 Circularly polarized transmittance and reflectance spectra of (a) sample-1, (b) sample-2 (c) sample-3. Transmittance difference (∆T) and reflectance difference spectra (∆R) of (d) sample-1, (e) sample-2 (f) sample-3.

Download Full Size | PDF

 

Fig. 4 Calculated values of g-factor as functions of wavelength for (a) sample-1, (b) sample-2 (c) sample-3.

Download Full Size | PDF

4. SERS characterization

To perform SERS characterization, the Raman probe molecule, 1, 2-Di(4-pyridyl)ethylene (BPE, TCI) was used, and a 4 μL droplet of BPE methanol solution with a concentration of $5.5\times {10}^{-5}M$ was dispersed on the surfaces of the three spiral-like Au NHAs. After the droplet had dried, the area over which it had spread on each of the substrates was observed to be approximately 5 mm². In SERS measurement, the fluorescence contribution has to be reduced [11]. Compared with other spectra by excited wavelengths lower than 785 nm, the spectra obtained at 785 nm showed better quality for all dyes, mostly because fluorescence was reduced. The Raman spectra were obtained using a Stroker 785L Raman Spectrometer from Wasatch Photonics, with an excitation wavelength of 785 nm, a power of 100 mW, a laser spot with a diameter of less than 50 μm, and a collection time of 30 ms..As shown in Fig. 5, all of the spectra include the following Raman characteristic peaks of BPE; $\Delta \text{ν}=1200 c{m}^{-1}$ (C = C stretching mode), $\Delta \text{ν}=1610 c{m}^{-1}$ (aromatic ring stretching mode), and $\Delta \text{ν}=1640 c{m}^{-1}$ (in-plane ring mode). The SERS intensity relative to the Raman intensity yielded an enhancement factor, EF [12]. $\text{EF}=(I_{sers}/C_{sers})/({\text{I}}_R/C_R)$, where $C_{sers}$ and $I_{sers}$ are the molecular density and Raman intensity in normal Raman measurement and $C_R$ and ${\text{I}}_R$ are the molecular density and SERS intensity. At $\Delta \text{ν}=1200 c{m}^{-1}$, the enhancement factors of the sample-1, sample-2, sample-3 are $E{F}_{sample-1}^{1200}$ = $2.22\times {10}^6$,$E{F}_{sample-2}^{1200}$ = $1.6\times {10}^6$, and $E{F}_{sample-3}^{1200}$ = $1.28\times {10}^6$, respectively. At $\Delta \text{ν}=1610 c{m}^{-1}$, the enhancement factors of the three samples are $E{F}_{sample-1}^{1610}$ = $2.1\times {10}^6$, $E{F}_{sample-2}^{1610}$ = $1.51\times {10}^6$, and $E{F}_{sample-3}^{1610}$ = $1.14\times {10}^6$ Samaple-1 yielded the strongest SERS signal of the three samples. Since most of the Au nanohelices were developed with large radius of curvature, they were expected to exhibit many more hot-spots than the other two samples. The SERS spectrum from the NHA is compared with those of slanted Ag and Au nanorod arrays (NRAs) that have been previously reported as highly sensitive SERS substrates [13]. Both Ag and Au NRAs were deposited under the deposition angle of 89° and deposition rate of 0.3 nm/s. The average length of Ag NRA was 624 nm and the column angle from the substrate normal is 75°. The average length of Au NRA was 432 nm and the column angle from the substrate normal is 64°. The SERS spectra of both NRAs are shown in Fig. 6. At $\Delta \text{ν}=1200 c{m}^{-1}$, the enhancement factors of the Ag NRA and Au NRA are $E{F}_{Ag NRA}^{1200}$ = 1.31 × 10⁵ and $E{F}_{Au NRA}^{1200}$ = $7.34\times {10}^5$, respectively. At $\Delta \text{ν}=1610 c{m}^{-1}$, the enhancement factors of the two samples are $E{F}_{Ag NRA}^{1610}=3.6\times {10}^4 \text{and }E{F}_{Au NRA}^{1610}=2.45\times {10}^5, \text{respectively}$. It is demonstrated that the Au NHAs own higher sensitivity than the previous reported NRAs.

 

Fig. 5 SERS spectra of sample-1(black line), sample-2(red line), sample-3(blue line).

Download Full Size | PDF

 

Fig. 6 SERS spectra of Au NRAs (black line), Ag NRAs(red line). The insects are the SEM images of Ag and Au NRAs.

Download Full Size | PDF

The relationship between the morphology of Au NHA and SERS signal is investigated with near field simulation. 3-D finite-difference time-domain (FDTD) simulations (Lumerical FDTD Solutions 8.7.11) are performed. Figure 7 shows the simulated extinctance spectra of three arrays of helixes with spacing S₁ = 96 nm, S₂ = 125 nm and S₃ = 162 nm shown in Table 1. It shows that the simulation result is in agreement with measured result.From the SEM images, the distribution of nanohelices in sample 1 is not so discrete as that of sample 2 and sample 3. Some nanohelices approach or connect to adjacent nanohelices to form a bundle nanohelix. There are pores and slits within the bundle nanohelix to support hot spots near the sharp features. The bundle morphology is more obvious for NHA grown with a lower spin rate because the average radius of curvature of helices is increased to cause overlap effect. A typical bundle nanohelices is selected for each sample to simulate the localized field intensity, as shown in Fig. 8. The set parameters for the FDTD calculations include a 3 nm-mesh and time step of 0.005 fs. The permittivity of gold was adopted from Johnson and Christy in the material library of the software [14]. An unpolarized light wave with a wavelength of 785 nm is normally incident onto the NHA and the electric field intensity defined as $|E/E_i{|}^2$ where E_i and E are the amplitudes of incident electric filed and localized electric field, respectively is simulated for its distribution on the typical helix. Figure 8 shows the field intensity distributions on xz-plane for the three samples. It is obvious that the sample-1 has the largest field intensity and most hot spots within the three samples. The field intensity and number of hot spot of sample-2 are higher than those of sample-3. The simulation can qualitatively explain the SERS measurement results.

 

Fig. 7 Measured and simulated extinctance spectra of (a) sample-1, (b) sample-2 (c) sample-3.

Download Full Size | PDF

 

Fig. 8 Simulated structure and corresponding intensity contour of electric field of nanohelix for (a) sample-1, (b) sample-2 and (c) sample-3. Note that, the insets of (a), (b), and (c) are the SEM image of simulated nanohelix.

Download Full Size | PDF

5. Conclusions

Three different sized NHAs were successfully fabricated on a smooth substrate with GLAD. By cooling the substrate during deposition, an Au NHA can be well developed by controlling substrate spin rate during deposition. The one-step shadowing deposition offers a general route to mass product various metal nano-helixes. Within the substrate spin rates which lead to spiral-like NHAs, the NHA with larger radius of curvature can be reached using lower substrate spin rate. As a typical three-dimensional structure, the three samples are potentially to act as a substrate for SERS. The lower spin rate of substrate causes larger pitch and radius of curvature to support more hot spots caused by the narrow gaps between adjacent nanohelics. The size dependent SERS from NHAs are then demonstrated. In the future, a finely 3D nano-structured plasmonic film can be developed based on this result by dynamic manipulating two axis rotation stages