Hot-Volumes as Uniform and Reproducible SERS-Detection Enhancers in Weakly-Coupled Metallic Nanohelices

Reproducible and enhanced optical detection of molecules in low concentrations demands simultaneously intense and homogeneous electric fields acting as robust signal amplifiers. To generate such sophisticated optical near-fields, different plasmonic nanostructures were investigated in recent years. These, however, exhibit either high enhancement factor (EF) or spatial homogeneity but not both. Small interparticle gaps or sharp nanostructures show enormous EFs but no near-field homogeneity. Meanwhile, approaches using rounded and separated monomers create uniform near-fields with moderate EFs. Here, guided by numerical simulations, we show how arrays of weakly-coupled Ag nanohelices achieve both homogeneous and strong near-field enhancements, reaching even the limit forreproducible detection of individual molecules. The unique near-field distribution of a single nanohelix consists of broad hot-spots, merging with those from neighbouring nanohelices in specific array configurations and generating a wide and uniform detection zone (“hot-volume”). We experimentally assessed these nanostructures via surface-enhanced Raman spectroscopy, obtaining a corresponding EF of ~107 and a relative standard deviation <10%. These values demonstrate arrays of nanohelices as state-of-the-art substrates for reproducible optical detection as well as compelling nanostructures for related fields such as near-field imaging.

Scientific RepoRts | 7:45548 | DOI: 10.1038/srep45548 Currently, the generation of near-fields satisfying such demanding conditions is attempted by regular arrays of rounded and conveniently spaced individual nanostructures (monomers) 3,11,12 . So far, there are two basic shapes studied to design monomer-based substrates: nanospheres 11,12 and -rods 3,5 . Both nanostructures have broad hot-spots, however, their maximum near-field intensity enhancement is moderate even when placed in an array configuration 3,5,11,12 . Such enhancement values are still low to be appropriate for very low concentrations of analytes, in particular for the case of single-molecule detection 6,7 . Here, we demonstrate how arrays of weakly-coupled Ag nanohelices have a reproducible, uniform and high near-field enhancement, reaching SERS EF values which are sufficient for detection of molecules in low concentrations, even down to the single-molecule level for the here tested SERS-active molecule 6,7 . With these near-field qualities, these substrates represent the first monomer-based substrate able to reach the limit of single molecule detection. This is possible due to the unique near-field distribution of a single nanohelix consists of broad hot-spots, merging with those from the neighbouring nanohelices in specific array configurations and generating a wide, uniform a highly enhanced detection zone, which we refer here as "hot-volume". We obtain a SERS EF of ~10 7 in our fabricated arrays of metallic nanohelices and a relative standard deviation (RSD) < 10%. These values validate these substrates as state-of-the-art substrates for highly enhanced and reproducible SERS detection 4,6,7 , as well as functional nanostructures for related fields such as near-field imaging 9 .

Results
Homogeneous near-field enhancement in metallic nanohelices. According to far-field studies, the optical response of a nanohelix is governed by localized surface plasmon resonances (LSPR) [13][14][15][16][17][18][19][20] . Therefore, regular metallic nanohelices bear the potential of converting incoming light into strongly enhanced and localized optical near-fields 18 . Figure 1 compares the calculated (see Methods) electric field intensity distribution of the commonly used nanorods and -spheres with the one of a nanohelix. At a wavelength close to the position of their LSPR, the ratio between the maximum near-field intensity from single nanohelices I helix max ( ) with respect to the incident radiation I inc is ~85, a value which is more than three times higher than the value for nanorods 1 ) and is over twice that of the nanospheres 11,12 ). Particularly, in a single helix (Fig. 1c) three electric field antinodes appear per turn due to its longitudinal LSPR 13,20 (that is, two consecutive dipoles per turn Fig. 1c, inset). Each one of these 'hotspots' cover a large area of ~100 nm in length and a plasmon penetration depth of ~40 nm. Furthermore, multiple-turn nanohelices 13 have a noticeable near-field enhancement of ~10% (turquoise colour, Fig. 1c) due to dipole-dipole interaction between different turns, which extends along the distance of one pitch of the helix, p. In addition, in this unique near-field distribution, periodic dipoles are formed with an effective length of half a pitch 13,20 , p/2 (Fig. 1c). Thus, surface charges of one sign are accumulated on a side of the nanohelix at integer (n) positions np; whereas the opposite charge is accumulated on the other side of the nanohelix at positions shifted by p/2: p(n + 1/2). This peculiar charge distribution plays a determining role for the intercoupling strength in arrays of individual metallic nanohelices separated by a distance (see discussions below). To assess this, we calculated ( Fig. 2) the maximum near-field intensity enhancement I max /I inc of a nanohelix placed in an array for different ratios /(2r w ), were r w is the radius of the wire forming the nanohelix.
While decreasing /(2r w ), a significant continuous increase in I max /I inc is found below /(2r w ) ~ 0.7, attributed to a strong coupling regime [21][22][23] . Meanwhile, in the limit of large separations ( /(2r w ) > 2), I max /I inc approaches the value of an individual nanohelix ( ) as expected for a residual coupling regime [21][22][23] . For intermediate values of /(2r w ), a weakly interacting interparticle coupling regime exists. Figure 2b-d show representative near-field distributions for the three described regimes. In the strong and residual-coupling regimes only smaller and rather localised hot-spots are found. In contrast, in the weak coupling regime, large regions of high and uniform electromagnetic enhancement are generated, which, as aforementioned, we refer as "hot-volumes". For an easier visualisation of these homogeneous enhanced near-field regions, the isosurface defining the plasmon penetration length 3,24 at (1/e)·I max /I inc from the near-field distributions in Fig. 2 are depicted in the insets of Fig. 3. Importantly, this ability of the weakly coupled regime to generate hot-volumes stems from the exclusive (regular) hot-spot distribution along an individual nanohelix (Fig. 1c) allowing a unique interparticle coupling  Hotvolumes are estimated by the ratio of volume enclosed by the isosurface of (1/e)·I max /I inc (IV) and the volume between neighbouring helices (VNH) in arrays of Ag nanohelices for different aspect ratios /(2r w ). This ratio IV/VNH shows a clear maximum at /(2r w ) ~ 1 (weak interparticle coupling) whereas it falls rapidly off when entering into the strong coupling regime ( /(2r w ) < 0.7). At this point, the IV/VNH ratio is ~8%, that is, 8% of the total volume between neighbouring nanohelices is filled by the isourface. Insets: (left) Visualization of the isosurfaces at between neighbouring nanohelices at a distances p/2 (Fig. 2a, inset). For a better observation of the formation of the different hot-volumes generated when varying the interparticle distance, we plot (Fig. 3) the ratio between the volume enclosed by the isosurfaces at (1/e)·I max /I inc , IV, and the total volume existing between neighbouring helices VNH for several values of /(2r w ). The histogram shows that there is a steady increase of IV/VNH coming from large interparticle distances peaking in the weakly coupled regime for /(2r w ) ~ 1. Thereafter IV/VNH quickly falls off close to zero below /(2r w ) ~ 0.7, that is when entering into the strong coupling regime. The combined results in Figs 2 and 3 underpin that indeed the generation of hot-volumes can be deliberately achieved by proper choice of the nanohelix's parameters and their separations in an array. We further emphasise that the hot-volume at /(2r w ) ~ 1 is larger than 3 × 10 5 nm 3 and therefore allows for the uniform excitation of even large biomolecules 3 . Moreover, this value comprises ~8% of the free space in between neighbouring nanohelices; exceeding considerably any reported volume fraction (< 0.1% 10 ) in current substrates [3][4][5][6][7][8][9][10][11][12] . Importantly, weakly coupled nanohelices with /(2r w ) between 0.8 and 1.1 possess similar I max /I inc and normalized hot-volumes IV/VN, a fact suggesting robustness against fabrication tolerances of these nanostructured substrates 13 .

Assessment of SERS detection in metallic nanohelices.
To underpin our theoretical considerations on the use of arrays of weakly coupled nanohelices for optical detection of molecules in low concentrations, and thereby confirming their principle applicability, we fabricated 19 a Ag nanohelix array with the aspect ratio of ~1 (Fig. 4a, inset). We opted for SERS as a commonly used optical technique for single molecule detection 3-10 and measured their corresponding EF and RSD at 785 nm incident laser light with the widely used probing molecule 3,8 pMA as analyte.
The choice of the 785 nm laser line is due to the position of the Ag nanohelix LSPR position at ~760 nm (Supplementary Note 4). After being fabricated, nanohelices were covered with the analyte by immersion into a diluted aqueous solution of pMA (see Methods). Through this, an adsorbed monolayer of these molecules forms on the nanostructured metal surface, as reported in the literature 3,8 . Figure 4a shows the Raman spectra of pMA covering Ag nanohelix arrays and bulk-powder pMA used as reference. The presence of Ag nanohelices enhances considerably the Raman signal of the monolayer of pMA molecules deposited on their surface. The estimated EF of the present samples is at least 10 7 as extracted from the two characteristic pMA Raman lines 3,8 at around 1080 and 1180 cm −1 (cf. Methods and Supplementary Note 1). The main contribution is of electromagnetic origin as we have confirmed through additional measurements of pMA on an Ag thin film (Supplementary Note 1). Our observed EF is one order of magnitude larger than state-of-the-art, monomer-based, reproducible SERS substrates such as regular Ag nanorods 3 . The latter can be understood from both, the higher near-field intensity achieved by individual nanohelices with respect to nanorods (I I / helix r od max ( ) max ( ) > 3), and the existence of a hot-volume rather than hot-spots in our samples. Furthermore, our EF compares to state-of-the-art, single-molecule SERS substrates with homogeneous near-fields, which, however, require additional filling (blocking) molecules 4 to ease the inhomogeneity of the near-field of highly localised hot-spots for the analytes and bears the possibility of interference of the individual Raman responses 4 of filling and analyte molecules. The performance of the arrays of nanohelices, however, does not involve any additional treatment.
Finally, we assessed the reproducibility of the SERS signals in our Ag nanohelix arrays measuring 10 random spectra separated ~100 μ m across the entire sample area (1 mm × 100 μ m). We obtained highly repeatable spectra (Fig. 4b) and estimated RSD values of ~10% and ~6% for the two aforementioned Raman lines (1080 and 1180 cm −1 , respectively; cf. Supplementary Note 2). Again these values are within the range of state-of-the-art substrates [3][4][5] . We reconfirmed these RSD values by using graphene as an alternative probing molecule (Supplementary Note 2).

Discussion
Altogether, the unique optical near-field of an individual nanohelix allows the creation of large and homogeneous near fields (hot-volumes) in arrays of nanohelices for aspect ratios /(2r w ) ~ 1 via weak interparticle plasmon coupling. These volumes comprise up to 8% of the free space inbetween neighbouring nanohelices. By using SERS as a common optical detection technique and without any further optimization, these arrays can reach high EF ~ 10 7 and RSD ≤ 10%, as exemplified on an array of Ag nanohelices. These values are superior to other (well separated) monomer-based substrates 3,5 , reaching the limit for reproducible single-molecule detection 6,7 for the studied analyte (pMA). In addition, they are comparable to substrates containing narrow hot-spots with interfering filling molecules to achieve a homogeneous near-field 4 . More generally, the present results provide principle design strategies to achieve uniform, spatially extended and intense near-fields via weak interparticle coupling of plasmonic nanostructures. Thus, these systems are not only suitable for molecular detection and analytics as demonstrated here, but also for other related near-field techniques such as fluorescence 1 or tip-enhanced Raman spectroscopy (TERS) imaging 2,9 .

Methods
Fabrication of metallic nanohelices. Periodic arrays of regular metallic nanohelices are engineered at the nanoscale by oblique angle deposition on rotating substrates 19 . For the present study (Supplementary Information), we fabricated arrays of Ag nanohelices (size 1 mm × 100 μ m) keeping for each array similar helical parameters (pitch p ~ 128 nm, diameter D ~ 75 nm, number of turns N ~ 2.8, and wire radius r w ~ 35 nm). The centre-to-centre distance between nanohelices is 150 nm, thus the separation distance between helices is ~ 75 nm and the achieved aspect ratio is /(2r w ) ~ 1. Samples were kept under vacuum after fabrication prior to the Raman measurements.
Measurements and determination of EF. Raman spectra acquisition was done at ambient conditions in a Renishaw micro-Raman spectrometer using a × 50 objective lens. The laser wavelength used was 785 nm. The calibration was carried out with the aid of the Rayleigh and Si bands at 0 cm −1 and 521 cm −1 , respectively. The micro-Raman resolution is ~2 cm −1 . Acquisition time was 1 min, and the laser power was 1 mW. Experimentally, the EF can be determined by the expression 3 where ξ is the focal depth of the laser spot (12 μ m for 785 nm wavelength), ρ ~ 6.5 × 10 27 m −3 (1.18 g cm −3 ) is the density of pMA molecules 25  pMA thin layer on nanohelices. We exposed the Ag nanohelices for ~12 hours to an aqueous solution (10 −5 M) of pMA. Through this we achieved a monolayer of these molecules attached to the nanohelices' surface due to the common chemisorption process 3,8 of pMA to silver.

Numerical simulations.
We simulate the near-field intensity of individual and arrays of nanohelices using the commercial finite element modelling (FEM) package, COMSOL (Supplementary Note 3). In all the cases, and adaptive mesh was used with maximum element sizes of 5 nm and a perfect conductor (PC) boundary condition in the plane z = 0 is used. The latter condition models the metallic Ag thin film existent in our samples underneath the nanohelices 19 . Before applying this PC approximation a comparison was made by simulating the full experimental substrate (50 nm silver, 300 nm SiO 2 and Si base wafer); no differences were observed. Light is incident along the negative direction in the z axis. The helices used have a pitch of 130 nm, a diameter of 75 nm and a diameter of the helical wire of 70 nm. All these values are close to those of the experimentally fabricated nanohelices. The scattered field is absorbed by perfectly matched layers (PML) regions. For all (individual and arrays) of nanohelices, the simulation domain is enclosed by five PMLs. Five elements were needed across the PML region to accurately absorb scattered field with less than 1% back reflection. In the case of arrays of nanohelices, we estimated the intercoupling effect in helical arrays by taking into account the contribution from the four nearest neighbours at different particle separation distances . Finally, the optical properties of the nanohelices are defined through the real and imaginary components of the refractive index 26 .