1. Introduction

Optical detection techniques, such as a terahertz (THz) time-domain spectroscopy (TDS) [1–3], surface-enhanced Raman spectroscopy [4], gas chromatography [5], and near-infrared spectroscopy [6, 7] are quite promising detection methods in terms of molecular sensing in both quantitative and qualitative manners. In particular, the THz-TDS provides great benefit for fast and precise detection of intra- and intermolecular features through broad THz spectral bands [8–10]. Despite their advantages in molecular sensing, however, the detectability of THz-TDS studies is severely limited by extremely small amount of target molecules at THz regime, and has remained as the most critical and challenging issue. Several strategies have been introduced to overcome these issues. For example, metamaterials based on nano scale gap structures have been used to increase the THz absorption cross section, in turn, enhance the detection sensitivity [11–14]. Furthermore, nano gap structures designed to target the intrinsic vibrational- and intra-molecular vibrational modes of the samples have been introduced as highly sensitive and selective detection tools [15–17]. In most cases, target subjects were prepared in solution state and drop casted onto the sensing chip, unavoidably accompanying molecule aggregations and ion stains such as crystallization and a coffee-ring effect [18, 19]. Clearly, this unwanted surface status can cause random errors in transmission/reflection result as well as decrease in detection efficiency.

Here, we suggest a unique sample collecting method to improve sensing capability in THz measurements, enabling both accurate transmission control and increased detection efficiency. Based on nano slot antenna arrays with a field enhancement of 140 at the resonance frequency of 1.0 THz [20], we measured THz transmission spectra for 80-nm-diameter gold nanoparticles (AuNPs) that are spread over the nano antennas. A THz transmission was controlled by a sweeping of randomly dispersed AuNPs into the nano antenna region where the THz field is strongly enhanced. With two different widths of the slot antenna, the changing behaviors of transmission spectra were discussed quantitatively and qualitatively in terms of distinctive resonance frequency shift and reduced transmission amplitude.

2. Results and discussion

Schematic of the sweeping method using PDMS film with nano slot antenna structure is shown in Fig. 1(a). Gold nano slot antennas are composed of two-dimensional (2-D) punctured rectangular slots of h = 150 nm, l = 60 μm, p = 10 μm, and d = 40 μm, where h, l, p, and d indicate the height, length of slot, and adjacent distances between neighboring slots, respectively, fabricated by combinatorial process of an e-beam lithography and a lift-off technique on polished silicon wafers [21]. Nano slot antenna patterns were fabricated on high-resistivity Si wafer (500 μm in thickness and 20,000 Ωcm in resistivity) by a general E-beam lithography process consisting of resist coating (AR-N 7520.18), exposure and development. And the lift-off process was carried out by socking in an acetone bath to wash away the patterned resist with unwanted metal. More than 1000 nano slots are beneficial for consistent measurement result for the high throughput sensing performance (Fig. 1(b)). The Au solution was a 30 μL solution (BBI^TM solutions) of water containing 80-nm-diameter AuNPs, with a concentration of 2.75 × 10⁹/mL, was dropped over the nano slot antenna surface and dried for avoiding any unwanted THz water absorption. A ‘squeegee’, made of the Polydimethylsiloxane (PDMS), was prepared by mixing Sylgard 184 silicone elastomer (Dow Corning Corp.) with the curing agent in a 10:1 ratio, followed by thermal cure process for 3 hours at 80 °C. Completely dried 5-mm-thick bulk PDMS film was then sliced into 5 mm × 10 mm pieces. The maximum intensity of THz near-field, however, are still dominantly concentrated inside the nano slots as illustrated Fig. 1(c) obtained from the finite-difference time-domain (FDTD) calculation. Figure 1(c) represents THz near-field distributions at a single nano slot using physical values of w = 500 nm and h = 150 nm without taking into account the existence of AuNPs in the nano antenna. This clearly implies that additional gathering of AuNPs inside nano antennas where the strongest field are focused is expected to support more robust and quantitative analysis of transmission reduction (ΔT) and a qualitative resonance frequency shifts (Δλ), respectively (Fig. 1(d)).

 

Fig. 1 (a) Schematic illustration of concept of sweeping AuNPs inside the nano slot area using a PDMS squeegee. (b) THz detection using nano-slot-antenna array sensing chip. (c) FDTD-calculated cross-sectional THz near-field maps around a nano antenna, producing highly enhanced local fields inside the slot. The bottom gray area represents the silicon substrate, and the nanoparticles were drawn over the simulation image for the conceptual purpose. (d) Concept of THz transmission signal analysis in terms of a transmission reduction (ΔT) and a frequency shift (Δλ) by insertion of nanoparticles (NPs) inside the nano antenna.

Download Full Size | PDF

We first find the optimized procedure for efficient sweeping of AuNPs with a concentration of 2.75 × 10⁹/mL into the nano antenna region using PDMS ‘squeegee’. To check the validity of the sweeping procedure, we performed a negative control experiment; a solution drop-casting, followed by a solvent drying process on two identical nano slot antennas (Fig. 2(a)). Note that all sweeping processes, perpendicular to the long-axis of nano antennas, were performed in wet environment by means of the ‘fine water spraying’ prior to the PDMS sweeping (Fig. 2(b)). We found that an additional fine water spraying over AuNP, spread on nano slot antenna surface, facilitates to slide a PDMS squeegee smoothly as well as eliminate any possible surface damage, i.e., scratches under drying conditions. Smooth sliding is attributed to thin water layers which provides a buoyancy force during the sweeping process. The bright-field microscopic image of AuNP (1.375 × 10¹⁰/mL by successive five cycles of solution drop-casting) spread over nano slot antenna is shown in Figs. 2(c) and 2(d). The dispersed AuNPs having light scattering in yellow color show how they are distributed by corresponding sample preparation method. Here, the samples prepared by drop-casting intrinsically show AuNPs strains at the edge of solution (Fig. 2(c)), while those by sweeping was mostly free from the coffee-ring effects near the nano slot regimes (Fig. 2(d)). The scanning electron microscope (SEM) images of AuNPs spread by drop-casting were investigated in two different nano slot antennas (w = 130 nm for Fig. 2(e) and w = 500 nm for Fig. 2(f)). This apparently shows random distribution of AuNPs over the entire nano antenna surface, while swept particles lead mostly positioned within nano slots (w = 130 nm for Fig. 2(g) and w = 500 nm for Fig. 2(h)). Direct comparison of SEM images and measured analysis of the portion of AuNPs inside nano slot clearly show that the PDMS rubbing not only cleans the surface of the sensing chips, but also concentrates AuNPs into the sensing of the interest area in a well-controlled manner (Fig. 2(i)). We found that drop-casting results in less than 10% of AuNP migration to nano slots, in strongly contrast to sweeping with a more than 90% of success rate. The physical sweeping of AuNPs with a squeegee, therefore, can be considered as an excellent supporting technique on highly ordered and controlled sensing experiments.

 

Fig. 2 Schematic illustrations of the drop-casting (a; top line) and the sweeping (b; bottom line) techniques. Microscopic image of AuNPs spread over the nano slot array after five cycles of solvent drop and evaporation (c), and corresponding SEM images of a nano slot at w = 130 nm (e) and w = 500 nm (f). Microscopic image of AuNPs over the array after 5 sets of drop-casting procedure, followed by sweeping procedures (d) and corresponding SEM images of each sample at w = 130 nm (g) and w = 500 nm (h). Counted AuNP ratio in nano slot using two techniques (i).

Download Full Size | PDF

Now we compare THz transmission spectra for the AuNPs-filled nano slot antennas in two different supporting techniques. We repeated the sweeping process, followed by THz transmittance measurements with repeated additions of 2.75 × 10⁹/mL of AuNP in each cycle (Fig. 3(a)). Control experiments which skipped the sweeping in every cycle are performed as shown in Figs. 3(b) and 3(d). As we expected, THz transmissions for nano slot antenna by sweeping show gradual signal reductions until more than 10 times of sweepings, while THz transmissions from control experiments are saturated within an initial few cycles. For example, w = 130 nm slot shows a transmission reduction (ΔT_drop≈18%) as well as a red shift of resonance frequency (Δf_res≈0.06 THz) within couple of cycles (Fig. 3(b)), but immediately being saturated. Surprisingly, the swept slots show drastic transmission reduction down to ΔT_sweep≈70% and Δf_res gradually increases from zero to 0.062 THz at every cycle (Fig. 3(c)).

 

Fig. 3 (a) Schematic illustrations of a single cycle of sweeping methods, followed by THz detection of AuNPs. THz transmission spectra measured for AuNPs over the w = 130 nm (b, c), and w = 500 nm (d, e) slots by drop casting method only (b, d), and by additional sweeping method (c, e) in each cycle, respectively.

Download Full Size | PDF

Additionally, to exploit the role of sensitivity dependent on the nano slot width, we measured THz transmittances for equal amount of AuNPs dropped on 500 nm slots, which show similar trends (Figs. 3(d) and 3(e)). That is to say, drastic reduction of transmittance was observed (ΔT_sweep≈55%; Fig. 3(e)) in swept case, while drop-casted samples show saturated signal with ΔT_drop≈6%. We also observed gradual Δf_res in sweeping case from Δf_res≈0.00 to 0.062, meanwhile very quick saturation was monitored in drop-casting (Δf_res≈0.03). Distinguished ΔT values in terms of AuNP concentrations confirm our assumption that the filling factor of AuNPs inside the nano antenna gradually increases until no more enough spaces are allowed. That is to say, a saturation of the ΔT is attributed to full AuNPs inside slots to a certain cycle. On the other hand, the drop-casting and spontaneous solvent evaporation, lacks of particular strategies to manipulate AuNPs into the sensing of the interest area, resulting in random distribution of particles. The ΔT value as function of particle filling rates clearly demonstrate how effectively our nano slot antenna works for nano particle sensing in quantitative manner. Further, a direct comparison of THz transmission by the sweeping and drop-casting manifests that the sweeping is absolutely more sensitive in particle sensing.

For a quantitative analysis of our measurements, we plot ΔT and Δf_res as a function of AuNP concentration (Fig. 4). As shown in Fig. 1(c), nano slot antenna generates strong THz near-field that causes a significant increase in the absorption cross section, allowing detection of target molecules in low concentration [11, 17]. However, simple drop-casting clearly lacks of a sensing capability, as this method shows ΔT_drop up to 18% within initial two cycles of AuNP introductions but no more quantitative analysis available afterward (Fig. 4(a), pink dots). On the other hand, the sweeping enables ΔT_sweep up to 70% within 12 cycles (Fig. 4(a), blue dots) and very similar curve was achieved for w = 500 nm slot antenna (55%; Fig. 4(c), blue dots), whereas only ΔT_sweep≈6% in drop-casted sample (Fig. 4(c), pink dots) were monitored.

 

Fig. 4 Quantitative analysis of ΔT (a, c) and Δf (b, d) as a function of AuNP concentration. The AuNPs over the w = 130 nm (a, b), and w = 500 nm (c, d) slots and their corresponding THz transmissions are also represented. After evaporation of last cycles of experiments in both methods, SEM images for drop-casted sample (e), and swept sample (f) in w = 500 nm slots were taken.

Download Full Size | PDF

Transmittance spectra can be interpreted by adapting a fitting function with sigmoidal curve equation [22, 23], ΔT = a/$(1+e^{-k(x-c)})$, where a is a maximum value in transmittance change, k is a sensitivity of the measurement (steepness of the curve), and c is an inflection point: the point on the S shaped curve halfway between maximum transmittance change and zero transmittance change (Fig. 4, blue dotted-line). A transmittance spectrum curve by simple drop-casting was also verified by the Michaelis-Menten function represented as ΔT = $V_{max} x/(k_m+x)$ [24], where V_max is a maximum of transmittance change, and k_m is also slope of a curve depicting sensitivity (Fig. 4, pink dotted-line). Data in appropriate ranges are employed for better fitting. On the one hand, commercial 80-nm-diameter AuNPs were uniformly dispersed and suspended in water having trace amounts of ions, such as citrate, tannic acid and potassium carbonate. Unfortunately, the exact specifications of the dissolved ions, affecting the red shift of f_res, was hardly accessible. In practical, we easily observe crystallized solid residuals right after evaporation processes. Figure 4(e) clearly shows randomly formed solid residuals within and near the slots when the sample were prepared by the drop-casting, meanwhile swept sample doesn’t show any crystallized residual over the entire surface of antenna array (Fig. 4(f)). Those facts imply that sweeping has advantages not only gathering target materials into a sensing hot spot, but also contributing to a removal of significant amount of uninterested ions during the rubbing process.

3. Conclusion

We propose a robust transmission control of THz electromagnetic wave by tuning filling factors of sub-hundred diameter AuNPs inside nano slot antennas where the THz field is dominantly enhanced and localized. Sweeping the dropped nano particles into the nano antenna enabled quantitative and qualitative analysis of THz signal with greatly enhanced sensitivity and accuracy. A direct comparison of THz signals between the sweeping and drop-casting manifests that the sweeping is far sensitive with more than about 3.5 times THz transmittance reduction as compared to the other. Furthermore, we found that the sweeping significantly reduces the ion adsorption near the nano antennas, reducing unwelcome THz signal difference from ions. These presents important clues in reinterpreting in conventionally prepared-sample deposition to meet further steady demands in developing ultrasensitive terahertz (THz) molecule sensors with higher sensitivity and accuracy. This method is expected to provide insight into a precise and quick detection in vast range bio-application including nanoscale cell (e.g., exosome), proteins assembly, and nano solid-materials.