SERS Active Hierarchical Nanopillar-huddle Array Fabricated via the Combination of Nanoimprint Lithography and Anodization

Metallic nanostructures and their fabrication methods have been studied for over decades as they are crucial in developing plasmonic sensing platforms. In this work, a hierarchical nanopillar huddle structure fabricated by thermal nanoimprint lithography with anodic porous alumina as template is presented. By utilizing this scheme, nanopillars (branches) rooted on regularly deployed substructures (footings) can be easily produced/reproduced for large working area at low-cost with high-throughput. After metal deposition for plasmon activation, tiny nanogaps were generated within each single huddle. The as-fabricated substrates are also tunable by varying the anodizing conditions and metal deposition material/thickness. Substrates produced using this scheme were evaluated by absorption spectra measurements and SERS detection of series of adsorbed molecules. Finite-difference time-domain (FDTD) simulation was conducted to validate the promising feature of the higher electric field energy density stimulated at the tiny nanogaps which resulted in a regular distribution of “hot-spots”. Finally, biosensing potentials were demonstrated by conducting measurements of four different nucleotides (i.e. AMP, CMP, TMP, GMP at 10M) using silver sputtered substrate without any modification. Its SERS performance in the micron level was also evaluated via line-scan in two orthogonal direction in 10M AMP solution. © The Author(s) 2020. Published by ECSJ. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium provided the original work is properly cited. [DOI: 10.5796/electrochemistry.19-00072]. Uploading "PDF file created by publishers" to institutional repositories or public websites is not permitted by the copyright


Introduction
Raman spectroscopy is a powerful technique in investigating molecular structures and properties based on their vibrational transitions by applying an external electric field. As one of its applications, SERS (Surface-enhanced Raman Spectroscopy) substrates have been studied as a label-free tool for advanced molecule sensing and detection. Preserving the advantages of Raman spectroscopy being a fundamental molecular spectroscopy, SERS can directly investigate molecular structures through amplified spectral fingerprints. 1 Owing to the interplay between the incident light and nanostructured metallic surface, Raman signals can be further boosted for several orders of magnitudes. Thus, a small amount of molecules under the detection limit of normal Raman spectroscopy can be sensitively identified when attached to a SERS substrate surface. This is attributed to the extremely enhanced "hot spots" 2 generated by strongly interacting polarized nanoparticles 3 and the effect of metal-molecule charge transfer mechanism (i.e. "chemical" enhancement 4 ). In molecular biosensing strategies, SERS has been a strong candidate along with other optical methods, such as fluorescence spectroscopy and FT-IR. In comparison, SERS is more versatile than fluorescence spectroscopy considering that it provides additional molecular structural information of the subject. 5 Due to shorter vibrational relaxation times as to the electromagnetic relaxation times in fluorescence-based measurements, 10 2 -10 3 times more photons can be expected to be emitted by a molecule under saturated condition in Raman detection than in fluorescence detection in the same amount of time. 6 More importantly, photobleaching or self-quenching issues of reporter dyes are avoided due to the intrinsic sensing scheme of SERS that investigates the analyte molecule directly. 7 In addition, in terms of sample preparation, Raman spectroscopy including SERS only requires minimal to no preparation and even applicable for direct measurement, which is advantageous compared to FT-IR. 8 With these attributes, SERS has been expected to be able to sensitively detect scant biomolecules directly or with minimal modification.
Advanced SERS sensing applications are principally realized in the form of SERS substrates or SERS tags. 9,10 For low concentration analytes that are difficult to detect in conventional Raman spectroscopy analysis, SERS substrates are expected to enhance the Raman signal through the excitation of the local electromagnetic field generated by localized surface plasmon resonance (LSPR). The remarkably amplified extinction cross-section, which is proportional to absorbance, produced by the LSPR occurrence would induce a greatly increased Raman scattering cross-section of molecules attached to the metallic surface. 11 Thus, enhanced Raman signals from a low-concentrated sample can be recognized.
LSPR has been studied extensively for boosting local electric field intensity in producing Raman scattering enhancement. Since the nature of stimulating LSPR is harnessed by the size, shape, and the distribution pattern of the metallic nanostructures, 12 this in effect prominently influences the sensitivity of the substrate's SERS detection application. 13 As documented in numerous previous studies, extremely close-packed nanostructure that holds narrow gaps between nanoparticles/architectures is the key to realize such "hot-spots" features. 14 Thus, efficient fabrication methods are highly desirable to serve this purpose.
The challenges in achieving highly sensitive SERS substrates possessing nanostructure regularity, reproducibility, overall stability and bio-compatibility for large working area at low-cost still remain. 15 The top-down nanostructure patterning methods (i.e. electron beam lithography (EBL)) and the bottom-up assembling methods (i.e. nanoparticle/colloid immobilization, thermal dewetting based silver plating 16 ) are the well-established means 17 for such needs. Specifically, template-assisted fabrication methods have been combined with both top-down and bottom-up methods to further improve their cost-efficiency in producing/reproducing large working area (up to centimeter-level) with high definition (down to ³10 nm) at low cost. 18,19 Nanoimprint lithography (NIL), as one of the template-assisted fabrication methods, has been reported to produce roughened metallic surface for LSPR induction. 20 There have been reports about SERS substrates produced via NIL [21][22][23][24] and the nanoimprinted nanofinger/nanopillar array is one of the wellstudied structures. By exposing the high aspect-ratio nanostructures to liquid, micro-capillary force promotes the collapse of the nanofingers and in result decrease the distance between two adjacent nanopillars. 21,25 The introduced narrow gaps between the nanostructures promotes the enhancement of the electric field due to the strong coupling, which ensued the enhancement of SERS signal at these nanogaps. 26 Therefore, it is important to produce regularly aggregated nanofingers/nanopillars or huddle array to ensure the even distribution and density of induced narrow-gaps. However, to produce porous pattern as template for nanofinger structure via nanoimprinting, high-cost and time-consuming patterning techniques are still commonly utilized. In this work, the mold pattern is fabricated using anodization; a cost-efficient alternative approach for mold patterning, that also provides the possibility of fabricating hierarchical array readily.
By utilizing anodization, anodic porous alumina (APA) template can be produced that has larger area but at lower fabrication cost compared to EBL method. APA has proven its potential for nanofabrication in various perspectives. This self-organized, tunable porous structure has been utilized for producing plasmonic substrates either directly, or as an assistant template, i.e. evaporation mask, 27 or a nanoimprint mold. 28 The wide branched applications are attributed to its advantages in large area fabrication, structure dimension controllability, and durability as a nanoimprint template. In this study, APA templates were utilized in NIL for nanopatterning the substrates that were subsequently sputtered with gold or silver for plasmon activation purposes. Another advantage of choosing this method for SERS substrate fabrication was the observed formation of a hierarchical porous structure, which consists of multiple subpores within single main-pore, when processing the un-annealed aluminum at general purity for anodization in a relatively short duration time (i.e. less than 6 hours). This phenomenon has been documented before. 29 Since the applied potential linearly governs the main pore diameter and inter-pore distance, the distribution pattern of the huddle array is substantially predictable and wellregulated.
Using APA mold for NIL has been well-established for LSPR substrate preparation in our previous work. 30 The thermal resin utilized in this work is cyclo-olefin-polymer (COP), which is chosen for its preferable characteristics such as high-transparency, highfluidity (in favor of achieving high definition) and also for its hightolerance in different solution, 31 which is considered compatible for targeting possible bio-analytes in different media. Taking advantage of the thermal polymer resin's physical properties, high aspect-ratio nanopillar huddle structure has been fabricated and observed in this work. The structures exhibit nanogaps within the same huddle and the substrate has comparable enhancement factors to other reports. 22,32 Given this method is low-cost, time-saving, and reliable for large area fabrication, an alternative method for fabricating SERS substrate that has huddle array generating nanogaps with relative regularity has been realized.
To demonstrate its applicability and advanced sensitivity in SERS detection, a well-known surface modification molecule, 4-ATP has been successfully carried out on the primarily optimized substrate. This may widen the SERS biosensing capability since 4-ATP can act as SAM (self-assembled monolayers) for bio-analytes immobilization. Moreover, label-free investigation for a group of nucleotides dissolved in ultrapure water without further modification was also performed. Corresponding peaks to each nucleotide adsorbates have been clearly identified from their SERS spectra measured using this primarily optimized substrate after washing and nitrogen purging. As a collection of essential biomolecules, nucleotides are the subunits of nucleic acid (i.e. DNA and RNA). 33 Being able to detect and identify different nucleotides will lead to plentiful biosensing applications such as sequencing and genotyping and for the advancement of this technology. In addition, line-scan in AMP aqueous solution (10 ¹2 M) was also carried out to evaluate its applicability when dealing with a different sensing situation and its regular distribution of "hot-spots" in micron level.
To produce metallic surfaces for plasmonic purposes, nanoimprinted chips (denoted as chip/substrate I, II, and III respectively) were sputtered with approximately 50 nm Au or Ag (for chip I only) using compact sputter machine ACS4000 (ULVAC, Japan). This deposition thickness was set to match our previous LSPR study. 30 SERS spectra from the laser spots were obtained through an objective lens (10©, N.A. = 0.3) via a spectrometer (SR303i with DU970N, Andor) using a 532 nm laser (SUWTECH LDC 1500) for excitation. Laser power was set to be 0.6 mW, all Raman/SERS data were measured through 10 s acquisition time.

Porous mold fabrication for SERS substrates preparation
Three APA templates were employed in this study (denoted as template I, II, and III, respectively). The two-step anodizing strategy has been elaborated in our previous work and presented in Fig. 1. 30 The time duration for the first-step of anodization were all conducted at 80 V in 0.3 M Oxalic Acid solution as electrolyte for 1 hour, at 0°C; followed by an etching session by immersing the templates into an aqueous solution containing phosphoric acid (1.16%, w/v) and chromic acid (5%, w/v) at 70°C for over 20 minutes to remove the aluminum oxide layer. The second-step of anodization sessions in 0.3 M Oxalic Acid solution at 0°C and a pore-widening session via phosphate acid (1.16%, w/v) etching at 40°C procedures were performed following the conditions listed in Table 1. Higher anodizing voltages will result in higher oxide layer growth rate, 34 and shorter pore widening session will result in deeper sub-pore with Electrochemistry, (in press) thicker walls, namely, preserving more sub-pore features. Thus, template I and II were intentionally fabricated using the anodizing condition that will bring hierarchical nanoporous structures while remaining the comparable dimensions. Thus, all 3 templates share the same pitch of main pores; for template I the sub-pores are expected to be smaller and deeper than template II; on the other hand, template III is considered as a reference substrate in this study, the sub-pore feature has been etched off in pore widening session.
The as-prepared APA templates were applied as porous molds in the following thermal nanoimprinting process along with COP film (T g = 136°C) employed as the thermal resin material in this study. The APA templates were rinsed in a mold releasing agent beforehand; 1% OPTOOL (DAIKIN Industries Ltd.) diluted by perfluorohexane (Demnumsolvent, DAIKIN Industries Ltd.). The nanoimprinting process has been described in our previous work. 30 For precaution of imprinting nanopillars with high aspect-ratio, cool-down curing temperature was set to be at 60°C.

Substrates characterization
To compare the different substrates prepared in the previous section, surface features before and after metal deposition were characterized via morphology evaluation using a scanning electron microscope (SEM) (DB 235, FEI Company). Substrates were coated osmium using a coater (Neo-ST, Meiwafosis Co., Ltd. Japan).
To gain a rough reference of the electric field distribution of imprinted nanostructures, a finite-difference time-domain (FDTD) simulation was conducted on FullWave 6.2 (RSoft Design Group, Inc., USA) for a rough reference. The parameters used for FDTD simulation is measured from obtained SEM images using DIPP-Image (DITECT Corporation, Japan).

Substrates screening via R6G detection
Firstly, R6G aqueous solution at 10 ¹5 M was dropped onto 50 nm Au sputtered substrates, chip I-Au, II-Au and, III-Au; and incubated for 30 minutes. The substrates were then washed with running ultrapure water and were dried with nitrogen gas thoroughly. Multiple arbitrary spots from each substrate were measured and the SERS signal intensities were compared. After this screening session, chip I was utilized for further study.

Sensitivity evaluation
Approximately 50 nm silver was sputtered onto chip I and subsequently applied for the detection of 4-ATP aqueous solution. The 4-ATP is well known Raman target molecule 35 and also a bifunctional molecule to form a SAM for biosensing. Different concentrations of 4-ATP aqueous solution in a range of 10 ¹5 -10 ¹8 M were introduced to surface of chip I and incubated for 30 minutes, then MiliQ washed and nitrogen gas dried subsequently.
The results were compared with the performance of as-fabricated chips sputtered with 50 nm gold utilized in 10 ¹5 M 4-ATP detection. For background references, gold and silver sputtered non-patterned COP films were also examined (data not shown).
For enhancement factor calculation, 20 µl of 10 ¹5 M 4-ATP aqueous solution was dropped onto 0.5 cm © 0.5 cm substrate and dried overnight. Five arbitrary spots were measured. In comparison, powder 4-ATP was examined under the same Raman set-up.
To explore the possible applications in biosensing, 4 different nucleotides aqueous solution at 10 ¹2 M were incubated on silversputtered chip I for 1 hour, thoroughly washed, and nitrogen gas blown dried before measurements.

Nanostructure regularity evaluation
Furthermore, for the attempt of testing our chip when dealing with various situations, detection of 10 ¹2 M AMP aqueous solution sample was performed in an orthogonal line scanning manner by employing a digital piezo controller (PI, E-710 4CL). The line-scan was done by collecting SERS signal from 15 spots in each orthogonal direction at 1 µm step.
All analytes were introduced to the sputtered chip surface directly at room temperature without further modification.

Structure evaluation using SEM
Both images taken from bare chips and chips sputtered with Au film are presented in Fig. 2a-c. The metal deposition thickness was set to be approximately 50 nm. The formation of huddles and the subsequent result of nanogaps were confirmed based on careful visual inspection to the SEM images similar to a report but the actual measurement of the gap size was not performed. 36 The difference between chip I, II and III has been determined. Chip I and II both withhold hierarchical structures; sub-pores  Hence, the nanogap generated within the same huddle ( Fig. 2d: chip I, II). On the other hand, chip III only has the hexagonal substructure (footings) but no such secondary pillars (branches) (Fig. 2d: chip III). Chip I exhibited lengthened nanopillar huddle structure compared to chip II that presented more sphere-aggregation like features. Longer anodizing duration increases pore depth hence taller pillar height and more pore-widening etching time enlarges the pore diameter i.e. pillar diameter. This is coherent with the obtained SEM images for both chips that match the anodizing condition performed in porous molds preparation session. From Fig. 2, as the nanopillars grown from the same hexagonal base are closer to each other than to those from different huddles, and due to the lengthened nanopillars in chip I and the flexibility of COP material, the pillars demonstrated slightly tip-bending tendency, which brings subsequently sputtered Au caps within one single huddle even closer thus resulting in narrower gaps. This feature is expected to further enhance the SERS signals.

SERS enhancement comparison via R6G detection
To serve the purpose of verifying the nanoimprinted chips as candidates for SERS substrates, 10 ¹5 M R6G solution was introduced onto each 50 nm thick gold-film sputtered chip: The SERS measurements using each chip (shown in Fig. 3a) were carried out after 30 min of incubation at room temperature, thoroughly washed with miliQ, and then dried by Nitrogen gas purging. R6G as an analyte, has been widely utilized as a SERS performance indicator. 37 Averaged signal intensity (N ² 4) from each gold-sputtered chip, chip I-Au, II-Au and III-Au, has been summarized in Fig. 3b. Judging from the R6G SERS signal, chip I-Au and chip II-Au, which contains nano-pillar huddle feature showed distinguished stronger enhancement compared to chip III-Au, which does not share such hierarchical structure. This result has clearly proven that the nano-pillar huddle array, which generates predetermined nanogaps, is the essential key element to boost the enhancement in such nanoprinted nanopillar chips.
No significant peak was observed on non-patterned COP films either sputtered with gold or silver (data not shown). This confirms that our produced nanoimprint chips can serve as a SERS substrate.
In this preliminary screening session, Chip I showed better signal enhancement than chip II (Fig. 3a, b), which is considered due to its slightly different dimensions (pillar size, inter-pillar distance within the same huddle, etc.). To verify this speculation, an FDTD simulation was performed to roughly recreate the E-field enhancement situation in each structure.

LSPR and FDTD
A basic simulation was conducted to obtain a more straightforward understanding of the electric field energy density spatial distribution stimulated on these two chips, chip I and chip II. To recreate the nanostructure for simulation, parameters such as pillar diameter and inter-pillar distance within the same huddle were measured from the obtained SEM images using an image analyzing software (DIPP-image, DITECT, Tokyo, Japan).
The inter-hexagonal-cell distance D int (Fig. 3c) is generalized as linear to the applied anodizing voltage, U, 38 where k is the proportionality constant of approximately 2.5 nm/V. Hence, the inter-cell distance for substrates fabricated using 80 V voltage is approximately 200 nm, which is in agreement with our SEM observation results. Noticing that our substrates only own relative regularity (inter-cell distance μ 200 nm), and there are several factors that would affect the measured dimensions, such as Electrochemistry, (in press) Os coated layer thickness, SEM resolution, and analyzing software resolution, the measured parameters are only considered as rough references to the actual fabricated nanostructures. The measurement errors and the structural inhomogeneity are admissible since the simulation was solely to investigate the tendency of electric field energy density distribution. In addition to the dimension estimation, LSPR spectra (Fig. 3d) were measured and compared with FDTD simulated absorption spectra results to ensure an approximately simplified structure recreation (Fig. 3c, e) drawn via CAD, as the absorbance spectra of substrates indicate their capability of plasmon excitation, i.e. SERS excitation, in certain wavelength range, and this property is determined by the nature of their metallic nanostructure. Furthermore, as silver nanospheres have been reported providing better enhancement with 532 nm laser than gold nanospheres, 39,40 for the attempt of further boosting the enhancement factor of chip I for current 532 nm laser set up, 50 nm thick silver-film sputtered chip I was also included in the LSPR measurement and FDTD simulation. The LSPR absorbance spectra were measured via an optical set-up using transmission mode (Fig. 3d). According to the experimental spectra, substrate I sputtered with gold presented absorption peak at 593 nm; while the same imprinted substrate sputtered with 50 nm silver stimulated absorption peak at 465 nm, and substrate II-Au showed a peak at 660 nm. The comparison between experimental data and simulated spectra will be discussed later in this section. Taking consideration of how the dimensional features of the nanostructures determine their LSPR spectra; there is a slight shoulder appeared at around 530 nm in the spectrum of chip II, which is a result of the coupling of multiple plasmonic modes due to having extreme close-packed formation or regionally conjunct sphere-aggregation-like structure. Therefore, the simulated parameters were further fitted to ensure the successful recreation of the spectra taken in experiments. The tipbending nanopillar huddle structure is simplified as closer packed dimer array, with slightly narrower inter-pillar distance (Fig. 3c). In chip I, metal-cap diameter was estimated as 70 nm and inter-pillar distance within one huddle is estimated as 80 nm. In chip II, Au-cap diameter was estimated as 84 nm and inter-pillar distance was estimated as 85 nm. The inter-cell distance is estimated as 200 nm in both cases. Thus, the spatial distribution patterns of electric field density of both structures at 532 nm have been revealed via FDTD simulation (Fig. 3d).
The results have shown that clear enhancement has been confirmed within nanopillar huddles rather than between two huddles; Ag sputtered chip I presented higher enhancement factors compared to chip I-Au and chip II-Au. These results are consistent to previous studies. 21,25 However, the 50 nm novel metal deposition seemed to be slightly excessive for chip II, some gaps in adjacent pillars were observed fully covered by gold which limited the formation of nanogaps. These overfilled gaps decreased the signal enhancement and the available surface for molecule attachment. This is considered to be the reason why chip II-Au did not present a strong enhancement as with chip I-Au in the previous R6G detection.

4-ATP detection
Based on the speculation in FDTD simulation, Ag sputtered (approximately 50 nm) chip I was introduced to further measurements (Fig. 4a, b). For sensitivity evaluation, 4-ATP aqueous solution with concentrations in a range of 10 ¹5 -10 ¹8 M have been incubated on as-prepared chip I-Ag for 30 min then thoroughly washed and nitrogen gas dried before measurements. Energydispersive X-ray (EDX) spectroscopy was performed on bare and Figure 3. Nanoimprinted chip I, II and III were screened via R6G detection comparison and analyzed through FDTD simulation. (a) 10 ¹5 M R6G detection using chip I, II and III (n > 4) and the comparison of their averaged spectra (b); (c) Image of hierarchical structure and approximately recreated 2 D model structure for FDTD simulation (d) LSPR absorbance spectra measured using an optical set-up (experimental) and calculated via FDTD simulation (simulated); (e) Simulation results of electric field distribution for chip I prepared with gold or silver, and chip II prepared with gold; noticing the magnitude bar for chip I-Ag is different for clarity.
Electrochemistry, (in press) treated substrates to confirm the adsorption of 1 mM 4-ATP (Fig. S1). The SERS measurement results of adsorbed 4-ATP molecules conducted on each chip are summarized in Fig. 4a and b. The intensity showed concentration dependency 37 (Fig. 4d). Also, upon response comparison at 10 ¹5 M 4-ATP, Ag sputtered chip I has proven to stimulate further enhanced Raman intensity (Fig. 4c) than I-Au, and has yielded out a limit of detection of 10 ¹8 M. Judging from the LSPR spectra and 4-ATP detection results, Ag sputtered substrate is more preferable for excitation laser at 532 nm. This is in good agreement with the simulated electric field energy density spatial distribution results.
The detection limit at 10 ¹8 M for 4-ATP has proven that chip I-Ag is qualified to be considered as SERS active substrate with moderated sensitivity. 41 In addition, to calculate enhancement factor (EF) of our nanopillar-huddle structured chip, 20 µl of 10 ¹5 M 4-ATP aqueous solution was dropped onto 0.5 cm © 0.5 cm substrate and naturally dried overnight. SERS spectra from 5 arbitrary spots were measured and compared to the Raman spectrum of 4-ATP powder measured (Fig. 3e) using the same Raman set-up.
As the spectra shown in Fig. 4c, Raman peak positions redshifted to 1096 cm ¹1 and 1175 cm ¹1 compared to SERS positions at 1078 cm ¹1 and 1145 cm ¹1 , this phenomenon is considered as the indication of the formation of strong Ag-S band. 42 For 4-ATP detection, both two kinds of enhancement mechanisms (i.e. electromagnetic effect and chemical effect) can be studied. As Ag-S band forming through HS-group binding to the silver surface, charge transfer induced chemical enhancement (b 2 type mode) 35 can be observed at 1145 cm ¹1 , 1307 cm ¹1 , 1395 cm ¹1 , 1435 cm ¹1 , 1580 cm ¹1 . For electric field effected enhancement, a 1 type mode intensity can be obtained at 1078 cm ¹1 and 1196 cm ¹1 .
Enhancement factor (EF) estimation was conducted through Eq. (2), 43 in which SERS intensities at about 1078 cm ¹1 and at about 1145 cm ¹1 is roughly 10 times and 60 times, respectively, stronger than the corresponding Raman intensities measured. Electrochemistry, (in press) EF ¼ For the sampling molecule amount comparison ( ARaman ASERS ), in the measuring area (887 nm in diameter), A SERS was estimated to be 2.96 © 10 6 . Considering the laser spot diameter d = 887 nm, the penetration depth is estimated to be l = 2d = 1774 nm; with the density of 4-ATP at 1.2 g/ml, A Raman is calculated at about 6.3 © 10 9 within the detected solid sample area. 42 The molecule amount difference is estimated to be at ARaman ASERS % 2:1 Â 10 3 . Therefore, the EF is expected to be at least 2 © 10 4 ³ and 1.2 © 10 5 ³ for a 1 -type vibration mode and b 2 -type vibration mode, respectively; given the other peaks (i.e. 1395 cm ¹1 , 1435 cm ¹1 ) are apparently far more enhanced than the calculated values, and the limit of detection for 4-ATP molecule was yielded at 10 ¹8 M. Herein, this substrate is considered to present a moderate SERS performance. 16,41,44 Further enhancement can be expected by altering excitation laser wavelength (Eq. (3) 45 ), varying anodization voltage and duration or varying metal deposition thickness. 46 Given the results that the optimized laser wavelength should be slightly blue-shifted to the peak position appeared in LSPR spectrum, hence for our current SERS-active chip, stronger electric field enhancement can be expected once switching to a shorter laser wavelength (i.e. 473 nm).
As a bifunctional molecule, 4-ATP is applicable for label-free high-density protein nanoarray to play the role as a reporter in biosensing. 47 With the success in detecting 4-ATP at a low concentration, more applications based on protein recognition can be attempted to perform on this substrate.

Nucleotides detection
The application of Raman spectroscopy for nucleotides and DNA structure characterization has been attempted for over decades. It has always been a strong appeal to researchers to realize all kinds of detection and analysis of DNA molecules intrinsically with minimal modification work. Here described the use of our imprinted hierarchical SERS substrate (chip I sputtered with silver) that provides profuse scattering cross-section at an easy manner. Mili-Q washed and Nitrogen gas dried chips absorbed with 4 different types of nucleotide molecules after 1-hour incubation in corresponding nucleotide solution (dissolved in ultra pure water) were detected without further preparation. As the results shown in Fig. 5, relevant significant peaks are marked for each nucleotide. 48 These marked spectra features assign to relevant DNA base are in good agreement to the results listed in mentioned literature (Table 2). Also, as the peak of CMP at around 796 cm ¹1 and TMP at around 799 cm ¹1 (ring breathing mode) are too close to each other. Therefore, additional peaks were marked for identification purposes. Similar measured SERS signal was obtained at multiple arbitrary spots then averaged as basis of a normalized reference. Although for some measurement spots, there are tiny wavenumber shifts that have been observed, which can be considered as the result of different molecule surface orientation at different spots. 49 Since these small  Electrochemistry, (in press) shifts do not compromise the spectra identifying features, the validation of this SERS substrate applying for DNA detection has been established. Furthermore, in regards to realizing a more straightforward and rapid detection, AMP sample was also measured in ambient solution. The signal is detectable in less than 5 min of incubation (data not shown). A total of 29 spots with horizontal and vertical line scan measurement at 1 µm step was performed to showcase the in-solution detection performance and to verify the regularity of our nanoimprinted substrate. Judging from the obtained line-scan measurement results and their corresponding calculated RSD (relative standard deviation) (Fig. 6), which is less than 20%, the regularity of the nanogap formation in the aforementioned nanoimprinted SERS substrates using self-organized hierarchical APA template has been verified. 50 According to previous reports, employing shorter excitation laser wavelength (i.e. UV-SERS at 260 nm 51 ) is better for nucleotide detection, therefore again by switching to shorter laser wavelength (i.e. 473 nm), further enhancement on DNA base detection can also be expected.
The sensitivity of the fabricated SERS device for the model analytes was lower than other reports. [52][53][54] This is related to the low occurrence of interaction to huddle array as the concentration of target analyte decreases. Fabricating a different nanopillar array pattern that increases the area density of huddles can increase the sensitivity. Conversely, if the molecular orientation can be controlled, the SERS signal can be tuned. This can be made possible through electrical or chemical intervention which can be an extension of this study. More importantly, our substrate can be proven more versatile for detection compared to the reported methods. The dependence to gold particle concentration in a solution to form aggregates in the liquid or on a substrate requires careful and technical handling of reagents. Also, an enrichment step is needed in these steps that need 10 min to 1 hr. In this study, the fabricated substrate can detect the target faster (less than 5 min) with simple handling procedure; simply pour the solution and then dry. In addition, the large area of the fabricated substrate can be exploited for parallel and multiple detection of several targets which cannot be done in other reports.

Conclusion
A cost-efficient deployment that supplies large area fabrication by exploiting the self-organized hierarchical APA template with a one-step nanoimprint lithography utilizing a durable thermal resin (COP) has been demonstrated in this work. The nanopillar huddle sub-structure fabricated via this straightforward method has been validated as an active SERS substrate for adsorbates detection that demands no further modification. Their electric field enhancement behavior stimulated by LSPR has been studied via FDTD simulation. The calculated energy distribution patterns have proven the distinguished electric field enhancement occurs within each huddle rather than between two huddles. This feature indicates a regular distribution of "hot-spots" that can be expected via this fabrication method. Comparison between the two substrates with different dimensions revealed that chip I, with lengthened nanopillar-huddle structure sputtered with silver, demonstrated remarkable SERS signals for 532 nm laser. Label-free detections of different molecules in different concentrations have been successfully achieved using chip I-Ag. It is considered as a moderately sensitive SERS substrate based on its performance in 4-ATP detection: limit of detection is tested at 10 ¹8 M, and the EF is estimated to be roughly 10 4 -10 5 . The regularity of such a substructured substrate has been verified with horizontal and vertical line scan in detecting AMP solution.
The EF is not considered possible for single molecule level detection and the homogeneity at nano level is difficult to achieve with the current method. However, based on the tuneable, highthroughput, reproducible and cost-efficient merits of this fabrication scheme, along with the detection results for various molecules, we Figure 6. Line-scan in AMP solution ambient with 15 points each (29 points in total) in two orthogonal directions are measured for nanostructure regularity verification. The RSD value calculated for SERS signal intensity collected at 29 points is under 20% level, which indicates a reliable nanostructure regularity.
Electrochemistry, (in press) believe that this nanoimprinted huddle array is a strong candidate for general analytical purposes.

Supporting Information
The Supporting Information is available on the website at DOI: https://doi.org/10.5796/electrochemistry.19-00072.

Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.