Guided Wrinkling of Hierarchically Structured Nanoporous Gold Films for Improved Surface‐Enhanced Raman Scattering Performance

Plasmonic nanostructured metals have many advantages for applications in high‐performance surface‐enhanced Raman scattering (SERS) spectroscopy. In particular, unique designing nanostructures with bicontinuous ligaments surrounded by cylindrical voids with tunable dense pores from a few to hundreds of nanometers can be utilized for the high‐performance SERS‐active substrate. Here, a fabrication strategy is reported to prepare hierarchically arranged micro/nanostructures of wrinkled nanoporous gold (WNPG) films, which involves laminating of the dealloyed Au film on the heat‐shrinkable shape‐memory polymer film and geometrical modulation of the substrate. As a result, the various types of WNPG films are crafted with a remarkable density of cracks in the structured surface area. Specifically, the WNPG films consisting of multilayered overlapping features are explored and used as the SERS‐active substrate. This dual porosity coupled with localized surface plasmon resonance estimated by numerical simulation in a suitable model of bicontinuous ligaments is found to be the core mechanism for the enhancement of SERS sensitivity, which quantitatively characterizes the “hot spots” from the surface to interlayers. These suggested characteristic features are fully assessed by applying a series of dye molecules and DNA strands on the prepared SERS substrate, demonstrating the enhanced intensity of the Raman scattering signals on the optimized WNPG surface.


Introduction
The subject of dealloyed metal produced by selective dissolution involves a broad range of solid solution alloy systems. Dealloying has long been investigated as a selective dissolution method that often results in the formation of open porous sponge-like structures with three-dimensionally percolated inert materials from the removal of the templated elements. [1][2][3][4] Among the different model systems, dealloyed nanoporous gold (NPG) is a commonly used nanostructured material, which contains bicontinuous ligaments surrounded by cylindrical void channels, extending the open pore size from a few to hundreds of nanometers. [5,6] Owing to the excellent characteristic features, such as electrochemical properties, active catalytic behavior, chemical stability, biocompatibility, and high specific surface area, the NPG material provides useful functionalities for a wide range of fascinating applications. [7][8][9][10] Moreover, intrinsic structural nanogap-enabled configuration exhibits unusual optical responses and influences the propagation of electromagnetic waves. In principle, the light-driven collective oscillations of electrons on the NPG surface allow one to achieve unexpected physical properties not only from the dealloying technique but also from the reconstruction of the surface morphology. [11,12] In this context, as one practical strategy, surface-enhanced Raman spectroscopy (SERS) can be combined with the tailored NPG film and inelastic scattering of the molecules from the interaction of incident light to enable the detection of specific molecules in the proximity of the nanostructured Au surface. [13] Since the enhanced Raman scattering offers a unique opportunity to probe the bond vibrations of adsorbing molecules, molecular detection can be used for sensors as a single molecule analytical tool, especially in the bioengineering research field at the range of sufficiently low concentration. [13] The SERS effect is mainly attributed to the close interaction of the light source with the nanostructured surface, generating localized surface plasmon resonance (LSPR) that confines an enhanced electromagnetic field in the "hot spots". [14] Therefore, the design of the well-studied "hot spots" in the NPG film is crucial to preparing the SERSactive substrate with an affordable and tunable capability. [15] Although some other strategies have been presented for SERSactive substrate by conventional lithographic methods (i.e., electron beam, interference, and nanoimprint lithography), [16][17][18][19] these customized manufacturing process of engraving nanostructured surface generally involves equipment-based expensive techniques and should consider the trade-off in resolution with throughput. [16,20,21] Compared to this, the improved SERS properties can be demonstrated with the enhancement of the LSPR in the bicontinuous porous structure by dealloying the NPG film for a suitable structural formation using appropriate fabrication processes. [2,10,13] However, there are still some shortcomings to overcome. As previously reported, the results of numerical simulations suggested that it is difficult to optimize the signal enhancement at specific wavelengths on the structured NPG films. [22][23][24][25] Because the optimization of enhanced scattering Raman signals highly depends on the pore size, shape, and composition of the metal alloy, the changes in the internal pore distance according to the ligament structural elements should be well matched to the incident light optical windows. While exploiting the coupling efficiency of molecules for electromagnetic excitation at the pumping source wavelengths is difficult, the use of the short wavelength in spectroscopy may be one option for nanostructure NPG films. [26][27][28] For example, the sensitive wavelength in the Raman spectroscopic system can be implemented using a light wavelength of 532 nm, as probed in charge-coupled devices incident on nonperiodic materials, in which the enhanced Raman signals were obtained more than seven times compared to the wavelength of 785 nm transformed from the fourth power of the excitation energy. Thus, new types of NPG material with open porosity can be promising candidates to serve as SERS-active substrates, comparable to other materials produced by lithographic, synthetic, or self-assembly approaches. In general, because the Raman spectroscopy indicates a weak scattering signal acquisition at the sized molecules as low as the cross-section of ≈10 −30 -10 −25 cm 2 , the intensity of enhanced Raman scattering signals in a specific re-gion can be improved by up to ≈10 10 times, thereby even single molecules can be detected for a wide range of niche applications, such as high-sensitivity biosensors. [29] Here, we report a developed strategy to fabricate hierarchically arranged NPG films with highly wrinkled micro/nanostructures in a controllable manner via dealloying of Au-Ag bimetallic alloy by corrosive electrolyte and subsequent transfer on the shapememory polymer film for implementation of SERS-active substrate. By fully utilizing the shape-memory polymer (SMP) film as a thermally shrinkable working substrate, the attached dealloyed NPG films were reconfigured into macroscopically wrinkled structures that contain nanoscale open porosity with a remarkable density of cracks and buckling on the structured surface area. [30][31][32][33][34] The main experimental concept of our presented work is to craft intriguing topological features by controlling the external field of mechanical force onto the NPG films underlying the SMP substrate, in which the sophisticated surfaces are covered with SERS-enabled nanoporous structures and adjacent broken ligaments. Importantly, several parametric factors for the generation of the wrinkled surface of NPG films were carefully tuned by the strict guidance of polymeric carrier film and geometrical modulation in the shrinkage process. The dealloying time and relevant selective dissolution process were also one of the important conditions to understand porosity formation during the sample preparation, which ensured the reproducibility and sensitivity of the SERS substrate. [35,36] A dual porosity coupled with LSPR was able to compensate for the non-uniformity of sensitivity in the quasi-periodic structure by increasing the SERSactive surface. Moreover, we performed a numerical simulation using the finite-difference time-domain (FDTD) method to investigate the correlation between the size distribution of nanopores and the enhancement of the electromagnetic field, elucidating the characteristics of the "hot spot" at the surface of the structured NPG film. In particular, the multilayer overlapping features were formed inside the cracks as a result of the shrinkable NPG film that exhibited a nanogap-enabled configuration; we interpreted the enhancement of SERS sensitivity in this intrinsic structure through the FDTD method. Notably, by applying the dye molecules on the prepared SERS substrate, we confirmed the improved intensity of enhanced Raman scattering signals of more than ≈9.68 × 10 6 for a specified wrinkled NPG surface ( = 632.8 nm). In addition, we evaluated SERS activity using fluorescent dye-modified DNAs to determine immobilization and subsequent hybridization, demonstrating potential use for biomolecule detection. We believe that our simple yet robust strategy is highly advantageous to producing SERS-active substrate in a rapid and scalable high-throughput production for multiscale nanostructure.

Manufacturing Process and Characteristics of Wrinkled NPG Structures
Figure 1a describes a schematically illustrated sequential process (from (i) to (vii)) to prepare wrinkled NPG structures with a regularly ordered configuration. In our main experimental design, we used a binary alloy metal film (i.e., forged metal film, Au 37.4 -Ag 62.6 at.%, Sepp Leaf) to create a porous Au structure by Figure 1. a) Schematically illustration of the sequential process to produce hierarchical arranged micro/nanostructures of WNPG Films; (i,ii) selective dissolution of Au-Ag bimetallic alloy, (iii-v) PMMA solution spin-casting, the undercut of SiO 2 by HF solution, subsequent transfer on SMP film, (vi) placing with fixture jig, and (vii) PMMA removal after the heat treatment. b) WNPG structures generated by three different in-plane compressive modes from isotropic contraction (Mode I), uniaxial contraction (Mode II), and two-step contraction (Mode III). c) Representative SEM images of the aligned arrays of the controlled wrinkled PMMA film and WNPG film; bilayer of PMMA/WNPG film and magnified surface image of the WNPG (lower images). www.advancedsciencenews.com www.advmatinterfaces.de facilitating the selective dealloying of the Ag component in the main Au matrix. At the initial stage of the experiment, the binary alloy film was floated over a ≈70% nitric acid to dissociate Ag with a slightly elevated temperature and optimized time, (i); the inset digital image shows an originally provided material, cut into the size of 1 × 1 cm 2 . Next, the appropriately dealloyed NPG film was conformally placed onto a SiO 2 /Si substrate, being flattened with high integrity to prevent noticeable cracking, which also allowed for easy delivery of NPG films, (ii). The NPG film/SiO 2 /Si substrate was then coated with polymeric carrier film (≈1 μm) by using a spin casting of polymethylmethacrylate (PMMA) toluene solution, (iii). In the following, the PMMA/NPG film could be separated from the solid substrate via undercut of sacrificial SiO 2 layer (300 nm) through the etching with the HF solution, so thus the floated freestanding PMMA/NPG film readily transferred to the SMP working substrate, as shown in (v). Finally, highly wrinkled structures consisting of NPG film were produced by the use of a fixture jig to firmly fasten the ends of the PMMA/NPG film/SMP substrate, as schematically drawn in (vi). In this fabrication setup, we intentionally confined the wrinkled surface area (the original width of SMP: 14-20 mm) where the ordered features were expected, as previously reported. [37] For example, this simple tool controlled the uniaxial compressive force on the laminated films in the thermal shrinkage step that induced wrinkled microstructure with fixed dimensions (vii). In addition, to fully adjust the shrinkable ability of the SMP substrate with mechanically constrained NPG film, three different modes were engaged, depending on the geometrical modulation of the in-plane compressive forces, as illustrated in Figure 1b. In fact, since the production of the commercial SMP involves an arrangement of polymer chains (i.e., disentangled polystyrene chains) that are not completely compressed, [30][31][32] we postulated that free-shrinkage (i.e., isotropic Mode I), mechanical confinement (i.e., uniaxial Mode II) in specific directions, or additional contraction after the uniaxial mode (i.e., Mode III) might affect the final morphological features of the NPG films. Therefore, various types of wrinkled NPG (WNPG) structures were easily crafted by the guidance of PMMA film, associated with the directions of the compression. As appeared in Figure 1c, a set of representative scanning electron microscope (SEM) images shows the regularly controlled surface wrinkles of NPG film as a result of Mode II after the heat treatment of the PMMA/NPG/SMP film and subsequent removal of PMMA. The guidance effect of the PMMA was clearly observed during the successive processes (upper panels), and it was found that the highly defective wrinkles formed without applying PMMA film (see the details in, Figure S1, Supporting Information). Moreover, it should be noted that the selection of organic solvent is critically important in removing PMMA because the common solvent can simultaneously degrade the SMP film and induce deterioration of NPG films formed on the SMP substrate. [37] Thus, we used selective acetic acid at this crucial step to avoid the wanted swelling or degradation of the working SMP substrate. As one of the collected information, the side-cut cross-sectional SEM image obviously reveals the thickness level of each film in the wrinkled structure from surface to bottom (i.e., PMMA/NPG/SMP film in the lower left panel). Notably, a highly magnified SEM image also represented nanoscale porous features all over the surface area at the wrinkled structure.
As shown in Figure 2, we observed the characteristic features of the surface on the wrinkled PMMA/NPG/SMP film to carefully scrutinize the wrinkle formation process with some other controllable parametric correlations. The SEM images in Figure 2a represent the PMMA surface geometries at the shrunk states right after the heat treatment, according to the proposed different modes, to produce wrinkled NPG structures with the aid of PMMA guidance; the PMMA thin films were wrinkled in arbitrarily oriented configuration (Mode I) and highly aligned arrays that have similar features with a certain wavelength in the case of Modes II and III. In particular, the wrinkled structure of the PMMA/NPG/SMP film was attracted to our attention for systematic analysis as a long range of regular surface structures were found over the entire surface area ( Figure S2, Supporting Information). Thus, we evaluated the tunability of the wrinkled structure as a function of wavelength and the provided PMMA film thickness, in which the sample width can also affect the subtle in-plane compressive force due to Poisson's ratio during the thermal-induced shrinkage process. As tested and measured, a graph in Figure 2b summarized the wavelength of WNPG on the SMP film at each PMMA film thickness range after the removal of PMMA, where the concentration of the PMMA solution for spin casting depicts the corresponding film thickness. As the PMMA film thickness increases, the wavelength of the wrinkles linearly increases in the measured regime with the increased sample width. As reported earlier, these results can be attributed to the different distributions of the engaged compressive forces, depending on the laminated conditions. For example, the contraction of SMP film was ≈73% in the central region when the sample width was provided as 14 mm ( Figure S3, Supporting Information). In contrast, as the increased sample width reached 20 mm, the centered contraction could be decreased up to ≈60% mainly due to the lower applied stress on the SMP film under the same constrained condition (i.e., the distance between the ends).
In our experimental scheme, because the PMMA guidance is a critically important parameter to regulate the wrinkled structures and plays a key role in the development of the periodically ordered arrays of NPG films, we focused on the physical analysis, especially for the results from the Mode II, based on the previous studies. First, we considered a stress instability-driven thermal shrinkage behavior on the SMP substrate, compressing the PMMA/NPG film. The well-known energetic analysis of a sinusoidal vertical deflection model was adopted for our materials system, so thus the critical stress for wrinkling ( w ) can be defined as [38] (1) whereĒ f andĒ s are the plane-strain moduli of PMMA/NPG film and SMP substrate; here, the plane-strain modulus was defined asĒ = E∕(1 − v 2 ), (E: Young's modulus, : the Poisson's ratio). The Poisson's ratio of each SMP substrate and PMMA/NPG film was estimated as ≈0.33 and ≈0.35, respectively, [37] which was collected by the nanoindentation tests and standard micrographic measurements. In this physical condition, the wrinkles of the PMMA/NPG layer attached to the SMP substrate can be spontaneously generated when the shrinkage-induced stress ( ) exceeds w . [38] Whereas, if the is higher than buckling critical stress ( B ), a buckle-delamination of the PMMA/NPG film occurs by the delamination. In other words, by applying the compressive stress to the SMP substrate, the PMMA/NPG film was naturally buckled, forming freestanding wrinkles across the film surface. Thus, the critical stress for buckling initiation is equal to that for the freestanding PMMA/NPG wrinkle, which can be described by the following equation: [38]  where t f and b are the thickness of the PMMA/NPG film and the half-width of the delamination, respectively. Indeed, as the compressive stress on the PMMA/NPG/SMP film was f > B , only the buckle-delamination can occur as presented in Figure S1 (Supporting Information); this parametric condition is also independent of the substrate properties. Therefore, the shape and the wavelength of the PMMA structure directly affect the formation of the microstructure of the WNPG layer. Based on the equation suggested by Whitesides and Stafford, the elastic modulus of a WNPG with PMMA/NPG film can be extracted, [39,40] where is the critical wavelength by the calculation of assuming a sinusoidal waveform z (x) = Asin(2 x/ ) of wrinkle (or buckling) instability guided by PMMA film. [39] At a low strain field regime (≈<10%), the sinusoidal model can be estimated by the linear function of t f (i.e., the thickness of PMMA/NPG film, as follows: Conclusively, if the modulus of elasticity of the PMMA/NPG film does not change significantly, the wavelength of the WNPG wrinkles can be determined by the thickness of the spin-coated PMMA film. As displayed in Figure 2c, we fitted the wavelength of the formed wrinkles as a function of the PMMA/NPG film thickness according to Equation (4), in which the linear fitting curves reflect the modulated strain by the SMP width. Compared to the case of the isotropic shrinkage (i.e., ≈50% strain), the maximum strain in the transverse direction on the SMP film was increased by confining the film with a fixing jig. For example, under the condition of 20 mm (solid red line), the strain applied to the PMMA/SMP film can be increased by ≈17% from the original state, which was used as the correlation coefficient. As a consequence, the graph indicates that the measured wavelengths of the wrinkles deviate slightly from the calculated regime, but the specific case of SMP width of 20 mm was in good agreement with the proposed model. The deviated fitted results might be derived from the topographic shape of PMMA/NPG/SMP film due to the high compression of up to ≈73%, as appeared in Figure 2d, which is close to the sharp curvature rather than the originally suggested sinusoidal waveform. In other words, we acknowledge that modeling the sinusoidal vertical deflection as a function of the WNPG/SMP film requires a linear combination, such as (z (x) = Asin(2 x/ ) + Bsin 2 (2 x/ )…), from which solution gives the curvature non-linearity. The first term of the equation can be defined as vertical deflection caused by the planestrain moduli, [40] but the second term is difficult to specify the main source in our materials system at this moment. However, it can be presumed that this occurred due to the subtle competition between the tensile and compressive stresses in the microenvironment of the formed wrinkles. [41] By close observation of the cross-sectional SEM images, we found the clear guide capability of PMMA film on the WNPG formed on the SMP substrate. Surprisingly, as shown in Figure 2e (colorized in yellow (i.e, WNPG layer) and green (i.e., wrinkled SMP film)), the PMMA film pulled both WNPG and the surface of the SMP film in the thermally induced shrinkage stage; see also inset SEM image, blue-colorized PMMA was captured right before the removal process. In the step of transferring the PMMA/NPG film to the SMP film, the contact area between the WNPG and the SMP film was completely flat although there were some surface undulations, but no other interfacial chemistry or pre-pattern transfer was involved. This observation was obviously distinct from the results from the shrinkage process without the use of the PMMA film as presented in Figure  S1c (Supporting Information), in which delaminated NPG on the SMP film surface was only observed. It should be noted that the formation of the wrinkled surfaces of NPG film with a certain wavelength was highly dependent upon the guidance of the PMMA film. For this, clear evidence was presented in Figure 2e (middle panel), as measured at the boundary between the regions of WNPG film and SMP surface, where the surface of SMP film was rigorously guided to the wrinkled structure with sharp contrast. A simple further experiment, presented in the right panel of Figure 2e, confirmed this assumption by thermal treatment of the PMMA/SMP film (without NPG film), where the partially removed PMMA on the SMP film exhibited fully covered wrinkles at the boundary area. Therefore, the newly developed experimental approach applying the interfacial guidance of PMMA film enabled the production of intriguing structural features of wrinkled NPG films.
The overall trend of the morphologies from the results of three different modes was observed by SEM as presented in Figure 2fh. As noted, the hierarchically arranged structures were featured within the geometrical characteristics. In the case of Mode I, similar to the initial morphology of the PMMA guidance, flower-like winding peaks were produced as a result of the isotropic contraction of ≈50% from the originally provided size (Figure 2f), in which a magnified SEM image indicates nanoporous surface structure formed on the edge of the peak with a large amount of the propagated cracks. Interestingly, after the complete shrinkage by heat treatment, the increased thickness levels of the SMP film were measured (i.e., ≈100%) ( Figure S4, Supporting Information). On the other hand, as carefully characterized earlier, the uniaxial mode restricted the engaged samples in one direction by the thermally induced shrinkage up to ≈73% at the in-plane with a width of 14 mm (vertical expansion of ≈115%), as shown in Figure 2g. As systemically measured, the regularity of wrinkles was diminished by the decrease in the wavelengths that are defined by PMMA thickness ( Figure S5, Supporting Information), which is mechanically understandable due to the anisotropic nature of the SMP substrate and NPG film. [30,34] A highly magnified SEM image indicates that the major cracks were formed along the wrinkle-tip and in between the wrinkles with the highly packed configuration of the pores. In the case of Mode III, after separating the initially heat-treated working substrate from the fixture jig, additional thermal plastic deformation was applied to intentionally force it to a more contracted state. In other words, the isotropic contraction was imposed on the unentangled regions, inducing the directional shrinkage of the polymer chains. Since the initial shrinkage was almost processed in the structure, the contractile stress was mainly involved in parallel to the wrinkles, which destroyed pre-existed wrinkles, as presented in Figure 2h. As expected, a large number of bucklings or partially pop-up distortion was observed from the surface with propagating cracks in the delaminated structure (inset and lower panel). Note that a shorter secondary heat treatment time was applied for Mode III because the size of pores can be changed by thermal diffusion.

Numerical Simulation of Hot Spots for SERS-Active Substrate Consisted of WNPG Films
With the ease of access and facile steps in preparation, the micro/nanostructured WNPG substrates were facilitated as a SERS substrate that may be applicable to specific sensors with high sensitivity by the enhanced LSPR on the unique nanostructures. As displayed in Figure 3a, the example prototype of the SERS substrates, consisting of WNPG film, were fabricated on the handling slide glass, attached with a size of 5 × 5 mm 2 . As characterized earlier, the main concept of our structural approach was centered on creating spontaneously formed rough nanoscale cracks in the wrinkled or buckled geometries, including the featured rearrangement of porous structures, which are expected to induce an evolution of the improved physical properties. Notably, due to the propagating cracks, the nanoscopic creased forms were hierarchically configured on the surface of the WNPG. The intrinsic mechanical property was not easily predictable because the NPG film was a 3D networked matrix in conjunction with the necked cylindrical pores (i.e., Au ligaments in the open-cell foam). However, as an experimental result, the modulus mismatch between the NPG film and SMP substrate in the shrinkage process derived copious dangling protrusions that may act as LSPR-enabled hot spots. Although some variations naturally existed with location, most of the surface areas were fully covered with several types of cleaved architectures in a nanoscopic view, as classified in Figure 3b, (i-iii). By large compressive stress, the major trend of the cracks was observed with random orientation following the ridges and tips of the wrinkles within the transformed film at a distance of ≈500 nm or less, (i). Interestingly, the parallelly formed cracks in the uniaxial mode collapsed into crevasse-like structures with the competition of the global compressive force in the shrinkage and expanding tensile stress at the tips of wrinkles, which cleaved the tips of the wrinkles, as appeared in (ii). Most importantly, the cleaved edges of the NPG film in most structures were embedded adjacently into a layered configuration, depicting the unique reconstructed folding step-shaped surface of the NPG film (iii), as marked by the red-dotted box. This multilayered configuration might be attributed to the lateral dissolution by considering the relatively rapid penetration of nitric acid during the dealloying stage within the bulk film, simultaneously undergoing the layer-through vertical etching. [35,42,43] It should be noted that the dealloying time (i.e., 5 min) was one of the important parameters to produce an optimized thickness of NPG in this structural reconfiguration (i.e., delamination to the single layer and overlapping). Therefore, based on the above observed structural formation, we postulated that the multilayered structure at the light-irradiated NPG surface as schematically illustrated in (iv), at which the representative top surface and thickness of a single layer were respectively imaged in (v). As shown in Figure 3c, the three different contraction modes generated a large number of cracks and cleaved layers with the delamination of the NPG films. Thus, the SERS effect on this defined surface area may be improved with the unique morphological features. We believe that this nano/microstructure of the layered NPG film, locally appearing in rough steps with cracks, can be greatly advantageous in increasing the specific surface area and enhancing the Raman scattering signals by responding as a passage for the targeting molecules.
To scrutinize the layered NPG film, we designed a numerical analysis, considering the dual porosity coupled with LSPR in the quasi-periodic structure, as displayed in Figure 3d. [1,22] For this approach, the thickness of a single NPG layer was set as ≈110 nm in the total thickness of ≈300-350 nm; the entire multilayered NPG was composed of a three-layer configuration. Based on the experimentally confirmed information, we performed the FDTD numerical analysis assisted by 3D mesh modeling of simplified geometrical elements to minimize the uncertainties, such as rounding error, data uncertainty, and truncation (e.g., numerical computation error at discontinuous inflection points). The geometrical elements in the layered NPG film were composed of the ellipsoidal islands with connected necking features (i.e., bicontinuous ligaments), as graphically described in (i), where the enlarged SEM micrograph and the modeled image are compatible in the structural features; here, each dimension of ellipsoidal islands was settled by Gaussian random distribution to systematically compare with pore positions in the surface of the WNPG film ( Figure S6, Supporting Information). As seen in the background analysis (ii), the uniform mesh generation in Cartesian grids allowed the XY plane to set the object position in the simulation box, thereby appropriate binarization of the NPG was visualized; the green and blue colored areas represent Au and air, respectively. Since the cross-sectional dimension ranges of wrinkles were ≈10 μm or more, the interference between the wrinkles was negligible for the wavelength of 632.8 nm in this situated condition. Therefore, the simulation box depicted the wellmatched layer (i.e., artificial absorption layer for the wave equations) as the boundary condition (Note S1, Supporting Information described the detailed technical process of simulation). In the following, an optimized analysis of the electric field distributions was performed by applying the experimentally measured dimensional factors (e.g., ellipsoidal island: ≈80 nm, necking: ≈15 nm), as shown in (iii), in which the local field enhancement factor was rendered in the backscattered E along the −k direction (inset). As denoted, the local field enhancement factor (|E/E 0 | 2 ) describes the relationship between the intensity of the Raman polarizability tensor of the probe and the scattering geometry; typically assumed square of the localized electric field in the normalized state. [29] Furthermore, the enhancement factor from the SERS signal is proportional to the fourth power of the excitation energy (i.e., |E/E 0 | 4 ), which can be the product of the excitation (i.e., local field enhancement factor) and the emission, this is analogous to the modified spontaneous emission induced by dipole-dipole coupling. [44] The contour plot of the electric field indicated that randomly distributed but strong "hot spots" were excited from the quasi-periodic structure, allowing the prepared material to perform accurate SERS activity; the computed maximum enhancement factor from the SERS signal was found to be ≈4.35 × 10 7 on the SERS-enabled surface of the WNPG film. Although this demonstration confirmed that the surface areas of the WNPG film are possibly covered with dense "hot spots", an optimized open porosity can be determined by tuning the Figure 3. a) Prototype of the WNPG-based SERS substrates. b) hierarchically configured nanoscopic architectures; (i) cracks, (ii) crevasse-like structures, and (iii-v) layered configuration of the NPG film. c) SEM images of WNPG films with cracks, cleaved layers, and buckles. d) Modelling of 3D mesh for a multilayer overlapping feature of dual porosity coupled with LSPR in the quasi-periodic structure; binarization of the nanopores (blue) and Au surface (green) in the simulation grid, and the corresponding electric field distribution of dealloyed structure. e) A graph of the maximum electric field enhancement at the hot spots and integrated |E/E 0 | 4 intensity in the selected area as a function of the pore concentration.
experimental elements for the SERS application associated with the dealloying conditions. Thus, the pore size and concentration in the configured WNPG structure were considered, according to the degrees of dealloying time. In the measured surface area, the pore concentration appeared constant after a specific time (i.e., ≈24 h) although the diminished thickness by the excessive dealloying process was accompanied, as summarized in Figure  S7 (Supporting Information) (e.g., the pore concentration of the dealloyed samples for 30 min was estimated to be ≈32.47%). A graph in Figure 3e indicates the correlation of the maximum Figure 4. a) 3D FDTD simulations of |E| 2 electric field distributions of WNPG film and b) cross-section of the simulation box. c) Representative Raman spectra on the R6G bulk, pure Au thin film, NPG thin film, and WNPG-based SERS substrate. d) Reproducibility test measured at the different locations along with wrinkle ridges. e) Representative SERS spectra for CV from the WNPG films produced from the three different modes. f) Representative SERS spectra for BCB depending on the dealloying time. enhancement factor of the SERS signal (i.e., |E/E 0 | 4 ) at the "hot spots" and integrated |E/E 0 | 4 intensity in the selected area (i.e., 600 × 600 nm 2 ), as a function of the pore concentration in the layered WNPG films. Based on this result, the optimized pore concentration can be in the range of ≈21-37% (the solid black line in the graph), which could be produced under dealloying conditions of ≈5-30 min, corresponding to the maximum SERS signal enhancement at the hot spots. The Z-axis-dependent variations of the layered WNPG films were described by the error bars, and the simulation parameters were adjusted to the diameter of the ellipsoids and the width of the interconnected necks in the range of ≈60-120 nm and ≈10-20 nm, respectively. We compared the generation of "hot spots" in all conditions with the same variance of the NPG distribution model and the coordinate values of the ellipsoids, as suggested in Figure 3d. The lower panel in Figure 3e visualized the FDTD results according to the changes in the percentage of the pore volume (vol. %) at the simulated surface area. At the boundary condition, most of the ellipsoidal islands (120 nm) were connected without necking with decreased internal pores to 5 vol. %; at the same time, the calculated |E/E 0 | 4 was sharply dropped to the level of single layer NPG film. The hot spot sites appeared from the specific pore concentration (i.e., 21 vol. %) and progressed up to the occupied volume range of 56 vol. %. Based on these results, it is important to note that similar intensity levels of the "hot spots" were observed within the pore concentration range from 21 to 46 vol. %, which is approximately matched with the maximized integration values of the |E/E 0 | 4 , as shown in the graph. The simulation grid on the contour plot of the electric field (|E/E 0 | 2 ) was added to the opacity as guidance with the open porosity variations. Notably, the individual contour plots indicate that the location of the excited "hot spots" dramatically emerges although the coordinate values of the ellipsoid are fixed in electric field simulations according to the changes in the pore concentrations. The results of FDTD simulations related to a range of pore concentrations so far clearly suggest that the controlled dealloying process can progress buckling dissolution to produce a specific level of pore distribution used as a SERS substrate with appropriate signal enhancement.

Evaluation of Selected SERS Substrates Designed for Molecular Detection
The SERS-enabled substrates consisting of WNPG films were characterized in a view of multilayered overlapping features, assisted with the thermally shrinkable SMP film. Figure 4a describes a simulated 3D-stacked plot result that visualizes the electric field of the entire nanostructure of WNPG film, in which the "hot spots" exist even in the interlayer; here, the electric field associated with vertically coupled plasmonic response was also integrated from the top to the central layer (refer to the cross-section of the simulation box in Figure S8, Supporting Information). In this specific microenvironment, enhanced Raman scattering can be expected when the target molecules are adsorbed at the provided "hot spots" in the widespread areas between the vertically layered NPG films. Although the density of hop spot sites becomes decreased to the lower surface layer due to electromagnetic wave reflection, the hierarchically quasi-periodic nanostructures of the NPG layer on the top exposed surface are mostly involved in SERS activity. [45] The vertical contour plot of the |E/E 0 | 2 electric fields in Figure 4b illustrated the supporting role of the back side of the NPG layers created by wrinkling formation. The cross-sectional nanogap was randomly distributed because the elliptical pores of the overlapped NPG layer were located at oblique coordinates in the Y-axis. Indeed, the vertical coupling of LSPR appeared over a widespread area of ≈50-100 nm in the Z-axis. However, we demonstrated that the electromagnetic field between the interfaces from the top to the underlying layer was significantly enhanced, allowing WNPG films to enable layer-based hotspots.
For a set as this precisely configured structure, we evaluated the SERS activity for the quantitative analysis by using molecules diluted in a certain concentration. For example, Figure 4c shows the SERS activity on the Rhodamine 6G (R6G, 10 −6 m) by applying pure Au thin film, flat NPG film, and WNPG film produced from Mode I. The Raman intensity measured from the surface of the WNPG film was ≈32 times higher compared to the non-structured NPG film and ≈2440 times higher than that of the Au thin film. On this, the SERS enhancement factor (EF) of the WNPG structure was calculated as ≈9.68 × 10 6 (refer to the details in Note S2, Supporting Information), which indicates a slightly lower value rather than the prior simulation results. To interpret this phenomenon, it should be acknowledged that the hot spots of an electric field in the WNPG structure are placed in the gap formed by two Au ellipsoids with dielectric properties. As previously reported, when the Au ellipsoids with a radius of curvature of 30 nm approach a gap of 2 nm, a hot spot of ≈10 8 level occurs in the center. [29] We postulate that the steric hindrance between the NPG layer and the target molecule makes it difficult for the fluorescent group to penetrate through these confined spaces, so thus partial SERS activity is likely to be generated at the hot spot in the vertical area. Another example molecule was also implemented on the WNPG-based SERS substrate produced from Mode I, using crystal violet (CV), as shown in Figure 4d, where excellent reproducibility was evaluated when the Raman spectra were collected at multiple locations along the ridge of the wrinkles. As measured at the Raman shift of 1615 cm −1 , the average signal intensity of ≈950 000 counts was detected for the C-C ring stretching mode, and the standard deviation was calculated as ≈110 000 counts within the error range of ≈87%. The additional SERS substrates prepared from Mode II and Mode III also showed high reproducibility ( Figure S9, Supporting Information), which confirmed the reproducibility of ≈88.4% through repeated experiments. Figure 4e represents Raman spectra with CV molecule (10 −8 m) applied on different types of WNPG substrates, suggesting excellent hot spot-derived SERS activities. Interestingly, in the same experimental condition, the extracted Raman spectra on the WNPG structures crafted from different fabrication methods yielded slight variations in a signal enhancement mode in molecular detection. These results can be attributed to the mechanical stress-induced nanoscale plastic deformation of the internal porous structures that delicately affect the Raman scattering signals although the originally provided NPG films were in similar shapes and distributions. In addition to the structural effects, we surveyed the influences on the Raman spectra by applying brilliant cresyl blue (BCB) molecule that has high chemical stability in water and superior biocompatibility, easily combining with acidic DNA or RNA. [36] Since an absorbance peak for BCB is at ≈630 nm with an outstanding extinction coefficient (over 20 000 m −1 cm −1 ), some other aspects of SERS activity could be tested in this experiment. Spanning the dealloying time as displayed in Figure 4f, the highest SERS signals were measured from the short period of the dealloying process (i.e., 5 min), and the Raman scattering signals gradually decreased as increased the dealloying time with the observation of the morphological degradation in all WNPG films. As addressed in the previous section, a key role to control the nanoporous structure critically lies in the dealloying time by the corrosive electrolyte, correlated with the composition ratio of Au-Ag in a binary metal alloy. The relatively long dealloying time generally derived aggregation and an unwanted coagulated state of remaining Au, as reported previously. [35] However, in our case, the SERS activity was correlated not only with the original porous features but also with the structured WNPG films, which was an important parameter to maintaining a consistent Raman scattering signal enhancement when designing an NPG-based SERS substrate. Thus, our approach suggests that the improved SERS effect (i.e., EF = ≈9.68 × 10 6 ) can be attributed to the multilayered overlapping characteristics of WNPG films with numerous cracks. Compared to other recent reports that showed a signal enhancement mechanism using NPG structures, [20,46] our results imply that independent optimization on the mechanical external field by using SMP films and pore concentration by dealloying time should be delicately controlled.

WNPG-Based SERS Substrates for DNA Hybridization
To extend our approach in biomolecular sensing, we performed the detection of DNA hybridization utilizing the WNPG-based SERS-active substrate. [47] For a simple demonstration, a target single-stranded DNA was complimented with thiol-modified probe DNA by the immobilization on the Au surface, as schematically described in Figure 5a. The first step was to immobilize a thiol-terminated DNA on the WNPG substrates with a specific sequence designed as 5″-AGTAC CGTGA GGGAA AGGCG-SH-3″. After the reduction of the sulfide bonds and column purification, the probe DNA strands dispersed in potassium phosphate buffer solution were self-assembled on the SERS substrates for ≈24 h in a sealed chamber; the excessive unbound DNA strands were rigorously washed with sodium dodecyl sulfate (SDS) buffer solution after incubation. [48,49] Next, the complemental target DNA strands (3″-TCATG GCACT CC-CTT TCCGC-Cy5-5″) were readily hybridized to the probe DNAimmobilized substrate to detect the spontaneous paring by the Raman scattering enhancement with the help of the tagged dye molecule (i.e., Cy5). After the rinsing step, the NPG surface was observed by optical microscope to ensure Cy5-labeled hybridization. A representative fluorescent optical micrograph displays tiny spots of anchored dye molecules located all over the surface area as a result of the complemental DNA binding on the provided NPG film ( Figure S10, Supporting Information). Based on this method, the optimization of pore density with a specific shape was performed to be used as a reliable SERS substrate by tuning the dealloying time, as shown in Figure 5b. As appeared, the inset SEM images indicate apparent differences in the surface morphologies, and the SERS signals evaluated the basic performance according to the level the pore density. Similar to the previous cases, a successful enhancement was detected for the sample only from 5 min dealloying time. As indicated in the graph, assigned main peaks validated the vibrational modes of the central p-conjugated chain (i.e., ring vibration mode) at 798 and 913 cm −1 , C-H in-plane bends at 1173 cm −1 , aromatic ring modes at 1296 cm −1 , the methane chain deformation mode at 1366 cm −1 , and stretch modes of C-N stretch mode overlapped with aromatic ring vibrations at 1603 cm −1 , respectively; a lowlevel injection of the applied Cy5 shows a broadening of Raman scattering on the aromatic ring vibration mode at 1480 cm −1 . [50] The relatively long dealloying (i.e., 60 min) degraded the arrangement of pores with coarsening and aggregation due to the excessive expansion, which may block the formation of hot spots and not allow the extraction of the SERS signals from the labeled DNA hybridization. [51] With this background information, the hot spots can be ideal when located in the layered structure in the WNPG film. As illustrated in Figure 5a, by the fact that SERS signals can only be generated when Cy5 dyes are anchored in the nanogap without steric hindrance hanging on the DNA strands, careful measurements were carried out on the SERSactive WNPG films (Figure 5c). Surprisingly, different Raman signals were observed depending on the wrinkled structures, and as shown in the graph, significantly increased signals of the major peaks (approximately fivefold) were found from the SERS substrate produced from Mode I compared to the case of Mode II and III. Repeated tests showed the same trend observed for hotspot-responsive DNA hybridization. These slightly different Raman scattering signals suggest that the delicate positioning of the DNA strands at the hot spots presented in the vertical nanogap is important. On this, we postulated that surface properties such as wettability for the DNA-containing liquid could also be critical in the hybridization step and affect the SERS sensitivity.
As previously reported, appropriate immobilization of the target molecules is a crucial step to verify the SERS activity on the specific substrates. In our case, we used a drying droplet deposition to observe the DNA hybridization, manifesting hot spotenabled WNPG films where the colloidal DNA strands can wet into the wrinkled surface with a specified structure (i.e., Wenzel state). To consider the wrinkled structure and related parametric morphological features, we briefly scrutinized wetting conditions with a droplet, as presented in Figure 5d. The main idea is to categorize the surface state associated with hot spots. As well-known, the wettability of a microstructure can be mainly determined by capillary pressure, which varies with geometric interfacial parameters such as height or periodic features. [52,53] For estimating a capillary pressure in the WNPG structure, the surface tension involved in the wrinkled structure can be considered from the existence of the internal void space. According to the Young-Laplace relation, the static capillary pressure at the surface of the microstructure can be described by the following equation that includes an interfacial relationship between the structured surface and the air cushion in pores: [54] where ΔP is the Laplace pressure (i.e., the capillary pressure at the liquid interface), LA is surface tension at the interface of liquid/air (i.e., 72.8 × 10 −3 N m −1 for water), and r y and r x are the principal radii in the direction of each axis (upper right panel). Assuming a simplified repetitive structure with apparent wrinkles in a similar range over specific directions, each capillary pressure on the different modes of the WNPG films can be classified by the Young-Laplace relation. For example, the shape of wrinkles for Mode I is close to a randomly distributed post structure, so thus r x ≈ r y . On the other hand, in the case of Mode II, the value of r y can be approximated to 1/r y ≈ 0 due to the guidance wetting of highly aligned microstructures. The value of r for Mode III is ranged in an intermediate level between Mode I and Mode II. Under this condition, the wettability is slightly varied, depending on the spatial arrangement of WNPG film, with the external surface curvature of the droplets. Thus, when the spatial distance on the microstructured WNPG film is distributed in a similar range, the differences in the capillary force determine the magnitude of the advancing contact angle ( adv ), as illustrated in the lower left panel of Figure 5d. Macroscopically, when the average distance of internal voids underlying the WNPG films is set to d, the capillary pressure (P c ) can be roughly developed as: [53] where is a structural constant determined by a curvature of the meniscus at a given point (e.g., in the case of an ideal structure in Mode II, the value of ≈ 2). Together with the Young-Laplace relation, it can be inferred that in Mode I has a higher value than in Mode II (i.e., 2 < < 4 in Mode I). Through measuring the values of the adv for each sample ( Figure S11, Supporting Information), we indirectly estimated the effect of height (h) in the formed wrinkles approaching the capillary equilibrium state affected by the existence of underlying void space, which may influence on the water-repellent properties (i.e., hydrophobicity). During wetting on the provided surface, the contact areas between liquid/solid were shrunken by the change of the adv . If the internal voids in the WNPG film were partially filled with the applied liquid, the adv is lower than 90 o , that is, Wenzel or metastable Cassie-Baxter state, as illustrated in Figure 5d. Therefore, the degree of penetration of the meniscus into the internal voids can be derived from the Cassie-Baxter equation for hydrophobic/hydrophilic transition (i.e., cos app to be positive or app ≤ 90 o ). [55] cos app = rf s ( cos adv + 1 ) − 1 ( 7 ) where app , f s , and r are the macroscopic apparent contact angle and the fraction of solid at the surface, and the roughness factor is defined by the Wenzel equation, respectively. Under these criteria, we guide the range of the wettability with the coefficient . As shown in Figure 5e, the relationship between the solid fraction and the coefficient in each WNPG structure was summarized as a function of geometric factors that were extracted by the SEM images and optical micrographs ( Figure S12, Supporting Information). Guided fitting curves indicate the coefficient by the modulated solid fraction and the fixed r; in which the representative values are marked with asterisks in the graph, such as ≈76°( Mode I), ≈104°(Mode II), and ≈100°(Mode III); for more information, refer to the relation between the advancing contact angle and coefficient , as described in Figure S13 and Note S3 (Supporting Information). As discussed above, in view of the app , the effective wettability of the surface properties for Mode I was superior to other samples, so thus DNA hybridization was processed with high efficiency by propagating DNA-containing meniscus into the "hot spots". It is worth noting that the ring-shaped sedimentation (i.e., coffee-ring effect) was suppressed when the sessile droplets are dried in a sealed chamber by controlling the receding meniscus on the WNPG structure. [56,57] Because the coffee stain is a capillary flow in the edge of a droplet caused by the difference in evaporation rate at the contact line, slow evaporation at the liquid-gas interface in the evaporating fixed droplet derives endothermic heat flow that circulates the colloidal DNA strands from the droplet surface to the interior (i.e., Marangoni convection). [58,59] The surface tension-driven recirculation prohibited unwanted aggregates at the contact line, resulting in uniform deposition of the target DNA strands to the immobilized probe DNA strands on the WNPG substrate. [51] On the basis of the brief structural analysis of the wettability of the WNPG surface, we confirmed that specified structures can be suitable for an optimized surface state that utilize droplet deposition based on homogeneous DNA hybridization method and successful biological molecule detection.

Conclusion
In summary, we report facile yet robust strategies to produce hierarchically arranged NPG films with wrinkled micro/nanostructures by utilizing selective dealloying and transfer methods on the thermally shrinkable substrate (i.e., SMP film) for preparing SERS-active substrates. In the fabrication process, a polymeric carrier PMMA film strictly guided the wrinkled structure of NPG films and reconfigured them into SERSenabled nanoporous structures. Particularly, controllable pore size with densely constructed structures was successfully tuned with remarkable cracks and buckling on the structured surface area geometrically and modulated by mechanical in-plane www.advancedsciencenews.com www.advmatinterfaces.de compressive force during the shrinkage step. Parametric physical factors such as optical FDTD simulation were thoroughly evaluated, prior to the use of the WNPG films as SERS-enabled substrates. This result suggested that multilayered overlapping features at the cracks played a decisive role in the SERS-activity of WNPG films that underwent the optimized lateral etching in a selective dissolution before the shrinkage process. Specifically, an experiment of numerical simulation estimated a dual porosity coupled with LSPR in a suitable binarized network modeling of bicontinuous ligaments for the enhancement of SERS sensitivity, which quantitatively characterized the "hot spots" from the surface to interlayers. Finally, the prepared SERS substrates were fully evaluated by applying a series of dye molecules, determining the improved intensity of enhanced Raman scattering signals on the optimized WNPG surfaces. We also facilitated the prepared WNPG-based SERS-active substrates to use in biomolecule detection (i.e., DNA hybridization), demonstrating potential use for biosensors by tailoring topographic features with the optimized nanoporosity. We envision that our proposed strategy in this work toward the tunable micro/nanostructured WNPG can be extended its potential to other applications such as the detection of biomarkers or biofunctional analysis of small molecules, supported by the rapid and scalable high-throughput production for multiscale nanostructure. [60,61]

Experimental Section
Fabrication of WNPG Structure: Commercially available Ag-Au alloy film (Monarch, white gold-Au 37.4 Ag 62.6 at.%, Sepp Leaf, New York) was used as received; the thickness (≈150 nm) and the surface were measured by the AFM (NX10, Park Systems Co., Suwon, Korea). The grain structure of the Ag-Au alloy foil, as measured by electron backscatter diffraction, had a strong (001) texture. To dealloy the Ag-Au film, it was floated over 70% nitric acid (Junsei Chemical Co. Ltd., Japan) at 30°C for 5 min to 5 days; the bimetallic alloy films were transformed into an NPG structure after the dealloying process. The NPG was rinsed with deionized water to remove residual nitric acid and then transferred to a SiO 2 /Si wafer. As a sacrificial substrate, a SiO 2 /Si wafer was cut into 15 cm × 15 mm 2 , which was cleaned in ethanol, acetone, and deionized water and then dried with nitrogen gas. PMMA dissolved in toluene was spin-coated on NPG/SiO 2 /Si wafer by casting at 3000 rpm for 30 s (PMMA concentration: 10 wt.%). A sacrificial SiO 2 /Si wafer was undercut in 5% hydrofluoric acid solution, and PMMA/NPG was transferred to SMP substrate (polystyrene, Grafix Plastics, Cleveland, OH, USA, t = 250 μm). Next, the PMMA/NPG/SMP films were placed on a fixture jig in which the width of the SMP substrates was varied from 14 to 20 mm (the length of the SMP substrates was fixed at 30 mm). The compressive force was engaged by thermal treatment at 150°C for 40 min. After the shrinkage, the PMMA film was then removed by the selective organic solvent (i.e., acetic acid) not to dissolve the polystyrene SMP substrate. The wrinkle wavelength and height of micro/nano wrinkled structures were confirmed by SEM (Carl Zeiss Supra 40VP, Oberkochen, Germany).
Numerical Simulations: The electromagnetic field in NPG layers was simulated by the FDTD solver from ANSYS/Lumerical FDTD software. As the constituent material of the simulation structure, a sponge structure composed of pure Au was assumed. Information on the physical properties of Au such as complex permittivity was used in the library, the CRC Handbook of Chemistry and Physics, inside the FDTD software. The Au sponge structure was composed of a mixture of ellipsoid and necking of a cylindrical bar. The curvature equation of the cross-section of the cylindrical necking was assumed to be f (x) = ax 2 − bx + c. The size of the necked ellipsoid was randomly arranged to have a Gaussian distribution. The center position of the necking and the curvature variables were artificially modified. In the x-, y-, and z-directions, a perfectly matched layer was set as the boundary condition of the simulation box. Background material was set as air. The spacing of the simulation grids was set for the x, y, and z axes to be a maximum of 2 nm. The pump pulse was injected from the Z-axis backward, and the electric field's vector was in the direction parallel to the x-direction. For total-field scattered-field source was an incident in the z-direction with an area of 600 × 600 × 300 nm 3 (i.e., N x × N y × N z = 300 × 300 × 150 of total grid points) at room temperature. The wavelength of the pumped laser pulses was 632.8 nm.
Raman Measurements and Detection of DNA Hybridization: R6G, CV, and the BCB dyes were used in SERS quantitative analysis (all from Sigma-Aldrich Co., St Louis, MO). The dye material was dissolved in an ethanol solute to prepare a solution. The concentration of the dye solution was adjusted from 10 −6 to 10 −8 m. After dropping 20 μl of the dye material solution, it was completely dried in a desiccator. Raman spectra (UNIRAM II system, UniNanoTech Co. Ltd, South Korea) were measured using a wavelength of 632.8 nm laser. The Raman microscopy laser spot size was ≈8 μm in diameter. The measurement time was 30 s, and a total of 10 times were accumulated and proceeded for 300 s. To evaluate the applicability as a biosensor, the Raman spectra were investigated according to adsorption, and hybridization was measured using dye-attached DNA on WNPG. For DNA, two types of complementary 20-mer oligonucleotides DNA which were custom-made (Genotech Co., Ltd., Korea) were used. Among them, a thiol functional group was attached to one of ssDNA to facilitate a reaction with the WNPG sheet, and the other was custom-made to have Cy5, a fluorescent dye that exhibits Raman signals, as a terminal group. After the thiol-modified probe DNA (10 μm) was treated with dithiothreitol to reduce the disulfide bond and was purified using NAP 5 columns, the probe DNA strands dispersed in potassium phosphate buffer solution were self-assembled on the SERS substrates for ≈24 h in a sealed chamber (All reagents were obtained from Sigma-Aldrich).
Wetting Properties Characterization: Wetting properties of the WNPG surface were analyzed using a contact angle analyzer (SmartDrop, Femtofab Co. Ltd., Seongnam, South Korea). The WNPG sheet was placed on a flat place, and ≈2 μL of deionized water was dropped each time and then observed through a measuring camera, and the average value was obtained by measuring at least 10 times for each specimen. Advancing contact angles for water droplet was measured on the tilting plate of rate 0.16 o s −1 . Since the contact angle was very sensitive to surface contamination, tertiary deionized water was used, and the wrinkled structure was washed with ethanol to prevent contamination of the contact surface and dried, and then measured.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.