1. Introduction

Raman scattering is a fast and powerful analysis tool for uniquely identifying chemical and biological molecules, where a small fraction of incident photons experience inelastic scattering processes with energy shifts after interacting with molecular species. The energy differences between the incident and scattered photons are determined by the vibrational states of molecular species, and therefore can be directly utilized to identify the molecules [1]. Due to the intrinsically weak Raman scattering effect, methods have been pursued to enhance the intensity of the Raman scattered light and facilitate the molecular identification of much lower concentrations of molecules. Most commonly, Raman active molecules are attached on a rough metal surface to interact with the localized strong electromagnetic field and take advantage of plasmonic effects to achieve surface enhanced Raman scattering (SERS) [2]. The localized surface plasmon (LSP) effect, an effect of collective electron oscillations, leads to a strong localized electromagnetic field in the vicinity of metallic nanoparticles, and is therefore commonly used to enhance the Raman signal. Numerous LSP based metallic nanostructures have been intensively investigated for SERS, such as silver or gold-coated silicon nanopores [3, 4] and nanopilars [5], arrays of nanoparticles [6, 7], nanodomes [8], and nanogaps [9], where the significantly enhanced Raman signals rely on the laterally confined strong electromagnetic field in the deep sub-wavelength regions of the various structures. Such small lateral features (e.g., within a few nanometers) require either a relatively complex fabrication process or a search for ‘hot spots’ on the substrates.

Surface plasmon polaritons (SPP), propagating plasmonic modes at a planar noble metal and dielectric interface, are vertically confined on a planar surface and result in an overall strong field at all the molecular adsorption sites on the surface. Therefore, activation of a SPP enables large area uniformity of the SERS signal. A strong SPP effect can be engineered in a straightforward manner at designated wavelengths via optimization of the coupling configuration. Several SERS templates have taken advantage of the SPP field enhancement in combination with a LSP effect to further amplify the SERS response [10–12].

Nanoporous gold (NPG) has become a particularly attractive SERS substrate due to its low-cost and straightforward fabrication, as well as its relatively uniform SERS enhancement across large areas [13, 14]. As-prepared NPG consists of a nanoscale sponge-like porous structure that provides a large surface area for molecule binding and supports a LSP over a broad wavelength range [15]. Various modified NPG films, such as NPG with nanoscale wrinkles [16, 17], fractures [18], and ultra-small pores [13, 14], have been reported to promote improved SERS intensity due to the increased electromagnetic field intensity, or improved LSP activity, in the vicinity of the nanoscale pores and sharp features. Smooth as-prepared NPG films fabricated as metamaterials have appropriate optical constants to support a SPP in the near infrared (NIR) wavelength range, and both prism- and grating-coupled SPPs on NPG films have been reported as refractive index-based plasmonic sensors with high detection sensitivity [19, 20]. Recently, we demonstrated 2D grating patterned NPG (P-NPG) films that, for the first time, utilized both LSP and SPP effects to dramatically enhance the already strong SERS intensity of as-prepared NPG [21]. In the present work, we carefully study the contribution of the SPP effect on the strong SERS intensity measured from the grating-type P-NPG substrates. We theoretically and experimentally demonstrate that the strength and position of the SPP resonance and the resulting SERS intensity can be systematically controlled via tuning of the grating depth and duty cycle on a given NPG film. Activation of a strong SPP effect and its strong coupling with the LSP effect on P-NPG samples with appropriate grating dimensions gives rise to SERS enhancements up to 0.5 × 10⁸ compared to non-enhancing substrates for benzenethiol molecules.

2. Experiments and methods

2.1 Synthesis of NPG

The synthesis of NPG films was carried out in accordance to methods reported elsewhere [21, 22]. Briefly, a 160 nm thick Ag₅₀Au₅₀ (molecular weight ratio) alloy film is dealloyed in 70% HNO₃ for 15 min to remove the silver, followed by a rinsing procedure with deionized water. The free-standing NPG film is then transferred to a 1,6 hexanedithiol-modified gold film on a silicon substrate for stable support and the subsequent imprinting process. SEM images reveal the the NPG films have 15-25 nm pore openings after the dealloying process.

2.2 Fabrication of P-NPG SERS substrates

P-NPG SERS substrates consisting of as-prepared NPG regions (denoted as ‘NPG’ in Fig. 1.) and densified-NPG regions (denoted as ‘D-NPG’ in Fig. 1.) were fabricated via the recently developed technique ‘direct imprinting of porous substrates’ (DIPS), which enables high resolution and high throughput patterning of porous materials, such as NPG, in a simple and low-cost process [23]. DIPS was performed on NPG substrates using pre-fabricated silicon stamps, ~15 mm² in area, containing inverted 2D grating patterns. DIPS produces a locally densified NPG continuous network surrounding 2D NPG grating ridges (unstamped regions), thus converting the original planar NPG film into a 3D NPG structure with subwavelength gratings (Fig. 1.). The role of 2D grating patterns is to enable the direct coupling of free space incident light into a SPP mode at normal incidence; this coupling approach is compatible with standard Raman microscopes and other compact SERS measurement tools. Importantly, 2D P-NPG substrates are polarization-symmetric at normal incidence, and therefore a SPP mode can be launched on the substrate surface with unpolarized light and no strict sample alignment.

 

Fig. 1 (a) Schematic configuration of 2D grating-type P-NPG substrates consisting of NPG regions and D-NPG regions (f: duty cycle, Λ: period, h: depth). (b) Plan-view SEM image showing a large area P-NPG film with 650 nm period and ~60% duty cycle. (c) – (e) Zoomed-in plan-view SEM images of 650 nm period P-NPG films with decreasing duty cycles of ~90%, 75%, and 60%, respectively.

Download Full Size | PDF

2.3 SERS measurements

SERS measurements were performed with benzenethiol, a Raman-active molecule that readily forms in a self-assembled monolayer (SAM) on gold surfaces. All P-NPG samples were immersed in 10 mM benzenethiol in ethanol solution for 1 hour to form a SAM of benzenethiol, and the samples were subsequently rinsed with ethanol and dried with nitrogen. A standard 500 µm thick polydimethylsiloxane (PDMS) cell containing pure benzenethiol was also prepared as the control, non-enhancing substrate for the calculation of the enhancement factor (EF) of the SERS substrates [11]. SERS spectra were measured at normal incidence with a DXR Raman microscope (Thermo Scientific) utilizing a 780 nm diode laser at 0.9 mW power. A 10 × magnification objective (NA = 0.25) was used to focus the laser beam and collect the SERS signal. The low magnification objective is preferred due to the larger laser illumination spot size (~3.1 µm), which contains many periods of the P-NPG pattern and promotes high reproducibility and uniformity of the SERS measurements. Data was collected with an integration time of 10 s and averaging of 5 scans.

2.4 Available surface area for thiol binding

Reductive desorption of benzenethiol molecules from gold into a deaerated 0.5 M KOH (aq) solution was performed in order to quantify the accessible surface area provided by a stamped, nanoporous gold substrate when compared to a planar gold substrate. Voltammetric scans were performed with a Gamry Instruments Reference 600TM Potentiostat using a Ag/AgCl reference electrode, a platinum mesh counter electrode, and a working electrode [22] that consisted of either a smooth gold-coated silicon wafer (control) or stamped, nanoporous gold (15 min dealloying) on a gold-coated silicon wafer, both of which had been exposed to a 1 mM solution of benzenethiol in ethanol for at least 1 hour. The potential was swept from 0 to −1.2 V with a scan rate of 200 mV/s.

3. Results and discussion

3.1 P-NPG grating design parameters

A schematic of the P-NPG substrate with 2D gratings is shown in Fig. 1(a), where Ʌ is grating period, h is grating depth, and f is duty cycle, which is defined as the ratio between grating ridge width and the grating period. In order to investigate the effect of tuning the grating parameters on the SERS response, 2D P-NPG samples with fixed grating period Ʌ = 650 nm, various duty cycles ranging from ~15% to 90%, and various grating heights ranging from ~20 to 70 nm were fabricated. The grating period of 650 nm was chosen based on prior studies that demonstrated that this period most effectively excites a SPP on 2D P-NPG samples using excitation light near 780 nm [21]. Note that the grating period needs to be correspondingly tuned for other excitation wavelengths based on the dielectric constant of NPG.

Figure 1(b) shows a representative top-view scanning electron microscopy (SEM) image of a 2D P-NPG substrate with Ʌ = 650 nm and f = 60%, which reveals the uniformity of the P-NPG substrates over a large area. Figures 1(c), 1(d), and 1(e) show a series of zoomed-in SEM images of 2D P-NPG substrates with duty cycles of ~92%, 77% and 65%, respectively, revealing the surface morphology of both the densified-NPG (D-NPG) grating grooves and unstamped NPG grating ridges.

3.2 Role of plasmonic effects

The LSP effect on similar as-prepared NPG films has previously been experimentally demonstrated via measurement of the far-field absorption [15]; supporting simulations of the near-field distribution, where a regular porous structure was used to approximate the random porous structure, confirmed a strong electromagnetic field inside the pores [14]. In order to theoretically examine the presence of a SPP, in addition to the LSP, on the proposed 2D P-NPG structures at normal incidence, the specific material optical properties and grating geometry must be considered. In contrast to the prior study focused solely on the LSP effect where NPG was treated as a complex nanostructure, in the present work, both the NPG and D-NPG films are considered as uniform metamaterials in the investigation of the SPP effect since the pore size of NPG and D-NPG is much smaller than the excitation wavelength. Accordingly, the dielectric constants (ε_r + iε_i) of bulk Au, NPG, and D-NPG were first measured using a J.A. Woolam M-2000 VI spectroscopic ellipsometer in a wavelength range of 400-900 nm. Here, large area D-NPG samples were prepared by imprinting a NPG film with an unpatterned 9 mm² silicon stamp. The measured complex dielectric constants of the bulk Au (dashed curve) and NPG (solid curve) film are shown in Fig. 2(a). Compared to bulk Au, the NPG is found to exhibit weaker metallic behavior, especially in the visible wavelength range where the real part of its dielectric constant is not sufficiently negative to support a SPP mode. A SPP mode requires that the real part of the dielectric constant of NPG must be negative with an absolute value exceeding that of the air. Based on the measured dielectric constant of NPG, Maxwell-Garnett theory suggests a value of ~65% porosity for the NPG film; the measured dielectric constants of D-NPG (not shown) suggest lower porosities and therefore stronger metallic behavior than NPG.

 

Fig. 2 (a) Measured dielectric constants of Au (dashed curve) and NPG (solid curve) as a function of wavelength. (b) Simulated far field absorbance curves for 2D grating-type P-NPG substrates (Λ = 650 nm, f = 60%) with various grating depths (h = 20, 40, and 60 nm). Resonances indicate the presence of SPPs. (c) Simulated near field intensity |E|²/|E₀|² within a single grating period and simulated absorbance spectra for 2D grating-type P-NPG substrates (Λ = 650 nm, h = 60 nm) with various duty cycles (f = 30%, 60%, and 90%). (d) Measured absorbance spectra of NPG, D-NPG, and P-NPG. All films exhibit a LSP resonance but only P-NPG exhibits a SPP resonance.

Download Full Size | PDF

Using the measured dielectric constants of NPG and D-NPG films, we accurately defined the NPG and D-NPG regions in the simulation, and then utilized the 3D finite-difference time-domain (FDTD) method to simulate the far-field absorbance of P-NPG substrates with 2D gratings (2D P-NPG) under TM polarized light at 780 nm. As an example, simulated wavelength-interrogated absorbance curves from three 2D P-NPG substrates at normal incidence with fixed f = 60% duty cycle and different grating depths h = 20, 40, and 60 nm are shown in Fig. 2(b). Clear resonances located at 700-800 nm were observed from all samples, with the deeper grating samples showing stronger resonances, which is strong evidence indicating SPP mode activation on each substrate. It is important that the SPP resonance occurs at or near normal incidence so that it can be efficiently activated via a standard Raman microscope. The microscope objective with NA = 0.25 utilized in this work provides a θ = 0 – 14.5° angular window for activating the SPP resonance and collecting the SERS signal. It can be seen from Fig. 2(b) that the resonance strength decreases with decreasing grating height, implying reduced SPP activity and therefore reduced electromagnetic field intensity on the P-NPG surface [24]. We also note that the coupling resonance wavelength is slightly shifted for the different heights, which indicates a shifted wave vector of the activated SPP mode. These wave vector shifts can be explained as the change of the near field distribution of the activated SPP modes. For relatively shallow gratings, the SPP mode can be approximated as a propagating wave existing on a continuous NPG surface, while for relatively deep gratings, the activated SPP modes in fact exist on top of the grating-modulated meta-surfaces. This series of SPP modes associated with relatively deep gratings are due to the localized Fabry-Perot resonances formed in consecutive vertical metallic cavities between neighboring grating ridges [25, 26]. Similar observations of SPP resonance shifts and different resonance depths with a change of grating height have also been reported in previous studies [10, 27].

In order to analyze the influence of duty cycle on the SPP effect, we used the 3D FDTD method to simulate the wavelength-interrogated absorbance for P-NPG samples with a fixed grating depth (60 nm) and various duty cycles (f = 30%, 60%, and 90%), as shown in Fig. 2(c). Note that the SPP resonance continuously red shifts with the increase of the duty cycle. For P-NPG samples with 30% (dashed curve) and 90% (dot dashed curve) duty cycle, SPP resonance peaks are located at ~700 and 1000 nm, respectively, which is not spectrally overlapped with the wavelength of excitation (780 nm) and most of the Raman bands of benzenethiol (800 – 900 nm). The 60% (solid curve) duty cycle sample results in a strong and broad SPP resonance occurring at ~800 nm, which completely covers the wavelengths of excitation and all Raman bands of benzenethiol. The clear SPP resonance or wave vector shift with different grating duty cycles can be explained by the change of near field distributions on meta-surfaces with various duty cycles. The near field intensity of the P-NPG samples with different duty cycles was also simulated by using the 3D FDTD method. The inset of Fig. 2(c) shows the corresponding near field intensities of the three samples at the Stokes wavelength of the 1070 cm⁻¹ band (850 nm). It is clear that the 60% duty cycle sample leads to the strongest near field intensity over a large area molecular adsorption site, which is consistent with our observation from the far-field absorbance spectra and proves that a SPP mode can be efficiently achieved when the grating parameters are designed appropriately. Hence, correct specification of the grating period, height, and duty cycle gives rise to efficient scattering and coupling of light to a SPP mode for a particular wavelength of light. Gratings with inappropriate dimensions will result in a significantly decreased scattering and therefore decreased coupling efficiency of a SPP mode [27]. The theoretical calculations shown in Figs. 2(b) and 2(c) show that both the grating depth and duty cycle influence the resonance wavelength as well as the activation efficiency of the SPP mode, and are correspondingly expected to strongly influence the achievable SERS intensity. Absorption spectra and near electric field intensity were also simulated for these and similar P-NPG structures at an excitation wavelength of 532 nm, another commonly used SERS laser wavelength known to promote LSP activation in NPG films; however, no resolvable resonance was observed, suggesting that the SPP effect could not be efficiently activated on our P-NPG substrates at this wavelength. As shown in Fig. 2(a), the real part of the dielectric constant of NPG is not sufficiently negative at 532 nm to exhibit the strongly metallic behavior required to support efficient activation of a SPP [20].

In order to further elucidate the roles of both SPP and LSP effects in the SERS response of our P-NPG substrates, we experimentally measured the absorption spectra for NPG, D-NPG, and P-NPG films at normal incidence. In Fig. 2(d), broad absorption resonances with similar peak wavelength near 510 nm are observed from all three films indicating the existence of the LSP effect due to the nanoscale porous structure of NPG, which is consistent with prior experiments and simulations [14, 15]. Since the broad LSP resonance ranging from 500 to 900 nm encompasses the excitation laser wavelength of 780 nm, the LSP effect contributes to the enhanced SERS response of NPG-based SERS substrates. Note that although both the D-NPG and P-NPG films show overall reduced absorption intensities due to the densification effect, the LSP condition and its coupling in the smaller densified pores leads to a larger SERS enhancement compared to as-prepared NPG films. A second clear absorption resonance located at ~800 nm is only observed from the P-NPG sample (Λ = 650 nm, f = 67%, h = 70 nm), which indicates the activation of SPP mode that also contributes to the SERS enhancement of P-NPG substrates. The activated SPP subsequently introduces a stronger LSP effect via strong coupling between the two modes since the SPP resonance occurs within the broad range of the LSP resonance [19]. The width of the SPP resonance is strongly dependent on the dielectric constant of NPG; NPG films with a relatively small imaginary part and large negative real part of the dielectric function exhibit narrower and stronger SPP resonances.

3.3 Controlling of SERS response

Based on the promising simulation studies demonstrating that the efficiency of activating a SPP mode and its resonance wavelength can be tuned via the control of various grating parameters, we next experimentally examined the influence of these grating parameters on SERS intensity by fabricating 2D P-NPG samples with varied parameters (duty cycle and grating depth) using the DIPS imprinting technique [23]. With this fabrication method, the grating depth can be easily tuned by applying different pressures ranging from 75 N mm⁻² to 300 N mm⁻², which results in a range of corresponding grating depths from approximately 20 to 75 nm (as determined by atomic force microscopy). The grating duty cycle is adjusted using different stamps. We note that, under the same pressure, the grating depth or porosity of densified regions could be slightly different due to the different grating duty cycles. Therefore, in the grating depth-dependent study, we intentionally reduced the patterned area to 1 mm², while keeping the entire stamp area the same (15 mm²). Since the patterned area is relatively small compared to the surrounding area, there will be negligible variation in the grating depth for samples with different grating structures. In all of the experiments, the grating period was fixed at 650 nm, as was the case for our simulation studies. Figures 3(a), 3(b), and 3(c) show the experimentally measured relationship between SERS intensity, duty cycle, and grating depth at three standard Raman bands of benzenethiol (414, 1070, and 1570 cm⁻¹). Error bars were calculated based on 5 – 10 measurements on different spots across each sample. We found a 10-15% SERS intensity variation across a 0.01 mm² area, and we believe this variation can be mostly attributed to the non-uniformity of the initial NPG film.

 

Fig. 3 (a) - (c) Relationship between SERS intensity and duty cycle with respect to various grating depths for benzenethiol bands 414 cm⁻¹, 1070 cm⁻¹, and 1570 cm⁻¹, respectively. (d) Summary of SERS spectra of benzenethiol adsorbed on three optimized substrates with various grating depths. (e) SERS intensity distribution of benzenethiol adsorbed on 2D P-NPG substrates with h = 70 nm and various duty cycles ranging from ~15% to 90%. The three bands shown in (a) - (c) are indicated with dotted boxes.

Download Full Size | PDF

A clear variation in SERS intensity with duty cycle was observed from each respective Raman band. As explained previously, for a given grating period and grating height, there is an optimal duty cycle that leads to the most efficient activation of a SPP mode. For the relatively deep 70 nm gratings (black triangle), the SERS response is considerably increased with increasing duty cycle (or grating ridge width) within the range from ~15% to 67%. After reaching a maximum SERS intensity near 67%, for each of the three chosen Raman peaks, further increasing the duty cycle results in reduced SERS intensities. The maximized SERS intensity from the sample with a duty cycle of ~67% shows good agreement with our simulation result in Fig. 2(c), where the sample with ~60% duty cycle shows a clear resonance spectrally overlapped with both excitation and Stokes wavelengths. The observed phenomenon indicates not only a continuous red shift of the SPP resonance with increased grating duty cycle, as shown in Fig. 2(c), but also the importance of the resonance wavelength to the resulting SERS intensity. Similar phenomenon showing an optimized duty cycle for a given grating depth has been predicted in previous study [10]. It also can be seen that decreased grating depth leads to a reduced maximum SERS intensity, which agrees well with our calculations shown in Fig. 2(b). Note that decreasing the grating height also shifts the optimum grating duty cycle towards larger values, which is required for shallow gratings to enable strong coupling of SPPs. For the 40 nm (red circle) and 20 nm (blue square) deep gratings, maximum SERS intensities were achieved at larger duty cycles of ~77% and 92%, respectively. The SERS spectra from benzenethiol on the 2D P-NPG samples with different grating depths and their corresponding optimized duty cycles are shown in Fig. 3(d). A color plot of SERS intensity for all Raman peaks as a function of duty cycle is shown in Fig. 3(e) for the 70 nm deep grating sample, which indicates that f = 67% is the optimal design across all measured Raman bands. The optimal grating design parameters for our NPG films are: Λ = 650 nm, f = 67%, h = 70 nm. Based on the measured trend curves of the SERS signals, it is clear that the SPP effect indeed plays a very important role in enhancing the Raman signal, and efficient activation of a SPP resonance at a given excitation wavelength requires proper design of the grating structure.

In order to determine the SERS enhancement effect of the 2D P-NPG substrates, a reference sample of pure benzenethiol contained in a 500 µm thick PDMS cell was also measured with the same experimental setup [11]. Figure 4 shows the comparison of the SERS intensity from a 2D P-NPG substrate (Λ = 650 nm, h = 70 nm, f = 67%) and the reference sample. The enhancement factor (EF) of the optimized 2D P-NPG substrate was calculated using the following equation [28]:

 

Fig. 4 Comparison of SERS spectra between optimized 2D P-NPG substrate (black curve) and a reference sample (500 µm thick pure benzenethiol in PDMS cell, red curve). 2D P-NPG shows an EF of ~0.5 × 10⁸.

Download Full Size | PDF

In order to determine the number of benzenthiol molecules within the detection volume of the 2D P-NPG substrate, we first performed voltammetric scans for the reductive desorption of benzenethiol molecules to estimate the accessible surface area of the 2D P-NPG film [21]. Figure 5 shows cyclic voltammograms for the reductive desorption of benzenethiol self-assembled monolayers from planar gold and stamped nanoporous gold films in a 0.5 M KOH (aq) electrolyte solution. The planar gold curve exhibits two cathodic current peaks at −0.61 V and −1 V that we ascribe to thiolates bound at different surface sites on these gold surfaces [29]. Integration of these peaks is used to obtain a charge that is proportional to the number of benzenethiol molecules desorbed from the surface. Accordingly, for planar gold, the number of benzenethiol molecules per unit area is found to be 3.5 × 10¹⁴ cm⁻². The stamped nanoporous gold electrode shows the combination of two cathodic current peaks from −0.95 to −1.1 V. The average integrated charge ratio of stamped nanoporous gold to planar gold, both shown on a basis of 1 cm² of geometric area under the same solution conditions, is approximately 1.5, suggesting that the stamped electrode has ~1.5 times more bound benzenethiol molecules than the planar gold electrode. Therefore, the estimated number of benzenethiol molecules per unit area on the 2D P-NPG substrate is 5.25 × 10¹⁴ cm⁻². Note that the available surface area of this P-NPG substrate is not significantly increased compared to planar gold, likely due to the relatively large pore size (15-25 nm) of the NPG film as well as the post-stamping process utilized to form the grating pattern. At the same porosity, NPG films with larger pores provide smaller available surface area than NPG films with smaller pores. Furthermore, the densification effect of the DIPS process crushes almost 60% of the NPG region, resulting in deformed gold ligaments and reduced pore openings and thus significantly reducing the available surface area of the P-NPG substrate. Next, we consider that the strong electromagnetic field on the surface associated with the SPP effect cannot penetrate the entire 160 nm thickness of the NPG film, and hence all of the benzenthiol molecules on the 2D P-NPG substrate do not contribute to the measured SERS signal. The penetration depth (h) in NPG is found to be ~20 nm at 780 nm excitation, based on the relation, h = c/2kω [30], where c is the speed of light, ω is the frequency of the light, and k is the measured imaginary part of the refractive index of NPG at 780 nm excitation (k = 3.06). Considering this penetration depth and the nearly collimated illumination in the vicinity of the focus plane of the Gaussian beam, only ~12.5% of the total adsorbed benzenethiol molecules contribute to the SERS signal. Thus, N_{2D P-NPG} is determined to be 0.125 × (5.25 × 10¹⁴) × A, where A is the laser spot size.

 

Fig. 5 Voltammetric scans for the reductive desorption of benzenethiol molecules from planar gold and stamped NPG. The scan rate was 200 mV/s. Desorption was performed in a 0.5 M KOH electrolyte solution.

Download Full Size | PDF

The total number of probed molecules within the detection volume of the 500 µm thick cell, N_ref, is estimated to be (1.19 × 10¹⁹) × A cm⁻² based on N_ref = N_AρAd/M [31], where N_A is Avogadro’s number, ρ is the benzenethiol density (1.08 g/cm³), M is the molar mass of benzenethiol (110.19 g/mol), and d is the laser probe length in the pure benzenethiol sample (d = nλ/(NA)²), which is estimated to be ~20 µm at λ = 780 nm excitation based on the 10 × magnification objective (NA = 0.25) and refractive index of benzenethiol (n = 1.59) [32].

From Fig. 4, the measured intensity ratio (I_{2D P-NPG}/I_ref) at the 1070 cm⁻¹ band is 275. Therefore, the spatially averaged EF of the 2D P-NPG substrate with Λ = 650 nm, h = 70 nm, f = 67% is estimated to be 0.5 × 10⁸, which shows a more than two orders of magnitude higher SERS signal enhancement than that of as-prepared NPG films. This significant enhancement is attributed to the simultaneous activation of both LSP and SPP effects on NPG. Importantly, the P-NPG substrates presented here using a straightforward and repeatable fabrication approach provide an effective and economical way to design and manufacture high-through output and low-cost SERS substrates based on porous metallic materials.

4. Conclusions

In summary, 2D P-NPG substrates were utilized to demonstrate the influence of grating parameters on the SERS enhancement. The importance of a strongly activated SPP for achieving large SERS enhancement has been theoretically and experimentally demonstrated. By precisely and systematically tuning the gratings parameters, the SERS response of 2D P-NPG substrates was controlled and a SERS enhancement factor of 0.5 × 10⁸ was reported based on a 2D P-NPG substrate with appropriate grating period, duty cycle, and grating height. Unlike the solely LSP-based SERS substrates, 2D P-NPG is found to provide SERS enhancement relying on both the LSP and SPP of NPG with no requirement for using sophisticated lithography to achieve ultra-small lateral features or the need to search for a hot spot. Further refinement of the NPG morphology or the grating geometry, for example use of blazed grating design, may lead to additional increase of the EF. The straightforward and low-cost of fabrication, large area uniformity, and high SERS enhancement factor make the 2D P-NPG substrate attractive as a high performance SERS substrate for chemical and biological molecule identification with high detection sensitivity.