Enhanced Dye Degradation through Multi‐Particle Confinement in a Porous Silicon Substrate: A Highly Efficient, Low Band Gap Photocatalyst

A platform is introduced for fabrication of a reusable and highly efficient low band gap photocatalyst by confining gold nanoparticles (AuNPs) in the pores of a nanopatterned Si monolith (AuNSM). Due to their size, a maximum of two AuNPs can assemble in a single pore, thus preventing agglomerations. Their access to the analyte provides more active sites for redox reaction, leading to enhanced efficiency. While proximity of nanoparticles enhances coupling efficiency, confinement prevents rapid recombination of photogenerated charge carriers, a major factor contributing to low efficiency of photocatalytic materials. Degradation of methyl orange (MO) is used to determine the photocatalytic efficacy of AuNSM compared to 1) bare silicon and 2) AuNPs randomly dispersed on silicon. After 90 min of exposure to UV light (λ = 353 nm) in the AuNSM, the MO absorption is <1%, indicating near complete degradation, while it is still 85% and 70% for systems (1) and (2), respectively. Finite element method simulations of the confined structure suggest that the AuNPs act as a mediator/receptacle for photogenerated charges rather than a source of them at this wavelength and thus enhance the performance of the photocatalyst by creating more effective Schottky junctions—preventing recombination of electrons and holes—rather than by a localized surface plasmonic resonance effect.


Introduction
Heterogeneous photocatalysis technology is always looking for an ideal photo catalyst, one that is reusable and that possesses high photocatalytic efficiency, a large specific surface area, and a response to the broadband spectrum of sunlight. Semiconductor photocatalysts have found various applications and are a potential candidate for future sustainable fuel gene ration from solar light. However, semi conductor heterostructures typically have drawbacks such as rapid recombination of photogenerated charge carriers, nar rowband response to light, and lower efficiency. These drawbacks could be over come by various methods such as nano structuring of the semiconductor surface to increase absorption density to create more charge carriers for photocatalysis, and formation of heterostructure with plasmonic materials to increase photo catalytic efficiency across the broad solar spectrum. Recently, photocatalysis has received lots of attention as a possible solution to some of the most fundamental issues in sustainable development, such as renewable energy production, water treatment, and air purifica tion. [1][2][3] In one scheme, several semiconductor photocatalysts are deposited alongside metallic cocatalysts: holes are confined in the semiconductor rods, while the delocalized electrons get transferred onto the metal cocatalyst. [4,5] The advantage of this system is good control over the charge separation. These cocatalysts show great promise for water splitting, due to their lower activation potentials for hydrogen generation and enhanced photocatalytic function. [3,5] Here, we introduce a new recyclable heterogeneous cocatalyst made of gold nanoparticles (AuNPs) confined in a nanoporous Si monolith (AuNSM) using blockcopolymer nanopatterning (Figure 1). The novel aspect of our design takes advantage of confinement without compromising the exposure of metal to the solution as the nanoparticles are not fully embedded in the semiconductor. Otherwise, the deeply embedded nanoparti cles in the semiconductor photocatalyst having limited contact with the solution could act as charge recombination centers and lower the efficiency of the photocatalyst. Moreover, metal A platform is introduced for fabrication of a reusable and highly efficient low band gap photocatalyst by confining gold nanoparticles (AuNPs) in the pores of a nanopatterned Si monolith (AuNSM). Due to their size, a maximum of two AuNPs can assemble in a single pore, thus preventing agglomerations. Their access to the analyte provides more active sites for redox reaction, leading to enhanced efficiency. While proximity of nanoparticles enhances coupling efficiency, confinement prevents rapid recombination of photogenerated charge carriers, a major factor contributing to low efficiency of photocatalytic materials. Degradation of methyl orange (MO) is used to determine the photocatalytic efficacy of AuNSM compared to 1) bare silicon and 2) AuNPs randomly dispersed on silicon. After 90 min of exposure to UV light (λ = 353 nm) in the AuNSM, the MO absorption is <1%, indicating near complete degradation, while it is still 85% and 70% for systems (1) and (2), respectively. Finite element method simulations of the confined structure suggest that the AuNPs act as a mediator/receptacle for photogenerated charges rather than a source of them at this wavelength and thus enhance the performance of the photocatalyst by creating more effective Schottky junctions-preventing recombination of electrons and holes-rather than by a localized surface plasmonic resonance effect.
nanoparticles in confined space volumes could also cause synergistic interactions between the nanoparticles and the support, thereby changing the electronic properties of the heterostructure. [6][7][8] Due to the high surface energy of metal nanoparticles, they have a strong tendency to agglomerate into clusters, [9,10] and as a result, control of their assembly and applications cannot be attained without appropriate designs. Therefore, it is imperative to design a hybrid photocatalyst system, which can suppress agglomeration and facilitate better photocatalyst activity. In this work, we address the agglomera tion problem by confining limited numbers of the nanoparti cles in similar size cavities of a nanopatterned Si substrate. One of the major factors contributing to low efficiency of photocat alytic materials is the recombination of excited electrons and holes; our design allows for better charge separation because of the large interaction area of the AuNPs and the silicon sub strate within the pores. Even though this heterostructure was not experimentally tested at visible wavelengths range, our finite difference time domain (FDTD) simulations predict that this system has significantly greater electric field enhancement which potentially leads to a significantly faster redox reaction. Compared to semiconductor CdS [11] nanocrystals and TiO 2 [12] nanoparticles with 95% and 97% efficiency of methyl orange (MO) degradation after 300 and 420 min UV light illumination respectively, our heterostructure photocatalyst achieved near complete MO degradation after 90 min illumination.
Blockcopolymer soft nanolithography is a versatile, robust, and costeffective technique to make nanostructured surfaces and produce various types of surface morphologies. [13][14][15][16][17][18] The structural dimensions such as pore size, pore depth, and sur face area are adjustable by simply employing different block copolymers with varying interaction parameters (χ), volume fraction of blocks (f), and finetuning microphase separation parameters. [15,[18][19][20] In this study, we have used poly(styrene) blockpoly(methyl methacrylate) (PSbPMMA) blockcopolymer to define an etch mask for producing a nanoporous Si substrate, with AuNPs encapsulated in the nanopores for photocatalytic applications. As one of the most prevalent organic pollutants in waste water, MO has become a great concern due to its poten tial carcinogenic, mutagenic, and bactericide properties. There fore, MO deposited on AuNSM was chosen as a model reaction to examine the photocatalytic activity of the catalyst platform.
Si is a cheap and abundant semiconductor that has revo lutionized the microelectronics industry. [21] Si quantum dots, nanowires, mesoporous silica and other composites have been used in photocatalytic applications. [22,23] However, to the best of our knowledge, porous nanostructured Si as a template for confining metal nanoparticle cocatalysts to produce field enhancement has not been investigated. The advantages of our heterogeneous system are 1) its high surface area; [24] 2) the Si photocatalytic monolith can be easily recycled and separated from the reaction system (whereas, homogeneous powder cata lysts cannot be easily reused); and 3) the close proximity of the metal nanoparticles confined in the nanopores enhances cou pling efficiency and prevention of electron-hole recombination. Moreover, metal cocatalysts embedded in nanopores control the reaction kinetics efficiently by adjusting the absorption and loading capacity, where the accessibility of analyte to the active sites is high. These materials also allow for controlling the con tact time, which is a key parameter for increasing selectivity and limiting the occurrence of secondary reactions.

Morphological Analysis of the Nanoporous Silicon Monolith
To fabricate the NSM, we employed PSbPMMA blockcopol ymer with the PMMA as the cylinder forming microdomain and the PS as the majority matrix of the blockcopolymer film. Blockcopolymer systems have been explored as an alternative route to overcome conventional lithography methods. Due to immiscibility between two blocks, joined by a covalent bond, blockcopolymers selfassemble to periodic nanosize structures, typically with repeat distance of 10-100 nm. Figure 1A shows a PSbPMMA phase separated film on a silicon substrate. The centertocenter pitch in the blockcopolymer film measured by atomic force microscopy (AFM) is 48 ± 2 nm and the diameter of the cylinders is 25 ± 5 nm. The nanopatterned template was employed as the etch mask to produce nanopore arrays in Si through the pattern transfer process. The details on preparation of the film and the etch process can be found in the experi mental section. The PMMA block (the dots in Figure 1A) was selectively removed through a dry etch process. [25] The resulting porous PS matrix ( Figure 1B) was employed as an etch mask for the fabrication of the nanoporous Si substrate. Pattern transfer into the Si substrate was performed by induc tively coupled plasma (ICP) etching using C 4 F 8 and SF 6 process gases. This resulted in a welldefined nanoporous Si substrate, with pore depth of ≈60 nm ( Figure 1D). No considerable disor dered or missing patterns were observed for the Si substrate after etching. The inset in Figure 1C is a highmagnification scan ning electron microscopy (SEM) image, indicating wide range ordering of the nanopores. The diameters of the nanopores in the film varied from 25 ± 5 nm with a spacing of 45 ± 3 nm and depth of 60 nm, which is a close match with the original template mask. To study the effect of confinement on photocatalysis, we synthesized AuNPs and confined them in the pores of the NSM. Figure S2, Supporting Information, shows SEM images of the thiolcapped AuNPs with an average diameter of 18 nm. AuNSMs were prepared by spin coating of the AuNP solution on the nanoporous Si substrate. The insertion procedure was repeated twice to ensure the nanopores were fully loaded with AuNPs. Figure 1D is a crosssectional SEM image of the AuNSM, where the bright spots represent AuNPs in the nanopores.

The Effect of Confinement on the Kinetics of Photocatalytic Reaction
To verify the effect of enhanced coupling in AuNPs when con fined in Si nanopores on photocatalysis, MO was chosen for dye degradation on UVlight exposure. The AuNSM system is believed to improve the degradation rate due to the greater interaction area of the porous silicon substrate with the MO dye and the AuNPs compared to flat silicon. In the AuNSM, contact between each AuNP and the nanoporous Si creates a Schottky junction and thus an internal electric field that prevents recom bination of electrons and holes. Based on experimental data, we believe the AuNPs confined in nanoporous Si are important for separating the electrons and holes generated upon light absorp tion in the AuNSM and thus these electrons are driving the degradation of the dye. We have also verified our hypothesis by two control experiments, where UV light absorbed on 1) a flat silicon substrate and 2) AuNPs on flat silicon produced signifi cantly less degradation of the dye. This indicates that electronhole separation is facilitated effectively in AuNSM compared to flat silicon and to randomly dispersed AuNPs on Si. Figure 2A represents a schematic of AuNPs deposited on bare Si (top image, Figure 2A) and AuNPs embedded in the NSM (bottom image, Figure 2A). The photocatalytic activity of three systems was tested: 1) unpatterned Si; 2) AuNPs randomly dispersed on Si ( Figure 2B); 3) AuNSM where synthesized AuNPs were deposited and confined in the pores of a NSM ( Figure 2C). All the substrates were cleaned with deionized (DI) water. After deposition of AuNPs, the samples were left overnight to dry then rinsed with DI water to remove any surface adducts or AuNPs hinged outside of the pore structures. Xray photoelec tron spectroscopy (XPS) spectra of AuNSM samples before and after catalytic use are shown in Figure 3. The nearly identical amount of gold on the surface before and after use indicates that AuNPs are confined in the pores, otherwise they would have been washed away through the rinsing process.
MO dye (0.25 mmol mL -1 ) in DI water was deposited onto all three sets of substrates: bare Si, randomly dispersed AuNPs on Si, and AuNSM. The substrates were irradiated with UV light (353 nm) for a range of time intervals (5-90 min). The irradi ated substrates were rinsed in equal volumes of DI water and this "wash" (active spectra) was used in the UV-vis spectrom etry. In Figure 2D-F, the UV-vis spectra of the degraded MO dye wash from the different types of modified Si is shown after 10, 50, and 90 min of exposure. Scheme S1, Supporting Infor mation, describes the steps. We observe that after 15 minutes, the active spectra-i.e., the absorption of light by the washis lowest for the AuNSM template, indicating the MO dye has already been degraded. This suggests that catalytic efficiency is higher in the AuNSM than in other systems. Specifically, after 90 min of irradiation time ( Figure 2F), most of the dye on the AuNSM has been degraded completely, whereas the dye on the bare Si substrate and on the AuNPs on bare Si substrate were degraded partially or very little. These results further support our hypothesis that the confinement in AuNSM photocatalyst enhanced the photochemical activity and/or the synergistic interactions between the NPs and the Si support that changed the electronic properties of heterostructure, which enhanced the dye degradation. From the active spectra ( Figure 2D-F), it can be inferred that the addition of AuNPs to bare Si also improves the photocatalytic performance (slightly better than bare Si, but significantly less than the AuNSM system). This may be due to an increased number of photogenerated elec trons in the AuNPs compared to bare Si or the possibility of charge separation by transfer from the substrate to the AuNP. For comparison, a solution of dye containing AuNPs was exposed for 90 min in the same experimental setup but the degradation was negligible. This indicates that simple UV expo sure is not enough to degrade the dye and also that AuNPs in solution (most likely isolated) do not degrade the dye either.
In the absorption spectrum of MO, an absorption peak was observed at 464 nm ( Figure 4A). The decrease in the intensity of the absorption peak at 464 nm indicates the possible cleavage of the azo group (NN), which is a chromophoric functional group. In Figure 4B, the peak absorption is plotted against exposure length for the four substrates. The dashed line is a fit to a single exponential. For the AuNSM system, the fit gives a decay constant of 4 min (the necessary exposure time for the absorption to decrease to 1/e of its original value). (The fits for the other substrates would require more data points.) The pro gressive decolorization of the MO dye on AuNSM at different time intervals is illustrated in Figure 4C. After 90 min irradia tion time, most of the dye was decolorized indicating complete degradation. Figure 4D shows the reflectivity measurements of the three different modified Si substrates, where the reflectivity in the AuNSM is comparatively lower than nanoporous Si and bare Si. This indicates AuNSM has higher absorption density  The plausible mechanism of the MO degradation due to UV exposure is proposed to be the following: Degradation of MO solution by the cleavage of the NN bond to two different chemical amines structures (NR 2 ) (Scheme S1). The NN double bond in azo dyes is the chromophoric functional group for color, which is prone to decomposition upon photoexcita tion. Photocatalysis via light absorption induces the generation of charge carriers such as hot electrons. These electrons are very effective at causing the degradation of the dye. [26] In addition, we presume the acceleration of the photocatalysis could be through the coupling of AuNPs and the silicon substrate. The degrada tion process leveled off after ≈90 min, indicating that all of the MO dye had been degraded ( Figure 4A). These results indicate that the AuNSM photocatalyst support enhanced the photo chemical activity. From Figure 4B, it can be seen that the MO dye degradation on the AuNSM at different time intervals of UVlight exposure is higher than bare Si (the data in Figure 4B were normalized to the dye absorption for zero exposure time). This might be due partially to the increased surface area of the porous Si substrate (≈3× more than bare Si) providing more active sites for photoreaction and leading to a greater interaction area with the MO dye. In addition, AuNPs confined in Si nano pores could also cause synergistic interactions between the NPs and the Si support, and thus change the electronic properties of the heterostructure, this also facilitates charge separation which enhances the dye degradation. [6,26,27] These interactions also take advantage of the high surface area of the nanoporous Si com pared with the bare Si to increase the photocatalytic efficiency. [24] To understand the mechanism of the photocatalysis, our control experiments with different substrates showed that: 1) photocata lytic degradation of MO dye on bare Si had a slow reaction rate, this could be ascribed to the unavoidable rapid recombination of photogenerated charge carriers in the Si, that decreases the photocatalytic MO degradation. 2) Deposition of the noble metal NPs on semiconductors suppressed the electronhole recombi nation. This semiconductor/metal heterostructure, pumps in photogenerated electrons, and facilitates an interfacial charge transfer process due to its high Schottky barrier at the metal/ semiconductor interface. [28,29] However, efficiency of the ran domly distributed NPs over the Si substrate on the photocatalytic MO degradation is too small. We presume the low efficiency is because of fewer photogenerated charge carriers to induce the MO degradation. Finite element method (FEM) simulations, in Figure 5B-D, show that there is a greater absorption density around the pore of the AuNSM which suggests more charge carriers are being generated and that they are nearer to the dye thus improving the likelihood of transfer and interaction. 3) In our system, confinement of the AuNPs in the Si nanocavities enlarges the excitation space and encourages transfer of hot carriers from the Si support to enhance the photocatalytic effi ciency. Moreover, it could also be that the nanopores serve as a template to stack AuNPs resulting in a short separation and a higher enhancement factor. This could be ascribed to the gen eration of hot carriers, which combined with Si semiconductor cause excitation of electronic or vibrational transitions in MO dye molecules to accelerate the photodegradation. [30]

Electromagnetic Simulations
FEM simulations were performed for the three different sys tems used in our experiment. The simulations were used to gain insight into the relative photocatalytic performance of each system. The sizes of the pore and AuNP were taken from the SEM images (25 ± 5 nm and 18 ± 3 nm, respectively) and the geometry was set up to simulate a periodic triangular lat tice for each substrate. While it is unlikely that the AuNPs on bare Si formed a triangular lattice, it is expected that they were isolated individual particles on the surface and thus the simulated geometry was an adequate approximation. FEM simulations were performed using the software COMSOL [31] and used plane wave excitation at a wavelength of 353 nm. Maps of the power loss density are plotted in Figure 5B-D; it can be seen that the AuNSM has strong absorption in the Si substrate around the pore ( Figure 5D). It is proposed that this absorption and the resulting charge carriers generated, along with the increased surface area, is the source of the improved photocatalytic activity. However, our simulations showed that there was not much absorption by the AuNPs in the AuNSM system, suggesting that the AuNPs act as a mediator/ receptacle for photogenerated charges rather than a source of them at this wavelength. For the other systems, there is some absorption in the Si substrate, i.e., 62% and 64% of the AuNSM, respectively ( Figure 4B,C). However, since these sys tems were less effective in experiment, it suggests that the confinement of the AuNPs and dye in the pores improves the transfer of charge. From our simulation outcome, it is likely that the main reason for fast photocatalytic degradation of MO is due to cre ating a large volume of Schottky junctions near the AuNP/Si interface, which provide extra paths for electron-hole separa tion and fast charge lane to the active sites.
It has been reported that metallic NPs that are separated by distances of a few nanometers show high amplitude of elec tromagnetic field [26,[32][33][34] that is ascribed to the strong coupling between adjacent NPs. Moreover, the synergy of noble metals (such as Au and Ag) and semiconductor nanopores in photo catalytic heterostructures enhances photocatalytic efficiency by significant suppression of the electron-hole recombina tion generated upon photoexcitation. [28] Indeed, our original hypothesis was that confining multiple nanoparticles (AuNPs) in nanostructured spaces (Si pores) would increase the near field enhancement significantly in comparison to randomly dispersed nanoparticles on the same substrate. In a photo catalytic system, this enhancement should be able to drive the reactions at significantly higher rates. Therefore, the AuNSM system is expected to be photocatalytically active at visible wavelengths due to its plasmonic response. To investigate the extent to which the system could be affected by the range of exposure wavelength, FDTD simulations were carried out on AuNP ensembles in a Si nanopore. We simulated the electric field enhancement for multiple plasmonic AuNPs confined in nanostructured spaces, in visible wavelength range and dem onstrated that the enhancement of nearfield is due to plas monic coupling. We found that the electric field enhancement due to plasmonic coupling for multiple AuNPs embedded in a nanopore is greater than for dispersed and/or individual NPs; this enhancement contributes strongly to the plasmondriven photocatalysis efficiency. The generation of plasmoninduced hot charge carriers is directly related to the local electric field |E(r)| 2 inside the plasmonic particles [35] where V MFP is the volume within a distance of the mean free path from the surface inside (r p ) up to active interface (r MFP ).
Due to the small size of our particles, we can consider all the electrons of the entire volume of the particle. [34] Smaller particles also have the advantage that they scatter less com pared to larger ones and therefore their plasmon absorption yield is higher, [34] so more hot charge carriers can be injected into the semiconductor Si substrate, and direct transfer of the hot charges from the AuNPs (20 nm in diameter) to MO dye can occur simultaneously. For example, Boerigter et al. reported direct charge excitation as the dominant role in the degrada tion of methylene blue. [36] In this case, the electric field at the surface of the plasmonic system is the determining factor. Figure 5 shows the local electric field enhancement of a single AuNP ( Figure 5E) in a nanopore (28 nm diameter), compared to a dimer of touching AuNPs ( Figure 5F). FDTD calculations indicated the significant increase in near field enhancement as a function of wavelength, shown in Figure 5G, and red shift from (≈630 nm) to (≈735 nm) due to the dimer of AuNPs with a closepacked hotspot in the Si nanopore. Experimental verifica tion of visible range will be the focus of future work.

Conclusions
In summary, we have studied how confining AuNPs in a porous Si template can significantly enhance the photocatalytic activity of MO. The pores prevent agglomeration of nanoparticles and eliminate the need for any functionalization. Confinement of the AuNPs in the Si nanocavities prevents electron-hole recom bination and facilitates the transfer of hot carriers from the Si support to accelerate the photocatalytic efficiency. Since the nan oparticles are not fully buried in the semiconductor matrix, the number of active sites for redox reaction is plentiful. All these factors contribute to higher efficiency of our heterogeneous photocatalyst. We have developed a blockcopolymer templated nanoporous Si monolith (NSM) to achieve the confinement of the AuNPs and their arrangement in closepacking set. Close packing enhances the electromagnetic field and increases the synergistic interactions between the nanoparticles and the nanoporous support, thereby changing the electronic properties of the photocatalytic heterostructure. Full degradation of the MO dye on the AuNSM was successfully achieved on exposure to UV light for 90 min, at least three times faster than reported in the literature, compared to unmodified or bare Si substrates, where no or minimal degradation was observed. FEM simula tion demonstrates in our system where the Si pore and AuNPs have similar dimension, nanocavities have large absorption density around the pore of the AuNSM compared to bare Si and porous Si. At a wavelength of 353 nm, the interparticle interac tion in the pores seems negligible, indicating that the AuNPs act as a mediator/receptacle for photogenerated charges rather than a source of them at this wavelength. This suggests con fining the AuNPs in Si pores enhances the performance of the photo catalyst by creating more effective Schottky junctions and thus preventing recombination of electrons and holes rather than localized surface plasmonic resonance effect. Nonetheless, we believe that this proposed mechanism of electron transfer between the AuNPs and Si warrants further study-in par ticular utilizing characterization such as Kelvin probe micro scopy to study the work function of the materials. [37,38] The recyclable, lowband gap photocatalytic system has economic and environmental advantages that promote implementation of catalytic and separation processes in continuous flow mode, with the advantages associated with easier phase separation and product recovery, enhanced safety, and easier operation. We anticipate that this system can be applied to fabrication of various other nanomaterials for photocatalytic applications.
Preparation of the Nanoporous Silicon Template-Preparation of Block-Copolymer Film: The preparation of self-assembled diblock-copolymer thin films on Si wafers was performed in four steps: wafer cleaning and oxidation, polymer brush formation, thin film formation, and annealing.
Step 1 involved wafer cleaning and oxidation. A (100) Si wafer was treated with fresh piranha solution (H 2 SO 4 /H 2 O 2 with a volume ratio of 3:1 at 90 °C for 45 min) to form a clean Si oxide surface. The samples were then extensively rinsed with DI H 2 O and dried under N 2 flow.
Step 2 involved polymer brush formation. A polymer brush was applied to the Si wafer in order to produce a neutral or non-preferential wetting surface. PS-r-PMMA (Arkema) with 1 wt% in PGMEA solution was spincoated onto the cleaned Si wafer at 1500 rpm for 30 s. The coated substrates were annealed at 170 °C for 6 h, allowing the hydroxyl end groups to diffuse and react with the oxide layer, anchoring the random copolymer to the substrate. The unreacted random polymer units were removed by rinsing and sonicating the wafers in toluene for 5 min for three times.
Step 3 involved diblock-copolymer thin film formation. The thin diblock copolymer films of PS-b-PMMA were produced by depositing the polymer onto the brush-treated Si wafer surfaces by spin-coating (1500 rpm, 30 s) a 1% solution of PS-b-PMMA in PGMEA.
Step 4 involved annealing. The block copolymer thin films were annealed at 250 °C for either 10, 20, 40, or 60 min to allow reorganization. The phase separations were characterized by AFM (Figure 1).
Preparation of the Nanoporous Silicon Template-Fabrication of the Block-Copolymer Nanopatterned Etch Mask: This method produced remarkably high-quality masks with minimum disordered or missing patterns over a large area, observed in the nanoporous template after the PMMA removal step ( Figure 1B). The PS:PMMA selective etch was run in an Oxford System 100 ICP etcher for 100 s, using CHF 3 and O 2 (40 and 5 sccm), 10 mTorr, 100 W RF, with a 10 Torr helium gas cooling. An oxygen plasma treatment was performed to cross-link the PS chains and remove the PS-r-PMMA at the bottom of the pores. The resulting porous PS matrix was employed as an etch mask to create nanopores in the Si substrate. Pattern transfer to the Si substrate was performed using C 4 F 8 and SF 6 (90 and 30 sccm), 600/15 W (ICP/RF), with a 10 Torr helium gas cooling. Samples were mounted on a 100 mm Si carrier wafer with Krytox vacuum oil. After the reactive ion etching process, any residue of the polymer mask was removed in a cyclohexane solution. See Scheme S2, Supporting Information.
Synthesis of AuNPs: Functionalized AuNPs were prepared in a one-pot procedure. In a typical synthesis, 0.058 mmol of hydrogen tetrachloroaureate(III) trihydrate (H[Au(Cl 4 )]·3H 2 O) and 11-mercaptoundecanoic acid (0.068 mmol, 15 mg) were dissolved in 5 mL of anhydrous ethanol. The reaction mixture was magnetically stirred at 70 °C for 2 h. Subsequently, sodium borohydride (NaBH 4 ) (0.26 mmol, 10 mg) in ethanol was added dropwise into the reaction mixture and left on stirring for 24 h at 70 °C. The solution mixture turned light pink in color, indicating the formation of the AuNPs ( Figure S1, Supporting Information). The AuNPs dispersed in ethanol were then collected by a centrifugation at 20 000 rpm and dried in a desiccator and used for catalytic experiments ( Figure S2, Supporting Information).
Fabrication of Nanopore Si Substrate Embedded with AuNPs: Before embedding the AuNPs in the porous Si, the substrate was cleaned with DI water. AuNP solution was then deposited onto the porous Si wafer through spin coating. The insertion procedure was repeated twice to ensure the nanopores were fully embedded with AuNPs. The solution was left on the substrate overnight to dry and then rinsed with DI water to remove any adducts. The surface morphologies of the substrates were examined by SEM ( Figure S3, Supporting Information).
Photocatalytic Degradation of MO Dye: A total of 50 µL at 0.25 mmol mL -1 of MO was prepared in DI water. The dye solution was stirred for 30 min before use. There was no agglomeration or particulate formation in the solution, exhibiting good homogeneity in the solution. The dye solution was spin cast at 3000 rpm onto the AuNSM. The samples were irradiated under UV lamp (15 W, 353 nm) for different time durations (5-90 min). The irradiated substrates were washed in equal volumes of distilled water to extract the dye for UV-vis absorption measurements. The degradation of the dye was analyzed by the decrease in the absorption of MO.
X-Ray Photoelectron Spectroscopy: An XPS/UPS Omicron MultiProbeXP X-ray photoelectron spectrometer was used to determine the elements present in the samples. The system uses an aluminum source (1486.6 eV, 300 W) and an XM 1000 monochromator.
Step-sizes of 0.5 eV and 0.2 eV were used for broadband and single peak scans, respectively. To make sure AuNPs are confined in the Si pores, XPS spectra of samples were collected before and after photocatalytic use (Figure 3). XPS was also performed on nanopatterned Si (without gold) to show the stability of the substrate ( Figure S4, Supporting Information). SEM image of the samples after use confirms the pores are intact ( Figure S5, Supporting Information).
Optical Characterization: Absorption measurements of the extracted dye were taken using a PerkinElmer LAMBDA 1050 UV-Vis-NIR spectrophotometer fitted with an integrating sphere. The samples were placed on a holder within the sphere and scanned over the wavelength range 250-700 nm in steps of 3 nm with an integration time of 0.7 s. Reference measurements of distilled water and of the original dye solution were also taken.
FEM Simulation: The absorption of the different substrates was simulated with the FEM using COMSOL Multiphysics. A hexagonal unit cell of side length 28 nm (to correspond to a pitch of 48 nm) was used with spherical AuNPs that had a diameter of 22 nm. Periodic boundary conditions on sides perpendicular to the substrate surface, absorbing boundary conditions on sides parallel to the substrate surface, and a plane wave light source with a wavelength of 353 nm were used. The substrates simulated were bare silicon, silicon with a single AuNP, and nanoporous silicon containing two AuNPs. For porous silicon, a pore diameter of 25 nm and a depth of 60 nm were used. The built-in permittivity Au (Rakic) and Si (Aspnes) were used for gold and silicon, respectively.
FDTD Simulation: The electric field enhancement due to multiple AuNPs in a Si nanopore was simulated with the FDTD method, through the open-source software package MEEP. [39] The dielectric function of AuNPs (20 nm in diameter) was approximated by a Lorentz-Drude model using experimental data obtained from Rakić et al. [40] Periodic boundaries were placed on all sides of the simulation cell (pore size = 28 nm, pitch = 48 nm, depth = 60 nm) and an artificial absorber layer was placed parallel to the plasmonic array at the end of the simulation cell to block transmission through the cell. The plasmonic arrays and nanopores were hexagonally arranged.

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