Dewetted Silver Nanoparticle-Dispersed WO3 Heterojunction Nanostructures on Glass Fibers for Efficient Visible-Light-Active Photocatalysis by Magnetron Sputtering.

Fabrication of hybrid-heterojunction nanostructures comprising the Z-scheme and localized surface plasmon resonance is essential for enhancing the photocatalytic degradation of organic compounds to enable environmental remediation. This study focuses on the dispersion of dewetted Ag nanoparticles over the 3D network-like silica glass fibers (SGFs) coated with a Cu-doped WO3 heterojunction system by a high-throughput and cost-effective method using magnetron sputtering, followed by solid-state dewetting. The influence of Cu doping on the crystal structure, growth direction, and morphology of WO3 and the effect of localized surface diffusion-driven dewetted Ag nanoparticles on the photocatalytic performance were investigated. The Cu doping changed the optical band gap, and the 2Cu-WO3/SGF exhibited excellent photocatalytic activity. The surface dispersion of dewetted Ag nanoparticles over Cu-WO3/SGFs exhibited lowest photoluminescence intensity, indicating the effective separation of photogenerated electrons-holes, which led to highest efficiency (∼98%) in photocatalytic degradation of methylene blue among all the fibers with a degradation rate constant (k = 0.0205 min-1) that was ∼18.6 times higher than that of pure WO3 (k = 0.0011 min-1). The findings of this study can provide insights for designing low-cost and efficient visible-light-active photocatalysts for organic dye degradation, enabling environmental remediation.

Introduction

The 21st century brought significant awareness about devastating environmental impacts and environmental consciousness in the society. Thus, the scientific community has developed substantial interest in the utilization of renewable energy, especially inexhaustible clean energy showered by the sun, for environmental remediation through the development of efficient visible-light-active photocatalysts to decontaminate air and water.[1,2] Tungsten oxide (WO3) is a chemically stable non-toxic n-type semiconductor with a controllable optical band gap (2.4–2.9 eV) and good resistance against photocorrosion in aqueous media.[3] Therefore, WO3 has potential applications in hydrogen production, virus deactivation, and harmful pollutant degradation,[4,5] which make it an appealing candidate for harvesting visible light to degrade organic compounds. However, the widespread application of WO3 is inhibited due to the relatively low conduction band (CB) edge and the rapid recombination of photogenerated electrons and holes, which lead to poor photocatalytic reaction efficiency.[6] The photocatalytic efficiency can be significantly improved by improving the visible light absorption capacity, increasing the surface area, and reducing the recombination rate of photogenerated electron–hole pairs. To date, many scholars have invested a lot of energy to accomplish enhanced photocatalytic activity of WO3 by implementing various strategies such as doping, metal loading, and heterojunction formation.[7−9] Among them, the Z-scheme heterojunction system has attracted increasing attention, which promotes the separation of charge carriers in addition to possessing high redox ability.[9−11] Nevertheless, most of the existing Z-scheme photocatalysts are powder-based materials that often tend to agglomerate because of their high surface energy and involve complexities in the separation process of photocatalytic nanoscale powders from aqueous solutions. Moreover, these residues can cause secondary contamination. As a result, exploring the immobilized functional materials with high photocatalytic activity, which does not require any additional separation process, is easily recyclable, and low cost is imperative for advanced and sustainable photocatalysis. In this context, it is expected that the exploitation of network-like fibers as a substrate material or a supporting skeleton for the photocatalyst could be a viable solution for environmental friendliness. The fibers possess potential advantages such as good adsorption ability, high surface area, good physicochemical and thermal stability, low cost, reusability, and recyclability.[12] Fukugaichi reported TiO2 nanoparticle (NP) dispersions on glass fibers with enhanced photocatalytic degradation of organic dyes.[13] Heo et al. exploited filter fibers for antimicrobial applications using TiO2 nanoparticle coverage through a simple aerosol deposition method, which is impressive in this COVID-19 pandemic situation.[14]

Recently, the mere dispersion of plasmonic metal nanoparticles (such as Ag, Au, Cu, Pt, and Pd) onto a dielectric surface that can radically alter the localized surface plasmon resonance (LSPR) has been considered as the state of the art in the areas of photocatalysis, surface-enhanced Raman scattering, plasmon-enhanced fluorescence, photovoltaics, chemical, and biological sensing applications.[15,16] However, fabrication of these nanoparticles with appropriate structure, size, density, and arrangement is crucial as they influence the LSPR phenomenon. To date, chemical synthesis and lithography were reported to fabricate plasmonic metal nanoparticles,[17,18] which have limitations such as high cost, complexity in processing, high temperature, reproducibility, and time-consuming. In addition, most of these studies reported on the preparation of both patterned and randomly arranged nanoparticles on flat substrates rather than non-flat substrates. Silver has unique advantages over other plasmonic metals, such as low cost, bactericidal properties, high work function, and special behavior for oxygen adsorption.[19] Moreover, the dispersion of evenly distributed plasmonic Ag nanoparticles improves the visible light absorption range, in addition to substantially ameliorating the photocatalytic degradation efficiency of organic pollutants, by suppressing electron–hole recombination.[20] The loading of Ag on metal oxides was generally achieved by photoreduction, conventional impregnation, or chemical reduction methods for photocatalysis.[21−23] Solid-state dewetting, a spontaneous physical phenomenon that refers to the decomposition of a thin film into small particles or nanostructures, leaving voids on a substrate, driven by excess energy due to high fraction of grain boundaries, interfaces, surfaces, or residual stresses is another approach for the dispersion of Ag nanoparticles.[24−26] Although solid-state dewetting is the simplest method, the template or seed-mediated patterning is inevitable,[15,17,18] which again involves complex processing (such as lithography/shadow-mask and so forth) for the 3D network-like substrates.

In this work, we report the fabrication of evenly dispersed Ag nanoparticles with a high-throughput and cost-effective method using magnetron sputtering, followed by solid-state dewetting. To the best of our knowledge, Ag dewetting-driven homogeneous Ag nanoparticle dispersion over 3D network-like fibers has rarely been reported. Utilization of low-cost, reusable, and recyclable 3D-network-like silica glass fibers (SGFs) as supporting materials and the preparation of hybrid nanostructures CuO–WO3/SGF and Ag nanoparticle dispersion based on the Z-scheme and LSPR phenomena for high-performance, practicable, visible-light-active photocatalysts are demonstrated. CuO is a direct band gap p-type semiconductor photocatalyst with a band gap of 1.3–1.7 eV, which is mostly coupled with other semiconductors to form a heterojunction that can potentially inhibit the recombination of photogenerated electron–hole pairs.[27,28] In this work, the synergistic properties arising from the hybrid nanostructure composed of CuO–WO3 composite coated on SGFs and the solid-state dewetting-driven surface dispersion of Ag nanoparticles are thoroughly investigated for an efficient visible-light-active photocatalyst.

Results and Discussion

Crystal Structure and Microstructural Changes in Cu–WO3 Fibers

Figure a shows the X-ray diffraction (XRD) results of the undoped and Cu-doped WO3-coated fibers annealed at 873 K. The undoped WO3 fibers exhibited a monoclinic-WO3 phase (JCPDS no. 01-083-0950) with characteristic peaks of (002), (020), and (200) facets at 2θ = 23.11, 23.55, and 24.24°, respectively. Among these three characteristic peaks, the (200) facet exhibited high intensity, indicating the preferred direction of growth. The 1Cu–WO3 and 2Cu–WO3 fibers also exhibited a monoclinic WO3 phase (JCPDS no. 01-083-0950). However, with increasing Cu concentration, the crystal structure and growth direction of WO3 significantly changed. The XRD intensity of the (002) peak increased with increasing Cu concentration, whereas the intensity of the (200) peak decreased. The change that occurred in the preferred orientation of WO3 could be explained as follows: the as-coated fibers exhibited an amorphous phase (data not shown here), which recrystallized during the rapid thermal annealing (RTA) through surface reconfiguration, by evolving the crystal planes with low surface energy to lower the total surface energy. The literature suggested that the surface energies of monoclinic WO3 followed the order (002) > (020) > (200) with corresponding surface energies of 1.56 > 1.54 > 1.43 J/m2, which implied that the (200) plane was a highly stable phase, while the (002) plane was unstable.[29] Among the three characteristic facets (002), (020), and (200) of WO3, the (002) facet was reported to exhibit significantly enhanced adsorption of reactive species during the photocatalytic reaction in comparison with the other facets.[30] Cu doping could be effective for the recrystallization of the Cu–WO3 fibers in the (002) plane, which became predominant in 2Cu–WO3. A similar observation was reported by Kalanur and Seo where the WO3-preferred orientation was systematically tuned by controlling the Al doping;[30] a higher concentration of dopant changed the crystal structure, as the dopant took W positions in the lattice and was distributed uniformly throughout the crystal structure. Therefore, a further increase in the Cu content revealed the formation of the triclinic CuWO4 (JCPDS no. 01-080-1918) phase due to the replacement of W by Cu in the lattice to form the CuWO4 phase. All the Cu–WO3-coated fibers annealed at 873 K exhibited XRD peaks at 2θ of 35.5 and 38.6° attributed to the (−111) and (111) planes of the monoclinic CuO phase (JCPDS no. 45-0937), and a gradual increase in the CuO peak intensity was observed with an increase in Cu content. It has been reported that the thermodynamic equilibrium oxide phase (Cu–O) was wholly determined by the applied oxidation temperature, rather than growth kinetics, and as a result, CuO phase formation was ensured because of rapid O2 diffusion above its critical temperature of 600 K.[31,32] Thus, in this study, a higher oxidation temperature (873 K) could drastically enhance the O2 diffusion process to form CuO. Consequently, the XRD results revealed that the 1Cu–WO3- and 2Cu–WO3-coated fibers annealed at 873 K comprised the WO3 and CuO composite phases, and the 3Cu–WO3 fibers consisted of CuWO4 and CuO composite phases.

Figure 1

(a) XRD spectra of WO3 and Cu–WO3-coated SGFs annealed at 873 K; FE-SEM images (magnification: 30 000×) of (b) WO3, (c) 1Cu–WO3-, (d) 2Cu–WO3-, and (e) 3Cu–WO3-coated SGFs annealed at 873 K.

The surface morphologies of the undoped and Cu-doped WO3-coated fibers annealed at 873 K were observed by field-emission scanning electron microscopy (FE-SEM), as shown in Figure b–e. The as-coated fibers are shown in Figure S1, where all the fibers were aligned like the 3D network, with the existence of macropores; they exhibited a glossy and homogeneous surface morphology. As shown in Figure b, the pure WO3-coated fibers exhibited a flake-like surface microstructure and the formation of a few nanoplates with well-grown facets and sharp facet edges, which indicated the typical morphology of the crystalline WO3 nanoplates. The 1Cu–WO3 and 2Cu–WO3 fibers exhibited densely covered large grains that could be composed of the CuO–WO3 composite as shown in Figure c,d, respectively. In addition, the 3Cu–WO3 fibers exhibited a porous microstructure and agglomerated grains. It could be understood that in the case of high-concentration Cu-doped WO3 fibers, the Cu atoms could rapidly diffuse into the adjacent WO3 phase during the RTA process to occupy the W atom sites.[33] This would lead to the formation of localized pores, as shown in Figure e, forming the CuWO4 phase, as confirmed by the XRD results.

Chemical Composition and Optical Properties of Cu–WO3 Fibers

X-ray photoelectron spectroscopy (XPS) measurements were conducted to investigate the surface chemical composition and state of the WO3 and Cu–WO3-coated fibers annealed at 873 K. As shown in Figure , the survey spectrum showed the peaks of W, Cu, and O in Cu–WO3-coated fibers. With an increase in Cu concentration, the Cu–WO3 fibers exhibited a very intense signal of Cu along with the presence of strong satellite peaks, and the W signal was significantly decreased. The high-resolution W 4f spectra of WO3, 1Cu–WO3, and 2Cu–WO3 were deconvoluted into three peaks with binding energies of 35.4, 37.5, and 41.0 eV, corresponding to W 4f7/2, W 4f5/2, and the satellite peak of WO3, respectively, which were consistent with the W6+ state.[34] Moreover, the splitting energy of the W 4f doublet at 2.1 eV also confirmed the W6+ state. The peak at 41.0 eV could be attributed to the binding energy of W 5p3/2.[35,36] However, a slight shift in the W 4f spectra toward a higher energy of about 0.4 eV for 3Cu–WO3 was observed, which is in good agreement with a similar observation reported by Thongpan et al. for the CuWO4 phase.[37] The W 4f spectra of 3Cu–WO3 were fitted into two peaks at 35.8 and 37.9 eV corresponding to W 4f7/2 and W 4f5/2 of W6+ in the form of CuWO4, respectively. The deconvolution of high-resolution XPS spectra of Cu 2p for 1Cu–WO3 and 2Cu–WO3 showed that two major peaks centered at 933.65 and 953.45 eV are assigned to Cu 2p3/2 and Cu 2p1/2 of Cu2+, respectively, which confirmed the CuO phase.[38,39] The binding energy difference between the two peaks Cu 2p3/2 and Cu 2p1/2 was 19.8 eV, which strongly indicated the presence of Cu2+.[40] In addition, the presence of strong satellite peaks around 940–944 and 962 eV in the Cu–WO3/SGFs corresponds to Cu 2p3/2 and Cu 2p1/2, respectively, which indicate an open Cu 3d9 shell corresponding to the bivalent (Cu2+) state (CuO phase).[27,39,40] However, the peaks centered at 931.6 and 951 eV are attributed to Cu 2p3/2 and Cu 2p1/2 of Cu+,[41] which indicates the presence of a small amount of Cu2O phase in the 1Cu–WO3-coated fibers. The relative concentrations of Cu+ and Cu2+ species present on 1Cu–WO3 was quantified to be about 18% (Cu+) and 82% (Cu2+) by considering the curve-fitting of the total Cu 2p emission peaks including the shake-up satellite peaks as reported by Biesinger et al. using the following equations[42,43]where B is the area of the shake-up peak and A is the total area of the Cu 2p main peak. The main emission line (A) contains both Cu2+ (A1) and Cu+ (A2) contributions, but the satellite intensity (B) is entirely from Cu2+.[42] In the case of 3Cu–WO3, the Cu 2p peaks were slightly shifted (about 0.7 eV) toward higher binding energies, and they were deconvoluted into two major peaks centered at 934.4 and 954.2 eV, attributed to Cu 2p3/2 and Cu 2p1/2, respectively, due to Cu2+ in the lattice sites of CuWO4.[44] The observed peak shifting in the Cu 2p spectra of the CuWO4 phase is in good agreement with the previous reports.[45] The O 1s exhibited peaks at about 531 and 530 eV in both WO3 and Cu–WO3 fibers, which were attributed to the surface-adsorbed oxygen and the characteristic peak of metal ion-associated oxygen (W–O and Cu–O), respectively.[46]

Figure 2

XPS survey spectra and high-resolution spectra of WO3 and Cu–WO3-coated SGFs annealed at 873 K.

Raman spectroscopy is a well-known non-destructive method that is commonly used to investigate material properties including the structure and phase change owing to its high sensitivity to the electronic structure. Therefore, Raman spectroscopy was further utilized to evaluate the chemical bonds of Cu–WO3. The Raman spectra displayed in Figure revealed five active modes in 1Cu–WO3 and 2Cu–WO3/SGF centered at about 134, 272, 326, 712, and 806 cm–1. The peak that appeared at 134 cm–1 is ascribed to the lattice vibration of WO3; 272 and 326 cm–1 are assigned to the δ(O–W–O) bending modes. The other two peaks centered at 712 and 806 cm–1 correspond to the ν(O–W–O) vibrational stretching modes of the monoclinic WO3.[47,48] The peak at 978 cm–1 in the 2Cu–WO3/SGF is attributed to the symmetric stretching mode of W=O bonds resulting from the adsorbing water molecules in the air.[37] Based on group theory calculations, the triclinic CuWO4 exhibits 36 vibrational modes (ΓRaman+infrared = 18Ag + 18Au).[49] For the 3Cu–WO3/SGF, we have observed different Ag active Raman modes, among which the highest intense peak is observed at about 905 cm–1, which is ascribed to the W–O stretching vibration ν1(Ag). The other peaks, 313 cm–1 attributed to the scheelite tungstate vibration ν2(Ag) in the triclinic lattice and 780 cm–1 associated with the scheelite structure vibration ν3(Ag), confirm the triclinic CuWO4 phase.[50] The peaks at 357 and 398 cm–1 correspond to the deformation modes of r(Bg) and δ(Ag), respectively. The Raman peaks observed in 3Cu–WO3-coated fibers are in good coincidence with the previous reports of the CuWO4 phase.[50] Consequently, the Raman spectroscopy studies are well corroborated with the XRD and XPS results, which confirm that the higher doping concentration of Cu leads to CuWO4 phase formation.

Figure 3

Raman spectra of the Cu–WO3/SGFs annealed at 873 K: (a) 1Cu–WO3, (b) 2Cu–WO3, and (c) 3Cu–WO3.

Visible light absorption was an important precondition for efficient visible-light-active photocatalysts. Figure shows the UV–visible absorption spectra of the WO3 and Cu–WO3-coated fibers annealed at 873 K. The absorption spectra vary with Cu doping. It was observed that the red shift in the absorption spectra occurred with an increase in the Cu content, which was favorable for visible light photocatalysis. The optical band gaps of the WO3 and Cu–WO3 fibers were determined using Tauc’s equation as followswhere α is the absorption coefficient, hϑ is the incident photon energy, A is an arbitrary coefficient, n = 1/2 for direct and n = 2 for indirect recombination, and Eg is the optical band gap. The optical band gap of WO3, 1Cu–WO3, 2Cu–WO3, and 3Cu–WO3 was determined to be approximately 2.95, 2.82, 2.51, and 2.36 eV, respectively, by the linear extrapolation of Tauc’s plot (αhϑ)2 versus hϑ (Figure ). The variation in the optical band gap implied that Cu doping reduced the optical band gap, which was significant for visible-light-active photocatalysis. Similar observations were reported for Cu-doped WO3.[51] Narrowing the band gap was presumed to enhance the absorbance of visible light. The 3Cu–WO3 fibers showed the lowest band gap among all the fibers (∼2.36 eV), which was due to the formation of CuWO4, as confirmed from the XRD results. Moreover, the obtained band gap of 3Cu–WO3 was in agreement with the other reports on CuWO4 nanostructures, whose band gap was approximately 2.3 eV.[52,53]

Figure 4

UV–visible absorption spectra and Tauc’s plot of undoped and Cu-doped WO3 structures annealed at 873 K.

Photocatalytic Activity of Cu–WO3 Fibers

The photocatalytic activities of the WO3 and Cu–WO3-coated fibers annealed at 873 K were evaluated through methylene blue (MB) degradation under visible light irradiation. As shown in Figure a, it could be distinctly observed that the 2Cu–WO fibers exhibited the highest efficiency of photocatalytic degradation compared to other fibers. The undoped WO3 fibers degraded about 18.5%, and 2Cu–WO3 fibers degraded 88.0% of MB in 180 min. It is known that the efficiency of photogenerated electron–hole pair separation plays a crucial role in increasing the photocatalytic degradation rate of the organic dye. Langmuir–Hinshelwood’s first-order kinetic model was used to determine the photocatalytic activity for MB degradation as follows[54]where C0 is the initial concentration of MB and C is the concentration of MB in the presence of fibers after visible light exposure. The MB degradation kinetics of undoped WO3 and Cu–WO3 fibers are shown in Figure b, and the fitted data exhibited a linear behavior for all the fibers, indicating that the MB degradation reactions followed the pseudo-first-order kinetics model. The photocatalytic activity rate constants (k) of WO3, 1Cu–WO3, 2Cu–WO3, and 3Cu–WO3 fibers were 0.0011, 0.0081, 0.0118, and 0.0023 min–1, respectively. The WO3-coated fibers exhibited the lowest rate constant, which could be due to rapid electron–hole recombination. The 2Cu–WO3 fibers exhibited approximately 10.7 times higher activity rate constant than pure WO3. In fact, the CB of WO3 is more positive than the O2/O2•– redox potential (−0.33 eV), which indicates that the adsorbed O2 molecules on the surface of WO3 cannot be reduced to superoxide radicals O2•–. The valence band (VB) of CuO (+1.43 eV) is less positive than OH–/•OH (+2.38 eV), which indicated that CuO could not directly produce the •OH radicals. Hence, the observed enhanced photocatalytic activity could not be explained by the conventional p–n heterojunction formation of CuO–WO3. A more credible mechanism could be the Z-scheme heterojunction formation; in the case of Cu–WO3 samples, the photogenerated electrons from the CB of WO3 could recombine with the holes in the VB of CuO. As the CB of CuO was more negative than the redox potential, the electrons could participate in the redox reactions to form the superoxide radical O2•–. The holes in WO3 could contribute to the oxidation process to produce the •OH radicals because the VB of WO3 was more positive than the oxidation potential to yield •OH. Thus, the electrons in the CB of CuO and holes in the VB of WO3 could effectively perform redox/oxidation reactions. 2Cu–WO3 possessed a higher concentration of Cu than 1Cu–WO3. This led to the formation of higher phase fraction of CuO while forming the CuO/WO3 interfaces. This resulted in higher photocatalytic activity. Thus, the enhanced photocatalytic activity conformed to the CuO–WO3 Z-scheme heterojunction mechanism, which greatly enhanced the separation of photogenerated electrons–holes. Moreover, the (002) crystal orientation of WO3 in 2Cu–WO3 predominantly accelerated the transfer of photoinduced charge carriers to ascertain high reactivity, compared to other orientations, as reported by several researchers.[55,56] It is to be noted that the surface area of the photocatalyst greatly influenced the photocatalytic activity; herein, the 3D network-like fiber substrates obviously possess a high surface area compared to the 2D substrates. In the case of 3Cu–WO3 fibers, the lower photocatalytic activity could be due to the CuWO4 phase formation and porosity, which could be attributed to the smaller volume fraction of CuO. In general, the CuWO4 phase has a CB edge of approximately 0.2–0.44 V versus NHE, which is more positive than the O2 reduction potential;[57] hence, the lower photocatalytic activity can be obviously observed in 3Cu–WO3 fibers. Overall, the 2Cu–WO3-coated fibers exhibited better photocatalytic activity among the Cu–WO3-coated SGFs.

Figure 5

(a) Change in concentration vs reaction time for the degradation of MB solution under visible light irradiation and (b) MB degradation kinetics of the WO3 and Cu–WO3-coated SGF.

Silver Dewetting on CuO–WO3 Fibers

Ag nanoparticles were dispersed over CuO–WO3 (2Cu–WO3) fibers to further enhance the photocatalytic activity of the fibers, as shown in Figure a. A 10 nm-thick Ag film was sputtered on the CuO–WO3 fibers and subsequently vacuum-annealed by RTA at 873 K for 5 min to obtain the island formation and growth of Ag in the form of isolated nanoparticles. The crystal structure of the Ag-dispersed CuO–WO3 fibers was studied by XRD, as shown in Figure b. The XRD peaks observed at 38.2, 44.4, and 64.6 were attributed to the (111), (200), and (220) planes of silver (JCPDS no. 01-087-0720). In addition, the other peaks were similar to the diffraction pattern observed in 2Cu–WO3 composed of the CuO–WO3 composite phase, where the most intense peak of the (002) facet was observed, indicating that the growth of the WO3 crystal was along the [001] direction. The high-magnification (30 000×) surface microstructure of the Ag-dispersed CuO–WO3 fibers clearly indicated the formation of evenly distributed Ag nanoparticles over the surface of CuO–WO3, as shown in Figure c. The corresponding low-magnification (5000×) image is shown in Figure S2. The size distribution histogram of the Ag nanoparticles is shown in Figure d, where the solid line represents the Gaussian fitting to a normal distribution curve. Based on this fitting, the mean size (D) and standard deviation (σ) of the Ag nanoparticles were 78.8 ± 3 and 46.2 nm, respectively. The surface coverage of the Ag nanoparticles was determined to be approximately 24.9% using ImageJ software. The high-resolution XPS spectrum of Ag 3d is shown in Figure e, where two peaks with binding energies centered at 368.55 and 374.55 eV were observed with an energy gap of 6.0 eV, which corresponded to Ag 3d5/2 and Ag 3d3/2 of metallic Ag, respectively.[58] This result was in good accordance with the results from the XRD analysis, indicating the metallic nature of Ag.

Figure 6

(a) Schematic illustration of the fabrication of Ag-dispersed CuO–WO3/SGFs (Ag@CuO–WO3/SGFs), (b) XRD pattern, (c) high-magnification (30 000×) FE-SEM surface morphology of Ag-dispersed CuO–WO3, (d) Ag nanoparticle distribution histogram, and (e) high-resolution XPS spectra of Ag 3d of Ag-dispersed CuO–WO3.

Additional structural information of Ag-dispersed CuO–WO3 fibers was examined using high-resolution transmission electron microscopy (HR-TEM), and TEM sampling was performed using the focused ion beam (FIB) H-bar technique, as shown in Figure a. Before milling, the surface of the target region of the Ag-dispersed CuO–WO3 fiber was coated with a thin carbon layer, followed by a platinum protective coating to avoid milling of the target region. Milling was performed with a high-energy Ga ion beam. The milled specimen was picked and mounted on a TEM specimen grid. Subsequently, the specimen was milled again to obtain a very thin piece (electron transparent) with a sample thickness of less than 100 nm. The backscattered electron (BSE) images of the high-energy milled specimen (inset of Figure b) and final sample preparation (the rectangular region shown in the inset) are shown in Figure b. The cross-sectional TEM image (Figure c) shows the existence of Ag nanoparticles on the surface, and the thickness of the Ag-dispersed CuO–WO3 coating was approximately 380 nm. From the HR-TEM image (Figure d), a small near-spherical silver nanoparticle in the range of ∼50 nm was observed, which formed an interface with the monoclinic WO3 phase. The crystal structure of silver was determined by the lattice fringes, and the corresponding fast Fourier transformation (FFT) and inverse FFT (IFFT) images of the marked region are depicted in Figure d. The interplanar distance of the lattice fringes was determined to be d = 0.235 nm, which corresponded to the d-spacing of the Ag(111) plane. Similarly, the FFT and IFFT images of WO3 are shown in Figure d, which corresponded to monoclinic WO3 with an interplanar distance d = 0.354 nm of the (002) plane. Based on the above results, Ag nanoparticle dispersion through Ag solid-state dewetting was successfully obtained. As shown in the schematic in Figure , the as-deposited Ag layer on the CuO–WO3-coated fibers comprised numerous Ag adatoms. When subjected to RTA, the interfacial free energy between Ag and CuO–WO3 spontaneously induced the separation and isolation of Ag (Ag solid-state dewetting). More specifically, from a thermodynamic point of view, dewetting occurred when the surface energy (Es) was positive, and the driving force was defined as followswhere γAg, γ(CuO–WO, and γAg,(CuO–WO are the surface energy densities of Ag, CuO–WO3, and Ag and CuO–WO3, respectively. The evolution of isolated Ag nanoparticles in an Ag thin film sputtered on a glass substrate is depicted in Figure S3 for clearly understanding the formation of the Ag dewetting-induced Ag nanoparticle. The as-deposited Ag layer showed unstable or metastable Ag adatoms with a relatively smooth microstructure, which was ascribed to insufficient surface diffusion to obtain a low-energy configuration. The Ag adatoms started to diffuse on the surface with an increase in temperature and finally formed isolated island-like spherical or near-spherical nanoparticles owing to the lower surface energy. Surface diffusion was the major driving parameter for the Ag dewetting behavior, and the diffusion coefficient (Ds) of the atom could be related to the temperature (T) as follows[25]where D0 is the pre-exponential factor of the surface diffusion coefficient, Ea is the activation energy, k is the Boltzmann constant, and T is the applied temperature. From the above equation, it is obvious that Ds is related to T. In the process of surface diffusion, voids were eventually formed, and the growth rate of the surface diffusion-induced voids accelerated with temperature.[59] Thus, the surface diffusion coefficient increased with an increase in temperature, which indicated that the Ag adatoms had a high probability of coalescence with the adjacent atoms to form a 3D island-like nanoparticle. In addition, the Ag nanoparticle density decreased, and the voids grew larger. The 873 K annealed films exhibited homogeneously distributed Ag island-like nanoparticles. As the CuO–WO3-coated fibers were composed of large grains and a seemingly rough surface, the Ag nanoparticle growth was physically limited to each grain of the substrate (CuO–WO3 surface), and they could not physically migrate (hopping) from one grain to another, as illustrated in the schematic in Figure . As a result, the formation of large nanoparticles, unlike the Ag nanoparticles which formed over a flat and smooth surface (for instance, glass substrate), was restrained. Therefore, in the case of CuO–WO3-coated fibers, small island-like Ag nanoparticles were formed owing to the localized surface diffusion-driven Ag dewetting, as observed in the FE-SEM image (Figure c).

Figure 7

(a) Schematic cross-sectional view of the Ag-dispersed CuO–WO3/SGF and the FIB-milled TEM sampling, (b) BSE SEM micrographs of FIB-milled Ag-dispersed CuO–WO3 fiber cross-sectional view, (c) TEM cross-sectional micrograph, and (d) HR-TEM image of the Ag-dispersed CuO–WO3/SGF.

Figure 8

Cross-sectional and plane-view schematic representation for before and after RTA of a thin Ag layer-coated CuO–WO3/SGF, illustrating the localized surface diffusion-driven Ag dewetting (3D island-like Ag nanoparticle formation).

Influence of Dewetted Ag NPs on Photocatalytic Activity

The photocatalytic activity and MB degradation kinetics of the Ag-loaded CuO–WO3 fibers are shown in Figure a. It was observed that 58% of MB was decomposed within the first 30 min and 98% degradation was observed after 180 min, which was much higher than that of the Cu–WO3 fibers. The photocatalytic activity rate constant of WO3, Cu–WO3, and Ag-dispersed CuO–WO3 fibers is shown in Figure b, where the Ag-dispersed CuO–WO3 fibers exhibited remarkable photocatalytic activity among all the fibers with the highest MB degradation rate constant (k = 0.0205 min–1), which is approximately 1.73 times higher than that of 2Cu–WO3 fibers and 18.6 times higher than that of pure WO3. The photocatalytic MB degradation stability of the Ag-dispersed CuO–WO3 fibers was determined under the identical conditions and shown in Figure S4. The efficiency was slightly changed with the degradation efficiency of about 90% during the third cycling, while the first cycling experiment showed about 98% efficiency. The XRD pattern and the FE-SEM image of the Ag@CuO–WO3 sample after the photocatalytic MB degradation stability test for the third cycling are shown in Figure S5.

Figure 9

(a) Photocatalytic activity of Ag-dispersed CuO–WO3; (b) MB degradation rate constant of WO3, Cu–WO3, and Ag-dispersed CuO–WO3 photocatalysts; (c) PL spectra of the WO3, Cu–WO3, and Ag-dispersed CuO–WO3/SGF; and (d) schematic illustration of the proposed LSPR and Z-scheme hybrid heterojunction mechanism in the Ag-dispersed CuO–WO3/SGF photocatalyst.

The photocatalytic efficiency relies critically on the generation and recombination rates of electron–hole pairs upon light excitation. Photoluminescence (PL) spectroscopy is an important characterization tool for the investigation of surface processes involving photogenerated electron–hole pair recombination in semiconductors. The lower PL intensity implied more effective separation of electron–hole pairs; thus, higher photocatalytic activity could be achieved. PL was measured for WO3, Cu–WO3, and Ag-dispersed CuO–WO3 fibers, as shown in Figure c, and the fibers showed emission peaks around 530 nm. The pure WO3 fibers exhibited the strongest PL intensity due to the severe electron–hole recombination, while the Cu-doped WO3 fibers exhibited lower PL intensity than WO3. 2Cu–WO3 exhibited the lowest PL intensity among all the Cu-doped WO3 fibers, indicating the effective retardation of electron–hole recombination. The PL intensity drastically decreased in Ag-dispersed CuO–WO3 fibers, which indicated that Ag loading greatly inhibited the recombination rate of the electron–hole pairs, resulting in effective charge separation at the respective interfaces of Ag/CuO/WO3. The mechanism for the enhanced photocatalytic activity of the Ag-dispersed CuO–WO3 fibers could be understood as follows: the schematic is shown in Figure d on the synergetic effect of the Z-scheme and LSPR. The fast electron–hole pair recombination rate was greatly inhibited by the Z-scheme system of CuO–WO3, and the strong LSPR effect of dewetted Ag nanoparticles functions as visible light harvesting and electron-generating centers. The LSPR-excited electrons would be enriched on the surface of the dewetted Ag nanoparticles. As a result, the electrons transferred from Ag to the CB of the adjacent semiconductor (WO3 or CuO at the respective interface with Ag) and increased the steady-state concentration of “chemically useful” (i.e., charge carriers at the surface of the semiconductor that participated in photocatalytic reactions) energetic charge carriers.[60] In this study, the dispersed Ag nanoparticles were approximately 79 nm, implying that the quadrupolar LSPR could be the most probable mode in which photon absorption was predominant rather than hot carrier generation. As a result, the excess electrons could also contribute to the production of more O2•– (as shown in the schematic in Figure d). Bastús et al. and Seemala et al. reported that the Ag nanoparticle size and distribution greatly influenced the LSPR effect.[61,62] According to their reports, hot carrier generation was predominantly observed with relatively higher hot carrier density in relatively small Ag nanoparticles (∼67 nm), which exhibited dipolar plasmon resonance. Quadrupole moments could be observed above this critical size (∼67 nm), as the size of the Ag nanoparticle increased until ∼189 nm. This facilitated O2 dissociation via direct interactions with the strong electric fields at the Ag surface due to the predominant photon absorption that effectively promoted photocatalysis.[62] Thus, the strong visible light absorption and charge separation ability of this Ag-dispersed CuO–WO3 structure greatly enhanced the photocatalytic activity. Consequently, the Ag dispersion onto the fibers could remarkably improve the absorption of visible light due to the LSPR effect and significantly increased the probability of separation of charge carriers. Thus, the photogenerated electron–hole pairs were accompanied by the reduction/oxidation process for the degradation of the organic dye.

Conclusions

In summary, a novel and effective hybrid composite photocatalyst comprising a plasmonic metal (Ag)-dispersed CuO–WO3 Z-scheme heterojunction was successfully coated on reusable and recyclable 3D network-like SGFs by an industrially well-adopted sputtering process. The XRD results revealed that the high concentration of Cu in 3Cu–WO3 led to CuWO4 phase formation at 873 K. Homogeneously dispersed, localized surface diffusion-driven Ag-dewetted nanoparticles of approximately 79 nm were successfully obtained. The MB photocatalytic degradation efficiency of Ag-dispersed CuO–WO3/SGF was about 18.6 times higher than that of pure WO3-coated fibers. The construction of the Z-scheme heterojunction system decreased the number of photogenerated electron–hole pairs. The dewetted Ag nanoparticle dispersion facilitated effective electron–hole transfer across the respective interfaces of Ag/CuO and Ag/WO3, and it was certainly beneficial to retard the rapid recombination of electron holes from WO3 and CuO. This work provides insights into the practical applications of highly efficient fiber-supported visible-light-active photocatalysts for environmental remediation.

Experimental Section

Fabrication of Cu–WO3 and Ag-Dispersed CuO–WO3 Photocatalysts on SGFs

A 3D network-like commercially available SGF sheet was used as a substrate for the deposition of WO3 and Cu-doped WO3 by radio frequency magnetron sputtering at room temperature. A 2 in. planar tungsten oxide target and a Cu chip (99.999% purity, chip size: 1 × 1 × 0.5 mm)-embedded mosaic-type tungsten oxide target were used for the deposition of WO3 and Cu–WO3 over the SGF, respectively, under the same conditions. An Ar gas flow of 230 sccm, a target power of 150 W, a substrate rotation of 23 rpm, a target–substrate distance of 90 mm, and a deposition time of 30 min were maintained as deposition parameters. Before sputtering, the chamber was evacuated to a base pressure below 2 × 10–6 Torr and pre-sputtered for 10 min to remove any adsorbed contaminants on the surface of the sputter target. Finally, the WO3 and Cu–WO3-coated fibers were annealed at 873 K in ambient air for 30 min using the RTA system to obtain crystalline CuO–WO3 composite nanostructures. The Cu concentration in the Cu–WO3-coated fibers was controlled by the mosaic structure of the WO3 target by varying the number of Cu chips (1, 2, and 3) on the WO3 target, which were named as 1Cu–WO3, 2Cu–WO3, and 3Cu–WO3, respectively. The Cu doping concentration was approximately 3–10 at. % based on the area fraction of Cu over the WO3 target, respectively. Finally, an ultrathin Ag layer (∼10 nm) was sputtered over CuO–WO3 (which was obtained by annealing 2Cu–WO3 fibers annealed at 873 K) by the DC magnetron sputtering system using a 2 in. diameter Ag target of 99.999% purity with a target power supply of 30 W, and these fibers were subsequently annealed under vacuum for 5 min at 873 K to obtain Ag nanoparticle dispersion over the CuO–WO3/SGF.

Material Characterization and Organic Dye Degradation

The crystal structure was examined by XRD with a Rigaku SmartLab X-ray diffractometer using Cu Kα radiation at a scanning speed of 4° min–1 over the 2θ range of 15–70°. The surface morphology was investigated using FE-SEM (TESCAN-MIRA3, Korea). XPS measurements were carried out with a Thermo Scientific K-Alpha spectrometer using an Al Kα X-ray source with a constant analyzer mode to study the chemical composition of the constituent elements. Raman measurements were carried out by HORIBA LabRAM HR-800 with 514 nm laser excitation. TEM observations were performed using an FEI TECNAI G2 F20 (operating voltage: 200 kV) to determine the structure and cross-sectional microstructure of the dewetted Ag nanoparticles dispersed on CuO–WO3-coated fibers. The TEM sample was prepared by the H-bar technique using a TESCAN LYRA3 high-resolution FIB-SEM. The optical absorbance spectra were recorded using a UV–vis–NIR spectrophotometer (UV-3600, Shimadzu). The PL properties were studied using a HORIBA LabRAM HR-800, HORIBA France SAS. The photocatalytic activity was investigated by measuring the degree of decomposition of MB solution in a double-walled glass beaker (40 mL MB solution, 10 ppm) under visible light illumination using a 500 W Xe lamp with a 400 nm filter.