Introduction
Titanium dioxide is a semiconductor that has attracted increasing interests in photocatalytic and photovoltaic applications [
1‒
4] because of the great oxidative potential of its positive holes (
E = +3.0 V versus standard hydrogen electrode) [
5], especially in the anatase form. However, due to the wide band-gap of 3.2 eV (for the anatase form) [
6], TiO
2 has trivial response to the visible light (wavelength: 390–700 nm) [
7,
8] hindering the development of visible-light-driven photocatalysts based on TiO
2 for energy and environmental applications [
6,
9]. Various strategies [
8,
10‒
14] such as (i) metal (e.g., transition metals Cu, Co) [
11] with and non-metal doping (e.g., p-block elements N, B, F) [
12,
15,
16] and (ii) crystal structure and morphology engineering [
13,
14] are being developed to adjust the band structure and trap states of TiO
2 with an aim to enhance the response of the TiO
2-based photoelectrochemical materials in the visible region. Recently, the integration of plasmonic gold or silver nanoparticles (NPs) with TiO
2 [
8,
17‒
22] have been proposed to enhance the photocatalytic and photovoltaic activity in the visible region due to the strong surface plasmon resonance (SPR) excitation of such metal NPs, by which can enhance the concentration of charge carriers. One specific example presented by Zhang et al. [
22] is the hybrid plasmonic Au nanoparticle (NP)-loaded hierarchical hollow porous TiO
2 spheres, which showed the enhanced the overall catalytic activity in 4-nitrophenol reduction [
22]. In general, the energy-transfer mechanisms SPR were concluded as: (i) direct charge injection from metal to semiconductor, (ii) near-field electromagnetic and (iii) resonant photon-scattering mechanisms [
21], depending largely on the geometric configuration of the composites. The synergic effect of plasmonic metal NPs and TiO
2 semiconductor can be manipulated by controlling the relative arrangement of the two building blocks. For instance, the electric field intensity enhancement was found dropped significantly by more than one order of magnitude when the distance between adjacent plasmonic Ag NPs were increased from 1 to 10 nm [
21]. However, to stabilise nanosized metal particles on TiO
2 absorber against their high tendencies to aggregate presents a challenge to design stable hybrid nanostructures for practical settings. TiO
2 hollow spheres (HSs) [
20,
23,
24] have attracted increasing interests due to their tunable structural properties (e.g., size, shell thickness and crystallinity) and hence their photoelectrochemical properties [
23,
24]. The unique void of TiO
2 HSs was also employed to encapsulate and stabilise plasmonic metal NPs [
17,
20,
25‒
27] yielding yolk (metal NP)-shell (TiO
2) arrangements, e.g., Au(TiO
2 HSs (‘(’ represents the encapsulation of Au NP in HSs), with improved thermal stability of the metal cores. Though few excellent samples of the yolk-shell Au(TiO
2 HSs nanoreactors have shown good performance in photocatalytic oxidation of CO [
20] and water splitting [
26], their applications in liquid phase environmental remediation with bigger organic molecules are still limited due to (i) the restricted accessibility of the metal core and (ii) the inadequate availability of metal species (usually the single core configuration was designed to avoid aggregation).
Molecular dynamics simulations have shown that the surface layer of a metal NP possessed liquidlike properties [
28] favouring the aggregation of adjacent metal NPs. In fact, the outer surface of TiO
2 HSs is convex in all directions perfect to provide spatial confinement for separating metal drops. To date, the confinement effect of metal NPs created by the curvature of the TiO
2 HSs has been mostly neglected, and only a few examples showed the effectiveness of using the outer surface of TiO
2 HSs to stabilise plasmonic metal NPs [
8]. Herein, we demonstrate the design and synthesis of plasmonic Au-TiO
2 HSs nanostructure with Au NPs deposited on both sides of the anatase TiO
2 HS, i.e
., a single encapsulated Au core and multiple Au NPs confined by the convex curved outer surface of the TiO
2 HS, denoted as Au@Au(TiO
2 (@ represents the deposition of Au NP on HSs), to improve the photocatalytic efficiency of the system.
Materials and methods
Synthesis of core-shell Au(SiO2
All glassware was thoroughly cleaned with aqua regia (three parts HCl, one part HNO3) for 12 h and rinsed with deionised water (DI). 5 mL sodium citrate (1%, Tianjin Chemical Reagent No 1 Plant) was dispersed into 50 mL HAuCl4∙3H2O (J&K Scientific) aqueous solution (4 × 10-4 mol/L) under sonication and then heated to 95°C. After 5 min, 37.5 mL of 400 mmol/L myristyltrimethylammonium bromide (TTAB, Sigma-Aldrich,≥99%) was added to the solution and the mixture was maintained at 95°C for 15 min under vigorous stirring. The obtained TTAP-capped Au aqueous solution (20 mL), tetraethyl orthosilicate (28%, Tianjin Chemical Reagent No 1 Plant, 0.96 mL), dehydrated ethanol (99.8%, Tianjin Guangfu Fine Chemical Institute, 23 mL) and ammonia solution (26%, Tianjin Guangfu Fine Chemical Institute, 0.62 mL) were then mixed and stirred for 4 h at room temperature. The precipitated silica nanoparticles were extracted by centrifugal separation and washed with ethanol then re-dispersed in 5 mL of ethanol.
Synthesis of yolk-shell Au(TiO2
Au(SiO2 was dispersed in a mixture of hydroxypropyl cellulose (HPC, Tokyo Chemical Industry Co., Ltd., 0.1 g), ethanol (20 mL) and DI (0.1 mL) and stirred for 40 min. Titanium tert-butoxide (98%, Tianjin Chemical Reagent No 1 Plant, 1 mL) in ethanol (5 mL) solution was added in the mixture at a rate of 0.5 mL·min-1. The mixture was stirred at 900 r·min-1 and 85°C under refluxing condition for 100 min. The final product was separated by centrifugation, washed with ethanol, and kept in 5 mL of ethanol to form Au (SiO2(TiO2 nanocomposites. The Au(SiO2(TiO2 was calcined in air at 500°C (2°C·min-1) for 2 h to remove all organic compounds and crystallise the amorphous TiO2. The calcined samples were dispersed in 20 mL water under sonication and heated to 50°C. 1.5 mL aqueous NaOH solution (2.5 mol/L, Tianjin Guangfu Fine Chemical Institute) was added in the aqueous Au(SiO2 (TiO2 solution and the solution was heated to 70°C. Finally, aqueous NaOH solution (2.5 mol/L, 1 mL) was added and the resulting solution was stirred for 6 h to remove the SiO2 core and obtain yolk-shell Au(TiO2.
Synthesis of Au@Au(TiO2
Yolk-shell Au(TiO2 (0.1 g) and sodium citrate (1%, Tianjin Chemical Reagent No 1 Plant, 5 mL) were dispersed into 50 mL HAuCl4∙3H2O aqueous solution (4 × 10-4 mol/L) under sonication. The mixture was heated to 95°C and maintained at 95°C for 20 min under vigorous stirring. The product was separated by centrifugal separation, washed with DI and dried at 100°C for 10 h. The same procedure was used for depositing Au NPs on other supports.
Characterisation of materials
N2 adsorption-desorption at -196°C was performed using a Micromeritics Tristar 3000 analyser to determine the specific surface area (the Brunauer–Emmett–Teller method) and the pore size distribution (the Barrett-Joyner-Halenda method). X-ray diffraction (XRD) patterns of materials were recorded on a Bruker D8 Focus diffractometer (nickel-filtered CuKa1 radiation, l = 1.5406 Å, 40 kV, 40 mA) at a scanning rate of 0.02° per step and 0.15 s per step in a 2q range of 10°‒85° (and 5°‒65°). Transmission electron microscopy (TEM) imaging was completed using an FEI Tecnai G2 F20 operating at 100 kV. Samples were dispersed in ethanol and pipetted onto the copper grid-supported transparent carbon foil and dried prior to imaging. Scanning electron microscopy imaging of materials was undertaken using a JEM-2100F. Diffuse reflectance spectra (DRS) of materials were obtained using a SHIMADZU UV-2550 spectrophotometer equipped with a 60 nm diameter integrating sphere using BaSO4 as the reflectance sample.
Evaluation of photocatalytic activity
The photocatalytic activity of materials was evaluated by photodegradation of methylene blue (MB, 98%, Tianjin Damao Chemical Co., Ltd.) and phenol (Tianjin Damao Chemical Co., Ltd.). 50 mg photocatalyst was firstly added into a 100 mL quartz photoreactor containing the solution with the substrate (12 mmol·L-1) and ultrasonicated for 2 min. The mixture was magnetically stirred for 30 min in the dark to ensure good dispersion of the photocatalyst in the solution the adsorption-desorption equilibrium between substrate molecules and the photocatalyst. The same 50 mg photocatalyst was also used for phenol solution (100 µmol·L-1). A visible light source (420≤l≤780 nm) was introduced by a 300 W xenon lamp (Beijing Perfectlight Technology Co. Ltd, LS-SXE300CUV, equipped with an AM 1.5-filter power intensity= 100 mW·cm-2) 10 cm above the liquid surface. The ambient temperature of the reaction was maintained by water cooling during the reaction. Every 30 min, a 2 mL sample was taken and analysed for the concentration of the substrate. For MB and phenol, the absorbance intensity at their maximum absorbance wavelength of lmax = 664, 270 nm was monitored using a SHIMADZU UV-2550 spectrophotometer.
Results and discussion
The synthesis and morphological evolution of the yolk-shell Au(TiO
2 nanostructure and the corresponding TEM images were illustrated in Figs. 1 and 2. A three-step method was used (Fig. 1) to synthesise the yolk-shell Au(TiO
2 nanostructure, i.e., (i) Synthesising Au(SiO
2 nanoparticles using the Stöber method (Fig. 2(a)) [
29]; (ii) Coating Au(SiO
2 with anatase TiO
2 via hydrolysis-condensation and calcination (Fig. 2(b)) [
17,
23]; and (iii) Removing SiO
2 core to render the yolk-shell Au(TiO
2 with NaOH-based chemical etching (Fig. 2(c)). The method enabled the solitary inclusion of one myristyltrimethylammonium bromide-capped Au NP of 7±2 nm. The average shell thickness and diameter of the anatase TiO
2 shell are about 25±2 and 140±10 nm, respectively. The developed TiO
2 shell was highly porous with a specific surface area of 187 m
2·g
-1 (Figs. S1, S2 and Table S1, cf. Electronic Supplementary Material (ESM)).
To further utilise the outer surface of the porous TiO
2 shell, Au NPs (ca
. 1 wt-%) were deposited on the yolk-shell Au(TiO
2 yielding the Au NP core-TiO
2 shell-Au NPs hybrid structure as shown in Fig. 2(d). Au NPs with similar diameters were found well-dispersed on the outer surface of anatase TiO
2 shell in the as-prepared Au(TiO
2 (Fig. 2(d)). The convex outer surface of TiO
2 HSs was believed to avoid the aggregation of Au NPs. High-resolution transmission electron microscopy (HR-TEM) analysis also showed the well-resolved crystalline Au NPs with a fine crystal lattice spacing of 0.2355 nm (inset of Fig. 2(d)), corresponding to (111) plane of the face-centred cubic gold. For the purpose of comparison, TEM images of pure SiO
2 core, core-shell SiO
2(TiO
2, TiO
2 HSs and Au@TiO
2 HSs were prepared and shown in Figs. 2(e‒h). It was found that the hollow spherical structure of porous TiO
2 remained largely intact during the synthesis (Fig. S1). XRD patterns of TiO
2 HSs, Au@TiO
2, Au(TiO
2 and Au@Au(TiO
2 (Fig. S3, cf. ESM) showed characteristic diffraction peaks of the anatase TiO
2 phase (JCPDS file no. 21-1272) and the Au metal phase (JCPDS file no. 01-1174). As compared to Au(TiO
2, the intensity of Au peaks in Au@Au(TiO
2 are much stronger, which can be attributed to the extra Au NPs loaded on TiO
2 shell. UV-Vis DRS analyses (Fig. S4, cf. ESM) showed that there was no absorption observed in the visible region for TiO
2 HSs. With the decoration of plasmonic Au NPs, the surface plasmon resonance (SPR) peaks at around 550 nm emerged [
30,
31]. In comparison with pure Au nanocrystal (ca
. 520 nm), the redshift in SPR peaks of Au-based TiO
2 materials can be ascribed to the effect of the high-refractive-index anatase TiO
2 shell [
32,
33]. In addition, the variation in SPR intensities found in Au-based TiO
2 composites (Au@Au(TiO
2, Au(TiO
2 and Au@TiO
2) suggests that the optical properties of TiO
2 HSs can be tuned by controlling the deposition location of Au NPs. Yolk-shell Au@Au(TiO
2 showed the strongest SPR intensity among these samples, which is related to a substantial increase in its scattering coefficient promoted by the Au NPs located on both sides of TiO
2 shell. The photocatalytic activity of materials was evaluated under visible light irradiation (420≤
l≤780 nm) using MB and phenol as probe molecules shown in Figs. 3 and 4.
Comparative photocatalytic studies were carried out using materials developed alongside naked Au NPs and Au NPs supported on P25 catalyst (Au@P25) and results were shown in Figs. 3(a) and 4(a). The chemical catalysis effect of Au NPs was trivial because naked Au NPs were found inactive in both cases. Commercial P25 and TiO
2 HSs showed the poor catalytic activity with both MB and phenol (i.e
.,<7% conversions after 180 min illumination) suggesting that the plasmon-enhanced light absorption and plasmonic sensitisation of Au NPs indeed improved the photocatalytic activity of TiO
2 materials. Au@TiO
2 HSs showed better activities than Au@P25 in both photocatalytic model reactions. This is attributed to the high specific surface area of TiO
2 HSs (187 m
2·g
-1) providing abundant active sites for converting the probing molecules, while only 36 m
2·g
-1 for commercial P25 (Table S1). In addition, the light harvesting and scattering ability were also improved by the HSs structure as explained in the previous work [
17]. For the Au-based TiO
2 hybrid nanomaterial, the decolorisation of MB and degradation of phenol were improved for the same period of the time, following the order of activity as: Au@Au(TiO
2>Au@TiO
2 HSs>yolk-shell Au(TiO
2.
The degradation rate of both probing molecules followed first-order kinetics concerning their concentrations. The rate can, therefore, be interpreted using the pseudo-first-order reaction model (Figs. 3(b) and 4(b)). The previous study by the action spectrum has shown that the degradation of MB was induced mainly by the SPR excitation of Au NPs in Au-TiO
2 materials rather than the sensitisation of the dye molecules [
30]. By comparing the rate constants in the decolourisation of MB, Au@TiO
2 HSs showed a better activity (0.030 min
-1) than that of yolk-shell Au(TiO
2 (0.013 min
-1). In the yolk-shell Au(TiO
2, e‒h pairs are produced inside the TiO
2 shell, whereas the catalytic reactions take place mostly at the outer surface of TiO
2 HSs. Therefore, the transportation of e‒h pairs through the TiO
2 shell increases the probability of e-h recombination leading to the relatively low activity in photocatalysis in comparison with that of Au@TiO
2 HSs. To decorate the inner and outer surface of TiO
2 HSs simultaneously, the photocatalytic activity of Au@Au(TiO
2 increased significantly with a constant of 0.044 min
-1 for MB decolourisation, more than three times higher than that of yolk-shell Au(TiO
2. Based on the results above, the utilisation of both sides of TiO
2 shell was proven to be efficient to boost the photocatalytic activity of Au NP-TiO
2 HS nanostructures. This can be attributed mainly to the improved light-harvesting capacity of TiO
2 HSs caused by the multiple intensified local electromagnetic fields across the TiO
2 shell. Accordingly, the interfacial hot electron transfer from Au NPs to TiO
2 shell upon the excitation of the LSPR (plasmonic sensitisation) was also intensified effectively as the synergistic effect. Additionally, due to the confinement of Au NPs both within the TiO
2 HSs and on their convex external surfaces, the developed Au@Au(TiO
2 photocatalyst also demonstrated good stability in the model reaction. The photocatalyst remained sufficiently active even after three successive recycles in the decolourisation of MB in the aqueous phase (Fig. S5, cf. ESM)
Conclusions
In conclusion, a hybrid structure of Au nanoparticles (NPs) and anatase TiO2 HSs was synthesised, in which Au NPs were confined on both sides of the TiO2 HSs enabled by the silica templating-NaOH etching and convex surface-induced confinement. The sandwich yolk-shell structures of Au NPs and TiO2 HSs, termed Au@Au(TiO2, were determined by TEM characterisation, in which multiple Au NPs decorated on a single Au yolk-TiO2 shell nanostructure were found. According to the UV-Vis DRS analysis, Au@Au(TiO2 showed the strongest SPR intensity at about 550 nm compared to Au(TiO2 and Au@TiO2. Comparative studies of Au@Au(TiO2, Au(TiO2, Au@TiO2 and Au@P25 in the photocatalytic degradation of two model molecules (i.e., MB and phenol) under visible light irradiation were carried out. Au@Au(TiO2 exhibited the enhanced activity for both reactions with the degradation rate of MB and phenol as 99% and 78%, respectively, after 180 min due to the multiple intensified local electromagnetic fields across the TiO2 shell. The strategy of sequential decorating both sides of TiO2 shells is generic and applicable to another semiconductor hollow spherical materials (e.g., BiVO4 and SnO2), which is beneficial to develop highly efficient photocatalysts for photoelectrochemical applications such as solar cells and water splitting.