Regulating thermal diffusion of gold thin films at solid-state interfaces for site-selective decoration of gold nanoparticles on titania nanotubes as an efficient SERS sensing platform

Abstract

Decorating plasmonic Au nanoparticles on TiO₂ nanotube arrays with high specific surface area is an efficient strategy for constructing high-performance surface-enhanced Raman scattering (SERS) substrates. Direct deposition of a Au thin film on a TiO₂ support, followed by dewetting at elevated temperatures, is economical for scale-up production of Au nanoparticles. The thermal diffusion behavior of Au on a high-curvature surface, however, makes it difficult to meet an optimized distribution of Au nanoparticles that is required for forming electromagnetic SERS hotspot regions. Herein, we proposed a layer-by-layer deposition method to prepare Au-decorated TiO₂ nanotube arrays using ZnO nanowires as structural templates. Au nanoparticles could be loaded on either the interior or exterior of the TiO₂ nanotubes or buried in the walls simply by switching the sequence of magnetron sputtering of Au and atomic layer deposition of TiO₂. We revealed that regulating the thermal diffusion of the Au thin film in the confined ZnO/TiO₂ interlayer favors an optimal distribution of the dewetted Au nanoparticles at the interstices. By virtue of the size and interspace of the Au nanoparticles balancing each other, Au–TiO₂ featuring Au nanoparticles dispersed on the inner walls of TiO₂ nanotubes provided abundant localized electromagnetic field hot spots with a small amount of Au consumption. This Au–TiO₂ substrate presented high SERS activity, e.g. a detection limit of 10⁻⁹ M for the R6G molecule. This substrate could also achieve self-cleaning through photocatalytic degradation of adsorbed organic analytes, which is required for repeated SERS sensing.

---

1. Introduction

Surface-enhanced Raman scattering (SERS) spectroscopy has attracted intensive attention due to its enormous potential in the fields of medicine, biology, and analytical chemistry for molecular-level trace detection.^1–3 Initially, many studies were based on noble metals Au, Ag, and Cu, where enhanced Raman signals were achieved by exploiting electromagnetic (EM) enhancement.^4–8 Electrons on the surface of rough metallic nanostructures are excited to generate localized surface plasmon resonance (SPR), which amplifies the Raman signal by intensifying the induced dipole moments of analyte molecules. With the prospect of reducing the cost of noble metal SERS substrates, researchers have devoted their attention to various semiconductor SERS active substrates like TiO₂, ZnO and MoS₂.^9–12 The effective charge transfer (CT) between the semiconductor band gap and the adsorbed molecule as well as the coupling of the charge transfer path to the incident light contributes to the selective enhancement of the target molecule vibration mode by the chemical mechanism (CM). Recently, hybrid nano-heterostructures composed of plasmonic metals and semiconductors have been fabricated to achieve integrated multiple functions with the advantages of low cost, high stability, reproducibility, self-cleaning, and easy manufacturing.^13–17 Because of the high photocatalytic activity, good biocompatibility, and ultra-high stability of TiO₂,¹⁸ various types of TiO₂ and plasmonic metal hybrid nanostructures have been studied as recyclable, ultra-stable, and self-cleaning SERS substrates.^11,19–22

SERS sensitivity depends on the target molecules that are distributed in the hotspot region and in good contact with the substrate.²³ Constructing reasonable support nanostructures with a high specific surface, like Au-coated TiO₂ nanotubes, plays a crucial role in realizing efficient SERS performance.^19,24 Using 1-D nanostructures as a structure-directing and removable template is one common strategy to fabricate TiO₂ nanotubes. The charged surface of TiO₂ nanotubes under basic conditions or of those further modified by a polyelectrolyte is inclined to be electrostatic bound to Au nanoparticles of opposite charges, facilitating the organization of Au nanoparticles on TiO₂.²⁵ Otherwise, it is of low cost and high throughput to directly deposit Au thin film on the outer walls of TiO₂ nanotubes by physical vapor deposition using a Au target. Previous studies have shown that the sputtered Au thin film on flat substrates will agglomerate and grow into large Au particles upon thermal treatment. A thicker Au film usually leads to larger dewetted individual Au nanoparticles, however, a broader size distribution and wider particle interspacing.^26,27 It is evident that the morphology, size, density of Au nanoparticles, and the hybrid pattern with semiconductors have a great influence on the hotspot region and SERS activity of the composite substrate.²⁸ In view of the shadow effect and high curvatures of TiO₂ nanotubes in the array, the coating of a uniform Au thin film is still challenging. Besides, the Au nanoparticles formed by dewetting of the Au thin film cannot be distributed closely enough to form a large number of hot spots throughout the TiO₂ nanotube array.²⁹ Therefore, it is significant to tailor the organization modes of the dewetted Au nanoparticles on TiO₂ nanotubes for aiming at specific functions of the hollow composites.

In this work, we proposed an efficient solid-state route for site-selective decoration of Au nanoparticles on TiO₂ nanotubes in a single system setup. ZnO nanowire arrays were selected as a 1-D structural template for sequential deposition of Au and TiO₂ thin films via a layer-by-layer (LbL) mode. The solid-state interfaces constructed by the LbL mode were exploited to provide a unique platform to regulate the thermal diffusion of Au at elevated temperatures, which is closely related to ZnO nanowire template characteristics.³⁰ After the Au thin film was converted to individual nanoparticles through dewetting and interface-tuned diffusion, the ZnO nanowire template was selectively removed by a mild acidic etching, forming Au-coated TiO₂ composites with a tubular structure. We illustrated that Au nanoparticles could be easily loaded on either the interior or exterior of the TiO₂ nanotubes or buried in the walls by switching the sequence of magnetron sputtering of Au and atomic layer deposition (ALD) of TiO₂.

Furthermore, we investigated the different thermal diffusion behaviors of the Au thin film in confined and unconfined environments and their effect on the morphology and spatial distribution of the dewetted Au nanoparticles. For a very thin magnetron-sputtered Au film, the dewetted Au nanoparticles, which were loaded in the interior of the TiO₂ nanotubes (Au–TiO₂), exhibited the best SERS performance for Rhodamine 6G (R6G) probe molecules with the lowest detection limit of 10⁻⁹ M. The outstanding SERS performance of Au–TiO₂ originated from the synergetic contribution of TiO₂ and the incorporated Au nanoparticles with a suitable size and an optimized spatial density, which were achieved through dewetting of the Au thin film at the confined ZnO/TiO₂ interstices. The enriched hot spots could induce a large near-field enhancement effect, which was also confirmed by finite-difference time-domain (FDTD) simulation. In addition, the Schottky junction formed between Au and TiO₂ improved the photo-generated charge separation efficiency and granted Au–TiO₂ outstanding performance in degrading organic pollutants, enabling Au–TiO₂ with self-cleaning function to be recycled and reused as SERS substrates. This design strategy will thus allow us to tailor optimistic composite structures constructed by dewetted Au nanoparticles and TiO₂ nanotubes for targeted applications.

2. Experimental section

2.1. Materials

Zinc nitrate hexahydrate (Zn(NO₃)₂·6H₂O) was procured from Sigma Aldrich. Hexamethylenetetramine (C₆H₁₂N₄, HMTA) was procured from Shanghai Lingfeng Chemical Reagent Co., Ltd. Hydrofluoric acid (HF) was procured from Shanghai No. 4 Reagent & H.V. Chemical Co., Ltd. Glycerin was procured from Xilong Scientific Co., Ltd. Rhodamine 6G (R6G), Rhodamine B (RhB) and methylene blue (MB) were procured from Shanghai Yuanye Co., Ltd. All the chemical reagents were used as received. Deionized water (DI water) was consumed to produce the solutions of the experiments.

2.2. Preparation of TiO₂ nanotube arrays decorated with Au nanoparticles

ZnO nanowire arrays used as templates were grown on FTO/glass substrates (3 cm × 1 cm) through a chemical solution reaction following a standard procedure.³¹ The obtained ZnO samples were annealed in air at 500 °C for 1 h to remove organic residuals on the nanowire surface.

The samples with Au nanoparticles loaded on the interior and exterior of the TiO₂ nanotubes, and embedded in the TiO₂ nanotube wall were denoted as Au–TiO₂, TiO₂–Au and TiO₂–Au–TiO₂, respectively. For Au–TiO₂, a 3 nm thick Au thin film was sputtered onto the as-prepared ZnO nanowire array template using an ultra-high vacuum magnetron system (AGAR Sputter Coater) at ambient temperature. The operating current was 15 mA, and it was maintained for 90 seconds. Subsequently, a TiO₂ shell layer with a thickness of 15 nm was deposited by ALD (MNT-100) using TiCl₄ and water as the precursors, which confined the Au thin film to the interface between the ZnO nanowires and TiO₂ conformal layer. With regard to TiO₂–Au, the sputtering of the Au thin film was conducted using the same parameters after the deposition of 15 nm thick TiO₂. To synthesize TiO₂–Au–TiO₂, a 5 nm layer of TiO₂ was first coated onto the ZnO template by ALD, followed by sputtering of the Au thin film. Finally a 10 nm layer of TiO₂ was coated on the sample surface by another ALD operation.

The above products were annealed at 500 °C for 1 h to trigger the dewetting process of the Au thin film. Afterward, a mixed solution of HF, glycerin, and water was employed to etch the TiO₂ tips for further removing the internal ZnO nanowire template. After washing with DI water and ethanol, three types of TiO₂ nanotube arrays with site-selective decoration of Au nanoparticles were obtained.

2.3. Characterization

The morphologies of the samples were investigated by field emission scanning electron microscopy (SEM, Hitachi S-4800). A transmission electron microscope (TEM, JEM-2100 plus) operated at 200 kV was used to determine the detailed nanotube morphology and the size of the Au nanoparticles in each sample. The crystal phase of the samples was determined by X-ray diffraction (XRD, Rigaku D/MAX-2400). The absorption spectra were obtained with a UV-vis spectrophotometer (Shimadzu UV-3600). The SERS measurements were carried out at room temperature using a confocal microscopy Raman spectrometer with a 50× objective field lens (LabRAM HR Evolution, HORIBA Jobin Yvon, France). The laser beam power was about 1 mW and the laser beam diameter was around 1 μm. An excitation wavelength of 532 nm, a scan time of 10 s, and accumulation of one time were applied. The SERS substrates were immersed into the solution of target molecules for 30 min for reaching the adsorption equilibrium, then taken out, rinsed with DI water and dried under ambient conditions.

3. Results and discussion

The fabrication schematic of three types of TiO₂ nanotubes with Au nanoparticles decorated in different positions is illustrated in Fig. 1a. In brief, a ZnO nanowire array prepared by a solution method was used as the structural template of the nanotube substrate. Then, the Au thin film was deposited as the inside layer, interlayer, and outside layer, respectively, relative to the TiO₂ shell manipulatively through varying the sequence of the TiO₂ ALD and Au sputtering processes. When the Au thin film was thermally annealed at 500 °C below the Au melting point, it was prone to split open and retract into distinct Au nanoparticles by solid-state dewetting.³² The primary Au nanoparticles continued to grow and ripen by adatom attachment via surface diffusion at elevated temperature. After the samples gradually cooled down to room temperature, we removed the top edge of TiO₂ and the inner ZnO nanowires by chemical etching and obtained the Au-decorated TiO₂ nanotube arrays.

Fig. 1 (a) Fabrication procedure for site-selective decoration of Au nanoparticles on TiO₂ nanotubes by dewetting Au thin film. SEM images of Au–TiO₂ (b), TiO₂–Au–TiO₂ (d), and TiO₂–Au (f). TEM images of Au–TiO₂ (c), TiO₂–Au–TiO₂ (e), and TiO₂–Au (g); inset in each panel is the corresponding size distribution histogram of the loaded Au nanoparticles.

The SEM images of the three samples intuitively depict the hollow structure of TiO₂ nanotubes in the arrays. The Au nanoparticles are uniformly distributed inside the nanotubes for Au–TiO₂ (Fig. 1b), embedded in the nanotube wall for TiO₂–Au–TiO₂ (Fig. 1d), and outside the nanotubes for TiO₂–Au (Fig. 1f). Thus the site-selective localization of Au nanoparticles relative to the TiO₂ shell is also clear. Although the thickness of the initial Au thin film was identical, there is a significant difference in the average size of Au nanoparticles for the three samples by analyzing the particle size based on TEM observations. The average sizes of the Au nanoparticles for Au–TiO₂, TiO₂–Au–TiO₂, and TiO₂–Au are 15.8 nm (Fig. 1c), 8.3 nm (Fig. 1e), and 19.8 nm (Fig. 1g), respectively.

We took Au–TiO₂ as an example to further investigate the structural features of the Au-decorated TiO₂ composite nanotubes. Fig. S1a (ESI†) shows a SEM image of the ZnO–Au–TiO₂ core–shell nanowire array annealed at 500 °C for 1 h in air. The inhomogeneous contrast in the TEM image (Fig. S1b, ESI†) indicates the presence of a discontinuous Au particulate interlayer. The ALD-deposited TiO₂ external layer is uniform with a thickness of 15 nm, agreeing with the value predetermined by the ALD cycles. Notably, the TiO₂ shell is conformal to the substrate, which completely encapsulates the underlying ZnO–Au nanowire. Fig. 2a is a typical TEM image of the Au–TiO₂ nanotubes that were converted from the annealed ZnO–Au–TiO₂ nanowires by acidic etching. The nanotube wall slightly cracked due to the phase transition of TiO₂ at 500 °C and the pressing of the Au nanoparticles subjected to an anisotropic pressure upon thermal expansion. In addition, Au nanoparticles of various sizes are densely distributed in the interior of the nanotube. In the high-resolution TEM image exhibited in Fig. 2b, the lattice spacing of 0.23 and 0.35 nm can be, respectively, attributed to the (111) plane of the Au nanoparticles and the (101) plane of the TiO₂ nanotubes of the anatase phase. We can also observe the shrinkage of Au-enclosing TiO₂ sections due to grain boundary diffusion during the sintering process. Fig. 2c shows the individual element mappings of the Au–TiO₂ nanotubes. The elemental densities of Ti and O follow a decreasing trend along the edge to the interior, confirming its nanotube-like structure, while the dispersed nanoparticles only consist of Au. EDX elemental analysis in Fig. S2 (ESI†) reveals that the atomic ratio of Ti to Au is 2.83. The above microstructure characterizations verify the successful preparation of the Au–TiO₂ nanotubes.

Fig. 2 (a) TEM image of a selected Au–TiO₂ nanotube. (b) HRTEM image of an area including the Au–TiO₂ interface. (c) EDX elemental mappings of Ti, O, Au, and Ti + Au for the composite nanotube.

Fig. 3a shows the XRD patterns of the annealed ZnO–Au–TiO₂ sample and the different products of Au–TiO₂, TiO₂–Au–TiO₂ and Au–TiO₂. All the samples have a diffraction peak located at 25.5°, which corresponds to the (101) lattice plane of the anatase phase TiO₂.³⁰ This result reveals that the amorphous phase of the ALD-deposited TiO₂ shells was transformed into anatase upon annealing at 500 °C. The distinct peaks of ZnO–Au–TiO₂ at 31.8°, 34.4°, 35.2°, 47.5° and 62.8° can be assigned to (100), (002), (101), (102), and (103) lattice planes of wurtzite ZnO, which are absent from Au–TiO₂, TiO₂–Au–TiO₂ and Au–TiO₂, indicating the successful removal of the ZnO templates by acidic etching. However, the diffraction peaks of Au are not found in all the XRD patterns, which is probably caused by the small size, high dispersion and low quantity of the loaded Au nanoparticles.³³

Fig. 3 (a) XRD patterns of ZnO–Au–TiO₂ annealed at 500 °C, Au–TiO₂, TiO₂–Au–TiO₂ and TiO₂–Au. (b) Raman spectra of single TiO₂ and Au–TiO₂ nanotubes. (c) UV-vis diffuse reflectance spectra of Au–TiO₂, TiO₂–Au–TiO₂ and TiO₂–Au.

Fig. 3b exhibits the Raman spectra of the single TiO₂ and Au–TiO₂ nanotubes. The identical vibration modes of E_g (142 cm⁻¹), E_g (197 cm⁻¹), B_1g (397 cm⁻¹), A_1g + B_1g (517 cm⁻¹), and E_g (636 cm⁻¹) suggest that the TiO₂ in the Au–TiO₂ nanotubes is of tetragonal anatase structure. Its conventional cell consists of two primitive cells, each containing two TiO₂ units, which is consistent with the results of the XRD patterns.³⁴ The Raman scattering intensity of Au–TiO₂ is significantly enhanced and slightly shifted compared to the single TiO₂, which may be attributed to the plasmon resonance-assisted amplification of Raman scattering caused by the Au nanoparticles.³⁵

The optical properties of the different Au-decorated TiO₂ composite nanotubes were investigated by absorption spectroscopy. Fig. 3c shows the UV-vis diffuse reflectance spectra of Au–TiO₂, TiO₂–Au–TiO₂, and TiO₂–Au. As expected, the three samples show the characteristic absorption of TiO₂ in the UV region, where the signal of Au–TiO₂ is the strongest. In addition, a blue-shift of the absorption edge is observed for Au–TiO₂ from 380 to 360 nm, different from those of TiO₂–Au–TiO₂ and TiO₂–Au. The above phenomena demonstrate that the light absorption of the TiO₂ nanotubes was affected by the size and site of the loaded Au nanoparticles, e.g. Au–TiO₂ with Au nanoparticles decorated in the interior of TiO₂ nanotubes, could reduce the shielding effect and ensure adequate UV light absorption by TiO₂. Moreover, the enhanced light absorption in a wide visible region is observed for the three samples on account of the SPR absorption generated by the dewetted Au nanoparticles with various sizes.

For investigating the site-selective effect of the Au nanoparticles on the SERS performance of the Au-decorated TiO₂ nanotubes, we employed R6G and MB as prototype probe molecules to explore the SERS behavior of Au–TiO₂, TiO₂–Au–TiO₂ and TiO₂–Au under 532 nm excitation. As shown in Fig. 4a and c, the three composite substrates provided enhanced Raman signals for both R6G and MB with a low concentration of 10⁻⁶ M, while the single TiO₂ nanotube substrate was unresponsive (data not shown). The obtained Raman peaks agree well with the reported literature.^36,37 Specifically, the characteristic peaks of R6G at 612, 774, 1186 and 1314 cm⁻¹ can be individually assigned to C–C–C in-plane ring vibration, C–C–C out-of-plane bending vibration, C–H in-plane bending and C–O–C stretching motion. The other peaks observed in the positions of 1364, 1506 and 1652 cm⁻¹ correspond to the stretching of the aromatic ring.³⁶ In the case of MB, the weak intensity peaks around 669, 767 and 1071 cm⁻¹ are derived from the out-of-plane and in-plane bending of C–H vibration. The peaks around 1394 cm⁻¹ are caused by C–N symmetrical stretching vibration. The C–C ring stretching mode is attributed to the most intense peak located at around 1623 cm⁻¹.³⁷ Evidently the enhanced SERS signals depend on the decoration of Au nanoparticles on the TiO₂ nanotube arrays. However, we also notice that the performance drastically varies with the loading sites of the dewetted Au nanoparticles, although the sputtered amount of Au is ideally the same for Au–TiO₂, TiO₂–Au–TiO₂ and TiO₂–Au. Fig. 4b and d summarize the spectral intensities of the three substrates by monitoring the characteristic peaks of R6G and MB, where Au–TiO₂ presents a superior SERS activity to Au–TiO₂–Au and TiO₂–Au.

Fig. 4 SERS spectra of R6G (a) and MB (c) with a concentration of 10⁻⁶ M on Au–TiO₂, TiO₂–Au–TiO₂ and TiO₂–Au under 532 nm excitation. SERS intensities of R6G at peaks of 612, 774, and 1506 cm⁻¹ (b), and MB at peaks of 767, 1394, and 1623 cm⁻¹ (d) for the three samples. (e)–(g) FDTD simulations of the optimized models of Au–TiO₂, TiO₂–Au, and TiO₂–Au–TiO₂ composite nanotubes. The EM-field enhancement was obtained by excitation at 532 nm.

As verified by the microstructure analysis shown in Fig. 1, the site-selective decoration of the Au nanoparticles on the TiO₂ nanotubes by dewetting the Au thin film of the same thickness resulted in a difference in the average size, which accordingly adjusted the interspaces between the Au nanoparticles distributed on the equal-area supports. It is known that Au atoms can diffuse to long distances on a heterogeneous surface at high temperatures.^30,38 During the annealing process, the heat input first induced the disintegration of the Au thin film into isolated particles. Subsequently, the large particles grew and ripened by diffusion of Au atoms from the neighboring smaller particles according to the Gibbs–Thomson equations. For Au–TiO₂, the spatial confinement between the internal ZnO nanowire and the TiO₂ shell could retard the migration of Au atoms for formation of larger Au nanoparticles in the thermal diffusion process, whereas the Au nanoparticles formed in the TiO₂ interlayer exhibited a much smaller size, suggesting the stronger binding force of the TiO₂ bilayer constructed by ALD with the sandwiched Au thin film. With respect to TiO₂–Au, the Au thin film was directly deposited on the external side of the ZnO/TiO₂ core–shell nanowires, the dewetting process of which was merely influenced by the TiO₂ support. As a result, the formed Au nanoparticles outside the nanotubes featured the largest size.

The 3D FDTD simulation was carried out to numerically estimate the EM field distribution in the three Au-decorated TiO₂ nanotubes. The simulation models of Au nanoparticles at different locations of TiO₂ nanotubes (Fig. S3, ESI†) were constructed by ideal geometric parameters based on the diverse particle size evolution from the identical Au thin film. Fig. 4e–g exhibit the EM field images of Au–TiO₂, TiO₂–Au and Au–TiO₂–Au at an excitation wavelength of 532 nm. The dipole resonance intensity of the Au nanoparticles is known to be simultaneously correlated with their size and distribution, inducing a strong electric field, which favors the SERS signal of the adsorbed probe molecules. From the FDTD simulations, it can be observed that numerous EM hot spots are formed at the nanogaps of the Au nanoparticles in Au–TiO₂via multiple plasmonic coupling (Fig. 4e). In contrast, fewer plasmonic coupling regions emerge between the Au nanoparticles in TiO₂–Au, although the electric field intensity at the Au/TiO₂ interface is greatly enhanced due to the larger size of the Au nanoparticles (Fig. 4f). In the case of TiO₂–Au–TiO₂ where Au nanoparticles are surrounded by TiO₂, the change in the dielectric electric field environment causes Au–TiO₂–Au to present the weakest EM field intensity (Fig. 4g).

Additionally, we simulated the EM field distributions in the three Au-decorated TiO₂ nanotubes by taking the size distribution of the Au nanoparticles into account. The number of dewetted Au nanoparticles still stayed with the identical thickness of the initial Au thin film. Fig. S4 (ESI†) exhibits the EM field images of Au–TiO₂, TiO₂–Au and Au–TiO₂–Au with a 20% size distribution of the Au nanoparticles. From the FDTD simulations, it is observed that more EM hot spots can form at the nanogaps of the Au nanoparticles in Au–TiO₂ due to multiple plasmonic coupling, consistent with the results obtained from ideal geometric parameters based on the diverse particle size evolution. In order to evidence the effect of configuration on the formation of the hotspot region, we further conducted simulations under conditions of identical Au nanoparticle size and interspace to investigate the EM field associated with these three structural configurations. The results in Fig. S5 (ESI†) show that Au–TiO₂ still exhibits the most significant EM field enhancement and the broadest hotspot region compared to TiO₂–Au and TiO₂–Au–TiO₂ configurations. This observation suggests that the Au–TiO₂ configuration offers distinct advantages for SERS performance. The multiple scattering of incident light within the nanotube cavity might contribute to amplifying the electric field intensity and expanding the coverage within the hotspot region. It can be concluded that the superior SERS performance of Au–TiO₂ does not solely stem from the optimized spatial density of the Au nanoparticles but also from the unique Au–TiO₂ configuration. Obviously, the probe molecules adsorbed on Au–TiO₂ have a higher probability to be distributed in the electric field coupling region, thus achieving an effective enhancement of their Raman signals. Furthermore, Au–TiO₂ could benefit from the capillary effect of TiO₂ nanotubes, the cavities of which are capable of serving as molecular sinks to effectively enrich the probe molecules in the hotspot region distributed in the nanotube interior.

We further used Au–TiO₂ as a substrate for Raman detection of the probe molecule R6G of different molecular concentrations. Fig. 5a shows the SERS spectra of R6G with a concentration ranging from 10⁻⁵ down to 10⁻⁹ M. The Raman peaks of R6G slowly decrease in intensity with decreasing concentration of R6G, and the signals are clearly observable at a concentration as low as 10⁻⁹ M, indicating a high sensitivity. The calculated analytical enhancement factor (AEF) of Au–TiO₂ for R6G at 612 cm⁻¹ is 6.64 × 10⁶ (ESI†). For investigating the reproducibility and stability of the SERS sensing that are important for the practicability of the substrate, Raman signals of R6G with a concentration of 10⁻⁵ M were collected by randomly taking 25 points on the Au–TiO₂ substrate. As shown in Fig. 5b, the fluctuation of the peak intensity at 612 cm⁻¹ is slight with a relative standard deviation (RSD) value of ∼0.12 (Fig. 5c), implying a uniform distribution of SERS hot spots on the entire substrate. In addition, after the as-prepared Au–TiO₂ was stored in a wet environment for 10 days, the SERS spectrum of R6G obtained from the substrate is almost unchanged (Fig. S6, ESI†). The above results clearly demonstrate that the Au–TiO₂ substrate possesses high stability, good uniformity and a low detection limit in SERS sensing, with potential in trace detection for molecules of interest.

Fig. 5 (a) SERS spectra of different concentrations of R6G molecule deposited on the Au–TiO₂ substrate. (b) SERS spectra of R6G with a concentration of 10⁻⁵ M collected from 20 random sites on the Au–TiO₂ substrate. (c) R6G signal intensities at 612 cm⁻¹, presenting an RSD of 0.12.

Considering the photocatalytic property of TiO₂, we studied the self-cleaning ability of Au–TiO₂, which is indispensable for a recyclable SERS substrate. The experiment of photocatalytic degradation of MB dye molecules in bulk solution was first applied under Xe lamp irradiation (100 mW cm⁻²) using Au–TiO₂ (3 cm × 1 cm) as the photocatalyst. Fig. 6a shows the absorption spectra of MB (20 μM, 20 ml) during this process. As illustrated in Fig. 5b, the concentration of MB was reduced to 8% of the initial value within 120 min. The intimate contact between Au and TiO₂ in Au–TiO₂ is supposed to form a Schottky barrier, which could suppress the recombination of electrons and holes photogenerated in TiO₂ and improve its photocatalytic activity for dye degradation.³⁹ The cycling stability of Au–TiO₂ as the photocatalyst was further investigated. The results summarized in Fig. 6b show that the degradation performance towards MB did not deteriorate after three degradation cycles operating under the same experimental conditions.

Fig. 6 (a) UV-vis spectra of MB solution (20 μM, 20 ml) under Xe lamp irradiation for different times in the presence of Au–TiO₂. (b) Cycle test results of photocatalytic degradation of MB. Reversible SERS behaviors of R6G (c) and (d) and RhB (e) and (f) by using the Au–TiO₂ substrate during three cycles.

On the basis that Au–TiO₂ can effectively and continuously clean itself by photocatalytic degradation of the target molecules adsorbing on the surface, we examined its recyclable application as a self-cleaning SERS substrate. Fig. 6c shows the Raman spectra of R6G (10⁻⁵ M) measured on the Au–TiO₂ SERS substrate via a procedure of detecting/cleaning in three consecutive cycles. Characteristic Raman peaks of R6G could be clearly identified by the initial SERS detection. After the used substrate was immersed in DI water and illuminated with a Xe lamp for 30 min, the Raman signals from R6G vanished, indicating a complete removal of the analyte by the cleaning treatment. As displayed in Fig. 6d, this reversible SERS behavior can sustain at least three cycles by tracking the intensity of the Raman peak at 612 cm⁻¹. When RhB was used as the target probe molecule, we came to the same conclusion (Fig. 6e and f). Therefore, the Au–TiO₂ substrate with efficient self-cleaning performance is expected to showcase its reversibility for a wide variety of organic analytes.

4. Conclusions

Au-decorated TiO₂ nanotube arrays with different spatial distributions of Au nanoparticles have been successfully prepared via a top-down LbL deposition method using ZnO nanowires as structural templates. We demonstrate that regulating the thermal diffusion of the Au thin film in the confined ZnO/TiO₂ interlayer favors an optimal distribution of the dewetted Au nanoparticles at the interstices with a small amount of Au consumption. By virtue of the size and interspace of the Au nanoparticles balancing each other, Au–TiO₂ featuring Au nanoparticles dispersed on the inner walls of TiO₂ nanotubes exhibited abundant localized EM field hot spots relative to those of TiO₂–Au–TiO₂ and TiO₂–Au, which was supported by FDTD simulations. Accordingly, Au–TiO₂ showed the best SERS activity, e.g. a detection limit of 10⁻⁹ M with an RSD of 0.12 for the R6G molecule. Meanwhile, the Au–TiO₂ SERS substrate could achieve self-cleaning through photocatalytic degradation of the adsorbed organic analyte, which is required for repeated usage.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work is financially supported by the National Natural Science Foundation of China (21878157 and 22078156), and the Project of Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

References

1. N. Guarrotxena and G. C. Bazan, Adv. Mater., 2014, 26, 1941–1946.
2. X. Wang, S. C. Huang, T. X. Huang, H. S. Su, J. H. Zhong, Z. C. Zeng, M. H. Li and B. Ren, Chem. Soc. Rev., 2017, 46, 4020–4041.
3. J. F. Li, Y. F. Huang, Y. Ding, Z. L. Yang, S. B. Li, X. S. Zhou, F. R. Fan, W. Zhang, Z. Y. Zhou, D. Y. Wu, B. Ren, Z. L. Wang and Z. Q. Tian, Nature, 2010, 464, 392–395.
4. Z. Q. Tian, J. S. Gao, X. Q. Li, B. Ren, Q. J. Huang, W. B. Cai, F. M. Liu and B. W. Mao, J. Raman Spectrosc., 1998, 29, 703–711.
5. M. Dendisová-Vyškovská, V. Prokopec, M. Člupek and P. Matějka, J. Raman Spectrosc., 2012, 43, 181–186.
6. K. Kim and H. S. Lee, J. Phys. Chem. B, 2005, 109, 18929–18934.
7. D. Li, J. Q. Liu, H. B. Wang, C. J. Barrow and W. R. Yang, Chem. Commun., 2016, 52, 10968–10971.
8. J. P. Lee, D. C. Chen, X. X. Li, S. Yoo, L. A. Bottomley, M. A. El-Sayed, S. Park and M. L. Liu, Nanoscale, 2013, 5, 11620–11624.
9. S. Xie, D. Chen, C. Gu, T. Jiang, S. Zeng, Y. Y. Wang, Z. Ni, X. Shen and J. Zhou, ACS Appl. Mater. Interfaces, 2021, 13, 33345–33353.
10. S. Cong, Y. Yuan, Z. Chen, J. Hou, M. Yang, Y. Su, Y. Zhang, L. Li, Q. Li, F. Geng and Z. Zhao, Nat. Commun., 2015, 6, 7800.
11. K. Huang, C. Li, X. Zhang, X. Meng, L. Wang, W. Wang and Z. Li, Appl. Surf. Sci., 2020, 518, 146219.
12. G. Song, W. Gong, S. Cong and Z. Zhao, Angew. Chem., Int. Ed., 2021, 60, 5505–5511.
13. J. Lin, Y. Shang, X. Li, J. Yu, X. Wang and L. Guo, Adv. Mater., 2017, 29, 1604797.
14. X. Tang, X. Fan, L. Yao, G. Li, M. Li, X. Zhao, Q. Hao and T. Qiu, J. Phys. Chem. Lett., 2022, 13, 7816–7823.
15. J. Huang, T. X. Zhou, W. S. Zhao, S. C. Cui, R. Guo, D. Li, N. R. Kadasala, D. L. Han, Y. H. Jiang, Y. Liu and H. L. Liu, J. Colloid Interface Sci., 2022, 628, 315–326.
16. J. Dong, Y. Wang, Q. Wang, Y. Cao, Q. Han, W. Gao, Y. Wang, J. Qi and M. Sun, Nanotechnology, 2022, 33, 135501.
17. M. Sakir, E. Yilmaz and M. S. Onses, Microchem. J., 2020, 154, 104628.
18. P. Lv, C. Y. Zhao, W. J. Lee, S. P. Huo, S. H. Kwon, J. Fang and Y. Yang, Appl. Surf. Sci., 2019, 494, 508–518.
19. X. Li, G. Chen, L. Yang, Z. Jin and J. Liu, Adv. Funct. Mater., 2010, 20, 2815–2824.
20. Z. Dai, G. Wang, X. Xiao, W. Wu, W. Li, J. Ying, J. Zheng, F. Mei, J. Wang and C. Jiang, J. Phys. Chem. C, 2014, 118, 22711–22718.
21. J. Prakash, P. Kumar, R. A. Harris, C. Swart, J. H. Neethling, A. J. van Vuuren and H. C. Swart, Nanotechnology, 2016, 27, 355707.
22. C. Xu, H. Geng, R. Bennett, D. A. Clayton and S. Pan, J. Phys. Chem. C, 2013, 117, 1849–1856.
23. J. Langer, D. J. de Aberasturi and J. Aizpurua, et al. , ACS Nano, 2022, 14, 28–117.
24. S. Wang, J. Yao, Z. Ou, X. Wang, Y. Long, J. Zhang, Z. Fang, T. Wang, T. Ding and H. Xu, Nanoscale, 2022, 14, 4705–4711.
25. J. Yu, G. A. Rancea and A. N. Khlobystov, J. Mater. Chem., 2009, 19, 8928–8935.
26. C. V. Thompson, Annu. Rev. Mater. Res., 2012, 42, 339–434.
27. K. Sugawara, M. Kawamura, Y. Abe and K. Sasaki, Microelectron. Eng., 2007, 84, 2476–2480.
28. Y. F. Wang, S. Ma, H. Yu, Y. Liu, J. Gao, L. Yang, M. Zhang, G. He and Z. Q. Sun, Nanotechnology, 2020, 32, 75708.
29. Y. Cai, F. Y. Yuan, F. Li, H. H. Kang, D. X. Xue, S. P. Huo, F. J. Yu, J. Fang and Y. Yang, CrystEngComm, 2021, 23, 6551–6558.
30. Y. Yang, R. B. Yang, H. J. Fan, R. Scholz, Z. Huang, A. Berger, Y. Qin, M. Knez and U. Gösele, Angew. Chem., Int. Ed., 2010, 49, 1442–1446.
31. Z. Y. Hu, H. J. Huang, F. Y. Yuan, Y. Cai, H. B. Shu, F. J. Yu, J. Fang and Y. Yang, J. Colloid Interface Sci., 2023, 644, 466–477.
32. S. Karim, M. E. Toimil-Molares, W. Ensinger, A. G. Balogh, T. W. Cornelius, E. U. Khan and R. Neumann, J. Phys. D: Appl. Phys., 2007, 40, 3767–3770.
33. W. F. Yan, V. Petkov, S. M. Mahurin, S. H. Overbury and S. Dai, Catal. Commun., 2005, 6, 404–408.
34. C. Engelbrekt, K. T. Crampton, D. A. Fishman, M. Law and V. A. Apkarian, ACS Nano, 2020, 14, 5061–5074.
35. O. L. Stroyuk, V. M. Dzhagan, A. V. Kozytskiy, A. Y. Breslavskiy, S. Y. Kuchmiy, A. Villabona and D. Zahn, Mater. Sci. Semicond. Process., 2015, 37, 3–8.
36. X. Z. Wang, Z. Wang, M. Zhang, X. S. Jiang, Y. F. Wang, J. G. Lv, G. He and Z. Q. Sun, J. Alloys Compd., 2017, 725, 1166–1174.
37. H. Chen, Z. Yang and S. Lee, Langmuir, 2016, 32, 10184–10188.
38. E. Florez, F. Vines, J. A. Rodriguez and F. Illas, J. Chem. Phys., 2009, 130, 244706.
39. Q. Deng, X. W. Duan, D. H. L. Ng, H. B. Tang, Y. Yang, M. G. Kong, Z. K. Wu, W. P. Cai and G. Z. Wang, ACS Appl. Mater. Interfaces, 2012, 4, 6030–6037.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3tc03702a

---