Abstract

A 3D surface-enhanced Raman scattering (SERS) substrate based on titanium oxide nanorods (TiOx-NRs) coated with gold nanoparticles (Au-NPs) was fabricated by a simple hydrothermal, no-template process. The nanostructure of TiOx-NRs influenced by the concentrations of hydrochloric (HCl) acid and sodium chloride (NaCl) was studied in detail. The substrate showed the strongest Raman enhancement, when the diameters of Au-NPs were around 40 nm and the gaps of Au-NPs were in the range of 5 nm to 10 nm. The surface electric field of our substrate was examined by finite-different time-domain (FDTD) solutions. Rhodamine 6G (R6G) was chosen as the probe molecule to study the SERS performance of the substrates. The Raman signal of 10−10 M R6G was detected clearly by the substrate with the enhancement factor of 2.64 × 108. All relative standard deviation (RSD) values of the major peaks for R6G were within the scope of 10.4% to 16.7%. The substrate could work efficiently even after immersed in water for one month.

1. Introduction

SERS substrates have drawn tons of attention in the fields such as analytical chemistry [1], environment protection [2], medical remedy [3], biology detection [4], and so on, for its high sensitivity, quick response, and universal properties to most organic molecules. Research on the SERS substrates combined with 3D semiconductor nanostructure and noble metal nanoparticles [5, 6] is currently a hot research area due to their high local field enhancement (LFE) [7]. A series of semiconductor nanostructures coated by noble metal has been made, like Au-TiO2 [8], Ag-TiO2 [9], Ag-GaN [10], Au-ZnO [11], Ag-ZnO [12], Ag-Si [13], and so on.

Surface plasmon resonance [14, 15] and charge transfer mechanism [16, 17] are considered as the major contributions in the noble metal-semiconductor SERS system. In the theory of surface plasmon resonance, when exciting laser interacts with the electrons in the metal conduction band, there would be a strong electric field in the metal surface. The high electric field would interact with the absorbed molecule. In this condition, Raman signal is sharply enhanced at the junction between the nanoparticles [18], atomic clusters [19], or sharp edges [20], which are called hotspots. Therefore, a mass of uniformly distributed hotspots is important for high-sensitive SERS substrates. Meanwhile, 3D nanostructures (e.g., nanowire, nanoshell, and nanorods) could provide sufficient hotspots forming locations because of its large specific surface area [21, 22].

Based on the charge transfer mechanism, when the molecules bond to semiconductor appropriately [17], charge would transfer from semiconductor to absorbed molecules under exciting laser, which could enhance the Raman signal obviously. Semiconductor with lower bandgap would realize this mechanism easier. In consideration of the dramatic SERS enhancement from semiconductor nanostructure, semiconductors TiO2, ZnO, and Si are always chosen as the substrate material [2325].

Compared with conventional Si or ZnO, TiO2 is widely used in photocatalytic degradation and polluted water purification [26] due to its perfect chemical, thermal stability, better biocompatibility, and low cost [27]. Hydrothermal synthesis [28, 29] rather than anodic oxidation [30] and template-based synthesis [31] are used to fabricate TiOx for its simple process, high yield, and low cost. The lattice structure of rutile TiO2 is the same as that of fluorine-doped SnO2 (FTO), and the lattice mismatch is just about 2% [32], so FTO glass was selected as the substrate to improve the nucleation and growth of hydrothermal-synthesized TiO2 nanorods.

Au and Ag are widely used in SERS substrates due to their perfect surface plasmon resonance mechanism [3335]. Au is more suitable to make recyclable SERS substrates for its chemical stability.

In this work, a kind of highly ordered TiOx-NR (x < 2) SERS substrate coated with Au-NPs was demonstrated experimentally. A SERS substrate stacked by disordered TiO2 nanowires supporting gold nanocrystals had been proposed in our previous research work firstly [36]. TiOx (x < 2) SERS substrate would realize charge transfer mechanism easier than TiO2. Meanwhile, compared with TiO2 powder SERS substrate [9], ordered TiOx-NR based on FTO glass SERS substrate is more suitable in recycling application. The pH and chloridion concentration ([Cl]) in our reaction were optimized to obtain better morphology of TiOx-NRs. We also changed sodium citrate (Na3Ct)/chloroauric acid (HAuCl4) concentration ratio to adjust size and density of Au-NPs. R6G was chosen as the probe molecule for its strong fluorescence effect and large Raman scattering cross section.

2. Materials and Experimental

2.1. Materials

(3-Aminopropyl)triethoxysilane (APTES, 98%) was purchased from Aladdin. Titanium isopropoxide (TTIP, Ti 16.6–16.9%), chloroauric acid (HAuCl4, Au ≥ 47.8%), rhodamine 6G (95%), concentrated sulfuric acid (AR, 95.0%–98.0% by weight), hydrogen peroxide (AR, 30%), acetone (AR), ethanol (AR), hydrochloric acid (AR, 36.5%–38% by weight), sodium chloride (AR), and sodium citrate (Na3Ct, AR) were purchased from Sinopharm. Fluorine-doped tin oxide (FTO) glass (FTO-P001, <10 Ω/sq) was purchased from Kaivo Optoelectronic Technology Co.

2.2. Instrument

Au-NPs were deposited by the ion sputtering equipment (HTCY Technology Co. JS-1600) or by the citrate reduction method with the help of the thermostat magnetism stirrer (Jintan Jiangsu Xinbao 90-2). Morphologies of the samples were characterized with the field emission scanning electron microscope (FESEM, Zeiss Ultra Plus). Annealing was processed by high-temperature electric oven (Shanghai Xunbo SX2). The phase of titanium oxide was analyzed by X-ray diffraction (XRD, Bruker D8 ADVANCE). Raman spectra of R6G were measured with the dispersive Raman microscope (Horiba HR Evolution). Ultraviolet light condition was realized by an ultraviolet analyzer (Shanghai Jihui ZF-7).

2.3. Hydrothermal Synthesis of TiOx-NRs

Before experiment, a piece of FTO glass was cleaned in the mixture of concentrated piranha solution (H2SO4 : H2O2=3 : 1) at 80°C for 10 min. Piranha solution should be operated in ventilated hood, and the operator should wear protection suit, rubber gloves, and goggles. Then, the FTO glass was washed by deionized (DI) water and dried with nitrogen.

An experiment was designed to explore the effect of pH value and concentration of [Cl] on the orientation growth of TiOx-NRs. Firstly, 0.7 M NaCl and 0.042 M TTIP (C12H28O4Ti) were added into the diluted HCl solution with different concentrations, for example, 7.0 M, 6.0 M, 5.0 M and 4.0 M. Then, the concentration of [Cl] can be further tuned by adding NaCl salt, for example, 0 M, 0.55 M, 0.7 M, and 0.85 M. Thereafter, the solution was transferred into a sealed Teflon hydrothermal reaction vessel. The FTO glass was placed at an angle against the inner wall of vessel with the conductive side facing down. The vessel was sealed and put in oven at 150°C for 18 h. After that, the vessel was cooled to room temperature naturally. The sample was taken out and rinsed with deionized water for three times to wash out the residual reactant. Finally, the crystallization of TiOx-NRs film was carried out by annealing at 500°C for 30 min.

2.4. Decoration of Gold Nanoparticles

Two different deposition methods for Au-NPs attaching on TiOx-NRs were used in this work. One was ion sputtering with ion current 20 mA for 30 s. Another was the citrate reduction method. In citrate reduction, the density and size of Au-NPs were varied by changing the initial HAuCl4 concentrations. TiOx-NRs substrate was amino-modified in the fresh-made amination solution (40 μL of APTES, 0.5 mL of DI water, and 0.5 mL of ethanol) for 20 min firstly. Thereafter, the samples were dried at 60°C for 10 min and then cleaned in ultrasonic cleaning with deionized water for 1 min. After that, the sample was put in Au-NPs citrate reduction solution, which consisted of 1.5 × 10−2 M Na3Ct and special concentration HAuCl4 (e.g., 1.0 × 10−4 M, 2.0 × 10−4 M, 3.0 × 10−4 M, and 4.0 × 10−4 M). The reactant should be kept on boiling and stirring condition for 10 min. When finishing, the solution was cooled to room temperature naturally. At last, the sample was taken out and rinsed with DI water.

2.5. XRD Detection

The crystallization of sample was detected by XRD with the following parameters: 0.154056 nm of wavelength of X-ray, 40 kV of working voltage, 30 mA of working current, 0.2° of scanning step, and 0∼90° of scanning range.

2.6. Measurement of Raman Spectrum

All Raman spectra in this paper were measured with dispersive Raman microscope of Horiba HP Evolution. The wavelength of excitation laser was set as 532 nm, the power was 0.2 mW, the diameter of the focused laser beam was 1 μm, and accumulation time was 10 s. Before measurement, the samples were dyed by R6G solution for half an hour. Then, the sample was taken out and dried at room temperature naturally in dark condition. Three different places were detected for each sample.

3. Results and Discussion

3.1. Synthesis of TiOx-NRs

After hydrothermal synthesis reaction, a white thin film of TiOx-NRs was coated on the conductive FTO glass surface. The effect of pH value on the growth of TiOx-NRs is shown in Figure 1. It can be found that the TiOx-NRs are cuboid and nearly vertical to the substrate with a small angle ranging from 0 to 15 degrees. The top surface of nanorods is in sidestep shape composed of dozens of nanowires, whereas sidewalls are quite flat. The length of nanorods is about 2 μm. Diameters of nanorods decrease with the increase in concentration of HCl.

For the low HCl concentration (e.g., 4 M shown in Figure 1(a)), the titanium resultants grow as nanorods with small intervals, some even combine together and form a huge one. For the high concentration (e.g., 6 M shown in Figure 1(c)), the titanium resultants are too small to gather in block mass. When the HCl concentration is 5 M, the diameter and intervals of TiOx-NRs are around 200 nm and 100 nm, respectively (shown in Figure 1(b)), which is suitable for Au-NPs deposition.

This result can be explained by the fact that high concentration of HCl (low pH) can suppress the rapid hydrolysis of TTIP in aqueous solution [37]. The morphology of TiOx would change from nanorods to nanowires. When the concentration of HCl reaches 7 M, there would be no TiOx-NRs.

As for [Cl], TiOx-NRs get looser and smaller with the increase in [Cl] concentration, as shown in Figure 2. The reason for this phenomenon can be explained by the fact that TTIP precursor would form [Ti(OH)2Clx(OH2)y]n± (x + y = 4) with the attendance of [Cl]. The intermediate products would suppress TTIP precursor hydrolysis [38] and result in smaller TiOx-NRs.

In view of 3D structure role in SERS substrates, TiOx-NRs with the largest surface area should be chosen. The samples shown in Figures 1(b) and 1(c) have much larger surface area. We decorated the Au-NPs on the two substrates with 3 × 10−4 M HAuCl4 and 1.5 × 10−2 M Na3Ct. The Raman spectra of the two samples dyed by 10−6 M R6G are shown in Figure 3. The Raman signal on sample synthesized with 5 M HCl showed much higher Raman enhancement. So the nanorod substrate synthesized with 5 M of HCl, 4.2 × 10−2 M of TTIP, and 0.7 M of NaCl was chosen as the basement in the following experiments.

The XRD spectrum for the optimal TiOx-NRs is shown in Figure 4, in which Au-NPs substrate shows that only one type of titanium oxide TiO1.04 rather than TiO2 is found in our substrate after annealing. The other diffraction peaks mainly come from SnO2 FTO glass substrate. There is hardly a diffraction peak of Au due to its low proportion. According to the work by Gruber and Krautz, the conductivity of TiOx is changed with x. When 1.0 ≤ x ≤ 1.03, TiOx shows metallic conductivity; when 1.04 ≤ x ≤ 1.24, TiOx shows semiconductivity [39].

Based on the charge transfer mechanism between semiconductor and absorbed molecules reported by Yang et al. [17], the energy is difficult to excite electrons crossing the bandgap of TiO2 (3.4 eV) or the energy gap between LUMO (lowest unoccupied molecular orbital) and HOMO (highest occupied molecular orbital) for absorbed molecules under 514.5 nm (2.4 eV) laser illumination. So charge transfer of TiO2 could only be realized by electrons excited from VB (valence band) to surface state energy levels firstly and then transfer to LUMO. Mott and Friedman reported that the bandgap of TiO1.5 is of order ∼1 eV [40]. This can be explained by the fact that there is less O atom vacancy in TiOx (x < 2) with lower x. Therefore, TiO1.04 with less than 1.0 eV bandgap is helpful in charge transferring across bandgap to LUMO easily under 532 nm laser (2.33 eV) illumination and can be considered as a kind of SERS substrate with high-detection sensitivity (Figure 5).

3.2. Decoration of Au-NPs

The morphology of Au-NPs attached by two methods has shown great differences. The Au-NPs were decorated by physical sputtering mainly gathered on the top surface of nanorods, as shown in Figure 6(a). The nanoparticles are spheroidicity with size about 20 nm. There is a shortage of Au-NPs to form hotspots, and the gathering phenomenon would also prevent electromagnetism enhancement between Au-NPs.

A large number of Au-NPs attached by chemical reduction methods are uniformly distributed on all surfaces of TiOx-NRs as shown in Figures 6(b)6(e). Compared with the physical sputtering, this kind of SERS substrate has more hotspots. Four kinds of SERS substrates with 1.0 × 10−4 M, 2.0 × 10−4 M, 3.0 × 10−4 M, and 4.0 × 10−4 M HAuCl4 were prepared, respectively. In nucleation-growth process, the size and nucleation rate of Au increase with the Na3Ct/HAuCl4 ratio decrease, when the amount of HAuCl4 is fixed [41]. In our experiment, we changed the HAuCl4 concentration intended to change the size and density of Au-NPs at the same time. As shown in Figures 6(b)6(e), the density of Au-NPs increases with HAuCl4 concentration. When HAuCl4 concentration was varied from 1.0 × 10−4 M to 3.0 × 10−4 M, the size and density of Au-NPs increase with the decrease in Na3Ct/HAuCl3 ratio. Once, HAuCl4 concentration reaches 4.0 × 10−4 M, Au-NPs size decrease. In consideration of the nucleation-growth process, nucleation, aggregation, and intraparticle ripening [41], this phenomenon can be attributed to the excessive nucleuses in the nucleation stage, resulting in HAuCl4 shortage in the aggregation stage. In Figure 6(d), the diameters of Au-NPs are around 40 nm, and the nanogaps between Au-NPs are in the range of 5 nm to 10 nm. After image analyzing with Image Pro, it shows that total area percentage of Au nanoparticles attaching on the nanorod sidewall surface is about 16%, while that on the top surface is about 12.28%.

3.3. Raman Scattering Measurement

The Raman spectra of 10−6 M R6G on four substrates are shown in Figure 7. According to the Raman signal of R6G, sample d shows the highest Raman enhancement. Although sample e has the highest density of Au-NPs, the Raman signal of sample e is lower than samples c and d. So the density and the size of hotspots should be considered at the same time when making SERS substrates. On the strongest substrate d, the diameter and gap of Au-NPs are around 40 nm and 5 nm–10 nm, respectively. The sample d was considered to be the optimal substrate for further research.

The surface plasmon resonant mechanism was verified by finite-different time-domain solutions in numerical soft Lumerical solutions. According to the results of the substrate we have made, three FDTD simulations were processed. The exciting light was 532 nm, and the light rotated 15 degrees to normal line in accordance with TiOx-NRs-based angle on FTO glass. It was assumed that Au-NPs were uniformly distributed on our substrate. Simulation results are shown in Figure 8. Comparing Figure 8(a) with Figure 8(b), the SERS substrate with 40 nm Au-NPs had more intensive electric field between Au-NPs than the substrate with 20 nm Au-NPs. Comparing Figure 8(a) with Figure 8(c), when the distance between nanorods was close enough, the electric field enhanced for Au-NPs on different nanorods would occur. Through simulation, when the diameter of Au-NPs and gap between TiOx-NRs are 40 nm and 100 nm, respectively, there would be the highest surface electric field enhancement. The distribution of Au-NPs and TiOx-NRs in experiment was not as uniform as the simulation model, but the simulation can reveal the tendency.

After that, the Raman spectra of SERS substrates dyed by R6G with concentration ranging from 10−6 M to 10−11 M are detected, and the results are shown in Figure 9. Sharp Raman signal peaks of R6G (e.g., 1650.6, 1573.7, 1508.5, 1362, 1313, 1186, 774, and 611 cm−1) could be still found on the sample dyed by 10−10 M R6G. The peak location in the spectra was about 2 cm−1 shift compared with Peter Hildebrandt’s work [42]. Both R6G chemical absorption on Au-NP and R6G interaction with semiconductor substrate could cause Raman shift [17, 43].

The enhancement factor of 10−6 M and 10−10 M R6G on the substrates was calculated by using the formula EFSERS=[(ISERS/CSERS)/(IRef/CRef)] [44], respectively, where ISERS is the Raman signal intensity of R6G on our SERS substrate, IRef is the Raman signal intensity of R6G on the TiOx-NRs array without Au-NPs, and C represents R6G concentration. The Raman signal intensity of 10−1 M R6G was chosen as the reference group. The Raman intensity at peak 1650.62 cm−1 was chosen to calculate. The Raman enhancement factors of 10−6 M and 10−10 M R6G were calculated as 6.65 × 105 and 2.64 × 108, respectively.

From Figure 9, it could be found that the Raman enhancement of the substrate is hardly weakened even after immersed in water for one month. The relative standard deviation (RSD) of all main Raman peaks of R6G on the substrate was calculated by detecting ten different points randomly on the same sample. The values are ranging from 10.4% to 16.7%. As stated above, it suggests that this SERS substrate is effective, stable, and uniform.

4. Conclusions

A novel SERS substrate based on semiconductor TiOx-NRs decorated with Au-NPs on FTO glass was prepared by hydrothermal synthesis and citrate reduction.

For the hydrothermal synthesis of TiOx-NRs, higher HCl concentration and NaCl concentration could result in the looser and smaller TiOx-NRs. Our substrate material is TiO1.04 rather than TiO2. TiO1.04 with charge transfer mechanism could perform better than TiO2 in the SERS detection. Compared with sputtering Au-NPs, those formed by citrate reduction shows better distribution uniformity and sidewall attaching ability.

TiOx-NRs with the diameter from 150 nm to 200 nm, Au-NPs with the diameter 40 nm, and 5∼10 nm nanogaps between Au-NPs show the highest Raman sensitivity. For this kind of SERS substrate, the detection limit of R6G reaches 10−10 M, and the enhancement factor is about 2.64 × 108. Meanwhile, the Raman signal on the substrate shows good area uniformity and long-time stability.

In summary, this kind of TiOx-NRs SERS substrate decorated with uniform Au-NPs shows high sensitivity and has great application potential in trace-level organic molecule detection.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

The authors gratefully acknowledge the financial support from the National High-Tech R&D Program of China (2012AA02A104).