Surface Plasmon-Enhanced Photoelectrochemical Sensor Based on Au Modified TiO2 Nanotubes

Based on the enhanced charge separation efficiency of the one-dimensional structure and strong surface plasmon resonance (SPR) of gold, a gold modified TiO2 nanotube (Au/TiO2NTs) glucose photoelectrochemical (PEC) sensor was prepared. It could be activated by visible red light (625 nm). Under optimal conditions, the Au/TiO2NTs sensor exhibited a good sensitivity of 170.37 μA·mM−1·cm−2 in the range of 1–90 μM (R2 = 0.9993), and a detection limit of 1.3 μM (S/N = 3). Due to its high selectivity, good anti-interference ability, and long-term stability, the fabricated Au/TiO2NTs sensor provides practical detection of glucose. It is expected to be used in the construction of non-invasive PEC biosensors.


Introduction
Glucose is one of the body's primary energy sources and an important biomarker in clinical diabetes diagnosis. Diabetes is a severe disease that can lead to severe complications, such as blindness, lower limb amputations, and cardiovascular disease. Sometimes, it can even lead to death. According to the latest report, the number of adults with diabetes worldwide reached 537 million in 2021, with one person dying from diabetes every five seconds on average. Although there is no effective cure for diabetes, complications associated with the disease can be reduced by tightly controlling blood sugar levels. Therefore, it is significant to explore fast and reliable glucose concentration monitoring methods in the treatment of diabetes [1]. Among the reported methods for detecting glucose, photoelectrochemical (PEC) technology has been widely considered for its simple operation, high sensitivity, low detection limit, high specificity, low cost, and so on [2][3][4]. The PEC process is generally composed of photoelectric conversion and an electrochemical process [5]. The analyte changes the properties of the photosensitive material or the electrolyte environment, resulting in a change in the PEC signal intensity; thereby realizing quantitative analysis of the analyte concentration [6].
As the central part of PEC biosensors, the semiconductor materials play the role of molecular fixation, signal generation, and transmission in the system [7,8]. Therefore, the photoelectric properties of the used semiconductors are vital for the performance of PEC sensors. Many photoactive semiconductors with excellent PEC performance have been investigated and applied [9,10].
TiO 2 is considered to be the optimal and typical semiconductor nanomaterial for the construction of PEC biosensors, due to its great photocatalytic performance, good range. The Au/TiO 2 NTs sensor exhibited high sensitivity, excellent linearity, low detection limit, good selectivity, and stability, and it is expected to be used for non-invasive blood glucose detection.

Preparation of Titanium Dioxide Nanotubes (TiO 2 NTs)
The hydrogen titanate nanotubes (HTNTs) and stabilized TiO 2 NTs were prepared according to our previous work [45]. Briefly, pure anatase TiO 2 powder was treated by the hydrothermal method in 10 M NaOH solution at 150 • C for 12 h, and then washed with ultrapure water to pH = 7. Then, 0.1 mol/L HNO 3 was added to the solution until pH = 2~3. Then, the obtained HTNTs were washed with ultrapure water until the solution was neutral. After centrifugation, the obtained sample was dried at 80 • C for 10 h.
Stabilized TiO 2 NTs were synthesized using the sol impregnation method, with HTNTs as the precursors. HTNTs were added in titanium sol and stirred for 4 h. After drying at room temperature, the nanotubes were calcined at 400 • C in the air for 2 h to obtain TiO 2 NTs.

Preparation of Au/TiO 2 NTs
The gold-modified TiO 2 NTs were prepared by a photoreduction process. First, 0.3 g TiO 2 NTs were dispersed in a mixed solution of 30 mL anhydrous ethanol and 30 mL deionized water, stirred for 0.5 h, and ultrasonically rinsed for 0.5 h. The above suspension was then bubbled with nitrogen for 15 min, to remove the dissolved oxygen. Then, different volumes of 0.01 mol/L HAuCl 4 ·4H 2 O solution were added, and the mass percentage of gold was 0.25, 0.50, 0.75, and 1.0 wt.%, respectively. Subsequently, the suspension was irradiated under ultraviolet light (300 W Hg lamp) for 2.5 h. Finally, the prepared samples were centrifugated and washed several times, and then dried at 80 • C for 8 h. According to the mass percentage of Au, the obtained samples were marked as 0.25, 0.50, 0.75, and 1.0 wt.% Au/TiO 2 NTs, respectively. The actual gold loadings in Au/TiO 2 NTs were 0.193, 0.356, 0.519, and 0.746 wt.%, respectively. This can be attributed to the gold losses in the preparation process.

Fabrication of Au/TiO 2 NTs Photoelectrode
The 4-mg as-prepared samples were first dissolved in the mix solution of 720 µL deionized water, 240 µL anhydrous ethanol, and 40 µL 5% Nafion solution. Then the content was ultrasonicated for 0.5 h, to obtain a homogenous suspension. Then, 10 µL of suspension was dropped on the surface of a glassy carbon electrode (GCE, 4 mm in diameter) and dried at room temperature. The coated GCE was denoted as Au/TiO 2 NTs photoelectrodes.

Characterization
X-ray diffraction (XRD) was characterized using a diffractometer (Rigaku SmartLab, Rigaku Corporation, Tokyo, Japan). Transmission electron microscope (TEM) images were Nanomaterials 2022, 12, 2058 4 of 14 observed by a Talos F200X G2 instrument. X-ray Photoelectron Spectroscopy (Kratos Axis Ultra DLD, Kratos Analytical Ltd., Manchester, UK) recorded the chemical composition and oxidation state of the elements of samples. The actual Au loadings on the catalysts were examined using an inductive coupled plasma emission spectrometer (ICP-OES, Spectroblue, Spectro, Kleve, Germany). The UV-Vis diffuse reflection spectra (UV-Vis DRS) were recorded on a UV-Vis spectrophotometer (UV3600-Plus, Shimadzu, Kyoto, Japan). Photoluminescence spectra (PL) were recorded on a spectrophotometer (PTI, New York, NY, USA). Electrochemical measurements were performed on a Zahner Zennium electrochemical workstation (Kronach-Gundelsdorf, Germany). An XPA-7 photocatalytic device was from XuJiang Electromechanical Plant, Nanjing, China, which was used in the photoreduction process.

Photoelectrochemical Measurements
The photoelectrochemical measurements were performed on a workstation via a threeelectrode system at room temperature. The platinum wire electrode and Ag/AgCl electrode were used as counter electrode and reference electrode, respectively, while Au/TiO 2 NTs nanocomposite modified GCE (4 mm diameter) was utilized as a working electrode. The electrolyte was composed of 0.1 M NaOH solution in a 50-mL electrolytic cell. A LED lamp (625 nm, 30.98 W/m 2 ) was selected as the illumination source during the PEC testing process.

Structure Characterization by XRD
The XRD measurement was carried out to explore the materials' phase. Figure 1 shows the XRD patterns of the pure TiO 2 NTs and the Au/TiO 2 NTs with different gold contents (0.25, 0.5, 0.75, and 1.0 wt.%). As shown in Figure 1, the as-prepared TiO 2 NTs were anatase phase, with principal diffraction peaks at 2θ = 25. were observed, and no diffraction peak related to metal Au was observed, which could be attributed to the relatively low loading of Au and the high dispersion in TiO 2 NTs. In addition, the diffraction peaks of the photocatalyst were sharp and high, indicating that the samples had good crystallinity. There were no diffraction peaks of other phases in Figure 1, indicating that the synthesized photocatalysts had good purity.

Morphology Characterization of Samples by TEM
The microstructure of the as-prepared Au/TiO2NTs was characterized by TEM. From the TEM images, nearly uniform nanotube-like structures could be clearly seen. The Au/TiO2NTs

Morphology Characterization of Samples by TEM
The microstructure of the as-prepared Au/TiO 2 NTs was characterized by TEM. From the TEM images, nearly uniform nanotube-like structures could be clearly seen. The Au/TiO 2 NTs were hundreds of nanometers long, and the outer diameter was about 10 nanometers. As seen from Figure 2a, the black dots were confirmed to be AuNPs, with a size of about 8-9 nm. Figure 2b was the high-resolution HRTEM image of Au/TiO 2 NTs, in which the lattice fringes of 0.347 nm and 0.233 nm correspond well to anatase TiO 2 (101) and Au (111) interfaces, respectively. It could be observed that the Au modification did not influence the nanotubes' structure. Furthermore, EDS element mapping diagrams of Au/TiO 2 NTs were performed (Figure 2c-f). Except for the large gold particles, large amounts of gold particles with very small sizes can also be observed. They were highly dispersed on the nanotubes. As a result, Au/TiO 2 NTs with sufficient gold loading were obtained.

Morphology Characterization of Samples by TEM
The microstructure of the as-prepared Au/TiO2NTs was characterized by TEM. From the TEM images, nearly uniform nanotube-like structures could be clearly seen. The Au/TiO2NTs were hundreds of nanometers long, and the outer diameter was about 10 nanometers. As seen from Figure 2a, the black dots were confirmed to be AuNPs, with a size of about 8-9 nm. Figure 2b was the high-resolution HRTEM image of Au/TiO2NTs, in which the lattice fringes of 0.347 nm and 0.233 nm correspond well to anatase TiO2 (101) and Au (111) interfaces, respectively. It could be observed that the Au modification did not influence the nanotubes' structure. Furthermore, EDS element mapping diagrams of Au/TiO2NTs were performed (Figure 2c-f). Except for the large gold particles, large amounts of gold particles with very small sizes can also be observed. They were highly dispersed on the nanotubes. As a result, Au/TiO2NTs with sufficient gold loading were obtained.

The Chemical State and Composition Analysis Using XPS
To further study the electronic interaction between TiO 2 and Au, XPS was performed. The full XPS spectrum is shown in Figure 3a, in which the electronic states of Ti 2p, O 1s, C 1s, and Au 4f can be observed. The mass percentage of Au in the 0.5 wt.% Au/TiO 2 NTs was 0.369 wt.%, which is consistent with the result obtained by ICP. The C signal might come from the XPS measurement itself. Figure 3b shows the Ti XPS spectrum. The binding energy of Ti 2p 1/2 and Ti 2p 3/2 were located at 464.3 eV and 458.6 eV, respectively, indicating that Ti exists in the +4 valence state [46,47]. It corresponded to the existence of stoichiometric TiO 2 . The O 1s XPS spectrum showed two peaks at a binding energy at 531.2 eV and 529.8 eV (Figure 3c), respectively, and belonging to Ti-OH and lattice oxygen [Ti-O 6 ] species, respectively [48,49]. Figure 3d shows the Au 4f XPS spectrum of Au/TiO 2 NTs. It can be clearly observed that the samples showed a main Au 4f 7/2 peak at 83.2 eV, with an analogous Au 4f 5/2 peak at 87.0 eV, which indicated that gold in the Au/TiO 2 NTs was presented in the metallic state [50]. In addition, the peak of Au 4f was negatively shifted compared with the characteristic peak of standard Au (83.7 eV), indicating that Au was electron-rich, which may have been caused by e − transfer from the carrier to gold. This showed that there was a strong interaction between TiO 2 and gold. Peaks of oxidized gold species were not observed, which should be around 86.3 and 85.5 eV. It seemed that the trivalent gold in chloroauric acid had been reduced to elemental gold after the photoreduction process. pared with the characteristic peak of standard Au (83.7 eV), indicating that Au was electron-rich, which may have been caused by e − transfer from the carrier to gold. This showed that there was a strong interaction between TiO2 and gold. Peaks of oxidized gold species were not observed, which should be around 86.3 and 85.5 eV. It seemed that the trivalent gold in chloroauric acid had been reduced to elemental gold after the photoreduction process.

Optical Absorption Properties Analysis by UV-Vis DRS
To further analyze the optical absorption ability of the material, the UV-Vis DRS test was selected to characterize the samples. It could be clearly seen from the UV-Vis DRS spectrum that both TiO2NTs and Au/TiO2NTs had a strong absorption peak at 230-380 nm. It was confirmed that the loading of noble metals did not change the inherent absorption of TiO2. Combined with Figure 4a,b, it could be seen that the absorption edge of TiO2 nanotubes had a redshift after being loaded with gold. The corresponding bandgap widths of TiO2NTs and Au/TiO2NTs were 3.2 and 3.0 eV, respectively. As the absorption edge expanded, the bandgap

Optical Absorption Properties Analysis by UV-Vis DRS
To further analyze the optical absorption ability of the material, the UV-Vis DRS test was selected to characterize the samples. It could be clearly seen from the UV-Vis DRS spectrum that both TiO 2 NTs and Au/TiO 2 NTs had a strong absorption peak at 230-380 nm. It was confirmed that the loading of noble metals did not change the inherent absorption of TiO 2 . Combined with Figure 4a,b, it could be seen that the absorption edge of TiO 2 nanotubes had a redshift after being loaded with gold. The corresponding bandgap widths of TiO 2 NTs and Au/TiO 2 NTs were 3.2 and 3.0 eV, respectively. As the absorption edge expanded, the bandgap decreased; that is, the spectrum range that TiO 2 NTs could respond to after loading the noble metal became significantly wider. This could improve the deficiency of TiO 2 catalyst and help the photocatalyst to capture more wavelengths of light; thus, generating more photogenerated electrons to participate in a photoelectric chemical reaction. Moreover, obvious absorption peaks in the visible light range of 500-600 nm could be observed in Au/TiO 2 NTs catalysts, which were attributed to the SPR effect of the loaded AuNPs [51]. As a result, the loaded AuNPs could significantly improve the absorption of TiO 2 NTs under visible light and improve the PEC performance of TiO 2 NTs.

The Optical Properties Analysis by Photoluminescence (PL) Spectroscopy
The recombination ability of e − and h + was characterized by photoluminescence intensity. As displayed in Figure 5, the photoluminescence spectra of the samples were evaluated at the excitation wavelength of 300 nm. The PL intensity of Au/TiO 2 NTs was lower than that of pure TiO 2 NTs, which indicated that the loading of gold was beneficial for inhibiting the recombination of e − and h + and improving the photoelectric performance. As indicated in the XPS spectra, the formation of the Schottky junction enabled the gold to become electron-rich. Thus, the e − and h + pairs could be effectively separated, and the charge recombination could be avoided. Therefore, the 0.5 wt.% Au/TiO 2 NTs have a higher photogenerated carrier separation rate and mobility than TiO 2 NTs, which was beneficial for generating more photogenerated e − and h + , and exhibiting better PEC performance than the pure TiO 2 NTs. the photocatalyst to capture more wavelengths of light; thus, generating more photogener electrons to participate in a photoelectric chemical reaction. Moreover, obvious absorp peaks in the visible light range of 500-600 nm could be observed in Au/TiO2NTs cata which were attributed to the SPR effect of the loaded AuNPs [51]. As a result, the lo AuNPs could significantly improve the absorption of TiO2NTs under visible light and prove the PEC performance of TiO2NTs.

The Optical Properties Analysis by Photoluminescence (PL) Spectroscopy
The recombination ability of e − and h + was characterized by photoluminescenc tensity. As displayed in Figure 5, the photoluminescence spectra of the samples were uated at the excitation wavelength of 300 nm. The PL intensity of Au/TiO2NTs was lo than that of pure TiO2NTs, which indicated that the loading of gold was beneficia inhibiting the recombination of e − and h + and improving the photoelectric performa As indicated in the XPS spectra, the formation of the Schottky junction enabled the to become electron-rich. Thus, the e − and h + pairs could be effectively separated, and charge recombination could be avoided. Therefore, the 0.5 wt.% Au/TiO2NTs ha higher photogenerated carrier separation rate and mobility than TiO2NTs, which was eficial for generating more photogenerated e − and h + , and exhibiting better PEC pe mance than the pure TiO2NTs.

The Optical Properties Analysis by Photoluminescence (PL) Spectroscopy
The recombination ability of e − and h + was characterized by photoluminescence intensity. As displayed in Figure 5, the photoluminescence spectra of the samples were evaluated at the excitation wavelength of 300 nm. The PL intensity of Au/TiO2NTs was lower than that of pure TiO2NTs, which indicated that the loading of gold was beneficial for inhibiting the recombination of e − and h + and improving the photoelectric performance. As indicated in the XPS spectra, the formation of the Schottky junction enabled the gold to become electron-rich. Thus, the e − and h + pairs could be effectively separated, and the charge recombination could be avoided. Therefore, the 0.5 wt.% Au/TiO2NTs have a higher photogenerated carrier separation rate and mobility than TiO2NTs, which was beneficial for generating more photogenerated e − and h + , and exhibiting better PEC performance than the pure TiO2NTs.  3.6. Photoelectrochemical Measurements 3.6.1. The Influence of Potential and Au Load on the Performance of Biosensing A photocurrent-time curve can characterize the separation efficiency of the photogenerated carriers. Experimental parameters may affect PEC performance. Thus, the experimental parameters, such as bias voltages and gold contents, must be optimized. The application of bias voltages on the photoelectrode can facilitate the separation of e − and h + and reduce their recombination rate, which can be directly reflected in the enhancement of photocurrent. The influence of voltages on the photocurrent of Au/TiO 2 NTs electrode under the irradiation of 0.1 M NaOH solution at 625 nm was investigated. In the range of 0-0.5 V, the maximum photocurrent response was at 0.3 V (Figure 6a). Therefore, 0.3 V was determined as the optimum bias voltage. In addition, it is necessary to optimize the effect of gold content on photocurrent response. Under the irradiation of 625 nm, the influence of gold loading content was studied in 0.1 M NaOH solution. It can be seen from Figure 6b that no significant photocurrent was observed on the bare TiO 2 NTs, while an obvious photocurrent could be observed on the gold-modified ones. Due to its wide bandgap, TiO 2 only could absorb short-wavelength ultraviolet light. As the gold contents increased, the photocurrent intensity first increased, reaching a maximum value of 0.5 wt.% Au/TiO 2 NTs, and then gradually decreased. Figure 6b that no significant photocurrent was observed on the bare TiO2NTs, while an obvious photocurrent could be observed on the gold-modified ones. Due to its wide bandgap, TiO2 only could absorb short-wavelength ultraviolet light. As the gold contents increased, the photocurrent intensity first increased, reaching a maximum value of 0.5 wt.% Au/TiO2 NTs, and then gradually decreased. In the Au/TiO2NTs, as a result of the high work function of gold, the electrons migrate from TiO2 to gold. When the Fermi levels of the gold and TiO2 are equal, a Schottky barrier will form. Consequently, the electrons and holes generated after illumination can be localized on gold and TiO2, respectively, be separated, and undergo oxidation-reduction reactions at different positions. When glucose exists in the system, the holes on the electrode surface will react with the glucose in the solution, and glucose is easily oxidized. When the gold content is too low, there are not enough electron receiving centers for effective e − and h + separation. However, when the gold content is too high, the gold coating on the surface of TiO2 has a shielding effect and affects the acceptance of electrons by TiO2. Moreover, the further deposition of gold results in the growth of AuNPs, which leads to the aggregation of gold atoms, the reduction of active sites, and the formation of recombination centers for e − and h + on gold particles. In addition, with too much gold loading, the color of the sample is too dark and this leads to light avoidance, resulting in a decrease in the photocurrent. Therefore, Au/TiO2NTs with 0.5 wt.% gold content exhibited the best PEC performance in this system. Figure 7a shows the photocurrent curve on the Au/TiO2NTs electrode in different concentrations of glucose (1-90 μM). It can be observed that the intensity of photocurrent In the Au/TiO 2 NTs, as a result of the high work function of gold, the electrons migrate from TiO 2 to gold. When the Fermi levels of the gold and TiO 2 are equal, a Schottky barrier will form. Consequently, the electrons and holes generated after illumination can be localized on gold and TiO 2 , respectively, be separated, and undergo oxidation-reduction reactions at different positions. When glucose exists in the system, the holes on the electrode surface will react with the glucose in the solution, and glucose is easily oxidized. When the gold content is too low, there are not enough electron receiving centers for effective e − and h + separation. However, when the gold content is too high, the gold coating on the surface of TiO 2 has a shielding effect and affects the acceptance of electrons by TiO 2 . Moreover, the further deposition of gold results in the growth of AuNPs, which leads to the aggregation of gold atoms, the reduction of active sites, and the formation of recombination centers for e − and h + on gold particles. In addition, with too much gold loading, the color of the sample is too dark and this leads to light avoidance, resulting in a decrease in the photocurrent. Therefore, Au/TiO 2 NTs with 0.5 wt.% gold content exhibited the best PEC performance in this system. Figure 7a shows the photocurrent curve on the Au/TiO 2 NTs electrode in different concentrations of glucose (1-90 µM). It can be observed that the intensity of photocurrent increased with the increase of the concentrations of glucose. In essence, they exhibited a linear regression equation: j (nA/cm 2 ) = 4011.7782 + 170.3722 Cglucose (µM) with a correlation coefficient of 0.9993, where j is photocurrent density, and C is the concentration of the glucose (Figure 7b). The corresponding sensitivity of the Au/TiO 2 NTs biosensor was 170.37 µA·mM −1 ·cm −2 . The detection limit (LOD) of the biosensor was 1.3 µM, which was obtained from LOD = 3sb/S, (sb represents standard deviation of blank signal, and S represents sensitivity). Table 1 summarizes the comparison of different glucose biosensors. Compared with the reported PEC sensors, the Au/TiO 2 NTs in this work exhibited a long excitation wavelength, high sensitivity, and low LOD. Most excitation light sources in PEC sensor research were ultraviolet or visible light with short wavelengths, which could cause cell damage or degeneration. In our work, visible red light with a wavelength of 625 nm was selected as the excitation wavelength. This low-energy long-wavelength light is more conducive to noninvasive blood glucose monitoring than short-wavelength light.

PEC Biosensing Application for Glucose
At present, blood analysis is the most commonly used method to monitor blood glucose, which has the characteristics of invasiveness, infection, bleeding, complex operation, and low efficiency. Compared to the traditional blood glucose technique, the high sensitivity and low LOD of the Au/TiO 2 NTs PEC sensors are beneficial to the realization of in vitro detection and provide the possibility of non-invasive glucose detection for humans, such as using interstitial fluid and sweat. lation coefficient of 0.9993, where j is photocurrent density, and C is the concentration of the glucose (Figure 7b). The corresponding sensitivity of the Au/TiO2NTs biosensor was 170.37 μA·mM −1 ·cm −2 . The detection limit (LOD) of the biosensor was 1.3 μM, which was obtained from LOD = 3sb/S, (sb represents standard deviation of blank signal, and S represents sensitivity).  Table 1 summarizes the comparison of different glucose biosensors. Compared with the reported PEC sensors, the Au/TiO2NTs in this work exhibited a long excitation wavelength, high sensitivity, and low LOD. Most excitation light sources in PEC sensor research were ultraviolet or visible light with short wavelengths, which could cause cell damage or degeneration. In our work, visible red light with a wavelength of 625 nm was selected as the excitation wavelength. This low-energy long-wavelength light is more conducive to noninvasive blood glucose monitoring than short-wavelength light. At present, blood analysis is the most commonly used method to monitor blood glucose, which has the characteristics of invasiveness, infection, bleeding, complex operation, and low efficiency. Compared to the traditional blood glucose technique, the high sensitivity and low LOD of the Au/TiO2NTs PEC sensors are beneficial to the realization of in vitro detection and provide the possibility of non-invasive glucose detection for humans, such as using interstitial fluid and sweat.

PEC Biosensing Selectivity and Stability Analysis
Selectivity is an important parameter of biosensors. Figure 8 shows the photocurrent comparison of a 0.5 wt.% Au/TiO 2 NTs electrode in glucose, sucrose, lactose, ascorbic acid, saccharose, and fructose solutions (the concentration of the solution was 90 µM). It was obvious that the photocurrent generated on the Au/TiO 2 NTs electrode in the glucose solution was much greater than in other solutions, indicating that the Au/TiO 2 NT PEC sensor had a good selectivity.

PEC Biosensing Selectivity and Stability Analysis
Selectivity is an important parameter of biosensors. Figure 8 shows the photocurrent comparison of a 0.5 wt.% Au/TiO2NTs electrode in glucose, sucrose, lactose, ascorbic acid, saccharose, and fructose solutions (the concentration of the solution was 90 μM). It was obvious that the photocurrent generated on the Au/TiO2NTs electrode in the glucose solution was much greater than in other solutions, indicating that the Au/TiO2NT PEC sensor had a good selectivity. Furthermore, the photocurrent response stability of the Au/TiO2NT electrode was investigated by switching the i-t response 10 times. As can be seen from Figure 9, the photocurrent increased rapidly at the moment of turning on the light and became stable within a few seconds. With switching 10 times, the photocurrent remained stable, and the photocurrent intensity had no obvious change in each cycle. The results showed that the Furthermore, the photocurrent response stability of the Au/TiO 2 NT electrode was investigated by switching the i-t response 10 times. As can be seen from Figure 9, the photocurrent increased rapidly at the moment of turning on the light and became stable within a few seconds. With switching 10 times, the photocurrent remained stable, and the photocurrent intensity had no obvious change in each cycle. The results showed that the Au/TiO 2 NTs had a good photocurrent response stability. Furthermore, the photocurrent response stability of the Au/TiO2NT electrode was investigated by switching the i-t response 10 times. As can be seen from Figure 9, the photocurrent increased rapidly at the moment of turning on the light and became stable within a few seconds. With switching 10 times, the photocurrent remained stable, and the photocurrent intensity had no obvious change in each cycle. The results showed that the Au/TiO2NTs had a good photocurrent response stability.  Long-term stability is another important parameter of sensor performance. The 0.5 wt.% Au/TiO 2 NT electrode was stored in a room temperature atmosphere, and the photocurrent-time curve was continuously tested every day in a 50 µM glucose solution. Figure 10 shows the data for 25 days. It can be observed that the photocurrent intensity of the Au/TiO 2 NTs electrode maintained 96% of its original photocurrent intensity, indicating that the Au/TiO 2 NTs glucose sensor had a good long-term stability. Long-term stability is another important parameter of sensor performance. The 0.5 wt.% Au/TiO2NT electrode was stored in a room temperature atmosphere, and the photocurrent-time curve was continuously tested every day in a 50 μM glucose solution. Figure 10 shows the data for 25 days. It can be observed that the photocurrent intensity of the Au/TiO2NTs electrode maintained 96% of its original photocurrent intensity, indicating that the Au/TiO2NTs glucose sensor had a good long-term stability. To clarify the PEC process of Au/TiO2NTs, we elucidate the mechanism of action of the glucose sensor in Scheme 1. The modification of AuNPs plays a key role in the PECgenerated e − and h + in TiO2. The SPR effect of AuNPs enhances the light absorption of TiO2 under visible light. In addition, the large gold work function causes the Au/TiO2NTs interface to form a Schottky barrier. Therefore, the photoinduced electrons are easily transferred to the gold particles, thereby decreasing the recombination of eand h + and accelerating the separation efficiency of photoinduced carriers. Assisted by a bias voltage between the counter electrode and the working electrode, electrons are transferred along AuNPs to the glassy carbon electrode and eventually to the counter electrode via an external circuit. The holes on the Au/TiO2NTs electrode directly oxidize the glucose mole- To clarify the PEC process of Au/TiO 2 NTs, we elucidate the mechanism of action of the glucose sensor in Scheme 1. The modification of AuNPs plays a key role in the PECgenerated e − and h + in TiO 2 . The SPR effect of AuNPs enhances the light absorption of TiO 2 under visible light. In addition, the large gold work function causes the Au/TiO 2 NTs interface to form a Schottky barrier. Therefore, the photoinduced electrons are easily transferred to the gold particles, thereby decreasing the recombination of eand h + and accelerating the separation efficiency of photoinduced carriers. Assisted by a bias voltage between the counter electrode and the working electrode, electrons are transferred along AuNPs to the glassy carbon electrode and eventually to the counter electrode via an external circuit. The holes on the Au/TiO 2 NTs electrode directly oxidize the glucose molecules. During this process, the glucose concentration changes, resulting in a change in the photocurrent. Consequently, the linear relation between photocurrent and glucose concentration is presented.
To clarify the PEC process of Au/TiO2NTs, we elucidate the mechanism of action of the glucose sensor in Scheme 1. The modification of AuNPs plays a key role in the PECgenerated e − and h + in TiO2. The SPR effect of AuNPs enhances the light absorption of TiO2 under visible light. In addition, the large gold work function causes the Au/TiO2NTs interface to form a Schottky barrier. Therefore, the photoinduced electrons are easily transferred to the gold particles, thereby decreasing the recombination of eand h + and accelerating the separation efficiency of photoinduced carriers. Assisted by a bias voltage between the counter electrode and the working electrode, electrons are transferred along AuNPs to the glassy carbon electrode and eventually to the counter electrode via an external circuit. The holes on the Au/TiO2NTs electrode directly oxidize the glucose molecules. During this process, the glucose concentration changes, resulting in a change in the photocurrent. Consequently, the linear relation between photocurrent and glucose concentration is presented.

Conclusions
To sum up, a high-sensitivity and SPR-enhanced PEC sensor based on TiO 2 NTs was fabricated, in which Au NPs were decorated on TiO 2 NTs using the photoreduction method. Via gold modification, the absorption of TiO 2 under visible red light was promoted, attributable to the SPR effect, and the separation of e − and h + was promoted through the formation of the Schottky junction. The resultantly enhanced electrochemical performance produced the high sensitivity of the Au/TiO 2 NTs for glucose under visible red light. In addition, the low detection limit, high selectivity, and long-term stability for glucose provide the possibility of the noninvasive detection of glucose. In addition to glucose, we predict that this PEC sensor has potential applications for other substances, such as glutathione and dopamine.