Sensitive and Selective Detection of Tartrazine Based on TiO2-Electrochemically Reduced Graphene Oxide Composite-Modified Electrodes

TiO2-reduced graphene oxide composite-modified glassy carbon electrodes (TiO2–ErGO–GCE) for the sensitive detection of tartrazine were prepared by drop casting followed by electrochemical reduction. The as-prepared material was characterized by transmission electron microscopy (TEM) and X-ray diffraction (XRD). Cyclic voltammetry and second-order derivative linear scan voltammetry were performed to analyze the electrochemical sensing of tartrazine on different electrodes. The determination conditions (including pH, accumulation potential, and accumulation time) were optimized systematically. The results showed that the TiO2–ErGO composites increased the electrochemical active area of the electrode and enhanced the electrochemical responses to tartrazine significantly. Under the optimum detection conditions, the peak current was found to be linear for tartrazine concentrations in the range of 2.0 × 10−8–2.0 × 10−5 mol/L, with a lower detection limit of 8.0 × 10−9 mol/L (S/N = 3). Finally, the proposed TiO2–ErGO–GCEs were successfully applied for the detection of trace tartrazine in carbonated beverage samples.


Introduction
With the rapid development of modern life, artificially sweetened beverages attract a lot of children because of their colorful appearance and sweet taste. Food colorants are commonly used in these artificially sweetened beverages as additives. Tartrazine is an important food colorant used in various kinds of drinks and foods [1,2]. However, the azodi group and aromatic ring in tartrazine are found to be a potential risk to human health; thus, the dosage of tartrazine used should be monitored constantly. The acceptable daily intake (ADI) of tartrazine recommended by the FDA is 3.75 mg/kg for humans. Moreover, the World Health Organization (WHO) suggest that the ADI could be limited to 2.5 mg/kg. The excessive intake of tartrazine could cause various diseases, such as allergy, asthma, eczema, anxiety, migraine, and even cancer [3,4]. More importantly, the accidental abuse of tartrazine (such as from tainted steamed buns) has often happened in recent years. Thus, detecting the content of tartrazine in foods and drinks is important for improving food security in the modern society. our knowledge, TiO 2 -graphene nanocomposite-modified electrodes for tartrazine have rarely been reported. A novel electrochemical sensor based on graphene-mesoporous TiO 2 -modified carbon paste electrodes was developed for the detection of trace tartrazine and exhibited a wide linear detection range (0.02-0.18 µM) and a low detection limit (8.0 nM) using square wave voltammetry (SWV). However, the graphene was obtained by chemically reduction of graphene oxide, which requires a poisonous reductant and several steps in the reduction process. Herein, TiO 2 -electrochemically reduced graphene oxide-modified glassy carbon electrodes (TiO 2 -ErGO-GCEs) were prepared by a facile hydrothermal and electrochemical reduction method. Second-order derivative linear sweep voltammetry was employed to detect tartrazine, because of its advantages, including higher sensitivity and better selectivity than differential pulse voltammetry (DPV) and square wave voltammetry (SWV) [30]. The morphologies and structures of these samples were analyzed by transmission electron microscopy (TEM) and powder X-ray diffraction (XRD). Moreover, the electrochemical behavior of tartrazine on TiO 2 -ErGO-GCE was investigated in detail. Furthermore, various electrochemical parameters (pH, scan rate, accumulation potential, and time) were discussed. Finally, the TiO 2 -ErGO-GCE was successfully applied for tartrazine detection in a carbonate beverage.

Synthesis of TiO 2 Nanoparticles (NPs)
The TiO 2 NPs were synthesized according to a published method [31]. Typically, 4.899 g of Ti(SO 4 ) 2 were dissolved in 50 mL of water under stirring for 30 min. This solution was transferred to a Teflon-lined stainless-steel autoclave (100 mL) and reacted at 200 • C for 4 h. After cooling to room temperature, the TiO 2 reactants were centrifugated at 7920 rcf. Then, the samples were washed with water and ethyl alcohol for several times, and the TiO 2 NPs were obtained by drying at 60 • C in vacuum for 10 h.

Synthesis of TiO 2 -GO Composite Nanomaterials
Graphene oxide (GO) was synthesized by the modified Hummers' method [32]. In a typical process, concentrated H 2 SO 4 was cooled down to 0 • C, and 0.5 g of graphite powder and NaNO 3 were added subsequently under stirring. Then, 3.0 g of KMnO 4 was added slowly at 5 • C. After that, the temperature was raised to 35 • C for 2 h under stirring to form a mash, and 40 mL of water was added at 50 • C; then, the temperature was increased to 95 • C for 0.5 h. The above solution was added to 20 mL of 30% H 2 O 2 in batches. The as-obtained precipitate was washed with 150 mL of hydrochloric acid (1:10) and 150 mL of H 2 O. Then, it was vacuum-dried at 50 • C for 12 h to obtain the graphite oxide. Subsequently, 100 mg of GO were dispersed in 100 mL of water under ultrasound exfoliation for 2 h. Finally, 2 mg of TiO 2 NPs was added to 5 mL of GO supernatant solution (1 mg/mL) under ultrasounds for 2 h to obtain the TiO 2 -GO composite.

Fabrication of TiO 2 -ErGO-Modified GCE
Firstly, the polished GCE was immersed in ethyl alcohol and water under ultrasounds for 1 min. Then the TiO 2-GO-GCEs were fabricated via drop casting of the TiO 2 -GO dispersion on the GCE, followed by an electrochemical reduction process. Specifically, 5 µL of TiO 2 -GO dispersion was dropped and casted on the surface of bare GCE to prepare the TiO 2 -GO-GCE. Then, the TiO 2 -GO-GCE was reduced by electrochemical reduction under the potential of −1.2 V for 120 s for the formation of TiO 2 -ErGO-GCE. Reduced graphene oxide-modified GCEs (ErGO-GCE) were also prepared for comparison.

Electrochemical Experiments
Both cyclic voltammetry (CV) and second-order derivative linear sweep voltammetry (SDLSV) were carried out with a standard three-electrode system. The bare or nanomaterial-modified GCEs were used as working electrodes. A platinum electrode and a saturated calomel electrode (SCE) acted as counter electrode and reference electrode, respectively. The CVs were measured by CHI 660E electrochemical workstation (Chenhua Instrument Co. Ltd., Shanghai, China), and the SDLSV was tested by a JP-303E Polarographic Analyzer (Chengdu Instrument Company, Chengdu, China). Fresh PBS, 0.1 M, was used as a supporting electrolyte for all electrochemical tests. Unless stated otherwise, all electrochemical tests were recorded at a scan rate of 100 mV/s, after a suitable accumulation period under stirring at 500 rpm and a 5 s rest. The potential scan ranges were 0.4-1.2 V for CV and 0.6-1.2 V for the SDLSV.

Analysis of Real Samples
The carbonate beverage was purchased from a local supermarket. The CO 2 was eliminated by an ultrasound process. An amount of 1 mL of sample was diluted to 6 mL with 1.0 M PBS (pH 3.7). Then, carbonate beverage samples at various concentration were prepared by dilution. The content of tartrazine in the carbonate beverage was measured using SDLSV by the standard addition method under the optimal detection conditions.

Morphologic and Structural Characterization of TiO 2 -GO Nanocomposites
The morphologies of the TiO 2 and TiO 2 -GO composite samples were characterized by TEM. As shown in the TEM images ( Figure 1A), the TiO 2 NPs was cube-like with an average particle size of 50 nm. The TiO 2 NPs aggregated with each other, and their dispersibility could be improved. Moreover, GO sheets were obviously observed in the surrounding of the TiO 2 NPs ( Figure 1B), indicating that the TiO 2 NPs were well combined with the GO nanosheets. The XRD pattern of the TiO 2 NPs is also presented ( Figure 1C

Electrochemical Behavior of Tartrazine on Different Electrodes
The cyclic voltammograms of bare GCE, GO-GCE, ErGO-GCE, andTiO2-ErGO-GCE in 2.5 × 10 −3 mol/L [Fe(CN)6] 3−/4− solution were investigated, and the results are shown in Figure 2. As expected, a pair of reversible redox peaks appeared on all electrodes. However, the intensities of the redox peaks increased in the following order, GO-GCE, bare GCE, ErGO-GCE, and TiO2-ErGO-GCE. The redox peak current on the GO-GCE was the smallest because of the poor electrical conductivity. When GO was reduced to ErGO, the redox peak current enhanced greatly because of the restoration of the conductive carbon conjugate network and large surface area. When ErGO was combined with TiO2 NPs, the redox peak current further increased because of a synergistic enhancement between ErGO and TiO2 NPs. The reduction peak currents of bare GO, GCE, RGO-GCE, and TiO2-RGO-GCE were 3.37 × 10 −5 A, 4.32 × 10 −5 A, 8.49 × 10 −5 A, and 1.17 × 10 −4 A, respectively. According to Randles-Sevcik equation, their electroactive area were estimated as 0.058 cm 2 , 0.074 cm 2 , 0.145 cm 2 , and 0.199 cm 2 , respectively. The electrochemical active area of bare GCE coincided with the geometric area (Φ 3.0 mm, 0.071 cm 2 ), and the electrochemical active area of ErGO-CCE and TiO2-ErGO-GCE were approximately 2.0 and 2.7 times that on the bare GCE, in relation to the large specific surface area of TiO2 NPs and ErGO. The large electrochemical active area of the TiO2-ErGO nanocomposites will enhance the adsorption capacity of tartrazine and offer more catalytic sites for tartrazine oxidation. The electrochemical behavior of tartrazine on (1.0 × 10 −5 mol/L) the surface of the bare and modified GCEs were investigated by SDLSV. The results are shown in Figure 3A. A wide and short peak appeared on the surface of the GCE at 1000 mV (curve a), and the peak current was 1.594 μA. The peak current of tartrazine on the surface of GO-GCE was 1.006 μA (curve b); the lower current

Electrochemical Behavior of Tartrazine on Different Electrodes
The cyclic voltammograms of bare GCE, GO-GCE, ErGO-GCE, andTiO 2 -ErGO-GCE in 2.5 × 10 −3 mol/L [Fe(CN) 6 ] 3−/4− solution were investigated, and the results are shown in Figure 2. As expected, a pair of reversible redox peaks appeared on all electrodes. However, the intensities of the redox peaks increased in the following order, GO-GCE, bare GCE, ErGO-GCE, and TiO 2 -ErGO-GCE. The redox peak current on the GO-GCE was the smallest because of the poor electrical conductivity. When GO was reduced to ErGO, the redox peak current enhanced greatly because of the restoration of the conductive carbon conjugate network and large surface area. When ErGO was combined with TiO 2 NPs, the redox peak current further increased because of a synergistic enhancement between ErGO and TiO 2 NPs. The reduction peak currents of bare GO, GCE, RGO-GCE, and TiO 2 -RGO-GCE were 3.37 × 10 −5 A, 4.32 × 10 −5 A, 8.49 × 10 −5 A, and 1.17 × 10 −4 A, respectively. According to Randles-Sevcik equation, their electroactive area were estimated as 0.058 cm 2 , 0.074 cm 2 , 0.145 cm 2 , and 0.199 cm 2 , respectively. The electrochemical active area of bare GCE coincided with the geometric area (Φ 3.0 mm, 0.071 cm 2 ), and the electrochemical active area of ErGO-CCE and TiO 2 -ErGO-GCE were approximately 2.0 and 2.7 times that on the bare GCE, in relation to the large specific surface area of TiO 2 NPs and ErGO. The large electrochemical active area of the TiO 2 -ErGO nanocomposites will enhance the adsorption capacity of tartrazine and offer more catalytic sites for tartrazine oxidation.

Electrochemical Behavior of Tartrazine on Different Electrodes
The cyclic voltammograms of bare GCE, GO-GCE, ErGO-GCE, andTiO2-ErGO-GCE in 2.5 × 10 −3 mol/L [Fe(CN)6] 3−/4− solution were investigated, and the results are shown in Figure 2. As expected, a pair of reversible redox peaks appeared on all electrodes. However, the intensities of the redox peaks increased in the following order, GO-GCE, bare GCE, ErGO-GCE, and TiO2-ErGO-GCE. The redox peak current on the GO-GCE was the smallest because of the poor electrical conductivity. When GO was reduced to ErGO, the redox peak current enhanced greatly because of the restoration of the conductive carbon conjugate network and large surface area. When ErGO was combined with TiO2 NPs, the redox peak current further increased because of a synergistic enhancement between ErGO and TiO2 NPs. The reduction peak currents of bare GO, GCE, RGO-GCE, and TiO2-RGO-GCE were 3.37 × 10 −5 A, 4.32 × 10 −5 A, 8.49 × 10 −5 A, and 1.17 × 10 −4 A, respectively. According to Randles-Sevcik equation, their electroactive area were estimated as 0.058 cm 2 , 0.074 cm 2 , 0.145 cm 2 , and 0.199 cm 2 , respectively. The electrochemical active area of bare GCE coincided with the geometric area (Φ 3.0 mm, 0.071 cm 2 ), and the electrochemical active area of ErGO-CCE and TiO2-ErGO-GCE were approximately 2.0 and 2.7 times that on the bare GCE, in relation to the large specific surface area of TiO2 NPs and ErGO. The large electrochemical active area of the TiO2-ErGO nanocomposites will enhance the adsorption capacity of tartrazine and offer more catalytic sites for tartrazine oxidation. The electrochemical behavior of tartrazine on (1.0 × 10 −5 mol/L) the surface of the bare and modified GCEs were investigated by SDLSV. The results are shown in Figure 3A. A wide and short peak appeared on the surface of the GCE at 1000 mV (curve a), and the peak current was 1.594 μA. The peak current of tartrazine on the surface of GO-GCE was 1.006 μA (curve b); the lower current The electrochemical behavior of tartrazine on (1.0 × 10 −5 mol/L) the surface of the bare and modified GCEs were investigated by SDLSV. The results are shown in Figure 3A. A wide and short peak appeared on the surface of the GCE at 1000 mV (curve a), and the peak current was 1.594 µA. The peak current of tartrazine on the surface of GO-GCE was 1.006 µA (curve b); the lower current could be attributed to the inferior electrical conductivity of GO. Moreover, a wide and short peak (1004 mV, 1.987 µA) were detected on the surface of TiO 2 -GO-GCE, probably because of the catalytic properties of TiO 2 with mesoporous structure and the poor electrical conductivity of GO. However, a manifest oxidation peak located at 1036 mV was observed on the ErGO-GCE, and the peak current increased significantly to 20.10 µA (curve d). This phenomenon could be ascribed to the higher electrical conductivity of ErGO due to the restoration of the conductive carbon-conjugated structure. Moreover, the large surface area could promote the adsorption of tartrazine onto the electrodes. More importantly, the peak current further increases to 26.98 µA when TiO 2 -ErGO-GCE acted as the work electrode (curve e). The peak current on the TiO 2 -ErGO-GCE was 18 times higher than that on the bare GCE, because of the synergistic effects of TiO 2 and ErGO that enhanced the electrochemical oxidation of tartrazine. could be attributed to the inferior electrical conductivity of GO. Moreover, a wide and short peak (1004 mV, 1.987 μA) were detected on the surface of TiO2-GO-GCE, probably because of the catalytic properties of TiO2 with mesoporous structure and the poor electrical conductivity of GO. However, a manifest oxidation peak located at 1036 mV was observed on the ErGO-GCE, and the peak current increased significantly to 20.10 μA (curve d). This phenomenon could be ascribed to the higher electrical conductivity of ErGO due to the restoration of the conductive carbon-conjugated structure. Moreover, the large surface area could promote the adsorption of tartrazine onto the electrodes. More importantly, the peak current further increases to 26.98 μA when TiO2-ErGO-GCE acted as the work electrode (curve e). The peak current on the TiO2-ErGO-GCE was 18 times higher than that on the bare GCE, because of the synergistic effects of TiO2 and ErGO that enhanced the electrochemical oxidation of tartrazine. The CV curves of bare and modified GCEs recorded in 1.0 × 10 −5 mol/L of tartrazine solution are presented in Figure 3B. Only the oxidation peak can be observed on all the electrodes, suggesting that the electrochemical oxidation of tartrazine is an irreversible process. The order of peak current for the different electrodes was consistent with the results of SDSLV. As expected, the largest peak current was obtained on the TiO2-ErGO-GCE, further confirming that the synergistic effect of TiO2 and ErGO improved the electrochemical oxidation of tartrazine.

The Influence of pH
Since the pH is an important parameter influencing the electrochemical oxidation of tartrazine, it is important to evaluate the optimal pH value for tartrazine detection. As shown in Figure 4A, the largest current intensity (ipa) of tartrazine was observed when the pH was 3.7. The electro-oxidation of tartrazine was performed better in more acidic media, whereas in neutral to alkaline media, the anodic peak current was considerably decreased. Moreover, the oxidation peak potential Ep is linear to the pH in the pH range of 2.5-7.5 ( Figure 4B). The linear equation was Ep = −0.0560 pH + 1.024 (R 2 = 0.999), and the slope (−63 mV/pH) was very close to the theoretical value (−59 mV/pH), indicating that the same electron and proton number participate in the electrochemical oxidation process. The CV curves of bare and modified GCEs recorded in 1.0 × 10 −5 mol/L of tartrazine solution are presented in Figure 3B. Only the oxidation peak can be observed on all the electrodes, suggesting that the electrochemical oxidation of tartrazine is an irreversible process. The order of peak current for the different electrodes was consistent with the results of SDSLV. As expected, the largest peak current was obtained on the TiO 2 -ErGO-GCE, further confirming that the synergistic effect of TiO 2 and ErGO improved the electrochemical oxidation of tartrazine.

The Influence of pH
Since the pH is an important parameter influencing the electrochemical oxidation of tartrazine, it is important to evaluate the optimal pH value for tartrazine detection. As shown in Figure 4A, the largest current intensity (i pa ) of tartrazine was observed when the pH was 3.7. The electro-oxidation of tartrazine was performed better in more acidic media, whereas in neutral to alkaline media, the anodic peak current was considerably decreased. Moreover, the oxidation peak potential E p is linear to the pH in the pH range of 2.5-7.5 ( Figure 4B). The linear equation was E p = −0.0560 pH + 1.024 (R 2 = 0.999), and the slope (−63 mV/pH) was very close to the theoretical value (−59 mV/pH), indicating that the same electron and proton number participate in the electrochemical oxidation process.

Effect of Accumulation Conditions
Accumulation potential and time are other two important factors that influence the oxidation current of tartrazine. After accumulation for 180 s with different accumulation potentials (−0.3 to 0.4 V), the oxidation peak currents in 1 × 10 −5 mol/L tartrazine were measured. When the accumulation potential was −0.2 V, the largest oxidation peak current was obtained ( Figure 5A), indicating that −0.2 V was the best accumulation potential. In addition, the accumulation at −0.2 V for various times was also investigated. As plotted in Figure 5B, the oxidation peak currents increased with the accumulation time between 0 and 180 s. Afterward, the ipa remained stable because of the saturation adsorption of tartrazine on the surface of TiO2-ErGO-GCE. Thus, 180 s was the optimal accumulation time.

The Influence of the Scan Rate
The electrochemical response of tartrazine is strongly dependent on the scan rate, thus this parameter was also considered. The CVs were scanned at different scan rates (30~300 mV/s) in PBS (0.1 M, pH 3.7) solution containing 1 × 10 −5 mol/L of tartrazine, and the results are presented in Figure  6A. The oxidation peak current increased gradually with the increase of the scan rate. As shown in Figure 6B, a good linear relationship between oxidation peak currents (ipa) and scan rate (v) was obtained, and the corresponding linear equation was ipa = 58.89v + 16.59 (R 2 = 0.990). This result indicates that the electrochemical oxidation of tartrazine was an adsorption-controlled process. Thus, the accumulation method was adopted in the subsequent experiments in order to enhance the sensitivity. However, the background currents were also increased correspondingly. Considering the best signal to noise ratio (SNR) and the lowest background current, a suitable scan rate was found to be 100 mV/s.

Effect of Accumulation Conditions
Accumulation potential and time are other two important factors that influence the oxidation current of tartrazine. After accumulation for 180 s with different accumulation potentials (−0.3 to 0.4 V), the oxidation peak currents in 1 × 10 −5 mol/L tartrazine were measured. When the accumulation potential was −0.2 V, the largest oxidation peak current was obtained ( Figure 5A), indicating that −0.2 V was the best accumulation potential. In addition, the accumulation at −0.2 V for various times was also investigated. As plotted in Figure 5B, the oxidation peak currents increased with the accumulation time between 0 and 180 s. Afterward, the i pa remained stable because of the saturation adsorption of tartrazine on the surface of TiO 2 -ErGO-GCE. Thus, 180 s was the optimal accumulation time.

Effect of Accumulation Conditions
Accumulation potential and time are other two important factors that influence the oxidation current of tartrazine. After accumulation for 180 s with different accumulation potentials (−0.3 to 0.4 V), the oxidation peak currents in 1 × 10 −5 mol/L tartrazine were measured. When the accumulation potential was −0.2 V, the largest oxidation peak current was obtained ( Figure 5A), indicating that −0.2 V was the best accumulation potential. In addition, the accumulation at −0.2 V for various times was also investigated. As plotted in Figure 5B, the oxidation peak currents increased with the accumulation time between 0 and 180 s. Afterward, the ipa remained stable because of the saturation adsorption of tartrazine on the surface of TiO2-ErGO-GCE. Thus, 180 s was the optimal accumulation time.

The Influence of the Scan Rate
The electrochemical response of tartrazine is strongly dependent on the scan rate, thus this parameter was also considered. The CVs were scanned at different scan rates (30~300 mV/s) in PBS (0.1 M, pH 3.7) solution containing 1 × 10 −5 mol/L of tartrazine, and the results are presented in Figure  6A. The oxidation peak current increased gradually with the increase of the scan rate. As shown in Figure 6B, a good linear relationship between oxidation peak currents (ipa) and scan rate (v) was obtained, and the corresponding linear equation was ipa = 58.89v + 16.59 (R 2 = 0.990). This result indicates that the electrochemical oxidation of tartrazine was an adsorption-controlled process. Thus, the accumulation method was adopted in the subsequent experiments in order to enhance the sensitivity. However, the background currents were also increased correspondingly. Considering the best signal to noise ratio (SNR) and the lowest background current, a suitable scan rate was found to be 100 mV/s.

The Influence of the Scan Rate
The electrochemical response of tartrazine is strongly dependent on the scan rate, thus this parameter was also considered. The CVs were scanned at different scan rates (30~300 mV/s) in PBS (0.1 M, pH 3.7) solution containing 1 × 10 −5 mol/L of tartrazine, and the results are presented in Figure 6A. The oxidation peak current increased gradually with the increase of the scan rate. As shown in Figure 6B, a good linear relationship between oxidation peak currents (i pa ) and scan rate (v) was obtained, and the corresponding linear equation was i pa = 58.89v + 16.59 (R 2 = 0.990). This result indicates that the electrochemical oxidation of tartrazine was an adsorption-controlled process. Thus, the accumulation method was adopted in the subsequent experiments in order to Sensors 2018, 18, 1911 8 of 12 enhance the sensitivity. However, the background currents were also increased correspondingly. Considering the best signal to noise ratio (SNR) and the lowest background current, a suitable scan rate was found to be 100 mV/s. Furthermore, only a positive oxidation peak potential (Epa) shifts with the rising of the scan rates (v) was observed, meaning that the oxidation process of tartrazine was irreversible. The liner relationship between Epa and the Napierian logarithm of the scan rate (ln v) is also presented ( Figure  6C). The linear equation was Epa = 0.0172 ln v + 1.0924 (R 2 = 0.995). According to Lavrion equation [33], where E 0′ is the formal potential (V), α is the charge transfer coefficient, n is the electron transfer number, F is the Faraday constant (96,480 C·mol −1 ), R is the ideal gas constant (8.314 J·mol −1 ·K −1 ), T is the Kelvin temperature (K), D is the diffusion coefficient, and k 0 is the heterogeneous electron transfer rate. The value of α is always supposed to be 0.5 in an irreversible process, and the n value is calculated as 1. Thus, the oxidation of tartrazine is an irreversible process with one electron and one proton. This result is consistent with those of a previous report [34]. The electrochemical oxidation mechanism of tartrazine is summarized in Figure 7.

Calibration Curve and Detection Limit
Under the optimal detection conditions, the SDLSV response of tartrazine at various concentration (range of 2.0 × 10 −8 -2.0 × 10 −5 mol/L) was measured, and the results are presented in Figure 8A. With the increase of tartrazine concentration, the peak currents ipa were enhanced linearly. Moreover, the linear relationship between the peak currents ipa and the concentration of tartrazine was calculated as ipa (μA) = 3.450c (μmol/L) + 0.486 (R 2 = 0.991, with a standard error of slope: 0.104 and standard error of intercept: 0.0704) ( Figure 8B); the detection limit (S/N = 3) was calculated as 8.0 × 10 −9 mol/L. This results were comparable and even better than those reported in the literature [35,36]. Furthermore, only a positive oxidation peak potential (E pa ) shifts with the rising of the scan rates (v) was observed, meaning that the oxidation process of tartrazine was irreversible. The liner relationship between E pa and the Napierian logarithm of the scan rate (ln v) is also presented ( Figure 6C). The linear equation was E pa = 0.0172 ln v + 1.0924 (R 2 = 0.995). According to Lavrion equation [33], where E 0 is the formal potential (V), α is the charge transfer coefficient, n is the electron transfer number, F is the Faraday constant (96,480 C·mol −1 ), R is the ideal gas constant (8.314 J·mol −1 ·K −1 ), T is the Kelvin temperature (K), D is the diffusion coefficient, and k 0 is the heterogeneous electron transfer rate. The value of α is always supposed to be 0.5 in an irreversible process, and the n value is calculated as 1. Thus, the oxidation of tartrazine is an irreversible process with one electron and one proton. This result is consistent with those of a previous report [34]. The electrochemical oxidation mechanism of tartrazine is summarized in Figure 7. Furthermore, only a positive oxidation peak potential (Epa) shifts with the rising of the scan rates (v) was observed, meaning that the oxidation process of tartrazine was irreversible. The liner relationship between Epa and the Napierian logarithm of the scan rate (ln v) is also presented ( Figure  6C). The linear equation was Epa = 0.0172 ln v + 1.0924 (R 2 = 0.995). According to Lavrion equation [33], where E 0′ is the formal potential (V), α is the charge transfer coefficient, n is the electron transfer number, F is the Faraday constant (96,480 C·mol −1 ), R is the ideal gas constant (8.314 J·mol −1 ·K −1 ), T is the Kelvin temperature (K), D is the diffusion coefficient, and k 0 is the heterogeneous electron transfer rate. The value of α is always supposed to be 0.5 in an irreversible process, and the n value is calculated as 1. Thus, the oxidation of tartrazine is an irreversible process with one electron and one proton. This result is consistent with those of a previous report [34]. The electrochemical oxidation mechanism of tartrazine is summarized in Figure 7.

Calibration Curve and Detection Limit
Under the optimal detection conditions, the SDLSV response of tartrazine at various concentration (range of 2.0 × 10 −8 -2.0 × 10 −5 mol/L) was measured, and the results are presented in Figure 8A. With the increase of tartrazine concentration, the peak currents ipa were enhanced linearly. Moreover, the linear relationship between the peak currents ipa and the concentration of tartrazine was calculated as ipa (μA) = 3.450c (μmol/L) + 0.486 (R 2 = 0.991, with a standard error of slope: 0.104 and standard error of intercept: 0.0704) ( Figure 8B); the detection limit (S/N = 3) was calculated as 8.0 × 10 −9 mol/L. This results were comparable and even better than those reported in the literature [35,36].

Calibration Curve and Detection Limit
Under the optimal detection conditions, the SDLSV response of tartrazine at various concentration (range of 2.0 × 10 −8 -2.0 × 10 −5 mol/L) was measured, and the results are presented in Figure 8A. With the increase of tartrazine concentration, the peak currents i pa were enhanced linearly. Moreover, the linear relationship between the peak currents i pa and the concentration of tartrazine was calculated as i pa (µA) = 3.450c (µmol/L) + 0.486 (R 2 = 0.991, with a standard error of slope: 0.104 and standard error of intercept: 0.0704) ( Figure 8B); the detection limit (S/N = 3) was calculated as 8.0 × 10 −9 mol/L. This results were comparable and even better than those reported in the literature [35,36].

Interference Studies
The interference studies on species and other food colorants coexistent with tartrazine was also investigated. Different kinds of interfering species were added into 0.45 mol/L PBS (pH 3.7) containing 1.0 × 10 −5 mol/L of tartrazine, and the peak currents were tested and compared. The peak currents ipa of 10 μmol/L of tartrazine in the presence of interferents are listed in Table 1. Under an acceptable error range, 100-times higher (than the concentration of tartrazine) concentrations of glucose, benzoic acid, citric acid, Na + , K + , and Fe 3+ , and 10-times higher concentrations of sunset yellow and amaranth did not interfere with the detection of tartrazine. Moreover, the peak currents intensities of tartrazine under the influence of interferents were still similar to those of pure Ttrtrazine, meaning that no oxidation peaks of the interfering species appeared or that the oxidation peaks separated well from those of tartrazine. This indicates that the TiO2-ErGO-GCE showed superior anti-interference performance and presents great prospects for the detection of tartrazine in various real samples. It is noteworthy that the oxidation peaks of sunset yellow and amaranth did not overlay with those of tartrazine, in spite of their similar structure. Hence, the TiO2-ErGO-GCE shows great potential for the simultaneous detection of tartrazine, sunset yellow, and amaranth. This work will be carried out in the future.

Reproducibility of the Detection
Under the optimal detection conditions, tartrazine standard solutions (1.0 × 10 −5 mol/L) were detected for seven times by using the same TiO2-ErGO-GCE as the work electrode. After every test, the electrode was washed in 0.1 mol/L of nitric acid for two times under cyclic voltammetry. The reproducibility results are listed in Table 2. The relative standard deviation (RSD) was 0.45%, indicating that the TiO2-ErGO-GCE exhibited good reproducibility for tartrazine detection.

Interference Studies
The interference studies on species and other food colorants coexistent with tartrazine was also investigated. Different kinds of interfering species were added into 0.45 mol/L PBS (pH 3.7) containing 1.0 × 10 −5 mol/L of tartrazine, and the peak currents were tested and compared. The peak currents i pa of 10 µmol/L of tartrazine in the presence of interferents are listed in Table 1. Under an acceptable error range, 100-times higher (than the concentration of tartrazine) concentrations of glucose, benzoic acid, citric acid, Na + , K + , and Fe 3+ , and 10-times higher concentrations of sunset yellow and amaranth did not interfere with the detection of tartrazine. Moreover, the peak currents intensities of tartrazine under the influence of interferents were still similar to those of pure Ttrtrazine, meaning that no oxidation peaks of the interfering species appeared or that the oxidation peaks separated well from those of tartrazine. This indicates that the TiO 2 -ErGO-GCE showed superior anti-interference performance and presents great prospects for the detection of tartrazine in various real samples. It is noteworthy that the oxidation peaks of sunset yellow and amaranth did not overlay with those of tartrazine, in spite of their similar structure. Hence, the TiO 2 -ErGO-GCE shows great potential for the simultaneous detection of tartrazine, sunset yellow, and amaranth. This work will be carried out in the future.

Reproducibility of the Detection
Under the optimal detection conditions, tartrazine standard solutions (1.0 × 10 −5 mol/L) were detected for seven times by using the same TiO 2 -ErGO-GCE as the work electrode. After every test, the electrode was washed in 0.1 mol/L of nitric acid for two times under cyclic voltammetry.
The reproducibility results are listed in Table 2. The relative standard deviation (RSD) was 0.45%, indicating that the TiO 2 -ErGO-GCE exhibited good reproducibility for tartrazine detection.

Real Sample Detection
SDLSV shows high resolution and sensitivity in electrochemical analyses, thus it is widely applied in food additives detection. In this section, this method was used to analyze carbonated beverage samples. These carbonated beverage samples at various concentration were measured by using SDLSV under the optimal conditions. A shown inn Table 3, the concentration of tartrazine detected in samples 1-3 were 0.2, 2.24, and 4.26, respectively, which were well consistent with the standard values. Moreover, the corresponding RSD was 1.05-2.32%, and the recovery rate of the samples 2 and 3 was 112.0% and 106.5%, respectively. These results indicate that the TiO 2 -ErGO-GCE could be an efficient system for tartrazine detection in carbonate beverage samples. Table 3. The results of the determination of tartrazine in a soft drink at different adding concentrations (n = 3).

Conclusions
In summary, a TiO 2 -ErGO-GCE was successfully fabricated by hydrothermal and electrochemical reduction and it was used for practical tartrazine detection. After electrochemical reduction, the TiO 2 NPs were coated by ErGO. The oxidation peak current on the TiO 2 -ErGO-GCE increased to 26.98 µA, which was 18 times higher than that on the bare GCE. Moreover, the electrochemical results revealed that the electrochemical oxidation of tartrazine is an adsorption-controlled process with one electron and one proton. A wide linear range (from 2 × 10 −8 mol/L to 2 × 10 −5 mol/L) and a low detection limit (S/N = 3) of 8.0 × 10 −9 mol/L were also obtained with the TiO 2 -ErGO-GCE. Finally, in practice, the TiO 2 -ErGO-GCE also showed a good detection sensitivity in the detection of tartrazine in a carbonated beverage. This detection system shows great application prospects for the sensitive detection of food additives in real samples, due to its prominent advantages including facile fabrication, rapid response, good selectivity, low detection limit, and wide linear range of detection.
Author Contributions: Q.H., P.D., and D.C. conceived and designed the experiments; J.L., X.L., and L.J. performed the experiments; G.L. and X.L. analyzed the data; Q.H. and D.C. contributed reagents/materials/analysis tools; J.L. and G.L. wrote the paper.