Facile Fabrication of CeO2/Electrochemically Reduced Graphene Oxide Nanocomposites for Vanillin Detection in Commercial Food Products

In this paper, CeO2 nanoparticles were synthesized by the solvothermal method and dispersed uniformly in graphene oxide (GO) aqueous solution by ultrasonication. The homogeneous CeO2-GO dispersion was coated on the surface of a glassy carbon electrode (GCE), and the CeO2/electrochemically reduced graphene oxide modified electrode (CeO2/ERGO/GCE) was obtained by potentiostatic reduction. The results of X-ray diffraction (XRD), energy dispersive X-ray spectroscopy (EDS), scanning electron microscopy (SEM), and transmission electron microscopy (TEM) showed that CeO2 nanocrystals were uniformly coated by gossamer like ERGO nanosheets. The electrochemical behavior of vanillin on the CeO2/ERGO/GCE was studied by cyclic voltammetry (CV). It was found that the CeO2/ERGO/GCE has high electrocatalytic activity and good electrochemical performance for vanillin oxidation. Using the second derivative linear sweep voltammetry (SDLSV), the CeO2/ERGO/GCE provides a wide range of 0.04–20 µM and 20 µM–100 µM for vanillin detection, and the detection limit is estimated to be 0.01 µM after 120 s accumulation. This method has been successfully applied to the vanillin detection in some commercial foods.


Introduction
The problem of food safety has caused public concern all over the world. Unsafe food can lead to many acute and lifelong diseases, from diarrhea to various cancers [1]. Vanillin (4-hydroxy -3-methoxybenzaldehyde), a food additive, is widely used to contribute to the fragrance of various foods, such as ice-cream, cookies, pudding, beverages, custard, and chocolate [2]. However, the production cost of natural vanillin from vanilla pods is very high. Low-cost materials, such as eugenol, 2-methoxyphenol, and lignin, can also be used to synthesize vanillin. Although the synthetic version is cheaper and widely produced, it can lead to headaches, nausea, vomiting, and may affect the functions of the kidneys and liver when large quantities of vanillin are ingested. Therefore, for the sake of human health, the content of vanillin in food should be strictly controlled.
In order to determine vanillin sensitively and effectively, gas chromatography [3], high performance liquid chromatography [4], thin layer chromatography [5], ultraviolet visible spectrophotometry [6], chemiluminometry [7], and capillary electrophoresis [8] were introduced. However, most of these methods need large and expensive instruments, and the operation is complex and time-consuming. The electrochemical method has been one of the research hotspots due to its advantages of easy

Chemicals
Cerium nitrate hexahydrate was purchased from Chengdu Aikeda Chemical Reagent Co., Ltd. (Chengdu, China). Vanillin, graphite powder, potassium permanganate, sodium nitrate, ethylene glycol, hydrogen peroxide solution (30 wt %), ammonia solution (25 wt %), and hydrazine solution (80 wt %) were obtained from Sinopharm Group Chemical Reagent Co., Ltd. (Shanghai, China). The accurately weighed vanillin was dissolved in a proper amount of ethanol, diluted with water to obtain a 1.0 mM stock solution. Other chemicals used in the experiment are of analytical reagent grade. The water used is ultrapure water.

Apparatus
Electrochemical experiments (cyclic voltammetry and second-order derivative linear sweep voltammetry) were carried out on a chi660e electrochemical workstation (Shanghai Chenhua Instrument Co., Ltd., Shanghai, China) and JP-303E polarographic analyzer (Chengdu Instrument Co., Ltd., Chengdu, China) respectively. A three electrode system was used, i.e., modified GCE with diameter of 3 mm was used as working electrode, and asaturated calomel electrode (SCE) and platinum wire electrode were used as reference electrode and auxiliary electrode respectively. The scanning electron microscope (SEM) images were obtained using a scanning electron microscope (EVO10, ZEISS, Jena, Germany). Transmission electron microscope (TEM) images and energy dispersive X-ray spectroscopy (EDS) were obtained by a transmission electron microscope (JEOL JEM-2100, Tokyo, Japan). The X-ray diffraction data (XRD) were collected on a powder X-ray diffractometer (Rigaku, Tokyo, Japan) (Cu Kα radiation λ = 0.154056 nm). All measurements were performed at room temperature.

Synthesis of CeO 2 Nanoparticles
CeO 2 nanoparticles were synthesized via a hydrothermal method. In brief, 1 mL of 0.5 M cerium nitrate aqueous solution was added to 30 mL of ethylene glycol, followed by the addition of 1 mL of ultrapure water, the mixed solution was stirred for 30 min at room temperature and transferred to a 50 mL autoclave for solvothermal reaction at 140 • C for 18 h. After the reaction, the mixture was cooled down to room temperature. The product was centrifuged, washed with water and ethanol, and dried in an oven at 50 • C.

Preparation of GO and CeO 2 -GO Composite
Graphene oxide (GO) was prepared using a modified Hummers method, which involves the steps of graphite oxidation and subsequent exfoliation [42]. A homogeneous CeO 2 -GO dispersion was obtained by mixing 1.0 mg of the CeO 2 nanoparticles and 1.0 mL GO solution (1 mg mL −1 ) directly by ultrasonication for 20 min.

Fabrication of Modified Electrodes
Before modification, a GCE was polished on a polishing cloth with 0.3 µm alumina powder, and then ultrasonic treatment was carried out in ethanol and ultrapure water successively. A treated GCE was modified with 5.0 µL of CeO 2 -GO dispersion and the solvent on the electrode surface was evaporated under an infrared lamp. Then the electrode was put into a 0.1 M phosphate buffer (pH 6.5) and reduced at the constant potential of −1.2 V for 120 s to obtain the CeO 2 /ERGO/GCE. For comparison, GO/GCE, CeO 2 /GCE and ERGO/GCE were also fabricated in a similar way. The brief fabrication process of the CeO 2 /ERGO/GCE and its use in vanillin sensing is shown in Scheme 1.

Electrochemical Measurement
For electrochemical measurements, 10 mL 0.10 M HCl solution containing a certain concentration of vanillin was added to a 10 mL cell, and a three-electrode system was installed in the test solution. After accumulation at 0.1 V for 120 s, the cyclic voltammograms was recorded from 0.1 V to 1.2 V and the second-order derivative linear sweep voltammograms were recorded from 0.6 V to 1.2 V. After every measurement, the CeO2/ERGO/GCE was regenerated by two successive voltammetric sweeps in 0.6-1.2 V in 0.1 M H2SO4 solution. Sample analysis was carried out under the best conditions.

Morphological and Structural Characterizations
The surface morphology of GO nanosheets, CeO2 nanoparticles, and CeO2/ERGO nanocomposites were studied by SEM. As shown in Figure 1A, the GO layer produced a rough, wrinkled and folded surface. The typical morphology of CeO2 nanoparticles is shown in Figure 1B. Most of CeO2 nanoparticles were spherical with a particle size distribution of about 20-200 nm. On the contrary, when CeO2 and ERGO were mixed together ( Figure 1C), it was observed that CeO2 nanoparticles were wrapped inside by gossamer like ERGO, indicating that CeO2 and ERGO are well combined. The combination of CeO2 and ERGO provides a rich conductive channel for the electron transfer of Trp on the electrode surface. The XRD pattern of CeO2 is shown in Figure 1D

Electrochemical Measurement
For electrochemical measurements, 10 mL 0.10 M HCl solution containing a certain concentration of vanillin was added to a 10 mL cell, and a three-electrode system was installed in the test solution. After accumulation at 0.1 V for 120 s, the cyclic voltammograms was recorded from 0.1 V to 1.2 V and the second-order derivative linear sweep voltammograms were recorded from 0.6 V to 1.2 V. After every measurement, the CeO 2 /ERGO/GCE was regenerated by two successive voltammetric sweeps in 0.6-1.2 V in 0.1 M H 2 SO 4 solution. Sample analysis was carried out under the best conditions.

Morphological and Structural Characterizations
The surface morphology of GO nanosheets, CeO 2 nanoparticles, and CeO 2 /ERGO nanocomposites were studied by SEM. As shown in Figure 1A, the GO layer produced a rough, wrinkled and folded surface. The typical morphology of CeO 2 nanoparticles is shown in Figure 1B. Most of CeO 2 nanoparticles were spherical with a particle size distribution of about 20-200 nm. On the contrary, when CeO 2 and ERGO were mixed together ( Figure 1C), it was observed that CeO 2 nanoparticles were wrapped inside by gossamer like ERGO, indicating that CeO 2 and ERGO are well combined. The combination of CeO 2 and ERGO provides a rich conductive channel for the electron transfer of Trp on the electrode surface. The XRD pattern of CeO 2 is shown in Figure 1D. The diffraction peaks located at 2θ = 28.  [37].
The morphology of the as-prepared CeO 2 /ERGO nanocomposite was further examined by TEM. Figure 2A showed the TEM micrograph of the composite. It can be seen that most CeO 2 nanoparticles were well wrapped in the ERGO sheets, and no free CeO 2 nanoparticles were observed outside of the ERGO sheets, indicating a good interfacial interaction between the CeO 2 nanoparticles and the ERGO sheets. The element composition and distribution of CeO 2 -ERGO nanocomposite were analyzed by EDS. As shown in Figure 2B-D, C, O, and Ce elements were observed in the element mapping images of the CeO 2 /ERGO composite. It was worth noting that the three elements were uniformly distributed throughout the CeO 2 /ERGO composite, indicating that the electrochemical reduction method is efficient to synthesize the CeO 2 /ERGO nanocomposite. The morphology of the as-prepared CeO2/ERGO nanocomposite was further examined by TEM. Figure 2A showed the TEM micrograph of the composite. It can be seen that most CeO2 nanoparticles were well wrapped in the ERGO sheets, and no free CeO2 nanoparticles were observed outside of the ERGO sheets, indicating a good interfacial interaction between the CeO2 nanoparticles and the ERGO sheets. The element composition and distribution of CeO2-ERGO nanocomposite were analyzed by EDS. As shown in Figure 2B-D, C, O, and Ce elements were observed in the element mapping images of the CeO2/ERGO composite. It was worth noting that the three elements were uniformly distributed throughout the CeO2/ERGO composite, indicating that the electrochemical reduction method is efficient to synthesize the CeO2/ERGO nanocomposite.

Electrochemical Characterization of the Modified Electrodes
] as a probe, The properties of the surface of different electrodes were characterized by CV. Figure 3 exhibited the CV results obtained on bare GCE (a), GO/GCE (b), CeO2/GCE (c), and CeO2/ERGO/GCE (d) in the potential range of −0.2~0.6 V. On bare GCE a pair of well-defined redox peaks was observed with the peak-to-peak potential separation (ΔEp) of 85 mV and the redox peak current of 14.80 µA (Ipa) and 15.87µA (Ipc) (curve a). It was shown that the electron transfer process is quasi-reversible. However, the redox peak current decreased obviously on  6 ] as a probe, The properties of the surface of different electrodes were characterized by CV. Figure 3 exhibited the CV results obtained on bare GCE (a), GO/GCE (b), CeO 2 /GCE (c), and CeO 2 /ERGO/GCE (d) in the potential range of −0.2~0.6 V. On bare GCE a pair of well-defined redox peaks was observed with the peak-to-peak potential separation (∆E p ) of 85 mV and the redox peak current of 14.80 µA (I pa ) and 15.87µA (I pc ) (curve a). It was shown that the electron transfer process is quasi-reversible. However, the redox peak current decreased obviously on GO/GCE (I pa =3.994 µA, I pc =4.016 µA), ∆E p was 167 mV (curve b). This may be due to the poor conductivity of GO and the strong negative charge repulsion force between [Fe(CN) 6 ] 3−/4− and the ionized groups such as COO − in GO. On the other hand, the current response of [Fe(CN) 6 ] 3−/4− on the CeO 2 /GCE (curve c) increased slightly compared with that of bare GCE. On CeO 2 /ERGO/GCE (curve d), the ∆E p value decreased to 78 mV and the highest redox peak current was observed, suggesting a more reversible electron transfer process of [Fe(CN) 6 ] 3−/4− occurred on CeO 2 /ERGO/GCE. Therefore, the electron transfer rate was greatly improved due to the coexistence of ERGO and CeO 2 nanoparticles.

Electrochemical Oxidation of Vanillin at Different Electrodes
The catalytic properties of CeO2/ERGO/GCE for vanillin oxidation are confirmed in Figure 4, which showed the CV responses of 10 µM vanillin recorded at different electrodes in 0.1 M HCl solution. On the CeO2/ERGO/GCE, a well-defined and sensitive vanillin oxidation peak (P1) appeared at 0.942 V [12,13]. In addition to the oxidation peak at 0.942 V, a redox couple P2/P3 (Epa = 0.688 V, Epc = 0.630 V) was also observed, corresponding to the redox reaction of the oxidized intermediate of vanillin. The electrochemical reaction mechanism was illustrated in Scheme 2 [12,13]. Table 1 compared the electrochemical data of vanillin obtained on different electrodes. It can be clearly observed that an obvious oxidation peak (Ip = 25.24 µA) appeared at 0.949 V at the ERGO/GCE compared with other electrodes, which may be related to the good catalytic activity, high conductivity, and large surface area of ERGO. The maximum oxidation peak current was obtained at CeO2/ERGO/GCE, which is about two times higher than that of ERGO/GCE, demonstrating the synergistic effect of CeO2 nanoparticles and ERGO nanosheets were existed on the electrode surface. As shown in Table 1, the order of the ability to enhance the oxidation signals of vanillin is CeO2/ERGO/GCE > ERGO/GCE > CeO2/GO/GCE > GO/GCE > CeO2/GCE > bare GCE. In particular, ERGO nanosheets have excellent conductivity and high surface area. Also the electron exchange between CeO2 nanoparticles and vanillin promoted the electrocatalytic reaction. In addition, ERGO is a good supporter for CeO2 nanoparticles, which can effectively prevent the agglomeration of CeO2 nanoparticles and give full play to its catalytic property. Therefore, the interface conductivity of the modified electrode was greatly improved and the sensitivity of the determination of vanillin was improved.

Electrochemical Oxidation of Vanillin at Different Electrodes
The catalytic properties of CeO 2 /ERGO/GCE for vanillin oxidation are confirmed in Figure 4, which showed the CV responses of 10 µM vanillin recorded at different electrodes in 0.1 M HCl solution. On the CeO 2 /ERGO/GCE, a well-defined and sensitive vanillin oxidation peak (P 1 ) appeared at 0.942 V [12,13]. In addition to the oxidation peak at 0.942 V, a redox couple P 2 /P 3 (E pa = 0.688 V, E pc = 0.630 V) was also observed, corresponding to the redox reaction of the oxidized intermediate of vanillin. The electrochemical reaction mechanism was illustrated in Scheme 2 [12,13]. Table 1 compared the electrochemical data of vanillin obtained on different electrodes. It can be clearly observed that an obvious oxidation peak (I p = 25.24 µA) appeared at 0.949 V at the ERGO/GCE compared with other electrodes, which may be related to the good catalytic activity, high conductivity, and large surface area of ERGO. The maximum oxidation peak current was obtained at CeO 2 /ERGO/GCE, which is about two times higher than that of ERGO/GCE, demonstrating the synergistic effect of CeO 2 nanoparticles and ERGO nanosheets were existed on the electrode surface. As shown in Table 1, the order of the ability to enhance the oxidation signals of vanillin is CeO 2 /ERGO/GCE > ERGO/GCE > CeO 2 /GO/GCE > GO/GCE > CeO 2 /GCE > bare GCE. In particular, ERGO nanosheets have excellent conductivity and high surface area. Also the electron exchange between CeO 2 nanoparticles and vanillin promoted the electrocatalytic reaction. In addition, ERGO is a good supporter for CeO 2 nanoparticles, which can effectively prevent the agglomeration of CeO 2 nanoparticles and give full play to its catalytic property. Therefore, the interface conductivity of the modified electrode was greatly improved and the sensitivity of the determination of vanillin was improved.

Effect of Scan Rate
In order to further investigate the oxidation mechanism, the CVs of 10 µM vanillin recorded on the CeO2/ERGO/GCE under different scan rates were illustrated in Figure 5A. Obviously, the

Effect of Scan Rate
In order to further investigate the oxidation mechanism, the CVs of 10 µM vanillin recorded on the CeO2/ERGO/GCE under different scan rates were illustrated in Figure 5A. Obviously, the oxidation peak increased gradually with the increase of scan rate in the range of 0.03-0.3 V s −1 . A good Scheme 2. Oxidation mechanism of vanillin on the CeO 2 /ERGO/GCE.

Effect of Scan Rate
In order to further investigate the oxidation mechanism, the CVs of 10 µM vanillin recorded on the CeO 2 /ERGO/GCE under different scan rates were illustrated in Figure 5A. Obviously, the oxidation peak increased gradually with the increase of scan rate in the range of 0.03-0.3 V s −1 . A good linear relationship between the peak current (I p ) and scan rate (ν) was observed in Figure 5B. The corresponding regression equation was I p (µA) = 0.1929 v (V s −1 ) − 0.4991 (R = 0.9959), It shows that the oxidation process of vanillin on the CeO 2 /ERGO/GCE is controlled by adsorption. By plotting the relationship between logarithm I p and logarithm v, the adsorption control behavior of vanillin was further confirmed: log I p (µA) = 1.0371logv (V s −1 ) + 2.3049 (R = 0.9987). The obtained slope of 1.0371 is close to 1.0. In Figure 5C, it is worth noting that the E p of vanillin moved to a positive value with the increase of scan rate, and E p changed linearly versus Napierian logarithm of scan rate (ln v). The linear regression equation was E p (V) = 0.0214 ln v (V s −1 ) + 0.9966 (R = 0.9981). For a completely irreversible adsorption control process, E p can be expressed by the Laviron equation [43]: Nanomaterials 2020, 10, x 9 of 17 linear relationship between the peak current (Ip) and scan rate (ν) was observed in Figure 5B. The corresponding regression equation was Ip (µA) = 0.1929 v (V s −1 ) − 0.4991 (R = 0.9959), It shows that the oxidation process of vanillin on the CeO2/ERGO/GCE is controlled by adsorption. By plotting the relationship between logarithm Ip and logarithm v, the adsorption control behavior of vanillin was further confirmed: log Ip (µA) = 1.0371logv (V s −1 ) + 2.3049 (R = 0.9987). The obtained slope of 1.0371 is close to 1.0. In Figure 5C, it is worth noting that the Ep of vanillin moved to a positive value with the increase of scan rate, and Ep changed linearly versus Napierian logarithm of scan rate (ln v). The linear regression equation was Ep (V) = 0.0214 ln v (V s −1 ) + 0.9966 (R = 0.9981). For a completely irreversible adsorption control process, Ep can be expressed by the Laviron equation [43]: According to equation (1), the slope of the straight line is equal to RT/αnF, so αn was calculated as 1.196. α is approximately 0.5 in a completely irreversible electrode process, thus n is about 2, which is consistent with the results obtained on silver nanoplate/GR composite modified GCE [12] and graphene-polyvinylpyrrolidone modified acetylene black paste electrode [13].

Chronocoulometric Studies
Using chronocoulometry, the diffusion coefficient D and Faradic charge Qads of vanillin at the CeO2/ERGO/GCE were calculated. Figure 6A showed the plot of Q-t obtained in the 0.1 M HCl solution with and without 0.1 mM vanillin. After background subtraction and the plot of charge (Q) against the square root of time (t 1/2 ) ( Figure 6B), a good linear relationship was observed with a slope of 1.729 × 10 −5 C s −1/2 and Qads of 1.679 × 10 −5 C. Using Equation (2) given by Anson [44], D can be obtained: According to equation (1), the slope of the straight line is equal to RT/αnF, so αn was calculated as 1.196. α is approximately 0.5 in a completely irreversible electrode process, thus n is about 2, which is consistent with the results obtained on silver nanoplate/GR composite modified GCE [12] and graphene-polyvinylpyrrolidone modified acetylene black paste electrode [13].

Chronocoulometric Studies
Using chronocoulometry, the diffusion coefficient D and Faradic charge Q ads of vanillin at the CeO 2 /ERGO/GCE were calculated. Figure 6A showed the plot of Q-t obtained in the 0.1 M HCl solution with and without 0.1 mM vanillin. After background subtraction and the plot of charge (Q) against the square root of time (t 1/2 ) ( Figure 6B), a good linear relationship was observed with a slope of 1.729 × 10 −5 C s −1/2 and Q ads of 1.679 × 10 −5 C. Using Equation (2) given by Anson [44], D can be obtained: where Q dl is double layer charge, A is the electrode surface area, and c is the substrate concentration. According to Figure 3, using Randles-Sevcik Equation (3) [45], A was estimated to be 0.01478 cm 2 .
where Qdl is double layer charge, A is the electrode surface area, and c is the substrate concentration. According to Figure 3, using Randles-Sevcik Equation (3) [45], A was estimated to be 0.01478 cm 2 .
For the completely irreversible oxidation of vanillin at the CeO2/ERGO/GCE, the standard heterogeneous rate constant (ks) can be obtained based on Equation (4) [46,47]: In the formula, Ep represents the peak potential, Ep/2 is the potential at which I = Ip/2. Other symbols have their usual meanings. In our experiment, Ep − Ep/2 =34 mV, D = 2.9 × 10 −5 cm 2 s −1 , v = 100 mV s −1 , and T = 298 K. The value of ks was obtained as 1.02 × 10 −2 cm s −1 , confirming a relative rapid electrode reaction process.

Optimal of Some Determination Conditions
The electrochemical response of vanillin on the CeO2/ERGO/GCE was investigated in different types of supporting electrolytes, such as HAc-NH4Ac buffer solution (pH 4.0), phosphate buffer solution (pH 3.0 and 6.0), HAc-NaAc buffer solution (pH 4.0), HCl, H2SO4, HNO3, and H3PO4 (each Thus, D of 2.9 × 10 −5 cm 2 s −1 was obtained. According to the equation Q ads = nFA Γ s , the adsorption capacity Γ s was obtained as 5.88 × 10 −10 mol cm −2 . For the completely irreversible oxidation of vanillin at the CeO 2 /ERGO/GCE, the standard heterogeneous rate constant (k s ) can be obtained based on Equation (4) [46,47]: In the formula, E p represents the peak potential, E p/2 is the potential at which I = I p /2. Other symbols have their usual meanings. In our experiment, E p − E p/2 =34 mV, D = 2.9 × 10 −5 cm 2 s −1 , v = 100 mV s −1 , and T = 298 K. The value of k s was obtained as 1.02 × 10 −2 cm s −1 , confirming a relative rapid electrode reaction process.

Optimal of Some Determination Conditions
The electrochemical response of vanillin on the CeO 2 /ERGO/GCE was investigated in different types of supporting electrolytes, such as HAc-NH 4 Ac buffer solution (pH 4.0), phosphate buffer solution (pH 3.0 and 6.0), HAc-NaAc buffer solution (pH 4.0), HCl, H 2 SO 4 , HNO 3 , and H 3 PO 4 (each 0.1 M). It was found that the peak current was the largest and the peak shape was sharp in HCl solution. In the range of 0.02-0.5 M, the effect of HCl concentration on the peak current of 10 µM vanillin was studied. It was found that the maximum peak current of vanillin was obtained when the concentration of HCl reached 0.1-0.2 M (Figure 7). In this experiment, 0.1 M HCl solution was selected for vanillin detection.  Since the electrode process of vanillin on the CeO2/ERGO/GCE is controlled by adsorption, the accumulation conditions have a great influence on the peak current of vanillin. As showed in Figure  8A, when the accumulation potential changed in the range of −0.3-0.3 V, the peak current of vanillin increased first and then decreased, and reached the maximum value at 0.1 V. As shown in in Figure  8B, the peak current was greatly affected by the accumulation time. In the initial 120 s, the peak current increased significantly, and then the peak current tended to be stable when the time exceeded 120 s. This may be due to the saturated adsorption of vanillin. Therefore, the accumulation conditions (0.1 V, 120 s) were selected for vanillin detection. Since the electrode process of vanillin on the CeO 2 /ERGO/GCE is controlled by adsorption, the accumulation conditions have a great influence on the peak current of vanillin. As showed in Figure 8A, when the accumulation potential changed in the range of −0.3-0.3 V, the peak current of vanillin increased first and then decreased, and reached the maximum value at 0.1 V. As shown in in Figure 8B, the peak current was greatly affected by the accumulation time. In the initial 120 s, the peak current increased significantly, and then the peak current tended to be stable when the time exceeded 120 s. This may be due to the saturated adsorption of vanillin. Therefore, the accumulation conditions (0.1 V, 120 s) were selected for vanillin detection.

Linear Range and Detection Limit
Compared with the traditional CV method, the second-order derivative linear sweep voltammetry (SDLSV) has the advantages of low detection limit and high sensitivity, and is a more effective electrochemical technique for quantitative analysis. Therefore, SDLSV was used for the determination of vanillin in this study. Figure 9A,B showed the voltammetric curves of vanillin at different concentrations on the CeO 2 /ERGO/GCE under the optimized conditions. The proportional of well-defined peak current to vanillin concentration was observed in Figure 9C,D and the calibration curve in the form of I p versus concentration of vanillin showed two linear regions from 0.04 to 20 µM and 20-100 µM with the equations of I (µA) = 3.003c (µM) + 0.7587 (R = 0.9941) and I (µA) = 0.4582c (µM) + 48.159 (R = 0.9836), respectively. The detection limit was 0.01 µM. In order to estimate the analytical characteristics of the developed sensor, the CeO 2 /ERGO/GCE was comprehensively compared with other published electrochemical methods for vanillin detection. From the results in Table 2, it was found that, compared with other electrodes, the developed CeO 2 /ERGO/GCE has better detection performance with a wider dynamic range and lower detection limit. Since the electrode process of vanillin on the CeO2/ERGO/GCE is controlled by adsorption, the accumulation conditions have a great influence on the peak current of vanillin. As showed in Figure  8A, when the accumulation potential changed in the range of −0.3-0.3 V, the peak current of vanillin increased first and then decreased, and reached the maximum value at 0.1 V. As shown in in Figure  8B, the peak current was greatly affected by the accumulation time. In the initial 120 s, the peak current increased significantly, and then the peak current tended to be stable when the time exceeded 120 s. This may be due to the saturated adsorption of vanillin. Therefore, the accumulation conditions (0.1 V, 120 s) were selected for vanillin detection.

Linear Range and Detection Limit
Compared with the traditional CV method, the second-order derivative linear sweep voltammetry (SDLSV) has the advantages of low detection limit and high sensitivity, and is a more effective electrochemical technique for quantitative analysis. Therefore, SDLSV was used for the determination of vanillin in this study. Figure 9A,B showed the voltammetric curves of vanillin at different concentrations on the CeO2/ERGO/GCE under the optimized conditions. The proportional of well-defined peak current to vanillin concentration was observed in Figure Table 2, it was found that, compared with other electrodes, the developed CeO2/ERGO/GCE has better detection performance with a wider dynamic range and lower detection limit.

Reproducibility, Stability and Selectivity of CeO 2 /ERGO/GCE
Seven CeO 2 /ERGO/GCEs were fabricated by the same method to test the reproducibility, and 10 µM vanillin solution was measured under the same conditions. The relative standard deviation (RSD) of 4.62% (n = 7) indicated excellent reproducibility. In addition, to test the repeatability, the vanillin solution was determined by a single CeO 2 /ERGO/GCE for 10 times. After each measurement, to regenerate the electrode surface, two successive voltammetric sweeps were carried out in 0.1 M H 2 SO 4 solution in the range of 0.2-1.2 V. The good repeatability was reflected by the RSD of 2.74%. The storage stability of the modified electrode was also evaluated by storing the modified electrode in air. After two weeks, the response of the electrode is still 91.6% of the initial current, which showed that the electrode has long-term stability. In addition, it was found that when a 1000-fold amount of sucrose, glucose, fructose, K + , Ca 2+ , Na + , Mg 2+ , Zn 2+ , Al 3+ , and a 500-fold amount of citric acid, tartaric acid, lactic acid, and caffeine were present, the peak current change of 10 µM vanillin was less than 5%. Ascorbic acid (AA), uric acid (UA) and dopamine (DA) as three kinds of important biological substances in the human fluid. Figure 10 exhibited the SDLSVs obtained at the CeO 2 /ERGO/GCE in the presence of 1.0 mM AA, 10 µM DA, 10 µM UA, and 1.0 µM vanillin. As shown in Figure 10, four oxidation peaks were well separated in 0.1 M HCl solution, the oxidation peak potentials of AA, DA, UA and vanillin were 0.248 V, 0.473 V, 0.620 V, and 0.918 V, respectively. It was found that there was no obvious interference for the oxidation signal of vanillin (signal change below ±5%).

Application
The determination of vanillin in commercial food products such as biscuit, chocolate and pudding powder was performed using the developed method. The sample solutions were prepared according to reference [13], and the results were listed in Table 3. The contents of vanillin in these samples were calculated as 46.25 µg g −1 (for biscuit), 69.38 µg g −1 (for chocolate) and 132.98 µg g −1 (for pudding powder), respectively. The recoveries were 97.0-104.0%, which showed that the method can be used for the accurate and rapid detection of vanillin in commercial food samples.

Conclusions
In this paper, CeO2 nanoparticles were synthesized by the hydrothermal method and dispersed uniformly in graphene oxide (GO) aqueous solution by ultrasonication. The homogeneous CeO2-GO dispersion was coated on the surface of a glassy carbon electrode (GCE), and GO was converted into ERGO by potentiostatic reduction to obtain the modified electrode (CeO2/ERGO/GCE). This method offers several advantages over other techniques, including being green, efficient, inexpensive, and rapid. The electrochemical behaviors of vanillin on the CeO2/ERGO/GCE were studied carefully and the electrochemical parameters were calculated. The obtained CeO2/ERGO nanocomposite exhibited excellent performance for vanillin oxidation due to its strong electrocatalytic ability, good conductivity, and large surface area. Using the SDLSV technique, vanillin can be detected in the concentration range of 0.04-20 µM and 20-100 µM with the detection limit of 0.01 µM. Furthermore, the developed method has the advantages of high sensitivity, good selectivity, simple electrode preparation, and low cost. It has a wide application prospect in the sensitive detection of vanillin in commercial foods.

Application
The determination of vanillin in commercial food products such as biscuit, chocolate and pudding powder was performed using the developed method. The sample solutions were prepared according to reference [13], and the results were listed in Table 3. The contents of vanillin in these samples were calculated as 46.25 µg g −1 (for biscuit), 69.38 µg g −1 (for chocolate) and 132.98 µg g −1 (for pudding powder), respectively. The recoveries were 97.0-104.0%, which showed that the method can be used for the accurate and rapid detection of vanillin in commercial food samples.

Conclusions
In this paper, CeO 2 nanoparticles were synthesized by the hydrothermal method and dispersed uniformly in graphene oxide (GO) aqueous solution by ultrasonication. The homogeneous CeO 2 -GO dispersion was coated on the surface of a glassy carbon electrode (GCE), and GO was converted into ERGO by potentiostatic reduction to obtain the modified electrode (CeO 2 /ERGO/GCE). This method offers several advantages over other techniques, including being green, efficient, inexpensive, and rapid. The electrochemical behaviors of vanillin on the CeO 2 /ERGO/GCE were studied carefully and the electrochemical parameters were calculated. The obtained CeO 2 /ERGO nanocomposite exhibited excellent performance for vanillin oxidation due to its strong electrocatalytic ability, good conductivity, and large surface area. Using the SDLSV technique, vanillin can be detected in the concentration range of 0.04-20 µM and 20-100 µM with the detection limit of 0.01 µM. Furthermore, the developed method