3.1. Characterization of the materials
The morphologies of the obtained materials (hollow carbon spheres and the NC@ZIF-8 composite) were observed using TEM. Figure 2(a) shows that the surface of the hollow carbon sphere is very regular and does not have other structures attached. The hollow carbon spheres are distinguishable, indicating their successful synthesis using SiO2 particles as the template. The TEM image of the NC@ZIF-8 composite (Fig. 2(b)) reveals that some crystals are attached to the surface of the hollow carbon spheres, and the crystals present a regular octahedral structure, which is consistent with the typical morphology of ZIF-8 crystals [9]. It can therefore be concluded that the synthesis of ZIF-8 crystals on hollow carbon spheres by the in situ growth method was successful, and the synthesized crystals grew irregularly on the surface of the hollow carbon spheres. To further investigate the crystal phase, the morphologies of the synthesized materials were characterized by SEM. The SEM image of the carbon spheres in Fig. 2(c) shows that the surface of the hollow carbon sphere is covered with crystals, and the crystals are evenly distributed, not only on the surface of the hollow carbon sphere but also inside it. This is because some of the spheres cracked during the synthesis, and the ZIF-8 crystals were adhered to the inner walls of such spheres. Figure 2(d) reveals that the silica particle template used to synthesize the carbon sphere was completely etched by sodium hydroxide, resulting in a uniform hollow carbon sphere with a neat surface, which corresponds to the TEM results. These SEM results are in good agreement with the TEM results.
Figure 3 shows the XRD pattern of the NC@ZIF-8 composite, in which the peaks observed at 2θ = 7.29, 10.34, 12.69, 14.66, 16.38, and 17.99 are attributable to the (001), (002), (112), (022), (013), and (222) planes of the ZIF-8 crystal, respectively. These peaks indicate the successful formation of the MOF crystals on the hollow carbon sphere [10]. However, the diffraction peaks of the hollow carbon sphere were not observed in the XRD pattern, probably because of their relatively low peak intensity compared with that of the ZIF-8 crystal peaks, which caused the diffraction peak of the hollow carbon sphere to be hidden. The XRD pattern of the NC@ZIF-8 composite is consistent with that of the standard card, indicating that the desired material was successfully produced [11].
The functional groups of the NC@ZIF-8 composites were analyzed using X-ray photoelectron spectroscopy (XPS). As shown in Fig. 4(a), the composite contains four elements: C, N, O, and Zn. Figure 4(b) shows that the C1s peak can be deconvoluted into five characteristic peaks at 284.6, 285.4, 286.6, 287.8, and 289 eV, corresponding to C = C, C-N, C-O, C = O, and O-C = O linkages [12]. Figure 4(c) shows that the O1s peak can be deconvoluted into three peaks that represent C = O, C-OH, and C-O [13] bonds at 531.4, 532.3, and 533.4 eV, respectively. Figure 4(d) shows the fitting of the N1s peak with three components corresponding to pyridinic, pyrrolic, and graphitic N atoms at 398.2, 399.9, and 500.9 eV, respectively [14].
3.2. Brunauer-Emmett analysis of the composite
To determine the specific surface area and material type, the BET analysis of the nitrogen adsorption-desorption curve of the NC@ZIF-8 composite was performed (see Fig. 5(a)). The adsorption capacity of the material increased rapidly under low relative pressures and tended to be stable after reaching a certain pressure, but it increased again at higher pressures. Compliance with type IV isotherms indicates that the material has a mesoporous structure [15]. Through BET analysis, the specific surface area of the composite was determined to be 175.8756 m²/g. As shown in Fig. 5(b), the pore size of the NC@ZIF-8 composite was mainly concentrated in the range of 2–16 nm, and the pore size range of this mesoporous material was 2–50 nm, which is consistent with the isothermal adsorption results.
3.3. Optimization of electrochemical conditions
For the electrochemical detection of luteolin using the NC@ZIF-8/GCE, the most important experimental parameters, such as the amount of the material used to modify the GCE, enrichment potential, enrichment time, and pH of the buffer were optimized.
Figure 6(a) shows the effect of the amount of NC@ZIF-8 used to modify the GCE on the peak current in DPV measurements. In DPV, the peak current is considerably affected by the amount of the material used for GCE modification. For NC@ZIF-8/GCE, the peak current increased with the increase in the amount of NC@ZIF-8 used in GCE modification, reaching the maximum value at 3 mg/mL. However, when the amount of NC@ZIF-8 used in GCE modification was increased further, the thick layer of the composite formed on the electrode hindered electron transfer, causing the peak current to decrease. Therefore, an NC@ZIF-8 concentration of 3 mg/mL was selected as the optimal one for GCE modification.
To explore the influence of luteolin enrichment time on luteolin detection using the NC@ZIF-8/GCE, the enrichment time was varied between 60 and 180 s (the enrichment potential was fixed at 0.2 V). As shown in Fig. 6(b), the enrichment time affected the peak current. As the enrichment time increased from 60 to 120 s, the peak current increased gradually. When the enrichment time reached 150 s, the electrode surface was saturated. Nevertheless, the peak current increased at a smaller rate. Considering the comprehensive time cost and efficiency, an enrichment time of 120 seconds was selected for further experiments.
Figure 6(c) shows the effect of the enrichment potential on the peak current in the detection of luteolin using the optimized NC@ZIF-8/GCE. The enrichment potential in the range of 0.1–0.5 V was explored; as the enrichment potential was increased from 0 to 0.2 V, the peak current increased significantly and reached the maximum value at 0.2 V. When the enrichment potential was increased further, the peak potential decreased. The optimal enrichment potential in this experiment was 0.2 V, and an enrichment potential of 0.2 V was used in subsequent experiments.
Next, the effect of the buffer pH on the experimental peak current was investigated. As shown in Fig. 6(d), the peak current in the DPV curve increased with an increase in the buffer pH from 3 to 5, reaching the highest value at pH 5. When the buffer pH was increased further, the peak current decreased because of the decreased acidity of the solution and thus the H+ concentration, which slowed the electron transport. Subsequent experiments were performed in a buffer solution with pH 5 to optimize the electrochemical detection of luteolin.
3.4. Effects of pH and scan rate
Figure 7(a) shows the DPV curves corresponding to the detection of 10 µM luteolin with an NC@ZIF-8/GCE sensor at different pH values. The solution pH strongly affected the intensity of the detected electrochemical signals. As shown in Fig. 7(a), Epa changed with the solution pH, and when the pH increased from 3 to 7, Epa shifted in the negative direction. As shown in Fig. 7(b), the peak potential has a good linear response relationship with pH, and the corresponding linear equation is Epa = − 0.0582pH + 0.6958 (R2 = 0.992). The slope of the line is 58.2 mV/pH, which is close to the theoretical value of 59 mV/pH based on the Nernst equation, indicating that the number of protons involved in the reaction is equal to the number of electrons transferred in the reaction [16].
To explore the redox reaction mechanism of luteolin, the electron transfer number of the reaction was calculated as 2.3 RT/nF, where R is the gas constant, T is the temperature, F is the Faraday constant, and n is the number of transferred electrons. The electron transfer number calculated based on the peak potential difference in Fig. 7(a) is n = 1.87, which is close to 2 [17]. These results indicate that luteolin is involved in the transfer of two protons and electrons in the redox reaction during the electrochemical reaction on the NC@ZIF-8/GCE sensor, and the corresponding electrochemical reaction is shown in Fig. 8 [18].
To further study the electrochemical behavior (redox reaction) of luteolin on the NC@ZIF-8/GCE sensor, the effects of different sweep velocities in the range of 30–210 mV/s on the electrochemical peak current were studied by CV. As shown in Fig. 7(c), the oxidation and reduction peak currents corresponding to the redox reaction of luteolin showed a good positive trend with scan rate v. This trend indicates that the electrochemical behavior of luteolin on the NC@ZIF-8/GCE sensor is mainly controlled by the adsorption process [19]. Figure 7(d) shows the linear relationship between scan rate v and oxidation peak current, and the linear equation is Ipa (Ipa = 0.07489v − 0.49465 (R2 = 0.995)). Similarly, a good linear relationship between the scan rate and reduction peak current was observed, and the linear equation is Ipa = − 0.0456v − 0.03747 (R2 = 0.998).
3.5. Electrochemical behavior of luteolin on different electrodes
Figure 9(c) shows the CV curves of the bare GCE and three modified electrodes, viz., hollow carbon sphere/GCE, ZIF-8/GCE, and NC@ZIF-8/GCE. As shown, the NC@ZIF-8/GCE has the largest redox peak, indicating that its electrochemical performance is better than those of the bare electrode and the other two electrodes modified with the components of the composite (hollow carbon spheres and ZIF-8).
To determine the effective areas of the different modified electrodes, the linear relationship between charge Q and t1/2 was obtained by evaluating the bare GCE and three differently modified electrodes by the CC method (see Fig. 9(a, b)). The effective areas (A) of the four different electrodes were calculated using the Anson equation:
Q = 2nFAcD1/2t1/2/п1/2 + Qdl + Qads [20] (1)
Here, F is the Faraday constant (96500 C/mol), c is the substrate concentration, which is 5×10− 6 mol/cm3, D is the standard diffusion coefficient, for which a value of 7.6×10− 6 cm2/s was used, Qdl is the double-layer charge, Qads is the Faraday charge, and n is the number of electrons transferred, which was calculated to be 2. Therefore, according to Eq. (1), the active surface areas of GCE, NC/GCE, ZIF-8/GCE, and NC@ZIF-8/GCE are 0.0072, 0.0472, 0.0910, and 0.0943 cm2, respectively. Thus, the effective area of the NC@ZIF-8/GCE electrode is significantly larger than those of the other three electrodes, which is more conducive to the enrichment of luteolin.
3.6. Establishment of standard curves
After the optimization experiments, a standard curve was plotted to verify the performance of the composite sensor in luteolin detection via DPV electroanalysis under the optimized conditions. As shown in Fig. 10(a), the oxidation peak current in the DPV experiment increased as the luteolin concentration increased from 0.05 to 30 µM. The oxidation peak current (Ipa) exhibited a good linear relationship with the concentration of luteolin (c), as shown in Fig. 10(b), and the linear equation is Ipa (µA) = 0.99433 c (µM) + 0.92699, R2 = 0.9997. The NC@ZIF-8/GCE sensor has a limit of detection (LOD, S/N = 3) of 0.011 µM for luteolin. The experimental results indicate that the prepared NC@ZIF-8/GCE has a good linear concentration range and a low LOD. It is therefore suitable for specific practical applications.
3.7. Repeatability and reproducibility testing of the NC@ZIF-8/GCE
The repeatability and reproducibility of the NC@ZIF-8/GCE sensor were evaluated by conducting DPV under the optimized conditions (Fig. 11). To study the repeatability of the electrode, a freshly prepared NC@ZIF-8/GCE was used in the detection of 10 µM luteolin up to 8 times. To study the reproducibility of the NC@ZIF-8/GCE sensor, eight replicate NC@ZIF-8/GCE sensors were prepared to detect luteolin at 10 µM concentration. The relative standard deviations (RSDs) for repeatability and reproducibility are 2.62 and 1.26%, respectively. These results reveal that the NC@ZIF-8/GCE sensor has good repeatability and reproducibility, and thus has good application potential.
3.8. Anti-interference capability of NC@ZIF-8/GCE in luteolin detection
Under optimal experimental conditions, the anti-interference ability of the NC@ZIF-8 composite in luteolin detection was evaluated by DPV. K+, Cl−, Mg2+, Na+, and NO3− were added at 500-fold, while glucose, fructose, ascorbic acid, mannitol, and citric acid were added at 300-fold to a PBS buffer solution containing 10 µM luteolin (0.1 M), and DPV was conducted under optimal conditions. As shown in Fig. 12, compared with the blank control group without interfering substances, the change range of the peak current for the solutions added with various interfering species was less than 5%, which falls in the normal range. This experimental result demonstrates that the presence of these ions and substances did not interfere with luteolin detection. Therefore, the NC@ZIF-8/GCE chemical sensor can be used for the accurate, reliable, and safe detection of luteolin in actual samples.
3.9. Detection of actual samples
To evaluate the ability of the NC@ZIF-8/GCE sensor to detect luteolin in actual samples, a standard addition method was used for its detection in honeysuckle extract and watermelon juice. Honeysuckle was purchased from a local pharmacy, and watermelons were obtained from a local supermarket. For extraction, 1 g of honeysuckle was suspended in 10 mL of absolute ethanol, the mixture was sonicated for 1 h, and the filtrate was collected. The volume of the test sample was then made up with ethanol solution in a 100 mL volumetric flask. The watermelon was manually squeezed into watermelon juice and filtered, and the clear filtrate was diluted 100 times with aqueous ethanol. The recovery rates of luteolin in honeysuckle extract and watermelon juice samples were 95.41–100.06% and 98.09–101.20%, respectively. These results prove that the NC@ZIF-8/GCE chemical sensor can effectively detect luteolin in actual samples. Thus, the NC@ZIF-8/GCE chemical sensor can be used accurately, reliably, and safely to detect luteolin in actual samples.
Table 1
Detection of luteolin in watermelon juice and honeysuckle extract using the NC@ZIF-8/GCE sensor
Sample | Added (µM) | Found ± SD by DPV (µM) | Recovery (%) |
Watermelon juice | 0 | - | - |
3 | 3.036 ± 0.146 | 101.22 |
10 | 9.809 ± 0.261 | 98.09 |
20 | 19.903 ± 0.750 | 99.52 |
Honeysuckle | 0 | 1.338 ± 0.066 | - |
3 | 4.139 ± 0.066 | 95.41 |
10 | 11.086 ± 0.228 | 97.78 |
20 | 21.351 ± 0.297 | 100.06 |