3.1 SEM characterization of ZIF8, GO, ErGO, and ZIF8/ErGO
ZIF8, GO, ErGO, and ZIF8/ErGO were modified at the surface of glassy carbon (GC) sheet according to the method of Experiment 2.4. The SEM images of them are exhibited in Fig. 2. It can be seen from Fig. 2a and 2b that there are obvious wrinkles on the surface of GO and ErGO modified GC sheet, which is a distinctive feature of graphene. Figure 2c indicates that ZIF8 particles with uniform size are distributed on the surface of GC sheet. While the composite film of ZIF8 and ErGO is fixed onto the GC sheet, it clearly shows that ZIF8 material looks like being cladded by ErGO (shown in Fig. 2d). That pleated protuberance of graphene and porous structure of ZIF8 work in conjunction to increase the surface area and conductivity of GCE. Thus, it is conjectured that the ZIF8/ErGO GCE can improve the electrochemical response of NHDC by their synergistic effect.
3.2 Raman Characterization
As shown in Fig. 3, Raman spectroscopy was used to test and analyze the characteristic bands of carbon material at the surface of modified electrode, and the results showed that D and G peaks of the carbon atom crystal appeared at 1353.71 cm− 1 and 1592.79 cm− 1both for ZIF8/GO GCE and ZIF8/ErGO GCE [28], while there was almost no peak observed for ZIF8 GCE. In general, the D peak is the disordered vibration of the carbon atom crystal, which represents the defect of the carbon atom lattice. The G peak originates from the plane stretching vibration of sp2 hybrid carbon atom. The peak intensity ratio of ID/IG for ZIF8/GO GCE is calculated to be 1.08. After electrochemical reduction, an increase intensity ratio for ID/IG with a value of 1.78 is obtained for ZIF8/ErGO GCE, indicating that some sp3 hybrid carbon group such as carbon hydroxyl, carboxyl, other functional groups in GO is reduced and there are many defects in the ErGO, which demonstrates the successful reduction of GO in situ at ZIF8/GO GCE by cyclic voltammetric method.
3.3 Electrochemical impedance spectroscopy characterization
Figure 4 shows the electrochemical impedance spectroscopy (EIS) of different electrodes in 0.1 M KCl containing 5 mM Fe[(CN)6]3−/4−. The typical impedance spectra are observed at all electrodes in Nyquist plots of Fig. 4. That is, the characteristic impedance semicircle appears in the high frequency region, indicating a charge transfercontrolled process for the redox of Fe[(CN)6]3−/4− at the electrode surface; while the straight line appeared in the low frequency region reveals a diffusioncontrolled process. Ordinarily, the diameter of semicircle mirrors the charge transfer resistance (Rct). After fitting by equivalent circuit diagram, the Rct values of different electrodes are 137.31 Ω, 182.52 Ω, 11.12 Ω, and 23.85 Ω for bare GCE, ZIF8 GCE, ErGO GCE and ZIF8/ErGO GCE, respectively, manifesting that the larger impedance for ZIF8 shows poor conductivity while ErGO displays excellent conductivity. Concurrently, the smaller value of 23.85 Ω for ZIF8/ErGO GCE also exhibits favorable conductivity by compositing ZIF8 and ErGO, which becomes an ideal modified electrode material in this study.
3.4 Electrochemical response of NHDC at different electrodes
The CV responses of NHDC at bare GCE, ZIF8 GCE, ErGO GCE and ZIF8/ErGO composite film electrode are exhibited in Fig. 5. From yellow curve of inset, no peaks appear, indicating that there are no response for NHDC at bare GCE. After GCE is modified with ZIF8, ErGO and ZIF8/ErGO composite film, respectively, the electrochemical response of NHDC at the electrodes improves notably in comparison with bare GCE. Especially for the ZIF8/ErGO composite film modified electrode (red curve), three obvious redox peaks are observed during two cycles of CV measurements. For the first cycle, a strong oxidation peak (Pa1) at 0.506 V and a reduction peak (Pc) at 0.198 V are obtained. Moreover, the peak current of Pa1 at ZIF8/ErGO composite film electrode is about 3.8 times as much as that of ErGO GCE and 42 times of ZIF8 GCE. The probably reason is stemed from the synergistic effect between good adsorption of ZIF8 and high conductivity of ErGO. Thus, the sensitivity for NHDC is improved drastically at ZIF8/ErGO GCE. For the second cycle, another new oxidation peak (Pa2) appears at 0.227 V, which is one of a couple of redox peaks with Pc, caused by reversible redox process. It can also be concluded that Pa1 is produced by an irreversible oxidation process. Strangely, the peak current of Pa1 remarkably diminishes, which is due to the blocked pore channel caused by the oxidation product of NHDC in the first cycle. In the subsequent procedure, Pa1 in the first cycle is used for the determination of NHDC.
3.5 Calculation of reaction kinetic constant
In order to study the overall redox process of NHDC at the ZIF8/ErGO modified electrode surface and calculate the related kinetic parameters, the CV tests of NHDC was performed by two successive cycles at different scan rates, as shown in Fig. 6a and 6b, where 6a is the data for the first cycle and 6b for the second cycle. It can be seen that with the increase of the scan speed, three redox peaks current of NHDC also increases, two oxidation peaks potential shifts positively and one reduction peak potential slightly shift negatively. Figure 6c ~ 6f show the relationships between peak current and scan rate, or between peak potential and scan rate. Firstly, Fig. 6c gives good linear relationship between peak current of Pa1 (iPa1) and square root of scan rate (υ1/2), which reveals that the first irreversible oxidation of NHDC is a diffusioncontrolled process. In the same way, the linear fit between peaks current of Pc (iPc) and Pa2 (iPa2) and its scan rate (υ) is manifested in Fig. 6d. But unlike Fig. 6c, the linear relationship demonstrates that the reversible redox of the first oxidation product of NHDC is an adsorptioncontrolled process in Fig. 6d. Secondly, for the irreversible redox process, some kinetic reaction parameters can be calculated according to the Laviron equation[29]:\(\)
\({ E}_{P}={E}^{\theta }+\frac{RT}{\alpha nF}ln(\frac{RT{k}_{s}}{\alpha nF}\) ) + \(\frac{RT}{\alpha nF}ln\upsilon\) (1)
where α is the electron transfer coefficient, ks is the apparent rate constant, n is the electron transfer number, υ is the scan rate, Eθ is the standard potential, and other constants have their common sense. Figure 6e and 6f display the relationships between peaks potential (EPa1, EPc, EPa2) and narural logarithm of scan rate (lnυ), respectively. The linear regression equation of EPa1 and lnυ is EPa1=0.03037lnυ + 0.6149 (R2 = 0.9908). In terms of Laviron Eq. (1), the value of αn is calculated to be 0.85. For the irreversible diffusioncontrolled process, usually α = 0.5, that is, n = 1.70 ≈ 2, indicating that the first irreversible oxidation of NHDC is a twoelectron process. Finally, Eq. (1) is transformed into the following Equations according to Laviron theory, that is:
E pc =E θ \( \frac{RT}{\alpha nF}\) lnυ (1–1)
E pa =E θ \( \frac{RT}{(1\alpha )nF}\) lnυ (1–2)
lgk s = \(\alpha {lg}\left(1\alpha \right)+\left(1\alpha \right)lg\alpha lg\frac{RT}{nF\upsilon }\)  \(\alpha (\) 1 \(\alpha )\frac{nF\varDelta Ep}{2.3RT}\) (1–3)
where all symbols have the same sense as Eq. (1). By linear fit for Fig. 6f, the corresponding linear regression equations of EPa2 (or EPc) ~ lnυ are as follows: EPa2=0.0195lnυ + 0.3088 (R2 = 0.9957); EPc=0.0294lnυ + 0.1086 (R2 = 0.9901). The results of calculation express α with a value of 0.40 (α = 0.40) and n with a value of 2.19 (n ≈ 2) according to Eq. (1–1) and (1–2), indicating that the reversible redox of NHDC’s oxidation products is a twoelectron process. According to Eq. (1–3), it can also be calculated that the apparent rate constant ks of NHDC’s oxidation products at a scan rate of 0.05 V/s is 0.88 s− 1at ZIF8/ErGO modified electrode surface.
3.6 Optimization of experimental conditions
3.6.1 Effects of supporting electrolytes
0.1 M phosphate buffer solution (PBS), MBS, NH4ClNH3·H2O, HAcNaAc and HClC6H5O7Na3 were prepared, respectively. In this study, all the above buffer solutions with an approximate pH value was obtained by mixing a certain solution with volume ratio, and then 80 µM NHDC solution was added for the first cycle of CV test. The results imply that the largest oxidation peak current (ipa1) is achieved while the supporting electrolyte is MBS, so the buffer solution is chosen as MBS.
3.6.2 Effects of ratio and volume of ZIF8/ErGO
For the initial dispersion of ZIF8 and GO, its ratio and volume determine the ratio of ZIF8 to ErGO and the composite film thickness at the electrode, and ultimately influence the sensitivity of modified electrode. Thus, it is necessary to explore the effects of ratio and volume of ZIF8/ErGO on determination of NHDC. The different mass ratios of ZIF8 to GO with 5:1, 2:1, 1:1, 2:3 and 1:2 were mixed by ultrasonic dispersion, and then 8 µL of each dispersion was dropped onto the surface of electrode, the CV response of each modified electrode containing 80 µM NHDC was recorded (shown in Fig. S2a). The experimental results show that when the mass ratio of ZIF8 to GO is 2:3, the ipa1 value in the oxidation process of NHDC was the largest among all ratios, so the mass ratio of ZIF8 to GO (same as ZIF8 to ErGO) of 2:3 was selected as the optimal ratio.
Soon afterwards, each different amount of ZIF8/ GO dispersion with a fixed mass ratio of 2:3 were carefully transferred onto the surface of the glassy carbon electrode, including 2 µL, 4 µL, 6 µL, 8 µL and 10 µL. According to the same method as introduced in Experiment 2.4, the asprepared ZIF8/ErGO modified electrode with different thickness was applied to separately detect NHDC, as shown in Fig. S2b. The experimental results show that ipa1 of NHDC increases continuously with the increase of the modification volume, until the peak current reaches the maximum while the modification volume is 8 µL, and later the peak current decreases as the modification amount continuing rising. The probable reason is that quantities of ZIF8 with poor conductivity will hinder the electron transport of NHDC on the electrode surface as the ZIF8/ErGO film thickens with increasing the volume of dispersion. Therefore, 8 µL is selected as the best modification volume during the experiment.
3.6.3 Effect of pH
pH of solution also influences the peak current and peak potential of NHDC redox. The effect of pH on the redox of NHDC was tested in MBS in the range of pH 3 ~ 7, as exhibited in Fig. 7a (the first cycle) and 7b (the second cycle). As pH increases, peak current of Pa1 increases firstly and then decreases (seen in star symbols of Fig. 7c). Therefore, pH 5.0 MBS is chosen for further study in the experiment. Figure 7c also separately depicts the dependence of each redox peak potential of NHDC on pH. It is vividly found that each peak potential (Ep) of Pa1, Pc and Pa2 shifts to the negative direction with the increase of pH, indicating that H+ was involved in the redox process of NHDC. In addition, each peak potential of redox of NHDC shows a good linear relationship with pH. The linear regression equations are expressed as follows: Epa1=0.0343pH + 0.7139, R2 = 0.9964; Epc=0.0531pH + 0.4469, R2 = 0.9933; Epa2= 0.0604pH + 0.5482, R2 = 0.9922. According to the Nernst equation:
$$\frac{d{E}_{p}}{d\text{p}\text{H}}=\frac{2.303mRT}{nF}$$
2
Where m and n are the transfer number of proton and electron, respectively, and other symbols have the conventional meanings. Thus, for Pa1, it can be calculated that m/n = 0.57 ≈ 0.5, and m = 1 according to the result of n = 2 provided in Step 3.5,suggesting that the first oxidation of NHDC on the surface of the modified electrode is a oneproton and twoelectron process. For Pc and Pa1, it can also be obtained that the transfer number of proton and electron is equal in the subsequent redox process in the light of the similar computing method, that is, m = n = 2, so the redox process of the second step for NHDC is a twoproton and twoelectron transfer process. Figure 7d reveals the whole redox mechanism of NHDC at the ZIF8/ErGO GCE, which is consistent with the result reported in the literature [30].
3.6.4 Effect of enrichment time
As described in the above discussion, the first oxidation of NHDC is an adsorptioncontrolled process, thus enrichment time will affect the current response of NHDC. In this experiment, 80 µM NHDC was added into MBS with pH = 5 and the CV responses at the ZIF8/ErGO GCE were recorded by enabling preconditioning of time with opencircuit under magnetic stirring. The enrichment time was started from 30 s and raised every 30 s. The experimental results show that, with the increase of enrichment time, ipa1 gradually increases until it reaches the maximum at 240 s, and finally ipa1 tends to be stable (seen in Fig. S3), so the enrichment time is set as 240 s in this experiment.
3.7 Standard curve for NHDC detect
Under the optimized experimental conditions, LSV method was used to determine NHDC in the concentration range of 0.08 ~ 80 µM, as shown in Fig. 8a. The oxidation peak current of NHDC climbs with its concentration. Figure 8b unveils the linear relationship between the oxidation peak current (ipa1) of NHDC and its concentrations (cNHDC) from 80 nM to 80 µM, and the fitting linear equation is stated as: ipa1(µA) = 0.4892 + 0.7674 cNHDC(µM) (R2 = 0.9922). The detection limit was 31.5 nM (S/N = 3), according to Eq. (3):
LOD= \(\frac{3\sigma }{R}\) (3)
Where σ represents the standard deviation of ipa1 at the lowest concentration under the operating curve (σ = 0.008057 µA after repeated measurement for 11 times); R represents the slope of the fit line.
3.8 Antiinterference and reproducibility of ZIF8/ErGO GCE
In order to test the antiinterference ability of ZIF8/ErGO GCE, Fig. 9 records the oxidation peak current response of NHDC by LSV method under optimized conditions while some inorganic ions or organic substances are individually added to MBS containing 10 µM NHDC. .Considering the possible interference in food such as milk or drink, the NHDC solutions coexisting with 400fold concentration of NaCl, MgCl2, FeCl3, KNO3 and CuSO4, respectively, were detected. The experimental results show that the relative deviation of ipa1 was less than ±5%, indicating that the modified electrode has preeminent ability of antiinterference for NHDC detection in the presence of the above ions. In the same way, 50fold concentration of sucrose and glucose, and 10fold concentration of dopamine, ascorbic acid and pnitrophenol were also added, respectively, there is still no interference for the detection of NHDC. Therefore, ZIF8/ErGO GCE has strong antiinterference property.
To characterize the reproducibility of ZIF8/ErGO GCE, the same modified electrode was parallel tested for 7 times by LSV method under optimized conditions. Due to strong adsorption of NHDC and its redox product on the electrode, the modified electrode should be first placed in blank neutral PBS buffer solution for 10 cycles of CV after each test, and then the next test was conducted. The relative standard deviation (RSD) of ipa1 was 4.67% for 10 µM NHDC, indicating that the modified electrode has good reproducibility. Furthermore, the 7 batches of ZIF8/ErGO GCE were attained according the same procedure for fabricating electrode, and the RSD of ipa1 was only 3.56%, also demonstrating good reproducibility.
3.9 Detection of NHDC in milk samples
The milk samples of 100 µL, 500 µL and 1000 µL, pretreated as stated in Experiment 2.6, were diluted to constant volume of 10 mL with pH = 5 MBS, respectively. Under the optimized experimental conditions, the LSV method was used for several tests, and the results show that no NHDC was detected in the milk samples or the content of NHDC in the samples was quite low.
After that, the standard addition method was chosen to test the practical application of the modified electrode by adding a certain amount of NHDC solution to the above diluted milk samples. While NHDC of 1.6 µM, 2.4 µM and 4.8 µM was added for parallel measurements of three times, respectively, its recoveries are determined in the range of 98.3% and 106.3% (shown in Table 1). The results were reliable. Therefore, this method can be used for the detection of NHDC content in milk.
Table 1 Detection of NHDC in milk samples (n=3)
Samples of milk

Found/μM

Added/μM

Detected/μM

Recovery/%

1
2
3

_
_
_

1.60
2.40
4.80

1.70
2.36
4.76

106.3
98.3
99.2
