Construction of a Molecularly Imprinted Sensor Modified with Tea Branch Biochar and Its Rapid Detection of Norfloxacin Residues in Animal-Derived Foods

Norfloxacin (NOR) is a common antibiotic used in humans and animals, and its high levels can cause intolerance or poisoning. Therefore, NOR levels in animal-derived foods must be monitored due to potential side effects and illegal use phenomena. This research centered on the development of an environmentally friendly electrochemical sensor for NOR detection. Potassium carbonate activated tea branch biochar (K-TBC) as an efficient use of waste was coated on the surface of glassy carbon electrode (GCE), and a molecular-imprinted polymer (MIP) layer was subsequently electropolymerized onto the modified electrode. NOR was used as template molecule and o-phenylenediamine (o-PD) and o-aminophenol (o-AP) were used as bifunctional monomers. The electrochemical sensor was built and its electrochemical behavior on NOR was investigated. The sensor demonstrated an excellent linear current response to NOR concentrations in the ranges of 0.1–0.5 nM and 0.5–100 nM under ideal experimental circumstances, with a detection limit of 0.028 nM (S/N = 3). With recoveries ranging from 85.90% to 101.71%, the designed sensor was effectively used to detect NOR in actual samples of milk, honey, and pork. Besides, the fabricated sensor had low price, short detection time, good selectivity and stability, which can provide a theoretical and practical basis for the actual monitoring of NOR residues.


Introduction
Norfloxacin (NOR), a third-generation quinolone antibiotic, is a low-cost, highly effective broad-spectrum antimicrobial agent, which has been widely used in livestock and poultry farming and veterinary clinics [1,2]. However, NOR remains in meat, milk, honey, and other animal-derived foods, which may lead to pathogenic bacteria developing drug resistance, as well as joint pain, aortic aneurysms, and other diseases, posing a serious threat to human health [3][4][5]. In order to better guarantee the quality and safety of food of animal origin. China's Ministry of Agriculture and Rural Affairs has issued a notice prohibiting the use of NOR in food animals [6]. However, some farmers or businesses continue to use banned antibiotics for profit. The results of veterinary drug residue detection of animalderived foods in China (2020) showed that the residual amount of NOR was 163 µg/kg in honey samples, which seriously exceeded the standard [7]. As a result, NOR must be detected in animal-derived foods for risk assessment using appropriate, precise, and handy analytical methods.
At present, the conventional methods for NOR detection mainly include high performance liquid chromatography (HPLC) [8], high performance liquid chromatography-mass spectrometry (HPLC/MS) [9], microbial detection methods [10], and enzyme-linked immunosorbent assay (ELISA) [11]. These chromatographic methods have high detection accuracy and sensitivity, but they usually take a long time and require expensive and sophisticated analytical instruments. Meanwhile, the antigens and antibodies used in ELISA are expensive and easily cross-react, and microbial detection methods suffers from a high level of detection error, which significantly limits their widespread application. In contrast, the electrochemical sensor was considered to be a sensitive electrochemical analytical method with simple equipment, short analysis time and good repeatability, which can easily realize online detection of target substances, and it has broad development prospect [12,13].
In the last few years, porous materials based on electrochemical sensors have been brought into focus. Amongst multifarious porous materials, biochar (BC) is a potential material for sensor preparation due to its unique physical chemical and green environmental protection. BC is a highly aromatic porous material obtained by carbonization and activation at high temperature using carbon sources such as animals and plants, typically has a large specific surface area, high porosity (mostly mesoporous or microporous), good electrical conductivity, and abundant oxygen-containing functional groups (-COOH, -OH) [14][15][16], indicating that BC has excellent potential as electrode modification materials to construct electrochemical sensors. As a kind of biomass raw material, the fixed carbon content of tea waste is generally more than 10% [17], which is similar to the commercialized coconut shell and walnut shell [18,19], indicating that tea waste has the application potential of biochar preparation. Furthermore, China, the world's largest tea producer, can generate approximately 1 million tons of pruned tea branches (TB) annually, which is not effectively used [20]. Consequently, it is worthwhile to research the conversion of wasted tea branches into high-value biological products. However, there are few reports on the application of tea tree branch biochar (TBC) in the direction of electrochemical sensing, it is commonly used to remove tetracycline from wastewater [20,21]. However, small selective adsorption capacity may occur when TBC is used alone, molecularly imprinted polymers (MIPs) were introduced to further improve the selectivity of the electrochemical sensor for NOR.
The benefits of molecular imprinting and electrochemical techniques are combined in molecularly imprinted electrochemical sensors (MIECS) with high selectivity and rapid detection capabilities. By means of bulk polymerization, suspension or precipitation polymerization (PPm), molecular self-assembly electropolymerization (EPm), and other processes, MIPs have been deposited on MIECSs [22,23]. EPm has the advantage of allowing the thickness and density of the polymer layer to be controlled by selecting appropriate functional monomers and adjusting electrochemical parameters, which eliminates the drawbacks of conventional MIPs such as incomplete template removal and low binding capacity [24]. The choice of functional monomer is critical for the EPm construction of an MIP-based sensor. Recently, aniline-based copolymers (such as o-phenylenediamine) have been synthesized and applied to novel MIPs [24]. Since this particular copolymer has novel properties brought about by other monomers in addition to the properties of Poly-aniline (PAn). It is crucial to remember that adding aminophenol to PAn-based copolymers expand the copolymer's pH range since the hydroxyl functional groups on the benzene ring can undergo reversible oxidation and reduction. For instance, a poly(o-phenylenediamineco-o-aminophenol) (PoPD-oAP)/GCE sensor was used to detect ascorbic acid with an LOD of 36.4 µmol/L [25] and an MIP (PoPD-oAP)/PEDOT:PSS@Fc/SPCE electrochemical sensor with an LOD of 8.3 × 10 −8 ng/mL in the presence of prostate specific antigen (PSA) [26]. Compared with the MIP of a single functional monomer, the molecularly imprinted copolymer exhibits stronger sensitivity and selectivity to the target molecule, which may be because multiple functional groups can be recognized or interact with the template molecule in various ways. Therefore, MIP based on copolymer synthesis is beneficial to the development of MIECSs.
Herein, a simple and efficient MIP electrochemical sensor for the detection of NOR was developed by stepwise modification of K-TBC and MIP films using binary functional monomers on glassy carbon electrodes, Figure 1 depicts the synthesis mechanism. Considering the superior porosity and effective catalytic activity, TBCs were prepared by facile pyrolysis and chemical activation methods to increase the sensor's sensitivity to detection. By using the cyclic voltammetry (CV) approach, MIP films were electropolymerized onto TBC/GCE surfaces using o-PD and o-AP as binary functional monomers and NOR as a template molecule. Functional monomers and template molecules are connected by hydrogen bonding interactions because both o-PD and o-AP have amino groups. Therefore, the prepared MIP films have high template adsorption selectivity. The molar ratio of template to functional monomer, the proportion of bifunctional monomers, pH of polymerization solution, elution time, and incubation time were all optimized as experimental parameters affecting MIP electrochemical sensor performance. Furthermore, the sensor's applicability in animal-derived food samples such as milk, honey, and pork were investigated. The proposed technology is the first use of an MIP electrochemical sensor by bifunctional monomers for NOR detection that we are aware of.

Apparatus and Reagents
Apparatus: The ZEISS Sigma 300 apparatus was used to perform scanning electron microscopy (SEM) at a voltage of 10.0 kV. The phase and elemental analyses were performed on an Ultima IV X-ray diffractometer (XRD, Rigaku, Saitama, Japan) and X-ray photoelectron spectroscope (XPS, K-Alpha, Thermo Scientific, Waltham, MA, USA). A CHI660E electrochemistry workstation was used for all electrochemical measurements (CH Instruments, Shanghai, China). A glass carbon electrode (GCE, 3.0 mm in diameter) functioned as the working electrode in the traditional three-electrode system. A calomel electrode (3 M KCl) served as the reference electrode, and a platinum wire served as the counter electrode. At room temperature, all electrochemical studies were conducted.

Preparation of Tea Tree Branch Biochar
The tea branch biochar was prepared by high-temperature carbonization treatment and chemical activation method with slight modification [27]. The dried tea branches were washed, crushed and sieved, and then coked for 2 h in an N 2 -atmosphere at 600 • C. The obtained biochar particles were marked as "TBC".
The as-prepared biochar particles were then chemically activated. For chemical activation, the TBC (1 g) were mixed with different concentrations of K 2 CO 3 (The quality ratio of samples and K 2 CO 3 from 1:1 to 1:3) at room temperature, and then dried in an oven at 80 • C. The dried samples are activated in a tubular furnace and the required burning temperatures (650 • C, 700 • C, 750 • C, 800 • C and 850 • C) and burning time (1 h, 1.5 h, 2 h, 2.5 h and 3 h) remain unchanged. After the sample was cooled to room temperature, it was boiled with 0.1 mol/L hydrochloric acid, and then repeatedly washed with hot distilled water until the filtrate became neutral, and then dried at 105 • C. The obtained product was marked as "K-TBC".

Preparation of the Biochar-Modified GCE
Prior to modification, the GCE was successively polished with 0.3 µm and 0.05 µm Al2O3 slurry on a polishing cloth, followed by a 5-min ultrasonic wash in ultrapure water [28]. The bare electrode was repeatedly cycled between −0.2 and 0.6 V (scan rate, 50 mV/s) prior to the voltammetric test in order to gather reliable responses. To obtain the TBC/GCE, 5 µL of the equally mixed 3 mg TBC and 1 mL of the chitosan-acetic acid solution (the mass concentration of chitosan ranged from 0.1% to 0.5%) was poured onto the naked glassy carbon electrode (the loading of TBC from 1 µL to 6 µL). The solution was then thoroughly dried at room temperature. under the potential range of 0-1.2 V to obtain electrically polymerized membranes [25]. After the electro polymerization, the polymer-modified electrode was incubated in the mixed solution of NaOH (0.1 mol·L −1 ) and 100% ethanol (V:V = 1:3) solution for 12 min to disconnect the hydrogen bond and remove NOR, thus obtaining the MIP/TBC/GCE sensor. As a blank control trial, the non-imprinted electrode (NIP/TBC/GCE) was prepared in the same way as the imprinted electrode except that no template molecule NOR was added.

Electrochemical Measurement and NOR Detection
Cyclic voltammetry (CV) and differential pulse voltammetry (DPV) were used to characterize the preparation process of the sensor. CV experiments were performed in 1 M KCl solution containing a 5 mM [Fe (CN) 6  where I 0 and I 1 represent the peak current before and after incubation with NOR, respectively.

Sample Preparation
In this study, samples such as milk, honey and pork were pre-processed and tested according to standard procedures to evaluate the suitability of the MIP/TBC/GCE sensor for the detection of NOR in food. Each sample was spiked with three different doses of NOR for spike recovery experiments (6.0, 30.0 and 60.0 nM).
We placed 2 g of spiked milk and pork samples in a 50 mL centrifuge tube, then 5 mL of 10% trichloroacetic acid-acetonitrile solution was added, followed by thorough mixing. After centrifuging at 10,000× g rpm for 3 min to remove the protein, the milk sample supernatant was concentrated in a nitrogen stream at 40 • C, redissolved in AB solution (pH = 5.0), and then filtered through a 0.22 µm filter to obtain samples for analysis.
Then, 2 g of spiked honey samples was placed in a 50 mL centrifuge tube, and 20 mL of phosphate buffer solution was added followed by thorough mixing. After centrifuging at 10,000× g rpm for 5 min, the supernatant of honey sample is purified in HLB solid phase extraction column, concentrated in nitrogen stream at 40 • C, and redissolved in AB solution (pH = 5.0). Finally, the samples were filtered through a 0.22 µm filter for analysis.
All the experiments were carried out in triplicate.

Statistical Analysis
The data were subjected to analysis of variance, and the means were evaluated using Duncan's multiple range test (ANOVA, Abacus Concepts, Inc., Piscataway, NJ, USA), with p < 0.05 considered significant. The results are reported as the mean ± standard deviation based on multiple samples (n = 3).

Characterization of Materials and Polymers
Scanning electron microscopy (SEM) was used to describe the morphology of TBC, K-TBC, K-TBC/ITO, NOR-MIP/K-TBC/ITO, and MIP/K-TBC/ITO. In Figure 2A, the TBC sample has a long tubular structure, which can maintain the basic skeleton and organizational structure of plant raw materials without many pores. After the activation of K 2 CO 3 , the K-TBC surface has the pores which were different sizes and different shapes, and the diameter is about 20 nm-10 µm. In addition, its pore structure extends to the interior, with more micropores and a small amount of granular texture ( Figure 2B,C). This may be attributed to the removal of hydrogen, oxygen atoms in the pyrolysis process, and the etching of potassium. Other researchers have observed similar findings [29][30][31].
Compared with Figure 2C,D, it can be seen that the surface of K-TBC/ITO is smoother than that of K-TBC, which may be due to the addition of chitosan as a fixative. However, when a molecularly imprinted membrane layer aggregates, the initially smooth surface becomes rough, the hole wall is slightly thickened, and the surface is bright and evenly distributed with white particles ( Figure 2E). MIP/K-TBC/ITO was obtained after the modified electrode was immersed in the elution solution to remove the template molecules. Cracks appeared on the surface of the imprinted membrane, and cavities were released ( Figure 2F). These results indicate that the preparation process of modified electrode is successful. The X-ray diffraction (XRD) was used to characterize the crystal structure of biochar. Figure 3A shows the X-ray diffraction (XRD) characterization of K-TBC and TBC. Two broad diffusional diffraction peaks on curve a and b are located at around 23.8 • and 43.6 • , which correspond respectively to the (002) and (101) plane reflection of graphite-carbon structure, it confirms the amorphous carbon structure of K-TBC and TBC accompanied by local graphitization [32,33]. The samples all contain strong miscellaneous peaks, indicating that there are a small number of foreign bodies in the samples, which may be caused by the sample preparation process being insufficiently rigorous. In addition, the intensity of peak (002) can reflect the degree of graphitization [34], indicating that the degree of graphitization of activated K-TBC decreases, indicating that the activated biochar graphite microcrystal disorder increases, the carbon structure tends to be disordered, and the activity of surface atoms is enhanced, which is easy to form a more developed pore structure [35]. In addition, the (101) planes show a higher degree of interlayer packing, which can effectively increase the electrical conductivity of the material. The surface element compositions and chemical valence states of TBC and K-TBC were studied by X-ray photoelectron spectroscopy (XPS). As can be seen from Figure 3B, the total XPS spectra of TBC and K-TBC are similar. K-TBC has three peaks at 285.17 eV, 400.69 eV and 533.02 eV, which are attributed to C1s, O1s and N1s, and their element ratios are 88.13%, 10.47% and 1.39%, respectively. The data showed that the activation of potassium carbonate increased the mass fraction of O on the surface of tea branch biochar, but decreased the mass fraction of C and N at the same time. The reason may be that the potassium carbonate and its decomposition products react with the C in TBC to generate CO 2 , CO and other gases, which escape, consuming a large amount of C element, increasing the relative content of O element and increasing the oxygen-containing groups on the surface [36].
In order to further explore the effect of activation on the functional groups of biochar, XPS spectra of TBC and K-TBC were fitted by peak separation. Figure 4A shows that there are three peaks in the C1s spectrum of TBC, which are 284.8, 285.47 and 288.83 eV respectively, and the corresponding functional groups are C-C/C=C, C-N/C-O and C=O [37]. In its O1s spectrum ( Figure 4B), C=O (531.85 eV) and C-O (533.49 eV) can be observed [38].
The N1s spectrum of TBC in Figure 4C shows that nitrogen is mainly in the form of pyridine nitrogen (398.52 eV) and pyrrole nitrogen (400.72 eV) [39]. After activation by K 2 CO 3 , C-C/C=C and C=O functional groups increased significantly ( Figure 4D), and the presence of a peak of O-C=O (534.57 eV) was observed in the O1s spectrum of K-TBC ( Figure 4E) [40]. In addition, the nitrogen functional groups in Figure 4F also changed significantly, with the proportion of pyrrole nitrogen increasing significantly and the presence of graphite nitrogen (403.32 eV) [41]. The high proportion of graphite nitrogen helps to increase the conductivity of the material, and the pyrrole nitrogen and C=O, C-O and -COOH oxygen-containing groups on the surface of the material are conducive to increasing the wettability of porous active biochar in electrolyte solution, thus increasing the contact surface area of ions and enhancing the interaction between porous biochar and electrolyte [42].   6 ] is the most commonly used analytical reagent in electrochemical methods. Due to its excellent oxidized and reduced properties, K 3 [Fe (CN) 6 ] can be used to indirectly characterize the changes of conductance properties during the preparation of sensors by cyclic voltammetry (CV) and differential pulse voltammetry (DPV). As shown in Figure 5A, the electric and oxidized and reduced current and current area of the modified K-TBC (curve b) are larger than those of the bare electrode (curve a), because BC has the advantages of multiple active sites and large specific surface area in the electrochemical oxidation process, and thus has strong conductive properties. The CV current response after polymerized MIP (curve d) is lower than that before polymerization (curve c), because MIP is a poorly conductive polymer, which hinders the electron conduction of [Fe (CN) 6 ] 3−/4− [43]. The NOR in MIPs eluted with organic solvents formed moleculeimprinted pores that facilitated electron conduction, so the current response of the eluted MIPs (curve e) was stronger than that of the pre-eluted MIPS (curve d). Curves e and c show that the current response of the eluted MIPs is still lower than that of K-TBC/GCE of the unpolymerized MIPs, which indicates that there is still polymer on the surface of the eluted electrode. This is illustrated by the DPV in Figure 5B.

Optimization of Conditions for the Fabrication of the TBC/GCE
The structure and properties of biomass carbon materials vary with the products obtained by different preparation processes and different conditions, including pretreatment methods, carbonization temperature, heating rate, cracking time and gas atmosphere [20,44]. The ratio of biochar to activator, the calcination temperature, the calcination time, the amount of chitosan added, and the loading of TBC were optimized in order to increase the electrical conductivity of TBC/GCE, with the current difference between the electrode before and after modification being measured by CV (−0.2 V-0.6 V, 50 mV·s −1 ) in probe solution as the index. According to the experimental results, when the ratio of biochar to activator is 1:2 ( Figure 6A), the roasting temperature of TBC is 700 • C ( Figure 6B), the roasting time of TBC is 2.5 h (Figure 6C), the addition of chitosan is 0.2% ( Figure 6D), and the loading of TBC is 5 µL (Figure 6E), the prepared TBC/GCE has the highest current response. In addition, with the increase of the concentration of TBC suspension, the response current value also increases, but the increase of the concentration is not conducive to the subsequent binding with the molecular-imprinted copolymer. Therefore, 3 mg/mL TBC suspension concentration was selected to prepare an MIP/TBC/GCE sensor.

Optimization of Conditions for the Fabrication of the MIP/TBC/GCE
It is important to optimize sensor preparation conditions, including background pH of electropolymerization, ratio of template molecule to functional monomer, ratio of composite functional monomer, scanning segment, scanning rate and elution time. The modified electrodes under various circumstances were incubated in 0.1 µM NOR solution for DPV detection in the probe solution.
The interaction between the template molecule and the functional monomer, as well as the structural stability of the membrane bearing the molecular imprint, are both impacted by the pH level of the polymerization solution. Therefore, choosing a suitable pH level electropolymerization solution has a certain influence on the performance of the sensor. Polymerization solutions with pH levels of 3.0, 3.5, 4.0, 4.5 and 5.0 were created to study the effect of pH values on the performance of composite molecularly imprinted membranes. The experimental results show that the current response difference is the largest when pH = 3.5 ( Figure 7A), indicating that this pH is most favorable for the interaction of oPD, oAP and NOR, which is conducive to the formation of more binding sites [26].
An important condition for the preparation of molecularly imprinted electrochemical sensors by electropolymerization is to determine the amount of template molecules and functional monomers. Keeping other conditions unchanged, with a fixed total functional monomer concentration of 30 mM (oPD:oAP = 1:1), the developed sensor response signal was tested at different ratios of NOR and total functional monomer (1: Figure 7B), when the ratio is 5:15:15, the largest current difference, but when the ratio was lower or higher than 5:15:15, the current response decreased significantly, probably because during the polymerization process, if the concentration of functional monomers was low, the bound NOR was less, and the number of imprinted sites decreased. With a high concentration of functional monomers, the copolymer film becomes denser, impeding template molecule removal and adsorption as well as electron transfer on the electrode surface [45]. In addition, the performance of MIP is also affected by the monomer molar ratio. In order to obtain the optimal copolymer for forming imprinted sites, keeping the rest of the conditions the same, a fixed total functional monomer concentration of 30 mM with different molar ratios of monomers (oPD:oAP, 2:1, 3:2, 1:1, 2:3, 1:2) is maintained. Figure 7C is the current difference before and after the electrode adsorption, and it can be seen that the maximum current response is obtained under the monomer with the ratio of 3:2. Both the scan segment and electropolymerization scan rate are significant factors in the manufacturing of MIP/TGC/GCE, affecting the thickness and density of the molecularly imprinted polymer, respectively. It was found in the experiment that the 25 scan segments showed relatively high sensitivity ( Figure 7D), which may be due to the small number of polymerizations, which resulted in fewer imprinted sites for NOR binding, and higher polymerization times would lead to more polymer. A thick film leaves the imprinted site incompletely exposed and affects NOR recombination. The effect of scan rate on electropolymerization is illustrated in Figure 7E. For one thing, at slightly slower scan rates, dense MIP layers are formed, making the removal of NOR to form imprinted sites less feasible. For another thing, the non-compact and rough MIP layers formed at higher scan rates may be affected by the eluent.
The removal of all template molecules is a critical step in the fabrication of molecularly imprinted electrochemical sensors. The elution time was investigated by immersing the MIP sensor in NaOH (0.1 mol·L −1 ) and 100% ethanol (V:V = 1:3) solution for different times. Figure 7F depicts the elution profile of MIP/TGC/GCE. It is not difficult to see that as the elution time increases, the current increases evenly until it stabilized after more than 12 min. This indicates that the template molecule has been entirely removed from the MIP. Thus, 12 min is the optimal time for template removal.

Optimization of Conditions for the Fabrication of the MIP/TBC/GCE
The same concentration of NOR solution was used to study the effect of incubation time and pH on the electrochemical performance of the sensor. As shown in Figure 8A, as the pH increased from 3.0 to 7.0, the peak current difference first increased and then decreased, and the maximum peak current difference (∆I) was obtained at pH 5.0, which was considered to be the optimum pH for the experiment. The modified electrodes were incubated in 0.1 µM NOR solution for different times, and DPV detection was performed in K 3 [Fe (CN) 6 ] solution. Figure 8B shows that as absorption time increases, so does the peak current difference, and the current is almost stable after 4 min. This trend indicates that the adsorption capacity has reached its limit. As a result, an incubation time of 4 min is ideal.

Linear Range and Limit of Detection
To monitor the effectiveness of the constructed sensor, the MIP/TBC/GCE sensor was tested for different concentrations of NOR solutions (0.1, 0.3, 0.5, 1, 3, 5, 10, 30, 50, 100 nM) under optimal experimental conditions. DPV experiments were performed and Figure 9A shows that the peak current decreases with increasing NOR concentration. While Figure 9B further indicates that a two-stage linear association between ∆I and NOR concentration. In the low concentration NOR solution range (0.1-0.5 nM), the regression equation is ∆I (µA) = 9.76136 c (nM) + 14.56717 (R 2 = 0.99893), and at high concentration (0.5-100 nM), the regression equation is ∆I (µA) = 0.11396 c (nM) + 24.332642 (R 2 = 0.99235), ∆I is the difference in current between before and after NOR adsorption by the MIP/TBC/GCE sensor, and c represents the concentration of NOR standard solution. This indicates that the local concentration on the sensor surface is rapidly depleted at low NOR solution concentration, resulting in a high sensor response sensitivity, while saturation occurs at high concentration conditions, and the kinetic adsorption rate is significantly reduced, resulting in a decrease in the slope to 0.11396 [46]. When calculated according to formula 3S b /K, the limit of detection (LOD) is about 0.028nM, where S b is the standard deviation of the blank value's six measured values (∆I) (NOR concentration close to the blank value but still current response), and K is the calibration curve's slope for low concentrations.

Selectivity, Reproducibility and Stability
Shape, size and functional group are the most critical factors for the recognition of template molecules by the imprinted cavity [47]. Therefore, it is useful to examine enrofloxacin (ENR), ciprofloxacin (CIP), and danofloxacin (DAF), which are similar in structure to NOR, then ampicillin (AMP), tetracycline (TCY), oxytetracycline (OCY), chloramphenicol (CHL) and thiamphenicol (THI) -which are quite different from NOR in structure-to study selectivity. As shown in Figure 10A, among different antibiotics at the same concentration, the sensor exhibited the highest current response to NOR, and a slight current response to structural analogs, which were slightly higher than those of non-structural analogs. It shows that the sensor has good recognition specificity and selectivity for NOR. Interference experiments were conducted to study the effects of common components such as metal ions, carbohydrates and amino acids in food matrices on the detection of NOR by MIP/TBC/GCE sensors. When the change of ∆I is less than 5%, it can be considered that there is no significant impact on the test object [48,49]. The results are shown in Table 1. When the concentration of NOR is 0.1 µM, high concentrations of Na + , K + , Ca 2+ , Fe 2+ , Al 3+ , NO 3 − , Cl − , as well as glucose, phenylalanine, cysteine and lysine, etc, are seen. The ∆I of NOR was not significantly affected, and the relative errors were all less than ±4.1%. The reproducibility and stability of the MIP/TBC/GCE sensor were investigated by the change in ∆I in 0.1 µM NOR standard solution. The relative standard deviation (RSD) for different sensors (n = 5) and the same sensor (n = 5) were 0.92% and 1.36%, respectively ( Table 2). According to the above results, the MIP/TBC/GCE sensor exhibited satisfactory reproducibility. To study the stability of the constructed sensor, the MIP/TBC/GCE sensor was stored for 0, 5, 10, 15, 20, and 25 days for the detection of the same concentration of NOR, and the ∆I was recorded by DPV. After preparation, the sensor was placed in a nitrogen-filled airtight bag and stored in a 4 • C refrigerator. As the storage days increased, the peak current of the modified electrode for detecting NOR decreased, and on the 25th day, the peak current was 84.63% of the initial current ( Figure 10B), indicating that the developed MIP/TBC/GCE sensor has excellent stability. The decrease in current response may be due to the damage of the recognition site due to oxidation of the film, or the film peeling off the electrode surface due to poor adhesion of the film [50].

Application of Real Sample Detection
Three animal-derived foods, milk, honey and pork, which are prone to NOR residues in actual production, were selected as samples for NOR detection to evaluate the applicability of the sensor in practical applications. It can be seen from Table 3 that no NOR residues were detected in the commercially available milk, honey and pork samples, indicating that the samples were not contaminated by NOR. At the same time, the standard addition method was used for quantitative analysis, and the recoveries of milk samples were 95.2-102.0%, the recoveries of honey samples were 96.6-102.0%, and the recoveries of pork samples were 85.9-101.7%. All results showed that the recovery and precision of this sensor were ideal and it is feasible to detect NOR in milk, honey and pork samples.  Comparing the MIP/TBC/GCE sensor prepared in this study with published NOR detection methods in Table 4, it can be noted that the proposed sensor exhibits higher sensitivity than most other sensors and its incubation time was at a medium level. MIP/Cu-N-C/GCE DPV drug capsules 1 × 10 −8 -5 × 10 −7 , 5 × 10 −7 -1 × 10 −4 3.33 × 10 −9 1.5 [50] Sensor in this work DPV milk, honey and pork 1 × 10 −10 -5 × 10 −10 , 5 × 10 −10 -1 × 10 −7 2.8 × 10 −11 4 This work

Conclusions
In summary, a simple and low-cost biochar molecular-imprinted electrochemical sensor fabricated from discarded tea branch precursors was constructed for the detection of NOR. Tea branch biochar provides porous structure and large specific surface area to enhance the conductivity of the sensor. A molecular-imprinted layer provides specific recognition ability to further improve the sensitivity of the sensor, which makes the lower limit of detection (LOD) 0.028 nM (S/N = 3) and the linear range wider (0.1-100 nM). At the same time, it also shows excellent performance, including short detection time, good selectivity and stability. In addition, the sensor was able to detect NOR in milk, honey and pork, with recovery rates between 85.90% and 101.71%, and RSD less than 5%, which proved its potential in the detection of NOR residues in animal-derived foods. Efforts should be made to commercialize the sensor and provide a new idea for food quality monitoring.

Conflicts of Interest:
The authors declare no conflict of interest.