A screen-printed carbon electrode modified with gold nanoparticles, poly(3,4-ethylenedioxythiophene), poly(styrene sulfonate) and a molecular imprint for voltammetric determination of nitrofurantoin

A molecularly imprinted polymer (MIP) and a nanocomposite prepared from gold nanoparticles (AuNP) and poly(3,4-ethylenedioxythiophene)/poly(styrene sulfonate) (PEDOT:PSS) were deposited on a screen-printed carbon electrode (SPCE). The nanocomposite was prepared by one-pot simultaneous in-situ formation of AuNPs and PEDOT:PSS and was then inkjet-coated onto the SPCE. The MIP film was subsequently placed on the modified SPCE by co-electrodeposition of o-phenylenediamine and resorcinol in the presence of the antibiotic nitrofurantoin (NFT). Using differential pulse voltammetry (DPV), response at the potential of ~ 0.1 V (vs. Ag/AgCl) is linear in 1 nM to 1000 nM NFT concentration range, with a remarkably low detection limit (at S/N = 3) of 0.1 nM. This is two orders of magnitude lower than that of the control MIP sensor without the nanocomposite interlayer, thus showing the beneficial effect of AuNP-PEDOT:PSS. The electrode is highly reproducible (relative standard deviation 3.1% for n = 6) and selective over structurally related molecules. It can be re-used for at least ten times and was found to be stable for at least 45 days. It was successfully applied to the determination of NFT in (spiked) feed matrices and gave good recoveries.Schematic representation of a voltammetric sensor for the determination of nitrofurantoin. The sensor is based on a screen-printed carbon electrode (SPCE) modified with an inkjet-printed gold nanoparticles-poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) nanocomposite and a molecularly imprinted polymer.

Antibiotic; Biomimetic sensor; Electrochemical sensor; Inkjet-printed electrode; Molecularly imprinted polymer; Nanocomposite

Nitrofurantoin belongs to a group of broad-spectrum antimicrobial drugs called nitrofurans. They are widely used as feed additives for growth promotion, as well as for prevention and treatment of bacterial and protozoan infections in animal husbandry and aquaculture [[1] ]. After ingestion by animals, nitrofurans are rapidly absorbed and distributed into body fluids. Some are found to be excreted intact in urine which might become water and soil contaminants [[2] ]. Some are broken down into their corresponding metabolic products and bind to tissue proteins [[3] ]. Previous studies in animals have shown suspected potential carcinogenic and mutagenic risks of these metabolites [[4] ]. Furthermore, these toxic tissue-bound residues can bind and persist in animal tissues for months and can also be transferred to secondary species through the ingestion of contaminated animal products [[5] ]. As a result, nitrofurans have been banned from use in food-producing animals by the European Union, USA and China as well as many other countries [[6] ]. However, these drugs are still being illicitly used across the world. To control the illegal use, one can monitor either the intact compound in the animal feeds or the protein-bound metabolite in the animal products. Since the metabolite is covalently bound, it is not readily available for examination. Additional steps such as acid digestion and derivatization of the releasing compound are therefore needed. In contrast, determination of the parent drug frankly in the feed samples would be an alternative means to shorten and simplify the sample preparation process and to straightforwardly monitor and control the abuse of the drug.

Various analytical methods including luminescence spectroscopy [[7] , [8] ], electrochemistry [[9] , [10] ] and chromatography [[11] , [12] ] have been described for the determination of nitrofurantoin. LC/MS and LC/MS/MS provide high-precision mass detection with a low limit of quantification. However, they are considerably bulky and costly, making them inapplicable for on-site screening. Additionally, they also require time-consuming sample pretreatment steps including extraction, enrichment and clean-up. To address these challenges, ready-to-use test kits and sensors are highly desired by offering a portable size, effective cost, fast response and straightforward and quantitative analysis. Enzyme-linked immunosorbent assay (ELISA) kits have been successfully launched to the market for analysis of nitrofuran metabolites in animal products with considerably low detection limits and cross-reactivity [[6] , [13] ]. Unfortunately, neither a kit nor an antibody for the intact form of nitrofurantoin is available.

Among various existing sensor devices, electrochemical-based sensors have gained tremendous interest because they provide certain benefits of portable size, high sensitivity, cost-effectiveness and simple instrumentation. However, previously reported electrochemical sensors still did not afford sufficient discrimination between each derivative of nitrofuran drugs. To improve selectivity of these electroanalytical sensors, synthetic molecularly imprinted polymers (MIPs) are of high interest [[14] ]. Besides their specific recognition ability, MIPs also offer exceptional stability, low production cost and a rapid fabrication process. To develop more sensitive devices, incorporation of nanomaterials in electrochemical MIP sensors has become an interesting option [[15] -[22] ].

Nitrofurantoin-binding MIPs were previously reported by Athikomrattanakul et al. as ground monolith particles for thermometric sensing [[23] -[25] ]. Since nitrofurantoin itself is not stable under heat treatment and UV exposure, a heat-stable analogue of nitrofurantoin, carboxyphenyl aminohydantoin (CPAH), was used as a template instead. To use nitrofurantoin as a direct template, conventional polymerization processes which are normally induced by light or heat are not applicable. Electropolymerization, which offers milder ambient conditions, seems to be a more promising method for this reason. Using electropolymerization, homogeneous films can be directly coated in a precise area on the electrode surface with finely tuned film thickness and morphology. These beneficial features make electrosynthesized MIP films a promising platform for sensor applications [[26] , [27] ].

Screen-printed carbon electrodes (SPCEs) have been broadly exploited in the field of electrochemical sensors due to their large-scale production feasibility with reasonable cost and good reliability. They also hold great promise for on-site screening in terms of field portability and a dramatic reduction in sample volume [[28] ]. Various materials including carbonaceous nanomaterials, metal nanoparticles and conducting polymers have been widely used to improve the sensing performance of SPCEs [[29] -[31] ]. Among them, gold nanoparticles (AuNPs) have been extensively studied due to their particular properties such as a large surface area, high catalytic activity and effective electron transport [[32] ]. Of all known conducting polymers, PEDOT:PSS is often used for fabricating electrochemical sensors owing to its high electrical conductivity and exceptional stability. These materials can be used individually or in combination with each other as high-performance nanocomposites. Conventional electrode modification processes such as drop casting, physical attachment, dip coating, etc., often lead to the material peeling off and consequently suffer from instability and poor reproducibility. In contrast, electrode modification using a material inkjet printer has attracted a considerable amount of attention by offering a high-precision manufacturing process with greater reproducibility [[33] , [34] ].

According to a literature survey, there are no previous reports on an electrochemical MIP sensor for nitrofurantoin. Therefore, we demonstrate a strategy to prepare the first MIP-based electrochemical sensor for the sensitive and selective detection of nitrofurantoin. We herein combined the outstanding electrochemical characteristic of the AuNP-PEDOT:PSS nanocomposite with high selectivity of the MIP as a hybrid bilayer. Firstly, the AuNP-PEDOT:PSS layer was inkjet-printed onto an SPCE. Thereafter, a thin MIP film was prepared through the electrocopolymerization of o-phenylenediamine (o-PD) and resorcinol in the presence of nitrofurantoin directly on top of the nanocomposite electrode (Scheme 1). The principle behind this electrochemical detection is that upon the rebinding of NFT to the imprinted sites, the number of cavities available for the redox probe to reach the electrode surface underneath decreases, and consequently the signal drops off. Decrease of the K3[Fe(CN)6]/K4[Fe(CN)6] signal depends linearly on the amount of bound nitrofurantoin in the sample solution. As the imprinted polymer can selectively capture the target analyte from the complex matrix, interfering signals will be diminished. Furthermore, no additional extraction, enrichment or sample clean-up steps are needed. This also means that the process is less time-consuming.Schematic representation of the electrochemical MIP sensor for nitrofurantoin (NFT) detection: a Preparation of electrosynthesized MIP on the nanocomposite modified screen-printed carbon electrode (SPCE); b Indirect voltammetric determination of NFT making use of K3[Fe(CN)6]/K4[Fe(CN)6] as electrochemical probe

Nitrofurantoin, nitrofurazone and furazolidone were acquired from TCI chemicals. o-Phenylenediamine and resorcinol were obtained from Acros. Hydrogen tetrachloroaurate(III) trihydrate (HAuCl4·3H2O), 3,4-ethylenedioxythiophene (EDOT), sodium-4-polystyrenesulfonate (Na-PSS), potassium hexacyanoferrate(III) (K3Fe(CN)6), potassium hexacyanoferrate(II) trihydrate (K4Fe(CN)6·3H2O), acetic acid, nifuroxazide, furaltadone, nifurtimox, N-benzyl-2-nitro-1H-imidazole-1-acetamide and pimonidazole were obtained from Sigma-Aldrich (http://www.sigmaaldrich.com). All buffer salts and organic solvents were purchased from Merck (http://www.merckmillipore.com). SPCE electrodes were fabricated in-house at the Thai Organic and Printed Electronics Innovation Center (TOPIC) using carbon graphite and silver/silver chloride pastes from Gwent Group (http://www.gwent.org).

The nanocomposite ink was printed onto SPCEs by a material inkjet printer from FUJIFILM (Dimatix Materials Printer, DMP-2800, http://www.fujifilm.com). The morphology of the AuNPs dispersed in the solution was examined by a transmission electron microscope (TEM; JEOL JEM-2010, http://www.jeol.com). The surface morphology and functional groups of the printed composite film were characterized by a scanning electron microscope (SEM; Hitachi S-3400 N, http://www.hitachi-heightech.com) and a Fourier transform infrared spectrometer (Perkin Elmer System 2000, http://www.perkinelmer.com), respectively. XRD measurement was carried out on an X-ray diffractometer (PANalytical X’Pert PRO, http://www.panalytical.com). All electrochemical measurements were conducted using a potentiostat from Palmsens (http://www.palmsens.com).

A three-electrode configuration is comprised of a working screen-printed carbon electrode printed from a carbon-based paste, a silver/silver chloride pseudo-reference electrode from a silver/silver chloride paste and a counter electrode from a carbon paste. To prepare a nanocomposite, EDOT was dissolved in 1% (w/v) Na-PSS solution. Subsequently, HAuCl4 solution was slowly added and the reaction mixture was stirred for 12 h. Afterward, five layers of the nanocomposite ink were printed onto the working area of the SPCE in which each printed layer was dried at 60 °C for 15 min prior to subsequent printing steps.

The imprinted electrodes were prepared in a mixture solution of 5 mM o-PD, 5 mM resorcinol and 2.5 mM NFT in 100 mM phosphate buffer (pH 7.0) by cyclic voltammetry sweeping between 0 and 1.2 V at a scan rate of 50 mV·s−1 for 20 scans. Thereafter, the template was removed by incubating the electrodes in a mixture of methanol-water-acetic acid (8:1:1, v/v/v) at room temperature for 15 min. Non-imprinted electrodes were prepared in the same manner, but in the absence of the template.

A stock solution of NFT (10 mM) was prepared in DMSO and standard solutions were prepared by appropriate dilution of the stock solution with phosphate buffer (pH 7.0). The imprinted electrodes were allowed to interact with the sample for 12 min. Afterward, the electrodes were rinsed and replaced in 5 mM K3[Fe(CN)6]/K4[Fe(CN)6] solution containing 100 mM KCl in a 2 mL homemade electrochemical cell and the electrochemical measurements were conducted. Cyclic voltammetry (CV) measurements were conducted by sweeping the potential from −0.3 V to 0.5 V at a scan rate of 50 mV·s−1 and differential pulse voltammetry (DPV) measurements were recorded by scanning the potential from −0.3 to 0.6 V at a scan rate of 10 mV·s−1.

Feed samples were obtained from retail markets in Thailand. To the thoroughly minced sample (2 g), a sufficient volume of NFT stock solution and methanol (20 mL) were added. The mixture was sonicated for 20 min and centrifuged for 10 min at 3000 rpm. The supernatant was filtered and the filtrate was diluted with phosphate buffer (pH 7.0) to obtain the sample for detection.

Conducting polymers including polypyrrole (PPy), polyaniline (PANI) and PEDOT:PSS have been widely applied to improve electrochemical performances of bare electrodes. In this work, PEDOT:PSS is chosen as a material of choice because it has greater thermal and electrochemical stability than polypyrrole and polyaniline. This can be ascribed to the presence of ethylenedioxy bridging group in the 3 and 4 positions on its thiophene ring. To further enhance the electrochemical performances of PEDOT:PSS, nanomaterials have been introduced into the aforementioned matrix. In comparison to the carbon-based nanomaterials (i.e., carbon nanotube, graphene, carbon dot), the preparation of metallic nanoparticles is more facile. Especially in the case of the preparation of metal nanoparticles incorporated in conducting polymer, a simultaneous in situ formation of both components can be achieved within a single step. Among the more commonly used metal nanoparticles, gold is the most widely studied metal due to its high catalytic activity which facilitates the electron transfer and decrease the electrode overpotentials. Therefore, AuNP-PEDOT:PSS is chosen herein as a promising electrode modifying-material.

In this work, o-phenylenediamine and resorcinol are chosen due to their ability to create arrays of non-covalent interactions with the template molecule through their diamine, diol and aromatic moieties. This can be realized by a computational simulation. The calculation revealed that nitrofurantoin formed a stable 1:2 complex with o-phenylenediamine and resorcinol with the lowest interaction energy of −126.68389760 kJ·mol−1 (Fig. S1). It also showed that H-bonding and π-π stacking are the principal interactions within the ternary complex, and these interactions will be responsible for producing the microenvironment for the recognition of the target molecule.

A basic way to generate signal in the MIP-based electrochemical sensors is through the direct signal generation from the electro-active target itself. However, this method is only limited to the target molecules which are electro-active. On the other hand, the signal generation can also be based on the “gate effect” in which the electrode is covered with the non-conducting MIP layer with the imprinted cavities as the controlled gate. In this case, the signal comes from a redox probe which is typically [Fe(CN)6]3−/4-. Binding of the analytical target to the MIP film impedes the transfer of the electro-active probe to the underlying electrode surface, thereby reducing the current signal. In contrast to the direct method, this approach is more versatile and theoretically feasible to detect any substances regardless of their electroactivity. As a result, it is often applied in the MIP-based electrochemical sensors. It is important to note that when the indirect method is used to detect the electro-active molecule, the signal generated by the analytical target should not interfere with the measuring signal of the redox probe. In this case, nitrofurantoin exhibits reduction peaks below −0.8 V (at the neutral pH of 7.0) [[35] ] which are not in the potential range used for measuring the oxidation signal of the redox marker. Furthermore, it is also known that PEDOT in the oxidized state is conductive but it turns electrically insulating after being reduced by applying negative potentials. This may limit the application of PEDOT in the negative potential window only until the potential of −0.5 V [[36] , [37] ]. As a result, the indirect method is considered to be a suitable choice in this work.

An aqueous dispersion of AuNP-PEDOT:PSS was prepared by simultaneous formation of PEDOT and AuNPs via the in-situ oxidative polymerization and the in-situ reduction of HAuCl4. This can be directly observed by a gradual change in solution color from yellow to a dark blue solution. To further confirm the presence of PEDOT in the nanocomposite, an FTIR spectrum was recorded for a scan range of 3800 to 600 cm−1. Relevant data and spectrum are given in the Electronic Supporting Material (Fig. S2). As shown in the TEM image (Fig. S3), AuNPs were clearly observed as dark spots. XRD was also used to characterize the chemical composition and crystallographic structure of the AuNP-PEDOT:PSS nanocomposite and its pattern is shown in Fig. S4. The diffraction peaks at 38.29°, 44.32°, 64.46° and 77.62° were assigned to the (111), (200), (220) and (311) planes of the cubic lattice structure of Au.

The resulting aqueous dispersion of AuNP-PEDOT:PSS was used to modify the working area of the SPCE by a material inkjet printer. The morphology of the resulting electrode was characterized using SEM. As illustrated in Fig. S5a, a smooth film with consistently dispersed bright nanoparticles was clearly observed. This surface morphology was completely different from that of the bare SPCE (Fig. S5b), implying the successful deposition of the composite film. Using EDS elemental mapping, these bright spots were assigned to metallic gold as shown in Fig. S6.

CV measurements were used to characterize the modified and unmodified electrodes. Related data and voltammograms are given in the Electronic Supporting Material (Fig. S7). The results indicated that the AuNP-PEDOT:PSS modified SPCE possessed the highest catalytic activity in comparison to the PEDOT:PSS/SPCE and the unmodified electrode. This is possibly due to a profound synergistic effect between AuNP and PEDOT.

Similar voltammograms were obtained during the electropolymerization in the presence (MIP) and absence (NIP) of the template (Fig. S8). A prominent decrease in the current intensity under successive scans was observed. This evidently suggested that a non-conducting polymer was progressively formed and gradually covered the electrode surface, yielding to the suppression of the voltammetric response. The electrochemical behavior of the MIP modified electrode after electropolymerization, template removal and rebinding was monitored using K3[Fe(CN)6]/K4[Fe(CN)6] (Fig. 1). In comparison to the AuNP-PEDOT:PSS electrode, the response of the redox probe after electropolymerization was almost suppressed. This was explained by the fact that the non-conductive film fully covered the electrode surface. Hence, the signaling probe cannot access the electroactive area. The template removal was performed in a mixture of methanol-water-acetic acid. Under aqueous acidic conditions, the azomethine bond (-N=CH-) of NFT was labile and readily hydrolyzed into 5-nitrofurfural and 1-aminohydantoin [[38] ]. After extracting these molecules from the polymeric matrix, the imprinted cavities were created so that the redox signal reappeared. This signal was again suppressed after rebinding of NFT to the imprinted cavities of the MIP layer.Cyclic voltammograms of 5 mM K3[Fe(CN)6]/K4[Fe(CN)6] in 100 mM KCl recorded between −0.3 V and 0.5 V (vs. Ag/AgCl) at a scan rate of 50 mV·s−1 for (a) AuNP-PEDOT:PSS/SPCE, b MIP electrode before template removal, c MIP electrode after template removal and (d) MIP electrode after rebinding

Surface morphologies of the prepared MIP and NIP films after the template removal were visualized by SEM (Fig. 2). In comparison to the relatively compact NIP layer, the MIP film revealed a slightly rougher surface, which might be ascribed to the successful deposition and extraction of the template molecule in and from the MIP layer.SEM images of (a) MIP and (b) NIP modified AuNP-PEDOT:PSS/SPCE

The following parameters were optimized: (a) incubation time; (b) incubation pH of rebinding buffer; (c) regeneration time. Respective data and figures are given in the Electronic Supporting Material (Fig. S9). The following experimental conditions were found to give best results: (a) incubation time of 12 min; (b) pH of rebinding buffer of 7.0; (c) regeneration time of 15 min.

Under optimal conditions, DPV responses of the MIP electrodes recorded in 5 mM K3[Fe(CN)6]/K4[Fe(CN)6] from −0.3 V to 0.6 V (vs. Ag/AgCl) at a scan rate of 10 mV·s−1 after incubation in the samples containing NFT at different concentrations are shown in Fig. 3a. Due to the rebinding of NFT to the imprinted cavities, the peak current of the redox probe decreased as the concentration of NFT increased. As depicted in Fig. 3b, a linear relationship between the current change (∆I) and NFT concentration was observed from 1 nM to 1000 nM with a correlation coefficient (R2) of 0.997 and a detection limit of 0.1 nM based on the signal-to-noise characteristics (S/N = 3). The electrochemical sensitivity of the detection method in term of the slope of the calibration curve was found to be 1203 μA·μM−1·cm−2. To demonstrate the enhancement effect of the AuNP-PEDOT:PSS nanocomposite on the sensor performance, a control sensor (MIP/SPCE) was prepared by direct growth of a MIP layer on top of the bare SPCE. As depicted in Fig. S10, a linear relationship between the relative current change and the concentration was obtained from 75 nM to 500 nM with R2 of 0.992 and the sensitivity of 606 μA·μM−1·cm−2. The resulting concentration range was narrower than that observed on the MIP/AuNP-PEDOT:PSS/SPCE. Furthermore, a limit of detection of the control MIP/SPCE was found to be 50 nM, which is two orders of magnitude higher than that obtained from the MIP/AuNP-PEDOT:PSS/SPCE. A better performance of the MIP/AuNP-PEDOT:PSS/SPCE can be ascribed to the enhancement effect of the nanocomposite layer, which can efficiently improve the electron-transfer rate, and eventually amplify the sensor signal.a Differential pulse voltammograms of MIP/AuNP-PEDOT:PSS/SPCE recorded in 5 mM K3[Fe(CN)6]/K4[Fe(CN)6] at scan range from −0.3 V to 0.6 V (vs. Ag/AgCl) and scan rate of 10 mV·s−1 after incubation in different concentrations of NFT. b Calibration plot of NFT at the MIP/AuNP-PEDOT:PSS/SPCE. (Inset: the magnified calibration plot at the low concentration of NFT) The current responses were evaluated at the potential of ~ 0.1 V (vs. Ag/AgCl). Error bars represent standard deviations from triplicate measurements

To identify the MIP affinity, the binding isotherm was fitted using the Langmuir isotherm model (Fig. S11). The affinity constant (Ka) of the MIP sensor was found to be 1.73 × 106 M−1, which is four orders of magnitude greater than that of the former MIP-based thermistor (1.40 × 102 M−1) reported by Athikomrattanakul et al. [[25] ]. The significant difference in binding affinity of the two MIP-based sensors reveals the effectiveness of the present work in creating a high affinity synthetic receptor for nitrofurantoin.

Cross-reactivity of the imprinted electrode was evaluated using structurally related nitro compounds including nitrofurazone (NFZ), furazolidone (FZD), furaltadone (FTD), nifuroxazide (NFX), nifurtimox (NFM), pimonidazole (PMZ) and benznidazole (BNZ) (Fig. S12). The current change (∆I) of the sensor toward different analytes at a concentration of 1 μM was measured and is shown in Fig. 4. For all the analytes, the current response of the non-imprinted sensor was almost equal and changed only slightly, and this change can be ascribed to the non-specific binding of the compounds to the polymeric matrix. The current response of the imprinted sensor toward NFT was higher than those of the other molecules, clearly due to its perfect match to the imprinted cavities in size, shape and orientation of functionalities. Although FTD, NFX and NFM hold a nitrofuran motif similar to NFT, their molecular size is considerably too large and bulky in comparison to the imprinted cavities. PMZ and BNZ, on the other hand, contain neither nitrofuran nor hydantoin rings which can interact with the conserved functional groups within the imprinted site. As a result, they were bound to the MIP sensor in considerably lower amounts than NFT. In the case of FTD, NFX, NFM, PMZ and BNZ, the higher degree of surface roughness of the MIP would be responsible for a slightly higher binding capacity of the MIP in comparison to the control NIP, rather than a real imprinting effect. In contrast to those control compounds, NFZ and FZD bind preferentially to the MIP sensor because they have a higher degree of structural resemblance to the template molecule. Additionally, their molecular size is also smaller than that of the template, making them fit readily into the imprinted pocket. However, the amount of NFZ and FZD bound was still not as high as that of NFT. This is possibly due to the lack of a hydantoin ring, one of the crucial parts that can interact with the functional monomer via double hydrogen bonds in their molecular structures (Fig. S1).Selectivity of the MIP/AuNP-PEDOT:PSS/SPCE. Current change (∆I) at the potential of ~ 0.1 V (vs. Ag/AgCl) of the MIP and NIP modified electrodes in 5 mM K3[Fe(CN)6]/K4[Fe(CN)6] after incubation in NFT and other structurally related compounds (at 1 μM). Error bars represent standard deviations from triplicate measurements

The inter-electrode reproducibility of the sensor was evaluated using six independent MIP electrodes fabricated using the same procedure. These electrodes were used to determine NFT at 1 μM. The current response showed a relative standard deviation (RSD) of 3.1% for six independent measurements, indicating excellent sensor-to-sensor reproducibility. To investigate the repeatability, a methanol-water-acetic acid mixture solution was used to regenerate the MIP sensor. The same MIP sensor was used to determine 1 μM NFT for ten repeated times with regeneration in between. The RSD of the current change (∆I) was found to be 2.9% (Table S1), representing good reusability of the established MIP sensor for at least ten times. The stability of the MIP sensor was investigated by storing the electrode at 4 °C under an N2 atmosphere for 45 days. Then, the sensor was used to detect NFT at a concentration of 1 μM. The response remained at 93.0% of its initial value, suggesting good stability of the prepared MIP sensor.

To identify its practical applications, the prepared MIP sensor was used to determine NFT in real animal feed samples. As shown in Table 1, the recoveries at three concentration levels of each sample were found to be in the range of 95.0-104.0% and the RSDs were less than 3.2%. LC/MS/MS was used as a reference method to validate the reliability of the sensor. The results obtained by both methods were consistent with each other as shown in Table S2. This indicates that the prepared sensor can be used for reliable determination of NFT in real samples.

The analytical performances in terms of the linear range and detection limit of the established MIP sensor were compared with other previously reported electrochemical sensors (Table S3). Most of them relied on mercury based electrodes (i.e., mercury modified silver solid amalgam and hanging mercury drop). Others were based on boron doped diamond film (BDDF), carbon nanofiber (CNF) and DNA-poly(5-amino-2-mercapto-1,3,4-thiadiazole) modified electrodes. As shown in Table S3, our MIP sensor shows a lower limit of detection and a much wider linear response in comparison to those electrochemical sensors reported so far. The improvement in sensor performances can be attributed to the incorporation of the conductive AuNP-PEDOT:PSS nanocomposite into the MIP sensor. It is also worth mentioning that the MIP based screen-printed carbon electrode provides relatively low cost and better portability in comparison to most of those reported electrode materials. Further comparison to other analytical methods can be found in Table S4. The present MIP based electrochemical sensor shows much better limit of detection in comparison to the MIP-based thermistor, luminescence assays and HPLC. It is also smaller in size and less costly. Although the limit of detection of the established sensor is comparable to that obtained from the ELISA-based technique, the MIP sensor offers a broader linear range of detection and can be reused. These advantages make the established MIP sensor a good analytical method for NFT detection.

In this work, the first electrochemical MIP sensor for NFT was successfully established. This can overcome such a limitation of the enzyme-linked immunosorbent assay (ELISA) where neither a kit nor an antibody for the parent form of nitrofurantoin is available until currently. The functional MIP layer was prepared by co-electrodeposition of o-PD and resorcinol in the presence of NFT on top of the AuNP-PEDOT:PSS modified SPCE. Due to a synergistic effect of AuNP and PEDOT:PSS, a great enhancement in current signal was attained, and this apparently lowered the detection limit down to 0.1 nM. In addition to its exceptional limit of detection, the established sensor also shows good selectivity, reproducibility and rather good stability. The sensor can be reused at least 10 times and can be successfully used to determine NFT in spiked feed matrices with good recoveries. Additionally, such a sensor also requires no multiple tedious steps for sample pretreatment. However, it should be noted that the established MIP sensor is only applicable for the determination of the intact form of NFT, not its protein bound metabolite. As a result, application of this sensor in food control is only limited to feed samples rather than contaminated food products (e.g., meat, fish, shrimp, egg, honey, milk).

The online version of this article (10.1007/s00604-018-2797-3) contains supplementary material, which is available to authorized users.

The authors gratefully acknowledge the financial support of the Thailand Research Fund (TRF, Grant No. 5880243), the Kasetsart University Research and Development Institute (KURDI) and the Faculty of Science, Kasetsart University. D. Dechtrirat also wishes to thank Specialized Center of Rubber and Polymer Materials for Agriculture and Industry (RPM), Faculty of Science, Kasetsart University for the publication support.

The authors declare no conflict of interest.

1 Vass M, Hruska K, Franek M, Nitrofuran antibiotics: a review on the application, prohibition and residual analysis, Vet Med (Praha), 2008, 53, 469, 500, 10.17221/1979-VETMED

• 2 Manzetti S, Ghisi R, The environmental release and fate of antibiotics, Mar Pollut Bull, 2014, 79, 7, 15, 10.1016/j.marpolbul.2014.01.005
• 3 Nouws JFM, Laurensen J, Postmortal degradation of furazolidone and furaltadone in edible tissues of calves, Vet Q, 1990, 12, 56, 59, 10.1080/01652176.1990.9694243
• 4 Jung R, Le JY, Wengenmayer F, Wolf E, Kramer M, Mutagenicity studies of a carcinogenic nitrofuran and some analogues, Biomed Biochim Acta, 1985, 44, 485, 492
• 5 Gottschall DW, Wang R, Depletion and bioavailability of [C-14] furazolidone residues in swine tissues, J Agric Food Chem, 1995, 43, 2520, 2525, 10.1021/jf00057a037
• 6 Jiang W, Luo P, Wang X, Chen X, Zhao Y, Shi W, Wu X, Wu Y, Shen J, Development of an enzyme-linked immunosorbent assay for the detection of nitrofurantoin metabolite, 1-amino-hydantoin, in animal tissues, Food Control, 2012, 23, 20, 25, 10.1016/j.foodcont.2011.05.014
• 7 Zhang F, Yao H, Chu T, Zhang G, Wang Y, Yang Y, A lanthanide MOF thin-film fixed with Co, 3O, 4 nano-anchors as a highly efficient luminescent sensor for nitrofuran antibiotics, Chem Eur J, 2017, 23, 10293, 10300, 10.1002/chem.201701852
• 8 Wang Y, Chen T, Zhuang Q, Ni Y, Label-free photoluminescence assay for nitrofurantoin detection in lake water samples using adenosine-stabilized copper nanoclusters as nanoprobes, Talanta, 2018, 179, 409, 413, 10.1016/j.talanta.2017.11.014
• 9 Krejčová Z, Barek J, Vyskočil V, Voltammetric determination of nitrofurantoin at a mercury meniscus modified silver solid amalgam electrode, Electroanalysis, 2015, 27, 185, 192, 10.1002/elan.201400410
• 10 Aydoğdu G, Günendi G, Zeybek DK, Zeybek B, Pekyardımcı Ş, A novel electrochemical DNA biosensor based on poly-(5-amino-2-mercapto-1,3,4-thiadiazole) modified glassy carbon electrode for the determination of nitrofurantoin, Sensors Actuators B Chem, 2014, 197, 211, 219, 10.1016/j.snb.2014.02.083
• 11 Chu PS, Lopez MI, Determination of nitrofuran residues in milk of dairy cows using liquid chromatography−tandem mass spectrometry, J Agric Food Chem, 2007, 55, 2129, 2135, 10.1021/jf063391v
• 12 Viñas P, Campillo N, Carrasco L, Hernández-Córdoba M, Analysis of nitrofuran residues in animal feed using liquid chromatography and photodiode-array detection, Chromatographia, 2007, 65, 85, 89, 10.1365/s10337-006-0112-7
• 13 Wang Q, Liu YC, Chen YJ, Jiang W, Shi JL, Xiao Y, Zhang M, Development of a direct competitive chemiluminescent ELISA for the detection of nitrofurantoin metabolite 1-amino-hydantoin in fish and honey, Anal Methods, 2014, 6, 4414, 4420, 10.1039/C4AY00487F
• 14 Wulff G, Forty years of molecular imprinting in synthetic polymers: origin, features and perspectives, Microchim Acta, 2013, 180, 1359, 1370, 10.1007/s00604-013-0992-9
• 15 Alizadeh T, Amjadi S, Indirect voltammetric determination of nicotinic acid by using a graphite paste electrode modified with reduced graphene oxide and a molecularly imprinted polymer, Microchim Acta, 2017, 184, 2687, 2695, 10.1007/s00604-017-2296-y
• 16 Alizadeh T, Hamidi N, Ganjali MR, Rafiei F, Determination of subnanomolar levels of mercury (II) by using a graphite paste electrode modified with MWCNTs and Hg(II)-imprinted polymer nanoparticles, Microchim Acta, 2018, 185, 1, 9, 10.1007/s00604-017-2534-3
• 17 Beluomini MA, Silva JL, Stradiotto NR, Amperometric determination of myo-inositol by using a glassy carbon electrode modified with molecularly imprinted polypyrrole, reduced graphene oxide and nickel nanoparticles, Microchim Acta, 2018, 185, 1, 10, 10.1007/s00604-018-2710-0
• 18 Li S, Liu C, Yin G, Luo J, Zhang Z, Xie Y, Supramolecular imprinted electrochemical sensor for the neonicotinoid insecticide imidacloprid based on double amplification by Pt-In catalytic nanoparticles and a Bromophenol blue doped molecularly imprinted film, Microchim Acta, 2016, 183, 3101, 3109, 10.1007/s00604-016-1962-9
• 19 Liu Y, Liu J, Liu J, Gan W, Ye B-C, Li Y, Highly sensitive and selective voltammetric determination of dopamine using a gold electrode modified with a molecularly imprinted polymeric film immobilized on flaked hollow nickel nanospheres, Microchim Acta, 2017, 184, 1285, 1294, 10.1007/s00604-017-2124-4
• 20 Wu T, Wei X, Ma X, Li J, Amperometric sensing of L-phenylalanine using a gold electrode modified with a metal organic framework, a molecularly imprinted polymer, and β-cyclodextrin-functionalized gold nanoparticles, Microchim Acta, 2017, 184, 2901, 2907, 10.1007/s00604-017-2281-5
• 21 Zhang J, Liu J, Zhang Y, Yu F, Wang F, Peng Z, Li Y, Voltammetric lidocaine sensor by using a glassy carbon electrode modified with porous carbon prepared from a MOF, and with a molecularly imprinted polymer, Microchim Acta, 2018, 185, 1, 8, 10.1007/s00604-017-2562-z
• 22 Zhang L, Li J, Zeng Y, Molecularly imprinted magnetic nanoparticles for determination of the herbicide chlorotoluron by gate-controlled electro-catalytic oxidation of hydrazine, Microchim Acta, 2015, 182, 249, 255, 10.1007/s00604-014-1326-2
• 23 Athikomrattanakul U, Katterle M, Gajovic-Eichelmann N, Scheller FW, Development of molecularly imprinted polymers for the binding of nitrofurantoin, Biosens Bioelectron, 2009, 25, 82, 87, 10.1016/j.bios.2009.06.003
• 24 Athikomrattanakul U, Katterle M, Gajovic-Eichelmann N, Scheller FW, Preparation and characterization of novel molecularly imprinted polymers based on thiourea receptors for nitrocompounds recognition, Talanta, 2011, 84, 274, 279, 10.1016/j.talanta.2010.12.049
• 25 Athikomrattanakul U, Gajovic-Eichelmann N, Scheller FW, Thermometric sensing of nitrofurantoin by noncovalently imprinted polymers containing two complementary functional monomers, Anal Chem, 2011, 83, 7704, 7711, 10.1021/ac201099h
• 26 Malitesta C, Mazzotta E, Picca RA, Poma A, Chianella I, Piletsky SA, MIP sensors - the electrochemical approach, Anal Bioanal Chem, 2012, 402, 1827, 1846, 10.1007/s00216-011-5405-5
• 27 Sharma PS, Pietrzyk-Le A, D’Souza F, Kutner W, Electrochemically synthesized polymers in molecular imprinting for chemical sensing, Anal Bioanal Chem, 2012, 402, 3177, 3204, 10.1007/s00216-011-5696-6
• 28 Taleat Z, Khoshroo A, Mazloum-Ardakani M, Screen-printed electrodes for biosensing: a review (2008-2013), Microchim Acta, 2014, 181, 865, 891, 10.1007/s00604-014-1181-1
• 29 Sinha A, Dhanjai, Jain R, Zhao H, Karolia P, Jadon N, Voltammetric sensing based on the use of advanced carbonaceous nanomaterials: a review, Microchim Acta, 2018, 185, 89, 10.1007/s00604-017-2626-0
• 30 Wang J, Electrochemical biosensing based on noble metal nanoparticles, Microchim Acta, 2012, 177, 245, 270, 10.1007/s00604-011-0758-1
• 31 Aydemir N, Malmström J, Travas-Sejdic J, Conducting polymer based electrochemical biosensors, Phys Chem Chem Phys, 2016, 18, 8264, 8277, 10.1039/C5CP06830D
• 32 Saha K, Agasti SS, Kim C, Li X, Rotello VM, Gold nanoparticles in chemical and biological sensing, Chem Rev, 2012, 112, 2739, 2779, 10.1021/cr2001178
• 33 Li J, Rossignol F, Macdonald J, Inkjet printing for biosensor fabrication: combining chemistry and technology for advanced manufacturing, Lab Chip, 2015, 15, 2538, 2558, 10.1039/C5LC00235D
• 34 Komuro N, Takaki S, Suzuki K, Citterio D, Inkjet printed (bio)chemical sensing devices, Anal Bioanal Chem, 2013, 405, 5785, 5805, 10.1007/s00216-013-7013-z
• 35 Sridevi C, Jayarama Reddy S, Electrochemical behavior of nitrofurantoin and assay of its formulations, Electroanalysis, 1991, 3, 435, 438, 10.1002/elan.1140030432
• 36 Park HS, Ko SJ, Park JS, Kim JY, Song HK, Redox-active charge carriers of conducting polymers as a tuner of conductivity and its potential window, Sci Rep, 2013, 3, 2454, 10.1038/srep02454
• 37 Kurra N, Wang R, Alshareef HN, All conducting polymer electrodes for asymmetric solid-state supercapacitors, J Mater Chem A, 2015, 3, 7368, 7374, 10.1039/C5TA00829H
• 38 Ertan G, Karasulu Y, Güneri T, Degradation and gastrointestinal stability of nitrofurantoin in acidic and alkaline media, Int J Pharm, 1993, 96, 243, 248, 10.1016/0378-5173(93)90233-6

PHOTO (COLOR)

PHOTO (COLOR)

PHOTO (COLOR)

PHOTO (COLOR)

PHOTO (COLOR)

PHOTO (COLOR)

PHOTO (COLOR): (PDF 1526 kb)