A Surface Plasmon Resonance-Based Optical Fiber Probe Fabricated with Electropolymerized Molecular Imprinting Film for Melamine Detection

Molecularly imprinted polymer (MIP) films prepared by bulk polymerization suffer from numerous deficiencies, including poor mass transfer ability and difficulty in controlling reaction rate and film thickness, which usually result in poor repeatability. However, polymer film synthesized by electropolymerization methods benefit from high reproducibility, simplicity and rapidity of preparation. In the present study, an Au film served as the refractive index-sensitive metal film to couple with the light leaked out from optical fiber core and the electrode for electropolymerizing MIP film simultaneously. The manufactured probe exhibited satisfactory sensitivity and specificity. Furthermore, the surface morphology and functional groups of the synthesized MIP film were characterized by Atomic Force Microscopy (AFM) and Fourier transform infrared microspectroscopy (FTIR) for further insights into the adsorption and desorption processes. Given the low cost, label-free test, simple preparation process and fast response, this method has a potential application to monitor substances in complicated real samples for out-of-lab test in the future.


Introduction
Melamine (2,4,6-triamino-1,3,5-triazine, MEL), a triazine analog with three amino groups, is widely used in the production of melamine formaldehyde resins for surface coatings, laminates, and adhesives [1]. Containing a substantial amount of nitrogen (66.7%) by mass, MEL has been illegally added to dairy products to boost the apparent total nitrogen content reading. Melamine combined with cyanuric acid can cause renal failure in human beings and animals due to precipitation of insoluble crystals of melamine-cyanurate adduct in the kidneys [2][3][4]. In 2004 and 2007 large outbreaks of nephrotoxic renal failure occurred in dogs and cats that were attributed to overdosing on melamine in adulterated pet food [4,5]. In September 2008, infant formulas illegally adulterated with MEL were found, which caused thousands of infants in China to suffer from kidney stones with subsequent renal failure and death if the infants were not treated [6][7][8]. Standard limits of 1 ppm (8 mM) for MEL in infant formula and 2.5 ppm (20 mM) in other milk products have been introduced by many countries [9,10]. In Canada, the maximum levels of melamine allowed in dried infant formula and other products containing milk are 0.5 ppm and 2.5 ppm, respectively [2]. Therefore, determination of melamine is of biological, clinical, and food industry importance.
Numerous instrument analytical methods have been employed for the detection of melamine, such as gas chromatography (GC) [11], high performance liquid chromatography (HPLC) [12], hyphenated atomic force microscopy (AFM) were employed in order to verify the functional groups of MIP films. Different concentrations of melamine solutions were chosen for the characterization of optical fiber probes. The specificity test was carried out against acrylamide, tetracycline, urea, glucose solutions. To the best of our knowledge, we are reporting for the first time the synthesis of electropolymerized MIP film on the surface of OF-SPR probe.

Reagents and Apparatus
Commercial multimode polymer clad silica fiber (NA 0.4, core and cladding diameter 600 µm and 660 µm, respectively) used for the fabrication of the probes, was purchased from Wyoptics (Shanghai, China). Melamine and o-aminophenol was used as template molecule and functional monomer, respectively. Phosphate buffered saline (PBS, pH = 7) was prepared using 0.1 mol/L K2HPO4 and 0.1 mol/L KH 2 PO 4 . Methanol, acetonitrile and acetic acid were used for the removal of template molecules from the MIP polymer film. Urea, glucose acrylamide and tetracycline were chosen for the specificity test. All the chemicals used were of analytical grade and were purchased from Sinopharm Chemical Reagent Co., Ltd. (Beijing, China). All aqueous solutions were prepared with purified water (Merck Chemicals (Shanghai) Co., Ltd., Shanghai, China). All electrochemical experiments were carried out at room temperature. High purity nitrogen was purchased from local vendor and was used for deaeration and blow-drying.
Molecular imprinting polymerization was performed with a three-electrode system connected to a CHI 620E electrochemical workstation (Chenhua Instruments Co., Shanghai, China). The actual pH values were determined with a microprocessor pH-meter (LE438, Mettler-Toledo, Shanghai, China). Uniform coating of Cr and Au film were performed inside a magnetron sputtering equipment. Surface topography of the MIP and non-imprinted polymer (NIP) films before and after template removal were carried out with atomic force microscopy (SPA-400, SII NanoTechnology (Shanghai) Inc., Shanghai, China). Functional groups on the surface of MIP films were characterized with FTIR microspectroscopy (Spotlight200, PerkinElmer (Shanghai) Inc., Shanghai, China). Fiber optic probes were characterized by a tungsten halogen lamp (DH2000-BAL, Ocean Optics Asia, Shanghai, China), which exhibits a wavelength emission ranging from 360 nm to 2500 nm and a spectrometer with detection resolution of 0.3 nm (Avaspec-3648, AVANTES B.V., Eerbeek, The Netherlands). To measure the change of normalized transmitted power, the resonance wavelength was obtained through transmittance mode, which follow the equation: where sample n is the current sample value at pixel n, dark n is the value at pixel n without exposure to light, re f n is the value at pixel n without any sample.

Probe Fabrication
About 12 cm long fibers were used for the fabrication of different types of probes. Both endfaces were carefully polished. The coated layer of the multimode fibers were unclothed using a sharp blade. Then the cladding of sensing sections (about 13 mm long) were completely removed by immersing into hydrogen fluoride (HF) acid to make the optic power leak out from the core. After a certain time, cleanout the exposed length to keep the etched area smooth. The etching time is related to the optical power leak out and fiber diameter.
Uniform coating of Cr (5 nm) and Au (50 nm) film were performed by fixing the probe into a homemade rotating (along the axis of optical fiber) apparatus inside a magnetron sputtering system. The chromium film served as adhesive layer while the gold film acted as sensitive layer as well as conducting layer.

Electrochemical Synthesis of MIP
Molecular imprinting polymerization was performed with a three-electrode system. The electrochemical reaction solution was made by the following procedure: o-aminophenol (1.5 mM) was dissolved in 25 mL of prepared PBS solution. Then the pH value was buffered close to 6.8. The imprinted molecule melamine (0.08 mM) and 10mL of MilliQ water were then added into the solution. We transferred this solution to a 50 mL volumetric flask, diluted with acetonitrile to volume, and then agitated with ultrasonic instrument over 5 min to remove oxygen.
Electrochemical polymerization for MIP probes was performed in a 70 mL volume self-made cylindrical electrolytic cell accommodating three electrodes, using the optical fiber probe coated with Cr/Au film as working electrode. A potassium chloride (KCl)-saturated Ag/AgCl electrode and a platinum plate electrode were used as reference and counter electrodes respectively. The optical fiber probe coated with Cr/Au was then dipped into the reaction solution and incubated for 30 min. The cylindrical electrolytic cell needed to be tightly closed and light blocking.
The electropolymerization was performed by cyclic voltammetry (CV) for 30 cycles in the potential range from −0.3 V to 1.2 V with a scan rate of 50 mV/s. The NIP probes were prepared in the same way but without the addition of template MEL. The removal of the template MEL was carried out by immersing the probes prepared above in desorption solution, an acetic acid: methyl alcohol: acetonitrile (3:4:4) solution for over 4 h at room temperature under continuous agitation. Then the probes were washed with ethanol to remove the template molecules and adsorbates on the surface of the imprinted polymer film, and dried under pure nitrogen flow for further use. The finished imprinted probes were stored at 4 • C in dry condition. A schematic diagram of the probe preparation is shown in Figure 1.

Electrochemical Synthesis of MIP
Molecular imprinting polymerization was performed with a three-electrode system. The electrochemical reaction solution was made by the following procedure: o-aminophenol (1.5 mM) was dissolved in 25 mL of prepared PBS solution. Then the pH value was buffered close to 6.8. The imprinted molecule melamine (0.08 mM) and 10mL of MilliQ water were then added into the solution. We transferred this solution to a 50 mL volumetric flask, diluted with acetonitrile to volume, and then agitated with ultrasonic instrument over 5 min to remove oxygen.
Electrochemical polymerization for MIP probes was performed in a 70 mL volume self-made cylindrical electrolytic cell accommodating three electrodes, using the optical fiber probe coated with Cr/Au film as working electrode. A potassium chloride (KCl)-saturated Ag/AgCl electrode and a platinum plate electrode were used as reference and counter electrodes respectively. The optical fiber probe coated with Cr/Au was then dipped into the reaction solution and incubated for 30 min. The cylindrical electrolytic cell needed to be tightly closed and light blocking.
The electropolymerization was performed by cyclic voltammetry (CV) for 30 cycles in the potential range from −0.3 V to 1.2 V with a scan rate of 50 mV/s. The NIP probes were prepared in the same way but without the addition of template MEL. The removal of the template MEL was carried out by immersing the probes prepared above in desorption solution, an acetic acid: methyl alcohol: acetonitrile (3:4:4) solution for over 4 h at room temperature under continuous agitation. Then the probes were washed with ethanol to remove the template molecules and adsorbates on the surface of the imprinted polymer film, and dried under pure nitrogen flow for further use. The finished imprinted probes were stored at 4 °C in dry condition. A schematic diagram of the probe preparation is shown in Figure 1.  Figure 1 shows a schematic of experimental setup used for the characterization of the sensing probe. The prepared optical fiber probe was fixed in a flow cell with sample inlet and outlet channel on the cylindrical surface so that the sample could interact with the sensing region. Meanwhile, optical fiber splices were set at both endfaces of the flow cell. The flow cell was then connected to the halogen lamp and spectrometer through Optical fiber patch cables and optical fiber flanges for the maximum launching of the light into the probe. SPR response for various concentrations of Melamine samples were recorded immediately after 30 s of incubation. The sensing surface of the fiber was washed with MilliQ water between two consecutive samples.  Figure 1 shows a schematic of experimental setup used for the characterization of the sensing probe. The prepared optical fiber probe was fixed in a flow cell with sample inlet and outlet channel on the cylindrical surface so that the sample could interact with the sensing region. Meanwhile, optical fiber splices were set at both endfaces of the flow cell. The flow cell was then connected to the halogen lamp and spectrometer through Optical fiber patch cables and optical fiber flanges for the maximum launching of the light into the probe. SPR response for various concentrations of Melamine samples were recorded immediately after 30 s of incubation. The sensing surface of the fiber was washed with MilliQ water between two consecutive samples.

Electropolymerization of Molecularly Imprinted Film
A typical cyclic voltammogram recorded during electropolymerization in the presence of functional monomer oAP and template MEL under a scan rate of 50 mV/s is shown in Figure 2. It is observed that a distinct irreversible broad anodic oxidation peak appeared at +1.1 V (versus Ag/AgCl) on the anodic branch at the first scan. During successive scans (up to 30 cycles), the peak current decreased and shifted to more negative potentials (stabilized around +0.65 V), which was in accordance with the feature reported by Evrim et al. [51]. Meanwhile a dark brown and homogeneous PoAP film was observed on the gold surface of optical fiber probe. Here, the oxidation peak shifted from +1.1 V to +0.65 V may related to the dimer oxidation to a semi-oxidized state followed by the total-oxidized state process [52]. No reduction peak was observed during the polymerization of oAP. Electropolymerization under a scan rate of 5 mV/s is shown in Figure S1. At a lower scan rate, the oxidation peak stabilized around +0.6 V without peak shifting, which might indicate that a total-oxidized state process directly took place on the gold surface. It did not have significant differences in comparison with the cyclic voltammogram obtained under the same conditions in the absence of MEL template.

Electropolymerization of Molecularly Imprinted Film
A typical cyclic voltammogram recorded during electropolymerization in the presence of functional monomer oAP and template MEL under a scan rate of 50 mV/s is shown in Figure 2. It is observed that a distinct irreversible broad anodic oxidation peak appeared at +1.1 V (versus Ag/AgCl) on the anodic branch at the first scan. During successive scans (up to 30 cycles), the peak current decreased and shifted to more negative potentials (stabilized around +0.65 V), which was in accordance with the feature reported by Evrim et al. [51]. Meanwhile a dark brown and homogeneous PoAP film was observed on the gold surface of optical fiber probe. Here, the oxidation peak shifted from +1.1 V to +0.65 V may related to the dimer oxidation to a semi-oxidized state followed by the total-oxidized state process [52]. No reduction peak was observed during the polymerization of oAP. Electropolymerization under a scan rate of 5 mV/s is shown in Figure S1. At a lower scan rate, the oxidation peak stabilized around +0.6 V without peak shifting, which might indicate that a totaloxidized state process directly took place on the gold surface. It did not have significant differences in comparison with the cyclic voltammogram obtained under the same conditions in the absence of MEL template.

FTIR Microspectroscopy of MIP Film
Since the MIP film on the surface of OF-SPR probe was synthesized by cyclic voltammetry (30 cycles) in the potential range from −0.3 V to 1.2 V under a scan rate of 50 mV/s, the MIP film was too thin to generate an infrared signal (data not shown). To characterize the surface groups on PoAP-MEL film, a thicker film sample was synthesized on the surface of a glass slide (1 cm × 1.5 cm) coated with Au film by cyclic voltammetry (150 cycles) in the potential range from −0.4 V to 1.3 V under a scan rate of 5 mV/s ( Figure S1). Figure 3 shows the FTIR spectrum of (a) PoAP-MEL and (b) oAp. All the FTIR peaks assignment for PoAP-MEL and oAP were listed in Table 1. The typical peaks at 3377 and 3307 cm in Figure 3, represented the absorption bands of the asymmetrical and symmetrical N-H stretching vibrations.

FTIR Microspectroscopy of MIP Film
Since the MIP film on the surface of OF-SPR probe was synthesized by cyclic voltammetry (30 cycles) in the potential range from −0.3 V to 1.2 V under a scan rate of 50 mV/s, the MIP film was too thin to generate an infrared signal (data not shown). To characterize the surface groups on PoAP-MEL film, a thicker film sample was synthesized on the surface of a glass slide (1 cm × 1.5 cm) coated with Au film by cyclic voltammetry (150 cycles) in the potential range from −0.4 V to 1.3 V under a scan rate of 5 mV/s ( Figure S1). Figure 3 shows the FTIR spectrum of (a) PoAP-MEL and (b) oAp. All the FTIR peaks assignment for PoAP-MEL and oAP were listed in Table 1. The typical peaks at 3377 and 3307 cm −1 in Figure 3, represented the absorption bands of the asymmetrical and symmetrical N-H stretching vibrations. C=C stretching of aromatic ring [51,[53][54][55] 1512 cm C=C stretching of aromatic ring [51,55] 1478 cm Asymmetrical N-H stretching vibrations [51] 3377 cm Symmetrical N-H stretching vibrations [51] 3307 cm C-O-H deformation vibrations of phenols [51,53,54,56] 1407 cm O-H deformation vibration and C-O stretching vibration of phenols [52,53] 1270 cm C-O stretching vibration of phenols [51][52][53] 1216 cm    [52,53] 1274 cm −1 C-O stretching vibration of phenols [51] 1207 cm −1 Vibration assignments for oAP Wave number, cm −1 N-H bending in primary amines [54,55] 1585 cm −1 C=C stretching of aromatic ring 1612 cm −1 C=C stretching of aromatic ring [51,[53][54][55] 1512 cm −1 C=C stretching of aromatic ring [51,55] 1478 cm −1 Asymmetrical N-H stretching vibrations [51] 3377 cm −1 Symmetrical N-H stretching vibrations [51] 3307 cm −1 C-O-H deformation vibrations of phenols [51,53,54,56] 1407 cm −1 O-H deformation vibration and C-O stretching vibration of phenols [52,53] 1270 cm −1 C-O stretching vibration of phenols [51][52][53] 1216 was in accordance with the liner PoAP polymer theory during electropolymerization via N-C bonding process [57,58], which might suggested that the MEL adsorpt and desorpt from the binding sites through hydrogen bond with -OH and -NH. Since the template content was too low to be detected by Fourier transform infrared microspectroscopy, no remarkable peak was observed standing for MEL.

Surface Morphology Characterization by AFM
AFM was employed to characterize the surface morphology change of PoAP-MEL film before and after template removed. Glass slides (1 cm × 1.5 cm) coated with Au film served as work electrodes. MIP and NIP film samples were synthesized by cyclic voltammetry (30 cycles) in the potential range from -0.3 V to 1.2 V under a scan rate of 50 mV/s. Figure 4 shows typical topographic images by AFM of (a) PoAP-MEL film, (b) PoAP-MEL removed template MEL and (c) NIP film. It reveals that the surfaces were uniform with root-mean-square roughnesses (RMS) of 1.04 nm, 0.87 nm and 0.61 nm respectively. It was quite obvious that PoAP-MEL film was rougher than PoAP-MEL film removed template MEL and NIP film. It was noteworthy that after removed MEL template, the upper surface became flat. Meanwhile the vertical axis revealed that film valleys become much deeper. This might suggest that the binding cavities were exposed. It was speculated that when the imprinted template MEL rebind into the cavities, the film becomes rough, and the effective refractive index increase, leading to a shift in SPR absorption peak. As contrast, the NIP film appeared to be much smoother and thinner than the PoAP-MEL film, and this could be explained by the fact that no MEL template was trapped in the film. Similar features have been reported in previous studies [59,60]. surface of PoAP, which was able to bond up with "imprinted" molecules (melamine) to constitute binding sites. The IR spectrum obtained was in accordance with the liner PoAP polymer theory during electropolymerization via N-C bonding process [57,58], which might suggested that the MEL adsorpt and desorpt from the binding sites through hydrogen bond with -OH and -NH. Since the template content was too low to be detected by Fourier transform infrared microspectroscopy, no remarkable peak was observed standing for MEL.

Surface Morphology Characterization by AFM
AFM was employed to characterize the surface morphology change of PoAP-MEL film before and after template removed. Glass slides (1 cm × 1.5 cm) coated with Au film served as work electrodes. MIP and NIP film samples were synthesized by cyclic voltammetry (30 cycles) in the potential range from -0.3 V to 1.2 V under a scan rate of 50 mV/s. Figure 4 shows typical topographic images by AFM of (a) PoAP-MEL film, (b) PoAP-MEL removed template MEL and (c) NIP film. It reveals that the surfaces were uniform with root-mean-square roughnesses (RMS) of 1.04 nm, 0.87 nm and 0.61 nm respectively. It was quite obvious that PoAP-MEL film was rougher than PoAP-MEL film removed template MEL and NIP film. It was noteworthy that after removed MEL template, the upper surface became flat. Meanwhile the vertical axis revealed that film valleys become much deeper. This might suggest that the binding cavities were exposed. It was speculated that when the imprinted template MEL rebind into the cavities, the film becomes rough, and the effective refractive index increase, leading to a shift in SPR absorption peak. As contrast, the NIP film appeared to be much smoother and thinner than the PoAP-MEL film, and this could be explained by the fact that no MEL template was trapped in the film. Similar features have been reported in previous studies [59,60].

Melamine Detection Based on Resonance Wavelength Modulation
Melamine solutions of different concentrations ranging from 10 M to 10 M were chosen for the characterization of PoAP-MEL OF-SPR probe. SPR spectrum was recorded immediately after 30 s after each sample solution was injected. Sensing region of the probe was cleaned with MilliQ water between two consecutive samples to remove the residual molecule from previous solution. The SPR spectrum of the MilliQ water was recorded as a reference. The SPR spectra recorded for different concentrations of melamine are shown in Figure 5a. To avoid overlapping between SPR curves, all SPR spectra are not shown in the figure.

Melamine Detection Based on Resonance Wavelength Modulation
Melamine solutions of different concentrations ranging from 10 −10 M to 10 −2 M were chosen for the characterization of PoAP-MEL OF-SPR probe. SPR spectrum was recorded immediately after 30 s after each sample solution was injected. Sensing region of the probe was cleaned with MilliQ water between two consecutive samples to remove the residual molecule from previous solution. The SPR spectrum of the MilliQ water was recorded as a reference. The SPR spectra recorded for different concentrations of melamine are shown in Figure 5a. To avoid overlapping between SPR curves, all SPR spectra are not shown in the figure. An evident SPR resonance wavelength shifted towards the higher wavelength with the increase in the concentration of melamine solutions, which could be observed from Figure 5b. As previously studied, when a sample molecule comes near the vicinity of the polymer, the non-template molecule would be randomly absorbed around the surface of MIP film, however the template molecule would bind with one of the binding sites with weak interactions due to their complementary shape and size. The binding of template molecule with polymer causes the change in dielectric nature of the gold layer-MIP interface [61]. This change is recognized by change in resonance wavelength in the SPR spectrum since the position of resonance wavelength depends upon the dielectric constant of the polymer. Hence the peak shift in resonance wavelength of SPR represented the melamine molecule amount absorbed onto the binding sites of the PoAP-MEL film [43]. Since the adsorption and desorption kept dynamic equilibrium under thermal equilibrium condition, the shift distance of each SPR resonance wavelength reflected the melamine concentration in each solution. An evident SPR resonance wavelength shifted towards the higher wavelength with the increase in the concentration of melamine solutions, which could be observed from Figure 5b. As previously studied, when a sample molecule comes near the vicinity of the polymer, the non-template molecule would be randomly absorbed around the surface of MIP film, however the template molecule would bind with one of the binding sites with weak interactions due to their complementary shape and size. The binding of template molecule with polymer causes the change in dielectric nature of the gold layer-MIP interface [61]. This change is recognized by change in resonance wavelength in the SPR spectrum since the position of resonance wavelength depends upon the dielectric constant of the polymer. Hence the peak shift in resonance wavelength of SPR represented the melamine molecule amount absorbed onto the binding sites of the PoAP-MEL film [43]. Since the adsorption and desorption kept dynamic equilibrium under thermal equilibrium condition, the shift distance of each SPR resonance wavelength reflected the melamine concentration in each solution.
For melamine concentration of 10 −10 M, resonance wavelength obtained was 545.80 nm while that for 10 −2 M was 554.87 nm. This confirmed a red shift of 9.07 nm in the resonance wavelength for the change in concentration from 10 −10 M to 10 −2 M. As a reference, the resonance wavelength obtained for MilliQ water was 539.91 nm, which meant a total 14.96 nm resonance wavelength shift was obtained in 30 s incubation time for each sample.
However, it can be noted from Figure 5b that the resonance wavelength shift tended to slow down at higher concentrations of melamine. This could be explained by the limited number of binding sites on the MIP film. At lower concentration, the number of binding sites far surpassed the total melamine molecules in solution sample. Most melamine molecules could be absorbed onto the surface of MIP film, which would cause a maximum shift. As the melamine concentration increased, the binding sites started getting crowed, and quite a number of binding sites were filled. The remaining binding sites and the melamine molecules in solution sample maintained dynamic equilibrium, which resulted in a lower shift in resonance wavelength, and appeared a saturation trend.
According to the definition of sensitivity under resonance wavelength modulation mode, which is the derivative of the resonance wavelength curve versus refractive index, we calculated the shift in resonance wavelength per unit change in the logarithmic scale of melamine concentration (nm/log M). From Figure 6, maximum sensitivity of the probe has been found to be 5.89 nm/log M at 10 −10 M melamine sample (the same probe in Figure 5). It decreased with the increase in melamine concentration due to the limited number of binding sites available in the sensing layer mentioned above. It was considered that the probes are more valuable for low concentrations of melamine. The error bar showed the maximum probable error in the measurement for 10 different probes.
To confirm the recognition ability of PoAP-MEL film, resonance wavelength shift of Au/MIP probe was compared with probe covered with bare Au film probe and Au/NIP probe during the detection of melamine solutions with concentration from 0 M to 10 −7 M. All three probes were carefully made to ensure the sensing region and the Au film thickness equal. The comparison of observed shift in resonance wavelength was shown in the histogram (Figure 7). A small shift of 0.95 nm in resonance wavelength was observed for bare Au coated probe since melamine solution concentration from 0 M to 10 −7 M varies little in refractive index. A shift of 1.87 nm in resonance wavelength was observed for Au/NIP coated probe due to barely any binding sites on NIP films. In the case of Au/MIP probe, a remarkable shift of 11.79 nm was observed for the same concentration change of melamine solution. This suggested it is the cavities formed by templates trapped during the polymerization rather than poly(o-aminophenol) that had the ability to absorb template molecule.
Specificity is the fundamental feature of a MIP sensor. In this study, acrylamide, tetracycline, urea, glucose and melamine were chosen to verify the specificity of PoAP-MEL film. Figure 8 shows the comparison of resonance wavelength shift for the change in concentration of analytes from 0 M to 10 −7 M. Compared to other analytes, the observed shift in resonance wavelength reached over 11 nm, which showed a very high specificity for melamine. Meanwhile for acrylamide, tetracycline, urea, glucose they were 0.31 nm, 0.78 nm, 2.18 nm and 0.31 nm respectively. Removal of templates from MIP film exposed cavities with complementary size and shape related to the templates for the subsequent rebinding process, which formed a "lock and key" relationship [62]. Thus, only template molecules could enter these binding sites, while other molecules were unable to form a mass accumulation on the surface of MIP film, which resulted in limited change in dielectric index at the gold layer-MIP interface. Therefore, only little shift in resonance wavelength for other analytes was observed.

Limit of Detection (LOD)
Limit of detection is another crucial parameter of a sensing device, which represents the detectable analyte concentration near zero concentration. According to Harshit et al. [43], the mathematical expression for LOD is: where C represents the limit of detection of the probe, stands for the sensitivity near blank (zero) concentration, and R represents spectrometer resolution. Since the spectrometer has a resolution of 0.3 nm and the sensitivity of the probe near zero concentration is 5.89 nm/log M as mentioned above, the calculated value of LOD of the probe is 5.1 × 10 M. To further investigate the performance of the proposed probe, we compared our results with previous studies in Table 2. It could be seen that the probe exhibited remarkable advantages, such as higher sensitivity, wider linear range and lower detection limit. It is worth to note that the LOD in present study is much lower than the safe standard of 2.5 ppm in food products.

Limit of Detection (LOD)
Limit of detection is another crucial parameter of a sensing device, which represents the detectable analyte concentration near zero concentration. According to Harshit et al. [43], the mathematical expression for LOD is: where C LOD represents the limit of detection of the probe, S stands for the sensitivity near blank (zero) concentration, and R represents spectrometer resolution. Since the spectrometer has a resolution of 0.3 nm and the sensitivity of the probe near zero concentration is 5.89 nm/log M as mentioned above, the calculated value of LOD of the probe is 5.1 × 10 −12 M. To further investigate the performance of the proposed probe, we compared our results with previous studies in Table 2. It could be seen that the probe exhibited remarkable advantages, such as higher sensitivity, wider linear range and lower detection limit. It is worth to note that the LOD in present study is much lower than the safe standard of 2.5 ppm in food products. Table 2. Limit of detection and operating range of various approaches for melamine reported in previous study.

Conclusions
In summary, we have fabricated an OF-SPR probe with PoAP-MEL MIP film synthesized by electropolymerization as sensing layer. An Au film served as the refractive index sensitive metal film coupling with the light leaked out from optical fiber core and the electrode for electropolymerizing MIP film at the same time. o-Aminophenol was chosen as functional monomer to trap melamine during the electropolymerizing process. FTIR spectrum revealed that -NH and -OH might be the functional groups bonded up with "imprinted" molecules (melamine) to constitute reactive sites. Surface morphology was characterized by AFM through contact mode. AFM 3D images showed that after template molecules were removed, the upper surface of PoAP-MEL film fattened, while the film valleys became much deeper, suggesting that the binding cavities were exposed. The sensitivity and specificity of the probe were all found to be satisfactory. Compared with previously reported studies, our method had the ability to detect MEL with concentration lower than 10 −10 M, and had a calculated detection limit of 5.1 × 10 −12 M, which showed significant combined advantages of sensitivity and convenience. Given the low cost, portable size, label-free test, simple preparation process and fast response, this method has potential application to monitor substances in complicated real samples for field tests in the near future.