Colorimetric detection of melamine in milk by using gold nanoparticles-based LSPR via optical fibers

A biosensing system with optical fibers is proposed for the colorimetric detection of melamine in liquid milk samples by using the localized surface plasmon resonance (LSPR) of unmodified gold nanoparticles (AuNPs). The biosensing system consists of a broadband light source that covers the spectral range from 200 nm to 1700 nm, an optical attenuator, three types of 600 μm premium optical fibers with SMA905 connectors and a miniature spectrometer with a linear charge coupled device (CCD) array. The biosensing system with optical fibers is low-cost, simple and is well-proven for the detection of melamine. Its working principle is based on the color changes of AuNPs solution from wine-red to blue due to the inter-particle coupling effect that causes the shifts of wavelength and absorbance in LSPR band after the to-be-measured melamine samples were added. Under the optimized conditions, the detection response of the LSPR biosensing system was found to be linear in melamine detection in the concentration range from 0μM to 0.9 μM with a correlation coefficient (R2) 0.99 and a detection limit 33 nM. The experimental results obtained from the established LSPR biosensing system in the actual detection of melamine concentration in liquid milk samples show that this technique is highly specific and sensitive and would have a huge application prospects.


Introduction
Melamine (C3H6N6; molecular weight: 126.12) is an organic compound with a 1, 3, 5-triazine and 2, 4, 6-triamine skeleton and the aqueous solution of melamine is weakly alkaline. Generally, melamine is often a basic chemical that can be used to make, for example, fire retardant, water-reducing agent and formaldehyde cleaner. Due to its biotoxicity, excessive intake of melamine will cause great harm to human health and lead to diseased kidney or urinary system [1,2]. Melamine is therefore not approved by governments as a food additive in animal feed or a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 human food. However, melamine was sometimes illegally added to protein-rich food by unethical manufactures to increase nitrogen level (66% nitrogen by mass). The most famous food safety accidents related to melamine occurred in China in 2008. Melamine was illegally added into the infant milk powder by a dairy company that caused about 40,000 children affected with diseases. As reported in 2012, international experts in Europe and USA have limited the melamine level in food at the concentration less than 2.5 mg/kg [3]. Thus there is an increasing demand for effective and reliable methods proposed to detect the concentration of melamine in food.
In recent years, a growing body of researches has focused on the detection of melamine in food and drinking water. The most commonly used methods for melamine detection are the chromatographic analysis methods, including high performance liquid chromatography (HPLC) [4], mass spectrometry (MS) [5], gas chromatography/mass spectrometry (GC/MS) [6], high performance liquid chromatography/mass spectrometry (HPLC/MS) [7], capillary zone electrophoresis/mass spectrometry [8], and electrochemical methods [9,10]. Enzymelinked immunosorbent assay (ELISA), surface enhanced Raman spectroscopy (SERS), twophoton photoluminescence, dynamic light scattering, hyper-rayleigh scattering, and visible and near-infrared bulk optical properties of raw milk have also been utilized for the detection of melamine [11][12][13][14][15]. However, most of these techniques have the disadvantages such as complicated pre-treatment, time-consuming steps and high-cost instruments, which cannot meet the requirement for rapid detection. Hence, there is an urgent need to develop a fast, low-cost, highly sensitive and selective biosensing system to determine the trace melamine in food samples. Recently, AuNPs-based colorimetric sensors have attracted great attention due to the unique properties of AuNPs, such as the excellent optical property, electronic property, biocompatibility, stability and high extinction coefficients [16][17][18]. Various standard protocols and recent advances for the shape-controlled synthesis of nanocrystals have been reported [19]. Several studies have already been conducted for the colorimetric detection of melamine in milk based on AuNPs localized surface plasmon resonance (LSPR) that offers an easy and rapid way for melamine detection [20][21][22][23]. The influence produced by the size and shape of AuNPs in LSPR biosensing system has been investigated [24][25][26][27][28]. Gold nanostars immobilized onto flexible PDMS substrates via a simple solution process based on electrostatic self-assembly have been demonstrated [29,30]. However, most of the colorimetric detection methods rely on an ultraviolet-visible (UV-Vis) spectrophotometer, which is expensive and not portable.
In this paper, we report a localized surface plasmon resonance (LSPR) biosensing system that utilizes AuNPs for the colorimetric detection of melamine in liquid milk. When the AuNPs solutions are with different melamine concentrations, the absorbance ratios of light at wavelength 520 nm are different due to the AuNPs changed from dispersion to aggregation. Based on this phenomenon, a standard model is established to detect the melamine concentrations in liquid milk samples and the melamine at low concentration can be detected rapidly and sensitively by the proposed LSPR biosensing system with optical fibers.

Principle of LSPR and the experimental setup
The nanoparticle-based optical sensing techniques have been found to be effective for quantitative detection of chemical and biological targets. The principle of this method is based on the analysis of LSPR spectrum of noble metal nanoparticles, which is pertaining to the refractive index changes caused by the surrounding environment (see Fig 1a). In considering a spherical nanoparticle of radius a and irradiated by a light at wavelength λ (a is much smaller than the wavelength of λ), the LSPR resonance occurs when the frequency of incident electromagnetic wave is coincident with the oscillation frequency of the oscillating electrons on the surface of nanoparticle. The extinction spectrum of the gold sphere can be calculated by solving Maxwell's equations using the quasi-static approximation [31]: Where ε in and ε out are the dielectric constants of the gold nanoparticle and the external environment, respectively. Since the gold metal is dispersive, the complex dielectric constant depends on the wavelength of incident light, λ, and can be expressed by ε(λ) = ε r (λ) + ε i (λ), where ε r (λ) and ε i (λ) represent the real and imaginary parts, respectively. N is the electron density. χ is the shape factor of the metal nanoparticle for a typical value of 2 for the case of a sphere.
The occurrence of LSPR depends on the size, shape, interparticle space and dielectric properties of the material, as well as the dielectric properties of the surrounding environment. The maximum wavelength (λ max ) of LSPR extinction for nanoparticles is highly sensitive to the refractive index n and can be measured by using ultraviolet-visible extinction spectroscopy. Therefore, λ max will shift when the adsorbate is bound to AuNPs due to the change of refractive index of local environment. The LSPR shift is mainly attributed to the adsorbate and the shift of LSPR extinction wavelength maximum Δλ max is given as follows: where m is the bulk refractive-index response of the nanoparticle, Δn is the change of refractive index induced by the adsorbate, d is the effective adsorbate layer thickness, and l d is the electromagnetic field decay length (approximated as an exponential decay) [32,33]. In theory, the extinction spectrum is determined by both the absorption and scattering characteristics of AuNPs and obtained by measuring the ratio of light transmitting through the AuNPs solution to the incidence light. In this work, optical fibers and spectrometer are utilized to record the transmission ultraviolet-visible spectra in a straightforward mode, as shown in Fig 1(b). Since the diameters of AuNPs are usually less than 10 nm and the concentration of AuNPs is very low in the solution, the scattering effect of AuNPs is negligible as compared with the absorption effect. Thus, the absorbance can be approximatively calculated by the Lambert-Beer's law. The schematic diagram of the optical fiber-based LSPR detection system is depicted In Fig  1(b), where the tungsten halogen lamp is used as the broadband light source with spectrum ranging from 200nm to 1700 nm and the optical attenuator is used to adjust and optimize the intensity of light source that was delivered to the cuvette for spectral measurement. The light from the source is transmitting through an optical fiber of diameter 600 μm before entering the cuvette filled with sample solution where the interaction between the sample molecules and the incident light photons occurs. After that, the transmitted light enters the spectrometer through another optical fiber and the output light is converted into electrical signals via the CCD detector of spectrometer. The values of light absorbance can be calculated by the Beer-Lambert Law, AðlÞ ¼ log 10 ð I R ðlÞÀ I B ðlÞ I S ðlÞÀ I B ðlÞ Þ, where I R is the reference signal intensity, I B is the background signal intensity (or dark noise) and I S is the sample signal intensity.

Experimental results and discussion Reagents
Chloroauric acid (HAuCl 4 ), sodium citrate, and sodium carbonate (Na 2 CO 3 ) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Glucose, fructose and lactose were purchased from Tianjin Chemical Co., Ltd. Melamine was obtained from Tianjin Reagent Chemical Reagent Co., Ltd. (Tianjin, China). Liquid milk was purchased from the local supermarket. All solvents and reagents were of analytical grade and were used without further purification. Ultrapure water was used throughout the whole experiment. All glassware used in the experiment were soaked in nitrohydrochloric acid and rinsed thoroughly in water and dried in air before use.

Preparation of AuNPs
The AuNPs used in the experiments were prepared by the reduction of chloroauric acid using sodium citrate. 1 mL of sodium citrate (1%) was rapidly injected into 0.01% HAuCl 4 (100 mL) boiling aqueous solution and further refluxed for 15 min. The mixture was then cooled with distilled water and continuously stirred before its temperature returned to room temperature. The formed wine-red AuNPs solution was then stored at 4˚C for further use. According to the measured UV-visible spectra of AuNPs solution obtained from this preparation, it is found that the maximum absorption wavelength of AuNPs is 520 nm. Moreover, it also can be inferred that the sizes of AuNPs were about 13 nm from the transmission electron microscopy (TEM) images [34].

Establishment of detection method
Firstly, 100 μL of AuNPs solution was taken using a centrifugal tube (full scale, 500 μL) and diluted with 300 μL of ultrapure water. Then, 100 μL of melamine solutions (standard level) with different concentrations were poured into the AuNPs solutions to form the sample solutions. These sample solutions were evenly mixed by the Vortex-Qilinbeier 5 vibrator before being individually transferred to the cuvette. After 30 min reaction time, the absorption spectrum was recorded continuously every 15s and 120 sets of spectra data were stored totally. Then the calibration curve was established according to the relationship between the absorbance at the maximum absorption wavelength 520 nm and the concentrations of melamine solution.

Pretreatment of liquid milk samples
Firstly, 1.2 mL of 10% trichloroacetic acid and 1.5 mL of chloroform were added into 4 mL of liquid milk. The mixture was placed into the centrifugal tube and agitated for 2 min with the vibrator before being placed into an ultrasonic bath for 15 min to reach equilibrium. After that, the liquid milk suspensions were homogenized and then centrifuged for 10 min at a speed of 13000 rpm. The clear supernatant was taken into another centrifugal tube and the 1.0 mol/L Na 2 CO 3 solution was added to adjust the pH to 4.3~4.5. Again, this supernatant was centrifuged for 3 min at a speed of 3000 rpm. Finally, the supernatant was stored in the refrigerator for further use.

Melamine measurement based on colorimetric detection
The mechanism for the colorimetric detection of melamine is illustrated in Fig 2(a). The synthesized AuNPs are stable in aqueous solution due to the repulsion force between citrate anions on the surface of AuNPs which prevents the AuNPs from aggregating. The citrate-stabilized AuNPs solution shows a wine-red color. When the melamine solution is added, the electrostatic repulsion between AuNPs is screened, which leads to the aggregation of AuNPs and causes the solution color to turn blue. The reason is that three amine groups (-NH 2 ) of melamine molecules would interact with the AuNPs and the dissociation of citrates ions from the surface of AuNPs makes them losing the ability to stabilize AuNPs.
In order to further validate the above detection mechanism, a series of UV-Visible spectra of AuNPs are measured by using the optical fiber-based LSPR detection system under different experimental conditions. When the melamine is added into the AuNPs solution, the induced aggregation effect makes the localized surface plasmon resonance (LSPR) absorption band of AuNPs red-shifted and broadened. As shown in Fig 2(b), the absorption spectra of AuNPs solution have maximum absorption peaks at 520 nm. The maximum absorption peak of the solution decreases when the added melamine is with a concentration of 0.5 μM. Further, when 2 μM melamine is added into the AuNPs solution, a new absorption peak at about 640 nm appears. From the inset of Fig 2(b), it can be seen clearly that the color of AuNPs solution

Establishment of standard curve
The standard melamine solutions with different concentrations, 0.05, 0.2, 0.4, 0.6 and 0.9 μM, are added into the AuNPs solutions, respectively. As can be seen from Fig 4, the increment of melamine concentration leads to the decreasing absorbance of AuNPs solutions at wavelength 520nm. In order to quantitatively study the relationship between the melamine concentration and the AuNPs absorbance, we set the absorbance difference (ΔA) between the bare AuNPs solution and the AuNPs solutions added with melamine of different concentrations as the abscissa variable and the melamine concentration as the ordinate variable. Then a good linear curve is obtained with an R square of 0.99 between the concentration and absorbance. The fitting equation of the linear curve is y = 0.026x + 0.01 and the detection limit is 33 nM as calculated according to the formula 3a slope , where α is the standard deviation and slope is the slope of the fitting straight line. The obtained linear calibration curve is shown in the inset of Fig 4. As compared with some other methods with the detection limits shown in Table 1, the optical fiber-based LSPR method for melamine detection has a high sensitivity. Furthermore, from Table 1, the linear relation between the absorbance ratio and the concentration of melamine in the range of 2.0-250.0 μM [35], a linear relation between the absorbance and the melamine concentration in the range of 0.4-4μM [36], a linear correlation between the absorbance and the melamine concentration ranging from 0.9μM to 3.1μM [37] have been validated. Therefore, it can be seen that the proposed method not only provides a sensitive colorimetric detection of melamine at a low concentration in the range of 0~0.9 μM with a correlation coefficient (R2) of 0.99 and a detection limit of 33 nM, but also makes a compact and inexpensive spectroscopic instrumentation possible.

Experiment of specificity
In order to investigate the specificity of this method for the detection of melamine concentrations in liquid milk samples, several interfering compounds commonly present in liquid milk,  such as glucose, fructose, lactose and ammeline (C3H5N5O), which is the hydrolysis product of melamine, were chosen for the absorption spectra experiment as the same to those of melamine. The specificity study was estimated by measuring the absorption difference ΔA in the presence of melamine and other interfering compounds. During the experiment of specificity, the concentration of standard melamine solution was 0.5 μM, while the interfering compounds are with a much high concentration, 5 μM. As shown in Fig 5, the absorption difference ΔA from the melamine measurement is more than ten times as big as the effect from saccharide (the interfering compounds in milk), however, it is less than twice the absorption difference ΔA from ammeline. The phenomenon demonstrates that the saccharides do not interfere the melamine detection. The results imply that the presence of these interfering compounds cannot cause much aggregation effect to AuNPs and hasn't much impact on the melamine detection by using this optical fiber-based LSPR biosensing system.

Detection of melamine in liquid milk samples
Further experiments are conducted in this section to validate the feasibility of the proposed method for detecting melamine in real milk samples mixed with different amounts of melamine. The milk samples with different melamine concentrations are prepared following the procedure described above in Section 4.2. The melamine solutions with concentrations 0.5 μM, 1.5 μM and 2.5 μM were agitated for 0.5 min with the Vortex-Qilinbeier5 vibrator, respectively. Precisely, 100 μL of melamine solutions was taken separately from above 0.5 μM, 1.5 μM and 2.5 μM of melamine solutions, and then 400 μL of AuNPs was added to reach the five-fold dilution. Afterwards, 100 μL of the mixtures were taken and their absorption spectra are measured by using this optical fiber-based LSPR detection system. The experiment results show that response relationship between the difference of absorbance values at wavelength 520 nm and the melamine concentrations of the liquid milks is still linear in the concentration range from 0.1μM to 0.9 μM. The calculated recovery rate of the colorimetric detection of melamine samples according to the measurement results is shown in Table 2. A recovery rate of 99.2%~111% was obtained with the detection limit of 33 nM. This recovery experiment shows that the optical fiber-based localized surface plasmon resonance of AuNPs for the colorimetric detection of melamine in liquid milk is also suitable for the rapid melamine detection in practice.

Conclusions
An optical fiber-based biosensing system using the optical fiber-based localized surface plasmon resonance of AuNPs for colorimetric detection of melamine in liquid milk has been investigated. It has obvious advantages over the conventional ultraviolet-visible spectrophotometer, not only because it is fast and accurate, but also is stable in the effective detection of melamine in real milk samples in practice. The experimental results demonstrate a good linear relationship between the difference of absorbance values at the wavelength of 520 nm and the concentrations of melamine in the liquid milk in the range from 0.1μM to 0.9μM and within the detection limit 33 nM. Moreover, the recovery rate of 99.2%~111% of melamine in liquid milk samples is also successfully obtained. The proposed methodology would have a huge application prospects because it also can be applied to other molecules by just selecting appropriate ligands to the to-be-detected molecules.