Smartphone-based colorimetric determination of glucose in food samples based on the intrinsic peroxidase-like activity of nitrogen-doped carbon dots obtained from locusts

2.1 Instrumentation and reagents

TEM images of N-CDs were obtained by a transmission electron microscope (JEM-2011, JEOL, Japan). The elemental composition of N-CDs was determined with an X-ray photoelectron spectrometer (Escalab 250Xi, Thermo Scientific, USA). Electron spin resonance (ESR) experiments were performed on an EMX-8/2.7 spectrometer (Bruker, Germany). A fluorescence spectrometer (RF-6000, Shimadzu, Japan) was employed to obtain the fluorescence spectra. Absorption spectra were recorded using a UV–vis spectrophotometer (UV-2600, Shimadzu, Japan). Fourier transform infrared (FTIR) absorption spectra were obtained using KBr pellets on an IR spectrometer (WQF-510, Beijing Beifen-Ruili Analytical Instrument (Group) Co. ltd., Beijing, China). The solid powder of N-CDs was obtained by a lyophilizer (FD-IA-50, Beijing Songyuan Huaxing Technology Development Co., ltd., Beijing, China). An iPhone 7 smartphone was used to obtain photos for the tests. All the images were analyzed by a free application (Color Analyzer).

Hydrogen peroxide (30 %), nitric acid (AR), 3,3′,5,5′-tetramethylbenzidine (TMB), diethylenetriamine (DETA), and glucose oxidase (GOx) were purchased from Aladdin (Shanghai, China). Amino acids, sugars (including methionine (Met), histidine (His), arginine (Arg), lysine (Lys), glycine (Gly), glutamic acid (Glt), cysteine (Cys), Vitamin C (Vc), sucrose (Suc), fructose (Fru), and glucose (Glu)), and sulfates or nitrites of metals were purchased from Sangon Biotech (Shanghai) Co., ltd. (Shanghai, China). The pH of the testing solutions was controlled by acetic acid-sodium acetate buffer solutions. Locusts were obtained from a local supermarket (Zhengzhou, China). Deionized water (18 MΩ cm⁻¹) obtained from a Millipore system was used in all the experiments.

2.2 Synthesis of N-CDs

N-CDs were prepared according to our previously reported method^[47], and the details are provided in the supplementary information.

2.3 Peroxidase-like catalytic activity of N-CDs

The peroxidase-like catalytic activity of N-CDs was determined by investigating the oxidation of TMB in the presence of H₂O₂. Typically, N-CDs-TMB, TMB-H₂O₂, and N-CDs-TMB-H₂O₂ were mixed in three colorimetric tubes, sequentially. After reactions in a water bath for 15 min, the absorbance (654 nm) of the above-mentioned solutions was recorded.

2.4 Steady-state kinetic study

Steady-state kinetic experiments were carried out in the time course mode by monitoring the absorbance variation at 654 nm on a UV-2600 spectrophotometer. The experiments were performed at 50 °C with 0.45 mg mL^-1N-CDs in buffer solutions (pH 3.0) by varying the concentration of TMB (0.1–1.0 mM) at a fixed H₂O₂ concentration (0.1 mM) and by varying the concentration of H₂O₂ (0.05–10 mM) at a fixed TMB concentration (0.6 mM). The apparent kinetic parameters were calculated through a Lineweaver-Burk plot: $$\frac{1}{V_0}=\frac{K_m+\left[s\right]}{V_{\mathit{max}}·\left[s\right]}=\frac{K_m}{V_{\mathit{max}}}·\frac{1}{\left[s\right]}+\frac{1}{V_{\mathit{max}}}$$

Here, V₀ and V_max are the initial and maximum velocities, respectively, [S] is the substrate concentration, and K_m is the Michaelis-Menten constant, which is an indicator of the affinity of an enzyme toward the substrate.

2.5 Quantitative detection of glucose by a smartphone

First, 200 μL GOx solution (1.0 mg mL⁻¹) and 200 μL Glu with different concentrations in NaAc-HAc buffer (pH 3) were incubated at 40 °C in a water bath for 90 min. Then, 225 μL N-CD solution (10 mg mL⁻¹), 0.75 mL 4 mM TMB, and 0.5 mL NaAc-HAc buffer (pH 3) were sequentially added into the mixed solution. After diluting to volume with deionized water, the final solution was further incubated at 50 °C in the water bath for 15 min. Finally, photos of the resulting solutions were taken by a smartphone (iPhone 7), and the 1/L values of the solutions images (HSL color space: H stands for hue, S stands for saturation, and L stands for lightness) were obtained from the images by a free application (Color Analyzer); furthermore, quantitative analysis was based on the excellent linear relationship between the 1/L values and the concentrations of Glu. For comparison, the absorbance (654 nm) of the resulting solutions was detected by a UV-2600 spectrophotometer. Moreover, the corresponding quantitative analysis was performed by the conventional absorbance method.

2.6 Sample treatment

Different food samples (including fruits, vegetables, and cereals purchased from the local supermarket) with different Glu concentrations were chosen as the representative samples for evaluating the analytical performance of the developed method. The extraction of Glu from the food samples was carried out according to a previously reported method (GB 5009.8-2016). For fruit and vegetable samples, 10.0 g peeled fruits or vegetables were chopped and homogenized; for the grain samples, 10.0 g milled rice and husked beans were ground into flour. During extraction, an aliquot of the food sample (1.0 g of the homogenate or cereal flour) was weighed and extracted with 60 mL deionized water. The mixture was kept in an ultrasonic equipment at 60 W for 20 min. After being centrifuged at 5000 rpm for 10 min, the supernatant was collected. Finally, the obtained supernatant was filtered through a 0.45 μm filter and diluted to 100 mL for the subsequent tests.

3.1 Characterization of N-CDs

The N-CDs prepared using locust powder were characterized by HRTEM, XPS, FTIR spectroscopy, UV–vis spectroscopy, and fluorescence spectroscopy. The morphology of the N-CDs was investigated by TEM; the TEM images (Fig. 1a-b) indicate that the N-CDs are uniformly distributed with an average diameter of 2.59 ± 0.3 nm. The HRTEM image (inset of Fig. 1b) indicates high crystallinity of N-CDs; moreover, the lattice fringe spacing of 0.206 nm corresponded to that of the sp² (1120) graphitic crystal phase of graphene (Alam et al., 2015). The elemental composition and chemical bonds in N-CDs were characterized by XPS. Fig. 1c exhibits the survey scan of the N-CDs, where the three major peaks at 286.3, 401.0/406.5, and 532.2 eV can be ascribed to C 1 s, N 1 s, and O 1 s with the atom ratios of 41.14 %, 25.41 %, and 32.83 %, respectively. The high-resolution spectrum of C 1 s, N 1 s, and O 1 s is shown in Fig. S1. It could be concluded that there were C—N/C⚌C, C—C, C—O—C, C⚌O, C—N, N—H, and N—O groups on the surface of N-CDs (Atchudan et al., 2018; Kumar et al., 2017; Li et al., 2015; Zhang and Chen et al., 2014). FTIR spectroscopy was conducted to confirm the surface groups of N-CDs. As shown in Fig. 1d, the peak at 769 cm⁻¹ is ascribed to the C⚌C stretching vibration. The absorption peaks at 1384 cm⁻¹ and 1766 cm⁻¹ were attributed to the C—N and C⚌O bending vibrations, respectively; the peaks at 3016 cm⁻¹ and 3431 cm⁻¹ corresponded to the N—H and O—H stretching vibrations (Qi et al., 2019). The information about elemental composition and chemical bonds obtained from XPS was consistent with the results of FTIR spectroscopy.

Close

Fig. 1

TEM image (a) and HRTEM image (b) of N-CDs. XPS survey spectrum (c) and FTIR spectrum (d) of N-CDs.

Export to PPT

Under excitation in the range of 350–450 nm, the wavelength and intensity of the fluorescence emission were investigated, and the strongest emission peak of N-CDs was observed at 470 nm at the excitation wavelength of 390 nm (Fig. S1d). The fluorescence quantum yield of N-CDs was measured in a solution (PB buffer, pH 4) using quinine sulfate as the standard, and the value was approximately 3.1 %.

The optical properties of N-CDs were studied using fluorescence spectroscopy and UV–vis spectroscopy. Fig. S1d shows that the N-CDs exhibit two typical UV–vis absorption peaks at 300 nm and 350 nm: The peak at 300 nm was ascribed to the π-π* transition for the C⚌C bond, in which the orbitals are sp²-hybridized clusters; the peak at 350 nm corresponded to the n-π* transition of the C⚌O and N—H groups (Yang et al., 2018).

3.2 Peroxidase-like activity of N-CDs

The peroxidase-like activity of N-CDs was investigated by choosing TMB as the chromogenic substrate for oxidation by H₂O₂, and the UV–vis spectra of TMB were obtained under different reaction conditions. As shown in Fig. 2, there is no obvious absorption peak at 654 nm and no significant color change in the TMB solution for TMB + N-CDs and TMB-H₂O₂ (curves a and b, respectively); this indicated that N-CDs could not oxidize TMB and H₂O₂ oxidized TMB very slowly without the catalyst. TMB + H₂O₂ + N-CDs showed a remarkable absorption peak at 654 nm and produced a blue-green color (Fig. 2 inset, photo c), suggesting that N-CDs exhibited peroxidase-like activity for the oxidation of TMB by H₂O₂.

Close

Fig. 2

UV–vis absorption spectra of (a) TMB + N-CDs, (b) TMB + H₂O₂, and (c) TMB + H₂O₂ + N-CDs in acetate buffer at pH 3.0 (TMB: 0.6 mM, H₂O₂: 50 μM). Inset: The corresponding photographs of the above-mentioned reaction solutions.

Export to PPT

3.3 Optimization of catalytic conditions

To further evaluate the peroxidase-like activity of N-CDs, the effects of pH, reaction temperature, and concentrations of TMB and N-CDs on the absorptance at 654 nm were investigated.

The oxidation reaction between H₂O₂ and TMB usually takes place under acidic conditions, and pH is the key factor for the reaction and the color of the reaction products. The catalytic behavior of N-CDs for the oxidation of TMB by H₂O₂ at pH values ranging from 1 to 6 was investigated. As shown in Fig. 3a, the maximum absorptance at 654 nm was achieved at pH 3, and the absorptance decreased considerably beyond this pH. This result indicated that the peroxidase-like activity of N-CDs was lower under weak acidic conditions. Moreover, the lower absorptance at pH 1 and 2 was because the blue oxidized product of TMB was easily converted to another yellow compound at strong acidic conditions. Therefore, pH 3 was chosen as the best pH. The effect of the reaction temperature was studied between 20 °C and 70 °C. The maximum absorptance at 654 nm was obtained when reaction temperature higher than 50 °C (Fig. 3b), and considering the decomposition of H₂O₂ at higher temperature, 50 °C was selected as the optimal reaction temperature.

Close

Fig. 3

Optimized catalytic conditions for the oxidation of TMB by H₂O₂ (pH (a), temperature (b), and concentrations of TMB (c) and N-CDs (d)).

Export to PPT

The concentrations of the substrate (TMB) and the catalyst (N-CDs) are also important factors for the catalytic oxidation reaction. Fig. 3c shows the effect of TMB concentrations on the absorptance at 654 nm in the range of 0.1–1.0 mM. The absorptance at 654 nm increased with the increase in the TMB concentration (0.1–0.4 mM) and then, the absorptance reached a plateau (0.4–0.8 mM). Based on these results, 0.6 mM was chosen as the optimal substrate concentration. The effect of the catalyst (N-CDs) concentration on the absorptance at 654 nm was also investigated. As shown in Fig. 3d, the absorptance at 654 nm increases with the increase in the concentration of N-CDs from 0.2 to 0.5 mg mL⁻¹; this indicates that the catalytic activity of N-CDs depends on their concentration and a higher concentration leads to higher catalytic activity. When the concentration of N-CDs was greater than 0.45 mg mL⁻¹, the solution appeared light yellow, which interfered with the color of the solution after the catalytic reaction. Therefore, it was inferred that the optimal concentration of N-CDs was 0.45 mg mL⁻¹.

3.4 Steady-state kinetic study

The study of the effect of N-CDs as a peroxidase mimic on the reaction kinetics was based on Michaelis-Menten kinetics according to the procedure reported in section 2.4. The initial velocity was calculated from the absorptance at 654 nm, and the molar absorption coefficient (ε) for the oxidized product of TMB (TMB_ox, ε = 39000 M⁻¹ cm⁻¹) was calculated by the Lambert-Beer law. A typical Michaelis-Menten curve was obtained by fixing the H₂O₂ concentration at 0.1 mM (Fig. 4a) and by varying the concentration of TMB from 0.1 to 1.0 mM. Similarly, Fig. 4b shows the Michaelis-Menten curve obtained by fixing the TMB concentration at 0.6 mM and by varying the concentration of H₂O₂ from 0.05 to 10 mM. Then, the Lineweaver-Burk plots were obtained, and they are shown in Fig. 4c and Fig. 4d.

Close

Fig. 4

Steady-state kinetic study of the peroxidase-like activity of N-CDs for TMB oxidation by varying concentrations of (a) TMB and (b) H₂O₂ and (c, d) corresponding Lineweaver-Burk plots.

Export to PPT

The Michaelis-Menten constant (k_m) reflects the affinity between the enzyme and the substrate; a lower value of k_m indicates stronger affinity of the enzyme for the substrate and higher catalytic performance. For N-CDs as a peroxidase mimic, the calculated k_m and V_max values are listed in Table 1 along with the values of horseradish peroxidase (HRP) and other peroxidase mimics reported in the literature. The k_m values of N-CDs toward both TMB and H₂O₂ were significantly lower than those of HRP, indicating that N-CDs exhibited considerably higher affinity than HRP. The k_m value of N-CDs toward H₂O₂ was comparable to those of other previously reported CDs-based peroxidase mimics, whereas the k_m value of N-CDs with TMB as the substrate was the smallest among those of the previously reported CD-based peroxidase mimics; this suggested that the affinity between N-CDs and TMB was considerably higher than that between other previously reported CD-based peroxidase mimics and TMB. Furthermore, it should be pointed out that the N-CDs were obtained via only a self-exothermic reaction using locust powder as the carbon source within 10.0 min. On the contrary, other CD-based peroxidase mimics have been prepared by complicated and time-consuming synthesis procedures with high energy consumption.

Generally, the catalytic mechanism of peroxidase mimics can be ascribed to different factors such as the generation of reactive oxygen species including ·OH and enhanced electron transfer processes (Shi et al., 2019). To investigate the catalytic mechanism of N-CDs, electron spin resonance (ESR) spectroscopy was conducted to confirm the generation of hydroxyl radicals in the catalytic system. 5,5-Dimethyl-1-pyrroline N-oxide (DMPO) was used as the spin trapping agent to react with ·OH; this resulted in the generation of the stable DMPO/·OH spin adduct, which exhibited a typical four-line ESR spectrum with relative intensities in the ratio of 1:2:2:1. The typical ESR spectrum of the DMPO/·OH spin adduct shown in Fig.S3 suggests the generation of hydroxyl radicals in the catalytic oxidation of the substrate TMB by N-CDs.

3.5 Analytical performance of the smartphone-based method

To verify the analytical performance of the smartphone-based analytical method for Glu, images of the testing solutions were taken by a smartphone and then, the color space parameter values were read through an iPhone application (Color Analyzer). To determine the most suitable relationship for quantification by the images of the testing solutions, different color space parameters, including RGB, CMYK, grayscale and HSL, were analyzed. Among all the dependence between the concentration of Glu and the color space parameters, the linear relationship observed between the concentration of Glu and 1/L (here, L stands for lightness in HSL color space) was found to be excellent (Fig. 5a). The linear regression equation was 1/L = 0.000184C (μM) + 0.0148 with a correlation coefficient of 0.9984. The detection limit was calculated to be 1.09 μM (3σ/k), and the relative standard deviation (RSD) was 2.0 % (c = 5 × 10^-5 M). With the increase in the Glu concentration, the color of the testing solutions gradually changed from colorless to blue-green (Fig. 5a, inset). When the concentration of Glu was higher than 20 μM, obvious blue color of the testing solutions was visible. Thus, the developed method has significant potential for the visual detection of Glu.

Close

Fig. 5

Linear relationship between 1/L and the concentration of Glu (a). Linear relationship between the concentration of Glu and absorptance at 654 nm (b).

Export to PPT

For comparison, a standard calibration curve (the absorptance at 654 nm versus Glu concentrations) was also obtained using a UV–vis spectrophotometer under the same optimal reaction conditions. As shown in Fig. 5b, the linear regression equation is A = 0.00789C (μM) + 0.0223 in the range of 5–100 μM with a detection limit of 0.20 μM.

A comparison of some of the features of the developed method with those of other previously reported Glu assays based on CDs with peroxidase-like activity is presented in Table S1. It is clear that the smartphone-based method has a lower detection limit and comparable linear range, which can be explained by the higher affinity between N-CDs and TMB. Furthermore, it should be noted that serum samples are the most popular real samples used for Glu assays. In this study, the smartphone-based method was successfully applied for the analysis of different food samples, including fruits, vegetables, and grains, which can widen the application fields.

3.6 Selectivity of the smartphone-based method

To evaluate the selectivity of the smartphone-based method for Glu detection, the influence of the potential interfering substances, such as metal ions, common amino acids, sugars and Vc, on the absorptance at 654 nm was investigated. The catalytic system containing Glu (100 µM) and selected potential interfering substances was tested according to the recommended procedures. The tolerance concentration of a potential interfering substance was defined as the maximum concentration that causes < 10 % variation in the absorptance at 654 nm for the testing solutions. As evident from Fig. 6a, none of the metal ions (Hg²⁺, Pb²⁺, Na⁺, and K⁺, 100.0 µg mL⁻¹; Ni²⁺, Co²⁺, Mn²⁺, Ca²⁺, Cr³⁺, and Zn²⁺, 50.0 µg mL⁻¹) exhibit significant influence on the absorptance at 654 nm. Amino acids, sugars, and Vc were also used to investigate the selectivity of the method for Glu detection (Fig. 6b). The results indicated that the tolerance concentration of Vc was 1.0 µM, which was considerably lower than that of amino acids and sugars (100 µM). This could be explained by the fact that blue TMB_ox could be reduced to colorless TMB by Vc (Li and Zeng et al., 2021). Fortunately, the amount of Vc in many food samples, such as fruits, vegetables, and grains, is often 1000 times lower than that of glucose or even less. Therefore, the concentration of Vc in the testing solutions used in this method for Glu detection (Glu at the µM level) was considerably lower than 1.0 µM. From the above-mentioned results, it can be concluded that N-CDs can be employed as a highly selective colorimetric probe for Glu in food samples.

Close

Fig. 6

Selectivity investigation of Glu detection in the presence of (a) mental ions and (b) organic compounds in HAc-NaAc buffer (pH 3). Concentrations of organic compounds are 100 μM; concentrations of metal ions are 100 μg mL⁻¹ (Concentrations for substances indicated with * are 50 μg mL⁻¹ and for substances indicated with ** are 1 μM.).

Export to PPT

3.7 Food sample analysis

The excellent feasibility of the smartphone-based colorimetric assay was verified by analyzing different food samples. The analytical results and recoveries for the spiked samples are given in Table 2. The recoveries were in the range of 88.5–109.0 % with the RSD values varying from 0.40 to 6.30 %.

At the same time, the same samples were also analyzed based on the absorptance at 654 nm using a UV–vis spectrophotometer. Both the analytical results are shown in Fig. 7; it is clear from Fig. 7 that the analytical results obtained by the smartphone-based colorimetric method are in good agreement with those obtained by the UV–vis method. Moreover, the t values calculated by using the paired t-test were 0.66, 1.08, and 0.77, and these values were less than t_{95%, 4} = 2.78 (t-test with 95 % confidence level); this indicated that there were no statistically significant differences between the analytical results of the smartphone-based colorimetric method and the UV–vis method. To demonstrate the practical applications of the developed smartphone-based assay, the amounts of Glu in different fruits, vegetables, and grains were analyzed, and the analytical results are given in Table S2. From the above-mentioned results, it can be concluded that the smartphone-based colorimetric assay has excellent accuracy, sensitivity, and repeatability; moreover, the analytical process is very simple, and the analytical cost is considerably low.

Close

Fig. 7

The analytical results for Glu in food samples by smartphone-based colorimetric and UV–vis methods.

Export to PPT