1 Introduction

The use of malachite green (Scheme 1a) has been widespread as colorants in industry and as antimicrobial agents due to low cost and ready availability. Since 1990, researchers have found that this azo dye is highly toxic, carcinogenic, and mutagenic [1–3]. Therefore, it is very significant to monitor malachite green in aquatic products.

Scheme 1
scheme 1

Chemical structures of MG (a) and PMG (b)

Full size image

Quantitative analysis of malachite green in fish tissue today is usually performed by liquid chromatography, and liquid chromatography with mass and tandem mass spectrometry [4–6]. In the recent years, immunoassay method has also been used as a rapid, specific, and sensitive method [7–9]. These approaches possess high sensitivity and excellent selectivity. However, these methods require expensive instruments and time-consuming pretreatment processes. Electrochemical methods are alternative quantitative methods and have received an increasing attention in the detection of harmful residues, due to its high sensitivity, good selectivity, rapid response, and low cost. Up to date, the electrochemical determination of MG is rarely reported [10]. So, new modified electrodes with good catalytic performance for the determination of MG are still needed. Graphene is a two-dimensional honeycomb network of sp2-hybridized carbon atoms [11]. Due to the remarkable conductivity, high surface-to-volume ratio, and good biocompatibility, several works of electrochemical sensors and biosensors have been introduced based on graphene materials [12, 13]. Graphene sheets, which are smaller than 100 nm, are called graphene quantum dots (GQDs). Owing to quantum confinement and edge effects, GQDs have various electronic and optoelectronic properties and can be an excellent candidate for construction of electrochemical sensor. A few of electrochemical sensors constructed with GQDs have been reported [14, 15]. It has been reported that the integration of carbon-based materials and metal nanoparticles [16, 17] can offer synergistic effect in electrocatalytic applications. The carbon-based materials can provide an ideal graphitic carbon surface for the assembly, dispersion, and stabilization of metal nanoparticles. So we have reason to expect the GQDs–Au composite would have the same effect.

In this study, a new graphene quantum dots–gold nanoparticles multilayers–modified glassy carbon electrode was successfully fabricated to detect MG. Its preparation involved two steps: (1) Electrodes were modified with GQDs and (2) electrodes were electrodeposited with Au nanoparticles in a HAuCl4 solution. By layer-by-layer methods, the anodic potential of MG was reduced and the anodic current was obviously increased.

2 Experimental

2.1 Reagents and apparatus

High-purity graphite rods were purchased from the Qingdao Tennry Carbon Co. (China). Malachite green was obtained from Tianjin Basf Chemical Co., Ltd. (China). By dissolving 146.0 mg of MG into 100.0 mL redistilled deionized water, 4 mmol L−1 of MG stock solution was prepared. Chloroauric acid and sulfuric acid were purchased from Pure Crystal Reagent Ltd. (China) and used as received. Other chemicals were of analytical reagent grade, and all the solutions were prepared with redistilled deionized water. Phosphate-buffered saline (PBS) was prepared by dissolving 8.94 g sodium chloride, 0.77 g disodium hydrogen phosphate, and 0.18 g potassium dihydrogen phosphate in 1,000 mL redistilled deionized water and adjusted to the proper pH 7.4 using 1 mol L−1 hydrochloric acid and 1 mol L−1 sodium hydroxide.

Electrochemical experiments were performed with CHI 660C electrochemical workstation (Shanghai Chenhua Co., China) with a conventional three-electrode cell. The working electrode is a bare glassy carbon electrode (d = 3 mm, GCE) or modified glassy carbon electrode. A saturated calomel electrode (SCE) and a platinum wire were used as reference electrode and auxiliary electrode, respectively. The differential pulse voltammetry (DPV) was carried out with the parameters of increment potential, 0.004 V; pulse amplitude, 0.05 V; pulse width, 0.05 s; sample width, 0.0167 s; pulse period, 0.2 s; and quiet time, 2 s. The transmission electron microscope (TEM) image was obtained at JEM-2100TEM (Japan). The scanning electron microscopy (SEM) was performed on a Hitachi S-3000 N instrument (Japan).

2.2 Preparation of GQDs

GQDs were prepared according to the previous report [18]. Briefly, graphene oxide sheets were prepared from natural graphite powder by a modified Hummers method [19, 20]. Micrometer-sized graphene sheets (GSs) were obtained by thermal deoxidization of graphene oxide sheets in a tube furnace at 200–300 °C for 2 h with a heating rate of 5 °C min−1 in nitrogen atmosphere. GSs (0.05 g) were oxidized in a concentrated H2SO4 (10 mL) and HNO3 (30 mL) mixed solution for 20 h under mild ultrasonication (500 W, 40 kHz). The mixture was then diluted with deionized water (250 mL) and filtered through 0.22-μm microporous membrane to remove the acids. Purified oxidized GSs (0.2 g) were redispersed in deionized water (40 mL), and the pH was adjusted to 8 with NaOH (1 mol L−1). The suspension was transferred to a poly(tetrafluoroethylene) autoclave (50 mL) and heated at 200 °C for 10 h. After cooling to room temperature, the resulting black suspension was filtered through a 0.22-μm microporous membrane and a brown filter solution was separated. The colloidal solution still contained some large graphene nanoparticles (50–200 nm). So, the colloidal solution was further dialyzed in a dialysis bag (retained molecular weight 3,500 Da) overnight.

2.3 Fabrication of modified electrodes

Before modification, the bare GCE was polished to mirror with 0.3 and 0.05 μm alumina slurry on microcloth pads, rinsed thoroughly with redistilled deionized water, and then washed successively with anhydrous alcohol and redistilled deionized water in an ultrasonic bath, respectively. Finally, it was dried in air before use. For preparation of modified electrode, 2 mg mL−1 GQDs solution was first prepared by dispersing GQDs in redistilled deionized water, following ultrasonication for 1 h. With a microinjector, 10 μL of GQDs solution was dropped on the surface of GCE to obtain GQDs/GCE. The electrodeposition of Au nanoparticles at the GQDs/GCE was performed in a HAuCl4 (3 mmol L−1) solution by applying a negative potential of −0.2 V (vs. SCE) for 100 s, which was negative enough for the electrochemical reduction of chloroauric acid to Au nanoparticles. Multilayers of GQDs/Au were then constructed by repeated alternative dropping of the GQD solution and electrodeposition of Au nanoparticles. The modified electrode was denoted in the text as (GQDs/Au) n /GCE where n stands for the number of bilayers. Au/GCE was constructed by direct electrodeposition of Au nanoparticles at the GCE. The modified GCE was stored at 4 °C in a refrigerator when it was not in use. The same cleaned procedure was applied to the electrode before it was modified every time.

2.4 Preparation of fish sample

The salmon was purchased from local supermarket. The fish samples were prepared according to the previous report [9]. The fish was filleted, the skin and bones were removed, and the muscles were minced and deep-frozen before the detection. To the muscle samples (2 g), 10 μL MG solution (4 mmol L−1) was added. Samples were allowed to equilibrate for 15 min before extraction. The sample was mixed with 4 mL 2 mol L−1 NaCl in PBS 7.4 by a vortex mixing for 1 min. Afterward, 2 mL HCl solution (1 mol L−1) was mixed with the sample by another vortex mixing for 1 min. After centrifugation (2,000 rpm, 5 min), 4 mL solution was pipetted from underneath the fat layer into a tube and 5 mL dichloromethane was added. The tube was shaken overhead for 1 min. After centrifugation for 5 min, at 2,000 rpm, 5 mL of the upper layer was removed. Three milliliters of the dichloromethane phase was pipetted into a 10-mL glass tube and evaporated with nitrogen until dried at 50 °C. Residues were dissolved with 25 mL 100 % methanol. Further, 10 mL 0.05 mol L−1 H2SO4 solution was added and mixed by a vortex mixing for several minutes. The attained solution was used as DPV detection.

3 Results and discussion

3.1 Characterization of GQDs and surface morphological studies on GQDs/Au multilayers

Figure 1a showed the TEM image of the colloidal solution of GQDs. Their diameters are mainly distributed in the range of 5–15 nm (average diameter 10 nm). From Fig. 1a, it can be seen that most of GQDs were uniformly dispersed, and a small amount of GQDs aggregated. Figure 1b showed the SEM image of (GQDs/Au)1 multilayers. The diameters of gold nanoparticles are mainly distributed in the range of 150–300 nm.

Fig. 1
figure 1

a TEM image of GQDs. b SEM image of (GQDs/Au)1 multilayers

Full size image

3.2 Characterization of electrochemical behavior of (GQDs/Au) n /GCE

Cyclic voltammograms of 5 mmol L−1 Fe(CN) 3−/4−6 at GCE (a), GQDs/GCE (b), (GQDs/Au)1/GCE (c), and (GQDs/Au)4/GCE (d) were shown in Fig. 2. A couple of well-defined redox peaks were observed at bare GCE with peak-to-peak separation (ΔE p) of 98 mV. But when the electrode was coated with GQDs, an increase in ΔE p and a decrease in peak current (I p) were observed. This phenomenon indicated there were electrostatic repulsive interactions between GQDs and [Fe(CN)6]3−/4− anions. But when the electrode was coated with Au nanoparticles, an obvious increase in peak current was observed. This observation proved that the modified electrode showed fine catalytic activity. Furthermore, the current signal at the (GQDs/Au)4/GCE was stronger than at the (GQDs/Au)1/GCE.

Fig. 2
figure 2

Cyclic voltammograms of 5 mmol L−1 Fe(CN) 3−/4−6 (1:1) in 0.1 mol L−1 KCl solution recorded at GCE (a), GQDs/GCE (b), (GQDs/Au)1/GCE (c), and (GQDs/Au)4/GCE (d). Scan rate: 100 mV s−1

Full size image

3.3 Cyclic voltammetric behaviors of MG

Figure 3 showed cyclic voltammograms of GCE (a), GQDs/GCE (b), Au/GCE (c), and (GQDs/Au)4/GCE (d, e) in the presence (a–c, e) and absence (d) of 4.0 × 10−5 mol L−1 MG in 0.05 mol L−1 H2SO4 solution at a scan rate of 100 mV s−1. When MG was absent, no redox peak was observed at the (GQDs/Au)4/GCE, suggesting that GQDs and Au nanoparticles were electroinactive in the selected potential region. The peaks of oxidation and reduction appeared 0.502 and 0.446 V, respectively, at the (GQDs/Au)4/GCE. The redox reaction may be the mutual transform of MG and poly(malachite green) (PMG) (Scheme 1b) [21–26]. The oxidation peak current at the (GQDs/Au)4/GCE was about 2 times higher than at the GQDs/GCE in 4.0 × 10−5 mol L−1 MG. The reduction peak current at the (GQDs/Au)4/GCE was about 5 times higher than at the GQDs/GCE. This observation may be attributed to the synergistic effects of GQDs and Au nanoparticles.

Fig. 3
figure 3

Cyclic voltammograms of GCE (a), GQDs/GCE (b), Au/GCE (c), and (GQDs/Au)4/GCE (d, e) in the absence (d) and presence (a–c, e) of 4.0 × 10−5 mol L−1 MG in 0.05 mol L−1 H2SO4 solution. Scan rate: 100 mV s−1

Full size image

3.4 Optimization parameters

3.4.1 Influence of the number of bilayers

The effect of the number of bilayers on the oxidation of MG was investigated by DPV. Figure 4 showed the effect of the number of bilayers on the oxidation of MG. It revealed that the oxidation peak current increased gradually up to the number of films 4 and then decreased. The oxidation peak current at the (GQDs/Au)4/GCE was about 2 times higher than at the (GQDs/Au)1/GCE in 1.0 × 10−5 mol L−1 MG. This observation may be attributed to two aspects: With the increasing of n, the effective surface area of the electrode increased; at the same time, the electron transfer resistance increased.

Fig. 4
figure 4

Influence of the number of bilayers on the oxidation response of 1.0 × 10−5 mol L−1 MG at the modified electrode. Accumulation time: 400 s; open-circuit accumulation

Full size image

3.4.2 Influence of accumulation time

The effect of accumulation time on the oxidation of MG was investigated by DPV. Figure 5 illustrated the relationship between the oxidation peak current of MG and accumulation time under open-circuit condition. The oxidation peak current increased gradually with accumulation time up to 400 s and then leveled off. This phenomenon could be attributed to saturated adsorption of MG at the modified electrode surface. So, in the following experiments, the accumulation time was fixed at 400 s.

Fig. 5
figure 5

Differential pulse voltammograms of 1.0 × 10−5 mol L−1 MG in 0.05 mol L−1 H2SO4 solution at the (GQDs/Au)4/GCE at different accumulation time (from a to f: 0, 100, 200, 300, 400, and 500 s) under open-circuit condition

Full size image

3.4.3 Influence of scan rate

The influence of scan rate on the electrochemical responses of MG was studied by cyclic voltammetry, and the results were shown in Fig. 6a. Fig. 6b showed the peak current increased linearly with the scan rate in the range of 40 to 800 mV s−1 and the linear regression equation was calculated as follows: I pa = −0.0853 v − 6.817 (μA, mV s−1, R = 0.998), I pc = 0.0737 v + 4.080 (μA, mV s−1, R = 0.998). The result indicated that electrode process was controlled by adsorption.

Fig. 6
figure 6

a Cyclic voltammograms of 4.0 × 10−5 mol L−1 MG in 0.05 mol L−1 H2SO4 solution at the (GQDs/Au)4/GCE at different scan rates (from a to l: 40, 100, 160, 200, 240, 320, 400, 480, 560, 640, 720, and 800 mV s−1). b The plot of peak current versus scan rate. c The relationship between E p and the logarithm of scan rate (log v)

Full size image

From Fig. 6c, it can be seen that E pa changed linearly versus lnν with a linear regression equation of E pa = 0.0595 log v + 0.5413 (V, mV s−1, R = 0.994), and E pc changed linearly versus lnν with a linear regression equation of E pc = −0.063 log v + 0.4112 (V, mV s−1, R = 0.993), in the range of 300 to 800 mV s−1. According to the following Eq. (1) [27], the anodic and cathodic peak potentials were linearly dependent on the logarithm of the scan rates (v) with slopes of 2.3 RT/(1 − α)nF and −2.3 RT/αnF (T = 298.15 K, R = 8.314 J mol−1 K−1, and F = 96,480 C mol−1), respectively. Thus, the charge-transfer coefficient (α) was calculated to be 0.47.

$$ \log \frac{{v_{a} }}{{v_{c} }} = \log \frac{\alpha }{1 - \alpha }{\kern 1pt} $$
(1)

3.5 Electrochemical effective surface area

Figure 7 showed the plots of Q − t and Q − t 1/2 at bare GCE and (GQDs/Au)4/GCE in 1 × 10−4 mol L−1 K3[Fe(CN)6] solution containing 1 mol L−1 KCl. From the slope of the plot of Q versus t 1/2, the electrochemical effective surface area for bare GCE and (GQDs/Au)4/GCE can be calculated by chronocoulometry using 1 × 10−4 mol L−1 K3[Fe(CN)6] as model complex (the diffusion coefficient D of K3[Fe(CN)6] is 7.6 × 10−6 cm2 s−1 [28]) based on Eq. (2) given by Anson [29]:

$$ Q(t) = \frac{{2nFAcD^{1/2} t^{1/2} }}{{\pi^{1/2} }} + Q_{\text{dl}} {\kern 1pt} {\kern 1pt} + {\kern 1pt} {\kern 1pt} {\kern 1pt} Q_{\text{ads}} $$
(2)

where n is electron transfer number, A is the surface area of the working electrode, c is the concentration of substrate, D is the diffusion coefficient, Q dl is double-layer charge which could be eliminated by background subtraction, and Q ads is adsorption charge. Based on the slope of the linear relationship between Q and t 1/2, A can be calculated to be 0.093 and 0.460 cm2 for GCE (Fig. 7 inset a) and (GQDs/Au)4/GCE (Fig. 7 inset b), respectively. The results indicated that the electrode effective surface area was increased after modification of GCE.

Fig. 7
figure 7

Plot of Q−t curve of GCE (a) and (GQDs/Au)4/GCE (b) in 1 × 10−4 mol L−1 K3[Fe(CN)6] containing 1 mol L−1 KCl. Inset: plot of Q − t 1/2 curve at GCE (a) and (GQDs/Au)4/GCE (b)

Full size image

3.6 Calibration and limit of detection

Figure 8a showed the DPV curves of MG at various concentrations. The oxidation peak current was proportional to the concentration of MG in the range of 4.0 × 10−7–1.0 × 10−5 mol L−1 with a linear regression equation of I pc = −2.040 × 106 c − 5.483 (μA, mol L−1, R = 0.992) (Fig. 8b), and the detection limit was estimated to be 0.1 μmol L−1 at a signal-to-noise ratio of 3, indicating an easy method to detect MG was obtained. Hence, it can be concluded that the modified electrode showed excellent sensitivity for MG.

Fig. 8
figure 8

a Differential pulse voltammograms of MG at the (GQDs/Au)4/GCE with various concentrations (from a to f: 4.0 × 10−7, 2.0 × 10−6, 4.0 × 10−6, 6.0 × 10−6, 8.0 × 10−6, and 1.0 × 10−5 mol L−1). b Calibration curve for MG at (GQDs/Au)4/GCE

Full size image

3.7 Determination of MG in real sample

In order to further evaluate the practicality of the proposed method, standard addition method was adopted to estimate the accuracy. The recoveries ranged from 96.25 to 98.00 % as shown in Table 1. Therefore, (GQDs/Au)4/GCE was able to determine the concentration of MG in fish samples.

Table 1 Recovery values of MG obtained for three salmon samples (n = 3)
Full size table

3.8 Stability, reproducibility, and selectivity of the modified electrodes

The stability and reproducibility were evaluated by measuring the electrochemical response of 4.0 × 10−6 mol L−1 MG at the (GQDs/Au)4/GCE. The modified electrode retained 92 % of its initial response after it was kept in refrigerator at 4 °C for 10 days, which indicated that the modified electrode had good stability. The reproducibility was investigated by five parallel modified electrodes. The relative standard deviation was 6.90 %, which suggested that the (GQDs/Au)4/GCE displayed good reproducibility.

To evaluate the selectivity of the fabricated electrode, the influence of some complexes and inorganic ions was examined in 0.05 mol L−1 H2SO4 solution containing 4.0 × 10−6 mol L−1 MG. The results suggested that 200-fold concentration of ascorbic acid, uric acid, dopamine, caffeine, vitamin E, xanthine, hypoxanthine, and leukomalachite green had no influence on the signals of MG with deviations below 10 %. Additionally, some inorganic ions such as 400-fold concentration of Na+, Ca2+, Mg2+, Fe3+, Al3+, Zn2+, Cu2+, Cl−, \( {\text{SO}}_{4}^{2 - } ,{\text{ PO}}_{4}^{3 - } \), and \( {\text{NO}}_{3}^{ - } \) had no influence on MG determination.

4 Conclusions

In this paper, a new graphene quantum dots–gold nanoparticles multilayers–modified electrode was successfully fabricated to detect MG. The electrode effective surface area was increased four times after modification. Owing to the synergistic effects of GQDs and Au nanoparticles, the oxidation peak potential of MG was decreased, and the oxidation peak current was increased. Furthermore, by layer-by-layer methods, the oxidation peak current at the (GQDs/Au) n /GCE was increased by 132 % in 1.0 × 10−5 mol L−1 MG. The application of this modified electrode in analysis of real samples was also evaluated with good performance. This proposed method was a promising electrochemical approach for tracing MG.