1 Introduction

Melatonin (MT) is an amine hormone produced primarily in the pineal gland of mammals and humans, and its synthesis is regulated by the light cycles so that its levels in the body vary in a circadian rhythm [1,2,3,4]. Normal rhythms are important for good health. When melatonin levels are in a normal rhythm, it has many physiological functions, such as promoting sleep, regulating jet lag, anti-aging, regulating immunity and anti-tumor [5,6,7,8,9,10]. However, when the circadian rhythm is disrupted by jet lag, night shift, etc., melatonin levels will be abnormal and it is also easy to induce emotional abnormalities and cognitive disorders and other problems. The study found that both serotonin(5-HT) [11, 12] and dopamine (DA) [13], which were important for the functions such as control and memory, were significantly altered. The individual detection of DA, 5-HT and MT have been made several attempts [14,15,16]. However, the three substances coexist in the biological systems and all play an important role in the activity of circadian rhythm. Therefore, it is necessary to detection of DA, 5-HT and MT simultaneously, which can not only cause patients little pain and provide low-cost and rapid diagnoses, but also is of great significance to deeply understand circadian rhythms and solve the problems caused by circadian rhythm disorder in mammals comprehensively [17, 18].

Currently, simultaneous detection of mine hormones and other substances include spectroscopy, chromatography, mass spectrometry, enzyme-linked immunosorbent and electrochemical methods [19,20,21]. Electrochemical sensors have the advantages of low cost, high sensitivity, fast detection speed, good portability and strong operability when detecting multiple substances simultaneously [22,23,24,25]. Moreover, many electrochemical sensors have been reported to simultaneous detect multiple substances, including DA, 5-HT or MT. Shahrokhian et al. achieved the simultaneous detection of dopamine (DA) and uric acid (UA) on a gold electrode modified with cysteamine self-assembled monolayer and functionalized multi-walled carbon nanotubes (MWCNTs) [26]. Zhan et al. [27] synthesized a hierarchical nano-porous (HNP) PtTi alloy material for electrochemical simultaneous detection of ascorbic acid (AA), dopamine (DA), and uric acid (UA). Deng et al. [28] constructed an electrochemical sensor for simultaneous detection of AA, DA and UA by a three-dimensional graphene-like carbon frameworks (3DGLCFs), which were facilely prepared via copy-rolysis of polyaniline and nickel nitrate powder, followed by acid etching. However, there are some drawbacks including high cost and difficulty of operation originated from the complicated and time-consuming preparation procedures, which may largely limit these sensors’ wide applications, especially now that research is pursuing green and safe synthesis methods [29]30. And electrochemical peaks overlap or interfere with each other often occur when more substances are detected simultaneously or the electrochemical activity of substances is similar. Therefore, it is necessary to use green and safe methods to synthesize electrodes with better electrocatalysis to solve these problems.

Carbon materials, metal compounds, polymer materials et al. have good catalytic oxidation effect [31, 32], and are widely used in a variety of electrochemical sensors for simultaneous detection [33]. Multi-walled carbon nanotubes (MWCNTs) are a kind of novel carbon-based nanomaterials, which have the good electrical conductivity and can improve the electron-transfer efficiency. The effect of electrochemical detection can be improved by increasing the conductivity of electrode. Sun et al. used a novel NiO/ Multi-walled carbon nanotube (MWCNT)/poly(3,4-ethylenedioxythiophene) (PEDOT) composite to detect DA, 5-HT and tryptophan (Try) simultaneously [34]; when MWCNT was added to other materials, it not only promoted the simultaneous oxidation of DA, which is similar to 5-HT electroactivity, but also promoted the oxidation of the third substance Try. Furthermore, MWCNTs, which are tubes made from many layers of graphene sheets rolled up, have better mechanical properties, more active sites and can be modified with different functional groups. Guan et al. used a functional hybrid carbon nanotube composite, prepared by ultrasonic assembly of carboxylated multi-walled carbon nanotube (MWCNT-COOH) and hydroxylated single-walled carbon nanotube (SWCNT-OH), to build an electrochemical sensor for the simultaneous detection of dopamine and uric acid [35]. Chitosan is a kind of natural polysaccharide, and have the advantages of good film-forming property, biocompatibility and non-toxicity [36], which is often used as an adhesive to prevent the modified material from falling off the electrode surface. Furthermore, the surface of chitosan is rich in hydroxyl and amino functional groups, which can be chemically modified and use for sensor fabrication. When chitosan is combined with MWCNTs, MWCNTs can provide more adhesion sites for chitosan and enhance the mechanical properties of composite materials, and chitosan can increase the functional groups of the composites to have better electrocatalytic performance [35, 37]. In addition to selecting appropriate materials to improve the electrocatalysis of electrodes, the electrocatalysis of electrodes can also be improved by selecting the appropriate material modification method. Zhang et al. [38] compared the electrocatalysis of different electrodes in the simultaneous detection of hydroquinone (HQ), catechol (CC) and resorcinol (RC), and found that the electrode constructed by electrodeposition was best, and the deposition method also had an effect on the electrocatalysis. Khan et al. [37]constructed an electrochemical sensor for simultaneous detection of AA, DA, 5-HT, and L-Trp by modifying poly(L-arginine), reduced graphene oxide and gold nanoparticle layer by layer on the electrode by cyclic voltammetry.

In this article, we developed a composite coated glassy carbon electrode (GCE), which contained carboxylated multi-walled carbon nanotubes (MWCNTs) and chitosan, to detect MT, 5-HT and DA simultaneously, based on the fast electron transfer of CNTs and electrocatalysis of functional groups on chitosan. In addition, the catalytic oxidation effects of the electrodes modified with CS and MWCNTs composites obtained by different methods were analysed. The redox mechanism of the three substances on working electrode was analysed. And the sensor was applied to the detection of MT, 5-HT and DA in human saliva samples. Overall, this study successfully developed an electrochemical sensor for simultaneous detection of circadian related markers such as melatonin, serotonin and dopamine. The detection method presented in this study was portable, fast, low cost and easy to implement.

2 Materials and methods

2.1 Reagents and apparatus

All reagents and solvents adopted in this work were used as received. Carboxylated multi-walled carbon nanotubes (MWCNTs, > 98%), chitosan (CS, Degree of deacetylation ≥ 95%), dopamine (DA, 98%), serotonin (5-HT, AR) and melatonin (MT, 98%) were purchased from Aladdin Holdings Group Co., Ltd (Shanghai, China). Acetic acid (Hac, AR) was purchased from Beijing Tong Guang Fine Chemicals Company (Beijing, China)0.0.01 M phosphate buffer solutions (PBS) were prepared using 0.01 M Na2HPO4 (AR) and 0.01 M NaH2PO4 (AR), which were purchased from Innochem Technology Co., Ltd (Beijing, China). Ultrapure water (18.2 MΩ × cm, 25 ℃) was used throughout the experiment.

The images of surface morphology were gained by field emission scanning electron microscope JSM-7500F (JOEL, Japan). All the electrochemical measurements were performed on a CHI 6043E Electrochemical Workstation (Chenhua Instrument Company of Shanghai, China). A conventional three-electrode cell was used with a glassy carbon electrode (GCE, with a diameter of 3 mm) as the working electrode, an Ag/AgCl electrode as the reference electrode and a platinum wire as the counter electrode. The electrochemical measurements were carried out in 0.01 M PBS at room temperature (20–25 ℃).

2.2 Preparation of modified electrodes

The specific preparation scheme was shown in Scheme 1. Firstly, the GCE was carefully hand-polished with alumina-water slurry (0.3 μm and 0.05 μm) in sequence and then washed successively in ultrapure water and ethanol, and then dried at room temperature. Secondly, aqueous solution of MWCNTs (1.0 mg/mL) was carefully moved to the cleaned GCE surface with a pipette, and then dried in air for 2 h to form MWCNTs/GCE. Next, CS (2.0 mg/mL), dissolved in acetic acid with 1% volume fraction, was modified onto MWCNTs/GCE by cyclic voltammogram (CV) to form e-CS/MWCNTs/GCE. The potential of CV was scanned from − 2.2 V to 1.4 V[3739].

Scheme 1.
scheme 1

Schematic illustration of the construction procedure of e-CS/MWCNTs/GCE and its application in simultaneous detection of DA, 5-HT and MT based on DPV

Furthermore, to compare the influence of modification methods on the detection effect, another three kinds of working electrodes were prepared. The first kind was coated with a mixture of CS and MWCNTs, and the electrode was named CS/MWCNTs/GCE. The other two kinds of working electrodes were to modify CS onto MWCNTs/GCE by I-T methods. And according to the different potentials used, I-T (0.5 V) and I-T(− 0.45 V) were used to distinguish the two kinds of working electrodes.

2.3 Real samples preparation and detection

All samples were collected from the healthy adult, and were obtained by separating the SARSTEDT collection tube at 3000 rpm/min for 5 min at 4 °C. And saliva samples were diluted with 0.01 M PBS. The electrochemical measurement was recorded by differential pulse voltammograms (DPV).

3 Results and discussion

3.1 Morphological characterization of working electrode

Scanning electron microscopy (SEM) was utilized to evaluate the morphological characteristics of different working electrodes. As shown in Fig. 1A, MWCNTs had a curved tubular structure, and showed a network shape due to irregular staggered distribution after being dripped onto the electrode. When MWCNTs and CS were combined by different methods, the morphology of working electrode was different. Figure 1B represented the morphology after the mixture of CS and MWCNTs was dripped on the electrode, where many granular materials were distributed on the surface. In the mixed solution of CS and MWCNTs, MWCNTs used contained carboxyl group and CS was rich in amino group, so there was electrostatic attraction between the two materials, which enabled the reticular MWCNTs to aggregate into granules under the action of CS and CS formed a good film on the particle surface. In addition, since the content of CS in the mixed solution was limited, the CS encapsulation of particles was not complete (Fig. 1C). The surface of e-CS/MWCNTs/GCE (Fig. 1D) obtained by CV electrochemical technology was relatively smooth, which was obviously different from the network structure of MWCNTs surface and the granular materials of CS/MWCNTs/GCE surface. It was thought that a large number of CS wrapped on the surface of MWCNTs, and electrochemical modification method may can cause continuous CS polymerization to form a dense membrane on MWCNTs surface. And the detailed CS polymerization process can be found in Sect. 3.2.

Fig. 1
figure 1

SEM micrographs of MWCNTs/GCE (A), CS/MWCNTs/GCE (B, C) and e-CS/MWCNTs/GCE (D)

3.2 Electrochemical behaviour of DA, 5-HT and MT on working electrode

DPV was used to compare and determine the effect of different working electrodes for simultaneous detection of DA, 5-HT and MT, and the test solutions used in this work, unless otherwise specified, had a default substance concentration of 100 μM and were prepared with 0.01 M PBS (pH = 7). As shown in Fig. 2A, when different working electrodes were used to detect the mixed solution of DA, 5-HT and MT, only e-CS/MWCNTs/GCE showed three distinct electrochemical peaks, which showed that the electrochemical performance of e-CS/MWCNTs/GCE was good. Then the relationship between the electrochemical peaks and the substances was determined by the control variable method. As shown in Fig. 2B, the electrochemical peaks on e-CS/MWCNTs/GCE represented the oxidation of DA, 5-HT and MT from left to right. Furthermore, the peak-to-peak separation potential (Δ Ep) between DA (0.28v) and 5-HT (0.45 V) in simultaneous detection was greater than that of single detection (0.30 V for DA, 0.43 V for 5-HT), which reflected the interference between different substances when simultaneous detection.

Fig. 2
figure 2

DPV curves of different working electrodes in the mixed solution of DA, 5-HT and MT (A). DPV curves of e-CS/MWCNTs/GCE in different solutions (B). CV curves of CS deposited on different working electrodes. Scan rate: 0.05 V/s, cycle number: 6 (C). DPV curves of different working electrodes (Depositing CS onto MWCNTs/GCE in different electrochemical methods) in the mixed solution of DA, 5-HT and MT (D)

In addition, we compared the electrochemical deposition curves of CS on GCE and MWCNTs/GCE to determine the deposition process of CS. As shown in Fig. 2C, by comparing the electrochemical deposition curves of CS on GCE and MWCNTs/GCE, it was found that the reduction peaks appeared at the negative potential of both deposition curves, which was speculated that CS was positively charged by protonation under acidic conditions, so it will be adsorbed to the surface of the working electrode by strong electrostatic attraction at negative potential to form a reduction peak [40, 41]. While only when the electrochemical deposition curve of CS on MWCNTs/GCE appeared the oxidation peak, which was speculated that the amino groups on CS reacted with the carboxyl group on MWCNTs and was similar to poly-amino acids [37, 39]. According to all of the above, we analysed that CS was first adsorbed to MWCNTs in the reduction peak, and then was continuously reacted with the carboxyl group on MWCNTs to form a dense membrane, which can be seen from the morphological characterization. Furthermore, the effects of these two reactions (electrostatic attraction at negative potential and reaction of amino and carboxyl groups at positive potential) on electrochemical simultaneous detection were analysed, that was, I-T was used to electrodeposit chitosan at these two potentials. The time of I-T was 300 s, which was consistent with the time of CV, so the influence of time on the electrochemical treatment technology itself was excluded. As shown in Fig. 2D, CV method was still the best electrochemical treatment method to modify CS onto MWCNTs. Therefore, the e-CS/MWCNTs/GCE described in subsequent articles was prepared by CV method.

3.3 Optimization of operating conditions

3.3.1 Optimization of electrode preparation conditions

The main function of MWCNTs was to improve the electrical conductivity, so the droplet coating volume was optimized in solution of 5.0 mM [Fe (CN)6]3− containing 0.1 M KCl and evaluated by electrochemical impedance spectroscopy (EIS). As we can see from Fig. 3A, the lowest resistance could be obtained at 10 μL of the MWCNTs modified GCE. It was proposed that when the coating amounts were excessive, material spilled and hindered the effective electron-transfer between electrode surface and the biomolecules; while when the amount of coating was too small, the film was too thin to enhance the effective electron transfer between the electrode surface and the biomolecules. Hence, 10 μL of MWCNTs (1 mg/ml) was used to modify GCE in the following experiments.

Fig. 3
figure 3

EIS of MWCNTs/GCE with different volumes of MWCNTs (A). DPV peak currents of DA, 5-HT and MT under various CV cycle numbers (B) and scan rates (D). The last cycle of CV curve with different cycle number (C)

When optimizing the cycle number of CV, we found that when the cycle number was six, the oxidation peak current of DA, 5-HT and MT were best (Fig. 3B), and CV curve has the lowest oxidation peak current (Fig. 3C). It was concluded that when the cycle number was six, the chitosan that can reacted with MWCNTs was the lowest and the electron-transfer efficiency of MWCNTs reached the highest. The same method was used to optimize scan rate of CV. And it was found that when the scan rate was 0.05 v/s (Fig. 3D), the oxidation peak currents of DA, 5-HT and MT were highest.

3.3.2 Optimization of detection conditions

DPV method was used to study the effect of pH on the simultaneous detection of DA, 5-HT and MT, and the pH of the detection solution was changed by adjusting the PBS. As shown in Fig. 4 A, the DPV peak currents of DA, 5-HT and MT increased with an increasing pH from 5.8 to 7.4, and then stabilized when pH was higher than 7.4. Hence, the following experiments were carried out in the detection solution with a pH value of 7.4, which was close to the pH of human body fluid. Meanwhile, the linear relationship of oxidation peak potential and pH was obtained, suggested the direct involvement of protons in the oxidation process (Fig. 4B). And the obtained slope values were very close to the theoretical slope values of − 59 mV/pH, which meaned that the equal number of protons and electrons were transferred during the electrochemical reaction [42]. Referring to earlier reports, two-protons and two-electrons processes were involved in the electrochemical reaction of DA, 5-HT and MT [34, 43]. The proposed mechanisms of DA, 5-HT and MT oxidation reactions were shown in Scheme 1 [34, 43]. With the increase of electrolyte pH, more and more OH were contained in the solution, which promoted the oxidation of substances, so the oxidation potential of these substances decreased continuously.

Fig. 4
figure 4

A DPV curves of e-CS/MWCNTs/GCE in different pH solutions containing DA, 5-HT and MT. B The linear relationship between DPV peak potentials and pH values

3.4 Electrochemical active surface area

The electrochemical active surface area of the electrode was not a geometric area. In this paper, the large specific surface area of MWCNTs used can provide abundant active sites. Therefore, to better describe the effective contact area between the catalytic site on the electrode and the electrolyte, it was necessary to calculate the electrochemical active surface area of the electrode, rather than simply using the geometric area of the electrode to represent the number of catalytic sites. The chronocoulometry (CC) technique was adopted to detect the electrochemical active surface area of GCE, CS/MWCNTs/GCE and e-CS/MWCNTs/GCE in solution of 5.0 mM [Fe (CN)6]3− containing 0.1 M KCl. The results were shown in Fig. 5A, and the relationships between time(t) and charge(Q) were plotted according to the integrated Cottrell expression (Eq. (1)) [44].

$$Q = \left( {\frac{{2nFAcD^{\frac{1}{2}} t^{\frac{1}{2}} }}{{\pi^{\frac{1}{2}} }}} \right) + Q_{dl} + nFAT_{0}$$
(1)

where n stands for the number of transferred electrons from [Fe (CN)6]3− to [Fe (CN)6]4− per molecule, that means n = 1; F is the Faraday constant; A is the effective surface area of electrode; c is the concentration of the electrolyte, which is 5.0 mM in this paper; D is the diffusion coefficient, which is equal to 7.6 × 10−6 cm2/s for 5.0 mM [Fe (CN)6]3−. Besides, Qld is the capacitive charge, and nFAГ0 is the charge from the reduction of adsorbed redox marker. As shown in Fig. 5B, the linear relationship between Q and t1/2 were obtained with the slopes of 98.82 μC/s1/2 for the GCE, 542.3 μC/s1/2 for the CS/MWCNTs/GCE and 699.2 μC/s1/2 for the e-CS/MWCNTs/GCE. The electrochemical active surface area could be calculated to be 0.0641 cm2, 0.3614 cm2 and 0.4659 cm2 for the GCE, CS/MWCNTs/GCE, and e-CS/MWCNTs/GCE, respectively. Obviously, the electrochemical active surface area of e-CS/MWCNTs/GCE was larger than that of the CS/MWCNTs/GCE, and approximately 7 times as large as that of GCE, resulted from the chitosan functionalized on the multi-walled carbon nanotubes matrix by electrochemical method. The largest electrochemical active surface area of e-CS/MWCNTs/GCE indicated that there were more active sites on the electrode, so the electrode had better electrocatalytic activity. Therefore, in terms of the electrochemical detection performance, e-CS/MWCNTs/GCE can oxidize three substances at the same time, especially MT at high potential.

Fig. 5
figure 5

A CC curves of different working electrodes in [Fe (CN)6]3−. Initial potential: − 0.1 V; final potential: 0.7 V; number of steps: 2; pulse width: 0.25 s; sample interval: 0.002. B The linear relationship between Q and t1/2

3.5 Effect of scan rate

The effect of scan rate (in the range of 10–300 mV/s) on the electrochemical detection of DA, 5-HT and MT was investigated via CV for e-CS/MWCNTs/GCE in a mixture solution of DA, 5-HT and MT (Fig. 6A). The results showed that peak currents(I) vary linearly with scan rates(v) (Fig. 6B) for three substances. According to the Randle’s Sevcik equation, the electrochemical oxidation of DA, 5-HT and MT on e-CS/MWCNTs/GCE were an adsorption-controlled process [42]. And obviously, DPV can exhibit a much more defined peak and higher peak current than CV.

Fig. 6
figure 6

A CV curves of e-CS/MWCNTs/GCE in the mixed solution of DA, 5-HT and MT at different scan rates. B The linear relationship between current(I) and scan rate (v)

3.6 Simultaneous detection of DA, 5-HT and MT

Based on the above analysis, simultaneous quantification determination of DA, 5-HT and MT were conducted under the optimum experimental conditions. As shown in Fig. 7A, the oxidation peak currents of DA, 5-HT and MT increased simultaneously with the increasing concentrations. Thus, the linear relationships between peak current (I) and concentration(c) were obtained. For 5-HT (Fig. 7C), the relationship between c and I exhibited two linear sections with the slopes of 0.04319 and 0.01193 μA/μM, and the corresponding concentrations ranged from 9 to 120 μM and from 120 to 1000 μM, respectively. Figure 7B and D showed the calibration curves of DA and MT with the linear range from 20 to 1000 μM. The LODs were 12, 10, 22 μM (S/N = 3) for DA, 5-HT and MT respectively. The electrochemical sensors reported previously for the simultaneous detection of DA, 5-HT and MT were few, so electrochemical sensors for detecting DA, 5-HT or MT were listed in Table1. Compared with these modified electrodes, e-CS/MWCNTs/GCE could detect MT, 5-HT and DA simultaneously and two of them have oxidation potential as low as 0.17 V, which means that more substances may can be detected simultaneously within the same potential range. Besides, the sensor was convenient in both material synthesis and electrode preparation.

Fig. 7
figure 7

A DPV curves of e-CS/MWCNTs/GCE in the mixed solutions of DA, 5-HT and MT with different concentrations. Calibration curves of DA (B), 5-HT (C) and MT (D)

Table 1 Comparison of differently modified electrodes for detection of DA, 5-HT or MT

3.7 Reproducibility, repeatability and stability

The reproducibility of e-CS/MWCNTs/GCE was essential to the mass production of sensors. Ten of e-CS/MWCNTs/GCEs were prepared using the same preparation process, and investigated by comparing the peak currents of DA, 5-HT and MT. The relative standard deviations (RSD) of the peak currents were calculated to be 2.52%, 1.42% and 2.36% respectively, indicating a good reproducibility of e-CS/MWCNTs/GCE. However, it was recommended that the proposed e-CS/MWCNTs/GCE can be utilized only once because of the contamination of oxide, which was why the affordable and abundant materials were selected for modification in the beginning. To evaluate the long-term stability of the proposed sensors, the response currents of e-CS/MWCNTs/GCEs to DA, 5-HT and MT after storage at room temperature for 7 days and 14 days was monitored. The peak currents of DA, 5-HT and MT were basically stable in the range of 98.61%-99.51%, 96.52%-98.83% and 99.43%-100.14% of the initial values respectively, demonstrated that the sensor had excellent stability.

3.8 Real sample analysis

Human saliva samples were selected as biological samples to evaluate the practical of e-CS/MWCNTs/GCE. The subjects were healthy adults with regular schedules. And to ensure that the saliva collected were not contaminated, the subjects were asked to brush teeth first, and then no food for half an hour prior to collection, and no water for ten minutes prior to collection. The time of collection was fixed in the morning and noon. The results were shown in Table 2. According to the rhythm of DA, 5-HT and MT, saliva collected in the morning was used as blank samples of DA, and the saliva collected at noon was used as blank samples of 5-HT and MT [12, 48]. It was noted that the calculated recovery of 79.19–119.78% are obtained, and RSD of MT, 5-HT and DA were calculated to be 2.64%, 3.35% and 1.14% respectively, indicating a good satisfactory practical application effect of e-CS/MWCNTs/GCEs.

Table 2 Determination of DA, 5-HT and MT in real saliva sample

4 Conclusion

A glassy carbon electrode modified with a composite consisting of electrodeposited chitosan and carboxylated multi-walled carbon nanotubes (e-CS/MWCNTs/GCE) was used as a working electrode for simultaneous determination of dopamine (DA), serotonin (5-HT) and melatonin (MT), which were based on the fast electron transfer of MWCNTs and the electrocatalysis of functional groups on chitosan. And it was found that electrochemical modification method may can cause continuous CS polymerization on MWCNTs surface to form a dense membrane with more active sites on the electrode, so a largest electrochemical active surface area and better electrocatalytic activity of e-CS/MWCNTs/GCE. The electrochemical sensor e-CS/MWCNTs/GCE was a rapid, simple and green method for the simultaneous detection of DA, 5-HT, and MT with a wide detection range of 20-1000 μM, 9-1000 μM and 20-1000 μM. And the actual detection effect of this sensor in human saliva samples was analysed. The substance content in saliva at different time conformed to the characteristics of rhythm change. The proposed sensor was expected to monitor the effects of rhythm on DA, 5-HT, and MT, which was of great significance for understanding the rhythm disorder and restoring rhythm.