Carbon nanotube/carbon fiber electrodes via chemical vapor deposition for simultaneous determination of ascorbic acid, dopamine and uric acid

2.1 Materials

DA, AA, UA and Ni(II) nitrate hexahydrate were purchased from Sigma-Aldrich. Carbon fibers without sizing agents (T700, 12 K) were prepared by Shandong University (Jinan, China). Dopamine Hydrochloride Injection (DHI) was purchased from Shandong provincial hospital health physical examination center, P. R. China. Silver conductive glue (ZB2567) and epoxy adhesive were purchased from Electronics Nanjing Zhongbei co., LTD. All the other chemicals were analytical reagent grade and used without further purification. Phosphate buffer solutions (PBS, Na₂HPO₄/NaH₂PO₄ 0.1 M) of pH 6–8 were used as supporting electrolyte. High purity distilled-deionized water was used throughout the experiments. Solutions were deoxygenated by purging with pre-purified nitrogen gas. Pt and Ag/AgCl electrodes were purchased from CH Instruments.

2.2 Synthesis of CNTs on the surface of carbon fiber

The homogeneous carbon nanotubes (CNTs) grew on the surface of carbon fibers via chemical vapor deposition (CVD) by tip-growth mechanism, shown in Fig. 1(a). Previously, electrochemical anodic oxidation was used to increase reactive oxygen species (carbonyl, carboxyl and hydroxyl group) on the surface of carbon fibers (Fan et al., 2015; Yuan et al., 2012). The catalyst for CVD method in this work stemmed from metallic salt solution by dissolving Ni(II) nitrate hexahydrate (0.1 mol·L⁻¹) in ethanol and impregnating CF for 10 min. The CF bundles were taken out and dried at 80 °C for 1 h. Then they were placed in the middle of a vertical graphite barrels reactor, heated to 450 °C at 10 °C·min⁻¹ in N₂ and kept for 1 h with H₂ at a flow of 5 L·min⁻¹ to convert the layer of catalyst precursor into metallic Ni nanoparticles. The reactor temperature was subsequently raised to 600 °C at 10 °C·min⁻¹ in N₂ and CNTs were synthesized directly on the surface of carbon fibers by replenished with a mixture of C₂H₂, H₂, and N₂ for 5, 25, 35 and 40 min (CNTs graft rate of weight gain ratio was 15%, 48%, 65% and 70%). The feeding rates of these gases were 5, 5, and 10 L·min⁻¹ for C₂H₂, H₂, and N₂, respectively. Then, the gas flow was replaced by nitrogen at cooling phase with a flow rate of 5 L·min⁻¹ until the reactor was cool to room temperature. During the whole CVD process, the pressure of the reactor was kept at 0.02 MPa. The CNTs grafted carbon fibers (CNTs/CF) were ultimately cleaned in the ultrasonic bath with acetone.

Close

Fig. 1

(a) Schematic illustration for CNTs growing on carbon fiber surface and (b–e) SEM images of different surfaces ((b): bare CF; (c): Ni nanoparticles depositing on the carbon fibers; (d) and (e): CNTs modified CF (inset was TEM image of CNT).

Export to PPT

2.3 Preparation of the CNTs/CFE

The preparation procedure of CNTs grafted carbon fibers electrode (CNTs/CFE) was as following: carbon fiber was firstly cleaned by ultrasonic for 5 min successively in acetone, alcohol and double-distilled water. A small bunch (diameter: 30 μm) of carbon fiber was used for each CFE. One end of the carbon fiber (length: ∼1.5 cm) was connected to a silver wire with silver conductive glue. After the glue was fully solidified, the carbon fiber-silver wire was inserted into a polypropylene micro pipette with about 1 mm length of the carbon fiber exposed over its tip. And the tip was placed on the outer flame of the alcohol lamp for a very short time (∼0.5 s). The tip was fused and the carbon fiber was tightly sealed in the polypropylene micro pipette. At the other end of the silver wire was fixed to the polypropylene micro pipette by epoxy adhesive. Finally, the electrode was rinsed thoroughly again with acetone and copious amounts of double-distilled water for 3 min, then dried under N₂ atmosphere.

2.4 Equipment and measurements

The morphology of carbon fibers was investigated by the scanning electron microscope (SEM, JEOL, SU-70) operated at 15.0 kV. LabRam-1B Raman spectrometer were used to analyze carbon fiber surface structure after modification with He-Ne laser, power 4.3 mW, wavelength 632.8 nm, resolution 1 cm⁻¹ and time 50 s. An automated adsorption apparatus (V-Sorb 2800P) was employed for the measurement of surface characteristics of carbon fibers by nitrogen adsorption. Specific surface areas of the carbon samples were evaluated with the application of Brunauer–Emmett–Teller (BET) equation.

Electrochemical measurements were carried out on a CHI-832C electrochemical analyzer (CH Instruments, China). A three-electrode cell was used with a CFE with a diameter of 30 μm as working electrode, a Pt wire as counter electrode and an Ag/AgCl electrode as reference electrode, respectively. The pH values of solutions were measured with a PB-10 pH meter (Renhe Instruments, China). All the measurements were performed at room temperature.

3.1 Morphological and structural characterization of electrodes

The surface of carbon fibers was modified by electrochemical anodic oxidation (bare CF was shown in Fig. 1(b)) and then coated with the catalyst precursor by dip coating; there was a formation of uniform and fine particles with a narrow size distribution observed on the surface of carbon fibers, about 20.5 ± 4.5 nm (Fig. 1(c)), which was the prerequisite for obtaining uniform aligned CNTs by tip growth mechanism (Gohier et al., 2008). Fig. 1(d, e) showed the morphology of CNTs growing on carbon fibers at 600 °C. The CF surface was wrapped with carbon nanotubes. This was attributed to the large specific surface area of Ni nanoparticles which provided countless active sites for the decomposition of hydrocarbon, and meanwhile had a longer diffusion length which was favorable for the diffusion of carbon atoms (Fan et al., 2015). Besides, The CNT was multi-walled with a diameter of 15–30 nm, shown in the inset image of Fig. 1(e).

Laser Raman scattering obtained to identify microstructure of fibers before and after CNTs growth, was given in Fig. 2(a–c). The spectra showed two major bands in about 1340 cm⁻¹ and 1590 cm⁻¹, which represented disordered structure (D peak) and graphite structure (G peak), respectively. The D peak belonged to graphite crystallite A_1g vibration mode; the G peak belonged to the C—C bond stretching vibration within graphite lattice plane and vibration model was E_2g (Yuan et al., 2012). The Raman spectra in Fig. 1(e) showed that the D band intensity of CNTs/CF was much higher than the bare carbon fibers. D line was also observed on the G band which raised due to disorders resulting from finite size effect or lattice distortion (Sharma and Lakkad, 2011). Moreover, the samples showed peak intensity increased and FWHM decreased in case of CNTs grown carbon fibers, led to the orderly structure of graphite, which was corresponded with the SEM results above.

Close

Fig. 2

Raman spectra of carbon fiber surface (a) and curve fitting of Raman spectrum for carbon fiber: (b) bare CF; (c) CNTs modified CF.

Export to PPT

G peak and D peak overlapped seriously in Raman spectrum curve, so Gaussian-Lorentz mixed function was used to peak-differentiation-imitating analysis, shown in Fig. 2 (b, c). Finally, R (D line and G line intensity ratio, I_D/I_G) could characterize sp² bonding carbon atoms of carbon fiber. The R of CNTs modified CF were greater than that of bare carbon fiber was due to the presence of 2-D disordered structures in the basal plane which was quite common in pyrolytic carbon materials synthesized using CVD, shown in Table 1. The I_D/I_G ratio of CF-CNTs (1.882) increased compared with that of the bare CF spectrum (1.138). This change of R value could be assigned to more sp³ defects / disorders and smaller average size (or more amount) of the in-plane graphitic crystallite sp² domains upon formation of the CNTs. This meant that both the disorder degree and the unsaturated carbon atom number on the CF surface increased.

In the D line, low wavenumber side (about 1250 cm⁻¹) of the acromion was called D″ line, and the D″ line was the stretching vibration of C—C bond in the structure of fat and the structure of alkenes, or the vibration of sp³ bond of diamond like structure. The D′ line near the 1500 cm⁻¹ was generally considered to be caused by the presence of an amorphous carbon or some organic functional groups. The increscent of I_D″/I_G ratio and I_D′/I_G radio could be assigned to the damage of fat and alkenes structure (Yuan et al., 2012). CNTs had high crystallization degree, which had excellent electrical conductivity and high dielectric loss and beneficial to electrochemical signal.

3.2 Cyclic voltammetric behaviors of AA, DA and UA on modified electrodes

Cyclic voltammetry (CV) behaviors of the CNTs/CFE were characterized by adding DA, UA and AA in 0.1 M PBS ranging from −0.2 to 0.8 V. Compared to bare CF electrodes, the CNTs/CFE showed an apparently larger background current and well-defined; and resolved voltammetry responses for the direct oxidation of DA, AA and UA in Fig. 3a–c. The oxidation peak current at the modified electrode was almost dozens of times larger than oxidation current at unmodified CFE. The amperometry response significantly increased, showing stronger conductivity and effective electrocatalyst for the oxidation of DA, UA and AA in the surface of the CNTs/CFE.

Close

Fig. 3

(a) CV responses at the bare CF electrode and CF-CNTs electrodes in 0.1 M PBS (pH 7.0) containing 0.8 mM DA, (b) CV in 0.1 M PBS (pH 7.0) containing 10 mM AA, (c) CV in 0.1 M PBS (pH 7.0) containing 2.5 mM UA, (d) CVs in 0.1 M PBS (pH 6.0) containing the mixture of 5 mM AA, 0.2 mM DA and 1 mM UA scan rate: 50 mV/s and (e) BET surface area.

Export to PPT

Fig. 3d showed the CVs of DA, AA and UA mixture solution at CFE and CNT_S/CFE. At CFE, the oxidation peak of DA, AA and UA was broad. The cause might be that the oxidation peaks of AA, DA and UA overlapped together and came into large peak. Compared to CFE electrode, three well-defined and remarkable oxidation peaks with obviously enhanced peak currents were observed at approximately −0.05 V for AA, 0.15 V for DA and 0.37 V for UA at CNTs/CFE.

Above all, the modified electrode showed excellent electrocatalytic properties and perfect selectivity. When the surface grafting rate increased, a great improvement of oxidation peak intensity was achieved. There was a remarkable increase in the BET surface area which increased with the increase of surface grafting rate, shown in Fig. 3e. The results were ascribed to the following reason: CNTs had a catalytically active surface and a very high aspect ratio. The area of the electrode surface was larger, and the electrical conductivity was stronger. It was shown that the oxidation process of dopamine on the modified electrode was reversible for the sharp peak, thus the coulomb efficiency was improved, the reaction cycle performance was better, and the detection was more practical.

3.3 EIS analysis

The EIS technique was used to elucidate the relationship between electron transfer resistance and electrode surface morphology (Fei et al., 2016). Fig. 4 showed the EIS results of the pristine CFE and CNTs/CFE with the redox couple of [Fe(CN)₆]³⁻/[Fe(CN)₆]^4 − in the range of 0.01 Hz to 100 kHz under open circuit potential conditions. It was well-known that a representative impedance spectrum was written of a semi-circle portion and a linear portion. The semicircle diameter at higher frequencies corresponded to the electron-transfer resistance (R_ct), and the linear part at lower frequencies reflected the diffusion process (Filik et al., 2016).

Close

Fig. 4

The EIS patterns of bare CFE, 15%CNTs/CFE and 70% CNTs CFE in 5 mM [Fe(CN)₆]³⁻/⁴⁻ (1:1) containing 0.1 M KCl.

Export to PPT

The values of charge transfer resistance were 715, 70 and 46 Ω at CFE, 15% CNTs/CFE and 70% CNTs/CFE respectively. This data demonstrated that the enhanced redox activity of [Fe(CN)₆]³⁻/[Fe(CN)₆]^4 − probe at CNTs/CFE. This decrease was attributed to the conductive properties of the CNTs, which would benefit electron transfer of the electrochemical probe. The result displayed a small semicircle in high frequency region indicating small interface impedance, which was closely related to the interfacial charge-transfer resistance between the electrode and the detection solution (Dou et al., 2018) (Wang et al., 2019). The modified material could decrease the electron transfer resistance and improve the electron transport process. The conclusions of EIS were conformed to CV curves in Fig. 3.

3.4 Effect of pH and scan rate on the electro-oxidation of DA

Proton was always involved in the electrochemical reaction of organic compound and exerts significant impact on the reaction speed (Yu et al., 2014). Therefore, the effect of pH was investigated by CV method in 0.1 M PBS from pH 6.0 to 8.0. Fig. 5 showed the CV results in 0.1 M PBS containing 200 μM DA at different pH values. It was clear that the electro-oxidation behavior of DA was dependent on the pH of the solution. The peak potential for oxidation of DA shifted to more positive potentials with the changes of the pH of the solutions from alkaline to acidic values. It was apparent that the highest peak currents were obtained at pH 7.0. Besides, the value of human blood and urine was close to 7.0. Therefore, pH 7.0 was chosen for the subsequent simultaneous determination of AA, DA and UA.

Close

Fig. 5

(a) Effect of pH on the electro-oxidation of DA: CV responses of 0.2 mM DA on SWNT modified electrodes in 0.1 M PBS at different pH values (6.0, 6.5, 7.0, 7.5, and 8.0, respectively), Scan rate: 50 mV/s; (b) CV responses of 0.15 mM DA on CNTs/CFE in pH 7.0 PBS at various scan rates (20.0, 40.0, 65.0, 95.0, 120.0, 140.0, 160.0, 180.0 and 200.0 mV/s, respectively); (c) Relationship between the peak currents and the scan rates (the linear regression equation was expressed as: I_pa = 0.4050 v + 3.238 (R = 0.9975, red) and I_pc = −0.4541 v - 12.10 (R = 0.9987, black), where v was the scan rate.).

Export to PPT

The effect of scan rate (in the range of 20–200 mV·s⁻¹, pH 7.0) of electrochemical behaviors of CNTs/CFE was researched in 0.1 mol/L PBS containing 0.15 mmol/L DA. From Fig. 5(b), we could see that the oxidation peak currents of DA increased along with increasing of scan rates. The plot of cathodic and anodic peak current was linear (R² = 0.99, Fig. 5(c)), indicating that the electrochemical process was a diffusion-controlled reaction. All above results clearly validated the significant selectivity of CNTs/CFE toward DA.

3.5 Separation of electrochemical responses to DA, AA and UA at the CNTs modified CFE

Fig. 6(a–c) depicted the DPV behaviors of the CNTs/CFE in the presence of AA, DA and UA in 0.1 M PBS (pH 7.0), respectively. The oxidation peak of AA, DA and UA appeared at about 0 V, 0.20 V and 0.02 V, respectively. The results showed that peak current increases with rising levels of dopamine (linear relationship: I_DA =  −0.4584 C − 21.4329 (R² = 0.9932)). The linear relationship between the peak current and concentration of AA was I_AA =  −0.0129 C − 4.7243 (R² = 0.9905). The peak current for the UA was found to be proportional to their concentration in the range of 0.02–0.8 mM, with correlation coefficients of 0.9850. The detection limits were 10.0 μM, 0.03 μM and 0.6 μM for AA, DA and UA (S/N = 3), respectively.

Close

Fig. 6

DPV curves of CNTs/CFE in 0.1 M PBS (pH = 7.0) on the electrooxidation of (a) 25.6–2000.3 μM AA, (b) 5.0–120.6 μM DA, (c) 20.0–800.0 μM UA and (d) Selective detection of DA in the presence of UA and AA, Scan rate: 50 mV/s.

Export to PPT

Detection of the individual concentration of these compounds from the overlapped oxidation peak was impossible. Differently, on the CNTs/CFE, three well-defined peaks by DPV were well resolved at about −20 mV, 220 mV and 360 mV, corresponding to the oxidation of AA, DA and UA, respectively, indicating that the simultaneous determination of the three species was feasible, shown in Fig. 6(d). These facts proved the good selectivity of the proposed method and exhibit excellent electrocatalytic effect towards DA, in preference to AA and UA.

This selective detection might be due to the high density of oxygen-containing groups on the CNTs providing a selective interface via hydrogen bonds with the proton-donating group of AA, DA, and UA, beneficial to the catalysis of the oxidation of three substances at the surface of the electrode. Also, it had been discovered that the conducting loosen structure on the CNTs/CFE surface can change the mass transport, facilitating the discrimination of many species which oxidized or reduced at similar potentials, resulting in promoting the electron transfer between the interface of analyte and electrode just like a nanoscale electrode (Chen, 2018). They also indicated that the electrode had a large electroactive surface area, according to the Randles–Sevcik equation ( $I\mathrm{p}=2.69\times {10}^5{\mathit{AD}}^{1/2}{n}^{3/2}{v}^{1/2}C$ ), and the average value of CNTs/CFE was (9.15 ± 0.21) × 10⁻² cm² (n = 6) (Kang et al., 2007). Another reason might be the π–π interactions between the CNTs and these molecules which can promote the charge transfer of AA, DA and UA. Furthermore, this detection level would overlap with upper half of clinically relevant ranges which were linked with several diseases.

3.6 Study of stability, reproducibility and interferences

To examine the repeatability of the biosensor, by measuring the DPV response in two individual experiments with a 10-day interval, good stability was shown with the result that 89.04% of its initial signal was retained after 10 days of storage period at 4 °C.

Under pH = 7.0 and Scan rate = 50 mV/s, the prepared sensor was measured by CVs for a 20-cycle successive scan, and a less than <1% deviation of initial responds was observed. The results indicated that the peak currents of electrode changed less than 2.44%.

Additionally, the modified electrodes were stored for 30 days at room temperature before the experiment. No obvious change in peak current was observed; the peak current intensity of DA, AA and UA decayed by 1.05%, 2.80% and 2.51%, indicating an acceptable stability of the biosensor.

3.7 Sample analysis

Fig. 7 showed the practical analytical utility of the method was demonstrated by the measurement of DA·HCl in the real samples of dopamine hydrochloride injection by using the standard addition method without any preliminary treatment. All samples were diluted with 0.1 M PBS (pH 7.0) before the measurement.

Close

Fig. 7

Selective detection of DA·HCl in real sample of dopamine hydrochloride injection. (a) DPV responses at CNTs/CFE in 0.1 mol/L PBS (pH 7.0) containing different concentrations of DA·HCl from 25 μmol/L to 125 μmol/L (25, 30, 45, 80 and 125 μmol/L, respectively). Scan rate: 50 mV/s. (b) Relationship between the oxidation peaks current versus concentration of DA·HCl.

Export to PPT

As presented in Fig. 7a, the DPV response decreased with the increase of DA·HCl concentration. The recovery of the spiked samples ranged between 93.2% and 104.0%. Therefore, the proposed method might be applied to the determination of DA in real biological samples with satisfactory results. The linear regression equation was I_p = −28.46–0.3258 C with R² = 0.9942. The detection limit (LOD) calculated was 0.1 uM (S/N = 3). The obtained LOD was comparable with the sensitive sensor reported previously (Nsabimana et al., 2017).

To test the uric acid and ascorbic acid by CNT/CFE in real samples, human urine and the standard addition method were used and the results showed in Table 2. The recovery of real urine samples with different concentrations of UA/AA was between 98.68% and 104.20% with relatively low divergence.