Different Electrochemical Behavior of Cationic Dopamine from Anionic Ascorbic Acid and DOPAC at CNT Yarn Microelectrodes

CFMEs were fabricated by inserting a single T650 carbon fiber into a glass capillary (0.68 mm ID × 1.2 mm OD), and were then pulled to form a sharp tip with fiber exposed. The exposed fiber was then cut to be 100 μm, and sealed with Epoxy Resin 828 (Miller-Stephenson, Danbury, CT). The insulated electrodes were left overnight, and then were placed in 100 °C oven for 2 h and then 150 °C for 12 h to cure.

CNTYMEs are disk electrodes made with commercial carbon nanotube yarns with diameter of 50 μm (Nanoworld Lab, Department of Chemical and Environmental Engineering, University of Cincinnati). A glass capillary was pulled by a glass puller (model PE-21, Narishige, Tokyo, Japan) to form a cylinder tip, which was then cut to form an open-tip with diameter of approximately 70 μm. A piece of yarn with 2 cm length was soaked in isopropanol, and inserted into the glass capillary from the open tip carefully during soaking. With the assistance of surface tension, the CNT yarn was placed at the glass tip without structure deformation. After insulation with Epon epoxy Resin 828, the electrode was then polished to a 45° angle with a fine diamond abrasive plate (Sutter Instruments model BV-10, Novato, CA). The polishing process for an electrode was control to be less than 3 min and the electrode was then soaked in isopropanol to remove surface impurities. The whole structure of the disk electrode is shown in Fig. S1 (available online at stacks.iop.org/JES/169/026506/mmedia).

Dopamine hydrochloride, L-ascorbic acid and 3,4-Dihydroxyphenylacetic acid (DOPAC) were purchased from Sigma-Aldrich (St. Louis, MO). Compounds were dissolved in 0.1 M HClO₄ to make 10 mM stock solution, and were diluted with PBS buffer (131.25 mM NaCl, 3.00 mM KCl, 10 mM NaH₂PO₄, 1.2 mM MgCl₂, 2.0 mM Na₂SO₄, and 1.2 mM CaCl₂) to corresponding concentration (dopamine 1 μM, DOPAC 20 μM and L-ascorbic acid 20 μM). Solutions were further adjusted to different pH (pH 4, pH 7.4 and pH 8.5) for FSCV detection.

Fast-scan cyclic voltammetry was performed in a flow cell with dual syringe pump pumping buffer and dopamine into a cell with a six-port switching valve (Valco Instruments Co. Inc. Houston, TX). The potentiostat was a Dagan with a 10 MΩ headstage (Dagan, Minnesota). Electrodes were filled with KCl solution to be connected to Ag/AgCl reference electrode, and dopamine waveform (−0.4 V to 1.3 V at a scan rate of 400 V/s from frequencies from 10 Hz to 100 Hz) was applied. The data was analyzed with HDCV software developed at University of North Carolina, Chapel Hill.

SEM images (spot size 2, accelerating voltage 2 kV) of CFMEs and CNTYMEs was performed on ZEISS Merlin High- Resolution Scanning electron microscope (Oberkochen, Germany), provided by Center for Nanophase Materials Science (CNMS), Oak Ridge National Laboratory.

The simulations of electrostatic effects with analytes with different charges were performed in COMSOL Multiphysics (Burlington, MA). The 2D model of CFMEs was simplified as a line segment, since CFMEs are cylinder electrodes with little surface structure. The model of CNTYMEs was composed of dense CNT arrays standing vertically to show surface roughness of a disk electrode. Further details can be found in support information (Fig. S2). "Electrostatics" and "Transport of Diluted Species" modules were coupled to simulate mass transport and analyte migration in an electric field.

Figure 1 shows SEM images of the electrodes used in this study: a cylindrical CFME (Fig. 1A) with a diameter of 7 μm and a disk CNTYME (Fig. 1B) using a 50 μm diameter CNT yarn., which was made by twisting long CNT arrays into bundles. ^19,24 The CFME has smooth striations on the surface, whereas the CNTYME has micron-depth surface roughness with twisted CNT bundles. ^19,24 After polishing, a CNT forest is formed of vertically aligned CNT arrays, and crevices are observed between the CNT bundles.

Zoom In Zoom Out Reset image size

First, we studied CFMEs by comparing dopamine, DOPAC, and ascorbic acid at different FSCV repetition frequencies. For FSCV flow cell experiments, we run buffer past the electrode for 5 s, then analyte for 5 s, and finally switch back to buffer for 5 s. The CVs shown are collected at the end of the analyte injection, approximately the 50th CV at 10 Hz, and the 500th CV at 100 Hz. Figure 2A shows the CVs of dopamine at 10 Hz and 100 Hz. Dopamine is reversibly oxidized to dopamine-o-quinone (DOQ) via two-electron transfer (Scheme 1A), and the oxidation peak occurs at 0.6 V and reduction at −0.2 V at the back scan. The peak-to-peak separation is larger than the conventional cyclic voltammetry because of the fast scan rate (400 V s⁻¹), so that the voltage ramp is faster than electron transfer. The reduction peak current is smaller than the primary oxidation one is due to the desorption of DOQ molecules. ^10,43,44 The primary peak current at 100 Hz (11 ± 3 nA) is much smaller than that at 10 Hz (26 ± 2 nA) because there is less time at the holding potential for dopamine to be adsorbed on the electrode. ²⁵ In Fig. 2B, 20 μM DOPAC has a similar CV shape as dopamine at 10 Hz, with similar oxidation peak potentials. In Fig. 2C, ascorbic acid has a wider oxidation peak at around 0.5 V and a reduction peak at −0.2 V in the back scan. At CFMEs, both DOPAC and ascorbic acid have broadened CV peaks at 100 Hz, which makes selective detection more challenging because the peaks may overlap. Figure 2D plots the primary current vs frequency and shows that the current for dopamine dramatically decays with increasing FSCV repetition frequency. In contrast, DOPAC and ascorbic acid have increasing peak currents at high frequencies, which is explained by electrostatic effects. The anions will be repelled during the holding time at a negative voltage. At higher FSCV repetition frequencies, there are secondary oxidation peaks at a lower potential (approximately 0.15 to 0.2 V) for dopamine and DOPAC due to cyclization reactions. ²⁶

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Next, the repetition frequency experiment was performed at CNTYMEs. The primary oxidation current for dopamine (Fig. 3A) at 0.6 V is only slightly smaller at 100 Hz than 10 Hz, and the secondary peak current increases with frequency. DOPAC (Fig. 3B) has a similar CV shape to dopamine, and similar trends, with a small drop in primary peak current (at 0.6 V) and an increase in secondary peak current (at around 0 V) at 100 Hz. Ascorbic acid (Fig. 3C) has an oxidation peak at approximately 0.5 V, and the signal drops dramatically for the 100 Hz repetition frequency. Both DOPAC and ascorbic acid have a peak current at the switching potential 1.3 V, which may be due to changes in the background current after increased adsorption of compounds with more oxide groups. ²⁷ Secondary oxidation peaks of dopamine and DOPAC that appear at a lower potential are enhanced at CNTYMEs and are easier to distinguish at 100 Hz than at CFMEs.

Zoom In Zoom Out Reset image size

Figure 3D shows that ascorbic acid has a large decrease in primary oxidation current with frequency at CNTYMEs but that the current declines only slightly for DOPAC and dopamine. This pattern is a contrast to CFMEs, where dopamine decreased with FSCV repetition frequency but AA and DOPAC did not. Given the possibility of ascorbic acid fouling, the response at different frequencies was tested for AA, then the electrode was exposed to AA for 25 cycles, and the current responses at different frequencies tested again. Figure S3 shows the currents did not drop after 25 cycles, which indicates that no fouling happens during AA redox cycling at CNTYMEs. The decreasing currents for AA at the CNTYME at high repetition frequencies may be due to restricted diffusion, as the solution may not easily reach into the crevices with rapid repetition frequencies. In contrast, DOPAC redox is adsorption controlled so it has a less current drop and may be less susceptible to effects of restricted diffusion. AA is a diffusion-controlled process and the oxidation is irreversible and products are consumed in the following hydration reaction, whereas DOPAC with a quasi-reversible redox mechanism that partially recycles the primary product back to DA. Thus, restricted diffusion and irreversibility may cause the larger current drop for AA than for DOPAC.

To understand the electrostatic effects for different charged molecules, a model of the electric field was built with COMSOL Multiphysics; details can be found in the supporting information. In Fig. 4, an electrode (white rectangle representing half CFME) was placed in an electrochemical cell with zero potential. When applying a potential of −0.4 V to the electrode, an electric field was generated around the microelectrode, as shown in Fig. 4A. Then electroactive species were introduced to the system from the bulk solution on the right side. To simply show the effect of migration in the analyte movements, convection was not considered in the model. The concentration is 1 μM everywhere in the electrochemical cell when there is no migration (arrows represent the solution flux in the electrochemical system). Figure 4C shows the concentration profile of a cationic species, which not only diffuses but also migrates to the electrode. The positively charged molecule is attracted by the negative holding potential applied at the electrode by electrostatic forces; thus, concentrations of cations are enriched near the electrode surface. Figure 4D is the concentration of an anionic species, and the concentration near the electrode is smaller because anions are repelled by the negative potential applied to the electrode. With an opposite direction of migration due to repulsion, fewer molecules approach to the electrode surface so the amount that undergo redox reactions is less. In the real experiment, a flow injection system will introduce convection with molecules continuously flowing to the electrode.

Zoom In Zoom Out Reset image size

The concentration of anions was 20 μM instead of 1 μM, so the real concentration near the electrode is larger than the results in the model. In this study, dopamine (pKa = 9.27) is cationic, whereas DOPAC (pKa = 4.25) and ascorbic acid (pKa = 4.2) are anionic at physiological pH 7.4. ^27,45 Thus, dopamine is present at a higher concentration on the surface (such as Fig. 4C), while DOPAC and ascorbic acid are at lower concentrations at the electrode surface (such as Fig. 4D). The electrostatic effect explains the different current trends of dopamine, DOPAC, and ascorbic acid with increasing frequencies. At 10 Hz, the holding time at −0.4 V between each cycle is 91.5 ms, and the time dramatically decreases when repetition frequency increases. With less time at negative potential, there is less attraction of dopamine to the electrode, so the current goes down as the result at CFMEs shows (Fig. 2A). But for DOPAC and ascorbic acid that are repelled by the electrode at holding potential, the shorter time is less time for repulsion, so the current trends higher (Figs. 2C, 2D).

The electrostatic effect is the same at both CFMEs and CNTYMEs, and both electrodes show fast time responses of three compounds based on the current vs time plots (Fig. S4), however, the frequency effects are different at CNTYMEs (Fig. 3D). Dopamine has less primary current decay at CNTYMEs (Fig. 3D) than at CFMEs (Fig. 2D) and DOPAC has a slight decrease at CNTYMEs. AA decreases with frequency at CNTYMEs in contrast to its increasing trend at CFMEs. The reason for the different current response at CNTYMEs is due to the trapping effect introduced by the creviced surface structure as shown in SEM image (Fig. 1B). At CNTYMEs, the crevices on the surface with micron depth act as many thin layer cells that trap molecules in the CNT forests, limiting diffusion on the FSCV time scale. ^21,29 Thus, cations such as dopamine are trapped near the surface and can be readily detected after redox cycling at high repetition frequencies. Anions will be repelled by electrostatic forces, but those that are at the surface will also be trapped in the crevices and consumed in the redox reactions. The current for ascorbic acid decreases with frequency at CNTYMEs because it is an irreversible redox reaction (Scheme 1C) and AA is consumed more quickly with high repetition frequency in the thin layer cell. Dopamine and DOPAC have less current decay than AA, because they undergo quasi-reversible redox reactions (Schemes 1A, 1B), and side products are recycled in the thin layer cell. As a result, current trends at CNTYMEs with high surface roughness are different from CFMEs with smooth surface.

At CFMEs only one oxidation peak is typically present for FSCV detection at 10 Hz, but additional peaks can be detected at CNTYMEs, especially at high repetition frequencies (Figs. 2 and 3). For DA and DOPAC, these extra peaks are attributed to the cyclization reactions following an ECE (electron transfer-chemical-electron transfer) mechanism as shown in Scheme 1. Dopamine (Scheme 1A) is oxidized to dopamine-o-quinone (DOQ) via two-electron transfer, the primary oxidation reaction, and the peak appears at 0.6 V in FSCV. After forming DOQ, it can either be reduced back to dopamine or undergo a cyclization reaction by intramolecular Michael addition to generate leucodopamine. Leucodopamine is then electrochemically oxidized to dopamine-chrome, which causes a secondary oxidation current that appears at 0.15 V. ^26,30 DOPAC, a major metabolite of dopamine, has same catechol structure and undergoes a similar redox pathway, with an initial oxidation reaction that is then cyclized to benzofuran-2,5,6-triol, which can be oxidized to furanoquinone. ^14,31 The redox pathway of L-ascorbic acid (Scheme 1C) is irreversible because the hydration reaction after oxidation largely consumes the oxidant dehydro-L-ascorbic acid to generate a more stable product. ^32,33 However, side reactions have been reported, such as ring cleavage. ⁶

At CNTYMEs, secondary oxidation peaks are enhanced (Fig. 3), because the cyclization products that are made in the crevices are trapped in the crevices on the FSCV time scale and therefore are detected electrochemically. ²¹ At 100 Hz, there is even less time for molecules to escape (1.5 ms between scans), leading to further enhancement. In addition, there is an increasing number of reactions in a unit time at high frequencies, and more primary products are consumed since the cyclization reaction is irreversible. With 10 times more reactions at 100 Hz, there are more secondary products accumulated in the crevices, thus the secondary peak currents are higher. The different CV responses of CFMEs and CNTYMEs caused by the different surface structure facilitates the discrimination of analytes by different secondary peaks, but it is not enough to provide selectivity among three analytes with FSCV.

For in vivo studies, detection of neurochemicals is performed at physiological pH (7.4), but the redox mechanisms are pH dependent and worth understanding for detection ex vivo. Figure 5 shows dopamine, DOPAC, and AA detection at CFMEs at pH 4 and 8.5. For all three molecules, reduction peaks are right shifted in acid and left shifted in base, because the reactions are pH dependent, and more protons in solution favors reduction, while fewer protons favors oxidation. ^34,2 Dopamine (pKa = 9.27) is a cation at both pH 4 and 8.5 and is attracted by the electrode at a negative holding potential, so the current is frequency dependent. DOPAC (pKa = 4.25) is cationic at pH 4, so it has a larger current due to migration to the surface, but the current decays at 100 Hz. However, at pH 8.5 DOPAC is an anion that is normally repelled by the electrode at the holding potential, so the overall current is much smaller (due to differences in migration) but the current increases with repetition frequency as there is less time at negative potentials. Ascorbic acid has a pKa of 4.2, and it is positively charged at pH 4, so it has a larger current due to migration and a signal that decays with repetition frequency. Dopamine and DOPAC have enhanced secondary peaks at acidic pH, that become more apparent at 100 Hz. The normalized trends in Figs. 5G, 5H show different electrostatic effect of three analytes at acidic or basic conditions. Cationic dopamine has decreasing currents at both pH values, whereas the currents for DOPAC and AA decrease with frequency at pH 4 but increase at pH 8.5.

Zoom In Zoom Out Reset image size

Figure 6 shows CVs from CNTYMEs for dopamine, DOPAC, and AA at acidic and basic pH. Similar to CFMEs, even in CNTYMEs, the peaks shift due to the pH dependence of the reactions, as the reduction peaks are right shifted in acid and left shifted in base. For dopamine (Figs. 6A–6B), the primary peak currents at CNTYMEs are smaller at 100 Hz, and secondary peaks are enhanced at both pH values. DOPAC has the same peak positions as dopamine, but a high current at 1.3 V at CNTYMEs because of the background current change introduced by the adsorption of molecules. Secondary peak currents for DOPAC are enhanced at both pH values, especially at pH 4, where the secondary peak dominates at 100 Hz. Secondary peak currents for both dopamine and DOPAC are larger at pH 8.5 than at pH 7.4, because nitrogen and oxygen in the chains become more nucleophilic and cyclization reactions are more facile. ^35,36 At pH 4, AA has primary current decreasing at 100 Hz, and secondary peaks are seen in the cyclic voltammogram. At pH 8.5, the primary peak has nearly disappeared at 100 Hz, and only a secondary reaction is observed. The normalized trends of three analytes at CNTYMEs (Figs. 6G, 6H) are mostly the same as those at CFMEs (Figs. 5G, 5H), but the decreasing trends are flatter. However, AA has dramatically decreasing currents at pH 8.5 which is opposite that at CFMEs.

Zoom In Zoom Out Reset image size

Ascorbic acid is interesting because it has a different redox mechanism at different pH values (Scheme 1D). ^2,32,37 As Scheme 1C shows, in acid or neutral solution, ascorbic acid is oxidized to dehydroascorbic acid via two-electron transfer. The redox reaction is followed by a hydration reaction to form a stable product, which causes it to be an irreversible process. In Fig. 6E, the CV shape at pH 4 (10 Hz) is more reversible at CNTYMEs because dehydroascorbic acid that does not undergo hydration is trapped and can be cycled back to AA. The additional wave at the lower potential in the cyclic voltammogram of ascorbic acid is attributed to the background change since trapped molecules are easily adsorbed on the CNTYMEs. When pH is above 8 (Scheme 1D), ascorbic acid opens its lactone ring and is oxidized to 2,3-diketogulonic acid, the primary reaction. However, in basic solutions, 2,3-diketogulonic acid dissociates and is enolized. ^2,33 Two-electron oxidation of the enolate generates 2,3,4-triketogulonic acid, which causes a secondary oxidation peak at a lower potential. ³³ As an unstable compound acid with four adjoining carbonyl groups, 2,3,4-triketogulonic acid degenerates quickly. The CVs obtained at CNTYMEs in Fig. 6F show a large difference between 10 Hz and 100 Hz at pH 8.5. At 10 Hz, the primary oxidation appears at around 0.6 V and secondary peak is not easily observed but at 100 Hz, the largest peak in the CV is due to a secondary reaction and the primary peak is small. Note that CV shapes vary in time with FSCV and these CVs were taken at the end of a 5 s flow injection. Figure S5 plots CVs of the three compounds at 100 Hz, pH 8.5 at different time during a 5 s flow cell injection. The results show that the secondary peaks are not observed at the beginning of the injection, but grow in with time, while primary peak currents decrease over time as the primary products are consumed.

Multiple analyte detection is difficult in FSCV due to the extremely fast voltage ramp that leads to peak broadening or overlapping when analytes have similar formal potentials. Codetection experiments are most often performed with conventional cyclic voltammetry or differential pulse voltammetry, because peak separation is easier at slower scan or step rates. ^8,38 Here, our strategy is to use secondary peaks at different pH values to identify dopamine in the mixture solutions. Figure 7 plots CVs of three analytes at CNTYMEs and mixtures solutions at pH 8.5, 100 Hz. In this condition, dopamine (Fig. 7A) has a favored cyclization reaction while DOPAC (Fig. 7B) has similar peak potentials, but a much smaller secondary peak, as well as an extra wave at the switching potential. At pH 8.5, ascorbic acid (Fig. 7C) has a very large secondary peak at 0.1 V as well as a peak at the switching potential, but little primary peak because the product is used up. Figure 7D is a CV obtained in solutions of 1 μM dopamine with 20 μM DOPAC and the while the primary peak is a combination of dopamine and DOPAC, the large secondary peak is attributed to dopamine. The additional peak at switching potential also indicates the existence of DOPAC in the mixture solution. Figure 7E is a CV of 1 μM dopamine with 20 μM ascorbic acid and the primary peak at 0.6 V is evidence of dopamine, while ascorbic acid is identified by the peak at 1.3 V. Thus, at CNYMEs, we can identify the components of the mixture solution, although quantitation might be difficult. To achieve that goal, we would need machine learning ³⁹ or principle component analysis ⁴⁰ for accurate quantification.

Zoom In Zoom Out Reset image size

The codetection of dopamine with DOPAC and ascorbic acid at pH 8.5 indicates the good selectivity of CNTYMEs by showing distinctive CV shapes without losing sensitivity. While animal and brain slice experiments are conducted at physiological pH 7.4, ^41,42 ex vivo studies, such as electrochemical detection of neurotransmitters after HPLC separation are not limited to physiological pH. ⁹ This study shows that manipulating the pH can amplify secondary peaks that are trapped by CNTYMEs, which could be an important tool for tuning selectivity when primary peaks overlap. Thus, the trapping nature of the surface structure at CNTYMEs allows redox reactions to be observed that are not typically detected and these insights are useful for assay development in FSCV.