Transparent Carbon Ultramicroelectrode Arrays - ACS Publications

Article pubs.acs.org/ac

Transparent Carbon Ultramicroelectrode Arrays: Figures of Merit for Quantitative Spectroelectrochemistry for Biogenic Analysis of Reactive Oxygen Species Jonathon Duay,† Janine Elliott,† Jason B. Shear, and Keith J. Stevenson*

Downloaded by UNIV OF CAMBRIDGE on September 15, 2015 | http://pubs.acs.org Publication Date (Web): September 14, 2015 | doi: 10.1021/acs.analchem.5b02804

Department of Chemistry, Center for Nano and Molecular Science and Technology, The University of Texas at Austin, Austin, Texas 78712, United States ABSTRACT: Opaque and transparent carbon ultramicro- to nanoelectrode arrays were made using a previously reported facile versatile fabrication method (Duay et al. Anal. Chem. 2014, 86, 11528). First, opaque carbon ultramicroelectrode arrays (CUAs) were characterized for their analytical response to hydrogen peroxide (H2O2) oxidation using cyclic voltammetry. The alumina blocking layer was found to contribute to the noise and thus had undesirable effects on the array’s limit of detection (LOD) for H2O2 at fast scan rates. Nonetheless, at slower scan rates (ν ≤ 250 mV s−1), the LODs for H2O2 for both opaque (O-CUAs) and transparent arrays (T-CUAs) were found to be lower than previously reported levels for array-based UMEs. LODs as low as 35 nM H2O2 are obtained for T-CUA at a 2.5 mV s−1 scan rate. Furthermore, the transparent arrays were analyzed for their spectroelectrochemical response during the oxidation/reduction of ferrocenemethanol. Results showed very good correlation between the optical and electrochemical response for ferrocenemethanol at a UV wavelength of 254 nm. Thus, these electrodes allow for the in situ mechanistic and kinetic characterization of heterogeneous electrochemical and intermediate homogeneous chemical reactions with high electroanalytical sensitivity, low detection limits, and wide dynamic range.

T

surface of PSS arrays and electrode via atomic layer deposition (ALD). The electrodes are then sonicated to remove the PSSs. Bare carbon areas with ultramicro to nano radii discs remain wherever the PSSs contact the PPF. These exposed regions form the ultramicro to nano CUAs.10 Specifically for CUAs, the PPF used as the electrode material is similar to that of glassy carbon. It is hard and inert and has a high hydrogen evolution overpotential and thus a wide electrochemical window of stability. Additionally, it is highly conductive, chemically inert, and biocompatible and can be easily modified as well as made optically transparent. To date, reported microelectrodes and CUA electrodes have been made from nontransparent materials, preventing their utilization in transmission-based spectroelectrochemical measurements. The conventional optically transparent electrode used for spectroelectrochemistry has been costly indium tin oxide (ITO) films on glass substrates. However, ITO is known to be chemically unstable at cathodic potentials and has limited transparency below 350 nm.11,12 As a result, ITO is insufficient at detecting electronic transitions of redox species in the UV region. Alternatively, PPFs can be made relatively cheap and have a high transparency window over a wide range of wavelengths, including well into the UV region.11,13−15

o probe nano- and picomolar concentrations of analytes, there has been considerable research in the utilization of ultramicroelectrodes for electroanalysis.1−5 These electrodes allow for larger current densities due to enhanced radial diffusion and smaller background currents due to their small electroactive areas, thus resulting in higher signal-to-noise ratios and lower detection limits. Additionally, ultramicroelectrodes have unique physical, chemical, and electronic properties that offer other advantages over bulk electrodes. Owing to their small size, ultramicroelectrodes have decreased ohmic drop, allowing for less supporting electrolyte to be used; rapid acquisition times for steady-state signals; and increased mass transport to the electrode boundary.2,6 However, their small size can produce currents in the picoamp range or lower, which can become noise-limited. To enhance signal-to-noise ratios, arrays of ultramicroelectrodes have been fabricated to function in parallel.2,7−9 In a previous technical note,10 we demonstrated a versatile way to make carbon ultramicro- to nanoelectrode arrays (CUAs) with controllable electrode radii and spacing. In comparison to other fabrication methods, our approach to manufacturing these CUAs is very facile, cheap, and tunable.10 The fabrication consists of dropcasting polystyrene spheres (PSSs) of varying diameters (1.54, 11, and 90 μm) onto a conductive carbon pyrolized photoresist film (PPF) electrode. A hexagonal close-packed two-dimensional ordered network of PSS is formed. A conformal layer of Al2O3 is applied to the © XXXX American Chemical Society

Received: July 24, 2015 Accepted: September 3, 2015

A

DOI: 10.1021/acs.analchem.5b02804 Anal. Chem. XXXX, XXX, XXX−XXX

Article


Analytical Chemistry

with a total exposed geometric area of 0.495 cm2. All voltammograms are displayed with a positive cathodic current and a negative anodic current. For the hydrogen peroxide detection experiments, a saturated calomel electrode (SCE) was used as the reference electrode, and a platinum mesh, as the counter electrode. Experiments were performed in a Faraday cage. Cyclic voltammetry (CV) was performed at scan rates of 2.5, 25, 250, and 2500 mV s−1 between the potentials of 0.45 and 1.15 V vs SCE. The background solution consisted of 5 mL of a 0.1 M KCl and 0.1 M sodium phosphate buffer solution (PBS), pH 7. After a background CV curve was collected, an aliquot (20 μL of a 25 mM H2O2 solution for each 0.1 mM addition) of H2O2 was added and homogeneously mixed before each subsequent CV curve was collected. The CV curves shown are for the fifth CV cycle. All spectroelectrochemical measurements were performed in a 250 μM ferrocenemethanol aqueous solution with 0.1 M PBS to maintain a neutral pH of 7. A platinum (Pt) electrode was used as the reference electrode, and a stainless steel electrode was used as the counter electrode. CVs were performed for 2 cycles at scan rates of 1, 10, 25, and 50 mV s−1 between potentials of −0.1 and 0.3 V vs Pt. Absorbance spectra were acquired with an Agilent Instrument 8453 UV−vis−NIR spectrometer. A homemade Teflon spectroelectrochemical cell was used, and contact to the working electrode was made with copper tape. A CHI 440 potentiostat (CH Instruments Inc.) was used to perform the CVs.

The analytical response for these CUAs was done using a notable biogenic species of reactive oxygen in the form of hydrogen peroxide (H2O2) since much interest has been garnered around the detection and monitoring of this species within biological samples. The attention that this molecule has acquired lies in the fact that it initiates and moderates a variety of key cellular functions, yet it is a ubiquitous cytotoxin.16 H2O2 is considered to modulate signal transduction in vascular remodeling, immune response mechanisms, and apoptosis.17−21 H2O2 can be generated by the spontaneous or enzymatic dismutation of superoxide, O2−, that “leaks” from the mitochondria. It can then undergo a Fenton-like process, creating hydroxyl radicals (·OH) in the presence of iron species and causing damage to the cell and surrounding tissue. It can also be generated as a product of several oxidative enzymatic processes from a variety of oxidase enzymes, including glucose oxidase, alcohol oxidase, lactate oxidase, and various others.22 H2O2 is conventionally detected via fluorescence, chemiluminescence, and spectrophotometry; however, such methods are relatively costly and time-consuming.22 Detection of H2O2 utilizing electrochemical methods is currently on the rise as a superior alternative to the aforementioned detection techniques, as it allows for real time, continuous monitoring of H2O2 with fast response times, easy sensing-platform fabrication, and low cost, as well as ease of modification for greater selectivity. To test the CUAs’ spectroelectrochemical response, we chose the biologically relevant oxidation of ferrocenemethanol (FcMeOH), as it is a relatively hydrophobic redox probe, which allows it to penetrate the lipid bilayer of a cell membrane. Thus, it has been used to determine mass transfer rates across the cell membrane, evaluate the membrane potential, investigate electron transport through the membrane, and visualize receptor proteins on the living cell membrane.23−26 Furthermore, the oxidative product of FcMeOH, ferrocenium methanol (FcMeOH+), is UV-active, which allows the UV transparency of our T-CUAs to be examined.27 Here, we report on the characterization and use of opaque and transparent carbon ultramicroelectrode arrays (T-CUAs). These electrodes are characterized for their analytical figures of merit toward H2O2 oxidation, via cyclic voltammetry, and their spectroelectrochemical response to the oxidation of ferrocenemethanol (FcMeOH).

■

RESULTS Geometry of CUAs. The carbon ultramicroelectrode arrays (CUAs) are here labeled as 1.54CUA, 11CUA, and 90CUA, with the numeric value specifying the micron-sized diameter of the polystyrene spheres (PSS) used during their fabrication, 1.54, 11, and 90 μm, respectively. Figure 1 graphically displays

■

EXPERIMENTAL SECTION Reagents and Materials. All chemicals were used as received. Photoresist AZ 1518 was purchased from Microchemicals. Polystyrene spheres (Polybead) with diameters of 1.54, 11, and 90 μm were purchased from Polysciences, Inc. Ferrocenemethanol, 1-methoxy-2-propanyl acetate, and potassium chloride were acquired from Sigma-Aldrich Co. Monosodium phosphate, disodium phosphate, and hydrogen peroxide (30%) were purchased from Thermo Fisher Scientific Inc. Preparation of Pyrolized Photoresist Film (PPF) Electrodes. Opaque and transparent pyrolized photoresist films (PPFs) were prepared by a previously reported procedure.14,28−31 Fabrication of Carbon Ultramicroelectrode Arrays (CUAs). Preparation of T-CUAs and CUAs follows a procedure described in a previous technical note.10 Electrochemical Measurements. All electrochemical measurements were done using a three-electrode cell. The ultramicroelectrode arrays were used as the working electrode

Figure 1. (a) Schematic of the cross-section of the carbon ultramicroelectrodes, where d is the interelectrode distance, a is the electrode radius, and t is the Al2O3 layer thickness. Top-view scaled representations of (a): (b) macro, (c) 1.54CUA, (d) 11CUA, and (e) 90CUA electrodes with their d/a values indicated.

the cross-sectional as well as the scaled unit-cell geometry of these electrodes, where d is the interelectrode distance, a is the electrode radius, and t is the alumina layer thickness. As t ≪ a, it can be assumed that the electrochemical behavior of these electrodes will resemble that of a coplanar electrode. The scaled representations in Figure 1c−e indicate that the alumina blocking layer dominates the arrays’ geometry, with only a minute area occupied by the exposed carbon electrodes, which is indicated by the black dots (circles represent the residual alumina that had coated the PSS). As the schematic is a scaled geometric representation, it can be seen from Figure 1c− B


Article

Analytical Chemistry Table 1. Geometric Parameters for the Macro, 1.54CUA, 11CUA, and 90CUA Electrodes PPF carbon electrode

diameter of PSS used, d (μm)*

radius of UME, a (μm)

d/a (unitless)

total exposed carbon area (cm2)

percent exposed carbon

macroelectrode 1.5CUA 11CUA 90CUA

N/A 1.54 ± 0.06 11 ± 1 90 ± 2

N/A 0.097 ± 0.009 0.32 ± 0.04 1.3 ± 0.2

N/A 16 ± 2 34 ± 5 70 ± 10

0.495 0.0071 0.0015 0.00038

100% 1.4% 0.31% 0.076%

*


Statistics from manufacturer. Radii were measured by AFM in a previous report.10

e that the exposed carbon area is very small compared to the alumina layer for the CUAs. In fact, it is difficult to even resolve the individual electrodes for 90CUA in this scaled representation. The geometric parameters for each of these arrays are displayed in Table 1. The table indicates that upon increasing the PSS diameter the radius of the electrodes increases; however, the overall exposed carbon surface area decreases. This decrease in exposed carbon area is associated with the larger diameter spheres having a projected area proportional to their radii squared, R2, whereas the individual microelectrode area increases by R4/3 according to Johnson−Kendall−Roberts contact mechanics for the elastic contact of a sphere on a planar substrate.30 This results in an overall theoretical decrease for the total exposed carbon area of R2/3 (proportional increase in projected area divided by proportional increase in individual electrode area (R2/R4/3)) as the PSS radius is increased. Thus, an increase in PSS diameter from 1.54 to 90 μm should result in about a 15 times decrease in exposed carbon area. Experimentally, it is shown to decrease by about 19 times, which is in fair agreement with the theoretical estimate obtained from the geometrical model. The variance presumably comes from the large size distribution of the PSS microspheres and thus large electrode radii standard deviation of the 90 μm PSSs. CUA Geometry Effects for the Analytical Response of H2O2 Oxidation. As previously reported using an outer sphere redox probe,10 a characteristic of these and all microelectrode arrays is that their CV behavior response transitions from a peak-shaped response, where the diffusion layers overlap, at slower scan rates, to a sigmoidal-shaped response, where each electrode behaves as an individual microelectrode, at faster scan rates. Therefore, it was decided to analyze their analytical behavior based on scan rate. The opaque PPFs are utilized here in order to obtain a good understanding of the generalized trends. The hydrogen peroxide oxidation reaction is used without electrode modification to analyze the performance of the different CUAs. The raw cyclic voltammetry data for the different arrays utilizing a 10 nm alumina insulating layer (t = 10 nm) at three different scan rates (ν = 25, 250, and 2500 mV s−1) is represented in Figure 2, where, for each electrode and scan rate, 0.1 mM aliquots of H2O2 were added sequentially. As the PPF carbon electrode has analogous electrochemical behavior to that of glassy carbon electrodes,31 H2O2 oxidation on unmodified PPF is kinetically unfavorable, thus pushing its oxidation peak to extreme potentials. Accordingly, only an increase in the current response with increasing H2 O2 concentration with no redox peak is observed at the potentials used here (0.45 and 1.15 V vs SCE). Nonetheless, the CV current response for each electrode is linear with H2O2 concentration. Qualitatively assessing these CV curves indicates that the analytical signal compared to the peak-to-peak noise level of the individual CV traces or, more conveniently, the signal-to-noise

Figure 2. CV response curves upon addition of hydrogen peroxide (red = 0.0 mM, orange = 0.1 mM, yellow = 0.2 mM, green = 0.4 mM, and blue = 0.6 mM H2O2) for the Macro, 1.54CUA, 11CUA, and 90CUA electrodes utilizing a 10 nm Al2O3 insulating layer at scan rates of 25, 250, and 2500 mV s−1, as indicated.

ratio (SNR) for the arrays decreases with increasing scan rate (left to right) as well as decreasing ultramicroelectrode density from the 1.54CUA to the 90CUA (Table 1). The decrease in SNR with increasing scan rate is usually explained by the signal and background dependence on scan rate. The signal, a faradaic reaction, should increase with the square root of scan rate, whereas the background, a nonfaradaic reaction, should increase proportionally with scan rate. These dependencies indicate that upon increasing scan rate the background increases faster than the signal, which is what is qualitatively observed in Figure 2. However, a slightly more complicated depiction of the scan rate dependence on SNR for the CUAs is described later. The decrease in SNR with decreasing electrode density is not as straightforwardly explained and is investigated here quantitatively. Figure 3 displays the signal and noise for each array, where the signal is the current response for a solution of 0.1 mM H2O2 at the 25 mV s−1 scan rate and a potential of 1.15 V, whereas the noise is the peak-to-peak noise of the background CV trace. Here, it can be seen that the signal decreases about an order of magnitude when going from the 1.54CUA to the 11CUA and additionally from the 11CUA to the 90CUA, which is indicative of the decreasing exposed carbon mentioned above. However, the noise level decreases only marginally (about half from the 1.54CUA to the 90CUA). C


Article


previously and shown schematically in Figure 1. Its effect on the total capacitance is graphically shown in Figure 4, where the


Figure 3. Signal and noise for the 1.54CUA, 11CUA, 90CUA, and macro electrodes. The left y-axis (in mA) is associated with the signal and noise represented by the bars, and the right y-axis is associated with the signal-to-noise ratio (SNR) represented by the blue diamond points.

Figure 4. Partitioning of the capacitance between the exposed carbon area (hashed bars) and the alumina layer (solid bars) for the 1.54CUA, 11CUA, and 90CUA electrodes at different thicknesses of the alumina layer (red = 10 nm, blue = 20 nm, and green = 50 nm Al2O3 layer). *Polystyrene spheres could not be removed due to the relatively thick ALD deposited alumina layer compared to the sphere diameter.

Furthermore, the SNR becomes less as the array density decreases to the point where the SNR for 90CUA is, in fact, lower than that for the macroelectrode. The SNR may be considered an analogous way of analyzing the quantification limits of the electrode. Thus, according to the SNR, the macroelectrode ought to exhibit a lower limit of detection (LOD) than that of 90CUA. Indeed, the calculated 3σ LOD (utilizing standard approach 1 (SA1) as defined by IUPAC and ACS32) at the 25 mV s−1 scan rate for 90CUA and the macroelectrode are 2.5 and 1.4 μM, respectively. This result is counter to the electrochemical theory of microelectrode arrays, as these arrays should exhibit a better SNR than macroelectrodes due to their enhanced radial diffusion and their relatively small areas. This anomaly can be understood by the knowledge that voltammetric noise is proportional to electrode capacitance and not to electrode area.33 This aspect is an important distinction to make, as the capacitance of the CUAs is not proportional to their exposed carbon area. The reason for this is that the alumina layer contributes to the total capacitance and thus the noise (background) as well. The thin layer of alumina on conductive carbon can be thought of as a dielectric layer such as that in an aluminum electrolytic capacitor. Therefore, the theoretical capacitance of the alumina layer can be theoretically calculated. For example, the theoretical specific capacitance (F cm−2), Cspec, for a 10 nm ALD deposited alumina layer can be obtained from the following equation εε Cspec = 0 r (1) d

capacitance value obtained experimentally from the arrays was proportioned between that from the exposed carbon (hashed bars) and from the alumina layer (solid bars). The exposed carbon capacitance values were calculated using the experimental specific capacitance obtained from the bare PPF carbon and the total exposed carbon area from Table 1 for each CUA. While the alumina capacitance values were obtained using the specific capacitance of alumina coated PPFs with varying thicknesses of 10, 20, and 50 nm and the alumina’s areal coverage for each CUA, it is assumed here that the capacitances are additive since they act as capacitors connected in parallel. For the 10 nm alumina layer (red bars), it is shown that the alumina layer contributes to most of the total capacitance, with the exposed carbon contributing relatively little. This result elucidates the basis for the larger SNR for the macroelectrode when compared to that for 90CUA. Hence, for the macroelectrode, the entire surface that contributes to the capacitance and thus the noise also contributes to the analytical signal. Conversely, for 90CUA, only a small portion of the area that contributes to the capacitance contributes to the signal. This consequence can be mitigated two ways. First, the thickness of the alumina layer, t, can be increased as alumina’s capacitance is inversely related to this value, as indicated by eq 1. Thus, arrays with thicker alumina layers (t = 20 and 50 nm) were fabricated, and their capacitance was experimentally determined. The blue and green bars in Figure 4 represent the proportioned capacitance values for the 20 and 50 nm alumina layers, respectively. (Note that the 50 nm alumina layer coats the 1.54 μm PSSs so well that they adhere to the PPF carbon strongly and that the alumina layer is also pinhole free so that neither sonication nor chemical dissolution can remove them.) In the figure, it can be clearly seen that the capacitance decreases for all three CUAs as the alumina layer thickness is increased. Furthermore, alumina’s proportion to the electrode capacitance decreases to about half of its level at 10 nm when its thickness is doubled and to about a fifth of this level when the thickness is increased by 5 times. The second way to mitigate the noise (background) from the alumina layer is to decrease the scan rate. As mentioned previously, the capacitance or nonfaradaic current increases

where ε0 is the permittivity of free space (8.854 × 10−14 F cm−1), εr is the dielectric constant of Al2O3 (7 for ALD grown Al2O334), and d is the thickness of the Al2O3 layer (1.0 × 10−6 cm), resulting in a specific capacitance of 6.2 × 10−7 F cm−2. Accordingly, a 10 nm alumina coated PPF was electrochemically characterized for its specific capacitance by integrating the area of its CV curve in background solution (0.1 M PBS), and it was found to be 5.6 × 10−7 F cm−2, which is within a 10% error of the expected theoretical value indicated above. Although this specific capacitance is small compared to that obtained from a bare PPF carbon electrode (1.1 × 10−5 F cm−2), its contribution to a CUA’s total capacitance is significant due to alumina’s relatively large area when compared to that of the exposed carbon electrode area, as mentioned D


Article


can withstand increased ALD process temperatures for faster deposition rates, as polystyrene’s glass transition temperature is only 90 °C.35 Additionally, we are investigating other materials as insulating layers with lower dielectric constants that are amenable to faster deposition rates. After observing the LOD trends on opaque PPFs (O-CUAs), transparent PPFs (T-CUAs) were made and patterned to form a transparent 1.54CUA (1.54 T-CUA) with the 10 nm alumina insulating layer to test its behavior. These transparent arrays were electrochemically tested for their LOD using the same previous experimental parameters (Figure 5). The experimen-


linearly with scan rate or alternatively decreases linearly with decreasing scan rate, whereas the signal increases to the square root of scan rate or, again, alternatively decreases with the square root of scan rate. For the CUAs, the nonfaradaic current decreases relative to geometric area, whereas the faradaic current decreases only relative to the exposed carbon area. Thus, the SNR increases upon a decrease in the scan rate more for the CUAs than for the macroelectrode. Therefore, these electrodes were characterized for their limits of detection at a slower scan rate of 2.5 mV s−1. At this slow scan rate, the LOD for 90CUA becomes lower than that for the macroelectrode (0.25 and 0.72 μM, respectively), indicating that, indeed, the SNR increases more rapidly as the scan rate is decreased for the CUAs as opposed to the macroelectrode. The values for all electrodes and scan rates as well as alumina layer thicknesses can be found in Table 2. Table 2. Experimentally Determined Limits of Detection (LOD) for the Macro, 1.54CUA, 11CUA, and 90CUA Electrodes at Variable Scan Rates of 2.5, 25, 250, and 2500 mV s−1 and at Different Thicknesses of the Al2O3 layer of 10, 20, and 50 nma scan rate (mV/s) Macro

1.54CUA

11CUA

90CUA

2.5 25 250 2500

LOD (μM) 0.72 ± 0.08 1.4 ± 0.2 1.6 ± 0.3 3.2 ± 0.3 LOD (μM)

Figure 5. Log−log plot of the limit of detection versus scan rate for 1.54O-CUA (red squares) and 1.54T-CUA (black circles) utilizing a 10 nm Al2O3 insulating layer.

scan rate (mV/s)

10 nm Al2O3

20 nm Al2O3

50 nm Al2O3

2.5 25 250 2500 2.5 25 250 2500 2.5 25 250 2500

0.03 ± 0.02 0.15 ± 0.06 1.3 ± 0.6 9±6 0.16 ± 0.06 0.9 ± 0.4 3±1 43 ± 9 0.25 ± 0.08 2.5 ± 0.7 19 ± 6 140 ± 90

0.02 + 0.02 0.10 ± 0.07 1.0 ± 0.9 7±2 0.12 ± 0.07 0.19 ± 0.06 3±2 33 ± 6 0.17 ± 0.09 1±1 5±3 70 ± 30

N/A N/A N/A N/A 0.06 ± 0.04 0.11 ± 0.06 1.6 ± 0.9 14 ± 4 0.09 ± 0.05 0.7 ± 0.4 2±1 38 ± 9

tally determined LOD for the 1.54T-CUA electrode was within error of that for the opaque 1.54CUA at scan rates below 250 mV s−1; however, the 1.54 T-CUA electrode’s LOD deviates from that of the opaque 1.54CUA at the fastest 2500 mV s−1 scan rate. This deviation can be attributed to the increase in sheet resistance of the transparent PPFs compared to that of their opaque counterpart (1850 and 97 Ω □−1, respectively).36 The increase in total resistance can be additionally attributed to the thinness of the T-PPF carbon, since it is only 11 nm thick, whereas its opaque counterpart is 250 nm thick.36 It is considered to be very promising that the detection limit for the 1.54T-CUA electrode is 35 ± 30 nM at a 2.5 mV s−1. When compared to the literature values, this is the best detection limit so far for any electrochemical sensor transparent or opaque.22 However, we are tentative to tout this value because the PPF is nonselective to the analyte of interest, which is a key feature of successful electrochemical sensors. Therefore, future work will be focused on improving the selectivity of this electrode through modification of the PPF with a selective catalyst and/or the addition of a semipermeable membrane that is selective to small molecules such as H2O2. Spectroelectrochemistry of T-CUAs. To evaluate the spectroelectrochemical ability of the T-CUAs, especially in the UV region where common ITO electrodes fail, experiments involving the oxidation of ferrocenemethanol, FcMeOH, were used. FcMeOH is a well-known classic outer sphere redox probe. Its redox reaction is a diffusional, reversible one-electron electrochemical process. Additionally, its ferrocenium methanol oxidative product is stable and UV active and has two absorption peaks observed at 254 and 287 that correspond to ligand−metal charge transfer processes.34 In situ UV−vis spectroelectrochemistry of FcMeOH oxidation at the 1.54T-CUA, 11T-CUA, and T-macro electrodes was produced with the simultaneous corresponding CV

a

Each standard deviation is derived from the LOD measurement of three different electrodes.

Additionally, the electrodes with increased alumina layer thicknesses have improved detection limits. For example, 90CUAs are revealed to have a lower LOD than the macroelectrode when the alumina layer thickness is doubled to 20 nm. This finding is attributed to the aforementioned reduction in the capacitance/noise level of the CV response with the signal being relatively unchanged. Although 11CUA and 90CUA approach similar LODs at slower scan rates and thicker alumina layers, 1.54CUA exhibits the best LOD at all scan rates and alumina thicknesses, which is most likely due to its individual electrode radii being the smallest, thus allowing for enhanced radial diffusion. However, for 1.54CUA, there is not much advantage in increasing the alumina layer thickness, as it minimally improves the LOD (Table 2) with a substantial increase in the ALD deposition time (10 nm = 3 h10). Currently, we are investigating ways to reduce deposition times through the replacement of PSSs with silica microspheres that E


Article



Figure 6. In situ UV−vis spectroelectrochemical plots of FcMeOH oxidation at T-Macro (A), 1.5T-CUA (B), and 11T-CUA (C) electrodes at a potential scan rate 1 mV s−1 in a solution containing 0.1 M PBS. The corresponding cyclic voltammetric responses for each electrode (inset: TMacro, a; 1.5T-CUA, b; 11T-CUA, c) are shown with different scan rates (1, 10, 25, and 50 mV s−1). Right column: The correlating absorbance spectra for the various scan rates at each electrode (T-Macro, D; 1.5T-CUA, E; 11T-CUA, F); absorbance at 0.3 V is plotted.

response for each electrode at different scan rates (1, 10, 25, and 50 mV s−1) between −0.1 and 0.3 V vs Pt, as illustrated in Figure 6. (It should be noted that 90T-CUA was also analyzed for its spectroelectrochemistry, but due to its small electrode area and thus low ferrocenium methanol production, no systematic spectroscopic modulation was observed.) As the linearly varying scan rate is moved from negative to positive potentials, the simultaneous absorbance spectrum of ferrocenium methanol reaches a maximum value. This corresponds to the oxidative apex at the CV curve of 0.3 V vs Pt, which relates to the maximum concentration of ferrocenium methanol produced at the surface of the T-CUAs and the macroelectrode. Additionally, the absorbance behavior of the UV-active species is dependent on the scan rate. Since the absorbance of ferrocenium methanol is produced through a faradaic reaction, the current and/or flux of these ions is known to be related to the square root of the scan rate, ν1/2. Thus, a doubling of the scan rate increases the flux by the square root of 2, but it decreases the scan time by one-half, resulting in less total measured absorbance of the ferrocenium methanol product. Owing to this, the total absorbance change becomes smaller at faster scan rates for each of the T-CUAs and the macroelectrode, as depicted in Figure 6D−F. By taking the derivative of the absorbance with respect to the varying potential (ΔA/ΔE) and plotting it as a function of the potential (E), a derivative cyclic voltabsorptammogram (DCVA) plot is obtained. This differential ΔA/ΔE plot is the optical analogue to that of the CV.37 Figure 7 illustrates the

Figure 7. Cyclic voltabsorptammograms (CVAs), absorbance− potential (black squares) plots, and derivative cyclic voltabsorptammograms (DCVAs), colored squares (ΔA/ΔE), at the 254 nm wavelength showing the oxidation of 250 μM FcMeOH in 0.1 M PBS (scan rate, 1 mV s−1) at the T-Macro (A, green squares), 1.54TCUA (B, blue squares), and 11T-CUA (C, red squares) electrodes.

DCVA plots obtained at a scan rate of 1 mV s−1 for T-Macro, 1.54T-CUA, and 11T-CUA used here. It is demonstrated that the DCVA plots obtained at the 254 nm wavelength distinctly track the CV curves for the oxidation/reduction of FcMeOH. The steepest slopes in the absorbance−potential plots specify the absolute maximum diffusional flux of ferrocenium methanol ions to and from the electrodes’ surface. Thus, the derivative cathodic and anodic peaks of the DCVAs for each electrode align well with the peak currents in the CV, which is logical, as current is another quantitative measure of the diffusional flux. To make the interpretation of the dependence of the absorbance on the scan rate more candid, the dependence of the maximum peak height of the DCVA, (dAO/dE)p, on the scan rate, ν, is used here. This relationship is defined by an optical analogue of the Randles−Sevcik equation, established by Bancroft et al.37 F


Analytical Chemistry (dAR /dE)p = βn1/2 ϵR D1/2C0v−1/2

■

(2)

AUTHOR INFORMATION

Corresponding Author

where β is a constant (−0.08810 mV−1/2), n is the number of electrons, εR is the molar extinction coefficient, D is the diffusion coefficient, and C0 is the concentration. Thus, the peak of the DCVA should scale linearly with the inverse square root of the scan rate. The ΔA/ΔE anodic peak response vs the inverse square root of the scan rate for the T-CUAs and macro electrode is shown in Figure 8. The relationship is, indeed, linear for the macro


Article

*E-mail: [email protected]. Author Contributions †

J.D. and J.E. contributed equally to this work.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS J.D. acknowledges support from the program “Understanding Charge Separation and Transfer at Interfaces in Energy Materials (EFRC:CST)”, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under award no. DESC0001091. J.D. thanks Jacob Goran for helpful discussions. J.B.S. is a fellow at the Institute for Cellular and Molecular Biology. J.E. and K.J.S. acknowledge support from the Welch Foundation, grant nos. F-1331 and F-1529, respectively. J.E. thanks Daniel Redman for helpful discussions.

■

REFERENCES

(1) Ordeig, O.; Banks, C. E.; Del Campo, J.; Munoz, F. X.; Compton, R. G. Electroanalysis 2006, 18, 573−578. (2) Fleischmann, M.; Pons, S. Anal. Chem. 1987, 59, 1391A−1399A. (3) Forster, R. J. Chem. Soc. Rev. 1994, 23, 289−302. (4) Achterberg, E. P. TrAC, Trends Anal. Chem. 1996, 15, 550−558. (5) Dudin, P. V.; Snowden, M. E.; Macpherson, J. V.; Unwin, P. R. ACS Nano 2011, 5, 10017−10025. (6) Trouillon, R.; Passarelli, M. K.; Wang, J.; Kurczy, M. E.; Ewing, A. G. Anal. Chem. 2012, 85, 522−542. (7) Compton, R. G.; Wildgoose, G. G.; Rees, N. V.; Streeter, I.; Baron, R. Chem. Phys. Lett. 2008, 459, 1−17. (8) Chen, R.; Li, Y.; Huo, K.; Chu, P. K. RSC Adv. 2013, 3, 18698− 18702. (9) Kissinger, P. T.; Heineman, W. R. Laboratory Techniques in Electroanalytical Chemistry; Marcel Dekker: New York, 1996; Vol. 15. (10) Duay, J.; Goran, J. M.; Stevenson, K. J. Anal. Chem. 2014, 86, 11528−11532. (11) Lyon, J. L.; Eisele, D. M.; Kirstein, S.; Rabe, J. P.; Vanden Bout, D. A.; Stevenson, K. J. J. Phys. Chem. C 2008, 112, 1260−1268. (12) Dai, Y.; Swain, G. M.; Porter, M. D.; Zak, J. Anal. Chem. 2008, 80, 14−22. (13) Donner, S.; Li, H. W.; Yeung, E. S.; Porter, M. D. Anal. Chem. 2006, 78, 2816−2822. (14) Walker, E. K.; Vanden Bout, D. A.; Stevenson, K. J. J. Phys. Chem. C 2011, 115, 2470−2475. (15) Walker, E. K.; Vanden Bout, D. A.; Stevenson, K. J. Anal. Chem. 2012, 84, 8190−8197. (16) Dröge, W. Physiol. Rev. 2002, 82, 47−95. (17) Baumbach, G. L.; Ghoneim, S. Scanning Microsc. 1993, 7, 137− 142. (18) Geiszt, M.; Leto, T. L. J. Biol. Chem. 2004, 279, 51715−51718. (19) Giorgio, M.; Trinei, M.; Migliaccio, E.; Pelicci, P. G. Nat. Rev. Mol. Cell Biol. 2007, 8, 722−728. (20) Halliwell, B.; Clement, M. V.; Long, L. H. FEBS Lett. 2000, 486, 10−13. (21) Veal, E. A.; Day, A. M.; Morgan, B. A. Mol. Cell 2007, 26, 1−14. (22) Chen, W.; Cai, S.; Ren, Q.-Q.; Wen, W.; Zhao, Y.-D. Analyst 2012, 137, 49−53. (23) Cannes, C.; Kanoufi, F.; Bard, A. J. J. Electroanal. Chem. 2003, 547, 83−91. (24) Sun, P.; Laforge, F. O.; Abeyweera, T. P.; Rotenberg, S. A.; Carpino, J.; Mirkin, M. V. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 443−448. (25) Borland, L. M.; Kottegoda, S.; Phillips, K. S.; Allbritton, N. L. Annu. Rev. Anal. Chem. 2008, 1, 191−227.

Figure 8. Variation of the maximum DCVA anodic peak response vs the inverse square root of the scan rate for 250 μM FcMeOH at the 254 nm wavelength. The R2 values for TPPF, 1.54T-CUA, and 11TCUA are 0.92, 1.00, and 0.89, respectively.

electrode and 1.54T-CUA given that the R2 values are 0.92 and 1.00, respectively. However, 11T-CUA deviates from linearity at scan rates at or above 10 mV s−1. This occurrence is attributed to the CV behavior of 11T-CUA, which becomes more sigmoidal at faster scan rates (insets in Figure 6a−c). Therefore, the current and thus the diffusional flux of ferrocenium methanol approaches steady state at fast scan rates. At scan rates above 970 mV s−1, a purely steady-state DCVA for 11CUA should be observed according to our previous CUA results.10 However, currently, we are limited by the optical sensitivity and temporal resolution of the spectrometer. Future work will report on how the electrochemical and optical responses can be matched with appropriate temporal resolution.

■

CONCLUSIONS Both opaque and transparent carbon ultramicroelectrode arrays (CUAs and T-CUAs) fabricated with polystyrene spheres of varying diameter (1.54, 11, and 90 μm) were characterized here for their analytical performance as well as their voltabsorptive response. The alumina insulating layer between the electrodes was found to act as a capacitive dielectric that adversely affected the analytical performance of the arrays at fast scan rates. However, at slow scan rates, extremely low detection limits in the nanomolar range were found for H2O2 for both the opaque and transparent CUAs. Additionally, the spectroelectrochemical capabilities of these electrodes proved to be extremely adept, specifically in the UV region where traditional transparent electrodes fail. By acquiring derivative cyclic voltabsorptammograms (DCVAs) at the 254 nm wavelength, it is shown that they exhibit good morphological correlation with the CV curves. It is thus evident that the 1.54 and 11 T-CUAs can be used to make spectroelectrochemical measurements and should be applicable to the in situ study of a wide range of complex electrochemical and intermediate chemical processes. G


Article


Analytical Chemistry (26) Takahashi, Y.; Miyamoto, T.; Shiku, H.; Asano, R.; Yasukawa, T.; Kumagai, I.; Matsue, T. Anal. Chem. 2009, 81, 2785−2790. (27) Weber, C. M.; Eisele, D. M.; Rabe, J. P.; Liang, Y.; Feng, X.; Zhi, L.; Müllen, K.; Lyon, J. L.; Williams, R.; VandenBout, D. A.; Stevenson, K. J. Small 2010, 6, 184−189. (28) Donner, S.; Li, H. W.; Yeung, E. S.; Porter, M. D. Anal. Chem. 2006, 78, 2816−2822. (29) Tian, H.; Bergren, A. J.; McCreery, R. L. Appl. Spectrosc. 2007, 61, 1246−1253. (30) Johnson, K. L.; Kendall, K.; Roberts, A. D. Proc. R. Soc. London, Ser. A 1971, 324, 301−313. (31) McCreery, R. L. Chem. Rev. 2008, 108, 2646−2687. (32) Mocak, J.; Bond, A. M.; Mitchell, S.; Scollary, G. Pure Appl. Chem. 1997, 69, 297−328. (33) Long, J. T.; Weber, S. G. Anal. Chem. 1988, 60, 2309−2311. (34) Käar̈ iäinen, T.; Cameron, D.; Käar̈ iäinen, M.-L.; Sherman, A. Atomic Layer Deposition; John Wiley & Sons, Inc.: Hoboken, NJ, 2013. (35) Wünsch, J. R. Polystyrene: Synthesis, Production and Applications; iSmithers Rapra Publishing: London, 2000. (36) Walker, E. K.; Vanden Bout, D. A.; Stevenson, K. J. Langmuir 2012, 28, 1604−1610. (37) Bancroft, E. E.; Sidwell, J. S.; Blount, H. N. Anal. Chem. 1981, 53, 1390−1394.

H


Transparent Carbon Ultramicroelectrode Arrays - ACS Publications

Recommend Documents