1666 Hwi Yong Lee Cedrick Barber Adrienne R. Minerick Department of Chemical Engineering, Michigan Technological University, Houghton, MI, USA
Received December 5, 2014 Revised May 12, 2015 Accepted May 20, 2015
Electrophoresis 2015, 36, 1666–1673
Research Article
Platinum electrode modification: Unique surface carbonization approach to improve performance and sensitivity Many microfluidic devices, also known as lab-on-a-chip devices, employ electrochemical detection methods using microelectrodes. Miniaturizing electrodes inevitably reduces electrode sensitivity and decreases the S/N, which limits applications within microfluidic devices. However, microelectrode surface modification can increase the surface area and sensitivity. In the present work, we report substantial improvement in platinum electrode performance and sensitivity by coating with carbon from red blood cells. The larger goal of this work was to measure DC electrical resistances of red blood cell suspensions in a microchannel for hematocrit determination. It was observed that as current responses of red blood cell suspensions were measured, the platinum electrode performance (reproducibility and S/N) improved with time. The platinum electrode electrocatalytic activity for red blood cell current measurements improved by 140%. Systematic experimentation revealed that red blood cells adsorb and carbonize the platinum electrode surfaces. The electrode surfaces before and after performance improvements were analyzed by field emission scanning electron microscopy, energy dispersive spectrometry, and Raman spectrometry. The formed carbon layers on the electrode surfaces were found to be proteomic and increased surface area with a porous three-dimensional structure, thus improving performance and stabilizing currents. Keywords: Electrode carbonization / Electrode surface area / Microfluidic device / Microelectrode / Signal to noise ratio DOI 10.1002/elps.201500227
1 Introduction Microfluidic devices are considered to be a key, gamechanging platform for many applications such as pointof-care diagnostics, single cell or molecule analysis and manipulation, biosensors, pharmaceutical tests, and many other analysis applications [1–4]. They can achieve laboratory functions on a chip with high-resolution, low cost, design versatility, and portability. Many microfluidic devices employ electrochemical detection using microelectrodes aiming to achieve high sensitivity detection of low concentration analytes. However, miniaturizing electrodes inevitably reduces sensitivity and decreases the S/N [5] because electrical double layer impedance is inversely proportion to electrode surface area (SA). As SA decreases, double layer impedance increases and the signal current and electrode sensitivity decreases [6]. This limits microdevice
Correspondence: Dr. Adrienne R. Minerick, Department of Chemical Engineering, Michigan Technological University, 1400 Townsend Drive, Houghton, MI 49931, USA E-mail: [email protected].
Abbreviations: EDS, energy dispersive spectrometry; FESEM, field emission scanning electron microscopy; RBC, red blood cell; RS, Raman spectrometry; SA, surface area C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
applications, especially applications requiring larger current signals. Thus, a demand exists for microelectrode surface modifications to increase surface area, increase sensitivity, and improve performance. Microfluidic devices developed for biological applications such as biosensors and point-of-care medical diagnostics can experience fouling of electrode surfaces via adsorbed biomolecules. The interface between the electrode surface and biomolecule solution has a higher free energy than the bulk solution. Biomolecules from solution readily adsorb on the electrode surface to lower the interfacial-free energy thus stabilizing the system [7]. Adsorbed biomolecules foul electrode surfaces by covering active surface sites causing significant interferences to electrical responses and decreasing electrode sensitivity and reproducibility [8–11]. The relative length scale of electrode SA to biomolecule size causes more substantial adsorption interferences on microelectrodes. Platinum and gold electrodes are chemically inert, biocompatible, and demonstrate favorable electrocatalytic activity but readily experience biomolecular fouling [12]. Carbon-based electrodes have been extensively investigated and exhibit minimal fouling [13], lower overpotentials, and larger electrochemical potential ranges [14–16].
Colour Online: See the article online to view Figs. 1-4 in colour.
www.electrophoresis-journal.com
Electrophoresis 2015, 36, 1666–1673
For example, graphite-epoxy composite electrodes for dopamine and catechol detection [17] and palladium-coated screen-printed carbon electrodes for hydrazine separation and detection [18] were investigated. Further, electrodes modified with carbon nanotubes exhibit high surface-to-volume nanoscale structures, excellent electrocatalytic activity, high chemical stability, and minimal fouling [19, 20]. Carbon nanotube/copper composite electrodes embedded into microfluidic devices achieved carbohydrate detection [21] and gold deposited carbon nanotube array electrodes were embedded for impedance measurement of prostate cancer cells [22]. Both showed enhanced sensitivity and performance. Carbon nanotube films cast on platinum, gold, and glassy carbon electrodes [20] improved electrode performance and resolution as electrochemical detectors. These carbon modifications maximize the effective electrode/electrolyte interface with 3D porous carbon structures where size, structure, and pore distribution are critical to electrode performance [23–26]. In the present work, we report improvement of platinum electrode performance and sensitivity after what originally presented as red blood cell (RBC) fouling of the electrode, but progressed into beneficial carbonization of the electrode surface under ambient conditions. The larger goal was to measure DC electrical resistances of RBC suspensions in a microchannel for hematocrit determination. As a consequence of this exploration, this paper details irreversible RBC adsorption onto the electrode surfaces, which positively affects the reproducibility of current signals across the microchannel. Initial current responses displayed irreproducible current measurements between runs and erratic changes in current within a single run. However, when current responses were repeatedly measured, platinum electrode performance improved over time. Systematic experimentation revealed that RBCs adsorb and carbonize the electrode surfaces. The carbon layers formed on the electrode surfaces are composed of denatured proteins and hemoglobin, which create porous 3D structures that increase SA to improve performance and stabilize currents. Platinum electrode surfaces before and after performance improvement were analyzed by field emission scanning electron microscopy (FE-SEM), energy dispersive spectrometry (EDS), and Raman spectrometry (RS).
2 Materials and methods 2.1 Microfluidic device Figure 1 presents the microfluidic device used for current response measurements of RBC solutions. Microchannel dimensions were 180 m by 70 m by 10 mm long. Brand new platinum wires (0.10 mm diameter, 99.99%, Sigma-Aldrich) were used as anode and cathode electrodes without any pretreatment. Platinum wires were cut via scalpel and bent by tweezers to immerse 2 mm into each reservoir. General-purpose epoxy (Henkel Corporation, Rocky Hill, CT, USA) attached platinum wire electrodes to a C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Microfluidics and Miniaturization
1667
photolithographically fabricated PDMS microfluidic layer, as previously described [27].
2.2 Sample preparation Blood samples obtained from voluntary donors via venipuncture were stored in vacutainers (Becton, Dickinson, and Company, Franklin Lakes, NJ, USA) containing 1.8 mg K2 EDTA (dipotassium ethylenediaminetetraacetic acid) per mL of blood at 4°C. RBCs were isolated from plasma by centrifugation at 1400 rpm (rcf = 110 g) for 5 min. Isolated RBCs were rinsed twice with 0.9 wt% NaCl solution by spinning at 1400 rpm for 5 min. RBCs were then mixed with 1000 S/cm isotonic PBS (2.57 mM KH2 PO4 , 2.57 mM K2 HPO4 , 2.57 mM NaCl, 274.3 mM dextrose) to form a 60 vol% RBC suspension. To test pH dependence of the carbonized electrodes, 2000 S/cm PBS at pH 3 (1.79 mM KH2 PO4 , 1.79 mM K2 HPO4 , 1.79 mM NaCl, 0.12 mM dextrose, 4.80 mM HCl), pH 7 (5.06 mM KH2 PO4 , 5.06 mM K2 HPO4 , 5.06 mM NaCl, 0.68 mM dextrose), and pH 11 (1.94 mM KH2 PO4 , 1.94 mM K2 HPO4 , 1.94 mM NaCl, 0.23 mM dextrose, 7.32 mM NaOH) were prepared. Lastly, 320 M Triton X-100 (T8532, Sigma-Aldrich, Saint Louis, MO, USA) was added to the 60 vol% RBC suspension and the 1000 S/cm PBS solution to stabilize current responses. Triton X-100 attenuates electrolysis bubble effects that destabilize current responses by facilitating formation of smaller bubbles. It was shown that 320 M Triton X-100 does not compromise RBC membrane integrity [27, 28]. 2.3 Current measurements Separate current responses of 1000 S/cm isotonic PBS and 60 vol% RBC suspension were measured in alternating cycles. RBCs served as the only carbon source in the system. Anode and cathode platinum electrodes were connected to an HVS 448 high voltage sequencer (LabSmith, Livermore, CA, USA). PBS or RBC suspensions were injected via BD Safety-LokTM 3 mL syringes (Becton, Dickinson and Company) through a bonded anode well port connector (LabSmith) and pushed through the microchannel to the cathode well. A 100 V DC signal was applied for 60 s while concurrently saving current responses at 39.06 Hz using Sequence Software (LabSmith). After each measurement, the entire microfluidic device and microchannel were flushed with e-pure water (Millipore Simplicity 185, Millipore Corporation, Billerica, MA, USA, Resistivity: 18.2 MΩ). 2.4 Sample characterization Platinum electrodes before and after current measurements were imaged using a Hitachi S-4700 FE-SEM operated with a 5.0 keV electron beam. Platinum wire electrodes were cut off the microfluidic device using a scalpel and placed, uncoated, www.electrophoresis-journal.com
1668
H. Y. Lee et al.
Electrophoresis 2015, 36, 1666–1673
Figure 1. Schematic drawing of the single channel microfluidic device. (A) Device configuration illustrating anode and cathode platinum wires (0.10 mm in diameter) submerged in fluidic wells connected by a microchannel 180 m by 70 m by 10 mm long as shown in (B). Not drawn to scale.
Figure 2. (A) Current responses of 1000 S/cm PBS (blue traces) and 60 vol% RBC suspension (red traces) at 100 V DC versus time. Each bar represents one 60 s measurement, begun with new solution. Note the broken time axis to enlarge initial and final current responses. (B) Average current responses in the boxes labeled as before and after in (A). Error bars represent 95% confidence intervals.
on carbon-tape stubs using tweezers. Electrode surface composition was determined via EDS with a 15.0 keV electron beam. Raman spectrometric spectrums were collected and analyzed from a carbonized anode excited with 632 nm laser using a Horibia Jobin Yvon LabRAM HR 800 RS.
3 Results and discussion 3.1 Current responses Figure 2A shows current responses sampled at 39.06 Hz of 1000 S/cm PBS (blue traces) and 60 vol% RBC suspensions in PBS (red traces) at 100 V. The plot time-axis is discontinuous to emphasize initial and final responses. Average current responses of the first and last eight measurements of each solution are highlighted by green boxes in Fig. 2A and compared in Fig. 2B. RBC solution current responses gradually increased with each cycle while the PBS baseline remained C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
fairly constant. PBS fluctuations, such as from 1440 to 1500 s and from 1560 to 1620 s, were attributed to residual blood cells after manual device flushing with e-pure water. Fewer large fluctuations occurred as the trials progressed indicating that the platinum electrodes incrementally stabilized. These observations were repeated with ⬎10 electrode pairs, although averages reported are for a representative experiment cycle. Current response fluctuations of both PBS and RBC solutions were attributed to isolated and thus unstable RBC adsorption. It was previously documented that unwanted molecule adsorption fouls biosensor electrode surfaces resulting in irreproducible electrochemical responses [8–11]. We thus hypothesized that RBC adsorption on platinum electrode surfaces was irregular, yet incrementally accumulating during measurements. After ten cycles of RBC current measurements and ten cycles of PBS, the platinum wire electrodes lost their luster; after 20 cycles of RBC current measurements and 20 cycles of PBS, the anode appeared black to the naked eye. RBC current responses decreased during a cycle, such as www.electrophoresis-journal.com
Microfluidics and Miniaturization
Electrophoresis 2015, 36, 1666–1673 Table 1. S/Ns of current responses before and after in Fig. 2A
PBS 60 vol% RBC
Before
After
55.5 51.8
65.5 170
from 900 to 960 s and from 1740 to 1800 s. This phenomenon, which decreased as the electrode aged, was attributed to either carbonized RBC clusters fragmenting off the electrode, which was experimentally observed, or to collapse of densely carbonized RBCs onto the electrodes. In Fig. 2B, 60 vol% RBC current responses before and after were 25.3 ± 2.13 A and 60.2 ± 0.67 A, respectively. This indicated that, as current signal responses were measured, the platinum electrode electrocatalytic activity (term used here to reflect the relative amount of active surface sites participating in the electron transfer reaction) for the RBC current measurements was increased by 140%. Over time, current responses became more stable and reproducible as indicated by reduced error bars in Fig. 2B. Their values before and after in Fig. 2B were 10.5 ± 1.13 A and 12.6 ± 0.20 A. It is presumed that the RBC layers modified the platinum electrode surface properties to improve performance similar to how other carbon-coated electrodes improve performance [20, 29]. S/Ns before and after were compared in Table 1. S/N ratios of both PBS and RBC suspension current responses increased with measurement cycles. The average PBS current response signals were nearly constant, so the increased S/N was attributed to reduced noise. The decrease in RBC suspension S/N was predominantly due to increased current responses and decreased SD, which before was 0.922 and after was 0.483. The entire sequence in Fig. 2A was reproduced more than ten times (N ⬎ 10) beginning each time with new platinum electrodes. The result—improved RBC current responses stabilizing between 60 and 63 A—was observed each time. The most common sequence length was 30–50 cycles of the 60 vol% RBC solution current measurements (plus the repeated PBS current measurements). Once current responses were stabilized at 60–63 A, the entire immersed anode electrode was completely black and fluctuations of current responses were minimal. The black color was attributed to carbon layers covering the electrode surface, consistent with prior literature reporting minimal biomolecule adsorption onto carbon-based electrodes [21, 30, 31]. The black carbon layers were very brittle and could be easily removed from the electrode surface mechanically by tweezers or high-pressure air. If the carbon layers were kept intact, the electrodes retained improved performance. However, if the carbon layers were damaged, the current responses decreased below 60 A. Then, by repeating the same PBS and 60% RBC solution measurement cycles, current responses could be restored to 60–63 A again. These carbonized platinum electrodes demonstrated sufficient activity and stability to reproducibly measure RBC hematocrit to 2 vol% resolution [28]. C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1669
3.2 Electrode surface characterization To directly investigate the black layers on the electrode surfaces, unused platinum electrode wire surfaces and the carbonized platinum electrode surfaces showing the improved performance were imaged by FE-SEM. Black layer compositions were analyzed via EDS and the results shown in Table 2. Due to platinum wire electrode curvature, EDS analysis locations along the same ordinate were chosen for quantitative comparison. Figure 3A, B, and C illustrate unused 0.10 mm, 99.99% pure platinum wire electrode surfaces at three magnifications. EDS elemental analysis was performed on both the dominant bright regions and the sporadic dark regions marked by A and B, respectively, in Fig. 3C and correlated to values in Table 2. The dark regions were carbon rich, while the bright regions were nearly pure platinum. Overall, the unused platinum surface contained measurable amounts of carbon and oxygen. Figure 3D, E, and F illustrate SEM images of the cathode surface after 31 cycles with 60 vol% RBC and 31 intermittent PBS current measurements. This electrode demonstrated stable 60 A signals in 60 vol% RBC solutions. Darker regions were more prominent compared to unused platinum electrodes as shown in Fig. 3D. EDS analysis revealed these dark regions were carbon rich as cataloged in Table 2. In Fig. 3E, the left most region labeled C had similar composition to the unused platinum electrode surface in Table 2. However, the dark region D contained 31.9% carbon. Figure 3F is a 10 000× magnification of region D. Another cathode was imaged after 49 current measurement cycles with PBS and 49 cycles with 60 vol% RBC as shown in Fig. 3G, H, and I. Figure 3G shows a scalpel-induced curved cut (labeled E) and F marks the carbon tape used to adhere the electrode to the FE-SEM stub. In contrast with Fig. 3D, E, and F, the darker regions more extensively cover the electrode surface (region G) in Fig. 3H. EDS of G indicated more extensive carbonization, as catalogued in Table 2. However, the bright region marked as H in Fig. 3H retained composition similar to unused platinum. These results indicate that with increasing measurement cycles, carbon coverage of the cathode electrode surface increased. The source of the carbonization was attributed primarily to irreversible RBC adsorption onto the platinum surface because this is the predominant source of carbon within the solution. The solution also contains 320 M Triton X-100, which is a nonionic surfactant comprised of a hydrophilic polyethylene oxide chain and an aromatic hydrocarbon hydrophobic group ((C2 H4 O)n C14 H22 O, N = 9 – 10). It was observed that repeated current measurements in only PBS solution containing Triton X-100 did not change electrode surfaces. Other experimental impurities that could serve as sources of carbon include the PDMS channel itself and the bonded port connector composed of Ultem (polyetherimide) although these solids have negligible solubility in water [32]. The corresponding treated anode electrode surfaces were also imaged. Figure 3J, K, and L shows the anode surface R
www.electrophoresis-journal.com
1670
H. Y. Lee et al.
Electrophoresis 2015, 36, 1666–1673
Table 2. Composition of platinum electrode surfaces in Figs. 3 and 4, and 5 analyzed by EDS
Electrode
New Cathode
Anode
Labela)
A B C D H G L K M N
Surface status
Bright Dark Bare Carbonized Bare Carbonized Bare Carbonized Bare Carbonized
Element (weight%) C
O
Pt
N
14.8 ± 1.37 20.0 ± 5.35 14.5 ± 1.05 31.9 ± 14.9 13.3 ± 0.26 38.6 ± 5.77 15.9 ± 1.32 56.8 ± 2.59 12.4 55.9 ± 3.34
0.82 ± 0.29 1.39 ± 0.67 1.09 ± 0.43 2.36 ± 1.24 0.88 ± 0.04 3.18 ± 0.55 1.37 ± 0.01 5.94 ± 0.83 1.83 9.67 ± 0.54
84.4 ± 1.62 78.6 ± 6.01 84.4 ± 1.44 65.7 ± 16.1 85.8 ± 0.23 58.2 ± 6.26 84.3 ± 0.17 36.1 ± 3.23 85.8 28.9 ± 3.28
4 3 3 4 3 3 2 2 1 3
a) The capital letters in the label column, corresponding to the labels in Figs. 3 and 4, and 5, indicate representative regions for the analyses.
Figure 3. FE-SEM images of platinum wire electrodes. First column is 500X magnification, second column is 1000X, and third column is 10 000X. Unused platinum wire electrodes in (A), (B), and (C). Cathode in (D), (E), and (F) after 31 cycles of the 60 vol% RBC suspension current measurement (62 cycles if including the PBS current measurements). Cathode in (G), (H), and (I): after 49 cycles of 60 vol% RBC. Anode in (J), (K), and (L) after 31 cycles of the 60 vol% RBC and (M), (N), and (O) after 49 cycles of 60 vol% RBC suspension. Yellow boxes indicate representative regions for EDS analysis reported in Table 2.
after 31 current measurements cycles with PBS and 31 cycles with 60 vol% RBC. This anode was paired with Fig. 3D, E, and F and demonstrated a stable 60 A in 60 vol% RBC solution. Figure 3K, a magnified view of the I region in Fig. 3J, is a side view of the coated layers, which illustrated a rough 3D anode surface, which increases SA. A top view of the C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
layers, an enlarged image of region J, is shown in Fig. 3L. EDS analysis revealed that the K region was a carbon rich layer containing 56.8 ± 2.59 wt% carbon as reported in Table 2. In K, 1.16 ± 0.19 wt% chlorine was also detected. Smoother regions, like L, had similar composition to untreated platinum.
www.electrophoresis-journal.com
Electrophoresis 2015, 36, 1666–1673
Microfluidics and Miniaturization
1671
Figure 4. (A) Raman spectrum of an anode surface after 60 cycles of 60 vol% RBC current measurements (120 cycles if including PBS cycles). (B) Current responses of 2000 S/cm PBS at pH 3, pH 7, and pH 11 at 100 V for 1 min. Error bars represent 95% confidence intervals (N = 4 7).
Figure 3J, K, and L indicate that the coated carbon layers developed into 3D porous structures, and differed structurally from the cathodes. RBCs have negative zeta potential in PBS, which would electrophoretically drive RBCs to the positively charged anode surface to deposit into 3D structures. RBC zeta potential is –31.8 ± 1.1 mV [33] in PBS with 1.7 mM KH2 PO4 , 5.2 mM Na2 HPO4 , 150 mM NaCl, and pH 7.4, which is similar to our PBS medium condition of 2.57 mM KH2 PO4 , 2.57 mM K2 HPO4 , 2.57 mM NaCl, 274.3 mM dextrose, and pH 7.2. RBC adsorption is energetically favorable because adsorption reduces the interfacial energy at the electrodesolution interface. In contrast, the negatively charged cathode would repel intact RBCs, so cathodic 3D adsorption was not observed as shown in Fig. 4. The anode electrode was examined after 49 cycles of RBC current measurements and 49 cycles of PBS and shows much more extensive carbon layers than the 31 cycles in Fig. 3J, K, and L. The FE-SEM images shown in Fig. 3M, N, and O represent anode status right after current measurements in Fig. 2, which displayed 60.2 ± 0.67 A with the 60 vol% RBC suspension at 100 V DC. Handling of the completely coated anode electrodes was challenging because the black carbon layers were brittle and easily damaged. Thus, it is inferred that the carbon layers were deposited onto the platinum electrode surface via electrostatic interactions and at most, the bonding between the carbon layers and the platinum surface must be weak noncovalent bonding. Figure 3M illustrates a region, labeled M, where the 3D carbon layer peeled off; this region had the same composition as unused platinum as reported in Table 2. The carbon layer, marked by N in Fig. 3N was 4.46 ± 2.14 wt% of chlorine and 1.08 ± 0.58 wt% of potassium were also detected and would have originated from the PBS solution. This PBS microfluidic system previously demonstrated electrolysis bubble generation [27]; thus, electrode electrolysis reactions occur concurrent with carbon adsorption. During current measurements, electrolysis of water generates oxygen
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
and hydrogen gas bubbles at the anode and cathode electrode surfaces, respectively [34]. Electrolysis bubble behavior could impede RBC and carbon adsorption, or, as shown in this work, could lead to a porous carbonaceous deposition. Adsorption interference was reported in an electrophoretic colloidal coating process whereby charged particles in suspension were deposited onto polarized electrode surfaces with opposite charges under an externally applied electric field. In this colloidal electrophoretic deposition, electrolysis bubbles incorporated into the deposit layers and led to porous and inhomogeneous deposited structures [30, 31, 35]. In our system, the size and number of electrolysis bubbles depends upon the applied voltage, medium conductivity (sample conductivity), and Triton X-100 concentration [27]. This work demonstrates an effective composition that favors deposition of RBCs into sufficiently thick and porous carbon layers that stabilize the current. Deposition rate would likely change with solution composition and loading percentage of RBCs (60 vol%); since RBCs are not an ideal carbon source, these further experiments were not conducted. It should be considered that the evolution, growth, and detachment rate of electrolysis bubbles is not spatially uniform even at the same experimental conditions due to local variations in platinum morphology [36]. This could be the reason for slight variations in the number of cycles necessary to achieve optimized electrode performance. In summary, the applied 100 V DC was high enough to generate electrolysis bubbles, which is a viable explanation for the resulting porous and inhomogeneous carbon structures illustrated in Fig. 4A, B, and C. Limitations of EDS analysis include that it is not reliable for minor element analysis (less than 1 wt%) and it cannot detect hydrogen. Thus, the coated layer composition was analyzed via RS. RS has been widely used to investigate the composition of structure of biological molecules and systems, such as proteins and blood [37–42]. Protein analysis via RS has focused on 3D protein structures and amino acid side chains [40–43].
www.electrophoresis-journal.com
1672
H. Y. Lee et al.
Figure 4A shows the Raman spectrum of the carbonized anode surface. RBCs are comprised of membrane proteins, phospholipids, and hemoglobin. Typical protein RS display a prominent amide I band near 1650 cm−1 and an amide III band near 1300 cm−1 reflecting polypeptide backbone conformations in proteins [40, 42, 43]. In Fig. 4A, the amide bands, especially the prominent amide I, did not appear indicating structural changes and denaturation of the membrane proteins and hemoglobin [44]. However, residues of hemoglobin are confirmed through the bands near 565 cm−1 by Fe-O2 bonding and 1360 and 1543 cm−1 due to the Soret band excitation of heme proteins [38, 39]. Amino acid side chains are also identified by two bands: tryptophan at 850 cm−1 and phenylalanine at 1000 and 1170 cm−1 [37, 41, 42]. These indicate that the electrode carbon layers are composed of denatured proteins and hemoglobin from lysed RBCs. The denatured proteins would act similar to structureless polymer chains (random coils) [45] on the electrode surfaces. Proteins are conductive, but their electrical properties depend upon pH as a consequence of conformational changes [45–47]. Thus, current responses at different pH values (2000 S/cm PBS solutions of pH 3, 7, and 11) were measured for 1 min at 100 V. Figure 4B shows that the current responses of the PBS of pH 11 (20.6 ± 0.40 A) were lower than pH 3 (23.7 ± 1.37 A) and pH 7 (23.8 ± 0.88 A). This evidence also supports the conclusion that the coated carbon layers were comprised of protein fragments. Based on the data presented herein, we have attributed improved electrode performance to the formation of carbon layers on the anode surface and carbon films on the cathode surfaces. During current measurements, adsorbed RBCs lysed onto the electrode surfaces, proteins denature, and progressively carbonized to form the observed black carbon layers. As the RBC current measurement cycles progressed, RBC adsorption compounded and eventually formed thick black carbon layers, which were especially pronounced on the anode. The anode carbon layers developed into 3D porous structures, likely facilitated by electrolysis, which increased the electrode surface area. These carbon layers were composed of the conductive denatured membrane proteins and hemoglobin from lysed RBCs. Consistent with findings in other carbon electrode systems, the critical factor with carbon-based modification of electrode surfaces is to achieve maximum effective surface area with well-optimized porous structures. The stabilization of current and increased electrocatalytic activity of the platinum electrodes in our microfluidic device suggests that carbon layers formed via RBC adsorption developed into optimized porous structures for stable RBC current measurements.
4 Concluding remarks Many microfluidic devices employ microelectrodes with small surface areas yielding weaker electrical signals. Prior biosensor research has revealed that irreversible biomolecule
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Electrophoresis 2015, 36, 1666–1673
adsorption destabilizes these electrical signals [8–10]. Successful strategies to reduce biomolecule adsorption have included bonding carbon nanotubes or graphene on electrode surfaces. The present work is an inadvertent extension of this approach with red blood cells as the carbon source. Systematic studies in RBC solutions suspensions revealed that RBCs offer a carbon source for carbon deposition on both the cathode and anode, although the structure and composition differed greatly between the electrodes. The platinum electrodes with 3D carbon layers formed from RBC adsorption reached a steady state as indicated by increased current responses with smaller S/N. These electrode carbon-coating results using RBCs are consistent with other electrode carbonization literature, which reports that (1) increasing electrode SA improves the electrode performance, and (2) carbon-based or carbon-coated electrodes experience less fouling. This suggests that biological sources of carbon may be utilizable to carbon-coated electrodes, and once coated reduce further biomolecule adsorption. The broader implications are that the ambient condition approach described here may serve as a lower cost alternative to carbon nanotubes or graphene binding to electrode surfaces. The authors gratefully acknowledge ACMAL (MTU) and Dr. Yoke Khin Yap for surface characterization. Additional gratitude is extended to Joe Halt, Zhichao Wang, and Maryam Khaksari for RBC sample preparation and valuable discussions. The authors also gratefully acknowledge prior discussions and efforts years ago with Soumya Srivastava and David O. Wipf. The authors also gratefully acknowledge recent funding support from NSF PFI:AIR IIP 1414331 and an NSF STTR IIP 1417187 subcontract. The authors declare no conflict of interest.
5 References [1] Yeo, L. Y., Chang, H.-C., Chan, P. P. Y., Friend, J. R., Small 2011, 7, 12–48. [2] Mark, D., Haeberle, S., Roth, G., vonStetten, F., Zengerle, R., Chem. Soc. Rev. 2010, 39, 1153–1182. [3] Whitesides, G. M., Nature 2006, 442, 368–373. [4] Minerick, A. R., AIChE J. 2008, 54, 2230–2237. [5] Naka, K., Hayashi, H., Senda, M., Shiraishi, H., Konishi, S., Micro Electro Mechanical Systems, 2007. MEMS. IEEE 20th International Conference on. 2007, pp. 195–198. [6] Zheng, S., Liu, M., Tai, Y.-C., Biomed. Microdevices 2008, 10, 221–231. [7] Nakanishi, K., Sakiyama, T., Imamura, K., J. Biosci. Bioeng. 2001, 91, 233–244. [8] Zourob, M., Recognition Receptors in Biosensors, Springer, New York 2010. [9] Atta, N. F., Galal, A., Ahmed, R. A., Bioelectrochemistry 2011, 80, 132–141. [10] Marino, A., Brajtertoth, A., Anal. Chem. 1993, 65, 370–374.
www.electrophoresis-journal.com
Electrophoresis 2015, 36, 1666–1673
[11] Nassar, A. E. F., Willis, W. S., Rusling, J. F., Anal. Chem. 1995, 67, 2386–2392.
Microfluidics and Miniaturization
1673
[30] Besra, L., Uchikoshi, T., Suzuki, T. S., Sakka, Y., J. Eur. Ceramic Soc. 2010, 30, 1187–1193.
[12] Baldwin, R. P., J. Pharm. Biomed. Anal. 1999, 19, 69–81.
[31] Ammam, M., RSC Adv. 2012, 2, 7633–7646.
[13] Gawron, A. J., Martin, R. S., Lunte, S. M., Electrophoresis 2001, 22, 242–248.
[32] Varaprath, S., Frye, C. L., Hamelink, J., Environ. Toxicol. Chem. 1996, 15, 1263–1265.
[14] Vandaveer, W. R., Pasas-Farmer, S. A., Fischer, D. J., Frankenfeld, C. N., Lunte, S. M., Electrophoresis 2004, 25, 3528–3549.
[33] Bondar, O. V., Saifullina, D. V., Shakhmaeva, II, Mavlyutova, II, Abdullin, T. I., Acta Nat. 2012, 4, 78–81.
[15] Lacher, N. A., Garrison, K. E., Martin, R. S., Lunte, S. M., Electrophoresis 2001, 22, 2526–2536. [16] McCreery, R. L., Cline, K. K., in: Kissinger, P., Heineman, W. R. (Eds.), Laboratory Techniques in Electroanalytical Chemistry, Marcel Dekker, New York 1996, pp. 293–332.
[34] Minerick, A. R., Ostafin, A. E., Chang, H. C., Electrophoresis 2002, 23, 2165–2173. [35] Chavez-Valdez, A., Shaffer, M. S. P., Boccaccini, A. R., J. Phys. Chem. B 2012, 117, 1502–1515. [36] Vogt, H., Balzer, R. J., Electrochim. Acta 2005, 50, 2073–2079.
[17] Pumera, M., Merkoc¸i, A., Alegret, S., Electroanalysis 2006, 18, 207–210.
´ ´ [37] Carmona, P., Monleon, E., Monzon, M., Badiola, J. J., Monreal, J., Chem. Biol. 2004, 11, 759–764.
[18] Wang, J., Chatrathi, M. P., Tian, B., Polsky, R., Electroanalysis 2000, 12, 691–694.
[38] Torres Filho, I. P., Terner, J., Pittman, R. N., Proffitt, E., Ward, K. R., J. Appl.Physiol. 2008, 104, 1809–1817.
[19] Choong, C.-L., Milne, W., Teo, K. K., Int. J. Mater. Form. 2008, 1, 117–125.
[39] Lin, J., Lin, J., Huang, Z., Lu, P., Wang, J., Wang, X., Chen, R., J. Innov. Opt. Health Sci. 2014, 07, 1350051.
[20] Pumera, M., Merkoc¸i, A., Alegret, S., Electrophoresis 2007, 28, 1274–1280.
[40] Maiti, N. C., Apetri, M. M., Zagorski, M. G., Carey, P. R., Anderson, V. E., J. Am. Chem. Soc. 2004, 126, 2399–2408.
[21] Wang, J., Chen, G., Wang, M., Chatrathi, M. P., Analyst 2004, 129, 512–515.
[41] Overman, S. A., Thomas, G. J., Biochemistry 1999, 38, 4018–4027.
[22] YeoHeung, Y., Zhongyun, D., Vesselin, N. S., Mark, J. S., Nanotechnology 2007, 18, 465505.
[42] Tuma, R., Journal of Raman Spectrosc. 2005, 36, 307–319.
[23] Zhang, D. W., Li, X. D., Li, H. B., Chen, S., Sun, Z., Yin, X. J., Huang, S. M., Carbon 2011, 49, 5382–5388.
[43] Williams, R. W., in: Hirs, C. H. W., Timasheff, S. N. (Eds.), Methods in Enzymology, Academic Press, New York 1986, pp. 311–331.
[24] Zhang, L. L., Zhao, X. S., Chem. Soc. Rev. 2009, 38, 2520–2531. [25] Long, J. W., Dunn, B., Rolison, D. R., White, H. S., Chem. Rev. 2004, 104, 4463–4492. [26] Chen, Y., Zhang, X., Yu, P., Ma, Y., J. Power Sources 2010, 195, 3031–3035.
[44] Borgognoni, C. F., Tattini Junior, V., Ayrosa, A. M. I. B., Polakiewicz, B., Leirner, A. A., Maizato, M. J. S., Higa, O. Z., Beppu, M. M., Pitombo, R. N. d. M., Brazilian Arch. Biol. Technol. 2009, 52, 1493–1504.
[27] Lee, H. Y., Barber, C., Minerick, A. R., Electrophoresis 2014, 35, 1782–1789.
[45] Tanford, C., in: Anfinsen, C. B. (Ed.), Advances in Protein Chemistry, Academic Press, New York 1968, pp. 121–282.
[28] Lee, H. Y., Barber, C., Rogers, J. A., Minerick, A. R., Electrophoresis 2015, 36, 978–985.
[46] Bazinet, L., Ippersiel, D., Mahdavi, B., J. Agricult. Food Chem. 2004, 52, 1980–1984.
[29] Wang, Y., Li, Q., Hu, S., Bioelectrochemistry 2005, 65, 135–142.
[47] Ron, I., Pecht, I., Sheves, M., Cahen, D., Accounts Chem. Res. 2010, 43, 945–953.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.electrophoresis-journal.com
Received December 5, 2014 Revised May 12, 2015 Accepted May 20, 2015
Electrophoresis 2015, 36, 1666–1673
Research Article
Platinum electrode modification: Unique surface carbonization approach to improve performance and sensitivity Many microfluidic devices, also known as lab-on-a-chip devices, employ electrochemical detection methods using microelectrodes. Miniaturizing electrodes inevitably reduces electrode sensitivity and decreases the S/N, which limits applications within microfluidic devices. However, microelectrode surface modification can increase the surface area and sensitivity. In the present work, we report substantial improvement in platinum electrode performance and sensitivity by coating with carbon from red blood cells. The larger goal of this work was to measure DC electrical resistances of red blood cell suspensions in a microchannel for hematocrit determination. It was observed that as current responses of red blood cell suspensions were measured, the platinum electrode performance (reproducibility and S/N) improved with time. The platinum electrode electrocatalytic activity for red blood cell current measurements improved by 140%. Systematic experimentation revealed that red blood cells adsorb and carbonize the platinum electrode surfaces. The electrode surfaces before and after performance improvements were analyzed by field emission scanning electron microscopy, energy dispersive spectrometry, and Raman spectrometry. The formed carbon layers on the electrode surfaces were found to be proteomic and increased surface area with a porous three-dimensional structure, thus improving performance and stabilizing currents. Keywords: Electrode carbonization / Electrode surface area / Microfluidic device / Microelectrode / Signal to noise ratio DOI 10.1002/elps.201500227
1 Introduction Microfluidic devices are considered to be a key, gamechanging platform for many applications such as pointof-care diagnostics, single cell or molecule analysis and manipulation, biosensors, pharmaceutical tests, and many other analysis applications [1–4]. They can achieve laboratory functions on a chip with high-resolution, low cost, design versatility, and portability. Many microfluidic devices employ electrochemical detection using microelectrodes aiming to achieve high sensitivity detection of low concentration analytes. However, miniaturizing electrodes inevitably reduces sensitivity and decreases the S/N [5] because electrical double layer impedance is inversely proportion to electrode surface area (SA). As SA decreases, double layer impedance increases and the signal current and electrode sensitivity decreases [6]. This limits microdevice
Correspondence: Dr. Adrienne R. Minerick, Department of Chemical Engineering, Michigan Technological University, 1400 Townsend Drive, Houghton, MI 49931, USA E-mail: [email protected].
Abbreviations: EDS, energy dispersive spectrometry; FESEM, field emission scanning electron microscopy; RBC, red blood cell; RS, Raman spectrometry; SA, surface area C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
applications, especially applications requiring larger current signals. Thus, a demand exists for microelectrode surface modifications to increase surface area, increase sensitivity, and improve performance. Microfluidic devices developed for biological applications such as biosensors and point-of-care medical diagnostics can experience fouling of electrode surfaces via adsorbed biomolecules. The interface between the electrode surface and biomolecule solution has a higher free energy than the bulk solution. Biomolecules from solution readily adsorb on the electrode surface to lower the interfacial-free energy thus stabilizing the system [7]. Adsorbed biomolecules foul electrode surfaces by covering active surface sites causing significant interferences to electrical responses and decreasing electrode sensitivity and reproducibility [8–11]. The relative length scale of electrode SA to biomolecule size causes more substantial adsorption interferences on microelectrodes. Platinum and gold electrodes are chemically inert, biocompatible, and demonstrate favorable electrocatalytic activity but readily experience biomolecular fouling [12]. Carbon-based electrodes have been extensively investigated and exhibit minimal fouling [13], lower overpotentials, and larger electrochemical potential ranges [14–16].
Colour Online: See the article online to view Figs. 1-4 in colour.
www.electrophoresis-journal.com
Electrophoresis 2015, 36, 1666–1673
For example, graphite-epoxy composite electrodes for dopamine and catechol detection [17] and palladium-coated screen-printed carbon electrodes for hydrazine separation and detection [18] were investigated. Further, electrodes modified with carbon nanotubes exhibit high surface-to-volume nanoscale structures, excellent electrocatalytic activity, high chemical stability, and minimal fouling [19, 20]. Carbon nanotube/copper composite electrodes embedded into microfluidic devices achieved carbohydrate detection [21] and gold deposited carbon nanotube array electrodes were embedded for impedance measurement of prostate cancer cells [22]. Both showed enhanced sensitivity and performance. Carbon nanotube films cast on platinum, gold, and glassy carbon electrodes [20] improved electrode performance and resolution as electrochemical detectors. These carbon modifications maximize the effective electrode/electrolyte interface with 3D porous carbon structures where size, structure, and pore distribution are critical to electrode performance [23–26]. In the present work, we report improvement of platinum electrode performance and sensitivity after what originally presented as red blood cell (RBC) fouling of the electrode, but progressed into beneficial carbonization of the electrode surface under ambient conditions. The larger goal was to measure DC electrical resistances of RBC suspensions in a microchannel for hematocrit determination. As a consequence of this exploration, this paper details irreversible RBC adsorption onto the electrode surfaces, which positively affects the reproducibility of current signals across the microchannel. Initial current responses displayed irreproducible current measurements between runs and erratic changes in current within a single run. However, when current responses were repeatedly measured, platinum electrode performance improved over time. Systematic experimentation revealed that RBCs adsorb and carbonize the electrode surfaces. The carbon layers formed on the electrode surfaces are composed of denatured proteins and hemoglobin, which create porous 3D structures that increase SA to improve performance and stabilize currents. Platinum electrode surfaces before and after performance improvement were analyzed by field emission scanning electron microscopy (FE-SEM), energy dispersive spectrometry (EDS), and Raman spectrometry (RS).
2 Materials and methods 2.1 Microfluidic device Figure 1 presents the microfluidic device used for current response measurements of RBC solutions. Microchannel dimensions were 180 m by 70 m by 10 mm long. Brand new platinum wires (0.10 mm diameter, 99.99%, Sigma-Aldrich) were used as anode and cathode electrodes without any pretreatment. Platinum wires were cut via scalpel and bent by tweezers to immerse 2 mm into each reservoir. General-purpose epoxy (Henkel Corporation, Rocky Hill, CT, USA) attached platinum wire electrodes to a C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Microfluidics and Miniaturization
1667
photolithographically fabricated PDMS microfluidic layer, as previously described [27].
2.2 Sample preparation Blood samples obtained from voluntary donors via venipuncture were stored in vacutainers (Becton, Dickinson, and Company, Franklin Lakes, NJ, USA) containing 1.8 mg K2 EDTA (dipotassium ethylenediaminetetraacetic acid) per mL of blood at 4°C. RBCs were isolated from plasma by centrifugation at 1400 rpm (rcf = 110 g) for 5 min. Isolated RBCs were rinsed twice with 0.9 wt% NaCl solution by spinning at 1400 rpm for 5 min. RBCs were then mixed with 1000 S/cm isotonic PBS (2.57 mM KH2 PO4 , 2.57 mM K2 HPO4 , 2.57 mM NaCl, 274.3 mM dextrose) to form a 60 vol% RBC suspension. To test pH dependence of the carbonized electrodes, 2000 S/cm PBS at pH 3 (1.79 mM KH2 PO4 , 1.79 mM K2 HPO4 , 1.79 mM NaCl, 0.12 mM dextrose, 4.80 mM HCl), pH 7 (5.06 mM KH2 PO4 , 5.06 mM K2 HPO4 , 5.06 mM NaCl, 0.68 mM dextrose), and pH 11 (1.94 mM KH2 PO4 , 1.94 mM K2 HPO4 , 1.94 mM NaCl, 0.23 mM dextrose, 7.32 mM NaOH) were prepared. Lastly, 320 M Triton X-100 (T8532, Sigma-Aldrich, Saint Louis, MO, USA) was added to the 60 vol% RBC suspension and the 1000 S/cm PBS solution to stabilize current responses. Triton X-100 attenuates electrolysis bubble effects that destabilize current responses by facilitating formation of smaller bubbles. It was shown that 320 M Triton X-100 does not compromise RBC membrane integrity [27, 28]. 2.3 Current measurements Separate current responses of 1000 S/cm isotonic PBS and 60 vol% RBC suspension were measured in alternating cycles. RBCs served as the only carbon source in the system. Anode and cathode platinum electrodes were connected to an HVS 448 high voltage sequencer (LabSmith, Livermore, CA, USA). PBS or RBC suspensions were injected via BD Safety-LokTM 3 mL syringes (Becton, Dickinson and Company) through a bonded anode well port connector (LabSmith) and pushed through the microchannel to the cathode well. A 100 V DC signal was applied for 60 s while concurrently saving current responses at 39.06 Hz using Sequence Software (LabSmith). After each measurement, the entire microfluidic device and microchannel were flushed with e-pure water (Millipore Simplicity 185, Millipore Corporation, Billerica, MA, USA, Resistivity: 18.2 MΩ). 2.4 Sample characterization Platinum electrodes before and after current measurements were imaged using a Hitachi S-4700 FE-SEM operated with a 5.0 keV electron beam. Platinum wire electrodes were cut off the microfluidic device using a scalpel and placed, uncoated, www.electrophoresis-journal.com
1668
H. Y. Lee et al.
Electrophoresis 2015, 36, 1666–1673
Figure 1. Schematic drawing of the single channel microfluidic device. (A) Device configuration illustrating anode and cathode platinum wires (0.10 mm in diameter) submerged in fluidic wells connected by a microchannel 180 m by 70 m by 10 mm long as shown in (B). Not drawn to scale.
Figure 2. (A) Current responses of 1000 S/cm PBS (blue traces) and 60 vol% RBC suspension (red traces) at 100 V DC versus time. Each bar represents one 60 s measurement, begun with new solution. Note the broken time axis to enlarge initial and final current responses. (B) Average current responses in the boxes labeled as before and after in (A). Error bars represent 95% confidence intervals.
on carbon-tape stubs using tweezers. Electrode surface composition was determined via EDS with a 15.0 keV electron beam. Raman spectrometric spectrums were collected and analyzed from a carbonized anode excited with 632 nm laser using a Horibia Jobin Yvon LabRAM HR 800 RS.
3 Results and discussion 3.1 Current responses Figure 2A shows current responses sampled at 39.06 Hz of 1000 S/cm PBS (blue traces) and 60 vol% RBC suspensions in PBS (red traces) at 100 V. The plot time-axis is discontinuous to emphasize initial and final responses. Average current responses of the first and last eight measurements of each solution are highlighted by green boxes in Fig. 2A and compared in Fig. 2B. RBC solution current responses gradually increased with each cycle while the PBS baseline remained C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
fairly constant. PBS fluctuations, such as from 1440 to 1500 s and from 1560 to 1620 s, were attributed to residual blood cells after manual device flushing with e-pure water. Fewer large fluctuations occurred as the trials progressed indicating that the platinum electrodes incrementally stabilized. These observations were repeated with ⬎10 electrode pairs, although averages reported are for a representative experiment cycle. Current response fluctuations of both PBS and RBC solutions were attributed to isolated and thus unstable RBC adsorption. It was previously documented that unwanted molecule adsorption fouls biosensor electrode surfaces resulting in irreproducible electrochemical responses [8–11]. We thus hypothesized that RBC adsorption on platinum electrode surfaces was irregular, yet incrementally accumulating during measurements. After ten cycles of RBC current measurements and ten cycles of PBS, the platinum wire electrodes lost their luster; after 20 cycles of RBC current measurements and 20 cycles of PBS, the anode appeared black to the naked eye. RBC current responses decreased during a cycle, such as www.electrophoresis-journal.com
Microfluidics and Miniaturization
Electrophoresis 2015, 36, 1666–1673 Table 1. S/Ns of current responses before and after in Fig. 2A
PBS 60 vol% RBC
Before
After
55.5 51.8
65.5 170
from 900 to 960 s and from 1740 to 1800 s. This phenomenon, which decreased as the electrode aged, was attributed to either carbonized RBC clusters fragmenting off the electrode, which was experimentally observed, or to collapse of densely carbonized RBCs onto the electrodes. In Fig. 2B, 60 vol% RBC current responses before and after were 25.3 ± 2.13 A and 60.2 ± 0.67 A, respectively. This indicated that, as current signal responses were measured, the platinum electrode electrocatalytic activity (term used here to reflect the relative amount of active surface sites participating in the electron transfer reaction) for the RBC current measurements was increased by 140%. Over time, current responses became more stable and reproducible as indicated by reduced error bars in Fig. 2B. Their values before and after in Fig. 2B were 10.5 ± 1.13 A and 12.6 ± 0.20 A. It is presumed that the RBC layers modified the platinum electrode surface properties to improve performance similar to how other carbon-coated electrodes improve performance [20, 29]. S/Ns before and after were compared in Table 1. S/N ratios of both PBS and RBC suspension current responses increased with measurement cycles. The average PBS current response signals were nearly constant, so the increased S/N was attributed to reduced noise. The decrease in RBC suspension S/N was predominantly due to increased current responses and decreased SD, which before was 0.922 and after was 0.483. The entire sequence in Fig. 2A was reproduced more than ten times (N ⬎ 10) beginning each time with new platinum electrodes. The result—improved RBC current responses stabilizing between 60 and 63 A—was observed each time. The most common sequence length was 30–50 cycles of the 60 vol% RBC solution current measurements (plus the repeated PBS current measurements). Once current responses were stabilized at 60–63 A, the entire immersed anode electrode was completely black and fluctuations of current responses were minimal. The black color was attributed to carbon layers covering the electrode surface, consistent with prior literature reporting minimal biomolecule adsorption onto carbon-based electrodes [21, 30, 31]. The black carbon layers were very brittle and could be easily removed from the electrode surface mechanically by tweezers or high-pressure air. If the carbon layers were kept intact, the electrodes retained improved performance. However, if the carbon layers were damaged, the current responses decreased below 60 A. Then, by repeating the same PBS and 60% RBC solution measurement cycles, current responses could be restored to 60–63 A again. These carbonized platinum electrodes demonstrated sufficient activity and stability to reproducibly measure RBC hematocrit to 2 vol% resolution [28]. C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1669
3.2 Electrode surface characterization To directly investigate the black layers on the electrode surfaces, unused platinum electrode wire surfaces and the carbonized platinum electrode surfaces showing the improved performance were imaged by FE-SEM. Black layer compositions were analyzed via EDS and the results shown in Table 2. Due to platinum wire electrode curvature, EDS analysis locations along the same ordinate were chosen for quantitative comparison. Figure 3A, B, and C illustrate unused 0.10 mm, 99.99% pure platinum wire electrode surfaces at three magnifications. EDS elemental analysis was performed on both the dominant bright regions and the sporadic dark regions marked by A and B, respectively, in Fig. 3C and correlated to values in Table 2. The dark regions were carbon rich, while the bright regions were nearly pure platinum. Overall, the unused platinum surface contained measurable amounts of carbon and oxygen. Figure 3D, E, and F illustrate SEM images of the cathode surface after 31 cycles with 60 vol% RBC and 31 intermittent PBS current measurements. This electrode demonstrated stable 60 A signals in 60 vol% RBC solutions. Darker regions were more prominent compared to unused platinum electrodes as shown in Fig. 3D. EDS analysis revealed these dark regions were carbon rich as cataloged in Table 2. In Fig. 3E, the left most region labeled C had similar composition to the unused platinum electrode surface in Table 2. However, the dark region D contained 31.9% carbon. Figure 3F is a 10 000× magnification of region D. Another cathode was imaged after 49 current measurement cycles with PBS and 49 cycles with 60 vol% RBC as shown in Fig. 3G, H, and I. Figure 3G shows a scalpel-induced curved cut (labeled E) and F marks the carbon tape used to adhere the electrode to the FE-SEM stub. In contrast with Fig. 3D, E, and F, the darker regions more extensively cover the electrode surface (region G) in Fig. 3H. EDS of G indicated more extensive carbonization, as catalogued in Table 2. However, the bright region marked as H in Fig. 3H retained composition similar to unused platinum. These results indicate that with increasing measurement cycles, carbon coverage of the cathode electrode surface increased. The source of the carbonization was attributed primarily to irreversible RBC adsorption onto the platinum surface because this is the predominant source of carbon within the solution. The solution also contains 320 M Triton X-100, which is a nonionic surfactant comprised of a hydrophilic polyethylene oxide chain and an aromatic hydrocarbon hydrophobic group ((C2 H4 O)n C14 H22 O, N = 9 – 10). It was observed that repeated current measurements in only PBS solution containing Triton X-100 did not change electrode surfaces. Other experimental impurities that could serve as sources of carbon include the PDMS channel itself and the bonded port connector composed of Ultem (polyetherimide) although these solids have negligible solubility in water [32]. The corresponding treated anode electrode surfaces were also imaged. Figure 3J, K, and L shows the anode surface R
www.electrophoresis-journal.com
1670
H. Y. Lee et al.
Electrophoresis 2015, 36, 1666–1673
Table 2. Composition of platinum electrode surfaces in Figs. 3 and 4, and 5 analyzed by EDS
Electrode
New Cathode
Anode
Labela)
A B C D H G L K M N
Surface status
Bright Dark Bare Carbonized Bare Carbonized Bare Carbonized Bare Carbonized
Element (weight%) C
O
Pt
N
14.8 ± 1.37 20.0 ± 5.35 14.5 ± 1.05 31.9 ± 14.9 13.3 ± 0.26 38.6 ± 5.77 15.9 ± 1.32 56.8 ± 2.59 12.4 55.9 ± 3.34
0.82 ± 0.29 1.39 ± 0.67 1.09 ± 0.43 2.36 ± 1.24 0.88 ± 0.04 3.18 ± 0.55 1.37 ± 0.01 5.94 ± 0.83 1.83 9.67 ± 0.54
84.4 ± 1.62 78.6 ± 6.01 84.4 ± 1.44 65.7 ± 16.1 85.8 ± 0.23 58.2 ± 6.26 84.3 ± 0.17 36.1 ± 3.23 85.8 28.9 ± 3.28
4 3 3 4 3 3 2 2 1 3
a) The capital letters in the label column, corresponding to the labels in Figs. 3 and 4, and 5, indicate representative regions for the analyses.
Figure 3. FE-SEM images of platinum wire electrodes. First column is 500X magnification, second column is 1000X, and third column is 10 000X. Unused platinum wire electrodes in (A), (B), and (C). Cathode in (D), (E), and (F) after 31 cycles of the 60 vol% RBC suspension current measurement (62 cycles if including the PBS current measurements). Cathode in (G), (H), and (I): after 49 cycles of 60 vol% RBC. Anode in (J), (K), and (L) after 31 cycles of the 60 vol% RBC and (M), (N), and (O) after 49 cycles of 60 vol% RBC suspension. Yellow boxes indicate representative regions for EDS analysis reported in Table 2.
after 31 current measurements cycles with PBS and 31 cycles with 60 vol% RBC. This anode was paired with Fig. 3D, E, and F and demonstrated a stable 60 A in 60 vol% RBC solution. Figure 3K, a magnified view of the I region in Fig. 3J, is a side view of the coated layers, which illustrated a rough 3D anode surface, which increases SA. A top view of the C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
layers, an enlarged image of region J, is shown in Fig. 3L. EDS analysis revealed that the K region was a carbon rich layer containing 56.8 ± 2.59 wt% carbon as reported in Table 2. In K, 1.16 ± 0.19 wt% chlorine was also detected. Smoother regions, like L, had similar composition to untreated platinum.
www.electrophoresis-journal.com
Electrophoresis 2015, 36, 1666–1673
Microfluidics and Miniaturization
1671
Figure 4. (A) Raman spectrum of an anode surface after 60 cycles of 60 vol% RBC current measurements (120 cycles if including PBS cycles). (B) Current responses of 2000 S/cm PBS at pH 3, pH 7, and pH 11 at 100 V for 1 min. Error bars represent 95% confidence intervals (N = 4 7).
Figure 3J, K, and L indicate that the coated carbon layers developed into 3D porous structures, and differed structurally from the cathodes. RBCs have negative zeta potential in PBS, which would electrophoretically drive RBCs to the positively charged anode surface to deposit into 3D structures. RBC zeta potential is –31.8 ± 1.1 mV [33] in PBS with 1.7 mM KH2 PO4 , 5.2 mM Na2 HPO4 , 150 mM NaCl, and pH 7.4, which is similar to our PBS medium condition of 2.57 mM KH2 PO4 , 2.57 mM K2 HPO4 , 2.57 mM NaCl, 274.3 mM dextrose, and pH 7.2. RBC adsorption is energetically favorable because adsorption reduces the interfacial energy at the electrodesolution interface. In contrast, the negatively charged cathode would repel intact RBCs, so cathodic 3D adsorption was not observed as shown in Fig. 4. The anode electrode was examined after 49 cycles of RBC current measurements and 49 cycles of PBS and shows much more extensive carbon layers than the 31 cycles in Fig. 3J, K, and L. The FE-SEM images shown in Fig. 3M, N, and O represent anode status right after current measurements in Fig. 2, which displayed 60.2 ± 0.67 A with the 60 vol% RBC suspension at 100 V DC. Handling of the completely coated anode electrodes was challenging because the black carbon layers were brittle and easily damaged. Thus, it is inferred that the carbon layers were deposited onto the platinum electrode surface via electrostatic interactions and at most, the bonding between the carbon layers and the platinum surface must be weak noncovalent bonding. Figure 3M illustrates a region, labeled M, where the 3D carbon layer peeled off; this region had the same composition as unused platinum as reported in Table 2. The carbon layer, marked by N in Fig. 3N was 4.46 ± 2.14 wt% of chlorine and 1.08 ± 0.58 wt% of potassium were also detected and would have originated from the PBS solution. This PBS microfluidic system previously demonstrated electrolysis bubble generation [27]; thus, electrode electrolysis reactions occur concurrent with carbon adsorption. During current measurements, electrolysis of water generates oxygen
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
and hydrogen gas bubbles at the anode and cathode electrode surfaces, respectively [34]. Electrolysis bubble behavior could impede RBC and carbon adsorption, or, as shown in this work, could lead to a porous carbonaceous deposition. Adsorption interference was reported in an electrophoretic colloidal coating process whereby charged particles in suspension were deposited onto polarized electrode surfaces with opposite charges under an externally applied electric field. In this colloidal electrophoretic deposition, electrolysis bubbles incorporated into the deposit layers and led to porous and inhomogeneous deposited structures [30, 31, 35]. In our system, the size and number of electrolysis bubbles depends upon the applied voltage, medium conductivity (sample conductivity), and Triton X-100 concentration [27]. This work demonstrates an effective composition that favors deposition of RBCs into sufficiently thick and porous carbon layers that stabilize the current. Deposition rate would likely change with solution composition and loading percentage of RBCs (60 vol%); since RBCs are not an ideal carbon source, these further experiments were not conducted. It should be considered that the evolution, growth, and detachment rate of electrolysis bubbles is not spatially uniform even at the same experimental conditions due to local variations in platinum morphology [36]. This could be the reason for slight variations in the number of cycles necessary to achieve optimized electrode performance. In summary, the applied 100 V DC was high enough to generate electrolysis bubbles, which is a viable explanation for the resulting porous and inhomogeneous carbon structures illustrated in Fig. 4A, B, and C. Limitations of EDS analysis include that it is not reliable for minor element analysis (less than 1 wt%) and it cannot detect hydrogen. Thus, the coated layer composition was analyzed via RS. RS has been widely used to investigate the composition of structure of biological molecules and systems, such as proteins and blood [37–42]. Protein analysis via RS has focused on 3D protein structures and amino acid side chains [40–43].
www.electrophoresis-journal.com
1672
H. Y. Lee et al.
Figure 4A shows the Raman spectrum of the carbonized anode surface. RBCs are comprised of membrane proteins, phospholipids, and hemoglobin. Typical protein RS display a prominent amide I band near 1650 cm−1 and an amide III band near 1300 cm−1 reflecting polypeptide backbone conformations in proteins [40, 42, 43]. In Fig. 4A, the amide bands, especially the prominent amide I, did not appear indicating structural changes and denaturation of the membrane proteins and hemoglobin [44]. However, residues of hemoglobin are confirmed through the bands near 565 cm−1 by Fe-O2 bonding and 1360 and 1543 cm−1 due to the Soret band excitation of heme proteins [38, 39]. Amino acid side chains are also identified by two bands: tryptophan at 850 cm−1 and phenylalanine at 1000 and 1170 cm−1 [37, 41, 42]. These indicate that the electrode carbon layers are composed of denatured proteins and hemoglobin from lysed RBCs. The denatured proteins would act similar to structureless polymer chains (random coils) [45] on the electrode surfaces. Proteins are conductive, but their electrical properties depend upon pH as a consequence of conformational changes [45–47]. Thus, current responses at different pH values (2000 S/cm PBS solutions of pH 3, 7, and 11) were measured for 1 min at 100 V. Figure 4B shows that the current responses of the PBS of pH 11 (20.6 ± 0.40 A) were lower than pH 3 (23.7 ± 1.37 A) and pH 7 (23.8 ± 0.88 A). This evidence also supports the conclusion that the coated carbon layers were comprised of protein fragments. Based on the data presented herein, we have attributed improved electrode performance to the formation of carbon layers on the anode surface and carbon films on the cathode surfaces. During current measurements, adsorbed RBCs lysed onto the electrode surfaces, proteins denature, and progressively carbonized to form the observed black carbon layers. As the RBC current measurement cycles progressed, RBC adsorption compounded and eventually formed thick black carbon layers, which were especially pronounced on the anode. The anode carbon layers developed into 3D porous structures, likely facilitated by electrolysis, which increased the electrode surface area. These carbon layers were composed of the conductive denatured membrane proteins and hemoglobin from lysed RBCs. Consistent with findings in other carbon electrode systems, the critical factor with carbon-based modification of electrode surfaces is to achieve maximum effective surface area with well-optimized porous structures. The stabilization of current and increased electrocatalytic activity of the platinum electrodes in our microfluidic device suggests that carbon layers formed via RBC adsorption developed into optimized porous structures for stable RBC current measurements.
4 Concluding remarks Many microfluidic devices employ microelectrodes with small surface areas yielding weaker electrical signals. Prior biosensor research has revealed that irreversible biomolecule
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Electrophoresis 2015, 36, 1666–1673
adsorption destabilizes these electrical signals [8–10]. Successful strategies to reduce biomolecule adsorption have included bonding carbon nanotubes or graphene on electrode surfaces. The present work is an inadvertent extension of this approach with red blood cells as the carbon source. Systematic studies in RBC solutions suspensions revealed that RBCs offer a carbon source for carbon deposition on both the cathode and anode, although the structure and composition differed greatly between the electrodes. The platinum electrodes with 3D carbon layers formed from RBC adsorption reached a steady state as indicated by increased current responses with smaller S/N. These electrode carbon-coating results using RBCs are consistent with other electrode carbonization literature, which reports that (1) increasing electrode SA improves the electrode performance, and (2) carbon-based or carbon-coated electrodes experience less fouling. This suggests that biological sources of carbon may be utilizable to carbon-coated electrodes, and once coated reduce further biomolecule adsorption. The broader implications are that the ambient condition approach described here may serve as a lower cost alternative to carbon nanotubes or graphene binding to electrode surfaces. The authors gratefully acknowledge ACMAL (MTU) and Dr. Yoke Khin Yap for surface characterization. Additional gratitude is extended to Joe Halt, Zhichao Wang, and Maryam Khaksari for RBC sample preparation and valuable discussions. The authors also gratefully acknowledge prior discussions and efforts years ago with Soumya Srivastava and David O. Wipf. The authors also gratefully acknowledge recent funding support from NSF PFI:AIR IIP 1414331 and an NSF STTR IIP 1417187 subcontract. The authors declare no conflict of interest.
5 References [1] Yeo, L. Y., Chang, H.-C., Chan, P. P. Y., Friend, J. R., Small 2011, 7, 12–48. [2] Mark, D., Haeberle, S., Roth, G., vonStetten, F., Zengerle, R., Chem. Soc. Rev. 2010, 39, 1153–1182. [3] Whitesides, G. M., Nature 2006, 442, 368–373. [4] Minerick, A. R., AIChE J. 2008, 54, 2230–2237. [5] Naka, K., Hayashi, H., Senda, M., Shiraishi, H., Konishi, S., Micro Electro Mechanical Systems, 2007. MEMS. IEEE 20th International Conference on. 2007, pp. 195–198. [6] Zheng, S., Liu, M., Tai, Y.-C., Biomed. Microdevices 2008, 10, 221–231. [7] Nakanishi, K., Sakiyama, T., Imamura, K., J. Biosci. Bioeng. 2001, 91, 233–244. [8] Zourob, M., Recognition Receptors in Biosensors, Springer, New York 2010. [9] Atta, N. F., Galal, A., Ahmed, R. A., Bioelectrochemistry 2011, 80, 132–141. [10] Marino, A., Brajtertoth, A., Anal. Chem. 1993, 65, 370–374.
www.electrophoresis-journal.com
Electrophoresis 2015, 36, 1666–1673
[11] Nassar, A. E. F., Willis, W. S., Rusling, J. F., Anal. Chem. 1995, 67, 2386–2392.
Microfluidics and Miniaturization
1673
[30] Besra, L., Uchikoshi, T., Suzuki, T. S., Sakka, Y., J. Eur. Ceramic Soc. 2010, 30, 1187–1193.
[12] Baldwin, R. P., J. Pharm. Biomed. Anal. 1999, 19, 69–81.
[31] Ammam, M., RSC Adv. 2012, 2, 7633–7646.
[13] Gawron, A. J., Martin, R. S., Lunte, S. M., Electrophoresis 2001, 22, 242–248.
[32] Varaprath, S., Frye, C. L., Hamelink, J., Environ. Toxicol. Chem. 1996, 15, 1263–1265.
[14] Vandaveer, W. R., Pasas-Farmer, S. A., Fischer, D. J., Frankenfeld, C. N., Lunte, S. M., Electrophoresis 2004, 25, 3528–3549.
[33] Bondar, O. V., Saifullina, D. V., Shakhmaeva, II, Mavlyutova, II, Abdullin, T. I., Acta Nat. 2012, 4, 78–81.
[15] Lacher, N. A., Garrison, K. E., Martin, R. S., Lunte, S. M., Electrophoresis 2001, 22, 2526–2536. [16] McCreery, R. L., Cline, K. K., in: Kissinger, P., Heineman, W. R. (Eds.), Laboratory Techniques in Electroanalytical Chemistry, Marcel Dekker, New York 1996, pp. 293–332.
[34] Minerick, A. R., Ostafin, A. E., Chang, H. C., Electrophoresis 2002, 23, 2165–2173. [35] Chavez-Valdez, A., Shaffer, M. S. P., Boccaccini, A. R., J. Phys. Chem. B 2012, 117, 1502–1515. [36] Vogt, H., Balzer, R. J., Electrochim. Acta 2005, 50, 2073–2079.
[17] Pumera, M., Merkoc¸i, A., Alegret, S., Electroanalysis 2006, 18, 207–210.
´ ´ [37] Carmona, P., Monleon, E., Monzon, M., Badiola, J. J., Monreal, J., Chem. Biol. 2004, 11, 759–764.
[18] Wang, J., Chatrathi, M. P., Tian, B., Polsky, R., Electroanalysis 2000, 12, 691–694.
[38] Torres Filho, I. P., Terner, J., Pittman, R. N., Proffitt, E., Ward, K. R., J. Appl.Physiol. 2008, 104, 1809–1817.
[19] Choong, C.-L., Milne, W., Teo, K. K., Int. J. Mater. Form. 2008, 1, 117–125.
[39] Lin, J., Lin, J., Huang, Z., Lu, P., Wang, J., Wang, X., Chen, R., J. Innov. Opt. Health Sci. 2014, 07, 1350051.
[20] Pumera, M., Merkoc¸i, A., Alegret, S., Electrophoresis 2007, 28, 1274–1280.
[40] Maiti, N. C., Apetri, M. M., Zagorski, M. G., Carey, P. R., Anderson, V. E., J. Am. Chem. Soc. 2004, 126, 2399–2408.
[21] Wang, J., Chen, G., Wang, M., Chatrathi, M. P., Analyst 2004, 129, 512–515.
[41] Overman, S. A., Thomas, G. J., Biochemistry 1999, 38, 4018–4027.
[22] YeoHeung, Y., Zhongyun, D., Vesselin, N. S., Mark, J. S., Nanotechnology 2007, 18, 465505.
[42] Tuma, R., Journal of Raman Spectrosc. 2005, 36, 307–319.
[23] Zhang, D. W., Li, X. D., Li, H. B., Chen, S., Sun, Z., Yin, X. J., Huang, S. M., Carbon 2011, 49, 5382–5388.
[43] Williams, R. W., in: Hirs, C. H. W., Timasheff, S. N. (Eds.), Methods in Enzymology, Academic Press, New York 1986, pp. 311–331.
[24] Zhang, L. L., Zhao, X. S., Chem. Soc. Rev. 2009, 38, 2520–2531. [25] Long, J. W., Dunn, B., Rolison, D. R., White, H. S., Chem. Rev. 2004, 104, 4463–4492. [26] Chen, Y., Zhang, X., Yu, P., Ma, Y., J. Power Sources 2010, 195, 3031–3035.
[44] Borgognoni, C. F., Tattini Junior, V., Ayrosa, A. M. I. B., Polakiewicz, B., Leirner, A. A., Maizato, M. J. S., Higa, O. Z., Beppu, M. M., Pitombo, R. N. d. M., Brazilian Arch. Biol. Technol. 2009, 52, 1493–1504.
[27] Lee, H. Y., Barber, C., Minerick, A. R., Electrophoresis 2014, 35, 1782–1789.
[45] Tanford, C., in: Anfinsen, C. B. (Ed.), Advances in Protein Chemistry, Academic Press, New York 1968, pp. 121–282.
[28] Lee, H. Y., Barber, C., Rogers, J. A., Minerick, A. R., Electrophoresis 2015, 36, 978–985.
[46] Bazinet, L., Ippersiel, D., Mahdavi, B., J. Agricult. Food Chem. 2004, 52, 1980–1984.
[29] Wang, Y., Li, Q., Hu, S., Bioelectrochemistry 2005, 65, 135–142.
[47] Ron, I., Pecht, I., Sheves, M., Cahen, D., Accounts Chem. Res. 2010, 43, 945–953.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.electrophoresis-journal.com
