Subscriber access provided by CORNELL UNIVERSITY LIBRARY
Article
Carbon Nanotube Based Flow-Through Electrochemical Cell for Electroanalysis Andrea Buffa, Yigal Erel, and Daniel Mandler Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.6b02827 • Publication Date (Web): 17 Oct 2016 Downloaded from http://pubs.acs.org on October 19, 2016
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 11
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
Carbon Nanotube Based Flow-Through Electrochemical Cell for Electroanalysis Andrea Buffa†, Yigal Erel*§ and Daniel Mandler*† †
Institute of Chemistry, The Hebrew University of Jerusalem, Jerusalem 9190401, Israel
§
Institute of Earth Sciences, The Hebrew University of Jerusalem, Jerusalem 9090401, Israel Corresponding Author: Daniel Mandler, Fax: +972 2 6585319; Email:
[email protected] ABSTRACT: A flow-through electrode made of a carbon nanotubes (CNT) film deposited on a polytetrafluoroethylene (PTFE) membrane was assembled and employed for the determination of low concentration of copper as a model system by linear sweep anodic stripping voltammetry (LSASV). CNT films with areal mass ranging from 0.12 to 0.72 mg cm−2 were characterized by measurement of sheet resistance, water permeation flux and capacitance. Moreover, CNT with two different sizes and PTFE membrane with two different pore diameters (0.45 µm and 5.0 µm) were evaluated during the optimization of the electrode. Thick layers made of small CNT exhibited the lowest sheet resistance and the greatest analytical response, whereas thin layers of large CNT had the lowest capacitance and the highest permeation flux. Electrodes made of 0.12 mg cm−2 of large CNT deposited on 5.0 µm PTFE enabled sufficiently high mass transfer and collection efficiency for detecting 64 ppt of Cu(II) within 5 minutes of deposition and 4.0 ml min−1 flow rate. The analytical response was linear over four order of magnitude (10−9 to 10−5 M) of Cu(II). The excellent performance of the flow-through CNT membrane integrated in a flow cell makes it an appealing approach not only for electroanalysis but also for the electrochemical treatment of waters, such as the removal of low concentrations of heavy metals and organics.
electrode. Numerous studies have been reported where mass transfer, and as a result sensitivity, was markedly increased using electrochemical detection.
INTRODUCTION Electrochemistry is an inherently sensitive analytical tool as technology nowadays enables measuring very low currents, which are due to the oxidation or reduction of electroactive species. Hence, the accumulation of minute concentrations of species on the electrode surface allows their determination by electrochemical means. Sub mono-atomic layers of, for example, heavy metals can be readily detected. We have shown that the formation of self-assembled monolayers that interact selectively with various heavy metals made it possible to determine ppb and ppt levels of these metals in aquatic systems1-5. In most cases stripping voltammetry is employed whereby the metal is deposited (by applying negative potentials) on the electrode surface followed by its stripping (oxidation).
Yet, this is insufficient and to achieve very low detection limits, i.e. in the ppt level, relatively long accumulation times are frequently required. Moreover, the collection efficiency that is the ratio between the charge required to deposit the analyte onto the electrode surface divided by the charge of the ions that passed through the electrode, is very low and typically does not exceed a few percent. Evidently, if a substantial larger fraction of the analyte could be accumulated on the electrode surface (by passing the same charge) the sensitivity could increase by orders of magnitude and the time needed for the analysis could be shortened. The essence of this study is to develop a new approach whereby the solution flows through the electrode, i.e. flow-through electrode, and not parallel to the surface, i.e. flow-by electrode. In the past, there have been a few attempts to generate different types of flow-through electrodes based on highly porous materials such as reticulated vitreous carbon (RVC)15-17, carbon felt18,19 or glassy carbon particles.20-22 It should be noted that these materials find applications other than in electroanalytical systems.
The detection limit in electroanalytical methods is usually governed by mass transfer, namely, by diffusion, migration and convection of the species from the solution to the electrode surface. Different approaches have been developed to enhance mass transfer, such as employing microelectrodes6,7, introducing various methods of convection8-11 and increasing the temperature12. Convection can be driven by flowing the solution parallel or vertical to the electrode surface, or by rotating or vibrating13,14 the
Carbon nanotubes easily form a self-standing, porous and water permeable film called buckypaper23-27 whose
1
ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
filtration properties have been characterized in previous studies.28-32 The electrical conductivity and micro-porosity make buckypaper an ideal flow-through electrode. Indeed, in the last decade many investigations conducted by Vecitis et.al. studied buckypaper performance for flowelectrolysis.33-35 Most of the recent studies using buckypaper as a flow-through electrode principally target chemical and electrochemical water purification36-43 by virtue of the adsorptive properties and large surface area of CNT. Surprisingly and to the best of our knowledge, studies where buckypaper flow-through electrodes are applied in electroanalytical chemistry are still very limited.
Page 2 of 11
Instrumentation. All the electrochemical experiments were performed at room temperature using a CHI 750B and a CHI 630B electrochemical analyzer (CH Instruments, Austin, TX, USA). Vacuum was measured by a pressure gauge model DPG-3.0-500PSI-2P (Atlantis, Taipei, Taiwan). Pressure in the flow cell was measured by a polypropylene diaphragm pressure gauge model PDSPLS-Viton (Atlantis, Taipei, Taiwan). The sheet resistance was determined using a Jandel RM3000 four point probe test unit. The electrochemical flow cell was constructed using a 13 mm diameter Swinnex™ filter holder type SX0001300 (Merck Millipore). The flow cell was sealed with 13 mm diameter silicon gaskets type SX0001301 (Merck Millipore). In all experiments the analyte solution was pumped through the system by a SGE 50 ml syringe model GL50LL-S installed on a NE-500 OEM Syringe pump (New Era Pump Systems, Inc. Farmingdale, NY, USA). The counter electrode (CE) was a 0.5 mm thick Papyex® graphite sheet (Mersen) that was configured in a ring shape with an inner and outer diameter of 6 and 12.8 mm, respectively. The electrical contact to the CNT membrane working electrode (WE) was made using an oring having an outer and inner diameter of 12.8 and 8 mm, respectively, made of an electrically conductive carbon filled silicone sheet 0.5 mm thick (J-Flex Rubber Products, UK). The CE and WE were connected externally with 0.5 mm platinum wires (Good Fellow). The Ag/AgCl reference electrode was supplied by CH Instruments and filled with 0.1 M NaCl and 0.01 M HCl solution. Multi walled CNT type SKU030403 (10-2CNT, 10-20 nm outer diameter (OD), 0.5-2.0 µm length and 95% purity), and type SKU030107 (50-20CNT, >50 nm OD and 10-20 µm length and 95% purity) were purchased from Cheap Tubes Inc. (Grafton, VT, USA). CNT/NMP dispersions were prepared using a Sonics VCX-750 ultrasonic processor with a 13 mm probe. Hydrophilic PTFE Omnipore™ (Merck Millipore) membrane with 0.45 and 5.0 µm pores and 47 mm diameter were used into a 47 mm vacuum glass filter holder (Merck Millipore). In all the experiments a Vaccubrand MZ-2C vacuum pump was used. EpoThin™ 2 Epoxy System, SamplKlip® plastic sample support and SamplKup® mounting mold were purchased from Buehler and utilized for the preparation of the cross section specimen.
Stripping voltammetry is well known to be an intrinsically sensitive technique where the analyte is concentrated on the electrode surface by reduction or adsorption from a solution.44 The deposited analyte is then either oxidized or reduced while recording the current that is proportional to its surface concentration. Hence, this technique is capable of detecting, in particular, metals at very low concentration (40 m2 g−1 for 50-20CNT and >233 m2 g−1 for 10-2CNT.
Figure 1. Scheme of the flow cell: a) CNT membrane WE, b) PTFE substrate membrane, c) syringe filter case, d) Ag/AgCl RE, e) plastic membrane holder grid, f) conductive silicon oring, g) insulating silicon o-ring, h) graphite ring CE, i) CE connection, j) analyte solution inlet, k) analyte solution outlet, l) WE connection.
Type of CNT. CNT can be found in a variety of diameters lengths, purity and surface functionalization. Evidently, these are expected to influence the flow and sheet resistance of the membranes made of CNT. Therefore, we decided to examine the membranes composed of two quite different CNT, i.e. 50-20CNT with a measured OD of 37 ± 8 nm and relatively long ca. 20 µm and 10-2CNT that consist of tubes with a measured OD of 17 ± 3 nm and length of ca. 0.5-2 µm. Figure 2 shows the SEM images of membranes made of 50-20CNT and 10-2CNT with different magnification. An appreciable difference in the size of the two types of CNT is evident and it is confirmed by the measurements of the diameter in figures 2B and 2D. As a consequence of the larger diameter of the 50-20CNT, the void volume in the membrane made of 10-2CNT is significantly smaller than that of the 50-20CNT membrane. This entails a superior hydrodynamic permeability for membranes made of 50-20CNT as is well documented for fibrous materials52. The absence of agglomeration of CNT is evident in both types of membrane, hence this testifies to the stability of the CNT/NMP dispersion which was used for deposition.
Electrochemical experiments. Unless specified otherwise, all Cu(II) solutions used in the electrochemical experiments were prepared using 0.1M HNO3 as supporting electrolyte. The reference electrode was placed at the outlet of the electrochemical cell (Figure 1). Every new membrane was primed by flowing 0.4 ml of ethanol to enable flow of aqueous solution through the hydrophobic net of CNT. The flow was stabilized before applying potential to the cell by waiting that at least 1 ml of solution flowed through the cell after the pump was started in order to reach the working pressure. LSASV was used for depositing the analyte. The flow of analyte solution was not interrupted during the deposition, stripping and cleaning steps. The electrode was cleaned by linear sweep voltammetry (LSV) after each LSASV experiment by scanning once from 0 to +0.7 V at 200 mV s−1. A subsequent measurement could be performed after the cleaning step if the flow was not interrupted. In all experiments deaeration of the solutions was not performed. In all the plots the error bars are calculated as the t-based 95% confidence interval for the mean of three measurement and 2 degree of freedom.
RESULTS AND DISCUSSION The CNT based membranes, which were formed by filtering a dispersion of CNT across a membrane, were characterized prior to being applied for copper determination. The parameters affecting its performance, videinfra, can be divided into internal, i.e. type and AM of the CNT and external such as the flow rate, time and potential. The following sections will deal with the characterization and optimization of these parameters. CNT characterization. The diameter of the CNT was measured by SEM imaging obtaining 37 ± 8 nm for 5020CNT that is smaller than manufacturer specifications of >50 nm. The diameter of 10-2CNT was indeed 17 ± 3 nm that meets the specifications of the manufacturer. CNT characterization must include the determination of the O/C ratio and residual catalyst which affect the properties of CNT. XPS analysis provided an O/C ratio of 0.013 and 0.020 for 10-2CNT and 50-20CNT respectively but residues of catalyst such as Fe, Co, and Ni were not detected because they are located too deep in the core of CNT to be reached by XPS analysis.39 EDX analysis was able to detect 2.0 ± 0.6 wt% Co in 10-2CNT and 3.0 ± 0.3 wt% Ni in 50-20CNT as residual catalysts. TGA analysis provided a residual mass of 3.3% for 50-20CNT and 2.5% for 102CNT. The surface area measured by BET was 64.3 m2 g−1 for 50-20CNT and 133.8 m2 g−1 for 10-2CNT. According to these results only 50-20CNT met the specification of the
Figure 2. High and low magnification SEM images of the CNT membranes made of 50-20CNTs (A, B) and 10-2CNTs (C, D).
A typical cross section of 0.48 ± 0.01 mg cm−2 of 5020CNT layer deposited on 0.45 µm membrane is shown (in two magnifications) in figure 3A-B. The cross section shows a compact and homogenous CNT layer (upper layer) with a constant thickness of 7.7 ± 0.6 µm. The larger magnification clearly shows that the CNT created a very distinct interface with the PTFE membrane (lower layer). The thickness of a 0.96 ± 0.03 mg cm−2 50-20CNT membrane was also measured, providing a result of 18 ± 1 µm. The composite membrane (CNT/PTFE), in fact, forms an asymmetric membrane where the skin is the CNT and is of great importance in water treatment and is commonly used nowadays in ultra- and nanofiltration.53 Furthermore, and as we are going to show, we were able to as-
4
ACS Paragon Plus Environment
Page 5 of 11
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
semble membranes as thin as 0.120 ± 0.003 mg cm−2 with a thickness of 1.9 ± 0.2 µm deposited on a 5.0 µm PTFE membrane.
varying with the thickness. Specifically, it drops as predicted by the theory of percolation for thin films composed of conductive nanoparticles.54,55 The plot of the sheet resistance as a function of the flux gives a straight line for 50-20CNT deposited on 5.0 µm membrane in a range of AM from 0.12 to 0.72 mg cm−2 (Figure S-2). The 50-20CNT exhibit faster flow rate than 10-2CNT in particular when deposited on 5.0 µm pore membrane with AM < 0.5 mg cm−2, whereas 10-20CNT had a lower sheet resistance. The best performance in terms of flux was 10200 ± 600 L m−2 h−1 and was reached by depositing 0.120 ± 0.003 mg cm−2 of 50-20CNT on a 5.0 µm pores substrate. These conditions provided membranes with a sheet resistance of 240 ± 30 Ω/sq and a resulting conductivity of 2200 ± 400 S m−1 whereas the conductivity of the 0.48 ± 0.01 mg cm−2 and 50-20CNT membranes was 2600 ± 300 S m−1. These values of sheet resistance and conductivity agree with those previously reported for CNT buckypaper prepared by vacuum filtration.56,57
Figure 3. SEM image of the cross section of a 50-20CNTs −2 membrane with a AM of 0.48 ± 0.01 mg cm at low (A) and high (B) magnification.
Sheet resistance and water permeation flux. The CNT membrane flow-through electrodes have to maximize the flux while minimizing the sheet resistance. This will allow large volume of analyte to be flowed through the electrode in a short time while maintaining a uniform current density throughout the entire membrane.
The sheet resistance of 50-20CNT membranes with AM from 0.12 to 0.72 mg cm−2 was measured before and after that the membranes were wetted by deionized water. Figure S-1 shows that when the membranes are wetted by deionized water their sheet resistance is ca. twice that one measured on the dry membrane. Our results demonstrate as expected that the size of CNT and the pore size of the substrate membrane affect buckypaper's sheet resistance and water permeation flux. Nevertheless previous studies demonstrated that the electrical and hydraulic properties of buckypaper are also determined by the amount of impurities and extent of surface oxidation and functionalization of CNT39. Moreover, the transmembrane differential pressure at the moment of deposition is critical for the fabrication of membranes with repeatable features.
Figure 4. Dependence of the flux at constant transmembrane pressure difference (A) and the sheet resistance (B) on the AM of membranes made of 50-20CNT deposited on 0.45 µm (red line) and 5.0 µm (black line) membranes and 10-2CNT deposited on 0.45 µm (blue line).
Figure 4 compares the water flow rate (A) and the sheet resistance (B) under constant transmembrane pressure. Two different types of CNT as well as two substrate membranes differing in their pore size were employed. Measurements were performed with four different AM using triplicates. The results in Figure 4 show, as expected, that the sheet resistance and flux decrease as the AM of the CNT layer decreases. According to Darcy’s law the flux of a Newtonian fluid through a porous membrane is proportional to the hydrodynamic permeability and the transmembrane pressure difference and inversely proportional to the thickness of the membrane.52 The inset in Figure 4A shows that flux and thickness are indeed hyperbolically dependent only when the 5.0 µm membrane was used. In the case of 0.45 µm membranes a linear dependence of the flux on the reciprocal of the thickness of the membrane was not obtained because the hydrodynamic permeability was not constant over the experimental range of AM. The plot of the sheet resistance as a function of the reciprocal of the membrane thickness (inset in figure 4B) shows that the sheet resistance was inversely proportional to the thickness only in a range of AM from 0.12 to 0.72 mg cm−2 whereas at 0.06 mg cm−2 the conductivity begins
Flow rate in the cell. The flow rate strongly affects the sensitivity and time efficiency of the analytical method, therefore the determination of the maximum flow rate in the flow cell (enabled by different types of membranes) is important for establishing the experimental conditions. According to the specifications of the supplier the maximum pressure (P) allowed in the syringe filter case is 345 kPa. Therefore, membrane with different hydrodynamic permeability should allow different maximum flow rates within this pressure value as shown in table 1. Nevertheless, the maximum applied pressure in our experiments was dictated by the pump, which was ca. 207 kPa. The filtering area of the membrane when installed in the flow cell was 0.5 cm2 and was determined by the inner diameter of the conductive silicon ring used as electric contact. The flow rates specified in table 1 were considered maximum for the respective membrane. Thus, flow rates of up to 5 ml min−1 were applicable with our instrumentation providing that the pressure did not increase due to membrane fouling. Table 1. Maximum flow rate in the cell.
5
ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
CNT type
PTFE pore size (µm)
AM −2 (mg cm )
10-2CNT
0.45
10-2CNT
5.0
P (kPa)
0.48 ± 0.01
Flow rate (ml −1 min ) 0.5
0.12 ± 0.003
1.5
207
Analytical response. The analytical signal of 10−5 M Cu(II) was evaluated by LSASV using two membranes made of 50-20CNT with AM of 0.48 and 0.12 mg cm−2 and a sheet resistance of 50 ± 4 Ω/sq and 200 ± 6 Ω/sq, respectively. Figure 5A-B shows the LSASV with different deposition times for both membranes. It can be seen that the peak currents gradually increase with the deposition time, which is evidently due to the increase of Cu deposition on the CNT. The voltammograms obtained by the 0.48 mg cm−2 electrode (Figure 6A) have a higher peak current as compared with those recorded by the 0.12 mg cm−2 electrode (Figure 6B) under the same operating conditions. This can be explained by the lower resistance of the membrane with higher AM. These experiments were repeated with 10-2CNT membranes providing similar results.
138
50-20CNT
5.0
0.48 ± 0.01
2.0
172
50-20CNT
0.45
0.12 ± 0.003
4.5
179
50-20CNT
5.0
0.12 ± 0.03
5.0
186
Page 6 of 11
Effect of the areal mass on capacitive currents. Minimizing the non-faradaic current is very important to reduce the detection limits of an electroanalytical method. The high specific surface area of CNT implies high capacitance of the electrodes made of this material. In fact, the applicability of CNT for the construction of supercapacitors has been widely demonstrated.58-61 The capacitance of four membranes made of 50-20CNT deposited on 0.45 µm substrate was measured by cyclic voltammetry (CV) at different scan rates (Figure 5B). The peaks appearing on all voltammograms at ca. +0.2 V are ascribable to reversible redox reaction of oxygen containing groups present on the surface of the CNT.62-65 The inset of Figure 5B shows that a linear dependence of the capacitive current (measured in the non-faradic region) on the scan rate is obtained. The slope of this straight line provides the capacitance, which is plotted as a function of the total surface area of the electrode in the inset of Figure 5A. Clearly, the capacitive current increases as a function of the AM of the membranes implying that the surface area increases with the AM. The specific capacitance calculated for 50-20CNT and 10-2CNT in 0.01M HNO3 was 4.9 F g−1 and 8.0 F g−1, respectively. The difference in specific capacitance is attributed to the different surface area of the two types of CNT. The results above do not diverge significantly from the previously reported results of 12.2 F g−1 for non-functionalized CNT with a surface area of 148.6 m2 g−1 64 and 4 F g−1 for CNT with a surface area of 128 m2 g−1.59 We recall that the surface area of our 5020CNT and 10-2CNT is 64.3 m2 g−1 and 133.8 m2 g−1, respectively. These results suggest that thin membranes perform better than the thicker because of their lower capacitive current.
−5
Figure 6. Background subtracted stripping peaks of 10 M −1 Cu(II) obtained at a flow rate of 0.5 ml min and different −2 −2 deposition times, using 0.48 mg cm (A) and 0.12 mg cm (B) made of 50-20CNT on 0.45 µm PTFE substrate.
For both AM, there is a continuous shift of the peak potential to more positive values with increasing deposition time. This is not because of higher iR drops as compensating for the resistance of the solution (which was
