Communication pubs.acs.org/JACS
Electron-Transfer Gated Ion Transport in Carbon Nanopipets Dengchao Wang and Michael V. Mirkin* Department of Chemistry and Biochemistry, Queens College, City University of New York, Flushing, New York 11367, United States S Supporting Information *
ABSTRACT: Coating the inner wall of a quartz nanopipet with a thin layer of carbon yields a nanopore with tunable surface charge and chemical state for resistivepulse and rectification sensing. Herein we report the experimental study and modeling of the electron-transfer gated ion transport processes in carbon nanopipets. The potential of the unbiased carbon layer can be tuned by adding very low (sub-nM) concentrations of redox species to the solution via bipolar electrochemistry. The potential of the carbon layer determines the electrical double-layer structure that, in turn, affects the ionic transport processes. The ion current rectification decreased when redox species with a relatively positive formal potential (e.g., Fe(CN)63/4−) were added to the solution and increased upon adding redox species with a negative formal potential (e.g., Ru(NH3)63/2+). Additionally, the ion current displays high sensitivity to redox species, suggesting the possibility of trace-level analysis.
Figure 1. Schematic representation of the charge transport in a CNP. The voltage applied between the internal and external reference electrodes can drive bipolar electron-transfer processes at the carbon surface, while the carbon potential affects the ion current flowing through the nanopipet.
the external voltage is applied between them to drive the transport of ions and analytes through the nanopipet. The carbon film can be connected to the potentiostat and serve as a working electrode or be left floating with no direct ohmic contact, as shown in Figure 1.18 If the thickness of the electrical double layer (EDL) at the carbon/solution interface is comparable to the radius of the pipet orifice, the ionic transport in its narrow shaft is strongly affected by the surface charge density1,7 and therefore, by the carbon potential.18 However, the surface area of carbon exposed to the solution is by many orders of magnitude larger that of the orifice, and applying voltage between the carbon film and one of the reference electrodes results in significant faradaic current interfering with the ion current through the pipet. Here we demonstrate the possibility of controlling the ionic transport processes in a CNP by adding low concentrations of redox species to the solution via bipolar electrochemistry30 at the carbon surface. As many common redox mediators (i.e., Fe(CN)63−/4−, Ru(NH3)63+/2+) are multivalent ions, the adsorption effect of these species needs to be investigated and separated from the ET effect on ion current through CNPs. A TEM image of the 170 nm CNP and a current−voltage (i− V) curve are shown in Figure 2. CNP i−V curves show stronger ion current rectification (ICR) than those obtained with comparably sized quartz pipets. The ICR is due to the combination of the voltage-induced concentration polarization with the conical geometry and negative surface charges.1,31−35 In Figure 2B, the rectification factor at ±0.5 V is ∼6 as compared to 1.4 for a 130 nm quartz pipet (Figure S1B), suggesting that the negative charge density on the carbon surface is significantly higher. Additionally, when the carbon layer is polarized by the applied voltage, the resulting surface charge distribution can also
I
n addition to being of considerable fundamental interest,1,2 mass/charge transport in nanoscale channels and pores is essential for numerous applications, including nanopore sensors,3−7 nanofluidics,8 membrane transport,9 and highefficiency energy storage and conversion systems.10 To control the transport of ions and neutral analytes through the nanochannels and improve selectivity of nanopore-based sensors, their inner surfaces have been chemically modified, e.g., by covalent attachment of charged molecules.11−13 Smart nanopore sensors responding to the pH, potential, temperature, and light were fabricated by modifying the interior surface with suitable stimuli-response materials.14 Alternatively, the interior wall of the nanochannel can be functionalized by depositing conductive materials such as gold, 15−17 carbon,18,19 or conductive polymers20 or embedding electrodes21 to control the electrostatic interactions between the surface and solution ions. Novel conductive nanopore platforms such as carbon nanotubes22 and 2-D graphene23 can also offer electric conductance24 and optical measurement25 of the analytes. Although conductive nanopores and pipets are potentially useful for DNA sequencing26 and protein detection,27 little is known about the interplay between the electron-transfer (ET) processes at the conductive surface and the mass transport processes in the nanocapillary. The charge transport processes in a carbon nanopipet (CNP) are shown in Figure 1. The interior surface of the quartz nanopore is coated with a conductive carbon layer by chemical vapor deposition (Supporting Information).18,28,29 Two reference electrodes are placed in the inner and outer solutions, and © 2017 American Chemical Society
Received: May 16, 2017 Published: August 11, 2017 11654
DOI: 10.1021/jacs.7b05058 J. Am. Chem. Soc. 2017, 139, 11654−11657
Communication
Journal of the American Chemical Society
Figure 2. (A) TEM image of the 170 nm CNP and (B) i−V curve in 10 mM KCl solution. Potential scan rate, v = 0.1 V/s.
affect the ICR.17 The hysteresis loops in i−V curves indicative of slow ionic dynamics inside the pipet on the given experimental time scale (Figures 2B and S1B) are similar those reported earlier for conical pores and quartz nanopipets.1,36 The multivalent ions are known to adsorb on the interior surface (either bare or chemically modified) of a glass nanopore and affect the ionic transport processes.37,38 The binding of double-charged cations (i.e., Ca2+ or Mg2+) to the poly-acidmodified surface resulted in the screening of excess surface charge and decrease in the ionic conductance.37 Moreover, the polarity of the charged surface can be inverted at high concentration of multivalent cations.39 The effect of Ca2+ (Figure 3) and citrate3− (Figure S2) adsorption on the CNP Figure 4. Effect of (A) K3Fe(CN)6 and (B) Ru(NH3)6Cl3 concentration on the CNP i−V curves in 10 mM KCl. (A) From top to bottom, experimental (solid lines) and simulated (symbols) i−V curves for [K3Fe(CN)6], nM: 0 (black), 100 (red), 200 (blue), and 300 (green); and surface charge density, σ (mC/m2) = −23, −10, −4, and −2. (B) [Ru(NH3)6Cl3], pM: 0 (black), 100 (red), 200 (blue), and 300 (green); and fitted σ (mC/m2) = −5, −10, −30, and −60. The magenta curve was obtained with [Ru(NH3)6Cl3] = 100 nM.
of K3Fe(CN)6 were low in comparison to the 10 mM KCl and the effect of adsorption of anionic Fe(CN)63− species should be negligible (see above), the decrease in ICR can be attributed to ET process at the carbon layer involving Fe(CN)63−. The degree of ICR in the CNP is determined by the total surface charge density (σ) on the carbon wall. The two components of σ are the charge due to the deprotonated surface carboxylic groups and the charge stored in the EDL related to the potential of the carbon layer. With no K3Fe(CN)6 added to solution, the negative open-circuit potential (EOCP) of the CNP (−280 mV vs Ag/AgCl; Figure S3B) reflects the high negative charge density on the carbon surface. As the Fe(CN)63− concentration increases, the OCP becomes less negative, gradually shifting toward the formal potential of Fe(CN)63−/4− (E0′ = 150 mV vs Ag/AgCl). The change in σ caused by the addition of a redox mediator to the solution can be evaluated using a simplified double-layer capacitor model as Δσ = ΔEOCPεrε0/d, where ΔEOCP is the change in OCP, ε0 is the vacuum permittivity, and εr and d are the dielectric constant and thickness of the Helmholtz layer, respectively. Thus, the Δσ can be related to the change in the OCP. In Figure S3B, ΔEOCP ≈ +100 mV for the change in Fe(CN)63− concentration from 100 nM (red curve in Figure 4A) to 300 nM (green curve). Assuming εr = 6 and d = 0.6 nm,40 the corresponding Δσ = +8 mC/m2. This number is in excellent agreement with the simulations in Figure 4A, where σ = −10 and −2 mC/m2 were used to fit the red and green i−V curves, corresponding to Δσ = +8 mC/m2. One should notice that only the portions of i−V curves corresponding to the high-conductance state are fitted to the theory in Figure 4.
Figure 3. i−V curves recorded at a 130 nm CNP in 10 mM KCl solution containing different concentrations of CaCl2. v = 0.1 V/s. [CaCl2], nM: 0 (black), 100 (red), 200 (blue), and 400 (green).
inner wall can be seen in the i−V curves. The ion current in the high-conductance state started to decrease when the concentration of Ca2+ reached ∼100−150 nM. This can be attributed to the Ca2+ binding to the negatively charged (−COO−) groups on the carbon surface and the resultant decrease in the net surface charge. By contrast, the ion current through CNP was essentially independent of citrate anion concentration in solution (up to ∼600 nM; Figure S2). The apparent insensitivity of the ion current to sub-μM concentrations of multivalent anions allows the effect of anionic redox species on the ionic transport in CNPs to be probed without worrying about their adsorption. The i−V responses of a 50-nm-radius CNP in 10 mM KCl containing different concentrations of K3Fe(CN)6 are shown in Figure 4A. The strong ICR can be seen in KCl (black curve; the rectification factor at ±0.5 V is ∼10, i.e., higher than that in Figure 2B, as expected for a smaller CNP radius). When the increasing concentration of K3Fe(CN)6 was added to the solution (from 100 nM, red curve, to 300 nM, green), the ion current in the high conductance state decreased, while the low conductance state current remained largely unchanged. Because the concentrations 11655
DOI: 10.1021/jacs.7b05058 J. Am. Chem. Soc. 2017, 139, 11654−11657
Communication
Journal of the American Chemical Society Modeling the low-conductance response is more complicated (see Figure S4 and related discussion in Supporting Information). The ET gated ionic transport was observed with different CNPs, redox species, and solvents (some additional data are shown in Figure S5). Figure S6 shows the effect of concentration of neutral redox species, i.e., aqueous ferrocenemethanol and ferrocene in acetonitrile, on the shape of the i−V curves. The formal potentials of both mediators are slightly more positive than that of Fe(CN)63−/4−, and the ion current in the high conductance state decreased with increasing concentration of each redox species due to the OCP shift toward more positive values. Both redox species exerted a stronger effect on the ion current than Fe(CN)63−. For instance, a significant decrease in the ICR can be seen at ferrocene concentration as low as 200 pM. The difference is likely to be due to the faster ET rate constants of both ferrocene mediators. The addition of Ru(NH3)63+ redox species, whose formal potential (−160 mV vs Ag/AgCl) is much more negative than that of ferro/ferricyanide, produced an opposite effect on the OCR and ion current through the CNP (Figure 4B). The degree of the ICR increased with the Ru(NH3)6Cl3 concentration, corresponding to the OCP shift toward more negative values (Figure S3C) and more negative σ. The σ values obtained from the fit agree reasonably well with the changes in EOCP. For instance, the σ values at 0 and 100 pM Ru(NH3)63+ are −5 and −10 mC/m2 (i.e., Δσ = −5 mC/m2), while the Δσ value calculated from the ΔEOCP is about −10 mC/m2. The ion current in the high-conductance state increases significantly in response to pM-range concentrations of Ru(NH3)6Cl3. This redox gating effect is opposite to that produced by adsorption of multivalent cations (cf. Figure 3, where the current in the high-conductance state and the degree of ICR decreased with increasing concentration of Ca2+). However, when at high concentrations of Ru(NH3)63+ (e.g., 100 nM; magenta curve in Figure 4B), the effect of the ionic adsorption becomes significant and the rectification factor stops increasing with [Ru(NH3)6Cl3]. Two parallel pathways for the current flow through a CNP with the floating carbon electrode (Figure 1) are the ion current through the orifice and the electronic current through the conductive carbon layer. The results of finite-element simulations (COMSOL Multiphysics 5.2a; see Supporting Information for the model and simulation details) of the ionic transport coupled with the bipolar electrochemistry in the CNP are shown in Figure 5. The 0.5 V voltage applied between the external and internal reference electrodes produced the potential profile inside a 50 nm radius CNP (Figure 5A). Due to the conical CNP geometry, the bias potential drops mainly within the ∼2 μm long portion of the pipet shaft adjacent to the orifice, which also is the most resistive (or transport limiting) zone. The potential drop in the carbon layer is negligible because of its high electric conductivity (∼103 S/m)19 and sub-nA electronic current. The difference between the carbon potential (Ec) and the local solution potential at the carbon surface (Esol) provides the driving force for the local ET reaction, η = Ec − Esol − E0′. The distributions of the η and ET current density (j) along the carbon layer surface are shown in Figure 5B. With the applied voltage dropping mostly near the CNP tip, η changes sharply within ∼2 μm distance from the pipet orifice, and j is very high in that region. Following the Butler−Volmer equation, j = 0 at the same point where η = 0 V and where the local ET current changes its polarity. The bipolar electrochemistry at the carbon surface
Figure 5. Finite-element simulations of ion current and bipolar ET in a CNP. (A) Simulated 2D potential distribution near the pore orifice inside the CNP. The arrows indicate the electric field direction. (B) Overpotential and ET current density distributions at the carbon surface. The pore orifice is at z = 0. The applied voltage is +0.1 V. (C) Experimental (solid line) and simulated (black open circles) i−V curves for [K3Fe(CN)6] = 10 μM. Red and blue open circles represent electronic and ionic components of the simulated CNP current, respectively.
mostly occurs near the CNP orifice, and the resulting electronic current is mainly determined by the applied voltage and the mass transport processes inside the nanopipet. The higher the concentration of the redox mediator in solution, the more significant the contribution of the electronic current to the total current through the CNP. An essentially sigmoidal experimental i−V curve reminiscent of a steady-state ET voltammogram (solid black line in Figure 5C; see also red curve in Figure S7) was recorded with [K3Fe(CN)6] = 10 μM and fitted to the theory (open black circles). The two components of the total simulated current, i.e., the ion current (open blue circles) and electronic current (open red circles), are also shown in Figure 5C. Unlike the ET current at the carbon solution interface (Figure 5B), whose total value over the CNP surface in the bipolar regime must be zero, the nonzero electronic current along the carbon layer is driven by the applied voltage (red arrow in Figure 5A). The direction of this current corresponds to the voltage polarity (red curve in Figure 5C), and it equals zero at 0 V, regardless of the E0′ of the redox mediator. The sub-nA limiting currents are controlled by the ET reaction occurring at the CNP tip (i.e., reduction at positive voltages and oxidation at negative voltages). The different 11656
DOI: 10.1021/jacs.7b05058 J. Am. Chem. Soc. 2017, 139, 11654−11657
Communication
Journal of the American Chemical Society
(6) Wang, Y.; Kececi, K.; Mirkin, M. V.; Mani, V.; Sardesai, N.; Rusling, J. F. Chem. Sci. 2013, 4, 655. (7) Howorka, S.; Siwy, Z. Chem. Soc. Rev. 2009, 38, 2360. (8) Cheng, L. J.; Guo, L. J. Chem. Soc. Rev. 2010, 39, 923. (9) Sun, L.; Crooks, R. M. J. Am. Chem. Soc. 2000, 122, 12340. (10) Guo, W.; Cao, L. X.; Xia, J. C.; Nie, F. Q.; Ma, W.; Xue, J. M.; Song, Y. L.; Zhu, D. B.; Wang, Y. G.; Jiang, L. Adv. Funct. Mater. 2010, 20, 1339. (11) Wang, G. L.; Zhang, B.; Wayment, J. R.; Harris, J. M.; White, H. S. J. Am. Chem. Soc. 2006, 128, 7679. (12) Vlassiouk, I.; Kozel, T. R.; Siwy, Z. S. J. Am. Chem. Soc. 2009, 131, 8211. (13) Liu, S. J.; Dong, Y. T.; Zhao, W. B.; Xie, X.; Ji, T. R.; Yin, X. H.; Liu, Y.; Liang, Z. W.; Momotenko, D.; Liang, D. H.; Girault, H. H.; Shao, Y. H. Anal. Chem. 2012, 84, 5565. (14) Hou, X.; Guo, W.; Jiang, L. Chem. Soc. Rev. 2011, 40, 2385. (15) Xu, X. L.; He, H. L.; Jin, Y. D. Anal. Chem. 2015, 87, 3216. (16) Freedman, K. J.; Otto, L. M.; Ivanov, A. P.; Barik, A.; Oh, S. H.; Edel, J. B. Nat. Commun. 2016, 7, 10217. (17) Yang, C.; Hinkle, P.; Menestrina, J.; Vlassiouk, I. V.; Siwy, Z. S. J. Phys. Chem. Lett. 2016, 7, 4152. (18) Hu, K. K.; Wang, Y. X.; Cai, H. J.; Mirkin, M. V.; Gao, Y.; Friedman, G.; Gogotsi, Y. Anal. Chem. 2014, 86, 8897. (19) Singhal, R.; Bhattacharyya, S.; Orynbayeva, Z.; Vitol, E.; Friedman, G.; Gogotsi, Y. Nanotechnology 2010, 21, 015304. (20) Perez-Mitta, G.; Marmisolle, W. A.; Trautmann, C.; ToimilMolares, M. E.; Azzaroni, O. J. Am. Chem. Soc. 2015, 137, 15382. (21) Nam, S. W.; Rooks, M. J.; Kim, K. B.; Rossnagel, S. M. Nano Lett. 2009, 9, 2044. (22) Munzer, A. M.; Michael, Z. R.; Star, A. ACS Nano 2013, 7, 7448. (23) Schneider, G. F.; Kowalczyk, S. W.; Calado, V. E.; Pandraud, G.; Zandbergen, H. W.; Vandersypen, L. M. K.; Dekker, C. Nano Lett. 2010, 10, 3163. (24) Garaj, S.; Hubbard, W.; Reina, A.; Kong, J.; Branton, D.; Golovchenko, J. A. Nature 2010, 467, 190. (25) Liu, S.; Zhao, Y.; Parks, J. W.; Deamer, D. W.; Hawkins, A. R.; Schmidt, H. Nano Lett. 2014, 14, 4816. (26) Banerjee, S.; Shim, J.; Rivera, J.; Jin, X. Z.; Estrada, D.; Solovyeva, V.; You, X.; Pak, J.; Pop, E.; Aluru, N.; Bashir, R. ACS Nano 2013, 7, 834. (27) Rutkowska, A.; Freedman, K.; Skalkowska, J.; Kim, M. J.; Edel, J. B.; Albrecht, T. Anal. Chem. 2015, 87, 2337. (28) Kim, B. M.; Murray, T.; Bau, H. H. Nanotechnology 2005, 16, 1317. (29) Vitol, E. A.; Schrlau, M. G.; Bhattacharyya, S.; Ducheyne, P.; Bau, H. H.; Friedman, G.; Gogotsi, Y. Chem. Vap. Deposition 2009, 15, 204. (30) Fosdick, S. E.; Knust, K. N.; Scida, K.; Crooks, R. M. Angew. Chem., Int. Ed. 2013, 52, 10438. (31) Wei, C.; Bard, A. J.; Feldberg, S. W. Anal. Chem. 1997, 69, 4627. (32) Zangle, T. A.; Mani, A.; Santiago, J. G. Chem. Soc. Rev. 2010, 39, 1014. (33) Wang, D. C.; Liu, J.; Kvetny, M.; Li, Y.; Brown, W.; Wang, G. L. Chem. Sci. 2014, 5, 1827. (34) Guerrette, J. P.; Zhang, B. J. Am. Chem. Soc. 2010, 132, 17088. (35) Momotenko, D.; Cortes-Salazar, F.; Josserand, J.; Liu, S. J.; Shao, Y. H.; Girault, H. H. Phys. Chem. Chem. Phys. 2011, 13, 5430. (36) Wang, D. C.; Kvetny, M.; Liu, J.; Brown, W.; Li, Y.; Wang, G. L. J. Am. Chem. Soc. 2012, 134, 3651. (37) Ali, M.; Nasir, S.; Ramirez, P.; Cervera, J.; Mafe, S.; Ensinger, W. ACS Nano 2012, 6, 9247. (38) He, Y.; Gillespie, D.; Boda, D.; Vlassiouk, I.; Eisenberg, R. S.; Siwy, Z. S. J. Am. Chem. Soc. 2009, 131, 5194. (39) Besteman, K.; Zevenbergen, M. A. G.; Heering, H. A.; Lemay, S. G. Phys. Rev. Lett. 2004, 93, 170802. (40) Xiong, J. W.; Chen, Q. J.; Edwards, M. A.; White, H. S. ACS Nano 2015, 9, 8520.
positive and negative plateau heights are closely related to the ICR (blue curve). Despite a reasonably good fit between the simulated and experimental i−V curves in Figure 5C, the presented theoretical description is only semiquantitative because the employed double-layer model is not exact. Simulated carbon layer and solution potentials, current density and ion concentration profiles at the carbon surface at different applied voltages are shown in Figure S8. One should notice that the electronic current in Figure 5C is on the pA scale at [K3Fe(CN)6] as high as 10 μM. With the nM (or pM in Figures 4 and S5) concentrations of redox species, this current should be immeasurably low. However, by changing the OCP, the redox species affect the much larger ion current through the CNP. This amplification effect enables the detection of very low concentrations of redox species. In conclusion, our CNP experiments and simulations revealed a new phenomenon: the ET gated ion transport in conductive nanopores. The open-circuit potential of the carbon layer can be controlled by low (e.g., pM) concentrations of redox species in the solution via bipolar electrochemistry. The redox mediators with positive E0 tend to decrease the negative charge density on the carbon surface, thus, decreasing the ion current in the highconductance state and the degree of the ICR, while the redox species with a negative E0 produce the opposite effect. In addition to rectification sensing of redox analytes, these findings suggest the possibility of redox control of the ICR in conductive nanopores that can be useful for resistive-pulse sensing. The improved understanding of the bipolar electrochemistry on the nanoscale is important for various applications such as electrodeposition, corrosion, and separations in nanofluidic devices.
■
ASSOCIATED CONTENT
* Supporting Information S
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b05058. Experimental and simulation details, multivalent anion adsorption effect, OCP measurements, and i−V responses with other redox mediators, including Scheme 1 and Figures S1−S8 (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*
[email protected] ORCID
Michael V. Mirkin: 0000-0002-3424-5810 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The support of this work by the National Science Foundation (CHE-1300158) is gratefully acknowledged. We thank Keke Hu for obtaining a TEM image (Figure S1A).
■
REFERENCES
(1) Lan, W. J.; Edwards, M. A.; Luo, L.; Perera, R. T.; Wu, X. J.; Martin, C. R.; White, H. S. Acc. Chem. Res. 2016, 49, 2605. (2) Schoch, R. B.; Han, J. Y.; Renaud, P. Rev. Mod. Phys. 2008, 80, 839. (3) Bayley, H.; Martin, C. R. Chem. Rev. 2000, 100, 2575. (4) Stoloff, D. H.; Wanunu, M. Curr. Opin. Biotechnol. 2013, 24, 699. (5) Morris, C. A.; Friedman, A. K.; Baker, L. A. Analyst 2010, 135, 2190. 11657
DOI: 10.1021/jacs.7b05058 J. Am. Chem. Soc. 2017, 139, 11654−11657
