- Email: [email protected]
Selective ion transport in functionalized carbon nanotubes
Applied Surface Science 423 (2017) 154–159
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
Selective ion transport in functionalized carbon nanotubes Olga N. Samoylova a , Emvia I. Calixte b , Kevin L. Shuford b,∗ a b
Department of Biochemistry and Molecular Biology, University of Texas Medical Branch, 301 University Blvd, Galveston, TX 77555-0304, United States Department of Chemistry and Biochemistry, Baylor University, One Bear Place #97348, Waco, TX 76798-7348, United States
a r t i c l e
i n f o
Article history: Received 24 February 2017 Received in revised form 16 May 2017 Accepted 10 June 2017 Available online 15 June 2017 Keywords: Modified carbon nanotube Selectivity Ion current Molecular dynamics
a b s t r a c t Ion transport through functionalized carbon nanotubes in an external electric field is studied using all atom molecular dynamics simulations. The surface of carbon nanotubes has been functionalized with hydrogens and hydroxyl groups, and ionic current passing through the nanochannels has been examined with respect to the extent of surface modification. We are able to dramatically increase the ionic current passing through the nanotube via the appropriate surface modification. An analysis of the electrostatic potential within the tube shows higher ionic currents result from an increase in accessible pathways coupled with a global shift toward more direct ion passage. Moreover, through judicious choice of structure, the current can be modulated to a large degree with ion selectivity. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Structure and dynamics of confined fluids is drastically different from that of bulk due to large internal surfaces and geometric restriction. Variations in the local environment can lead to altered physical and chemical properties, and in some cases yield exotic system behavior. Examples include modified fluid structure, different thermodynamic, kinetic and correlation properties, size dependence and shift of critical parameters like temperature, pressure, and density, as well as spatial limitation and dimensional crossover effects on the dynamic properties such as diffusion and relaxation [1–5]. In our previous investigations of brines under confinement, we observed significant changes in fluid properties as the encapsulating pores decreased in size [6–9]. Specifically, we found reduced water densities, increased water and ion structuring at interfaces, reduced ion hydration, and disruption to the overall hydrogen bonded network. The increased hydrogen bond lifetimes coupled with higher hydrogen bond activation energies were due to unfavorable reorientation of water molecules as pore diameter decreased [9]. There is great interest in elucidating the characteristics of confined fluids and further developing systems that can actively control these properties on the nanoscale. Carbon nanotubes (CNTs) are promising candidates for such novel technologies due to their unique mechanical, electrical, and optical properties [10–12]. In particular, applications that incorporate nanotubes (sensing,
∗ Corresponding author. http://dx.doi.org/10.1016/j.apsusc.2017.06.120 0169-4332/© 2017 Elsevier B.V. All rights reserved.
nanofluidics, electrochemistry, etc.) may benefit greatly from controllable ion transport coupled to favorable surface interactions that can be manipulated in some way [13–18]. For example, the development of water purification technologies using CNTs is very attractive owing to their chemical and mechanical properties, low mass, large surface area, high aspect ratio, low cost and low impact on the environment [19]. These properties along with high flux and potential for selectivity make them highly favorable for next generation membrane technology [20,21]. However, pristine nanotubes tend to aggregate and fouling can also limit desalination. Surface modifications to CNT sidewalls can address these limitations and may also improve adsorption of contaminants and intensify water flux [19]. Other studies on ion entry have reported routes for selectivity based upon judicious choice and placement of functional groups [22,23]. More generally, it is expected that uncapped CNTs are inherently tip modified, and investigations into different forms of functionalization are highly relevant [24]. It is thus important to fully understand fluid transport through CNTs as you reduce tube size, and in turn how the surface chemical properties affect this. In a previous molecular dynamics (MD) study, we examined ion and solvent transport through CNTs of different diameters in an external electric field [25]. For large diameter CNTs, ionic current was essentially linear with voltage. However, deviations from this behavior emerged as the CNT diameter decreased, until eventually no ions were able to pass. The aim of the current study is to increase the ionic current through small diameter CNTs by modifying the tube surface with different functional groups. The results are meant to emphasize the effects of surface chemistry and demonstrate how these interactions can be utilized to control transport, especially
O.N. Samoylova et al. / Applied Surface Science 423 (2017) 154–159
155
Table 1 Carbon nanotube systems of interest. Columns are labeled by CNT chiral vector indices (n, m), CNT radius, the number of carbon atoms in the CNT, and the number of ions and water molecules, respectively. The tube length is 40 A˚ in all cases. (n, m)
Radius, A˚
Carbon
Ions
Water
(6,6) (7,7) (8,8) (9,9)
4.09 4.76 5.44 6.12
480 559 640 720
36 34 34 32
932 898 857 839
Table 2 CNT modification by rings with the numbers of H/OH functionalizations bound to carbon atoms. Columns are labeled by number of rings and CNT type.
Fig. 1. Schematic representation of CNT modification from the outside or inside (left) and by rings (right). Spheres denote functional groups (OH or H atoms) bound to carbon atoms of the CNT. Green/blue rings represent odd/even number of modification rings (for visualization purposes only). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
in small CNTs that do not readily permit ion passage otherwise. We find that by altering the conditions at the interface, the ionic current can be increased significantly. Our analysis also provides insight into the mechanism and underlying principles driving this observation, which can be leveraged to design fluid transport vessels with enhanced properties. Through surface chemistry, we are able to modulate the current and provide a degree of ion selectivity not previously present in pristine CNTs of similar size. This yields a chemically intuitive approach toward controlling fluid transport properties within CNTs via existing methods of synthesis. Although not the focus here, we note there is a vast literature describing synthetic routes to functionalize CNTs. Most relevant for the theoretical studies below are those that result in sidewall addition of H or OH groups, which have been realized experimentally [26–32]. 2. Theoretical methods 2.1. Model systems We have examined armchair CNTs with chiral indices ranging from (6,6) to (9,9) that are 40 A˚ in length. Our previous studies found NaCl transport through unfunctionalized (9,9) CNTs to be substantially altered by confinement effects [25], thus CNTs of this size were specifically targeted in this work. In particular, we wanted to induce significant ionic transport in systems that did not readily allow ions to pass previously, while providing some degree of selectivity. CNTs were constructed using the tcl code [33] with the C C bond equilibrium distance set to 1.42 A˚ and an equilibrium angle between carbon atoms of 0 = 120◦ . To model functional groups on the surface, pristine CNTs were modified by adding H or OH groups to selected carbon atoms. Experimental routes of sidewall addition lead to varying degrees of functionalization that are not saturated but can vary greatly (1–10%, perhaps more) [34]. To account for this broad range and extrapolate to levels beyond, we varied the extent of functionalization by modifying the CNT incrementally, one “ring” at a time starting from the top. We denote a ring as a series of connected benzene-like molecules that roll up to encompass a segment of the overall CNT length; stacking all the rings forms the tube. For a (n, n) CNT, each ring has n hexagons of carbon atoms connected in a circle. Our model for the modified CNT adds functional groups to 2n carbon atoms composing the ring – every second carbon atom along the top of a hexagon cell and every second carbon along the bottom of each hexagon cell, staggered from the top. Fig. 1 shows schematically the method of CNT modification by rings. For example, a (9,9)
Rings
(6,6)
(7,7)
(8,8)
(9,9)
1 2 . . . 10
12 24 . . . 120
14 28 . . . 140
16 32 . . . 160
18 36 . . . 180
Table 3 Bond parameters. Atom pair
kstretch , kcal/mol A˚ 2
r0 , A˚
C C C O
305.0 340.0 334.3 545.0
1.375 1.083 1.411 0.96
C H O H
CNT with 1 ring of modified carbon atoms has 18 functional groups of atoms bound to certain carbons. The H or OH groups were added to the outside of the CNT, leading to different configurations for each CNT diameter examined. Note that functionalizing the CNT alters the charge distribution, which is represented in the simulations by a change in force field parameters at those sites (see Section 2.2). All modifications of pristine CNTs were performed using Avogadro [35], followed by an initial geometry optimization using MMFF94 force field [36]. The functionalized CNTs were then solvated with NaCl solution. Details about the model systems – the number of carbon atoms, OH/H groups added, sodium and chloride ions, and water molecules – can be found in Tables 1 and 2. We have chosen to utilize a rather idealized model throughout these studies. The goal was not to represent a particular experimental system, but rather to explore the range of effects that could be achieved by introducing new interactions in a controllable manner. This approach allows us to glean insight from the model systems, which can then be used to develop a fundamental understanding of complex transport processes and routes to control them. Model variations that probe the vast complexities present in real experimental systems are under development currently. 2.2. Simulation approach We used all atom molecular dynamics (MD) to simulate ion transport through CNTs of various radii in the presence of an electric field. Each tube was solvated by ∼1 M NaCl solution in an ˚ orthorhombic simulation box with dimensions 24 A˚ ×24 A˚ ×70 A. TIP3P water molecules [37] have been used for the simulations. Carbon atoms of the tube were harmonically restrained. The tube was aligned along the z-axis, corresponding to the long axis of the hexagonal system. All MD simulations have been performed using the program NAMD2.9 [38] with the CHARMM36 force field [39]. Force field parameters are shown in Tables 3–6. The system was minimized and then equilibrated for 5 ns at 300 K using the NPT ensemble and a 1 fs time step. Periodic boundary conditions were applied in 3 dimensions, and the Particle
156
O.N. Samoylova et al. / Applied Surface Science 423 (2017) 154–159
Table 4 Angle parameters. Atom triplet C C C C
C C C O
C H O H
kangle , kcal/mol rad2
0 , ◦
40.0 29.0 45.2 65.0
120 120 120 108
Table 5 Dihedral parameters. Atom quadruplet
kdihedral , kcal/mol
0 , ◦
C C C C
3.10 3.50 3.10 0.99
180 180 180 180
C C C C
C C C O
C H O H
Table 6 Non-bonded interaction parameters and atom charges. Atom pair
ij , kcal/mol
ij , A˚
Pair charges
C C C O
0.070 0.103 0.039 0.084
3.98 3.76 3.31 1.99
0, 0 −0.09, 0.09 0.11, −0.53 −0.53, 0.42
C H O H
Fig. 3. Ionic current in the (9,9) CNT modified externally with H atoms for direct(+z) and reverse(−z) electric field, red and black traces respectively. Also shown are the levels of current passing through unmodified CNTs. Na ions are current carriers in these systems. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
type/location of functionalization, and direction of the applied field. For example, (9,9)-HoutRev denotes a (9,9) CNT modified with H atoms on the tube exterior biased with a 1 V electric field in the −z direction. Production MD runs of 50 ns including the biasing electrical field were performed on every system. A Lowe-Anderson thermostat was used to control the temperature at 300 K, and the extended system pressure method was used to maintain a constant pressure of 1 atm. Current was computed by counting every ion inside the carbon nanotube passing the CNT center of mass [25], given as
I t+
t 2
1 qi i (t + t) − i (t) tl N
=
(1)
i
where l = ztop − zbot is a nanotube length, ztop and zbot are the zcoordinates of the top and bottom position of the CNT, N is the number of ions, qi is an ion charge, zi is the z coordinate of the ion, t is the time between trajectory frames, and
i (t) =
⎧ z (t), if zbot ≤ zi (t) ≤ ztop ⎪ ⎨ i ⎪ ⎩
zbot ,
if zi (t) < zbot
ztop ,
if zi (t) > ztop
(2)
The first 2 ns of the trajectory were not included when calculating the mean value of the current to allow the system time to reach the steady state regime. 3. Results and discussion
Fig. 2. VMD screenshot of the simulated system containing a carbon nanotube modified externally with hydroxyl groups and solvated in water with chloride (cyan) and sodium (yellow) ions. Water is rendered here with a space filling representation. The direct electric field is applied along the positive z direction of the CNT, and the reverse electric field corresponds to negative z direction. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
Mesh Ewald summation was used with a grid size of 1 A˚ for longrange electrostatic interactions. The van der Waals interactions were calculated with a 12 A˚ cut-off. The system was biased by an external electric field in the z-direction (direct field) and −zdirection (reverse field), with voltage drops of 1 V (e.g. see Fig. 2). Throughout the manuscript, results will be labeled by the CNT size,
We computed the ionic current migrating through surfacemodified CNTs with sizes ranging from (6,6) to (9,9). The smallest ones, (6,6) and (7,7), did not register any statistically significant current. H-functionalized (8,8) CNTs on the exterior display a slightly increased ionic current compared to larger, pristine tubes (see Supplementary material); however, the most dramatic effect from our studies is observed for the (9,9) CNT with H atoms bound to the exterior surface. As a result, functionalized (9,9) CNT systems will be the primary focus of the paper, initially with H atoms on the exterior and then hydroxyls. Fig. 3 shows the ionic current driven through functionalized (9,9) CNTs with H atoms added to the outside of the tube surface. Adding H atoms to a single ring increases the current 5–20 fold with respect to an unmodified CNT of the same size, whereas a fully decorated CNT (10 rings) generates relative current changes of
O.N. Samoylova et al. / Applied Surface Science 423 (2017) 154–159
approximately 70–95. The latter represents a substantial increase in the number of ions passing through the CNT when H atoms uniformly cover the exterior surface. Amazingly, we find it is not only possible to increase the current compared to the pristine tube of the ˚ but to reach ionic same size (e.g., (9,9) CNT radius is about 6.12 A), current levels similar to an unmodified (12,12) CNT with radius ˚ In this case, the altered conditions at the interface effec8.15 A. tively increase the CNT diameter by ∼4 A˚ from the perspective of current passed. As shown in Fig. 3, current trends behave differently depending upon the direction of the applied field. For this system, hydrogen atoms are bound to carbon on the external surface. The negatively charged chloride ions are concentrated in the bulk water, which is largely due to the external CNT modification with hydrogens as well as the anions larger size. Therefore, the current carriers through the (9,9)-Hout CNT are almost exclusively sodium ions. For the reverse field case (electric field in the −z direction, black trace), cations enter the tube from the top and move down. The functionalized CNTs were created by adding rings of modified carbon atoms bound with hydrogens starting at the top. So for the case where the sodium ion is driven from the top, there is a more favorable Coulomb interaction with the carbon atoms that have been functionalized with hydrogen due to bond polarity. The partially negative carbons attract sodium ions into the tube by decreasing the energetic barrier for ions to leave bulk water and move into the confined area within the tube. As shown in Fig. 3, the ionic current through the (9,9)-Hout modified tube with 1 ring is already much higher for the reverse field than for the pristine (9,9) CNT (green line). For the direct field case (+z direction, red trace), sodium ions will enter the tube from the bottom and migrate up, corresponding to the electric field direction. The situation is not much different from an unmodified CNT if there are only a few functionalized carbon rings on the top of the tube. The ions only weakly feel the electrostatic attraction from the functionalized carbons until the modified tube section is closer to the entrance point (achieved within our model by adding more rings). When sodium cations can more easily interact with the functionalized regions, the additional Coulomb forces facilitate ion passage. As a result, there is a significant increase in the total current as more functionalized rings are included starting from 7. For the fully modified tube, the values for reverse and direct current are close but not equivalent due to slight variations in the structure. Note the CNT has some asymmetry in the position of functionalized carbons. The first ring (top) has modified carbons at the edge of CNT, whereas the last ring of carbons (bottom) has unmodified carbons at the edge (see Fig. 1). The interactions between ions and these critical regions of the structure is the reason for the difference in reverse and direct currents for fully modified CNTs. To glean insight into current trends, we studied ion passage through the modified (9,9)-Hout tubes with 1, 5, and 10 rings. Different drift regimes within these tubes were identified by tracking the time dependence of z-position of any ion inside the tube. We only included events where the z-coordinate of the ion changed sign and then eventually traversed the CNT interior (i.e., a translocation event). We did not incorporate times when an ion changed its location from the bulk water to the tube and back multiple times before an actual translocation initiated. Events were ordered by ascending passage time and separated into small time bins. Fast, moderate, and slow regimes were then identified arbitrarily, based upon passage time. Translocation events that occurred in less than 0.4 ns were designated as fast, while events over ∼0.8 ns were considered slow. Everything in between these extremes was denoted moderate. Fig. 4 displays the mean value of z (averaged over all events in the particular bin) as a function of time for different
157
Fig. 4. Ion drift regimes and passage time for the modified (9,9)-Hout tubes with the reverse electric field applied. Panels A–C correspond to 1, 5, and 10 rings of functionalized carbons. The axial coordinate, z, is averaged over all translocation events. (For interpretation of the references to color in the text, the reader is referred to the web version of the article.)
regimes. There is a quasi-linear trend for ion translocation in the tube with one ring of modified carbon atoms (Panel A); 61% of ions drift with the moderate pace, and 39% in the fast regime. In the case with 5 modification rings (half of the CNT, Panel B), we see a similar rapid passage through the functionalized part of the tube, more relaxed motion in the pristine part of CNT, and an approximately equal number of ions in the fast and moderate regime. However, a deviation from this behavior emerges, where some rare events (5%) of ion trapping at the boundary of the modified/unmodified region occur. These ions fluctuate around the middle of the tube at z = 0
158
O.N. Samoylova et al. / Applied Surface Science 423 (2017) 154–159
Fig. 5. Electrostatic potential (in kT/e units = 0.0258 V for T = 300 K) for the (9,9)Hout CNT with a reverse applied electric field. Panels A–C correspond to CNTs with 1, 5, and 10 modified carbon atoms respectively.
before passing, which leads to the slow translocation regime. For the completely modified tube (10 rings, Panel C), the vast majority (78%) of ions traverse rapidly through the tube within 0.4 ns. 21% of events are in moderate regime, and only 1% of ions are in slow regime. An important factor that affects the ion passage is position inside the CNT. This occurs because drastically different Coulomb interactions are present between carbon sites and the ion based upon internal location. To illustrate this, we provide electrostatic potential maps (Fig. 5) using the VMD PMEpot plugin [40]. Note the product of charge times electrostatic potential can be interpreted as the potential energy of the charge in an electrostatic field. From these maps one can infer different potentials are present depending upon sodium ion’s radial and z position within the CNT, which can be used to understand the translocation times in Fig. 4. For 1 ring (Fig. 5A), there is a clear path through the middle (due to the circular symmetry in the electric field in XY plane) leading to fast, linear transport. Any cation near the tube edge is pushed to the center, potentially bouncing back and forth, yielding approximately linear transport over slightly longer times. The short modified part of the tube has a small influence on the ion motion only at the beginning of the passage. Beyond that point, ion transport would be similar to being in a pristine CNT under the influence of an external electric field. The electrostatic environment changes as more charged rings are added to the tube. Fig. 5B shows the cation feels a strong attraction from the functionalized part of the tube initially and then experiences a very different environment moving through the CNT in the external electric field. There is a negative potential inside the top portion of the tube with 5 modified rings, which becomes neutral in the central region before turning positive. Sodium ions readily enter the CNT from the top and easily migrate half-way through before experiencing the positive potential barrier. Ions that do not traverse exclusively via the center of the CNT feel more resistance, but can begrudgingly migrate toward the exit under the influence of the applied field. This scenario lends itself to larger radial force components, causing more ion motion in the XY plane of the tube and delaying translocation along z. We can see oscillations and more spread in axial position (green line, Fig. 4B) that emerge from the white neutral zone in the tube, which persist until the ion exits the CNT. For 10 rings (Fig. 5C), we see a similar electrostatic potential pattern inside the tube; however, the positive barrier is shifted to the CNT exit and considerably smaller. The vast majority of sodium ions can easily traverse the barrier rapidly, and 78% of translocation events are fast. So we see from this analysis that external H modification of CNTs leads to a negative electrostatic environment inside, which facilitates cation transport by increasing accessible pathways through the CNT and inducing a shift toward faster ion passage from the other drift regimes. Analogously, anion passage can be enhanced via the appropriate surface functionalization. While the absolute current values are
Fig. 6. Ionic current in a (9,9) CNT externally modified with hydroxyl groups for direct(+z) and reverse(−z) electric field, red and black traces respectively. Also shown are the levels of current through unmodified CNTs. Chloride ions are the current carriers in these systems. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
significantly less, due to the larger size of chloride, the relative gains compared to pristine CNTs are noteworthy, and this provides a route toward selectivity of ion transport. Functionalizing the CNT exterior with hydroxyl groups creates an environment inside that favors chloride, while the partially negative oxygen atoms draw sodium to the CNT exterior. Thus, the ionic current carriers through the nanotube are Cl− , and they are moving in the direction opposite to an applied external electric field. So for the direct electric field chloride ions are entering the top (i.e., where the modified rings start), and for the reverse field chloride is moving in from the bottom. As a result, the direct ionic current for the partially modified tube is greater than the reverse field case (see Fig. 6). Note, the situation is exactly opposite to what was described previously for external functionalization with H atoms, substantiating that chloride is the carrier. The maximum relative current change for (9,9)-OHout is approximately 12 compared to the pristine (9,9) CNT. So while the currents are still quite small, we are able to pass some carriers through CNTs that essentially did not permit chloride transport before functionalization. More generally, we find it is possible to get ≈ 5 times more current through a (9,9) CNT using external surface modification with H atoms compared to OH groups. The primary differentiator is the size of the ion carrier (sodium for the former and chloride for the latter). To be complete, we briefly note results obtained for the cases of interior functionalization, even though such systems would be challenging to realize experimentally. Adding hydrogens to the interior favors chloride passage, creating an environment analogous to hydroxls on the exterior described above. The radius of ˚ We can estimate the effective the unmodified (9,9) CNT is 6.12 A. radius of available space using the bonds and angle parameters for the functional group. The internal radius for the modified (9,9)-Hin ˚ which is less than the radius of an unmodified (8,8) is about 5.04 A, ˚ but the ionic current can reach values greater CNT (R(8,8) = 5.44 A), than the current passed in an unmodified (9,9) CNT (see Supplementary material). The ionic current attained is comparable to the (9,9)-OHout CNT discussed above. Lastly, we did not register any current for hydroxyl modification on the inside of (9,9) CNTs. The internal radius of available cylindrical space for the modified (9.9)˚ which is less than the radius of an unmodified OHin is about 4.74 A, ˚ where we did not register any ionic per(7,7) CNT (R(7,7) = 4.76 A) meation events in a previous study [25]. In principle, both ions could have favorable interactions with the interior regions to some degree; however, the lack of available space for the solvent and ions to traverse the CNT is strongly limited in this case.
O.N. Samoylova et al. / Applied Surface Science 423 (2017) 154–159
4. Conclusions Our simulations show that ionic current through functionalized carbon nanotubes can be modulated by different functional groups bound to the carbon atoms. Using modified nanotubes, it is possible to not only dramatically increase ionic current through the tube, but also provide selectivity of the current carriers. We have shown that direct and reverse currents in the partially modified nanotubes are not identical, and there is a direct correlation between ionic current and the extent of surface modification in CNT. Thus it may be possible to actively manipulate the time ions spend inside a carbon nanotube (e.g., trap the ion inside and/or release it) by adjusting field polarity in modified tubes, which could be potentially useful for multiple applications like nanosensors, chemical reactions under confinement, and fluidics. Acknowledgements This work is supported by the Chemical Sciences, Geosciences, and Biosciences Division, Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy under Award Number DESC0010212. KLS thanks Baylor University for sabbatical funding. Appendix A. Supplementary data Supplementary data associated with can be found, in the online cle http://dx.doi.org/10.1016/j.apsusc.2017.06.120.
this artiversion, at
References [1] H.E. Stanley, Introduction to Phase Transition and Critical Phenomena, Clarendon Press, Oxford, 1971. [2] M.A. Anisimov, Critical Phenomena in Liquids and Liquid Crystals, Gordon & Breach, Philadelphia, 1991. [3] J.-C. Wang, K.A. Fichthorn, Influence of molecular structure on the properties of confined fluids by molecular dynamics simulation, Colloids Surf. A: Physicochem. Eng. Aspects 206 (1–3) (2002) 267–276. [4] D. Morineau, C. Alba-Simionesco, Liquids in confined geometry: how to connect changes in the structure factor to modifications of local order, J. Chem. Phys. 118 (2003) 9389. [5] A.B. Farimani, N.R. Aluru, Existence of multiple phases of water at nanotube interfaces, J. Phys. Chem. C 120 (2016) 23763–23771. [6] M.C.F. Wander, K.L. Shuford, Molecular dynamics study of interfacial confinement effects of aqueous NaCl brines in nanoporous carbon, J. Phys. Chem. C 114 (2010) 20539–20546. [7] M.C.F. Wander, K.L. Shuford, Electrolyte effects in a model system for mesoporous carbon electrodes, J. Phys. Chem. C 115 (2011) 4904–4908. [8] M.C.F. Wander, K.L. Shuford, Alkali halide interfacial behavior in a sequence of charged slit pores, J. Phys. Chem. C 115 (2011) 23610–23619. [9] E.I. Calixte, O.N. Samoylova, K.L. Shuford, Confinement and surface effects of aqueous solutions within charged carbon nanotubes, Phys. Chem. Chem. Phys. 18 (2016) 12204–12212. [10] M.D. Volder, S.H. Tawfick, R.H. Baughman, A.J. Hart, Carbon nanotubes: present and future commercial applications, Science 339 (2013) 535–539. [11] S. Iijima, T. Ichihashi, Single-shell carbon nanotubes of 1-nm diameter, Nature 363 (1993) 603–605. [12] J.N. Coleman, U. Khan, W.J. Blau, Y.K. Gunko, Small but strong: a review of the mechanical properties of carbon nanotube-polymer composites, Carbon 44 (9) (2006) 1624–1652. [13] J.J. Gooding, Nanostructuring electrodes with carbon nanotubes: a review on electrochemistry and applications for sensing, Electrochim. Acta 50 (2005) 3049–3060. [14] R. Martel, T. Schmidt, H. Shea, T. Hertel, P. Avouris, Single- and multi-wall carbon nanotube field-effect transistors, Appl. Phys. Lett. 73 (1998) 2247–2249.
159
[15] T. Chen, L. Dai, Carbon nanomaterials for high-performance supercapacitors, Mater. Today 16 (2013) 272–280. [16] J. Su, D. Huang, Coupling transport of water and ions through a carbon nanotube: the role of ionic conduction, J. Phys. Chem. C 120 (2016) 11245–11252. [17] B. Corry, Designing carbon nanotube membranes for efficient water desalination, J. Phys. Chem. B 112 (5) (2008) 1427–1434. [18] K. Sint, B. Wang, P. Král, Selective ion passage through functionalized graphene nanopores, J. Am. Chem. Soc. 130 (2008) 16448–16449. [19] R. Das, M.E. Ali, S. Hamid, S. Ramakrishna, Z.Z. Chowdhury, Carbon nanotube membranes for water purification: a bright future in water desalination, Desalination 336 (2014) 97–109. [20] S. Kar, R.C. Bindal, P.K. Tewari, Carbon nanotube membranes for desalination and water purification: challenges and opportunities, Nano Today 7 (2012) 385–389. [21] R. Das, S. Hamid, M.E. Ali, A.F. Ismail, M.S.M. Annuar, S. Ramakrishna, Multifunctional carbon nanotubes in water treatment: the present, past and future. desalination, Desalination 354 (2014) 160–179. [22] S. Joseph, R.J. Mashi, E. Jakobsson, N.R. Aluru, Electrolytic transport in modified carbon nanotubes, Nano Lett. 3 (10) (2003) 1399–1403. [23] M. Majumder, N. Chopra, B.J. Hinds, Effect of tip functionalization on transport through vertically oriented carbon nanotube membranes, J. Am. Chem. Soc. 127 (2005) 9062–9070. [24] P.S. Goh, A.F. Ismail, B.C. Ng, Carbon nanotubes for desalination: performance evaluation and current hurdles, Desalination 308 (2013) 2–14. [25] O.N. Samoylova, E.I. Calixte, K.L. Shuford, Molecular dynamics simulations of ion transport in carbon nanotube channels, J. Phys. Chem. C 119 (2015) 1659–1666. [26] A. Tagmatarchis, V. Georgakilas, M. Prato, H. Shinohara, Sidewall functionalization of single-walled carbon nanotubes through electrophilic addition, Chem. Commun. (2002) 2010–2011. [27] S. Santangelo, Controlled surface functionalization of carbon nanotubes by nitric acid vapors generated from sub-azeotropic solution, Surf. Interface Anal. 48 (2016) 17–25. [28] I. Gerber, M. Oubenali, R. Bacsa, J. Durand, A. Goncalves, M. Fernando, R. Pereira, F. Jolibois, L. Perrin, R. Poteau, P. Serp, Theoretical and experimental studies on the carbon-nanotube surface oxidation by nitric acid: interplay between functionalization and vacancy enlargement, Chem. Eur. J. 17 (2011) 11467–11477. [29] G. Miller, J. Kintigh, E. Kim, P. Weck, S. Berber, D. Tomanek, Hydrogenation of single-walled carbon nanotubes using polyamine reagents: combined experimental and theoretical study, J. Am. Chem. Soc. 130 (2008) 2296–2303. [30] A. Dillon, K. Jones, T. Bekkedahl, C. Kiang, D. Bethune, M. Heben, Storage of hydrogen in single-walled carbon nanotubes, Nature 386 (1997) 377–379. [31] Y.S. Nechaev, A. Yurum, A. Tekin, N.K. Yavuz, Y. Yurum, T.N. Veziroglu, Fundamental open questions on engineering of “super” hydrogen sorption in graphite nanofibers: relevance for clean energy applications, Am. J. Anal. Chem. 5 (2014) 1151–1165. [32] A. Fujiwara, K. Ishii, H. Suematsu, H. Kataura, Y. Maniwa, S. Suzuki, Y. Achiba, Gas adsorption in the inside and outside of single-walled carbon nanotubes, Chem. Phys. Lett. 336 (3–4) (2001) 205–211. [33] F.Q. Zhu, K. Schulten, Water and proton conduction through carbon nanotubes as models for biological channels, Biophys. J. 85 (2003) 236–244. [34] M. Burghard, K. Balasubramanian, Chemically functionalized carbon nanotubes, Small 1 (2005) 180–192. [35] M.D. Hanwell, D.E. Curtis, D.C. Lonie, T. Vandermeersch, E. Zurek, G.R. Hutchison, Avogadro: an advanced semantic chemical editor, visualization, and analysis platform, J. Chem. Inf. 4 (2012) 4–17, article number 17. [36] T.A. Halgren, Merck molecular force field. I. Basis, form, scope, parameterization, and performance of MMFF94, J. Comput. Chem. 17 (1996) 490–519. [37] W.L. Jorgensen, J. Chandrasekhar, J.D. Madura, R.W. Impey, M.L. Klein, Comparison of simple potential functions for simulating liquid water, J. Chem. Phys. 79 (1983) 926–935. [38] J.C. Phillips, R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa, C. Chipot, R.D. Skeel, L. Kale, K. Schulten, Scalable molecular dynamics with NAMD, J. Comput. Chem. 26 (2005) 1781–1802. [39] B.R. Brooks, C.L. Brooks, A.D. Mackerell, L. Nilsson, R.J. Petrella, B. Roux, Y. Won, G. Archontis, C. Bartels, S. Boresch, A. Caflisch, L. Caves, Q. Cui, A.R. Dinner, M. Feig, S. Fischer, J. Gao, M. Hodoscek, W. Im, K. Kucczera, T. Lazaridis, J. Ma, V. Ovchinnikov, E. Paci, R.W. Pastor, C.B. Post, J.Z. Pu, M. Schaefer, B. Tidor, R.M. Venable, H.L. Woodcock, X. Wu, W. Yang, D.M. York, M. Karplus, Charmm, The biomolecular simulation program, J. Comput. Chem. 30 (2009) 1545–1614. [40] A. Aksimentiev, K. Schulten, Imaging alpha-hemolysin with molecular dynamics: ionic conductance, osmotic permeability and the electrostatic potential map, Biophys. J. 88 (2005) 3745–3761.
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
Selective ion transport in functionalized carbon nanotubes Olga N. Samoylova a , Emvia I. Calixte b , Kevin L. Shuford b,∗ a b
Department of Biochemistry and Molecular Biology, University of Texas Medical Branch, 301 University Blvd, Galveston, TX 77555-0304, United States Department of Chemistry and Biochemistry, Baylor University, One Bear Place #97348, Waco, TX 76798-7348, United States
a r t i c l e
i n f o
Article history: Received 24 February 2017 Received in revised form 16 May 2017 Accepted 10 June 2017 Available online 15 June 2017 Keywords: Modified carbon nanotube Selectivity Ion current Molecular dynamics
a b s t r a c t Ion transport through functionalized carbon nanotubes in an external electric field is studied using all atom molecular dynamics simulations. The surface of carbon nanotubes has been functionalized with hydrogens and hydroxyl groups, and ionic current passing through the nanochannels has been examined with respect to the extent of surface modification. We are able to dramatically increase the ionic current passing through the nanotube via the appropriate surface modification. An analysis of the electrostatic potential within the tube shows higher ionic currents result from an increase in accessible pathways coupled with a global shift toward more direct ion passage. Moreover, through judicious choice of structure, the current can be modulated to a large degree with ion selectivity. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Structure and dynamics of confined fluids is drastically different from that of bulk due to large internal surfaces and geometric restriction. Variations in the local environment can lead to altered physical and chemical properties, and in some cases yield exotic system behavior. Examples include modified fluid structure, different thermodynamic, kinetic and correlation properties, size dependence and shift of critical parameters like temperature, pressure, and density, as well as spatial limitation and dimensional crossover effects on the dynamic properties such as diffusion and relaxation [1–5]. In our previous investigations of brines under confinement, we observed significant changes in fluid properties as the encapsulating pores decreased in size [6–9]. Specifically, we found reduced water densities, increased water and ion structuring at interfaces, reduced ion hydration, and disruption to the overall hydrogen bonded network. The increased hydrogen bond lifetimes coupled with higher hydrogen bond activation energies were due to unfavorable reorientation of water molecules as pore diameter decreased [9]. There is great interest in elucidating the characteristics of confined fluids and further developing systems that can actively control these properties on the nanoscale. Carbon nanotubes (CNTs) are promising candidates for such novel technologies due to their unique mechanical, electrical, and optical properties [10–12]. In particular, applications that incorporate nanotubes (sensing,
∗ Corresponding author. http://dx.doi.org/10.1016/j.apsusc.2017.06.120 0169-4332/© 2017 Elsevier B.V. All rights reserved.
nanofluidics, electrochemistry, etc.) may benefit greatly from controllable ion transport coupled to favorable surface interactions that can be manipulated in some way [13–18]. For example, the development of water purification technologies using CNTs is very attractive owing to their chemical and mechanical properties, low mass, large surface area, high aspect ratio, low cost and low impact on the environment [19]. These properties along with high flux and potential for selectivity make them highly favorable for next generation membrane technology [20,21]. However, pristine nanotubes tend to aggregate and fouling can also limit desalination. Surface modifications to CNT sidewalls can address these limitations and may also improve adsorption of contaminants and intensify water flux [19]. Other studies on ion entry have reported routes for selectivity based upon judicious choice and placement of functional groups [22,23]. More generally, it is expected that uncapped CNTs are inherently tip modified, and investigations into different forms of functionalization are highly relevant [24]. It is thus important to fully understand fluid transport through CNTs as you reduce tube size, and in turn how the surface chemical properties affect this. In a previous molecular dynamics (MD) study, we examined ion and solvent transport through CNTs of different diameters in an external electric field [25]. For large diameter CNTs, ionic current was essentially linear with voltage. However, deviations from this behavior emerged as the CNT diameter decreased, until eventually no ions were able to pass. The aim of the current study is to increase the ionic current through small diameter CNTs by modifying the tube surface with different functional groups. The results are meant to emphasize the effects of surface chemistry and demonstrate how these interactions can be utilized to control transport, especially
O.N. Samoylova et al. / Applied Surface Science 423 (2017) 154–159
155
Table 1 Carbon nanotube systems of interest. Columns are labeled by CNT chiral vector indices (n, m), CNT radius, the number of carbon atoms in the CNT, and the number of ions and water molecules, respectively. The tube length is 40 A˚ in all cases. (n, m)
Radius, A˚
Carbon
Ions
Water
(6,6) (7,7) (8,8) (9,9)
4.09 4.76 5.44 6.12
480 559 640 720
36 34 34 32
932 898 857 839
Table 2 CNT modification by rings with the numbers of H/OH functionalizations bound to carbon atoms. Columns are labeled by number of rings and CNT type.
Fig. 1. Schematic representation of CNT modification from the outside or inside (left) and by rings (right). Spheres denote functional groups (OH or H atoms) bound to carbon atoms of the CNT. Green/blue rings represent odd/even number of modification rings (for visualization purposes only). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
in small CNTs that do not readily permit ion passage otherwise. We find that by altering the conditions at the interface, the ionic current can be increased significantly. Our analysis also provides insight into the mechanism and underlying principles driving this observation, which can be leveraged to design fluid transport vessels with enhanced properties. Through surface chemistry, we are able to modulate the current and provide a degree of ion selectivity not previously present in pristine CNTs of similar size. This yields a chemically intuitive approach toward controlling fluid transport properties within CNTs via existing methods of synthesis. Although not the focus here, we note there is a vast literature describing synthetic routes to functionalize CNTs. Most relevant for the theoretical studies below are those that result in sidewall addition of H or OH groups, which have been realized experimentally [26–32]. 2. Theoretical methods 2.1. Model systems We have examined armchair CNTs with chiral indices ranging from (6,6) to (9,9) that are 40 A˚ in length. Our previous studies found NaCl transport through unfunctionalized (9,9) CNTs to be substantially altered by confinement effects [25], thus CNTs of this size were specifically targeted in this work. In particular, we wanted to induce significant ionic transport in systems that did not readily allow ions to pass previously, while providing some degree of selectivity. CNTs were constructed using the tcl code [33] with the C C bond equilibrium distance set to 1.42 A˚ and an equilibrium angle between carbon atoms of 0 = 120◦ . To model functional groups on the surface, pristine CNTs were modified by adding H or OH groups to selected carbon atoms. Experimental routes of sidewall addition lead to varying degrees of functionalization that are not saturated but can vary greatly (1–10%, perhaps more) [34]. To account for this broad range and extrapolate to levels beyond, we varied the extent of functionalization by modifying the CNT incrementally, one “ring” at a time starting from the top. We denote a ring as a series of connected benzene-like molecules that roll up to encompass a segment of the overall CNT length; stacking all the rings forms the tube. For a (n, n) CNT, each ring has n hexagons of carbon atoms connected in a circle. Our model for the modified CNT adds functional groups to 2n carbon atoms composing the ring – every second carbon atom along the top of a hexagon cell and every second carbon along the bottom of each hexagon cell, staggered from the top. Fig. 1 shows schematically the method of CNT modification by rings. For example, a (9,9)
Rings
(6,6)
(7,7)
(8,8)
(9,9)
1 2 . . . 10
12 24 . . . 120
14 28 . . . 140
16 32 . . . 160
18 36 . . . 180
Table 3 Bond parameters. Atom pair
kstretch , kcal/mol A˚ 2
r0 , A˚
C C C O
305.0 340.0 334.3 545.0
1.375 1.083 1.411 0.96
C H O H
CNT with 1 ring of modified carbon atoms has 18 functional groups of atoms bound to certain carbons. The H or OH groups were added to the outside of the CNT, leading to different configurations for each CNT diameter examined. Note that functionalizing the CNT alters the charge distribution, which is represented in the simulations by a change in force field parameters at those sites (see Section 2.2). All modifications of pristine CNTs were performed using Avogadro [35], followed by an initial geometry optimization using MMFF94 force field [36]. The functionalized CNTs were then solvated with NaCl solution. Details about the model systems – the number of carbon atoms, OH/H groups added, sodium and chloride ions, and water molecules – can be found in Tables 1 and 2. We have chosen to utilize a rather idealized model throughout these studies. The goal was not to represent a particular experimental system, but rather to explore the range of effects that could be achieved by introducing new interactions in a controllable manner. This approach allows us to glean insight from the model systems, which can then be used to develop a fundamental understanding of complex transport processes and routes to control them. Model variations that probe the vast complexities present in real experimental systems are under development currently. 2.2. Simulation approach We used all atom molecular dynamics (MD) to simulate ion transport through CNTs of various radii in the presence of an electric field. Each tube was solvated by ∼1 M NaCl solution in an ˚ orthorhombic simulation box with dimensions 24 A˚ ×24 A˚ ×70 A. TIP3P water molecules [37] have been used for the simulations. Carbon atoms of the tube were harmonically restrained. The tube was aligned along the z-axis, corresponding to the long axis of the hexagonal system. All MD simulations have been performed using the program NAMD2.9 [38] with the CHARMM36 force field [39]. Force field parameters are shown in Tables 3–6. The system was minimized and then equilibrated for 5 ns at 300 K using the NPT ensemble and a 1 fs time step. Periodic boundary conditions were applied in 3 dimensions, and the Particle
156
O.N. Samoylova et al. / Applied Surface Science 423 (2017) 154–159
Table 4 Angle parameters. Atom triplet C C C C
C C C O
C H O H
kangle , kcal/mol rad2
0 , ◦
40.0 29.0 45.2 65.0
120 120 120 108
Table 5 Dihedral parameters. Atom quadruplet
kdihedral , kcal/mol
0 , ◦
C C C C
3.10 3.50 3.10 0.99
180 180 180 180
C C C C
C C C O
C H O H
Table 6 Non-bonded interaction parameters and atom charges. Atom pair
ij , kcal/mol
ij , A˚
Pair charges
C C C O
0.070 0.103 0.039 0.084
3.98 3.76 3.31 1.99
0, 0 −0.09, 0.09 0.11, −0.53 −0.53, 0.42
C H O H
Fig. 3. Ionic current in the (9,9) CNT modified externally with H atoms for direct(+z) and reverse(−z) electric field, red and black traces respectively. Also shown are the levels of current passing through unmodified CNTs. Na ions are current carriers in these systems. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
type/location of functionalization, and direction of the applied field. For example, (9,9)-HoutRev denotes a (9,9) CNT modified with H atoms on the tube exterior biased with a 1 V electric field in the −z direction. Production MD runs of 50 ns including the biasing electrical field were performed on every system. A Lowe-Anderson thermostat was used to control the temperature at 300 K, and the extended system pressure method was used to maintain a constant pressure of 1 atm. Current was computed by counting every ion inside the carbon nanotube passing the CNT center of mass [25], given as
I t+
t 2
1 qi i (t + t) − i (t) tl N
=
(1)
i
where l = ztop − zbot is a nanotube length, ztop and zbot are the zcoordinates of the top and bottom position of the CNT, N is the number of ions, qi is an ion charge, zi is the z coordinate of the ion, t is the time between trajectory frames, and
i (t) =
⎧ z (t), if zbot ≤ zi (t) ≤ ztop ⎪ ⎨ i ⎪ ⎩
zbot ,
if zi (t) < zbot
ztop ,
if zi (t) > ztop
(2)
The first 2 ns of the trajectory were not included when calculating the mean value of the current to allow the system time to reach the steady state regime. 3. Results and discussion
Fig. 2. VMD screenshot of the simulated system containing a carbon nanotube modified externally with hydroxyl groups and solvated in water with chloride (cyan) and sodium (yellow) ions. Water is rendered here with a space filling representation. The direct electric field is applied along the positive z direction of the CNT, and the reverse electric field corresponds to negative z direction. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
Mesh Ewald summation was used with a grid size of 1 A˚ for longrange electrostatic interactions. The van der Waals interactions were calculated with a 12 A˚ cut-off. The system was biased by an external electric field in the z-direction (direct field) and −zdirection (reverse field), with voltage drops of 1 V (e.g. see Fig. 2). Throughout the manuscript, results will be labeled by the CNT size,
We computed the ionic current migrating through surfacemodified CNTs with sizes ranging from (6,6) to (9,9). The smallest ones, (6,6) and (7,7), did not register any statistically significant current. H-functionalized (8,8) CNTs on the exterior display a slightly increased ionic current compared to larger, pristine tubes (see Supplementary material); however, the most dramatic effect from our studies is observed for the (9,9) CNT with H atoms bound to the exterior surface. As a result, functionalized (9,9) CNT systems will be the primary focus of the paper, initially with H atoms on the exterior and then hydroxyls. Fig. 3 shows the ionic current driven through functionalized (9,9) CNTs with H atoms added to the outside of the tube surface. Adding H atoms to a single ring increases the current 5–20 fold with respect to an unmodified CNT of the same size, whereas a fully decorated CNT (10 rings) generates relative current changes of
O.N. Samoylova et al. / Applied Surface Science 423 (2017) 154–159
approximately 70–95. The latter represents a substantial increase in the number of ions passing through the CNT when H atoms uniformly cover the exterior surface. Amazingly, we find it is not only possible to increase the current compared to the pristine tube of the ˚ but to reach ionic same size (e.g., (9,9) CNT radius is about 6.12 A), current levels similar to an unmodified (12,12) CNT with radius ˚ In this case, the altered conditions at the interface effec8.15 A. tively increase the CNT diameter by ∼4 A˚ from the perspective of current passed. As shown in Fig. 3, current trends behave differently depending upon the direction of the applied field. For this system, hydrogen atoms are bound to carbon on the external surface. The negatively charged chloride ions are concentrated in the bulk water, which is largely due to the external CNT modification with hydrogens as well as the anions larger size. Therefore, the current carriers through the (9,9)-Hout CNT are almost exclusively sodium ions. For the reverse field case (electric field in the −z direction, black trace), cations enter the tube from the top and move down. The functionalized CNTs were created by adding rings of modified carbon atoms bound with hydrogens starting at the top. So for the case where the sodium ion is driven from the top, there is a more favorable Coulomb interaction with the carbon atoms that have been functionalized with hydrogen due to bond polarity. The partially negative carbons attract sodium ions into the tube by decreasing the energetic barrier for ions to leave bulk water and move into the confined area within the tube. As shown in Fig. 3, the ionic current through the (9,9)-Hout modified tube with 1 ring is already much higher for the reverse field than for the pristine (9,9) CNT (green line). For the direct field case (+z direction, red trace), sodium ions will enter the tube from the bottom and migrate up, corresponding to the electric field direction. The situation is not much different from an unmodified CNT if there are only a few functionalized carbon rings on the top of the tube. The ions only weakly feel the electrostatic attraction from the functionalized carbons until the modified tube section is closer to the entrance point (achieved within our model by adding more rings). When sodium cations can more easily interact with the functionalized regions, the additional Coulomb forces facilitate ion passage. As a result, there is a significant increase in the total current as more functionalized rings are included starting from 7. For the fully modified tube, the values for reverse and direct current are close but not equivalent due to slight variations in the structure. Note the CNT has some asymmetry in the position of functionalized carbons. The first ring (top) has modified carbons at the edge of CNT, whereas the last ring of carbons (bottom) has unmodified carbons at the edge (see Fig. 1). The interactions between ions and these critical regions of the structure is the reason for the difference in reverse and direct currents for fully modified CNTs. To glean insight into current trends, we studied ion passage through the modified (9,9)-Hout tubes with 1, 5, and 10 rings. Different drift regimes within these tubes were identified by tracking the time dependence of z-position of any ion inside the tube. We only included events where the z-coordinate of the ion changed sign and then eventually traversed the CNT interior (i.e., a translocation event). We did not incorporate times when an ion changed its location from the bulk water to the tube and back multiple times before an actual translocation initiated. Events were ordered by ascending passage time and separated into small time bins. Fast, moderate, and slow regimes were then identified arbitrarily, based upon passage time. Translocation events that occurred in less than 0.4 ns were designated as fast, while events over ∼0.8 ns were considered slow. Everything in between these extremes was denoted moderate. Fig. 4 displays the mean value of z (averaged over all events in the particular bin) as a function of time for different
157
Fig. 4. Ion drift regimes and passage time for the modified (9,9)-Hout tubes with the reverse electric field applied. Panels A–C correspond to 1, 5, and 10 rings of functionalized carbons. The axial coordinate, z, is averaged over all translocation events. (For interpretation of the references to color in the text, the reader is referred to the web version of the article.)
regimes. There is a quasi-linear trend for ion translocation in the tube with one ring of modified carbon atoms (Panel A); 61% of ions drift with the moderate pace, and 39% in the fast regime. In the case with 5 modification rings (half of the CNT, Panel B), we see a similar rapid passage through the functionalized part of the tube, more relaxed motion in the pristine part of CNT, and an approximately equal number of ions in the fast and moderate regime. However, a deviation from this behavior emerges, where some rare events (5%) of ion trapping at the boundary of the modified/unmodified region occur. These ions fluctuate around the middle of the tube at z = 0
158
O.N. Samoylova et al. / Applied Surface Science 423 (2017) 154–159
Fig. 5. Electrostatic potential (in kT/e units = 0.0258 V for T = 300 K) for the (9,9)Hout CNT with a reverse applied electric field. Panels A–C correspond to CNTs with 1, 5, and 10 modified carbon atoms respectively.
before passing, which leads to the slow translocation regime. For the completely modified tube (10 rings, Panel C), the vast majority (78%) of ions traverse rapidly through the tube within 0.4 ns. 21% of events are in moderate regime, and only 1% of ions are in slow regime. An important factor that affects the ion passage is position inside the CNT. This occurs because drastically different Coulomb interactions are present between carbon sites and the ion based upon internal location. To illustrate this, we provide electrostatic potential maps (Fig. 5) using the VMD PMEpot plugin [40]. Note the product of charge times electrostatic potential can be interpreted as the potential energy of the charge in an electrostatic field. From these maps one can infer different potentials are present depending upon sodium ion’s radial and z position within the CNT, which can be used to understand the translocation times in Fig. 4. For 1 ring (Fig. 5A), there is a clear path through the middle (due to the circular symmetry in the electric field in XY plane) leading to fast, linear transport. Any cation near the tube edge is pushed to the center, potentially bouncing back and forth, yielding approximately linear transport over slightly longer times. The short modified part of the tube has a small influence on the ion motion only at the beginning of the passage. Beyond that point, ion transport would be similar to being in a pristine CNT under the influence of an external electric field. The electrostatic environment changes as more charged rings are added to the tube. Fig. 5B shows the cation feels a strong attraction from the functionalized part of the tube initially and then experiences a very different environment moving through the CNT in the external electric field. There is a negative potential inside the top portion of the tube with 5 modified rings, which becomes neutral in the central region before turning positive. Sodium ions readily enter the CNT from the top and easily migrate half-way through before experiencing the positive potential barrier. Ions that do not traverse exclusively via the center of the CNT feel more resistance, but can begrudgingly migrate toward the exit under the influence of the applied field. This scenario lends itself to larger radial force components, causing more ion motion in the XY plane of the tube and delaying translocation along z. We can see oscillations and more spread in axial position (green line, Fig. 4B) that emerge from the white neutral zone in the tube, which persist until the ion exits the CNT. For 10 rings (Fig. 5C), we see a similar electrostatic potential pattern inside the tube; however, the positive barrier is shifted to the CNT exit and considerably smaller. The vast majority of sodium ions can easily traverse the barrier rapidly, and 78% of translocation events are fast. So we see from this analysis that external H modification of CNTs leads to a negative electrostatic environment inside, which facilitates cation transport by increasing accessible pathways through the CNT and inducing a shift toward faster ion passage from the other drift regimes. Analogously, anion passage can be enhanced via the appropriate surface functionalization. While the absolute current values are
Fig. 6. Ionic current in a (9,9) CNT externally modified with hydroxyl groups for direct(+z) and reverse(−z) electric field, red and black traces respectively. Also shown are the levels of current through unmodified CNTs. Chloride ions are the current carriers in these systems. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
significantly less, due to the larger size of chloride, the relative gains compared to pristine CNTs are noteworthy, and this provides a route toward selectivity of ion transport. Functionalizing the CNT exterior with hydroxyl groups creates an environment inside that favors chloride, while the partially negative oxygen atoms draw sodium to the CNT exterior. Thus, the ionic current carriers through the nanotube are Cl− , and they are moving in the direction opposite to an applied external electric field. So for the direct electric field chloride ions are entering the top (i.e., where the modified rings start), and for the reverse field chloride is moving in from the bottom. As a result, the direct ionic current for the partially modified tube is greater than the reverse field case (see Fig. 6). Note, the situation is exactly opposite to what was described previously for external functionalization with H atoms, substantiating that chloride is the carrier. The maximum relative current change for (9,9)-OHout is approximately 12 compared to the pristine (9,9) CNT. So while the currents are still quite small, we are able to pass some carriers through CNTs that essentially did not permit chloride transport before functionalization. More generally, we find it is possible to get ≈ 5 times more current through a (9,9) CNT using external surface modification with H atoms compared to OH groups. The primary differentiator is the size of the ion carrier (sodium for the former and chloride for the latter). To be complete, we briefly note results obtained for the cases of interior functionalization, even though such systems would be challenging to realize experimentally. Adding hydrogens to the interior favors chloride passage, creating an environment analogous to hydroxls on the exterior described above. The radius of ˚ We can estimate the effective the unmodified (9,9) CNT is 6.12 A. radius of available space using the bonds and angle parameters for the functional group. The internal radius for the modified (9,9)-Hin ˚ which is less than the radius of an unmodified (8,8) is about 5.04 A, ˚ but the ionic current can reach values greater CNT (R(8,8) = 5.44 A), than the current passed in an unmodified (9,9) CNT (see Supplementary material). The ionic current attained is comparable to the (9,9)-OHout CNT discussed above. Lastly, we did not register any current for hydroxyl modification on the inside of (9,9) CNTs. The internal radius of available cylindrical space for the modified (9.9)˚ which is less than the radius of an unmodified OHin is about 4.74 A, ˚ where we did not register any ionic per(7,7) CNT (R(7,7) = 4.76 A) meation events in a previous study [25]. In principle, both ions could have favorable interactions with the interior regions to some degree; however, the lack of available space for the solvent and ions to traverse the CNT is strongly limited in this case.
O.N. Samoylova et al. / Applied Surface Science 423 (2017) 154–159
4. Conclusions Our simulations show that ionic current through functionalized carbon nanotubes can be modulated by different functional groups bound to the carbon atoms. Using modified nanotubes, it is possible to not only dramatically increase ionic current through the tube, but also provide selectivity of the current carriers. We have shown that direct and reverse currents in the partially modified nanotubes are not identical, and there is a direct correlation between ionic current and the extent of surface modification in CNT. Thus it may be possible to actively manipulate the time ions spend inside a carbon nanotube (e.g., trap the ion inside and/or release it) by adjusting field polarity in modified tubes, which could be potentially useful for multiple applications like nanosensors, chemical reactions under confinement, and fluidics. Acknowledgements This work is supported by the Chemical Sciences, Geosciences, and Biosciences Division, Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy under Award Number DESC0010212. KLS thanks Baylor University for sabbatical funding. Appendix A. Supplementary data Supplementary data associated with can be found, in the online cle http://dx.doi.org/10.1016/j.apsusc.2017.06.120.
this artiversion, at
References [1] H.E. Stanley, Introduction to Phase Transition and Critical Phenomena, Clarendon Press, Oxford, 1971. [2] M.A. Anisimov, Critical Phenomena in Liquids and Liquid Crystals, Gordon & Breach, Philadelphia, 1991. [3] J.-C. Wang, K.A. Fichthorn, Influence of molecular structure on the properties of confined fluids by molecular dynamics simulation, Colloids Surf. A: Physicochem. Eng. Aspects 206 (1–3) (2002) 267–276. [4] D. Morineau, C. Alba-Simionesco, Liquids in confined geometry: how to connect changes in the structure factor to modifications of local order, J. Chem. Phys. 118 (2003) 9389. [5] A.B. Farimani, N.R. Aluru, Existence of multiple phases of water at nanotube interfaces, J. Phys. Chem. C 120 (2016) 23763–23771. [6] M.C.F. Wander, K.L. Shuford, Molecular dynamics study of interfacial confinement effects of aqueous NaCl brines in nanoporous carbon, J. Phys. Chem. C 114 (2010) 20539–20546. [7] M.C.F. Wander, K.L. Shuford, Electrolyte effects in a model system for mesoporous carbon electrodes, J. Phys. Chem. C 115 (2011) 4904–4908. [8] M.C.F. Wander, K.L. Shuford, Alkali halide interfacial behavior in a sequence of charged slit pores, J. Phys. Chem. C 115 (2011) 23610–23619. [9] E.I. Calixte, O.N. Samoylova, K.L. Shuford, Confinement and surface effects of aqueous solutions within charged carbon nanotubes, Phys. Chem. Chem. Phys. 18 (2016) 12204–12212. [10] M.D. Volder, S.H. Tawfick, R.H. Baughman, A.J. Hart, Carbon nanotubes: present and future commercial applications, Science 339 (2013) 535–539. [11] S. Iijima, T. Ichihashi, Single-shell carbon nanotubes of 1-nm diameter, Nature 363 (1993) 603–605. [12] J.N. Coleman, U. Khan, W.J. Blau, Y.K. Gunko, Small but strong: a review of the mechanical properties of carbon nanotube-polymer composites, Carbon 44 (9) (2006) 1624–1652. [13] J.J. Gooding, Nanostructuring electrodes with carbon nanotubes: a review on electrochemistry and applications for sensing, Electrochim. Acta 50 (2005) 3049–3060. [14] R. Martel, T. Schmidt, H. Shea, T. Hertel, P. Avouris, Single- and multi-wall carbon nanotube field-effect transistors, Appl. Phys. Lett. 73 (1998) 2247–2249.
159
[15] T. Chen, L. Dai, Carbon nanomaterials for high-performance supercapacitors, Mater. Today 16 (2013) 272–280. [16] J. Su, D. Huang, Coupling transport of water and ions through a carbon nanotube: the role of ionic conduction, J. Phys. Chem. C 120 (2016) 11245–11252. [17] B. Corry, Designing carbon nanotube membranes for efficient water desalination, J. Phys. Chem. B 112 (5) (2008) 1427–1434. [18] K. Sint, B. Wang, P. Král, Selective ion passage through functionalized graphene nanopores, J. Am. Chem. Soc. 130 (2008) 16448–16449. [19] R. Das, M.E. Ali, S. Hamid, S. Ramakrishna, Z.Z. Chowdhury, Carbon nanotube membranes for water purification: a bright future in water desalination, Desalination 336 (2014) 97–109. [20] S. Kar, R.C. Bindal, P.K. Tewari, Carbon nanotube membranes for desalination and water purification: challenges and opportunities, Nano Today 7 (2012) 385–389. [21] R. Das, S. Hamid, M.E. Ali, A.F. Ismail, M.S.M. Annuar, S. Ramakrishna, Multifunctional carbon nanotubes in water treatment: the present, past and future. desalination, Desalination 354 (2014) 160–179. [22] S. Joseph, R.J. Mashi, E. Jakobsson, N.R. Aluru, Electrolytic transport in modified carbon nanotubes, Nano Lett. 3 (10) (2003) 1399–1403. [23] M. Majumder, N. Chopra, B.J. Hinds, Effect of tip functionalization on transport through vertically oriented carbon nanotube membranes, J. Am. Chem. Soc. 127 (2005) 9062–9070. [24] P.S. Goh, A.F. Ismail, B.C. Ng, Carbon nanotubes for desalination: performance evaluation and current hurdles, Desalination 308 (2013) 2–14. [25] O.N. Samoylova, E.I. Calixte, K.L. Shuford, Molecular dynamics simulations of ion transport in carbon nanotube channels, J. Phys. Chem. C 119 (2015) 1659–1666. [26] A. Tagmatarchis, V. Georgakilas, M. Prato, H. Shinohara, Sidewall functionalization of single-walled carbon nanotubes through electrophilic addition, Chem. Commun. (2002) 2010–2011. [27] S. Santangelo, Controlled surface functionalization of carbon nanotubes by nitric acid vapors generated from sub-azeotropic solution, Surf. Interface Anal. 48 (2016) 17–25. [28] I. Gerber, M. Oubenali, R. Bacsa, J. Durand, A. Goncalves, M. Fernando, R. Pereira, F. Jolibois, L. Perrin, R. Poteau, P. Serp, Theoretical and experimental studies on the carbon-nanotube surface oxidation by nitric acid: interplay between functionalization and vacancy enlargement, Chem. Eur. J. 17 (2011) 11467–11477. [29] G. Miller, J. Kintigh, E. Kim, P. Weck, S. Berber, D. Tomanek, Hydrogenation of single-walled carbon nanotubes using polyamine reagents: combined experimental and theoretical study, J. Am. Chem. Soc. 130 (2008) 2296–2303. [30] A. Dillon, K. Jones, T. Bekkedahl, C. Kiang, D. Bethune, M. Heben, Storage of hydrogen in single-walled carbon nanotubes, Nature 386 (1997) 377–379. [31] Y.S. Nechaev, A. Yurum, A. Tekin, N.K. Yavuz, Y. Yurum, T.N. Veziroglu, Fundamental open questions on engineering of “super” hydrogen sorption in graphite nanofibers: relevance for clean energy applications, Am. J. Anal. Chem. 5 (2014) 1151–1165. [32] A. Fujiwara, K. Ishii, H. Suematsu, H. Kataura, Y. Maniwa, S. Suzuki, Y. Achiba, Gas adsorption in the inside and outside of single-walled carbon nanotubes, Chem. Phys. Lett. 336 (3–4) (2001) 205–211. [33] F.Q. Zhu, K. Schulten, Water and proton conduction through carbon nanotubes as models for biological channels, Biophys. J. 85 (2003) 236–244. [34] M. Burghard, K. Balasubramanian, Chemically functionalized carbon nanotubes, Small 1 (2005) 180–192. [35] M.D. Hanwell, D.E. Curtis, D.C. Lonie, T. Vandermeersch, E. Zurek, G.R. Hutchison, Avogadro: an advanced semantic chemical editor, visualization, and analysis platform, J. Chem. Inf. 4 (2012) 4–17, article number 17. [36] T.A. Halgren, Merck molecular force field. I. Basis, form, scope, parameterization, and performance of MMFF94, J. Comput. Chem. 17 (1996) 490–519. [37] W.L. Jorgensen, J. Chandrasekhar, J.D. Madura, R.W. Impey, M.L. Klein, Comparison of simple potential functions for simulating liquid water, J. Chem. Phys. 79 (1983) 926–935. [38] J.C. Phillips, R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa, C. Chipot, R.D. Skeel, L. Kale, K. Schulten, Scalable molecular dynamics with NAMD, J. Comput. Chem. 26 (2005) 1781–1802. [39] B.R. Brooks, C.L. Brooks, A.D. Mackerell, L. Nilsson, R.J. Petrella, B. Roux, Y. Won, G. Archontis, C. Bartels, S. Boresch, A. Caflisch, L. Caves, Q. Cui, A.R. Dinner, M. Feig, S. Fischer, J. Gao, M. Hodoscek, W. Im, K. Kucczera, T. Lazaridis, J. Ma, V. Ovchinnikov, E. Paci, R.W. Pastor, C.B. Post, J.Z. Pu, M. Schaefer, B. Tidor, R.M. Venable, H.L. Woodcock, X. Wu, W. Yang, D.M. York, M. Karplus, Charmm, The biomolecular simulation program, J. Comput. Chem. 30 (2009) 1545–1614. [40] A. Aksimentiev, K. Schulten, Imaging alpha-hemolysin with molecular dynamics: ionic conductance, osmotic permeability and the electrostatic potential map, Biophys. J. 88 (2005) 3745–3761.