Role of surfactant structure in aqueous dispersions of carbon nanotubes

Accepted Manuscript Title: Role of Surfactant Structure in Aqueous Dispersions of Carbon Nanotubes Author: N. Poorgholami-Bejarpasi B. Sohrabi PII: DOI: Reference:

S0378-3812(15)00100-4 http://dx.doi.org/doi:10.1016/j.fluid.2015.02.032 FLUID 10468

To appear in:

Fluid Phase Equilibria

Received date: Accepted date:

15-2-2015 25-2-2015

Please cite this article as: N.Poorgholami-Bejarpasi, B.Sohrabi, Role of Surfactant Structure in Aqueous Dispersions of Carbon Nanotubes, Fluid Phase Equilibria http://dx.doi.org/10.1016/j.fluid.2015.02.032 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Role of Surfactant Structure in Aqueous Dispersions of Carbon Nanotubes N. Poorgholami-Bejarpasi, B. Sohrabi* Department of Chemistry, Surface Chemistry Research Laboratory, Iran University of Science and Technology, P.O. Box 16846-13114, Tehran, Iran. *

Corresponding author. Fax: +98-(0)21-77491204, Tel.: +98-(0)21-77240540 E-mail address: [email protected]

Graphical abstract

1

Highlights: 

The single-chain surfactants, particularly at low surface coverage, tend to align parallel to the nanotube axis.



The random adsorption model can be appropriately used to describe the single-chain surfactants adsorption onto SWNT surface at low surface coverage.



The Gemini surfactant [12-2-12]Br2 heads protrude more pronouncedly toward the aqueous phase.



The Gemini surfactant [12-6-12]Br2 with a longer spacer tends to yield more disordered structures as compared to[12-2-12]Br2.

Abstract Gemini surfactants typically consist of two single-chain surfactantschemicallyconnected by a spacer chain. Recently, it has been shown experimentally that they are able to effectively disperse and stabilize carbon nanotubes (CNTs)in aqueous solutions at very low surfactantconcentration.To aid elucidate the role of surfactant structure in the CNT dispersion process, we report hereinthe results of fully atomistic molecular dynamics (MD) simulations ofthe adsorption and surface self-assembly ofa cationic single-chain surfactant,dodecyltrimethylammonium

bromide

(C12TAB),

and

its

related

Gemini

surfactantdimethylene-α,β-

bis(dodecyldimethylammonium bromide) [12-2-12]Br2, on (5,7), and (10,14) single-walled carbon nanotubes (SWNTs) in aqueous solutionat ambient conditions. We find that the morphology of surfactant aggregates on the SWNT is influenced by the surfactant structure. The same number of [12-2-12]Br2 Gemini surfactants adsorbed are able to cover a larger surface area of SWNT and their head-groups are protruded moreextensively toward the aqueous phase, 2

preventing water molecules from accessing the nanotube surface.These morphological results propose that[12-2-12]Br2should be more effective thanC12TABat stabilizingaqueous dispersions of carbon nanotubes. Furthermore, the influence of SWNT diameter and length of the Gemini surfactant spacer on the structure of aggregates formed onto nanotubes are investigated.These simulation results advance our understanding of the mechanism of CNTsolubilization via Gemini surfactants.

Keywords:Single-walled Carbon Nanotube (SWNT); Gemini Surfactants; Molecular Dynamics Simulations; Morphology; Adsorption.

1. Introduction The successful preparation of stable single-walled carbon nanotube (SWNT) dispersions represents an important steptoward the effective use of this unique material inmany technological applications. As-produced SWNT samplesconsist of bundles or ropes containing nanotubes ofdifferent diameters and chiralities, lengths, and electronicproperties [1]. Due to the strongintertube van derWaals interactions, the bundles are insoluble in common organic solvents and water, which limits their utilization [2,3]. Remarkable advances have been made over the last years to suspend individual SWNTs in several solvents by thenon-covalent functionalization approach [1,2].In this method, the tube surface can be modified via van der Waalsforces and π-π interactions, by adsorption or wrapping ofsurfactants, polymers, orbiomolecules [4]. Surfactants are widely used to disperse SWNTs in water.Although they are expected to attach on the nanotube surface withtheir hydrophobic tails, while the hydrophilic heads of surfactants orient towardthe water,a continuing debate attempts to clarify the structure of self-assembled

3

surfactant aggregates on nanotubes [5]. A wide variety of conventional surfactants, such as sodium

dodecylbenzenesulfonate

(SDBS),

sodium

dodecyl

sulfate

(SDS),hexadecyltrimetylammonium bromide (C16TAB), and sodium cholate (SC)have been shown to stabilize SWNT dispersions. Tan et al. used optical absorption spectroscopy to compare the ability of different surfactants for dispersing of carbon nanotubes [6]. They found that the structure of hydrophobic tail and the charge of the hydrophilic head play important roles in the dispersion process. Geminisurfactants are a new class of surfactants consisting of two hydrophobic chains and two polarhead groups covalently attached through a spacer at the level ofthe head groups or very close to these groups [7].These surfactants have a greatpotential because of the enormous variability that characterizes their structuresand a number of novel aggregationproperties in comparison toconventional single-chain surfactants, such aslow critical micelle concentration (CMC), high surface activity, prevailingly micellar aggregate morphology spanning from spherical to disk-like or rod micelles (due to their high surfactant packing parameter), stronghydrophobic microdomain, and strong dependenceon spacer structure [8,9]. These properties make Gemini surfactants very promising candidates for non-covalent carbon nanotube (CNT) dispersion methods.Thesimplest and most studied cationic Gemini surfactants are the quaternary ammonium compounds, represented by the generalstructure [C m H2m+1 (CH3)2N+ (CH2)s N+ (CH3)2 C m H2m+1] 2Br− These surfactant molecules, often simplified to m-s-m, where m refers to the number of C-atoms in the alkyl chain and s to the C-atoms in the spacer group [10].Wang et al. studied the suspending behaviors of multi-walled carbon nanotubes (MWNTs), stabilized by cationic singlechain surfactant, dodecyltrimethylammonium bromide (C12TAB)and cationic Gemini surfactant 4

hexyl-α,β-bis(dodecyldimethylammonium bromide)[12-6-12]Br2[9].Their results showed that [12-6-12]Br2 can suspend the MWNTs effectively even at the surfactant concentration well below its CMC while, C12TABis not capable of stabilizing the MWNTs below CMC.They also found that the adsorption of these two surfactants reaches equilibrium at twice the CMC (2 mM for [12-6-12]Br2, and 30 mM for C12TAB).Further, they observedthat after adsorption equilibrium, the maximum amount of the suspended CNTs in[12-6-12]Br2solution is about twice as muchas its relative single-chain cationic surfactant (C12TAB).In a similar study, Di Crescenzo and co-workers showed that Gemini surfactants having an aromatic spacer are able todisperse SWNTs at surfactant/CNTs weight ratios fromfour to seven times lower than conventional surfactantssuch as C16TAB, SDS or SDBS [8].Using Raman spectroscopy, theyalso found that the Gemini surfactants show a slightly pronounced selectivity towards semiconducting SWNTs.Liu et al. used UV-vis-NIR and transmission electron microscopy (TEM) to study dispersion of the MWNTs in ionic liquid-type Gemini imidazolium surfactant solutions [11]. They observed the ionic liquid-based Gemini surfactantshave stronger ability for dispersing MWNTs as compared to its relative single-chain ionic liquid-based surfactant.Their results also demonstrated the Gemini surfactant with a longer hydrocarbon chain has a greater dispersing capacity.Bagonluri and Foldvaristudied the effects of the methylene spacer chain length of cationic Gemini surfactants on the stability of nanotube dispersions, and found that Gemini surfactants having shorter spacers yielded increased stability for aqueous nanotube dispersions [12]. The high charge capacity and strong adsorption ability of Gemini surfactants along with its ability to form more compact aggregates on the CNT surface are believed to provide the superior ability of Gemini surfactants to disperse CNTs [9,13].

5

The experimental studies available for CNT dispersion using Gemini surfactants are supported by very few, if any,computer simulation studies.Computer simulations are a convenient tool to studythe adsorption process of surfactant molecules onnanotube surface because they provide a microscopicpicture of the self-assembly process. They avoid some ofthe intrinsicexperimental difficulties associated with the observation ofthe adsorbed structures [14].Molecular dynamics (MD) simulationshave presented the atomistic model of the interactionmechanism between SWNTs and surfactant molecules.Tummala et al. used all-atom MD simulations to study the SDS aggregates morphology on the SWNTs [15]. They found that the morphology of the surfactantaggregates strongly depends on the nanotube diameter, as well as on the surface coverage. A large-scale MD simulation has been used by Xu and co-workers to investigate the structure of SDS/SWNT aggregates in aqueous solution under various surface coverageand nanotube curvatures [16]. The results showed that the counterion plays a vital dual role in providing both the vdWrepulsion, which assists the SWNT dispersion, and theCoulombic attraction, which hinders the dispersion.Aqueous dispersions of SWNTs stabilized using the bile salt surfactant SC have been studied via MD simulations by Lin et al. [17]. Their results demonstrated that the cholate ions wrap around the tubes with a small tendency to orient perpendicularly to the nanotube axis.By computing the potential of mean force (PMF) between two parallel SC-covered SWNTs as a function of the intertube separation, they concluded that SC is a better stabilizer than SDS at the saturated surface coverages [17].Tummala et al. used MD simulations to study adsorption of flavin mononucleotide (FMN) on the SWNTs [18]. They found that the aggregation morphology of aqueous FMN on SWNTs depends on nanotube diameter. Additionally, the calculation of PMF between two SWNTs in the presence of FMN showed that FMN surfactants are far superior to SDS in stabilizing aqueous dispersions of

6

SWNTs.Suttipong et al. used all-atomistic molecular dynamics (MD)simulations to study the adsorption and surface self-assembly ofSDBS surfactant onto SWNT [5]. They suggested that the surfactant molecular structurestronglyaffects the packing of surfactants on the nanotubes.MD simulations have also been used by this group to explore the effect of the counter-ion on the morphology of dodecyl sulfate surfactants adsorbed on carbon nanotubes [19].Their simulations demonstrated that when Na+ ions are substituted with Cs+ ones the self-assembled surfactant aggregate

morphology

changes

significantly

in

the

presence

of

narrow

carbon

nanotubes.Recently, all-atom MD simulations have been used in our group to investigate the influence of the surfactant tail length on the adsorption of cationic surfactants onto SWNTs[20]. Our results demonstrated that a longer chain yields the higher packed aggregates inwhich the surfactant heads are extended far into the aqueousphase, which in turn may increase the SWNTs stabilization inaqueous suspensions. In this work, the adsorption and the self-assembly of one conventional single-chain surfactant,dodecyltrimethylammonium

bromide(C 12TAB),

and

one

related

Gemini

surfactant,dimethylene-α,β-bis(dodecyldimethylammonium bromide) [12-2-12]Br2 on (5,7) and (10,14) SWNTs are investigated via all-atomistic MD simulations. The effects of the surfactant structure, surfactant surface coverage, and the length of spacer of Gemini surfactant on the aggregate morphology are discussed. Because of computational limitations, we limit our study torelatively dilute systems.

2. Models and simulation methods Molecular dynamics simulations of the self-assembly of C12TAB and Geminisurfactantson the nanotube surface in aqueous solution were carried out using the GROMACS 4.5 software package [21]. Two SWNTs [(5,7) and (10,14)] of 4.45 nm long and with diameters of 0.82 and 7

1.64 nm , respectively, were considered. In order to study the effect of surfactant surface coverage on the structure of surfactant aggregates formed on the SWNT surface, for both C12TAB and [12-2-12]Br2 surfactants two surface coverages were considered: 1.05 surfactant molecules /nm2 (low surface coverage) and 2.27 surfactant molecules /nm2 ( high surface coverage). Both surface coverages are comparable to those used in recent simulation studies [15,17].Note that we consideredthe surface coverages of both surfactant molecules (C12TAB and [12-2-12]Br2) equal to study the role of the surfactant structure on the morphology of surfactant aggregates

formed

on

the

SWNT

surfaces

of

given

diameter

under

the

samecircumstances.Because Gemini surfactants are found with different spacer length, we chose to simulate the adsorption of [12-6-12]Br2 onto (5,7) SWNT at high surface coverage to clarify the effect of spacer length on the surfactant aggregate morphology. The SWNTs were generated using TubeGen tool[22]and kept rigid throughoutthe simulations, with all the carbon atoms in the nanotube treatedas uncharged Lennard-Jones (LJ) spheres using the LJ non-bondedinteraction parameters which corresponded to the naphthalene OPLS-AA (allatom optimized molecular potential for liquid simulation)carbon atoms [23].Water molecules were modeled using the simple point charge extended (SPC/E) model [24]. C12TAB and Gemini surfactant molecules ([12-2-12]Br2 and [12-6-12]Br2), which were assumed to completely dissociate into bromide ions and C12TA+ or [12-2-12]2+ or [12-6-12]2+ions were modeled using the OPLS-AA force field and the atomic charges were determined by RESP fit using the RED server [25-28]. The equations of motion were integrated with a time step of 2 fs using the Verlet (Leap-Frog) algorithm [29,30]. All MD simulations were performed in an NPT ensemble at 300 K and 1.0 bar with

a

velocity-rescaled

Berendsen

thermostat 8

and

a

Parrinello-Rahman

barostat

[31,32].Thelong-range electrostatic interactions were handled with the particle mesh Ewald (PME) method [33,34]. The van der Waals interactions (vdW) were treated with cut off at 1.2 nm. The vdWinteractions betweendifferent atoms were calculated from the LJ potential using thestandard geometric averaging rule which is implemented in theOPLS-AA force field.Periodic boundary conditions were appliedin all three directions.The trajectories, velocities, and forcescorresponding to all the atoms in the system were saved everyeither 1000steps (2 ps) or 10000 steps (20 ps) to satisfy the ergodicity criterion for dataanalysis [35]. For all simulated systems, one SWNT was maintained at the center ofone simulation box of dimensions 7×7×7 nm3. The cylindrical axis of the SWNT was aligned along the z-directionof the simulation box and maintained fixed during the simulations.Because most of the nanotubes were opened after sonication, it is likely that the interior of the nanotube would be accessible to the environment, and potentially filled by the solvent, ions or other molecules. Therefore, the length of the simulation box was chosen to be about 2.5 nm longer than the nanotubes length (4.45 nm) to allow the water molecules or ions to fill the nanotubes. In the initial configurations the desired numbers of surfactants were placed around the nanotubes in a surrounding configuration and then the simulation box was filled withwater molecules. In order to maintain electroneutrality, anappropriate numbers of water molecules werereplaced by bromide counterions resulting from the added eitherdodecyltrimethylammoniumionsor the cationic part of Gemini surfactants. The simulated systems, including the SWNTs, the total number of surfactants and water molecules, and the total number of atoms are presented in Table 1. Before initiating the MD simulations, an energy minimization was performed using the steepest decent method to relax the systems. Each system was equilibrated for 110 ns, and only the last 10 ns of simulation were used for data analysis.To show the simulated systems have reached the stable

9

minimum, we plotted the variation of solvent accessible surface (SAS) areas of the C12TA+, [122-12]+2, and [12-6-12]+2ions as a function of simulation time (see Figure S1 in the Supporting Information).Furthermore, we performed simulations with two different initial configurations for two of our simulated systems to ensure that the results were not affected by the initial configuration.In Figure S2, we compared the simulation results computed from the new configurations (configuration 2) to those obtained from the primary configurations (configuration 1). For two systems considered, we did not observe any significant difference in the radial distribution functions (RDFs) obtained from the two different initial configurations selected here. Visualizations of all molecular configurations were done using VMD [36].

3. Results and discussion 3.1. Self-Assembly of C12TAB and [12-2-12]Br2Surfactants on SWNT. Representative simulation snapshots of C12TAB and [12-2-12]Br2 surfactants adsorbed on a (5,7) SWNT at low and high surfactant surface densities are shown in Figure 1. The snapshots suggest that the morphology of surfactant aggregates depends on both the surface coverage, as expected, and the surfactant structure. At low surface coverage, C12TAB surfactant tends to either lie flat on the nanotube surface or wrap around the SWNT. The tail segments and most head-groups are found near the nanotube surface and only few C12TAB heads are positioned further from the nanotube surface and exposed to water.Upon increasing the surface coverage,due to the reduction in tube surface area available per surfactant molecule, some of the C12TAB molecules prefer to rotate away from the nanotube surface such that fewer tail particles are in direct contact with the SWNT surface. The C12TAB head-groups at high surface coverage, as can be observed in Figure 1b, protrude morepronouncedly toward the aqueous phase.For the

10

adsorption of [12-2-12]Br2onto nanotube, visualinspection of the simulation snapshots indicate qualitative differences compared to results obtained for C12TABparticularly at high surface coverage. At low surface coverage, most of the alkyl chains of the Gemini molecules prefer to lie parallel to each other and parallel to the nanotube axis such thattheir entire hydrophobic tails are in direct contact with the SWNT surface. Hydrophobic interactions between the tail segments and the nanotube surface are responsible for this result. The higherhydrophobicity of the Gemini molecule(because of having two hydrocarbon alkyl chain)compared to itssingle-chain homologues (C12TAB), increases the repulsion between tail−head neighboring Gemini molecules and make a larger number of Gemini head-groups to be extended to theaqueous phase in this system. In addition, the higher charge capacity per single moleculeof Gemini surfactant (because of having two charged head-groups)increases the head-head repulsions and facilitates protrusion of the heads toward water.The larger size of a Gemini molecule give rises to the reductionin tube surface area available per particles,therefore, few molecules tend to adsorb onto nanotube surface such that only their terminal tailparticles are now in direct contact with the nanotube.As the surface coverage increases, a greater number of Gemini tails stand up on the nanotube surface and the head-groups project into water. This orientation of Gemini molecules enables more surfactant to adsorb onto SWNT, thereby reducing exposure of the hydrophobic surface of the SWNT and the Gemini tails to the aqueous environment.A great number of bromidecounterions accumulate near the Gemini head-groups because ofthe strong electrostatic attraction between the oppositelycharged groups.To quantify the orientations both of C12TAB and Gemini surfactants adsorbed on (5,7) SWNT at low and high surface coverage, we calculated the probability distribution of the angle betweenthevector identified by the surfactant tail-to-head (defined in Figure 2) and the SWNT axis. In the case of [12-2-12]Br2 Gemini surfactant, we

11

defined two tail-to-head vectors for two side chains of one surfactant molecule. The tail-to-head vectors considered here indicate either the overall orientation of the whole surfactant with respect to the carbon nanotube (for C12TAB) or the orientation of two side chains of a Gemini molecule with respect to the carbon nanotube.Note that when the angle between the tail-to-head vector and the SWNT axis is the angle is

or

, the vector is parallel to the nanotube axis, while when

, the vector is perpendicular to the nanotube axis.

The orientation probabilitydistributions for C12TAB and [12-2-12]Br2molecules on (5,7) SWNT at low and high surface coveragesare illustrated in Figure 2.The Figure suggests that the orientation distribution depends on thesurface coverage and more significantly on the surfactant structure. In Figure 2a, we observe that C 12TAB surfactant has a preference for orientingparallel to the nanotube axis at low surface coverage. Upon increasing the surface coverage, the probability of parallel orientation of C12TAB declines and surfactant molecules tend to wrap around the nanotube with various angles. Figure 2a indicates that C12TAB molecules also have a small tendency to orient perpendicular to the nanotube axis at high surface coverage. As can be observed in Figure 2b, there is only a slight difference in orientation distribution between two side chains of Gemini surfactantswith respect to the SWNT axis for both low and high surface coverages. At low surface coverage, the noticeable peaks around

and

suggest that both

side chains of the Gemini molecules tend to orient nearly parallel with respect to the nanotube axis. Compared to the C12TAB case at low surface coverage, the angles formed between the tailto-head vectors and the SWNT axisare larger which can be attributed to the fact that Gemini heads protrude pronouncedly toward the water. Similar to the C12TAB at high surface coverage, small tendency to orient almost perpendicular to the nanotube axis is observed for Gemini molecules in this system.This tendency ismore pronounced at high surface coverage (Figure 2b), 12

where we observe two broad peaks around molecules. The obvious peaks at

for the orientation of two side chains of Gemini

and

indicate somepreference for nearly parallel

orientation on the SWNT at high surface density. In four systems discussed so far, the structures of surfactant aggregates formed on the SWNT are dependent upon both the surface coverage and structure of surfactant. With increasing the surfactant surface coverage or adding a chain to the surfactant structure, the surfactant headgroups alone oralong with some of tail particles protrude morepronouncedly toward the aqueous phase. The protrusion of heads and tails toward the water molecules improve theSWNT isolation from the aqueous environment reflected in the radial distribution function (RDF) profiles of water molecules around the SWNT (see Figure S3 in the Supporting Information) . The results obtained so far propose that the morphology of the C12TAB and [12-212]Br2surfactants aggregates on the nanotube surface are different at comparable surface coverages. The different structure of the C12TAB and [12-2-12]Br2aggregates on the SWNT surface canbe seen more clearly by comparing the radial distributionfunction (RDF) of the tail and head segments of these two surfactants withrespect to the axis of the nanotube at low and high surface coverages.A commonfeature shared by the twoC 12TABtail segment RDF curves in Figures 3a and 3b is thatboth display one strong peak at ~ 0.8 nm, indicating the C12TAB tails adsorb on the SWNT surfaces and form an adsorptionmonolayer as well.Upon increasing the surface coverage, the thickness of this adsorbed monolayer is increased. Most of the C 12TAB heads for both surface coveragesare located near the nanotube surface, next to the monolayer formed by thetail segments (Figures3a and 3b). At low surface coverage, a few head-groups are found away from the SWNT surface, yielding a shoulder at ~ 1.2 nm in Figure 3a. The shoulder becomes stronger and broader at high surface coverage, indicating a greater number of C12TAB 13

heads extend to water.For [12-2-12]Br2surfactant at low surface coverage, we find that two tailgroups are strongly adsorbed on the SWNT as a monolayer,manifesting two strong peaksat ~ 0.8 nmfor the two tails of the Gemini surfactant (Figure 3c).The peaks at ~ 0.9 and 1.2 nm and the shoulders found at~ 1.4 nm in Figure 3c indicate that some of the Gemini head-groups are located near the SWNT surface and some of the heads are positionedaway from the SWNT toward the aqueous phase.At high surface coverage, we observe two obvious peaks at~ 0.8 nm for the two tails of the Gemini surfactant molecules, as illustrated in Figure 3d, which indicate that most of the tail segments remain in contact with the nanotube surface. The two weaker peaks around 1.2 nm clearly show that some of the Gemini tails are positioned fartheraway from the cylindrical axis of the SWNT,as was indicated by thesimulation snapshots of Figure 1d.In this system, the Gemini head-groups are located either near the nanotube and tail segments, yielding the broad peaks at around 0.9 nm, or further away from the SWNT surface (Figure 3d). In the latter case, the peaks at ~ 1.6 nm and the shoulders at ~ 1.9 nm confirm the protrusion of the Gemini heads toward water. The results of our four studied systems suggest thatwith increasing the surfactant surface coverage or adding a chain to the surfactant structure, the morphology of aggregates on the SWNT surface can gradually change. For the C12TAB single-chain surfactant at low surface coverage, most of the surfactant tails and head are located at the same position on the nanotube surface which is in qualitativeagreement with the random adsorption model [34]. While, the simulation results for thedouble-chain surfactant at high surface coverage demonstrate that some of the Gemini molecules are adsorbed on the SWNT such that only the end of their tail particle are in direct contact with the nanotube. In this case, the structure of surfactant aggregates is more consistent with the cylindrical model [38].

14

3.2.SWNT Diameter Effect To examine the influence of the nanotube diameter on the morphology of the surfactant aggregates on the SWNT surface,the self-assembled aggregates formed by either single-chain cationic surfactant (C12TAB) or double-chain cationic surfactant ([12-2-12]Br2) on a (10,14) SWNT were studied at low surface coverage (1.05 molecule/nm2).

Figure 4 shows the

representativesimulation snapshots of both surfactant molecules adsorbed on the SWNTs. For C12TAB adsorption on a (10,14) SWNT, the simulation snapshot of Figure 4 indicates that the surfactant tails either align with the SWNT axis or wrap around the tube, with most of the surfactant heads lying adjacenton the tube surface. The only observable difference between the structure of C12TAB aggregates on the (5,7) and (10,14) SWNTs is that due to a lower energetic cost for the C12TABsurfactant to wrap around a(10,14) nanotube, a larger number of C 12TAB molecules tend to wrap around this nanotube. The comparison of the orientation between the C12TAB and [12-2-12]Br2 surfactants on the (10,14) SWNT reveals that some of the Gemini head-groups are nearly vertically oriented against the nanotube surfaceand extend to the aqueous phase. Note that the orientation of the Gemini heads toward watermolecules can increase theirinteractions with water molecules and improves the nanotubedispersion. To quantify the orientations both of C12TAB and [12-2-12]Br2adsorbed on the (10,14) SWNT surface, we calculated the probability distribution of the angle between the tail-to-head vectors identified by one surfactant molecule and the SWNT axis. The results of the calculations are shown in Figure 5. As can be seen in Figure 5, there is no significant difference in the orientation of C12TAB molecules between a (5,7) SWNT and a (10,14) SWNT at low surface coverage.However, the difference in the orientation of the side chains of [12-2-12]Br2molecules between the two nanotubes ((5,7) and (10,14)) at low surface coverage is considerable. In Figure 15

5, one broad peak around

is observed for each side chains of the Gemini molecules,

suggesting that the tendency of the Gemini chains to orient perpendicular to the axisof the (10,14) SWNT is increased in comparison to the (5,7) nanotube. In addition, the obvious peaks around

and

indicate that the Gemini chains tend to orient almost parallel with respect to

the (10,14) nanotube axis, similar to what has been observed for the (5,7) nanotube. From the results shown in Figures 4 and 5, it is evident that the orientation of the Gemini surfactant aggregates on the (10,14) SWNT at low surface coverage is different from that of C12TAB surfactant. This different orientation may increase the effectiveness of the Gemini surfactants at shielding thenanotube surface from being in contact with water molecules(see Figure S4 in the Supporting Information). Focusing on the adsorbed surfactant molecules, the organization of the C12TAB and [12-212]Br2 surfactants on the (10,14) SWNT surface can be quantified by the RDF profiles of the tail and head segments respect to the SWNT axis. In Figure 6, we show the RDF profiles of the tails and heads of both C12TAB and [12-2-12]Br2surfactants at low surface coverage.A shared feature of the tail segment RDF curves in Figure 6 is that the tail-groups are strongly adsorbed on the nanotube surface, as indicated by the peaks around 1.2 nm (Figures 6a and 6b). Note that the two small peaks at around 0.4 nm in Figure 6barebecauseone of chains of two Gemini molecules enters the interior of theSWNT. The head segments RDF curves for C12TAB surfactant displays one strong peak at ~ 1.3 nm which overlaps with the tail segment profiles, identifyingalmost all head-groups are located near the SWNT surface and at the same position of the tail-groups. In the case of the Gemini surfactant, it is possible to find some Gemini heads near the nanotube surface, but a larger numberof the head-groups arepositioned further from the SWNT surface and protruded toward the water, as indicated by the peaks found at ~ 1.5 and 1.8 nm (Figure 6b). 16

The analysis of the results obtained so farsuggest that only in the case of the Gemini surfactant, the morphology of aggregates formed on the (10,14) SWNT at low surface coverage depends to some extent on the SWNT diameter.The reason for this difference between C 12TAB and [12-2-12]Br2surfactants may be that the Gemini surfactant has the higher hydrophobicity and also the higher charge capacity per single molecule of surfactant which these factors increase the tail-head and head-head repulsions between adsorbed surfactant on the SWNT surface and therefore, some of the surfactant head-groups protrude extensively toward the aqueous phase, effectively pulling some tail particles away from the SWNT surface. It is worth mentioning at this point that the number of the Gemini surfactants adsorbed onto both (5,7) and (10,14) SWNTs at low and high surface converges are either equal orlarger than those for C12TAB surfactant at the end of 100 ns of the simulations (after 100 ns, the number of surfactants adsorbed onto nanotubes remains constant). Because of the higher charge capacity per single molecule of surfactant, the effective surface chargeof Gemini-coated nanotube is expected to be higher than C 12TAB-coated nanotubes. The higher surface charge of Geminicoated nanotube generates a repulsive barrier preventing the nanotubes from aggregating. This result is consistent with the experimental observations [9]. 3.3. Effect of the Spacer Chain Length In this section, we investigate whether or not the length of spacer influences the morphology of surfactant aggregates formed on the (5,7) SWNT under high surface coverage. For this purpose, we position 26 [12-6-12]Br2 surfactants around a (5,7) SWNT, corresponding to a surface coverage 2.27 molecule/nm2. Representative simulation snapshots of [12-612]Br2surfactants adsorbed on the nanotube surface is shown in Figure 7. The snapshot indicates that the two tail-groups of most of the [12-6-12]Br2molecules are rotated away from the 17

nanotube surface such that either their terminal tail particles or some of tail particles are in direct contact with the SWNT while their hydrophilic head-groups are projected towardsthe aqueous phase.However, we note that,some of the [12-6-12]Br2molecules tend to lie on the SWNT surface with their head-groups adsorbed near the nanotube surface.To determine the precise orientations of the two side chains of the [12-6-12]Br2surfactants with respect to the SWNT axis, we calculated the distribution of the angle between the tail-to-head vectors identified by one surfactant molecule (defined in Figure 8) and the SWNT axis. As can be seen in Figure 8, the angle distribution profiles are more distributed for this Gemini surfactant, indicating that the two side chains of the [12-6-12]Br2surfactant tend to form any angle to the nanotube axis. Compared to the [12-2-12]Br2case, the difference in orientation distribution between two side chains of the [12-6-12]Br2surfactants is notable. The reason for this feature in the angle distribution profiles of the [12-6-12]Br2side chains might be that its spacer contains four CH2 groups more than [12-212]Br2and, therefore, is significantly more hydrophobic which causes the side chains of adsorbed [12-6-12]Br2surfactants need to adapt various configurations to accommodate the spacer hydrophobicity. Thus, [12-6-12]Br2surfactants yield more disordered structures as compared to [12-2-12]Br2surfactants ( see Figures 1d and 7).It should be noted that both Gemini surfactants are able to isolate the SWNT from the environment to the same degree (see Figure S5 in the Supporting Information). The results of the C 12TAB, [12-2-12]Br2, and [12-6-12]Br2 adsorption on the (5,7) SWNT at high surface coverage clearly show that the tendency of two side chains of Gemini surfactants to orient perpendicular to the nanotube axis are higher thantheir single-chain homologues; C12TAB. The previous calculations for the potential of mean force between nanotubes in aqueous surfactants solutions have demonstrated that the ability of surfactant heads to orient

18

perpendicularly to the SWNTs axis increases the repulsive forces between the SWNTs which in turn would enhance the stabilization of aqueous dispersions. The results obtained so far indicate the only difference in structure of aggregates formed on the (5,7) nanotube between the two Geminisurfactants at high surface coverage is the fact that the orientations of two side chains of [12-6-12]Br2surfactants are different, whereas the [12-212]Br2sidechains have almost the same orientations. A more detailed picture of the structure of [12-6-12]Br2surfactant aggregates formed onto nanotube may be obtained by analyzingthe RDFs of tail and head segments of [12-6-12]Br2surfactant with respect to the nanotube axis. Figure 9 shows the RDFs of [12-6-12]Br2tails and heads as a function of distance from the SWNT axis at high surface coverage.The RDF profile of the tail segments displays one strong peak at ~ 0.8 nm, indicating the adsorptionmost of the two tail-groups of [12-6-12]Br2 next to the nanotube surface. Note that the thickness of these two peaks is different, confirming their different orientations. Some of the two tail-groups are located further from the nanotube surface, yielding the two broad peaks at around 1.2 and 1.3 nm. The thickness of these two peaks is also different. The RDF profile of the head segments show that a majority of both head-groups are predominantlyfound away from the SWNT surface, with a broad peak at ~ 1.4 nm and a shoulder found at around 1.9 nm. However, some of the head-groups remainnear the nanotube surface, as demonstrated by the first peak at ~ 0.9 nm. The results obtained for [12-6-12]Br2adsorption on the SWNT surface demonstrate that [12-612]Br2surfactant tends to coat the nanotube in a more disordered manner at high surface coverage. This aggregate structure is more consistent with the structurelessrandom adsorption which was proposed by Yurekli et al. [37] to model the SDS adsorption on the CNT surfaceand obviously cannot be described in terms ofthe ordered micellar structuressuggested by Wang and 19

co-workers [9]. It should be noted that our simulation is conducted at low surfactant coverage. As the surfactant density increases,it is likely that a reorientation of [12-6-12]Br2structure happens at the nanotube surface. This possibility can be examined in future work.

4. Conclusions In this paper, the large-scale all-atom molecular dynamics simulationshave been conductedto explore the self-assembly andthe aggregate morphology of three cationic surfactants, one singlechain and two Gemini surfactants, adsorbed on theSWNTs in aqueous solution.The effect of the surfactant structure, surface coverage, SWNT diameter, and the length of the spacer have been examined. Our simulation results have been quantified using representative simulation snapshots,orientation probability distributions, and radial distribution functionsof the water molecules and surfactant tails and heads around the axis of the SWNT. The results have demonstrated that single-chain surfactant C12TAB, particularly at low surfacecoverage, tends to align parallel to the nanotube axis, with the tails and head-groups located almost at same position on the nanotube surface. The random adsorption model can be appropriately used to describe the C12TAB adsorption onto SWNT surface at low surface coverage. Our simulation results have shown that the Gemini surfactant [12-2-12]Br2heads especially at high surface coverage, protrude more pronouncedly toward the aqueous phase, effectively dragging the tail particles away from the SWNT surface such that only their terminal tailparticles are in direct contact with the nanotube. This orientation of [12-2-12]Br2is more consistent with cylindrical model. We have also found that changing the SWNT diameter affects the orientation of [12-2-12]Br2adsorbed on (10,14) SWNT at low surface coverage which may 20

increase the separation of nanotubesbased on diameter. In addition, we have studied the morphology of self-assembled aggregates formed by Gemini surfactant [12-6-12]Br2onto SWNT at high surface coverage to inspect role of the spacer length in adsorption process. Our findings have indicated that the Gemini surfactant [12-6-12]Br2with a longer spacer tends to yield more disordered structures as compared to[12-2-12]Br2. We believe that theresults presented here offer new and valuable insights into the importance of surfactant structure in determining the morphology of surfactantsaggregates formed onto nanotubes. Further, this studymay shed light on the future design of novel Gemini surfactantscapableof more efficientlystabilizing aqueous SWNTsdispersions. Acknowledgment. Wewould like to thank theAmirkabir University of Technology and High Performance Computing Research center for generous allocations ofcomputing time. Supporting Information File: Figures of (i)Variation ofsolvent accessible surface (SAS) area of: (a) the C12TA+ ions, (b) the [12-2-12]+2 ions, and (c) the [12-6-12]+2 ions under different surfactant surface coveragesas a function of simulation time (Figure S1), (ii) The comparison of the simulated radial distribution functions (RDFs) of the surfactant tail and head segments around the CNTbetween the two different initial configurations (Figure S2),(iii)Simulated radial distribution functions (RDFs) of water molecules around the SWNT as a function of the distance from the axis of the nanotube (5,7) in the presence of either aqueous C12TAB or [12-2-12]Br2 surfactants at low and high surface coverages(Figure S3) , (iv) Simulated radial distribution functions (RDFs) of water molecules around the SWNT as a function of the distance from the axis of the nanotube (10,14) in the presence of either aqueous C12TAB or [12-2-12]Br2 surfactants at low surface coverage (Figure S4), and (v) Simulated radial distribution functions (RDFs) of water molecules around the SWNT as a function of the distance from the axis of the 21

nanotube (5,7) in the presence of either aqueous [12-6-12]Br2 or [12-2-12]Br2 surfactants at high surface coverage(Figure S5).

22

References [1] J. R. Alves da Cunha, C. Fantini, N. F. Andrade, P. Alcantara, Jr., G. D. Saraiva, A. G. Souza Filho, M. Terrones, and M. C. dos Santos, Enhanced Solubilization of Carbon Nanotubes in Aqueous Suspensions of Anionic–Nonionic Surfactant Mixtures, Journal of Physical Chemistry C 117 (2013) 25138-25145. [2] C. Backes, A. Hirsch, Noncovalent Functionalization of Carbon Nanotubes, in Chemistry of Nanocarbons, T. Akasaka, F. Wudl and S. Nagase, Eds, John Wiley & Sons, Ltd, Chichester, UK, (2010) DOI: 10.1002/9780470660188.ch1. [3] M. Suttipong, N. R. Tummala, A. Striolo, C. S. Batista, J. Fagan, Salt-specific Effects in Aqueous Dispersions of Carbon nanotubes, Soft Matter 9 (2013) 3712-3719. [4] N. Karousis, N. Tagmatarchis, D. Tasis, Current Progress on the Chemical Modification of Carbon Nanotubes, Chemical Review 110 (2010) 5366-5397. [5] M. Suttipong, N. R. Tummala, B. Kitiyanan, A. Striolo, Role of Surfactant Molecular Structure on Self-Assembly: Aqueous SDBS on Carbon Nanotubes, Journal of Physical Chemistry C 115 (2011) 17286-17296. [6] Y. Tan, D. E. Resasco, Dispersion of Single-Walled Carbon Nanotubes of Narrow Diameter Distribution, Journal of Physical Chemistry B 109 (2005) 14454-14460. [7] L. Liu, X. Fei, S. Zhu, L. Yu, and B. Zhang, Self-Assembly of Anionic Gemini Surfactant: Fluorescence Resonance Energy Transfer and Simulation Study,Langmuir 29 (2013) 5132-5137. [8] A. Di Crescenzo, R. Germani, E. Del Canto, S. Giordani, G. Savelli, and A. Fontana, Effect of Surfactant Structure on Carbon Nanotube Sidewall Adsorption, European Journal of Organic Chemistry 2011(2011) 5641–5648. [9] Q. Wang, Y. Han, Y. Wang, Y. Qin, and Z-X. Guo, Effect of Surfactant Structure on the Stability of Carbon Nanotubes in Aqueous Solution, Journal of Physical Chemistry B 112 (2008) 7227-7233. [10] J. A. S. Almeida, H. Faneca, R. A. Carvalho, E. F. Marques, A. A. C. C. Pais, DicationicAlkylammonium Bromide Gemini Surfactants. Membrane Perturbation and Skin Irritation, PLoS ONE 6 (2011), e26965. DOI:10.1371/journal.pone.0026965. [11] Y. Liu, L. Yu, S. Zhang, J. Yuan, L. Shi, L. Zheng, Dispersion of Multiwalled Carbon Nanotubes by Ionic Liquid-type Gemini Imidazolium Surfactants in Aqueous Solution, Colloids and Surfaces A: Physicochemical and Engineering Aspects 359 (2010) 66–70.

23

[12] M. Bagonluri, M. Foldvari, Dispersion and Debundling of Carbon Nanotubes (CNTs) Using Pharmaceutical Excipients.

2007 AAPS Annual Meeting and Exposition, San Diego, USA,

November (2007) 11– 15. [13] C. Lamprecht , J. T. Huzil, M. V. Ivanova, and M. Foldvari, Non-Covalent Functionalization of Carbon Nanotubes with Surfactants for Pharmaceutical Applications – A Critical Mini-Review, Drug Delivery Letters 1(2011) 45-57. [14] P. Angelikopoulos, H. Bock, Directed Self-Assembly of Surfactants in Carbon Nanotube Materials, Journal of Physical Chemistry B 112 (2008) 13793–13801. [15] N. R. Tummala, A. Striolo, SDS Surfactants on Carbon Nanotubes: Aggregate Morphology, ACS Nano 3 (2009) 595-602. [16] Z. J. Xu, X. N. Yang, Z. Yang, A Molecular Simulation Probing of Structure and Interaction for SupramolecularSodium dodecyl sulfate/Single-wall Carbon Nanotube Assemblies, Nano Letter 10 (2010) 985-991. [17] S. Lin, D. Blankschtein, Role of the Bile Salt Surfactant Sodium Cholate in Enhancing the Aqueous Dispersion Stability of Single-Walled Carbon Nanotubes: A Molecular Dynamics Simulation Study, Journal of Physical Chemistry B 114 (2010) 15616–15625. [18] N. R. Tummala, B. H. Morrow, D. E. Resasco, A. Striolo, Stabilization of Aqueous Carbon Nanotube Dispersions Using Surfactants: Insights from Molecular Dynamics Simulations. ACS Nano 4 (2010) 7193–7204. [19] M. Suttipong, N. R. Tummala, A. Striolo, C. S. Batista, J. Fagan, Salt-specific Effects in Aqueous Dispersions of Carbon Nanotubes. Soft Matter 9 (2013) 3712-3719. [20] N. Poorgholami-Bejarpasi, B. Sohrabi, Self-assembly of Cationic Surfactants on the Carbon Nanotube Surface: Insights from Molecular Dynamics Simulations, Journal of Molecular Modeling 19 (2013) 4319–4335. [21] B. Hess, C. Kutzner, D. van der Spoel, E. Lindahl GROMACS 4:Algorithms for Highly

Efficient, Load-balanced, and Scalable Molecular Simulation, Journal of Chemical Theory and Computation 4 (2008) 435-447. [22] J. T. Frey, D. J. Doren, TubeGen 3.2; University of Delaware: Newark, DE, 2003. [23] W. L. Jorgensen, D. S. Maxwell, J. Tirado-Rives, Development and Testing of the OPLS Allatom Force Field on Conformational Energetics and Properties of Organic Liquids, Journal of the American Chemical Society 118 (1996) 11225-11236. [24] H. J. C. Berendsen, J. R. Grigera, T. P. Straatsma, The Missing Term in Effective Pair Potentials, Journal of Physical Chemistry 91(1987) 6269-6271.

24

[25] E. Vanquelef, S. Simon, G. Marquant, E. Garcia, G. Klimerak, J. C. Delepine, P. Cieplak, F-Y. Dupradeau, R.E.D. Server: A Web Service for Deriving RESP and ESP Charges and Building Force Field Libraries for New Molecules and Molecular Fragments, Nucleic Acids Research 39 (2011) W511-517. [26] F-Y. Dupradeau, A. Pigache, T. Zaffran, C. Savineau, R. Lelong, N. Grivel, D. Lelong W. Rosanski, P. Cieplak, The R.E.D. Tools: Advances in RESP and ESP Charge Derivation and Force Field Library Building, Physical Chemistry Chemical Physics 12 (2010) 7821-7839. [27] C. I. Bayly, P. Cieplak, W. Cornel, P. A. Kollman, A Well-behaved Electrostatic Potential Based Method Using Charge Restraints for Deriving Atomic Charges: the RESP Model, Journal of Physical Chemistry 97 (1993) 10269-10280. [28] M. W. Schmidt, K. K. Baldridge, J. A. Boatz, S. T. Elbert, M. S. Gordon, J. H. Jensen, S. Koseki, N. Matsunaga, K. A. Nguyen, S. Su, T. L. Windus, M. Dupuis M, J. A. Montgomery, General Atomic and Molecular Electronic Structure System, Journal of Computational Chemistry 14(1993) 1347-1363. [29] R. W. Hockney, S. P. Goel, J. W. Eastwood, Quiet High Resolution Computer Models of a Plasma, Journal of Computational Physics 14(1974) 148-158. [30] L. Verlet, Computer Experiments on Classical Fluids. I.ThermodynamicalProperties of LennardJones Molecules, Physical Review 159(1967) 98-103. [31] G. Bussi, D. Donadio, M. Parrinello, Canonical Sampling through Velocity Rescaling, Journal of Chemical Physics 126 (2007) 014101-1-014101-7. [32] M. Parrinello, A. R. Rahman, Polymorphic Transitions in Single Crystals: A New Molecular Dynamics Method, Journal of Applied Physics 52 (1981) 7182-7190. [33] T. Darden, D. York, L. Pedersen, Particle Mesh Ewald: An N-log(N) Method for EwaldSums in Large Systems, Journal of Chemical Physics 98(1993) 10089-10092. [34] U. Essmann, L. Perera, M. L. Berkowitz, T. Darden, H. Lee, L. G. Pedersen, A Smooth Particle Mesh EwaldMethod, Journal of Chemical Physics 103(1995) 8577-8592. [35] A. R. Leach, Molecular Modelling: Principles and Applications, 2nd ed Prentice Hall Harlow, (2001) England. [36] W. Humphrey, A. Dalke, K. Schulten, VMD - Visual Molecular Dynamics, Journal of Molecular Graphics 14(1996) 33-38. [37] K. Yurekli, C. A. Mitchell, R. Krishnamoorti, Small-angle Neutron Scattering from SurfactantAssisted Aqueous Dispersions of Carbon Nanotubes, Journal of the American Chemical Society 126(2004) 9902-9903.

25

[38] O. Matarredona, H. Rhoads, Z. R. Li, J. H. Harwell, L. Balzano, D. E. Resasco, Dispersion of Single-walled Carbon Nanotubes in Aqueous Solutions of the anionic Surfactant NaDDBS, Journal of Physical Chemistry B 107 (2003) 13357-13367.

Table1. Simulation Details for the Systems Studied in This Work System

SWNT

N C12TAB

1

(5,7)

2

(5,7)

3

(5,7)

4

(5,7)

5

(10,14)

6

(10,14)

7

(5,7)

N [12-2-12]Br2

N [12-6-12]Br2

Nwater

N total atoms

12

11002

34054

26

10749

34009

12

10837

34147

26

10390

34206

10581

33839

10222

33938

10267

34149

24 24 26

26

27

28

Figure 1. Representative simulation snapshots of a (5,7)SWNT covered with surfactants, showing the surface structures of the two surfactants at two different surface packing densities considered here. (a) C12TAB at 1.05 molecule/nm2, (b) C12TAB at 2.27 molecule/nm2, (c) [12-2-12]Br2 at 1.05 molecule/nm2, and (d) [12-2-12]Br2 at 2.27 molecule/nm2. The three plots on the right are side views, and the three plots on the left are corresponding front views. Water molecules are not shown for clarity. Color code: blue,nitrogen; yellow, bromide counterion; cyan, carbon; white, hydrogen;gray, carbon atoms in the SWNT. All snapshots are at the 110 ns in the MD simulations.

29

Figure 2 . Simulated probability distribution of the angle formed between the vector of the surfactant tailto-head and the SWNT axis (a) C12TAB and (b) [12-2-12]Br2. In the C12TAB and [12-2-12]Br2 molecular structures shown, the dotted lines connecting the carbon atom in surfactant tails with the nitrogen atom in surfactant heads define the surfactant tail-to-head vectors.

30

a

b

31

c

d

Figure 3. Simulated radial distribution functions (RDFs) of surfactant tail and head segments

around the SWNT as a function of the distance from the axis of the tube (a) C 12TAB at low

32

surface coverage, (b) C 12TAB at high surface coverage, (c) [12-2-12]Br2 at low surface coverage, and (d) [12-2-12]Br2 at high surface coverage.

Figure 4. Representative simulation snapshots of a (10,14)SWNT covered with surfactants, showing the surface structures of the two surfactants at low surface packing density. (a) C12TAB at 1.05 molecule/nm2, and (b) [12-2-12]Br2 at 1.05 molecule/nm2. The two plots on the right are side views, and the two plots on the left are corresponding front views. Water molecules are not shown for clarity. The color code is the same as that used in Figure 1. All snapshots are at the 110 ns in the MD simulations.

33

Figure 5. Simulated probability distribution of the angle formed between the vector of the surfactant (C12TAB or [12-2-12]Br2) tail-to-head and the (10,14) SWNT axis at low surface coverage. The tail-tohead vectors of the surfactants are defined in Figure 2.

34

a

b

Figure 6. Simulated radial distribution functions (RDFs) of surfactant tail and head segments around the SWNT as a function of the distance from the axis of the (10,14 ) nanotube (a) C12TAB, and (b) [12-2-

12]Br2 at surface packing densities 1.05 molecule/nm2.

35

Figure 7. Representative simulation snapshots of a (5,7)SWNT covered with [12-6-12]Br2, showing the surface structures of the [12-6-12]Br2 at high surface coverage (2.27 molecule/nm2). The plot on the right is side view, and the plot on the left is corresponding front view. Water molecules are not shown for clarity. The color code is the same as that used in Figure 1. All snapshots are at the 110 ns in the MD simulations.

36

Figure 8. Simulated probability distribution of the angle formed between the vector of the [12-6-12]Br2 tail-to-head and the SWNT axis at high surface coverage. In the [12-6-12]Br2 molecular structure shown, the dotted lines connecting the carbon atom in surfactant tails with the nitrogen atom in surfactant heads define the surfactant tail-to-head vectors.

37

Figure 9.Simulated radial distribution functions (RDFs) of the [12-6-12]Br2 tail and head segments around the SWNT as a function of the distance from the axis of the (5,7) nanotube at high surface coverage.

38