Optimizing Surfactant Concentrations for Dispersion of Single

It has been reported that the dispersion efficiencies of some surfactants are ... diameters compared to arc nanotubes (d ∼ 1.4 ± 0.2 nm),(17) diffe...
0 downloads 0 Views 4MB Size
Download PDF
J. Phys. Chem. B 2010, 114, 9805–9811

9805

Optimizing Surfactant Concentrations for Dispersion of Single-Walled Carbon Nanotubes in Aqueous Solution Adam J. Blanch, Claire E. Lenehan, and Jamie S. Quinton* School of Chemical and Physical Sciences, Flinders UniVersity, GPO Box 2100, Adelaide, SA 5001, Australia ReceiVed: May 5, 2010; ReVised Manuscript ReceiVed: June 16, 2010

The sonication-centrifugation technique is commonly used for dispersing single-walled carbon nanotubes (SWCNTs) in aqueous surfactant solutions. However, the methodologies and materials used for this purpose are widely varied, and few dispersive agents have been studied systematically. This work describes a systematic study into the ability of some well-known (and some less common) surfactants and polymers to disperse SWCNTs fabricated by two different techniques. UV-vis-NIR absorbance spectra of their supernatant solutions showed that the smaller ionic surfactants were generally more effective dispersants, with larger polymer and surfactant molecules exhibiting a reduced performance for ensembles of carbon nanotubes of smaller average diameter. Optimal surfactant concentrations were established for dispersions of carbon nanotubes produced by the electric arc method in aqueous solutions of sodium dodecylbenzene sulfonate, sodium deoxycholate, Triton X-405, Brij S-100, Pluronic F-127, and polyvinylpyrrolidone. This optimum value was determined as the point at which the relative concentration of nanotubes dispersed is maximized, before flocculation-inducing attractive depletion interactions begin to dominate. The aggregation state of carbon nanotubes dispersed in sodium dodecylbenzene sulfonate was probed by AFM at different stages of rebundling, showing the length dependence of these effects. Introduction Carbon nanotubes (CNTs) have long been recognized as a promising material for a variety of applications due to their remarkable properties.1 However, their implementation has been hampered by the intractability and inhomogeneity of “as produced” CNT soot, where the CNTs adhere to each other strongly in ropes or bundles through van der Waals forces.2 The CNTs within this soot typically exhibit a broad distribution of both diameter and length. Furthermore, a high level of impurities such as non-nanotube carbonaceous material and residual metal catalyst particles from the synthesis process are often present.3–6 Consequently, difficulties are often encountered when attempting to purify CNT mixtures, first in isolating nanotubes from the impurities, and second in separating nanotubes possessing uniform properties from the ensemble. Many methods exist to purify raw CNT material, though most require an oxidative treatment that may damage or destroy at least a portion of the nanotubes or append functional groups to their sidewalls, thus disrupting the nanotube’s electronic structure.3 The majority of separation techniques to sort nanotubes by length, diameter, or electronic type require the nanotubes to be dispersed, preferably as individuals.7 As CNTs are inherently insoluble in water, aqueous dispersion requires modification of the CNT surface, and methodologies toward this end may be grouped into either chemical (covalent) or physical (noncovalent) approaches.8 Surface functionalization of CNTs has been shown to be effective for dispersion in both aqueous and organic media,9 though the intrinsic electronic and mechanical properties of the nanotubes are adversely affected.10 Dispersion via noncovalent methods has the advantage of preserving the nanotubes conjugated π system and hence their electrical properties, and CNTs have been successfully dispersed * To whom correspondence [email protected].

should

be

addressed.

E-mail:

in water with the aid of a vast number of different dispersive agents, such as surfactants,11–20 polymers,21–23 and singlestranded DNA,24,25 among others.26,27 Dispersion of the CNTs in such media is usually achieved via ultrasonic processing, followed by a consecutive (ultra)centrifugation step. Centrifugation removes the remaining larger CNT bundles, leaving primarily individual nanotubes and small bundles in the supernatant.28 This process has the added advantage of purifying the CNT soot by removing the heavier metal catalyst particles and larger non-nanotube carbonaceous material from the solutions of unpurified nanotube soot.3,5,6 The mechanism for dispersion is expected to be primarily due to hydrophilic and hydrophobic interactions, where attraction between the CNT surface and the surfactant’s hydrophobic segment facilitates adsorption while the hydrophilic group associates with water.10 Dispersions involving ionic surfactants are believed to be stabilized by electrostatic repulsion between the hydrophilic head groups, and both cationic and anionic surfactants are able to adequately disperse CNTs with neither showing superiority;8 however, a recent study by Xu et al. reports that the counterion essentially balances the electrostatic forces.29 The assembly of short chained amphiphilic surfactants around a CNT has been proposed to occur through encapsulation in cylindrical micelle,28,30 adsorption of hemimicelles,19,31 or random adsorption,32 though recent simulations suggest that all of these conformations are possible and are dependent on the surfactant concentration33,34 and nanotube diameter.29 The formation of micelles is not a prerequisite for CNT dispersion and many surfactants function as dispersants below their critical micelle concentration (CMC);35 however, the performance is greatly improved at concentrations above the CMC.36 Long-chained polymers and DNA are known to wrap around a CNT helically,25,37 though a nonwrapping polymer conformation is also possible,38 while block copolymers are more likely to decorate the CNT along

10.1021/jp104113d  2010 American Chemical Society Published on Web 07/12/2010

9806

J. Phys. Chem. B, Vol. 114, No. 30, 2010

its sidewall.23 For nonionic surfactants and polymers, steric hindrance prevents nanotube agglomeration.23,38 It has been reported that the dispersion efficiencies of some surfactants are sensitive to the diameter of the nanotubes.39 A large number of previous dispersion studies in the literature focus on CNTs formed by high pressure disproportionation of CO (HiPCO)14–16,18,19 or the similarly produced Co- and Mocatalyzed CoMoCAT nanotubes,13,40 with those produced by the electric arc and laser-vaporization methods studied to a lesser extent.17,19 As HiPCO (d ≈ 1 ( 0.3 nm)41 CNTs tend to have smaller diameters compared to arc nanotubes (d ∼ 1.4 ( 0.2 nm),17 differences in the effectiveness of each dispersive agent between nanotube types should be expected. Indeed, it has previously been shown that sodium cholate (SC) and sodium dodecyl sulfate (SDS) preferentially stabilize nanotubes of smaller diameter,39,42 and these surfactants have been utilized for CNT separation studies through density gradient ultracentrifugation.43 UV-vis-NIR absorbance spectroscopy is a common technique for characterizing CNT dispersions and has the capacity to probe all species of nanotube simultaneously. Absorbance spectra can provide a simple and rapid measure of the relative dispersion state through analysis of the characteristic nanotube absorbance peaks.13,40 These optical resonances were initially thought to arise from transitions between van Hove singularities in the valence and conductive bands of the nanotube’s electronic density of states,44 though such absorptions have been shown to be excitonic in nature.45 The absorbance spectrum of a nanotube ensemble (with a range of diameters) exhibits convoluted individual peaks that can be grouped according to their corresponding single-particle transition, ES,M ii , where i is the order of the allowed transition from the conducting band (ci) to the valence band (vi), and S and M denote transitions for nanotubes possessing semiconducting and metallic characteristics, respectively.44 Well-dispersed nanotube solutions exhibit distinct nanotube absorption peaks of greater intensity, with peaks from individual species becoming more resolved as the nanotubes are further exfoliated. Broader, red-shifted peaks are obtained for more bundled suspensions.28 In this work, a number of common surfactants were employed using relatively standard sample preparation procedures to directly compare between HiPCO and arc type nanotubes, and thus examine the relationship between the type of surfactant and the CNT diameter distribution. Although it has been demonstrated that optimal concentrations exist for nanotube dispersion,46,47 few surfactants have been examined to such an extent.11,15 Optimal surfactant concentrations for dispersion were therefore determined from analysis of UV-vis-NIR absorbance spectra for those surfactants that performed well for arc type CNTs. Experimental Methods Surfactants Brij S-100 (previously known as Brij 700), didodecyldimethylammonium bromide (DDAB), Pluronic F-127, polyvinylpyrrolidone (PVP-10 MW ≈ 10 000; PVP-55 MW ≈ 55 000), sodium dodecylbenzene sulfonate (SDBS), sodium dodecylsulfate (SDS), sodium cholate (SC), sodium deoxycholate (DOC), Triton X-100, Triton X-405, and Tween 60 were purchased from Sigma-Aldrich (Sydney, Australia) and used as received. Hexadecyltrimethylammonium bromide (or cetyl trimethylammonium bromide, CTAB) was obtained from Ajax (Sydney, Australia). Structures of these materials are shown in Figure S1 in the Supporting Information. As-produced electric arc (Carbon Solutions, Riverside, CA) or HiPCO (Carbon

Blanch et al. Nanotechnologies Inc., Houston, Texas, USA) CNTs were dispersed in 1% by mass (i.e., mass surfactant per total mass, 1 wt %; unless otherwise specified) aqueous surfactant solutions via 10 min of tip ultrasonication using a Sonics VCX 750W (Sonics, Newtown, CT) operating at 20 kHz with a 5 mm Ti microtip set to 20% of the maximum amplitude. Samples were cooled with ice water during exposure to ultrasound and were ultracentrifuged directly after sonication for 1 h at fixed angle at 43 000 rpm (∼122000g) in a Type 50 Ti rotor mounted in an Optima L-100XP ultracentrifuge (Beckman-Coulter, Sydney, Australia), with the upper ∼70-85% of the supernatant collected via pipet. UV-vis-NIR spectra were recorded on a Cary 5G spectrophotometer (Varian, Melbourne, Australia) at 600 nm/min with use of 10 mm path length quartz cuvettes and a surfactant solution baseline correction. Analysis of the spectra, including peak background subtractions and area integration calculations, was performed with IGOR Pro software (Wavemetrics, Inc., Portland, OR). For atomic force microscopy (AFM), 25 µL aliquots of the CNT solution were deposited onto Si wafers with a WS-400-6NPP-LITE spin coater (Laurell Technologies, North Wales, PA) operating at 1000 rpm and allowed to dry. Samples were washed with copious amounts of water to remove residual surfactant. Images were collected with a multimode head and Nanoscope IV controller (Digital Instruments, Veeco, Santa Barbara, CA) operating in tapping mode using a Si tip (NSC15, Mikromasch, San Jose, CA) under ambient conditions. Results and Discussion UV-vis-NIR absorbance spectra of both arc and HiPCO CNTs dispersed in 1% solutions of each surfactant are shown in Figure 1. A surfactant concentration of 1% was used as this is the most common approach reported in literature, likely since this was established as the best concentration for both SDS47 and Triton X-100.46 The concentration of nanotubes remaining in the supernatant after centrifugation was greater in most cases for HiPCO CNTs. This is likely to arise from two factors: (a) the lower purity of the raw arc nanotubes, having less CNT per weight of the starting material, and (b) the larger diameter arc nanotube/surfactant assemblies are expected to be heavier, leading to a faster sedimentation rate and hence less material remaining after centrifugation under the same conditions. It is clearly evident from the magnitude and observed resolution of the CNT peaks in the absorbance spectra that the amount and quality of the dispersion achieved by the different surfactants can vary markedly. As can be seen in panels A and B of Figure 1, the smaller anionic surfactants SDBS, DOC, and SC produce the most resolved spectra for both sets of nanotubes, thus indicating a greater fraction of individual nanotubes in the dispersion. While SDS performed well for the HiPCO nanotubes, fine structure for arc CNTs suspended in SDS was only observed immediately after centrifugation. This structure disappeared within 24 h of standing, thus reaggregation of the nanotubes seems to begin directly after sonication ceases48 and is quite rapid compared to that of the other surfactants (Supporting Information, Figure S2). Furthermore, it has previously been suggested that SDS preferentially dissolves impurities over CNTs,17 and hence the use of SDS to disperse arc nanotubes in particular is not recommended. For SDBS, the presence of a phenyl group in the surfactant is suggested to provide superior dispersive ability due to π-π stacking interactions,19,40 despite being at the hydrophilic end of the molecule. It is thought that the phenyl group may play a role in the initial separation of an individual nanotube from a bundle,

Dispersion of SWCNTs in Aqueous Solution

J. Phys. Chem. B, Vol. 114, No. 30, 2010 9807

Figure 1. UV-vis-NIR absorbance spectra for arc (left) and HiPCO (right) type CNTs dispersed in water by sonication-ultracentrifugation using 1% solutions of various anionic (A, B), cationic (C, D) and nonionic (E, F) surfactants. Asterisks indicate spectra reduced by 1/3 scalar multiplication for ease of comparison. Spectra are offset for clarity; see the Supporting Information for offset values.

adsorbing laterally in the narrow space between adjacent nanotubes where the surfactant cannot adsorb perpendicular to the nanotube surface.49 The absence of a hydroxyl group in DOC results in better dispersion of CNTs than with SC, which has been noted previously,13,17 though the difference is slight in these results. This hydroxyl group likely hinders the association of the SC molecule to the CNT sidewall through hydrophobic attraction in some way, thus its removal provides DOC with improved dispersive ability. DOC and SDBS seem to be comparable for both types of nanotubes, and dispersion is possible in cosurfactant mixtures of the two. However, when in competition, it appears that DOC preferentially adsorbs to the nanotubes over SDBS (see Figure S3 in the Supporting Information), probably due to stronger hydrophobic interactions with the linked cyclic rings in DOC compared to the single alkyl chain of SDBS. As can be seen in panels C and D of Figure 1, the cationic surfactants CTAB and DDAB are not as effective as DOC or SDBS. While CTAB resulted in a relatively large absorbance peak area, the individual absorptions for both arc and HiPCO nanotubes were not as well resolved as those for DOC or SDBS (Figure 1A,B), suggesting a lower fraction of individual tubes. Additionally, CTAB solutions were observed to form large crystalline aggregates during processing. The formation of this phase occurred in both the absence and presence of the CNTs and may hinder dispersion; however, these aggregates were completely removed upon centrifugation and the resulting spectra suggest CNTs remain in the supernatant. On the basis of the intensity of the spectra DDAB provides slightly fewer dispersed CNTs as CTAB for arc nanotubes and approximately 1 /3 the amount of HiPCO nanotubes, thus CTAB is clearly the superior dispersant of the two. In contrast to CTAB, where the crystallites formed were large and sedimented from the solution under gravity, DDAB solutions turned turbid over time. This

turbidity also occurred in the surfactant reference solution and hence is believed to be due to surfactant coagulation, which increases scattering in the UV-vis-NIR spectra as well as destabilizing the CNT dispersion. Of the nonionic dispersants (Figure 1E,F), the block copolymer Pluronic F-127 and longer linear polymers PVP-10 and PVP-55 appear to perform far better for arc CNTs than for HiPCO nanotubes, perhaps due to tighter “wrapping” or adsorption conformations being required for smaller diameter tubes. No CNTs were retained in the supernatant for PVP-10, making it the least effective surfactant for HiPCO nanotubes of the selection studied; however, the longer PVP-55 appears to disperse at least a small amount of HiPCO nanotubes. Similarly, Triton-X 100 was a relatively poor dispersant for both sets of CNTs (though especially ineffective for HiPCO tubes), while Triton X-405 dispersed both types reasonably well. This result supports the assertion that longer chain lengths are more effective for nonionic surfactants due to increased steric repulsion, regardless of the nanotube diameter.18 Brij S-100 also dispersed both types of nanotubes relatively well, though the resolution of the CNT peaks is slightly better for arc nanotubes with this surfactant. Again, the large molecule may be restricted in binding to a nanotube with a higher degree of curvature. Similarly, Tween 60 disperses arc CNTs to a greater degree, though the difference in the retained CNT concentration between tube types is much more pronounced, which accounts for the discrepancy observed in previous examinations of the Tween surfactant series for arc17 and HiPCO18 nanotubes. As with DDAB, the Tween 60 solution was found to turn turbid, which led to poor stability of the dispersion over time. It is probable that the overall enhanced ability of smaller surfactant molecules to disperse the CNTs is related to the debundling mechanism, postulated to occur through an “unzippering” of the CNT bundle during ultrasonic agitation.49 The

9808

J. Phys. Chem. B, Vol. 114, No. 30, 2010

Figure 2. UV-vis-NIR absorbance spectrum of arc type nanotubes dispersed in SDBS showing regions analyzed. CNT peak areas (solid shaded) were used as a relative measure of concentration, while the ratio of the peak to the subtracted background area (line shaded) was also calculated.

bulkier surfactant/polymer molecules would undoubtedly find it more difficult to enter the gaps between adjacent tubes and exfoliate the CNTs. As SDBS and DOC were established to be the most efficient dispersants, three separate sample series were produced at different initial CNT loadings to determine the optimal surfactant concentration for dispersing arc-type nanotubes. For UV-vis-NIR S M (S22) and E11 (M11) peak analysis the peak areas of the E22 groups (centered at ∼1000 and ∼689 nm for SDBS, respectively) were used as measures of the relative amount of nanotubes dispersed. Measurement of the maximum absorbance intensity only was avoided as it was observed to lead to exaggerated relative concentrations with the surfactants of larger

Blanch et al. molecular weight. This is likely due to an increasing surfactant content resulting in an increase in the solution viscosity, and hence a greater amount of the non-nanotube carbonaceous material remains in the supernatant for a fixed set of centrifugation parameters. Areal peak absorbance was used to negate this effect, and was calculated by subtraction of a linear background as shown in Figure 2. This linear subtraction removes absorption due to the carbon π-plasmon, residual carbon particulates, and other components that contribute to the background.4,50 Both S22 and M11 peaks were analyzed; however, since the CNT raw product contains approximately 2/3 semiconducting and 1/3 metallic nanotubes, we focus on the semiconducting S22 peak. The M11 areal absorbance trends are provided in Figure S7 of the Supporting Information for each of the different surfactants used. The S22:BG “resonance ratio” (the ratio of the nanotube peaks to the nonresonant background) was used as another indicator of the dispersion state of the solution. This value has been previously used as a relative measure of dispersion state13,40 and purity4 in arc CNTs. A larger such ratio may suggest a greater proportion of nanotubes suspended compared to nonnanotube material (or a less aggregated sample), as the absorbance of the nanotube peaks is expected to increase as the nanotubes become more disperse. Panels A and B of Figure 3 show the amount of CNT remaining in the supernatant (relative to the S22 areal absorbance) as a function of surfactant concentration for DOC and SDBS, respectively. Panels C and D of Figure 3 show the same data as a function of the surfactant:CNT mass ratio, using the weight of the raw CNT material (thus including all impurities). Some amount of scatter is obtained, particularly for low DOC concentrations and a starting CNT loading of 0.25 mg mL-1, which may be related either to the inhomogeneity of the raw

Figure 3. Relative amounts of arc CNTs remaining in the supernatant as a function of surfactant concentration for DOC (A) and SDBS (B) for three different initial CNT loadings, as measured by the S22 absorbance peak. The same data are reproduced as a function of the ratio of surfactant to CNT material by mass for DOC (C) and for SDBS (D). Separate scales are used for each series for ease of comparison.

Dispersion of SWCNTs in Aqueous Solution

J. Phys. Chem. B, Vol. 114, No. 30, 2010 9809

Figure 4. Relative arc CNT concentrations remaining in the supernatant as a function of surfactant concentration, measured via the S22 peak area of the absorbance spectra for six different dispersants. The ratio of the S22 peak area to the subtracted background absorbance is also shown in each case. Initial raw CNT mass loading was 0.75 mg mL-1 for all samples.

material or to mass errors from measuring such a low weight of CNTs. The maximum amount of nanotubes remaining is evidently more correlated with the absolute surfactant concentration in both cases,15,46 rather than the ratio of surfactant to CNT material,19,20 though only a small range of CNT concentrations has been examined. The changes in relative CNT concentration for arc nanotubes dispersed in four additional surfactants, Brij S-100, Triton X-405, Pluronic F-127, and PVP-55, are displayed Figure 4. The initial CNT loading was 0.75 mg mL-1 in each case. An optimal surfactant concentration is clear from the S22 areal absorbance for each dispersant except for PVP-55, where the apparent CNT concentration increases with surfactant concentration up to 17%, at which point absorbance is saturated. However, PVP-55 exhibits a decreasing S22:BG ratio above 3%, i.e., the background absorbance is increasing at a greater rate than that of the CNT peaks. Thus, it is likely that the cause of the increase in CNT concentration is not that more surfactant is available for dispersion, but rather it is due to the increase in solvent viscosity, resulting in more material (including large bundles and non-nanotube carbon particles) being retained in the supernatant for a fixed centrifugation speed and time. This effect was noted also for the larger surfactants TX-405 and Brij S-100 and is the reason for using the S22 peak area instead of the measured absorbance intensity. It is likely more pronounced in this case as PVP-55 is a long-chained linear polymer possessing a molecular weight five times greater than that of the next largest surfactant analyzed, which would increase the solvent viscosity at a faster rate. Therefore, the optimal dispersion concentration for PVP-55 is taken as the inflection in S22 peak area at 3%, corresponding to the maximum in the S22:BG ratio. Note that the break between 8% and 9% PVP55 (and 4-5% for Pluronic F-127) arose from two sample sets

Figure 5. UV-vis-NIR absorbance spectra of arc CNTs dispersed in 0.5%, 2%, and 4% SDBS before and after centrifugation (A). Spectra of sonicated samples are diluted 1:10 and scaled by 0.5 for comparison. An image of the diluted, uncentrifuged solutions after several weeks is shown (B) and an enhanced high contrast image of the highlighted segment (C). Visible sedimentation of CNTs is negligible for 0.5%, significant for 2%, and dominant for 4% SDBS.

9810

J. Phys. Chem. B, Vol. 114, No. 30, 2010

Blanch et al.

Figure 6. AFM height images (8 × 8 µm) of arc CNTs spin coated onto Si wafers from dispersions of 0.5% (A), 2% (B), and 3% (C) SDBS.

being centrifuged separately due to limited rotor slots, and a small amount of systematic deviation occurs between centrifugation runs. For the other surfactants examined, the S22:BG resonance ratio displays similar behavior to the S22 peak trend. From these results, the optimal surfactant concentrations for arc CNT dispersion are approximately 1.6% (DOC), 0.5% (SDBS), 3% (Triton X-405), 2% (Brij S-100), 5% (Pluronic F-127), and 3% (PVP-55). The value of 5% for Pluronic F-127 is similar to a previous result, where multiwalled CNTs were found to be less dispersed by this surfactant at around 5% (w/v).51 For SDBS, varied optimal concentrations have been reported, such as ∼0.09%,30 ∼0.74%19 (considering a ratio of 1:10 CNT to surfactant under the conditions used here), and as above ∼1.37%.36 In the determination of 0.09%, a significantly lower starting CNT concentration was used, which may account for the discrepancy, although the range examined did not exceed 0.45% and no peak was observed. The value of 0.74% agrees quite well with our results, falling within the peak range of areal absorbance values between ∼0.3% and 0.8%, while the suggestion that nanotubes are individualized only at concentrations greater than 1.37% does not account for the reduced dispersion obtained at higher surfactant content. More recently, the amount of CNTs remaining has been shown to fall away sharply after reaching around 1.5% SDBS for HiPCO nanotubes,15 with a similar trend. To our knowledge, although dispersions have been obtained by using a range of concentrations, optimal values for the other surfactants have not been previously reported. The CNT concentration remaining in the supernatant is thought to decrease at higher surfactant concentrations due to attractive depletion interactions.46,47,52 Simulations with rods (CNTs) and spheres (surfactant micelles) have shown these effects to depend on the length of the rod,53 with longer rods inducing greater depletion. Therefore, once the pressure exerted by the surfactant micelles is large enough to induce reaggregation of the nanotubes, larger nanotubes are forced together preferentially to provide a larger reduction in the osmotic pressure. The hydrodynamic radius of the micelle will also have an influence on the extent of the depletion attraction, thus each surfactant will induce a depletion effect at a different micelle volume fraction dependent on the nature of the molecule involved. This explains the variation in the surfactant concentration at which the onset of depletion is evident for the different dispersants in Figure 4. The depletion effect is not immediately apparent in sonicated dispersions, as shown in Figure 5. However, upon the solutions standing for a period of a few days, sedimentation is observed, with higher surfactant concentrations inducing a faster rate of flocculation. Figure 5B shows the uncentrifuged solutions after

approximately 4 months of incubation, where no visible flocculation has occurred for an SDBS concentration of 0.5%. A significant amount of the CNTs have reaggregated for 2%, while at 4% SDBS the majority of CNTs are no longer dispersed. Centrifugation greatly enhances the depletion effect, since once the nanotubes rebundle the agglomerates sediment out under the enhanced gravitational forces. Figure 6 shows AFM images of CNTs on Si wafers cast from SDBS dispersions having surfactant concentrations of 0.5%, 2%, and 3%. The CNTs appear to be well dispersed in 0.5% SDBS, existing primarily in small bundles or as individual tubes. Increasing the SDBS concentration to 2% appears to result in a slight reduction in CNT concentration, and a significant amount of rebundling. While this rebundling could potentially be an artifact of the spin coating and drying process, mostly individual tubes are observed for lower concentrations over a number of images, hence we believe it is directly related to the dispersion state in solution. There seems to be a close association of bundles to one-another, forming elongated networks. Such network formation may be a result of preferential stabilization of nanotube junctions, where the surfactant adsorbs to adjacent hydrophobic surfaces.33 Much shorter bundles remain in the supernatant when the SDBS concentration is further increased to 3%, though the nanotubes still appear to be tightly grouped. This length reduction is analogous to that observed in SDS,52 and may potentially be used as a simple method to separate out shorter nanotubes; however, the low yield and aggregated state of the CNTs are significant disadvantages. The AFM results confirm that the CNTs are well dispersed where the S22 absorbance area is maximized, and that increasing the surfactant concentration beyond the position of this maximum causes flocculation of the CNTs into long thin “ropes” or loosely associated bundles. It has been suggested that increasing the initial mass ratio of CNT can inhibit this effect as more surfactant is required to disperse the CNT, thus reducing the micelle concentration and shifting the onset of depletion effects to higher surfactant concentrations;23 however, this effect was not observed to any appreciable degree for the different CNT concentrations used in this work. Conclusions Many different dispersive agents are effective for producing aqueous dispersions of CNTs; however, the quality of the dispersion will depend on the type of surfactant/polymer etc. and the properties of the CNTs under investigation. Dispersion of arc CNTs was successfully achieved with all of the surfactants analyzed, though SDS was found to be the least effective. For

Dispersion of SWCNTs in Aqueous Solution the smaller diameter HiPCO nanotubes the discrepancy was greater between dispersive agents, with smaller ionic surfactants exhibiting superior performance, including SDS. Thus, an effective dispersant for nanotubes produced by a certain technique may not necessarily perform well for nanotubes prepared by a different method. The concentration of the surfactant was found to be a more influential parameter on the resulting dispersion than the ratio of surfactant to CNTs by mass for both DOC and SDBS, most likely as this determines the concentration of micelles in the solution volume. Optimal concentrations for dispersion of arc nanotubes were determined for the anionic surfactants SDBS and DOC as well as the nonionic Triton-X 405, Brij S-100, Pluronic F-127, and PVP-55. Increasing the surfactant concentrations above these optimal values is detrimental as it leads to flocculation of the CNTs, likely through attractive depletion interactions. Acknowledgment. The authors thank Dr. Milena GinicMarkovic for use of the Sonics VCX 750W. Supporting Information Available: Structures of the surfactants, stability analysis of the dispersions, SDBS/DOC cosurfactant results, Raman data, and M11 peak trends for the determination of optimal surfactant concentrations. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Dresselhaus, M. S.; Dresselhaus, G.; Jorio, A. Annu. ReV. Mater. Res. 2004, 34, 247. (2) Thess, A.; Lee, R.; Nikolaev, P.; Dai, H.; Petit, P.; Robert, J.; Xu, C.; Lee, Y. H.; Kim, S. G.; Rinzler, A. G.; Colbert, D. T.; Scuseria, G. E.; Tomanek, D.; Fischer, J. E.; Smalley, R. E. Science 1996, 273, 483. (3) Hou, P.-X.; Liu, C.; Cheng, H.-M. Carbon 2008, 46, 2003. (4) Itkis, M. E.; Perea, D. E.; Jung, R.; Niyogi, S.; Haddon, R. C. J. Am. Chem. Soc. 2005, 127, 3439. (5) Yu, A.; Bekyarova, E.; Itkis, M. E.; Fakhrutdinov, D.; Webster, R.; Haddon, R. C. J. Am. Chem. Soc. 2006, 128, 9902. (6) Wei, L.; Wang, B.; Wang, Q.; Li, L.-J.; Yang, Y.; Chen, Y. J. Phys. Chem. C 2008, 112, 17567. (7) Hersam, M. C. Nat. Nanotechnol. 2008, 3, 387. (8) Vaisman, L.; Wagner, H. D.; Marom, G. AdV. Colloid Interface Sci. 2006, 128-130, 37. (9) Kharisov, B. I.; Kharissova, O. V.; Leija Gutierrez, H.; Ortiz Mendez, U. Ind. Eng. Chem. Res. 2009, 48, 572. (10) Wang, H. Curr. Opin. Colloid Interface Sci. 2009, 14, 364. (11) Shin, J.-Y.; Premkumar, T.; Geckeler, K. E. Chem.sEur. J. 2008, 14, 6044. (12) Backes, C.; Schmidt, C. D.; Rosenlehner, K.; Hauke, F.; Coleman, J. N.; Hirsch, A. AdV. Mater. 2010, 22, 788. (13) Haggenmueller, R.; Rahatekar, S. S.; Fagan, J. A.; Chun, J.; Becker, M. L.; Naik, R. R.; Krauss, T.; Carlson, L.; Kadla, J.; Trulove, P.; Fox, D.; DeLong, H.; Fang, Z.; Kelley, S. O.; Gilman, J. W. Langmuir 2008, 24, 5070. (14) White, B.; Banerjee, S.; O’Brien, S.; Turro, N. J.; Herman, I. P. J. Phys. Chem. C 2007, 111, 13684. (15) Sun, Z.; Nicolosi, V.; Rickard, D.; Bergin, S. D.; Aherne, D.; Coleman, J. N. J. Phys. Chem. C 2008, 112, 10692. (16) Ishibashi, A.; Nakashima, N. Chem.sEur. J. 2006, 12, 7595. (17) Wenseleers, W.; Vlasov, I. I.; Goovaerts, E.; Obraztsova, E. D.; Lobach, A. S.; Bouwen, A. AdV. Funct. Mater. 2004, 14, 1105. (18) Moore, V. C.; Strano, M. S.; Haroz, E. H.; Hauge, R. H.; Smalley, R. E.; Schmidt, J.; Talmon, Y. Nano Lett. 2003, 3, 1379.

J. Phys. Chem. B, Vol. 114, No. 30, 2010 9811 (19) Islam, M. F.; Rojas, E.; Bergey, D. M.; Johnson, A. T.; Yodh, A. G. Nano Lett. 2003, 3, 269. (20) Rastogi, R.; Kaushal, R.; Tripathi, S. K.; Sharma, A. L.; Kaur, I.; Bharadwaj, L. M. J. Colloid Interface Sci. 2008, 328, 421. (21) Sinani, V. A.; Gheith, M. K.; Yaroslavov, A. A.; Rakhnyanskaya, A. A.; Sun, K.; Mamedov, A. A.; Wicksted, J. P.; Kotov, N. A. J. Am. Chem. Soc. 2005, 127, 3463. (22) Nish, A.; Hwang, J.-Y.; Doig, J.; Nicholas, R. J. Nat. Nanotechnol. 2007, 2, 640. (23) Xin, X.; Xu, G.; Zhao, T.; Zhu, Y.; Shi, X.; Gong, H.; Zhang, Z. J. Phys. Chem. C 2008, 112, 16377. (24) Tu, X.; Manohar, S.; Jagota, A.; Zheng, M. Nature 2009, 460, 250. (25) Zheng, M.; Jagota, A.; Semke, E. D.; Diner, B. A.; McLean, R. S.; Lustig, S. R.; Richardson, R. E.; Tassi, N. G. Nat. Mater. 2003, 2, 338. (26) Edri, E.; Regev, O. Langmuir 2009, 25, 10459. (27) Liu, Y.; Gao, L.; Zheng, S.; Wang, Y.; Sun, J.; Kajiura, H.; Li, Y.; Noda, K. Nanotechnology 2007, 18, 365702. (28) O’Connell, M. J.; Bachilo, S. M.; Huffman, C. B.; Moore, V. C.; Strano, M. S.; Haroz, E. H.; Rialon, K. L.; Boul, P. J.; Noon, W. H.; Kittrell, C.; Ma, J.; Hauge, R. H.; Weisman, R. B.; Smalley, R. E. Science 2002, 297, 593. (29) Xu, Z.; Yang, X.; Yang, Z. Nano Lett. 2010, 10, 985. (30) Matarredona, O.; Rhoads, H.; Li, Z.; Harwell, J. H.; Balzano, L.; Resasco, D. E. J. Phys. Chem. B 2003, 107, 13357. (31) Richard, C.; Balavoine, F.; Schultz, P.; Ebbesen, T. W.; Mioskowski, C. Science 2003, 300, 775. (32) Yurekli, K.; Mitchell, C. A.; Krishnamoorti, R. J. Am. Chem. Soc. 2004, 126, 9902. (33) Angelikopoulos, P.; Bock, H. J. Phys. Chem. B 2008, 112, 13793. (34) Calvaresi, M.; Dallavalle, M.; Zerbetto, F. Small 2009, 5, 2191. (35) Angelikopoulos, P.; Gromov, A.; Leen, A.; Nerushev, O.; Bock, H.; Campbell, E. E. B. J. Phys. Chem. C 2009, 114, 2. (36) Utsumi, S.; Kanamaru, M.; Honda, H.; Kanoh, H.; Tanaka, H.; Ohkubo, T.; Sakai, H.; Abe, M.; Kaneko, K. J. Colloid Interface Sci. 2007, 308, 276. (37) O’Connell, M. J.; Boul, P.; Ericson, L. M.; Huffman, C.; Wang, Y.; Haroz, E.; Kuper, C.; Tour, J.; Ausman, K. D.; Smalley, R. E. Chem. Phys. Lett. 2001, 342, 265. (38) Dror, Y.; Pyckhout-Hintzen, W.; Cohen, Y. Macromolecules 2005, 38, 7828. (39) McDonald, T. J.; Engtrakul, C.; Jones, M.; Rumbles, G.; Heben, M. J. J. Phys. Chem. B 2006, 110, 25339. (40) Tan, Y.; Resasco, D. E. J. Phys. Chem. B 2005, 109, 14454. (41) Huang, H.; Kajiura, H.; Maruyama, R.; Kadono, K.; Noda, K. J. Phys. Chem. B 2006, 110, 4686. (42) Okazaki, T.; Saito, T.; Matsuura, K.; Ohshima, S.; Yumura, M.; Iijima, S. Nano Lett. 2005, 5, 2618. (43) Arnold, M. S.; Green, A. A.; Hulvat, J. F.; Stupp, S. I.; Hersam, M. C. Nat. Nanotechnol. 2006, 1, 60. (44) Bachilo, S. M.; Strano, M. S.; Kittrell, C.; Hauge, R. H.; Smalley, R. E.; Weisman, R. B. Science 2002, 298, 2361. (45) Wang, F.; Dukovic, G.; Brus, L. E.; Heinz, T. F. Science 2005, 308, 838. (46) Wang, H.; Zhou, W.; Ho, D. L.; Winey, K. I.; Fischer, J. E.; Glinka, C. J.; Hobbie, E. K. Nano Lett. 2004, 4, 1789. (47) Vigolo, B.; Penicaud, A.; Coulon, C.; Sauder, C.; Pailler, R.; Journet, C.; Bernier, P.; Poulin, P. Science 2000, 290, 1331. (48) Eda, G.; Fanchini, G.; Kanwal, A.; Chhowalla, M. J. Appl. Phys. 2008, 103, 093118. (49) Strano, M. S.; Moore, V. C.; Miller, M. K.; Allen, M. J.; Haroz, E. H.; Kittrell, C.; Hauge, R. H.; Smalley, R. E. J. Nanosci. Nanotechnol. 2003, 3, 81. (50) Landi, B. J.; Ruf, H. J.; Evans, C. M.; Cress, C. D.; Raffaelle, R. P. J. Phys. Chem. B 2005, 109, 9952. (51) Monteiro-Riviere, N. A.; Inman, A. O.; Wang, Y. Y.; Nemanich, R. J. Nanomedicine Nanotechnol. 2005, 1, 293. (52) Bonard, J.-M.; Stora, T.; Salvetat, J.-P.; Maier, F.; Sto¨ckli, T.; Duschl, C.; Forro´, L.; Heer, W. A. d.; Chaˆtelain, A. AdV. Mater. 1997, 9, 827. (53) Schilling, T.; Jungblut, S.; Miller, M. A. Phys. ReV. Lett. 2007, 98, 108303.

JP104113D