J. Phys. Chem. C 2007, 111, 17827-17834
17827
Diameter Selection of Single-Walled Carbon Nanotubes through Programmable Solvation in Binary Sulfonic Acid Mixtures† Sivarajan Ramesh,*,‡,| Hongwei Shan,‡ Eric Haroz,‡ W. E. Billups,‡,§ Robert Hauge,‡ W. Wade Adams,‡ and Richard E. Smalley⊥ R. E. Smalley Institute for Nanoscale Science and Technology and Department of Chemistry, Rice UniVersity, Houston, Texas 77005-1827 ReceiVed: February 15, 2007; In Final Form: May 16, 2007
Pristine carbon nanotubes (CNTs) dissolve as polycarbocations in superacids through direct protonation. The solvating power of a superacid is determined by the stability of the conjugate base anion that competes with the CNTs for the dissociated proton. We have demonstrated that this equilibrium can be controlled in a predictable fashion, thus rendering the solvating power of a superacid system tunable. In this article, we show that the solvating power of chlorosulfonic acid can be changed in a desired fashion by forming binary mixtures in different proportions with a non-superacid such as methane sulfonic acid. Thus, the successive extraction of carbon nanotubes with binary acid mixtures of increasing solvating power leads to the differentiation of CNTs by their molecular geometry. We show that solvation by direct protonation is sensitive to the geometric strain at the carbon atom and, hence, to the nanotube diameter. In this respect, the direct protonation method is distinct from surfactant-based or electrical-field-based methods that distinguish metallic CNTs from semiconducting types mainly on the existence of finite density of states or not at the Fermi level. We have employed solid-state Raman spectroscopic analysis of the CNT radial breathing modes and UVvis absorption spectroscopy and a systematic mapping method to support our conclusions. We believe the concept demonstrated in this paper holds the potential to be developed into a chemical tool kit useful in the scaleable separation of CNTs by their (n, m) types, thus paving the way for molecular carbon nanotechnology.
Introduction Carbon nanotechnology constitutes a branch of science that has grown exponentially over the past decade, with very few parallels in all of modern scientific research.1-4 Despite this wide and deep growth, the goal of realizing experimental carbon nanotube research as a true molecular science still remains elusive. Ironically, the very interesting properties that provide immense application potential5-7 for carbon nanotubes (CNTs) turn out to be the principal reasons behind this elusiveness. As a molecule, a carbon nanotube is a seamlessly rolled graphene sheet, described by a set of two basic vectors defined on the hexagonal lattice of the graphene sheet that describe the molecular geometry1 of a single walled carbon nanotube (SWNT). If the SWNT is viewed as a polymer of elemental carbon, its diameter and chiral angle are comparable to the molecular characteristic of a monomer that constitutes a conventional polymer. The second most important molecular property is the length of the carbon nanotube. Even though the diameter and chirality of a SWNT influence its electronic and optoelectronic properties more than does its length, the length does influence SWNT dispersity in various solvents and, thus, serves as a limiting factor in experimentation. In this respect, the SWNT length can be compared to the molecular weight of a rigid rod polymeric system. The extreme Van der Waals †
Part of the special issue “Richard E. Smalley Memorial Issue”. * Corresponding author. E-mail:
[email protected]. R. E. Smalley Institute for Nanoscale Science and Technology. § Department of Chemistry. ⊥ Deceased October 28, 2005. | Current address: Nantero Inc, 25-E Olympia Avenue, Woburn, MA 01801. ‡
interaction energies at ∼500 meV for every nanometer of lateral CNT-CNT contact8 renders bundling of carbon nanotubes inevitable and adds further to the complexity of CNT dispersion or true dissolution in solvents. Primary purification processes designed to remove catalyst metal particles (necessary in the synthesis of CNTs) and amorphous carbon particles vary widely on the basis of the CNT synthesis method and usually leave a lasting mark on the surface characteristic of the pristine CNTs. Ionic and nonionic surfactants widely used to disperse CNTs into thin bundles or individuals in solvents and water9 more often leave a molecular monolayer around individual CNTs, thus affecting the surface chemistry of CNTs. The uphill task involved in the realization of molecular carbon nanotechnology, by which we mean the ability to work at will with a CNT sample of chosen length, type, diameter, and chiral angle, is represented graphically in Figure 1 in the form of a carbon nanotechnology value pyramid. Except for scaleable CNT synthesis, purification, and structural applications, there are no significant methods available to fulfill the goals on the upper portion of the value pyramid. Scaleable synthesis10,11 of CNTs shown at the bottom of the pyramid has by and large been achieved now, and the scales and homogeneity of the CNT products have improved several fold in the past decade. This was an important and necessary step in the proliferation of research on carbon nanotubes. As a natural second step, a wide range of first-pass purification techniques12-14 have been developed and are now available tailor-made for almost every type of known CNT synthetic technique. Purification of carbon nanotubes to the degree and form suitable for CMOS integration with trace metals at parts-
10.1021/jp071322t CCC: $37.00 © 2007 American Chemical Society Published on Web 07/04/2007
17828 J. Phys. Chem. C, Vol. 111, No. 48, 2007
Ramesh et al.
Figure 1. The carbon nanotechnology value pyramid.
per-billion levels has been accomplished by Nantero Inc, which has led to the integration of carbon nanotubes as part of standard CMOS operational flow in semiconductor fabrication facilities.15 Despite these major milestones, scaleable processes for true solubilization of CNTs that are suitable to climb further up the value pyramid have been fewer. Solubilization of CNTs in organic solvents through organic derivatization16,17 has been a very successful route which could find immediate application in the dispersion of CNTs in polymer matrixes. However, the delocalized π electrons responsible for many interesting properties of CNTs are lost in such methods. Similar has been the limitation with polymer wrapping methods18,19 or the dissolution of CNTs as alkali metal salt20 that makes it unsuitable for integration as part of standard microelectronic circuitry. The processes that remain higher in the pyramid are scaleable methods for true molecular dissolution, separation of metallic and semiconductor CNTs, length selection, diameter selection, and chiral angle controls. The “holy grail” in carbon nanotechnology still remains the ability to make a vial of pristine CNTs showing very narrow length dispersity and consisting of CNT tubes of a given (n, m) type, all dissolved individually without the aid of surfactants. At the Carbon Nanotechnology Laboratory at Rice University, the first step toward this ambitious goal was rightly identified as true molecular dissolution of CNTs in a suitable solvent. Inference from the solvation behavior21,22 of geometrically analogous, aramid or other rigid rod polymeric systems was taken as the clue, and for the first time, protic superacids were used as solvents for the true, individual solubilization of CNTs by direct protonation. A typical aramid polymer rigid rod consists of nitrogen- or oxygen-containing groups as part of the polymeric backbone, serving as specific centers for protonation.21 However, since pristine single-walled carbon nanotubes are amphoteric, they behave as a weak base and, hence, are directly protonated, resulting in the generation of a polycarbocation:
Cx + yAH f [Cxδ+ Hy(1-kδ)+] + yA-
carbon alewives,23,25 and finally, the first-ever demonstration of a wet spinning process, yielding continuous fibers of highly aligned, neat CNTs.26 A closer look at eq 1 suggests that the formation and stability of the polycarbocation rests solely on the stability of the conjugate base anion A-. This is a significant deviation, as compared to the conventional rigid rod polymeric molecules, and hence opens up an interesting possibility for the programmable protonation or solvation of the carbon nanotubes. This simple but powerful tool allows a chemist to design a solvent system with continuously variable solvating power of the solvent. In other words, carbon nanotubes of different diameters, helicity and electronic types with subtle variation in reactivity can be differentiated in a systematic fashion. In this paper, we introduce one such binary acid system consisting of chlorosulfonic acid and methane sulfonic acid mixed in different proportions, and hence, the system has a continuously variable solvating ability. Key characteristics of these completely miscible acids are listed in Table 1. We have employed Raman spectroscopy and UV-vis-NIR spectroscopies to obtain experimental data. An n, m assignment table27 depicting the periodic structural relationship between CNT tubes of different (n, m) types and holding key spectral data collected from the carbon nanotube literature28-33 in a readily comparable format was used to make tube assignments. Experimental Methods Single walled nanotubes (SWNT) produced by the HiPco process were purified by standard acid purification methods,14 followed by extensive washing in hexane as a final step of the wet chemical process, followed by drying under vacuum to a powdery material. The extraction process mentioned in this TABLE 1 physical property
(1)
where k ) x/y and δ is the fractional positive charge carried by each carbon atom, dependent on the oxidizing power of the superacid.23 On the basis of such true dissolution of CNTs, a series of experimental investigations in the Smalley group focused on the formation of liquid crystalline nematic phases,24 precipitation of the nematic phases into intrinsically aligned
mol wt (g/mol) mp, °C bp, °C density (g/cm3) pKa proton affinity of the conjugate base (kcal/mol) SWNT solubility mutual miscibility
Cl-SO3H 116.5 -80 151 1.75 -6.0 303 yes all proportions
CH3-SO3H 96.1 20 167 1.48 -1.92 323 no
Diameter Selection of SWNTs
J. Phys. Chem. C, Vol. 111, No. 48, 2007 17829
Figure 2. Experimental setup specially designed to carry out the binary acid extraction process.
section was carried out employing a specially designed extraction device shown in the photograph (Figure 2). The main component of the device is a glass extraction thimble placed in an airtight glass chamber as shown in the figure. In a typical experiment, a known weighed quantity of dry CNT powder (∼30 mg) was placed in the glass extraction thimble (3-4 µm mean pore size) and vacuum-dried at >100 °C overnight to eliminate all adsorbed water. All other glass components were kept in an air oven just prior to the assembly for the experiment. The glass thimble with dry CNTs was transferred to the extraction chamber and closed with an air-seal Teflon stirrer assembly. (Truebore, Ace glass). Immediately upon sealing, the extraction chamber was brought under a blanket of flowing argon through a standard Y junction. A 100 mL separation funnel connected to the bottom of the extraction setup was also kept under flowing argon by keeping the valve A open. In a separate set of experiments, chlorosulfonic acid and methane sulfonic acid (received as such) were mixed inside a dry box in the required proportions, placed in a dry transfer tube, and sealed with a septum. A 30 mL portion of the binary mixture was injected into the extraction thimble through a liquid inlet septum port employing a predried, long needle and syringe after closing valve A. The mixture of SWNT-acid was kept stirring slowly with the motorized glass shaft at room temperature. Within hours, the acid was seen seeping through the glass frit in the thimble and leveled with the acid slurry inside the extraction thimble, and the stirring process continued to allow continuous extraction of CNTs, typically ∼15 h, while the outlet gas port was connected to a silica gel moisture dryer trap to avoid back-diffusion of moist ambient air into the system. We emphasize the importance of keeping the conditions absolutely dry, because we noticed that even trace amounts of water (as few as 10 ppm) led to the precipitation of SWNT, forming alewives and hampering the extraction process, which is consistent with our previous experience in working with rigid rod polymer acid solutions. Once the extraction was completed, the binary mixture SWNT extract, which varied in color from golden yellow to dark brown, was collected in the separation funnel by draining through valve
A, while the argon flow and the opening were carefully adjusted. Immediately on draining the first extract, a second shot of a binary mixture with increasing solvating power was injected into the same thimble, and the process was allowed to continue. For the experiments described in this paper, the binary mixtures employed had proportions of chlorosulfonic acid mole fraction of 0.1, 0.15, 0.25, 0.4, and 1.0 with corresponding amounts of methane sulfonic acid. In the following sections, when we refer to a binary mixture as 0.25 m, it refers to a sample that has 0.25 m of chlorosulfonic acid. The last residue remaining in the thimble was marked as thimble residue and processed as described below in this section. The SWNT binary acid extract as obtained in each of the cases above was added dropwise into ∼200 mL of ice-cold DI water kept in a conical flask inside a fume hood. The solution appeared colorless and clear on completion of the addition. The flask was closed and allowed to stand overnight. On standing, a large number of tiny black aggregates that could be seen with the naked eye were formed in the flask. The aggregates were slowly filtered through a 25 mm polycarbonate filter disk (Millipore,