Adsorption on Carbon Nanotubes

Adsorption on Carbon Nanotubes

ADSORPTION ON CARBON NANOTUBES: EXPERIMENTAL RESULTS Aldo D. Migone Department of Physics, Southern Illinois University, Carbondale, IL, USA Contents...

2MB Sizes 0 Downloads 66 Views

ADSORPTION ON CARBON NANOTUBES: EXPERIMENTAL RESULTS Aldo D. Migone Department of Physics, Southern Illinois University, Carbondale, IL, USA

Contents 16.1 Introduction 16.2 Hydrogen Storage 16.3 Adsorption of Rare Gases and Simple Molecular Species 16.4 Conclusions Acknowledgments References

4°3 4°4 40 8

42 5 426 426

16.1 INTRODUCTION

Over the past ten years, interest in the study of gas adsorption on carbon has experienced resurgence and growth [1-3]. This revival ofinterest, in no small measure, followed as a result of the publication of two studies in the late 1990s on hydrogen adsorption on new forms of carbon. These were the Rodriguez's group work on graphite nanofibers (GNFs) [4] and the work on carbon nanotubes by Heben's group [5]. These two studies held the promise of providing new approaches to gas storage. Even though neither of them is now considered valid by most researchers in this field, their impact in stimulating further work in gas adsorption on novel carbon materials cannot be underestimated. Hydrogen storage continues to be one of the driving forces behind the research in gas adsorption on carbon nanotubes. The impetus for this work comes from the need to find an effective way to store hydrogen in order to make it possible for this gas to replace petroleum-based fuels in transportation Adsorption by Carbons ISBN: 978-0-08-044464-2

© 2008 Elsevier Ltd. All rights reserved.

403

404

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

applications [6]. Developing a commercially viable means of storing hydrogen is the most significant technological choke point in the path to the "hydrogen economy" [7-9]. Adsorption is one of a handful of alternative approaches that exist for storing hydrogen [9-11]. As the experimental and theoretical investigations of gas adsorption on carbon nanotubes progressed, a number of questions that are important from a fundamental perspective have emerged. These questions provide an independent scientific motivation for the study of adsorption of various gases on these novel carbon materials. This chapter summarizes some of the more important experimental developments that have occurred in the field of gas adsorption on carbon nanotubes. In the first section, we review results regarding hydrogen storage. In subsequent sections, we discuss experiments dealing mostly with the adsorption of rare gases and simple molecules on carbon nanotubes.

16.2 HYDROGEN STORAGE The experiments on hydrogen storage are focused on determining of how much hydrogen can be stored on a substrate at a specified pressure and temperature [12, 13]. Chambers et al. [4] described adsorption measurements on several types of GNFs produced by their research group. The measurements were conducted at room temperature and high pressures (up to 110 atm) in a volumetric adsorption apparatus. The storage determinations were performed by measuring the pressure decrease in the sample cell as a function of time. The study reported extraordinarily high levels of hydrogen adsorption for all the GNFs. The highest value found was a weight fraction of 0.68 (i.e., the ratio of the weight of the adsorbed hydrogen to the sum of the weights of the adsorbed hydrogen and the substrate was 0.68). Attempts by other groups [14-16] to reproduce these extraordinary hydrogen storage claims failed. In addition, the extremely high levels of hydrogen adsorption reported for the GNFs cannot be explained theoretically [17, 18]. Unlike the case for GNFs, where a consensus has now emerged that the extremely high storage reports are erroneous, the situation regarding hydrogen adsorption on single-walled carbon nanotubes (SWNTs) continues to be surrounded by controversy [5, 9, 15, 16, 18]. In 1997 Heben's group [5] presented results of temperature-programmed desorption measurements for hydrogen on samples that contained only a small fraction, on the order of 0.1-0.2 % by weight (wt %), of SWNTs in them. The SWNTs were prepared by electric arc discharge. The desorption experiments found that samples that been heated to 970 K under vacuum prior to being exposed to H 2 displayed a H 2 desorption peak at about 300 K [5]. The results were interpreted by attributing all of the hydrogen evolved in the measurements

16.2

Hydrogen Storage

to gas that had been adsorbed on the nanotubes. This resulted in an estimate of on the order of 5-10 atomic percent (at. %) of hydrogen adsorbed at room temperature on the carbon nanotubes. The estimate was a bold extrapolation from the data measured on samples that contained very few nanotubes. Ye et al. [21] reported on cryogenic adsorption on bundles of SWNTs that were subjected to purification and "cutting" treatments involving extended sonication in dimethylformamide. The measurements were conducted in a volumetric apparatus and were performed at 80 K and pressures of up to 12 MPa. At lower pressures, the data were compatible with adsorption on bundles; the levels of adsorption were much lower than those reported by Heben's group. As the pressure increased to approximately 40 atm; however, the amount of hydrogen adsorbed increased significantly. This increase was interpreted as signaling the "unbundling" of the nanotubes. As a result of unbundling, the entire surface of each of the tubes becomes accessible for adsorption [21]. The highest storage capacity reached in these measurements was 8 wt %, but the trend in the data pointed toward even higher capacities at pressures beyond the limits of the setup. It should be noted that to date no unbundling into individual nanotubes has been observed by other researchers working at high pressures. Heben's group also reported on hydrogen adsorption on SWNTs produced by laser ablation. These SWNTs were subjected to purification in nitric acid and were cut by a procedure involving high-power ultrasonication in nitric acid, with a titanium-based probe [9, 22]. They reported up to 7 wt % hydrogen adsorption at ambient conditions. They reported some differences with respect to their previous results [5]. Temperature-programmed desorption experiments on these samples provided evidence ofthe existence of two groups ofadsorption sites: one with a desorption peak 400 K (which accounted for approximately one-third of the adsorbed hydrogen) and another with a desorption peak extending from 475 to 800 K that accounted for the remainder two-thirds [9, 22]. It was noted that the cutting procedure introduced a titanium alloy to the sample, the result of the disintegration of the ultrasonication probe during the cutting process. Liu et al. [23] reported on measurements on 1.8S-nm diameter SWNTs yielding hydrogen storage capacities between 2.4 and 4.2 wt % (depending on the sample), at room temperature and at pressures of 10 MPa or less. The SWNTs were produced by arc discharge, and the measurements were conducted on as-produced tubes as well as on samples purified with hydrochloric acid. The amount of hydrogen adsorbed on the sample was determined volumetrically. By weighing the samples after desorption, they determined that close to 80 % of the hydrogen could be recovered at ambient temperature and pressure. The results reported in the set of four studies summarized above suggest a generally positive trend regarding the potential for using carbon nanotubes for hydrogen storage. These reports were followed a second group of studies that indicated, quite to the contrary, that SWNTs have extremely low hydrogen storage capacities. We look at these next. Hirscher and his collaborators [16, 24, 25] have studied hydrogen storage on carbon nanotubes. Their first report [24] focused on the effect of high-power

406

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

ultrasonication in nitric acid using a Ti alloy probe on the storage results. They used SWNTs produced by laser ablation, and by arc discharge. They measured the effect that duration of the ultrasonication had on the level of hydrogen storage. Their measurements were conducted using temperature-programmed desorption spectroscopy, using deuterium as the adsorbate. The dosing was performed exposing the sample to 0.08 MPa of gas at room temperature. They found that the amount of deuterium stored was directly proportional to the time that the sample was subjected to sonication [24]. The experiments were repeated using diamond powder instead of SWNTs, and similar storage results were obtained. X-ray diffraction analysis of the samples revealed peaks indicative of the presence of titanium hydride. They concluded that most of the hydrogen in the sample was stored as titanium hydride, which formed on titanium that was incorporated into the samples as a result of the aggressive purification process [24]. Hirscher's group [16] explored hydrogen storage on samples of GNFs and SWNTs (produced by arc discharge) that were previously subjected to ball milling in either an Ar or a D 2 atmosphere. Ball milling disrupts the carbon microstructure of these materials and may facilitate hydrogen intake. They found that when the ball milling is done in Ar, the maximum D 2 storage on SWNTs is below 0.1 wt %. This deuterium adsorption is reversible. When the experiments were performed on GNFs or on SWNTs that were not subjected to ball milling, no H 2 adsorption was found. Finally, when the ball milling is done in deuterium, the samples of SWNTs, GNFs, and graphite all showed significant D 2 uptake (>1 wt%). The desorption peaks for the GNFs and for the SWNTs were near 720 K, while that for graphite was above 900 K. When these samples were cooled back to room temperature and re-exposed to D 2 , they displayed no desorption up to 850 K, indicating that the process of adsorbing deuterium by ball milling in a deuterium atmosphere is nonreversible, and hence, deuterium is chemically attached to the carbon in that process [16]. Tibbetts et al. [15] studied the hydrogen storage capacity of various carbon materials, including SWNTs from two sources (MER Corporation and [email protected]), graphite, activated carbon, and several graphite fibers. The nanotube samples were not subjected to any purification treatment. The measurements were conducted in two different volumetric apparatus. Their conclusion was that less than 0.1 wt % of hydrogen adsorbed at room temperature and pressures of up to 10 MPa on all the forms of carbon studied. To understand the vastly different storage results reported, these authors explored potential sources of systematic errors [15]. They discussed the effect of small leaks on the results. They noted the need to perform desorption in addition to adsorption measurements, in order to verify results. They also discussed the effect on the storage results, of temperature variations of the gas-handling apparatus, and of the sample cell that results from the adsorption process. In sharp contrast with this second set of reports that find unfavorable hydrogen storage capacities in carbon nanotubes, a more recent group of experiments

16.2

Hydrogen Storage

4°7

dealing with activated carbon nanotubes reports higher levels of H 2 adsorption. We review these studies next. Eklund's group [26] studied the storage capacity of arc-discharge SWNTs subjected to different treatments (selective oxidation in dry air, followed by acid reflux with either HCl or HN0 3 ). The isotherms were conducted gravimetrically at 77 and 87 K, at pressures between 0 and 20 bar. The weight percentage of stored H 2 varied from 0.52 wt % for an as-prepared sample to 6.4 wt % for an acid-treated sample that was heated to 1000°C. Samples heated to 1000°C before hydrogen exposure generally showed greater storage capacities than those heated to 250°C. The results showed no correlation between the specific surface area of the samples and their hydrogen storage capacity. Adsorption-desorption cycles showed that the adsorption process was completely reversible. A significant difference between this report [26] and that of Ye et al. [21] is that for these samples most of the hydrogen was adsorbed at pressures of 1 atm or less. Smith et al. [27] investigated the storage capacities of as-received SWNTs as well as those of tubes subjected to various activation treatments. The measurements were conducted at 25°C using a tapered element oscillating mass analyzer. This type of microbalance determines the mass adsorbed on the substrate from changes in the frequency of oscillation of the tapered element that result from adsorption. In the setup used, only measurements at pressures above atmospheric were possible [27], i.e., the isotherms measured only the hydrogen that adsorbed on the SWNTs after the pressure reached 1 atm. The omission of the H 2 adsorbed between 0 and 1 atm may be significant in estimating the sorptive capacity of the tubes, in light of Eklund's group report of significant adsorption occurring at low pressures [26]. Smith et al. [27] found that H 2 adsorption either on "Raw-Material" SWNTs or on "Purified" SWNTs from Rice University reached a maximum of somewhere between 0.2 and 0.3 wt % at 48 bar [27]. When these materials were subjected to CO 2 oxidation, the fraction of hydrogen stored at 48 bar increased by a factor of between 3 and 4. Computer simulations performed as a companion to the experiments found that standard physisorption potentials were not able to account for the results on the activated nanotube samples. However, because the adsorption measurements were completely reversible, the experiments established that chemisorption did not play any significant role [27]. Rao's group [28] conducted a comparative study of hydrogen adsorption at 300 K and high pressures on arc-discharge as-produced and acid-treated (with HN0 3 ) SWNTs, as-produced and acid-treated MWNTs (produced by acetylene pyrolysis and produced by arc discharge), and as-produced and acidtreated aligned MWNTs (produced by two different approaches). The measurements were conducted volumetrically. This study found low adsorption on the SWNTs. However, 3.7 wt % storage was found for acid-treated, densely packed, aligned MWNTs, at a pressure of approximately 140 bar. In light of the cautionary notes listed by other workers [15], this group performed experiments for samples of different weights, as well as at various different pressures. In all cases the gas storage results were reproducible [28].

408

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

Prior to this report, there had been two other studies that indicated relatively high hydrogen storage capacities on aligned MWNTs [29, 30]. The storage values found in those reports were close to 3 wt % at 10 MPa and room temperature. As this summary of experimental reports clearly reveals, the situation regarding hydrogen storage on nanotubes remains unsettled. The set of low hydrogen storage capacity reports [14-16, 24, 25] are able to explain why some of the initial work reporting high storage values were in all likelihood incorrect [4, 5, 23]. The formation of hydrides, the effects of small leaks in volumetric adsorption setups, or those of uncorrected fluctuations in the temperature can explain the initial high storage values reported. On the other hand, it is difficult to attribute all reports of high hydrogen storage to systematic errors. In some cases, the high storage capacity reports come from groups that have also reported low levels of hydrogen storage for other samples, using essentially the same methods in both cases [14, 21]. In other cases [26, 27] the hydrogen storage determinations were conducted gravimetrically and should be less susceptible to the effect of leaks or temperature fluctuations. Finally, some reports of high hydrogen storage [28] come from groups that are aware of the potential systematic errors involved and that obtain in the same study low storage results for some samples and not for others. The data from the various groups reporting high levels of hydrogen storage, conducted under similar conditions, display significant differences in the pressure dependence of the amount of hydrogen adsorbed [21, 26]. The origin of these differences is not clear. Finally, we note that the samples used in almost all of the experiments discussed have been subj ected to different purification, cutting, and/or activation treatments. The effect that differences in treatment have on the storage capacities of the samples is not known. It is possible that treatment differences are responsible for part of the differences in the reported storage capacities. One way to compare directly experimental results would be to have different groups apply their experimental techniques to the same carbon nanotube samples. Unfortunately, as Hirscher [25] has noted, at least in some cases, this approach is not possible because contractual obligations prevent laboratories from exchanging samples.

16.3

ADSORPTION OF RARE GASES

AND SIMPLE MOLECULAR SPECIES

In the following sections we review experimental results obtained for rare gases and simple molecules adsorbed on SWNTs. These studies are devoted to explore the nature and properties of the films that these species form when adsorbed on SWNTs, mostly within the first layer.

16.3 Adsorption of Rare Gases and Simple Molecular Species

Some of the questions addressed in these studies are as follows: • On which sites on the nanotube bundles are the gases adsorbing? • What phases are present in the adsorbed film? Are any of these onedimensional? • What are the thermodynamic and structural properties of these films and how do they vary as a function of coverage?

16.3.1 Methane Methane adsorption on SWNTs has been investigated with a variety of techniques including adsorption isotherms [31-33], calorimetry [31], quasielastic neutron scattering [34], elastic neutron scattering [35, 36], temperatureprogrammed desorption [37], and NMR measurements [38]. Weber et al. [39] reported on the low coverage binding energy of CH 4 adsorbed on as-received carbon nanotubes produced by the arc-discharge process in Montpellier. Since the nanotubes were not subjected to purification or cutting, most of their caps were intact and their interior space was not available for adsorption. The binding energy of CH 4 on the SWNTs was determined from the isosteric heat of adsorption values obtained from isotherms conducted between 155 and 195 K, at coverages in the lowest one-tenth of the first layer. Low coverage measurements probe only the highest binding energy sites present in the bundles, because these sites are the ones that get occupied first. These experiments found that the binding energy of CH 4 on the highest binding energy sites on the SWNTs was 222 meV. This value is 1.76 times greater than the binding energy of CH 4 on planar graphite [40]. As is discussed in some detail in Chapter 9 in this book, there are three possible groups of adsorption sites on a nanotube bundle of close-ended tubes: the grooves, the interstitial channels (ICs) and the outer sites (OSE). The grooves are the convex valleys formed in the region on the outside surface of the bundle where two nearest neighbor tubes come closest together; the IC is the open space encircled by three nearest neighbor tubes at the interior of a bundle; and, the OSE are the outer surface of individual tubes located on the external surface of the bundle (see Chapter 9 in this book) [41]. The report by Weber et al. [39] provided two alternative interpretations for the data: that the high binding energy sites were the result of adsorption on the ICs (under less favorable conditions than those for adsorption of a smaller species) or, if the methane molecules were too large to fit in the ICs, that the high binding energies corresponded to adsorption on the grooves. Muris et al. [31] used adsorption isotherms and calorimetric measurements, to study CH 4 (and Kr) adsorbed on as-received carbon nanotubes produced by arc discharge, also from Montpellier. This study provided the first complete monolayer isotherm for any gas adsorbed on nanotube bundles. They found that for both CH 4 and Kr there were two substeps present in the first-layer data. These two substeps indicate that in the first layer adsorption occurs on two

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

410

different groups ofsites, each with a different binding energy. Their experimental results indicated that the lower coverage substep had an isosteric heat that was only 1.2 times larger than that for methane on planar graphite (considerably smaller than the 1.76 times found by Weber et al. [39]). The higher coverage substep, on the other hand, had an isosteric heat that was approximately 0.75 times that of the first layer of CH 4 on planar graphite. They interpreted their data as indicating that the lower coverage substep corresponded to adsorption in the ICs, and the higher coverage substep corresponded to adsorption on the OSEe Adsorption on the grooves was not considered when interpreting these data [31]. Bienfait's group [34] conducted a mobility study of methane molecules adsorbed on SWNT bundles using quasielastic neutron scattering. They found that the phase adsorbed at coverages corresponding to the higher binding energy sites had solid-like behavior up to at least 120 K. For coverages corresponding to the lower binding energy substep the behavior was solid-like at 50 K and became liquid-like above 70 K. The diffusion coefficients measured for this phase corresponded to a highly viscous liquid. Talapatra et al. [32] measured the binding energies for Ne, CH 4 , and Xe adsorbed on the highest binding energy sites of as-produced arc-discharge SWNTs. (The adsorption data from which these values are derived are shown in Fig. 16.1.) They found that for all three adsorbates the values of the binding energies on these sites were a factor of about 1.75 times greater than those for the same adsorbate on planar graphite. Since the increase in binding energy relative to planar graphite was the same for all three gases, the authors concluded that all three adsorbates were occupying the same type of sites on the SWNT bundles [32]. They argued that, owing to its large diameter, Xe was too large to fit in the ICs and concluded that, since all three adsorbates were occupying the same type of sites, none of the three gases was adsorbing in the ICs. The study

~ 600 .------..--......------.

--

(')5 :;400

€o / 'I' .,/ ~ 200 ./l/II~>)# / 0 -3

-2

-1 0 1 In pressure (torr)

(')~

i

I

.~.~' b /;'1 'I fl! 1)./' *

i

400

-e~ /!I:I~l. -g 200 /~ *~~«~.' C ~/••• / / / /~~ g ·.,..V

§0•" fl""" ~~~ -

E ~

600,..-------.....,.----, .9 600 ,r--------.--------. 1: ~

'C'

2

E 3~

0 -5

-4

-3 -2 -1 0 In pressure (torr)

1

(')5

i

-e~ -g §0

E 2~

400

200 0 ~--+---'--"---~--'---.l-~ -8 -6 -4 -2 0 2 In pressure (torr)

Figure 16.1 Low coverage adsorption data for left to right: xenon, methane, and neon on the single-walled nanotubes (SWNTs) at various temperatures. The temperatures are shown for xenon, from left to right: 220, 230, 240, 250, 260, 270, 280, and 295 K; for methane, from left to right: 159.88, 164.82, 169.86, 174.82, 179.84, 184.8, 189.85, and 194.68 K; and, for neon from left to right: 37.66, 40.13, 42.68, 45.11, 47.59,50.13,52.57,55.10, and 57.16K. The amount adsorbed in cm3 torr (1 cm3 torr == 3.54 x 10 16 molecules) is presented in y axis and logarithm of pressure in torr is given in x axis. The isosteric heats and the binding energies can be calculated from these data. (Reprinted figure with permission from Ref. [32]. Copyright (2000) by the American Physical Society.)

16.3 Adsorption of Rare Gases and Simple Molecular Species

411

also compared the values for the specific surface area of the SWNT sample measured using Ne and Xe isotherms and concluded also from these results that these gases were not adsorbing in the les. As had been noted previously in [1], Talapatra et al. had used incorrect values for the determination of the monolayer capacities in ref [32]. When the correct monolayer capacities are used, the SWNT sample had a specific area of 161.1 m 2 / g as determined with Xe, and of 173.1 m 2 /g as measured with Ne. Since these values are very close, their original argument remained unaltered, even after the reported values of the specific areas were revised. To identify the nature of the high binding energy sites, Talapatra et al. [32] compared their results to theoretical calculations [41]. These calculations determined the ground-state energies of single molecules of various adsorbates on the different adsorption sites present on homogeneous nanotube bundles. The binding energy on the OSE was approximated by adsorption on a graphene sheet (i.e., a single planar sheet of graphite) [41]. The ratio of the energy on the grooves to that on the graphene sheet was approximately 1.65 for the various gases studied in the calculations. Since this ratio was close to the ratio found in the experiments between the highest binding energy sites on the SWNT bundles and the experimental values for gases adsorbed on graphite, Talapatra et al. [32] concluded that the high binding energy sites on the SWNTs were the grooves. (The actual values for the binding energies measured in the experiments, however, were larger than those obtained in the theoretical calculations. The energies for one single graphene sheet are lower than those measured on graphite because graphite has contributions from many graphene sheets. Additionally, experiments on films include contributions from interactions between adsorbate molecules.) The issue of where on the nanotube bundles do different gases adsorb was examined further by Muris et al. [42]. They compared previous adsorption results for CH 4 [31] to those for Xe, CF 4 , and SF 6 on the same substrate. Specifically, this study compared the size of the two substeps present in the first-layer data for the different species (or, in the case of the larger adsorbates CF 4 and SF 6 , the absence of a lower pressure step) to identify the sites on which adsorption was occurring [42]. In a CH 4 isotherm at 77 K they found that the two first-layer substeps were nearly the same size (the higher pressure substep comprised a 1.3 times larger coverage interval than the lower pressure one). For Xe the lower pressure substep was about one-third the size of the higher pressure substep; for CF 4 the lower pressure substep was barely discernible; and for SF 6 it was altogether absent. On the other hand, the sizes of the higher pressure substeps (i.e., the ones that occur at pressures higher than those for the respective first layers on planar graphite) were all more or less comparable. The authors concluded that for methane, the lower pressure substep corresponded to adsorption inside the largest ICs present in the bundles and on the grooves, while the higher pressure substep corresponded to adsorption on the OSEe For Xe, the small low pressure substep corresponds to adsorption on the grooves only (this adsorbate was considered too large to penetrate in the lCs), and the larger high pressure substep corresponded to adsorption on the OSEe For CF 4

412

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

and SF 6 the substep corresponding to adsorption on the OSE is all that is present in the data. The smoothing of the lower coverage substep corresponding to adsorption on the grooves (to the point of disappearance in the case of SF 6 ) was attributed to the fact that larger molecules experience less corrugation in the substrate potential than do smaller adsorbates. It should be noted, however, that the presence of a relatively large low pressure substep in the methane data could be the result of lack of equilibrium, with data points accumulating above their equilibrium pressure, hence producing the appearance of a large substep. Talapatra and Migone [33] conducted a detailed study of CH 4 adsorbed on SWNTs. Their results confirmed several of the observations of Muris et al. [31]: There are two sub-steps in the first-layer data, corresponding to the existence of two groups of adsorption sites. The identification of the higher pressure substep as corresponding to adsorption on the OSE, proposed by Muris et al. [31], was confirmed. On the other hand, disagreement regarding the interpretation of the lower pressure step persisted. Talapatra and Migone [33] interpreted this feature as corresponding to adsorption on the grooves. They also explored the coverage dependence of the isosteric heat of adsorption and found that this quantity is a decreasing function of increasing coverage. With this result they were able to explain the apparent discrepancy between values of the binding energy on the low coverage, highest binding energy sites for CH 4 on SWNTs, which had been reported as being, respectively, 1.76 greater than on planar graphite by Weber et al. [39], and only 1.23 times greater than on planar graphite in Muris et aI.'s work [31]. The study concluded that this difference was the result of the data being measured over different coverage regimes in the two studies. Weber et al.'s data [39] extended to less than one-tenth of a monolayer, while Muris et al. [31] probed a somewhat higher coverage region, with correspondingly lower binding energies. Thus, no true disagreement existed in the experimental values for the binding energy. Temperature-programmed desorption measurements for CH 4 on SWNTs produced by [email protected] were conducted by Hertel's group [37]. The samples were vacuum-annealed at 1200 K. This study compared the results obtained for several adsorbates (including CH 4 , Xe, and SF 6 ), on highly oriented pyrolytic graphite (HOPG) and on SWNTs. For methane they found that the desorption peak on HOPG is near 50 K, while that of the highest binding energy sites on the nanotubes is at approximately 110 K. The binding energy on the nanotubes was estimated at approximately 140 % that on H 0 PG. They concluded that the data were consistent, with the grooves being the highest binding energy sites present [37]. Shi and Johnson [43] added a new theoretical perspective to the question of determining where on the nanotube bundles are the gases adsorbed, by proposing the heterogeneous bundle model. In this model, the bundles are constituted by rigid tubes of different diameters. This results in packing defects. Corresponding to these packing defects are a few, rather large, ICs. Gases as large as Xe can adsorb on these few, stacking defect-induced, large ICs. However, the majority

16.3 Adsorption of Rare Gases and Simple Molecular Species

4 13

of the ICs present on the nanotube bundles are small-diameter channels, and adsorbates such as Xe or CH 4 should not be able to penetrate them. It is not easy to get the type ofstructural information needed to unambiguously confirm the validity of the heterogeneous bundle model [44]. The best data in support of this model come from comparing the results of the coverage dependence of the isosteric heat obtained in the model to the isosteric heat of adsorption data for methane (as well as those of Xe and Ar) [43]. The heterogeneous model is better at reproducing the high values of the isosteric heat of adsorption measured at the lowest coverages than the homogeneous bundle model (in which all the tubes are of the same diameter). There have been two structural studies of CD 4 adsorbed on SWNTs by Bienfait's group [35,36]. The first was a preliminary investigation in which two coverages were explored [35]. This study noted that the main changes from the empty background diffraction pattern upon adsorption of CD 4 were an increase in the intensity of the diffraction peak at 0.4 A-1 , and the appearance of a broad peak at 1.8A-l. The peak at 0.4A-l was observed to shift by about 5% to lower Q values upon increase in coverage. This was interpreted as either the result of adsorption inside the ICs, which resulted in the deformation of the structure of the bundle (however, see the discussion concerning Ar diffraction measurements, in the next section in this chapter), or the result of adsorption on the grooves, in which the methane molecules sit at positions that are displaced by 1 A outward from the lattice hollow sites [35]. The second study is a recent report on combined neutron diffraction and computer modeling studies of CD 4 on SWNTs [36]. The elastic neutron scattering measurements were conducted at four different coverages between one-third of a layer and monolayer completion. The simulations were performed using three modules from the Cerius2 suite of molecular modeling programs. One of their main objectives was to reproduce the results of the scattering measurements; in particular, to account for the coverage dependence of intensity changes and position shifts of the peak at 0.4 A-1, which is related to the hexagonal packing of the nanotubes. These simulations adopted a new model, based on considering deformable nanotubes [36]. As a result of tube deformation, there are no obvious voids in the bundles (i.e., no structures such as the large defect-induced ICs of the rigid heterogeneous model [43]). The deformable bundles are constructed by placing tubes of different sizes at random in a hexagonal structure. This is followed by a geometry optimization step in which there is significant deformation of the individual tubes (ovalization). These authors found that the experimental data for the 0.4 A-1 peak, and its changes upon adsorption, were best reproduced using a heterogeneous bundle of deformable nanotubes [36]. Binding energies were obtained by calculating the energy of methane inside each of the ICs present in an ovalized bundle. This was done first introducing two methane molecules in a particular IC, filling the IC, and re-optimizing the structure. The binding energy was calculated by computing first the optimized energies for the tubes that make the chosen IC, and the adsorbates separately, and

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

calculating their sum; then calculating the energy of the combined structure of the methane molecules inside the IC; and finally subtracting this last value from the initial sum. The resulting difference is the binding energy of the methane in that IC [36]. These simulations found that the methane molecules induce an additional local deformation in the tubes. It would be highly desirable to have other quantities simulated in this model, as they would allow a more thorough testing of the model through comparisons with the available experimental data (e.g., quantities such as adsorption isotherms at several temperatures, or the coverage dependence of the isosteric heat of adsorption). These simulated data are not presently available because the heterogeneous deformable bundle model is very recent. The behavior of CH 4 (and that of ethane) adsorbed on purified, ultrasonically cut bundles of SWNTs produced by laser ablation has been studied using NMR at room temperature and pressures below 1 MPa [38]. The nuclear 1H spectrum shows distinct features corresponding to methane (and ethane) in the gas phase vs methane and ethane adsorbed in the interior of the cut tubes. Adsorption at the interior of the cut nanotubes (endohedral adsorption) dominates the behavior at the pressures and temperatures studied. The effects of having O 2 present during the adsorption measurements for methane and ethane were explored. It was determined that O 2 does not affect endohedral adsorption for either gas because the binding energy of O 2 is smaller than those of methane or ethane [38].

16.3.2 Argon Argon films have been investigated with adsorption isotherms [45-47], and elastic neutron scattering measurements [48]. Vilches' group [45] investigated Ar adsorption on as-produced arc-discharge nanotubes. Three temperatures were explored between 77.3 and 96 K. The coverage dependence of the isosteric heat of adsorption was obtained. As is the case for CH 4 and other gases, the isosteric heat is a decreasing function of coverage, having its maximum value as the coverage approaches o. In the first layer the isosteric heat exhibits a plateau region at value that is lower than that for the first layer of Ar on graphite. Migone's group [46] investigated the existence of different phases within the first layer for Ar and Xe films adsorbed on close-ended nanotubes with adsorption isotherms. The measurements were performed on as-produced arcdischarge nanotubes, between 59.91 and 87.14K. The data were measured at closely spaced coverage intervals to enable the reliable calculation of the derivatives of the isotherms (a quantity proportional to the isothermal compressibility of the film). Computer simulations of adsorption isotherms for Ar on rigid homogeneous bundles predict the existence of three different phases in the first layer [49]. The simulations found that, at low temperatures, the film grows by the formation of successive "lines" or "channels" of atoms on the outer surface of the SWNTs.

16.3 Adsorption of Rare Gases and Simple Molecular Species

At the lowest coverages, the Ar atoms form a 1-D "one-channel" phase, filling the grooves. This is followed by the "three-channel" phase in which the Ar atoms form two lines, on each side of the grooves, at somewhat higher pressures. (The three-channel phase is no longer resolvable in simulations above 60 K.) Monolayer completion is reached in the "six-channel" phase, where Ar atoms fill the remainder of the outer surface of the bundles [49]. The adsorption experiments found evidence to support this sequence [46]. The two highest temperatures investigated (82.17 and 87.14 K) displayed two substeps in the first layer. The pressure of the midpoint of the lower pressure substep corresponded well with the values of the pressure at the midpoint of the one-channel phase in the grooves from the simulations. These two isotherms were the only ones conducted at sufficiently low coverages to allow the investigation of the groove region. There were no distinct substeps indicative of the three-channel phase in the lowest two temperatures investigated (59.91 and 63.16 K). However, the derivatives of these two isotherms had two peaks in a coverage region corresponding to the three-channel and six-channel phases. The lower pressure peak in the derivative was identified with the three-channel phase and the higher pressure peak with the six-channel phase. The pressures at which these features were found corresponded well with those from the simulations. Isotherms measured at higher temperatures displayed only one peak in the derivative, consistent with the fact that the three-channel phase is not resolvable at higher temperatures [46]. In a separate set of measurements, Talapatra eta!' [47] explored the coverage dependence of the isosteric heat of adsorption of Ar in the first layer. Data for coverages between 3 and 8 % of the first layer were taken between 110 and 161 K (for lower temperatures the pressures for these coverages fall below the range accessible to the setup), while data for coverages above 40 % of a layer were measured between 57 and 87 K. This study found, consistent with previous reports [45], that the isosteric heat is a decreasing function of coverage, with a plateau region. The values of the isosteric heat at low coverages are larger than those on planar graphite by a factor of 1.7. Second-layer isotherms were measured between 48 and 55 K in this same study [47]. There is a barely resolvable substep at the foot of the second-layer Ar data [47]. This feature shows up more clearly as a peak in the derivative of the adsorption data. It was identified with the formation of a "groove" phase in the second layer. It corresponds to Ar atoms filling up a groovelike region in the second-layer film. A similar feature was found in computer simulations [50]. Bienfait's group [48] has reported on neutron scattering studies performed with 36 Ar and 40 Ar on SWNTs conducted to determine whether the nanotube bundle lattice "dilates" upon adsorption of the gas. 40 Ar is essentially "transparent" to neutrons, so studying 40 Ar adsorption allows the determination of the effects of adsorption on the SWNT substrate; 36 Ar, on the other hand, is quite visible to neutrons, and hence the structure of the adsorbed film can be studied with this isotope. As discussed for CH 4 , changes are observed upon

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

(a) 10

8

1

~6 ·c ::J

.ci ~4 ~

.~ 2 ....-..+~-f-L--+----+--+--+-~I--+--+----1

~

2

I

(b)

I I I

I I

.

~

o

.I

~.

I I

-_._

: .

.

I

I

-1

L.......L.._......a...-_........-..;.L._---I...._....a....--..L!_ - - & . . . . _ . . 1 . . - - - - - '

234

5

Wavevector transfer Q (A-1)

Figure 16.2

(a) Neutron diffraction spectra of the bare single-walled nanotube bundle (SWNTB) sample (broken line) and upon adsorption of 2.7mmol/g (solid line) of 36Ar. Vertical dotted lines: Positions of the Bragg peaks expected for a hexagonal packing of nanotubes with a lattice spacing of 17 A. (b) Solid trace: Difference between the diffraction spectra in (a), revealing the changes induced upon adsorption. Dotted trace: Same but for adsorption of 2: 7 mmol/g of 40 Ar. Vertical dashed and dotted lines indicate expected Ar peak positions. (Reprinted figure with permission from Re£ [48]. Copyright (2003) by the American Physical Society.)

adsorption in the positions of some of the SWNT diffraction peaks, and one possible interpretation is that they are the result of dilation in the nanotube lattice occurring when the Ar atoms penetrate in the les. The measurements were performed on arc-discharge nanotubes. Figure 16.2 presents the scattering results for the two Ar isotopes on the SWNTs. Four Bragg peaks were resolved in the data from the bare SWNT substrate, prior to the adsorption of Ar. In measurements performed with 36 Ar, shifts are detected in a scattering peak corresponding to the medium-range order of the tubes in the bundle. These shifts correspond to an increase in the "effective" lattice parameter of the bundle of about 3 %. When the measurements were repeated with 40 Ar, however, the apparent lattice dilation disappears: the data for 40 Ar and those for the bare nanotubes are essentially indistinguishable [48]. This indicates that there is no dilation in the SWNT lattice upon adsorption. The apparent shift in the lattice parameter is the result of diffraction from Ar atoms located on the outer surface of the bundle. These results prove that a previous X-ray diffraction study for N 2 and O 2 by Fujiwara et al. [51], in which similar shifts in the diffraction peaks had been

16.3 Adsorption of Rare Gases and Simple Molecular Species

attributed to dilation of the bundles as a result of adsorption in the ICs, was wrong in its identification of the reason for the shifts. That study [51] had failed even to consider the possibility of adsorption on the outer surface of the bundles in their analysis, incorrectly attributing all adsorption to the ICs (for close-ended tubes).

16.3.3 Helium Helium has been studied with temperature-programmed desorption [3, 52-55], adsorption isotherms [56], calorimetry [57-60], and NMR [38]. Hallock's group [52] conducted temperature-programmed desorption experiments for 4He on SWNTs. The substrate used was purified laser ablationproduced SWNTs. After vacuum degassing at 800 K, the SWNTs were kept under vacuum in a sealed glass container that was broken only to expose the nanotubes to 4He just prior to the measurements. After helium was dosed at 170 K, the sample was cooled to the starting temperature for the desorption measurements. The sample was pumped at this temperature to remove all the helium gas and the helium adsorbed on weaker binding sites prior to beginning the measurements. The binding energy was obtained from the dependence of the total amount of 4He desorbed as a function of temperature. A binding energy of 330 K was found, in good agreement with the value expected for helium inside the ICs [52]. Subsequently, the temperature scale used in the measurements was re-examined [53], resulting in a new fit to the data that yielded a binding energy at 210 K. Hallock's group [54] conducted a second set of thermal desorption experiments in which they extended upward the temperatures from which the desorption runs begin. They observed that there are two peaks in the desorption data: one near 140 K, which was identified as probably being produced by impurities in the sample, and another peak at lower temperatures. The temperature of this second peak varied depending on the amount of gas adsorbed on the sample. Recently, this group has explored the competitive adsorption of helium and hydrogen using temperature-programmed desorption [55]. The measurements compared the desorption spectra obtained when only helium was dosed into the cell, those when only hydrogen was dosed in, and those obtained when a 50:50 mixture of hydrogen and helium was dosed into the cell. The results obtained when only hydrogen is dosed are indistinguishable from those obtained when a 50:50 mixture of hydrogen and helium is dosed. The conclusion obtained was that the hydrogen binds more tightly to the SWNTs than helium. These experiments did not determine whether adsorption was occurring on the ICs or on the grooves [55]. Vilches' group [56] has reported adsorption isotherms for helium on SWNTs produced by arc discharge. The measurements were conducted between 2 and 14 K. The data measured at lower temperatures extended to the second layer. The isotherm at 2.1 K shows three layers in the data. This is the only case for which three distinct layers have been reported for adsorption on SWNTs.

418

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

This study determined the coverage dependence of the isosteric heat of adsorption for 4He. This quantity was found to be a decreasing function of coverage. The low coverage value of the isosteric heat approaches 240 K, significantly higher than on planar graphite. After a region of rather steep decrease at lower coverages, there is a plateau region in the first layer that occurs at an isosteric heat value lower than that for the first layer of helium on planar graphite. The plateau region extends for most of the first layer. As the coverage increases beyond the first layer, there are other plateaus present in the isosteric heat, corresponding to the formation of a second layer, and even a third layer [56]. There is a recent report, also from Vilches' group [60], of specific heat and adsorption isotherm data for 4He on HiPco SWNTs. The general features of the isotherms measured on the HiPco samples were similar to those on the arc-discharge nanotubes. However, the values of the isosteric heat of adsorption at low coverages were considerably higher on the HiPco tubes. Vilches' group [60] has performed calorimetric measurements for helium films at several different coverages in the first layer. The heat capacity of the film was obtained as the difference between the total heat capacity measured for a given coverage and the heat capacity of the calorimeter when no gas is adsorbed on it. They report that none of the coverages has a region where the heat capacity is constant with temperature. Hence, there are no regions in the data identifiable as either a 1-D or a 2-D ideal gas. In all cases the heat capacity was observed to increase with temperature. The data can be fitted to an expression of the form C = 0: T + r3 T 2 with both coefficients being coverage dependent. These authors proposed a possible scenario for the growth of the fist layer of helium on the SWNTs that involves first the occupation of the grooves, followed by the formation of lines of atoms parallel to the grooves in what was identified as a three-channel phase for Ar (there is a peak in the isosteric heat for 4He that could be associated with the latent heat of condensation of this phase), and finally first-layer completion concludes with the filling of the remaining surface of the nanotubes, at the outer surface of the bundle [60]. There have been other reports of heat capacity measurements of helium on carbon nanotubes [57-59]. Some of these resulted from studies in which the main aim was to investigate the heat capacity of the carbon nanotubes themselves (helium was used as the exchange gas to help cool down the carbon nanotubes) [57, 59]. One of the groups followed on their previous results with further investigations of 4He adsorption on SWNTs [58]. This second study used two SWNT samples: one prepared by arc discharge and the other by laser ablation. Through repeated cycles of heating and pumping the authors established the value for specific heat of the bare nanotube substrate. The film's specific heat was then determined by subtracting the contribution from the bare nanotube substrate. The heat capacities of films for two different coverages were determined for each one of the two substrates used [58]. In the highest coverages studied it was estimated that helium corresponded to 3 at. % (laser ablation sample) and 1.5 at. %

16.3 Adsorption of Rare Gases and Simple Molecular Species

(arc-discharge sample) of C, respectively. For these coverages the data follow a close to linear temperature dependence between 1 and 5 K, with a crossover to a stronger dependence at lower temperatures. Following a partial outgassing procedure, intermediate lower coverages were obtained. The coverages on these partially outgassed films are not very well determined [58]. They were estimated using published data from temperature-programmed desorption experiments [52-54]. The heat capacity of these films is quite different on the two samples: for the laser ablation sample, which was partially outgassed at 25-30 K, the film follows y2 behavior at lower temperatures and saturates to an estimated heat capacity of about lkB at higher temperatures (i.e., it exhibits 2-D behavior). By contrast, the film on the arc-discharge sample, which was outgassed at 15 K, follows a 1-D Einstein model behavior at lower temperatures and saturates to 0.5kB at higher temperatures (i.e., it displays 1-D behavior). The authors explained that the observed differences were due to differences in the size of the nanotube bundles. More careful coverage determinations are needed to support this interpretation. An NMR study [38] explored whether various gases can access the ICs at room temperature. The authors investigated access to the ICs through the effect that the presence of different gases has on the saturation recovery curves for 13C NMR on a sample of uncut laser ablation SWNTs. They found that exposure of the nanotubes to 0.17 MPa of H 2 , N 2 , or CO 2 leaves the recovery time unaffected relative to measurements of this quantity performed under vacuum. On the other hand, exposure of the SWNTs to 0.1 MPa of 4He noticeably reduces the recovery time. The interpretation of these results was that, unlike H 2 , N 2 , or CO 2 , helium can access the majority of the carbon atoms in the uncut nanotubes, and thus, it can provide a channel for the relaxation of the 13C nuclear spins. They concluded that helium has access to the ICs, while the other gases studied (H 2 , N 2' or CO 2 ) cannot access these sites [38].

16.3.4 Hydrogen In addition to the studies that deal with the capacity ofSWNTs to store H 2 , there are also studies that investigate the adsorption characteristics and properties of the first layer of H 2 adsorbed on SWNTs. We summarize some of these studies here. Vilches' group [45] has investigated with adsorption isotherms H 2 and D 2 films on as-produced arc-discharge SWNTs. Measurements were conducted over two temperature regimes: above 77 K, in order to determine the isosteric heats of adsorption of the low coverage, high energy binding sites present on the SWNTs, and below temperatures 45 K to explore adsorption at higher coverages, on the lower binding energy sites in the first layer. The values of the isosteric heat for D 2 are higher than those for H 2 , for all coverages. The isosteric heat is a decreasing function of coverage. It exhibits a rather large plateau region that corresponds to roughly 75 % of the first layer. In this plateau the isosteric heat values are lower than those on the first layer on planar graphite. At the

420

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

lowest coverages studied, the values of the isosteric heat are 1.5 times greater than on graphite for H 2 and 1.8 times greater than on graphite for D 2 . D 2 films have been explored by Bienfait's group [35] with neutron scattering. One peak corresponding to adsorption of D 2 is visible in the data. Analysis of this diffraction peak leads to a model in which, at low coverages, adsorption occurs in a zigzag chain in the grooves or in large interstitials; this then evolves to a quasihexagonal packing at higher coverages. Brown et al. [61] have reported on an inelastic neutron scattering study of H 2 on as-produced laser ablation nanotubes. The measurements were performed on a H 2 -loaded sample for temperatures between 25 and 65 K. By comparing the characteristics of the energy loss peaks with those observed from the gaseous phase, those for solid hydrogen, and those for H 2 adsorbed on graphite, the authors concluded that the data for H 2 on SWNTs indicated physisorption of this species on the curved exterior surface of the nanotubes. By analyzing the exponential decrease in the integrated intensity as a function of increasing temperature, they concluded that H 2 adsorption is occurring on the external cylindrical surface of the tubes, and not in the ICs. Sokol's group [62] conducted quasielastic neutron scattering measurements on H 2 adsorbed on commercially available arc-discharge nanotubes, to investigate the diffusion of H 2 between 20 and 45 K. No quasielastic component was observed below 30 K, indicating that if there is diffusion present, it is slower than the instrumental resolution. Between 30 and 45 K the data are well described by a liquid-like jump diffusion model. They interpreted the data as indicating that the H 2 molecules were preferentially adsorbed on the grooves, and that the molecules escaped from the grooves either directly to the gas phase without diffusion along the grooves, or from the grooves to the rest of the surface of the bundle, where they can freely diffuse along the surface. Schimmel et al. [63] have conducted adsorption isotherm and inelastic neutron scattering studies of hydrogen adsorbed on various carbon adsorbents including SWNTs. By comparing the inelastic neutron scattering results for H 2 on the SWNTs with those for interstitially loaded H 2 on C 60 , they conclude that H 2 cannot adsorb in the ICs of SWNTs. They explain that this occurs because the molecular size of H 2 is too large to fit in the estimated size for the ICs.

16.3.5 Xenon The first adsorption studies on Xe determined the binding energy of this gas on the highest energy binding sites in the nanotube bundles, as summarized earlier in the section on methane [32]. A binding energy 1.74 times larger than that for Xe on planar graphite was found. Complete isotherms measured 138 and 150 K show that the first layer for Xe consists of two substeps [64]. By comparing the coverages at the completion of each of these substeps to the results of geometric calculations conducted for different scenarios, the authors concluded that Xe does not adsorb on the ICs.

16.3 Adsorption of Rare Gases and Simple Molecular Species

421

Migone's group [46] also conducted more detailed studies of the different phases present on the Xe films. They compared their first- and second-layer data (between 112 and 150 K) to computer-simulated isotherms for this system. The experimentally measured values for the temperature dependence of the midpoint pressure of the two substeps in the first layer, and that for the midpoint pressure of the second-layer step agreed very well with the values for these same quantities obtained in the simulations. The lower pressure substep in the first layer was identified as a one-dimensional phase formed by Xe adsorbed in the grooves. The existence of a second-layer groove phase for Xe was also explored [65]. Computer simulations had found that the corrugation in the substrate potential is not sufficiently strong in the case of a larger adsorbate such as Xe to produce a distinct second-layer groove phase [66]. The experiments found that, unlike the cases for Ne [65] or Ar [47] where sharp features are present in the isotherm data and in their derivatives, for Xe there is only a rather weak feature found in the compressibility at the beginning of the second layer [65]. This result qualitatively agrees with the expectations from the simulations for Xe. Yates' group [67] used IR spectroscopy, temperature-programmed desorption, and mass spectrometry to study Xe adsorption on purified and cut SWNTs. The nanotubes were cut by subjecting them to a mixture of sulfuric and nitric acid treatment, followed by sonication with sulfuric acid and peroxide. Infrared (IR) measurements determined the presence of carboxylic acid and quinone groups on the treated tubes. Mass spectrometry of treated tubes heated under vacuum determined the evolution of different groups from the tubes as the temperature increased (CH 4 , CO, H 2 , and CO 2 ). This group determined that the nanotubes' capacity for adsorbing Xe was greatly enhanced as a result of the vacuum heating treatment at 1073 K [68]. The chemical treatment to which the tubes were subjected in the cutting process resulted in chemical groups being attached to, and blocking the entry ports to the open tubes. The heating process under vacuum removed or destroyed these groups, and, as a result made the interior of the tubes available for adsorption, resulting in an increased sorptive capacity [68]. This increased capacity was established through the performance of temperature programmed desorption measurements for Xe on the cut and vacuum heat-treated nanotubes. Hertel's group [69] has explored the desorption kinetics ofXe from SWNTs. The nanotubes used were commercially available purified buckypaper from [email protected] The samples were not subjected to opening treatment. The samples were outgassed at 1200 K under ultra-high vacuum in repeated annealing cycles prior to performing the adsorption experiments. From the desorption measurements the authors concluded that the low coverage binding energy ofXe is higher than that on graphite, in good agreement with previous results [32, 64]. By comparing their results to a model for the thermal desorption spectrum, they concluded that the data is consistent with the Xe atoms adsorbed on the grooves, forming a nearly ideal 1-D phase [69].

422

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

16.3.6 Neon Neon films have been explored by Migone's group in two sets of adsorption isotherm measurements conducted on as-produced arc-discharge nanotubes. As discussed in the methane section earlier, the low coverage value of the isosteric and the binding energy of Ne on the highest energy binding sites on the SWNT bundles were determined from a set of measurements conducted between 37.66 and 57.61 K, at coverages below one-tenth of a layer [32]. The binding energy was found to be greater than that on planar by a factor of approximately 1.7. Computer simulations had predicted the existence of a clearly resolvable second-layer groove phase for this adsorbate [66]. In the simulations neon occupies the second layer in two stages: first, Ne atoms form a second-layer groove phase (occupying the remnants of the grooves in the second layer), and then they fill up the rest of the surface of the second layer. An adsorption isotherm study has investigated second-layer Ne films [65]. The experimental results provide a confirmation of the qualitative picture presented by the simulations: there is a small substep present at the foot of the second layer in the adsorption data. This feature was identified as corresponding to the second-layer groove phase. The rest of the second layer fills up at higher coverages. The values for the pressure found at the location of these second-layer features in the experiments do not agree quantitatively with the simulations [65]. Very recent measurements, conducted on a sample of HiPco nanotubes, have confirmed the results obtained for the second-layer Ne films on the lower purity arc-discharge nanotubes [70].

16.3.7 Tetrafluoromethane There have been adsorption isotherm [42] and IR spectroscopy studies [71] of CF 4 adsorbed on SWNTs. The isotherm results have been partly described in the subsection on CH 4 [42]. Unlike the case for smaller adsorbates, where there are two substeps present in the first-layer data, for CF 4 the lower pressure step is barely discernible. There is, however, quite a considerable amount of CF 4 adsorption occurring on the substrate (corresponding to roughly 40 % of the first layer) prior to the formation of the higher pressure substep. The higher pressure substep present for CF 4 is comparable to those for other adsorbates. The isosteric heat for CF 4 on the higher pressure substep, which corresponds to adsorption on the OSE, has a value of 15.3 kJ/mo!. As expected, this value is lower than the corresponding value of 19 k]/mol for the first layer of CF 4 on graphite [40]. Yates' group [71] used CF 4 to investigate adsorption on the inside and outside surface ofSWNTs. These investigations were conducted using IR spectroscopy. This group has developed a method for opening the ends of the carbon nanotubes, and for cleaning up the residual species that result from the opening process that allows adsorbates to have access to the interior of the nanotubes.

16.3 Adsorption of Rare Gases and Simple Molecular Species

42 3

The laser ablation SWNTs used in these studies were purified and heat-treated under vacuum. To open the tubes, an oxidation process with ozone was used. It removes the caps at the ends of the tubes and opens sites on the walls of the tubes. Oxidation was followed by vacuum annealing in order to guarantee access to the interior of the tubes. The samples were subjected to successive cycles of ozone exposure and vacuum annealing to 873 K to achieve successive degrees of tube opening. The IR spectrum that results from dosing CF 4 onto the SWNTs was measured before and after the ozone opening treatment/vacuum heating cycle. There is a vibrational frequency in the spectrum of CF 4 that experiences different relative shifts with respect to the values that it has in the gas phase when the CF 4 molecules are adsorbed on the outside, or on the inside of the nanotubes. The shifted value of the frequency corresponding to adsorption on the interior of the nanotubes is only present on the etched nanotubes; that corresponding to adsorption on the outer surface is present on both etched and nonetched tubes. (The traces corresponding to the unetched and etched spectra as a function of increased CF 4 exposure are presented in Fig. 16.3.) Additionally, the intensity 0.006 C\JT'""

(a)

",I

0.005

C\J

T=133K

or-

E

0

0.004 0.003 0.002 0.001 Q)

0

c

External CF4 Before etching

0.000

~

-e0 en

.0

1400

1350

1300

1250

1200

1150

0.012 C\JT'""

cd::

~I

0.010

T=133K

"I

C\J or-

(b)

E

0

E

0 C\J

0.008

co C\J or-

0.006 .~

0.004 0.002

;'-

External + internal CF4~~ After etching

0.000 1400

1350

1300

1250

1200

1150

Wavenumber (cm- 1)

Figure 16.3 Development of CF4 infrared (IR) spectra for (a) external sites and (b) external plus internal sites. (Reprinted with permission from Re£ [71]. Copyright (2003) American Chemical Society.)

424

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

of the vibrational peak increased with exposure to CF 4 upon successive etching cycles. These two observations allowed the unambiguous identification of this shifted frequency as corresponding to CF 4 adsorbed on the inside of the etched nanotubes [71].

16.3.8 Nitrogen N 2 films on SWNTs have been explored by several investigators using adsorption isotherms [45, 72-76]. Vilches' group [45] studied the coverage dependence of the isosteric heat of adsorption of N 2 on as-produced arc-discharge nanotubes. They found a behavior similar to that for other gases: a sharp increase in the isosteric heat as the coverage decreases, and a plateau region, corresponding to adsorption of the lower energy OSE; however, the plateau region for N 2 was smaller than for that found for other gases. Kim's group [72] investigated N 2 adsorption isotherms on both close-ended and open-ended [73] laser ablation-produced nanotubes. The tubes were opened by sonication in a mixture of nitric and sulfuric acids. The treated tubes were heated to either 873 or 1073 K after the uncapping treatment and were transported, in air, to the cell in which the adsorption measurements were performed. Uncapped tubes heated to 1073 K have 50 % larger adsorptive capacity than those heated to 873 K, and 200 % larger adsorptive capacity than untreated tubes [73]. The highest value of the isosteric heat of adsorption (measured at low coverages) is a factor of 2 larger on the uncapped nanotubes than it is on the untreated tubes. Like it was reported for other gases, the isosteric heat is a decreasing function of coverage. Unlike previous reports, the data from this group for N 2 do not exhibit any plateau region in the first layer for either capped or uncapped nanotubes. There have been several studies in which N 2 adsorption has been used to determine the effect that purification treatments have on the specific surface area of the bundles [74-76]. Kaneko's group [74] used N 2 isotherms to study the effect that purification treatments with HCI, and HCI plus air oxidation, had on the effective specific surface areas of HiPco nanotube bundles. They found that the total area of the SWNTs increases from 524 m 2 / g for the pristine sample to 587 m 2 / g for the HCI-purified tubes, and to 861 m 2 / g for the air-oxidized and HCItreated nanotubes. There is also an increase in the size of the hysteresis loops in adsorption-desorption cycles, indicating that the pore volume increased as a result of the purification process. Du et al. [75] investigated the surface area of purified and pristine HiPco nanotubes by performing N 2 and Ar adsorption isotherms. Interestingly, this study found that there were significant differences in the specific surface areas of the pristine HiPco samples, even when their reported impurity levels were similar. These authors analyzed their data using the Horvath-Kawazoe equation

16.4 Conclusions

and found that the results of this analysis were also significantly different for the two pristine samples. Cinke et al. [76] have studied the effects on the specific surface area of the nanotubes of a purification process consisting of debundling the nanotubes (subjecting them to a dimethylformamide/ethylene diamide treatment), followed by an HCl treatment and wet oxidation. The area for the pristine HiPco nanotubes is 577 m 2 / g; that for the tubes subjected just to the wet oxidation and HCl treatment is 968 m 2 / g; and that for tubes subjected to the full two-step process is 1587 m 2 / g. This study also found an increase in the size ofthe hysteresis loops in adsorption-desorption cycles as a result of the purification process.

16.4

CONCLUSIONS

As a result of the work done on films adsorbed on SWNTs some points have become clear: • For close-ended nanotube bundles, adsorption on the outside surface of the bundle is of great importance; it accounts for either the totality or the majority of the adsorption, depending on the interpretation of the data. • There are different groups of adsorption sites present in the nanotube bundles, and their presence manifests itself as different substeps in the data for the first and second layers of gases with molecular diameters equal to or smaller than that ofXe. • The isosteric heat of adsorption reflects the different sites present. In the first layer it is a decreasing function of coverage. The highest values of the isosteric heat (corresponding to the highest binding energy sites on the SWNTs) are on the order of 1.5 to 1.8 times larger than the values found for the same adsorbate species on graphite. • There are several different phases present in the first layer of films adsorbed on SWNT bundles. Smaller adsorbates also have at least two different phases in the second layer. • Some of the phases that have been identified on the SWNTs are effectively one-dimensional. On the other hand, some of the questions that the experiments tried to address continue to be asked today. Chief among these is the question of where on the nanotubes is adsorption occurring. This is especially true for the high energy binding sites. Experiments have not yet resolved the question of what gases, if any, can adsorb on the ICs. The question of whether dilation occurs upon adsorption, or under which conditions will it occur if it does occur at all, is another one that requires further experimental investigation.

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

426

Collectively, experimentalists apply different techniques to study the same systems on similar substrates. Comparisons between experiments and simulations would be greatly enhanced if: in an analogous manner, the same bundle models were used in different sets of simulations to obtain a variety of properties. Comparing the results of simulations for the same bundle model for several properties at the same time (e.g., diffraction pattern, coverage dependence of the isosteric heat, monolayer and bilayer adsorption isotherm data, and heat capacity) would provide an exacting test for the models and would further our understanding of these systems. Much exciting work remains ahead.

ACKNOWLEDGMENTS

The author wishes to acknowledge financial support from the National Science Foundation, through Grant No. DMR-0089713. Helpful discussions with M.M. Calbi, M.W. Cole, J.K. Johnson, and O.E. Vilches are gratefully acknowledged. Assistance with the preparation of this manuscript was provided by V. Krungleviciute and S. Talapatra.

REFERENCES

1. Migone, A.D. and Talapatra, S. (2004). Gas adsorption on carbon nanotubes. Encyclopedia of Nanoscience and Nanotechnology, Vol. 4. American Scientific Publisher (and all the references cited therein). 2. Calbi, M.M., Cole, M.W., Gatica, S.M., et al. (2001). Colloquium: condensed phases of gases inside nanotube bundles. Rev. Mod. Phys., 73, 857-65. 3. Hallock, R.B. and Khang, Y.H. (2004). Adsorption of helium and other gases to carbon nanotubes and nanotube bundles.]. Low Temp. Phys., 134,21-30. 4. Chambers, A., Park, C., Baker, R.T.K., and Rodriguez, N.M. (1998). Hydrogen storage in graphite nanofibers.]. Phys. Chern. B, 102, 4253-6. 5. Dillon, A.C., Jones, K.M., Kiang, C.H., et al. (1997). Storage of hydrogen in single-walled carbon nanotubes. Nature, 386, 377-9. 6. Report on the Basic Energy Sciences Workshop on Hydrogen Production, Storage and Use. Basic Research Needs for the Hydrogen Economy, p. 31. Office of Science, US Department of Energy, 2003. 7. Report on the Basic Energy Sciences Workshop on Hydrogen Production, Storage and Use. Basic Research Needs for the Hydrogen Economy, p. 32. Office of Science, US Department of Energy, 2003. 8. Farrell, A.E., Keith, D.W., and Corbett, J.J. (2003). A strategy for introducing hydrogen into transportation. Enetgy Policy, 31, 1357-67.

References

9. Dillon, A.C. and Heben, M.J. (2001). Hydrogen storage using carbon adsorbents: past, present and future. Appl. Phys. A, 72, 133-42. 10. de la Casa-Lillo, M.A., Lamari-Darkrim, F., Cazorla-Linares, D., and LinaresSolano, A. (2002). Hydrogen storage in activated carbons and activated carbon fibers.]. Phys. Chem. B, 106, 10930-4. 11. Report on the Basic Energy Sciences Workshop on Hydrogen Production, Storage and Use. Basic Research Needs for the Hydrogen Economy, pp. 31-46. Office of Science, US Department of Energy, 2003. 12. Dresselhaus, M.S., Williams, K.A., and Eklund, P.C. (1999). Hydrogen adsorption on carbon materials. MRS Bull., 24(11), 45-50. 13. Bunger, U. and Zittel, W. (2001). Hydrogen storage in carbon nanostructures still a long road from science to commerce? Appl. Phys. A, 72, 147-51. 14. Ahn, C.C., Ye, Y., Ratnakumar, B.V., et al. (1998). Hydrogen desorption and adsorption measurements on graphite nanofibers. Appl. Phys. Lett., 73, 3378-80. 15. Tibbetts, G.G., Meisner, G.P., and Olk, C.H. (2001). Hydrogen storage capacity of carbon nanotubes, filaments and vapor-grown fibers. Carbon, 39, 2291-301. 16. Hirscher, M., Becher, M., Haluska, M., et al. (2002). Hydrogen storage in carbon nanostructures.]. Alloys Compd., 330-2, 654-8. 17. Wang, Q. andJohnson,J.K. (1999). Molecular simulation of hydrogen adsorption in single-walled carbon nanotubes and idealized carbon slit pores.]. Chem. Phys., 110, 577-86. 18. Simonyan, V.V. and Johnson, J.K. (2002). Hydrogen storage in carbon nanotubes and graphitic nanofibers.]. Alloys Compd., 330-2, 659-65. 19. Report on the Basic Energy Sciences Workshop on Hydrogen Production, Storage and Use. Basic Research Needs for the Hydrogen Economy, p. 44. Office of Science, US Department of Energy, 2003. 20. Dagani, R. (2002). Tempest in a tiny tube. Chem. Eng. News, January 14, pp. 25-8. 21. Ye, Y., Ahn, C.C., Whitham, C., et al. (1999). Hydrogen adsorption and cohesive energy of single-walled carbon nanotubes. Appl. Phys. Lett., 74, 2307-9. 22. Dillon, A.C., Gilbert, K.E.H., Alleman, J.L., et al. (2001). Carbon nanotube materials for hydrogen storage. Proceedings of the 2001 DOE Hydrogen Program Review, NREL/CP-570-30535. 23. Liu, C., Fan, Y.Y., and Liu, M. (1999). Hydrogen storage in single-walled carbon nanotubes at room temperature. Science, 286, 1127-9. 24. Hirscher, M., Becher, M., Haluska, M., et al. (2001). Hydrogen storage in sonicated carbon materials. Appl. Phys. A,72, 129-32. 25. Hirscher, M., Becher, M., Haluska, M., et al. (2003). Are carbon nanostructures an efficient hydrogen storage medium?]. Alloys Compd., 356-7, 433-7. 26. Pradhan, B.K., Harutyunyan, A.R., and Stojkovic, D. (2002). Large cryogenic storage of hydrogen in carbon nanotubes at low pressure.]. Mater. Res., 17, 2209-16. 27. Smith, M.R., Bittner, E.W., Shi, W., et al. (2003). Chemical activation of singlewalled carbon nanotubes for hydrogen adsorption.]. Phys. Chem. B, 107,3752-60. 28. Gundiah, G., Govindaraj, A., Rajalakshmi, N., et al. (2003). Hydrogen storage in carbon nanotubes and related materials.]. Mater. Chem., 13, 209-13. 29. Cao, A., Zhu, H., Zhang, X., et al. (2001). Hydrogen storage of dense-aligned carbon nanotubes. Chem. Phys. Lett., 342, 510-14. 30. Zhu, H., Cao, A., Li, X., et al. (2001). Hydrogen adsorption in bundles of wellaligned carbon nanotubes at room temperature. Appl. Surf. Sci., 178, 50-5.

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

31. Muris, M., Dufau, N., and Bienfait, M. (2000). Methane and krypton adsorption on single-walled carbon nanotubes. Langmuir, 16, 7019-22. 32. Talapatra, S., Zambano, A.J., Weber, S.E., and Migone, A.D. (2000). Gases do not adsorb on the interstitial channels of closed-ended single-walled carbon nanotube bundles. Phys. Rev. Lett., 85, 138-41. 33. Talapatra, S. and Migone, A.D. (2002). Adsorption of methane on bundles of closed-ended single-wall carbon nanotubes.Phys. Rev. B, 65, 045416 1-6. 34. Bienfait, M., Asmussen, B., Johnson, M., and Zeppenfeld, P. (2000). Methane mobility in carbon nanotubes. Surf. Sci., 460, 243-8. 35. Muris, M., Bienfait, M., Zeppenfeld, P., et ai. (2002). Diffraction study of CD 4 and D 2 adsorbed on carbon nanotubes. Appl. Phys. A, 74, S1293-5. 36. Johnson, M.R., Rols, S., Wass, P., et ai. (2003). Neutron diffraction and numerical modeling investigation of methane adsorption on bundles of carbon nanotubes. Chern. Phys., 293, 217-30. 37. Hertel, T., Kriebel, J., Moos, G., and Fasel, R. (2001). Adsorption and desorption of weakly bonded adsorbates from single-wall carbon nanotube bundles. In Nanonetwork Materials, Fullerenes, Nanotubes, and Related Systems (S. Saito, T. Ando, Y. Iwasa, et aI., eds). AlP Conference Proceedings, 590, 181-4. 38. Kleinhammes, A., Mao, S.-H., Yang, X.-J., et ai. (2003). Gas adsorption in single-walled carbon nanotubes studied by NMR. Phys. Rev. B, 68, 075418 1-6. 39. Weber, S.E., Talapatra, S., Journet, C., et ai. (2000). Determination of the binding energy of methane on single-walled carbon nanotube bundles. Phys. Rev. B, 61, 13150-4. 40. Vidali, G., Ihm, G., Kum, H.-Y., and Cole, M.W. (1991). Potential of physical adsorption. Surf. Sci. Rep., 12, 135-81. 41. Stan, G., Bojan, M.J., Curtarolo, S., et ai. (2000). Uptake of gases in bundles of carbon nanotubes. Phys. Rev. B, 62 2173-80. 42. Muris, M., Dupont-Pavlovsky, N., Bienfait, M., and Zeppenfeld, P. (2001). Where are the molecules adsorbed on single-walled nanotubes? Surf. Sci., 492, 67-74. 43. Shi, W. and Johnson, J.K. (2003). Gas adsorption on heterogeneous single-walled carbon nanotube bundles. Phys. Rev. Lett., 91, 015504 1-4. 44. Rols, S., Almairac, R., Henrard, L., et al. (1999). Diffraction by finite-size crystalline bundles of single wall nanotubes.]. Bur. Phys. B, 10, 263-70. 45. Wilson, T., Tyburski, A., DePies, M.R., et al. (2002). Adsorption of H 2 and D 2 on carbon nanotube bundles.]. Low Temp. Phys., 126,403-8. 46. Talapatra, S. and Migone, A.D. (2001). Existence of novel quasi-one-dimensional phases ofatoms adsorbed on the exterior surface of close-ended single wall nanotube bundles. Phys. Rev. Lett., 87, 206106 1-4. 47. Talapatra, S., Rawat, D.S., and Migone, A.D. (2002). Possible existence of a higher coverage quasi-one-dimensional phase of argon adsorbed on bundles of single-walled carbon, nanotubes.]. Nanosci. Nanotechnol., 2, 467-70. 48. Bienfait, M., Zeppenfeld, P., Dupont-Pavlovsky, N., et ai. (2003). Adsorption of argon on carbon nanotube bundles and its influence on the bundle lattice parameter. Phys. Rev. Lett., 91, 035503 1-4. 49. Gatica, S.M., Bojan, M.J., Stan, G., and Cole, M.W. (2001). Quasi-one and two-dimensional transitions of gases adsorbed on nanotube bundles.]. Chern. Phys., 114,3765-9. 50. Calbi, M.M. and Cole, M.W. (2002). Dimensional crossover and quantum effects of gases adsorbed on nanotube bundles. Phys. Rev. B, 66, 115413 1-12.

References

51. Fujiwara, A., Ishii, K., Suematsu, H., et al. (2001). Gas adsorption in the inside and outside of single-walled carbon nanotubes. Chem. Phys. Lett., 336, 205-11. 52. Teizer, W., Hallock, R.B., Dujardin, E., and Ebbesen, T.W. (1999). 4He desorption from single wall carbon nanotube bundles: a one-dimensional adsorbate. Phys. Rev. Lett., 82, 5305-8. 53. Teizer, W., Hallock, R.B., Dujardin, E., and Ebbesen, T.W. (2000). Errata: 4He desorption from single wall carbon nanotube bundles: a one-dimensional adsorbate. Phys. Rev. Lett., 84, 1844-5. 54. Kahng, Y.H., Hallock, R.B., Dujardin, E., and Ebbesen, T.W. (2002). 4He binding energies on single-wall carbon nanotube bundles.]. Low Temp. Phys., 126, 223-8. 55. Kahng, Y.H., Hallock, R.B., and Dujardin, E. (2003). Competitive adsorption of He-4 and H-2 on single-wall carbon nanotube bundles. Physica B, 329, 280-1. 56. Wilson, T. and Vilches, a.E. (2003). Adsorption of 4He on carbon nanotube bundles. Physica B, 329, 278-9. 57. Lasjaunias, ].C., Biljakovic, K., Benes, Z., et al. (2002). Low-temperature specific heat of single-wall carbon nanotubes. Phys. Rev. B, 65, 113409 1-4. 58. Lasjaunias, ].C., Biljakovic, K., Sauvajol, ].L., and Monceau, P. (2003). Evidence of lD behavior of 4He confined within carbon nanotube bundles.Phys. Rev. Lett., 91, 025901 1-4. 59. Hone, ]., Batlogg, B., Benes, Z., et al. (2002). Quantized phonon spectrum of single-wall carbon nanotubes. Science, 289(5485), 1730-3. 60. Wilson, T. and Vilches, a.E. (2003). Helium adsorbed on carbon nanotube bundles: one-dimensional and/or two-dimensional solids? Low Temp. Phys., 29, 732-5. 61. Brown, C.M., Yildrim, T., Neumann, D.A., et al. (2000). Quantum rotation of hydrogen in single-wall carbon nanotubes. Chem. Phys. Lett., 329, 311-16. 62. Narehood, D.G., Pearce,].V., Eklund, P.C., et al. (2003). Diffusion ofH 2 adsorbed on single-walled carbon nanotubes. Phys. Rev. B, 67,205409 1-5. 63. Schimmel H.G., Kearley, G.J., Nijkamp, M.G., et al. (2003). Hydrogen adsorption in carbon nanostructures: comparison of nanotubes, fibers, and coals. Chem. Bur. J., 9, 4764-70. 64. Zambano, A.]., Talapatra, S., and Migone, A.D. (2001). Binding energy and monolayer capacity of Xe on single-wall carbon nanotube bundles. Phys. Rev. B, 64,075415 1-6. 65. Talapatra, S., Krungleviciute, V., and Migone, A.D. (2002). Higher coverage gas adsorption on the surface of carbon nanotubes: evidence for a possible new phase in the second layer. Phys. Rev. Lett., 89, 246106 1-4. 66. Calbi, M.M., Gatica, S.M., Bojan, M.J., and Cole, M.W. (2001). Phases of neon, xenon and methane adsorbed on nanotube bundles.]. Chem. Phys., 115,9975-81. 67. Kuznetsova, A., Mawhinney, D.B., Naumenko, V., et al. (2000). Enhancement of adsorption inside of single-walled nanotubes: opening the entry ports. Chem. Phys. Lett., 321, 292-6. 68. Kuznetsova, A., Yates,].T.,]r, Liu,]., and Smalley, R.E. (2000). Physical adsorption of xenon in open single walled carbon nanotubes: observation of a quasi-onedimensional confined Xe phase.]. Chem. Phys., 112,9590-8. 69. Ulbricht, H., Kriebel,]., Moos, G., and Hertel, T. (2002). Desorption kinetics and interaction of Xe with single-wall carbon nanotube bundles. Chem. Phys. Lett., 363, 252-60.

430

Chapter 16 Adsorption on Carbon Nanotubes: Experimental Results

70. Krungleviciute, V., Heroux, L., Talapatra, S., and Migone, A.D. (2004). Gas adsorption on HiPco nanotubes: surface area determinations, and neon second layer data. Nano Lett., 4, 1133-7. 71. Byl, 0., Kondratyuk, P., Forth, S.T., et al. (2003). Adsorption of CF 4 on the internal and external surfaces of opened single-walled carbon nanotubes: a vibrational spectroscopy study. JACS, 125, 5889-96. 72. Yoo, D.-H., Rue, G.-H., Hwang, Y.-H., and Kim, H.-K. (2002). Study of nitrogen adsorbed on single-walled carbon nanotube bundles. j. Phys. Chern. B, 106,3371-4. 73. Yoo, D.-H., Rue, G.-H Chan, M.H.W., Hwang, Y.-H., and Kim, H.-K. (2003). Study of nitrogen adsorbed on open-ended nanotube bundles. j. Phys. Chern. B, 107, 1540-2. 74. Yang, C.M., Kaneko, K., Yudasaka, M., and Iijima, S. (2002). Effect ofpurification on pore structure of HiPco single-walled carbon nanotube aggregates.Nano Lett., 2,385-8. 75. Du, W.-F., Wilson, L., Ripmeester, J., et al. (2002). Investigation of the pore structure of as-prepared and purified HiPco single-walled carbon nanotubes by N 2 / Ar adsorption-implication for H 2 storage. Nano Lett., 2, 343-6. 76. Cinke, M., Li,J., Chen, B., et al. (2002). Pore structure of raw and purified HiPco single-walled carbon nanotubes. Chern. Phys. Lett., 365, 69-74.