Unusual Low-Vibrational C O Mode of COOH Can Distinguish

Nov 19, 2012 - Ajit K. Roy,. ‡ and Holger F. Bettinger. §. †. Department of Chemistry and Biochemistry, Utah State University, Logan, Utah 84322-...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/JPCC

Unusual Low-Vibrational CO Mode of COOH Can Distinguish between Carboxylated Zigzag and Armchair Single-Wall Carbon Nanotubes Tapas Kar,*,† Steve Scheiner,† Ajit K. Roy,‡ and Holger F. Bettinger§ †

Department of Chemistry and Biochemistry, Utah State University, Logan, Utah 84322-0300, United States Materials and Manufacturing Directorate, Air Force Research Laboratory, Wright-Patterson Air Force Base, Ohio 45433, United States § Institute for Organic Chemistry, University Tuebingen, Auf der Morgenstelle 18, 72076 Tuebingen, Germany ‡

S Supporting Information *

ABSTRACT: Vibrational frequency analysis using density functional theory resolves some ambiguities in interpretation of IR spectra of carboxylated single-wall carbon nanotubes (o-SWNTs). Armchair (n,n) and zigzag (m,0) tubes, with diameter varying between 0.5 and 1.4 nm, were populated with different numbers of COOH groups at one terminus. While armchair-COOH tubes exhibit CO stretching frequencies in the standard range of 1720−1760 cm−1, zigzag-COOH tubes display a new CO band around 1650 cm−1 with much higher intensity, in addition to the higher-frequency CO mode. The COOH groups exhibiting this low-frequency (lf-COOH) are displaced well off the lines of the tube walls, to the outside. This new band is in the same range as one observed experimentally, which has on occasion been attributed to quinone formation. The CO bond length in the relevant COOH groups is longer by about 0.01 Å than regular CO bonds. Such low-frequency CO band is unusual and not common for any stand-alone COOH group and seems characteristic of carboxylated zigzag tubes irrespective of their diameter and the number of acid groups at the terminus. The lower frequency is not a result of H-bonding, nor can it be reproduced by small models such as benzoic acid or extended carbon networks as in graphene-COOH. Its origin is attributed instead to the curvature of the zigzag tubes in addition to their structural arrangement.



INTRODUCTION

overcome by virtue of subtle differences, if any, among oSWNTs. It is thus imperative to have a thorough knowledge of the structure and intrinsic properties of individual o-SWNTs, as those −COOH groups are used as the anchor for further functionalization. Of several oxygen-containing functional groups, −COOH and −OH are the most widely observed,10 located primarily at the tips and also at the sidewall of SWNTs during various acid treatments. In general, routine characterizations of functional groups are derived from IR spectra, without critical analyses of data. However, there are landmark studies by Zhang et al.11 and

Carboxylated single-wall carbon nanotubes (SWNT-COOH, oSWNT), created during purification of raw SWNTs by oxidative treatments, are the gateway to the functionalized carbon nanotube arena. These functionalized nanotubes are a focus of nanotube research due to their wide range of applications, such as sensors, solar cells, drug delivery systems, and various electronic and optoelectronic devices.1−9 Despite significant progress in chemical modification of purified carbon nanotubes, inhomogeneity in raw as well as in oxidized SWNTs represents a bottleneck in progress toward application-oriented technological developments of individual nanotubes. Better understanding of structures and properties of different oSWNTs might lead to selective and controlled functionalization. In addition, a major hurdle in separating SWNTs might be © 2012 American Chemical Society

Received: September 30, 2012 Revised: November 15, 2012 Published: November 19, 2012 26072

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C

Article

Kim et al.,12 where the data were critically analyzed and difficulties pointed out in interpreting certain CO stretching bands, ν(CO), due to various intermingling factors. The latter include carboxylated carbonaceous fragments and broken carbon structures, in addition to different diameters and chirality of nanotubes, interaction among functional groups via H-bonding, coupling with other vibrational modes, and so on. Another key study by Andersson and Grennberg13 on the effect of purification process (using HNO3) clearly showed differences in COOH content in o-SWNTs, which depend on acid concentration, treatment time, reaction temperature, and even the amount of sample (10 mg vs 100 mg) and stirring efficiency. As the number of COOH groups differs depending on treatment conditions, their arrangements and hence characterization become complicated in a mixture of various o-SWNTs. The typical CO stretching mode of the −COOH group ranges between 1710 and 1760 cm −1 , and reported experimental IR spectra of the oxidized SWNTs exhibit C O bands in that range.3,12,14−21 Our recent density functional theory (DFT) study22,23 computed νCO at 1702−1750 cm−1 for armchair (5,5)-(COOH)x (x = 1 and 4) tubes, which is in the range of experimental data. However, the zigzag (10,0)(COOH)x (x = 1 and 4) tubes exhibit two distinct CO modes: one in the regular νCO range and the other around 1650 cm−1. The latter peak is unexpected for the CO mode of a stand-alone carboxylic group. This finding raises the question as to whether the vibrational mode close to 1650 cm−1 is characteristic of zigzag tubes. If the DFT prediction is correct, then there should be two IR peaks in some o-SWNT samples. And indeed many experimental spectra11,12,14,20,24−27 do in fact show peaks in the normal CO region and another around 1650 cm−1. In general, quinones (two CO double bonds are coupled to a benzene ring either at ortho- or at para-position) and their derivatives exhibit CO peaks in the 1630−1690 cm−1 region.28 Based on these standard IR data, most of the experimental studies assigned the ∼1650 cm−1 peak of o-SWNT as quinone CO. However, the earlier calculations offer a second interpretation, which is explored in more detail here. Adding to the difficulty in interpretation, the 1650 cm−1 peak is not common to all kinds of SWNT samples and purification processes. For example, KMnO4 -treated CVD samples exhibited a peak at 1650 cm−1, but not when the same sample was treated with HNO3 or H2SO4/HNO3 mixture.11 Laser ablation sample, treated with H2SO4/HNO3 mixture, 24 exhibited such peak. Dry oxidation of arc discharge sample did not show12 this peak, but it appeared when dry oxidized sample was further treated with H2O2. Thus, the origin of ∼1650 cm−1 is puzzling, and doubts have arisen in interpreting CO modes which we hope to resolve by thorough computational investigation. In the present theoretical study, we consider molecular models of SWNT-COOH (o-SWNTs) with a wide range of different diameters relevant to experimental studies. For example, diameters of pristine nanotubes obtained by arc discharge and laser ablation methods are in the range of 1.2− 1.6 nm, and smaller diameter tubes (0.8−1.2 nm) are produced in chemical vapor deposition (CVD) method of synthesis.29 We thus focus on armchair and zigzag SWNTs in this diameter range. Since tips of o-SWNTs contain primarily COOH groups, all pristine nanotube models considered were populated with COOH groups at one end. The number of such groups is

varied, so as to understand how this parameter might influence the spectra. The main objective of the present investigation is to disentangle complicated experimental IR results and to provide some insights into the properties of oxidized tubes.



MODELS AND METHOD OF CALCULATIONS Investigations on large system like nanotubes including calculation of vibrational frequencies are computationally highly demanding and so require a protocol that is both accurate and feasible. Kar and co-workers have22,30−32 shown the advantages of using the same level dif ferent basis (SLDB) prescription over other procedures, such as the widely used ONIOM33,34 method for studying SWNTs and their chemical modifications. In fact, several other studies also indicated ONIOM is not a reliable method for studying functionalized nanotubes due to disruption of the π-network.35−38 In the SLDB approach, atoms in defined active sites are provided with large sets of basis functions, while smaller sets are applied to the remaining atoms. Previous studies22,31 demonstrated that the use of a larger basis set for a small number of atoms at active sites is sufficient to reproduce structure, energetics, and vibrational spectra for sidewall and end-functionalized nanotubes. This DFT-SLDB technique was used throughout in the present study. Active atoms in the functional COOH groups were treated using the 6-31G* (5d functions) basis set, adding diffuse sp functions to this polarized split-valence double-ζ quality basis for proper description of electronegative O atom, denoted 6-31G*(O+). The remaining carbon atoms were treated with the smaller split-valence double-ζ quality 3-21G basis. The 6-31G* basis was used for terminal hydrogens at the functionalized end and 3-21G for hydrogens at the other end of SWNTs. This combination of basis functions has been assessed in previous studies and found reliable and adequate. The B3LYP variant of density functional theory (DFT)39,40 was employed in order to include correlation effects. The accuracy of normal mode calculations using the B3LYP functional is sufficiently high and includes an optimal cost-to-benefit ratio.41−44 Because computed harmonic vibrational frequencies are typically slightly overestimated (even for more accurate methods, such as MP2, CCSD(T), etc., as well as for larger basis sets), a scale factor is commonly used to better correspond to experimental spectra. For example, a value of 0.965 is recommended for B3LYP/3-21G and 0.960 for B3LYP/6-31G* methods.45 Since we are primarily interested in the spectra of attached functional groups (COOH in particular), application of a scale factor of 0.96 was deemed a reasonable choice. The selected normal modes (SNM) vibrational analyses were tested for smaller o-SWNTs46 and found to reproduce the CO mode of COOH groups quite accurately. In the SNM procedure47 only selected Cartesian nuclear coordinates, instead of all, are considered in the massweighted Cartesian Hessian matrix of second derivatives of the total electronic energy. Diagonalization of this matrix leads to the eigenvectors and eigenvalues that correspond to the selected motions. This approach reduces computation time significantly as one need not compute the entire Hessian. The specific models of o-SWNTs considered in the present study are armchair (n,n), n = 4, 5, 6, 8, 10, and zigzag (m,0), m = 8, 9, 10, 12, 15, 16, which cover most of the SWNTs reported in the literature to date. For example, the diameters of HiPco samples are within the 0.8−1.2 nm range, and most of these models occupy this range. The larger models, (10,10), (16,0), 26073

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C

Article

Figure 1. Optimized structures of (n,n)-(COOH)2n, o-SWNTs and diameter (in nm) of corresponding pristine (n,n) SWNTs.

Figure 2. Expanded view of normalized IR spectra (scaled by 0.960) of fully end-functionalized armchair (4,4) and zigzag (8,0) SWNTs. Lorentzian broadening with fwhm of 20 cm−1 was applied. All eight νCO values (in cm−1) and corresponding intensity (in km/mol) in parentheses are given. Bold numbers are the most intense CO vibrational modes in the group.

theoretical studies on those oxidized tubes are in progress, and results will be published separately.) Armchair (n,n) tubes contain 4n carbon atoms compared to 2m for zigzag (m,0) in each layer. Our SWNT models are

and (18,0), are close to the diameter of arc discharge and laser ablation samples. (It may be noted that besides zigzag and armchair variants, a significant portion of SWNTs might be chiral (n,m) tubes, where n and m are unequal. Follow-up 26074

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C

Article

Figure 3. Normalized IR spectra (scaled by 0.960) of fully end-functionalized armchair (n,n) (n = 4, 5, 6, 8, and 10) carbon nanotubes. None of these o-(n,n) tubes exhibited any imaginary frequency. Lorentzian broadening with fwhm of 20 cm−1 was applied. SNM stands for selected normal modes. νCO value (in cm−1) and intensity (in km/mol) of the most intense peak are shown for each spectrum.

saturated with hydrogens to occupy otherwise dangling bonds. There is some thought that due to steric effect, such high numbers of acid groups at their tip might distort the shapes of nanotubes. These fully carboxylated tubes may thus be considered as a gauge of such induced distortions, while also maximizing the possibility of H-bonding between one COOH group and its neighbors. Geometry optimizations were carried out without any constraints using the DFT(B3LYP)-SLDB method, where 4n (2m) carbons of armchair (zigzag) and all attached COOH groups are represented by 6-31G*(O+), while 3-21G was used for the remainder of atoms. o-Armchair SWNTs. End views of the optimized structures of o-(n,n) carboxylated armchair tubes, along with the diameter (in nm) of the corresponding pristine tubes, are shown in Figure 1, supplemented by side views in Figure S1. The COOH groups occupy sites around the outside of the perimeter of the tube, bent away from the central axis. For smaller (4,4) and (5,5) tubes, the high concentration of COOH groups does not appreciably distort the tube except for the first functionalized layer, where carbons at the tip are bent slightly outward. An increase in diameter causes some acid groups, along with their linked carbons, to bend in, affording the tube’s acidified end an oval shape. There is no definite pattern of number of acid

composed of four layers of carbons for both kinds of tube. The smallest (4,4) and (8,0) tubes thus contain 64 C atoms, whereas there are 160 and 144 atoms in the largest (10,10) and (18,0) tubes, respectively. Thus the length of the tube in all models is just less than 1.0 nm. As the present study is focused on the intrinsic properties of COOH groups present at the tips of SWNTs, their vibrational modes are not likely to be sensitive to small differences in length. In fact, in the case of the single acid group of SWNT-COOH, the length of the nanotube had practically no effect on CO vibrational modes.22 All calculations were performed using the Gaussian-0948 code. Theoretical vibrational modes were analyzed using Molden,49 and IR plots were generated by Gabedit50 programs. Pristine nanotube models were obtained using TubeVBS software,51 and in the o-SWNTs models COOH groups and terminal hydrogen atoms were placed using Chemcraft52 software, which was also used to generate o-SWNTs figures for geometry analyses.



RESULTS AND DISCUSSION The first set of calculations considered fully populated oSWNTs, viz., 2n and m COOH groups at one end of armchair (n,n) and zigzag (m,0) tubes, respectively. The other end was 26075

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C

Article

Table 1. Vibrational Frequenciesa (ν, cm−1) and Intensity (I, km/mol) of CO Modes of Fully Carboxylated SWNTs and Diameter (d, nm) of Pristine Tubes ν(CO)

I

o-(m,0) (d, nm)

ν(CO)

I

o-(m,0) (d, nm)

ν(CO)

I

(4, 4) (0.54)

1728

712

(8, 0) (0.63)

1752

785

(10, 0) (0.78)

(6, 6) (0.81)

1747

888

(16, 0) (1.25)

1109 754 1065 931 1009 791

(9, 0) (0.70)

(5, 5) (0.68)

1650 1740 1671 1723 1656 1738

(8, 8) (1.08)

1747

1100

1667 1746 1665 1753 1668 1750 1677 1755

1351 954 949 882 1382 1269 1641 1636

(10, 10) (1.36)

1749

1240

o-(n,n) (d, nm)

a

(12, 0) (0.94) (15, 0) (1.17) (18, 0) (1.41)

All ν(CO) values are from selected normal mode (SNM) analyses.

Figure 4. Optimized structures of (m,0)-(COOH)m o-SWNTs and diameter (in nm) of corresponding pristine (m,0) SWNTs. Carboxylic groups bent-out and parallel to the tube wall are shown by arrows. These groups exhibit CO peaks below 1700 cm−1.

groups for (8,0), five within the 1723−1752 cm−1 range and three between 1651 and 1670 cm−1. For purposes of nomenclature, later COOH groups are denoted as lowfrequency COOH (lf-COOH) and the others as regular (r) (details in following sections). After introducing appropriate broadening for each line, one sees a single band for the armchair tube and two separate bands for zigzag. The most intense peaks of the armchair o-(4,4) and o-(5,5) tubes (Figure 3) occur at 1728 and 1752 cm−1, respectively. These CO vibrational modes are pure (some coupling within the COOH groups) and no coupling with CC modes or any other modes. Such a scenario is favorable for application of the very efficient and affordable selected normal mode (SNM) approach, where only COOH groups and attached carbons of tube were selected for vibrational analyses. As may be seen in Figure 3, the SNM frequencies and intensities are very similar to the values obtained from a full molecule treatment in o-(4,4) and o-(5,5) tubes. Variations of the most intense CO stretching band of o(n,n) tubes with increasing diameter are summarized in Table 1. The νCO value is lowest for o-(4,4) and blue-shifted by 24 cm−1 in o-(5,5), after an increase in diameter from 0.54 to 0.68

groups bending out or bending in with increasing diameter. For example, (6,6)-(COOH)12 exhibits two pairs of C−COOH bent inward while the other four pairs are bent out, and this arrangement is in a symmetric patterneach inward pair is flanked by outward pairs. Moving to (8,8)-(COOH)16, the number of inward bent C−COOH groups did not change from its predecessor. The magnitude of the displacement of outward (inward) acid groups is more pronounced when flanked by inward (outward) pair. The tips and wall are significantly distorted for the largest (10,10)-(COOH)20. This indicates that high numbers of acid groups may be possible for tubes below ∼1.2 nm diameter (representing HiPco samples) and unlikely for larger tubes (arc discharge or laser ablation samples). Since the present study focuses on the CO stretching modes of COOH, IR spectra are reported in the 2000−1600 cm−1 range. The νCO frequencies of the eight COOH groups of (4,4)-(COOH)8 differ slightly in magnitude, and intensity varies from mode to mode as listed in Figure 2. Each vibrational mode of the CO stretching is fairly pure without extensive coupling to other CO vibrational modes. It is important to note that the eight lines for (4,4) all lie close to one another, within a span of 42 cm−1. In contrast, there are two separate 26076

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C

Article

Figure 5. Normalized IR spectra (scaled by 0.960) of fully end-functionalized zigzag (m,0) (m = 8, 9, 10, 12, 15, 16, and 18) carbon nanotubes. None of these o-(m,0) tubes exhibited any imaginary frequency. Lorentzian broadening with fwhm of 20 cm−1 was applied. SNM stands for selected normal modes. νCO value (in cm−1) and intensity (in km/mol) of the most intense peak are given for each spectrum along with number of lfCOOH units.

nm. Further increase in diameter from 0.68 to 1.36 nm has practically no further effect on CO bands. This near uniformity indicates νCO of o-(n,n) ought to be almost independent of the synthesis method that might yield different tube diameters, and the characteristic CO band should appear between 1740 and 1750 cm−1. Theoretical predictions of CO modes are within the range of reported experimental

data. In contrast to a stable frequency, the intensity of the C O mode rises almost linearly with the diameter which might be a useful feature for identifying o-armchairs based on diameter. o-Zigzag SWNTs. Fully optimized geometries of the most stable structures of zigzag (m,0)-(COOH)m nanotubes are summarized in Figures 4 and S2 (side views) and their IR spectra in Figure 5. Other minima on the potential energy 26077

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C

Article

Figure 6. DFT-optimized geometries and bond lengths (in Å) of fully end-functionalized o-(4,4) and o-(8,0) carbon nanotubes. Arrows marked COOH groups exhibiting frequencies around the 1650 cm range, the rest of the CO show peak around 1740 cm−1.

modes. Analyses of each CO mode reveal coupling of carbonyl stretching vibration among each set of carboxylic groups, but no coupling between lf-COOH and r-COOH groups. One of the two CO peaks is in the normal range of CO modes, i.e., around 1740 cm−1. These peaks correspond to the CO mode of the regular COOH groups. The lf-COOH groups exhibit a peak around 1650 cm−1 irrespective of tube diameter. As this frequency range is typical of quinones or their derivatives, most of the previous experimental studies assigned the peak accordingly. However, the results described here indicate that such a peak may be due instead to these lf-COOH groups. As another issue with implications for making this distinction, the intensity of each such low-frequency band is greater than the regular CO modes of COOH. Table 1 illustrates how νCO varies upon changing the diameter of zigzag (m,0) tubes. Results are listed separately for metallic (m/3 is an integer) and semiconducting (otherwise) tubes. Similar to metallic armchair tubes, the first CO band of purified metallic (m,0) (m = 9, 12, 15, and 18) tubes is close to 1750 cm−1, and its intensity tends to increase with diameter. For semiconducting (m,0) tubes, the average ν(CO) is 1735 cm−1, lower by about 15 cm−1 than metallic tubes. The outstanding distinction between carboxylated armchair and zigzag tubes is the band around 1650 cm−1 for the latter, which is absent in the former. The average frequency for the metallic zigzag tubes is 1669 cm−1, and about 15 cm−1 lower for semiconducting zigzag tubes. Also the intensities of the most intense peaks are, in general, lower in semiconducting zigzag tubes than their metallic sisters. Geometric Differences between o-(n,n) and o-(m,0) Tubes. Key geometric parameters of representative tubes are depicted in Figure 6. For o-(4,4) tube all CO bond lengths are between 1.208 and 1.217 Å, with linking C−C distances (about 1.5 Å) close to the single carbon−carbon bond length. Similar bond distances are found in o-(8,0) for regular carboxylic groups. In the cases of lf-COOH units, the CO distances are elongated (1.223−1.228 Å), and the linking C−C bond shows shortening toward double-bond character (1.45 Å). Bond length patterns of this type are found in all zigzag tubes. The CO elongation within lf-COOH is closely associated with the red shift of the band. A common cause of CO red shifting is H-bonding with a nearby group, in this case O−H··O interactions among COOH

surface and their IR spectra, not the most stable, are displayed in Figures S3 and S4, respectively. The top view in Figure 4 clearly shows two distinct arrangements of COOH groups irrespective of the diameter of tubes. One set of these groups (marked by arrows in figures) are displaced well off the lines of the tube walls, to the outside. Even the C atoms of the wall to which these groups are attached are displaced a bit in that direction. The planes in which these COOH groups lie make an angle of roughly 45° to the central axis direction of the tube. The other set of groups in o-(m,0) are situated similarly to those in the armchair tubes discussed above. Since the former group exhibits a CO peak well below regular values (Figure 2), those are designated as low-frequency COOH group (lfCOOH). The side views in Figure S2 clearly show that, unlike the armchairs, the zigzag tube shape is unperturbed by the functionalization, even for the largest o-(18,0) tube. It is possible, then, that zigzag arrangements might be better able to accommodate a higher concentration of functional groups. For smaller tubes, such as (8,0) and (9,0), various initial geometric arrangements were considered as starting points for optimizations which led to those structures shown in Figure 4. Thus the structures illustrated for these two tubes may be considered as global minima, and the number of lf-COOH groups is three for both tubes. (Similar structural arrangement of COOH groups in (9,0)-(COOH)9 was recently reported by Chelmecka et al.53.) Further increase in diameter to (10,0) resulted in two isomersthe most stable one also contains three lf-COOH groups. The structure with two lf-COOH groups (see Figure S3) is less stable by about 4.2 kcal/mol, a circumstance repeated in (12,0)-(COOH)12. It may be noted that the lf-COOH groups are not adjacent to one another, probably due to steric repulsion. In fact, these groups tend to space out so as to be as far apart from one another as possible, so that they are separated by two r-COOH groups in (9,0) for example, and by three such groups in (12,0). Such trend of stability and arrangements of COOH groups is also followed for larger tubes (Figure S3). As the arrangements of COOH groups of o-(m,0) tubes differ from armchairs, so do their IR spectra. Clearly there are two peaks for the CO modes for all zigzag tubes (Figures 2, 5, and S4). Results shown for o-(8,0), o-(9,0), and o-(10,0) in Figure 5 indicate that the SNM method is highly reliable in predicting CO frequencies and their intensities, due to noninterference of any other modes with CO stretching 26078

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C

Article

units. The calculated structures, however, do not find evidence of any such H-bonds even for larger tubes with highest concentration of acid groups. One set of criteria for the presence of such a bond would be a r(H···O) distance less than 2.0 Å and a θ(O−H···O) angle in the 150−180° range.54 The shortest distance between a hydroxyl H and the O atom of an adjacent carboxyl is greater than 2.2 Å; the θ(OH···O angles) are, in general, less than 120°, precluding such a H-bond. Natural bond orbitals (NBO) analysis55,56 of the o-(4,4) and o(8,0) tubes verified the absence of H-bonds by virtue of vanishingly small Olp → σ*(OH) second-order perturbation energies, E(2). Having ruled out the source of the low CO frequency from either intragroup H-bonding or coupling with other vibrational modes, it was thought that perhaps the bending of the COOH group away from the line of the nanotube might cause a change in the frequency. In order to test this notion, the COOH unit of benzoic acid was bent away from the phenyl plane in the same amount as observed in o-(m,0), 37° from planarity. Holding this angle fixed, the remainder of the geometry was optimized. This distortion resulted in only a 8 cm−1 lowering of the CO frequency, from 1716 to 1709 cm−1, so clearly rules out the bending as the prime source of red shift down to the 1650 cm−1 range observed in the nanotubes. One is left with the idea that the nature of the larger carbon network in a zigzag framework is likely responsible for the unusual CO peak. One can test this concept within the framework of a fully planar graphene sheet. As illustrated in Figure 7, a COOH group was added to both the armchair and

times and condition (such as temperature) intervals. In this process the concentration of acid groups incorporated into the nanotubes changes (either increases or decreases). To understand the effect of the concentration of COOH groups on the IR spectra, nanotubes with lower number of carboxylic groups were examined. Another reason is also to make sure the uncommon CO band of zigzag tube is not an artifact of higher concentration of acid group. (n,n)-(COOH)x. COOH groups were removed incrementally from (4,4)-(COOH)8, and the geometry reoptimized, followed by computation of IR spectra using the SNM formalism. Structures and IR spectra are summarized in Figure 8, emphasizing the r(CO) and r(C−COOH) bond lengths. These distances remain fairly stable as the number of COOH groups is reduced, with the average r(CO) becoming only slightly longer. The νCO frequency is likewise stable, dropping from 1728 to 1717 and then to 1715 cm−1 as the number of COOH groups lowers from 8 to 5 then to 3. This trend in red shift of CO mode with concentration of COOH group is also followed for the larger o-(10,10) tube. As the number of acid group decreases from 20 to 12, the CO frequency drops from 1749 to 1719 cm−1. The intensity of the most intense band decreases with increasing x for (4,4)-(COOH)x tube, but a reverse trend is found for larger (10,10) tubes. Thus, with decreasing concentration of acid groups, the carbonyl mode is red-shifted by 10−30 cm−1 for purified armchair tubes. (m,0)-(COOH)x. Analogous results are summarized in Figures 9 and S5 for zigzag tubes. For x = 3, all three COOH groups adopt a lf-COOH arrangement, which includes elongated r(CO) (>1.22 Å) and shorter r(C−COOH). The IR spectrum exhibits a single intense CO peak at 1662 cm−1. Placing two of the COOH groups consecutively in (8,0)(COOH)3 raises the energy by 9.0 kcal/mol and forces one of the two COOH groups to adopt a regular orientation. One may conclude that steric repulsions would tend to keep the COOH groups placed apart, confirming the results described above. When x is raised to 5, the additional two COOH groups of (8,0)-(COOH)5 positioned themselves as regular COOH, keeping arrangements of the existing three lf-COOH units unchanged. These regular arrangements manifest a CO peak of 1733 cm−1. When x is raised to 8, the spectrum shows two sharp CO peaks. The regular CO band at 1733 cm−1 is blue-shifted by only 7 cm−1, and the ν(CO) of lf-COOH is lowered by 8 cm−1. Intensity of the former band increases, but a sharp decrease is noted for the latter CO mode as concentration of acid group increases. Similar changes in geometric arrangements and IR spectra were found in the case of (12,0)-(COOH)x. The 1655 cm−1 peak of x = 3 is red-shifted by only 3 cm−1 for x = 7 and then blue-shifted by the same amount in x = 12; the intensity of the CO band drops as x increases, i.e., acid concentration increases. The CO mode of regular COOH is blue-shifted by 13 cm−1 with increased intensity as x changes from 7 to 12. For the largest (18,0)-(COOH)x tubes (see Figure S5), the intensity of the CO mode of lf-COOH groups remains constant for x = 5 and 18, but spectra show lower intensity for x = 10. However, the intensity of the CO band of regular COOH follows the same trend found in smaller tubes, and the peak is blue-shifted by 28 cm−1. As in smaller o-(12,0) tubes, ν(CO) of lf-COOH first decreases and then increases with increasing concentration of acid group. Due to steric effect between two adjacent COOH, the second isomer of (18,0)(COOH)5 (Figure S5) is less stable by 4.7 kcal/mol. Although

Figure 7. Optimized geometry and bond lengths (in Å) and normalized IR spectra (scaled by 0.960) of graphene-COOH. Lorentzian broadening with fwhm of 20 cm−1 was applied. νCO value (in cm−1) and intensity (in km/mol) are shown.

the zigzag edge of a graphene sheet. Just as in the nanotubes, a low-frequency mode, in this case 1690 cm−1, is observed for the zigzag COOH, as compared to a higher 1724 cm−1 peak at the armchair edge. The intensity pattern is also consistent with the tubes: that of the zigzag peak is more than three times higher than that of the armchair peak. These findings are unchanged when either armchair or zigzag edge is occupied separately, without the presence of the other. Note that the 1690 cm−1 peak remains higher than the 1650 cm−1 range of the nanotubes, leading one to conclude that the curvature of the tubes is a contributing factor as well. Effect of COOH Concentrations on IR Spectra. Raw nanotube samples were, in some studies, treated for a prolonged time in acid, and spectra were taken at different 26079

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C

Article

Figure 8. Optimized structures, CO and C−COOH bond lengths (in Å), IR spectra, and νCO (in cm−1) of the most intense peak of different concentrations of COOH groups of some o-(n,n) nanotubes. Italic numbers are the intensity (in km/mol). None of these o-SWNTs exhibited any imaginary frequency.

this isomer exhibits a single peak at 1650 cm−1, the ν(CO) of the single regular COOH is above 1700 cm−1 with low intensity. In summary, carboxylated zigzag single-wall carbon nanotubes, irrespective of their diameter and concentration of COOH groups, exhibit two CO stretching vibrational modes: one below and one above 1700 cm−1. The latter peak is common for any COOH group, but the former band is uncommon for any stand-alone carboxylated acid group. Such a CO band around 1650 cm−1 seems to be characteristic of zigzag tubes. At low concentration, only one CO band of purified zigzag tubes, around 1650−1670 cm−1, differentiates them from their cousin armchair tubes. O−H Modes of COOH Groups. In zigzag tubes, the OH bands of lf-COOF are slightly more intense compared to that of regular COOH. Also those lf-COOH groups exhibit a peak at 3556−3564 cm−1 compared to 3531−3542 cm−1 range in regular COOH groups. The O−H distances of lf-COOH are in general shorter by 1−3 mÅ than that of r-COOH, consistent with their higher frequency. In all cases those OH modes are pure in nature and no correlation exists with tube diameter. Had there been any H-bond, some of those O−H bands would have red-shifted significantly. Fully functionalized carboxylated armchair tubes have experimental ν(OH) values in the range of 3518−3557 cm−1, which overlaps the ranges of both armchair and zigzag tubes, so this mode is a poor candidate to distinguish these two structures. Since oxidized nanotubes also contain phenolic −OH groups, those may have influence on CO modes of COOH groups due to intergroup hydrogen bonding. To verify such possibilities, follow-up theoretical study on models of o-

SWNT with both acid and phenolic −OH at the end of tubes is in progress, and results will be published separately. Quinones within o-SWNTs. As mentioned above, the experimental IR spectra of several o-SWNTs exhibit a peak around 1650 cm−1, which has generally been assigned to a quinone CO band. However, the work described here suggests that such a peak could alternatively be associated with certain carboxylic groups of o-zigzag tubes. In order to differentiate these two possibilities, the frequencies of quinones embedded within nanotubes were calculated. Since quinones contain both carbonyl groups in the same ring, CO groups are only possible at the tips of armchair (n,n) tubes. As illustrated in Figure 10, the CO bond lengths (1.220 Å) of (4,4)(=O)2 are identical with those of o-benzoquinone. The asymmetric and symmetric CO stretches occur at 1701 and 1674 cm−1, respectively, in o-benzoquinone. Placement within the nanotube yields slight red shifts of these bands, including also a fourfold magnification of the intensity of the lowerfrequency band. When COOH groups are added to the nanotube, both CO bonds of quinone in (4,4)(COOH)6(O)2 are slightly shortened, and the frequency remains below 1700 cm−1. The band occurring at 1726 cm−1 is assigned to the carboxyl groups of this armchair tube. It was unnecessary to consider quinones at the tips of zigzag tubes as their presence would disrupt the conjugation (like mbenzoquinone) of the entire system. With respect, then, to the low-frequency CO stretches, these could arise first from quinone in o-armchair tubes or from lf-COOH of o-zigzag tubes, and the former frequencies are higher by about 20−30 cm−1 than the latter. Another key difference lies in their intensitiesthose of the lf-COOH 26080

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C

Article

Figure 9. Optimized structures, bond lengths (in Å), IR spectra, and νCO (in cm−1) of the most intense peak of different concentrations of COOH groups of o-(8,0) and o-(12,0) nanotubes. Italics numbers are the intensity (in km/mol). For the sake of comparison, IR spectra of (8.0)-(COOH)8 and (12,0)-(COOH)12 are also included in the figure. None of these o-SWNTs tubes exhibited any imaginary frequency.

early stage, and a 1650 cm−1 peak became visible after 10 min exposure time. The area of both peaks gradually increased (see Figure 1 of ref 23) as time wore on. Due to the presence of 1200 and 1040 cm−1 bands (even at the early stage of treatment), these authors assigned the 1740 cm−1 peak as an ester CO. An alternate explanation again takes the 1650 cm−1 peak to be a result of COOH groups in a zigzag tube. Yim and Johnson57 thoroughly investigated the etching surface of a (8,8) tube by O3 using DFT and ONIOM methods and did not find any quinone CO at the tube wall. However, some intermediate structures exhibit peaks at 1617 and 1677 cm−1, and both bands correspond to the same CO unit where a single oxygen is attached to a hexagon (ketone) of the defect site. The source of the 1650 cm−1 band could be partial oxidation at the ends of zigzag tubes with smaller diameter. A similar sample (diameter about 1.4 nm) was also used by Kim et al.,12 and H2O2-treated sample showed peaks at 1650 and 1623 cm−1. The authors expressed difficulties in explaining such a red shift of CO frequency, suggesting several possibilities such as H-bonding between OH and COOH groups (as in salicylic acid) or different coupling effects. Interestingly the same CVD sample used by Zhang et al.11 did not show a peak close to 1650 cm−1 when treated with either concentrated HNO3 or a mixture of HNO3 and H2SO4. Andersson and Grennberg13 also did not find any such peak in HNO 3 -treated HiPco sample. However, several studies24−27,58,59 clearly showed a peak around 1650 cm−1 for

species are more than double that of the quinone CO band, a factor which may help make a differentiation. On the other hand, the bands are very similar if there are no COOH groups on the tip along with the quinone. The latter situation may not occur as oxidation is likely to result in carboxylic groups as well as other oxygen-containing groups, such as quinone. Interpretation of Some Experimental Results. The results described above indicate that o-zigzag tubes, irrespective of size, at low acid concentration exhibit a single carbonyl peak around 1650 cm−1, and as the concentration increases the more common CO peak above 1700 cm−1 appears along with the low-frequency peak. On the other hand, at any concentration, o-armchair tubes exhibit a single carbonyl band above 1700 cm−1. The KMnO4-treated CVD sample (diameter in the range of 0.8−1.2 nm) of Zhang et al.11 indeed shows a peak at 1640 cm−1 at the initial stage of treatment (10 min; see Figure 7 in ref 10), along with a peak at 1740 cm−1, and the area of the latter peak gradually increases with treatment time, in contrast to the former which shows no such increase. This result could thus suggest that at the initial stage of treatment, both zigzag and armchair tubes are partially oxidized, and with increasing treatment time the concentration of COOH increases; these additional units adopt regular-COOH arrangements in both kinds of tubes, and hence the area of the higher frequency peak increases after 40 min. In another study, O3 treatment of arc discharge sample by Mawhinney et al.24 showed a prominent 1740 cm−1 peak at an 26081

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C

Article

Figure 10. Optimized structures (bond lengths in Å), IR spectra, and νCO (in cm−1) of o-benzoquinine, o-(4,4)-quinone, and carboxylated o-(4,4)quinone. Italics numbers are the intensity (in km/mol). s and as stand for symmetric and asymmetric stretching CO vibration.

geometric changes which require also connection to the curved surface of a nanotube. Some previous work had ascribed the low-frequency CO stretch to quinones at the end of nanotubes. And indeed, these entities do lead to a CO peak around 1680 cm−1. Thus, the CO band below 1700 cm−1 is possible from armchair oquinone or from the COOH groups of zigzag tubes, where the former mode is about 20−30 cm−1 higher than the carbonyl band of the latter. Besides, the frequency difference, νCO, of zigzag COOH has a higher intensity than does the quinone CO peak. The present study reveals differences between purified armchair and zigzag tubes based on COOH groups at the terminus. However, oxidized samples can be more complicated and contain other oxygen-containing groups, such as hydroxyl, ether, ketone, etc., and these groups may lie along the surface as well as at the tip. Future studies may take advantage of the combined SLDB and SNM approach applied here. However, the reliability of this prescription has been verified only for endfunctionalized SWNTs to this point so should be used with caution for other cases. For example, if the vibrational mode of interest couples with other modes, this approach may not be reliable and should be compared carefully with full vibrational analyses.

samples treated by such strong acids, irrespective of source of samples. It is hoped that the ideas expressed here may offer some guidance to workers attempting to resolve these discrepancies.



CONCLUSION

Theoretical IR spectra of a wide range of carboxylated armchair (n,n) and zigzag (m,0) nanotubes (o-SWNTs) were calculated along with their geometries. The models represent SWNTs synthesized from CVD, laser ablation, and arc-discharge methods. A completely new CO stretching mode was found around 1650 cm−1 for the COOH groups, without any H-bonding. This peak coincides with that observed by experimentalists, previously assigned as a quinone CO mode. This low-frequency CO mode of COOH appears only in oxidized zigzag tubes, irrespective of their diameter or number of acid groups. In addition to this uncommon CO band, both zigzag and armchair oxidized tubes exhibit a standard CO adsorption in the 1720−1760 cm−1 region. The low-frequency band is generally of greater intensity than the more common CO stretch. A closer look at the relevant geometric parameters and arrangements of COOH groups reveals the origin of this CO peak. Some COOH units of zigzag tubes are slightly bent out away from the tube wall. The CO bond length of these groups is about 0.01 Å longer than that of the more common COOH. Also, the C−C bond that links the COOH to the tube has some double-bond character. Consideration of simple benzoic acid and an extended planar carbon network of graphene-COOH shows that the bending of the COOH unit away from the aromatic ring cannot fully explain these



ASSOCIATED CONTENT

S Supporting Information *

Side views of o-SWNTs, optimized structures, and IR spectra of less stable isomers of several o-SWNTs. This material is available free of charge via the Internet at http://pubs.acs.org. 26082

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083

The Journal of Physical Chemistry C



Article

(26) Ramanathan, T.; Fisher, F. T.; Ruoff, R. S.; Brinson, L. C. Res. Lett, Nanotechnol. 2008, 2008, 296928. (27) Cortes, P.; Deng, S.; Camacho, L.; Smith, G. B. J. Sens. 2010, 2010, 691585. (28) Berger, S.; Hertl, P.; Rieker, A. Physical and Chemical Analysis of Quinones. In The Chemistry of Quinonoid Compounds; Patai, S., Rappoport, Z., Eds.; John Wiley & Sons Ltd.: New York, 1988; Vol. II, pp 29−86. (29) Hamon, M. A.; Itkis, M. E.; Niyogi, S.; Alvaraez, T.; Kuper, C.; Menon, M.; Haddon, R. C. J. Am. Chem. Soc. 2001, 123, 11292− 11293. (30) Kar, T.; Akdim, B.; Duan, X.; Pachter, R. Chem. Phys. Lett. 2004, 392, 176−180. (31) Kar, T.; Akdim, B.; Duan, X.; Pachter, R. Chem. Phys. Lett. 2006, 423, 126−130. (32) Kar, T.; Bettinger, H. F.; Scheiner, S.; Roy, A. K. J. Phys. Chem. C 2009, 112, 20070−20075. (33) Maseras, F.; Morokuma, K. J. Comput. Chem. 1995, 16, 1170. (34) Morokuma, K. Bull. Korean Chem. Soc. 2003, 24, 797−801. (35) Montoya, A.; Mondragon, F.; Truong, T. N. Carbon 2002, 40, 1863−1872. (36) Yim, W. L.; Lou, Z. F. Chem. Phys. Lett. 2004, 398, 297−303. (37) Chen, Z.; Nagase, S.; Hirsch, A.; Haddon, R. C.; Thiel, W.; Schleyer, P. v. R. Angew. Chem., Int. Ed. 2004, 43, 1552−1554. (38) Vreven, T.; Thompson, L. M.; Larkin, S. M.; Kirker, I.; Bearpark, M. J. J. Chem. Theory Comput. 2012, DOI: 10.1021/ ct300612m. (39) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652. (40) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785−789. (41) Stephens, P. J.; Devlin, F. J.; Cgabalowski, C. F.; Frisch, M. J. J. Phy. Chem. 1994, 98, 11623−11627. (42) Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502−16513. (43) Cheeseman, J. R.; Frisch, M. J.; Devlin, F. J.; Stephens, P. J. Chem. Phys. Lett. 1996, 252, 211−220. (44) Devlin, F. J.; Finley, J. W.; Stephens, P. J.; Frisch, M. J. J. Phys. Chem. 1995, 99, 16883−16902. (45) Johnson, R. D., III. NIST Standard Reference Database Number 101. In NIST Computational Chemistry Comparison and Benchmark Database; Johnson, R. D., III, Ed.; NIST: Gaithersburg, MD, 2006. (46) Kar, T.; Scheiner, S.; Roy, A. K. J. Phys. Chem. C 2012, DOI: 10.1021/jp3089947. (47) Reiher, M.; Neugebauer, J. J. Chem. Phys. 2003, 118, 1634− 1641. (48) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; et al. Gaussian09; Gaussian, Inc., Wallingford CT, 2009. (49) Schaftenaar, G.; Noordik, J. H. J. Comput.-Aided Mol. Des. 2000, 14, 123−134. (50) Allouche, A. R. J. Comput. Chem. 2011, 32, 174−182. (51) Veiga, R. G. A.; Tomanek, D. TubeVBS (http://k.1asphost. com/tubeasp/tubevbs.html), 2007. (52) Zhurko, G. A. http://www.chemcraftprog.com. (53) Chelmecka, E.; Pasterny, K.; Kupka, T.; Stobinski, L. J. Mol. Model 2012, 18, 2241−2246. (54) Scheiner, S. Hydrogen Bonding. A Theoretical Perspective; Oxford University Press: Oxford, U.K., 1997. (55) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88, 899−926. (56) Reed, A. E.; Weinhold, F.; Curtiss, L. A.; Pochatko, D. J. J. Chem. Phys. 1986, 84, 5687−5705. (57) Yim, W.-L.; Johnson, K. J. J. Phys. Chem. C 2009, 113, 17636− 17642. (58) Nogueira, A. F.; Lomba, B. S.; Soto-Oviedo, M. A.; Corio, P.; Furtado, C. A.; Hummelgen, I. A. J. Phys. Chem C 2007, 111, 18431− 18438. (59) Aizawa, M.; Shaffer, M. S. P. Chem. Phys. Lett. 2003, 2003, 121− 124.

AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected]; fax 1-435-797-3390. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the U.S. DoD High Performance Computing Modernization Program (HPCMP) and NSF (Grant CHE-1026826). H.F.B. thanks the Fonds der chemischen Industrie for financial support. We thank the AFRL/DSRC personnel for their support in using the DoD Supercomputing resources.



REFERENCES

(1) Niyogi, S.; Hamon, M. A.; Hu, H.; Zhao, B.; Bhowmick, P.; Sen, R.; Itkis, M. E.; Haddon, R. C. Acc. Chem. Res. 2002, 35, 1105−1113. (2) Lin, T.; Bajpai, V.; Ji, T.; Dai, L. Aust. J. Chem. 2003, 56, 635− 651. (3) Basiuk, V. A.; Basiuk (Golovataya-Dzhymbeeva), E. V. Chemical Derivatization of Carbon Nanotube Tips; American Scientific Publishers: Valencia, CA, 2004; Vol. 1. (4) Banerjee, S.; Hemraj-Benny, T.; Wong, S. S. Adv. Mater. 2005, 17, 17−29. (5) Tasis, D.; Tagmatarchis, N.; Bianco, A.; Prato, M. Chem. Rev. 2006, 106, 1105−1136. (6) Kauffman, D. R.; Star, A. Angew. Chem., Int. Ed. 2008, 47, 6550− 6570. (7) Cao, Q.; Rogers, J. A. Adv. Mater. 2009, 21, 29−53. (8) Lu, F.; Gu, L.; Meziani, M. J.; Wang, X.; Luo, P. G.; Veca, L. M.; Cao, L.; Sun, Y.-P. Adv. Mater. 2009, 21, 139−152. (9) Tian, B.; Kempa, T. J.; Lieber, C. M. Chem. Soc. Rev. 2009, 38, 16−24. (10) Gerber, I.; Oubenali, M.; Bacsa, R.; Durand, J.; Goncalves, A.; Pereira, M. F. R.; Jolibois, F.; Perrin, L.; Poteau, R.; Serp, P. Chem. Eur. J. 2011, 17, 11467−11477. (11) Zhang, J.; Zou, H.; Qing, Q.; Yang, Y.; Li, Q.; Liu, Z.; Guo, X.; Du, Z. J. Phys. Chem. B 2003, 107, 3712−3718. (12) Kim, U. J.; Furtado, C. A.; Liu, X.; Chen, G.; Eklund, P. C. J. Am. Chem. Soc. 2005, 127, 15437−15445. (13) Andersson, C.-H.; Grennberg, H. Eur. J. Org. Chem. 2009, xx, 4421−4428. (14) Wang, Y.; Iqbal, Z.; Mitra, S. J. Am. Chem. Soc. 2005, 128, 95− 99. (15) Chen, Z.; Kobashi, K.; Rauwald, U.; Booker, R.; Fan, H.; Hwang, W.-F.; Tour, J. M. J. Am. Chem. Soc. 2006, 128, 10568−10571. (16) Hu, C.; Chen, Z.; Shen, A.; Shen, X.; Li, J.; Hu, S. Carbon 2006, 44, 428−434. (17) Park, M. J.; Lee, J. K.; Lee, B. S.; Lee, Y.-W.; Choi, I. S.; Lee, S.g. Chem. Mater. 2006, 18, 1546−1551. (18) Martinez-Rubi, Y.; Guan, J.; Lin, S.; Scriver, C.; Sturgeon, R. E.; Simard, B. Chem. Commun. 2007, 5146−5148. (19) Umeyama, T.; Tezuka, N.; Fujita, M.; Matano, Y.; Takeda, N.; Murakoshi, K.; Yoshida, K.; Isoda, S.; Imahori, H. J. Phys. Chem. C 2007, 111, 9734−9741. (20) Shieh, Y.-T.; Liu, G.-L.; Wu, H.-H.; Lee, C.-C. Carbon 2007, 45, 1880−1890. (21) Imasaka, K.; Kato, Y.; Suehiro, J. Nanotechnology 2007, 18, 335602−335609. (22) Kar, T.; Scheiner, S.; Roy, A. K. Chem. Phys. Lett. 2008, 460, 225−229. (23) Kar, T.; Scheiner, S.; Patnaik, S. S.; Bettinger, H. F.; Roy, A. K. J. Phys. Chem. C 2010, 114, 20955−20961. (24) Mawhinney, D. B.; Naumenko, V.; Kuznetsova, A.; Yates, J. T., Jr.; Liu, J.; Smalley, R. E. J. Am. Chem. Soc. 2000, 122, 2383−2384. (25) Kuznetsova, A.; Mawhinney, D. B.; Naumenko, V.; Yates, J. T., Jr.; Liu, J.; Smalley, R. E. Chem. Phys. Lett. 2000, 321, 292−296. 26083

dx.doi.org/10.1021/jp309699z | J. Phys. Chem. C 2012, 116, 26072−26083