Noncovalent Interaction between Aniline and Carbon Nanotubes

Mar 11, 2010 - Beáta Peles-Lemli,†,§ Gergely Matisz,†,§ Anne-Marie Kelterer,‡ Walter M. F. ... Graz, A-8010, Austria, and Institute of Chemis...
0 downloads 0 Views 2MB Size
5898

J. Phys. Chem. C 2010, 114, 5898–5905

Noncovalent Interaction between Aniline and Carbon Nanotubes: Effect of Nanotube Diameter and the Hydrogen-Bonded Solvent Methanol on the Adsorption Energy and the Photophysics Bea´ta Peles-Lemli,†,§ Gergely Matisz,†,§ Anne-Marie Kelterer,‡ Walter M. F. Fabian,§ and Sa´ndor Kunsa´gi-Ma´te´*,† Department of General and Physical Chemistry, UniVersity of Pe´cs, Ifju´sa´g 6, Pe´cs, H-7624, Hungary, Institute of Physical and Theoretical Chemistry, Graz UniVersity of Technology, Technikerstr. 4/I, Graz, A-8010, Austria, and Institute of Chemistry, Karl-Franzens UniVersity of Graz, Heinrichstr. 28, Graz, A-8010, Austria ReceiVed: September 3, 2009; ReVised Manuscript ReceiVed: January 27, 2010

The adsorption of aniline on SWCNTs has been investigated using three different DFT methods, PW91LYP, the hybrid m-GGA MPWB1K, and the dispersion-corrected BP86-D functionals. The BSSE-corrected adsorption energies for top orientation of aniline to the nanotube surface are Eads ) 9.7 kcal mol-1 and Eads ) 1.4 kcal mol-1 with BP86-D/SVP and MPWB1K/6-311+G(d), respectively. Results validated that the adsorption energy of aniline depends on the diameter of the nanotube with a pronounced manner, especially for small diameters. The TDDFT calculations of UV/vis spectra predict two electronic transitions resulting from nanotube excitations in the visible spectra (>500 nm) and one imperceptible aniline f SWCNT excitation at ca. 1200 nm. Inclusion of solvent effects (MeOH) by explicit solvent molecules leads to a pronounced red shift of the weak longest wavelength aniline f nanotube transition, whereas the strong intratube excitations are only a little influenced. Fluorescence excitation spectra (λem ) 586 nm) of SWCNTs in aniline yield a broad structured absorption in the range of 22 000-18 000 cm-1, which is insensitive to the addition of methanol as cosolvent. Introduction The adsorption of aromatic molecules, such as benzene, phenol, dibenzothiophene, and cytosine, on single-walled carbon nanotubes (SWCNTs) and graphene sheets has attracted increased attention in the past few years.1–5 This is because carbon nanotubes are not soluble in most of the common solvents, and one of the known approaches to improve their solubility in aromatic solvents is based on π-π interactions between the aromatic “packer” compounds and the SWCNT.6 Thus, the π-π interactions between the aromatic solvent molecules and the nanotube are considered as key features in the solubilization process. However, this property of carbon nanotubes also influences their potential applications as adsorbent for solidphase extraction7 or in sensor applications.8 One known solvent of the SWCNTs is aniline.9–12 According to the importance of the solubilization also in their industrial application, much effort has been done to clarify the solubilization mechanisms, which, however, still have not been completely clarified yet. Recent ab initio calculations1,13,14 have also shown that, during the adsorption process of benzene derivatives onto the nanotube surface, the noncovalent π-π interaction plays the dominant role. Our previous experimental work15 concerning the possible carrier properties of aniline derivatives toward other common solvents supports the dominant role of this π-π interaction-based solubilization process of SWCNTs in aniline. The photoluminescence (PL) measurements * To whom correspondence should be addressed. Tel: +36-72503600(-4208). Fax: +36-72-501518. E-mail: [email protected]. † University of Pe´cs. ‡ Graz University of Technology. § Karl-Franzens University of Graz.

suggested that, although the aniline still has nice carrier behavior in the alcoholic solvents, the structure of the aniline-SWCNT complex shows significant changes in the presence of different alcoholic or carbon tetrachloride molecules. This behavior has also consequences on the solubilization processes. Therefore, because of its practical importance, the influence of solvents of different polarity on the noncovalent interaction between SWCNTs and aromatic packer molecules needs further investigations at the molecular level. Adsorption through π-π stacking of aniline on (8,0) SWCNTs has been investigated by Woods et al. with the local density approximation (LDA) method using plane waves.14 Contrary to the prevailing belief, this approach has been shown to underestimate dispersion interactions and, as a consequence, the adsorption energy.16 GGA functionals like PW91LYP have also been used for π-stacking molecules, but usually these also underestimate the adsorption energy.16 Other density functionals developed for a proper description of π-π stacking interactions include MPWB1K17 as well as dispersion-corrected DFT, introduced by Grimme.18 Recently, the noncovalent interaction between methanol and SWCNTs was investigated by Pankewitz et al. using the BP86-D dispersion-corrected functional.19 Using these latter two functionals (MPWB1K and BP86-D) and for comparison, PW91LYP, in this work, we perform quantum chemical calculations on the interaction between one aniline molecule and a fragment of rolled graphene surfaces with varying curvatures. The influence of MeOH as solvent on the electronic structure and, especially, the UV/vis spectroscopic properties of the aniline-SWCNT complexes are assessed by the supermolecule approach. Fluorescence excitation spectra are used to study experimentally the effect of methanol on the

10.1021/jp908505q  2010 American Chemical Society Published on Web 03/11/2010

Noncovalent Interaction between Aniline and CNTs

J. Phys. Chem. C, Vol. 114, No. 13, 2010 5899

photophysical properties of the aniline-SWCNT system. The results of these combined computational and experimental investigations should allow a rationalization of our previous experimental results regarding the solubilization of SWCNTs.15 Methods and Models Theoretical Methods. For the DFT calculations, the PW91LYP,20 MPWB1K,17 and the dispersion-corrected BP86D21 functionals were used. For a discussion of the reliability of the chosen computational procedures, see, for example, refs 4 and 13. In addition, benchmark calculations for the adsorption of benzene on graphene were performed. The model system to describe the aniline-SWCNT interaction was constructed in the following way: In the first step, an SWCNT tube with a molecular formula of C128H16 was fully optimized at the PW91LYP and MPWB1K level with the 3-21G basis set. From this nanotube, a fragment consisting of seven six-membered rings with a molecular formula of C42H16 was extracted and used in the calculation of the interaction with one molecule of aniline. The geometry of this fragment was frozen at the geometry obtained in the first step, and the dangling bonds were saturated by hydrogen atoms. The aniline molecule was allowed to move freely above the fragment of the SWCNT, and all of the atoms of aniline were allowed to relax during the calculation. To reduce the computational efforts, in the optimization of the complex between aniline and the C42H16 fragment, the so-called same level different basis set (SLDB) method for modeling SWCNTs was applied.22 In this approach, for the atoms of the central four pyrene-like rings of the C42H16 fragment as well as for those of the aniline molecule, the 6-311+G(d)23 and, for the remaining atoms, the STO 3-21G24 basis sets were used. In the final step, the adsorption energy, eq 1, was calculated using the 6-311+G(d) basis set on all atoms.

Eads ) Eaniline + ESWCNT - Eaniline/SWCNT

(1)

BP86-D optimizations were done with the SVP25 basis set, followed by TZVP26 single-point calculations for the (n,0) SWCNT (n ) 8, 10, 30). For selected complexes, the basis set superposition error was calculated. In analogy to the recent work of Woods et al,14 three π-stacked orientations of the aniline molecule with respect to the nanotube fragment, namely, top, hollow, and bridge, were considered as starting structures. In addition, either the hydrogen atoms of the NH2 group (dubbed “down”) or the nitrogen’s lone pair of aniline (“up”) can point toward the nanotube. Both possibilities were considered. In the interaction of nucleic acid bases with SWCNTs, T-shaped complexes of nucleic acid bases had quite low energies, at least with some density functionals.4 Consequently, such T-shaped orientations of the aniline molecule were also treated. The influence of the curvature (diameter) of the SWCNT was studied by using C42H16 fragments derived from several zigzag ((n,0) (n ) 8, 10, and 30)) SWCNTs as well as the corresponding graphene fragment with an infinite diameter. The possible effect of the size of the fragment representing the nanotube was investigated as follows: For the calculation of the adsorption energies of aniline, the C16H10, C42H16, and C80H22 fragments were adapted from the (8,0) SWCNT, where the whole (8,0) tube consists of C128H16 atoms. Preliminary MPWB1K/6-311+G(d) calculations for these fragments with different sizes and the whole C128H16 tube indicated that extending the fragment size beyond C42H16 has only a rather small effect on adsorption energies. Hence, this fragment size

was used for all further calculations, except those for aniline-methanol clusters. To completely accommodate the cluster here, the larger C64H20 fragment was used. D’Amore and co-workers showed the importance of the explicit consideration of the solvent molecules, especially in the first solvation shells, during the investigation of the conformation behavior of some nitroxide amino acid derivatives in aqueous solution.27 Accordingly, we have also calculated the interaction between the aniline moiety and explicit methanol solvent molecules. Electronic excitation energies were obtained by time-dependent density functional theory (TDDFT).28 For the quantum chemical calculations, the Gaussian 0329 and TURBOMOLE30 program packages were used. Preparation, manipulation, and visualization of coordinate files were done with MOLDEN31 and HyperChem Professional 7.32 Molecular orbitals were plotted with gOpenMol.33 Experimental Methods. The SWCNTs were purchased from Guangzhou Heji Trade Co. (China) and used as received. The end-capped tubes’ average diameter was 2.0 nm, and they were 5-15 µm in length with a purity of more than 95%. The aniline was purchased from Aldrich and was freshly distilled before preparing the aniline-SWCNT stock solution. The dissolution of SWCNTs in aniline was carried out as described earlier.12 The MeOH (HPLC grade) solvent was purchased from Panreac (Spain). A 10-fold dilution of the aniline-SWCNT stock solution was prepared by aniline or MeOH. The diluted solutions were homogenized by ultrasonic shaking during 30 s. A highly sensitive Fluorolog τ3 spectrofluorometer (JobinYvon/SPEX) was used to investigate the excitation spectra of the different solutions. For data collection, a photon counting method with a 0.1 s integration time was used. Excitation and emission bandwidths were set to 1 nm. A 2 mm thickness of the fluorescent probes with front face detection was used to eliminate the inner filter effect. The measurements were carried out at the temperature of 293.2 K and an emission wavelength of λem ) 586 nm. Results and Discussion Adsorption Energies. Carbon nanotubes are known to form with different diameters. To investigate the effect of diameter on the adsorption energy, calculations on C42H16 fragments with different curvatures, obtained from the respective zigzag ((n,0) (n ) 8, 10, and 30)) SWCNTs, and the planar graphene segment have been performed (Table S1 of the Supporting Information). In line with the increased binding energy calculated for π-π stacked cytosine-SWCNT complexes4a,5a,34 or decreased reactivity toward electrophilic additions35 with increasing SWCNT diameter, here, we also find an increase of adsorption energies with increasing nanotube diameter. A combined experimental and theoretical study of the interaction between acetone and SWCNTs also revealed a strong correlation between physisorption energy and nanotube diameter. Moreover, the dispersion forces were found be dominant.36 Accepting π-π stacking as the dominant contribution to the adsorption of aromatic molecules on the surface of carbon nanotubes, this increase of Eads with decreasing curvature is anticipated. Except for very small diameters, the variation of the SWCNT-aniline distance (for the most stable top-down arrangement measured by the distance between the aniline’s ipso-C and the SWCNT carbon beneath it) is not very pronounced. Slightly shorter distances are found with BP86-D as compared with MPWB1K calculations (Table S1 in the Supporting Information). Previously, Woods et al.14 have used LDA/plane wave DFT calculations to study the adsorption behavior of simple benzene

5900

J. Phys. Chem. C, Vol. 114, No. 13, 2010

Peles-Lemli et al.

TABLE 1: Calculated [BP86/SVP, MPWB1K/6-311+G(d), and PW91LYP/6-311+G(d)] Adsorption Energies (kcal mol-1) for the Aniline-(8,0) SWCNT Fragment Eads (BSSE-corrected value in parentheses) BP86-D top-down

13.8 (9.7) 11.3 (9.8)a

b top-up hollow-down 10.6 b bridge-down T-shaped parallel 6.2 T-shaped perpendicular 12.3

MPWB1K PW91LYP LDA14 6.2

3.5

5.2 (1.4) 2.8

6.5

b

3.0 (0.5) 2.9c

b

b

5.6 7.0

2.8 (1.4) 2.4 (1.5)

a BP86-D/TZVP results. b Collapses to top-down upon optimization. c Strongly distorted from the “true” hollow geometry.

Figure 1. Optimized structures (BP86-D/SVP) of aniline-(8,0) SWCNT fragment complexes. Top line: top-down (left) and hollowdown (right). Bottom line: T-shaped parallel (left) and T-shaped perpendicular (right).

derivatives, including aniline, on a semiconducting (8,0) SWCNT. Consequently, to be comparable with this investigation here, we also have used a C42H16 fragment of a (8,0) carbon nanotube to obtain adsorption energies for different orientations of the aniline molecule with respect to the SWCNT by the BP86-D/ SVP and MPWB1K/6-311+G(d) procedures. Calculated adsorption energies Eads are summarized in Table 1. Also included are the respective data obtained by Woods et al.14 and those resulting from PW91LYP calculations. Thermal desorption of benzene from graphite resulted in a binding energy of 11.5 kcal mol-1 (0.50 eV).37 This value is used to benchmark the three density functionals (PW91LYP, MPWB1K, and BP86-D) using the same C42H16 model system. Adsorption energies (BSSEcorrected data in parentheses) using the BP86-D/SVP geometries are: PW91LYP, -4.4 (-7.3); MPWB1K, 5.0 (-); BP86-D/SVP, 13.8 kcal mol-1 (10.9). When optimized by MPWB1K, no significant change is found; Eads ) 5.6 kcal mol-1. From these data, we conclude that the BP86-D aniline-SWCNT adsorption energies are the most reliable ones. Structures obtained by BP86-D calculations for these complexes are shown in Figure 1. In the top geometry of the aniline-SWCNT, the ipso- and the two m-carbon atoms of the aniline molecule are each located on top of a carbon atom of the SWCNT fragment. In the hollow conformation, all carbon atoms of the aniline are above a carbon atom of an SWCNT’s benzene ring. In both arrangements, the angle between the aniline ipso-C, the SWCNT-C below it, and the SWCNT long axis p-C is ca. 90°. Contrary to the LDA/plane wave investigation,14 where top, bridge, and hollow configurations of aniline-(8,0) SWCNT complexes were found, no true bridge

structure could be obtained by BP86-D or MPWB1K calculations. Despite several attempts with this latter functional, in addition, no hollow structure could be located. Instead, initial bridge arrangements collapsed to top complexes. Initial hollow conformations also collapsed to top structures when using the MPWB1K functional. All density functionals used resulted in the top structure as the most stable one (Table 1). With respect to the orientation of the NH2 hydrogen atoms, up or down, up and down structures were only found with MPWB1K and PW91LYP, whereas the BP86-D procedure resulted solely in down structures. In addition to these π-stacked arrangements, also T-shaped structures were considered. Initially, T-shaped parallel and perpendicular structures have the aniline axis, that is, the line connecting the nitrogen and p-carbon atom, oriented along and at right angle, respectively, to the SWCNT long axis. In both structures, initially, the angle between these two axes is 90°. BP86-D/SVP optimization reduced these angles to ∼60° (T-shaped parallel) and ∼0° (T-shaped perpendicular), but without change of the orientation between the two axes. Thus, this latter structure actually corresponds to a π-stacked arrangement, but with the aniline moiety still oriented perpendicular to the SWCNT long axis. The adsorption energy of this conformation (Figure 1, bottom right) is only 1.5 kcal mol-1 smaller than that for the top-down orientation (BP86-D/SVP, Table 1). In line with the results of Woods et al.,14 true T-shaped structures are less stable than π-stacked structures. All three functionals, BP86D/SVP, PW91LYP/6-311+G(d), and MPWB1K/6-311+G(d), lead to the top-down orientation of the aniline molecule as the lowest energy arrangement. Calculated adsorption energies are quite large with BP86-D/SVP, Eads ) 13.8 kcal mol-1 (9.7 with BSSE correction). Reoptimization with BP86-D/TZVP does not significantly change the results; see Table 1. For the interaction of aniline with the (30,0) SWCNT fragment, Eads ) 17.3 kcal mol-1. MPWB1K calculations yield slightly smaller adsorption energies than the LDA approach14 and somewhat larger ones as obtained with the PW91LYP functional (Table 1). Taking into account the basis set superposition error significantly (≈ 4 kcal mol-1) reduces the adsorption energies, yielding Eads ) 1.4 kcal mol-1 and Eads ) 9.7 kcal mol-1 with the MPWB1K/ 6-311+G(d) and BP86-D/SVP approaches, respectively. Although, experimentally,38 a very weak interaction between aniline and the nanotube surface has been found, ∆G ≈ 0 kcal mol-1, as pointed out above, from the benchmark calculations on the graphene-benzene interaction, we consider the BP86-D results more reliable. Theoretical analysis of the acetone-SWCNT interaction has shown that dispersion forces are dominant.36 Hence, our conclusions concerning the reliability of the functionals used are in line with this work. To model solvent effects (MeOH), we reoptimized the aniline-(30,0) SWCNT top-down structure, adding one and two explicit MeOH molecules at the BP86-D/SVP level. However, to accommodate the physisorbed aniline-methanol clusters, a larger fragment (C64H20) had to be used. Placement of the MeOH molecule was done according to the two published structures of the aniline-methanol cluster.39 In contrast to the bare aniline-MeOH clusters, interaction with the SWCNT fragment results in a diminished stability (∆E ) 4.7 kcal mol-1 with BP86-D/SVP) of the structure with MeOH on top of the aniline amino group. In this structure, the OH group is involved in a hydrogen bond with the aniline nitrogen atom (Figure S1 in the Supporting Information), r(N · · · H-O) ) 1.904 Å, and the methyl group is pointing toward the aniline ring. The second, more stable cluster involves a hydrogen bond between the oxygen atom of methanol and the aniline NH, r(N-H · · · O) )

Noncovalent Interaction between Aniline and CNTs 1.879 Å, with the methyl group of MeOH pointing away from the aromatic ring of the aniline molecule. Furthermore, the aniline moiety is twisted out of the pure top orientation. Similar to the structure calculated by Pankewitz and Klopper19 for the interaction of MeOH with SWCNTs, the O-H group of MeOH is pointing toward the nanotube surface, indicating an O-H · · · π interaction. Adding a second methanol molecule to this latter, more stable, structure resulted in a symmetrical as well as slightly more stable (∆E ) -0.6 kcal mol-1 with BP86-D/SVP) unsymmetrical arrangement of the two MeOHs with respect to the aniline moiety. In both structures, both OH groups of the methanol molecules point toward the nanotube surface (Figure S2 in the Supporting Information). Furthermore, the amino group pyramidalization is reduced and, in the more stable unsymmetrical arrangement, the hydrogen atoms of the NH2 group are pointing up from the SWCNT surface. During their investigation of the equilibrium geometries and binding energies of the MeOH-SWCNT interaction, Pankewitz and Klopper19 have found that the structure where the C-O bond of the MeOH molecules lies parallel with the nanotube surface has a greater binding energy than the structure where this bond is perpendicular to the surface and the -OH group is located farther from the tube. Although, in our calculations with two MeOH molecules, the MeOH molecules moved out from these “clear” orientations, the present results show that the C-O bonds of both MeOH molecules deviate from parallel orientation by an angle of ∼20° to the SWCNT. However, the alcoholic hydrogen atom points toward the SWCNT with a distance of 2.10 and 2.41 Å to the nearest SWCNT carbon atoms (see Figure S1 in the Supporting Information). At the same time, the aniline-SWCNT distance increases from 3.01 to 3.04 Å. Therefore, our BP86-D calculations including aniline and two solvent molecules are comparable with the structure obtained by Pankewitz and Klopper. The electron distribution of the DFT (BP86-D) orbitals for the C64H20 (30,0) SWCNT fragment, the complex with aniline, as well as with two additional methanol molecules in their unsymmetrical arrangement is shown in Figure 2. Although the presented DFT orbitals are influenced by the finite size of the fragment and tangling hydrogen atoms on the SWCNT, the main feature of the orbitals is their orientation along either the long or the short axis of the SWCNT. For instance, in the C64H20 (30,0) SWCNT fragment, HOMO and HOMO-1 are oriented along the long SWCNT axis, with HOMO-2 perpendicular to it. Interaction with the aniline molecule leads to the appearance of a high-lying occupied π-orbital (HOMO-1), which is nearly exclusively localized at the aniline moiety. Consequently, the HOMO-1 orbital of the (30,0) SWCNT fragment becomes HOMO-2 in the aniline-SWCNT complex. Addition of one or two methanol molecules raises the energy of the aniline π orbital even further. Thus, in both structures, it becomes the HOMO. The appearance of the lowest virtual orbitals, LUMO (oriented along the SWCNT long axis) and LUMO+1 (perpendicular to the SWCNT long axis), is barely influenced by the presence of aniline or MeOH moieties (Figure 2). On the basis of the orbitals in Figure 2, the excitation process will be discussed in the next section. UV/vis Spectra. The experimental excitation spectra of aniline solutions of carbon nanotubes in the absence and in the presence of 90% MeOH are shown in Figure 3. Excitation spectra of the pure aniline solvent after repeating the reflux applied during the solubilization process are also plotted on this figure for comparison. Characteristic is a broad structured band

J. Phys. Chem. C, Vol. 114, No. 13, 2010 5901

Figure 2. Molecular orbitals of the C64H20 (30,0) SWCNT fragment and its top-down aniline complex without and with two MeOH molecules. MOs are obtained by the BP86-D/SVP gas-phase calculations using surface values of 0.025 au, and 0.05 au for aniline orbitals.

Figure 3. Experimental fluorescence excitation spectra of refluxed aniline, SWCNTs in aniline, and the aniline/MeOH mixture, measured in a 2 mm cell with an emission wavelength of 586 nm.

in the range of 22 000-18 000 cm-1. Dissolution with methanol does not change the position of this band nor its fine structure.

5902

J. Phys. Chem. C, Vol. 114, No. 13, 2010

Peles-Lemli et al.

TABLE 2: BP86-D/SVP TDDFT Calculated Absorption Wavelengths and Oscillator Strengths for the C42H16 (30,0) Fragment and the Top-Down Aniline Complexes aniline-(30,0) SWCNT fragment

(30,0) SWCNT fragment

C42H16 λ/nm a

883 631 621 585 583 553 509 487 471 462 a

C42H16 f 0.003 0.219 0.003 0.000 0.000 0.000 0.002 0.005 0.090 0.225

a

λ/nm

f

630 590 554 498 462 461 430 428 413 411

0.242 0.000 0.000 0.000 0.362 0.000 0.000 0.000 0.000 0.085

Numbers in bold indicate excitations from the aniline moiety.

To aid in the interpretation of these experimental findings, TDDFT calculations on the model systems described above have been performed. First, the performance of BP86-D and MPWB1K functionals with respect to excitation energies is assessed via calculations of the first few electronic transitions of pyrene, coronene, and ovalene (Table S2 in the Supporting Information). Generally, the BP86-D functional underestimates ∆E, irrespective of whether the SVP or TZVP basis set is used (0.1-0.2 eV for Lb and ∼0.4 eV for La bands); in contrast, MPWB1K/ 6-311+G(d) calculations result in too high values for ∆E (0.3 eV for Lb and 0.1 eV for La bands; see Table S2 in the Supporting Information). To the best of our knowledge, for pentapheno[2,1,14,13,12,11-defghijkl]pyreno[2,1,10,9,8,7-pqrstuv]pentaphene (circumpyrene; the C42H16 graphene fragment), no experimental but only calculated [TDDFT-B3LYP/6-31G(d)] electronic excitation energies are available. These previous calculations [B3LYP/6-31G(d)] gave absorptions at 545, 420, and 380 nm.40 MPWB1K/6-311+G(d) and BP86-D/SVP calculations give 442, 339, and 310 nm and 565, 431, and 385 nm, respectively. Generally, density functionals without HF exchange underestimate electronic excitation energies, whereas purely HF-based calculations result in too large excitation energies.41 For comparison purposes, thus, B3LYP/6-31G(d) excitation energies are also included in Table S2 in the Supporting Information. Not surprisingly, BP86-D excitation energies are closer to the B3LYP results than those obtained by the MPWB1K functional. On the basis of this evidence, there is no clear preference for either one of the two computational procedures, BP86-D versus MPWB1K. For computational efficiency, especially for the larger C64H20 fragment, excitation energies preferentially were calculated by BP86-D/SVP. The experimental SWCNT diameter of 2.0 nm corresponds approximately to the (30,0) SWCNT fragment. Thus, in Table 2, BP86-D absorption wavelengths and oscillator strengths for this C42H16 (30,0) SWCNT fragment and its aniline complex are presented. More detailed results, CI vectors, orbitals involved in the various electronic transitions, and their dependence on nanotube diameter (curvature of C42H16 fragment) are provided in Table S3 (MPWB1K) and Table S4 (BP86-D) in the Supporting Information. BP86-D calculations predict two strong transitions for the (30,0) SWCNT fragment: λ ) 630 nm, f ) 0.242 and λ ) 462 nm, f ) 0.362. For the longest wavelength absorption, a blue shift with increasing nanotube diameter is obtained. MPWB1K/6-311+G(d) TDDFT yields quite comparable results. As expected from the PAH results, the MPWB1K excitation energies are considerably blue shifted compared with

those from the BP86-D calculations (Table S3 in the Supporting Information). Complexation of aniline by the SWCNT fragment leads to a new, albeit quite weak, long-wavelength transition, λ ) 883 nm, f ) 0.003, resulting from an aniline f nanotube excitation. However, the strong transitions at 630 and 462 nm are barely affected by the interaction of the nanotube fragment with aniline, λ ) 631 nm, f ) 0.22 and λ ) 462 nm, f ) 0.23 (Table 2). In contrast to the strong transition, the longest wavelength aniline f nanotube excitation predicted by the BP86-D calculations shows a significant red shift with increasing nanotube diameter: 816 versus 829 versus 883 nm for aniline-(8,0), aniline-(10,0), and aniline-(30,0) SWCNT fragments (Table S4 in the Supporting Information). It is interesting to compare the present results with previous, apparently contradictory, observations. Interaction of SWCNTs with electron-donor molecules (tetrathiafulvalene, aniline) or electron-withdrawing (tetracyanoethylene, nitrobenzene) has been found to cause significant changes in the electronic structure of the nanotubes.42 Specifically, the S22 band of SWCNTs appears at 1090 nm on interaction with aniline, whereas in the corresponding benzene complex, absorption occurs at 1054 nm.42 In stark contrast to these experimental results, adsorption of pyrene derivatives on (n,n) armchair nanotubes (n ) 3-13) did not at all change the calculated UV/ vis spectra that were dominated by the nanotube transitions.34 Our calculations of aniline-SWCNT complexes also indicate such a domination of their UV/vis spectra by the excitation on SWCNT intramolecular orbitals. Concerning the effect of nanotube diameter, experiments clearly show a decrease of the longest wavelength transition (bathochromic shift) with increasing diameter,43 whereas for the electronic transition in the UV region, a blue shift with increasing diameter is characteristic.44 The effect of the fragment size, C42H16 (30,0) SWCNT fragment versus C64H20 (30,0) SWCNT fragment, on calculated (BP86D) excitation energies is provided in Table S4 in the Supporting Information. Increase of the fragment size leads to a substantial bathochromic shift of the first calculated absorption band, 630 f 1009 nm for the fragment and 883 f 1207 nm for the corresponding aniline complex (Table S4 in the Supporting Information). Similar to the results for the smaller C42H16 SWCNT fragment, the two strong transitions are only a little influenced by interaction with aniline, λ ) 1009 f 1005 nm and λ ) 574 f 581 nm (Table S4 in the Supporting Information). Hence, we expect that, by this model system, the general features and influence of interaction with the aniline donor as well as explicit methanol solvent molecules are reasonably well described. To model possible specific solvent effects, one and two explicit methanol molecules were added to the aniline-(30,0) SWCNT fragment complex. As pointed out above, the larger C64H20 (30,0) SWCNT fragment was used to accommodate the adsorbed aniline-methanol cluster. For these, only BP86/SVP TD calculations were feasible because of computational restrictions. Calculated absorption wavelengths, oscillator strengths, and pertinent CI vectors for the 10 lowest electronic transitions are summarized in Table 3. Here, only results for the lower energy arrangements of the aniline + 1 MeOH and aniline + 2 MeOH clusters are given. The first strong transition involves excitations between orbitals almost exclusively localized on the SWCNT fragment (HOMO in the fragment and the aniline complex, HOMO-1 in the methanol clusters, and LUMO in all structures). Because the general form and energy of these orbitals are only (see above) influenced a little by the presence or absence of aniline and/or

Noncovalent Interaction between Aniline and CNTs

J. Phys. Chem. C, Vol. 114, No. 13, 2010 5903

TABLE 3: Effect of Explicit Solvent Molecules (MeOH) on Calculated BP86-D/SVP Electronic Transition Energies (Wavelength, λ/nm) for the C64H20-(30,0) Fragment + Aniline (Top-Down)a aniline-(30,0) SWCNT + 1 MeOH

aniline-(30,0) SWCNT f |CI vector|2*100

λ 1207 1005

984

815

795 720 627 622 592

584 a

0.0079 HOMO-1 f LUMO (88.1) HOMO f LUMO (11.4) 0.2375 HOMO f LUMO (73.6) HOMO-2 f LUMO+1 (13.4) HOMO-1 f LUMO (11.3) 0.0002 HOMO f LUMO+1 (56.1) HOMO-2 f LUMO (43.8) 0.0001

0.0001 0.0010 0.0028 0.0012 0.0901 HOMO-4 f LUMO (68.9) HOMO f LUMO+3 (12.5) HOMO-2 f LUMO+1 (11.5) 0.0045

f |CI vector|2*100

λ

aniline-(30,0) SWCNT+ 2 MeOH f |CI vector|2*100

λ

1606

0.0112 HOMO f LUMO (89.7)

2212

0.0059 HOMO f LUMO (99.5)

1064

0.0015 HOMO f LUMO+1 (96.4)

1295

0.0001 HOMO f LUMO+1 (99.7)

1025

1024

805 774 714 648 618

0.1837 HOMO-1 HOMO-1 HOMO-1 0.0512 HOMO-1 HOMO-2 HOMO-1 0.0001 0.0016 0.0017 0.0078 0.0020

893 804 728 719 679

0.1982 HOMO-1 f LUMO (75.0) HOMO-2 f LUMO+1 (11.0) HOMO f LUMO+1 (7.0) 0.0353 HOMO-1 f LUMO+1 (51.4) HOMO-2 f LUMO (36.7) HOMO-1 f LUMO (9.7) 0.0009 0.0000 0.0061 0.0019 0.0085

616

0.0022

642

0.0091

960

f LUMO (64.7) f LUMO+1 (10.1) f LUMO+1 (7.4) 975 f LUMO+1 (49.1) f LUMO (33.1) f LUMO (12.6)

Numbers in bold indicate excitations from the aniline moiety.

Figure 4. Calculated (BP86-D/SVP) electronic transitions (eV) for the C64H20 (30,0) SWCNT, (30,0) SWCNT + aniline, (30,0) SWCNT + aniline + one MeOH, and (30,0) SWCNT + aniline + two MeOH. Transitions with f < 0.025 are indicated by a dot beneath the abscissa.

methanol, the corresponding electronic transitions are quite constant (Table 3 and Figure 4). An analogous argument also holds for the higher-energy strong transitions (Figure 4). Interaction of the nanotube with aniline leads to the appearance of several additional, albeit quite weak (f < 0.01), transitions. Specifically, the longest wavelength absorption results from an aniline f SWCNT transition (HOMO-1 in the aniline-(30,0) SWCNT, HOMO in the methanol clusters; see Figure 2). In

contrast to the more intense excitations, these are very sensitive to the presence of an additional MeOH species; λ ) 1207, 1606, and 2212 nm in the aniline-(30,0) SWCNT, aniline-(30,0) SWCNT + 1 MeOH, and aniline-(30,0) SWCNT + 2 MeOH, respectively. The charge distribution obtained by the natural population analysis (NPA)45 distribution on the aniline molecule as well as in the central bond of the SWCNT below the aniline changes only slightly when adding two MeOH molecules near

5904

J. Phys. Chem. C, Vol. 114, No. 13, 2010

the aniline amino group. The negatively charged aniline nitrogen atom (qN ) -0.80 e) becomes more positive by only 0.03 e, whereas the amino hydrogen atoms get slightly more negative, decreasing the group charge by -0.017 e. The adjacent aniline carbon atom shows a small change in the charge (∆q ) -0.009 e). This only small change and the slight positively charged nitrogen atom are also reflected in the MO picture (Figure 2) and lead to a similar overlap of the aniline with the π orbitals on the SWCNT with a concomitant shift of the orbitals. The energy of the aniline orbital (HOMO-1 ) -4.57 eV without MeOH, and HOMO ) -4.03 eV with two MeOH) is raised when introducing two methanol molecules, which are connected to the aniline. This then leads to a vertical HOMO-1 f LUMO excitation at lower energies. The change of the charge at the SWCNT carbon atom (q ) 0.01 e) below the amino hydrogens is also not very pronounced, becoming more negative by 0.02 e. The influence on the central SWCNT carbon atoms, which lie below the aniline phenyl ring, shows almost no changes in the charge (∆q ) -0.006 e for the carbon atom located below the top-oriented aniline carbon and -0.001 e for the adjacent one) when introducing the methanol. The methanol molecules have negatively charged oxygen atoms (qO ) -0.75 and -0.74 e in the lower-energy unsymmetrical structure) and positive alcoholic hydrogen atoms (qH ) 0.488 and 0.478 e). Depending on the distance to the SWCNT surface (2.102 and 2.312 Å, respectively), the MeOH molecules induce the biggest increase of negative charge (∆q ) 0.059 and 0.24 e) on the closest SWCNT carbon atoms. The highest occupied SWCNT orbital (HOMO ) -4.28 eV without MeOH, and HOMO-1 ) -4.33 eV with two MeOH) and the lowest unoccupied SWCNT orbitals (LUMO ) -3.54 eV without MeOH and -3.56 eV with two MeOH) only slightly lower their energy. Almost no change in energy of these orbitals compared with the pure fragment (HOMO ) -4.27 eV, LUMO ) -3.52) results when the aniline is introduced. Therefore, no large change in the spectrum is expected, according to the first excitation after introduction of aniline and MeOH to an SWCNT (see Figure 4). This is also reflected in the very small changes in the charge distribution on the SWCNT fragment atoms. Conclusions Three density functionals, the GGA PW91LYP, the hybrid m-GGA MPWB1K, and the dispersion-corrected BP86-D, have been used to calculate the adsorption of the aromatic packer molecule, aniline, onto single-walled carbon nanotubes of different diameters. The SWCNTs were represented by C42H16 fragments derived from the optimized structure of full zigzag (n,0) nanotubes (n ) 8, 10, 30) with a molecular formula of C128H16. All functionals result in the top-down orientation of the aniline molecule as the lowest energy complex. The BSSEcorrected adsorption energies for this configuration are Eads ) 9.7 kcal mol-1 (BP86-D/SVP), Eads ) 0.5 kcal mol-1 [PW91LYP/ 6-311+G(d)], and Eads ) 1.4 kcal mol-1 [MPWB1K/6311+G(d)]. T-shaped structures with the amino group oriented toward the nanotube are higher in energy than π-stacked arrangements. On the basis of a comparison of the calculated adsorption energy of benzene onto graphene with the corresponding experimental value, the BP86-D results are considered to be the most reliable ones. TDDFT calculations on pyrene, coronene, and ovalene with the BP86-D/SVP and MPWB1K/ 6-311+G(d) procedures indicate an under- and overestimation by approximately the same amount of the electronic transition energies, respectively, by these two methods. BP86/SVP TD-

Peles-Lemli et al. DFT calculations predict two quite strong bands in the visible spectra at the long-wavelength region (>500 nm). Both are quite insensitive to the presence of one aniline molecule and its orientation (top-up, top-down, hollow-down), indicating that these electronic transitions mainly result from nanotube excitations. An aniline f single-walled carbon nanotube excitation at ca. 1200 nm is very weak (f ) 0.008) and, consequently, not likely to be discernible in the UV/vis spectra. Inclusion of explicit MeOH molecules to model solvent effects has only little influence on the strong electronic transitions resulting from intratube excitations. In contrast, the weak longest wavelength transitions resulting from an aniline f SWCNT excitation are very sensitive to the presence of MeOH, leading to a pronounced bathochromic shift. Although several of the vertical transition energies are significantly influenced by the solvent MeOH, the overall appearance of the calculated UV/vis spectra barely changes. This finding is completely in line with the experimentally observed insensitivity of the absorption spectra of SWCNT-aniline toward the MeOH solvent. The experimental absorption peaks at lower energies can be assigned to excitations from SWCNT π orbitals located along the long axis of the SWCNT and of aniline orbitals into the lower-lying unoccupied orbitals oriented along the diameter axis. Acknowledgment. The financial support of the Austrian¨ sterreich-Ungarn Hungarian Action Foundation (AKTION O Wissenschafts- und Erziehungskooperation, (72o¨u1)) and the Austrian-Hungarian Intergovernmental S&T Programme (AT23/08) is highly appreciated. The MPWB1K calculations were performed using the Gaussian 03, revision B.05, program package on the SunFire 15000 supercomputer located in the Supercomputer Center of the Hungarian National Infrastructure Development Program Office. We thank Prof. Elliot R. Bernstein, Department of Chemistry, Colorado State University, for the coordinates of the aniline-methanol cluster. Supporting Information Available: Tables for curvature dependence of calculated adsorption energies (Table S1), calculated excitation energies in the gas phase (Tables S2-S4), and drawings of BP86-D optimized structures for aniline-(30,0) SWCNT + nMeOH (n ) 1, 2; Figures S1 and S2). This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) (a) Tournus, F.; Latil, S.; Heggie, M. I.; Charlier, J.-C. Phys. ReV. B 2005, 72, 075431. (b) Tournus, F.; Charlier, J.-C. Phys. ReV. B 2005, 71, 165421. (2) Efremenko, I.; Sheintuch, M. Langmuir 2006, 22, 3614–3621. (3) Go´mez, B.; Martı´nez-Magada´n, J. M. J. Phys. Chem. B 2005, 109, 14868–14875. (4) (a) Wang, Y.; Bu, Y. J. Phys. Chem. B 2007, 111, 6520–6526. (b) Wang, Y. J. Phys. Chem. C 2008, 112, 14297–14305. (5) (a) Stepanian, S. G.; Karachevtsev, M. V.; Glamazda, A. Y.; Karachevtsev, V. A.; Adamowicz, L. Chem. Phys. Lett. 2008, 459, 153– 158. (b) Stepanian, S. G.; Karachevtsev, M. V.; Glamazda, A. Y.; Karachevtsev, V. A.; Adamowicz, L. J. Phys. Chem. A 2009, 113, 3621– 3629. (6) Nakashima, N. Sci. Technol. AdV. Mater. 2006, 7, 609–616. (7) Cai, Y.; Cai, Y.; Mou, S.; Lu, Y. J. Chromatogr., A 2005, 1081, 245–247. (8) Tang, X.; Bansaruntip, S.; Nakayama, N.; Yenilmez, E.; Chang, Y.-I.; Wang, Q. Nano Lett. 2006, 6, 1632–1636. (9) Zhang, J.; Wang, G.; Shon, Y.-S.; Zhou, O.; Superfine, R.; Murray, R. W. J. Phys. Chem. B 2003, 107, 3726–3732. (10) Wang, Z.; Yuan, J.; Li, M.; Han, D.; Zhang, Y.; Shen, Y.; Nin, L.; Ivaska, A. J. Electroanal. Chem. 2007, 599, 121–126. (11) Guo, L.; Peng, Z. Langmuir 2008, 24, 8971–8975. (12) (a) Sun, Y.; Wilson, S. R.; Schuster, D. I. J. Am. Chem. Soc. 2001, 123, 5348–5349. (b) Perepichka, D. F.; Wudl, F.; Wilson, S. R.; Sun, Y.; Schuster, D. I. J. Mater. Chem. 2004, 14, 2749–2752.

Noncovalent Interaction between Aniline and CNTs (13) (a) Irving, D. L.; Sinnott, S. B.; Lindner, A. S. Chem. Phys. Lett. 2004, 389, 96–100. Corrigendum to: (b) Irving, D. L.; Sinnott, S. B.; Lindner, A. S. Chem. Phys. Lett. 2004, 392, 567. (14) Woods, L. M.; Badescu, S. C.; Reinecke, T. L. Phys. ReV. B 2007, 75, 155415. ´ cs, P.; Kolla´r, L.; Kunsa´gi-Ma´te´, S. Fullerenes, (15) Peles-Lemli, B.; A Nanotubes, Carbon Nanostruct. 2008, 16, 247–257. (16) Hasegawa, M.; Nishidate, K. Phys. ReV. B 2004, 70, 205431. (17) (a) Zhao, Y.; Truhlar, D. G. Phys. Chem. Chem. Phys. 2005, 7, 2701–2705. (b) Zhao, Y.; Truhlar, D. G. J. Phys. Chem. A 2005, 109, 4209– 4212. (18) (a) Antony, J.; Grimme, S. Phys. Chem. Chem. Phys. 2006, 8, 5287– 5293. (b) Grimme, S.; Antony, J.; Schwabe, T.; Mu¨ck-Lichtenfeld, C. Org. Biomol. Chem. 2007, 5, 741–758. (19) Pankewitz, T.; Klopper, W. J. Phys. Chem. C 2007, 111, 18917– 18926. (20) (a) Burke, K.; Perdew, J. P.; Wang, Y. Electronic Density Functional Theory: Recent Progress and New Directions; Plenum: New York, 1998. (b) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785–789. (21) (a) Grimme, S. J. Comput. Chem. 2006, 27, 1787–1799. (b) Becke, A. D. Phys. ReV. A 1988, 38, 3098–3100. (c) Perdew, J. P. Phys. ReV. B 1986, 33, 8822–8824. (22) Kar, T.; Scheiner, S.; Roy, A. K. Chem. Phys. Lett. 2008, 460, 225–229. (23) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72, 650–654. (24) Binkley, J. S.; Pople, J. A.; Hehre, W. J. J. Am. Chem. Soc. 1980, 102, 939–947. (25) Schaefer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571– 2577. (26) Schaefer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829–5835. (27) D’Amore, M.; Improta, R.; Barone, V. J. Phys. Chem. A 2003, 107, 6264–6269. (28) (a) Bauernschmitt, R.; Ahlrichs, R. Chem. Phys. Lett. 1996, 256, 454–464. (b) Furche, F.; Ahlrichs, R. J. Chem. Phys. 2004, 121, 12772– 12773. (29) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.;

J. Phys. Chem. C, Vol. 114, No. 13, 2010 5905 Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision B.05; Gaussian, Inc.: Wallingford, CT, 2004. (30) TURBOMOLE V6.0 2009 is a development of the University of Karlsruhe and Forschungszentrum Karlsruhe GmbH, 1989-2007, and TURBOMOLE GmbH since 2007; it is available from http://www.turbomole.com. (31) Schaftenaar, G.; Noordik, J. H. J. Comput.-Aided Mol. Des. 2000, 14, 123–134. (32) HyperChem Professional 7; HyperCube, Inc.: Gainesville, FL, 2002. (33) (a) Laaksonen, L. J. Mol. Graphics 1992, 10, 33-34. (b) Bergman, D. L.; Laaksonen, L.; Laaksonen, A. J. Mol. Graphics 1997, 15, 301-306. (34) Fan, W.; Zhang, R. Sci. China, Ser. B: Chem. 2008, 51, 1203– 1210. (35) Wang, Z.; Irle, S.; Zheng, G.; Morokuma, K. J. Phys. Chem. C 2008, 112, 12697–12705. (36) Kazachkin, D.; Nishimura, Y.; Irle, S.; Morokuma, K.; Vidic, R. D.; Borguet, E. Langmuir 2008, 24, 7848–7856. (37) Zacharia, R.; Ulbricht, H.; Hertel, T. Phys. ReV. B 2004, 69, 155406. (38) Peles-Lemli, B.; Kollar, L.; Kunsagi-Mate, S. Fullerenes, Nanotubes, Carbon Nanostruct., in press. (39) Hu, Y.; Bernstein, E. R. J. Phys. Chem. A 2009, 113, 639–643. (40) Hammonds, M.; Pathak, A.; Sarre, P. J. Phys. Chem. Chem. Phys. 2009, 11, 4458–4464. (41) Frank, I.; Damianos, K. Chem. Phys. 2008, 343, 347–352. (42) Voggu, R.; Rout, C. S.; Franklin, A. D.; Fisher, T. S.; Rao, C. N. R. J. Phys. Chem. C 2008, 112, 13053–13056. (43) Kataura, H.; Kumazawa, Y.; Maniwa, Y.; Umezu, I.; Suzuki, S.; Ohtsuka, Y.; Achiba, Y. Synth. Met. 1999, 103, 2555–2558. (44) Takagi, Y.; Okada, S. Phys. ReV. B 2009, 79, 233406. (45) Reed, A. E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys. 1985, 83, 735–746.

JP908505Q