Self-Assembled Carbon Nanotubes on Gold: Polarization-Modulated

Feb 19, 2008 - Kevin R. J. Lovelock , Ignacio J. Villar-Garcia , Florian Maier , Hans-Peter Steinrück , and Peter Licence. Chemical Reviews 2010 110 ...
0 downloads 0 Views 176KB Size
Download PDF
Langmuir 2008, 24, 3235-3243

3235

Self-Assembled Carbon Nanotubes on Gold: Polarization-Modulated Infrared Reflection-Absorption Spectroscopy, High-Resolution X-ray Photoemission Spectroscopy, and Near-Edge X-ray Absorption Fine Structure Spectroscopy Study Natalia Kocharova,*,† Jarkko Leiro,‡ Jukka Lukkari,† Markku Heinonen,‡ Toma´sˇ Ska´la,§ Frantisˇek Sˇ utara,| Michal Skoda,| and Martin Vondra´cˇek⊥ The Laboratory of Materials Chemistry, Department of Chemistry, and The Laboratory of Materials Science, Department of Physics, UniVersity of Turku, FIN-20014 Turku, Finland, Sincrotrone Trieste SCpA, BasoVizza, Trieste, Italy, and Department of Surface and Plasma Science, Charles UniVersity, Prague, Czech Republic, and Institute of Physics, Academy of Sciences of the Czech Republic, Prague, Czech Republic ReceiVed October 10, 2007. In Final Form: December 25, 2007 Recently we reported noncovalent functionalization of nanotubes in an aqueous medium with ionic liquid-based surfactants, 1-dodecyl-3-methylimidazolium bromide (1) and 1-(12-mercaptododecyl)-3-methylimidazolium bromide (2), resulting in positively charged single-wall carbon nanotube (SWNT)-1,2 composites. Thiolation of SWNTs with 2 provides their self-assembly on gold as well as templating gold nanoparticles on SWNT sidewalls via a covalent -S-Au bond. In this investigation, we studied the electronic structure, intermolecular interactions, and packing within noncovalently thiolated SWNTs and also nanotube alignment in the bulk of SWNT-2 dried droplets and selfassembled submonolayers (SAMs) on gold by high-resolution X-ray photoemission spectroscopy (HRXPS), C K-edge X-ray absorption fine structure (NEXAFS) spectroscopy, and polarization-modulated infrared reflection-absorption spectroscopy (PM-IRRAS). HRXPS data confirmed the noncovalent nature of interactions within the nanocomposite of thiolated nanotubes. In PM-IRRAS spectra of SWNT SAMs on gold, the IR-active vibrational SWNT modes have been observed and identified. According to PM-IRRAS data, the hydrocarbon chains of 2 are oriented with less tilt angle to the bare gold normal in a SAM deposited from an SWNT-2 dispersion than those of 1 deposited from an SWNT-1 dispersion on the mercaptoethanesulfonic acid-primed gold. For both the dried SWNT-2 bulk and the SWNT-2 SAM on gold, the C K-edge NEXAFS spectra revealed the presence of CH-π interactions between hydrocarbon chains of 2 and the π electronic nanotube structure due to the highly resolved vibronic fine structure of carbon 1s f R*/σ*C-H series of states in the alkyl chain of 2. For the SWNT-2 bulk, the observed splitting and upshift of the SWNT π* orbitals in the NEXAFS spectrum indicated the presence of π-π interactions. In the NEXAFS spectrum of the SWNT-2 SAM on gold, the upshifted values of the photon energy for R*/σ*C-H transitions indicated close contact of 2 with nanotubes and with a gold surface. The angle-dependent NEXAFS for the SWNT-2 bulk showed that most of the molecules of 2 are aligned along the nanotubes, which are self-organized with orientation parallel to the substrate plane, whereas the NEXAFS for the SWNT-2 SAM revealed a more normal orientation of functionality 2 on gold compared with that in the SWNT-2 bulk.

Introduction Many applications of single-wall carbon nanotubes (SWNTs)1 dictate their integration into highly ordered macroscopic structures in a controlled way.2 However, controlled deposition of highly organized SWNTs onto solid substrates still remains a challenge. The stepwise fabrication of the SWNT multilayer structures using the electrostatic layer-by-layer (LbL) self-assembly technique requires their conversion into polyionic species, i.e., nanotube positive and negative charging. In addition, self-assembly of carbon nanotubes on gold or assembly of the transition-metal nanoparticles on nanotubes as templates requires thiolation of the nanotube sidewalls, i.e., SWNT functionalization with * To whom correspondence should be addressed. [email protected]. † Department of Chemistry, University of Turku. ‡ Department of Physics, University of Turku. § Sincrotrone Trieste SCpA. | Charles University. ⊥ Academy of Sciences of the Czech Republic.

E-mail:

(1) (a) Avouris, P.; Chen, Z.; Perebeinos, V. Nat. Nanotechnol. 2007, 2, 605. (b) Roth, S.; Carroll, D. One-Dimensional Metals; Wiley-VCH Verlag GmbH & Co.: Weinheim, Germany, 2004. (c) Reich, S.; Thomsen, C.; Maultzsch, J. Carbon Nanotubes; Wiley-VCH Verlag GmbH & Co.: Weinheim, Germany, 2004. (2) Nanoscale Science and Technology; Kelsall, R., Hamley, I., Geoghegan, M., Eds.; John Wiley & Sons, Ltd.: New York, 2005.

molecules containing a thiol group.3 Recently, taking these requirements into account, we demonstrated a novel simple noncovalent approach for preparation of positively charged carbon nanotubes in an aqueous medium from pristine nanotube material.4 For this purpose we used environmentally friendly, nonvolatile, and thermally stable ionic liquids (ILs), 1-dodecyl-3-methylimidazolium bromide (1) and 1-(12-mercaptododecyl)-3methylimidazolium bromide (2) (Chart 1). We have shown that long-chain 1-alkyl-3-methylimidazolium cations can be considered as multifunctional IL surfactants, since they not only impart water solubility to SWNTs but also provide a high degree of nanotube purification from carbonaceous particles, impart positive charge to SWNTs, and also, in the case of 2, provide direct self-assembly of SWNTs onto the gold surface. The noncovalent approaches of SWNT functionalization are based on adsorption of the hydrophobic part of the functional molecule on nanotube sidewalls through weak intermolecular van der Waals, electrostatic, π-π, CH-π, and other interactions, with aqueous solubility provided by the hydrophilic part of the (3) Isaacs, L.; Chin, D. N.; Bowden, N.; Xia, Y., Whitesides, G. M. In Supramolecular Materials and Technologies; Reinhoudt, D. N., Ed.; Wiley & Sons: New York, 1999; p 14. (4) Kocharova, N.; A ¨ a¨ritalo, T., Leiro, J.; Kankare, J.; Lukkari, J. Langmuir 2007, 23, 3363.

10.1021/la7030768 CCC: $40.75 © 2008 American Chemical Society Published on Web 02/19/2008

3236 Langmuir, Vol. 24, No. 7, 2008 Chart 1. Molecular Structures of the Ionic Liquid Surfactants 1-Dodecyl-3-methylimidazolium Bromide (1) and 1-(12-Mercaptododecyl)-3-methylimidazolium Bromide (2) Used in This Work

molecule. The π-π interactions have been utilized for noncovalent SWNT functionalization with electron-rich molecules.5 The adsorption of small molecules, including surfactants, on the graphene SWNT sidewalls can involve, besides electrostatic, van der Waals, and π-π interactions, other forces.6 In addition, it has been revealed that there are multiple weak molecular CH-π interactions in the carbon nanotube composites.7 The strength of the CH-π interactions is only 1/10 of that of the hydrogen bond; however, their cooperative multiple number significantly influences many chemical and biochemical phenomena.8 Although a proper characterization of SWNT composites and their films on molecular and electronic levels is highly important for modern nanotechnology, much work still remains to structurally characterize in detail the interactions, packing, and alignment within these films. Polarization-Modulated Infrared Reflection-Absorption Spectroscopy (PM-IRRAS) for Study of SWNT Films. Vibrational spectroscopy is an important analysis methodology used to study thin organic films and, in particular, nanoscaled self-assembled mono- and submonolayers. For metal substratesupported organic thin films, the most widely used Raman and IR techniques are surface-enhanced (resonance) Raman scattering (SERRS)9 and PM-IRRAS,10 respectively. Both are powerful tools for elucidation of the surface bonding and orientation of adsorbates on metal substrates.11 Because of its high sensitivity even to weakly adsorbed species or single molecules on metallic structures in the nanometer size range and to the electronic band structure of individual (n, m) nanotubes, SERRS is nowadays (5) (a) Chen, R. J.; Zhang, Y.; Wang, D.; Dai, H. J. Am. Chem. Soc. 2001, 123, 3838-3839. (b) Zhang, J.; Lee, J.-K.; Wu, Y.; Murray, R. W. Nano Lett. 2003, 3, 403-407. (c) Zhao, J.; Lu, J. P.; Han, J.; Yang, C.-K. Appl. Phys. Lett. 2003, 82, 3746. (6) (a) Tao, F.; Bernasek, S. L. Chem. ReV. 2007, 107, 1408. (b) Sumanasekera, G. U.; Pradhan, B. K.; Romero, H. E.; Adu, K. W.; Eklund, P. C. Phys. ReV. Lett. 2002, 89, 166801. (c) Islam, M. F.; Rojas, E.; Bergey, D. M.; Johnson, A. T.; Yodh, A. G. Nano Lett. 2003, 3, 269. (d) Moore, V. C.; Strano, M. S.; Haroz, E. H.; Hauge, R. H.; Smalley, R. E. Nano Lett. 2003, 3, 1379. (7) Baskaran, D.; Mays, J. W.; Bratcher, M. S. Chem. Mater. 2005, 17, 33893397. (8) (a) Nishio, M.; Hirota, M.; Umezawa, Y. CH/pi Interaction, EVidence, Nature and Consequences; Wiley-VCH: New York, 1998. (b) Kim, K. S.; Tarakeshwar, P.; Lee, J. Y. Chem. ReV. 2000, 100, 4145. (9) (a) Moskovits, M. J. Raman Spectrosc. 2005, 36, 485. (b) Kocharova, N.; Lukkari, J.; Viinikanoja, A.; A ¨ a¨ritalo, T.; Kankare, J. J. Phys. Chem. B 2002, 106, 10973. (c) Kocharova, N.; Lukkari, J.; Viinikanoja, A.; A ¨ a¨ritalo, T.; Kankare, J. J. Mol. Struct. 2003, 651, 75. (10) (a) Greenler, R. G. J. Chem. Phys. 1966, 44, 310. (b) Green, M. J.; Barner, B. J.; Corn, R. M. ReV. Sci. Instrum. 1991, 62, 1426. (c) Frey, B. L.; Corn, R. M., Weibel, S. In Handbook of Vibrational Spectroscopy; Chalmers, J., Griffiths, R., Eds.; John Wiley & Sons: New York, 2001; Vol. 2, p 1042. (d) Viinikanoja, A.; Areva, S.; Kocharova, N.; A ¨ a¨ritalo, T.; Vuorinen, M.; Savunen, A.; Kankare, J.; Lukkari, J. Langmuir 2006, 22, 6078. (11) Sue¨taka, W.; Yates, J. T., Jr. Surface Infrared and Raman Spectroscopy; Plenum Press: New York, 1995.

KocharoVa et al.

one of the most widely used analytical tools in SWNT science.4,12 We used SERRS in our previous work for demonstration of direct self-assembly of thiolated SWNTs on a gold substrate from their aqueous dispersion.4 In contrast to resonance Raman (RR) and SERRS studies, the general number of IR and, specifically, of PM-IRRAS studies on SWNTs is very much limited, the main obstacles being the poor sample quality of the nanotube materials produced and the difficulty to detect IR-active SWNT modes. The latter stems from very low intensities of the carbon nanotube IR modes, which is a consequence of the fact that SWNTs do not support a static dipole moment but generate a dynamic dipole moment, which is much weaker. The second reason for the limited number of carbon nanotube IR studies is the absence of reliable assignments of IR-active vibrational modes. However, recent work on vibrational assignment of 18 IR bands of SWNTs opens up further possibilities to study carbon nanotubes by IR methods.13 The experimental difficulties of depositing individual tubes on a metal surface and detecting them by IR means is the third reason for the almost complete absence of PM-IRRAS studies on SWNTs in the literature. Besides an earlier report of phonon nanotube modes observed by reflectance from unpurified powder containing SWNTs,14 there are only three works to date where PM-IRRAS is demonstrated to be a viable technique for spectroscopic study of SWNTs deposited on aluminum15 and acid-treated SWNTs on platinum surfaces.16 IRRAS was used for characterization of DNA-wrapped SWNTs on gold.17 The combination of the high electric field intensities in p-polarization and at a high angle of incidence (∼80°) makes IRRAS sensitive to layers of adsorbates ∼100 nm thick or less on a metallic substrate (so-called “metalsurface selection rule”). However, for ultrathin monolayer films, with thickness less than 50 nm, IRRAS is inefficient because of experimental drifts and instabilities between recordings of reflectance of the film on a metal substrate and the reference reflectance of the bare substrate. In contrast to IRRAS, PMIRRAS takes advantage of the polarization modulation, which allows eliminating these drawbacks inevitable in IRRAS and accounting for strong gas-phase absorption, thus significantly increasing sensitivity, which, in turn, facilitates the investigation of extremely thin films. Application of the metal-surface selection rule for self-assembled organic monolayers makes PM-IRRAS very sensitive for studying molecular orientation with respect to the surface normal and, thus, can be very informative for studies of self-assembled functionalized SWNTs on gold. However, to date, such studies have not been reported. Near-Edge X-ray Absorption Fine Structure (NEXAFS) for Study of SWNT Films. NEXAFS spectroscopy provides information on unoccupied electronic states of specific elements in organic substances in various environments by measuring (12) (a) Kneipp, K., Kneipp, H., Dresselhaus, M.; Lefrant, S. Philos. Trans. R. Soc. London, Ser. A 2004, 362, 2361. (b) Lefrant, S.; Baltog, I.; Balbarac, M. J. Raman Spectrosc. 2005, 36, 676 and references therein. (c) Baibarac, M.; Baltog, I.; Lefrant, S.; Gomez-Romero, P. Polymer 2007, 48, 5279. (d) Baibarac, M.; Baltog, I.; Frunza, S.; Lefrant, S.; Mevellec, J. Y.; Godon, C. J. Optoelectron. AdV. Mater. 2007, 9, 1422. (13) Kim, U. J.; Liu, X. M.; Funtado, C. A.; Chen, G.; Saito, R.; Jiang, J.; Dresselhaus, M. S.; Eklund, P. C. Phys. ReV. Lett. 2005, 95, 157402. (14) Kuhlmann, U.; Jantoljak, H.; Pfa¨nder, N.; Bernier, P.; Journet, C.; Thomsen, C. Chem. Phys. Lett. 1998, 294, 237. (15) Bermudez, V. M. J. Phys. Chem. B 2005, 109, 9970. (16) Rosario-Castro, B. I.; Contes, E. J.; Perez-Davis, M. E.; Cabrera, C. R. ReV. AdV. Mater. Sci. 2005, 10, 381. (17) Zangmeister, R. A.; Maslar, J. E.; Opdahl, A.; Tarlov, M. J. Langmuir 2007, 23, 6252.

Self-Assembled Carbon Nanotubes on Gold

absorption close to a core level ionization threshold.18 The peak positions and spectral line shape of the absorption fine structure in an NEXAFS spectrum are directly related to the nature of these electronic states; thus, the NEXAFS is extremely sensitive to the bonding environment of the specific absorbing atom. The C K-edge NEXAFS spectrum is often regarded as a “fingerprint” of carbon chemical bonds and the local environment. In many ways it is superior to X-ray photoemission spectroscopy (XPS), which does not provide local structural information. In carbon nanotube chemistry, the NEXAFS has not been widely used, and to date, very few NEXAFS studies on SWNTs exist.18b-24 The first report on the C K-edge NEXAFS of SWNTs19 was attributed to oxidation effects upon purification and heating of SWNTs. The NEXAFS has also been used to directly probe the electronic band structure of SWNTs20,22 and multiwalled carbon nanotubes22,21 and to study the nanotube structure and alignment in SWNT buckypaper.21,23 Angular dependence on the angle made by the π* and/or σ* orbitals with respect to the electric field vector E h of the incident polarized X-rays makes NEXAFS spectroscopy invaluable for orientation determination of functional groups on the surface. The angle-dependent C K-edge NEXAFS has been used to study helicity and defects in carbon nanotubes,20 for differentiation between local electronic structures of the tips and sidewalls of aligned SWNTs,24 and also for determination of the order and alignment in SWNT films and aligned MWNTs.23a However, there are no NEXAFS studies reported to date on functionalized SWNTs and on self-assembled SWNT submonolayers. In this work, we focus on characterization of noncovalently thiolated SWNTs and their self-assembled submonolayers (SAMs) on gold by spectroscopic methods rarely employed for these purposes, high-resolution XPS (HRXPS), PM-IRRAS, and NEXAFS/X-ray absorption near edge structure (XANES) spectroscopy. We demonstrate the potential of the combined use of these complementary and powerful spectroscopic methods to study the structure, interactions, order (packing), and alignment of noncovalently functionalized SWNTs within drop-cast films and SAMs on gold. To the best of our knowledge, this is the first report ever where noncovalently functionalized SWNTs and their SAMs have been studied by NEXAFS and PM-IRRAS techniques. Experimental Section Materials. Raw HiPco SWNTs were purchased from Carbon Nanotechnologies, Inc. and were used as received. Synthesis of 1 is described elsewhere.4 2 was synthesized according to Lee et al.25 (18) (a) Sto¨hr, J. NEXAFS Spectroscopy; Springer Series in Surface Sciences, Vol. 25; Springer: Heidelberg, Germany, 1992. (b) Hemraj-Benny, T.; Banerjee, S.; Sambasivan, S.; Balasubramanian, M.; Fischer, D. A.; Eres, G.; Puretzky, A. A; Geohegan, D. B.; Lowndes, D. H.; Han, W.; Misewich, J. A.; Wong, S. S. Small 2006, 2, 26. (19) Kuznetsova, A.; Popova, I.; Yates, J. T.; Bronikowski, M. J.; Hyffman, C. B.; Smalley, R. E.; Hwu, H. H.; Chen, J. G. J. Am. Chem. Soc. 2001, 123, 10699. (20) Tang, Y. H.; Sham, T. K.; Hu, Y. F.; Lee, C. S.; Lee, S. T. Chem. Phys. Lett. 2002, 366, 636. (21) Banerjee, S.; Hemraj-Benny, T.; Balasubramanian, M.; Fischer, D. A.; Misewich, J. A.; Wong, S. S. ChemPhysChem 2004, 5, 1416. (22) Schiessling, J.; Kjeldgaard, L.; Rohmund, F.; Falk, L. K. L.; Campbell, E. E. B.; Nordgren, J.; Bru¨hwiller, P. A. J. Phys.: Condens. Matter 2003, 15, 6563. (23) (a) Banerjee, S.; Hemraj-Benny, T.; Sambasivan, S.; Fischer, D. A.; Misewich, J. A.; Wong, S. S. J. Phys. Chem. B 2005, 109, 8489. (b) DettlaffWegikowska, U.; Skakalova, V.; Graupner, R.; Jhang, S. H.; Kim, B. H.; Lee, H. J.; Ley, L.; Park, Y. W.; Barber, S.; Toma´nek, D.; Roth, S. J. Am. Chem. Soc. 2005, 127, 5125. (24) Chiou, J. W.; Yueh, C. L.; Jan, J. C.; Tsai, H. M.; Pong, W. F.; Hong, I.-H.; Klauser, R.; Tsai, M.-H.; Chang, Y. K.; Chen, Y. Y.; Wei, S. L.; Wen, C. Y.; Chen, L. C.; Chuang, T. J. Appl. Phys. Lett. 2002, 81, 4189. (25) Lee, B. S.; Chi, Y. S.; Lee, J. K.; Choi, I. S.; Song, C. E.; Namgoong, S. K.; Lee, S. J. Am. Chem. Soc. 2004, 126, 480.

Langmuir, Vol. 24, No. 7, 2008 3237 Noncovalent SWNT Thiolation in an Aqueous Medium. The procedure of noncovalent thiolation of pristine HiPCO SWNTs in an aqueous medium is described in the Supporting Information. Sample Preparation. Gold Substrates. For HRXPS, NEXAFS, and PM-IRRAS measurements, gold substrates were prepared analogously to our previous works;4,10d,26 namely, a 100 nm thick gold layer was evaporated on silanized (1%, v/v, (3-aminopropyl)triethoxysilane in dry toluene, 4 min at 60 °C)27 microscopy glass slides using an Edwards E306A coating system. Sample Preparation for the HRXPS, NEXAFS, and PMIRRAS Experiments. For HRXPS and NEXAFS measurements, freshly made gold substrates were directly used for self-assembly of 2 and thiolated nanotubes (SWNT-2) and for making thick SWNT-2 films. For the thick drop-cast film, the drop of SWNT-2 stock solution (∼1 mg/mL SWNT) was placed directly onto the freshly made gold substrate and left to dry overnight. For acquisition of the C 1s core level spectrum of “pure” SWNTs, the sample was prepared in the following way. The purified HiPCO SWNTs dispersed in water were drop-cast onto a gold substrate and left to dry overnight. The HRXPS spectrum was measured. Then this sample was heated in the experimental chamber at 110 °C for 5 min at a base pressure of ∼10-9 Torr to get rid of possible surface contaminations, and the HRXPS spectrum was measured again. The HRXPS spectra of dropcast purified SWNTs before and after heating did not show any significant difference. For making SAMs of 2 and the SWNT-2 composite, gold substrates were immersed overnight in an aqueous solution of 2 and an aqueous dispersion of SWNT-2 (∼0.2 mg/mL SWNT) in tightly closed vessels deaerated with argon gas, followed by rinsing with deionized water and drying under a nitrogen gas stream. For PMIRRAS measurements, SAM films were prepared in the following way. For self-assembly of SWNT-2, a bare gold substrate was directly used. For preparation of the SWNT-1 SAM sample, the gold substrate was initially primed with a negatively charged mercaptoethanesulfonic acid (MESA) monolayer by immersion in an aqueous 1 mM MESA solution for 1 h followed by rinsing with deionized water and drying under a nitrogen gas stream. Then the SWNT-1,2 SAM samples were prepared for PM-IRRAS experiments analogously to those for HRXPS and NEXAFS experiments. Characterization. The UV-vis-NIR spectrum of an aqueous dispersion of thiolated SWNTs was recorded with a Varian Cary SE UV-vis-NIR spectrophotometer using quartz cuvettes with a path length of 1 mm. PM-IRRAS spectra were collected using a Nicolet Nexus 870 FTIR-Raman spectrometer (Madison, WI) equipped with an external tabletop optical mount, a liquid nitrogen-cooled mercury-cadmiumtelluride (MCT-A) detector, a photoelastic modulator (HINDS Instruments PM-90 with a II/ZS50 ZnSe 50 kHz optical head, Hillsboro, OR), and a synchronous sampling demodulator (GWC Instruments, Madison, WI). A Whatman laboratory gas generator (model 75-45) was used to purge the sample compartment with dry, CO2-free air, and all PM-IRRAS measurements were done at a constant humidity of 5%. PM-IRRAS spectra were collected with the PEM set for half-wave retardation at 3300 cm-1 for the CH stretching region and at 1500 cm-1 for the lower frequency region, with the infrared beam incident on the sample surface at an 80° angle to the surface normal. The Rp - Rs and R ) Rp + Rs signals from each polarization were acquired simultaneously with the inhouse Omnic 6.2 software and were used for calculation of differential reflectivity, ∆R/R ) (Rp - Rs)/R. All PM-IRRAS spectra were the result of 2048 scans at 4 cm-1 spectral resolution. To account for optical effects of the bare gold substrate and also infrared vibrations of the MESA SAM, the resulting ∆R/R spectra of both the SWNT2/Au sample and the SWNT-1/MESA/Au sample were normalized to ∆R/R spectra of the bare gold substrate and the MESA-primed (26) Paloniemi, H.; A ¨ a¨ritalo, T.; Liuke, H.; Kocharova, N.; Haapakka, K.; Terzi, F.; Seeber, R.; Lukkari, J. J. Phys. Chem. B 2005, 109, 8634. (27) Siqueira Petri, D. F.; Wertz, G.; Schunk, P.; Schimmel, T. Langmuir 1999, 15, 4520.

3238 Langmuir, Vol. 24, No. 7, 2008 gold substrate, respectively, which were recorded under the same experimental conditions, with final obtainment of the δ(∆R/R) value.10,28 The HRXPS and NEXAFS spectra were taken in the experimental chamber (base pressure ∼10-10 mbar) of the Materials Science beamline at the ELETTRA synchrotron radiation facility (Trieste, Italy). The C 1s core level spectra were measured at a photon energy of 400 eV with an energy resolution of 0.3 eV. After subtraction of Shirley background, the binding energy scale was calibrated with the Au 4f7/2 peak position (84.00 eV) and the C 1s peaks were then fitted using the Origin 6.0 Peak Fitting module (with the LevenbergMarquardt algorithm). The peak shapes were approximated by the Gaussian-Lorentzian sum function. For acquiring NEXAFS spectra, the partial electron yield (PEY) signal was collected using a carbon Auger line. A monochromator with a 600 line/mm grating, providing ∼0.2 eV resolution, was used for all NEXAFS spectra collected. To monitor the packing, surface order, and orientation of 2 molecules and nanotubes in SWNT-2 compounds on gold, the NEXAFS spectrum of the drop-cast SWNT-2 film was collected at different polarization angles by rotating the sample holder with respect to the incident X-ray beam in the plane of incidence. The NEXAFS spectrum of the SWNT-2 SAM was taken at 0° (normal to the substrate plane) polarization of the incident synchrotron radiation beam. For the dried droplet of thiolated SWNTs (SWNT-2) the monochromator energy scale was calibrated using the C K-edge π* transition of pyrolytic graphite located at 285.35 eV.29 To diminish the effect of monochromator absorption features and to probe the bulk of SWNT2, the PEY signals for the dried SWNT-2 droplet on gold were normalized by PEY signals of the SWNT-2 SAM on gold. For the sample of the SWNT SAM, the diode NEXAFS spectrum was used as the background intensity.

Results and Discussion In this work, as in our previous paper,4 major emphasis is given to the SWNT-2 composite because of the presence of a thiol group that can enable the chemisorption of nanotubes to transition-metal substrates or to nanoparticles. Noncovalent attachment of the surfactant 2 molecules to the pristine SWNT is made clearly evident in the high-resolution XPS, PM-IRRAS, and NEXAFS spectra of the resulting SWNT-2 composite and its SAM on gold. High-Resolution XPS Spectra. Previously,4 we analyzed the S 2p core level spectrum of the SWNT-2 SAM, which confirmed covalent Au-S bonding of SWNT-2 to the gold surface. Here, we use synchrotron radiation HRXPS to analyze more precisely the interactions and relative concentration of SWNT carbon and surfactant carbon in the SWNT-2 bulk. For acquiring the reference SWNT C 1s spectrum, we also made a drop-cast film from purified HiPCO nanotubes dispersed in water. Figure 1 displays core level C 1s emission spectra for three different samples. Figure 1a shows the C 1s spectrum for the dried SWNT-2 droplet, Figure 1b is the C 1s region of the dried droplet of purified SWNTs heated at 110 °C for 5 min in high vacuum (∼10-9 Torr), and Figure 1c displays the C 1s core level of the surfactant 2 SAM. To identify the C 1s HRXPS results, a curve-fitting procedure was employed. The sample of the SWNT-2 bulk (Figure 1a) gives a multicomponent C 1s spectrum with four binding energy (BE) components at 284.8, 285.3, 286.8, and 289.4 eV. The 284.8 eV value is the typical BE for the C 1s emission of graphitelike carbon atoms with sp2 hybridization characteristic of HOPG and SWNTs.4,30 In contrast to organic compounds, even smaller (28) (a) Buffeteau, T.; Desbat, B.; Turlet, J. M. Appl. Spectrosc. 1991, 45, 380. (b) Zawiza, I.; Bin, X.; Lipkowski, J. Langmuir 2007, 23, 5180. (29) Batson, P. E. Phys. ReV. B 1993, 48, 2608. (30) (a) Yudasaka, M.; Kilouchi, R.; Ohki, Y.; Yoshimura, S. Carbon 1997, 35, 195. (b) Banerjee, S.; Wong, S. S. J. Phys. Chem. 2002, 106, 12144. (c) Wang, Y. Q.; Sherwood, P. M. A. Chem. Mater. 2004, 16, 5427.

KocharoVa et al.

Figure 1. Core level HRXPS spectra of C 1s emission from the (a) dried droplet of SWNT-2, (b) “pure” defect-free SWNTs obtained after thermal treatment of the drop-cast, from a dispersion of purified HiPCO SWNTs in water, film, and (c) SAM of surfactant 2.

in diameter (0.4-0.5 nm) SWNTs are known to be stable up to at least 400 °C.31 To get a reference HRXPS C 1s spectrum from pure, adsorbate-free nanotubes, we also applied thermal treatment for purified HiPCO SWNTs at 110 °C, which desorbs possible contaminants with simultaneous retention of the nanotube population. To obtain a reference C 1s spectrum for adsorbatefree SWNTs, we also applied thermal treatment for purified HiPCO SWNTs, which desorbs possible contaminants with simultaneous retention of the nanotube population. In the resulting C 1s spectrum (Figure 1b) of these pure SWNTs, the single peak appears centered at 284.7 eV and is consistent with the BE of sp2-hybridized carbon previously reported for SWNTs.4,30 The presence of chemisorbed oxygen, in the form of hydroxyl, carbonyl, and/or carboxyl groups, on the nanotube damaged walls and/or ends leads to the appearance in the HRXPS C 1s spectrum of additional shoulders at 286.3 eV (hydroxyl carbon), 287.6 eV (carbonyl carbon), and 288.8 eV (carboxyl carbon).30b None of these shoulders are observed in any of the C 1s spectra presented in Figure 1, confirming the absence of the oxygen-containing contaminants and also the defect-free structure of the nanotube walls in the SWNT-2 composite. The BEs of the main maximum in the core level spectrum of the 2 SAM (Figure 1c) and the C 1s peak highest in intensity in the HRXPS spectrum of the SWNT-2 bulk (Figure 1a), observed at 285.3 eV, are very close to the respective value for imidazolium alkanethiols on gold reported at ∼285 eV.25 Whereas the peak at 285.3 eV can be attributed to the C atoms in the aliphatic dodecyl chain, the peak at 286.8 eV can be assigned to carbons bonded to nitrogen32,4 in the imidazolium ring of 2. Quantitative estimation of the relative atomic concentration of carbon in the SWNT-2 bulk (Figure 1a) gives a ∼1:2 ratio (31) Xu, Y.; Ray, G.; Abdel-Magid, B. Composites, Part A 2006, 37, 114. (32) Frey, S.; Shaporenko, A.; Zharnikov, M.; Harder, P.; Allara, L. J. Phys. Chem. B 2003, 107, 7716.

Self-Assembled Carbon Nanotubes on Gold

Langmuir, Vol. 24, No. 7, 2008 3239

the methyl asymmetric bending modes.34 The maxima of the νa(CH2) and νs(CH2) bands at 2926 and 2857 cm-1, respectively, indicate that most of the hydrocarbon chains of 1 and 2 in SWNT1,2 SAMs are disordered.34,35 However, the shoulders observed at 2916 and 2850 cm-1 of νa(CH2) and νs(CH2), respectively, also indicate the presence of the highly ordered hydrocarbon units in an all-trans conformation characteristic of SAMs of long alkanethiols on gold.36,37 The peak at 3030 cm-1 is assigned to the symmetric H-C-H stretch in the CH3 group terminal to the imidazolium ring.38

Figure 2. C-H stretch region in PM-IRRAS spectra: (a) SWNT-1 SAM and (b) SWNT-2 SAM. The SWNT-1 PM-IRRAS spectrum was corrected for the presence of the primer MESA SAM.

between SWNT carbon (284.8 eV) and surfactant 2 carbon (285.3 and 286.8 eV), respectively, which implies the presence of a large amount of free 2 molecules not adsorbed on the nanotube sidewalls. As one can see in Figure 1a, the C 1s photoemission of SWNT-2 maintains the BE localizations at 285.3, 286.8, and 284.8 eV for the 2 SAM (Figure 1c) and pure SWNTs (Figure 1b), respectively, which indicates that nearly no interaction between C atoms of the nanotubes and the molecules of 2 takes place or these interactions are very weak. However, the occurrence in the core level spectrum of SWNT-2 of a very small peak at 289.4 eV implies that some interactions between the SWNT and 2 molecules could take place. PM-IRRAS. To study self-assembly of thiolated SWNTs on gold, we also used PM-IRRAS. In PM-IRRAS, the so-called “metal-surface PM-IRRAS selection rule” 10a,33 is applied, according to which only those adsorbate vibrational modes that have a component of the dipole moment derivative normal to the surface with a p-polarized electric field are detectable and produce a p-polarized reflectance signal, Rp. At this condition, a strong correlation between the IR intensity and the orientation of the corresponding functional group is observed. At s-polarization, since the electric field vanishes at the metal surface, IR radiation does not couple to the adsorbate vibrational modes and the corresponding Rs reflectance signal is unaffected by the adsorbate and serves as a reference. For comparison reasons, we prepared an aqueous dispersion of SWNTs noncovalently functionalized with the imidazoliumbased surfactant 1, in which the cation structure differs from that of 2 by the absence of the thiol group. Both SWNT-1 and SWNT-2 were adsorbed on gold substrates from their aqueous dispersions in the form of self-assembled submonolayers and studied by the PM-IRRAS technique. In the case of SWNT-1, the surface was first negatively primed with a monolayer of MESA. Figure 2 shows PM-IRRAS spectra in the CH stretching region for freshly deposited SWNT-1,2 SAMs. This spectral region contains only CH stretching modes of surfactant 1 or 2. For the SWNT-1 SAM (Figure 2a), the νas(CH2) band at 2916-2926 cm-1 is asymmetric in shape with a shoulder at 2930 cm-1 originating from the Fermi resonance between the fundamental frequency of the symmetric methyl stretch and the overtones of (33) Chabal, Y. J. Surf. Sci. Rep. 1988, 8, 211.

We note that both SWNT-1,2 dispersions, from which all the films studied in this work were deposited, also contained free molecules of 1 or 2. When applying the metal-surface selection rule, it is necessary to remember that the intensity value in surface IR is also a measure of the adsorbate amount. The observed split of the νas(CH3) mode, with distinguished maxima at 2958 and 2965 cm-1, was attributed to restricted rotation of the terminal CH3 group36 and can indicate the presence of two kinds of 1 and 2 molecules in SWNT-1,2 SAMs: those that are adsorbed on nanotube walls and those that are not. With the restriction of the mobility of the methyl group bonded to the imidazolium ring, its C3V symmetry is lowered to C1 and the νas(CH3) band splits into two bands.35 The similarity in the position and relative intensities of CH stretch modes of 1 and 2 imply quite similar conformations of their hydrocarbon chains and an inclined orientation of these molecules on the gold substrates. The intensities of the symmetric H-C-H stretch bands of the CH3 groups terminal to the imidazolium ring at 3030 cm-1 are almost equal for both SWNT-1 and SWNT-2 SAMs, indicating similar amounts of the adsorbed material on the gold surface. Therefore, the overall higher intensity of the CH stretch modes of hydrocarbon chains of 2 for the SWNT-2 SAM compared with the SWNT-1 SAM in the 2840-2970 cm-1 region suggests that the molecular axis of the thiol group-containing 2 is oriented more perpendicular on the bare gold than the 1 molecules on the MESA-primed gold. The lower frequency region in PM-IRRAS spectra of SWNT1,2 (Figure 3) contains IR modes of 1 and 2 (see the IR spectrum of 2 in the Supporting Information)25,38 and also SWNT IR modes.13 The observed fundamental modes of 1 and 2 in the spectra in Figure 3 include ν(N-C-N) in the imidazolium ring and δs(CH3) of the methyl groups terminal to the imidazolium ring at 1559 and 1470 cm-1, respectively.38 Other fundamentals of imidazolium-based ionic liquids are very weak or absent in PM-IRRAS spectra of SWNT-1,2 SAMs. In contrast, some combination bands of 1 and 2 that are extremely weak in IR are activated in PM-IRRAS: the in-plane imidazolium ring bend with the alkyl CH2 twist at 1300 cm-1, the imidazolium ring ν(N-C) with the ring δ(CH) at 1070 cm-1, the alkyl ν(C-C) with the C-H rock at 1045 cm-1, and also the combination of (34) Snyder, R. G.; Strauss, H. L.; Elliger, C. A. J. Phys. Chem. 1982, 86, 5145. (35) MacPhail, R. A.; Strauss, H. L.; Snyder, R. G.; Elliger, C. A. J. Phys. Chem. 1984, 88, 334. (36) Zawisza, I.; Burgess, I.; Szymanski, G.; Lipsowski, J.; Majewski, J.; Satija, S. Electrochim. Acta 2004, 49, 3651. (37) Seshadri, K. S.; Jones, R. N. Spectrochim. Acta 1963, 19, 1013. (38) (a) Talaty, E. R.; Raja, S.; Storhaug, V. J., Do¨lle, A.; Carper, W. R. J. Phys. Chem. B 2004, 108, 13177. (b) Heimer, N. E.; del Sesto, R. E.; Meng, Z.; Wilkes, J. S.; Carper, W. R. J. Mol. Liq. 2006, 124, 84. (c) Chang, H-C.; Jiang, J-C.; Su, J-C.; Chang, C-Y.; Lin, S-H. J. Phys. Chem. A 2007, 111, 9201. (d) Katsyuba, S. A.; Zvereva, E. E.; Vidisˇ, A.; Dyson, P. J. J. Phys. Chem. A 2007, 111, 352.

3240 Langmuir, Vol. 24, No. 7, 2008

KocharoVa et al.

Figure 3. Lower frequency region in PM-IRRAS spectra: (a) SWNT-1, (b) SWNT-2. Table 1. Comparison of the SWNT IR Mode Frequencies (cm-1) Observed in PM-IRRAS Spectra of SWNT-1,2 SAMs with the Experimental and Theoretical Values Reported Earlier13 SWNT-1

SWNT-2

ref 13, exptla

ref 13, theory

810 850 956 985 1242 1455

810 848 957 987 ∼1247 1455

806 854 956 987 1262 1443

774-775 870-876 970 970 1248-1261 1455

first-order A2 first-order E1 second-order 2E25 second-order 2E25 first-order A2 second-order 2E12, 2E6, 2E9

1465 1539

1465 1539

1464 1541 1564

1499-1501 1582-1588

first-order A2 first-order E1

assignment

a For the purified SWNT sample after subjection to high-temperature vacuum annealing at 1400 °C.

the imidazolium ring bend with the ring ν(C-C) at 835 cm-1.38-40 This attenuation of strong IR fundamentals with simultaneous activation of other vibrational modes can be attributed to molecular symmetry breaking at the surface for organic molecules chemisorbed on gold and was reported earlier in the IRRAS spectrum of the self-assembled 2 on gold.25 The orientation of 1 and 2 in SWNT-1,2 SAMs is difficult to derive in this spectral region using the metal-surface PM-IRRAS selection rule applied above in the CH stretch spectral region because of extremely low intensity values. The following bands observed in PM-IRRAS spectra of SWNT-1,2 SAMs, which do not belong to 1 and 2, can be tentatively assigned to the SWNT IR-active modes: 1539, 1465, 1455, 1280, 1247 (1242), 987 (985), 957, 848 (850), and 810 cm-1. Recently, Kim et al. made assignments13 of the SWNT IR-active first- and second-order modes, which prompted us to attempt to identify the SWNT IR-active modes in PM-IRRAS spectra of self-assembled SWNTs. Table 1 presents the IR band frequencies for SWNT-1,2 and those previously reported.13 Earlier we verified self-assembly of SWNT-1,2 on gold by SERRS and by XPS data.4 Very good agreement between SWNT-1,2 frequencies observed in PM-IRRAS spectra in Figure 3 with IR experimental data for SWNTs13 additionally confirms self-assembly of noncovalently functionalized SWNTs on gold and the feasibility of PM-IRRAS for study of functionalized and self-assembled carbon nanotubes. The shoulder at 1280 cm-1 in the PM-IRRAS spectrum (Figure 3b) of SWNT-2 has the same energy as the Raman-active SWNT (39) Berg, R. W.; Deetlefs, M.; Seddon, K. R.; Shim, I.; Thonpson, J. M. J. Phys. Chem. B 2005, 109, 19018. (40) Heimer, N. E.; Del Sesto, R. E.; Meng, Z.; Wilkes, J. S.; Carper, W. R. J. Mol. Liq. 2006, 124, 84.

Figure 4. Angle-dependent C K-edge NEXAFS spectra for a dried SWNT-2 droplet on gold. The inset shows a scheme for the NEXAFS experiment and three nanotubes aligned along Cartesian axes. Spectra are taken at three angles (θ) after rotation of the sample in the plane of incidence. The direction of the resultant electric field vector E h changes in the xz plane, always remaining perpendicular to the direction of the incident X-ray beam.

disorder D-band that we previously observed in the RR spectrum of thiolated SWNT-2 tubes.4 The IR spectrum of 2 does not show IR modes nearby. As can be seen from Table 1, there are also no predicted SWNT IR-active fundamentals near this frequency. The two nearest IR-active A2 modes lie at 1262 and 1369 cm-1 (not shown in Table 1).13 However, the Ramanactive D-mode, due to reduction of its symmetry as a consequence of disorder, can become allowed in PM-IRRAS experiments. Such disorder, besides sp2 bonds of structural defect origin, can be caused also by the finite length of the tubes or the self-assembly process. Therefore, the shoulder at 1280 cm-1, observed in the PM-IRRAS spectrum of SWNT-2, can be ascribed to the D-band IR feature. NEXAFS Spectra. Dried Droplets of SWNT-2. One of the strengths of NEXAFS spectroscopy is its ability to simultaneously probe the electronic structure of the surface and to some extent the interior of the material. Due to the angular dependence on the angle made by π* and/or σ*C-C orbitals with respect to the electric field vector E h of the incident polarized X-rays, NEXAFS spectroscopy is able to determine the orientation of functional groups of organic molecules on a surface. To further study interactions between the SWNT and the functionality 2 and also to elucidate the packing and orientation of SWNT-2 tubes in the bulk of a dried droplet on gold, we performed angle-dependent NEXAFS measurements. Figure 4 shows the variation in the PEY intensity in C K-edge NEXAFS spectra of SWNT-2 on gold with incidence angle θ between the surface normal and the incident synchrotron radiation beam. All NEXAFS spectra presented in Figure 4 exhibit four main sets of sharp resonances at 284.8-286.2, 292-293.4, ∼297307, and 287.1-290.3 eV. The first two sets of lines observed are two edges of π and σ symmetries characteristic of the carbon line shape of carbon nanostructures.19,21,41 The intensity and the photon energy peak location of the π* resonance (1s f π* transition), as in XPS, are measures of the bond hybridization, whereas the shape and/or position of the σ* resonance (1s f σ* transition) peaks are directly related to the intramolecular bond length. The lines at 292-294 eV are associated with σ*C-C transitions in SWNTs at 292 and 292.4 eV19 and in alkanethiols (41) Terminello, L. J.; Shuh, D. K.; Himpsel, F. J.; Lapiano-Smith, D. A.; Stohr, J.; Bethune, D. S.; Meijer, G. Chem. Phys. Lett. 1991, 182, 491.

Self-Assembled Carbon Nanotubes on Gold

at 293.4 eV.42,43 The third set of σ + π absorption features at ∼297-307 eV is related to intramolecular excitation (π f π*, σ f σ*) states associated with adsorption of molecules on metal surfaces. The energetically lowest resonances at 284.8-286.2 eV can be assigned to π* resonances of free individual SWNTs (284.8 eV) as well as of nanotubes having π-π interactions either with adjacent tubes in small bundles or with the π electronic structure of the imidazolium ring of 2. Indeed, earlier Zhao et al. predicted44 and recently Gotovac et al. experimentally showed45 that coupling between π electrons in SWNTs and π electrons of aromatic molecules causes upshift of the main π* resonance SWNT peak. In addition, it was also reported that, as a result of π-π interactions upon formation of films, the molecular π orbitals split and form a band fine structure.46 The adsorption of amphiphilic molecules of 2 on the nanotube sidewalls in solution, as performed in our study, has an advantage of noncovalent nanotube functionalization via weak intermolecular forces including van der Waals and, probably, CH-π interactions. At the same time, positively charged imidazolium rings of 2 are electrostatically attracted by the negatively charged π systems of nanotubes, whereas the graphene nanotube π electronic structure can interact with the π electronic structure of the imidazolium ring of 2. The observed splitting and upshift (284.8 f 285.3 f 285.6 f 285.9 f 286.2 eV) of the π* SWNT orbitals for SWNT-2 are in accordance with these expectations and confirm the presence of weak π-π and, probably, CH-π interactions in the thick film of thiolated SWNTs. The effect of CH-π interactions is seen more significantly in the 287.1-290.3 eV range. The resonances at 287-289 and 289-291 eV earlier were attributed to the π*(CO) and σ*(CO) states, respectively, in NEXAFS spectra of oxidized SWNTs19 and in the NEXAFS of SWNT buckypaper (free-standing SWNTs).21 Here we note that for aqueous dispersion and thiolation of SWNTs with 2 we used pristine, untreated by acidic purification, HiPCo nanotube material. As was shown by HRXPS data above, we did not find any organic contaminants in the form of oxygen-containing groups chemisorbed on the nanotube sidewalls or ends. In addition, our previous work showed negligible concentrations of carbonaceous impurities and imperfections in the graphene lattice of SWNTs in SWNT-1,2 dispersions, from which all films in this work were deposited.4 The photon energy resolution (better than 0.2 eV) allows the identification of closely spaced sharp resonances between 287 and 288.7 eV to the transitions of core electrons to the lowest Rydberg/valence resonances (or mixed R*/σ*C-H series of states) in the hydrocarbon backbone32,42,43,47-52 of 2. Indeed, the exact positions of the observed resonances in this range are very similar (42) (a) Bagus, P. S.; Weiss, K.; Schertel, A.; Wo¨ll, Ch.; Braun, W.; Hellwig, C.; Jung, C. Chem. Phys. Lett. 1996, 248, 129-135. (b) Fu, J.; Urquhart, G. Langmuir 2007, 23, 2615. (c) Weiss, K.; Bagus, P. S.; Wo¨ll, Ch. J. Chem. Phys. 1999, 111, 6834. (43) (a) Dannenberger, O.; Weiss, K.; Himmel, H.-J.; Ja¨ger, B.; Buck, M.; Wo¨ll, Ch. Thin Solid Films 1997, 307, 183. (b) Dannenberger, O.; Weiss, K.; Wo¨ll, C.; Buck, M. Phys. Chem. Chem. Phys. 2000, 2, 1509-1514. (c) Yan, C., Go¨lzhauser, A., Grunze, M. Langmuir 1999, 15, 2414-2119. (d) Yan, C.; Zharnikov, M.; Goelzhauser, A.; Grunze, M. Langmuir 2000, 16, 6208. (e) Zwahlen, M.; Herrwerth, S.; Eck, W.; Grunze, M.; Ha¨hner, G. Langmuir 2003, 19, 93059310. (44) Zhao, J.; Lu, J. P. Appl. Phys. Lett. 2003, 82, 3746. (45) Gotovac, S.; Honda, H.; Hattori, Y.; Takahashi, K.; Kanoh, H.; Kaneko, K. Nano Lett. 2007, 7, 583-587. (46) Bre´das, J. L.; Calbert, J. P.; da Silva Filho, D. A.; Comil, J. Natl. Acad. Sci. U.S.A. 2002, 99, 5804. (47) Urquart, S. G.; Gilles, R. J. Phys. Chem. A 2005, 109, 215. (48) Urquart, S. G.; Gilles, R. J. Chem. Phys. 2006, 124, 234704. (49) Stohr, J.; Outka, D. A.; Baberschke, K.; Arvantis, D.; Horsley, J. A. Phys. ReV. B 1987, 36, 2976., (50) Hahner, G.; Kinzler, M.; Thummler, C.; Woll, C.; Grunze, M. J. Vac. Sci. Technol., A 1992, 10, 2758.

Langmuir, Vol. 24, No. 7, 2008 3241

to those of the R*/σ*C-H features observed in high-resolution C K-edge NEXAFS spectra of gaseous alkanes,47,48 small- and long-chain alkanes,49,32,42 self-assembled alkanethiols,43 Langmuir-Blodgett monolayers,50 and polymer films.51 Therefore, these observed resonances can be assigned to the R*/σ*C-H transitions in the hydrocarbon backbone of 2. The absorption edges that are below 288 eV were assigned to C 1s f 3s transitions; the edges that are between ∼288 and 288.7 eV were assigned to C 1s f 3p transitions and those above 288.7 eV to C 1s f 3d or C 1s f 4s transitions in hydrocarbon chains.47,52 The peak at 287.7 eV was recently assigned to a combination of the R*/σ*C-H and σ*C-S excitations,51c and the feature with a photon energy of 289.4 eV was attributed to the σ*C-S resonance.32 From a fundamental point of view, the understanding of the origin and orientation variation of the R*/σ*C-H resonance intensity in C K-edge NEXAFS spectra of nanocomposites is highly important since it provides insights into intermolecular effects within their films. However, the interpretation of these mixed states in NEXAFS spectra is varied. High-resolution NEXAFS spectra of simple gaseous alkanes show a progression of the carbon 1s f Rydberg transitions with rich vibronic fine structure,47,48 while NEXAFS spectra of the condensed organic phase contain substantial σ*C-H valence character and exhibit unresolved resonances in the ∼287-290 eV spectral range.32,42,43,49-51 Scattering of the Rydberg electron in the solid, where electron mobility is reduced, leads to the broadening of R*/σ*C-H states due to reduction of the excited-state lifetime of the Rydberg state and vice versa: the solids with a higher electron mobility and weaker electron-atom interactions will have less lifetime broadening of the R*/σ*C-H states and exhibit vibronic fine structure.48,52 This is highly important for interpretation of NEXAFS spectra of nanocomposite films based on carbon nanotubes. We believe that the very well resolved R*/σ*C-H series of states in the carbon NEXAFS spectra of the dried SWNT-2 droplet (Figure 4) imply the existence of close CH-π interactions between hydrocarbon chains of 2 and the π electronic nanotube structure. Indeed, the mobility of the electron excited to a Rydberg state in the condensed organic phase is limited by the collisions and scattering events. Due to a decreased core excited lifetime, NEXAFS spectra of these objects have more σ*C-H valence character and are blue-shifted and broadened.47,48 SWNTs are good electron acceptors and exhibit ballistic transport of electrons. When alkyl chains of the surfactant are in close contact with highly conductive carbon nanotubes in the SWNT-2 film and when C K-edge electrons in the hydrocarbon chain are excited by synchrotron radiation to R* states, instead of scattering and quenching, they transport through conductivity paths provided by SWNTs. In this case, R*/σ*C-H states take more Rydberglike character with a rich vibronic fine structure, which is observed in Figure 4. To derive qualitative information on the packing in the SWNT-2 composite and orientational order within the thick SWNT-2 film, the incidence angle of synchrotron radiation was varied from normal 0° incidence (electric field vector E h in the surface plane) to 60° incidence (E h at 30° from the surface normal). Here we recall that the PEY signals for the dried SWNT-2 droplet on gold were normalized by the PEY signals (51) (a) Smith, A. P.; Ade, H. Appl. Phys. Lett. 1996, 69, 3833. (b) Scho¨ll, A.; Fink, R.; Umbach, E.; Mitchell, G. E.; Urquhart, S. G.; Ade, H. Chem. Phys. Lett. 2003, 370, 834. (c) Gurau, M. C.; Delongchamp, D. M.; Vogel, B. M.; Lin, E. K.; Fischer, D. A.; Sambasivan, S.; Richter, L. J. Langmuir 2007, 23, 834. (52) Robin, M. B. Higher Excited States of Polyatomic Molecules; Academic: New York, 1974.

3242 Langmuir, Vol. 24, No. 7, 2008

of the SWNT-2 SAM on gold; thus, the spectra presented in Figure 4 represent the SWNT-2 bulk and free molecules of 2 self-assembled on the gold surface. The π*, R*/σ*C-H, and σ*C-C resonances in NEXAFS spectra of SWNT-2 show angular dependence of the resonant photoexcitation process on the orientation of the electric field vector E h of the linearly polarized light with respect to the molecular orbitals, i.e., linear dichroism. For linear alkanes, the R*/σ*C-H states exhibit the strongest excitation probability for E h perpendicular to the molecular axis (or parallel to C-H bonds) and the weakest excitation probability for E h parallel to the molecular axis; the intensity of the σ*C-C resonance for alkanes is strongest for the electric field vector E h along the molecular axis.32,42,43b For molecules of 2 (Figure 4) at 0° incidence, which corresponds to E h at 90° from the surface normal, the intensity of the σ*C-C resonance at 293.4 eV is the highest among three incidence angles applied whereas the intensities of the R*/σ*C-H transitions are weak. This observation indicates an almost lateral orientation of the C-C-C backbone of 2 in which the methylene chains are parallel to the substrate surface. At 20° incidence, corresponding to E h at 70° from the surface normal, the σ*C-C resonance of 2 at 293.4 eV is weakened while the R*/σ*C-H resonances are the highest observed, implying that some C-C-C chains of 2 are oriented more tilted from the surface plane than those probed with 0° incidence. The feature at 287.1 eV associated with the C-C stretch in linear alkanes48 is characteristic of gauche conformations in the alkanethiol chain43a that can arise due to incomplete formation of the film. The intensity of this peak is lowest at 60° incidence, very small at 0°, and highest at 20° incidence. This means that at 60° incidence, which corresponds to the electric field vector E h at 30° with respect to the surface normal, some molecules of 2 have predominantly trans conformation and an average tilt angle of 30° between the molecular axis and the surface normal. The decreased overall partial electron yield indicates that the amount of these molecules probed by 60° incidence is relatively small. These highly ordered C-C-C chains of 2 must have less conformational freedom; therefore, they are densely packed, and most likely, these chains belong to those free molecules of 2 which are not adsorbed on the nanotube sidewalls and which self-assemble on the gold surface with a tilt angle of ∼30° to the surface normal. This result is in agreement with the wellknown average tilt of ∼30° between the molecular axis and the surface normal of perfectly ordered alkanethiols on gold.54 The evaluation of SWNT orientations within the bulk SWNT-2 film is more complicated due to the random mixture of their possible positions within the xy, yz, and xz planes and also their intermediate positions between these planes. Additionally, the σ*C-C SWNT resonances do not exhibit systematic variation in intensity with the incidence angle; therefore, to determine the SWNT order by the NEXAFS, the behavior of π* resonances is primarily considered.23a Simplified possible positions of nanotubes are presented in the inset of Figure 4. Depending on the projections made by E h onto each individual SWNT π* orbital, they contribute to a different extent to the π* intensity at each incidence angle. For tubes A being along the x axis, at any incidence angle applied in the xz plane, the electric field vector E h also rotates in this plane. At glancing incidence (θ ) 90°), E h has the largest projection onto the yz plane, in which also π* orbitals lie, giving rise to the highest π* resonance intensity. A zero intensity of π* resonances is expected for the normal (0°) incidence. For tubes B standing upright, the highest π* resonance (53) Du¨rkop, T.; Getty, S. A.; Cobas, E.; Fuhrer, S. Nano Lett. 2004, 4, 35. (54) (a) Harder, P.; Grunze, M.; Dahint, R.; Whitesides, G. M.; Laibinis, P. E. J. Phys. Chem. B 1998, 102, 426. (b) Vanderah, D. J.; Valincius, G.; Meuse, C. W. Langmuir 2002, 18, 4674.

KocharoVa et al.

Figure 5. C K-edge NEXAFS spectrum for the SAM of thiolated SWNTs (SWNT-2) on gold.

intensity can be observed at normal (0°) incidence when E h is in the same xy plane as the SWNT π* orbitals. Upon increasing the incidence angle, the π* resonance intensity for these tubes decreases. Tubes C being along the y axis exhibit completely isotropic behavior: for all incidence angles, the electric field vector E h always rotates in the plane of the SWNT π* orbitals, giving rise to equally high intensities of π* resonances. Low anisotropy observed for the SWNT π* resonance in the SWNT-2 thick film (Figure 4) indicates the absence of upright-standing tubes exhibiting tube-B-like orientation, which, at 0° incidence, gives rise to an intensity of π* peaks much higher than that of σ*C-C resonances. The presence of C-like tubes also is least possible because they would give rise to the highest intensities of π* resonances at all incidence angles. The equal intensities of π* resonances at 20° and 0° incidence indicate almost parallel orientation of the bulk SWNT-2 nanotubes to the substrate plane, with the direction being in vectorial combination of A- and C-like tubes. These observations suggest that in the bulk of the dried droplet of noncovalently functionalized carbon nanotubes deposited from aqueous dispersions they self-organize into a well-aligned and ordered structure parallel to the substrate plane. Such self-organization of SWNTs in dried droplets into wellaligned ordered structures is in agreement with recently reported scanning electron microscopic data and can be assigned to the strong edge effect on the periphery of the droplet and to the capillary flow effect in the interior.55 SWNT-2 SAM. For probing molecular interactions and packing within self-assembled thiolated carbon nanotubes on gold, we also recordered the C K-edge NEXAFS spectrum of the SWNT-2 SAM at normal (0°) incidence. Figure 5 shows strong differences between the SWNT-2 SAM and the bulk of the thick SWNT-2 film (Figure 4). Similarly to the SWNT-2 bulk, the well-resolved R*/σ*C-H features in the NEXAFS spectrum in Figure 5 indicate high electron mobility within the SWNT-2 SAM. The presence of self-assembled SWNTs in the SWNT-2 SAM is evidenced by the sharp σ*C-C resonance at 292 eV, a π* transition at 285.6 eV, and also a new SWNT π* feature at 286.6 eV. The appearance of this new SWNT π* peak and the upshifted R*/σ*C-H resonances of 2 compared with those in NEXAFS spectra of the thick SWNT-2 film suggests strong interactions of the SWNT-2 composite with the gold surface. Almost equal intensities of nanotube π* resonances and σ*C-C resonances (Figure 5) imply that the long axis of self-assembled thiolated SWNTs is not as parallel to the substrate plane as in the SWNT-2 bulk (Chart 2). (55) Li, Q.; Zhu, Y. T.; Kinloch, I. A.; Windle, A. H. J. Phys. Chem. B 2006, 110, 13926.

Self-Assembled Carbon Nanotubes on Gold Chart 2. Schematic Illustration of (a) Self-Organization of Noncovalently Functionalized Nanotubes into an Aligned Structure in the SWNT-2 Bulk and (b) Self-Assembly of Thiolated Nanotubes on the Gold Surface via a Covalent S-Au Bonda

a Black and red lines represent carbon nanotubes and surfactant molecules, respectively. Note that free molecules of 2 and those adsorbed on the nanotube sidewalls (in the SWNT-2 composite) self-assemble on gold with different angles.

In contrast to almost lateral orientation of alkyl chains of 2 in the SWNT-2 bulk (Chart 2), the C-C-C backbones of 2 adopt inclined orientation in the SWNT-2 SAM. This is evidenced by the observed much higher partial electron yield of R*/σ*C-H resonances at 288.2 and 289.9 eV compared to that of the σ*C-C resonance at 293.1 eV and the C-C alkane resonance at 287.2 eV. This result well correlates with the earlier NEXAFS study of the octadecanethiol SAM on gold where the average tilt angle of 18° from the surface normal for alkanethiol adsorbed from solution was found.43c It should be recalled here that the films studied in this work were deposited from an aqueous dispersion of thiolated SWNT-2, which also contains dissolved free molecules of 2 that can adsorb on gold with a tilt angle different from that of the SWNT-2 composite. This can partly explain the unusual normal orientation of functionality 2 on gold compared with that in the SWNT-2 bulk (Figure 4). As was mentioned above, previous NEXAFS studies of self-assembled alkanethiols showed broadening of Rydberg/valence resonances resulting in smoothing and disappearance of vibronic fine structure, which is not observed in the NEXAFS spectrum of the SWNT-2 SAM (Figure 5). The observed sharpness of R*/σ*C-H resonances and upshifted value of the photon energy for these

Langmuir, Vol. 24, No. 7, 2008 3243

states indicate close contact of 2 molecules with nanotubes in the SWNT-2 SAM and with the gold surface.

Conclusions The combination of high-resolution XPS, NEXAFS, and PMIRRAS spectroscopy is shown to be a complementary and powerful set of methods for the study of electronic structure, molecular interactions, packing, alignment, and orientation within SWNT composites in thick films and self-assembled submonolayers on gold. High-resolution XPS confirmed that thiolation of SWNTs with the imidazolium-based multifunctional surfactant 2 bearing a thiol group occurs via noncovalent adsorption of 2 molecules on the nanotube sidewalls. In PM-IRRAS spectra of self-assembled nanotubes on gold, the IR-active SWNT modes have been observed and identified. According to PM-IRRAS data, the hydrocarbon chains of 2 are oriented with less tilt angle to the bare gold normal in the SAM deposited from the SWNT-2 dispersion than those of 1 deposited from the SWNT-1 dispersion on the mercaptoethanesulfonic acid-primed gold. The C K-edge NEXAFS results provided significant information on the nature of interactions within noncovalently thiolated SWNTs. For both the dried SWNT-2 droplet and the self-assembled SWNT-2 submonolayer, the highly resolved carbon 1s f R*/σ*C-H transitions in the alkyl chain of 2 imply the existence of close CH-π interactions between 2 and the π electronic nanotube structure. The splitting and shift of the π* SWNT orbital in NEXAFS spectra for the dried SWNT-2 droplet confirm the presence of weak π-π interactions within the SWNT-2 bulk. Angle-dependent NEXAFS spectra show that molecules of 2 adsorbed on the nanotube sidewalls are aligned along the nanotubes, which are self-organized in the SWNT-2 bulk parallel to the substrate plane. In the NEXAFS spectrum of the SWNT-2 SAM on gold, the upshifted values of the photon energy for R*/σ*C-H transitions indicate close contact of 2 with the nanotubes and with the gold surface. The PM-IRRAS and the carbon NEXAFS spectra of noncovalently functionalized SWNTs and also of self-assembled SWNTs on gold are reported for the first time. Acknowledgment. The Materials Science beamline is supported by the Ministry of Education of the Czech Republic under Grant No. LC 06058. Supporting Information Available: Procedure of the SWNT aqueous dispersion with surfactants 1 and 2 and UV-vis-NIR spectrum of the SWNT-2 composite in an aqueous medium. This material is available free of charge via the Internet at http://pubs.acs.org. LA7030768