Infrared Physics & Technology 53 (2010) 434–438 Contents lists available at ScienceDirect Infrared Physics & Technology journal homepage: www.elsevier.com/locate/infrared FTIR spectroscopy as a tool for nano-material characterization Charles Baudot a,⇑, Cher Ming Tan b, Jeng Chien Kong a a STMicroelectronics Asia Paciﬁc Pte Ltd., 5A Serangoon North Avenue 5, Singapore 554574, Singapore b School of EEE, Nanyang Technological University, Block S2, Nanyang Avenue, Singapore 639798, Singapore a r t i c l e i n f o a b s t r a c t Article history: Covalently grafting functional molecules to carbon nanotubes (CNTs) is an important step to leverage the Received 10 June 2010 excellent properties of that nano-ﬁber in order to exploit its potential in improving the mechanical and Available online 22 September 2010 thermal properties of a composite material. While Fourier Transform Infra Red (FTIR) spectroscopy can display the various chemical bonding in a material, we found that the existing database in FTIR library Keywords: does not cover all the bonding information present in functionalized CNTs because the bond between FTIR characterization the grafted molecule and the CNT is new in the FTIR study. In order to extend the applicability of FTIR Carbon nanotube to nano-material, we present a theoretical method to derive FTIR spectroscopy and compare it with Covalent functionalization Epoxy ﬁllers our experimental results. In particular, we illustrate a method for the identiﬁcation of functional mole- cules grafted on CNTs, and we are able to conﬁrm that the functional molecules are indeed covalently grafted on the CNTs without any alterations to its functional groups. Ó 2010 Elsevier B.V. All rights reserved. 1. Introduction load transfer and heat transfer between the epoxy matrix and CNT ﬁllers, covalent bonding between the polymer and the nano- Infrared spectroscopy is a crucial tool to characterize the struc- tubes must be present. Consequently, we intend to apply FTIR to ture of matter at the molecular scale. Based on the singular reso- examine the presence of such covalent bonds. nant vibrational modes of different branches or body parts of a Blake et al.  demonstrated that there is indeed a covalent molecule, one can reveal its constituting elements and its bonding bond between functional molecules and CNTs by grafting silicon arrangement. The reverse process is also exploited because a spe- and tin based molecules on the nanotubes and by looking for ciﬁc molecule possesses a unique infrared spectrograph equivalent C–Si and C–Sn vibration modes in the FTIR spectrum. However, to a ﬁnger print which can unveil its identity. FTIR spectroscopy such an afﬁrmation cannot be done if purely carbon based func- analysis is a method based on the principle of infrared spectros- tional molecules are grafted as it is the case in our present study. copy, and it has extended its area of application to the study of C–Si and C–Sn vibrations occur in a very speciﬁc narrow range, nano-scaled objects during the last decade. almost independently of the remaining molecular structure, A common approach to exploit the unique properties of nano- because of the relatively big difference in atomic masses of the particles and to improve its workability is to functionalize the sur- two atoms. However, carbon to carbon bonds exist in a very wide face of the latter [1–6]. However, the application of FTIR in the range of wavenumbers due to the huge amount of existing organic characterization of functionalized nanoparticles could be difﬁcult molecules and their various conﬁgurations. Coleman et al.  used because it may not reveal the chemical bonds between the nano- another alternative to demonstrate the presence of covalent bond- particles and the functional molecules based on a database of ing. They grafted thiolic functional molecules on the CNTs and known vibrational frequencies. This is because the chemical bond dispersed gold nanoparticles in a solution containing the function- between the functionalized molecules and the nano-material is alized CNTs (f-CNTs). Then, using an atomic force microscope to usually novel, and thus, a strategy must be developed in order to scan a sample of the dispersed f-CNTs on a surface, they showed provide evidence about the resonant vibrational information be- the CNTs decorated with nanoparticles. However, this method is tween the grafted molecules and the nano-sized particles. only usable to demonstrate that covalent bonds indeed occur on In this work, we study the covalent grafting of epoxide mole- CNTs. It is a destructive method that alters the characteristics of cules on CNTs . This strategy is for the exploitation of CNTs as the functional molecules and it is only compatible with the grafting ﬁllers in epoxy composite materials for its mechanical and heat of functional groups that can bond to nanoparticles which do not dispersion enhancements. In order to ensure a better mechanical necessarily suit the targeted application. The method that we present here is a combined experimental ⇑ Corresponding author. Tel.: +65 6427 7368; fax: +65 6427 7379. and theoretical FTIR analysis of f-CNTs to unveil the occurrence E-mail address: firstname.lastname@example.org (C. Baudot). of a chemical covalent bond between functional molecules and 1350-4495/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.infrared.2010.09.002 C. Baudot et al. / Infrared Physics & Technology 53 (2010) 434–438 435 Once n-butyllithium reagent starts attacking the CNTs, a natural dispersion of the functionalized nanotubes occurs by electro- repulsion. After stirring the homogeneous solution for 2 h, epichlorohydrin is added in controlled amount while stirring is continuing. During this reaction, while lithium chloride is formed as a by-product, epoxide branches are grafted on the CNTs by nucleophilic substitution. Thus, each functionalized site contains two branches, namely a butyl branch and an epoxide branch as shown in Fig. 1. Fig. 1. Sketch showing the grafting of functional molecules on the side wall of a The butyl branch contributes in producing unevenness on the carbon nanotube. CNT surface that reduces considerably the bundling effect which is due to both the Van der Waals forces of attraction and the high CNTs. Pristine CNTs show very weak resonant frequencies in the aspect ratio of the nanotubes. Moreover, the epoxide branch con- infrared spectrum in the 400–4000 cmÀ1 range. However, once tributes in improving the solubility of the CNTs in epoxy resins. molecules are grafted on it, we observe differential infrared Furthermore, the epoxy ring contained in the grafted functional absorption response due to the functional sites. The purpose of this molecule participates in the polymerization of the composite when work is to provide evidence that the observed differential response the curing agent is added. Thus a covalent bond is believed to be is not due to either adsorbed molecules, wrapping molecules or created between the CNT and the composite matrix. any other such kind of residual particles which are not covalently Using a Perkin–Elmer System GX 2000 FTIR spectrometer, an bonded to the CNTs, but it is indeed due to the covalent bonding extensive FTIR analysis is performed on samples of f-CNTs. The of the grafted molecules to the CNTs. samples are prepared by taking a small amount of f-CNTs in pow- der form and mix it with potassium bromide to make several thin and semi-transparent discs. The discs are then analyzed in the 2. Experimental spectrometer under ambient conditions using different wavenum- ber resolutions. First, a broad scan of the sample is performed from The chemical procedure for the grafting of the functional mole- 400 cmÀ1 to 4000 cmÀ1 with a step of 4 cmÀ1 (Fig. 2). Among the cules to CNTs consists of two steps in a one pot reaction. The ﬁrst various peaks present in the spectrum, several are already clearly step consists of initiating reactive sites on the CNT walls by using a identiﬁable as belonging to the functional molecules. For example, lithiated alkyl and the second step, bonding functional molecules we observe a peak at 830 cmÀ1 which corresponds to a mono- containing oxirane rings. The detailed reaction path and methodol- substituted epoxide ring vibration. We also observe peaks corre- ogy is described elsewhere . The CNTs that are used for this sponding to the straight chain alkanes or branches of the epoxide experiment are high purity multiwall nanotubes from Nanothinx molecules. The list of known wavenumber ranges obtained from S.A. The use of organo-metallic agents to initiate reaction sites on FTIR databases  is listed in Table 1. In order to obtain more de- CNTs has already been reported [10–12] and n-butyllithium is used tailed spectra of the sample, a higher precision FTIR analysis on a in our work here. This organo-lithium molecule is obtained from narrower range of 800–1800 cmÀ1 is performed with a step of Sigma–Aldrich at a concentration of 2 M in cyclohexane. The sol- 0.1 cmÀ1 (Fig. 3). In order to visualize the peaks that are masked vent in which the reaction takes place is high purity anhydrous because of the convolution, a ﬁrst order derivative of the spectrum cyclohexane from Sigma–Aldrich. The reactions are done at room is performed (Fig. 4). The information collected in those spectra re- temperature under inert atmosphere to prevent the very reactive veal peaks which are not identiﬁable in the database of the pure butyl alkyl from reacting with the humidity present in air. The initial constituents such as the peaks at wavelengths of glove box used during the experiments consists of an argon atmo- 1401 cmÀ1 and 1470 cmÀ1, but they are believed to be of crucial sphere with a humidity level of less than 0.1 ppm and an oxygen importance for the characterization of the grafting of functional level of at most 2 ppm. molecules on CNTs. Fig. 2. Experimental FTIR spectrum of the functionalized CNTs showing the changes in transmittance of the IR signal over the range of 400–1800 cmÀ1 (inset 2500– 4000 cmÀ1). 436 C. Baudot et al. / Infrared Physics & Technology 53 (2010) 434–438 Table 1 3. Theoretical calculations List of know IR resonant vibrations related to straight chain alkanes, epoxide rings and water. In order to identify the signiﬁcance of each peak, a model of the Wavenumber Deformation type Fragment f-CNTs must be done and the different vibrational modes of the de- range (cmÀ1) signed macro-molecule must be compared to the experimental re- 445–825 Water O–H out-of-plane bending sults. As CNTs are almost chemically inert, a chemical reaction is vibration more likely to occur on the reactive defect sites of the CNT walls. 455–455 Straight chain alkanes C–C skeleton vibration 485–540 Straight chain alkanes C–C skeleton vibration 805–880 Mono-substituted Ring vibration epoxide ring 1060–1150 Saturated aliphatic esters C–O asymmetric stretching 820–1140 Saturated aliphatic esters C–O symmetric stretching 1370–1390 R–CH3 C–H symmetric deformation vibration 1440–1480 –CH2– C–H scissor vibration 1475–1500 Mono-substituted CH2 deformation vibration epoxide ring 1595–1710 Water O–H in-plane bending vibration 2865–2885 R–CH3 C–H symmetric stretching vibration 3000–3700 Water O–H stretching vibration Fig. 5. (a) Positions 1 and 2 are the locations of the binding sites for the functional 3230–3550 Water Intermolecular hydrogen bond molecules on the C30 and (b) Cluster model of the functionalization of a C30 molecule. Fig. 3. High resolution FTIR spectrum of the functionalized CNTs over the range of 800–1800 cmÀ1. Fig. 4. First order derivative of the FTIR spectrum shown above over the range of 1300–1550 cmÀ1. C. Baudot et al. / Infrared Physics & Technology 53 (2010) 434–438 437 Fig. 6. Optimized conﬁgurations of the three conformers created by grid scan technique for the functionalized half-fullerene. Table 2 Table 3 Theoretical results of the covalently grafted bond states for different orientations. Vibrational modes comparison between experiment and theoretical work. Conﬁguration B.E. (eV)a L(CCNT–Cbutyl) (Å)b L(CCNT–Cepoxide) (Å) Wavenumber (cmÀ1) Fragment deformation 6a 2.58 1.685 1.836 Experimental Theoretical 6b 6.61 1.581 1.571 489 493 Straight chain alkanes C–C skeleton vibration 6c 6.47 1.577 1.577 830 843 Epoxide ring Ring vibration a Binding energy (absolute value). 1401 1403 CCNT–Cepoxide C–C stretching b Bond length, L. 1470 1475 CCNT–Cbutyl C–C stretching 1504 1504 Epoxide ring CH2 bending Thus, the most probable location to ﬁnd grafted molecules on the ration 6a remains at its original position, leading to an unstable CNTs will be a place where defects naturally occur, that is, the arrangement with a B.E. of about 4 eV less than the two other con- tip of the CNT. Therefore, the model that we develop consists of formations. Correspondingly, the calculated L (CCNT–Cepoxide) in 6a grafting molecules on both CNT tips and on lateral defect sites is somewhat larger than 6b and 6c by about 0.26 Å and this also re- and the different vibrational modes of the resulting structure are ﬂects a much weaker covalent bonding in that conﬁguration as then computed. compared to those in the 6b and 6c arrangements. Therefore, to The ﬁrst cluster model consists of half a fullerene (C30) to which model the covalent functionalization of CNTs, the conﬁguration are attached a butyl molecule and an epoxide molecule placed on 6b is chosen in order to determine the IR spectrum of the f-CNTs. atomic sites where the most active binding regions are deemed We examine the vibrational spectrum of the f-CNTs by means of to be located. Such locations are found at the junction between cluster models with methods based on the density functional the- pentagonal and hexagonal rings (Fig. 5). In order to avoid any ory (DFT) within the generalized gradient approximation (GGA) geometry reconstruction at the hemi-spherical edge, the dangling with the Perdew, Burke and Ernzerhof (PBE) correlation functional bonds are saturated with hydrogen atoms. After performing a . Our calculations are closed shell and are performed using the relaxation of the cluster, the Hessian matrix is calculated in order Dmol3 module of Materials Studio from Accelrys. The electronic to evaluate the resonant vibrational frequencies of the molecular wavefunctions are expanded in atom-centered basis functions de- model . The second model consists of a 7-5-5-7 Stone Wales ﬁned on a dense numerical grid. The chosen basis set is double defect site found on the side wall of a CNT  on which the same numerical plus d-polarization (DND). DND is an all-electron basis molecules are grafted and the same procedures are performed. set composed of two numerical functions per valence orbital, sup- We ﬁrst investigate the most stable conﬁguration of the models. plemented with a polarization d-function on all the non-hydrogen For example, in the case of the functionalized half-fullerene, we atoms. Each basis function is restricted to a cut-off radius. A global study the optimal arrangement of the molecule displayed in real space cut-off is selected as the maximum value of the cut-offs Fig. 5b. Grid scan conformer search technique  is employed speciﬁc for every element of the system. Real space cut-offs for by ﬁxing the half-fullerene and the butyl molecule while rotating atomic species are optimized by considering total energies of the epoxide ring at three different angles with respect to the butyl atoms. The chosen cut-off values lead to atomic energies with an molecule. The same procedure is then repeated for side-wall func- accuracy of 0.1 eV/atom, allowing calculations without a signiﬁ- tionalization. Three conﬁgurations are created and geometrically cant loss of accuracy. The self consistent ﬁeld (SCF) convergence optimized for each model. The ﬁnal conﬁgurations of the function- has been tested and a convergence threshold less than 10À6 Ha alized half-fullerene are shown in Fig. 6. has been evaluated to be satisfactory to achieve the structural The binding energy (B.E.) and the length (L) of the C–C bond optimization. linking the butyl molecule to the C30 (CCNT–Cbutyl) and the epoxide From the simulation of Dmol3, the harmonic wavenumbers ob- molecule to the C30 (CCNT–Cepoxide) are summarized in Table 2. The tained are listed in Table 3 together with the corresponding exper- values of B.E. are calculated based on the following equation: imental results. In agreement with the theoretical results, the B:E: ¼ EðC30 þ butyl þ epoxideÞ À EðC30 Þ À EðbutylÞ À EðepoxideÞ covalent bond formation of the epoxide molecule on C30 is dis- closed by the wavenumber 1403 cmÀ1. Also, the covalent bond for- ð1Þ mation of butyl molecule on C30 is attributed to the CCNT–Cbutyl where E(C30 + butyl + epoxide), E(C30), E(butyl) and E(epoxide) are stretch at around 1475 cmÀ1. Besides the two crucial wavenum- the respective total energies of the indicated systems. bers, we also observe that the theoretical results are in agreement The conformations 6b and 6c in Table 2 are very close to each with the experimental vibrational frequencies of the functional other in both the B.E. and L. The carbon atom located at position molecules such as the ring oscillation and the alkyl stretching. 1 for these two conﬁgurations is distorted and a tetrahedral sp3 However, there are still some peaks that have not been identiﬁed hybridization contributing to the overall structural stability is such as the peaks at 873 cmÀ1, 1297 cmÀ1, 1333 cmÀ1 and formed. On the other hand, the same carbon atom in the conﬁgu- 1358 cmÀ1. We believe that it is because the attachment of the 438 C. Baudot et al. / Infrared Physics & Technology 53 (2010) 434–438 molecules occurs on various defect sites bearing different molecu-  C.J. Tang, A.J. Neves, M.C. Carmo, Infrared absorption study of hydrogen incorporation in thick nanocrystalline diamond ﬁlms, Appl. Phys. Lett. 86 lar conﬁgurations and hence it is not feasible to unveil all of them (2005) 223107. using this method since we do not know the molecular structure of  J.M. Abad, S.F.L. Mertens, M. Pita, V.M. Fernndez, D.J. Schiffrin, all the grafted sites that actually exist. Moreover, the water content Functionalization of thioctic acid-capped gold nanoparticles for speciﬁc in the KBr powder has a masking effect on the peaks that would lie immobilization of histidine-tagged proteins, J. Am. Chem. Soc. 127 (2005) 5689–5694. within the intervals of 445–825 cmÀ1, 1550–1750 cmÀ1 and 3000–  L. Armelao, H. Bertagnolli, D. Bleiner, M. Groenewolt, S. Gross, V. Krishnan, C. 3700 cmÀ1 [18,19]. Sada, U. Schubert, E. Tondello, A. Zattin, Highly dispersed mixed zirconia and hafnia nanoparticles in a silica matrix: ﬁrst example of a ZrO2–HfO2–SiO2 ternary oxide system, Adv. Funct. Mater. 17 (2007) 1671–1681. 4. Conclusions  C. Baudot, C.M. Tan, Solubility, dispersion and bonding of functionalised carbon nanotubes in epoxy resins, Int. J. Nanotechnol. 6 (2009) 618–627.  R. Blake, Y.K. Gun’ko, J. Coleman, M. Cadek, A. Fonseca, J.B. Nagy, W.J. Blau, A In this work, we have successfully demonstrated the use of the generic organometallic approach toward ultra-strong carbon nanotube combination of an experimental FTIR analysis and a theoretical polymer composites, J. Am. Chem. Soc. 126 (2004) 10226–10227. FTIR model to characterize the functionalization of nano-sized ob-  K.S. Coleman, S.R. Bailey, S. Fogden, M.L.H. Green, Functionalization of single- walled carbon nanotubes via the Bingel reaction, J. Am. Chem. Soc. 125 (2003) jects. We successfully unveil the identity of the molecular confor- 8722–8723. mations present on the f-CNTs and prove the presence of covalent  A. Pénicaud, P. Poulin, A. Derré, E. Anglaret, P. Petit, Spontaneous dissolution of bonding between the grafted molecules and the CNTs. a single-wall carbon nanotube salt, J. Am. Chem. Soc. 127 (2005) 8–9.  G. Viswanathan, N. Chakrapani, H. Yang, B. Wei, H. Chung, K. Cho, C.Y. Ryu, P.M. Ajayan, Single-step in situ synthesis of polymer-grafted single-wall Acknowledgments nanotube composites, J. Am. Chem. Soc. 125 (2003) 9258–9259.  S. Chen, W. Shen, G. Wu, D. Chen, M. Jiang, A new approach to the This research work is done in collaboration between the School functionalization of single-walled carbon nanotubes with both alkyl and carboxyl groups, Chem. Phys. Lett. 402 (2005) 312–317. of Electrical and Electronic Engineering of Nanyang Technological  Science and Fun Infrared Spectrum Database. <http://www.science-and- University and STMicroelectronics Asia Paciﬁc Pte. Ltd. We would fun.de/tools/>. like to thank Dr. Andrea Di Matteo and Dr. Francesco Buonocore  F. Pascale, C.M. Zicovich-Wilson, F. López Gejo, B. Civalleri, R. Orlando, R. Dovesi, The calculation of the vibrational frequencies of crystalline compounds from STMicroelectronics for their help in developing the FTIR mod- and its implementation in the CRYSTAL code, J. Comput. Chem. 25 (2004) 888– el of the f-CNTs. 897.  A.J. Stone, D.J. Wales, Theoretical studies of icosahedral C60 and some related species, Chem. Phys. Lett. 128 (1986) 501–503. References  Accelrys Materials Studio Product Datasheets. <http://accelrys.com/products/ datasheets/conformers.pdf>.  D. Roy, J. Fendler, Reﬂection and absorption techniques for optical  J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made characterization of chemically assembled nanomaterials, Adv. Mater. 16 simple, Phys. Rev. Lett. 77 (1996) 3865–3868. (2004) 479–508.  A. Lasagabaster, M.J. Abad, L. Barral, A. Ares, FTIR study on the nature of water  V.K. Parashar, J.B. Orhan, A. Sayah, M. Cantoni, M.A.M. Gijs, Borosilicate sorbed in polypropylene (PP)/ethylene alcohol vinyl (EVOH) ﬁlms, Eur. Polym. nanoparticles prepared by exothermic phase separation, Nat. Nanotechnol. 3 J. 42 (2006) 3121–3132. (2008) 589–594.  M.A. Munoz, C. Carmona, M. Balon, FTIR study of water clusters in water–  S. Hinds, B.J. Taft, L. Levina, V. Sukhovatkin, C.J. Dooley, M.D. Roy, D.D. MacNeil, triethylamine solutions, Chem. Phys. 335 (2007) 37–42. E.H. Sargent, S.O. Kelly, Nucleotide-directed growth of semiconductor nanocrystals, J. Am. Chem. Soc. 128 (2006) 64–65.
Pages to are hidden for
"FTIR spectroscopy as a tool for nano-material characterization"Please download to view full document