- Email: [email protected]
Diameter sensitive effect in singlewalled carbon nanotubes upon acid treatment
Journal of Alloys and Compounds 486 (2009) 386–390
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jallcom
Diameter sensitive effect in singlewalled carbon nanotubes upon acid treatment S. Costa ∗ , E. Borowiak-Palen Centre of Knowledge Based Nanomaterials and Technologies, Institute of Chemical and Environment Engineering, West Pomeranian University of Technology, Szczecin, Poland
a r t i c l e
i n f o
Article history: Received 13 May 2009 Received in revised form 23 June 2009 Accepted 24 June 2009 Available online 4 July 2009 Keywords: Nanostructures Laser processing Vapour deposition Thermal analysis Photoelectron spectroscopies
a b s t r a c t Singlewalled carbon nanotubes (SWCNT) exhibit very unique properties. As an electronic system they undergo amphoteric doping effects (n-type and p-type) which can be reversed. These processes affect the optical and vibronic properties of the carbon nanotubes. The most common and widely used procedure which changes the properties of the SWCNT is acid treatment applied as a purification procedure. This effect has been widely studied but not fully understood so far. Here, we present a study, in which a diameter sensitive effect has been observed. Therefore, two kinds of SWCNT samples have been studied: (i) produced via chemical vapour deposition with a broad diameter distribution, and (ii) synthesised by the laser ablation technique which is commonly known to result in narrow diameter distribution bulk SWCNT samples. Resonance Raman spectroscopy, optical absorption spectroscopy, and Fourier transform middle-infrared spectroscopy have been applied for the characterisation of the samples. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Carbon nanotubes (CNT) have been investigated very intensely over the last years. Singlewalled carbon nanotubes (SWCNT), in particular, are currently the aim of research investigations due to their multiple interesting properties. SWCNT can be synthesised by different routes, being the most used arc discharge, laser ablation (LA), and chemical vapour deposition (CVD) [1]. When a metallic catalyst is required for the synthesis process the sample purity becomes a problem and further purification steps are mandatory in order to remove the catalyst particles. To obtain the bulk scale purity, acid treatments have often been reported e.g. nitric acid reflux [2–4]. However, this procedure leads to the formation of functionalised products in which the electronic properties are modified as molecules can penetrate the graphite layer, generating defects and charge transfer [5]. A characteristic behaviour of SWCNT is that they form bundles due to Van der Waals inter-tube bonding [6–9]. When the dopant molecules are intercalated into the SWCNT bundles an expansion of the inter-tube spacing causing changes in the diameter of the nanotubes can be observed. Kukovecz et al. have shown that the doping process is diameter selective and that the graphite lattice can expand to accommodate the molecules of the dopant [5]. To follow this behaviour in greater detail, spectroscopic studies, such as resonance Raman response and optical absorption spectroscopy (OAS), have been carried out. The Fourier transform infrared tech-
∗ Corresponding author at: Szczecin University of Technology, Pulaskiego St. 10, 70-322 Szczecin, Poland. Tel.: +48 91 449 48 72; fax: +48 91 449 46 86. E-mail address: [email protected] (S. Costa). 0925-8388/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2009.06.177
nique was employed to detect the formation of new functional groups on the surface of the nanotubes upon the acid treatment. These tools are very powerful to understand the changes occurring in the optical and vibronic properties of the functionalised SWCNT in respect to the pristine material [10–13]. Many efforts have been made to understand these modifications. Nevertheless, the full understanding of this field still remains a challenge. The importance of this study comes from the fact that carbon nanotubes are now being widely considered for medical applications and the acid purified samples are often used for further bio-functionalisation. Therefore, a better understanding of the influence of the purification procedure on the SWCNT properties is still essential. Here, we present a study on concentrated acid treated SWCNT which leads to a shortening of the C–C bonds in the carbon nanotube structure, and to a Fermi level shift which is reflected in drastic changes in the radial breathing mode region and in the response of the carbon nanotubes, respectively. These findings are observed for the selected tube diameters. 2. Experimental In the current work two different samples of SWCNT were studied. The first batch of SWCNT was produced by the CVD process (CVD-S1) [14]. The characteristic property of this sample is a broad diameter distribution (0.82–1.87 nm). The second batch of SWCNT was synthesised by a standard laser ablation technique [15,16], with a narrow diameter distribution (LA-S2 (1.22 ± 0.16 nm)). As a first purification step the samples were annealed in air at 300 ◦ C. Afterwards, the annealed samples were refluxed for 24 h in acid solution of diluted aqua regia composed of a mixture of (HNO3 :HCl):H2 O with a ratio of (1:3):4. Finally, the samples were filtered with a microfiltration system, and washed with distilled water and acetone. In order to remove any functional groups formed on the nanotube surface an annealing procedure in high vacuum (10−5 mbar) at 800 ◦ C was performed. A Renishaw in via Raman microscope spectrometer ( = 785 nm), a Jasco-570 UV–vis/NIR spectrophotometer,
S. Costa, E. Borowiak-Palen / Journal of Alloys and Compounds 486 (2009) 386–390 and a Nicolet-530 FTIR spectrophotometer were used. The samples were prepared by dispersing the products in acetone followed by a sonication step until a homogeneous suspension was reached. Subsequently, the suspension was dropped onto a heated KBr crystal (ca. 100 ◦ C). This procedure was repeated for pristine, purified and annealed samples. A typical SWCNT Raman spectrum contains the tangential mode G-band, which is reminiscent of in-plane modes in graphite, the disorder induced Dband, and the radial breathing mode (RBM), through which it is possible to calculate the diameter of the tubes [17]. The UV–vis/NIR spectra of SWCNT usually reveal the so called van Hove singularities (vHSs). The peaks observed are related to transitions between the densities of states (DOS) singularities in semiconducting and metallic tubes [18].
3. Results and discussion SWCNT have amazing electronic properties. By the introduction of electron donor molecules, referred to as n-type doping (e.g. K), or electron acceptor as p-type doping (e.g. Br2 ), into the hollows of the triangular SWCNT lattice, the Fermi level can be shifted resulting in modification of the properties without generating defects in the SWCNT bond network [5]. In our study, the dopant molecules (HCl and HNO3 ) are electron acceptors, and charge transfer from the SWCNT to the acceptor molecule is expected. Therefore, a shortening of the C–C bond which leads to upshifts in Raman modes e.g. in RBM features [1], can result. The RBM is composed of many features corresponding to the atomic vibration of the C atoms in the radial direction as if the tube was breathing. This mode is carbon nanotube diameter sensitive and it is very useful for the characterisation of the nanotube diameters up to 2 nm [17]. The resonance in Raman intensity depends on the density of electronic states available for the optical transitions, and this property is very important for one-dimensional (1D) systems, such as SWCNT. An observable Raman signal from a carbon nanotube can be obtained when the laser excitation energy is equal to the energy separation between vHSs in the valence and conduction bands, but only for allowed electronic transitions [17]. Therefore, the different laser energies result in different nanotube excitations, since their electronic structure varies with their chirality and diameter distribution. Under different excitation energies different types of nanotubes are in resonance. Here, the laser wavelength of 785 nm corresponds to the energy of 1.58 eV, and only carbon nanotubes with this value of the energy separation between vHSs will give a Raman response. All the Raman spectra were normalised at ca. 300 cm−1 . In Fig. 1a, the RBM spectra of the sample CVD-S1 before (solid line), after acid treatment (dashed line), and the acid treated sample after the annealing (dotted line) are presented. This sample exhibits a very wide diameter distribution, from 0.82 nm to 1.87 nm. Here,
387
one can observe two effects for the acid treated specimen: (i) a decrease of the intensity of all the components in general, and (ii) an upshift of the RBM modes of the acid treated sample in respect to the pristine one. For the lower wavenumber region an upshift of 17 cm−1 was found. This gradually dropped to about 8 cm−1 at 182 cm−1 , and to only ca. 2 cm−1 for the remaining RBM components at 204 cm−1 . In the case of the acid treated CVD-S1 sample, after the annealing process, the RBM position was almost fully reversed to its initial position, although a slight downshift of the position only in the lowest wavenumber region was observed (see dotted line in Fig. 1a). The intensity of the components of the RBM modes increased significantly but the total intensity of the starting material was not reached. Fig. 1b represents the curve of the Raman upshift of the RBM components of the acid treated sample versus the corresponding carbon nanotube diameters. It can easily be observed that the strongest shift occurs for tube diameters of 1.62 nm up to 1.87 nm. The shift gradually decreases with decrease in tube diameter, being less for the tubes with diameter of 1.36 nm and not very significant for tubes with diameter below 1.21 nm. Keeping in mind that the cocktail of nanotubes with a broad diameter distribution went into resonance here, it was possible to observe diameter sensitivity to the doping effects. Two important effects explain the upshifts in the Raman spectra: (i) the charge transfer from the nanotubes to the dopant molecules, and a consequence shortening of the C–C bonds [1], and (ii) an increase on the thickness of the SWCNT bundles [5], caused by the intercalation of the dopant molecules into the interstitial channels. Both effects are proportional to the diameter of the nanotubes. The reduction of the intensity of the RBM modes for the selected tubes has already been reported for the case of K and FeCl3 doped HiPCO SWCNT [19]. The observation was explained as a diameter selective doping process. The inter-tube channels in SWCNT bundles consisting of wider tubes are proportionally broader than those of thinner tubes. Therefore, the intercalation of the carbon nanotube cores and the interstitial channels is easier. However, the second effect of the acid treatment – exohedral functionalisation, should also be considered. The defects on the nanotube walls become an active site for the formation of –COOH/–OH groups (see later in Fig. 4b). These functional groups also interact with the carbon walls which could result in the stronger upshift of the RBM components corresponding to the wider tubes. This interaction results in the contraction of C–C bonds in the nanotube structure and it is more pronounced for the tubes more affected by the acid treatment, therefore those with a higher amount of functional groups. In this way, the bigger upshift in the case of the wider tubes can be explained by the higher number
Fig. 1. (a) RBM feature in Raman spectrum of CVD-S1 sample; (b) relation between the diameter of nanotubes and their values of the shifts of RBM components in CVD-S1.
388
S. Costa, E. Borowiak-Palen / Journal of Alloys and Compounds 486 (2009) 386–390
Fig. 2. RBM feature of Raman spectrum of LA-S2 sample; (b) relation between the diameter of nanotubes and the corresponding shifts of RBM components in LA-S2.
of defects, which leads to more functional groups formed on the sidewalls. In order to confirm these data the same experiments were performed on the sample produced via laser ablation. Fig. 2a presents the RBM modes of the acid treated sample LA-S2 (dashed line), in respect to the pristine sample (solid line), and the acid treated sample after the annealing (dotted line) with the diameter distribution from 1 nm to 1.63 nm. Here, one can observe similar effects to those described above. The intensity of all the peaks strongly decreases after acid treatment. However, after the annealing step, the RBM components shift back almost completely to the initial position detected for the pristine sample. In addition, Fig. 2b presents the curve with the diameter of the tubes and the corresponding upshift in RBM region for the acid treated sample in respect to the pristine one. The trend line clearly shows that the wider diameter tubes exhibited stronger upshifts while the smaller ones are much less affected by acid treatment. The D and G regions of Raman response of all the analysed samples and their G/D intensity ratios (as an inset) are presented in Fig. 3. As a measure of the relative sample quality the G and D bands were analysed. In particular, their intensity ratios (G/D) were calculated. In the case of the sample LA-S2 (Fig. 3b) an upshift of the
G-band position of 10 cm−1 was observed when the acid treatment was performed. Due to the introduction of functional groups on the graphitic walls an increase of the D-band intensity is clear. Afterwards, an annealing step was carried out and a downshift in the G-band was observed recovering the original position of the pristine material. In the inset of Fig. 3b one can see the representation of the G/D intensity ratios, and the higher quality of the annealed sample is clearly indicated by the much higher G/D ratio. This can be explained by a reduction of the amorphous carbon content which increased the purity of the sample. Moreover, the decrease of the number of defects in the nanotube walls during the annealing process also improves the quality of the material. It is also noteworthy the D-band intensity decrease after the annealing process, indicating once more a strong reduction of the number of defects of the nanotubes. In respect to the sample CVD-S1 (Fig. 3a), and based on the G/D ratio values, it can be easily noticed that the quality of the CVD-S1 sample, in regard to the relative number of the defects in the carbon walls, is not altered upon the acid treatment. However, after the annealing treatment the ratio strongly increased, and the D-band vanished. This is in agreement with the results of the previous sample. The increase of the D-band shows an increase of the defects in the walls of nanotubes by the presence
Fig. 3. Representation of G and D bands in Raman spectra on pristine (solid line), acid treated (dashed line) and annealed (dotted line) samples of (a) CVD-S1, (b) LA-S2. The insets show the relation between G and D-band intensity (G/D) ratios in the same samples.
S. Costa, E. Borowiak-Palen / Journal of Alloys and Compounds 486 (2009) 386–390
of functional groups. The annealing step provides a decrease in the D-band and consequently, the decrease of the number of functional groups, increasing the sample quality. However, the sample CVD-S1 shows a less doping sensitive G-band, once no upshift was observed after the acid treatment (Fig. 3a). The difference can be explained by the homogeneity of the CNT. The CVD process leads to materials less homogeneous, proved for example by the bigger diameter distribution, while samples prepared via laser ablation are more uniform. These various trends indicate that one has to be careful with the optimisation of the purification procedure via acid treatment for the carbon nanotubes prepared by different synthesis techniques. The purification of carbon nanotubes via acid treatment should be optimised individually for each sample, not being possible to define a universal protocol for the sample purification. The optical properties of acid treated samples were studied using the LA-S2 sample only. The choice of the sample was determined by the fact that the laser ablated material has narrower diameter distribution, presenting sharper values for the transition energies in the density of states. Therefore, the van Hove singularities are more pronounced and well defined in this sample (see Fig. 4a). Fig. 4a indicates the optical response of the modified sample (dashed line), in comparison with the reference sample (solid line). The acid treated sample is strongly influenced upon the interaction with the foreign molecules. The sample presents S and E S , related to transitions between the DOS sinthe peaks E11 22 gularities, in semiconducting tubes. However, the intensity of the S S E11 and E22 peaks, corresponding to the transitions between the vHS in semiconducting tubes, decreases after the acid treatment S peaks shows (Fig. 4a). The calculation of the areas under the E11 that its intensity in the acid treated sample decreases and it is four times smaller when compared to the pristine sample. The initial intensity of the peaks is partially recovered after the annealing procedure. The introduction of the electron acceptors into the hollows of the triangular SWCNT lattice, and the functional groups onto their surface shift the Fermi level of the carbon nanotubes, and the corresponding electronic properties are modified [5]. This generates the case that some of the optically allowed transitions in the pristine system become forbidden. In addition, the inten-
389
sity of the remaining peaks dropped. However, this transition can be recovered during the high temperature annealing at 600 ◦ C for 1 h (2.5 × 10−3 mbar) of the sample (dotted line of Fig. 4a). A supporting experiment of in situ annealing of the LA-S2 sample in the Raman microscope was performed. The shift of ca. 10 cm−1 in the RBM modes of the acid treated sample was fully recovered (data not shown here). The decrease in the intensity of the optical absorption spectra in both samples can be assigned to the introduction of dopant molecules, such as HNO3 , which leads to oxidation [6,7], or 1,3cycloaddition [9], by the attachment of carboxylic groups (–COOH) formed in the graphite layer. Modifications in the SWCNT structure can be explained by the H-bonding promoted by –COOH groups as suggested by Kukovecz et al. [7]. One of the main sources of reactivity of SWCNT is the curvature given by the non-planar geometry of sp2 carbon atoms. When the oxidation occurs a certain number of carbon atoms are forced to change into the sp3 state, leading to structural modifications [5–8,20]. During the thermal treatment (decarboxylation) a fraction of these atoms changes back to the sp2 state, recovering the pristine properties. This fact explains why the optical and vibronic response of annealed samples is similar to the pristine one. To confirm the presence of the functional groups on the SWCNT surface measurements via infrared spectroscopy were performed. Therefore, the FTIR spectra of pristine (dashed line), acid treated (solid line), and annealed (dotted line) samples of LA-S2 are presented in Fig. 4b. It can be observed that functional groups are not present in the pristine material. On acid treatment, the existence of vibrational modes corresponding to C C bonds at ∼1510 cm−1 is detected. The peak at 1375 cm−1 is assigned to the nanotube phonon modes as well as to C–O groups. The 1610 cm−1 band was tentatively assigned to the C O vibration of carboxylic groups taking part in intermolecular H-bonding, which might also be responsible for the increased stacking of oxidized SWCNT. A small shoulder at 1746 cm−1 , corresponding to the C O stretching vibrations of the –COOH groups formed on the nanotubes upon oxidation during the acid treatment of the samples was also observed. After annealing the sample in high vacuum the bands
Fig. 4. (a) OAS spectra (normalized and vertically translated) on pristine, acid treated and annealed SWCNT of LA-S2 sample; (b) FTIR spectra of pristine, acid treated and annealed sample of LA-S2.
390
S. Costa, E. Borowiak-Palen / Journal of Alloys and Compounds 486 (2009) 386–390
corresponding to the functional groups vanished. This confirmed the successful removal of the groups attached in the nanotubes surface, improving the purity of the sample as well, confirmed by the increase of G/D ratio (see inset of Fig. 2). 4. Conclusions Concluding, the study of the influence of acid treatment on the optical and vibronic properties of SWCNT with different diameter distributions, and produced via different techniques (CVD and LA routes), was presented. The data presented clearly showed that the diameter sensitive effect was observed. For wider carbon nanotubes stronger upshifts of RBM components were detected. In addition, the concentrated acid treatment led to modifications in the Fermi level and changed strongly the features of the OAS spectra of the samples. The high temperature annealing reversed these effects, confirmed by the RBM modes in the Raman spectra, and the optical response of SWCNT, which returned to their initial state. The relative purity of the sample in respect to the pristine tubes did not change consistently for both batches of the tubes. However, after the annealing step in high vacuum the number of defects strongly decreased and the relative purity increased significantly. Acknowledgements The authors are grateful to Oliver Jost for delivery of the sample LA-S2 and to Peter Carrott for the help during the manuscript preparation. Research was sponsored by Polish State Committee for Scientific Research grant 1 T09B 009 30 (EB-P) and by the European Community through the Marie Curie Reasearch Training Network CARBIO under Contract MRTN-CT-2006-035616.
References [1] A. Barreiro, D. Selbmann, T. Pichler, K. Biedermann, T. Gemming, M.H. Rümmeli, U. Schwalke, B. Büchner, Appl. Phys. A 82 (2006) 719–725. [2] B. Bendjemil, E. Borowiak-Palen, A. Graff, T. Pichler, M. Guerioune, J. Fink, M. Knupfer, Appl. Phys. A 78 (2004) 311–314. [3] L.A. Montoro, J.M. Rosolen, Carbon 44 (2006) 3293–3301. [4] R. Schonfelder, M.H. Rummeli, W. Gruner, M. Loffler, J. Acker, V. Hoffmann, T. Gemming, B. Buchner, T. Pichler, Nanotechnology 18 (2007) 375601. [5] A. Kukovecz, T. Pichler, R. Pfeiffer, C. Kramberger, H. Kuzmany, Phys. Chem. Chem. Phys. 5 (2003) 582–587. [6] C. Bower, A. Kleinhammes, Y. Wu, O. Zhou, Chem. Phys. Lett. 288 (1998) 481–486. [7] A. Kukovecz, C. Kramberger, M. Holzinger, H. Kuzmany, J. Schalko, M. Mannsberger, A. Hirsch, J. Phys. Chem. B 106 (2002) 6374–6380. [8] H. Kuzmany, A. Kukovecz, F. Simon, M. Holzweber, C. Kramberger, T. Pichler, Synth. Met. 141 (2004) 113–122. [9] L. Petaccia, A. Goldoni, S. Lizzit, R. Larciprete, J. Electron. Spectrosc. Relat. Phenom. 144–147 (2005) 793–797. [10] M.S. Dresselhaus, G. Dresselhaus, R. Saito, A. Jorio, Phys. Rep. 409 (2005) 47–99. [11] X. Liu, T. Pichler, M. Knupfer, J. Fink, Phys. Rev. B 67 (2003) 125403. [12] T. Pichler, A. Kukovecz, H. Kuzmany, H. Kataura, Synth. Met. 135–136 (2003) 717–719. [13] A.M. Rao, E. Richter, P.C. Eklund, S. Bandow, Thin Solid Films 331 (1998) 141–147. [14] A. Bachmatiuk, E. Borowiak-Palen, M.H. Rummeli, C. Kramberger, H.W. Hubers, T. Gemming, T. Pichler, R.J. Kalenczuk, Nanotechnology 18 (2007) 275610. [15] E. Borowiak-Palen, T. Pichler, X. Liu, M. Knupfer, A. Graff, O. Jost, W. Pompe, R.J. Kalenczuk, J. Fink, Chem. Phys. Lett. 363 (2002) 567–572. [16] A. Thess, R. Lee, P. Nikolaev, H. Dai, P. Petit, J. Robert, C. Xu, Y.H. Lee, S.G. Kim, A.G. Rinzler, D.T. Colbert, G.E. Scuseria, D. Tomanek, J.E. Fischer, R.E. Smalley, Science 273 (1996) 483–487. [17] A. Jorio, M.A. Pimenta, A.G.S. Filho, R. Saito, G. Dresselhaus, M.S. Dresselhaus, New J. Phys. 5 (2003) 139–1139. [18] X. Liu, T. Pichler, M. Knupfer, M.S. Golden, J. Fink, H. Kataura, Y. Achiba, Phys. Rev. B 66 (2002) 045411. [19] A. Kukovecz, T. Pichler, R. Pfeiffer, H. Kuzmany, Chem. Commun. (2002) 1730–1731. [20] H. Kuzmany, A. Kukovecz, C. Kramberger, T. Pichler, M. Holzinger, H. Kataura, Synth. Met. 135–136 (2003) 791–794.
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jallcom
Diameter sensitive effect in singlewalled carbon nanotubes upon acid treatment S. Costa ∗ , E. Borowiak-Palen Centre of Knowledge Based Nanomaterials and Technologies, Institute of Chemical and Environment Engineering, West Pomeranian University of Technology, Szczecin, Poland
a r t i c l e
i n f o
Article history: Received 13 May 2009 Received in revised form 23 June 2009 Accepted 24 June 2009 Available online 4 July 2009 Keywords: Nanostructures Laser processing Vapour deposition Thermal analysis Photoelectron spectroscopies
a b s t r a c t Singlewalled carbon nanotubes (SWCNT) exhibit very unique properties. As an electronic system they undergo amphoteric doping effects (n-type and p-type) which can be reversed. These processes affect the optical and vibronic properties of the carbon nanotubes. The most common and widely used procedure which changes the properties of the SWCNT is acid treatment applied as a purification procedure. This effect has been widely studied but not fully understood so far. Here, we present a study, in which a diameter sensitive effect has been observed. Therefore, two kinds of SWCNT samples have been studied: (i) produced via chemical vapour deposition with a broad diameter distribution, and (ii) synthesised by the laser ablation technique which is commonly known to result in narrow diameter distribution bulk SWCNT samples. Resonance Raman spectroscopy, optical absorption spectroscopy, and Fourier transform middle-infrared spectroscopy have been applied for the characterisation of the samples. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Carbon nanotubes (CNT) have been investigated very intensely over the last years. Singlewalled carbon nanotubes (SWCNT), in particular, are currently the aim of research investigations due to their multiple interesting properties. SWCNT can be synthesised by different routes, being the most used arc discharge, laser ablation (LA), and chemical vapour deposition (CVD) [1]. When a metallic catalyst is required for the synthesis process the sample purity becomes a problem and further purification steps are mandatory in order to remove the catalyst particles. To obtain the bulk scale purity, acid treatments have often been reported e.g. nitric acid reflux [2–4]. However, this procedure leads to the formation of functionalised products in which the electronic properties are modified as molecules can penetrate the graphite layer, generating defects and charge transfer [5]. A characteristic behaviour of SWCNT is that they form bundles due to Van der Waals inter-tube bonding [6–9]. When the dopant molecules are intercalated into the SWCNT bundles an expansion of the inter-tube spacing causing changes in the diameter of the nanotubes can be observed. Kukovecz et al. have shown that the doping process is diameter selective and that the graphite lattice can expand to accommodate the molecules of the dopant [5]. To follow this behaviour in greater detail, spectroscopic studies, such as resonance Raman response and optical absorption spectroscopy (OAS), have been carried out. The Fourier transform infrared tech-
∗ Corresponding author at: Szczecin University of Technology, Pulaskiego St. 10, 70-322 Szczecin, Poland. Tel.: +48 91 449 48 72; fax: +48 91 449 46 86. E-mail address: [email protected] (S. Costa). 0925-8388/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2009.06.177
nique was employed to detect the formation of new functional groups on the surface of the nanotubes upon the acid treatment. These tools are very powerful to understand the changes occurring in the optical and vibronic properties of the functionalised SWCNT in respect to the pristine material [10–13]. Many efforts have been made to understand these modifications. Nevertheless, the full understanding of this field still remains a challenge. The importance of this study comes from the fact that carbon nanotubes are now being widely considered for medical applications and the acid purified samples are often used for further bio-functionalisation. Therefore, a better understanding of the influence of the purification procedure on the SWCNT properties is still essential. Here, we present a study on concentrated acid treated SWCNT which leads to a shortening of the C–C bonds in the carbon nanotube structure, and to a Fermi level shift which is reflected in drastic changes in the radial breathing mode region and in the response of the carbon nanotubes, respectively. These findings are observed for the selected tube diameters. 2. Experimental In the current work two different samples of SWCNT were studied. The first batch of SWCNT was produced by the CVD process (CVD-S1) [14]. The characteristic property of this sample is a broad diameter distribution (0.82–1.87 nm). The second batch of SWCNT was synthesised by a standard laser ablation technique [15,16], with a narrow diameter distribution (LA-S2 (1.22 ± 0.16 nm)). As a first purification step the samples were annealed in air at 300 ◦ C. Afterwards, the annealed samples were refluxed for 24 h in acid solution of diluted aqua regia composed of a mixture of (HNO3 :HCl):H2 O with a ratio of (1:3):4. Finally, the samples were filtered with a microfiltration system, and washed with distilled water and acetone. In order to remove any functional groups formed on the nanotube surface an annealing procedure in high vacuum (10−5 mbar) at 800 ◦ C was performed. A Renishaw in via Raman microscope spectrometer ( = 785 nm), a Jasco-570 UV–vis/NIR spectrophotometer,
S. Costa, E. Borowiak-Palen / Journal of Alloys and Compounds 486 (2009) 386–390 and a Nicolet-530 FTIR spectrophotometer were used. The samples were prepared by dispersing the products in acetone followed by a sonication step until a homogeneous suspension was reached. Subsequently, the suspension was dropped onto a heated KBr crystal (ca. 100 ◦ C). This procedure was repeated for pristine, purified and annealed samples. A typical SWCNT Raman spectrum contains the tangential mode G-band, which is reminiscent of in-plane modes in graphite, the disorder induced Dband, and the radial breathing mode (RBM), through which it is possible to calculate the diameter of the tubes [17]. The UV–vis/NIR spectra of SWCNT usually reveal the so called van Hove singularities (vHSs). The peaks observed are related to transitions between the densities of states (DOS) singularities in semiconducting and metallic tubes [18].
3. Results and discussion SWCNT have amazing electronic properties. By the introduction of electron donor molecules, referred to as n-type doping (e.g. K), or electron acceptor as p-type doping (e.g. Br2 ), into the hollows of the triangular SWCNT lattice, the Fermi level can be shifted resulting in modification of the properties without generating defects in the SWCNT bond network [5]. In our study, the dopant molecules (HCl and HNO3 ) are electron acceptors, and charge transfer from the SWCNT to the acceptor molecule is expected. Therefore, a shortening of the C–C bond which leads to upshifts in Raman modes e.g. in RBM features [1], can result. The RBM is composed of many features corresponding to the atomic vibration of the C atoms in the radial direction as if the tube was breathing. This mode is carbon nanotube diameter sensitive and it is very useful for the characterisation of the nanotube diameters up to 2 nm [17]. The resonance in Raman intensity depends on the density of electronic states available for the optical transitions, and this property is very important for one-dimensional (1D) systems, such as SWCNT. An observable Raman signal from a carbon nanotube can be obtained when the laser excitation energy is equal to the energy separation between vHSs in the valence and conduction bands, but only for allowed electronic transitions [17]. Therefore, the different laser energies result in different nanotube excitations, since their electronic structure varies with their chirality and diameter distribution. Under different excitation energies different types of nanotubes are in resonance. Here, the laser wavelength of 785 nm corresponds to the energy of 1.58 eV, and only carbon nanotubes with this value of the energy separation between vHSs will give a Raman response. All the Raman spectra were normalised at ca. 300 cm−1 . In Fig. 1a, the RBM spectra of the sample CVD-S1 before (solid line), after acid treatment (dashed line), and the acid treated sample after the annealing (dotted line) are presented. This sample exhibits a very wide diameter distribution, from 0.82 nm to 1.87 nm. Here,
387
one can observe two effects for the acid treated specimen: (i) a decrease of the intensity of all the components in general, and (ii) an upshift of the RBM modes of the acid treated sample in respect to the pristine one. For the lower wavenumber region an upshift of 17 cm−1 was found. This gradually dropped to about 8 cm−1 at 182 cm−1 , and to only ca. 2 cm−1 for the remaining RBM components at 204 cm−1 . In the case of the acid treated CVD-S1 sample, after the annealing process, the RBM position was almost fully reversed to its initial position, although a slight downshift of the position only in the lowest wavenumber region was observed (see dotted line in Fig. 1a). The intensity of the components of the RBM modes increased significantly but the total intensity of the starting material was not reached. Fig. 1b represents the curve of the Raman upshift of the RBM components of the acid treated sample versus the corresponding carbon nanotube diameters. It can easily be observed that the strongest shift occurs for tube diameters of 1.62 nm up to 1.87 nm. The shift gradually decreases with decrease in tube diameter, being less for the tubes with diameter of 1.36 nm and not very significant for tubes with diameter below 1.21 nm. Keeping in mind that the cocktail of nanotubes with a broad diameter distribution went into resonance here, it was possible to observe diameter sensitivity to the doping effects. Two important effects explain the upshifts in the Raman spectra: (i) the charge transfer from the nanotubes to the dopant molecules, and a consequence shortening of the C–C bonds [1], and (ii) an increase on the thickness of the SWCNT bundles [5], caused by the intercalation of the dopant molecules into the interstitial channels. Both effects are proportional to the diameter of the nanotubes. The reduction of the intensity of the RBM modes for the selected tubes has already been reported for the case of K and FeCl3 doped HiPCO SWCNT [19]. The observation was explained as a diameter selective doping process. The inter-tube channels in SWCNT bundles consisting of wider tubes are proportionally broader than those of thinner tubes. Therefore, the intercalation of the carbon nanotube cores and the interstitial channels is easier. However, the second effect of the acid treatment – exohedral functionalisation, should also be considered. The defects on the nanotube walls become an active site for the formation of –COOH/–OH groups (see later in Fig. 4b). These functional groups also interact with the carbon walls which could result in the stronger upshift of the RBM components corresponding to the wider tubes. This interaction results in the contraction of C–C bonds in the nanotube structure and it is more pronounced for the tubes more affected by the acid treatment, therefore those with a higher amount of functional groups. In this way, the bigger upshift in the case of the wider tubes can be explained by the higher number
Fig. 1. (a) RBM feature in Raman spectrum of CVD-S1 sample; (b) relation between the diameter of nanotubes and their values of the shifts of RBM components in CVD-S1.
388
S. Costa, E. Borowiak-Palen / Journal of Alloys and Compounds 486 (2009) 386–390
Fig. 2. RBM feature of Raman spectrum of LA-S2 sample; (b) relation between the diameter of nanotubes and the corresponding shifts of RBM components in LA-S2.
of defects, which leads to more functional groups formed on the sidewalls. In order to confirm these data the same experiments were performed on the sample produced via laser ablation. Fig. 2a presents the RBM modes of the acid treated sample LA-S2 (dashed line), in respect to the pristine sample (solid line), and the acid treated sample after the annealing (dotted line) with the diameter distribution from 1 nm to 1.63 nm. Here, one can observe similar effects to those described above. The intensity of all the peaks strongly decreases after acid treatment. However, after the annealing step, the RBM components shift back almost completely to the initial position detected for the pristine sample. In addition, Fig. 2b presents the curve with the diameter of the tubes and the corresponding upshift in RBM region for the acid treated sample in respect to the pristine one. The trend line clearly shows that the wider diameter tubes exhibited stronger upshifts while the smaller ones are much less affected by acid treatment. The D and G regions of Raman response of all the analysed samples and their G/D intensity ratios (as an inset) are presented in Fig. 3. As a measure of the relative sample quality the G and D bands were analysed. In particular, their intensity ratios (G/D) were calculated. In the case of the sample LA-S2 (Fig. 3b) an upshift of the
G-band position of 10 cm−1 was observed when the acid treatment was performed. Due to the introduction of functional groups on the graphitic walls an increase of the D-band intensity is clear. Afterwards, an annealing step was carried out and a downshift in the G-band was observed recovering the original position of the pristine material. In the inset of Fig. 3b one can see the representation of the G/D intensity ratios, and the higher quality of the annealed sample is clearly indicated by the much higher G/D ratio. This can be explained by a reduction of the amorphous carbon content which increased the purity of the sample. Moreover, the decrease of the number of defects in the nanotube walls during the annealing process also improves the quality of the material. It is also noteworthy the D-band intensity decrease after the annealing process, indicating once more a strong reduction of the number of defects of the nanotubes. In respect to the sample CVD-S1 (Fig. 3a), and based on the G/D ratio values, it can be easily noticed that the quality of the CVD-S1 sample, in regard to the relative number of the defects in the carbon walls, is not altered upon the acid treatment. However, after the annealing treatment the ratio strongly increased, and the D-band vanished. This is in agreement with the results of the previous sample. The increase of the D-band shows an increase of the defects in the walls of nanotubes by the presence
Fig. 3. Representation of G and D bands in Raman spectra on pristine (solid line), acid treated (dashed line) and annealed (dotted line) samples of (a) CVD-S1, (b) LA-S2. The insets show the relation between G and D-band intensity (G/D) ratios in the same samples.
S. Costa, E. Borowiak-Palen / Journal of Alloys and Compounds 486 (2009) 386–390
of functional groups. The annealing step provides a decrease in the D-band and consequently, the decrease of the number of functional groups, increasing the sample quality. However, the sample CVD-S1 shows a less doping sensitive G-band, once no upshift was observed after the acid treatment (Fig. 3a). The difference can be explained by the homogeneity of the CNT. The CVD process leads to materials less homogeneous, proved for example by the bigger diameter distribution, while samples prepared via laser ablation are more uniform. These various trends indicate that one has to be careful with the optimisation of the purification procedure via acid treatment for the carbon nanotubes prepared by different synthesis techniques. The purification of carbon nanotubes via acid treatment should be optimised individually for each sample, not being possible to define a universal protocol for the sample purification. The optical properties of acid treated samples were studied using the LA-S2 sample only. The choice of the sample was determined by the fact that the laser ablated material has narrower diameter distribution, presenting sharper values for the transition energies in the density of states. Therefore, the van Hove singularities are more pronounced and well defined in this sample (see Fig. 4a). Fig. 4a indicates the optical response of the modified sample (dashed line), in comparison with the reference sample (solid line). The acid treated sample is strongly influenced upon the interaction with the foreign molecules. The sample presents S and E S , related to transitions between the DOS sinthe peaks E11 22 gularities, in semiconducting tubes. However, the intensity of the S S E11 and E22 peaks, corresponding to the transitions between the vHS in semiconducting tubes, decreases after the acid treatment S peaks shows (Fig. 4a). The calculation of the areas under the E11 that its intensity in the acid treated sample decreases and it is four times smaller when compared to the pristine sample. The initial intensity of the peaks is partially recovered after the annealing procedure. The introduction of the electron acceptors into the hollows of the triangular SWCNT lattice, and the functional groups onto their surface shift the Fermi level of the carbon nanotubes, and the corresponding electronic properties are modified [5]. This generates the case that some of the optically allowed transitions in the pristine system become forbidden. In addition, the inten-
389
sity of the remaining peaks dropped. However, this transition can be recovered during the high temperature annealing at 600 ◦ C for 1 h (2.5 × 10−3 mbar) of the sample (dotted line of Fig. 4a). A supporting experiment of in situ annealing of the LA-S2 sample in the Raman microscope was performed. The shift of ca. 10 cm−1 in the RBM modes of the acid treated sample was fully recovered (data not shown here). The decrease in the intensity of the optical absorption spectra in both samples can be assigned to the introduction of dopant molecules, such as HNO3 , which leads to oxidation [6,7], or 1,3cycloaddition [9], by the attachment of carboxylic groups (–COOH) formed in the graphite layer. Modifications in the SWCNT structure can be explained by the H-bonding promoted by –COOH groups as suggested by Kukovecz et al. [7]. One of the main sources of reactivity of SWCNT is the curvature given by the non-planar geometry of sp2 carbon atoms. When the oxidation occurs a certain number of carbon atoms are forced to change into the sp3 state, leading to structural modifications [5–8,20]. During the thermal treatment (decarboxylation) a fraction of these atoms changes back to the sp2 state, recovering the pristine properties. This fact explains why the optical and vibronic response of annealed samples is similar to the pristine one. To confirm the presence of the functional groups on the SWCNT surface measurements via infrared spectroscopy were performed. Therefore, the FTIR spectra of pristine (dashed line), acid treated (solid line), and annealed (dotted line) samples of LA-S2 are presented in Fig. 4b. It can be observed that functional groups are not present in the pristine material. On acid treatment, the existence of vibrational modes corresponding to C C bonds at ∼1510 cm−1 is detected. The peak at 1375 cm−1 is assigned to the nanotube phonon modes as well as to C–O groups. The 1610 cm−1 band was tentatively assigned to the C O vibration of carboxylic groups taking part in intermolecular H-bonding, which might also be responsible for the increased stacking of oxidized SWCNT. A small shoulder at 1746 cm−1 , corresponding to the C O stretching vibrations of the –COOH groups formed on the nanotubes upon oxidation during the acid treatment of the samples was also observed. After annealing the sample in high vacuum the bands
Fig. 4. (a) OAS spectra (normalized and vertically translated) on pristine, acid treated and annealed SWCNT of LA-S2 sample; (b) FTIR spectra of pristine, acid treated and annealed sample of LA-S2.
390
S. Costa, E. Borowiak-Palen / Journal of Alloys and Compounds 486 (2009) 386–390
corresponding to the functional groups vanished. This confirmed the successful removal of the groups attached in the nanotubes surface, improving the purity of the sample as well, confirmed by the increase of G/D ratio (see inset of Fig. 2). 4. Conclusions Concluding, the study of the influence of acid treatment on the optical and vibronic properties of SWCNT with different diameter distributions, and produced via different techniques (CVD and LA routes), was presented. The data presented clearly showed that the diameter sensitive effect was observed. For wider carbon nanotubes stronger upshifts of RBM components were detected. In addition, the concentrated acid treatment led to modifications in the Fermi level and changed strongly the features of the OAS spectra of the samples. The high temperature annealing reversed these effects, confirmed by the RBM modes in the Raman spectra, and the optical response of SWCNT, which returned to their initial state. The relative purity of the sample in respect to the pristine tubes did not change consistently for both batches of the tubes. However, after the annealing step in high vacuum the number of defects strongly decreased and the relative purity increased significantly. Acknowledgements The authors are grateful to Oliver Jost for delivery of the sample LA-S2 and to Peter Carrott for the help during the manuscript preparation. Research was sponsored by Polish State Committee for Scientific Research grant 1 T09B 009 30 (EB-P) and by the European Community through the Marie Curie Reasearch Training Network CARBIO under Contract MRTN-CT-2006-035616.
References [1] A. Barreiro, D. Selbmann, T. Pichler, K. Biedermann, T. Gemming, M.H. Rümmeli, U. Schwalke, B. Büchner, Appl. Phys. A 82 (2006) 719–725. [2] B. Bendjemil, E. Borowiak-Palen, A. Graff, T. Pichler, M. Guerioune, J. Fink, M. Knupfer, Appl. Phys. A 78 (2004) 311–314. [3] L.A. Montoro, J.M. Rosolen, Carbon 44 (2006) 3293–3301. [4] R. Schonfelder, M.H. Rummeli, W. Gruner, M. Loffler, J. Acker, V. Hoffmann, T. Gemming, B. Buchner, T. Pichler, Nanotechnology 18 (2007) 375601. [5] A. Kukovecz, T. Pichler, R. Pfeiffer, C. Kramberger, H. Kuzmany, Phys. Chem. Chem. Phys. 5 (2003) 582–587. [6] C. Bower, A. Kleinhammes, Y. Wu, O. Zhou, Chem. Phys. Lett. 288 (1998) 481–486. [7] A. Kukovecz, C. Kramberger, M. Holzinger, H. Kuzmany, J. Schalko, M. Mannsberger, A. Hirsch, J. Phys. Chem. B 106 (2002) 6374–6380. [8] H. Kuzmany, A. Kukovecz, F. Simon, M. Holzweber, C. Kramberger, T. Pichler, Synth. Met. 141 (2004) 113–122. [9] L. Petaccia, A. Goldoni, S. Lizzit, R. Larciprete, J. Electron. Spectrosc. Relat. Phenom. 144–147 (2005) 793–797. [10] M.S. Dresselhaus, G. Dresselhaus, R. Saito, A. Jorio, Phys. Rep. 409 (2005) 47–99. [11] X. Liu, T. Pichler, M. Knupfer, J. Fink, Phys. Rev. B 67 (2003) 125403. [12] T. Pichler, A. Kukovecz, H. Kuzmany, H. Kataura, Synth. Met. 135–136 (2003) 717–719. [13] A.M. Rao, E. Richter, P.C. Eklund, S. Bandow, Thin Solid Films 331 (1998) 141–147. [14] A. Bachmatiuk, E. Borowiak-Palen, M.H. Rummeli, C. Kramberger, H.W. Hubers, T. Gemming, T. Pichler, R.J. Kalenczuk, Nanotechnology 18 (2007) 275610. [15] E. Borowiak-Palen, T. Pichler, X. Liu, M. Knupfer, A. Graff, O. Jost, W. Pompe, R.J. Kalenczuk, J. Fink, Chem. Phys. Lett. 363 (2002) 567–572. [16] A. Thess, R. Lee, P. Nikolaev, H. Dai, P. Petit, J. Robert, C. Xu, Y.H. Lee, S.G. Kim, A.G. Rinzler, D.T. Colbert, G.E. Scuseria, D. Tomanek, J.E. Fischer, R.E. Smalley, Science 273 (1996) 483–487. [17] A. Jorio, M.A. Pimenta, A.G.S. Filho, R. Saito, G. Dresselhaus, M.S. Dresselhaus, New J. Phys. 5 (2003) 139–1139. [18] X. Liu, T. Pichler, M. Knupfer, M.S. Golden, J. Fink, H. Kataura, Y. Achiba, Phys. Rev. B 66 (2002) 045411. [19] A. Kukovecz, T. Pichler, R. Pfeiffer, H. Kuzmany, Chem. Commun. (2002) 1730–1731. [20] H. Kuzmany, A. Kukovecz, C. Kramberger, T. Pichler, M. Holzinger, H. Kataura, Synth. Met. 135–136 (2003) 791–794.