7084
Langmuir 2003, 19, 7084-7088
Preferential Solubilization of Smaller Single-Walled Carbon Nanotubes in Sequential Functionalization Reactions Weijie Huang,† Shiral Fernando,† Yi Lin,† Bing Zhou,† Lawrence F. Allard,‡ and Ya-Ping Sun*,† Department of Chemistry and Center for Advanced Engineering Fibers and Films, Howard L. Hunter Chemistry Laboratory, Clemson University, Clemson, South Carolina 29634-0973, and High-Temperature Materials Laboratory, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6062 Received May 5, 2003 A purified single-walled carbon nanotube (SWNT) sample was fractionated in sequential functionalization reactions with a diamine-terminated oligomeric poly(ethylene glycol) to produce five soluble fractions and a final solid residue. These fractions and the residue were characterized by using optical spectroscopy, electron microscopy, thermal gravimetric analysis, and other techniques. The results show that the soluble fractions all contain primarily functionalized SWNTs and that there are no obviously systematic changes from one fraction to the next. However, one exception made evident by the Raman spectroscopy results is that the functionalization reactions preferentially solubilize the SWNTs of smaller diameters, resulting in an enrichment of larger diameter nanotubes in the final solid residue. In addition, the nanotube contents in different soluble fractions obtained from the sequential functionalization reactions are apparently different, with higher nanotube contents in the later fractions. The difference between the fractionation of SWNTs and the fractionation of multiple-walled carbon nanotubes (MWNTs) reported previously is discussed.
Introduction There has been significant recent progress in the development of methods for solubilizing both single-walled (SWNT)andmultiple-walled(MWNT)carbonnanotubes.1-5 The chemical functionalization of carbon nanotubes can be either direct addition to the graphitic walls4-8 or attachment of functional groups at the defect sites.9-16 The latter typically involves the nanotube-bound carboxylic acids, which allow the formation of ionic, ester, or † ‡
Clemson University. Oak Ridge National Laboratory.
(1) Sun, Y.-P.; Fu, K.; Lin, Y.; Huang, W. Acc. Chem. Res. 2002, 35, 1096. (2) Niyogi, S.; Hamon, M. A.; Hu, H.; Zhao, B.; Bhowmik, P.; Sen, R.; Itkis, M. E.; Haddon, R. C. Acc. Chem. Res. 2002, 35, 1105. (3) Sinnott, S. B. J. Nanosci. Nanotech. 2002, 2, 113. (4) Hirsch, A. Angew. Chem., Int. Ed. 2002, 41, 1853. (5) Bahr, J. L.; Tour, J. M. J. Mater. Chem. 2002, 12, 1952. (6) (a) Mickelson, E. T.; Huffman, C. B.; Rinzler, A. G.; Smalley, R. E.; Hauge, R. H. Chem. Phys. Lett. 1998, 296, 188. (b) Khabashesku, V. N.; Billups, W. E.; Margrave, J. L. Acc. Chem. Res. 2002, 35, 1087. (7) (a) Georgakilas, V.; Kordatos, K.; Prato, M.; Guldi, D. M.; Holzinger, M.; Hirsch, A. J. Am. Chem. Soc. 2002, 124, 760. (b) Tagmatarchis, N.; Georgakilas, V.; Prato, M.; Shinohara, H. Chem. Commun. 2002, 2010. (8) (a) Bahr, J. L.; Yang, J.; Kosynkin, D. V.; Bronikowskim, M. J.; Smalley, R. E.; Tour, J. M. J. Am. Chem. Soc. 2001, 123, 6536. (b) Bahr, J. L.; Tour, J. M. Chem. Mater. 2001, 13, 3823. (c) Dyke, C. A.; Tour, J. M. J. Am. Chem. Soc. 2003, 125, 1156. (9) Chen, J.; Hamon, M. A.; Hu, H.; Chen, Y.; Rao, A. M.; Eklund, P. C.; Haddon, R. C. Science 1998, 282, 95. (10) (a) Hamon, M. A.; Chen, J.; Hu, H.; Chen, Y.; Itkis, M. E.; Rao, A. M.; Eklund, P. C.; Haddon, R. C. Adv. Mater. 1999, 11, 834. (b) Niyogi, S.; Hu, H.; Hamon, M. A.; Bhowmik, P.; Zhao, B.; Rozenzhak, S. M.; Chen, J.; Itkis, M. E.; Meier, M. S.; Haddon, R. C. J. Am. Chem. Soc. 2001, 123, 733. (c) Zhao, B.; Hu, H.; Niyogi, S.; Itkis, M. E.; Hamon, M. A.; Bhowmik, P.; Meier, M. S.; Haddon, R. C. J. Am. Chem. Soc. 2001, 123, 11673. (d) Hamon, M. A.; Hu, H.; Bhowmik, P.; Itkis, M. E.; Haddon, R. C. Appl. Phys. A 2002, 74, 333. (11) Chen, J.; Rao, A. M.; Lyuksyutov, S.; Itkis, M. E.; Hamon, M. A.; Hu, H.; Cohn, R. W.; Eklund, P. C.; Colbert, D. T.; Smalley, R. E.; Haddon, R. C. J. Phys. Chem. B 2001, 105, 2525. (12) Riggs, J. E.; Guo, Z.; Carroll, D. L.; Sun, Y.-P. J. Am. Chem. Soc. 2000, 122, 5879.
amide linkages with various functionalities.1,2 An interesting issue is the potential selectivity in the solubilization of carbon nanotubes via functionalization reactions. For example, Prato and co-workers17 recently suggested that a HiPco SWNT sample could be purified in terms of the difference in solubility between the functionalized nanotubes and impurities. During the course of this study, Papadimitrakopoulos and co-workers18 reported that the solubilization of SWNTs in the functionalization reaction with octadecylamine is selective against metallic nanotubes, yielding a solubilized sample rich in semiconducting SWNTs. In an effort to examine the selectivity issue we have attempted to fractionate a MWNT sample in repeated functionalization reactions.16 The functionalization was based on the esterification of the nanotube-bound carboxylic acids. However, results from detailed characterizations of the different soluble fractions suggested that there was no obvious preferential solubilization of MWNTs in terms of their sizes and also no significant selectivity in the removal of carbon impurities. Here we report a similar attempt to fractionate a SWNT sample in se(13) (a) Sun, Y.-P.; Huang, W.; Lin, Y.; Fu, K.; Kitaygorodskiy, A.; Riddle, L. A.; Yu, Y. J.; Carroll, D. L. Chem. Mater. 2001, 13, 2864. (b) Fu, K.; Huang, W.; Lin, Y.; Riddle, L. A.; Carroll, D. L.; Sun, Y.-P. Nano Lett. 2001, 1, 439. (c) Qu, L.; Martin, R. B.; Huang, W.; Fu, K.; Zweifel, D.; Lin, Y.; Sun, Y.-P.; Bunker, C. E.; Harruff, B. A.; Gord, J. R.; Allard, L. F. J. Chem. Phys. 2002, 117, 8089. (14) (a) Huang, W.; Lin, Y.; Taylor, S.; Gaillard, J.; Rao, A. M.; Sun, Y.-P. Nano Lett. 2002, 2, 231. (b) Lin, Y.; Rao, A. M.; Sadanadan, B.; Kenik, E. A.; Sun, Y.-P. J. Phys. Chem. B 2002, 106, 1294. (c) Hill, D. E.; Lin, Y.; Rao, A. M.; Allard, L. F.; Sun, Y.-P. Macromolecules 2002, 35, 9466. (15) Huang, W.; Fernando, S.; Allard, L. F.; Sun, Y.-P. Nano Lett. 2003, 3, 565. (16) Lin, Y.; Taylor, S.; Huang, W.; Sun, Y.-P. J. Phys. Chem. B 2003, 107, 914. (17) Georgakilas, V.; Voulgaris, D.; Vazquez, E.; Prato, M.; Guldi, D. M.; Kukovecz, A.; Kuzmany, H. J. Am. Chem. Soc. 2002, 124, 14318. (18) Chattopadhyay, D.; Galeska, I.; Papadimitrakopoulos, F. J. Am. Chem. Soc. 2003, 125, 3370.
10.1021/la0301893 CCC: $25.00 © 2003 American Chemical Society Published on Web 06/19/2003
Preferential Solubilization of Smaller SWNTs
Langmuir, Vol. 19, No. 17, 2003 7085
Scheme 1
quential functionalization reactions. There are several noticeable differences between the SWNT and MWNT samples. The MWNT sample used in the fractionation attempt was produced via the chemical vapor deposition method, thus containing considerably less carbon impurities. MWNTs are also much more complicated than SWNTs with respect to the nanotube diameters. In fact, the results reported here suggest that the functionalization reaction is somewhat selective, resulting in an enrichment in SWNTs of larger diameters in the residue sample from the repeated reactions. Experimental Section Materials. Diamine-terminated oligomeric poly(ethylene glycol), H2NCH2CH2CH2(OCH2CH2)nCH2NH2, with n ∼ 35 (PEG1500N) was purchased from Aldrich and used after removal of the residual water via azeotropic distillation. All solvents were either spectrophotometry-HPLC-grade or purified via simple distillation. Deuterated solvents for NMR measurements were supplied by Cambridge Isotope Laboratories. The SWNT sample was produced in Professor A. M. Rao’s laboratory (Department of Physics and Astronomy, Clemson University) by using the arc-discharge method with Ni/Y as catalyst. In the purification, the sample was refluxed in an aqueous HNO3 solution (2.6 M) for 48 h. Upon centrifugation to remove the liquid phase, the remaining solids were washed repeatedly with deionized water until neutral pH and then dried under vacuum. Measurements. UV/vis/NIR absorption spectra were recorded on Shimadzu UV2101-PC and UV3100 spectrophotometers and a Thermo-Nicolet Nexus 670 FT-NIR spectrometer. Raman spectra were obtained on a Renishaw Raman spectrometer equipped with a 50 mW diode laser source for 780 nm excitation and a CCD detector. NMR spectra were measured on a Jeol Eclipse +500 NMR spectrometer. Thermal gravimetric analysis (TGA) was performed on a Mettler-Toledo TGA/SDTA851e system. Dynamic light scattering characterization was carried out on a Beckman Coulter N4 plus particle size analyzer. Scanning electron microscopy (SEM) images were obtained on a Hitachi 4700 field-emission SEM system. Transmission electron microscopy (TEM) analyses were conducted on a Hitachi HF2000 TEM system equipped with a Gatan Multiscan CCD camera for digital imaging. Functionalization Reactions. The functionalization of SWNTs was accomplished via direct heating with PEG1500N to form ionic linkages (Scheme 1).11,15 In a typical experiment, a mixture of purified SWNT sample (265 mg) and PEG1500N (4.5 g) was heated to 100 °C and vigorously stirred under nitrogen protection for 4 days. The reaction was quenched by the addition of deionized water. The reaction mixture was repeatedly extracted with deionized water, coupled with centrifuging at 7500 rpm (∼3100g). The colored supernatant was collected and placed in a dialysis membrane tubing (cutoff molecular weight ∼ 12 000) for the removal of unreacted PEG1500N via dialysis against fresh deionized water for 3 days. The sample of PEG1500N-functionalized SWNTs thus obtained was designated as the first soluble fraction. On the other hand, the insoluble residue from the functionalization reaction was dried to be used as starting material in the next functionalization reaction. The same procedure was repeated several times to obtain the second, third, fourth, and fifth soluble fractions. The final solid residue was also collected for characterization. The soluble fractions were also obtained in the solid state by removing the water from aqueous solutions on a rotatory evaporator.
Figure 1. 1H NMR spectra of PEG1500N and different soluble fractions in CDCl3 solutions.
Results and Discussion The soluble fractions from the sequential functionalization reactions of the SWNT sample with PEG1500N are all dark-colored, forming stable homogeneous solutions in polar organic solvents and water. The solubility allowed solution-phase NMR characterizations. As compared in Figure 1, the proton NMR spectra of all soluble fractions are similar, exhibiting a broad signal at ∼3.6 ppm corresponding to the protons in nanotube-attached PEG1500N species. The significant broadening in the NMR signals is consistent with the considerably reduced mobility in the PEG1500N species upon their attachment to SWNTs in the functionalization reactions. The NMR results also suggest that there are no other meaningful reaction products. As shown in Figure 2, the spectra of the sample solutions in the UV/vis region are generally featureless curves of gradually decreasing absorptivity with increasing wavelength, without any fundamental differences between the different soluble samples. The near-IR spectra were obtained for the samples in the solid state, a thin film on glass substrate (Figure 2). For all five fractions and the purified starting SWNT sample, there are the near-IR absorption bands at ∼5400 cm-1 (0.7 eV) and ∼9700 cm-1 (1.2 eV), which may be assigned to transitions corresponding to the first and second van Hove singularities in the electronic density of states for semiconducting SWNTs.19,20 However, the absence of the electronic transition peak around 14 000 cm-1 (1.76 eV), corresponding to the absorption of metallic SWNTs, in the spectra of the soluble samples can hardly be considered as evidence for any selection against the metallic nanotubes in the functionalization reactions. It is known that the chemical modification of metallic SWNTs could significantly change their absorption properties.21,22 (19) (a) Wildoer, J. W. G.; Venema, L. C.; Rinzler, A. G.; Smalley, R. E.; Dekker, C. Nature 1998, 391, 59. (b) Odom, T. W.; Huang, J.-L.; Kim, P.; Lieber, C. M. Nature 1998, 391, 62. (20) Dresselhaus, M. S.; Dresselhaus, G.; Eklund, P. C. Science of Fullerenes and Carbon Nanotubes; Academic Press: San Diego, CA, 1996.
7086
Langmuir, Vol. 19, No. 17, 2003
Huang et al.
Figure 3. Raman spectra of the starting SWNT sample, different soluble fractions (after the thermal defunctionalization in slow TGA scans), and the final solid residue.
Figure 2. UV/vis (---, in room-temperature THF solutions) and FT-NIR (-, in solid films at room temperature) spectra of the different soluble fractions. For each spectrum, the NIR portion after the subtraction of a sloped baseline is shown as an inset.
There have been several reports on the dependence of the near-IR absorption spectral feature on the SWNT diameter.22-25 For example, Jost et al.24 suggested that the band-gap transition energy is proportional to the inverse of the nanotube diameter. Ruoff and co-workers25 also suggested that the shift of the near-IR absorption band around 6000 cm-1 with the dispersion of SWNTs could be explained by different solubilities for SWNTs of different diameters, among other effects. Banerjee and Wong22 reported that the resolution of the near-IR spectral peaks in the solubilized SWNT sample could be explained by suggesting that certain discrete diameter distributions of nanotubes might be preferentially solubilized. The nearIR absorption spectra reported here (Figure 2) are apparently broad and show no obviously systematic changes in spectral features that might differentiate the soluble fractions. On the other hand, the absorption spectra do not exclude the possibility of selective solubilization of SWNTs in the sequential functionalization reactions. In fact, Raman spectroscopy is a more sensitive technique for an evaluation of the different soluble fractions. The SWNTs in the soluble fraction are well-dispersed, which makes it essentially impossible for direct Raman measurements because of overwhelming luminescence (21) Boul, P. J.; Liu, J.; Mickelson, E. T.; Huffman, C. B.; Ericson, L. M.; Chiang, I. W.; Smith, K. A.; Colbet, D. T.; Hauge, R. H.; Margrave, J. L.; Smalley, R. E. Chem. Phys. Lett. 1999, 310, 367. (22) Banerjee, S.; Wong, S. S. J. Am. Chem. Soc. 2002, 124, 8940. (23) Kataura, H.; Kumazawa, Y.; Maniwa, Y.; Umezu, I.; Suzuki, S.; Ohtsuka, Y.; Achiba, Y. Synth. Met. 1999, 103, 2555. (24) Jost, O.; Gorbunov, A. A.; Pompe, W.; Pichler, T.; Friedlein, R.; Knupfer, M.; Reibold, M.; Bauer, H.-D.; Dunsch, L.; Golden, M. S.; Fink, J. Appl. Phys. Lett. 1999, 75, 2217. (25) Ausman, K. D.; Piner, R.; Lourie, O.; Ruoff, R. S.; Korobov, M. J. Phys. Chem. B 2000, 104, 8911.
Figure 4. Radial breathing mode Raman peaks of the starting SWNT sample, different soluble fractions (after the thermal defunctionalization in slow TGA scans), and the final solid residue.
interference.1 Thus, Raman measurements of the different fractions were carried out after the thermal defunctionalization of the samples. The principle and practice for the thermal defunctionalization of functionalized carbon nanotubes via slow TGA scans have been reported previously.14-16 In the TGA scans, the organic functional groups are selectively evaporated, leaving behind defunctionalized SWNTs. As compared in Figure 3, the Raman spectra of the soluble fractions have similar features, except for changes in the radial breathing mode structure around 170 cm-1. A closer look at the changes is provided in Figure 4. First of all, the radial breathing mode peaks of the final solid residue left from the repeated functionalization reactions are shifted to lower wavenumbers from those of the starting SWNT sample. The relative intensities of the two peaks are also reversed in the two samples (Figure 4). For the soluble fractions, the radial breathing mode peak profiles are between the two extremes of the starting sample and the final residue,
Preferential Solubilization of Smaller SWNTs
Langmuir, Vol. 19, No. 17, 2003 7087
Table 1. Results from Dynamic Light Scattering Analyses of the Soluble Fractions in Homogeneous Solutions 90° scattering angle
30° scattering angle
fraction
avg sizea (nm)
σa,b (nm)
avg sizea (nm)
σa,b (nm)
1 2 3 4 5
1115 761 937 859 1034
977 926 515 623 533
1020 1225 1011 1182 1410
300 807 265 596 889
a Obtained from the histogram in terms of Gaussian distribution analysis. b Standard deviation.
Figure 6. High-resolution TEM image of the soluble fraction 4 deposited on a holey carbon-coated copper grid. Table 2. SWNT Contents in the Soluble Fractions
Figure 5. TEM images of the different soluble fractions deposited on holey carbon-coated copper grids (all scale bars 100 nm).
consistent with the changes from one extreme to the other. Since the radial breathing mode frequency is correlated with the SWNT diameter in an inversely proportional relationship,26 the Raman results in Figure 4 suggest that the functionalization reactions are selective, resulting in preferential solubilization of SWNTs of smaller diameters. An attempt was made to apply the sizing technique based on dynamic light scattering to the characterization of SWNTs in aqueous solutions of the soluble fractions. (26) Bandow, S.; Asaka, S.; Saito, Y.; Rao, A. M.; Richter, E.; Eklund, P. C. Phys. Rev. Lett. 1998, 80, 3779.
fraction
SWNT content (wt %)
1 2 3 4 5
6 4.7 8.1 9.1 17
In the measurements, the solution concentration was adjusted according to the scattering intensity for readings in the range between 5 × 104 and 1 × 106 counts/s. As shown in Table 1, the average sizes thus obtained are similar for different fractions, all around 1000 nm, with generally broad size distributions. In principle, the dynamic light scattering is not a technique suitable for rod-shaped objects, especially for the nanotubes of such a large aspect ratio. The manufacturer of the particle size analyzer suggests that a lower scattering angle should be considered as a more favorable option for sizing rod-shaped objects. However, the results obtained by measuring the soluble fractions at different scattering angles show no meaningful difference (Table 1). The functionalized SWNTs were analyzed by TEM. A specimen for analysis was prepared by using an inoculating loop to place a drop of a diluted solution of the functionalized SWNTs onto a holey carbon-coated copper grid, followed by the solvent evaporation. As shown in Figure 5, the TEM images for the five soluble fractions from sequential functionalization reactions exhibit no systematic changes, though the nanotubes in the first fraction appear somewhat shorter. The functionalized SWNTs in the images are apparently bundled. The bundling is due at least in part to the TEM sample preparation process. In fact, TEM images at a higher resolution, such as the one shown in Figure 6, confirm the aggregation of functionalized SWNTs. The nanotube contents in the soluble fractions were estimated in TGA measurements with a slow scan rate of 5 °C/min in nitrogen atmosphere. As shown in Table 2, the different soluble fractions contain different weight percentages of SWNTs, with a general trend of higher nanotube contents in later fractions. This might be rationalized by assuming that the SWNTs containing on average more defect sites (thus for the attachment of more functional groups) are preferentially solubilized in the
7088
Langmuir, Vol. 19, No. 17, 2003
Figure 7. SEM images of the starting SWNT sample (A), different soluble fractions (B, fraction 2, and C, fraction 5, both after thermal defunctionalization in slow TGA scans), and the final solid residue (D). Scale bars = 500 nm.
sequential functionalization reactions. The estimated nanotube contents in the five soluble fractions can be added up to the total amount of solubilized SWNTs, which represents 70% of the starting SWNT sample. Separately, the remaining solid residue obtained after the five repeated functionalization reactions is 33.5% of the starting SWNT sample (after correction for any trapped functional groups via TGA analysis), suggesting that 66.5% of the starting SWNT sample is solubilized. The two different mass balances are in reasonable agreement. It seems understandable that the nanotube contents in the soluble factions are probably slightly overestimated in TGA analyses because the thermal defunctionalization can hardly be complete and quantitative (see below). However, the small discrepancy does not change the overall understanding of the sequential functionalization reactions.
Huang et al.
The presence of organic functionalities makes the direct SEM characterization of the soluble fractions difficult because of the severe surface-charging interference. The SEM imaging condition is significantly improved upon removal of the functional groups in the thermal defunctionalization. Shown in Figure 7 are some SEM images of the defunctionalized soluble samples. There are still surface-charging effects, probably due to residual functional groups that are not completely removed in the thermal defunctionalization. This is consistent with what has discussed above: that the nanotube contents in the soluble fractions are probably slightly overestimated in TGA analyses. A comparison between SEM images of the starting and final SWNT samples (Figure 7) seems to suggest that the final residue obtained after the five repeated functionalization reactions may be somewhat cleaner than the starting SWNT sample. In summary, the direct heating of SWNTs with PEG1500N is an effective way to functionalize and solubilize the nanotubes. The different soluble fractions obtained from the sequential functionalization reactions are generally similar, showing no obviously systematic changes from one reaction to the next. However, one clear exception is the solubilization of SWNTs of different diameters, with the reactions preferentially solubilizing the nanotubes of smaller diameters. Strong experimental evidence for the selectivity is primarily from Raman spectroscopy, though other experimental results are not inconsistent with the explanation. The mechanism behind the selectivity is not clear. It is possible that the preferential solubilization is related to the fact that carbon nanotubes of much smaller diameters are somewhat soluble even without the functionalization.27 In addition, the nanotube contents in different soluble fractions obtained from the sequential functionalization reactions are apparently different, with higher nanotube contents in later fractions. Further investigations on other functionalization reactions for SWNT samples produced by other methods will be pursued. Acknowledgment. We thank Professor A. M. Rao for supplying the carbon nanotube samples. Financial support from NSF, NASA, and the Center for Advanced Engineering Fibers and Films (NSF-ERC at Clemson University) is gratefully acknowledged. We also acknowledge the support by the Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Transportation Technologies, as part of the HTML User Program, managed by UT-Battle LLC for DOE under Contract DE-AC05-00OR22725. LA0301893 (27) Bahr, J. L.; Mickelson, E. T.; Bronikowski, M. J.; Smalley, R. E.; Tour, J. M. Chem. Commun. 2001, 193.
