Insight into the mechanism of sidewall functionalization of single-walled nanotubes: an STM study

Understanding the chemistry of carbon nanotubes is a crucial step towards their ultimate practical use. Here we report the Ž . first nanoscale images of single wall nanotubes SWNTs whose sidewalls have been chemically derivatized. Scanning Ž . tunneling microscopy STM images of fluorinated SWNTs reveal a dramatic banded structure which indicates broad continuous regions of fluorination terminating abruptly in bands orthogonal to the tube axis. This pattern is consistent with an energetically favored addition mechanism where fluorine atoms add around the circumference of the tube. STM images of sparsely butylated SWNTs, for which fluorinated SWNTs served as precursor, are also reported. q 1999 Elsevier Science B.V. All rights reserved. The fundamental chemistry of single walled nanoŽ . tubes SWNTs is still in its infancy. Initial speculation suggested that chemical functionalization of SWNTs would be most favorable at the ends of nanotubes, and that functionalization of SWNT sidewalls may be difficult to achieve. Successful controlled fluorination of the SWNT sidewalls leading to soluble products has recently been reported, providing a vital precursor for the subsequent attachment of a wide variety of functional groups to the w x exterior of the SWNTs 1 . Unfortunately, the unique characteristics of SWNTs that make them ultimately ) Corresponding author. Tel.: q1-713-7375611; fax: q1-7135245237; e-mail: [email protected] attractive for applications also limit our ability to study their chemical derivatives with conventional chemical methods. In this paper, we show that scanŽ . ning tunneling microscopy STM images of fluorinated SWNTs provide a crucial starting point for developing a detailed understanding of SWNT sidewall chemistry. We also obtain STM images of butylated SWNTs for which the fluorinated tubes were used as precursor. This substitution chemistry gives rise to electronic modulations in the vicinity of the attached butyl group, analogous with similar w x observations on graphite surfaces 2–4 . The synthesis of single-walled nanotubes has been w x described previously 5,6 . SWNTs in ‘bucky paper’ form were fluorinated by passing fluorine gas plus a helium buffer through a Monel flow reactor containing the SWNT sample. By varying the sample tem0009-2614r99r$ see front matter q 1999 Elsevier Science B.V. All rights reserved. Ž . PII: S0009-2614 99 00973-2 ( ) K.F. Kelly et al.rChemical Physics Letters 313 1999 445–450 446 perature and reaction time, various degrees of fluoriŽ . nation up to the saturation condition 2C:1F can be obtained. This procedure is described in detail elsew x where 1 . Following fluorination, the SWNT material was removed from the reactor and analyzed with electron microprobe analysis and transmission electron microscopy to determine the exact product stoichiometry. Approximately 1 mg of the fluorinated SWNT material was placed into a vial containing approximately 10 ml of either isopropanol or dimethyl formamide. Sonication of SWNTs in either solvent for approximately five minutes results in the selective Ž . w x solubilization of highly fluorinated isopropanol 7 Ž . w x or sparsely fluorinated DMF samples 8 . The solŽ . vated fluorotubes were then dispersed on a Au 111 on mica substrate by means of a rotary spinner and examined with STM. All STM images were obtained using a homebuilt STM instrument with commercial w x electronics under ambient conditions 9 . In addition to PtrRh STM tips, C -adsorbed PtrRh STM tips 60 were selectively utilized to provide enhanced atomic resolution and to image electronic modulations on w x the SWNT 2,10 . STM images of a SWNT that has been fluorinated under saturation conditions, and a pristine SWNT for comparison, are shown in Fig. 1. The most striking feature observed in these images is the presence of narrow, abrupt, dark bands that form around the circumference of the tube upon fluorination. While a range of scanning conditions provide atomic resoluw x tion for bare nanotubes 11 , including the parameters used in this experiment, in images of nanotubes fluorinated under saturation conditions atomic resolution is not readily observable. STM images of SWNTs prepared under a variety of fluorination Ž . conditions are shown in Fig. 2. Fig. 2 a shows another case of saturation fluorination: in addition to the strongly banded regions shown in Fig. 1, long uninterrupted regions of the tube are also observed. Ž Ž .. Images of partially fluorinated SWNTs Fig. 2 b show that the darker narrow regions remain abruptly terminated but typically form significantly broader bands. From a comparison between the STM images obtained for fluorinated SWNTs and the corresponding elemental analysis, we conclude that the narrower, darker bands correspond to unfluorinated regions of the nanotube. Even in the case of partial fluorination, the termination between the fluorinated and nonfluorinated bands typically remains quite abrupt and orthogonal to the SWNT axis. In Fig. Ž . 2 c , we see that partially fluorinated nanotubes under high resolution exhibit this observed variation in atomic resolution: the narrower regions of the tube correspond to regions where atomic resolution is obtainable under these scanning conditions. A comparison of these STM images with theory provides insight into a likely geometry and function̊ ̊ Ž . Ž . Fig. 1. a An image 860 A by 180 A of a carbon nanotube fluorinated at 2508C for 12 h. The bright regions correspond to areas on the Ž . tube covered by fluorine atoms. Scan parameters: 0.7 nA tunneling current and q500 mV sample bias voltage. b A high-pass filtered ̊ ̊ Ž . Ž . image 170 A by 25 A, 1 nA, q100 mV of a bare single-walled carbon nanotube deposited on an Au 111 on mica surface and imaged using a C -functionalized STM tip. 60 ( ) K.F. Kelly et al.rChemical Physics Letters 313 1999 445–450 447 ̊ ̊ Ž . Ž . Fig. 2. a An image 1500 A by 200 A, 0.3 nA, q400 mV of a carbon nanotube fluorinated at 2508C for 12 h. The approximate carbon to ̊ ̊ Ž . Ž . fluorine ratio by microprobe analysis is 2:1. b An image 1250 A by 175 A, 0.5 nA, q500 mV of a carbon nanotube fluorinated at 1508C ̊ Ž . Ž . Ž for 5 h 3.7:1, C:F . The darker area on the left side of the image appears to be a less fluorinated area. c A high resolution image 300 A ̊ . by 20 A, 1 nA, q300 mV of a carbon nanotube fluorinated at 2508C for 1 h. This image was also obtained using a C -functionalized STM 60 Ž . Ž . tip. Between the fluorinated areas e.g., at left-hand arrow , small regions of atomic resolution can be observed e.g., at right-hand arrow . alization mechanism for fluorinated SWNTs. Two Ž . possible isomers for fully fluorinated 2C:1F nanoŽ . tubes are proposed, and are illustrated in Fig. 3 a . In Ž . Ž . Ž . Fig. 3. a Proposed 1,4 left and 1,2 right 2C:1F fluorination Ž . Ž . isomers. Tube axis for 10,10 tube is horizontal in both cases. b Ž X . Geometry denoting possible secondary fluorination sites 2,4,4 Ž . Ž . relative to an initial fluorine addition site 1 . c Illustration of energetically favorable multiple addition scenarios for the 1,4 Ž . Ž . circumferential top and the 1,2 axial bottom fluorination isomers. one case, the fluorine atoms occupy the 1,4 positions on every other row of hexagons: we call this the 1,4 Ž . Ž . isomer shown on the left . On the right of Fig. 3 a , an arrangement where the fluorine atoms occupy the 1,2 positions of the hexagons of alternate rows while the double bonds form a conjugated system along the tube axis is shown. We call this the 1,2 isomer. Traditional ab initio computational approaches to the single-wall carbon nanotubes are limited because of their relatively large size. However, semiempirical calculations can be used to yield semi-quantitative information of these large systems. In this study, all AM1 and CNDO calculations are carried out using a development version of the Gaussian suite of prow x grams 12 . This software exploited the latest development in replacing diagonalization by CG-DMS and other techniques, which scale linearly with rew x spect to the molecular size 13 . To obtain insight into the SWNT fluorination geometries and the fluorination mechanism, calculations were performed Ž . with a short open-end tube C H F using AM1 160 40 4 Ž methods and a short capped tube C F , xs20, 360 x . 40, 60 using both CNDO and AM1 methods. Both Ž . tubes were constructed to display the 10,10 chirality. From the CNDO optimized geometries obtained using the capped model, the calculated C5C bond length for the conjugated chain of the 1,2 isomer was ( ) K.F. Kelly et al.rChemical Physics Letters 313 1999 445–450 448 ̊ approximately 1.37 A, and the C–C bond length was ̊ approximately 1.45 A. For the 1,4 isomer, the C5C ̊ bond length obtained was 1.35 A. The C–F bond ̊ length of the two isomers is approximately 1.377 A, which is quite close to the C–F bond length of w x C F 14 . 60 60 The AM1 single point energy calculations based on the CNDO optimized geometries showed that the energy differences between the 1,2 and the 1,4 isomers are very small: 76.02 kcalrmol, 25.22 kcalrmol, and 64.30 kcalrmol for C F , C F , 360 20 360 40 and C F , respectively. For the greatest degree of 360 60 fluorination studied, C F , there is only 1 360 60 kcalrmol difference per fluorine atom. All results support the 1,4 as the more stable isomer, but the energy difference between the two isomers is actually quite small, and it would be realistic to anticipate that both isomers may co-exist. In the STM images, the fluorinated regions typically appear to terminate abruptly, forming bands around the circumference of the tube. This feature allows us to infer that the addition of fluorine to the pristine SWNT may occur more favorably around the circumference of the tube, and not down the tube axis. We have therefore performed AM1 calculations in order to ascertain the most energetically favorable sequence for multiple fluorine addition. This is shown Ž . schematically in Fig. 3 b . First, a single fluorine Ž atom is added to a C H molecule shown as 160 40 . position 1 . Making the assumption that the second fluorine atom chooses an active position near 1, three potential active positions exist for the second fluoŽ . rine atom positions 2, 4, and 4 . AM1 results for C H F show that for the 1,2, 1,4, and 1,4 160 40 2 isomers, the energy difference is negligible: all are candidates for the starting point of multiple fluorine addition. We then placed the second pair of fluorine atoms close to the first pair as either a 1,2 1,4 or 1,4 isomer, and calculated the addition energy for the second pair of atoms as a function of the relative separation between fluorine atom pairs. Relative energies corresponding to fluorine addition at positions Ž . around the circumference of the tube C ns0–4 n Ž . and along the tube axis A ns1, 2 relative to the n position of the first pair of fluorine atoms were calculated. These results are shown in Table 1. The energies of the isomers where the two pairs of fluorine add directly adjacent to each other are referenced as 0. It is quite clear that for the 1,2 isomer, if the four fluorine atoms are allowed to add around the circumference of the tube, they will try to stay as far away from each other as possible to minimize fluorine-fluorine repulsion. By comparison, addition along the axis of the tube for the 1,2 isomer is about 30 kcalrmol lower in energy than circumferential addition. This is a strong indication that the 1,2 isomer, if allowed, would form via fluorine addition along the tube axis. For the case of the 1,4 isomer, the calculations indicate that addition around the circumference of the tube is approximately 10 kcalrmol more energetically favorable than growth along the tube axis. The circumferential growth mechanism of the 1,4 fluorination isomer would explain the abrupt bandlike boundaries ubiquitously observed in our STM images of fluorinated SWNTs. It is interesting to note that even under saturation fluorination conditions these dark bands are still observable. This would be consistent with the 1,4 isomer being initiated at multiple sites along the tube. Since fluorination occurs on alternate pairs of rows of the 1,4 isomer, one can imagine the existence of adjacent 1,4 isomeric domains where the fluorinated rows of one domain lie in registry with the double bond rows Table 1 Ž . Relative energies kcalrmol of four fluorine atom addition C C C C C A A 0 1 2 3 4 1 2 1,2 isomer 0 y2.7 y15.2 y17.3 y18.2 y37.9 y34.9 X 1,4 isomer 0 ;0 y0.6 y2.1 y2.8 10.8 9.5 X Y 1,4 isomer 0 y4.7 y5.4 y6.5 y6.9 9.5 6.5 C denotes fluorine addition around the tube circumference, A denotes growth along the tube axis. 0–4 denotes the interval between the two pairs of fluorine atoms. ( ) K.F. Kelly et al.rChemical Physics Letters 313 1999 445–450 449 of a neighboring domain. This would result in full fluorination of the nanotube but with abrupt boundaries and may be the origin of the observed structures. Defects in the tubes might also play a role in either initiating or terminating the various isomeric domains, and discrete regions of the 1,2 isomer may also be present on the nanotubes. Since the C–F bonds in fluorofullerenes are significantly weaker than the C–F bonds in traditional alkyl fluorides, sidewall fluorinated SWNTs can serve as a chemical starting point for subsequent w x addition of a variety of moieties 15,16 . We have also imaged, via STM, sidewall butylated SWNTs which utilize fluorinated SWNTs as a precursor w x 17,18 . Samples were prepared in a manner similar to the preparation of fluorinated SWNTs. STM images of butylated SWNTs are shown in Fig. 4. One immediately notices that the banded morphology of fluorinated SWNT is no longer visible. Instead, relå Ž . tively large ;10 A bright features with an average ̊ spacing of 50 A are distinctly apparent as one scans Ž Ž .. along the butylated nanotube Fig. 4 a . In higher resolution, the apparent atomic periodicity in the ̊ vicinity of the defect is 4.26 A, while in regions far ̊ from a defect this periodicity appears to be 2.46 A Ž Ž .. Fig. 4 b . This superlattice feature is an electronic modulation of the nanotube induced by the presence of the attached butyl group, and has been studied both experimentally and theoretically on graphite w x 2–4,19,20 . Using a modified version of the LCAO w x method for graphite 3 with periodic boundary conditions, we have calculated the density of states at Ef Ž . for a 10,10 nanotube perturbed by a strongly bound Ž Ž .. carbon atom Fig. 4 c . This is a simple one p electron per carbon atom, 2401 atom model with the Ž . appropriate boundary conditions for a 10,10 nanotube in which the resulting surface charge density is Ž ‘warped’ into a cylindrical shape. For a more thorough treatment of the superlattice phenomenon on w x . carbon nanotubes see Ref. 20 . Both the size of the modulated region and the superlattice periodicity are in good agreement with the experimental image. In conclusion, we have obtained the first STM images of sidewall functionalized SWNTs. This powerful local probe, in combination with theoretical analysis, yields detailed new insights into how addition and substitution chemistry may be occurring on nanotube sidewalls. ̊ ̊ Ž . Ž . Fig. 4. a An image 240 A by 35 A, 1 nA, q200 mV of a fluorinated nanotube treated with butyl lithium. The three bright spots along the Ž . tube appear to be due to butyl groups bonded to the nanotube sidewall. b A zoomed and filtered area of a higher resolution image taken in Ž . the vicinity of one of the butyl defects large up-arrow . Both this one and the proceeding image were obtained with a C -functionalized 60 Ž . STM tip. Near the defect a strong electronic modulation can be observed small up-arrows . This 63 modulation seems to decay away to the Ž . right of the image where atomic resolution is once again obtained slanted up-arrows . Scan parameters: 0.67 nA and q500 mV sample bias Ž . Ž . voltage. c Theoretical image of the perturbation in the local density of states at E caused by a carbon adatom on a 10,10 nanof w x tube. This theory is based on similar calculations performed by Mizes and Foster 3 for carbon adatoms on graphite. ( ) K.F. Kelly et al.rChemical Physics Letters 313 1999 445–450 450