Functionalization of carbon nanotubes by electrochemical reduction of aryl diazonium salts: A bucky paper electrode

Article abstract of DOI:10.1021/ja010462s

Small-diameter (ca. 0.7 nm) single-wall carbon nanotubes are predicted to display enhanced reactivity relative to larger-diameter nanotubes due to increased curvature strain. The derivatization of these small-diameter nanotubes via electrochemical reduction of a variety of aryl diazonium salts is described. The estimated degree of functionalization is as high as one out of every 20 carbons in the nanotubes bearing a functionalized moiety. The functionalizing moieties can be removed by heating in an argon atmosphere. Nanotubes derivatized with a 4-tert-butylbenzene moiety were found to possess significantly improved solubility in organic solvents. Functionalization of the nanotubes with a molecular system that has exhibited switching and memory behavior is shown. This represents the marriage of wire-like nanotubes with molecular electronic devices.

Full text of DOI:10.1021/ja010462s

6536
J. Am. Chem. Soc. 2001, 123, 6536-6542
Functionalization of Carbon Nanotubes by Electrochemical Reduction
of Aryl Diazonium Salts: A Bucky Paper Electrode
Jeffrey L. Bahr, Jiping Yang, Dmitry V. Kosynkin, Michael J. Bronikowski,
Richard E. Smalley, and James M. Tour*
Contribution from the Department of Chemistry and Center for Nanoscale Science and Technology,
MS 222, Rice UniVersity, 6100 Main Street, Houston, Texas 77005
ReceiVed February 20, 2001. ReVised Manuscript ReceiVed April 25, 2001
Abstract: Small-diameter (ca. 0.7 nm) single-wall carbon nanotubes are predicted to display enhanced reactivity
relative to larger-diameter nanotubes due to increased curvature strain. The derivatization of these small-
diameter nanotubes via electrochemical reduction of a variety of aryl diazonium salts is described. The estimated
degree of functionalization is as high as one out of every 20 carbons in the nanotubes bearing a functionalized
moiety. The functionalizing moieties can be removed by heating in an argon atmosphere. Nanotubes derivatized
with a 4-tert-butylbenzene moiety were found to possess significantly improved solubility in organic solvents.
Functionalization of the nanotubes with a molecular system that has exhibited switching and memory behavior
is shown. This represents the marriage of wire-like nanotubes with molecular electronic devices.
Introduction
Since their discovery in 1991,1 there has been a great deal of
interest in derivatization of single-wall carbon nanotubes
(SWNTs) to facilitate their manipulation, enhance their solubil-
ity, and make them more amenable to composite formation.
Success at covalent sidewall derivatization of purified SWNTs
has been limited in scope, and the reactivity of the sidewalls
has been compared to that of the basal plane of graphite.2 The
only truly viable route to direct sidewall functionalization of
purified SWNTs has been fluorination at elevated temperatures.3
These functionalized nanotubes may either be de-fluorinated
by treatment with hydrazine or allowed to react with strong
nucleophiles, such as alkyllithium reagents.4,5 Although fluori-
nated nanotubes may well provide access to a variety of
functionalized materials, the two-step protocol and functional
group intolerance to organolithium reagents make alternate
routes worth developing. Other attempts at sidewall modification
have been hampered by the presence of significant graphitic or
amorphous carbon contaminants6 or have required solubilization
via chemistry on the ends of cut nanotubes.7 A more accom-
modating and direct approach to functionalized nanotubes is
therefore desirable.
Figure 1. Electrochemical reduction of an aryl diazonium salt, giving
a reactive radical that covalently attaches to a carbon surface, as in
refs 10-14.
Aryl diazonium salts are known to react with olefins
(Meerwein reaction8) and have also been shown to arylate
aromatic compounds.9 In solution-phase reactions, diazonium
salt decomposition is typically catalyzed by a metal salt such
as copper(I) chloride, giving a reactive aryl radical. In some
cases, the reaction is believed to proceed through an aryl cation.
This type of chemistry has been successfully applied to the
modification of carbon surfaces via grafting of electrochemically
reduced aryl diazonium salts.10-14 Reduction gives an aryl
radical that covalently attaches to the carbon surface, as shown
in Figure 1. This technique has been applied to both highly
ordered pyrolitic graphite (HOPG) and glassy carbon (GC)
electrodes. Additionally, the theoretical aspects of energetic
radical collisions with carbon nanotubes have been investigated
and predicted to be a viable means of functionalization.15
* Corresponding author. Department of Chemistry and Center for
Nanoscale Science and Technology-MS 222, Rice University, PO Box 1892,
Houston, TX 77005. Fax: 713-348-6250. E-mail: tour@rice.edu.
(1) Iijima, S.; Ichihashi, T. Nature 1993, 363, 603-605.
(2) Aihara, J. J. Phys. Chem. 1994, 98, 9773-9776.
(8) Obushak, M. D.; Lyakhovych, M. B.; Ganushchak, M. I. Tetrahedron
Lett. 1998, 39, 9567-9570.
(9) Gadallah, F. F.; Elofson, R. M. J. Org. Chem. 1969, 34, 3335-3339.
(10) Delamar, M.; De´sarmot, G.; Fagebaume, O.; Hitmi, R.; Pinson, J.;
Save´ant, J.-M. Carbon 1997, 35, 801-807.
(3) Mickelson, E. T.; Huffman, C. B.; Rinzler, A. G.; Smalley, R. E.;
Hauge, R. H.; Margrave, J. L. Chem. Phys. Lett. 1998, 296, 188-194.
(4) Boul, P. J.; Liu, J.; Mickelson, E. T.; Huffman, C. B.; Ericson, L.
M.; Chiang, I. W.; Smith, K. A.; Colbert, D. T.; Hauge, R. H.; Margrave,
J. L.; Smalley, R. E. Chem. Phys. Lett. 1999, 310, 367-372.
(5) Mickelson, E. T.; Chiang, I. W.; Zimmerman, J. L.; Boul, P. J.;
Lozano, J.; Liu, J.; Smalley, R. E.; Hauge, R. H.; Margrave, J. L. J. Phys.
Chem. B 1999, 103, 4318-4322.
(11) Allongue, P.; Delamar, M.; Desbat, B.; Fagebaume, O.; Hitmi, R.;
Pinson, J.; Save´ant, J.-M. J. Am Chem. Soc. 1997, 119, 201-207.
(12) Ortiz, B.; Saby, C.; Champagne, G. Y.; Be´langer, D. J. Electroanal.
Chem. 1998, 455, 75-81.
(6) Chen, Y.; Haddon, R. C.; Fang, S.; Rao, A. M.; Eklund, P. C.; Lee,
W. H.; Dickey, E. C.; Grulke, E. A.; Pendergrass, J. C.; Chavan, A.; Haley,
B. E.; Smalley, R. E. J. Mater. Res. 1998, 13, 2423-2431.
(7) Chen, J.; Hamon, M. A.; Hu, H.; Chen, Y.; Rao, A. M.; Eklund, P.
C.; Haddon, R. C. Science 1998, 282, 95-98.
(13) Saby, C.; Ortiz, B.; Champagne, G. Y.; Be´langer, D. Langmuir 1997,
13, 6805-6813.
(14) Delamar, M.; Hitmi, R.; Pinson, J.; Save´ant, J. M. J. Am. Chem.
Soc. 1992, 114, 5883-5884.
(15) Sinnott, S. B.; Ni, B. Phys. ReV. B 2000, 61, 343-346.
10.1021/ja010462s CCC: $20.00 © 2001 American Chemical Society
Published on Web 06/14/2001

Functionalization of Carbon Nanotubes
J. Am. Chem. Soc., Vol. 123, No. 27, 2001 6537
Figure 3. Absorption spectra in dimethylformamide, illustrating
the loss of structure on functionalization; (a) pristine SWNT-p, (b)
SWNT-1. The absorbance intensities have been adjusted to clarify the
change in structure; there is no implication that derivatization lowers
the base optical density throughout this region.
Figure 2. Aryl diazonium salts used to derivatize single-wall carbon
nanotubes. Preparation and characterization of these salts is found in
the Experimental Section.
Scheme 1
Recently, a method for producing small-diameter (ca. 0.7 nm)
single-wall carbon nanotubes was developed by Smalley et al.16
As the diameter of these nanotubes is approximately the same
as that of C60, they are expected to display enhanced reactivity
relative to the larger-diameter tubes typically produced by laser
oven methods, since the reactivity of C60 has been attributed in
part to curvature strain.17,18 We report here the functionalization
of these small-diameter nanotubes via electrochemical reduction
of aryl diazonium salts, in a manner similar to that employed
for functionalization of other carbon surfaces.10-14 Although
electrochemically induced reaction is the most favorable process,
we also noticed that a thermal reaction is efficacious. A variety
of diazonium salts have been used, including those that provide
moieties conducive to further elaboration after attachment to
the nanotubes. Also, an oligo(phenylene ethylene) molecular
device similar to one that has been shown to exhibit memory
and room-temperature negative differential resistance19 has been
attached to the nanotubes.
derivatization experiments, a piece of bucky paper, formed by
filtration of a suspension, was used as the working electrode in
a three-electrode cell and immersed in an acetonitrile solution
containing the diazonium salt and an electrolyte. Reaction of
the diazonium salts in Figure 2 with SWNT-p generated
SWNT-x, where x is the functionalized phenyl moiety of
diazonium salt 1-9, respectively.
Characterization. The electronic structure and optical prop-
erties of single-wall carbon nanotubes have been investigated.21-23
The UV/vis/NIR absorption spectra of SWNT-p and SWNT-1
are shown in Figure 3. The features in the spectrum of SWNT-p
are due to singularities in the density of states (DOS) and in
this spectral region are attributed to the band gap transitions in
semiconducting nanotubes. The width of these features is due
to the overlap of features from tubes of different diameters and
chiral indices. These transitions are no longer visible for
SWNT-1, and the spectrum is essentially featureless. The
absorption spectra of SWNT-2-SWNT-7 are similar, with no
apparent features. The spectra of SWNT-8 (Figure 4) and
SWNT-9 retained some visible features, but these were
significantly reduced relative to SWNT-p. The loss of structure
Results
Materials. The purified single-wall nanotube samples (here-
after, SWNT-p) used in this investigation contained little
amorphous or other extraneous carbon contaminants (see
Experimental Section for purification procedure). This fact is
significant, as the presence of such material has hindered the
ability to determine whether previous sidewall derivatization
efforts were successful.6 The residual iron content (catalyst from
gas-phase growth technique) in the samples was ca. 0.3 atomic
%. The diazonium salts used to derivatize SWNT-p are shown
in Figure 2. Compounds 1-7 were prepared from the corre-
sponding aniline derivatives using nitrosonium tetrafluoroborate.
The synthesis of 8 has been previously reported.20 Compound
9 was prepared according to Scheme 1. Characterization of these
compounds is found in the Experimental Section. For the
(16) Nikolaev, P.; Bronikowski, M. J.; Bradley, R. K.; Rohmund, F.;
Colbert, D. T.; Smith, K. A.; Smalley, R. E. Chem. Phys. Lett. 1999, 313,
91-97.
(20) Kosynkin, D.; Tour, J. M. Org. Lett. 2001, 3, 993-995.
(21) Liang, W. Z.; Wang, X. J.; Yokojima, S.; Chen, G. H. J. Am. Chem.
Soc. 2000, 122, 11129-11137.
(17) Haddon, R. C. Science 1993, 261, 545-549.
(18) Haddon, R. C. J. Am. Chem. Soc. 1997, 119, 1797-1798.
(19) (a) Chen, J.; Wang, W.; Reed, M. A.; Rawlett, A. M.; Price, D.
W.; Tour, J. M. App. Phys. Lett. 2000, 77, 1224-1226. (b) Reed, M. A.;
Chen, J.; Rawlett, A. M.; Price, D. W.; Tour, J. M. App. Phys. Lett. 2001,
78, 3735-3737.
(22) 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. App. Phys. Lett. 1999, 75, 2217-2219.
(23) Wu, J.; Duan, W.; Gu, B.-L.; Yu, J.-Z.; Kawazoe, Y. App. Phys.
Lett. 2000, 77, 2554-2556.

6538 J. Am. Chem. Soc., Vol. 123, No. 27, 2001
Bahr et al.
Table 1. Disorder Mode Frequency and Intensity Ratios of Major
Peaks in Raman Scattering Experiments
intensity ratio
compd
ωd
(ωr:ωd:ωt)a,b
SWNT-p
SWNT-1
SWNT-2
SWNT-3
SWNT-4
SWNT-5
SWNT-6
SWNT-7
SWNT-8
SWNT-9
1291
1295
1294
1295
1290
1291
1292
1293
1292
1293
1.0:0.3:2.7
1.0:2.2:3.3
1.0:2.2:4.0
1.0:2.0:4.0
1.0:1.4:3.7
1.0:1.4:3.7
1.0:1.5:3.5
1.0:1.3:3.8
1.0:0.7:3.0
1.0:0.8:2.5
a ωr ) radial breathing mode, ωd ) disorder mode, ωt ) tangential
mode. b ωr intensity taken at 265 cm-1; other intensities taken at
maxima.
Figure 4. Absorption spectra in dimethylformamide; (a) SWNT-p,
(b) SWNT-8.
Figure 6. Raman spectra in the radial breathing mode region; (a)
SWNT-p, (b) SWNT-4.
Figure 5. Raman scattering from pure and derivatized samples; (a)
SWNT-p, (b) SWNT-1. Samples were pieces of bucky paper. Excita-
tion was at 782 nm.
hybridization. The Raman spectra of the other functionalized
materials display similar modifications but to different degrees.
The frequency of the disorder mode and the relative intensities
of the three major bands are shown in Table 1. In all cases,
there is no significant change in the frequency of the disorder
mode, and the intensity of this mode increased relative to the
intensity of the breathing and tangential modes. The intensity
of the tangential mode is also increased relative to that of the
radial breathing mode in most cases, and the overall intensity
is lower. In some cases, Raman spectra collected after func-
tionalization revealed changes in the relative intensities of the
peaks within the radial breathing mode region. For example,
the Raman spectra in this region are shown in Figure 6 for
SWNT-p and SWNT-4. We did not observe changes in the
Raman spectra that could be attributed to the functionalizing
moieties. The presence of bands due to nitrophenyl moieties
on modified GC electrodes has been detected by subtraction of
untreated GC spectra, revealing several bands between 1100
in the absorption spectra is indicative of significant electronic
perturbation of the nanotubes and disruption of the extended π
network. This effect is most consistent with covalent function-
alization rather than simple adsorption to the nanotube walls or
end caps.
Raman spectroscopy of single-wall carbon nanotubes is also
well developed both theoretically and experimentally.24-26 The
Raman spectrum of SWNT-p (Figure 5) displays two strong
bands: the so-called radial breathing (ωr ≈ 230 cm-1) and
tangential (ωt ≈ 1590 cm-1) modes. The multiple peaks seen
in the radial breathing mode are due to the distribution of tube
diameters in the sample. The weaker band centered at ca. 1290
cm-1 is attributed to disorder (ωd) or sp3-hybridized carbons in
the hexagonal framework of the nanotube walls. The minor band
at 850 cm-1 is also characteristic of these small-diameter
nanotubes, although its molecular origin is not clear. The
spectrum of SWNT-1 (Figure 5) is quite different. Most notably,
the relative intensity of the disorder mode is much greater. This
is also an expected result of the introduction of covalently bound
moieties to the nanotube framework, wherein significant
amounts of the sp2 carbons have been converted to sp3
and 1175 cm-1 27
However, the bands were quite weak and
.
difficult to resolve, even though monolayer coverage with the
nitrophenyl moiety was achieved. In the present case, the
probable less extensive degree of functionalization or random
orientation of the aryl appendages renders the direct detection
of moieties on the nanotubes with Raman scattering exceedingly
difficult. Their effect on the properties of the nanotubes is
abundantly clear, however.
Infrared spectroscopy (FT-IR, ATR) was also used to
characterize some of the derivatized materials. The spectrum
of SWNT-4 (Figure 7a) clearly shows significant C-H stretch-
ing from the tert-butyl moiety at ca. 2950 cm-1. In the spectrum
(24) Richter, E.; Subbaswamy, K. R. Phys. ReV. Lett. 1997, 79, 2738-
2740.
(25) Rao, A. M.; Richter, E.; Bandow, S.; Chase, B.; Eklund, P. C.;
Williams, K. A.; Fang, S.; Subbaswamy, K. R.; Menon, M.; Thess, A.;
Smalley, R. E.; Dresselhaus, G.; Dresselhaus, M. S. Science 1997, 275,
187-191.
(26) Li, H. D.; Yue, K. T.; Lian, Z. L.; Zhan, Y.; Zhou, L. X.; Zhang,
S. L.; Shi, Z. J.; Gu, Z. N.; Liu, B. B.; Yang, R. S.; Yang, H. B.; Zou, G.
T.; Zhang, Y.; Iijimas, S. App. Phys. Lett. 2000, 76, 2053-2055.
(27) Liu, Y.-C.; McCreery, R. L. J. Am. Chem. Soc. 1995, 117, 11254-
11259.

Functionalization of Carbon Nanotubes
J. Am. Chem. Soc., Vol. 123, No. 27, 2001 6539
Figure 7. Infrared spectra (attenuated total reflectance) of derivatized
nanotubes; (a) SWNT-4, (b) SWNT-6.
Figure 9. Raman spectra illustrating restoration of structure and relative
mode intensities for SWNT-2 after TGA; (a) SWNT-p, (b) SWNT-2,
(c) SWNT-2 after TGA.
Table 2. Weight Loss and Estimated Stoichiometries from TGA,
with Heating to 500 °C in Argon
observed %
weight loss
stoicheometry
ratioa
compd
SWNT-p
SWNT-1
SWNT-2
SWNT-3
SWNT-4
SWNT-5
SWNT-6
SWNT-7
SWNT-8
SWNT-9
5
NA
35
30
26
27
26
31
39
-
1/25
1/27
1/20
1/34
1/31
1/28
1/36
-
Figure 8. Thermogravimetric analysis data in argon. Temperature
profile: 1 h hold at 150 °C, ramp 5 °C min-1 to 500 °C, 2.5 h hold at
500 °C.
36
1/40
a Nanotube carbons bearing a functionalized phenyl moiety. These
values are compensated for weight loss at low temperatures due to
solvent evaporation and degassing (ca. 2-4% in all cases).
of SWNT-6 (Figure 7b), the carbonyl (CdO) stretch is apparent
at 1731 cm-1 (1723 cm-1 in precursor diazonium salt), along
with minor C-H stretching modes in the 2900 cm-1 region.
Electron microprobe analysis (EMPA) experiments revealed
2.7 atomic % chlorine, relative to 96.2 atomic % carbon, for
SWNT-2. Similar experiments revealed 3.5 atomic % fluorine,
relative to 94.5 atomic % carbon, for SWNT-3. These percent-
ages correspond to estimated stoichiometries of CR0.036 for
SWNT-2, and CR0.05 for SWNT-3, where C is a carbon in the
nanotube framework, and R is the functionalizing moiety. In
other words, approximately one out of every 20-30 carbons in
the nanotube bears a functionalized phenyl moiety. For the
remaining functionalized materials, either they do not contain
an element conducive to quantitative determination, or no stan-
dard was available for calibration, as in the case of SWNT-1.
In thermogravimetric analysis (TGA) of SWNT-2 (Figure
8), a total weight loss of ca. 30% was observed on heating to
500 °C. De-gassing and solvent evaporation at low temperatures
account for approximately 3% of the loss. The majority of the
weight loss occurs at temperatures >400 °C. After TGA of
SWNT-2, the Raman spectrum is restored to approximately that
of SWNT-p, as seen in Figure 9. This restoration suggests
removal of the functional moieties, leaving the nanotubes intact.
The stoichiometry estimated from the EMPA data predicts a
weight loss of ca. 25% in the case of such a removal. After
accounting for weight loss due to solvent evaporation, these
figures are in excellent agreement. The TGA and EMPA data
for SWNT-3 are also in good agreement. SWNT-p suffers only
a ca. 5% weight loss following the same temperature profile.
TGA data and estimated stoichiometries for the remaining
materials (with the exception of SWNT-8; we did not prepare
enough of this material for TGA) are shown in Table 2.
Analysis of the reaction products by scanning electron
microscopy (SEM) did not reveal any visible evidence of
functionalization or significant change from SWNT-p. Trans-
mission electron microscopy (TEM) imaging of SWNT-4,
however, revealed significant changes due to the functional-
ization. In images of SWNT-p (Figure 10a), the nanotube walls

6540 J. Am. Chem. Soc., Vol. 123, No. 27, 2001
Bahr et al.
Figure 10. Transmission electron microscopy images; (a) SWNT-p, (b) SWNT-4. Scale bar shown in (b) applies to (a) as well.
are essentially clean and uniform, and there is no overcoating
of graphitic carbon. Images of SWNT-4 (Figure 10b) revealed
the presence of bumps on the sidewalls of the tubes, on the
order of 2-6 Å in dimension. These bumps are seen on almost
all individual tubes and on the exterior of ropes, although the
resolution is not sufficient to see whether they are present on
the walls of tubes buried within the ropes. These features are
clearly a result of functionalization, but we cannot derive much
information concerning the regioregularity of attachment to the
tubes.
SWNT-1 to a variety of conditions involving additional
sonication and heating. After these treatments, no discernible
spectroscopic changes were observed. As an additional check,
SWNT-3 was reexamined by EMPA after additional sonication
in acetonitrile, followed by filtration and washing. The fluorine
content was 3.6 atomic %, as compared to 3.5 atomic % (vide
supra), and hence within experimental limits
Discussion
Derivatization Mechanism. The functionalization described
here is probably initiated in a manner similar to that shown in
Figure 1. The aryl radical generated on reduction of the
diazonium salts reacts with a nanotube, leaving an adjacent
radical that may further react or be quenched by solvent. The
propensity of the initial aryl radical to dimerize or abstract a
hydrogen atom from the solvent is minimized by the fact that
the radical is generated at the surface of the nanotube where
reaction is desired. Indeed, herein lies the principal advantage
of this method, as opposed to a solution-phase method in which
the diazonium salt reduction is catalyzed by copper or some
other metal. Since the nanotubes would be present in solution
at quite low concentration, most aryl radicals would be
disadvantageously quenched by some other species. Dimeriza-
tion of nanotubes in the present case is also unlikely, due to
lack of mobility in the solid state. A polymerization growth
mechanism was proposed by McDermott et al.31 for the
derivatization of HOPG with 4-nitrophenyl diazonium salts. In
McDermott’s proposed mechanism, only a few aryl radicals
attacked the carbon surface, and subsequent radicals attacked a
growing chain of polyphenylene emanating from the original
attack sites. The initial point of “nucleation” or reaction was
considered to likely be a defect site. This locality of function-
alization was investigated with spatially resolved Raman
spectroscopy.32 One may suggest a similar scenario in the
present case, where only a few radicals become covalently
attached to the SWNTs, and a growing polyphenylene chain
wraps around the nanotubes. Polymer wrapping of nanotubes
is known.33-35 Alternatively, one may suggest polymerization
of the diazonium salts in solution, followed by wrapping, with
no actual covalent attachment to the nanotubes. Several factors
The solubility of single-wall carbon nanotubes is of significant
interest to researchers in the field.28-30 The three solvents most
applicable for the underivatized small-diameter nanotubes are
dimethylformamide, chloroform, and 1,2-dichlorobenzene.28
SWNT-4 was the only material found to offer significantly
improved solubility in organic solvents. This material was even
found to be somewhat soluble in tetrahydrofuran, as opposed
to SWNT-p which had a complete lack of solubility for in this
solvent. After sonication for about 30 min, the solution was
found to contain approximately 50 mg L-1 of SWNT-4, with
no visible particulate. After 36 h, some visible particulate was
present, but the solvent was still almost black. This dark color
was retained for at least several weeks. Solubility in dimeth-
ylformamide, chloroform, and 1,2-dichlorobenzene was also
improved, with suspensions being formed much more rapidly
than in the case of SWNT-p, and higher concentrations being
achievable. This improvement in solubility is probably due to
the blocking effect of the bulky tert-butyl group, which could
inhibit the close contact necessary for “roping” of the nanotubes.
SWNT-5 and SWNT-8 were found to be more soluble in
dimethylformamide, but solubility in other solvents (tetrahy-
drofuran, toluene, 2-propanol, carbon disulfide, etc.) was not
improved. SWNT-9 was prepared in an effort to effect improved
solubility in water and other hydrogen-bonding solvents. This
functionalization, however, had quite the opposite result. We
were not able to disperse the material in water or water
containing 0.2% Triton X. We, in fact, encountered considerable
difficulty in suspending the material in dimethylformamide, and
other solvents did not even wet the material.
In an effort to assess the robustness of the functionalization
and preclude simple intercalation or adsorption, we subjected
(31) Kariuki, J. K.; McDermott, M. T. Langmuir 1999, 15, 6534-6540.
(32) Ray, K.; McCreery, R. L. Anal. Chem. 1997, 69, 4680-4687.
(33) Dalton, A. B.; Stephan, C.; Coleman, J. N.; McCarthy, B.; Ajayan,
P. M.; Lefrant, S.; Bernier, P.; Blau, W. J.; Byrne, H. J. J. Phys. Chem.
2000, 104, 10012-10016.
(28) Bahr, J. L.; Mickelson, E. T.; Bronikowski, M. J.; Smalley, R. E.;
Tour, J. M. Chem. Commun. 2001, 193-194.
(29) Ausman, K. D.; Piner, R.; Lourie, O.; Ruoff, R. S.; Korobov, M. J.
Phys. Chem. B 2000, 104, 8911-8915.
(30) 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, 733-734.
(34) Riggs, J. E.; Guo, Z.; Carroll, D. L.; Sun, Y.-P. J. Am. Chem. Soc.
2000, 122, 5879-5880.
(35) Tang, B. Z.; Xu, H. Macromolecules 1999, 32, 2569-2576.

Functionalization of Carbon Nanotubes
J. Am. Chem. Soc., Vol. 123, No. 27, 2001 6541
argue against these events in the present case, however. First,
polymer wrapping would not be a sufficient electronic perturba-
tion to induce the spectroscopic changes seen in the derivatized
materials.35 Second, in the case of SWNT-4, polymerization
after initial reaction with a nanotube would be severely inhibited
by the steric bulk of the tert-butyl substituent. It is thus most
likely that the functionalizations described here are the result
of reaction at numerous sites on the nanotube sidewalls.
Some portion of the observed differences in the degree of
functionalization is surely due to differences in reactivity of
the salts. Such differences may give more or less opportunity
for dimerization or otherwise disadvantageous quenching of the
aryl radicals. A greater portion, however, is probably due to
heterogeneity among the bucky paper electrodes, possibly in
terms of defect frequency or diameter distribution within each
bucky paper electrode. The lower degree of functionalization
observed for SWNT-8, as evidenced by the Raman and
absorption spectra, is due in part to a lower concentration of 8
in the reaction solution, necessitated by its poor solubility. In
general, the solubility of the different salts at the liquid-surface
interface region of the reaction may differ considerably. Finally,
as previously mentioned, the quality of the bucky paper affects
the success of the functionalizations in general. Hence, variance
in this parameter accounts for some variation in the degree of
functionalization.
Unique “Bucky Paper” Issues. The use of bucky paper as
a working electrode for the derivatization raises several issues.
We did encounter reproducibility problems in the early stages
of this investigation, and we have identified two factors that
are critical to success. Electrical contact between the source and
the bucky paper is of course critical. We found this difficulty
alleviated by application of colloidal silver paste to the alligator
clip used to hold the bucky paper. Additionally, the success of
the reaction is somewhat dependent on the quality of the bucky
paper. This perhaps hearkens to issue of sufficient electrical
contact within the bucky paper itself. We have found that the
quality of the paper is influenced by the quality of the suspension
prior to filtration. Hence, it was helpful to achieve a suspension
that contained little or no visible particulate prior to filtration.
This condition ensures, macroscopically at least, a smooth,
homogeneous paper that may have better electrical contact
between the nanotube ropes and bundles that make up the bucky
paper.
Conclusions
Electrochemical reduction of aryl diazonium salts using a
bucky paper electrode has been presented as a viable means of
derivatization of small-diameter single-wall carbon nanotubes.
Derivatization far beyond just the end caps is achieved, and
the covalent attachment is quite robust. The series of diazonium
salts used here give differing degrees of functionalization,
although a single reason is not clearly responsible for these
differences. With appropriately bulky moieties, solubility of the
nanotubes in organic solvents can be significantly improved.
The functionalizing moieties can be removed by heating to 500
°C in an argon atmosphere, restoring the pristine nanotubes.
The functionalization technique described here is an enabling
development that will permit modification of nanotubes to
facilitate their incorporation into polymer composite materials
and perhaps molecular electronic applications.
Degree of Functionalization. The characterization data
suggest differing degrees of functionalization among the materi-
als reported here. In Raman scattering, the most direct evidence
lies in the relative intensity of the disorder mode peak at ca.
1290 cm-1. Although no quantitative relationship between this
ratio and the degree of substitution has been developed, the
relationship is surely a direct one. The change in the relative
ratio of the peaks in the radial breathing mode region (Figure
6) is also worthy of note. Since this Raman mode is resonance-
enhanced, a lower intensity for a given-diameter nanotube may
indicate a greater degree of functionalization for that particular-
diameter nanotube. In other words, this observation may provide
evidence for a greater reactivity of some tube diameters.
Preferential interaction of polymers with certain tube diameters
has been reported.32 However, this effect is not consistent
throughout the functionalized materials reported here. Addition-
ally, it is the larger-diameter tubes (lower frequency shift) whose
transitions are most significantly affected. This is not the
situation one would expect in consideration of the predicted
inverse relationship between the tube diameter and reactivity.
The import of this observation is then not clear, and further
investigation is required.
Experimental Section
Methylene chloride and acetonitrile were distilled from calcium
hydride. Dimethylformamide was distilled and stored over molecular
sieves. Tetrahydrofuran was distilled from sodium/benzophenone ketyl.
All other reagents were obtained commercially and used without further
purification. SEM experiments were performed at an accelerating
voltage of 50 kV. Samples for TEM imaging were drop dried from
THF onto a 200 mesh lacey carbon grid on a copper support. The
accelerating voltage was 100 kV. Raman spectra were collected from
solid samples, with excitation at 782 nm. UV/vis/NIR absorption spectra
were collected in double-beam mode, with solvent reference. FT-IR
spectra were collected using an attenuated total reflectance (ATR)
accessory. TGA data were collected in argon. The instrument used for
the EMPA experiments was calibrated, and data were taken from several
different points on each sample (∆d ≈ 100 µM). The average of these
points is reported in the Results section. NMR chemical shifts are
reported in ppm downfield from TMS and referenced to solvent. Melting
points are not corrected.
The small-diameter single-wall carbon nanotubes used in this
investigation were produced by a gas-phase catalytic technique, using
carbon monoxide as the feedstock and iron carbonyl as the catalyst.16
The raw production material was purified by air oxidation at 150 °C
for a period of 12 h, followed by annealing in argon at 800 °C for 1 h.
Finally, this material was sonicated in concentrated hydrochloric acid
(ca. 30 mg in 60 mL), filtered, washed extensively with water and
2-propanol, and dried under vacuum. The purity of these samples was
verified by SEM, TEM, and EMPA.
Total mass loss figures from the TGA experiments also
indicate differing degrees of functionalization. Direct relation
of this number to the degree of functionalization carries the
caveat of assuming that all phenyl moieties are cleanly removed
and that no other surface functionalities are lost. However, the
good correlation between the mass loss and EMPA data for
SWNT-2 and SWNT-3 suggest this is a reasonable assumption.
The UV/vis/NIR absorption spectra indicate a lower degree
of functionalization for SWNT-8 and SWNT-9 but do not
provide a true means for gauging the magnitude of the
difference. As this technique probes the entire ensemble of
species in solution, however, the spectra do suggest that the
functionalizations in general are not merely a surface reaction,
which would only change the spectroscopic properties of a
smaller portion of the whole, dependent, of course, on the
thickness of the bucky paper.
General Procedure for Electrochemical Derivatization of
SWNT-p. The apparatus used for the electrochemical derivatization
experiments was a three-electrode cell, with Ag/AgNO3 reference
electrode and platinum wire counter electrode. A piece of bucky paper
(1-2 mg) served as the working electrode. The bucky paper was

6542 J. Am. Chem. Soc., Vol. 123, No. 27, 2001
Bahr et al.
prepared by filtration of a 1,2-dichlorobenzene suspension over a 0.2
µM PTFE (47 mm, Sartorius) membrane. After drying under vacuum,
the paper was peeled off the membrane, and a piece was excised for
use in the derivatization. The bucky paper was held with an alligator
clip that was previously treated with colloidal silver paste (Ted Pella,
Inc.) and immersed in an acetonitrile solution of the diazonium salt
(0.05 M for SWNT-1-SWNT-7 and SWNT-9; 0.01 M for SWNT-8)
and tetra-n-butylammonium tetrafluoroborate (0.05 M). Care was taken
not to immerse the alligator clip itself. A potential of -1.0 V was
applied for a period of 30 min. Care was taken for the exclusion of
light, and nitrogen was bubbled through the solution during the
experiment. After reaction, the portion of the “bucky paper” that was
not immersed in the solution was excised, and the remainder was soaked
in acetonitrile for 24 h and then washed with acetonitrile, chloroform,
and ethanol. After drying, this material was sonicated in acetonitrile
for 20 min, filtered, and washed again with acetonitrile, 2-propanol,
and chloroform. The reaction products were dried under vacuum at
room temperature prior to characterization. Control experiments without
a diazonium salt confirm that these conditions do not affect the
nanotubes, as verified by UV/vis/NIR, Raman, and TGA.
General Procedure for Diazotization of Aniline Derivatives. A
portion of nitrosonium tetrafluoroborate (1.2 mol equiv) was weighed
out in a glovebox and sealed. After removal from the glovebox,
acetonitrile was added (3 mL/mmol of aniline), and the solution was
cooled to -30 °C. A solution of the aniline derivative (1 mol equiv) in
acetonitrile (ca. 1 mL/mmol) was added dropwise while stirring. In
some cases, dry methylene chloride was used as a cosolvent for the
aniline derivative (vide infra). After complete addition, stirring was
continued for 30 min, at which time the cold bath was removed. After
stirring for a total of 1 h, the solution was diluted with a 2× volume
of ether and stirred. The precipitate was collected by filtration and
washed with ether.
2-(2-(2-Methoxyethoxy)ethoxy)ethyl p-Toluenesulfonate (10).
Sodium hydroxide (3.65 g, 91.3 mmol) and tri(ethylene glycol)mono-
methyl ether (10.0 g, 60.9 mmol) were dissolved in a mixture of
tetrahydrofuran and water (140 mL, 20 mL respectively). The solution
was cooled in an ice bath. A solution of p-toluenesulfonyl chloride
(12.76 g, 67.0 mmol) in 20 mL of tetrahydrofuran was added slowly.
The solution was stirred at 0 °C for 3 h and then poured into 50 mL of
ice water. The mixture was extracted several times with methylene
chloride. The combined organic layers were washed with dilute HCl
and then brine and dried over magnesium sulfate. After filtration, the
solvent was removed by distillation at reduced pressure to give 16.6 g
of the product (86% yield). IR (neat) 3503.3, 2878.5, 1597.9, 1453.1,
1356.3, 1292.0, 1247.0, 1177.2, 1097.5, 1019.0, 924.17, 818.0, 776.9,
1
664.5 cm-1. H NMR (400 MHz, CDCl3) δ 7.50 (ABq, J ) 7.9 Hz,
∆ν ) 179 Hz, 4H), 4.09 (app t, J ) 4.8 Hz, 2H), 3.61 (app t, J ) 4.9
Hz, 2H), 3.55 to 3.52 (m, 6H), 3.47 to 3.46 (m, 2H), 3.30 (s, 3H), 2.38
(s, 3H). 13C NMR (100 MHz, CDCl3) δ 145.21, 133.28, 130.21, 128.28,
72.20, 71.00, 70.85, 69.69, 68.95, 68.26, 59.31, 21.96.
4-(2-(2-(2-Methoxyethoxy)ethoxy)ethyl)nitrobenzene (11). A por-
tion of 10 (9.0 g, 28.3 mmol) was dissolved in 50 mL of dimethyl-
formamide. Potassium carbonate (11.75 g, 85.0 mmol) and 4-nitro-
phenol (3.82 g, 27.5 mmol) were added. The solution was stirred at 80
°C for 16 h. After cooling to room temperature, the solution was poured
into water and extracted three times with methylene chloride. The
combined organic layers were washed with water and then brine, dried
over magnesium sulfate, and filtered, and the solvent was removed by
distillation at reduced pressure. Chromatography (silica, hexane:ethyl
acetate, 1:2) was employed to isolate the product (5.71 g, 73% yield).
IR (neat) 3109.2, 3078.2, 2878.5, 1726.3, 1588.1, 1511.2, 1337.1,
1106.7, 1050.3, 932.6, 845.5, 753.3, 656.1 cm-1. 1H NMR (CDCl3) δ
8.07 (d, J ) 9.3 Hz, 2 H), 6.88 (d, J ) 9.3 Hz, 2 H), 4.12 (app t, J )
4.8 Hz, 2 H), 3.79 (app t, J ) 4.6 Hz, 2 H), 3.62 (m, 2 H), 3.58 to 3.53
(m, 4 H), 3.44 to 3.42 (m, 2 H), 3.26 (s, 3 H). 13C NMR (100 MHz,
CDCl3) δ 164.29, 141.93, 126.24, 114.99, 72.29, 71.29, 71.03, 70.98,
69.77, 68.60, 59.44.
4-Bromobenzenediazonium tetrafluoroborate (1): yield 85%; mp
1
138 °C. H NMR (400 MHz, CD3CN) δ 8.22 (ABq, J ) 9.1 Hz, ∆ν
) 102.1 Hz, 4 H).
4-(2-(2-(2-Methoxyethoxy)ethoxy)ethyl)aniline (12). A portion of
11 (5.77 g, 20.2 mmol) was dissolved in 40 mL of acidic ethanol, and
a catalytic amount of 10% palladium on carbon was added. The mixture
was hydrogenated on a Parr apparatus (60 psi, 70 °C) for 3 h. The
mixture was then filtered over Celite, washing with ethanol. Solid
sodium bicarbonate was added, and the mixture was stirred for 2 h
and then filtered. The solvent was removed by distillation at reduced
pressure, leaving a brown oil (5.0 g, 98% yield). IR (neat) 3441.8,
3349.6, 2893.9, 2238.4, 1634.4, 1516.4, 1449.8, 1234.7, 1101.6, 907.0,
4-Chlorobenzenediazonium tetrafluoroborate (2): yield 78%; mp
134 °C. H NMR (400 MHz, CD3CN) δ 8.24 (ABq, J ) 9.2 Hz, ∆ν
) 214.2 Hz, 4 H).
1
4-Fluorobenzenediazonium tetrafluoroborate (3): yield 79%; mp
1
160 °C. H NMR (400 MHz, CD3CN) δ 8.64 (dd, J ) 9.4, 9.5 Hz, 2
H), 7.69 (dd, J ) 9.4, 9.5 Hz, 2 H).
4-tert-Butylbenzenediazonium Tetrafluoroborate (4). The 4-tert-
butylaniline was dissolved in a 1:1 mixture of acetonitrile and dry
methylene chloride prior to addition to the nitrosonium tetrafluorobo-
rate: yield 78%; mp 91 °C. IR (KBr) 3364.8, 3107.3, 2968.6, 2277.2,
1579.2, 1482.0, 1418.0, 1373.5, 1269.8, 1056.9, 841.1, 544.6, 621.4
1
722.6 cm-1. H NMR (400 MHz, CDCl3) δ 6.65 (ABq, J ) 8.7 Hz,
∆ν ) 51.5 Hz, 4 H), 4.01 (t, J ) 5.4 Hz, 2 H), 3.77 (t, J ) 4.6 Hz, 2
H), 3.69 (app t, J ) 5.6 Hz, 2 H), 3.65 to 3.59 (m, 4 H), 3.51 (app t,
J ) 4.9 Hz, 2 H), 3.34 (s, 3 H), 3.0 (brs, 2 H). 13C NMR (100 MHz,
CDCl3) δ 152.30, 140.58, 116.75, 116.24, 72.31, 71.14, 71.02, 70.93,
70.30, 68.49, 59.44.
1
cm-1. H NMR (400 MHz, CD3CN) δ 8.16 (ABq, J ) 9.0 Hz, ∆ν )
298.7 Hz, 4 H), 1.30 (s, 12 H). 13C NMR (100 MHz, CD3CN) δ 168.85,
133.67, 130.43, 111.88, 37.86, 30.84.
4-Nitrobenzenediazonium tetrafluoroborate (5): yield 67%; mp
4-(2-(2-(2-Methoxyethoxy)ethoxy)ethyl)benzenediazonium Tet-
rafluoroborate (9). Compound 12 was subjected to the procedure
described above for diazotization. The product was not crystalline, but
rather a dark red, sticky material that was difficult to manipulate. The
residue was mixed three times with ether, decanting the solvent. This
material was sufficiently pure by 1H NMR and was used without further
1
142 °C. H NMR (400 MHz, CD3CN) δ 8.72 (ABq, J ) 9.4 Hz, ∆ν
) 65.4 Hz, 4 H).
4-Methoxycarbonylbenzenediazonium tetrafluoroborate (6): yield
80%; mp 113 °C. IR (KBr) 3103.8, 3042.4, 2955.3, 2294.7, 2310.1,
1731.4, 1582.9, 1439.5, 1306.4, 1045.23, 953.1, 860.9, 758.5, 666.3,
1
528.0 cm-1. H NMR (400 MHz, CD3CN) δ 8.51 (ABq, J ) 9.1 Hz,
1
purification or characterization (2.17 g, 52% yield). H NMR (400,
∆ν ) 77.9 Hz, 4 H), 3.97 (s, 3 H). 13C NMR (100 MHz, CD3CN)
165.02, 142.44, 134.12, 133.16, 119.77, 54.53.
MHz, acetone-d6) δ 8.12 (ABq, J ) 9.5 Hz, ∆ν ) 479.5 Hz, 4 H),
4.53 (app t, J ) 4.5 Hz, 2 H), 3.92 (t, J ) 4.4 Hz, 2 H), 3.68 to 3.66
(m, 2 H), 3.61 to 3.56 (m, 4 H), 3.46 (t, J ) 5.4 Hz, 2 H), 3.27 (s, 3
H).
4-Tetradecylbenzenediazonium Tetrafluoroborate (7). The 4-tet-
radecylaniline was dissolved in a 1:1 mixture of acetonitrile and dry
methylene chloride prior to addition to the nitrosonium tetrafluorobo-
rate: yield 69%; mp 82 °C. IR (KBr) 3103.8, 2919.5, 2289.6, 1577.8,
Acknowledgment. Ivana Chiang and Professor John L.
Margrave purified the raw SWNT material. The financial
support of NASA (NASA-JSC-NCC 9-77, OSR 99091801), the
NSF (NSR-DMR-0073046) and DARPA/ONR is gratefully
acknowledged.
1
1473.7, 1070.8, 1024.8, 844.5, 813.8, 716.9, 541.0, 510.2 cm-1. H
NMR (400 MHz, CDCl3) δ 8.02 (ABq, J ) 8.8 hz, ∆ν ) 370.6 Hz, 4
H), 2.76 (t, J ) 7.7 Hz, 2 H), 1.61 (quin, J ) 7.8 Hz, 2 H), 1.23 (s,
22H), 0.85 (t, J ) 7.0 Hz, 3 H). 13C NMR (100 MHz, CDCl3) δ 159.92,
133.26, 131.94, 110.96, 37.49, 32.34, 30.87, 30.12, 30.10, 30.07, 30.04,
29.91, 29.78, 29.75, 29.72, 23.11, 14.55.
JA010462S