Stable dispersion of single wall carbon nanotubes in polyimide: the role of noncovalent interactions

Abstract

Single wall carbon nanotubes (SWNTs) have been dispersed in a nitrile functionalized polyimide matrix and the resulting composite shows excellent stability with respect to reaggregation of the nanotubes. This contrasts with the behaviour of structurally similar polyimides in which the dispersion is only stable for short periods of time. Shifts in certain characteristic FTIR and Raman peaks which indicate a charge transfer interaction between the nanotubes and polymer matrix are observed. A simple model for charge transfer stabilization is presented and shown to be consistent with the experimental observations.

Full text

Page 1

Stable dispersion of single wall carbon nanotubes in polyimide:
the role of noncovalent interactions
Kristopher E. Wisea,*, Cheol Parka, Emilie J. Siochib, Joycelyn S. Harrisonb
aNational Institute of Aerospace, Hampton, VA, USA
bAdvanced Materials and Processing Branch, NASA Langley Research Center, MS226, Hampton, VA, USA
Received 12 December 2003
Available online 25 May 2004
Abstract
Single wall carbon nanotubes (SWNTs) have been dispersed in a nitrile functionalized polyimide matrix and the resulting
composite shows excellent stability with respect to reaggregation of the nanotubes. This contrasts with the behaviour of structurally
similar polyimides in which the dispersion is only stable for short periods of time. Shifts in certain characteristic FTIR and Raman
peaks which indicate a charge transfer interaction between the nanotubes and polymer matrix are observed. A simple model for
charge transfer stabilization is presented and shown to be consistent with the experimental observations.
? 2004 Elsevier B.V. All rights reserved.
1. Introduction
New avenues in the design of future aerospace vehi-
cles can be enabled by taking advantage of multifunc-
tionality in structures. State-of-the art lightweight
aerospace structures are built from graphite fiber com-
posites. It is envisioned that with its impressive suite of
properties, carbon nanotube (CNT) nanocomposites
have the potential to surpass the performance of con-
ventional graphite fiber composites by providing both
sensing and load bearing functionalities in vehicle
structures. However, this will not be possible until some
problems associated with the use of CNTs can be re-
solved. In particular, the issue of CNT dispersion in
polymer matrices typically used in composite fabrication
will be addressed here. As has frequently been noted in
the literature, achieving a high degree of dispersion of
single wall carbon nanotubes (SWNTs) in a polymer
matrix is quite difficult, due primarily to CNTs’ high
affinity for one another (i.e. their tendency to aggregate
in bundles and agglomerates) and their rather weak in-
teraction with common polymers. Early efforts, using a
combination of mechanical mixing, sonication, and in
situ polymerization techniques, yielded composite solu-
tions which were kinetically stable but which tended to
phase separate over a period of days or weeks, indicat-
ing thermodynamic instability [1]. In an attempt to
produce solutions having long term stability, a number
of polyimide compositions were screened for compati-
bility with SWNTs. This Letter describes the prepara-
tion and characterization of a composite which exhibits
good long term stability and a number of desirable
materials properties. A qualitative rationale for the
success of this particular polymer in stabilizing the
SWNT dispersion is proposed.
2. Experimental
The experiments described below were performed
using purified laser ablated (LA) and high pressure
carbon monoxide (CO) decomposition (HiPco) single
wall carbon nanotubes. The LA and HiPco SWNTs
were purchased from Rice University and Carbon
Nanotechnologies, Inc., respectively. The LA and HiPco
SWNTs were about 1.2–1.6 nm and 0.8 nm in diameter,
respectively. The concentration of the catalysts in both
the purified LA (Ni and Co) and HiPco (Fe) SWNTs
was less than 3 wt% based on elemental analysis (Desert
*Corresponding author. Fax: +1-757-864-8312.
E-mail address: k.e.wise@larc.nasa.gov (K.E. Wise).
0009-2614/$ - see front matter ? 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.cplett.2004.04.096
Chemical Physics Letters 391 (2004) 207–211
www.elsevier.com/locate/cplett

Page 2

Analytics, ICP-MS). The polymer chosen for this work
was (ß-CN)APB/ODPA polyimide, the structure of
which is shown in Fig. 1. Through energy-filtered
transmission electron microscopy (EELS), (ß-CN)APB/
ODPA polyimide has been shown to wet SWNTs very
well [2]. This particular polyimide was selected because
of the presence of nitrile functionalized aromatic moiety.
Nitrile bearing aromatic compounds are generally good
electron acceptors due to their ability to accommodate
excess charge in low lying unoccupied orbitals. Of par-
ticular relevance is a recent report describing the ad-
sorption of 9, 10-anthracenedicarbonitrile on SWNT [3].
These authors found that of a series of substituted an-
thracenes, the dicarbonitrile derivative exhibited the
highest adsorption coverage. This result was attributed
to a higher SWNT binding affinity promoted by stron-
ger charge transfer interactions.
For the purposes of forming a composite with well
dispersed SWNT reinforcement, it was necessary to
prepare the mixture in a particular sequence. The com-
plete synthetic procedure is described in detail elsewhere
[1], but will be briefly recounted here and is schemati-
cally depicted in Fig. 1. A dilute SWNT suspension,
typically around 0.05 wt%, in N,N-dimethylacetamide
(DMAc), was prepared by homogenizing for 10 min
(750 rpm with a 6 mm diameter rotor homogenizer) and
sonicating for 1 h at 47 kHz. The sonicated SWNT
suspension was used as a solvent for the poly(amic acid)
synthesis with the diamine, 2,6-bis(3-aminophenoxy)
benzonitrile ((ß-CN)APB), and the dianhydride, 4,4-
oxydiphthalic anhydride (ODPA) (Fig. 1). The entire
reaction was carried out with stirring in a nitrogen-
purged flask immersed in a 40 kHz ultrasonic bath until
the solution viscosity increased and stabilized. Sonica-
tion was stopped and stirring continued for several
hours to form a SWNT-poly(amic acid) solution. The
unimidized SWNT poly(amic acid) solutions exhibited
excellent stability, remaining in solution for over two
years in sealed bottles under refrigeration.
A series of SWNT-polyimide nanocomposite films
with SWNT concentrations of 0, 0.02, 0.1, 0.2 and 0.5
wt% were prepared in the following manner. The
SWNT-poly(amic acid) solution was cast onto a glass
plate and dried in a dry air-flowing chamber. Subse-
quently, the dried tack-free film was thermally imidized
in a nitrogen-circulating oven to obtain a solvent-free
SWNT-polyimide film. The transparent films contain-
ing SWNTs were deep green in color while pristine
films were pale yellow. This color change was not
observed in other polyimides used in previous work
[1].
3. Results and discussion
One mechanism consistent with both long term dis-
persion stability and the observed color change is the
formation of an electron donor–acceptor (EDA) com-
plex between the nanotube filler and the polymer matrix.
EDA complexes, formed between a molecule of high
electron affinity and another of low ionization potential,
are not covalently bound, but can nevertheless be quite
stable. Previous reports have shown that SWNTs be-
have amphoterically (in a Lewis acid/base sense), inter-
acting strongly with both electron donors and electron
acceptors. Many of these studies have focused on alkali
metals as donors and halogens as acceptors [4–9], due to
their low ionization potentials and high electron affini-
ties, respectively. Other work has employed various
small molecules as electron donors (NH3, H2) and
electron acceptors (NO2, O2) [10–13]. Finally, reports
describing doping by larger organic molecules (various
aromatic acceptors [3,14] and organic amine donors
[15,16]), as well as by polymers functionalized with do-
nor or acceptor groups [17] have appeared. The cited
literature suggests that the direction of any charge
transfer found in the present situation can be controlled
by the electron donating/electron accepting nature of the
polymer matrix.
SWNT composites prepared using related polyimides
synthesized from ODPA monomers and other diamines
showed no noticeable color change. Based on these
negative results, ODPA was provisionally eliminated as
the active participant in the proposed EDA complex
with the SWNT. To test for the involvement of the
(ß-CN)APB diamine monomer, a solution of the
monomer in DMAc was prepared and observed to be
pale yellow in color. Upon addition of SWNTs the so-
lution immediately turned green, as observed in the
composite described above. These qualitative observa-
tions support the idea of EDA complex formation be-
tween SWNTs and the polymer matrix, evidently via the
(ß-CN)APB monomers.
Fig. 1. Synthetic route to (ß-CN)APB/ODPA.
208
K.E. Wise et al. / Chemical Physics Letters 391 (2004) 207–211

Page 3

To further examine the role of EDA interactions in
stabilizing the SWNT/(ß-CN)APB/ODPA composite,
Raman spectroscopy was employed to probe the impact
of the EDA interaction on the electronic structure of the
SWNT. Raman scattering spectra were taken using an
AlmegaTMdispersive Raman spectrometer (Thermo
Nicolet). A 532 nm incident laser light excitation was
employed and the laser beam was focused on the sample
with the aid of an optical microscope. Low excitation
laser power (15 mW) was used to minimize heating of
samples, which often caused downshifting of the ob-
served peaks. The spectrum of a reference sample of
pure SWNT was monitored through the entire experi-
ment and Raman shift of the G band caused by heating
was less than 1 cm?1.
Previous experimental and theoretical work have
shown that doping SWNTs with either electron donors
or acceptors [4,8,18,19] or electrochemically [20] resulted
in noticeable shifts in certain characteristic vibrational
modes. Specifically, removing charge from a SWNT (i.e.
p-doping or oxidizing) resulted in an upshift in the G
band peak around 1592 cm?1, while adding charge (i.e.
n-doping or reducing) to a SWNT resulted in a down-
shift. The downshift observed upon n-doping is easily
understood: as the additional electron density is placed
in the antibonding conduction bands of the SWNT, the
average C–C bond strength is weakened, resulting in a
downshift or softening of the vibrational frequency. The
reason for the upshift that occurs upon p-doping is less
obvious. One would intuitively expect that removing
electron density from the fully occupied, bonding va-
lence band of a SWNT would weaken the C–C bonding,
resulting in a downshift in the G band frequency. This
was not, in fact, what was observed. One possible ex-
planation for this behaviour is that the addition of some
sp3character to the sp2hybridized orbitals, which re-
sults from the curvature of the graphitic structure re-
quired to form a tube, results in Coulomb repulsion,
particularly in small diameter tubes [21]. Removing
electron density from these orbitals reduces the repul-
sion, resulting in stronger net bonding and a higher G
band frequency.
Based on these considerations, if the SWNTs were to
lose charge to the polymer matrix, one would expect an
upshift in the G band and, conversely, a downshift is
expected if charge is gained from the matrix. Fig. 2a
shows the measured Raman spectra of the laser ablated
tubes before and after dispersion in the (ß-CN)APB/
ODPA matrix at a concentration of 0.5%. An upshift of
4 cm?1is observed in the G band. Similar results are
found for a 0.2% composite using HiPco tubes (4 cm?1
upshift), as shown in Fig. 2b. While the magnitude of
this peak shift is relatively small, it is virtually constant
across a range of concentrations and is very reproduc-
ible. In contrast to the disorder induced dispersive D
band and its related second-order harmonic G0band,
the G band is not highly sensitive to hydrostatic pressure
or strain. The upshift of the G band was not observed
with other polymers such as polystyrene, poly(methyl
methacrylate), or the structurally similar CP2 polyimide,
none of which have a strong electron withdrawing group
[1]. This indicates that hydrostatic pressure or strain,
caused by thermal expansion mismatch between SWNT
and the matrix, does not significantly influence the po-
sition of the G band peak.
To evaluate the participation of the polymer matrix
as an acceptor in the proposed EDA complex, an FTIR
Fig. 2. (a) Raman spectra of LA SWNTs in isolation and in 0.5 wt%
composite. (b) Raman spectra of HiPco SWNTs in isolation and in
0.2 wt% composite. All intensities in arbitrary units.
K.E. Wise et al. / Chemical Physics Letters 391 (2004) 207–211
209

Page 4

spectrometer was used to collect nitrile stretching mode
shift in an ATR mode with a Nicolet Continu lm IR
microscope. Fig. 3 shows the spectral region containing
the CN stretching mode for a pristine (ß-CN)APB/
ODPA film and for a 0.5% laser ablated SWNT/polyi-
mide composite. If the nitrile group in the (ß-CN)APB
monomer were acting as a Lewis acid and withdrawing
charge density from the SWNT, one would expect to
observe a downshift in the CN stretching mode due to
partial occupation of the low lying antibonding acceptor
orbital. A downshift of approximately 2 cm?1was in
fact observed in this case. Interestingly, spectra taken on
samples with higher SWNT loadings showed no increase
in the magnitude of the shift. This invariance indicates
that SWNT/matrix coordination is saturated, even at
the lowest loading level (0.02 wt%), although it is un-
clear why this should be the case. The downshift ob-
served, while small, is reproducible and consistent with
the EDA model of the interaction.
A computational study of this system initiated to
augment understanding of the nature of the SWNT-ma-
trix interaction in this composite will be briefly described
here. While it is not currently possible to perform quan-
tum chemical calculations for systems of this size, quali-
tative insight may be gained by performing calculations
onsmallanalogsofthepolymers(monomers,dimers,etc)
and extrapolating the results to larger systems. One way
of assessing the likelihood of electron transfer from the
SWNT to the polymer matrix is to consider the relative
electronicchemical potentials ðlÞofthetwocomponents.
When two systems of differing electronic chemical po-
tentialarebroughtintocontact,thecompositesystemwill
reach an intermediate potential, i.e. equalize, through a
process of charge transfer [22–24]. For periodic materials
with band type electronic structure, the chemical poten-
tial is simply the negative of the Fermi level which, for
metallic or small band gap semiconducting tubes, is es-
sentially the negative of the work function (neglecting the
dipole potential) [7,25]
lNT¼ ?EFffi ?WFNT;
WFNTffi 4:8 ? 5:0 eV:
The situation is similar for molecular materials with
localized electronic structure, except that the chemical
potential is defined as the negative of the electronega-
tivity [26,27]. The molecular electronegativity, within the
finite difference approximation, is calculated as the
negative of the average of the ionization potential and
the electron affinity
lP¼ ?vP;
vPffi ðIP þ EAÞ=2:
The geometry of a (ß-CN)APB/ODPA monomer was
optimized using the B3LYP density functional method
with a 6-31G* basis set. The geometries of the radical
cationandanionwereoptimizedstartingfromtheneutral
geometry. Finally, single point energy calculations were
performed at these geometries using the larger 6-31+G*
basisset.Thisbasissetaddsdiffusefunctionstotheheavy
atoms which are known to be particularly important in
anions.Whilethepermonomerchargetransferinthereal
polymer composite is much less than a full electron, this
calculation provides a limiting value. All calculations
were done using either GAMESSAMESS [28] or NWChem4 [29].
Using the B3LYP/6-31+G* calculated energies of the
neutral, radical cation, and radical anionic forms of the
(ß-CN)APB/ODPA monomer, the ionization potential
and electron affinity were found to be 8.64 and 1.58 eV,
respectively. The average of these numbers is the Mul-
liken electronegativity of the monomer, 5.11 eV. Rather
than calculating the chemical potential of a SWNT, the
experimentally derived value of 4.8–5.0 eV was adopted
[7,25]. This approach is more sound because the experi-
mental value reflects the statistical distribution of radii
and chiral indices found in real SWNT samples while a
calculated value would be biased by the selection of a
particular type of tube. Comparing the electronic chem-
icalpotentials of the polymer model
(lP¼ ?5:1) with the range determined for SWNT
(lNT¼ ?4:8 to ?5:0), it is apparent that chemical po-
tentialequalizationwilldrivepartialchargetransferfrom
the SWNT to the polymer. This result agrees with the
conclusionsdrawn fromthe experimentalwork described
above and supports the idea that charge is transferred
from the SWNT to the polymer matrix at equilibrium.
Finally, it is noted that other factors that have nec-
essarily been excluded from these calculations would
compound
Fig. 3. FTIR spectra of pure (ß-CN)APB/ODPA and 0.5 wt% LA
SWNT composite in nitrile stretching region. The peak is downshifted
by about 2 cm?1in the composite. Intensity in arbitrary units.
210
K.E. Wise et al. / Chemical Physics Letters 391 (2004) 207–211

Page 5

tend to further stabilize charge transfer interactions in
this system. First, it is well known that a polar, polar-
izable environment, such as the bulk polymer in the
present case, can significantly stabilize EDA complex
formation. This is known to occur in both liquid and
solid ‘solutions’, where reorganization of the surround-
ing media lowers the energy of the EDA complex rela-
tive to its unsolvated value, usually significantly. A
second mechanism for stabilizing the EDA complex is
geometric distortion of one or both components to a
structure which, while unfavorable in isolation, is actu-
ally lower in energy when in conjunction with the other
component of the complex. An example of this is the
narrowing of the HOMO/LUMO gap which occurs
upon deformation of aromatic p systems. An interesting
example of this was recently described by Morin and
coworkers [30] in a study of benzene adsorption on a
platinum surface. It was found that distorting the planar
aromatic core of the benzene molecule raised the
HOMO energy and lowered the LUMO energy, which
allowed for an improved match with the metal Fermi
energy, and therefore, a more stable interaction.
4. Conclusions
In summary, a new SWNT – polymer composite that
exhibits excellent dispersion and long term stability has
been produced. The polymer, (ß-CN)APB/ODPA, is
believed to stabilize the dispersion of SWNTs by way of
a donor acceptor interaction between the tubes and the
(ß-CN)APB subunit of the polymer. This mechanism is
supported by both Raman spectra of the SWNTs and
FTIR spectra of the CN stretching band of the polymer,
as well as by ab initio calculations on the (ß-CN)APB
monomer. Work is underway to develop analogous
polyimides with high electron affinity substituent groups
for further improvement in SWNT dispersion and the
resultant physical properties of the composites.
Acknowledgements
K.E.W. and C.P. acknowledge NASA University
Research, Engineering and Technology Institute on Bio
Inspired Materials (BIMat) under award No. NCC-1-
02037 for partial support.
References
[1] C. Park, Z. Ounaies, K. Watson, R.E. Crooks, J. Connell, S.E.
Lowther, E.J. Siochi, J.S. Harrison, T.L.St. Clair, Chem. Phys.
Lett. 364 (2002) 303.
[2] C. Park, R.E. Crooks, E.J. Siochi, J.S. Harrison, N. Evans, E.
Kenik, Nanotechnology 14 (2003) L1.
[3] J. Zhang, J.-K. Lee, Y. Wu, R.W. Murray, Nano Lett. 3 (2003)
403.
[4] A.M. Rao, P.C. Eklund, S. Bandow, A. Thess, R.E. Smalley,
Nature 388 (1997) 257.
[5] P. Petit, C. Mathis, C. Journet, P. Bernier, Chem. Phys. Lett. 305
(1999) 370.
[6] S. Kazaoui, N. Minami, R. Jacquemin, H. Kataura, Y. Achiba,
Phys. Rev. B 60 (1999) 13339.
[7] S. Suzuki, C. Bower, Y. Watanabe, O. Zhou, Appl. Phys. Lett. 76
(2000) 4007.
[8] A. Claye, S. Rahman, J.E. Fischer, A. Sirenko, G.U. Sumanase-
kera, P.C. Eklund, Chem. Phys. Lett. 333 (2001) 16.
[9] N. Bendiab, L. Spina, A. Zahab, P. Poncharal, C. Marliere, J.-L.
Bantignies, E. Anglaret, J.L. Sauvajol, Phys. Rev. B 63 (2001)
153407.
[10] J. Kong, N.R. Franklin, C. Zhou, M.C. Chapline, S. Peng, K.
Cho, H. Dai, Science 287 (2000) 622.
[11] P.G. Collins, K. Bradley, M. Ishigami, A. Zettl, Science 287 (2000)
1801.
[12] C.K.W. Adu, U. Sumanasekera, B.K. Pradhan, H.E. Romero,
P.C. Eklund, Chem. Phys. Lett. 337 (2001) 31.
[13] J. Zhou, A. Buldum, J. Han, J.P. Lu, Mat. Res. Soc. Symp. Proc.
633 (2001) A13.48.
[14] E. Jouguelet, C. Mathis, P. Petit, Chem. Phys. Lett. 318 (2000)
561.
[15] K.D. Ausman, R. Piner, O. Lourie, R.S. Ruoff, M. Korobov,
J. Phys. Chem. B 104 (2000) 8911.
[16] J. Kong, H. Dai, J. Phys. Chem. B 105 (2001) 2890.
[17] M. Shim, A. Javey, N.W.S. Kam, H. Dai, J. Am. Chem. Soc. 123
(2001) 11512.
[18] N. Bendiab, E. Anglaret, J.-L. Bantignies, A. Zahab, J.L.
Sauvajol, P. Petit, C. Mathis, S. Lefrant, Phys. Rev. B 64 (2001)
245424.
[19] N. Bendiab, E. Anglaret, J.-L. Bantignies, J.L. Sauvajol, P. Petit,
C. Mathis, Physica B 323 (2002) 259.
[20] P. Corio, P.S. Santos, V.W. Brar, Ge.G. Samsonidze, S.G. Chou,
M.S. Dresselhaus, Chem. Phys. Lett. 370 (2003) 675.
[21] T. Dumitrica, C.M. Landis, B.I. Yakobson, Chem. Phys. Lett. 360
(2002) 182.
[22] R.F. Nalewajski, J. Am. Chem. Soc. 106 (1984) 944.
[23] W.J. Mortier, S.K. Ghosh, S. Shankar, J. Am. Chem. Soc. 108
(1986) 4315.
[24] P. Geerlings, F. De Proft, W. Langenaeker, Chem. Rev. 103
(2003) 1793.
[25] S. Kazaoui, N. Minami, N. Matsuda, H. Kataura, Y. Achiba,
Appl. Phys. Lett. 78 (2001) 3433.
[26] R.G. Parr, R.A. Donnelly, M. Levy, W.E. Palke, J. Chem. Phys.
68 (1978) 3801.
[27] C.-G. Zhan, J.A. Nichols, D.A. Dixon, J. Phys. Chem. A 107
(2003) 4184.
[28] GAMESSAMESS Version 25 June 2001 M.W. Schmidt, K.K. Baldridge,
J.A. Boatz, S.T. Elbert, M.S. Gordon, J.J. Jensen, S. Koseki, N.
Matsunaga, K.A. Nguyen, S. Su, T.L. Windus, M. Dupuis, J.A.
Montgomery, J. Comput. Chem. 14 (1993) 1347.
[29] High Performance Computational Chemistry Group, NWChem,
A Computational Chemistry Package for Parallel Computers,
Version 4.0.1 (2001), Pacific Northwest National Laboratory,
Richland, WA.
[30] C. Morin, D. Simon, R. Sautet, J. Phys. Chem. B 107 (2003)
2995.
K.E. Wise et al. / Chemical Physics Letters 391 (2004) 207–211
211