Carbon 42 (2004) 2849–2854

www.elsevier.com/locate/carbon

Mechanical and morphological characterization of polymer–carbon

nanocomposites from functionalized carbon nanotubes
a,* a,b a,b a,b a,c a,b
M.C. Paiva , B. Zhou , K.A.S. Fernando , Y. Lin , J.M. Kennedy , Y.-P. Sun
a
Center for Advanced Engineering Fibers and Films, Clemson University, Clemson, SC 29634-0910, USA
b
Department of Chemistry, Clemson University, SC 29634, USA
c
Department of Mechanical Engineering, Clemson University, SC 29634, USA

Received 29 February 2004; accepted 22 June 2004

Available online 14 August 2004

Abstract

Carbon nanotubes were functionalized with poly(vinyl alcohol) (PVA). The water-soluble PVA–functionalized carbon nanotubes
were then embedded into PVA matrix via a wet-casting method, resulting in polymer–carbon nanocomposite films with homoge-
neous nanotube dispersion. Composites with pristine and functionalized nanotubes were tested in tension. It was found that the
mechanical properties of these nanocomposite films were significantly improved compared to the neat polymer film. Functionaliza-
tion allowed good distribution of the nanotubes in the matrix, leading to higher film strength. Scanning electron microscopy shows
an apparent good wetting of the nanotubes by the PVA matrix. These results are supportive of good interfacial bonding between the
functionalized carbon nanotubes and the hosting polymer matrix.

2004 Elsevier Ltd. All rights reserved.

Keywords: A. Carbon composites, Carbon nanotubes; C. Electron microscopy, Raman spectroscopy; D. Mechanical properties

1. Introduction presence of carbon nanotubes [5,6] were developed for

the homogeneous dispersion of the nanotubes in the
There has been intense recent research concerning the polymer matrix. Presently, the chemical functionaliza-
fabrication, characterization and application of poly- tion and solubilization of carbon nanotubes has
mer–carbon nanotube composites, driven by a growing emerged as an effective means for the homogeneous dis-
recognition of the unique mechanical, electrical, thermal persion of carbon nanotubes in solution and for the sub-
and other materials-related properties of carbon nanotu- sequent wet-casting of polymer–carbon nanocomposite
bes [1,2]. In the fabrication of high-quality polymer–car- films [7]. Improvement in the compatibility between
bon nanocomposites with predictable and optimal the functionalized carbon nanotubes and the hosting
performance, a homogeneous dispersion of the carbon polymer matrices have been discussed. For example,
nanotubes in the polymer matrix is required. Methodol- Tour, Krishnamoorti and coworkers reported the func-
ogies such as sonication of carbon nanotubes in a poly- tionalization of single-walled carbon nanotubes
mer solution [3,4] and in situ polymerization in the (SWNTs) with diazonium salts, and the functionalized
SWNTs were used to form polystyrene-based compos-
*
ites [8]. These nanocomposites were found to have a
Corresponding author. Present address: Department of Polymer lower percolation threshold than those obtained with
Engineering, Institute for Polymers and Composites, University of
Minho, Campus of Azurém, 4800-058 Guimarães, Portugal. Tel.: +351
pristine SWNTs. Sun and coworkers reported the func-
253510327; fax: +351 253510339. tionalization and solubilization of SWNTs and multiple-
E-mail address: [email protected] (M.C. Paiva). walled carbon nanotubes (MWNTs) with polystyrene

0008-6223/$ - see front matter
2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2004.06.031
2850 M.C. Paiva et al. / Carbon 42 (2004) 2849–2854

copolymers bearing hydroxyl or amine moieties for the exact amount of functionalized nanotube solution was
subsequent fabrication of polystyrene–carbon nanocom- added to the PVA solution, briefly sonicated and stirred
posite films using a wet-casting method. The films thus overnight for effective mixing and for the removal of
obtained were found to have excellent optical quality, trapped air bubbles. The resulting solution, viscous
and were free of any significant nanotube aggregation and homogeneous, was cast onto a glass slide and flat-
effects [9]. Recently, they succeeded in functionalizing tened to uniform thickness using a flat blade. The film
carbon nanotubes with polyvinyl alcohol (PVA) [10]. was dried at room temperature for 24 h, peeled off the
The resulting PVA–carbon nanocomposite films showed glass slide and dried overnight in a convection oven at
excellent homogeneity. The homogeneous dispersion of approximately 100 C. For comparison, a neat PVA film
carbon nanotubes in the nanocomposite films was con- and a composite film with purified SWNTs without
firmed by transmission electron microscopy (TEM) functionalization were prepared using the same proce-
imaging microtomed ultrathin slices of the films [10]. dure. The typical dimensions of the fabricated films were
The objective of the present work is to study the char- 15 mm · 30 mm with a thickness of 50–100 lm. The car-
acteristics of nanocomposite films reinforced with PVA bon nanotube content in the PVA–carbon nanocompos-
matrix and single-walled and multi-walled carbon ite films was varied from 2.5% and 5% (wt/wt).
nanotubes (MWNT). Films were prepared with pristine
SWNTs and SWNTs that were functionalized with a 2.3. Film characterization
low molecular weight PVA. Films with functionalized
MWNTs were also prepared for morphology studies. Raman spectra were obtained on a Renishaw
It is hypothesized that functionalization of the nanotu- RM1000 Raman Microscope equipped with a research
bes facilitates their dispersion in the polymer matrix, grade Leica microscope, a laser diode source emitting
improving the homogeneity of the composite film and at 785 nm with an output of 24 mW laser power, and
thus the tensile properties. a RenCam CCD detector.
Tensile tests were conducted on film samples at room
temperature on a Rheometrics Miniature Materials Tes-
2. Experimental ter Minimat 2000. The samples were rectangular with
approximate planar dimensions of 25 · 4 mm2. Films
2.1. Materials were cut into test samples using a razor blade. The sam-
ples were kept in a desiccator for at least 5 days prior to
Aldrich supplied PVA matrix with a MW of
70,000– testing. Tensile tests were performed on four types of
100,000 (99% hydrolyzed). The PVA used for nanotube films: PVA only, PVA composite with approximately
functionalization was obtained from Alfa Aesar, and 5% weight of functionalized SWNT, PVA composite
had a MW of
22,000 (87–89% hydrolyzed). The SWNT with approximately 2.5% weight of functionalized
and MWNT samples were produced in Professor A.M. SWNT, and PVA composite with approximately 2.5%
RaoÕs laboratory (Physics Department, Clemson Uni- weight of pure SWNT. The tensile tests were carried
versity) using the arc-discharge method and the chemical out at a rate of 5 mm/min.
vapor deposition (CVD) method, respectively [11,12]. The resulting cross-sections formed after rupture of
The samples were purified by refluxing in an aqueous ni- the tensile specimens were analyzed by scanning electron
tric acid solution. microscopy (SEM). The SEM images were obtained on
The carbon nanotubes were functionalized with low a S-4700 cold field emission scanning electron micro-
molecular weight PVA by the N,N 0 -dicyclohexyl carbo- scope from Hitachi.
diimide-activated esterification reaction, as reported
elsewhere [10]. The carbon nanotube contents in the sol-
ubilized samples of functionalized nanotubes prepared 3. Results and discussion
in this study were 10–20% by weight, as obtained by
the defunctionalization method using thermal gravimet- As mentioned in Section 1, the carbon nanotubes
ric analysis (TGA) [10]. were functionalized according to the reaction scheme
presented in Ref. [10]. The functionalized nanotubes
2.2. Film preparation were thoroughly characterized in that paper by TEM,
Raman, and other techniques. TEM clearly shows a
Using the wet-casting method, nanocomposite films good dispersion of the functionalized nanotubes. For
were fabricated from water-solutions of PVA–SWNT the single wall nanotubes, it was observed that the initial
and PVA–MWNT. In a typical procedure, a PVA poly- large ropes were split into smaller ropes and individual
mer sample (MW
70,000–100,000) was dissolved in nanotubes by functionalization. The Raman character-
water by heating at 80 C and stirring continuously ization of the functionalized SWNTs [10] show the char-
for
4 h to obtain a 20% (wt/wt) PVA solution. An acteristic radial breathing mode at
170 cm1, as well as
 M.C. Paiva et al. / Carbon 42 (2004) 2849–2854 2851

45000 scattering. Typically, the relative intensity of this peak

PVA/Funct SWNT
to the centre of the G-band peak, ID/IG, increases with
PVA/Pure SWNT
increasing structural disorder and with decreasing crys-
Intensity (arbitrary units)

40000
tallite size (for graphite type materials) [13]. In the pres-
35000 ent study, the intensity ratio ID/IG remains similar for
both systems, 0.89 for the films with pure SWNTs and
30000 0.90 for the films with functionalized SWNTs. This is
an indication that no significant change in structural
25000 disorder of the nanotubes is produced by functionaliza-
tion.
20000 The results of the tensile tests in terms of YoungÕs
1100 1300 1500 1700 1900 2100 2300 2500 2700 modulus, yield stress and strain of the films and respec-
Raman shift (cm ) -1 tive standard deviations, corresponding to an average of
at least four samples tested for each type of film, are re-
Fig. 1. Raman spectra of PVA composite films with SWNT.
ported in Table 1. Fig. 2 shows typical stress–strain
curves obtained for the films.
the peaks found in sp2 carbon materials in general Fig. 2 shows the increase in strength and modulus
[13,14], namely the D band at 1310 cm 1, the G band achieved by incorporating functionalized nanotubes in
at 1595 cm1 and the G 0 band, a second harmonic of the PVA. The addition of pristine SWNT improves the
the D band, at 2595 cm1. modulus but decreases the strength. A decrease in
For the present work, concentrated solutions of these strength in a reinforced polymer is typically observed
functionalized nanotubes were diluted in PVA, as de- if there is an increase in the number and/or severity of
scribed in Section 2.2, in order to prepare films with the flaws. It is well known that pristine SWNT tend to
an approximate content of 2.5% and 5% (weight) of agglomerate to form ropes [15], and these ropes tend
functionalized nanotubes. to agglomerate into larger domains. These ropes of
The Raman spectra were collected for the composite SWNT contain a number of nanotubes that are not in
films with pure and functionalized SWNTs (Fig. 1), in contact with the polymer and are only bonded to each
order to monitor possible structural changes in the func- other by weak interactions that will fail at low shear
tionalized nanotubes, relative to the pure nanotubes. stresses. These nanotube domains, because of poor
The D-band is defined as a disorder induced Raman bonding, effectively become holes in the PVA which

Table 1
Tensile properties of the polymer and composite films
Sample YoungÕs modulus (GPa) Yield stress (MPa) Yield strain (mm/mm) Strain-to-failure (mm/mm)
PVA 4.0 ± 0.1 83 ± 1 0.038 ± 0.004 >0.60
PVA + 2.5% func SWNT 5.6 ± 0.4 97 ± 15 0.036 ± 0.004 0.05 ± 0.02
PVA + 5% func SWNT 6.2 ± 0.1 128 ± 2 0.038 ± 0.002 0.06 ± 0.01
PVA + 2.5% pure SWNT 5.4 ± 0.4 79 ± 1 0.032 ± 0.001 0.09 ± 0.01

150.0 PVA
PVA/5% Funct SWNT
125.0 PVA/2.5% Funct SWNT
PVA/2.5% Pure SWNT
100.0
Stress (MPa)

75.0

50.0

25.0

0.0
0.000 0.020 0.040 0.060 0.080 0.100

Strain (mm/mm)

Fig. 2. Typical stress–strain curves for PVA and for the composite films.
2852 M.C. Paiva et al. / Carbon 42 (2004) 2849–2854

Fig. 3. SEM micrograph of the fracture surface of a PVA/function- Fig. 4. SEM micrograph of the fracture surface of a PVA/function-
alized MWNT composite film. alized SWNT composite film.

smaller ropes, of SWNTs. Most regions of the fracture

act as stress concentrators that lead to strength and surface of the tensile specimens looked smooth, for three
strain reduction relative to the monolithic PVA. different samples observed, throughout the cross section,
PVA films with 5% functionalized SWNT show an in- at different levels of magnification. This may be an indi-
crease in YoungÕs modulus and yield strength of approxi- cation of good separation and dispersion of the SWNTs
mately 55%, relative to pure PVA. The films containing by functionalization. As explained in reference [10],
2.5% functionalized nanotubes show similar YoungÕs TEM observation of samples of PVA functionalized
modulus to the films with pristine nanotubes; the SWNTs cast from solution showed good dispersion of
improvement is more than 35%. Yield strength, as previ- the nanotubes, supporting the hypothesis of rope break
ously observed, decreases to a lower value than obtained down by functionalization.
for PVA, for the films formed with pure nanotubes, Micrographs obtained from the fracture surfaces of
while it increases considerably for the films with func- composite films with pure SWNT show that the nanotu-
tionalized nanotubes. This result may be explained by bes are most probably agglomerated in large nanotube
the tendency of pure SWNTs to agglomerate and not ropes, and that these ropes do not disperse evenly
be able to disperse in the matrix, while the functional- through the polymer. This is clearly depicted in Fig. 5
ized SWNTs disperse well and can be wet by the matrix that also shows a good wetting of these nanotube ropes
polymer. by the polymer matrix, as well as their tendency to
Fig. 3 shows a SEM micrograph of the fracture sur- agglomerate. The aspect is similar to the cross sections
face of a PVA film containing functionalized MWNTs. obtained for the composites with functionalized
After functionalization the MWNTs disperse well
through the PVA, and exhibit an excellent wetting by
the matrix. These observations are in good agreement
with the TEM observations of ultra-microtomed slices
of the composite film [10]. The micrograph of the func-
tionalized MWNT composite film is presented here as
an example for good dispersion and wetting by the
PVA matrix. The small dimensions of the films prepared
did not allow mechanical property determination for the
composites with MWNTs.
For the SWNT composite, if a good dispersion and
break down of the nanotube ropes is attained by func-
tionalization, the individual nanotubes will not be obser-
vable with SEM, due to their small size and the level of
magnification achieved. In fact, SEM observation of the
fracture surfaces of these composites did not show any
considerable amount of nanotubes. The cross-sections
analyzed showed rare formations, such as illustrated in Fig. 5. SEM micrograph of the fracture surface of a PVA/pure SWNT
Fig. 4, which can be associated with aggregates, or composite film.
 M.C. Paiva et al. / Carbon 42 (2004) 2849–2854 2853

MWNTs, but the dispersion, in the case of pure geneous composite films and leading to a considerable
SWNTs, is considerably less homogeneous. improvement in the film mechanical properties.
Mechanical properties of composites formed by a
PVA matrix and carbon nanotubes were reported in
the literature [16,17]. Shaffer and Windle [16] observed Acknowledgments
an improvement of mechanical properties, as measured
by dynamic mechanical thermal analysis, for the com- This work was supported by the Center for Advanced
posite films relative to pure PVA. The improvement Engineering Fibers and Films at Clemson University un-
was only significant for composites containing more der the auspices of the ERC Program of the National
than 40% in weight of nanotubes. Cadek et al. [17] re- Science Foundation, Award No. EEC-9731680.
port an increase in YoungÕs modulus (obtained by nano-
indentation) of nearly 78% for composite films
containing 1% in weight of MWNT, but conclude that References
this increment is mainly due to an increase in crystallin-
ity from 14% to 27%. They also observed that compos- [1] Ajayan PM. Nanotubes from carbon. Chem Rev
ites formed with PVK, that is mostly amorphous and 1999;99:1787–99.
remains so after inclusion of the nanotubes, show no [2] Thostensona ET, Renb Z, Choua T-W. Advances in the science
considerable change in mechanical properties with the and technology of carbon nanotubes and their composites: a
review. Compos Sci Technol 2001;61:1899–912.
addition of 1% of MWNTs. In both cases the compos- [3] Qian D, Dickey EC, Andrews R, Rantell T. Load transfer and
ites were formed with as-produced nanotubes. More re- deformation mechanisms in carbon nanotube–polystyrene com-
cently, Zhang et al. prepared PVA composites with posites. Appl Phys Lett 2000;76(20):2868–70.
HiPco single wall nanotubes, from an aqueous disper- [4] Star A, Stoddart JF, Diehl M, Boukai A, Wong EW, Yang X,
sion, assisted by sodium dodecyl sulfate (SDS) and et al. Preparation and properties of polymer-wrapped single-
walled carbon nanotubes. Angew Chem Int Ed 2001;40(9):1721–5.
poly(vinyl pyrrolidone) (PVP) [18]. The composite was [5] Tang BZ, Xu H. Preparation, alignment, and optical properties of
formed by PVA/PVP/SDS/SWNT in the approximate soluble poly (phenylacetylene)-wrapped carbon nanotubes. Mac-
proportion 100:8:2:5. The improvement in mechanical romolecules 1999;32:2569–76.
properties achieved with the addition of 5% of SWNTs [6] Kumar S, Dang TD, Arnold FE, Bhattacharyya AR, Min BG,
to the system PVA/PVP/SDS was considerable, namely Zhang X, et al. Synthesis, structure, and properties of PBO/
SWNT composites. Macromolecules 2002;35:9039–43.
an increase of 60% in modulus. The presence of bundles [7] Sun Y-P, Fu K, Lin Y, Huang W. Functionalized carbon
of SWNT in the final composite films is reported. nanotubes: properties and applications. Acc Chem Res
In this context, the preliminary results described in 2002;35:1096–104.
the present paper are encouraging, as they indicate a [8] Mitchell CA, Bahr JL, Arepalli S, Tour JM, Krishnamoorti R.
considerable improvement in YoungÕs modulus and Dispersion of functionalized carbon nanotubes in polystyrene.
Macromolecules 2002;35:8825–30.
strength at very low nanotube content of the composite [9] Hill DE, Lin Y, Rao AM, Allard LF, Sun Y-P. Functionalization
films. Functionalization does not seem to destroy the of carbon nanotubes with polystyrene. Macromolecules
nanotube structure, and may be an efficient means of 2002;35:9466–71.
separation of the nanotube bundles. [10] Lin Y, Zhou B, Fernando KAS, Liu P, Allard LF, Sun Y-P.
The degree of crystallinity of the PVA matrix could Polymeric carbon nanocomposites from carbon nanotubes func-
tionalized with matrix polymer. Macromolecules
not be assessed by DSC because the polymer decom- 2003;36:7199–204.
poses before melting. X-ray diffraction measurements [11] Journet C, Maser WK, Bernier P, Loiseau A, de la Chapelle ML,
of crystallinity are being conducted on a new sample Lefrant S, et al. Large-scale production of single-walled carbon
set, which will be reported in a follow-on paper. nanotubes by the electric-arc technique. Nature 1997;388:756–8.
[12] Andrews R, Jacques D, Rao AM, Derbyshire F, Qian D, Fan X,
et al. Continuous production of aligned carbon nanotubes: a step
closer to commercial realization. Chem Phys Lett
4. Conclusions 1999;303(5):467–74.
[13] Saito R, Jorio A, Souza Filho AG, Grueneis A, Pimenta MA,
The incorporation of a small weight fraction of Dresselhaus G, et al. Dispersive Raman spectra observed in
SWNTs in the PVA matrix resulted in an increase in graphite and single wall carbon nanotubes. Physica B
2002;323:100–6.
YoungÕs modulus for all the composite systems. The [14] Rao AM, Ritcher E, Bandow S, Chase B, Eklund PC, Williams
strength increased considerably for the composites with KA, et al. Diameter-selective Raman scattering from vibrational
functionalized SWNT, but decreased for the composites modes in carbon nanotubes. Science 1997;275:187–91.
with pure SWNTs, compared with the pure PVA. This [15] Thess A, Lee R, Nikolaev P, Dai H, Petit P, Robert J, et al.
must be due to the difficulty in separating and spreading Crystalline ropes of metallic carbon nanotubes. Science
1996;273(5274):483–7.
homogeneously the pure SWNTs in the PVA matrix. [16] Shaffer MSP, Windle AH. Fabrication and characterization of
The functionalization appears to facilitate the disper- carbon nanotube/poly (vinyl alcohol) composites. Adv Mater
sion of the nanotubes in the polymer, forming homo- 1999;11(11):937–41.
2854 M.C. Paiva et al. / Carbon 42 (2004) 2849–2854

[17] Cadek M, Coleman JN, Barron V, Hedicke K, Blau WJ. [18] Zhang X, Liu T, Sreekumar TV, Kumar S, Moore VC, Hauge
Morphological and mechanical properties of carbon-nanotube- RH, et al. Poly(vinyl alcohol)/SWNT composite film. Nano
reinforced semicrystalline and amorphous polymer composites. Letters 2003;3(9):1285–8.
Appl Phys Lett 2002;81(27):5123–5.