Noncovalent solubilization of multi-walled carbon nanotubes in common low-polarity organic solvents with branched Pd-diimine polyethylenes: Effects of polymer chain topology, molecular weight and terminal pyrene group

Article abstract of DOI:10.1016/j.polymer.2014.05.026

Noncovalent nonspecific solubilization of carbon nanotubes with common polymers without having any specific functionality is an important strategy for rendering debundled nanotube solutions for their processing and technological applications. Among the va

Full text of DOI:10.1016/j.polymer.2014.05.026

Polymer 55 (2014) 3120e3129
Contents lists available at ScienceDirect
Polymer
journal homepage: www.elsevier.com/locate/polymer
Noncovalent solubilization of multi-walled carbon nanotubes in
common low-polarity organic solvents with branched Pdediimine
polyethylenes: Effects of polymer chain topology, molecular weight
and terminal pyrene group
,
Lixin Xu a, Zhibin Ye a *, Stefan Siemann b, Zhiyong Gu c
a Bharti School of Engineering, Laurentian University, Sudbury, Ontario P3E 2C6, Canada
b Department of Chemistry and Biochemistry, Laurentian University, Sudbury, Ontario P3E 2C6, Canada
c Department of Chemical Engineering, University of Massachusetts Lowell, Lowell, MA 01854, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Noncovalent nonspecific solubilization of carbon nanotubes with common polymers without having any
specific functionality is an important strategy for rendering debundled nanotube solutions for their
processing and technological applications. Among the various polymers investigated thus far for non-
covalent nonspecific nanotube solubilization, hyperbranched polyethylene (HBPE) featured with distinct
highly compact dendritic chain architecture has been discovered to show outstanding performance in
rendering stable nanotube solutions in common low-polarity organic solvents (including tetrahydro-
furan (THF) and chloroform) at surprisingly high concentrations. To understand the mechanism of the
nanotube solubilization with this unique class of polymers and to elucidate the effects of various
macromolecular structural parameters, we have designed and synthesized in this work four sets of
highly branched polyethylenes varying in chain topology, molecular weight, and end group. With these
polymers, we have systematically investigated and compared their performance for the solubilization of
multi-walled carbon nanotubes in common solvents including THF, chloroform, n-heptane, and toluene.
We have found that these macromolecular structural parameters as well as the solvent play complex but
sensitive roles in this noncovalent solubilization system. This work thus provides some valuable
guidelines towards the design of optimum polymers for efficient noncovalent nonspecific solubilization
of carbon nanotubes.
Received 24 March 2014
Received in revised form
6 May 2014
Accepted 7 May 2014
Available online 20 May 2014
Keywords:
Multi-walled carbon nanotubes
Hyperbranched polyethylene
Noncovalent nonspecific solubilization
© 2014 Elsevier Ltd. All rights reserved.
1. Introduction
far, two general strategies, covalent [11,12] and noncovalent [13,14],
have been extensively explored for surface functionalization of
Stable dispersion or solubilization of carbon nanotubes (CNTs)
in aqueous or organic solvents at high concentrations has been
critically important to the processing and applications of these
nanomaterials of outstanding properties in various emerging fields
[1e6]. CNTs always tend to aggregate in most solvents, originating
from their high aspect ratios and strong van der Waals interactions
between these nanomaterials. One of the solutions to this problem
is surface functionalization of CNTs with organic polymers, which
can impart modified CNTs with required surface functionalities and
significantly improved dispersibility in desired solvents [7e10]. So
CNTs with polymers [7e10]. Through the covalent strategy,
numerous polymers have been covalently grafted onto nanotube
surface via two different methodologies, including “grafting-from”
surface-initiated living/controlled polymerization techniques and
“grafting-to” reaction of tailor-made reactive polymers with func-
tional groups on nanotube surface [11,12]. Though efficient in
rendering soluble CNTs, this covalent functionalization strategy
inevitably leads to considerable disruption in the
p-orbital struc-
ture of CNTs and thus their compromised properties.
In the noncovalent strategy, CNTs are surface functionalized
with polymers by noncovalent sidewall adsorption or wrapping
through either specific or nonspecific interactions between the
polymers and nanotube sidewalls [7e10,13,14]. This strategy has
the advantage of preserving the structural integrity of CNTs and
* Corresponding author. Fax: þ1 705 675 4862.
E-mail address: zye@laurentian.ca (Z. Ye).
http://dx.doi.org/10.1016/j.polymer.2014.05.026
0032-3861/© 2014 Elsevier Ltd. All rights reserved.

L. Xu et al. / Polymer 55 (2014) 3120e3129
3121
consequently their uncompromised properties, and is thus
preferred. In the case of noncovalent specific functionalization,
numerous specially designed polymers bearing various specific
functionalities are used to establish noncovalent specific in-
compact hyperbranched topology. The presence of abundant
branch ends on its spherical surface is believed to render high-
density stronger CHe
p interactions between HBPE and the nano-
tube surface and thus high nanotube solubility [38,48]. Following
the general polymer structureeproperty relationships, the chain
topology and molecular weight of the polymer are reasoned to be
important macromolecular structural parameters governing the
performance of the polymer in nanotube solubilization in this
particular system. To further understand the mechanism and to
elucidate the intriguing roles of chain topology and molecular
weight in this unique system, we have tailor designed in this work a
broad range of well-defined highly branched polyethylenes (PEs) of
varying in topology and molecular weight, and have investigated
systematically their effects on the solubility of MWCNTs. Mean-
while, we have also synthesized narrow-distributed PEs bearing a
terminal pyrene group and investigated the possible synergistic
effect arising from these additional specific groups on the nanotube
solubilization.
teractions, such as
pep stacking with the use of conjugated poly-
mers [15e24] or pyrene-containing polymers [25e32], ionic
interactions with the use of polyelectrolytes [33e37], etc., between
the polymers and nanotube sidewalls. Synthesis of these specialty
polymers, however, often involves the use of specially designed
functional monomers/molecules and/or sophisticated polymeriza-
tion techniques, which is unfavorable for the large-scale industrial
applications of CNTs [38]. The noncovalent nonspecific function-
alization, on the other hand, often employs the nonspecific CHe
p
interactions between nanotube sidewalls and conventional poly-
mers without any specific functionalities. Given the need of only
conventional polymers made from common abundant monomer
stocks, this noncovalent nonspecific approach is particularly
desired for the solubilization of CNTs.
Baskaran et al. were the first to report the noncovalent nonspe-
cific solubilization of multiwalled carbon nanotubes (MWCNTs)
with the use of conventional polymers, including polybutadiene,
2. Experimental section
polyisoprene, polystyrene, and polymethylmethacrylate, via CHe
p
2.1. Materials and reagents
interactions [39]. They have suggested that the polymer adsorption
or wrapping around CNTs is a general phenomenon. However, the
solubility of MWCNTs in organic solvents (such as CHCl3) achieved in
their study is too low (below 0.020 mg/mL) for practical applica-
MWCNTs (a product of Arkema Inc.) were purchased from
Aldrich and used without further purification. These nanotubes
were reported to have a purity of >90%, external diameter of
tions, possibly due to the weakness of CHe
p
interactions. In addi-
10e15 nm, inner diameter of 2e6 nm, length of 0.1e10 mm, and
tion, some other conventional nonfunctionalized polymers,
including poly(N-isopropylacrylamide) [40], poly(acrylic acid) [41],
poly(dialkylsilane) [42], and poly(N-(2-(dimethylamino)ethyl)-
methacrylate) [43], were also demonstrated to functionalize and
thickness of 5e15 graphene layers. Pdediimine catalysts, [(ArN]
C(Me)e(Me)C]NAr)Pd(Me)(N≡CMe)]þSbF6ꢀ (1) and [(ArN]
C(Me)e(Me)C]NAr)Pd(CH2)3CO(O)Me]þSbF6ꢀ (2) (Ar
¼
2,6-
(iPr)2C6H3), were synthesized by following literature procedures
[49]. Polymer-grade ethylene was obtained from Praxair and pu-
rified by passing through 3 Å/5 Å molecular sieves and Oxiclear
columns in sequence to remove moisture and oxygen, respectively,
before use. Dichloromethane (>99.9%) and chlorobenzene (>99.5%)
were obtained from Fisher Scientific and purified through a solvent
treatment system (Innovative Technology Inc.) before use. THF (ACS
reagent, >99.0%), toluene (>99.5%), and methanol (>99.8%) were all
obtained from Fisher Scientific. Chloroform (>99.8%) and n-hep-
tane (>99.0%) were obtained from Aldrich. All these solvents were
used as received.
disperse CNTs in organic solvents via CHe
p interactions. Though
effective nanotube dispersions were obtained in these latter cases,
the nanotube solubility data, presumably low, were not disclosed.
Superior to these reported polymers in terms of nanotube solubility,
our group demonstrated in 2009 the use of a hyperbranched poly-
ethylene (HBPE) for the noncovalent nonspecific functionalization
and solubilization of MWCNTs in organic solvents (tetrahydrofuran,
THF, and chloroform), which rendered strikingly high nanotube
concentrations (up to ca. 1235 mg/L in chloroform and 920 mg/L in
THF) [38]. These nanotube solubility data achieved with HBPE are
comparable to or even greater than the highest values ever achieved
in organic solvents in the literature with the use of special conju-
2.2. Synthesis of pyrene-functionalized Pdediimine catalyst (3)
gating polymers capable of forming pep interactions with nanotube
sidewalls [23]. HBPE is a new grade of polyethylene synthesized
directly and solely from ethylene through the well-known PdeDii-
mine-catalyzed chain walking polymerization [44,45]. The adsorp-
tion of HBPE on nanotube sidewalls has been observed with
transmission electron microscopy (TEM) [38]. In subsequent studies
[46,47], HBPE-functionalized MWCNTs were also shown to have
significantly enhanced dispersion in ethyleneeoctene copolymer
due to the presence of HBPE around the nanotubes, which
dramatically improves the compatibility of the nanotubes with the
matrix polymer. Recently, we have further discovered that HBPE also
efficiently facilitates the production of stable dispersions of high-
quality few-layer graphene sheets at high concentrations in THF
and chloroform through ultrasonication-assisted liquid-phase
exfoliation of graphite [48]. Similarly, HBPE was found to adsorb on
The pyrene-functionalized Pdediimine catalyst, [(ArN]
C(Me)e(Me)C]NAr)Pd(CH2)3CO (O)(CH2)Py]þSbF6ꢀ (3) (Ar ¼ 2,6-
(iPr)2C6H3, Py ¼ pyrenyl), was synthesized by reacting 1 with pyr-
enemethyl acrylate (CH2]CHCOOCH2Py) that was synthesized by
following a similar literature procedure [26]. Catalyst 1 (0.95 g,
1.18 mmol) and pyrenemethyl acrylate (0.50 g, 1.75 mmol,
1.48 equiv.) were subsequently added into a 100 mL Schlenk flask
containing 40 mL of anhydrous dichloromethane under nitrogen
protection. The solution was stirred at room temperature for 96 h
under nitrogen atmosphere. After the reaction, the resulting solu-
tion was filtered using a 0.22 mm PTFE syringe filter. The filtrate was
concentrated in vacuo to a volume of ca. 5 mL, and anhydrous
diethyl ether (10 mL) was added to precipitate the product. The
precipitate was washed with anhydrous diethyl ether for 2 times
(20 mL/time), followed with drying in vacuo at room temperature
for 2 h to render 3 as a light yellow powder (0.70 g, 56.5% yield).
Anal. Calcd (found) for C49H57O2N2PdSbF6: C, 56.41(56.15); N,
the surface of exfoliated graphene sheets through CHe
p in-
teractions, preventing their restacking and rendering their stability
in the organic solvents.
CHe
p
interactions are generally much weaker compared to
interactions [13]. The remarkable
2.67(2.67); H, 5.78(5.48). 1H NMR (500 MHz, CD2Cl2, ppm)
d:
7.85e8.31 (m, 9H, pyrenyl protons), 7.36e7.43 (m, 6H, Haryl), 5.03 (s,
2H, pyrenyl-CH2O), 3.10 (septet, 2H, CHMe2), 3.05 (septet, 2H,
C0HMe2), 2.52 (t, 2H, PdCH2CH2CH2C(O)), 2.32 and 2.30 (s, 3 each,
hydrogen bonding and
p
ep
performance of HBPE in functionalizing and solubilizing MWCNTs
and graphene is attributed its unique globular-shaped highly-

3122
L. Xu et al. / Polymer 55 (2014) 3120e3129
N]C(Me)e(Me)C0 ]N), 1.52, 1.51, 1.44, 1.43, 1.35, 1.34, 1.29 and 1.27
(s, 3 each, CHMeMe0, C0HMeMe0), 1.40 (t, 2H, PdCH2CH2CH2C(O)),
0.74 (pentet, 2H, PdCH2CH2CH2 C(O)). 13C NMR (125 MHz, CD2Cl2,
different molecular weights, which were synthesized with catalyst
3 at 27 atm and 5 ꢁC with varying polymerization time. The poly-
merization and polymer purification procedures for the polymers
in Sets 3 and 4 were similar to those we used for synthesizing the
polymers in Set 2, which were reported in our earlier paper [51].
ppm) d
: 182.5 (PdCH2CH2 CH2C(O)), 179.0 and 172.0 (N]CeC0 ]N),
140.5 and 140.6 (Ar, Ar0, Cipso),138.4 and 138.0 (Ar, Ar0, Co),129.5 and
128.6 (Ar, Ar0, Cp), 124.5 and 124.2 (Ar, Ar0, Cm), 132.4, 131.2, 130.5,
128.8, 128.5, 128.0, 127.6, 127.2, 126.5, 126.1, 125.9, 122.1, 125.3,
124.7, 124.6, 124.3 (pyrenyl carbon), 68.5 (pyrenyl-CH2-O), 35.9 and
29.9 (PdCH2CH2CH2C(O)), 29.2 and 28.8 (CHMe, C0HMe), 23.9
(PdCH2CH2CH2C(O)), 23.6, 23.5, 23.0 and 22.9 (CHMeMe0,
C0HMeMe0), 21.4 and 19.6 (N]C(Me)e(Me)C0 ]N).
2.4. Noncovalent solubilization of MWCNTs with PEs in organic
solvents
Noncovalent solubilization of MWCNTs was carried out in
organic solvents (THF, chloroform, heptane and toluene) with
various PEs via a common ultrasonication process. Typically, a
mixture of MWCNTs, polymer, and solvent at
a prescribed
2.3. Synthesis of PEs of different topology, molecular weight, and
end group
composition was added into a 25 mL glass vial and sonicated in a
bath sonicator (Branson 3510 with a measured ultrasonication
power of 70 W) at room temperature for 30 min. The resulting
mixture was left undisturbed overnight and the supernatant was
then filtered through a short plug of glass wool to give a homoge-
neous solution of MWCNTs. UVeVis absorbance spectrum was
subsequently taken on the dispersion after proper dilution by
scanning from 190 to 900 nm (step size: 2 nm) within a 1 cm cell.
The nanotube solubility was calculated from the UVeVis absor-
bance at 500 nm via the BeereLambert law (A ¼ εbc) [38]. The
UVeVis extinction coefficients of 0.0488 mL/(g cm) in chloroform
and 0.0477 mL/(g cm) in both THF and heptane, determined in our
earlier work [38], were used for the calculation. The final nanotube
solubility is the average value obtained on five individually pre-
pared nanotube solutions.
Table 1 summarizes the four sets of PEs synthesized and used in
this work, along with their polymerization conditions and charac-
terization results. In Set 1, three polymers having similar molecular
weight but different chain topologies, LPE, MBPE and HBPE were
synthesized with catalyst 1 under different combinations of
ethylene pressure and temperature in our earlier work. Please refer
to our earlier paper for their detailed polymerization procedures
[50]. The polymers in Set 2 (MBPE27K ~ MBPE84K with the number
indicating number-average molecular weight (Mn) of the polymer)
are narrowly distributed PEs of moderately hyperbranched chain
topology with varying molecular weights, which were synthesized
in our earlier work via “living” ethylene polymerization with
catalyst 1 at 1 atm and 25 ꢁC [51]. Set 3 (M-LPE8K ~ M-LPE98K) is
comprised of narrowly distributed PEs of linear topology but
different molecular weights with a methyl ester end group at the
starting end. They were synthesized by “living” ethylene poly-
merization with catalyst 2 at 27 atm and 5 ꢁC with varying poly-
merization time. Set 4 (P-LPE9K ~ P-LPE89K) is comprised of
pyrene-ended narrow-distributed PEs of linear topology but
2.5. Characterizations and measurements
Nuclear magnetic resonance (1H NMR) spectra of 3 and various
polymers were obtained with a Varian Gemini 2000 spectrometer
or a Bruker AV500 spectrometer in CDCl3 as solvent at ambient
Table 1
Synthesis of four sets of PEs varying in chain topology, molecular weight, and terminal group.
Set
PE sample
Polymerization condition
Triple-detection GPC characterizationa
1H NMR analysis
Branchesc
b
Catalyst
Ethylene pressure
Time (h)
Mn
Mw
PDI
[
h]
Mn,NMR
w
and temperature
(kg/mol)
sc2(kg/mol)
(mL/g)
(kg/mol)
(per 1000C)
1
2
LPE
MBPE
HBPE
1
1
1
1
30 atm, 25 ꢁC
6 atm, 35 ꢁC
1 atm, 35 ꢁC
1 atm, 25 ꢁC
5
6
18
1
2
3.2
4
5
6
1
2
3
4
5
12
18
1
2
3
114
96
94
27
47
63
71
78
84
8
15
22
29
35
74
98
9
17
23
30
36
42
77
89
186
152
156
28
51
74
87
101
112
9
16
23
29
36
82
112
9
19
23
30
37
43
82
99
1.63
1.59
1.66
1.03
1.09
1.17
1.24
1.29
1.33
1.09
1.01
1.01
1.02
1.03
1.12
1.14
1.00
1.01
1.01
1.02
1.02
1.03
1.07
1.11
71
37
16
14
19
24
26
29
31
15
22
28
32
36
58
69
15
22
28
32
36
39
57
64
e
102
97
112
95
106
95
97
96
94
95
92
95
90
88
88
86
94
86
87
82
93
87
91
84
e
e
MBPE27K
MBPE47K
MBPE63K
MBPE71K
MBPE78K
MBPE84K
M-LPE8K
M-LPE15K
M-LPE23K
M-LPE29K
M-LPE35K
M-LPE74K
M-LPE98K
P-LPE9K
P-LPE17K
P-LPE23K
P-LPE30K
P-LPE36K
P-LPE42K
P-LPE77K
P-LPE89K
e
e
e
e
e
e
3
4
2
3
27 atm, 5 ꢁC
7
15
20
28
38
97
168
7
18
23
30
39
52
86
132
27 atm, 5 ꢁC
4
5
6
12
16
a
Number-average molecular weight (Mn), weight-average molecular weight (Mw), and polydispersity index (PDI) were determined with the light scattering detector within
the triple-detection GPC. Weight-average intrinsic viscosity ([h]w) were measured with the viscosity detector within the triple-detection GPC.
b
Number-average molecular weight (Mn,NMR) determined with 1H NMR spectroscopy based on terminal group analysis.
c
Total branch density determined with 1H NMR spectroscopy.

L. Xu et al. / Polymer 55 (2014) 3120e3129
3123
temperature. Triple-detection gel permeation chromatography
1000
100
10
(GPC) analysis of polymers was performed on a Polymer Labora-
tories PL-GPC 220 system equipped with a differential refractive
index (DRI) detector (from Polymer Laboratories), a three-angle
laser light scattering (LS) detector (high-temperature miniDAWN
from Wyatt Technology) and a four-bridge capillary viscosity de-
tector (from Polymer Laboratories). The laser detector had a
wavelength of 687 nm with three scattering angles of 45, 90 and
135ꢁ. The GPC separation system was composed of one guard col-
LPE
Set 3, M-LPE8K ~ M-LPE98K
Set 4, P-LPE9K ~ P-LPE89K
HBPE
HBPE
Set 2, MBPE27K ~ MBPE84K
umn (PL# 1110e1120) and three 30 cm columns (PL gel 10
mm
MIXED-B 300 ꢂ 7.5 mm). HPLC-grade THF was used as the mobile
phase with a flow rate of 1.0 mL/min and the whole GPC system was
maintained at 33 ꢁC. Two narrowly distributed polystyrene (PS)
standards (from Pressure Chemicals, with weight-average molec-
ular weight (Mw) of 30,000 and 200,000 g/mol, respectively) were
used for the normalization of light scattering signals and the
determination of inter-detector delay volume and band broad-
ening. Astra software from Wyatt Technology was used for collec-
tion and process of the data. A dn/dc value of 0.078 mL/g was
adopted for all the PE samples.
UVeVis absorbance spectra were recorded with an Ultra spec
2100 pro spectrophotometer. Transmission electron microscopy
(TEM) measurements were performed on a Philips EM400 micro-
scope operated at 100 keV. TEM samples were made by depositing
a few drops of dilute nanotube solutions onto copper grids coated
with a carbon film, followed with drying in vacuo for 5 min. Ther-
mogravimetric analysis (TGA) was performed on a TA Instruments
Q50 thermogravimetric analyzer under N2 atmosphere with a
heating rate of 10 ꢁC/min.
1
1000
10000
100000
1000000
10000000
Molecular Weight (g/mol)
Fig. 1. MarkeHouwink plots of the four sets of PE samples, constructed with the
intrinsic viscosity and molecular weight data measured with triple-detection GPC
using THF as the eluant at 33 ꢁC.
Noncovalent solubilziation of MWCNTs with the three PE sam-
ples was performed in four common low-polarity organic solvents
(chloroform, THF, n-heptane, toluene), in which all the polymers
show good and increasing solubility following the order at room
temperature. With each polymer in any given solvent, four different
polymer/MWCNT mass ratios (mPE/mNT ¼ 1, 2.5, 5, 7.5, respectively)
were adopted at a fixed MWCNT feed concentration (2 mg in 1 mL
solvent). A common procedure, including sonication, settlement,
and filtration under identical conditions, was used to obtain
nanotube solutions. Negligible nanotube solubility was achieved
when toluene or n-heptane was used as the solvent with all three
polymers. On the contrary, dark nanotube solutions were obtained
with the polymers in THF or chloroform. Some of solutions,
particularly those prepared at high mPE/mNT ratios were found
stable without nanotube settling even after two years. This distinct
solvent effect has been found in our earlier studies [38,48] and
should be attributed to high polymeresolvent interactions in
toluene or n-heptane since the polymers dissolve better in them,
which renders insufficient polymer adsorption on nanotubes and
thus ineffective nanotube solubilization.
3. Results and discussion
3.1. Effect of polymer chain topology
To examine the effect of polymer chain topology on the nano-
tube solubilization, we used a set of three branched PE homopol-
ymers varying in chain topology (Set 1 in Table 1), including LPE,
MBPE, and HBPE, for the noncovalent nonspecific solubilization of
MWCNTs. These three samples were synthesized via chain walking
The nanotube concentration (CNT) in the solutions obtained with
the three polymers in THF or chloroform was quantified from their
UVeVis absorbance at 500 nm after proper dilution by following
the BeereLambert law. The extinction coefficients of 0.0477 L/
(mg cm) in THF and 0.0488 L/(mg cm) in chloroform, which were
determined in our earlier work for the same MWCNTs used herein
[38], were used in the calculations. Fig. 2 compares CNT data
ethylene polymerization with Pdediimine catalyst
1 under
different combinations of ethylene pressure and temperature
(30 atm/25 ꢁC, 6 atm/35 ꢁC, and 1 atm/35 ꢁC, respectively) [50].
Following the well-established chain walking mechanism of
Pdediimine catalysts [44,45], the topology of resulting PEs in this
unique polymerization system depends sensitively on ethylene
pressure and temperature, with an increasingly hyperbranched
topology yielded with decreasing ethylene pressure and/or
increasing temperature. The three polymers should have increas-
ingly compact chain topology in the order from LPE to MBPE and to
HBPE, with LPE being typically linear in topology but containing
mainly short branches, MBPE being moderately hyperbranched,
and HBPE being hyperbranched, as well elucidated in our earlier
papers [50]. Their MarkeHouwink curves (intrinsic viscosity vs.
molecular weight) in Fig. 1 confirms the topological differences
among the three samples, with the intrinsic viscosity curve
consistently shifted down following the above order. Scheme 1(a)
depicts schematically the topological differences among the three
PEs. Despite their very different chain topologies, this set of poly-
mers have very similar average molecular weights and molecular
weight distribution with Mn being about 100 kg/mol and poly-
dispersity index (PDI) being about 1.6. Meanwhile, they also have
similar total branch density of about 100 branch ends per 1000
carbons as per characterization with 1H NMR spectroscopy (see
Table 1). These features make them well suited for the elucidation
of the effect of chain topology on nanotube solubility.
(average data of five solutions prepared individually at each mPE
/
mNT ratio) achieved with the three polymers differing in chain to-
pology. In both solvents, CNT increases significantly with the change
of polymer topology from linear to hyperbranched, i.e., from LPE to
MBPE and to HBPE, at nearly every mPE/mNT ratio. For example, in
THF at the mPE/mNT ratio of 5, CNT increases from 106 mg/L with LPE
to 255 mg/L with MBPE and to 653 mg/L with HBPE. In chloroform
at the mPE/mNT ratio of 5, CNT increases from 47 mg/L with LPE to
563 mg/L with MBPE and to 880 mg/L with HBPE. This clear trend of
change in CNT confirms the significant effect of chain topology on
nanotube solubilization with the use of this class of highly
branched PEs. Consistent with our earlier findings [38], the CNT data
are generally higher in chloroform than the corresponding ones
achieved in THF with each polymer at an identical mPE/mNT ratio.
Along with the negligible nanotube solubilization in toluene and n-
heptane noted above, this also indicates the dramatic effect of the
solvent on nanotube solubilization with this set of polymers by
changing the polymeresolvent interactions. With a given polymer
in either solvent, increasing the mPE/mNT ratio from 1 to 5 generally

3124
L. Xu et al. / Polymer 55 (2014) 3120e3129
Scheme 1. Synthesis of the four sets of PEs, (a) Set 1 and (b) Sets 2e4, differing in chain topology, molecular weight, and end group by ethylene polymerization catalyzed with
various Pdediimine catalysts.
1400
1200
1000
800
600
400
200
0
1400
1200
1000
800
600
400
200
0
in THF
(a)
in CHCl3
(b)
mPE/mNT ratio:
mPE/mNT ratio:
1
1
2.5
5
2.5
5
7.5
7.5
LPE
MBPE
HBPE
LPE
MBPE
HBPE
Polyethylene Samples
Polyethylene Samples
Fig. 2. Nanotube solubility achieved with PEs in Set 1 (i.e., LPE, MBPE, and HBPE) in THF (a) and chloroform (b), respectively, at different mPE/mNT ratios and but the same nanotube
feed concentration of 2 mg/mL.
leads to an increase in CNT, which should be attributed to the
increased adsorption of the polymer on the nanotubes. A continued
increase from 5 to 7.5 somehow results in a slight drop in CNT, a
phenomenon also observed in our earlier studies on nanotube and
graphene solubilization with HBPE at enhanced loadings [38,48].
Fig. 3 shows representative TEM images of the solubilized
nanotubes achieved with the use of LPE. Despite the lowest CNT
data achieved with LPE among the three polymers, the solubilized
nanotubes are well exfoliated and debundled from each other, with
the presence of individual nanotubes and the absence of large ag-
gregates. Similar images were also seen with the nanotubes solu-
bilized with the use of more efficient HBPE or MBPE. In contrast to
the severely aggregated pristine MWCNTs [38], these images
confirm the successful exfoliation and debundling of the nanotubes
with the use of the polymers.
Given the absence of any specific functionality in the polymers,
the mechanism of nanotube solubilization should be ascribed to the
adsorption of the polymers through CHe
p interactions between
Fig. 3. TEM images of solubilized nanotubes obtained with LPE in THF (a) and chloroform (b) at the mPE/mNT ratio of 1.

L. Xu et al. / Polymer 55 (2014) 3120e3129
3125
the polymers and nanotube sidewalls, which renders a polymer
barrier against the aggregation of the nanotubes. In the case with
HBPE, its adsorption on MWCNTs has been directly observed with
TEM and also confirmed with TGA in our earlier work. To also
confirm the adsorption of LPE on the nanotubes herein, two
nanotube solutions obtained with LPE at the mPE/mNT ratio of 5
were filtered and washed extensively with the fresh solvent (to
remove free nonadsorbed polymer) to render two LPE/MWCNT
composites. Fig. 4 shows the TGA curves of the composites, along
with that for pristine MWCNTs. A weight loss of about 20% mainly
in the temperature range of 400e500 ꢁC is observed in the TGA
curves for both composites, confirming the presence of adsorbed
LPE on nanotubes since pristine MWCNTs have no/negligible
weight loss within this temperature range. Similar weight loss was
also found in the HBPE/MWCNT composites prepared under iden-
tical conditions as shown in our earlier paper [38]. It should be
noted that the weight loss found in these composites should belong
to polymers irreversibly adsorbed on the nanotubes. Their
dramatically different CNT data, while at similar amounts of irre-
versibly adsorbed polymer, suggest that those reversibly adsorbed
polymer in the solution may play an important role in facilitating
nanotube solubilization. In the nanotube solutions, such reversibly
adsorbed polymers should be present. But their quantification is
not possible currently.
competition for adsorption between polymer and solvent. For
successful partial adsorption of the polymer to occur, the two
competing interactions, i.e., the polymeresurface interactions
leading to adsorption and the polymeresolvent interactions
resulting in desorption, should be similar [60].
Coleman et al. have modeled polymer adsorption and derived a
simple expression in terms of the Hildebrand solubility parameters
to quantify the effects of the polymer stabilizer and solvent on the
concentration of the exfoliated nanosheets [60]. Also applicable to
the nanotubes, it is accordingly modified as below.
"
#
ꢀ
ꢁ
2
ðdS ꢀ dPÞðdNT ꢀ dSÞ
CNTfexp
ꢀ
(1)
kT
where k is Boltzmann constant; T is the temperature; dS
,
dNT
,
dP are
the solubility parameter for the solvent, nanotube, and polymer,
respectively. With this expression, Coleman et al. have predicted
that the concentration of exfoliated nanosheets/nanotubes is
maximized if the three solubility parameters are very close, which
was supported by experimental evidences in their study of exfoli-
ated graphene [60]. We also attempted to use Eq. (1) to semi-
quantitatively examine the effect of solvent on the CNT value herein.
Given that they are constructed solely with ethylene sequences, we
take the solubility parameter of regular PE, 16.1 MPa1/2, for the
three branched PEs herein [48]. Following Eq. (1), the CNT value
achieved with the use of either polymer in the four different sol-
With the above results, we now discuss the mechanism in the
effect of solvent on CNT in this unique system. Extensive in-
vestigations have been performed in the literature on elucidating
the effects of solvent and the use of polymer on the solubilization of
carbon nanotubes and graphene [3,52e59]. In the case of the sol-
ubilization of carbon nanotubes or graphene in a pure solvent
without any additive (polymer, surfactant, etc.), theoretical ther-
modynamic derivations, confirmed with experimental evidences,
vents
herein
should
increase
in
the
order:
chloroform < THF < toluene < n-heptane. This differs from the
experimental results herein. Possibly, Eq. (1) is oversimplified and
does not fit this particular system herein.
The completely inefficient nanotube solubilization in toluene
and n-heptane with all three polymers indicates the absence of
sufficient polymer adsorption for steric stabilization. This should
result from the excessively strong polymeresolvent interactions in
these two solvents since these two solvents have closer solubility
parameters to that of PEs and are better solvents for PEs [25,32,38].
From the higher CNT values achieved in chloroform than in THF with
each polymer in this set, we reason that the polymeresolvent in-
teractions in chloroform are more appropriately suited for polymer
adsorption than in THF.
We further discuss the mechanism in the effect of polymer chain
topology on CNT in this system. Given their identical chemical
composition, the three PEs in this set, though having different chain
topologies, should have a similar/identical solubility parameter.
With this, Eq. (1) obtained through simplified modeling is futile in
explaining the effect of polymer chain topology on CNT in a given
solvent. We reason that chain topology affects the polymeresolvent
indicate that a good solvent should have a solubility parameter (dS
)
close to that (dNT ¼ 21.25 MPa1/2) of the nanotubes or graphene. In
such good solvents, the enthalpy of mixing is minimized so as to
render effective solubilization of the nanotubes or graphene at high
concentrations [3,58,59].
In the case of polymer-assisted noncovalent solubilization of
nanotubes/graphene, the solvent, like those used herein (dS ¼ 18.7,
18.5, 18.3, and 15.3 MPa1/2 for chloroform, THF, toluene, and n-
heptane, respectively), is often not a good one for the nanotubes/
graphene with an unmatched solubility parameter but is good for
dissolving the polymer. To achieve nanotube solubilization, the
polymer should partially adsorb on the surface at some sites while
with the rest of chain segments/loops protruding into the solvent to
provide steric stabilization [60]. For a given surface, there is a
120
interactions and subsequently the polymer adsorption and CNT
,
pristine MWCNTs
with more linearized chain topology rendering stronger polymer-
esolvent interactions and thus reduced polymer adsorption. As
proposed in our earlier work [38], the globular-shaped HBPE
(having a hydrodynamic diameter at about 10 nm) should adsorb
onto the nanotube sidewall surface at some of its spherical sites
with the rest of the globule protruding into the solvent (see Scheme
2). Due to the highly compact chain conformation and congested
spherical surface of HBPE, the polymeresolvent interactions are
expected to be weaker; meanwhile, the interactions between the
adsorbed and nonadsorbed HBPE globules via physical chain en-
tanglements should also be negligible. In the case with the adsor-
bed LPE, the linear protruding chain segments/loops are expected
to have stronger interactions with the solvent due to less compact
chain conformation and there is enhanced possibility of chain en-
tanglements between the nonadsorbed and adsorbed chains
(Scheme 2). On the other hand, increasing the chain compactness
100
80
LPE/MWCNT composite
prepared in chloroform
60
40
20
0
LPE/MWCNT composite
prepared in THF
0
200
400
600
800
1000
Temperature (°C)
Fig. 4. TGA curves for pristine MWCNTs, and the LPE/MWCNT composites prepared in
chloroform and THF, respectively. The free polymer was removed from both two
composites.

3126
L. Xu et al. / Polymer 55 (2014) 3120e3129
Scheme 2. Proposed noncovalent interactions between MWCNTs and globular HBPE (a) and linear LPE (b) in organic solvents.
from LPE to HBPE is also reasoned to lead to enhanced polymer-
enanotube interactions since the density of chain segments per
volume is enhanced.
PEs of controllable molecular weight with chain topology resem-
bling HBPE was not possible.
Noncovalent solubilization of MWCNTs was performed with
these two sets of narrow-distributed PEs at the mPE/mNT ratio of 5
by following the same procedure used above. Consistent with
above results, negligible nanotube solubilization was achieved in
toluene or n-heptane with any polymer in the two sets while stable
solutions were obtained in THF and chloroform. Fig. 6 plots and
compares the CNT data as a function of Mn obtained with the two
sets of polymers in THF and chloroform, respectively. In particular,
one can see a strong dependence of CNT on molecular weight with
polymers in Set 3. In both THF and chloroform, CNT increases
initially to a peak value, then followed with a consistent decrease,
with the gradual increase of Mn from 8 to 98 kg/mol in Set 3. The
peak-maximum CNT value (689 mg/L in THF and 655 mg/L in
chloroform) occurs at the Mn value of ca. 29 kg/mol in THF and
15 kg/mol in chloroform. At Mn > 74 kg/mol, the CNT values are very
low, i.e., ca. 50 mg/mL in THF and 17 mg/mL in chloroform, which in
good agreement with the low CNT values (106 mg/L in THF and
47 mg/L in chloroform) found with the broad-distributed LPE of
similar Mn in the above section. These results indicate that PEs of
Carbon nanotubes have been extensively demonstrated to
induce crystallization of crystalline polymers, such as high-density
polyethylene (HDPE), in the solution to form nanohybrid shish
kebabs morphology [61e63]. Due to their highly branched struc-
tures, the Pdediimine PEs used herein are noncrystalline under the
current conditions. As we have shown earlier [44], similar
Pdediimine PEs typically crystallize at around ꢀ30 ꢁC in the melt
state. The nanotube-induced polymer crystallization should be
absent in the current system with the polymers used herein, which
is otherwise expected to affect the nanotube solubility.
3.2. Effect of polymer molecular weight
In order to investigate the effect of polymer molecular weight on
CNT, two sets (Sets 2 and 3 in Table 1) of narrow-distributed PEs of
well-defined molecular weights were synthesized and used for the
noncovalent nanotube exfoliation. Their synthesis benefits from
the unique “living” feature of Pdediimine-catalyzed ethylene
polymerization at a low temperature (25 ꢁC or lower), where the
control of molecular weight was achieved simply by tuning the
polymerization time [45]. The polymers in Set 2,
MBPE27K ~ MBPE84K (with the number indicating Mn) synthesized
with 1 at 1 atm/25 ꢁC [51], have nearly identical moderately
hyperbranched chain topology as MBPE in Set 1 since their intrinsic
viscosity data locate on the curve for MBPE (see Fig. 1). The narrow-
distributed polymers in Set 3, M-LPE8K ~ M-LPE98K (with M
standing for methyl-ester-ended PEs and the number indicating
Mn) were synthesized with catalyst 2 at 27 atm/5 ꢁC. Following the
characteristic chain initiation chemistry of chelate catalyst 2 (see
Scheme 1(b)) [45], each polymer in Set 3 should contain about a
methyl ester functionality at the starting end (representatively, see
Fig. 5 for the 1H NMR spectrum of M-LPE23K), which is confirmed
by end group analysis. Meanwhile, the polymers in Set 3 have the
similar linear topology of LPE in Set 1 as per their intrinsic viscosity
data (see Fig. 1). From GPC characterization, these two sets of
polymers all have narrow molecular weight distribution with PDI
generally below or around 1.2. All the polymers have a similar
branching density of about 90 branch ends per 1000 carbons. Due
to the deteriorated “living” behavior of Pdediimine catalysts at the
elevated temperature of 35 ꢁC [51], synthesis of narrow-distributed
Fig. 5. 1H NMR spectra of methyl-ester-ended M-LPE23K (a) in Set
3 and
pyrenemethyl-ester-ended P-LPE23K (b) in Set 4.

L. Xu et al. / Polymer 55 (2014) 3120e3129
3127
Fig. 6. CNT as functions of polymer Mn for nanotube solutions obtained with the polymers in Set 2 (MBPE series) and Set 3 (M-LPE series), respectively, in THF (a) and chloroform (b)
at the mPE/mNT ratio of 5.
linear topology also can efficiently solubilize the MWCNTs in both
solvents at high concentrations, provided that they have appro-
priate molecular weights. The trend of change in CNT with this set of
polymers is hypothesized to result from the subtle change in pol-
ymeresolvent interactions with the increase of Mn in this set of
linear polymers. At very low Mn values (e.g., <8 kg/mol), we reason
that the chain segments (of an adsorbed chain) protruding into the
solvent are relatively too short to provide sufficient barrier against
nanotube aggregation. The initial increase in Mn thus leads to a
thicker barrier, which better help the stabilization of the dispersed
MWCNTs. On the contrary, too high Mn leads to the pulling-off or
desorption of the adsorbed chain due to the excessively enhanced
polymeresolvent interactions [25,32,38].
synthesized earlier [45]. End-group analysis with 1H NMR spec-
troscopy of the polymers (see Fig. 5 for a representative 1H NMR
spectrum) confirms that nearly each chain contains a pyrene end,
with their Mn values determined through NMR end-group analysis
generally in good agreement with those determined with light
scattering in triple-detection GPC analysis (see Table 1). Mean-
while, this set of polymers is also featured with low PDI (around or
below 1.1). From Fig. 1, the intrinsic viscosity data of this set of
polymers locate on the same line as those for polymers in Set 3,
indicative of the identical linear topology of these two sets of
polymers given the same polymerization conditions. Like those in
Set 3, polymers in Set 4 also have similar branching density with ca.
90 branch ends per 1000 carbons.
In the case with polymers in Set 2 of a more compact chain
topology, the dependence of CNT on Mn is much weaker without a
clear conclusive pattern at the low molecular weight end with the
range of polymers herein. Nevertheless, at the high molecular
weight end, CNT shows a trend of slight decrease with the increase
of Mn, which is also indicative of polymer desorption due to
excessively enhanced polymeresolvent interactions. However, at
Mn > 40 kg/mol, CNT data achieved with polymers in Set 2 (MBPE
series) are much higher than those achieved with the counter
polymers in Set 3 (M-LPE series) due to more compact chain to-
pology, which is also in good agreement with the results obtained
in the above section. These indicate that the polymeresolvent in-
teractions show a weaker dependence on Mn with Set 2 polymers of
more compact chain topology.
Noncovalent solubilization of MWCNTs was carried out with
polymers in Set 4 at the mPE/mNT ratio of 5 in all four solvents (THF,
chloroform, n-heptane, and toluene). Different from previous sets
of polymers, we expect the possible involvement of both nonspe-
cific CHep and specific pep interactions between the polymers in
Set 4 and the nanotubes when used for noncovalent solubilization
of the MWCNTs, with the former resulting from the polymer
backbone and the latter from the pyrene end group. The involve-
ment of the specific
pep interactions may possibly provide addi-
tional synergistic effects on CNT. This is confirmed to be the case
when nanotube solubilization was carried out with the polymers in
n-heptane. Dark nanotube solutions with significant CNT values (ca.
100e400 mg/L calculated by using the extinction coefficient of
0.0477 L/(mg cm) in THF) were successfully obtained while it was
not possible with all the other sets of polymers without having the
specific functionality. Fig. 7a shows the dependence of CNT on Mn
with the set of polymers in n-heptane. One can see an initial in-
crease of CNT, followed with a decrease, with the increase of Mn,
which reflect again the gradual increase of polymeresolvent in-
teractions. The highest CNT value of 407 mg/L was achieved with P-
LPE23K at Mn of 23 kg/mol. These CNT data clearly confirms the
3.3. Effect of terminal pyrene group
We further investigate the effect of the presence an additional
pyrene end group in the branched PEs on the nanotube solubility.
Pyrene and similar functionalities have been well demonstrated to
show noncovalent specific
pep interactions with nanotube side-
walls and polymers containing these functionalities have been
extensively applied to render nanotube solubilization in organic
solvents [1e10,13,14,25e32]. Herein, we designed a set of narrow-
distributed PEs (Set 4 in Table 1, P-LPE9K ~ P-LPE89K with P
standing for pyrene-ended chains and the number indicating Mn)
containing a pyrenemethyl ester group at the starting end by
“living” ethylene polymerization with catalyst 3 at 27 atm/5 ꢁC (see
Scheme 1b). Catalyst 3, having the pyrene group designed specif-
ically into its initiating end, was synthesized by reacting 1 with
pyrenemethyl acrylate (CH2]CHCOOCH2Py) following the same
reaction chemistry used for synthesis of 2 and other various func-
tionalized Pdediimine ester chelate catalysts our group
presence of specific
pyrene end group in n-heptane. Meanwhile, the
are reasoned to be the sole ones leading to nanotube solubilization
since CHe interactions, as demonstrated above, are ineffective to
render nanotube solubility in the solvent. Fig. 7b shows a TEM
image of the solubilized nanotubes obtained with P-LPE23K.
Debundled individual nanotubes are clearly seen, confirming the
successful nanotube exfoliation and solubilization. Meanwhile, it is
also noted from Fig. 7a that significant CNT values at ca. 200 mg/L
are still maintained even with polymers of high Mn's (77 and 89 kg/
mol). This is distinctly different from the negligible CNT values
achieved in THF or chloroform with M-LPE series (Set 3) of
p
e
p
interactions with these polymers having a
pep
interactions
p

3128
L. Xu et al. / Polymer 55 (2014) 3120e3129
Fig. 7. (a) CNT as a function of Mn for nanotube solutions obtained with the polymers in Set 4 (P-LPE series) in n-heptane at the mPE/mNT ratio of 5; (b) TEM image of solubilized
nanotubes obtained with the P-LPE23K in Set 4 in n-heptane at the mPE/mNT ratio of 1.
polymers at similar molecular weights. It confirms that specific
interactions between the pyrene end group and nanotube
sidewalls are much stronger than the nonspecific CHe in-
two sets of polymers, we reason that CHep interactions should still
be the predominant ones effecting the nanotube solubilization in
these two solvents with polymers in Set 4. With the absence of
pe
p
p
teractions. Despite the success in n-heptane, negligible nanotube
solubilization was achieved with any polymers in Set 4 in toluene.
This once again should result from insufficient polymer adsorption
due to the over-strong polymeresolvent interactions in toluene
since it is the best solvent among the four for the polymers.
We further investigate the nanotube solubilization with poly-
mers in Set 4 in THF and chloroform. Fig. 8 shows CNT data obtained
with polymers in Set 4 in THF and chloroform, respectively. Those
achieved above with polymers in Set 3 (i.e., M-LPE series) under
identical conditions are also included in Fig. 8 for comparison to
investigate the effect of end group, since these two sets of polymers
have identical linear chain topology and similar molecular weight
range but different end group. In general, there are no distinct
differences in the CNT data in both solvents with the polymers from
the two different sets at similar molecular weight. However, a weak
but notable trend with polymers in Set 4 is that high CNT values (ca.
676 mg/L) were obtained in THF even with polymers of low mo-
lecular weight (9 kg/mol). With the increase of Mn from 9 to 42 mg/
L, CNT does not show distinct changes in both solvents. But low CNT
data (ca. 42 mg/L in chloroform and ca. 30 mg/L in THF) were ob-
tained with P-LPE77K and P-LPE89K of higher molecular weights in
the set, which is similar with M-LPE set of polymers at similar Mn
values and is indicative of insufficient polymer adsorption due to
stronger polymeresolvent interactions at higher polymer molec-
ular weights. From the marginal differences in CNT data with the
distinguishable pep interactions in THF and chloroform, this is
different from the case seen above in n-heptane. We suspect that, in
chloroform and THF, the polymer coils are not fully relaxed and the
pyrene end groups may be buried within the coils without getting
exposed to the nanotube sidewalls, thus rendering insufficient pep
interactions. On the contrary, the polymer coils are more relaxed in
n-heptane with the pyrene group exposable to the nanotube sur-
face for interactions given that n-heptane is a better solvent for the
polymers than THF and chloroform.
4. Conclusions
Four sets of highly branched Pdediimine polyethylenes varying
in structural parameters (including chain topology, molecular
weight, and end group) were designed/synthesized in this work for
the noncovalent solubilization of MWCNTs in four different solvents
(THF, chloroform, n-heptane, and toluene). The three structural pa-
rameters as well as the solvent have been demonstrated to exert
complex sensitive effects on the nanotube solubility. With the
polymers in Set 1 of different chain topology but similar high mo-
lecular weight (Mn ¼ 100 kg/mol), we have found that changing the
chain topology from linear (LPE) to moderately hyperbranched
(MBPE) and to hyperbranched (HBPE) leads to significantly
enhanced CNT values in both THF and chloroform. With narrow-
distributed polymers of linear topology in Set 3 (M-LPE series), it is
Fig. 8. CNT as functions of polymer Mn for nanotube solutions obtained with the polymers in Set 3 (M-LPE series) and Set 4 (P-LPE series), respectively, in THF (a) and chloroform (b),
at a fixed mPE/mNT ratio of 5.

L. Xu et al. / Polymer 55 (2014) 3120e3129
3129
[17] Lee HW, Yoon Y, Park S, Oh JH, Hong S, Liyanage LS, et al. Nat Commun
2011;2:541.
[18] Nish A, Hwang J-Y, Doig J, Nicholas RJ. Nat Nanotech 2007;2:640.
[19] Star A, Stoddart JF, Steuerman D, Diehl M, Boukai A, Wong EW, et al. Angew
Chem Int Ed 2001;40:1721.
[20] Star A, Liu Y, Grant K, Ridvan L, Stoddart JF, Steuerman DW, et al. Macro-
molecules 2003;36:553.
[21] Chen Y, Malkovskiy A, Wang X-Q, Lebron-Colon M, Sokolov AP, Perry K, et al.
ACS Macro Lett 2012;1:246.
demonstrated that their Mn (8e98 kg/mol) shows a dramatic effect
on CNT in both THF and chloroform, with high CNT values (689 and
655 mg/L in THF and chloroform, respectively) achievable at opti-
mum Mn (29 and 15 kg/mol inTHFand chloroform, respectively). Too
lowor too high Mn leads to low CNT values dueto insufficient polymer
barrier layer or excessive polymeresolvent interactions. For the case
with narrow-distributed moderately hyperbranched polymers in Set
2 (MBPE series), their Mn (27e84 kg/mol) has no distinguishable
effects on CNT in both THF and chloroform, generally with high CNT
values (600e800 mg/L) maintained across the whole Mn range
except a slight decrease at the high molecular weight end. All the
polymers in Sets 1e3 were found ineffective for the nanotube solu-
bilization in n-heptane and toluene due to insufficient polymer
adsorption as a result of excessive polymeresolvent interactions in
these two solvents. With the narrow-distributed pyrene-ended
polymers in Set 4, nanotube solubilization was, however, made
[22] Lee JU, Huh J, Kim KH, Park C, Jo WH. Carbon 2007;45:1051.
[23] Zhao H, Yuan WZ, Tang L, Sun JZ, Xu H, Qin A, et al. Macromolecules 2008;41:
8566.
[24] Zou J, Khondaker SI, Huo Q, Zhai L. Adv Funct Mater 2009;19:479.
ꢀ
[25] Gomez FJ, Chen RJ, Wang D, Waymouth RM, Dai H. Chem Commun 2003;190.
[26] Lou X, Daussin R, Cuenot S, Duwez A-S, Pagnoulle C, Detrembleur C, et al.
Chem Mater 2004;16:4005.
[27] Yuan WZ, Mao Y, Zhao H, Sun JZ, Xu HP, Jin JK, et al. Macromolecules 2008;41:
701.
[28] Bahun GJ, Wang C, Adronov A. J Polym Sci Part A: Polym Chem 1941;2006:44.
[29] Bahun GJ, Androv A. J Polym Sci Part A: Polym Chem 2010;48:1016.
[30] Meuer S, Braun L, Zentel R. Chem Commun; 2008:3166.
[31] Meuer S, Braun L, Schilling T, Zentel R. Polymer 2009;50:154.
[32] Yang L-P, Pan C-Y. Macromol Chem Phys 2008;209:783.
[33] Rouse JH, Lillehei PT. Nano Lett 2003;3:59.
[34] Mamedov AA, Kotov NA, Prato M, Guldi DM, Wicksted JP, Hirsch A. Nat Mater
2002;1:190.
[35] Huang S-CJ, Artyukhin AB, Wang Y, Ju J-W, Stroeve P, Noy A. J Am Chem Soc
2005;127:14176.
[36] Kim JB, Premkumar T, Lee K, Geckeler KE. Macromol Rapid Commun 2007;28:
276.
[37] O'Connell MJ, Boul P, Ericson LM, Huffman C, Wang Y, Haroz E, et al. Chem
Phys Lett 2001;342:265.
[38] Xu L, Ye Z, Cui Q, Gu Z. Macromol Chem Phys 2009;210:2194.
[39] Baskaran D, Mays JW, Bratcher MS. Chem Mater 2005;17:3389.
[40] Wang D, Chen L. Nano Lett 2007;6:1480.
[41] Liu A, Honma I, Ichihara M, Zhou H. Nanotechnology 2006;17:2845.
[42] Naito M, Nobusawa K, Onouchi H, Nakamura M, Yasui K-I, Ikeda A, et al. J Am
Chem Soc 2008;130:16697.
[43] Meyer F, Minoia A, Raquez JM, Spasova M, Lazzaronia R, Dubois P. J Mater
Chem 2010;20:6873.
[44] Dong Z, Ye Z. Polym Chem 2012;3:286.
[45] Ye Z, Xu L, Dong Z, Xiang P. Chem Commun 2013;49:6235.
[46] Petrie K, Docoslis A, Vasic S, Kontopoulou M, Morgan S, Ye Z. Carbon 2011;49:
3378.
[47] Oaszuwa O, Petrie K, Kontopoulou M, Xiang P, Ye Z, Docoslis A. Compos Sci
Tech 2012;73:27.
possible in n-heptane through the stronger specific
pep in-
teractions. Therein, theCNT value shows a distinct dependenceon Mn,
with a high CNT value (407 mg/L) obtained at the optimum Mn (23 kg/
mol). Nevertheless, such
when THF or chloroform was used as the solvent, where nanotube
solubilizationwas achieved still through primarily nonspecific CHe
pep interactions were found negligible
p
interactions. Meanwhile, it was still not possible to disperse the
nanotubes in toluene with these pyrene-ended polymers also due to
excessively strong polymeresolvent interactions.
Acknowledgment
The financial support for this work from the Natural Sciences
and Engineering Research Council of Canada, the Canadian Foun-
dation for Innovation (CFI), and the Canada Research Chair (CRC) is
greatly appreciated. L.X. thanks the Ontario Ministry of Economic
Development and Innovation for awarding a postdoctoral fellow-
ship and the National Natural Science Foundation of China
(#21074117) for funding support.
[48] Xu L, McGraw J-W, Gao F, Grundy M, Ye Z, Gu Z, et al. J Phys Chem C 2013;117:
10730.
[49] Johnson LK, Killian CM, Brookhart M. J Am Chem Soc 1995;117:6414.
[50] Ye Z, Zhu S. Macromolecules 2003;36:2194.
References
[51] Xu Y, Xiang P, Ye Z, Wang W-J. Macromolecules 2010;43:8026.
[52] Ausman KD, Piner R, Lourie O, Ruoff RS, Korobov M. J Phys Chem B 2000;104:
8911.
[53] Bahr JL, Mickelson ET, Bronikowski MJ, Smalley RE, Tour JM. Chem Commun
2001;193.
[54] Furtado CA, Kim UJ, Gutierrez HR, Pan L, Dickey EC, Eklund PC. J Am Chem Soc
2004;126:6095.
[55] Landi BJ, Ruf HJ, Worman JJ, Raffaelle RP. J Phys Chem B 2004;108:17089.
[56] Maeda Y, Kimura S, Hirashima Y, Kanda M, Lian YF, Wakahara T, et al. J Phys
Chem B 2004;108:18395.
[57] Umek P, Vrbanic D, Remskar M, Mertelj T, Venturini P, Pejovnik S, et al. Carbon
2002;40:2581.
[58] Bergin SD, Nicolosi V, Streich PV, Giordani S, Sun Z, Windle AH, et al. Adv
Mater 1876;2008:20.
[59] Bergin SD, Sun Z, Rickard D, Streich PV, Hamilton JP, Coleman JN. ACS Nano
2009;3:2340.
[60] May P, Khan U, Hughes JM, Coleman JN. J Phys Chem C 2012;116:11393.
[61] Li CY, Li L, Cai W, Kodjie SL, Tenneti KK. Adv Mater 2005;17:1198.
[62] Li L, Li CY, Ni C. J Am Chem Soc 2006;128:1692.
[63] Laird ED, Li CY. Macromolecules 2013;46:2877.
[1] Singh P, Campidelli S, Giordani S, Bonifazi D, Bianco A, Prato M. Chem Soc Rev
2009;38:2214.
[2] Karousis N, Tagmatarchis N, Tasis D. Chem Rev 2010;110:5366.
[3] Coleman JN. Adv Funct Mater 2009;19:3680.
ꢀ
[4] Kharisov BI, Kharissova OV, Gutierrez HL, Mendez UO. Ind Eng Chem Res
2009;48:572.
[5] Kim SW, Kim T, Kim YS, Choi HS, Lim HJ, Yang SJ, et al. Carbon 2012;50:3.
[6] Tasis D, Tagmatarchis N, Bianco A, Prato M. Chem Rev 2006;106:1105.
[7] Mittal V, editor. Surface modification of nanotube fillers. Weinheim: Wiley-
VCH; 2011.
[8] Adeli M, Soleyman R, Beiranvand Z, Madani F. Chem Soc Rev 2013;42:5231.
[9] Sun J-T, Hong C-Y, Pan C-Y. Polym Chem 2011;2:998.
[10] Szleifer I, Yerushalmi-Rozen R. Polymer 2005;46:7803.
[11] Sakellariou G, Priftis D, Baskaran D. Chem Soc Rev 2013;42:677.
[12] Homenick CM, Lawson G, Adronov A. Polym Rev 2007;47:265.
[13] Bilalis P, Katsigiannopoulos D, Avgeropoulos A, Sakellariou G. RSC Adv 2014;4:
2911.
[14] Zhao Y-L, Stoddart JF. Acc Chem Res 2009;42:1161.
[15] Pang X, Imin P, Zhitomirsky I, Adronov A. Macromolecules 2010;43:10376.
[16] Imin P, Imit M, Adronov A. Macromolecules 2011;44:9138.