Synthesis of well-defined side chain fullerene polymers and study of their self-aggregation behaviors

Article abstract of DOI:10.1002/pola.26748

Monoalkynyl-functionalized fullerene was precisely synthesized starting with pristine fullerene (C₆₀) and characterized by multiple techniques. Methyl methacrylate and 6-azido hexyl methacrylate were then randomly copolymerized via reversible a

Full text of DOI:10.1002/pola.26748

JOURNAL OF
POLYMER SCIENCE
WWW.POLYMERCHEMISTRY.ORG
ARTICLE
Synthesis of Well-Defined Side Chain Fullerene Polymers and Study of
Their Self-Aggregation Behaviors
Junting Li, Brian C. Benicewicz
Department of Chemistry and Biochemistry, University of South Carolina, Columbia, South Carolina 29208
Correspondence to: B. C. Benicewicz (E-mail: benice@sc.edu)
Received 1 March 2013; accepted 20 April 2013; published online
DOI: 10.1002/pola.26748
ABSTRACT: Monoalkynyl-functionalized fullerene was precisely
synthesized starting with pristine fullerene (C60) and character-
ized by multiple techniques. Methyl methacrylate and 6-azido
hexyl methacrylate were then randomly copolymerized via re-
versible addition fragmentation chain transfer polymerization
to build polymer backbones with well-controlled molecular
weights and copolymer compositions. Finally, these two moi-
eties were covalently assembled into a series of well-defined
side chain fullerene polymers (SFPs) via the copper-mediated
click reaction which was verified by Fourier transform infrared
spectroscopy and 1H NMR. The fullerene loadings of the result-
ant polymers were estimated by thermogravimetric analysis
and UV–vis spectroscopy, demonstrating consistent and high
conversions in most of the samples. The morphology studies
of the SFPs were performed both in solution and on solid sub-
strates. Very intriguing self-aggregation behaviors were
detected by both gel permeation chromatography and dynamic
light scattering analyses. Furthermore, the scanning electron
microscopic images of these polymers showed the formation
of various supramolecular nanoparticle assemblies and crystal-
line-like clusters depending on the fullerene contents and poly-
C
V
mer chain lengths. 2013 Wiley Periodicals, Inc. J. Polym. Sci.,
Part A: Polym. Chem. 2013, 00, 000–000
KEYWORDS: click reaction; fullerenes; reversible addition frag-
mentation chain transfer (RAFT); self-assembly
situation occurs in fullerene star-shaped polymers.8 C60
containing dendrimers are another type of interesting
architecture, but are typically prepared as low-molecular-
weight materials.9–11
-
INTRODUCTION Since its discovery, fullerene has attracted
much attention due to its unique and interesting proper-
ties, such as superconductivity, ferromagnetism, anti-HIV
bioactivity, and optical nonlinearity. However, its applica-
tions are seriously limited because pristine fullerene has
very poor compatibility with most other materials. The
covalent combination of fullerene with polymers is an
effective strategy to overcome this disadvantage and cre-
ate novel fullerene-based architectures. Generally, fuller-
ene polymers can be classified into the following types
according to the different positions of C60 moieties in the
polymer structures: main chain fullerene polymers, side
chain fullerene polymers (SFPs), fullerene-capped poly-
mers, star-shaped fullerene polymers, and dendrimers.1,2
Synthesis of polymers with C60 units in the main chain
involves fullerene-based monomers having two reacting
sites, which are relatively difficult to produce and purify,
In contrast, SFPs can have relatively well-defined structures,
high C60 loadings and molecular weights, although their syn-
theses can be quite challenging. Wudl and coworkers12 first
tried to prepare SFPs by step polymerization using C60-con-
taining monomers. Because of the steric hindrance of the C60
moieties, the degree of polymerization was very low. Alterna-
tively, anchoring fullerene on preformed polymers can avoid
the steric hindrance during polymerization, but an effective
reaction is needed to achieve a controlled attachment. In a
recent publication by the same group, a “rod-coil” diblock co-
polymer containing poly(3-hexylthiophene) (P3HT) and full-
erene was synthesized by combining reversible addition-
fragmentation chain transfer (RAFT) polymerization and a
and
a small amount of multifunctionalized fullerene
subsequent polymer-analogous cycloaddition.13
A similar
impurities can result in severe crosslinking during the
polymerization.3 Fullerene-capped polymers usually ex-
hibit excellent solubility and compatibility, but high full-
erene loadings cannot be attained because the C60
block copolymer was reported by Jo et al., but the “coil”
block was formed by atom transfer radical polymerization,
and the fullerene moieties were attached via a carboxylic
acid–alcohol coupling reaction.14 Hawker et al.15 first
moieties only exist at the chain ends.4–7
A similar
Additional Supporting Information may be found in the online version of this article.
C
V
2013 Wiley Periodicals, Inc.
WWW.MATERIALSVIEWS.COM
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000
1

JOURNAL OF
POLYMER SCIENCE
ARTICLE
WWW.POLYMERCHEMISTRY.ORG
reported the cycloaddition between azides and C60, and used
it to produce styrene-based random copolymers with fuller-
ene on the side chains.16 Hadziioannou and coworkers syn-
thesized similar polymer backbones by nitroxide-mediated
radical polymerization and anchored C60 moieties via either
Chemical and used as received. Methyl methacrylate (MMA;
99%, Acros) was passed through a basic alumina column to
remove inhibitors before use. 2,20-Azobis(4-methoxy-2,4-di-
methyl valeronitrile) (V-70) was purchased from Wako
Chemicals and used as received. Unless otherwise specified,
all chemicals were purchased from Fisher Scientific and used
as received.
an atom transfer radical addition17 or a cycloaddition to
18
C60
.
Through a direct fullerenation, Celli et al. prepared
polysulfones with C60 randomly connected to the side
chains.19 Yang and coworkers postfunctionalized side chains
of a P3HT derivative with C60 by adding sarcosine to create
Instrumentation
NMR spectra were recorded on Varian Mercury 300 and 400
spectrometers using CDCl3 as the solvent. Matrix-assisted
laser desorption-ionization time-of-flight mass spectrometry
(MALDI-TOF-MS) was performed with a Bruker Ultraflex
MALDI tandem time-of-flight mass spectrometer. Fourier
transform infrared spectroscopy (FTIR) spectra were
recorded using a PerkinElmer Spectrum 100 FTIR Spectrom-
eter. UV–vis absorption spectra were taken on a Perkin-
Elmer Lamda 4C UV/vis spectrophotometer. Molecular
weights and polydispersity indices (PDI 5 Mw/Mn) were
determined using a Waters gel permeation chromatograph
equipped with a 515 HPLC pump, a 2410 refractive index
detector, and three Styragel columns (HR1, HR3, HR4 in the
effective molecular weight range of 100–5000, 500–30,000,
andꢀ 5000–500,000, respectively) with THF as the eluent at
30 C and a flow rate of 1.0 mL min21. The GPC system was
calibrated with poly(methyl methacrylate) from Polymer
Laboratories. The thermal stability of the polymers was
determined by thermogravimetric analysis (TGA) performed
on vacuum-dried polymer samples from 30 to 600 ꢀC using
a TA Instruments Q5000 with a nitrogen flow rate of 20 mL
min21 and heating rate of 10 ꢀC min21. UV–vis absorption
spectra were taken on a Perkin-Elmer Lamda 4C UV/vis
spectrophotometer. The DLS experiments were conducted
using a Zetasizer Nano S instrument. The laser wavelength
was 633 nm and the detector position was at 173ꢀ. SEM
images were captured using a Zeiss Ultraplus Thermal field
emission SEM, and the samples were prepared on silicon
wafers by spin-coating at 3000 rpm.
a phenyl linking bridge.20 Also, Rusen et al.21 made C60
-
grafted polyethylene at 100 ꢀC based on the reaction of C60
with amino groups which were introduced earlier along the
main chains. However, in most of these cases the architec-
tures and C60 loadings were not well controlled, because it
was difficult to prevent multiple reactions on the same C60
molecule when pristine fullerene was involved in the attach-
ment process. Generally, the methods of attachment were
not effective enough to make polymers possessing carefully
adjustable fullerene contents.
Herein, we describe our work on the fabrication of SFPs by
combining RAFT polymerization and the copper-mediated
click reaction. As both of these techniques feature good con-
trol, mild reaction conditions, and functional group tolerance,
the combination is expected to be a convenient approach to
prepare well-defined linear SFPs.22–24 Methacrylate-based
monomers were chosen to prepare the polymer backbones
because of their good compatibility with fullerene,14,25,26 and
also their mechanical properties, optical transparency, and
stability to photoageing.27 Moreover, the synthesis of a solu-
ble and monofunctionalized fullerene derivative for the post-
functionalization is depicted which prevented crosslinking of
the polymer chains.
Additionally, the assembly behaviors of the prepared poly-
mers were investigated in solution by gel permeation chro-
matography (GPC) and dynamic light scattering (DLS), and
on solid substrates using scanning electron microscopy
(SEM). The SFPs displayed a variety of self-aggregation
behaviors. The SEM images of the SFPs on silica wafers
showed the formation of various nanoparticle assemblies
and crystalline-like clusters depending on fullerene contents
and chain lengths of the SFP samples. The study of the self-
assembly of fullerene derivatives into supramolecular archi-
tectures is always a significant challenge.28–30 Although
Synthesis of 3,5-Bis(octyloxy)phenyl Methanol (2) and
3-(3,5-Bis(octyloxy)benzyloxy)-3-oxopropanoic Acid (3)
The syntheses of compounds 2 and 3 were carried out
according to the methods in the literature.33
Synthesis of 3,5-Bis(octyloxy)benzyl Propy-2-nyl
Malonate (4)
Compound 3 (6.31 g, 14.0 mmol), propargyl alcohol (788 mg,
14.0 mmol), and 4-(dimethylamino) pyridine (DMAP) (512
mg, 4.70 mmol) were dissolved in methylene chloride (100
mL). Dicyclohexylcarbodiimide (DCC) (2.89 g, 14.0 mmol) in
30 mL of methylene chloride was added dropwise into the
solution with stirring at 0 ꢀC. The mixture was allowed to
slowly warm to room temperature and, after stirring over-
night, filtered, and evaporated. Silica gel column chromatog-
raphy (3:2 mixture of hexane and methylene chloride)
yielded compound 4 as a colorless oil (5.41 g, 79%).
many such investigations were performed on fullerene-
capped polymers
best of our knowledge, detailed morphology studies on SFPs
have not been previously reported.
4,5,31
and fullerene dendrimers,11,32 to the
EXPERIMENTAL
Materials
Fullerene (C60) was purchased from SES Research and used
as received. Tetrahydrofuran (THF; 99.9%, Acros) was dried
over CaH2 overnight and distilled before use. 4-Cyanopenta-
noic acid dithiobenzoate (CPDB) was purchased from Strem
1H NMR (300 MHz, CDCl3): d 5 0.88 (t, J 5 6.6 Hz, 6H, CH3),
1.37 (m, 20H, CH3(CH2)5CH2CH2O), 1.74 (m, 4H, CH3(CH2)5
2
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000

JOURNAL OF
POLYMER SCIENCE
WWW.POLYMERCHEMISTRY.ORG
ARTICLE
1
CH2CH2O), 2.49 (t, J 5 2.3 Hz, 1H, CBCH), 3.48 (s, 2H,
OCCH2CO), 3.92 (t, J 5 6.5 Hz, 4H, CH3(CH2)6CH2O), 4.73
(d, J 5 1.2 Hz, 2H, CH2CBCH), 5.09 (s, 2H, PhCH2O), 6.40 (d,
J 5 2.4 Hz, 1H, Ar), 6.46 (d, J 5 2.1 Hz, 2H, Ar). 13C NMR
(400 MHz, CDCl3): d 5 14.06, 22.65, 26.04, 29.22, 29.24,
29.34, 31.81, 41.04, 52.75, 67.20, 67.93, 75.42, 75.50,
101.09, 106.31, 137.19, 160.41, 165.53, 165.75. FTIR: 1741
cm21 (C@O) and 3291 cm21 (CBCH). HRMS (EI): calcd. for
analyzed via H NMR. The feed ratios varied according to the
requirements for different random copolymers.
Typical Click Reactions Between Poly(MMA-r-AHMA) and
Alkynyl-Functionalized Fullerene (1)
A sample of poly(MMA-r-AHMA) with a known copolymer
composition was reacted with compound 1 for example: Pol-
y(MMA-r-AHMA) (Mn 5 15,718, PDI 5 1.15, [AHMA]:[MMA] 5
1:11) (200 mg, 1 equiv. of N3), compound 1 (184 mg, 1
equiv. of alkyne), and N,N,N’,N’,N"-pentamethyldiethylene tria-
mine (16 lL, 0.5 equiv.) were dissolved in toluene (50 mL).
The solution was degassed by flushing with nitrogen for 30
min and then CuBr (11 mg, 0.5 equiv.) was added. The mix-
ture was stirred under nitrogen protection at room tempera-
ture for one day. The mixture was then diluted with
methylene chloride and passed through neutral alumina to
remove the copper catalyst and unreacted compound 1. After
concentration by rotary evaporation, the product was pre-
cipitated in hexane, filtered, and dried under vacuum. The
feed ratios were adjusted when polymers with different co-
polymer compositions were used.
C
29H44O6 [M]1 488.3124; found 488.3138.
Synthesis of Alkynyl-Functionalized Fullerene (1)
1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU) (0.310
2.08 mmol) was added at room temperature to a solution of
compound (447 mg, 0.915 mmol), C60 (600 mg,
mL,
4
0.832 mmol), and iodine (264 mg, 1.04 mmol) in toluene
(600 mL), and the mixture was stirred for 7 h. The mixture
was filtered through a short plug of silica gel and washed by
methylene chloride (100 mL). Silica gel column chromatogra-
phy (1:1 mixture of hexane and toluene) yielded compound
1 as dark red glassy solids (547 mg, 54%).
1H NMR (300 MHz, CDCl3): d 5 0.89 (m, 6H, CH3), 1.36 (m,
20H, CH3(CH2)5CH2CH2O), 1.75 (m, 4H, CH3(CH2)5CH2CH2O),
2.60 (t, J 5 2.6 Hz, 1H, CBCH), 3.91 (t, J 5 6.5 Hz, 4H,
CH3(CH2)6CH2O), 5.04 (d, J 5 2.1 Hz, 2H, CH2CBCH), 5.45 (s,
2H, PhCH2O), 6.42 (d, J 5 2.7 Hz, 1H, Ar), 6.61 (d, J 5 2.1 Hz,
2H, Ar). 13C NMR (400 MHz, CDCl3): d 5 14.12, 22.68, 26.12,
29.27, 29.39, 31.83, 51.26, 54.41, 68.16, 69.16, 71.18,
101.70, 107.25, 136.42, 138.93, 139.38, 140.89, 140.95,
141.84, 141.86, 142.19, 142.20, 142.96, 142.99, 143.01,
143.05, 143.84, 143.87, 144.53, 144.67, 144.71, 144.90,
144.91, 145.02, 145.15, 145.17, 145.24, 145.29, 160.49,
162.86, 163.13. FTIR: 1749 cm21 (C@O) and 3302 cm21
(CBCH). MALDI-TOF-MS: calcd. for C89H42O6 [M 1 Na]1
1229.28; found 1229.3.
RESULTS AND DISCUSSION
Modification of Fullerene
A highly soluble fullerene derivative compound 1 with a
“clickable” functional group was designed to create an exclu-
sive reactive site on C60 for the side chain functionalization
of prepolymers. The synthesis is depicted in Scheme 1. Com-
pound 3 was produced by following Felder’s procedure with
3,5-dihydroxybenzyl alcohol and 1-bromooctane as the start-
ing materials.33 Then the coupling reaction between the car-
boxylic acid and propargyl alcohol in the presence of DCC
and DMAP generated compound 4, which was anchored to a
C60 molecule via a facile Bingel cyclopropanation.35
Synthesis of 6-Azidohexyl Methacrylate
The synthesis of 6-azidohexyl methacrylate (AHMA)
was carried out according to the methods published
previously.34
In the final step, an excess amount of compound 4 (1.1
equiv.) was reacted with C60 to afford compound 1 as the
major product, which was purified through a silica gel col-
umn and identified by NMR analyses. As multifunctionalized
fullerene can lead to reticulate structures as mentioned ear-
lier, byproducts carrying more than one alkynyl group were
highly undesirable. To ensure that the product did not con-
tain this type of impurity, MALDI-TOF-MS was performed for
the further characterization (Fig. 1). Two expected charged
peaks were displayed at m/z 5 1206.3 and 1229.3 (calcu-
lated m/z 5 1206.29 and 1229.28), corresponding to the mo-
lecular ion peak of compound 1 and [M1Na]1, respectively.
If a difunctionalized fullerene was present, peaks at approxi-
mately m/z 5 1692.6 and 1715.6 would be expected. There-
fore, the absence of these peaks demonstrated that only
monoalkynyl functionalized fullerene was obtained.
“Caution: special care should be taken to minimize the possi-
ble hazards in the preparation and handling of the azide
compounds.”
Typical RAFT Polymerization of 6-Azidohexyl
Methacrylate and Methyl Methacrylate
A solution of AHMA (0.22 g, 1.0 mmol), MMA (1.0 g,
10.0 mmol), CPDB (10.3 mg, 36.7 lmol g21), V-70 (1.14 mg,
3.70 lmol g21), and THF (1.1 mL) were prepared in a dried
Schlenk tube. The mixture was degassed by three freeze-
pump-thaw cycles, backfilled with nitrogen, and then placed
ꢀ
in an oil bath at 40 C for various intervals. The polymeriza-
tion solution was quenched in ice water and poured into an
aluminum boat. The solvent and monomer were removed by
evaporation in a fume hood overnight and then 1 day under
vacuum. Monomer conversion was determined by gravimet-
ric analysis, molecular weight characteristics were analyzed
by GPC, and the copolymer composition of each residue was
In addition, most reactions involving fullerene require a large
amount of solvent and produce relatively low yields because
of its poor solubility (ꢁ2.8 mg mL21 in toluene maximum).36
In our current design, the fullerene modification not only
grafted a monoalkynyl group on the C60 molecules for the
subsequent click reaction, but also introduced two long alkyl
WWW.MATERIALSVIEWS.COM
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000
3

JOURNAL OF
POLYMER SCIENCE
ARTICLE
WWW.POLYMERCHEMISTRY.ORG
SCHEME 1 Synthetic route for the mono-alkynyl functionalized fullerene (compound 1).
chains which significantly improved the solubility in organic
solvents, such as toluene, methylene chloride and THF. A
25 mg mL21 solution of compound 1 in toluene was kept in
a refrigerator for 1 year without precipitation.
fullerene moieties to ameliorate the rigidity imposed by the
backbone.12 Besides homopolymer, random copolymers pol-
y(AHMA-r-MMA) were prepared by adding MMA in the poly-
merization to vary the fullerene content along the SFP
chains. Thus, both fullerene content and chain length could
be adjusted independently to study the effects of these varia-
bles on the polymer properties.
RAFT Polymerization
Azido-containing monomers can be polymerized by living
radical polymerization with controlled molecular weight and
narrow polydispersity.34,37,38 A relatively low temperature
(40 ꢀC) was applied to minimize possible side reactions
between the azide and C@C bond of the monomer
(AHMA).38,39 The six-carbon side chains of the resultant
polymer provide relatively long and flexible tethers for the
To demonstrate the controllability of the RAFT copolymeriza-
tion of AHMA and MMA, a kinetics study (Supporting Infor-
mation Fig. S1) was performed with a fixed feed ratio of
AHMA to MMA (1:20). Figure 2(a) shows a pseudo-first-
order kinetics plot indicating a constant free radical concen-
tration in the polymerization process. The number average
molecular weights (Mn) increased linearly with monomer
conversion and were in agreement with the theoretical pre-
dictions [Fig. 2(b)]. Moreover, the PDIs were maintained
below 1.2 with conversions up to 82%. Generally, the RAFT
copolymerization of AHMA and MMA was well-controlled,
and the kinetics were similar to the polymerization of other
azido-functionalized methacrylate which we previously
studied.38
The fullerene loading of the SFPs was adjusted by simply
altering the feed ratios of AHMA to MMA in the syntheses of
the prepolymers. Consequently, a series of prepolymers with
variable molecular weights and compositions were prepared
for the subsequent click reactions. Table 1 shows that the
compositions of the prepolymers were consistent with the
feed ratios of the corresponding polymerizations, indicating
FIGURE 1 MALDI-TOF-MS spectrum of compound 1.
4
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000

JOURNAL OF
POLYMER SCIENCE
WWW.POLYMERCHEMISTRY.ORG
ARTICLE
was converted to triazole, the methylene protons next to the
original azido group shifted downfield from 3.30 ppm to
4.37 ppm, and the proton on the triazole ring was detected
at about 7.8 ppm (Supporting Information Fig. S4). A note-
worthy issue is that the 1H NMR signals were too weak to
provide quantitative integrations after the click reaction, pre-
sumably because of the shielding effect of the fullerene moi-
eties and the limited solubility of the polymers.
Calculation of Fullerene Loadings by TGA and UV–vis
Spectrometry
TGA scans in nitrogen of pristine fullerene, compound 1, the
prepolymers and the SFPs with diverse fullerene loadings
were studied. Pristine fullerene exhibits outstanding thermal
stability—96.7 wt % residue remained when heated to
600 ꢀC [Fig. 3(a)]. In comparison, compound 1 had 69.6 wt
% char yield at this temperature, which was higher than the
theoretical estimation if assuming that the “nonfullerene”
moiety had been completely decomposed and removed.
Hence, in practice, the excess char yield can be ascribed to
the residue of the “nonfullerene” moiety of compound 1.
TGA in nitrogen showed that theꢀprepolymers were almost
completely decomposed after 450 C. Similar inflection points
ꢀ
in the range from 435 to 485 C were also observed on the
TGA curves of the SFP samples [Fig. 3(b)]. The weight changes
became smoother and almost parallel with each other after
this range, suggesting that they (including compound 1)
decomposed at similar rates. Thus, it is reasonable to propose
that the inflection points indicated the disappearance of the
polymer backbones of the SFPs, and the remains after these
temperatures corresponded to the residues of compound 1.
On the basis of this hypothesis, fullerene loadings of the SFPs
and conversions of the click reactions were calculated using
the TGA curve of compound 1 as reference.18(a)
FIGURE 2 (a) Kinetics plot and (b) dependence of Mn and poly-
dispersity on the conversion for the RAFT polymerization of
AHMA and MMA (1:20) ([monomer]: [CPDB]: [V-70] 5 300:1:0.1,
40 ꢀC). The solid line represents the theoretical Mn.
The weight losses of the SFPs between 150 and 550 ꢀC
were analyzed for the calculation, because the influence of
that AHMA and MMA have similar relative reactivity ratios in
this composition range and can be randomly copolymerized.
The compositions of the prepolymers were determined by
the integrated areas of the corresponding peak of each
repeat unit in the 1H NMR spectra. Using prepolymer 2 as
an example (Supporting Information Fig. S2), the methylene
protons (next to the oxygen) of the AHMA repeat unit at
3.93 ppm and the methyl protons of MMA repeat unit at
TABLE 1 RAFT Copolymerization of AHMA and MMA in THFa
Feed Ratio
Mn
([AHMA]:
[MMA])
Prepolymer (g mol21
)
Mw/Mn
Compositionb
1
2
3
4
5
6
7
8
15,200
32,000
18,700
15,700
11,200
33,000
53,100
20,900
1.23
1.23
1.21
1.15
1.08
1.12
1.13
1.13
1:0
1:0
3.59 ppm were chosen to calculate the composition ratio,
1:1
1:1
3:93=2
3:59=3
which can be expressed as: r5 II
.
1:5
1:5
1:10
1:20
1:20
1:20
1:40
1:11
1:18
1:22
1:18
1:41
Click Reaction
The copper-mediated click reactions between the prepoly-
mers and compound 1 were performed at room temperature
with equivalent amounts of azide and alkyne (Scheme 2).
FTIR was used to monitor the reaction, as a strong and spe-
cific absorption at ꢁ2100 cm21 ascribed to the azido group
disappeared after the 1,3-cycloaddition (Supporting Informa-
tion Fig. S3), indicating a high conversion of the click reac-
a
In all the polymerizations, [monomer]: [RAFT]: [V-70] 5 300:1:0.1,
[monomer] 5 50 vol %, and all the polymerizations were conducted at
40 ꢀC.
1
b
tion. Also, H NMR spectra provided further evidence for the
This represents the experimentally measured average ratio of AHMA
reaction involving PAHMA (prepolymer 1). Since the azide
to MMA repeat units in each polymer chain, determined by 1H NMR.
WWW.MATERIALSVIEWS.COM
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000
5

JOURNAL OF
POLYMER SCIENCE
ARTICLE
WWW.POLYMERCHEMISTRY.ORG
SCHEME 2 Click reaction for side chain functionalization of the prepolymers.
solvents could be excluded by starting at 150 ꢀC, and 550
ꢀC was right in the “parallel” interval. For instance, poly-
mer 10 had 59.5 wt % char yield at 550 ꢀC compared
with that at 150 ꢀC, and compound 1 had 71.5 wt %.
Therefore, the content of compound 1 moiety in polymer
10 was calculated as 59.5 wt %/71.5 wt % 5 83.2 wt %.
Because the theoretical content with 100% conversion of
the click reaction is 85.1 wt %, the actual conversion
could be obtained as 81.0 wt %/85.1 wt % 5 97.8%. Fol-
lowing this method, Table 2 summarizes the conversions
of all the SFP samples and the average numbers of C60
per polymer chain.
Alternately, UV–vis spectrometry provided a more convenient
method to determine the fullerene contents because of its
fast measurement and nondestructive nature. Additionally,
compound 1 has only one substituent, which leads to a
unique molar extinction coefficient, so the measurement
should be more accurate than that involving fullerene struc-
tures with multiple substituents and/or multiple substitution
patterns.17(a),40
Figure 4 shows the UV–vis spectra of pristine fullerene, com-
pound 1 and polymer 10 in toluene. Strong absorption at
approximately 284 nm was observed in all the three sam-
ples. However, the peak at 330 nm of compound 1 was blue-
shifted and weaker in comparison to the absorption at 335
nm of the pristine fullerene, which is probably due to the
interaction between the C60 moiety and the adjacent aro-
matic ring in compound 1. The UV–vis spectra of compound
1 and polymer 10 are very similar, indicating the similar
chemical environment of the fullerene moieties. Therefore,
the concentrations of the compound 1 moiety of the SFP
samples in toluene were determined using compound 1 as
an external standard. Subsequently, the contents of com-
pound 1 moiety in the SFPs were calculated with the known
concentrations of the SFP samples. A standard dependence
of the absorbance at 284 nm on the concentration of com-
pound 1 in toluene is displayed in Supporting Information
Figure S5. Fullerene loadings of the SFPs and conversions of
FIGURE 3 TGA scans of (a) pristine C60, compound 1, prepoly-
mer 4 and 7; (b) compound 1 and side chain fullerene poly-
mers with different loadings from 30 to 600 ꢀC in nitrogen.
6
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000

JOURNAL OF
POLYMER SCIENCE
WWW.POLYMERCHEMISTRY.ORG
ARTICLE
TABLE 2 Click Conversion Efficiency and Fullerene Loadings Estimated by TGA and UV–vis Analyses
Theoretical
Loading (%)
# of C60 per
Actual Loading
Conversion
# of C60 per Chain
(TGA/UV–vis)
Samplesa
Chain (Theoretical)
(TGA/UV–vis) (%)
(TGA/UV–vis) (%)
Polymer 10
Polymer 20
Polymer 30
Polymer 40
Polymer 50
Polymer 60
Polymer 70
Polymer 80
85.1
79.5
62.9
47.9
37.5
33.4
37.5
21.9
71
103
26
12
6
83.2/62.5
82.2/60.0
62.7/55.6
38.7/38.2
31.9/32.3
25.3/23.7
31.3/18.7
18.3/16.7
97.8/73.4
103.4/75.5
99.7/88.4
80.8/79.7
85.1/86.1
75.7/71.0
83.5/49.9
83.6/76.3
69/52
107/78
26/23
10/10
5/5
14
26
5
11/10
22/13
4/4
a
The SFP samples correspond to the prepolymers listed in Table 1.
the click reactions estimated by UV–vis spectrometry are
also summarized in Table 2.
above, both steric hindrance and chain entanglement could
reduce the efficiency of the fullerene attachment. The rela-
tively low conversion of polymer 70 can be rationalized since
the effect of chain entanglement is expected to be greater.
The results from TGA and UV–vis were consistent for poly-
mers 30–60 and polymer 80. However, TGA indicated higher
fullerene loadings and conversions with high azide content
prepolymers (polymer 10 and 20) or when the molecular
weight was high (polymer 70). A reasonable explanation is
that steric hindrance of the attached fullerene moieties and
entanglement of the polymer chains reduced the efficiency of
the click reactions resulting in unreacted azido groups in the
resultant SFPs, which became severe with an increase of the
statistical incorporation of AHMA repeat units and/or
the polymer chain length. During the TGA tests at high tem-
peratures, reactions between these azido groups and the
fullerene moieties generated crosslinked structures,15
which could retard the degradation and result in higher test
values.
In addition, DSC studies showed that the addition of fuller-
ene moieties to the prepolymers increased the Tg of the SFPs
except at the lowest loading levels (polymer 70 and 80). At
the highest fullerene loadings (polymer 10-30), Tg’s were not
detected up to 150 ꢀC. Apparently, the fullerene-fullerene
attractions can limit the mobility of the polymer chains and
therefore, raise the Tg’s. However, for the low-loading sam-
ples, this factor may be counteracted by the side chain effect,
since the side chain fullerene moieties also generate free vol-
ume and improve the mobility of the polymer chains.
Morphology Studies
C60 moieties have very strong p–p stacking interactions with
each other (ꢁ4.2 kcal mol21 in direct contact),41 which are
similar in strength to hydrogen bonding. Thus, most fuller-
ene derivatives are described as solvophobic and often form
various aggregates in solution. Accordingly, it was also
expected that our SFPs would not exist in solution as indi-
vidual chains but assemble into nanocomplexes, with the sol-
vent-compatible polymer backbones at the exterior surface
to reduce direct fullerene–solvent interactions.
Except for polymer 70, the SFPs were prepared at high con-
versions (70–90%) as determined from UV–vis. As mentioned
GPC analysis was initially used to study the self-aggregation
behavior.18(b),42 A monomodal peak at 22 min, ascribed to
individual polymer chains, was detected in the GPC traces of
polymers 30, 40, and 80 (Fig. 5), demonstrating that crosslink-
ing did not occur in the fullerene grafting process. The PDI
of this peak remained narrow. The chromatogram of polymer
80 exhibited the individual chain peak exclusively, suggesting
little tendency of self-aggregation in solution, which is rea-
sonable due to its low fullerene content. In contrast, in the
GPC traces of polymers 10, 30, and 40, another group of broad
peaks were displayed at very early retention times, which
provided strong evidence for the formation of aggregates,
although the GPC does not allow measuring molecular
weights accurately at this interval. With an increase in the
fullerene content, the signals ascribed to the aggregates
FIGURE 4 UV–vis spectra of pristine fullerene (0.0151 mg
mL21), compound
1
(0.0181 mg mL21), and polymer 10
(0.0376 mg mL21) in toluene.
WWW.MATERIALSVIEWS.COM
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000
7

JOURNAL OF
POLYMER SCIENCE
ARTICLE
WWW.POLYMERCHEMISTRY.ORG
became stronger and the peaks of individual chains became
weaker. Finally, the aggregation signal was the only peak
observed for polymer 10 while the individual chain peak
disappeared.
Moreover, the prepolymers of polymers 10 and 40 had similar
molecular weights, but the DLS analyses of the two corre-
sponding fullerene-attached samples showed different size
distributions (Fig. 6). Polymer 10 displayed a bimodal size
distribution with a Z-average size of 220 nm, and both peaks
were larger than the diameter of the individual chains, which
further verified the aggregation behavior of the SFPs in solu-
tion. However, the DLS for polymer 40 showed a major peak
with a Z-average size of 16 nm, which corresponded to the
size of the individual chains.
FIGURE 5 GPC traces of side chain fullerene polymers 10, 30, 40,
and 80 recorded by a refractive index detector in THF.
In addition, the GPC data were also compared between the
SFPs and their prepolymers (Supporting Information
Fig. S6). Surprisingly, it appeared that the molecular weights
FIGURE 6 Statistical size distributions of (a) polymer 10 and (b) polymer 40 in toluene tested by DLS. Curves represent the average
of three separate measurements.
8
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000

JOURNAL OF
POLYMER SCIENCE
WWW.POLYMERCHEMISTRY.ORG
ARTICLE
FIGURE 7 SEM images of polymer 10–80.
of the SFPs did not increase significantly after the fullerene
grafting. The average molecular weight of polymer 30 was
even lower than that of its precursor. It is worth noting that
the GPC traces recorded by the RI detector indicate the
hydrodynamic volume of the polymer chains, and may not
reflect the actual changes in molecular weights of the SFPs.
On the basis of the backbone modification, the SFPs can be
considered as comb- or brush- polymers, and it has been
reported that the GPC-measured values underestimated the
true molecular weights of such branched polymers by up to
a factor of 10.43 More importantly, the GPC with RI detector
responds to size differences of polymers, and it was possible
the coils of the SFPs were much denser ascribed to the
intra-fullerene attractions.
WWW.MATERIALSVIEWS.COM
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000
9

JOURNAL OF
POLYMER SCIENCE
ARTICLE
WWW.POLYMERCHEMISTRY.ORG
The morphology of the SFPs on silicon wafers was analyzed
by SEM (Fig. 7). All the samples were dissolved in toluene
with identical concentrations (1.0 mg mL21) except polymers
10 and 20 (0.2 mg mL21) due to their poorer solubility. It was
observed that all the SFPs in the SEM images aggregated into
nanoparticles as the elementary units for further supramolecu-
lar assemblies. The sizes of the nanoparticles in all the batches
were generally uniform from 10 to 30 nm in diameter, indicat-
ing that they were independent of the fullerene contents or
the chain lengths of the polymers, and apparently only de-
pendent on the nature of the fullerene moiety itself.
SFPs on silica wafers were relatively small compared with
other nanoparticles assembled by fullerene derivatives that
have been previously reported.22(a),44,45 Interestingly, the
size of the nanoparticles was not determined by the fuller-
ene loadings or chain lengths of the SFPs. Because of the
interplay of several important molecular variables, a rich va-
riety of nanostructures and morphologies were formed.
CONCLUSIONS
In this work, the precise synthesis of a soluble and monoal-
kynyl functionalized fullerene derivative starting with pristine
C60 has been reported. By combining RAFT polymerization
and the copper-mediated click reaction, we have successfully
prepared a series of well-defined SFPs with variable molecular
weights and fullerene contents. The RAFT polymerization pro-
ceeded with good control over the molecular weight and co-
polymer composition, and the click reaction to attach fullerene
in a postfunctionalization reached high conversions despite
the steric hindrance. Because of the high efficiency of this
method, we are able to achieve a very high fullerene loading
for the SFPs (up to 78 C60 moieties per chain on average from
UV–vis determination), which has not been reported previ-
ously. Studies on the self-aggregation behaviors of these SFPs
were performed both in solution and on silicon wafers. Pri-
mary nanoparticles were formed by the strong fullerene–full-
erene interactions and further assembled into more complex
morphologies, and the development of the supramolecular
architectures depended on the fullerene contents and chain
lengths of the samples. This type of self-aggregation with well-
defined polymers may find applications in the design of new
functional materials in the future.
Further assembly into sheets of nanoparticles was detected
for polymers 10–30 while the nanoparticles of polymers 40–70
tended to form string-like assemblies. The nanoparticles of
polymer 80 appeared as individual particles or clusters of
several nanoparticles rather than micron-size complexes.
These complex architectures were likely formed by noncova-
lent attractions of the fullerene moieties, since both individ-
ual nanoparticles and the complex assemblies were observed
in many sample preparations thus implying that reversible
interactions are the probable driving force for the assembly.
A hypothesis was proposed to explain the relationship
between the observed morphology and the variable fullerene
loadings of the SFPs. With increasing fullerene content, there
was an increase in the amount of C60 exposed on the surface
of the resultant nanoparticles, which could act as active sites
for the fullerene–fullerene attraction between different nano-
particles. Hence, they were more likely to build-up more
complex structures, for example, nanoparticle strings or
sheets on the wafer. In contrast lower fullerene loadings
resulted in fewer or no active sites on the nanoparticle
surfaces since most of the fullerene moieties were encapsu-
lated inside of the nanoparticles and covered by the polymer
backbones, and individual nanoparticles or small nanopar-
ticle clusters were preferred.
ACKNOWLEDGMENTS
The authors acknowledge financial support from the South Car-
olina SmartState Centers of Economic Excellence program, and
thank Y. Xie and S. K. Kumar (Columbia) for the DLS tests and
useful discussions.
Although nanoparticle strings were formed in polymers 40–
70, they were not identical in appearance. Polymers 40 and 60
formed very similar branched nanoparticle networks in a
range of a few microns, which may correlate to their similar
number of C60’s (10) per chain. In comparison, nanoparticles
of polymer 70 formed less-branched strings possibly result-
ing from the lower fullerene content. Polymer 50 had only 5
REFERENCES AND NOTES
1 F. Giacalone, N. Martin, Chem. Rev. 2006, 106, 5136–5190.
2 C. Wang, Z.-X. Guo, S. Fu, W. Wu, D. Zhu, Prog. Polym. Sci.
2004, 29, 1079–1141.
3 R. Hiorns, E. Cloutet, E. Ibarboure, L. Vignau, N. Lematre, S.
Guillerez, C. Absalon, H. Cramail, Macromolecules 2009, 42,
3549–3558.
C
60’s per chain and formed “necklace-like” nanoparticle
structures that were less than 1 lm in size. This is probably
due to both its shorter chain length and fewer number of
C60 per chain.
4 G. Zhou, J. He, I. I. Harruna, K. E. Geckeler, J. Mater. Chem.
2008, 18, 5492–5501.
Polymers 10 and 20 not only showed nanoparticle assemblies
but also assembled into ordered or crystalline-like clusters.
These polymers are likely to have extended chain conforma-
tions due to the highly crowded pendant fullerene moieties
which may further facilitate associations that underlie the
formation of ordered regions.
5 Z. L. Yao, K. C. Tam, Langmuir 2011, 27, 6668–6673.
6 I. Natori, S. Natori, H. Sekikawa, T. Takahashi, H. Sato, J.
Appl. Polym. Sci. 2011, 121, 3433–3438.
7 Q. Tai, J. Li, Z. Liu, Z. Sun, X. Zhao, F. Yan, J. Mater. Chem.
2011, 21, 6848–6853.
8 I. Natori, S. Natori, J. Polym. Sci. Part A: Polym. Chem. 2008,
46, 3282–3293.
In summary, a general tendency of well-defined SFPs to
assemble was observed, and the nanoparticles formed by the
9 B. Dardel, D. Guillon, B. Heinrich, R. Deschenaux, J. Mater.
Chem. 2001, 11, 2814–2831.
10
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000

JOURNAL OF
POLYMER SCIENCE
WWW.POLYMERCHEMISTRY.ORG
ARTICLE
10 J. Kim, M. H. Yun, J. Lee, J. Y. Kim, F. Wudl, C. Yang,
Chem. Commun. 2011, 47, 3078–3080.
28 D. Bonifazi, O. Enger, F. Diederich, Chem. Soc. Rev. 2006,
36, 390–414.
11 A. Munoz, B. M. Illescas, M. Sanchez-Navarro, J. Rojo, N.
Martin, J. Am. Chem. Soc. 2011, 133, 16758–16761.
29 L. Sanchez, R. Otero, J. M. Gallego, R. Miranda, N. Martin,
Chem. Rev. 2009, 109, 2081–2091.
12 S. Shi, K. C. Khemanikc, Q. C. Li, F. Wudl, J. Am. Chem.
Soc. 1992, 114, 10656–10657.
30 S. S. Babu, H. Mohwald, T. Nakanishi, Chem. Soc. Rev.
2010, 39, 4021–4035.
13 C. Yang, J. K. Lee, A. J. Heeger,F. Wudl, J. Mater. Chem.
2009, 19, 5416–5423.
31 X. Yu, W.-B. Zhang, K. Yue, X. Li, H. Liu, Y. Xin, C.-L. Wang,
C. Wesdemiotis, S. Z. D. Cheng, J. Am. Chem. Soc. 2012, 134,
7780–7788.
14 J. U. Lee, A. Cirpan, T. Emrick, T. P. Russell, W. H. Jo, J.
Mater. Chem. 2009, 19, 1483–1489.
32 B. Schade, K. Ludwig, C. Bottcher, U. Hartnagel, A. Hirsch,
Angew. Chem. Int. Ed. 2007, 46, 4393–4396.
15 (a) C. J. Hawker, K. L. Wooley, J. M. J. Frechet, J. Chem.
Soc. Chem. Commun. 1994, 925–926; (b) C. J. Hawker, P. M.
Saville, J. W. White, J. Org. Chem. 1994, 59, 3503–3505.
33 D. Felder, H. Nierengarten, J.-P. Gisselbrecht, C. Boudon, E.
Leize, J.-F. Nicoud, M. Gross, A. van Dorsselaer, J.-F. Nieren-
garten, New J. Chem. 2000, 24, 687–695.
16 C. J. Hawker, Macromolecules 1994, 27, 4836–4837.
17 (a) U. Stalmach, B. de Boer, C. Videlot, P. F. van Hutten, G.
Hadziioannou, J. Am. Chem. Soc. 2000, 122, 5464–5472; (b) B.
de Boer, U. Stalmach, P. F. van Hutten, C. Melzer, V. V. Krasni-
kov, G. Hadziioannou, Polymer 2001, 42, 9097–9109.
34 Y. Li, B. C. Benicewicz, Macromolecules 2008, 41,
7986–7992.
35 C. Bingel, Chemische Berichte 1993, 126, 1957–1959.
36 R. S. Ruoff, D. S. Tse, D. C. Lorents, J. Phys. Chem. 1993,
97, 3379–3383.
18 (a) M. H. van der Veen, B. de Boer, U. Stalmach, K. I. van de
Wetering, G. Hadziioannou, Macromolecules 2004, 37,
3673–3684; (b) S. Barrau, T. Heiser, F. Richard, C. Brochon, C.
Ngov, K. van de Wetering, G. Hadziioannou, D. V. Anokhin, D.
A. Ivanov, Macromolecules 2008, 41, 2701–2710.
37 B. S. Sumerlin, N. V. Tsarevsky, G. Louche, R. Y. Lee, K.
Matyjaszewski, Macromolecules 2005, 38, 7540–7545.
38 Y. Li, J. Yang, B. C. Benicewicz, J. Polym. Sci. Part A: Polym.
Chem. 2007, 45, 4300–4308.
19 A. Celli, P. Marchese, M. Vannini, C. Berti, I. Fortunati, R.
Signorini, R. Bozio, React. Funct. Polym. 2011, 71, 641–647.
39 V. Ladmiral, T. M. Legge, Y. Zhao, S. Perrier, Macromole-
cules 2008, 41, 6728–7632.
20 M. Li, P. Xu, J. Yang, S. Yang, J. Mater. Chem. 2010, 20,
3953–3960.
40 F. Djojo, A. Herzog, I. Lamparth, F. Hampel, A. Hirsch,
Chem. Eur. J. 1996, 2, 1537–1547.
21 E. Rusen, B. Marculescu, N. Preda, L. Mihut, J. Polym. Res.
2008, 15, 447–451.
41 H. Kim, D. Bedrov, G. D. Smith, J. Chem. Theory Comput.
2008, 4, 335–340.
22 C. J. Hawker, K. L. Wooley, Science 2005, 309, 1200–1205.
23 P. Lundberg, C. J. Hawker, A. Hult, M. Malkoch, Macromol.
Rapid Commun. 2008, 29, 998–1015.
42 C. Chu, Y. Tsai, L. Hsiao, L. Wang, Macromolecules 2011,
44, 7056–7061.
24 B. S. Sumerlin, A. P. Vogt, Macromolecules 2010, 43, 1–13.
43 M. B. Runge, S. Dutta, N. B. Bowden, Macromolecules 2006,
39, 498–508.
25 (a) T. Kawauchi, J. Kumaki, E. Yashima, J. Am. Chem. Soc.
2006, 128, 10560–10567; (b) T. Kawauchi, J. Kumaki, A. Kitaura,
K. Okoshi, H. Kusanagi, K. Kobayashi, H. Shinahara, E.
Yashima, Angew. Chem. Int. Ed. 2008, 47, 515–519.
44 S.-I. Yusa, S. Awa, M. Ito, T. Kawase, T. Takada, K. Maka-
shima, D. Liu, S. Yamago, Y. Morishima, J. Polym. Sci. Part A:
Polym. Chem. 2011, 49, 2761–2770.
26 M. Wang, K. P. Pramoda, S. H. Goh, Macromolecules 2006,
39, 4932–4934.
45 J. Wang, Y. Shen, S. Kessel, P. Fernandes, K. Yoshida, S.
Yagai, D. G. Kurth, H. Mohwald, T. Nakanishi, Angew. Chem.
Int. Ed. 2009, 48, 2166–2170.
27 E. Badamshina, M. Gafurova, J. Mater. Chem. 2012, 22,
9427–9438.
WWW.MATERIALSVIEWS.COM
JOURNAL OF POLYMER SCIENCE, PART A: POLYMER CHEMISTRY 2013, 00, 000–000
11