Polymer 47 (2006) 19351945

www.elsevier.com/locate/polymer

Amphiphilic gradient poly(styrene-co-acrylic acid) copolymer prepared via

nitroxide-mediated solution polymerization. Synthesis, characterization in
aqueous solution and evaluation as emulsion polymerization stabilizer
Catherine Lefay a, Bernadette Charleux a,*, Maud Save a, Christophe Chassenieux b,
Olivier Guerret c, Stephanie Magnet c
a

Laboratoire de Chimie des Polyme`res, Universite Pierre et Marie CurieParis 6, UMR 7610 Associee au CNRS, Tour 44,
1er etage4, Place Jussieu, 75252 Paris Cedex 05, France
b
Physicochimie des Polyme`res et des Milieux Disperses (UMR 7615), UPMC-CNRS-ESPCI, 10 rue Vauquelin, 75231 Paris Cedex 05, France
c
ARKEMA, Groupement de Recherches de Lacq, B.P. no 34, 64170 Lacq, France
Received 21 October 2005; received in revised form 23 December 2005; accepted 13 January 2006
Available online 7 February 2006

Abstract
A well-defined, amphiphilic poly(styrene-co-acrylic acid) copolymer was synthesized in a single step by nitroxide-mediated controlled freeradical copolymerization of styrene and acrylic acid, without protection of the acid groups: MnZ6500 g molK1, Mw/MnZ1.5 and a composition
of FAAZ0.70G0.03 in acrylic acid. In addition to the good control over molar mass and molar mass distribution, the copolymer exhibited a
narrow composition distribution with a slight gradient. Such copolymer was an efficient stabilizer for the emulsion polymerizations of styrene and
of mixtures of methyl methacrylate and n-butyl acrylate, until 45 wt% solids. A low amount (typically 34 wt% based on the monomer(s)) was
needed for a good stabilization. This is approximately a decade lower than the required amount of random, amphiphilic copolymers prepared via
conventional free-radical polymerization. The performances were, however, below those of analogous diblock copolymers, but the great
advantage is the very easy synthetic procedure.
q 2006 Elsevier Ltd. All rights reserved.
Keywords: Styrene; Acrylic acid; Amphiphilic copolymer

1. Introduction
Amphiphilic, random copolymers are widely industrially
produced and used in different applications such as coatings,
adhesives, inks etc. Those copolymers are composed of
hydrophobic and hydrophilic monomer subunits, randomly
distributed along the chains. They are generally synthesized by
classical free radical copolymerization in different media and
are usually called ASR for alkali-soluble resins when the
hydrophilic comonomer possess a carboxylic acid, ionizable at
alkaline pH. In particular, the free radical copolymerization of
styrene (St) and acrylic acid (AA) is well described in the
literature. It was, for instance, performed either in bulk [1,2], in

* Corresponding author. Tel.: C33 1 44275502; fax: C33 1 44277089.

E-mail address: charleux@ccr.jussieu.fr (B. Charleux).

0032-3861/$ - see front matter q 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.polymer.2006.01.034

solution in benzene [3] or in emulsion [36]. Moreover, some

of the ASR containing St and AA are industrially well-known
like for instance the Joncryle from Johnson Polymer, the
Neocryle from Neoresins or the Moreze 101 from Rohm and
Haas. Different patents [79] describe in more details the
synthesis and use of such copolymers as stabilizers for
emulsion polymerizations. They were sometimes used in
combination with surfactants of low molar mass [1012].
One of the major drawbacks of the ASR in emulsion
polymerization is the requirement of rather high quantity to
stabilize the latexes [1318]. Indeed, and specially when used
as the sole surfactant, their concentration ranged from 25 to
45 wt% on the monomer basis, for a latex solids content up to
ca. 50% [1317]. Such low efficiency might be assigned to the
poor structural quality of the amphiphilic copolymers, owing to
the use of classical free-radical copolymerization for the
synthesis. This method, indeed, produces chains with both a
broad molar mass distribution and a broad composition
distribution. Consequently, even when the average

1936

C. Lefay et al. / Polymer 47 (2006) 19351945

composition and molar mass are appropriate for latex

stabilization, a large part of the macromolecules does not
efficiently stabilize the emulsion polymerization because of an
improper molar mass or an improper hydrophilic/lipophilic
balance.
In contrast to the conventional method, controlled freeradical copolymerization is able to yield copolymers that are
homogeneous in both molar mass and composition [1921].
Moreover, as the chains are created simultaneously in the early
polymerization stage, they reveal a composition gradient
(continuous variation of the comonomer composition along
the chain) when the comonomers exhibit a noticeable
difference in reactivity [2226]. Such feature might endow
the chains with specific properties, between those of a
statistical copolymer and those of a diblock copolymer.
The purpose of this work was thus to improve the structural
quality and hence the stabilization properties of amphiphilic
random copolymers by using controlled free-radical copolymerization as a synthetic tool. The selected hydrophobic and
hydrophilic comonomers were styrene and acrylic acid,
respectively. Nitroxide-mediated controlled free-radical
polymerization (NMP) was chosen to perform the synthesis,
with SG1 as a mediator (Scheme 1), since we recently
demonstrated that this nitroxide is well suited to control the
copolymerization of styrene and acrylic acid, over a wide
composition range, without protection of the acidic groups
[27].
The target characteristics of the selected copolymer were a
moderate number average molar mass (Mn ranging between
5000 and 10,000 g molK1) and quite a high hydrophilicity
(above 50 mol% of AA) to allow its solubilization in alkaline
aqueous solutions. After the full characterization of the
copolymer structure and the study by light scattering of its
organization in aqueous solution, its efficiency in stabilizing
latex particles was tested in emulsion polymerizations. It
should be emphasized that the copolymer chain length and
composition range were selected in order to closely match the
structure of analogous polystyrene-b-poly(acrylic acid) diblock
copolymers, which exhibited excellent stabilization properties
in the emulsion polymerization of styrene [28,29], but which
were comparatively more difficult to produce. Indeed, they
were prepared in two polymerization steps followed by
acidolysis of the ester group of the tert-butyl acrylate, used
as a hydrophobic precursor of acrylic acid.

O
HO

O
O

Alkoxyamine initiator (called MAMA or BlocBuilder)

N
P

Nitroxide SG1

Scheme 1. Chemical structure of the alkoxyamine initiator MAMA and of the

nitroxide mediator SG1.

2. Experimental part
2.1. Materials
The monomers, styrene (St, Aldrich, 99%), methyl
methacrylate (MMA, Acros, 99%) and n-butyl acrylate (BA,
Aldrich, 99%) were distilled under vacuum before use. Acrylic
acid (AA, purest grade, Arkema, stabilized with 200 ppm of
hydroquinone) was stored at room temperature and used
without further purification. The alkoxyamine initiator
(Scheme 1), called MAMA in the following (commercial
name is BlocBuildere, 99% purity) and the nitroxide SG1 (Ntert-butyl-N-(1-diethylphosphono-2,2-dimethylpropyl)-Noxyl, 85% purity) were kindly provided by Arkema. The
solvents, 1,4-dioxane (synthesis grade) and dichloromethane
(Normapur) were, respectively, supplied by SDS and by
Prolabo. Trimethylsilyldiazomethane (2 M solution in hexane)
and sodium hydrogen carbonate (NaHCO3, 99.7%) were
purchased from Aldrich and potassium persulfate
(purityO99%) from Acros.
2.2. Nitroxide-mediated copolymerization of styrene
and acrylic acid
The copolymerization reactions were carried out in 1,4dioxane solution at 120 8C, according to an already published
protocol [27]. The only major difference was the type of
alkoxyamine used as an initiator, since here the MAMA
(Scheme 1) was chosen instead of the MONAMS [27]. The
reactions were performed in a Parr reactor of 300 mL at a
stirring rate of 300 rpm. In a typical synthesis, the alkoxyamine
initiator MAMA (1.637 g, 4.3!10K3 mol) was dissolved in a
mixture of styrene (11.50 g, 0.111 mol) and acrylic acid
(32.00 g, 0.444 mol, initial molar composition of AA in the
comonomer mixture: fAA0Z0.80). Then, a volume of 145 mL
of 1,4-dioxane and 0.0638 g of free SG1 (2.2!10K4 mol,
5 mol% based on the initiator) were added. The polymerization
solution was transferred into the reactor and deoxygenation
was performed at room temperature by nitrogen bubbling for
5 min. Afterward, the reactor was heated at 120 8C and a 2 bar
pressure of nitrogen was applied. Time zero for the reaction
was arbitrarily taken when the mixture reached 90 8C.
Aliquots were periodically withdrawn during the reaction
and cooled in an iced water bath to quench the polymerization.
For all of them, the overall monomer conversion was
determined by 1H NMR analysis of the raw polymerization
medium in acetone d6 solution. The overall molar conversions
(xmol) and the individual conversions of each monomer (xAA
and xSt) were determined by integrating the peaks corresponding to the vinylic protons, using the broad peak between 6.5 and
7.5 ppm as an internal reference (five aromatic H for styrene
and polystyrene, and one vinylic H for the styrene monomer
that was subtracted before calculation). The overall weight
conversion was calculated from those experimental data [27].
The molar mass of the dried copolymers was determined by
size exclusion chromatography after methylation of the AA
subunits. The overall molar conversion was used for kinetic

C. Lefay et al. / Polymer 47 (2006) 19351945

analysis (conversion versus time plots), whereas molar masses

and polydispersity indexes were plotted as a function of the
overall weight conversion.
After 8 h, the reaction yielded a molar conversion of 80%
and the final polymerization batch was precipitated in
dichloromethane and filtered. The polymer was first dried for
2 days at 60 8C in an oven, and then under vacuum at room
temperature. For the purpose of emulsion polymerization, the
nitroxide end-group was not removed from the chain-end;
actually, the reactions were performed at 70 8C, a temperature
that is far too low for the alkoxyamine dissociation (typically
above 100 8C).
2.3. Titration of the carboxylic acid groups
The copolymer composition was determined by direct
titration of the carboxylic groups with a 0.1 M sodium
hydroxide aqueous solution. A precise amount of copolymer
(0.115 g) was dissolved in 30 mL of THF and 2 mL of
deionized water. Then, the NaOH solution was added dropwise
and the pH was followed by a combined pH electrode (XC 100
from Radiometer Analytical). From the mole number of acidic
functions in the solution, determined at the equivalent point,
the average molar composition of the copolymer in acrylic acid
(FAA) was calculated.
2.4. Batch emulsion polymerization of styrene at 10 wt% solids
The free-radical polymerizations of styrene in aqueous
emulsion were carried out at 70 8C in a double jacket 250 mL
glass reactor, stirred at 300 rpm. The poly(styrene-co-acrylic
acid) amphiphilic copolymer (weight ranging from 0.10 to
0.80 g; 18 wt% based on the styrene content), was added to a
1 M NaOH solution (from 0.81 to 6.48 g, i.e. from 8.1!10K4
to 6.5!10K3 mol; 1 equiv. NaOH based on the acrylic acid
groups) in order to fully neutralize the acidic functions. The
solution was then diluted with, respectively, 8882 mL of
deionized water, which already contained NaHCO3 in such an
amount that the overall concentration was 0.012 mol LK1. The
so-formed alkaline aqueous copolymer solution was then
heated at 70 8C in a water bath for 20 min under magnetic
stirring to ensure complete copolymer solubilization. After
cooling the solution, the pH (which was around 8.5) was
adjusted to 10 by the addition of a very small volume of NaOH
1 M. Afterward, the solution was poured into a 250 mL glass
reactor thermostated at 70 8C. The stirring rate was 300 rpm.
After a 45 mm deoxygenation time by nitrogen bubbling,
styrene (10 g, 10 wt% target solids content, previously
deoxygenated for 20 min) was added in one shot to the
aqueous solution and the mixture was stirred for an additional
period of 15 min. The initiator, K2S2O8, dissolved in a small
quantity of deionized water (0.15 g, 0.006 mol LK1 based on
the total aqueous volume) was finally poured into the reactor,
which triggered the beginning of the reaction. Samples were
periodically withdrawn for the 3 h of reaction time, cooled in
an iced water bath to stop the polymerization and then analyzed
by dynamic light scattering. The conversion was determined by

1937

gravimetry after addition of a small amount of hydroquinone to

prevent further polymerization upon heating.
2.5. Semi-continuous emulsion polymerizations
at 45 wt% solids
The emulsion polymerizations of styrene (45.5 g, 45 wt%
target solids content) and of a mixture of methyl methacrylate
and n-butyl acrylate (35/65 (wt/wt), 45 wt% target solids
content) were performed following a semi-continuous addition
of the monomers in the reactor already containing the aqueous
phase. The latter was composed of 44 g of deionized water,
1.37 g of the poly(styrene-co-acrylic acid) stabilizer (3 wt%
based on the monomers), 0.012 mol LK1 of NaHCO3 and
11.5 g of a 1 M NaOH solution (1.15!10K2 mol, 1 equiv.
NaOH based on the acrylic acid groups of the stabilizer). The
preparation of the amphiphilic copolymer solution was the
same as before. The monomers (previously deoxygenated for
45 min) were continuously fed during 4 h into the double jacket
250 mL glass reactor heated at 70 8C and stirred at 300 rpm.
The introduction in one shot of the K2S2O8 solution (0.09 g
dissolved in a small quantity of deionized water, 0.006 mol LK1
based on the total aqueous volume) was synchronized with
the beginning of the monomer addition and triggered the start
of the reaction. At the end of the monomer introduction, a new
portion of K2S2O8 (0.045 g; 1.66!10K4 mol; 0.003 mol LK1)
was added and the polymerization was performed for an
additional hour. The final latexes were cooled in an iced bath
and then analyzed by dynamic light scattering and by
gravimetry.
2.6. Analytical techniques
2.6.1. Size exclusion chromatography (SEC)
The average molar mass and molar mass distribution were
obtained by SEC in THF with 1 mL minK1 flow rate and a
polymer concentration of 5 g LK1, after full methylation of the
acrylic acid subunits of the copolymers. The experimental
procedure used for methylation has already been published [27].
The SEC apparatus is composed of two columns thermostated at
40 8C (PSS SDV, linear M, 8 mm!300 mm; bead diameter:
5 mm; separation limits: 4002!106 g molK1) and of two
detectors: RI (LDC Analytical, refractorMonitor IV) and UV
operating at 254 nm (Waters 484). The molar mass characteristics of the poly(styrene-co-methyl acrylate) copolymers (Mn,
the number average molar mass; Mw, the weight average molar
mass; and PDIZMw/Mn, the polydispersity index) were
calculated from the differential RI trace, using a calibration
curve based on polystyrene standards [30]. The molar masses of
the original poly(styrene-co-acrylic acid) copolymers were
systematically derived and are the ones given in the article.
2.6.2. NMR (250 MHz)
Analyses were performed in 5 mm tubes at room
temperature using an AC250 Brucker spectrometer. The
chemical shift scale was calibrated on the basis of the solvent
signal, i.e. deuterated acetone, at 2.05 ppm.

1938

C. Lefay et al. / Polymer 47 (2006) 19351945

2.6.3. Liquid adsorption chromatography (LAC)

The methylated copolymers (5 g LK1) were analyzed by
LAC in order to investigate the composition distribution since
the technique allows the chains to be separated according to
their composition, regardless of the molar mass [31].
Separation was carried out at 30 8C on a grafted silica
column with a gradient hexane/THF as an eluent at a flow
rate of 3 mL minK1. The detection was performed using a
UV detector (Waters 481) and an evaporative light scattering
detector (DEDL 21, Eurosep).
2.6.4. Light scattering
Alkaline aqueous solutions of the poly(styrene-co-acrylic
acid) amphiphilic copolymer at different concentrations were
analyzed by light scattering. The master solution at a
concentration of 25 g LK1 (3.8!10K3 mol LK1) was prepared
with the following components: deionized water (44 g),
copolymer (1.376 g, 2.12!10K4 mol, MnZ6500 g molK1,
FAAZ0.70), 1 M NaOH solution (11.2 g) and NaHCO3
(0.055 g, 0.012 mol LK1). The solution was prepared as
described before in the emulsion polymerization part except
that after adjustment of the pH to 10, the solution was heated
again in a water bath at 70 8C for 45 min (same duration as the
deoxygenation period in the emulsion polymerization). The
cooled solution was next analyzed either directly, or after
filtration, or after dilution by a solution composed of 250 mL of
deionized water, NaHCO3 (0.252 g, 0.012 mol LK1) and a
small quantity of 1 M NaOH solution to adjust the pH to 10.
Static (SLS) and dynamic (DLS) light scattering measurements were performed with a Malvern goniometer in
combination with a Spectra Physics laser operating at lZ
514.5 nm. The scattered photons were collected by an ALV
photomultiplier and analyzed by an ALV-5000 multibit,
multitau full digital correlator. The intensity autocorrelation
function, g2(t), and the mean scattered intensity, I, were
measured at several wave vector values qZ(4pn/l)sin(q/2),
with n being the refractive index of the solvent and q being
the observation angle ranging from 30 to 1408. The
measurements were done at 25 8C if not otherwise specified.
The electric field autocorrelation functions g1(t), related to
g2(t) via the Siegert relation [32], were analyzed using a
REPES routine, assuming a continuous distribution of the
relaxation times (Eq. (1))
g1 t Z

C
N

AtexpKt=tdt

(1)

KN

In the whole concentration range investigated, the

calculated distributions of relaxation times were bimodal.
The relaxation times that can be derived for the fast (tfast)
and the slow (tslow) relaxation processes were q2-dependent,
which allowed the respective translational diffusion coefficient to be calculated, according to: DiZ(tiq2)K1, where i
denotes fast or slow. The concentration dependence of Di is
given by: DiZD0i(1CkDiC) where kDi is the dynamic virial
coefficient and D0i is the mutual diffusion coefficient useful to

derive the value of the hydrodynamic radius according to

RHiZkT/(6phD0i), in which k is the Boltzman constant and h
the viscosity of the solvent.
2.6.5. Surface tension measurements
The surface tension of alkaline aqueous copolymer
solutions was measured with a drop tensiometer (ITC Concept)
using the pendant drop method at 25 8C. Solutions with varying
copolymer concentrations (from 1.56 to 25 g LK1) were
prepared as described in the previous section. The time
dependence of the interfacial tension between solution and
air was determined by the digital processing of the shape of a
drop of the first fluid formed within a quartz cell containing the
second fluid (i.e. air). The drop was illuminated by a CCD
camera driven by a microcomputer using the Windrop 1.1
software from ITC Concept. After one day of measurement, the
surface tension had decreased but never leveled off to a plateau
value. For a longer measurement time, the drop fell down and
the analysis was stopped.
2.6.6. Analytical techniques to measure the average latex
particle size and the particle size distribution
The z-average diameter (Dz) of the polystyrene latex
particles was measured with a Zetasizer 4 from Malvern,
using a 5 mW HeNe laser at 633 nm.
The polystyrene latex particles were visualized by transmission electron microscopy (TEM, JEOL JEM 100 cx II
CHR). The count of 500 particles allowed the number (Dn),
and weight (Dw) average diameters to be calculated. The
number concentration of latex particles per liter of latex was
calculated according to Eq. (2), where Dn is the number
average particle diameter expressed in centimeter, dp is the
polystyrene density (dpZ1.05 g cmK3) [33] and tp is the
polymer content (g LK1
latex)
1
Np LK
latex Z

6tp
dp pD3n

(2)

3. Results and discussion

3.1. Synthesis and characteristics of the poly(styrene-coacrylic acid) amphiphilic copolymer
As previously mentioned, the aim of this work was to
obtain, in a single polymerization step, a well-defined,
amphiphilic poly(styrene-co-acrylic acid) copolymer with a
moderate number-average molar mass, a narrow molar mass
distribution, and a selected hydrophilic/lipophilic balance in
order to get easy solubilization in alkaline water. In addition,
the copolymer had to exhibit a narrow composition distribution. The model structure was selected on the basis of
polystyrene-b-poly(acrylic acid) (PSt-b-PAA) diblock copolymers, which proved in the past to be excellent stabilizers in
emulsion polymerization [28,29]. Actually, one of the best
ones had 16 styrene units and 55 acrylic acid units (molar

C. Lefay et al. / Polymer 47 (2006) 19351945

fraction of AA in the copolymer, FAAZ0.77). For the

copolymerization reaction, we started from an initial molar
fraction of AA in the comonomer feed of fAA0Z0.80. Knowing
from the previously determined reactivity ratios (rStZ0.72G
0.04 and rAAZ0.27G0.07) [27] that, at such initial composition, styrene would be consumed faster than acrylic acid, we
expected to have an average copolymer composition, FAA,
below 0.80 at incomplete conversion.
To control the chain length and chain length distribution, we
used SG1-mediated copolymerization of styrene and acrylic
acid, with experimental conditions already established in a
previous work [27]. The initiator concentration was selected so
as to reach a final Mn of 10,100 g molK1 at full monomer
conversion. Fig. 1 displays the dependence with the overall
weight conversion of the number average molar mass (Mn) and
of the polydispersity index (PDIZMw/Mn). It appears that all
the criteria of a controlled polymerization were fulfilled: (i) Mn
increased linearly with the overall weight conversion until 75%
and (ii) the polydispersity indexes remained simultaneously
lower than 1.4. Moreover, the SEC chromatograms depicted in
Fig. 2 exhibited the expected shift toward higher Mn with the
conversion progress. Nevertheless, when the experimental
Mns are compared with the theoretical values, a clear deviation
from linearity is observed at high conversion, along with an
increase in the polydispersity indexes. Such tendency was
assigned to chain transfer to the solvent, 1,4-dioxane. Taking
this transfer reaction into account and using an average value of
CtrZ4!10K4 for the chain transfer constant in the copolymerization reaction allows the experimental data to be nicely
adjusted (see all calculation details in Ref. [34]). This value
does not differ so much from that already determined for the
homopolymerization of acrylic acid where Ctr was 6!10K4
[34]. To limit the impact of this side reaction, while recovering
a sufficient amount of product, the copolymerization was
stopped at 80% conversion. Before purification, the recovered
copolymer had MnZ6300 g molK1 and Mw/MnZ1.53. After
purification by precipitation, the final copolymer exhibited the

1939

Weight conversion from 0.26 to final

2.5

3.5

4.5

log(M)

Fig. 2. SEC chromatogram (RI traces) of the methylated copolymer at various

overall weight conversions.

following characteristics: MnZ6500 g molK1 and Mw/Mn was

approximately the same as before purification, that means close
to 1.5. Some small copolymer chains might have thus been
eliminated by precipitation.
The average molar fraction of AA in the purified final
copolymer was FAAZ0.70G0.03, as determined by carboxylic
group titration (in the following, the copolymer will be referred
to as poly(St30%-co-AA70%)). As expected, the composition
drift with respect to the initial comonomer feed composition
can be explained by the difference in reactivity of the
comonomers, i.e. by the reactivity ratio values (rStZ0.72 and
rAAZ0.27). On the basis of these reactivity ratios, we
simulated the change in copolymer and remaining comonomer
compositions with the conversion progress [35] (see Fig. 3). A
good agreement of the calculated values of the remaining
comonomer composition with the experimental ones obtained
1.0

12000

2.0

10000

6000

1.6

Mw/Mn

Mn (g/mol)

1.8
8000

Molar fraction of AA

0.9

0.8

0.7

0.6
4000
1.4
2000
0
0.0

0.5
0.0
0.2

0.4

0.6

0.8

1.2
1.0

overall weight conversion

Fig. 1. Number average molar mass, Mn (full symbols), and polydispersity

index, Mw/Mn (empty symbols), of the methylated copolymer versus the overall
weight conversion, as measured by SEC. The straight line corresponds to the
theoretical Mn in the absence of transfer reaction, whereas the curved line
corresponds to Mn calculated assuming chain transfer reaction to the solvent
1,4-dioxane, with the average chain transfer constant, CtrZ4!10K4.

0.2

0.4
0.6
Overall molar conversion

0.8

1.0

Fig. 3. Theoretical proportion of acrylic acid in the comonomer mixture ()

and in the copolymer (: instantaneous composition calculated from the
composition equation assuming a terminal copolymerization model; - - -:
average composition from conversion 0) as a function of the overall molar
conversion (from 1H NMR), on the basis of the reactivity ratios, rStZ0.72
and rAAZ0.27, for an initial comonomer composition of fAA0Z0.80.
: Experimental proportion of acrylic acid in the comonomer mixture (from
1
H NMR).

C. Lefay et al. / Polymer 47 (2006) 19351945

from 1H NMR was observed. This result confirms that the

previously determined reactivity ratios are applicable to this
system, although a different initiator was used in both studies.
This is actually a normal trend since the influence of the
initiator might be lost above a given conversion. Consequently, the simulation correctly describes the copolymer
composition too. As shown in Fig. 3, the calculated
copolymer average composition in acrylic acid starts at
FAAZ0.64 and continuously increases. The instantaneous
composition, also displayed in Fig. 3, shows the enrichment in
AA units in the copolymer with the conversion increase.
Since, in controlled polymerization, conversion is proportional to chain length, the consequence of this is a gradient
composition, i.e. a continuous change in local composition
along the chain. This was confirmed experimentally by the
LAC analyses, performed at different conversions, as reported
in Fig. 4 and showing: (i) the net composition drift of the
copolymer chains toward higher FAA values with the increase
in comonomer conversion, and (ii) the narrow composition
distribution.
At the final conversion of 80%, the theoretical average
composition is 0.75, which is different from the value FAAZ
0.70G0.03 determined experimentally for the purified copolymer by carboxylic acid titration. The purification by
precipitation contributed thus to eliminate poly(acrylic acid)rich chains, most probably formed in the later stage of the
polymerization by chain transfer to the solvent.
To summarize, in one polymerization step we achieved the
synthesis of a copolymer with the following characteristics,
corresponding well to the target: MnZ6500 g molK1, Mw/Mn
close to 1.5 and FAAZ0.70G0.03. Due to the difference in
reactivity of the comonomers, the copolymer exhibited a
gradient composition, with an enrichment in AA from the
initiator fragment to the alkoxyamine chain-end. Moreover, the
copolymer could be easily dissolved in alkaline aqueous
solution, which allowed the properties in the solution state to
be examined.

3.2. Study of the poly(St30%-co-AA70%) gradient copolymer in

aqueous solution
As already pointed out in Section 2, the autocorrelation
functions measured by dynamic light scattering displayed two
diffusive relaxation processes within the whole concentration
range investigated (see insert of Fig. 5). Fig. 5 shows the
concentration dependence for both diffusion coefficients;
the values for the hydrodynamic radii that can be derived
for the fast mode and the slow mode of relaxation at
infinite dilution are, respectively, Rfast
H Z 1:2G0:2 nm and
Z
30:6G0:6
nm.
It
is
likely
that
the fast mode of
Rslow
H
relaxation corresponds to free polymer chains (the so-called
unimers), whereas the slow mode of relaxation corresponds to
large size aggregates. The unimers might be deeply compacted
by intramolecular association of the styrene subunits, which
would explain their very small size. This structuration for the
amphiphilic copolymer chains in solution is very different from
that generally observed in the case of analogous block
copolymers. The particular case of polystyrene-b-poly(acrylic
acid) diblock copolymers has been previously studied in our
group [29]. The system composed of diblock copolymers is
often strongly cooperative and the unimers, if they exist, are
not easy to observe in solution because of the large proportion
of micelles. The majority of the chains of diblock copolymers
is thus associated into micelles (for instance, the neutralized
PSt16-b-PAA55 diblock copolymer micelles exhibited an
aggregation number of 330350 at a salt concentration similar
to that used here) [29], which is not the case with our gradient
copolymers. The chemical structure of the latter could account
for this lower cooperativity in terms of aggregation.

1e-5
1e-6
A()

1940

1e-7
conversion

D (m/s)

PS

1e-8
-1

1e-9

log(R H/nm)

1e-10
1e-11
1e-12

10

15

20

25

30

C (g/L)
Polystyrene

Elution volume

Poly(methyl
acrylate)

Fig. 4. Liquid adsorption chromatography (LAC) traces of the methylated

poly(styrene-co-acrylic acid) amphiphilic copolymer obtained at different
conversions; the results are compared with a polystyrene standard.

Fig. 5. Concentration dependence of the apparent diffusion coefficient for the

slow (open circles) and the fast (closed circles) mode of relaxation. The
straightlines correspond to least-squares fits of the data and allow the
determination of the hydrodynamic radius for each mode of relaxation (see
text). The insert plot displays a typical bimodal distribution of relaxation times
obtained from DLS, measured here for CZ25 g LK1 at qZ908.

C. Lefay et al. / Polymer 47 (2006) 19351945

It appeared that the aggregates (slow relaxation mode) were

not sensitive (in terms of size and contribution to the scattered
intensity) to neither heating time during the preparation of the
solutions (from 20 min to 61 h at 90 8C; from 3 h to 24 h at
70 8C), nor salt addition (0.5 or 1 M NaCl; 0.5 or 1 M
NaHCO3). It should be emphasized that the salt content added
to the solution during its preparation was high enough to screen
electrostatic interactions. Nevertheless, a puzzling result is that
the size of those aggregates was deeply affected by the porosity
of the membrane used for filtration (from 0.45 to 0.02 mm).
They could thus be disrupted into smaller species by filtration,
leading to a sharp decrease of the scattered intensity but
keeping the concentration of the solution constant (as measured
by UV spectrometry and differential refractometry). As a
consequence, it is very likely that the aggregates were not a
major component of the system and they implied only a small
fraction of the overall concentration. It is very difficult for us to
firmly conclude about the exact nature of these aggregates but
we could invoke the explanation raised by Sedlak [36] who has
attributed the slow mode detected for polyelectrolyte solutions
(in pure water and/or in brine) to the existence of metastable
multi-chain domains. Our major conclusion is that these
aggregates implied only a small fraction of polymer chains (as
detailed below from a quantitative analysis of light scattering
data). Again, the behavior in solution of our gradient
copolymer clearly differed from that of its diblock homologues
since varying the heating time during the preparation of the
solutions had no influence on their characteristics, which is not
the case when frozen structures based on amphiphilic diblocks
are considered [37].
A more quantitative picture of this latter fact may result
from SLS measurements. The Rayleigh ratio for the solution is
given by Eq. (3), where Isolution, Isolvent and Ireference are the
intensity scattered, respectively, by the solution, the solvent
(water) and the reference (toluene) with, for the latter, the
Rayleigh ratioZRreference
Isolution qKIsolvent q
slow
Rreference Z Rfast
q C Rq
Ireference q

K is an optical constant, which depends on the refractive

index of the reference (nref) and on the specific refractive
index of the polymer (dn/dCZ0.14 mL/g) [29] according to
Eq. (5), in which Na is the Avogadros number and Pi(q) is
the particles form factor
4p2 n2ref dn=dC2
l4 Na

8e-4

(3)

The Rayleigh ratio for the fast and the slow modes of
relaxation is given by Riq Z Ai qRq , where Ai(q) is the
contribution to the scattered intensity of each relaxation
process obtained from analysis of the DLS data (see Eq. (1)).
The Rayleigh ratio may be related to the weight average molar
mass of the scattering entities (Mw,i) according to:



KCi
1
Z
C
2A
C
(4)
2;i i Pi q
Mw;i
Riq

KZ

When considering the fast mode of relaxation, the Eq. (4)

may be simplified. Firstly, Pfast(q)Z1 since the particles
displayed a very small size in solution (this was actually
proven by the non angular dependence of Rfast
q ). Secondly,
since the aggregates poorly contributed to the concentration,
one can assume that CfastZC, the overall copolymer
concentration. The plot of KC=Rfast
versus C should thus
q
lead to the estimate of the weight average molar mass of the
unimers (see Fig. 6). The value was indeed MwZ12,700G
2300 g molK1 in fairly good agreement with the weight
average molar mass of the amphiphilic copolymer chains
(MwZ10,000 g molK1).
It can thus be concluded that the gradient copolymer with
average composition FAAZ0.70G0.03 did not cooperatively
micellize in solution at concentrations below 25 g L (3.8!
10K3 mol LK1) like its block copolymer analogues do, but
remained under the form of free unimers compacted by
intramolecular association of styrene subunits. A very low
amount was actually involved in the formation of loose
aggregates, easily broken by filtration, which deeply affected
light scattering data despite their small contribution to the
concentration. Such conclusions are in agreement with recently
reported results on the aqueous solution behavior of
poly(styrene-co-acrylic acid)-b-poly(acrylic acid) copolymers
[38].
Even though the studied copolymer had a poor tendency to
form aggregates, it exhibited a surface activity as shown by the
surface tension measurement displayed in Fig. 7. Whatever the
initial copolymer concentration (from 1.56 to 25 g LK1; from
2.4!10K4 to 3.8!10K3 mol LK1), the surface tension of the
water solution continuously decreased with the analysis time,
and no equilibrium plateau was reached even after more than 1
day (Fig. 7A). The surface tension was systematically lower
when the copolymer concentration was larger (Fig. 7B) and no
transition could be observed. In other words, no apparent

(5)

6e-4
KC/R fast (mol/g)

Rq Z

1941

4e-4

2e-4

0
0

10

15

20

25

30

C (g/L)
Fig. 6. Concentration dependence of the ratio KC/Rq for the fast mode
of relaxation (see text), the straightline corresponds to a least-square fit of
the data.

1942

C. Lefay et al. / Polymer 47 (2006) 19351945

Surface tension (mN/m)

A 75
70

65

60

55

50
0

3600

7200

10800 14400 18000 21600 25200 28800

time (s)

Surface tension (mN/m)

B 72

67

62

57
Time = 3 600 s
Time = 15 000 s
52
0

10

15

20

25

30

C (g/L)
Fig. 7. (A) Surface tension versus time for an aqueous solution of the
poly(St30%-co-AA70%) amphiphilic copolymer (pH 10, [NaHCO3]Z12 mM) at
a concentration of 6.25 g LK1 (9.6!10K4 mol LK1). (B) Surface tension
versus copolymer concentration after 3600 and 15,000 s measurement time.

critical micelle concentration could be detected in the studied

concentration range, which is fully compatible with the light
scattering study. Those results, however, point out the
amphiphilic character of the copolymer, which is a good
preliminary indication of its stabilization property.
3.3. Use of the poly(St30%-co-AA70%) gradient copolymer as a
stabilizer in the batch emulsion polymerization of styrene
The gradient copolymer was used as the sole stabilizer in
model emulsion polymerizations of styrene. The target solids

content of the latexes was 10 wt% and all the polymerizations

were performed ab initio, in a batch process. The conditions
were selected to allow an easy comparison with our previous
work using polystyrene-b-poly(acrylic acid) diblock copolymers [28,29]. A first criterion of efficiency was the amount of
amphiphilic copolymer necessary to stabilize the latex particles
during the polymerization process. A second criterion was the
final number of latex particles (Np) for a given copolymer
concentration, i.e. the total surface area stabilized by the
copolymer. All results are summarized in Table 1.
From 1 to 8 wt% of copolymer based on styrene, all the final
latexes were stable. Such low amount of stabilizer as compared
to the usually used ASR, or other amphiphilic random
copolymers [1317], is the good indication of an efficient
adsorption at the particle surface of a large majority of the
chains in the copolymer sample. This result fully confirms the
strong, favorable impact of the narrow composition distribution of the amphiphilic copolymer.
As shown in Table 1, the z-average diameter (from DLS)
actually decreased when the amount of stabilizer was increased
from 1 to 3 wt%, but then reached a plateau. When the particle
size distribution is rather broad as it is the case here (see the
TEM particle size distributions in Fig. 8; Dw/Dn taken from the
TEM analyses ranged from 1.17 to 1.46, see Table 1), DLS
cannot be considered as an appropriate method to measure the
average diameter as it leads to size discrimination (the smallest
particles are not fully taken into account). Consequently, we
decided to rely on the TEM analyses and to consider the
n-average diameter, Dn, derived from this technique. Dn
actually continuously decreased when the copolymer concentration was increased. Therefore, increasing the amount of
poly(St30%-co-AA70%) had a strong impact on Np, which
increased proportionally to [copolymer]0.9. Such trend was also
corroborated by the polymerization kinetics, as it appeared that
the increase in the stabilizer content led to an increase in the
conversion rate (Fig. 9), which is the expected behavior as the
polymerization rate is proportional to Np in emulsion
polymerization (for all experiments the average number of
radicals per particle was the same and close to 0.5) [39].
In conclusion, considering both the particle size and particle
size distribution, 34 wt% of macromolecular stabilizer seems

Table 1
Model emulsion polymerizations of styrene in the presence of various amounts of the amphiphilic poly(St30%-co-AA70%) copolymer
wt% copo versus
St

[copo]!104
(mol LK1
latex)

Dz (nm) (DLS)

Dn (nm)a (TEM)

Dw/Dna (TEM)

K16
Np (LK1
latex)!10

Nc

Ac (nm2)

1
2
3
4
8

1.5
3.1
4.6
6.2
12.3

193
170
130
134
145

130
123
92
94
75

1.35
1.18
1.17
1.18
1.46

7
10
23
22
43

1400
1900
1200
1700
1700

41
25
23
16
10

Styrene/waterZ10 wt%; [NaHCO3]Z12 mM; [NaOH]Z973 mM depending on the amount of copolymer; TZ70 8C; timeZ3 h; conversion: 100%. Np: number of
particles per liter of latex, Nc: number of copolymer chains per particle, Ac: surface area occupied by a single copolymer chain at the particle surface (all three values
were derived
P using Dn, TEM).
P
a

nD
nD
Dn Z Pi in i and Dw Z Pi ni Di3 :
i

C. Lefay et al. / Polymer 47 (2006) 19351945

30%

2 wt-%

1943

1.0

25%
20%

conversion

Number fraction

0.8

15%

0.6
0.4

10%

0.2
5%

0.0
0

60

0%
0

23

38 53 68

3 wt-%
25%

Number fraction

180

Fig. 9. Conversion versus time for the emulsion polymerizations of styrene with
different amounts of the poly(St30%-co-AA70%) amphiphilic copolymer as the
sole stabilizer. TZ70 8C; styrene/waterZ10 wt%; [NaHCO3]Z12 mM;
[NaOH]Z973 mM depending on the amount of copolymer. %: 1 wt%, ,:
2 wt%, :: 3 wt%, >: 4 wt% and C: 8 wt%.

30%

20%

15%

10%

5%

0%
0

23

38

53

68

83

98 113 128 143 158 173 188 203

53

68

83

98 113 128 143 158 173 188 203

30%

4 wt-%
25%

Number fraction

120

time (min)

83 98 113 128 143 158 173 188 203

20%

15%

10%

5%

0%
0

23

38

to be an optimal amount to obtain stable, 10 wt% solids,

polystyrene latexes, in quite a fast polymerization reaction.
The final number of particles in a latex is the direct
consequence of the nucleation step. With diblock copolymers,
it was concluded that the micelles were playing a direct role in
the nucleation mechanism. In some cases, the micelles had a
sufficiently long lifetime with respect to the nucleation step,
so that they were all turned into a latex particle. In such
systems the final number of particles matched the initial
number of block copolymer micelles [29]. This was for
instance the case for the PSt16-b-PAA55 diblock copolymer
studied before [29]. In contrast, the poly(St30%-co-AA70%)
gradient copolymer, which exhibits an overall composition
quite similar to that of the aforementioned diblock, mainly
dissolves as unimers in alkaline aqueous solution; from light
scattering results, the fraction of aggregates was actually very
small. A consequence is that the number of particles nucleated
and stabilized by the poly(St30%-co-AA70%) gradient copolymer was not as high as that with the PSt16-b-PAA55 diblock
copolymer, as illustrated in Fig. 10. The relationship between
1E+19

30%

Number of particles / L

8 wt-%

Number fraction

25%
20%

15%

1E+18

1E+17
: PSt16 -b-PAA55 ([K2 CO3] = 20 mM)

10%

: Poly(St30%-co-AA70%) (calculated from the TEM diameters)

1E+16

5%

wt % stabilizer
0%
0

23

38

53

68

83

98 113 128 143 158 173 188 203

Particle diameter (nm)

Fig. 8. Particle size number distributions (from TEM) for the polystyrene
latexes at 10 wt% solids, containing various amounts (wt% based on styrene) of
the poly(St30%-co-AA70%) amphiphilic copolymer as the sole stabilizer.

Fig. 10. Number of particles per liter of latex (Np) as a function of the weight
percent of copolymer stabilizer, based on the monomer, for the emulsion
polymerization of styrene at 10 wt% solids content (see Table 1 for the
experimental conditions and results concerning the gradient copolymer
poly(St30%-co-AA70%) and Refs. [28] and [29] for the diblock copolymer
PSt16-b-PAA55).

1944

C. Lefay et al. / Polymer 47 (2006) 19351945

the number of particles and the copolymer concentration was,

however, very similar for both species (for the gradient
copolymer, Np was proportional to [copolymer]0.9, while the
exponent ranged from 0.86 to 1.0 for the PSt16-b-PAA55
diblock copolymer, depending on the salt concentration). In
the latter case, the number of copolymer chains per particle
matched the aggregation number of the micelles (300350
with 0.020 mol LK1 of K2CO3) [29], which was a good
indication of a direct micellar nucleation mechanism.
Differently, one can estimate a number of 12001900 chains
per particle (Table 1) for the gradient copolymer, assuming
that all chains reside at the particle surface (neither burial in
the particle core, nor equilibrium solubilization in the aqueous
solution). In other words, the average surface area, Ac,
occupied by one stabilizer molecule at the particle surface was
always smaller for the gradient copolymer than for the diblock
(for instance, AcZ25 nm2 for the former, whereas it was
39 nm2 for the latter, when an amount of 2 wt% of copolymer
was used, based on the monomer; see Table 1). From the
differences and similarities of both systems, one might
propose two possible nucleation mechanisms for the
emulsion polymerizations conducted in the presence of the
gradient copolymer. (i) The nucleation exclusively takes
place in the aggregates, although in small concentration, and
the free unimers adsorb onto the continuously growing
particle surface to improve the coverage. This might explain
the large number of copolymer chains per particle (even
though the number of chains per aggregate might be much
smaller) as well as the significantly lower number of
particles than in the case of the diblock copolymer stabilizers
(Fig. 10). (ii) A homogeneous nucleation mechanism
prevails, in which the role of the amphiphilic copolymer
would be to adsorb onto the polystyrene surface of the nuclei
formed in the aqueous phase, to stabilize them during their
growth. Such an assumption is logical in the presence of a
large concentration of unimers, but does not clearly support
the dependence of the particle number on the copolymer
initial concentration. It is nevertheless very likely that both
mechanisms coexist in the system.

Table 2
Semi-continuous emulsion polymerizations of styrene and MMA/BA (35/65
wt/wt) in the presence of the amphiphilic poly(St30%-co-AA70%) copolymer
used as the sole stabilizer
Monomer(s)

Amount of stabilizer
(wt% based on the
monomers)

Latex
description

Dz (nm)
(DLS)

Polya
(DLS)

St

155

0.07

MIMA/BA

Stable no
coagulum
Stable no
coagulum

197

0.06

Solids contentZ45 wt%; [NaHCO3]Z12 mM; 1 equiv. NaOH based on the

AA groups; TZ70 8C; conversionZ100%.
a
The lower value, the narrower the particle size distribution.

3.4. Use of the poly(St30%-co-AA70%) gradient copolymer as a

stabilizer in the semi-continuous emulsion polymerization of St
and MMA/BA (35/65 wt/wt) at 45 wt% solids
The semi-continuous emulsion polymerizations performed
here were very close to industrial systems. The poly(St30%-coAA70%) gradient copolymer was used in quite a low amount, as
it was 3 wt% based on the monomers. At the end of the
polymerization, stable latexes were recovered with
the diameter given by DLS: DzZ155 and 197 nm for the
polystyrene latex and the poly(methyl methacrylate-co-n-butyl
acrylate) latex, respectively (Table 2). In both cases, the
particle size distribution was narrow: the polydispersity given
by the DLS apparatus was 0.060.07.
4. Conclusion
A well-defined, amphiphilic poly(St30%-co-AA70%) copolymer was synthesized in a single step by nitroxide-mediated
controlled free-radical copolymerization of styrene and acrylic
acid, without protection of the acid groups: MnZ6500 g molK1,
Mw/MnZ1.5 and FAAZ0.70G0.03. In addition to the good
control over molar mass and molar mass distribution, we showed
that the copolymer exhibited a narrow composition distribution
with a slight gradient. Such copolymer was an efficient stabilizer
for the emulsion polymerizations of styrene and of mixtures of
MMA and BA, till 45 wt% solids. Low amounts (typically
34 wt% based on the monomer(s)) were needed. This is
approximately a decade lower than the required amount of
classical random copolymers. The performances were, however,
below those of analogous diblock copolymers, but the great
advantage is the very easy synthesis. Owing to the broad range of
functional monomers able to be polymerized via controlled freeradical polymerization, a great variety of amphiphilic copolymers
with high stabilization efficiency is thus achievable.
Acknowledgements
Arkema is acknowledged for the financial support of CLs
PhD thesis. The authors thank P. Beaunier from UPMC for the
TEM pictures and B. Ouvrard from Arkema for the LAC
analyses.
References
[1] Kyritsis A, Pissis P, Grigorieva OP, Sergeeva LM, Brouko AA,
Zimich ON, et al. J Appl Polym Sci 1999;73:385.
[2] Shinzo O, Mamoru I, Rika T, Kazuyoshi K, Mariko T. J Polym Sci, Part
A: Polym Chem 1994;32:571.
[3] Switala M, Wojtczak Z. Makromol Chem 1986;187:2411.
[4] Switala M, Wojtczak Z. Makromol Chem 1986;187:2419.
[5] Switala M, Wojtczak Z. Makromol Chem 1986;187:2427.
[6] Switala M, Wojtczak Z. Makromol Chem 1986;187:2433.
[7] Hurley SM, Hansen FC. 2000, US Patent 6,020,061.
[8] Dragner LR, Harper DW, Thompson NH. 1998, US Patent 5,795,932.
[9] Shang-Jaw C, Miao-Hsun LS. 1994, European Patent 0, 597, 567, A2.
[10] Schmidt RE, Schultz HH, Wilson DM. 1985, US Patent 4,529,787.
[11] Hameliec AP, Lawless GP, Schultz HH. 1983, US Patent 4,414,370.

C. Lefay et al. / Polymer 47 (2006) 19351945

[12] Roth G, Smith G, Grinstein R, Whyzmuzis PD, Singhal S, Boucher S, et
al., 2000, US Patent 6,160,059.
[13] Lee DY, Kim JH. Am Chem Soc: Polym Prep 1997;38:418.
[14] Lee DY, Kim JH. J Appl Polym Sci 1998;36:2865.
[15] Lee DY, Kim JH. J Appl Polym Sci 1998;69:543.
[16] Lee DY, Kim JH, Min TI. Colloid Surf 1999;A153:89.
[17] Hwu HD, Lee YD. Polymer 2000;41:5695.
[18] Do Amaral M, Asua JM. Macromol Rapid Commun 2004;25:1883.
[19] Controlled radical polymerization. ACS symposium series 1998; 1998. p.
685.
[20] Controlled/living radical polymerization: progress in ATRP, NMP, and
RAFT. ACS symposium series 2000; 2000. p. 768.
[21] Advances in controlled/living radical polymerization. ACS symposium
series; 2003. p. 854.
[22] Hawker CJ, Bosman AW, Harth E. Chem Rev 2001;101:366188.
[23] Matyjaszewski K, Xia J. Chem Rev 2001;101:292190.
[24] Madruga EL. Prog Polym Sci 2002;27:1879924 [and references therein].
[25] Ziegler MJ, Matyjaszewski K. Macromolecules 2001;34:41524.
[26] Matyjaszewski K, Ziegler MJ, Arehart SV, Greszta D, Pakula T. J Phys
Org Chem 2000;13:77586.
[27] Couvreur L, Charleux B, Guerret O, Magnet S. Macromol Chem Phys
2003;204:2055.
[28] Burguie`re C, Pascual S, Bui C, Vairon JP, Charleux B, Davis K, et al.
Macromolecules 2001;34:4439.
[29] Burguie`re C, Chassenieux C, Charleux B. Polymer 2003;44:509.

1945

[30] The polystyrene calibration is appropriate for poly(styrene-co-methyl

acrylate) samples as shown by the MarkHouwinkSakurada parameters:
actually it leads to an error of about 5%, which is within the accepted
range for SEC analysis. The MHS parameters in THF at 30 8C are: KPSZ
11.4!10K5 dL gK1 and aPSZ0.716 for polystyrene [see Hutchinson RA,
Paquet DA Jr, McMinn JH, Beuermann S, Fuller RE, Jackson C.
Dechema Monographs 1995;131:467]; KPMeAZ9.5!10K5 dL gK1 and
aPMeAZ0.719 for poly(methyl acrylate) [Piteau G. unpublished results;
for experimental conditions see: Couvreur L, Piteau G, Castignolles P,
Tonge M, Coutin B, Charleux B, Vairon JP. Macromol Symp
2001;174:197].
[31] Degoulet C, Perrinaud R, Ajdari A, Prost J, Benoit H, Bourrel M.
Macromolecules 2001;34:2667.
[32] Berne B, Pecora R. Dynamic light scattering with applications to
chemistry, biology and physics. New York: Wiley; 1976.
[33] Brandrup J, Immergut EH, Grulke EA. Polymer handbook. 4th ed. New
York: Wiley; 1999.
[34] Couvreur L, Lefay C, Belleney J, Charleux B, Guerret O, Magnet S.
Macromolecules 2003;36:82607.
[35] Odian G. Principles of polymerization. 3rd ed. New York: Wiley; 1991.
[36] Sedlak M. Langmuir 1999;(15):404551.
[37] Khougaz K, Astafieva I, Einsenberg A. Macromolecules 1995;28:
713545.
[38] Bendejacq DD, Ponsinet V, Joanicot M. Langmuir 2005;21:17128.
[39] Gilbert RG. Emulsion polymerization. A mechanistic approach. London:
Academic Press; 1995.