Controlled free radical polymerization in water-borne dispersion using reversible addition-fragmentation chain transfer

Article abstract of DOI:10.1021/ma011617j

A novel approach to conducting controlled free radical polymerization in water-borne organic dispersions using reversible addition-fragmentation chain transfer (RAFT) has been studied. The novel approach in this study focused on eliminating monomer and oligomer transport and comprised two fundamental steps: the synthesis of dithiobenzoate-end-capped styrene oligomers in bulk followed by emulsification of these oligomers to yield a polymerizable water-borne dispersion. Dithioesters that act as chain transfer agents in the RAFT process were synthesized in situ. The free radical polymerization of the dithiobenzoate-end-capped styrene oligomers in the water-borne organic dispersion proceeded in a controlled manner: molar mass increased in a linear fashion with increasing conversion, while polydispersities remained low. The familiar red layer formation associated with RAFT polymerization in conventional emulsions was not observed under these conditions. The effects of changing the costabilizer (hydrophobe) and the degree of polymerization of the emulsified oligomers were investigated. Better control was achieved with a less hydrophilic costabilizer and for the shorter of the oligomers tested.

Full text of DOI:10.1021/ma011617j

4894
Macromolecules 2002, 35, 4894-4902
Controlled Free Radical Polymerization in Water-Borne Dispersion
Using Reversible Addition-Fragmentation Chain Transfer
J . J . Vosloo,† D. De Wet-Roos,‡ M. P . Ton ge,† a n d R. D. Sa n d er son *,†
Division of Polymer Science, Department of Chemistry, University of Stellenbosch, Private Bag X1,
De Beers Street, Matieland, 7602, South Africa, and Plascon Research Centre, Institute for Polymer
Science, University of Stellenbosch, Private Bag X1, De Beers Street, Matieland, 7602, South Africa
Received September 11, 2001; Revised Manuscript Received April 1, 2002
ABSTRACT: A novel approach to conducting controlled free radical polymerization in water-borne organic
dispersions using reversible addition-fragmentation chain transfer (RAFT) has been studied. The novel
approach in this study focused on eliminating monomer and oligomer transport and comprised two
fundamental steps: the synthesis of dithiobenzoate-end-capped styrene oligomers in bulk followed by
emulsification of these oligomers to yield a polymerizable water-borne dispersion. Dithioesters that act
as chain transfer agents in the RAFT process were synthesized in situ. The free radical polymerization
of the dithiobenzoate-end-capped styrene oligomers in the water-borne organic dispersion proceeded in
a controlled manner; molar mass increased in a linear fashion with increasing conversion, while
polydispersities remained low. The familiar red layer formation associated with RAFT polymerization in
conventional emulsions was not observed under these conditions. The effects of changing the costabilizer
(hydrophobe) and the degree of polymerization of the emulsified oligomers were investigated. Better control
was achieved with a less hydrophilic costabilizer and for the shorter of the oligomers tested.
In tr od u ction
ability of termination, since the rate of termination is
strongly dependent on the active radical concentration.
Most of the chains are dormant, which allows control
over molar masses, and subsequently the possibility of
synthesizing polymers with complex architectures, since
the suppression of termination reactions allows a “liv-
ing” system.
A vast range of synthetic materials are commercially
produced by employing conventional free radical polym-
erization. Up to 50% of these free radical polymerization
reactions are conducted in emulsions.1 This combines
the desirable features of the free radical process2
(compatibility with a wide range of monomers, conven-
tional reaction conditions, and tolerance of trace amounts
of water and impurities) with those of emulsions, e.g.,
high conversions and reaction rates, excellent reaction
heat dissipation, and low solvent content.
Termination of growing radicals and chain transfer
are, however, features inherent to the mechanism of free
radical polymerization. Chain transfer and termination
of propagating radicals lead to a loss of control over
molar mass and chain architecture as well as broaden-
ing of the molar mass distribution. These aspects may
adversely affect the mechanical properties of the final
material and lead to poor performance of the resulting
latex in its ultimate application. Furthermore, termina-
tion of growing radicals means that there is no possibil-
ity of reinitiation of these species at a later stage, for
example, in the preparation of block or gradient copoly-
mers.
These limitations have led to the research field of
controlled free radical polymerization (CFRP). Tech-
niques such as reversible addition-fragmentation chain
transfer (RAFT),3 atom transfer radical polymerization
(ATRP),4,5 catalytic chain transfer,6 nitroxide-mediated
radical polymerization,7 and degenerative chain trans-
fer8 have become common terminology among research-
ers worldwide. The common principle of these tech-
niques is that the concentration of active, propagating
radicals is kept very low by converting the propagating
radicals to various dormant states, depending on the
technique that is being used. This reduces the prob-
One of the most robust and versatile techniques used
to obtain control over the free radical polymerization
process is RAFT.3,9 In this technique organic compounds
such as dithioesters and trithiocarbonates are used as
chain transfer agents. The RAFT process is effective for
a wide range of functional monomers and does not
require stringent reaction conditions. The mechanism
of the RAFT process is shown in Scheme 1.
In this process, initiator-derived primary radicals [I•]
react with monomer units [M] to form oligomeric
radicals [R•], which undergo addition to the carbon-
sulfur double bond of the dithioester chain transfer
agents. The resulting species then loses the group P, a
good homolytic leaving group, as a radical capable of
initiating a polymerization reaction or propagating
further. Equilibrium is then established between active
(propagating) and dormant polymer chains via the
addition (rate coefficient kadd) and fragmentation (rate
coefficient k-add) steps.
To gain acceptance in industry, techniques such as
RAFT and other CFRP techniques must be successful
in emulsion systems. Emulsions used in industry today
are mostly water-based. This is in accord with the global
trend toward being more environmentally friendly, as
well as often offering additional advantages, such as
easy processing of the latexes. The kinetics of emulsion
polymerization is much more complex than that of con-
ventional solution polymerizations. Therefore, a great
deal of research effort has gone into improving the
efficiency of these CFRP techniques in aqueous systems,
in terms of obtaining a linear increase in molar mass
with conversion and narrow molar mass distribu-
tions.10-15
† Division of Polymer Science, Department of Chemistry.
‡ Plascon Research Centre, Institute for Polymer Science.
* Author for correspondence: e-mail rds@akad.sun.ac.za, fax
+27-21-808-4967.
10.1021/ma011617j CCC: $22.00 © 2002 American Chemical Society
Published on Web 05/22/2002

Macromolecules, Vol. 35, No. 13, 2002
Water-Borne Dispersion 4895
Sch em e 1. Elem en ta r y Rea ction s of th e RAF T P r ocess
Sch em e 2. F low Dia gr a m of th e Novel Ap p r oa ch to
P r ep a r in g a n Or ga n ic Wa ter -Bor n e Disp er sion
F ollow ed in Th is Stu d ya
RAFT polymerization in emulsions has also received
much attention.16,17 A common feature of these efforts
has been a phase separation resulting in the formation
of a red layer. Analysis of this layer has shown it to
consist of monomer-swollen, dithiobenzoate-end-capped
oligomers.18 It was suspected that, due to their water
insolubility, these oligomers have difficulty being trans-
ported across the aqueous phase to nucleation sites.
Two options, both intent on bypassing interval II in
conventional emulsion polymerization, are available for
researchers to pursue, in order to overcome the prob-
lems associated with water-phase transport. The first
is elimination of transport of water-insoluble compo-
nents by polymerizing in miniemulsion3,18,19 and by
using starved-feed conditions.20 The second option is the
facilitation of transport by using “associative vehicles”,
such as water-soluble sugars21 or solvents,22 to improve
transport across the aqueous phase in conventional
emulsion systems.
Our approach was to first synthesize the troublesome,
dithiobenzoate-end-capped oligomers in bulk. The bulk
reaction mixture was to be then transferred to a water/
surfactant/cosurfactant mixture and sheared to yield an
oligomer-containing, polymerizable miniemulsion. The
dithio moieties would then be situated on an oligomer
that was beyond its limiting degree of polymerization
for water solubility and encapsulated in a miniemulsion
droplet. Additional stability was provided through equi-
librium swelling of the oligomers by monomer, ensuring
further preservation of the droplets. The addition of
polymer to miniemulsions for additional stability has
been well described in the literature.23-25 Polymeriza-
tion could then recommence upon heating of the result-
ing miniemulsion. A flow diagram of this approach is
shown in Scheme 2.
a
RAFT end-capped oligomers are formed by bulk reaction
and cooled. The entire bulk reaction mixture is then im-
mediately mixed with surfactant and hydrophobe and emulsi-
fied by the application of shear. The dispersion is then heated
to reaction temperature to recommence RAFT polymerization.
RAFT agents were synthesized in situ by reacting bis-
(dithiobenzoyl) disulfide and 2,2′-azobis(isobutyronitrile)
(AIBN) in the presence of styrene monomer. This

4896 Vosloo et al.
Ta ble 1. Con ver sion a n d SEC Resu lts of
RAF T-En d -Ca p p ed Styr en e Oligom er s Syn th esized by in
Situ RAF T Bu lk Rea ction of AIBN (0.162 g),
Bis(d ith ioben zoyl) Disu lfid e (0.201 g), a n d Styr en e (26.25
g) a t 80 °C a fter 60 a n d 80 m in
Macromolecules, Vol. 35, No. 13, 2002
were removed by filtration. After removal of THF under
reduced pressure the reaction mixture was acidified to pH 1
by the dropwise addition of fuming HCl. The color changed
from brown-red to a permanent deep purple. The mixture was
extracted three times with 200 mL of diethyl ether. After
drying the organic phase over MgSO4, the ether was removed
under vacuum. The red oil so obtained was dissolved in 100
mL of absolute ethanol and reacted with DMSO (12.50 g, 0.160
mol) in the presence of a catalytic amount of iodine at room
temperature for 1 h. After keeping the reaction mixture at 0
°C for 12 h, the resulting precipitate was filtered, redissolved
in ethanol at 40 °C, and recrystallized. After drying, bis-
(dithiobenzoyl) disulfide (5.20 g, 82%, purity (1H NMR) >90%)
was obtained as a purple, crystalline product.
reaction time
bulk (min)
Mh n
Mh w
PDI
60
80
3400
7100
4200
9800
1.23
1.39
eliminated time-consuming and laborious purification
of the final dithioesters, which are usually viscous oils.
Exp er im en ta l Section
P r ep a r a t ion of St yr en e Oligom er s in Bu lk Usin g
RAF T. Oligomers of different lengths were prepared in
separate bulk reactions (a total of four batches, corresponding
to the four reactions after emulsification). A typical bulk
reaction was prepared as follows: styrene (ca. 26.25 g, 0.252
mol) was degassed for 30 min by bubbling N2 through the
monomer. AIBN (0.162 g, 0.0009 mol) and bis(dithiobenzoyl)
disulfide (0.201 g, 0.0007 mol) (the 1.5:1 molar ratio of AIBN
Ch em ica ls. Styrene (Acros Organics, 98%) was extracted
with two equal aliquots of a 0.3 M aqueous KOH solution to
remove inhibitors and distilled at reduced pressure. The
middle fraction was collected and stored over molecular sieves
(4 Å) at 2 °C for no longer than 1 week for later use.
Azobis(isobutyronitrile) (AIBN, Delta Scientific, 98%), was
recrystallized from methanol. The water that was used was
distilled and deionized (DDI). Sodium lauryl sulfate (SLS,
BDH, 90%), 1-hexadecanol (cetyl alcohol, Acros Organics, 96%),
n-hexadecane (Acros Organics, 99%), bromobenzene (Fluka,
99%), carbon disulfide (Aldrich, 99.9%), dimethyl sulfoxide
(DMSO, Saarchem, 99.5%), and hydroquinone (Aldrich, 99%)
were used as received.
to bis(dithiobenzoyl) disulfide compensated for cage effects27
)
were added, and the mixture was degassed for another 30 min.
The reaction mixture was kept under N2 and heated to 80 °C.
Thus, the RAFT agent was formed by heating the bis-
(dithiobenzoyl) disulfide in the presence of AIBN, as shown
in Scheme 3.27
The degrees of polymerization of the styrene oligomers
obtained in bulk were varied by changing the reaction time,
with some variability between batches. The reactions were run
for 60 min (60 min oligomers, 15-20% conversion) and 80 min
(80 min oligomers, similar, higher conversion) before stopping
by rapid cooling and exposure to the atmosphere. Samples
were taken and precipitated using a 2.5% (m/v) solution of
hydroquinone in methanol for SEC analysis. Samples for
MALDI-TOF analysis were removed and cooled and dried by
careful evaporation of excess monomer to avoid possible
complications with the end group analyses. The bulk SEC
results and final conversions for the bulk reactions are shown
in Table 1.
An a lysis. Molar masses were determined using size exclu-
sion chromatography (SEC). Samples were prepared for SEC
analysis by drying the polymer in vacuo and redissolving ca.
5 mg of the polymer in 1 mL of THF. The SEC instrument
consisted of a Waters 717plus autosampler, a Waters 600E
system controller, and a Waters 610 fluid unit. A Waters 410
differential refractometer was used at 35 °C as detector.
Tetrahydrofuran (THF, HPLC grade) sparged with IR grade
helium was used as eluent at a flow rate of 1 mL min-1 and
60 min per sample. The column oven was kept at 35 °C, and
the injection volume was 100 µL. Four Phenogel columns (300
mm × 7.80 mm) with respective pore sizes of 100, 103, 104,
and 105 Å were used in series. The system was calibrated using
six narrow polystyrene molar mass standards in the range of
4000-2000 000 g mol-1, supplied by Pressure Chemical. Molar
masses and structures of oligomers were determined using
matrix-assisted laser desorption/ionization time-of-flight mass
spectrometry (MALDI-TOF MS). Samples for analysis by
MALDI-TOF were prepared as follows: 10 mg of the polymer
sample was dissolved in 1 mL of THF, 1 mg of matrix in 1 mL
of THF, and 10 mg of dithranol in 1 mL of THF. These three
solutions were mixed in a ratio of 1:1:8 (sample:Ag-TFA:
dithranol). A 2 µL aliquot of this final solution was loaded into
a single well of a gold-coated MALDI plate and air-dried before
placement in the vacuum chamber of the instrument. MALDI-
TOF MS was carried out on a PerSeptive Biosystems Voyager-
DE Pro mass spectrometer operating in linear mode. The
instrument was equipped with a nitrogen laser (λ ) 337 nm)
and calibrated using a mixture of insulin (bovine, peak at
5734.59 g mol-1), thioredoxin (E. coli, peak at 11 674.48 g
mol-1), and apomyoglobin (horse, peak at 16 952.56 g mol-1).
Silver(I) trifluoroacetate (Ag-TFA) was used as a matrix, and
THF was used to dissolve both the matrix and the sample.
The accelerating voltage was 20 kV, the grid voltage 94.5%,
the guide wire voltage 0.050%, and the pressure 2.66 × 10-7
Torr. Spectra were collected as the averages of 256 scans.
P r ep a r a tion of Bis(d ith ioben zoyl) Disu lfid e. Synthesis
was based on a reaction of Rizzardo et al.26 Bromobenzene
(11.30 g, 0.072 mol) in 30 mL of dry THF was added over a
period of 30 min, at 0 °C, to magnesium turnings (2.0 g, 0.082
mol), a catalytic amount of iodine, and bromobenzene (1.26 g,
0.008 mol) in THF (10 mL). The reaction was started by mild
heating. CS2 (6.10 g, 0.080 mol) was added dropwise to the
reaction mixture, keeping the temperature below 20 °C. The
Grignard product was hydrolyzed with 100 mL of cold, distilled
water, and the formed salts, together with excess Mg turnings,
P r ep a r a tion of Disp er sion s Em p loyin g SLS a n d n -
Hexa d eca n e a s Em u lsifier s. SLS (0.433 g, 0.0015 mol) was
dissolved in water (150 g), and n-hexadecane (n-HD, 1.359 g,
0.0060 mol) was dissolved in the styrene bulk reaction mixture.
After combining the two mixtures, the resulting mixture was
sheared at 4000 rpm (Silverson L4R high shear mixer)
for 1 h.
P r ep a r a tion of Disp er sion s Em p loyin g SLS a n d Cetyl
Alcoh ol a s Em u lsifier s. SLS (0.433 g, 0.0015 mol) and cetyl
alcohol (1-HD) (1.455 g, 0.0060 mol) were mixed at 70 °C for
2 h to form an aqueous gel. After cooling, the gel was added
to the styrene bulk reaction mixture and sheared at 4000 rpm
for 1 h.
P olym er iza tion in Disp er sion s. After shear, the disper-
sions were degassed by bubbling through N2 for 30 min. The
temperature was then raised to the reaction temperature (80
°C). Samples were taken to determine the conversion gravi-
metrically and to determine the molar mass during the course
of the reaction by SEC.
A typical reaction example is as follows. n-Hexadecane
(1.359 g, 0.0060 mol) was dissolved in a solution consisting of
the entire reaction mixture of RAFT-end-capped oligomers
synthesized by the in situ RAFT bulk reaction of AIBN (0.162
g), bis(dithiobenzoyl) disulfide (0.201 g), and styrene (26.25 g)
at 80 °C for 60 min (containing about 20 wt % oligomer). The
mixture was dissolved in a solution of SLS (0.433 g, 0.0015
mol) in water (150 g). The resulting mixture was sheared at
4000 rpm for 1 h and heated to 80 °C. After 12 h, the reaction
was stopped, and the polymer was isolated by precipitation
in a 2.5% (m/v) solution of hydroquinone in methanol to give
polystyrene (67% conversion, Mh n ) 19 000, Mh w/Mh n ) 1.46 by
SEC).

Macromolecules, Vol. 35, No. 13, 2002
Water-Borne Dispersion 4897
Sch em e 3. Rea ction s th a t Ma y Occu r fr om th e Rea ction of Bis(d ith ioben zoyl) Disu lfid e w ith AIBN In itia tor
Resu lts a n d Discu ssion
An important factor in the performance of this system
involves the in situ generation of RAFT chains. Scheme
3 shows that the production of growing RAFT chains
by the in situ generation process does not result in loss
of radical activity due to the reaction of the initiating
radicals with the bis(dithiobenzoyl) disulfide. The spe-
cies generated during the reaction of the initiator
fragment with the bis(dithiobenzoyl) disulfide include
radical species (A). This species is probably quite stable
and has previously been suggested to be unreactive.27
However, it is possible that this species may undergo
side reactions other than directly producing RAFT-end-
capped chains. This is consistent with the MALDI-TOF
observations discussed later. The species produced by
these side reactions may have greater chain lengths
than the “typical” growing chain produced by normal
RAFT initiation. It should also be noted that the
formation of radical species (A) allows the formation of
chains containing two RAFT agents. Thus, chains with
architecture other than that typical of RAFT polymer-
ization reactions are possible. Also, induced decomposi-
tion of the bis(dithiobenzoyl) disulfide may occur, due
to a propagating radical reaction with the bis(dithioben-
zoyl) disulfide, and thus it is possible that the RAFT
agent is “released” into the polymerization reaction
faster than may be predicted on the basis of initiator
decomposition alone.
F igu r e 1. MALDI-TOF MS spectrum of the styrene oligo-
mers synthesized by the in situ RAFT bulk reaction of AIBN
(0.162 g), bis(dithiobenzoyl) disulfide (0.201 g), and styrene
(26.25 g) at 80 °C for 60 min.
80 min oligomers were analyzed using MALDI-TOF
MS. The spectrum obtained for the 60 min oligomers is
shown in Figure 1. MALDI spectra can be used to
determine whether the dormant polystyrene oligomers
are dithiobenzoate-end-capped.28
To prove the presence of a dithiobenzoate end group,
any mass peak in the MALDI-TOF spectrum should
fit the following expression:
Mpeak ) Minitiator + nMmonomer + Mend group
+
Mcounterion (1)
MALDI-TOF Ch a r a cter iza tion of Oligom er s.
The effect of oligomer length on the performance of the
resulting dispersions was determined using the 60 and
80 min polystyrene RAFT end-capped oligomers.
Obtaining useful MALDI-TOF spectra of the oligo-
mers was difficult, and resolution was poor. Both 60 and
where Mpeak is the molar mass value of the selected
peak, Minitiator the molar mass of the initiating group,
Mmonomer the molar mass of the monomer, n the number
of monomer repeat units, Mend
the molar mass of
group

4898 Vosloo et al.
Macromolecules, Vol. 35, No. 13, 2002
Ta ble 2. SEC Resu lts for th e P olym er iza tion a t 80 °C of
Tw o Differ en t Oligom er s Syn th esized by th e in Situ
RAF T Bu lk Rea ction of AIBN (0.162 g),
Bis(d ith ioben zoyl) Disu lfid e (0.201 g), a n d Styr en e (26.25
g) a t 80 °C for 60 a n d 80 m in a n d Em u lsified in Wa ter
(150 g) Em p loyin g a SLS (0.433 g)/n -HD (1.359 g)
Em u lsifier Com bin a tion
F igu r e 2. General expected structure of the oligomers
synthesized by in situ RAFT bulk polymerization.
conva (%)
Mh n
Mh w
PDI
Run A: SLS/n-Hexadecane + 60 min Oligomers
chain end group, and Mcounterion the molar mass of
counterion used for ionization. The calculated value for
n should be very close to a positive integer. The expected
general structure of the oligomers is shown in Figure
2. The corresponding masses for the major products are
Minitiator ) 68.09 g mol-1, Mdithiobenzoate end group ) 153.25
g mol-1, and Mcounterion(Ag+) ) 107.86 g mol-1. Figure 2
shows the expected structure of the oligomers that
contain an AIBN initiating group and a dithiobenzoate
end-cap.
The MALDI-TOF spectrum of the oligomers synthe-
sized in bulk for 60 min (Figure 1) shows that there is
a regular spacing of peaks, corresponding to the styrene
repeat unit (104.13 g mol-1), and the peak molar mass
lies at 4185.84 g mol-1. The peak molar mass corre-
sponds to a value of 37.04 for the number of styrene
repeat units (n) and is consistent with the expected
structure of the oligomers in Figure 2. When the same
is done for the oligomers synthesized in bulk for 80 min,
a value of 76.97 (with corresponding peak molar mass
8344.09 g mol-1) is obtained for n. The scatter about
an integral value of n is observed throughout the
spectrum, is within the uncertainties in peak positions
in the spectrum, and is due to poor resolution.
33
40
51
62
67
10 700
13 100
16 200
17 200
19 000
13 100
16 900
22 200
24 100
27 700
1.22
1.30
1.37
1.41
1.46
Run B: SLS/n-Hexadecane + 80 min Oligomers
31
35
43
51
63
71
8400
12 200
13 500
16 600
19 700
21 500
10 000
15 900
17 800
23 100
30 300
33 900
1.20
1.31
1.31
1.39
1.53
1.58
a
Conversion is with respect to the initial amount of monomer
used in the initial bulk reaction.
Ta ble 3. SEC Resu lts for th e P olym er iza tion a t 80 °C of
Tw o Differ en t Oligom er s Syn th esized by th e in Situ
RAF T Bu lk Rea ction of AIBN (0.162 g),
Bis(d ith ioben zoyl) Disu lfid e (0.201 g), a n d Styr en e (26.25
g) a t 80 °C for 60 a n d 80 m in a n d Em u lsified Em p loyin g a
SLS (0.433 g)/1-HD (1.455 g) Em u lsifier Com bin a tion
conva (%)
Mh n
Mh w
PDI
Run C: SLS/1-Hexadecanol + 80 min Oligomers
31
40
44
51
63
69
9 100
11 300
13 900
15 600
17 800
19 600
12 100
16 100
22 000
26 900
33 000
39 300
1.33
1.42
1.59
1.73
1.83
2.01
This confirmed that the polystyrene oligomers contain
dithiobenzoate end-caps, which were capable of being
reactivated for further propagation.
Other distributions with spacings corresponding to
the styrene repeat unit were also evident. All of these
distributions were assignable to species formed by the
RAFT process. Some of these are due to different
initiator end groups, such as the termination products
corresponding to chains initially formed by monomer
initiation (effectively with no initiator end group). A
significant secondary distribution corresponds to species
containing two RAFT groups, four initiator end groups,
styrene repeat units, and the Ag+ counterion. This
species is presumably due to peculiarities associated
with the in situ generation of the RAFT agent, which
allows multiple RAFT agents per chain (see Scheme 3),
each of which may be terminated by another chain
containing an initiator end group. These peaks could
thus correspond to the case of two midchain RAFT
groups being terminated by short-chain groups contain-
ing initiator end groups. Artifacts resulting from the
ionization process in MALDI-TOF have also been
reported in the literature.29 The most important of these
is the possibility of dimer formation and detection
during time-of-flight experiments. Such an effect could
also give rise to the second observed distribution,
although the second (artifact) distribution would then
be expected at twice the masses of the main distribution.
Thus, it is unlikely that this effect is the source of the
second observed distribution, although tests to eliminate
this possibility were inconclusive, due to the difficulties
in reliably obtaining useful spectra. Further investiga-
tions of higher resolution MALDI-TOF spectra are in
progress.
Run D: SLS/1-Hexadecanol+ 60 min Oligomers
31
36
43
50
65
8 600
10 300
12 100
15 500
19 000
17 700
19 800
24 600
33 600
44 200
2.06
1.94
2.03
2.17
2.32
a
Conversion is with respect to the initial amount of monomer
used in the initial bulk reaction.
decanol were used as cosurfactants in a 4:1 ratio with
the anionic surfactant, SLS, in the emulsification step.
For each of the two SLS/cosurfactant emulsifier com-
binations the effect of changing the molar mass of the
emulsified oligomers was also investigated. The results
of each emulsifier combination employed with each of
the two different oligomers are shown in Tables 2 and
3 and in Figures 3 and 4 (conversions given are with
respect to the initial amount of monomer used in the
bulk step). Final particle sizes for these systems were
much larger than for a typical miniemulsion (ap-
proximately 600-700 nm diameter as measured using
dynamic light scattering). This is presumably due to the
relatively low shear used in the preparation of the
dispersions but is also likely to be due to stability issues
due to swelling of droplets with the RAFT oligomers,
since the particle size for dispersions prepared without
the oligomers (150-200 nm) is much smaller.
The following discussions of the results of these
polymerizations will be compared what might be ex-
pected for miniemulsion polymerizations. This compari-
son is useful because of the basic similarities between
miniemulsions and the systems studied here, such as
Recom m en cin g of P olym er iza tion in Disp er sion .
After the initial bulk step, n-hexadecane and 1-hexa-

Macromolecules, Vol. 35, No. 13, 2002
Water-Borne Dispersion 4899
Results of the dispersions employing 1-hexadecanol
(Table 3, Figure 4) were not good, especially with regard
to molar mass distributions. Fatty alcohols have been
reported not to stabilize droplets in miniemulsions as
efficiently against Ostwald ripening as do the corre-
sponding alkanes.24
If this is the case, nucleation sites outside of the
preformed droplets will therefore be better supplied with
free monomer in a dispersion employing 1-hexadecanol
than one using n-hexadecane as a cosurfactant. This
may lead to uncontrolled, conventional emulsion po-
lymerization sites that will cause a broadening of the
molar mass distribution. This broadening was indeed
observed in our systems employing 1-hexadecanol as
cosurfactant, although the source was not directly
verified. More importantly, the molar mass distributions
in the 1-hexadecanol systems appeared to be bimodal,
with a new component of (approximately constant) much
higher molar mass than the original chains. This is
strongly suggestive of extensive polymerization in the
absence of the RAFT agent for the component corre-
sponding to the new distribution, which in turn suggests
polymerization outside the original dispersion droplets/
particles. This will be addressed further in the discus-
sion of the molar mass distributions below.
F igu r e 3. Number-average molar mass and polydispersity
vs conversion at 80 °C for the polymerization of two different
oligomers synthesized by the in situ RAFT bulk reaction of
AIBN (0.162 g), bis(dithiobenzoyl) disulfide (0.201 g), and
styrene (26.25 g) at 80 °C for 60 and 80 min and emulsified in
water (150 g) employing a SLS (0.433 g)/n-HD (1.359 g)
emulsifier combination. The dotted line indicates predicted
molar masses.
None of the emulsifier systems exhibited any colloidal
destabilization as the polymerization reaction pro-
ceeded, and no phase-separated red layer was observed.
This observation is different from some in previous
studies of systems containing oligomeric RAFT
agents,18,19,30 where the latexes were destabilized when
such RAFT agents were used. In those systems, the
monomer type or degree of polymerization of the oligo-
mers was different, and the miniemulsions were pre-
pared differently, also resulting in different (usually
smaller) particle sizes. An important difference here is
that the entire bulk phase was dispersed after the
period when a significant number of very short oligo-
mers were present in the system, which is a speculated
source of destabilization. Note also that in the current
case the oligomers were preswollen with monomer in
the bulk step, and there was no new batch of initiator
added, as is typically the case when oligomeric or
polymeric RAFT agents are used in miniemulsions.
Presumably one or more of these factors is responsible
for the observed differences. Many of the comparable
systems consisted of much smaller particles, which were
predicted to undergo superswelling by Luo et al.31 The
stability of the dispersions in this study is consistent
with the predictions of Luo et al.,31 who predict no
superswelling-based destabilization for larger droplets
and for high hydrophobe concentrations, such as is the
case here.
F igu r e 4. Number-average molar mass and polydispersity
vs conversion at 80 °C for the polymerization of two different
oligomers synthesized by the in situ RAFT bulk reaction of
AIBN (0.162 g), bis(dithiobenzoyl) disulfide (0.201 g), and
styrene (26.25 g) at 80 °C for 60 and 80 min and emulsified
employing a SLS (0.433 g)/1-HD (1.455 g) emulsifier combina-
tion. The dotted line indicates predicted molar masses.
the use of the hydrophobic cosurfactant, and the initial
droplet/particle formation by the application of shear.
In the SLS/n-hexadecane systems (Table 2, Figure 3)
the molar mass increased in a linear and predictable
fashion with increasing conversion, which is consistent
with the behavior of a living system. Furthermore, the
molar mass distribution remained relatively narrow,
with polydispersity gradually increasing with conver-
sion. Throughout the reaction, the observed molar
masses were larger than those predicted (shown as a
dotted line in Figure 3), with the deviation slowly
increasing with conversion. This deviation may be in
part attributed to incomplete conversion of the bis-
(dithiobenzoyl) disulfide to RAFT chains. This is cer-
tainly possible, since the purity of the bis(dithiobenzoyl)
disulfide was between 90 and 100%. The increasing
polydispersity and deviation of molar mass from predic-
tions indicates gradual loss of control; i.e., the system
is not ideal. The molar mass evolution is considered in
further detail later.
The rates of reaction are similar to those observed in
miniemulsion systems studied in this laboratory32 start-
ing with a normal RAFT agent. This suggests, but does
not prove, that loss processes due to inefficiencies in the
in situ generation of the RAFT agent are small, as may
be expected by examination of Scheme 3, which suggests
that most initiator-derived radicals result eventually in
growing chains, with some termination events. Com-
plete conversions seem to be unattainable in these
systems, as shown in Figure 5.
One of the characteristics of the RAFT process
(Scheme 1) is that the reversible deactivation reaction
is much faster than the propagation reaction of the
growing polymer chain (kadd . kp). The deactivation

4900 Vosloo et al.
Macromolecules, Vol. 35, No. 13, 2002
F igu r e 5. Conversion vs time at 80 °C for the polymerization
of two different oligomers synthesized by the in situ RAFT bulk
reaction of AIBN (0.162 g), bis(dithiobenzoyl) disulfide (0.201
g), and styrene (26.25 g) at 80 °C for 60 and 80 min and
emulsified in water (150 g) employing a SLS (0.433 g)/n-HD
(1.359 g) and a SLS (0.433 g)/1-HD (1.455 g) emulsifier
combination.
F igu r e 7. Molar mass distributions for the polymer produced
by the reaction at 80 °C of oligomers synthesized by the in
situ RAFT bulk reaction of AIBN (0.162 g), bis(dithiobenzoyl)
disulfide (0.201 g), and styrene (26.25 g) at 80 °C for 80 min
and emulsified in water (150 g) employing a SLS (0.433 g)/n-
HD (1.359 g) emulsifier combination.
F igu r e 8. Molar mass distributions for the polymer produced
by the reaction at 80 °C of oligomers synthesized by the in
situ RAFT bulk reaction of AIBN (0.162 g), bis(dithiobenzoyl)
disulfide (0.201 g), and styrene (26.25 g) at 80 °C for 80 min
and emulsified in water (150 g) employing a SLS (0.433 g)/1-
HD (1.455 g) emulsifier combination.
F igu r e 6. Molar mass distributions for the polymer produced
by the reaction at 80 °C of oligomers synthesized by the in
situ RAFT bulk reaction of AIBN (0.162 g), bis(dithiobenzoyl)
disulfide (0.201 g), and styrene (26.25 g) at 80 °C for 60 min
and emulsified in water (150 g) employing a SLS (0.433 g)/n-
HD (1.359 g) emulsifier combination.
The molar mass distributions for the two different
emulsifier systems are shown in Figures 6-9. In an
ideal system where the free radical polymerization
process is conducted in a controlled manner, the molar
mass distribution should become narrower as the po-
lymerization reaction progresses, and molar mass as a
function of conversion should be predictable. The curves
shown in Figures 6-9 show the signal obtained from
the SEC instrument vs the logarithm of the molar mass
of the two different emulsifier systems containing the
different oligomers. The different reactions are referred
to as follows:
Reaction A: SLS/n-hexadecane (n-HD) system con-
taining oligomers synthesized in bulk for 60 min.
Reaction B: SLS/n-HD system containing oligomers
synthesized in bulk for 80 min.
Reaction C: SLS/1-hexadecanol (1-HD) system con-
taining oligomers synthesized in bulk for 80 min.
Reaction D: SLS/1-HD system containing oligomers
synthesized in bulk for 60 min.
reaction is also faster than the reactivation of the
dormant chain (kadd > k-add). As conversion increases,
higher internal viscosities of the droplets may cause the
deactivation reaction of the RAFT process to become
diffusion-controlled and to become slower (since this
reaction is between long chains and the chemically
controlled rate for deactivation is very rapid). This
happens before the reactivation reaction becomes dif-
fusion-controlled, causing the ratio of the rate coef-
ficients of the deactivation and reactivation reactions
(kadd/ k-add) to decrease. The relative concentration of
the active species will then increase. Propagation will
therefore not necessarily be faster, just more likely to
take place (relatively) than before. The increase in the
concentration of the active species may increase the
probability of irreversible termination through bimo-
lecular coupling. The change in the ratio of rate coef-
ficients will probably also lead to broadening of the
molar mass distributions. This is proposed as a possible
explanation for the loss of control in the later stages of
the reactions and is currently under investigation.
In Figure 6 all curves show tailing toward the lower
molar mass region. This could be indicative of some
fraction of the oligomers not being reinitiated in the

Macromolecules, Vol. 35, No. 13, 2002
Water-Borne Dispersion 4901
study were a broad spread of droplet nucleation times,
leading to uneven distribution of RAFT agents between
particles, and the potential loss of living character of
the polymeric RAFT agents before the reaction starts
(leading to a low molar mass tail of unactivated chains).
A large spread of nucleation times is unlikely in this
system, because of the large particle sizes. The large
particle sizes lead to a large average number of propa-
gating radicals per particle for most of the reaction
(estimated to be about 40 from early in the reaction)
and a very low particle number. The low particle
number means that the probability of not nucleating a
droplet quite early in the reaction is extremely low, since
the rate of reaction is high, and thus the average
number of propagating species per particle is high. The
second of the events described above may occur in the
current study and may lead to some broadening of the
molar mass distribution. However, such an explanation
seems unable to account for the significant increase in
polydispersity with conversion, since the number of dead
chains required would have to be very high.
F igu r e 9. Molar mass distributions for the polymer produced
by the reaction at 80 °C of oligomers synthesized by the in
situ RAFT bulk reaction of AIBN (0.162 g), bis(dithiobenzoyl)
disulfide (0.201 g), and styrene (26.25 g) at 80 °C for 60 min
and emulsified in water (150 g) employing a SLS (0.433 g)/1-
HD (1.455 g) emulsifier combination.
The sample at 67% conversion in run A shows a
significant fraction of its distribution in the lower molar
mass region. This may also be an indication that at
higher conversions shorter chains form by transfer and
then terminate, since deactivation might have become
diffusion-controlled at this stage. In Figure 7 the tailing
toward the lower molar mass region is not as pro-
nounced as in Figure 6, although it is still evident. The
earlier samples (31-51% conversion) also show a sig-
nificant high molar mass shoulder, which is not appar-
ent for the later samples. This is qualitatively consistent
with the behavior of run A.
miniemulsion polymerization step. This tailing could be
due to a large fraction of termination products during
the polymerization, leaving a tail of shorter chains. The
tailing may also be due to delayed release of RAFT
agents (of lower molar mass, due to shorter growth
periods), due to the nature of their in situ formation,
which may occur until significant conversions are
reached. Another important factor that may be a
peculiarity of the in situ RAFT formation process is the
nature of some of the chains that may be formed by this
process (see Scheme 3). Some of these species can have
multiple growth sites when more than one RAFT agent
exists per chain and thus gives chains of higher molar
mass. If termination of the intermediate radical is
possible, the terminated product may have significantly
higher molar mass than the living chains. Some of these
terminated chains may cease growing (the situation
being complicated by the possibility of more than one
growth site per chain), leading to broadening of the
molar mass distribution with conversion. Moreover,
such termination reactions reduce the amount of RAFT
agent per chain, thus increasing the average molar mass
of the chains later formed, since there is now more
monomer per growing chain. Thus, deviation above the
predicted molar mass by this mechanism, while still
retaining much of the living character, is a strong
possibility. This is consistent with the observations. A
further factor that is consistent with such a mechanism
is the high molar mass peak beyond the main peak at
the lowest conversions. This peak is presumably due to
a termination process (giving a molar mass 2 or more
times the average, as observed). A similar peak is not
evident in later samples, which may be because the
formation and termination of these species only occurs
early in the polymerization, while a considerable amount
of unreacted bis(dithiobenzoyl) disulfide is present. This
would also cause a visible tailing toward low molar mass
at intermediate conversions, as observed. The formation
of significant amounts of polymer outside the droplets
appears unlikely, since throughout the reaction there
is no fraction of significantly different molar mass from
the main distribution.
There is very pronounced tailing (later characterized
as secondary peak formation) toward the high molar
mass region through the sample set in Figures 8 and 9
(much more distinct in Figure 9). Moreover, the polymer
in the higher molar mass region is often of significantly
higher molar mass than the main, “controlled” peak and
does not change significantly with respect to the peak
region with conversion. The relative amount of this
polymer increases with conversion. The peak molar
masses of the main peak are also much lower as a
function of conversion than in Figures 6 and 7. These
results strongly suggest that a large fraction of polym-
erization is occurring in an uncontrolled manner, in the
presence of much less RAFT agent than for the con-
trolled fraction.
This is consistent with polymerization reactions lead-
ing to high molar mass compounds at aqueous phase
nucleation sites outside of the original emulsified drop-
lets. Such polymerization sites probably contain little
RAFT agent, especially if formed after the initial
emulsification step, due to the likely poor transport of
the water-insoluble RAFT agent through the aqueous
phase. This will be more probable in the systems
employing 1-hexadecanol as cosurfactant and SLS as
surfactant than in the systems using SLS and n-
hexadecane, due to poorer droplet stability. The tailing
to high molar mass in Figures 8 and 9 may also be an
indication of loss of control in the droplets themselves
at high conversions, although Figure 9 is strongly
suggestive of two environments: one where there is good
molar mass control and another where there is very
little.
Broadening of the molar mass distribution in mini-
emulsion systems has been discussed by Butte´ et al.30
The two factors suggested for the broadening in that
Clearly, the control is much poorer for the 60 min
oligomers than for the 80 min oligomers in the poor

4902 Vosloo et al.
Macromolecules, Vol. 35, No. 13, 2002
(5) Robinson, K. L.; Khan, M. A.; De Paz Ba´n˜ez, M. V.; Wang,
X. S.; Armes, S. P. Macromolecules 2001, 34, 3155.
(6) Haddleton, D. M.; Morsley, D. R.; O’Donnell, J . P.; Richards,
S. N. J . Polym. Sci., Part A: Polym. Chem. 1999, 37, 3549.
(7) Georges, M.; Hamer, G.; Listigovers, N. Macromolecules 1998,
31, 9087.
1-HD system, suggesting that oligomer length is very
important for systems that are not particularly stable.
In the case of the better-performing n-HD systems, the
60 min oligomers appear to give better results (although
not significantly better), and neither case shows evi-
dence of growth outside the main locus of polymeriza-
tion.
(8) Gaynor, S. G.; Wang, J . S.; Matyjaszewski, K. Macromolecules
1995, 28, 8051.
(9) Chiefari, J .; Chong, Y. K.; Ercole, F.; Krstina, J .; Le, T. P.
T.; Mayadunne, R. T. A.; Meijs, G. F.; Moad, G.; Moad, C. L.;
Rizzardo, E.; Thang, S. H. Macromolecules 1998, 31, 5559.
(10) Bon, S. A. F.; Bosfeld, M.; Klumperman, B.; German, A. L.
Macromolecules 1997, 30, 324.
Con clu sion s
A simple, novel two-step approach has been taken to
face the challenges of conducting RAFT polymerization
in an aqueous medium. The objective of restricting all
polymerization reactions to within dispersion droplets,
to avoid red layer formation, was attempted using a
preceding bulk step in which dithiobenzoate-end-capped
oligomers were synthesized, followed by conducting the
second part of the polymerization reaction after emul-
sification of the resultant oligomers. No destabilization
of the latex, neither initially nor through the course of
the reaction, was observed. The relative stability of the
dispersions is consistent with the predictions of Luo et
al.31 The choice of cosurfactant (hydrophobe) proved to
be an important variable in the performance of the
dispersion, with alkanes proving to be more efficient
than the corresponding fatty alcohols. The molar mass
of the end-capped oligomers synthesized in bulk before-
hand also played a role in the overall results. Lower
molar mass oligomers gave better results, which is
indicative of a larger portion of the droplet population
being preserved against monomer diffusion23-25 (since
shorter chains will allow the preservation of smaller
particles due to the oligomers’ smaller radii of gyration).
(11) Matyjaszewski, K.; Shipp, D. A.; Qiu, J .; Gaynor, S. G.
Macromolecules 2000, 33, 2296.
(12) Qiu, J .; Gaynor, S. G.; Matyjaszewski, K. Macromolecules
1999, 32, 2872.
(13) Kukulj, D.; Davis, T. P.; Suddaby, K. G.; Haddleton, D. M.;
Gilbert, R. G. J . Polym. Sci., Part A: Polym. Chem. 1997,
35, 859.
(14) Lansalot, M.; Farcet, C.; Charleux, B.; Vairon, J .-P. Macro-
molecules 1999, 32, 7354.
(15) Butte´, A.; Storti, G.; Morbidelli, M. Macromolecules 2000, 33,
3485.
(16) Hodgson, M. Emulsion Polymerization of Styrene in the
Presence of Radical Addition-Fragmentation Chain Transfer
Agents. M.Sc. Thesis, University of Stellenbosch, South
Africa, 2000.
(17) Uzulina, I.; Kanagasabapathy, S.; Claverie, J . Macromol.
Symp. 2000, 150, 33.
(18) de Brouwer, H.; Monteiro, M. J .; Tsavalas, J . G.; Schork, F.
J . Macromolecules 2000, 33, 9239.
(19) Tsavalas, J . G.; Schork, F. J .; de Brouwer, H.; Monteiro, M.
J . Macromolecules 2001, 34, 3938.
(20) Monteiro, M. J .; Sjo¨rberg, M.; Van der Vlist, J .; Go¨ttgens, C.
M. J . Polym. Sci., Polym. Chem. Ed. 2000, 38, 4206.
(21) Rimmer, S.; Tattersall, P. Polymer 1999, 40, 5729.
(22) Prescott, S.; Fellows, C. M.; Ballard, M. J .; Rizzardo, E.;
Gilbert, R. G. Macromolecules, in press.
(23) Blythe, P. J .; Morrison, B. R.; Mathauer, K. A.; Sudol, E. D.;
El-Aasser, M. S. Macromolecules 1999, 32, 6944.
(24) Blythe, P. J .; Klein, A.; Sudol, E. D.; El-Aasser, M. S.
Macromolecules 1999, 32, 6952.
Ack n ow led gm en t. The authors gratefully acknowl-
edge the support of Plascon paints and the National
Research Foundation of South Africa for financial
support of this project. J . J . Vosloo also acknowledges
Anke Fiedler from the University of Cape Town for the
MALDI-TOF analyses of the oligomers.
(25) Blythe, P. J .; Klein, A.; Sudol, E. D.; El-Aasser, M. S.
Macromolecules 1999, 32, 4225.
(26) Rizzardo, E.; Thang, S. H.; Moad, G. PCT Int. Appl. WO 99/
05099, 1998.
(27) Thang, S. H.; Chong, Y. K.; Mayadunne, R. T. A.; Moad, G.;
Rizzardo, E. Tetrahedron Lett. 1999, 40, 2438.
(28) Dong, Y.; Vladimirov, N.; Fre´chet, J . M. J . Macromolecules
1999, 32, 5186.
Refer en ces a n d Notes
(1) Gilbert, R. G. Emulsion Polymerization: A Mechanistic
Approach; Academic: London, 1995.
(2) Rizzardo, E.; Chiefari, J .; Chong, Y. K.; Ercole, F.; Krstina,
J .; J effery, J .; Le, T. P. T.; Mayadunne, R. T. A.; Meijs, G. F.;
Moad, G.; Moad, C. L.; Thang, S. H. Macromol. Symp. 1999,
143, 291.
(3) Le, T. P.; Moad, G.; Rizzardo, E.; Thang, S. H. PCT Int. Appl.
WO 98/01478, 1998.
(4) Wang, J .-S.; Matyjaszewki, K. Macromolecules 1995, 28,
7901.
(29) Axelsson, J .; Scrivener, E.; Haddleton, D. M.; Derrick, P. J .
Macromolecules 1996, 29, 8875.
(30) Butte´, A.; Storti, G.; Morbidelli, M. Macromolecules 2001, 34,
5885.
(31) Luo, Y.; Tsavalas, J .; Schork, F. J . Macromolecules 2001, 34,
5501.
(32) McLeary, J . B.; Tonge, M. P.; de Wet Roos, D.; Sanderson, R.
D.; Klumperman, L. Manuscript in preparation.
MA011617J