Access to chain length dependent termination rate coefficients of methyl acrylate via reversible addition-fragmentation chain transfer polymerization

Article abstract of DOI:10.1021/ma047476d

The reversible addition - fragmentation chain transfer - chain length dependent - termination (RAFT-CLD-T) method is employed to map out the chain length dependence of the termination rate coefficient in methyl acrylate (MA) bulk free radical polymerizati

Full text of DOI:10.1021/ma047476d

Macromolecules 2005, 38, 2595-2605
2595
Access to Chain Length Dependent Termination Rate Coefficients of
Methyl Acrylate via Reversible Addition-Fragmentation Chain
Transfer Polymerization
Alexander Theis, Achim Feldermann, Nathalie Charton, Martina H. Stenzel,
Thomas P. Davis, and Christopher Barner-Kowollik*
Centre for Advanced Macromolecular Design, School of Chemical Engineering and Industrial
Chemistry, The University of New South Wales, Sydney, NSW 2033, Australia
Received December 7, 2004; Revised Manuscript Received January 14, 2005
ABSTRACT: The reversible addition-fragmentation chain transferschain length dependentstermination
(RAFT-CLD-T) method is employed to map out the chain length dependence of the termination rate
coefficient in methyl acrylate (MA) bulk free radical polymerizations at 80 °C. Methoxycarbonylethyl
phenyldithioacetate (MCEPDA)sa novel RAFT agent carrying a methyl acryl leaving groupsis identified
as suitable for the RAFT-CLD-T method applied to methyl acrylate, as interfering inhibition and rate
retardation effects are avoided. The chain length dependency of the termination rate coefficient was
constructed in a stepwise fashion since the MA/MCEPDA system displays hybrid behavior (between
conventional and living free radical polymerization), resulting in initial high molecular weight polymers
formed at low RAFT agent concentrations. The chain length dependency of kt in the MA system for chain
lengths, i, ranging from 5 to 800 at 80 °C may be described by a value for R of 0.36 ( 0.05 (where R is the
slope of the associated log kti,i vs log i plot). An alternative RAFT agent, dimethoxycarbonylethyl
trithiocarbonate (DMCETC), may not be as ideally applicable to map CLD dependent kti,i in MA
polymerizations. Since the leaving group of both RAFT reagents is identical to the propagating methyl
acrylate radical, the addition rate coefficient of the methyl acrylate propagating radicals to the initial
and polymeric RAFT agent, kâ, was determined and found to be close to 1.4 × 106 L mol-1 s-1 for MCEPDA
and 2.1 × 106 L mol-1 s-1 for DMCETC at 60 °C.
Introduction
ing radical concentration. With a relatively simple set
of kinetic equations, time dependent rate of polymeri-
zation data recorded during a RAFT polymerization
(e.g., accessible with a differential scanning calorimetry
(DSC) instrument) can be analyzed to yield the termi-
nation rate coefficient as a function of the chain lengths
of the terminating radicals, when the rate coefficient
for initiator decomposition, kd, and its efficiencies, f, as
well as the propagation rate coefficient, kp, are known.8
We previously demonstrated that the novel methodology
can be successfully applied to map out chain length
dependent termination rate coefficients in bulk poly-
merizations if the initial RAFT agent is chosen judi-
ciously.11 In the present study, we address the possi-
bility of mapping out chain length dependent termination
rate coefficients for acrylates using the RAFT-CLD-T
method. Acrylate free radical polymerization is beset by
a range of features that are markedly different to those
of e.g. styrene polymerizations: (i) The propagation rate
coefficient in acrylate polymerizations (i.e., methyl,
butyl, or dodecyl acrylate) is considerably larger than
that observed in corresponding styrene systems,12,13 (ii)
some acrylate systems show a marked dependence of
the (average) termination rate coefficient, kt , on the
monomer to polymer conversion,14-17 and (iii) acrylates
form highly reactive radicals, which can undergo side
reactions, e.g., radical transfers forming midchain radi-
cals that are capable of propagation.18
Obtaining reliable data on chain length dependent
termination rate coefficients in free radical polymeri-
zation has been the goal of a range of studies over the
past decades. Various researchers have devised meth-
odologies to access and map out the termination rate
coefficient, kt, as a function of the chain length of the
terminating radicals.1-6 The experimental methodolo-
gies devised are often highly complex, and the subse-
quent mathematical analysis is very sophisticated.3 In
the past, a limited number of studies have addressed
the chain length dependency of the termination rate
coefficient in acrylate systems via different techniques.
For example, de Kock analyzed full molecular weight
distributions generated at low conversions for methyl
acrylate (MA), ethyl acrylate (EA), and butyl acrylate
(BA) polymerizations.4,5 In addition, the single pulse-
pulsed laser polymerization (SP-PLP) has also been
employed to deducespartially in a rather indirect
fashionsinformation about the chain length dependency
of the termination rate coefficient.1,6,7
Recently, the CAMD team has developed a relatively
simple experimental approach to map out the termina-
tion rate coefficient as a function of the chain lengths
of the terminating radicals by using the reversible
addition-fragmentation chain transfer (RAFT) process.
We term this approach the RAFT chain length depend-
ent termination technique (RAFT-CLD-T).8 Under ideal
circumstances, the RAFT methodology9 allows for a
direct correlation of the macromolecular chain length,
i, with the monomer to polymer conversion withoutsat
least to a good approximation10saffecting the propagat-
For the present investigation, methyl acrylate (MA)
has been selected as monomer for applying the RAFT-
based methodology. In previous applications of the
RAFT-CLD-T technique in styrene bulk free radical
polymerization, the overlap of the chain length depen-
dence and the conversion dependence of the termination
rate coefficient became apparent at conversion exceed-
* Corresponding author. E-mail: camd@unsw.edu.au.
10.1021/ma047476d CCC: $30.25 © 2005 American Chemical Society
Published on Web 02/25/2005

2596 Theis et al.
Macromolecules, Vol. 38, No. 7, 2005
Table 1. Monomer Properties and Kinetic Parameters for
Methyl Acrylate (MA) at 60 and 80 °C Used in the
Evaluation of the On-Line FT-NIR and DSC
Polymerizationsa
initiator concentration and kd, whereas the propagating
radical concentration [Pn] can be calculated via the
propagation equation using the propagation rate coef-
ficient, kp, and the monomer concentration at time t.
Equation 1 can thus be rewritten in its non-steady-state
form (eq 4). It is important to note that whether the
time dependent rate of polymerization data, Rp(t),
collated in the present study is analyzed via eq 1 or 4
is almost inconsequential for the analysis output, i.e.,
the resulting kt vs t functionality, due to the fact that
the differential correction term is relatively small.
T/°C cM0 /mol L-1 kp/L mol-1 s-1
kd/s-1
∆H/kJ mol-1 24
60
80
10.2b
33 000c
8.4 × 10-6 d
78
78
9.7b
47 400c
1.1 × 10-4 d
a The initiator efficiency of AIBN in MA systems is discussed
in the text. b The MA concentrations are calculated from densities
estimated in ref 12. c The propagation rate coefficients are from
ref 12. d The initiator decomposition rate coefficient, kd, of AIBN
was determined via on-line UV/vis spectrometry in ethyl acetate
following the protocol described in ref 19.
Rp(t)
d
ing =30%.8,11 Some monomers from the acrylate class
(such as MA and n-butyl acrylate (BA)) displaysfrom
relatively low conversions onwardsa strong dependence
of kt on the monomer to polymer conversion.7,15 Never-
theless, the kinetic parameters that are required for
mapping out the termination rate coefficient as a
function of the terminating radical chain length via eq
1 (e.g., the propagation rate coefficient, kp12,13) are
available from the literature or accessible via additional
experiments (such as the rate coefficient of initiator
decomposition, kd). The relevant parameters for the
present studies are collated in Table 1 for 60 and 80
°C.
The considerably higher initial monomer concentra-
tion in bulk and lower viscosity of MA compared to
n-dodecyl acrylate19 suggests that the initiator efficiency
f, of the employed initiator 2,2′-azobis(isobutyronitrile)
(AIBN) is significantly higher in MA polymerizations
and maybe more in line with those reported for sty-
rene.20 However, up to date there are no studies avail-
able regarding the progression of the initiator efficiency
with monomer to polymer conversion in the AIBN/MA
system. While the evolution of the initiator efficiency
with conversion of AIBN in styrene free radical poly-
merizations was found to display a concave shape,20 the
initiator efficiency reported for the system AIBN/DA
showed a slightly more pronounced decrease at low
monomer to polymer conversion.19
t
(
)
k ([M] - R (t) dt)
∫
p
0
p
0
2f kd[I]0e-k t
-
d
dt
kt (t) )
(4)
2
Rp(t)
2
t
(
)
k ([M] - R (t) dt)
∫
p
0
p
0
The transformation from time dependent kt data to
chain length dependent kt data proceeds via either the
theoretical molecular weight evolution or the measured
Mn vs conversion data (as in the present study). How-
ever, the inherent polydispersity of the polymer samples
will lead to an averaging of kt values. Thus, strictly
speaking, the reported kti,i data are averages of the (very
narrow) distribution of chain lengths present at each
point in time.
In the present contribution, we provide experimental
data that illustrate the scope of the applicability of the
RAFT-CLD-T method to acrylate systems. The RAFT
agents selected for this study were chosen such that
they induce as little retardation as possible, with the
aim of eliminating all RAFT agent induced rate effects.
To achieve this aim, a destabilizing Z group, e.g. Z )
benzyl, was selected along with a variety of R groups
ranging from cumyl to those mimicking a propagating
methyl acrylate radical. We will demonstrate in detail
in the experimental part of the results and discussion
section how the choice of the R group can affect the
outcome of the RAFT-CLD-T method.
Further, the applicability of eq 1, which was used to
describe the styrene system, needs to be verified for the
use in methyl acrylate polymerizations.
Experimental Section
2
Materials. Methyl acrylate (99%, Aldrich, M ) 86.06 g
mol-1) was freed from the inhibitor by percolating over a
column of activated basic alumina. 2,2-Azobis(isobutyronitrile)
(AIBN, DuPont) was recrystallized twice from ethanol prior
to usage. Cumyl phenyldithioacetate (CPDA) was prepared as
described earlier, using n-hexane as the solvent.21
kp
2
t
d
kt (t) )
f kd[I]0e-k t([M] - R (t) dt)2 (1)
∫
0
p
0
Rp (t)
In this context it should be noted that eq 1 can also be
represented in a slightly different form: The change of
the radical concentration in a radical polymerization
system is given by the difference of radicals produced
and radicals being terminated at any point in time (eq
2).
Methoxycarbonylethyl phenyldithioacetate (MCEP-
DA) was synthesized starting from phenyldithioacetic acid,
which was prepared analogous to a method described earlier.21
15 g (90 mmol) of phenyldithioacetic acid were poured into a
solution of 3.6 g (90 mmol) of sodium hydroxide in 15 mL of
water. The mixture was stirred for 5 min. Subsequently, the
water was evaporated to give sodium phenyldithioacetate in
d[Pn]
) 2f kd[I] - 2kt[Pn]2
(2)
1
quantitative yield. 300 MHz H NMR (D2O): δ [ppm] ) 4.45
dt
(s 2H CH2), 7.28-7.47 (m 5H C6H5). 3.75 g (20 mmol) of sodium
phenyldithioacetate and 3.3 g (20 mmol) of methyl 2-bromo-
propionate were dissolved in 25 mL of anhydrous ethanol and
refluxed for 15 h. Then, 200 mL of diethyl ether was added,
and the mixture was washed once with water. The ether
fraction was dried with magnesium sulfate and evaporated in
a vacuum, and the residual was purified by chromatography
over silica gel with toluene to yield 2.7 g (54%) of the pure
methoxycarbonylethyl phenyldithioacetate. 300 MHz 1H NMR
(CDCl3): δ [ppm] ) 1.55 (d 3H J ) 7.2 Hz CH3), 3.72 (s 3H
O-CH3), 4.30 (s 2H CH2), 4.59 (q 1H J ) 7.2 Hz CH-CH3),
Solving eq 2 for kt leads to eq 3.
d[Pn]
2f kd[I] -
kt )
dt
(3)
2[Pn]2
Similar to eq 1, the initiator concentration, [I], at any
point in time, t, can be calculated from the initial

Macromolecules, Vol. 38, No. 7, 2005
Termination Rate Coefficients of Methyl Acrylate 2597
7.23-7.38 (m 5H C6H5). 75 MHz 13C NMR (CDCl3): δ [ppm]
) 16.30 (1C CH-CH3), 48.00 (1C CH-CH3), 52.68 (1C
O-CH3), 57.42 (1C CH2), 127.30 (1C para C6H5), 128.54/129.06
(4C ortho/meta C6H5), 136.48 (1C -C6H5), 171.33 (1C CdO),
232.89 (1C CdS). IR (ATR) λ-1 [cm-1] ) 3061, 3027 (ar C-H),
2950, 2840 (aliph C-H), 1734 (CdO), further signals at 1494,
1451, 1374, 1306, 1220, 1156, 1074, 983, 853, 696.
Polymerizations (DSC). Solutions of methyl acrylate (MA)
with AIBN and RAFT agent were thoroughly deoxygenized via
four subsequent freeze-pump-thaw cycles and handled inside
a glovebox or glovebag filled with dry nitrogen gas. The
individual species concentrations can be found in the figure
captions describing the associated experiments. Exactly weighed
amounts of sample (50-70 mg) were loaded to aluminum pans
that were sealed with aluminum lids. The polymerization heat
was determined isothermally at 80 °C via measuring the heat
flow vs an empty sample pan in a differential scanning
calorimeter (Perkin-Elmer DSC 7 with a TAC 7/DX thermal
analysis instrument controller). The DSC instrument was
calibrated with a standard indium sample of known mass,
melting point temperature, and known associated enthalpy
change. The rate of polymerization, Rp, was calculated using
literature values for the heat of polymerization of butyl
acrylate (BA, ∆H ) 78 kJ mol-1),24 which should be very
similar to the heat of polymerization for MA. The heat of
polymerization for MA bulk polymerizations was also inde-
pendently determined via DSC and determination of the final
conversion by SEC within the present study (∆H ) 78 kJ
mol-1), confirming that the heats of polymerization of BA and
MA are close to identical.
Dimethoxycarbonylethyl trithiocarbonate (DMCETC)
was synthesized analogous to the method described by Tamami
and Kiasat.22 20 g of dry Ambersep 900OH ion exchanger
(equivalent to 20 mmol of OH- ions) and 100 mL of carbon
disulfide were stirred for 5 min at room temperature, whereby
the color of the ion exchanger changed to intensive red. To
this suspension, 6.68 g (40 mmol) of methyl 2-bromopropionate
was added, and the mixture was stirred for another 5 h.
Subsequently, the ion exchanger was filtered off, and the
solution was evaporated in a vacuum. The residual was
purified by chromatography over silica gel with toluene to yield
8.0 g (71%) of the pure dimethoxycarbonylethyl trithiocarbon-
ate. 300 MHz 1H NMR (CDCl3): δ [ppm] ) 1.55 (d 6H J ) 7.4
Hz CH3), 3.69 (s 6H O-CH3), 4.75 (q 2H J ) 7.4 Hz CH-CH3).
75 MHz 13C NMR (CDCl3): δ [ppm] ) 16.80 (2C CH-CH3),
48.04 (2C CH-CH3), 52.73 (2C O-CH3), 171.05 (2C CdO),
219.33 (1C CdS). IR (ATR) λ-1 [cm-1] ) 2981, 2851 (aliph
C-H), 1733 (CdO), further signals at 1434, 1375, 1308, 1253,
1156, 1065, 983, 854, 809, 731, 695.
Simulations. All simulations have been carried out using
the program package PREDICI, version 5.36.5, on a Pentium
IV (HT) 2.6 GHz or Athlon XP 2500+ IBM-compatible com-
puter.
Polymerizations (FT-NIR). Polymerization under on-line
FT-NIR conditions were carried out to determine exact mono-
mer to polymer conversions for samples used in subsequent
SEC analysis to obtain number-average molecular weight, Mn,
vs conversion data. The process was as follows: Solutions of
methyl acrylate (MA), RAFT agent, and AIBN were prepared
and mixed thoroughly. (The individual species concentrations
are given in the corresponding figure captions.) The solutions
were subsequently subjected to four freeze-pump-thaw cycles
to remove any residual oxygen. A small amount of solution
was then transferred into a 2 mm optical path length Infrasil
cell (Starna Optical) and sealed with a rubber septum.
Monomer conversions were determined at 60 °C via on-line
Fourier transform near-infrared (FT-NIR) spectroscopy by
following the decrease of the intensity of the first vinylic
stretching overtone of the monomer (λ-1(MA) ) 6169 cm-1).
The FT-NIR measurements were performed using a Bruker
IFS66\S Fourier transform spectrometer equipped with a
tungsten halogen lamp, a CaF2 beam splitter, and a liquid
nitrogen cooled InSb detector. Each spectrum in the spectral
region of 8000-4000 cm-1 was calculated from the coadded
interferograms of 12 scans with a resolution of 2 cm-1. A
depiction of a typical FT-NIR spectrum of an acrylate can be
found in ref 19. For conversion determination, a linear baseline
was selected between 6240 and 6100 cm-1. The integrated
absorbance between these two points was subsequently used
to calculate the monomer to polymer conversion via Beer-
Lambert’s law. It should be noted that other integration
methods or methods using only the variation of the peak height
at 6169 cm-1 have been tested and yield identical results. In
regular intervals, a small sample was withdrawn from the
reaction mixture with an airtight syringe that had been
flushed three times with nitrogen gas. The sample was
transferred into a flask containing THF with hydroquinone
as inhibitor and immediately subjected to SEC analysis.
Molecular Weight Analysis. Molecular weight distribu-
tions were measured via size exclusion chromatography (SEC)
on a Shimadzu modular system, comprising an autoinjector,
a Polymer Laboratories 5.0 µm bead-size guard column (50 ×
7.5 mm), followed by three linear PL columns (105, 104, and
103 Å), and a differential refractive index detector. The eluent
was tetrahydrofuran (THF) at 40 °C with a flow rate of 1 mL
min-1. The system was initially calibrated using narrow
Results and Discussion
The theoretical requirements for an optimum ap-
plicability of the RAFT-CLD-T method are twofold: (i)
the RAFT process should not lead to rate retardation
or inhibition effects and thus provide an additional
pathway of potential radical loss, and (ii) the RAFT
process should not lead to hybrid behavior between
conventional and living free radical polymerization and
induce the formation of initial high molecular weight
polymer.21 The first criterion is necessary for the RAFT-
CLD-T method to function, whereas the second criterion
is desirable. Rate retardation and inhibition phenomenas
irrespective of their exact mechanistic underpinning25s
provide an additional radical loss pathway, which in the
context of the RAFT-CLD-T method leads to a severe
overestimation of the chain length dependent termina-
tion rate coefficient and, in addition, can lead to an
apparent chain length dependent termination (see
below). Even if rate retardation phenomena are absent
in a given RAFT system, the polymerization may still
display a rapid rise in molecular weight with monomer
conversion with living behavior following suit.21 While
such hybrid behavior does not invalidate the RAFT-
CLD-T method, it drastically reduces the information
content that can be obtained because the termination
rate information on the shorter chain lengths is lost.
In the initial studies where the RAFT-CLD-T method
has been applied to styrene systems,8,11 CDPA has been
found to be ideally suited. It is thus appropriate to
initially test the applicability of CPDA in the RAFT-
CLD-T method of acrylates.
Figure 1 depicts the evolution of the full molecular
weight distributions with monomer to polymer conver-
sion in a CPDA-mediated RAFT polymerization of MA
at 60 °C with the initial CPDA concentration close to 3
× 10-3 mol L-1. The distributions are monomodal and
shift linearly with conversion, as evidenced by the Mn
vs conversion evolutions depicted in Figure 2. While the
molecular weight evolution is linear for two initial RAFT
agent concentrations, it is evident that the molecular
polystyrene standards ranging from 540 to 2 × 106 g mol-1
.
The resulting molecular weight distributions have been re-
calibrated using the Mark-Houwink parameters for poly-
(methyl acrylate) (K ) 19.5 × 10-3 dL g-1, a ) 0.66).13 The
Mark-Houwink parameters for polystyrene read (K ) 14.1 ×
10-5 dL g-1 and a ) 0.70).23

2598 Theis et al.
Macromolecules, Vol. 38, No. 7, 2005
Figure 1. Evolution of full molecular weight distributions in
the CPDA-mediated bulk free radical polymerization of methyl
acrylate (MA) at 60 °C. The initial CPDA concentration was
2.96 × 10-3 mol L-1, and the AIBN concentration was 1.57 ×
10-3 mol L-1
.
Figure 3. Rp vs time curves (a) and the resulting conversion
vs time evolutions (b) in free and CPDA-mediated MA poly-
merizations at 80 °C, measured by recording the heat of
polymerization. The AIBN concentration was close to 3.0 ×
10-3 mol L-1
.
2 and 4 (see Scheme 2) in order to prevent rate
retardation effects. The stability of an acrylate propa-
gating free radical isscompared to a styryl free radicals
greatly reduced. Thus, the tendency of species 2 and 4
to fragment is greatly reduced, and it can therefore be
anticipated that acrylate polymerization are subject to
more pronounced inhibition and rate retardation effects
than the corresponding styrene systems. Indeed, severe
rate retardation and inhibition effects have been re-
ported in MA RAFT systems, employing dithiobenzoate
type RAFT agents.26,27 It can thus be concluded that a
RAFT agent which does not induce retardation effects
in styrene polymerizations (i.e., CPDA) may do so in the
corresponding MA system. However, it is very important
to note that slow reinitiation of the R group radicals
cansas may be the case in cumyl radicals reacting with
acrylatessequally lead to severe rate reductions early
in the polymerization. Figure 3 shows the experimental
Rp vs time curves (a) and the resulting conversion vs
time evolutions (b) in conventional and CPDA-mediated
MA polymerizations at 80 °C. Inspection of Figure 3b
clearly indicates that inhibition and rate reduction
effects can be observed.
The use of cumyl dithiobenzoate as RAFT agent in
acrylate polymerizations will indeed increase the reac-
tivity of the CdS double bondsthus avoiding hybrid
behaviorsbut by the same token induce strong rate
retardation effects.26 The aim of designing RAFT agents
for acrylates that do not induce rate retardation effects,
while having a highly reactive CdS double, may result
in a scientific Catch 22: either the stability of the
intermediate species is high (resulting in rate retarda-
tion effects) and thus the reactivity of the double bond
Figure 2. Evolution of the number-average molecular weight,
Mn, with monomer to polymer conversion for two initial CPDA
concentrations in MA bulk polymerizations. The theoretical
molecular weight evolutions are given by the dotted lines. The
full lines represent the best-fit functions, resulting in an initial
degree of polymerization of 120 for [CPDA]0 ) 2.96 × 10-3 mol
L-1 and 64 for [CPDA]0 ) 1.06 × 10-2 mol L-1. The AIBN
concentration was 1.6 × 10-3 mol L-1. The upper part of the
figure gives the corresponding polydispersity indices, PDI.
weights do not adhere to the theoretically predicted
values (dotted lines, Figure 2).
inst
The initial rapid increase in molecular weight (DPn
) 120 at [CPDA]0 ) 2.96 × 10-3 mol L-1 and DPn
)
inst
64 at [CPDA]0 ) 1.06 × 10-2 mol L-1) can be attributed
to a too low kâ,1 value, which will induce hybrid behavior
in conjunction with a high kp value. Compared to the
earlier studied styrene RAFT systems,8,11 the observed
hybrid behavior comes at no surprise since the propaga-
tion rate coefficient of MA at identical reaction condi-
tions is 97 times larger than the same value for styrene.
Thus, the addition rate coefficient governing the pre-
equilibrium, kâ,1, can be approximately a factor of 100
less in styrene systems without inducing hybrid behav-
ior. To fulfill the second (nonessential) criterion that an
ideally suited RAFT agent for the RAFT-CLD-T method
should display (i.e., no hybrid behavior), the initial
RAFT agent design has to ensure a high reactivity of
the CdS double bond.
A rational RAFT agent design aimed at achieving a
high reactivity of the CdS double bond is partly
antagonistic to destabilizing the intermediate radicals

Macromolecules, Vol. 38, No. 7, 2005
Termination Rate Coefficients of Methyl Acrylate 2599
Scheme 1. Monomer Methyl Acrylate (MA) and
Reversible Addition-Fragmentation Chain Transfer
(RAFT) Agents Employed in the Current Study:
Cumyl Phenyldithioacetate (CPDA),
bind two polymer chains during polymerization. In
terms of the evaluation procedure leading to chain
length dependent termination rate coefficient, the ex-
perimentally determined molecular weight of the poly-
meric RAFT reagent needs to be divided by two, since
the chain length of the terminating macro radicals is
only half the length of the macroRAFT agent.
Methoxycarbonylethyl Phenyldithioacetate
(MCEPDA), and Dimethoxycarbonylethyl
Trithiocarbonate (DMCETC)
Determination of the Addition Rate Coefficient
kâ. Besides being a potential candidate for the RAFT-
CLD-T method for probing the chain length dependence
of kt in acrylate polymerizations, the structure of
MCEPDA opens several opportunities. The fact that the
leaving group, R, of the initial RAFT agent is identical
to the propagating radicals in the polymerizing system
not only is eliminating the preequilibrium reaction but
also allows to derive a set of equations that can be
employed to deduce the addition rate coefficient, kâ, in
systems displaying hybrid behavior. The initial rate of
propagation, Rp, is given by eq 5.
is sufficient (resulting in Mn vs conversion plots adher-
ing to the theoretical values) or the stability of the
intermediate species is low (resulting in no retardation
effects) and thus the reactivity of the double bond is too
low (resulting in hybrid behavior).
Rp ) kp[Pn][M]0
(5)
[Pn] is the propagating radical and [M]0 is the monomer
concentration. The rate of radical addition to the
carbon-sulfur double bond of the initial (or polymeric)
RAFT agent, Radd, can be quantified in a similar fashion
via eq 6
In light of the above experimental and theoretical
observations for previously employed RAFT agents in
the RAFT-CLD-T method (i.e., CPDA), it is appropriate
to use a RAFT agent that represents a workable
comprise for determining chain length dependent kt
data in acrylate systems. The RAFT agent design is
driven by the realization that RAFT agent caused
retardation must be avoided (see theoretical assessment
section). A RAFT agent suitable for the present study
should thus carry a destabilizing Z group, similar or
identical to the benzyl group of CPDA. Further, it is
necessary to modify the leaving R group such that the
effects of slow reinitiation are avoided. Thus, a leaving
group identical in structure to the propagating acrylate
radical was selected. Such a selection for the leaving
group simplifies the RAFT polymerization kinetics by
effectively eliminating the preequilibrium sequence.
Methoxycarbonylethyl phenyldithioacetate (MCEPDA,
see Scheme 1) is a RAFT agent that conforms to the
above criteria set. Although to some extent it can be
anticipated that MCEPDA will cause an initial rapid
increase to high molecular weights (caused by hybrid
behavior), the magnitude of the initial increase can be
controlled by adjusting the RAFT agent concentration.
Thus, by a variation of the initial RAFT agent concen-
tration, a continuous chain length function can be
constructed by accessing various molecular weight
regimes (see below).
In addition, all experiments were also carried out
using dimethoxycarbonylethyl trithiocarbonate (DM-
CETC, see Scheme 1). This RAFT reagent is sym-
metrically assembled, having no Z but two R groups,
allowing fragmentation to both sides. On one hand, it
is reasonable to assume that the third sulfur atom in
the molecule leads to an increased radical stabilization
and higher reactivity of the CdS double bond, resulting
in a lower tendency to undergo hybrid behavior. On the
other hand, the increased intermediate radical stability
will make the polymerization process more susceptible
to rate retardation phenomena. Similar as in the
architecture of MCEPDA, the leaving group is identical
in structure to the propagating acrylate radical, ensur-
ing that no slow reinitiation phenomena can occur. In
contrast to the dithioester type RAFT reagents (which
have one R and Z group), the MCEPDA reagent will
Radd ) kâ[Pn][RAFT]
(6)
[RAFT] is the initial (or polymeric) RAFT agent con-
centration. In RAFT systems that display hybrid be-
havior, both rates are correlated by the degree of
polymerization of the polymer formed instantaneously,
DPninst. DPninst can be obtained via plotting the number-
average molecular weight, Mn, vs the monomer to
polymer conversion and extrapolating to zero percent
monomer conversion.
Rp
inst
DPn
)
+ 1
(7)
R
addφ
The fragmentation coefficient, φ, in the denominator of
eq 7 is associated with the fact that the reaction product
of the addition reaction, i.e., the intermediate macro-
RAFT radicals, can also fragment to yield the starting
materials. The probability of the macroRAFT radical
undergoing a radical transfer (in contrast to fragment-
ing back to the starting materials) is 50%, since both
fragmentation pathways have identical rate coefficients.
Thus, in order for eq 7 to yield the correct value for
DPninst, the addition rate, Radd, has to be multiplied by
a factor of 0.5. In the case of the trithiocarbonate
DMCETC, there are three different fragmentation
pathways with the probability of a radical transfer being
67%. This can be compensated by multiplying Radd by a
factor of 0.67. Following a similar rationale, a hypo-
thetical ideal RAFT agent that would only allow frag-
mentation of the intermediate radical to release the R
group (and never fragment back to the starting materi-
als) would not require multiplication of Radd by any
factor other than 1. In contrast, a RAFT agent that leads
to an intermediate radical that affects a preferential
fragmentation to the starting material side would
require a factor of smaller than 0.5 on Radd. It further
has to be noted that the first propagation step already
leads to a chain length of two. Therefore, one needs to
be added to the right-hand side of eq 7. Equation 7 can

2600 Theis et al.
Macromolecules, Vol. 38, No. 7, 2005
subsequently be rearranged to eq 8, which allows for
the deduction of kâ, when the propagation rate coef-
ficient, kp, the monomer concentration, [M], the instan-
taneous degree of polymerization, DPninst, and the initial
RAFT agent concentration are known.
kp[M]0
(DPninst - 1)[RAFT]0φ
kâ )
(8)
Since it issat least with the data presently at hand in
the literaturesvery difficult to quantify the individual
reaction channels of the intermediate radical, the only
reliable avenue to kâ determination is the use of a RAFT
agent that features an R group radical identical to the
propagating radical. While the above approach does not
give access to a (potential) chain length dependency of
kâ with regard to the macroRAFT agent, the length of
the adding macroradical varies with DPninst. Of course,
a similar rationale as outlined above can also be
employed to a macroRAFT agent that is chain extended
with a monomer of its own kind. However, this approach
will invariably give a value for the main equilibrium
addition rate coefficient kâ at the corresponding chain
length.
Figure 4 depicts the evolution of the number-average
molecular weight, Mn, with monomer to polymer con-
version for the two RAFT reagents employed in the
present study. In Figure 4asassociated with the MCEP-
DA-mediated MA polymerizationshybrid behavior is
clearly shown at all except the highest MCEPDA
concentrations. The molecular weight evolution further
shows some deviation from a linear rise, which was
taken into account in the fitting process. Figure 4b
shows the analogous molecular weight evolution for the
DMCETC mediated polymerization, resulting in a linear
behavior with lower initial molecular weights. The
molecular weight vs conversion plots given in Figure 4
serve a dual purpose: (i) The plots are used as calibra-
tion for the chain length axis for the mapping of the
termination rate coefficient with chain length i. While
the agreement with the theoretically calculated chain
length is acceptable, it is preferable to employ an
experimentally determined i axis when mapping chain
length dependent termination rate coefficients (see
below). (ii) The Mn vs conversion plots can be used to
deduce values for the addition rate coefficients kâ. As
Figure 4. Number-average molecular weight, Mn, vs mono-
mer to polymer conversion in MCEPDA-mediated (a) and
DMCETC-mediated (b) bulk free radical polymerization of
methyl acrylate (MA) at 60 °C. The associated initial RAFT
reagent concentrations alongside the employed AIBN concen-
trations are given within the figure. The AIBN concentration
was 3.0 × 10-3 mol L-1. The curves represent a best fit for
molecular weight vs time evolutions. The y-axis intercepts of
the best fits (i.e. the initial molecular weights) are collated in
Table 2. The upper part of the figure gives the corresponding
polydispersity indices, PDI.
Table 2. Summary of the Initial MCEPDA and DMCETC
Concentrations in Methyl Acrylate (MA) Bulk Free
Radical Polymerizations as Well as the Associated
Degrees of Polymerization of the Instantaneously
Generated Polymer, DPninst, at 60 °Ca
inst
RAFT reagent
c0RAFT/mol L-1
DPn
kâ/L mol-1 s-1
outlined above, the degree of polymerization of the
inst
MCEPDA
MCEPDA
MCEPDA
MCEPDA
DMCETC
DMCETC
DMCETC
DMCETC
4.37 × 10-3
1.27 × 10-2
3.68 × 10-2
1.07 × 10-1
4.31 × 10-3
1.05 × 10-2
3.04 × 10-2
1.02 × 10-1
103
46
(8)
1.5 × 106
1.2 × 106
(2.6 × 106)
3.1 × 106
1.5 × 106
1.7 × 106
(0.8 × 106)
instantaneously generated polymeric material, DPn
,
has to be known. For this purpose, the resulting
y-intercept () DPninst) of the fitted data points (full lines)
given in Figure 4 was determined. Each polymerization
performed at a distinct initial RAFT reagent concentra-
39
34
11
(7)
tion yields a unique number for DPninst. The obtained
inst
DPn
values for both RAFT reagents alongside the
deduced numbers for kâ (via eq 8) are collated in Table
2. The error of this value is largely dependent on the
accuracy of the SEC analysis of the generated polymer
(estimated as 30%), as well as errors associated with
extrapolation to 0% monomer to polymer conversions.
However, with increasing DPninst, the results become
much more precise. For that reason, the values obtained
for DPninst lower than 10 were not used for the calcula-
tion of the average value. The resulting (average) kâ for
the system MA/MCEPDA reads 1.4 × 106 L mol-1 s-1
a The initiator concentration (AIBN) was close to 3.0 × 10-3 mol
L-1. The kâ values that have been deduced via eq 8 are included
as well (for details see text and the associated Figure 5).
similar number had been determined earlier by Fukuda
and co-workers in a styrene/polystyrene dithiobenzoate
system; our number agrees reasonably well with the
earlier value of 4 × 106 L mol-1 s-1 28
.
However, since
the styrene/polystyrene dithiobenzoate system is mark-
edly different from the present MA/MCEPDA system
(in terms of both chemical structure and chain length),
the agreement should not be overemphasized. It is
and for the system MA/DMCETC reads 2.1 × 106
L
mol-1 s-1. Under the present circumstances, we believe
our current numbers are accurate to a factor of 2. A

Macromolecules, Vol. 38, No. 7, 2005
Termination Rate Coefficients of Methyl Acrylate 2601
Scheme 2. Reaction Sequence Used for the Kinetic Simulations of the RAFT Process via the PREDICI Program
Package
further not surprising that the value for the DMCETC
is higher in comparison to that of the MCEPDA, since
the reactivity of the CdS double bond is likely to be
increased by the third sulfur atom in this molecule.
Chain Length Dependent Rate Coefficients. In
the following section, we will demonstrate how the
RAFT agent MCEPDA can be employed to obtain
information about the chain length dependency of the
termination rate coefficient in MA bulk free radical
polymerizations. To ensure that the resulting log kti,i
vs log i plots are a valid representation of the chain
length dependency of kt, the MA/MCEPDA system was
analyzed on the basis of the reaction set given in Scheme
2. Details on the exact implementation of the model into
the PREDICI program package can be found in refs 29
and 30.
decomposition, and the addition rate coefficient, kâ)
available. Among the kinetic parameters not available
is the fragmentation rate coefficient of the intermediate
macroRAFT radicals, k-â, as well as a potential cross-
termination reaction with propagating chains. It thus
seems mandatory to test how these two key kinetic
features affect the outcome of the stepwise construction
procedure for the log kti,i vs log i plots. Figure 5 depicts
the resulting (apparent) log kti,i vs log i dependencies
deduced from simulated Rp(t) data. For a realistic
comparability to the experimental data, a chain length
dependent kt value (using the common expression kti,i
) k0t i-R with kt0 ) 109 L mol-1 s-1 and R ) 0.4)31 was
implemented into the PREDICI model. It should be
noted that the x-axis is associated with the chain length
of the actually generated molecular masses of the RAFT
polymer and is not defined via the theoretical molecular
weight evolution based on an ideal RAFT process.
Further, the parameters for the preequilibrium and
main equilibrium have been chosen identical (under the
assumption that the radical addition and fragmentation
rate coefficients for the intermediate macroRAFT radi-
cal are chain length independent) and the reinitiation
rate of the leaving R group krein was chosen identical to
kp (at 60 °C, see Table 1). The simulation assumes a
constant initiator efficiency of 0.7.
The set of reactions depicted in Scheme 2 comprises
the complete RAFT preequilibrium and main equilib-
rium with individual rate coefficients. The preequilib-
riumswhere the initial RAFT agent is transformed into
macroRAFT agentsis governed by the two rate coef-
ficients for addition and fragmentation of kâ,1 and k-â,1
,
respectively. The main equilibrium is governed by the
addition/fragmentation rate coefficients kâ and k-â. The
scheme also comprises additional side reactions of the
intermediate macroRAFT radical 4, such as a potential
cross-termination reaction.
The black line in Figure 5 represents the resulting
log kti,i vs log i plots for k-â ) 1 × 105 s-1, whereby no
rate retardation is observed. Inspection of Figure 5
clearly indicates that, irrespective of the initial
The MCEPDA/MA system is relatively well charac-
terized with key parameters (including the propagation
rate coefficient, the rate coefficient for the initiator

2602 Theis et al.
Macromolecules, Vol. 38, No. 7, 2005
Both effectssslow fragmentation and cross-termin-
ationsresult in a severe overestimation of the chain
length dependent kt and give inconsistent (i.e., non-
matching) data for the different RAFT reagent concen-
trations. The experimental chain length dependencies
given below will be compared to this simulation output.
As mentioned in the Introduction, some attention
needs to be paid to the fact that the average value of
the termination rate coefficient is dependent on the
monomer to polymer conversion, at least for conversions
exceeding 15% in bulk MA polymerizations.16 However,
it is most likely that the conversion dependency is
predominantly a result of the increasing viscosity of the
system. In a RAFT process this issue is less problematic,
since the chain lengths remain relatively small even at
low conversions and thus increases in viscosity are less
pronounced. It has also been reported that the initiator
efficiency of AIBN in various monomers in bulk de-
creases with increasing monomer to polymer conver-
sion.19,20 It thus seems justified to use a decreasing
initiator efficiency vs conversion function, f(conversion),
rather than a constant f for the data evaluation. For
styrene, a strong decrease of the initiator efficiency at
high monomer to polymer conversions was reported,20
while the efficiency of AIBN in dodecyl acrylate poly-
merizations displayed a decrease preferentially in the
early stages of the polymerization.19 Since there are no
data regarding the conversion dependency of the initia-
tor efficiency of AIBN in MA bulk polymerization, a
linear decrease with conversion was implemented,
starting at a level of 0.7, steadily decreasing to zero at
100% conversion.
Nevertheless, it is clear that both the conversion
dependency of kt and f may cause some uncertainty,
especially at high monomer to polymer conversions. It
is for this reason that we limit the data evaluation to
40% monomer to polymer conversion, where the conver-
sion dependency of these parameters at the investigated
RAFT system will not play a dominant important role.
In addition, we do not include the initial data up to a
monomer to polymer conversion of 5% to avoid that the
results are affected by temperature equilibration issues
and trace oxygen amounts remaining in the DSC sample
cells.
Figure 6a depicts the resulting log kti,i vs log i plots
obtained by employing MCEPDA as RAFT agent. The
time dependent Rp data recorded for varying initial
MCEPDA concentrations were used to construct a log
kti,i vs log i plot in a stepwise fashion. The chain length
vs monomer to polymer conversion data depicted in
Figure 4a were used as a calibration for the x-axis in
the log kti,i vs log i plots.
It is gratifying to note that the stepwise construction
of the chain length dependency yield a continuously
decreasing function (the full line illustrates the best
fitting function in Figure 6a). If rate retardation effects
(as described above) were operational in the MCEPDA/
MA system, one would expect that the individual log
kti,i vs log i obtained from different RAFT agent con-
centrations are not correlated with each other; i.e., it
would be impossible to construct the chain length
dependency of kt in a stepwise fashion by accessing
different chain length regimes (see below). The fact that
a small variation of the results for different RAFT
reagent concentrations is observed further provides
some evidence that the initiator efficiency vs conversion
assumption issat least approximatelysvalid and that
Figure 5. Simulated log kti,i vs chain length, i, plots obtained
by analyzing the time dependent rate of polymerization data
by eq 1 for different RAFT reagent concentrations at 60 °C.
The graph depicts the expected log kti,i vs log i plots under the
assumption of slow fragmentation (preequilibrium and main
equilibrium) as well as cross-termination. The concentrations
and coefficients were selected analogous to the experimental
conditions given in Tables 1 and 2. Further, a chain length
dependent termination rate coefficient (kti,i ) 109 L mol-1 s-1
i-0.4) and the average experimental determined kâ for the
system MA/MCEPDA of 1.4 × 106 L mol-1 s-1 was used.
MCEPDA concentration, the given chain length de-
pendent termination rate coefficient kti,i is returned. The
green line results for a value of 1 × 102 s-1 for k-â
(causing considerable rate retardation), which prefer-
entially manifests itself at high initial RAFT agent
concentrations and in the initial period of each polym-
erization. It is evident that the overestimation of the
termination level results from an additional, albeit
reversible, radical loss pathway inducing non-steady-
state conditions.
The possibility of cross-termination reactions (see
reaction step VI in Scheme 2) between the intermediate
macroRAFT radicals has been extensively discussed in
the literature. Since the conception of the RAFT process,
various research groups have attempted to obtain
information on the mechanism and kinetics of the
process, either through the interpretation of dynamic
kinetic and molecular weight data via computer-based
modeling strategies,32-35 or through the indirect or
direct observation of intermediate species,36-38 or through
high-level ab initio molecular orbital calculations.39-41
The scientific debate has focused particular attention
on the fate of the so-called intermediate macroRAFT
radicals (species 2 and 4 in Scheme 2) that are formed
in the preequilibrium and main equilibrium of the RAFT
process.42 Evidence has been put forward for a relatively
long lifetime, τ, of the intermediate radicals in cases
where high stabilization of such radicals is possible (i.e.,
τ in the order of seconds)21,25,27,32-34,39,43 as well as for a
cross-termination reaction (either reversible43,44 or ir-
reversible27,28,36,37,45) of the intermediate radicals with
themselves or with the propagating chains. A recently
suggested number for ktcross is half the size of the
conventional termination rate coefficient.28 Rate retar-
dation as an effect of cross-termination (see red lines
in Figure 5) should appear preferentially at high initial
RAFT agent concentrations, but in contrast to slow
fragmentation in the later stages of the polymerization.
This effect can be attributed to the initially increasing
concentration of 4, leading to higher overall termination
levels.

Macromolecules, Vol. 38, No. 7, 2005
Termination Rate Coefficients of Methyl Acrylate 2603
This inclusion is, however, rather desirable because all
methyl acrylate polymerization processes will feature
such reactions, and considering only linear chain ter-
mination would portray a misleading picture. In addi-
tion, a separation of both pathways is experimentally
very challenging. In fact, in the past, chain length
dependent termination data in acrylate systems have
been interpreted arguing with the reactivity of midchain
radicals in termination reactions.6
It is mandatory to analyze the observed chain length
dependence of kt in the MA system with the frequently
used expression kti,i ) kt0i-R 31
In an earlier study,
.
Buback et al. have determined values for R in dodecyl
acrylate, methyl acrylate, and other acrylate and meth-
acrylate systems.6 These authors analyzed conversion
vs time traces obtained from SP-PLP experiments at
elevated pressures and 40 °C and found a strong
dependence of R on the size of the ester side chain for
acrylate polymerizations and, for some systems, on
monomer conversion. For MA, Buback et al. reported a
value of R ) 0.15 at low conversions and R increases
strongly during the course of the polymerization.6 The
present investigation, when the obtained log kti,i vs log
i data for varying initial MCEPDA concentrations are
averaged and subsequently analyzed, a value of 0.36 is
obtained for R (short dotted line in Figure 6a). In an
earlier study, the same authors reported a value of 0.32
for R, which was deduced in an indirect fashion via SP-
PLP experiments.1 When comparing our present num-
ber to the values obtained earlier by Buback et al., it
must be noted that the SP-PLP technique predomi-
nantly gives access to long-chain length regimes (i up
to 5000).6 The present value for R is valid for i between
5 and 800. Equation 9 gives the evolution of kt with
chain length (i.e., the bold full line in Figure 6a)
Figure 6. Experimentally obtained log kti,i vs chain length,
i, plots from MA bulk polymerization resulting from a stepwise
application of the RAFT CLD-T technique. The AIBN concen-
tration was 3.0 × 10-3 mol L-1. (a) Four different MCEPDA
concentrations were employed. The bold full line represents
the best-fit function, whereas the short dotted line depicts the
corresponding R value as a function of the chain length. (b)
The same experiment using DMCETC as RAFT reagent.
log(kti,i/L mol-1
s
)
-1 80°C ) 8.99 - 0.36 log i +
0.00542(log i)-2 (9)
the change of kt with the conversion is almost negligible
up to a monomer to polymer conversion of 40%. How-
ever, a closer inspection of the functions derived from
each RAFT concentration shows a slight S-like shape.
The initial higher values may be attributable to the time
to fully reach the steady-state radical concentration,
whereas the decrease in kt toward 40% conversion is
stronger at the lower RAFT reagent concentrations (i.e.,
at higher molecular weights and viscosities) and can be
assigned to the onset of the conversion dependence of
kt. In this context, it must be mentioned without fail
that these effects only result in a variation of less than
0.1 logarithmic units to the line of best fit, which is
considerably less than the statistical errors of other
methods used for the determination of kt. However, it
should be pointed out that this error only reflects the
variation caused by the RAFT-CLD-T method itself and
does not include systematic errors from other param-
eters (such as kp) and assumptions used to calculate the
log kti,i vs log i dependencies. Considering these uncer-
tainties, we estimate an overall error in log kti,i of 0.2
logarithmic units in absolute value.
The R value is also beset with an error, which is
influenced by the initiator efficiency functionality in the
individual chain length segments from which the overall
function (eq 9) is constructed. However, the effect on
the reported entire function is minimal, since this
overall result is computed from the series of individual
segments. For example, in the most extreme (albeit very
unrealistic) case where each segment would show no
chain length dependence (caused by the selection of an
unrealistic initiator efficiency), the next segment would
invariably be at a lower absolute value of kt and so on.
Still, a linear fit of such a series of “plateaus” would
yield a continuously decreasing function very similar
to eq 9. Clearly, this extreme case is an exaggeration
and does not to occur. We therefore estimate the overall
error in R to be close to 0.05.
Figure 6b depicts the resulting log kti,i vs log i plot
when trithiocarbonate DMCETC is employed as the
mediating agent. Inspection of Figure 6b clearly indi-
cates that the result is significantly different to that
observed in the corresponding MCEPDA system. The
apparent chain length dependency displays significantly
higher values for kt at short chain lengths that rapidly
decrease. It seems plausible that an increased stability
of the intermediate macroRAFT radicalsin possibly
both the preequilibrium and main equilibriumsinduced
by the stabilizing effect of the additional sulfur atom
provides an additional radical loss pathway. Such a
In the Introduction it was pointed out that acrylates
can undergo side reactions in radical polymerization,
e.g., radical transfers forming midchain radicals which
are capable of propagation. The extent of such reactions
is dependent on the polymerization temperature and
can reach a significant portion at 80 °C. It is evident
that the presented kti,i values may also include termina-
tion reactions of (potentially present) midchain radicals.

2604 Theis et al.
Macromolecules, Vol. 38, No. 7, 2005
notion becomes even more evident when the resulting
chain length dependencies are compared with those
expected theoretically for the scenario of slow fragmen-
tation in the preequilibrium and main equilibrium (see
Figure 5). It is therefore reasonable to assume that
DMCETC may not be very suitable in the context of the
RAFT-CLD-T method of MA. A similar observation can
be made when a RAFT agent is employed that uses an
even more stabilizing Z group, such as dithiobenzoates
(Z ) phenyl).
Recently there has been a lively discussion within the
scientific community on whether the propagation rate
coefficient, kp, is equally beset by a chain length
dependence.46 A potential chain length dependence of
kp may alter the outcome of the above analysis proce-
dures for small chain lengths.47 It is important to note
that any technique that maps out chain length depend-
ent kt data will require at some point propagation rate
data in the course of the evaluation. Thus, all techniques
(including SP-PLP) are affected by CLD kp data to some
extent. While there is some agreement that kp is in all
likelihood chain length dependent, there is significant
disagreement to what extend. Most studies regarding
the chain length have been carried out for styrene and
methyl methacrylate (MMA), with no report on acry-
lates. The most significant decrease of kp with chain
length should occur at small i (i.e., when going from 1
to 5 units). However, this most critical region is omitted
in our analysis. In addition, every conceivable chain
length dependency of kp can be evaluated with our
present data with relative ease, resulting in the (po-
tentially) modified kti,i functions.
RAFT agent. The assessment of kâ is possible because
the MA/MCEPDA and MA/DMCETC systems display
a significant degree of hybrid behavior between con-
ventional and living free radical polymerizationsat least
in the initial RAFT agent concentration range under
investigation.
Acknowledgment. The authors are grateful for
financial support from the Australian Research Council
(ARC) in the form of a Discovery Grant. T.P.D. acknowl-
edges an Australian Professorial Fellowship (ARC), and
A.F. acknowledge receipt of an International Linkage
Fellowship (ARC). The authors thank Dr. Philipp Vana
(Institute of Physical Chemistry, University of Go¨ttin-
gen) for his assistance with the DMCETC synthesis. The
authors also thank Dr. Leonie Barner and Istvan
Jacenyik for their excellent management of CAMD.
References and Notes
(1) Buback, M.; Busch, M.; Kowollik, C. Macromol. Theory Simul.
2000, 9, 442-452.
(2) Olaj, O. F.; Kornherr, A.; Zifferer, G. Macromol. Rapid
Commun. 1997, 18, 997-1007.
(3) Olaj, O. F.; Vana, P. Macromol. Rapid Commun. 1998, 19,
433-439. (b) Olaj, O. F.; Vana, P.; Kornherr, A.; Zifferer, G.
Macromol. Chem. Phys. 1999, 200, 2031-2039. (c) Olaj, O.
F.; Vana, P. Macromol. Rapid Commun. 1998, 19, 533-538.
(d) Olaj, O. F.; Vana, P. J. Polym. Sci., Polym. Chem. 2000,
38, 697-705.
(4) de Kock, J. B. L.; van Herk, A. M.; German, A. L. J.
Macromol. Sci., Polym. Rev. 2001, C41, 199-252.
(5) de Kock, J. B. L.; Klumperman, B.; van Herk, A. M.; German,
A. L. Macromolecules 1997, 30, 6743-6753.
(6) Buback, M.; Egorov, M.; Feldermann, A. Macromolecules
2004, 37, 1768-1776.
(7) Kowollik, C. Ph.D. Thesis, University of Go¨ttingen, 1999,
ISBN 3-89712-705-9.
Conclusions
(8) Vana, P.; Davis, T. P.; Barner-Kowollik, C. Macromol. Rapid
Commun. 2002, 23, 952-956.
In the present contribution we have demonstrated
that the reversible addition-fragmentation chain
transferschain length dependentstermination (RAFT-
CLD-T) methodology can be applied to study the chain
length dependence of the termination rate coefficient
in methyl acrylate (MA) bulk free radical polymeriza-
tions, provided the RAFT agent is judiciously chosen.
Methoxycarbonylethyl phenyldithioacetate (MCEPDA)
provides an effective RAFT agent for the application of
the RAFT-CLD-T method in acrylate systems. Its
particular advantage stems from the fact that its leaving
group is identical to that of the propagating MA radical.
Under such circumstances, the preequilibrium is ef-
fectively eliminated from the RAFT process. The extent
of the chain length dependence in the studied chain
length regime was quantified by the parameter R, which
is obtained by analyzing the experimental log kti,i vs log
i functionalities via the widely employed functionality
kti,i ) k0t i-R. R was found to be close to 0.36 in the chain
length region between 5 and 800. An alternative trithio-
carbonate RAFT agent, DMCETC, may not be as suited
for extracting reliable chain length dependent termina-
tion rate coefficients from methyl acrylate polymeriza-
tions. The termination rate coefficients in the DMCETC
system can be significantly overestimated, which we
tentatively attribute to a more stabilized intermediate
macroRAFT radical, which makes the polymerizations
process more susceptible toward rate retardation phe-
nomena.
(9) Mayadunne, R. T. A.; Rizzardo, E.; Chiefari, J.; Chong, Y.
K.; Moad, G.; Thang, S. H. Macromolecules 1999, 32, 6977-
6980. (b) Moad, G.; Chiefari, J.; Chong, Y. K.; Krstina, J.;
Mayadunne, R. T. A.; Postma, A.; Rizzardo, E.; Thang, S. H.
Polym. Int. 2000, 49, 993-1001. Chong, Y. K.; Le, T. P. T.;
Moad, G.; Rizzardo, E.; Thang, S. H. Macromolecules 1999,
32, 2071-2074. (c) Barner-Kowollik, C.; Davis, T. P.; Heuts,
J. P. A.; Stenzel, M. H.; Vana, P.; Whittaker, M. J. Polym.
Sci., Polym. Chem. 2002, 41, 365-375. (d) Destarac, M.;
Charmot, D.; Franck, X.; Zard, S. Z. Macromol. Rapid
Commun. 2000, 21, 1035-1039. (e) Taton, D.; Wilczewska,
A.-Z.; Destarac, M. Macromol. Rapid Commun. 2001, 22,
1497-1503.
(10) The statement that the radical concentration is unaltered in
RAFT systems is only true for chain length independent
termination kinetics. However, a nonretarding RAFT system
comes very close to the situation of an unchanged radical
population compared to the corresponding conventional po-
lymerization. An identical approximation is true for the rate
of polymerization.
(11) Feldermann, A.; Stenzel, M. H.; Davis, T. P.; Vana, P.;
Barner-Kowollik, C. Macromolecules 2004, 37, 2404-2410.
(12) Buback, M.; Kurz, C. H.; Schmaltz, C. Macromol. Chem. Phys.
1998, 199, 1721-1727.
(13) Beuermann, S.; Paquet, D. A., Jr.; McMinn, J. H.; Hutchin-
son, R. A. Macromolecules 1996, 29, 4206-4215.
(14) Buback, M.; Kowollik, C. Macromol. Chem. Phys. 1999, 200,
1764-1770.
(15) Beuermann, S.; Buback, M.; Schmaltz, C. Ind. Eng. Chem.
Res. 1999, 38, 3338-3344.
(16) Buback, M.; Kuelpmann, A.; Kurz, C. Macromol. Chem. Phys.
2002, 203, 1065-1070.
(17) Buback, M.; Barner-Kowollik, C.; Kuelpmann, A. Macromol.
Chem. Phys. 2002, 203, 1887-1894.
(18) Busch, M.; Mueller, M. Macromol. Symp. 2004, 206, 399-
418 and literature cited therein.
Concomitantly, the use of MCEPDA and DMCETC
as mediating agent in MA free radical polymerizations
allowed for the determination of the addition rate
coefficient, kâ, of propagating MA radicals to the initial
(19) Charton, N.; Feldermann, A.; Theis, A.; Davis, T. P.; Stenzel,
M. H.; Barner-Kowollik, C. J. Polym. Sci., Polym. Chem.
2004, 42, 5170-5179.

Macromolecules, Vol. 38, No. 7, 2005
Termination Rate Coefficients of Methyl Acrylate 2605
(20) Buback, M.; Huckestein, B.; Kuchta, F.-D.; Russell, G. T.;
Schmid, E. Macromol. Chem. Phys. 1994, 195, 2117-2140.
(21) Barner-Kowollik, C.; Quinn, J. F.; Nguyen, T. L. U.; Heuts,
J. P. A.; Davis, T. P. Macromolecules 2001, 34, 7849-7857.
(22) Tamami, B.; Kiasat, A. R. J. Chem. Res., Synop. 1998, 454-
455.
(35) Zhang, M.; Ray, W. H. Ind. Eng. Chem. Res. 2001, 40, 4336-
4352.
(36) Monteiro, M. J.; de Brouwer, H. Macromolecules 2001, 34,
349-352.
(37) Kwak, Y.; Goto, A.; Tsujii, Y.; Murata, Y.; Komatsu, K.;
Fukuda, T. Macromolecules 2002, 35, 3026-3029.
(38) Ah Toy, A.; Vana, P.; Davis, T. P.; Barner-Kowollik, C.
Macromolecules 2004, 37, 744-751.
(39) Coote, M. L.; Radom, L. J. Am. Chem. Soc. 2003, 125, 1490-
1491.
(23) Strazielle, C.; Benoit, H.; Vogl, O. Eur. Polym. J. 1978, 14,
331-334.
(24) Madorsky, S. L.; Hart, V. E.; Strauss, S. J. Res. Natl. Bur.
Stand. 1956, 56, 343-349.
(25) Feldermann, A.; Coote, M. L.; Stenzel, M. H.; Davis, T. P.;
Barner-Kowollik, C. J. Am. Chem. Soc. 2004, 126, 15915-
15923.
(40) Coote, M. L. Macromolecules 2004, 37, 5023-5031.
(41) Coote, M. L.; Radom, L. Macromolecules 2004, 37, 590-596.
(42) Barner-Kowollik, C.; Coote, M. L.; Davis, T. P.; Radom, L.;
Vana, P. J. Polym. Sci., Polym. Chem. 2003, 41, 2828-2832.
Wang, A. R.; Zhu, S.; Kwak, Y.; Goto, A.; Fukuda, T.;
Monteiro, M. S. J. Polym. Sci., Polym. Chem. 2003, 41, 2833-
2839.
(26) Perrier, S.; Barner-Kowollik, C.; Quinn, J. F.; Vana, P.; Davis,
T. P. Macromolecules 2002, 35, 8300-8306.
(27) Chernikova, E.; Morozov, A.; Leonova, E.; Garina, E.; Gol-
ubev, V.; Bui, C.; Charleux, B. Macromolecules 2004, 37,
6329-6339.
(43) Vana, P.; Quinn, J. F.; Davis, T. P.; Barner-Kowollik, C. Aust.
J. Chem. 2002, 55, 425-431.
(28) Kwak, Y.; Goto, A.; Fukuda, T. Macromolecules 2004, 37,
1219-1225.
(44) Monteiro, M. J.; Bussels, R.; Beuermann, S.; Buback, M. Aust.
J. Chem. 2002, 55, 433-437.
(29) Vana, P.; Davis, T. P.; Barner-Kowollik, C. Macromol. Theory
Simul. 2002, 11, 823-835.
(45) Kwak, Y.; Goto, A.; Komatsu, K.; Sugiura, Y.; Fukuda, T.
Macromolecules 2004, 37, 4434-4440.
(30) Wulkow, M.; Busch, M.; Davis, T. P.; Barner-Kowollik, C. J.
Polym. Sci., Polym. Chem. 2004, 42, 1441-1448.
(31) Barner-Kowollik, C.; Vana, P.; Davis, T. P. In Handbook of
Radical Polymerization; Matyjaszewski, K., Davis, T. P., Eds.;
Wiley and Sons: New York, 2002; p 209 and literature cited
therein.
(46) Willemse, R. X. E.; Staal, B. B. P.; van Herk, A. M.; Pierik,
S. C. J.; Klumperman, B. Macromolecules 2003, 36, 9797-
9803. (b) Olaj, O. F.; Vana, P.; Zoder, M.; Kornherr, A.;
Zifferer, G. Macromol. Rapid Commun. 2000, 21, 913-920.
(c) Olaj, O. F.; Vana, P.; Zoder, M. Macromolecules 2002, 35,
1208-1214. (d) Zetterlund, P. B.; Busfield, W. K.; Jenkins,
I. D. Macromolecules 2002, 35, 7232-7237.
(32) Barner-Kowollik, C.; Quinn, J. F.; Morsley, D. R.; Davis, T.
P. J. Polym. Sci., Polym. Chem. 2001, 39, 1353-1365.
(33) Vana, P.; Davis, T. P.; Barner-Kowollik, C. Macromol. Theory
Simul. 2002, 11, 823-835.
(34) Perrier, S.; Barner-Kowollik, C.; Quinn, J. F.; Vana, P.; Davis,
T. P. Macromolecules 2002, 35, 8300-8306.
(47) Smith, G. B.; Russell, G. T.; Yin, M.; Heuts, J. P. A. Eur.
Polym. J. 2005, 41, 225-230.
MA047476D