a semi-batch process for nitroxide mediated radical polymerization
TRANSCRIPT
![Page 1: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/1.jpg)
A Semi-Batch Process for Nitroxide Mediated
Radical Polymerization
Yanxiang Wang, Robin A. Hutchinson,* Michael F. Cunningham*
Department of Chemical Engineering, Queen’s University, Kingston, Ontario K7L 3N6, CanadaE-mail: [email protected]
Received: September 28, 2004; Revised: January 26, 2005; Accepted: January 28, 2005; DOI: 10.1002/mame.200400273
Keywords: butyl acrylate; coatings; living; nitroxide mediated polymerization; semi-batch; stable free radical polymerization;styrene
Introduction
In order to improve the performance of polymer materials
and their applications, synthesis of polymers with narrow
molecular weight distribution and controlled composition
has long been a goal of polymer researchers. As recent as a
decade ago, low polydispersity polymers and well-defined
copolymers could only be produced by ionic polymeriza-
tion (anionic and cationic). However, the discovery of living
(controlled) free radical (LRP) polymerization has changed
this situation. There are three important LRP types; stable
free radical polymerization (SFRP) or nitroxide mediated
radical polymerization,[1] atom transfer radical polymer-
ization (ATRP),[2] and reversible addition fragmentation
transfer process (RAFT).[3] Living radical polymerization
is based on an alternating activation/deactivation process of
Summary: A semi-batch process using nitroxide mediatedpolymerization, was explored for the design of low molecularweight solvent-borne coatings, typical of those used in theautomotive industry. While living radical polymerization(LRP) offers many advantages in the control of polymer chainmicrostructure that may confer important physical and chem-ical property benefits to coatings, adapting LRP to a semi-batch process poses significant challenges in the design andoperation of the process. Using styrene monomer, varioustwo-component initiating systems (free radical initiator,4-hydroxy-TEMPO) were studied to understand the effects ofdifferent initiators on the course of polymerization. In addi-tion, an alkoxyamine was synthesized and used as theinitiating source. The initiators Luperox 7M75 and Luperox
231 give higher polymerization rates and reasonable controlover polymerization, while benzoyl peroxide (BPO), Vazo67, and the alkoxyamine are less effective. The number ofpolymer chains in the final product is always less than thetheoretical value, reflecting poor initiation efficiency, prob-ably resulting from undesirable termination reactions thatbecome important due to the nature of the semi-batchprocess. Adding camphorsulfonic acid (CSA) or charginginitiator concurrently with monomer during semi-batch feed,can increase the polymerization rate while maintaining theliving character of the polymerization. The copolymerizationof styrene and butyl acrylate is also shown to exhibit livingcharacter.
Schematic representation of the exchange reaction to produce N-TEMPO cappedpolymer chains.
Macromol. Mater. Eng. 2005, 290, 230–241 DOI: 10.1002/mame.200400273 � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
230 Full Paper
![Page 2: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/2.jpg)
the propagating chains. Details of LRP kinetics and mech-
anisms can be found elsewhere.[4,5] Recent years have
witnessed numerous publications on the synthesis of low
polydispersity polymers and polymers with various struc-
tures (block, brush, star, etc.,). As this technique, by its
nature, is a radical process it has much stronger tolerance to
functional groups, water, and protic media than ionic living
polymerizations.
Due to the versatility and synthetic ease of living radical
polymerization, it has attracted enormous attention for its
potential applications in industry. In the past twenty years,
the paint and coating industry has faced regulatory pressure
to reduce the amount of volatile organic compound (VOC)
in the products, and even stricter regulations on VOC are
anticipated in the future.[6] This has acted as an impetus to
develop new environmentally compliant products such as
water-borne coatings and powder coatings, to resolve the
problems concerning VOC emissions. However, automo-
tive coatings, which are still primarily solvent-based mate-
rials produced by solution polymerization, cannot be easily
replaced by other techniques because of the advantageous
product properties of solvent-borne materials. Past appro-
aches to decrease the VOC content of automotive coatings
include, reducing the molecular weight of polymers (to
2 000 Da) and increasing the solid content from about
20 wt.-% in older formulations to 70 wt.-% in current
formulations.[7]
In this paper, the 4-hydroxy-2,2,6,6,-tetramethylpiper-
idinyloxy) (4-hydroxy-TEMPO) mediated SFRP of styrene
and styrene-co-butyl acrylate, was explored using a semi-
batch process. This approach is used industrially because it
permits easier control of temperature and superior control
of copolymer composition. In a batch conventional radical
copolymerization, if the consumption rates of two mono-
mers are considerably different, the copolymer composition
will change with reaction time. Using a semi-batch process,
copolymers with a more homogeneous copolymer compo-
sition distribution can be produced. In a living radical
polymerization, there is another potentially important ad-
vantage in chain microstructure that may be realized. LRP
allows variation of the copolymer composition within the
same chain, thereby allowing polymer chains with tapered
gradients in composition to be produced. This could be
advantageous in product design as it would allow desired
monomers (e.g., functional monomers used for subsequent
cross-linking) to be placed at specific locations along the
chain. This study is one of the first reports of semi-batch
living radical polymerization.
Experimental Part
Materials
The alkoxyamine PE-T [1-(2,2,6,6-tetramethylpiperidiny-loxy)-1-phenylethane] and the nitroxide N-TEMPO [4-(1-
naphthoyloxy)-2,2,6,6-tetramethylpiperidine-1-oxyl] weresynthesized in our laboratory according to literature proce-dures.[8] 4-Hydroxy-TEMPO (Aldrich Chemical), BPO (97%,Aldrich Chemical), Vazo 67 (2,20-azo(2-methylbutyronitrile)(DuPont), Luperox 231 [1,1-bis(tert-butylperoxy)-3,3,5-tri-methylcyclohexane 92%] (Lupersol), Luperox 7M75 (tert-butyl peroxyacetate 75 wt.-% solution in aliphatic hydro-carbons) (Lupersol), styrene (Aldrich Chemical), butylacrylate (Aldrich Chemical), and xylenes (Aldrich Chemical)were used without further purification.
Polymerization Procedure
The polymerizations were conducted in a 1 L automaticlaboratory reactor (METTLER TOLEDO LabMaxTM) con-trolled by CAMILE TG software (Camille Products). For theconventional free radical polymerization experiments, 210 gxylenes were first added to the reactor. After the temperaturestabilized at 138 8C, a mix of 496 and 13.8 g of styrene andLuperox 7M75, respectively was fed to the reactor in semi-batch mode at a constant rate over 6 h, and then the solution washeld at this temperature for another 30 min. The thermallyinitiated polymerizations used the same recipe except thatno initiator was added. Samples for analysis were collect-ed from the bottom valve of the reactor at specified times tobottles containing ca. 1–1.5% inhibitor (4-methoxyphenol)solution.
For semi-batch SFRPs using the alkoxyamine (unimer) PE-T as initiator, 27.8, 28, 179 g of PE-T, styrene, and xylenes,respectively were charged to the reactor and the temperaturewas increased to 138 8C. 397 g styrene was then fed in semi-batch mode over 6 h. The other semi-batch experiments usingvarious two-component initiating systems were based on thefollowing basic recipe, which is comprised of three stages(Figure 1). In Stage I, 210 g xylenes was charged into the reac-tor at room temperature and the temperature increased to138 8C. 21.6 g of 4-hydroxy-TEMPO, 15.8 g of Luperox 7M75,and 34 g of styrene (mole ratio of 4-hydroxy-TEMPO/Luperox7M75¼ 1.39) were charged in sequence and held for 1 h. InStage II, 467 g styrene was semi-batch fed over 6 h. In Stage III,the reaction mixture was held at this temperature to increase themonomer conversion. The same 4-hydroxy-TEMPO/initiatormolar ratio of 1.39 was consistently used with all initiators. Inthe solution batch polymerizations, 141 g of xylenes, 334 g ofstyrene, 14.4 g of 4-hydroxy-TEMPO and Luperox 7M75,were charged concurrently to the reactor and then the tem-perature was increased to 138 8C. In experiments varying theLuperox 7M75/4-hydroxy-TEMPO ratio, the amounts of Lupe-rox 7M75 were 10.6, 9.3, and 7.9 g, respectively, which gavethe corresponding mole ratios of nitroxide/initiator of 1.39,1.59, and 1.87. A summary of experiments is shown in Table 1.
Livingness Analysis
In the exchange reaction between alkoxyamine and N-TEMPO,0.11 g dry polymer, 0.66 g N-TEMPO, and 1.98 mL chloro-benzene were charged in a 5 mL flask. After the solution wasdegassed by three freeze-pump-thaw cycles, the flask wasimmersed in oil of about 123 8C. The exchange reaction
A Semi-Batch Process for Nitroxide Mediated Radical Polymerization 231
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
![Page 3: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/3.jpg)
lasted 154 min, with stirring, under a nitrogen blanket. Theexchanged products were dried before further analysis.[9]
Characterization
The concentration of residual monomers was determined byusing a Varian CP-3800 gas chromatograph installed with aModel 8410 autosampler and a flame ionization detector set at250 8C. A 30M Chrompack Capillary Column (CP-Sil 8 CB)was used for the separation and the injector temperature washeld at 200 8C. The samples were diluted in acetone and exter-nal calibration was made before the measurements.
For gel permeation chromatography (GPC) measurements,the dry samples were dissolved in THF, which was also used aseluant at a flow rate of 1 mL �min�1. Before injection, thesolutions were filtered through a Chromspec syringe filter(25 mm nylon, 0.2 mm non-Sterile). The molecular weightmeasurements were performed using a Waters 2690 Separa-tions Module equipped with Waters Styragel HR columns(HR0.5, HR1, HR3, HR4, HR5) in THF at 35 8C and a Waters410 differential refractometer. The molecular weight valueswere reported relative to polystyrene standards using Millen-ium software. For the measurement of polymer livingness, theWaters 410 differential refractometer was operated in series
Figure 1. Schematic representation of the procedure of semi-batch 4-hydroxy-2,2,6,6,-tetramethylpiperidinyloxy (4-hydroxy-TEMPO) mediated radical polymerization usingbimolecular initiators.
Table 1. Summary of experiments. T¼ 138 8C for all runs. Complete details in the Experimental Part.
Experiment ID Monomer(s) Initiator OH-TEMPO:I(molar ratio)
Operating mode
1 (Conventional) Styrene Luperox 7M75 0 Semi-batch2 Styrene Luperox 231 1.39 Semi-batch3 Styrene Luperox 7M75 1.39 Semi-batch4 Styrene benzoyl peroxide (BPO) 1.39 Semi-batch5 Styrene Vazo 67 1.39 Semi-batch6 Styrene [1-(2,2,6,6-tetramethylpiperidinyloxy)-
1-phenylethane] (PE-T)– Semi-batch
7 (Thermal) Styrene None – Semi-batchModified 1a) Styrene Luperox 7M75 1.39 Semi-batchModified 2b) Styrene Luperox 7M75 1.39 Semi-batchModified 3 Styrene Luperox 7M75 1.39 Semi-batch8 Styrene Luperox 7M75 1.87 Batch9 Styrene Luperox 7M75 1.59 Batch10 Styrene Luperox 7M75 1.39 Batch11 Styrene/butyl acrylate
(50/50 mass ratio)Luperox 7M75 1.39 Semi-batch
a) 9.8 g camphorsulfonic acid (CSA) added in the beginning of Stage II.b) Additional 4.1 g Luperox 7M75 added concurrently with styrene during the semi-batch feed.
232 Y. Wang, R. A. Hutchinson, M. F. Cunningham
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
![Page 4: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/4.jpg)
with a Waters 474 scanning fluorescence detector, which usedan excitation wavelength of 280 nm and an emission wave-length of 355 nm. The responses of RI and fluorescencedetectors are related to sample mass and number of alkoxy-amine chains, respectively, and a calibration was made beforethe calculation of livingness.[9]
Results and Discussion
Although alkoxyamine initiators offer greater control of the
number of chains in an LRP, we conducted most of our
experiments using a two-component initiating system, as
this is the approach most likely to be used by the industry. In
Stage I (Figure 1), the initiator thermally decomposes to
give primary radicals, which can undergo a variety of side
reactions in addition to the intended propagation reaction
with monomer.[10] These side reactions include transfer to
solvent and monomer by hydrogen abstraction, termination
with other radicals, and decomposition to smaller radicals
and fragments. It may also be possible that reactive initia-
tors such as peroxides will react directly with nitroxides.
While nitroxides readily react with carbon-centered radi-
cals, they do not easily react with oxygen-centered radicals
such as those derived from peroxides. Therefore, Stage I
may be a complex radical process and, depending on the
initiator used, various species can be formed. A noteworthy
and potentially problematic feature of Stage I is the
unusually high initiator concentrations required. In LRP,
all the chains are created at the outset of polymerization,
while in the analogous semi-batch FRP process, chains are
created continuously throughout the process. Furthermore,
the low desired molecular weight in the final product gives a
high number of chains. Consequently, the initial initiator
and nitroxide concentrations in semi-batch SFRP are an
order of magnitude higher than those in batch SFRP. The
implications of this issue will become apparent later in
Results and Discussion.
In Stage II, the propagation of the chains takes place
during the reversible activation/deactivation process as
shown in Figure 2. Successful control over the polymeriza-
tion depends largely on an appropriate equilibrium between
the activation and deactivation process. Even though K is
very small, both kact and kdeact should be large enough to
provide a reasonable polymerization rate while still
providing good control. Bimolecular termination between
propagating radicals also occurs, albeit at a much lower rate
than in conventional free radical polymerization. At the
beginning of the reaction, the nitroxide and propagating
radical concentrations increases due to the reversible homo-
lytic cleavage of the newly initiated chains. After a short
time (less than a few minutes), the system enters a quasi-
stationary state, where the nitroxide and propagating radical
concentrations change slowly with reaction time. The time
required to achieve this state depends on a number of
factors, including the initiation rate, temperature, and the
activation/deactivation equilibrium constant. Simulation
and experimental results have shown other reactions exist in
this system,[11–14] including styrene thermal initiation, and
disproportionation of dormant chains to yield hydroxyla-
mine and terminally unsaturated (dead) polymer. Two
routes are shown in Figure 2 for hydroxylamine formation,
but the bimolecular is likely to be the dominant route. The
propagating radical can also abstract hydrogen from hydro-
xylamine to regenerate nitroxide. These side reactions are
the main factors contributing to the loss of livingness
and the increase of polymer polydispersity at high
conversion.[11]
In this paper, we explored five different initiating systems
in semi-batch nitroxide mediated styrene polymerization.
The bimolecular initiators were 4-hydroxy-TEMPO com-
bined with peroxides (Luperox 7M75, Luperox 231, BPO),
an azonitrile (Vazo 67), and a unimolecular alkoxyamine
(PE-T). The alkoxyamine is the simplest system because
many of the complex reactions related to initiator decom-
position in Stage I are avoided.
Figure 3 shows experimental monomer concentration
profiles. For comparison, besides the nitroxide mediated
runs, three additional runs are shown. These are: (1) semi-
batch conventional free radical polymerization (no nitr-
oxide); (2) thermally initiated polymerization; and (3) the
monomer concentration profile without polymerization.
For the recipes using the initiators BPO and Vazo 67, or the
alkoxyamine PE-T, the polymerization rate was very slow
and no conversion was observed in the semi-batch feed
stage. For recipes using Luperox 7M75 and Luperox 231,
some conversion was observed during the semi-batch feed
stage, with higher conversion observed for the recipe using
Luperox 231. In all cases, the polymerization rate is lower
than that of conventional radical polymerization. The diffe-
rent polymerization rate is also reflected from the temper-
atures of the reaction mass and reactor jacket. In semi-batch
conventional radical polymerization, the temperature of the
reaction mass was maintained at 138 8C but the jacket
temperature was about 2 8C lower during the semi-batch
feed due to the exothermic nature of polymerization. This
temperature difference became gradually smaller in the
30 min holding time, because the polymerization was
close to completion and hence monomer concentrations
were low. In the nitroxide mediated polymerization using
Luperox 7M75 as initiator, after the addition of 4-hydroxy-
TEMPO, styrene and initiator at 138 8C, the color of the
reaction mass changed gradually from reddish to light
brown during the first hour (4-hydroxy-TEMPO gives a
reddish solution). During the addition of styrene in the
following semi-batch feed, the jacket temperature was
always slightly higher than the reactor temperature, which
indicated that the polymerization is slow, and less heat was
evolved.
Figure 4 shows the number average molecular weight
(Mn) and polydispersity profiles. (The runs using Vazo 67
A Semi-Batch Process for Nitroxide Mediated Radical Polymerization 233
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
![Page 5: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/5.jpg)
and BPO are not shown as their rates were negligibly small.)
Compared to conventional free radical polymerization,
polydispersities were significantly lower, and molecular
weight increased approximately linearly with conversion.
These two features indicate the polymerization proceeded
in a living manner under reasonably good control. In an
ideal LRP, Mn is equal to the ratio of consumed monomer
divided by the number of chains, and the molecular weight
should increase linearly with conversion. The deviation
from linearity, especially at high conversion, indicates that
the polymerization was affected by side reactions. The
nature of the curvature, which is concave down, shows that
the number of chains increases as the polymerization pro-
gresses, while in the ideal case it should remain constant.
This is attributable to the presence of thermal initiation.
Another important consideration is how the experimentally
observed number of chains compares to the theoretically
expected number. Table 2 shows the number of polymer
chains in the final products, calculated using experimental
conversion and molecular weight. Theoretical values were
calculated, based on the moles of nitroxide added. These
values were generally low, typically in the range of 0.3–0.5.
The characteristics of semi-batch nitroxide mediated
polymerization using the five different initiators can be
interpreted as follows. First, the recipes using Luperox
7M75 and Luperox 231 as initiators give a reasonable
polymerization rate, while the number of polymer chains is
much lower than the number of nitroxide molecules added
in the start-up stage. The low number of polymer chains can
be caused by low initiator efficiency or the formation of
inactive alkoxyamine in the start-up stage. For initiator
Luperox 7M75, it is believed that decomposition occurs by
one-bond homolysis to an acyloxy and an alkoxy radical[15]
as shown in Figure 5. These radicals cannot be directly
capped (deactivated) by 4-hydroxy-TEMPO as they are
oxygen-centered. Possible reactions for these two radical
Figure 2. Reaction mechanisms for semi-batch 4-hydroxy-TEMPO mediated radicalpolymerization.
234 Y. Wang, R. A. Hutchinson, M. F. Cunningham
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
![Page 6: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/6.jpg)
species are: (1) propagation with styrene (desired); (2)
hydrogen abstraction from xylene (transfer to solvent); and
(3) decomposition to methyl radicals. Luperox 231 decom-
poses, to give two tert-butoxy radicals, which follow the
processes described for Luperox 7M75. If a methyl radical
is capped by 4-hydroxy-TEMPO, it will probably be too
stable to contribute to the polymerization since the equili-
brium constant K is too small.[16]
The recipes using the initiators Vazo 67, or BPO, or the
alkoxyamine PE-T give little conversion in the semi-batch
feed stage and during the holding period in comparison to
the other runs, suggesting high residual nitroxide levels.
The initiator efficiencies of these runs (Table 2) are compa-
rable to the Luperox 231 and Luperox 7M75 runs. Similar
number of chains are thus formed for all initiators, yet with
Vazo 67, BPO, and PE-T the rates remain distinctly lower.
Therefore, the lower rates with these three initiators com-
pared to the Luperox 231 and Luperox 7M75 cannot be
explained solely by lower initiator efficiencies. It is true that
within the series of runs using Vazo 67, BPO, and PE-T,
higher initiator efficiencies (PE-T>Vazo 67>BPO) cor-
respond to lower rates, which is consistent with higher
initial nitroxide levels. The high nitroxide concentrations in
these recipes could be a result of bimolecular termination,
because of the very high initial concentration of initiator or
alkoxyamine (10� higher than is commonly used in batch
polymerizations) and low concentration of monomer. Ter-
mination leads to high amount of nitroxide being built-up in
the reactor at the early stage of polymerization, with the
consequent suppression of the polymerization rate [Equa-
tion (1) and (2)].
K ¼ kact
kdact
¼ ½P�n�½T��
½Pn-T�ð1Þ
Rp ¼ kp½P��½M� ¼ Kkp½M�½Pn-T�½T�� ð2Þ
In the recipes with Vazo 67, BPO, and PE-T, it is probable
that the high concentration of nitroxide built-up in the early
stages as a result of bimolecular termination, causes the rate
to be low for the duration of the polymerization. We ob-
served that during these runs, the solution was initially
colorless and a light red color appeared shortly after the
temperature was increased to 138 8C, which indicates the
accumulation of free nitroxide. Only limited polymeriza-
tion occurred after the semi-batch feed stage. Strong
evidence for the role of termination, early in the polymeri-
zation, is the low number of chains observed when the
alkoxyamine PE-T was used. In this case, we are beginning
with a known number of chains, and previous experience
with alkoxyamines in batch systems suggests that there
Figure 3. Monomer concentration profiles for semi-batch nitr-oxide mediated styrene polymerization using different initiationsystems. Conventional radical polymerization (*), Luperox 231/4-hydroxy-TEMPO (~), Luperox 7M75/4-hydroxy-TEMPO(^), benzoyl peroxide/4-hydroxy-TEMPO ( ), Vazo 67/4-hydroxy-TEMPO ( ), unimer PE-T (&), thermal initiation( ), without polymerization (�).
Figure 4. Number average molecular weight (solid symbols)and polydispersity index (open symbols) profiles for semi-batchnitroxide mediated styrene polymerization using different initia-tion systems. Conventional radical polymerization (*),Luperox1 231/4-hydroxy-TEMPO (~), Luperox1 7M75/4-hydroxy-TEMPO (^), unimer PE-T (&).
Table 2. Experimental polymer chain number in the final product and theoretical value for the recipes using five different initiationsystems.
Formula Luperox 7M75/4-hydroxy-2,2,6,6,-tetramethylpiperidi-
nyloxy) (4-hydroxy-TEMPO)
Luperox 231/4-hydroxy-TEMPO
Vazo 67/4-hydroxy-TEMPO
BPO/4-hydroxy-TEMPO
PE-T
Experimental/theoretical(mol/mol)
0.04/0.125 0.04/0.107 0.05/0.107 0.03/0.125 0.06/0.106
A Semi-Batch Process for Nitroxide Mediated Radical Polymerization 235
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
![Page 7: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/7.jpg)
should be no more than 10% deviation from the expected
chain number. That so many chains could be ‘‘lost’’ with
PE-T, can only be reasonably explained by high initial
termination rates, caused by the high PE-T concentration. A
recent report by Georges et al.[17] offers an explanation for
the BPO results. At low temperatures, promoted initiator
dissociation of the BPO, causes the loss of a significant
portion fraction of the nitroxide. At the unusually high
initiator and nitroxide concentrations used in this study (a
necessary condition of operating in semi-batch mode), this
problem is even more severe. For the other initiator systems,
we believe that high rates of primary radical termination
lead to an excess of free nitroxide in the system, and
consequently a prolonged induction period.
In nitroxide mediated styrene polymerizations, thermal
initiation plays an important role. The polymerization rate
is often seen to be independent on the concentration of
alkoxyamine and instead depends on the thermal initiation
rate.[11,12] In Figure 3, it can be seen that the polymerization
rate of the run recipe using Luperox 7M75/4-hydroxy-
TEMPO is comparable to the corresponding thermally
initiated run, consistent with literature findings for batch
systems.[11,12] It is also common for nitroxide mediated
polymerizations to be slower than conventional thermal
polymerization, if there is excess nitroxide present which
acts to suppress the active radical concentration (and
therefore the rate) by shifting the equilibrium toward the
dormant state. This can happen with two-component initi-
ating systems if initiation efficiency is low, resulting in an
initial excess of nitroxide. However, the recipe using Lupe-
rox 231/4-hydroxy-TEMPO gives a much higher polymer-
ization rate. That the rate is higher for the run using Luperox
231 indicates the active radical concentration, which is
higher for this run than even for the thermally initiated run
(without nitroxide). This observation implies that there is a
deficiency of nitroxide after the initiation stage. It is un-
clear, why there should be such a deficiency, given that the
nitroxide/initiator ratio used (1.39), which is typical to give
optimal results. For example, if the initiator has an effi-
ciency of �0.7, we have added just enough nitroxide to cap
(deactivate) the newly formed chains. One possible expla-
nation is that the efficiency for the Luperox 231 is much
greater that �0.7, thereby leading to a nitroxide deficiency.
However, examination of the number of chains (Table 2)
indicates that in fact the efficiency is lower with the Luperox
231 compared to Luperox 7M75, disqualifying that expla-
nation. Therefore an unexplained question is why the
Luperox 231 and Luperox 7M75 runs do progress reason-
ably well. Their initiation efficiencies are within the range
seen for the Vazo 67, BPO, and PE-T runs, and there-
fore they should have comparable residual nitroxide levels.
The higher rates observed with both Luperox initiators may
be caused by consumption of nitroxide by side reactions
involving the initiators. The chemistry of peroxide initiators
is complex, and they are known to be very reactive. Direct
reaction between the Luperox initiators and the hydroxyl
group of 4-hydroxy-TEMPO, would explain the faster than
expected rates with these two initiators. Despite the loss of
nitroxide, however, the polydispersity remains low (com-
parable to Luperox 7M75), and a well-controlled poly-
merization is maintained.
Based on the above results, we used Luperox 7M75 as the
initiator of choice for further experimental work, as it most
closely followed the expected rate (equal to the thermal
Figure 5. Reaction mechanisms in the start-up stage with Luperox 7M75/4-hydroxy-TEMPO initiation.
236 Y. Wang, R. A. Hutchinson, M. F. Cunningham
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
![Page 8: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/8.jpg)
initiation rate). Luperox 231 would have also been a suit-
able choice, although the unexplained high initial rate and
the lower initiation efficiency led us to choose the Luperox
7M75. As previously discussed (Figure 4), a characteristic
of these semi-batch living radical polymerizations is that
the polydispersity is low at the beginning of reaction and
increases with conversion. This is different from the batch
LRP experiments where the polydispersity is higher at the
beginning of polymerization and decreases with conver-
sion. This can be interpreted by the polymerization degree
(run length) obtained by a propagating radical each time it is
activated as shown in Equation (3),
Run length per activation cycle ¼ Rp
Rdeact
¼ kp½M�½P��kdeact½T��½P��
¼ kp½M�kdeact½T�� ð3Þ
which is the ratio of its propagation and deactivation rate.
This equation represents the mean polymerization degree
increase of a chain for each activation/deactivation cycle,
and is related to the monomer and nitroxide concentrations.
In a semi-batch solution polymerization, the monomer
concentration is much lower than that in a batch bulk poly-
merization, and therefore the increase in polymerization
degree, for each activation cycle is smaller than that for the
batch case. Consequently, the polydispersity in semi-batch
solution polymerization is low at the beginning. The
observed increase in polydispersity at higher conversions
is caused by the increasing number of chains due to thermal
initiation and the decrease in nitroxide concentration due to
dilution.
Effect of Using Modified Recipes
A high polymerization rate is mandatory for the application
of LRP in industrial applications. In order to overcome the
low polymerization rate of the basic recipe using 4-
hydroxy-TEMPO/Luperox 7M75, process modifications
were made. In ‘‘modified recipe I,’’ about 9.8 g camphor-
sulfonic acid (CSA),[18] was added in the beginning of Stage
II. CSA is known to enhance the rate of TEMPO mediated
polymerizations, primarily by reducing the free nitroxide
concentration. The color of the reaction mass changed
immediately from light brown to deep black after the addi-
tion of CSA. In ‘‘modified recipe II,’’ an additional 4.1 g
Luperox 7M75 was added concurrently with styrene during
the semi-batch feed. The additional initiator feed was
intended to reduce the nitroxide concentration by generat-
ing additional radicals at a controlled rate. Because the
initiator half-life is short at the reaction temperature, the
Luperox 7M75 decomposition rate is approximately equal
to its addition rate. Finally, in order to improve the effi-
ciency of 4-hydroxy-TEMPO (or increase the number of
polymer chains in the final product), in ‘‘modified recipe
III,’’ we designed a new charging procedure for the initia-
tors in Stage I designed to maintain a low 4-hydroxy-
TEMPO concentration. In this recipe, during Stage I, all
xylenes were first charged to the 1 L reactor and, after
increasing the temperature to 138 8C, a mixture of 4-
hydroxy-TEMPO, Luperox 7M75, and 107 g styrene was
fed by semi-batch over 80 min and then the solution was
held for another 10 min. 397 g styrene was charged by semi-
batch feed over 5 h in Stage II. The differences with the
original procedure are; (1) the 4-hydroxy-TEMPO,
Luperox 7M75, and styrene mixture was charged slowly
(over 80 min) in the new process, and quickly in the original
process; and (2) 107 g styrene was used in the new process
versus 34 g in the original process. These modifications are
intended to promote the formation of TEMPO-capped
styrene oligomers (increased styrene charge) during Stage I
while reducing the occurrence of side reactions that
consume initiator (slower initiator addition rate).
Figure 6 shows the free monomer concentration profiles
of the basic and modified recipes. For comparison, the
conventional free radical polymerization (bottom) and the
semi-batch process without polymerization (top) are also
shown. Addition of CSA or feeding Luperox 7M75 concur-
rently with monomer improves the polymerization rate
considerably, with the initiator addition having a more
pronounced effect. The unreacted monomer concentration
profile of modified recipe II (Luperox 7M75 addition)
appears similar to the conventional free radical polymer-
ization but with a higher free monomer concentration. In the
run using modified recipe I (CSA addition), the free
monomer concentration curve displays a differently shaped
profile, rising to 300 mg � g�1 and then staying at that level
for approximately 300 min before beginning to decline.
These differences between adding CSA and Luperox 7M75
reflect their differing roles. Luperox 7M75 initiates new
Figure 6. Monomer concentration profiles for the modifiedrecipes. Conventional radical polymerization (*), basic recipe( ),modified I (^), modified II (~), modified III (&), withoutpolymerization (�).
A Semi-Batch Process for Nitroxide Mediated Radical Polymerization 237
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
![Page 9: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/9.jpg)
chains, and will therefore cause a direct and immediate
decline in the nitroxide concentration. CSA is known to
have only a pronounced effect when there is considerable
excess of nitroxide, and to have a much smaller effect when
free nitroxide levels are low.[19] In these runs, the free
nitroxide concentrations are likely, not initially high enough
for the CSA addition to exhibit a significant effect. Modified
recipe III slightly increases the rate, however recall its pri-
mary purpose that was to enhance the initiation efficiency
rather than increase rate.
Figure 7 shows the molecular weight and polydispersity
profiles for the basic and modified profiles. It can be seen
that for all the LRP runs the molecular weight still increases
with conversion and the polydispersity is also lower than
that of conventional radical polymerization, which indica-
tes that the living characteristic of the reaction is maintained
in the modified recipes. It is interesting to observe that the
basic recipe and all modified recipes have similar poly-
dispersity at the same conversion. However, there are major
differences in the Mn profiles between the various recipes,
affecting both the linearity of the curves and the slope.
Modified recipe I (CSA addition) shows the steepest slope
with some downward curvature at higher conversions.
Modified recipe II (Luperox 7M75) has a slope slightly
steeper than the basic recipe, also with some downward
curvature at higher conversions. At low conversions
however, Mn is noticeably higher than the basic recipe,
which corresponds to fewer chains. Modified recipe III,
which was intended to increase the initiation efficiency,
displayed an Mn profile very similar to the basic recipe.
Recipe III was thus not successful in improving efficiency;
if it had been the observedMn values would have been lower
than the basic recipe at a given conversion. The downward
curvature seen at higher conversions in modified recipes I
and II, is a result of thermally generated chains and
additional chains arising from added initiator. The steeper
slopes, which signify fewer chains present in the mixture,
have a different underlying cause. Bimolecular termination
reduces the number of chains if termination is predomi-
nantly by combination, as it is with styrene. With modified
recipe I, the rapid increase in Mn at low conversions
followed by a linear growth suggests a burst of termination
occurs when the CSA is added, and after that the number of
chains remains approximately constant. With modified
recipe II, the slow Luperox 7M75 addition results in a more
gradual loss of chains by termination.
Effect of Recipe Modifications on Polymer Livingness
In living radical polymerization, the livingness of polymer
chains (fraction of polymer chains that are capped by nitro-
xide) is an important issue. Recently, our group developed
an approach to study the livingness of polymer chains.[9] In
this method, 4-hydroxy-TEMPO capping the polymer
chains end is replaced by the fluorescent N-TEMPO
through an exchange reaction, as shown in Figure 8. The
percentage of livingness can then be obtained by using GPC
coupled with calibrated RI and fluorescence detectors.
Figure 9 shows the mole percent livingness for recipe modi-
fication II (basic recipe using Luperox 7M75 as initiator).
Both conditions show that, as expected, the livingness
decreases as the conversion increases. Interestingly, the
slopes of the two curves are very different, with modified
recipe II declining more slowly. However the livingness of
the basic recipe is higher at low conversions. Extrapolating
to high conversions, the livingness is expected to be com-
parable for the two recipes. It is surprising that at the higher
reaction rate (modified recipe 2), the slope of livingness
versus conversion is so shallow. This finding may at first
consideration appear anomalous; higher rates imply higher
active radical concentrations and therefore higher termina-
tion rates. However, at these temperatures, disproportiona-
tion of the growing chains to yield a dead unsaturated chain
and a hydroxylamine is an important reaction at long
reaction times. We have previously shown that the majority
of dead chains are formed by disproportionation and not
termination.[20] Because disproportionation is a first order
reaction, longer reaction times (i.e., lower polymerization
rates) promote its effect. Therefore, the disadvantage of
increasing the rate (greater termination rates) is partially or
wholly offset by reduced disproportionation rates. We have
not optimized the initiator addition process; it may in fact be
possible to achieve higher livingness with the modified
recipe, as has been found in mini-emulsion SFRP.[21]
Batch Solution Polymerization With Varying4-Hydroxy-2,2,6,6,-tetramethylpiperidinyloxyLuperox 7M75 Ratio
One of the issues noted in previously described experiments
is the low initiation efficiency. A possible solution is to
Figure 7. Number average molecular weight (solid symbols)and polydispersity index (open symbols) for modified recipes.Symbols as in Figure 6.
238 Y. Wang, R. A. Hutchinson, M. F. Cunningham
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
![Page 10: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/10.jpg)
increase the amount of nitroxide present. Batch solution
polymerizations were run using three different 4-hydroxy-
TEMPO/Luperox 7M75 ratios to better understand the
effect of this variable on the course of polymerization. We
elected to use batch, and not semi-batch, conditions because
it simplifies interpretation of the results. The monomer
concentration profiles, molecular weight and its polydis-
persity are shown in Figure 10 and 11, respectively. In the
recipe using a nitroxide/initiator ratio of 1.39, the poly-
merization proceeded vigorously. The rate is high at the
beginning and slows down gradually. Mn increases rapidly
with conversion initially, and then subsequently displays
linear growth. This behavior is indicative of poor control at
an early stage, followed by a well-controlled polymeriza-
tion. The polydispersity remains constant at �1.4. At a
nitroxide/initiator ratio of 1.59, the rate is somewhat slower,
and the final conversion lower than at the 1.39 ratio. How-
ever, Mn is again seen to increase rapidly with conversion
initially, and then exhibit linear growth. Polydispersity is
much improved at 1.2. In the recipe using a nitroxide/
initiator ratio of 1.87, a long induction period was observed,
and the final conversion lower than in runs with a lower
nitroxide/initiator ratio. TheMn profile more closely resem-
bled the ideal curve, and polydispersity was 1.1. Along with
differences in the kinetics and polydispersities, the number
of chains initiated at the three nitroxide/initiators also
differed, with more chains being initiated at higher nitro-
xide/initiator ratios. Initiation efficiencies are 65, 55, and
46% for nitroxide/initiator ratios of 1.87, 1.59, and 1.39,
respectively, based on 4-hydroxy-TEMPO. Better control
of the polymerization was also achieved at higher nitroxide/
initiator ratios, as reflected by lower polydispersities. These
runs illustrate that better control of the molecular weight
distribution can be achieved by increasing the amount of
nitroxide used. This benefit unfortunately comes at the
expense of a significantly reduced rate. The relative impor-
tance of each of these attributes would have to be assessed
prior to deciding on an appropriate nitroxide/initiator ratio.
Copolymerization of Styrene and Butyl Acrylate
Copolymerization experiments were also performed to
investigate system behavior in the presence of acrylates, a
common monomer in coatings formulations. Acrylates do
Figure 8. Schematic representation of the exchange reaction to produceN-TEMPO cappedpolymer chains.
Figure 9. Comparison of livingness (mol-%) for the basic recipe(^), and modified recipe II (&).
Figure 10. Monomer concentration profiles for batch polymer-izations. 4-Hydroxy-TEMPO/TBPA¼ 1.87 (*), 4-hydroxy-TEMPO/TBPA¼ 1.59 (&), 4-hydroxy-TEMPO/TBPA¼1.39 (^).
A Semi-Batch Process for Nitroxide Mediated Radical Polymerization 239
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
![Page 11: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/11.jpg)
not polymerize easily with TEMPO-derived nitroxides,
primarily because the near absence of thermal initiation in
these system results in the gradual accumulation of nitro-
xide and subsequent rate suppression. Copolymerizations
with styrene are possible since the styrene still contributes
to thermally generated radicals but the rates tend to be low.
Using a recipe similar to the basic recipe for styrene
homopolymerization, a mixture of 50/50 (mass basis)
styrene/butyl acrylate was used in the copolymerization
experiments.
The monomer concentration profiles in Figure 12, reveal
the overall concentration of BA and ST in copolymerization
is similar to the monomer concentration profile in styrene
homopolymerization. During copolymerization, styrene is
consumed slightly faster than butyl acrylate because of their
different copolymerization reactivity ratios. Figure 13
shows that the mass of styrene in the copolymer is about
1.2 times that of butyl acrylate. Figure 14 shows that Mn
increases approximately linearly with conversion to about
40% conversion, after which downward curvature in the
plot is seen. This indicates an increasing number of chains
caused by thermally generated radicals, similar to what was
observed in styrene homopolymerization. The molecular
weight distribution is slightly broader than that of the
corresponding styrene homopolymerization, and increases
gradually during polymerization to a final value of 1.5.
Although no attempt was made to optimize this formula-
tion, these results demonstrate the feasibility of conducting
nitroxide mediated styrene/acrylate copolymerizations in a
semi-batch process.
Figure 11. Number average molecular weight (solid symbols)and polydispersity index (open symbols) for batch polymeriza-tions. Symbols as in Figure 10.
Figure 12. Monomer concentration profiles for homopolymer-ization of styrene (*) and styrene/butyl acrylate copolymeriza-tion: overall monomer (~), styrene (&) and butyl acrylate (^).
Figure 13. Weight ratio of styrene and butyl acrylate in solution(^) and in polymer (&).
Figure 14. Number average molecular weight (solid symbols)and polydispersity index (open symbols) for homopolymerization(^) and copolymerization (&).
240 Y. Wang, R. A. Hutchinson, M. F. Cunningham
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
![Page 12: A Semi-Batch Process for Nitroxide Mediated Radical Polymerization](https://reader035.vdocument.in/reader035/viewer/2022080307/575005ab1a28ab1148a5b279/html5/thumbnails/12.jpg)
Conclusion
It has been demonstrated that semi-batch nitroxide medi-
ated polymerization is feasible. The unique operating
conditions in semi-batch, present challenges that do not
exist with batch processes, including high initial initiator/
alkoxyamine concentrations (�10� greater than batch
processes) and low monomer concentrations throughout the
process. The high initiator/alkoxyamine concentrations,
result in increased rates of bimolecular termination in the
very early stages of the polymerization, and are reflected in
low initiation efficiencies. Low monomer concentrations
lead to reduced reaction rates. The five different initiating
systems (four two-component systems and one alkoxya-
mine) displayed distinctly different kinetic behavior.
Among the five recipes using different initiation appro-
aches, Luperox 7M75/4-hydroxy-TEMPO and Luperox
231/4-hydroxy-TEMPO give the highest rates with reason-
able control over the polymerization, although they are
slower than conventional free radical polymerization. The
slow polymerization rate can be partially overcome by
adding CSA or charging initiator concurrently with mono-
mer during the semi-batch feed stage, with initiator addition
being the preferred route. Using a higher ratio of nitroxide/
initiator can increase the initiating efficiency but at the
expense of a lower rate. The copolymerization of styrene
and butyl acrylate (50/50) was also demonstrated, and was
shown to yield a well-controlled polymerization with
the molecular weight increasing linearly with conversion.
The type of process reported in this work is suitable for the
preparation of low molecular weight coatings such as those
used in the automotive industry.
[1] [1a] M. K. Georges, R. P. N. Veregin, P. M. Kazmaier, G. K.Hamer, Macromolecules 1993, 26, 2987; [1b] C. J. Hawker,A. W. Bosman, E. Harth, Chem. Rev. 2001, 101, 3661.
[2] [2a] K. Matyjaszewski, T. E. Patten, J. Xia, J. Am.Chem. Soc.1997, 119, 674; [2b] K. L. Robinson, M. V. de Paz-Banez,X. S. Wang, S. P. Armes, Macromolecules 2001, 34, 5799.
[3] [3a] J. Chiefari, Y. K. Chong, F. Ercole, J. Krstina, J. Jeffery,T. P. T. Le, R. T. A. Mayadunne, G. F. Meijs, C. L. Moad, G.Moad, E. Rizzardo, S. H. Thang, Macromolecules 1998, 31,5559; [3b] A. Goto, K. Sato, Y. Tsujii, T. Fukuda, G. Moad,E. Rizzardo, S. H. Thang, Macromolecules 2001, 34, 402.
[4] T. Fukuda, A. Goto, K. Ohno, Macromol. Rapid Commun.2000, 21, 151.
[5] V. Percec, D. A. Tirrell, J. Polym. Sci., Part A: Polym. Chem.2000, 38, 1705.
[6] [6a] Superintendent of Documents, Title 1, US GovernmentPrinting Office, Washington, DC, 1990, p. 1; 6b] Super-intendent of Documents, Clean Air Act Amendments of1990, Title 111, US Government Printing Office, Washing-ton, DC, 1990, p. 236; [6c] R. S. Reisch, Chem. Eng. News1993, 71(Oct), 34.
[7] [7a] K. Adamsons, G. Blackman, B. Gregorovich, L. Lin, R.Matheson, Prog. Org. Coat. 1998, 34, 64; [7b] M. C. Grady,W. J. Simonsick, Jr., R. A. Hutchinson, Macromol. Symp.2002, 182, 149.
[8] [8a] K. Matyjaszewski, B. E. Woodworth, X. Zhang, S. G.Gaynor, Z. Metzner, Macromolecules 1998, 31, 5955; [8b]M. J. Jones, G. Moad, E. Rizzardo, D. Solomon, J. Org.Chem. 1989, 54, 1607.
[9] M. E. Scott, J. S. Parent, S. L. Hennigar, R. A. Whitney, M. F.Cunningham, Macromolecules 2002, 35, 7628.
[10] G. Moad, D. H. Solomon, ‘‘The Chemistry of Free RadicalPolymerization’’, Pergamon, Oxford, UK 1995.
[11] [11a] D. Greszta, K. Matyjaszewski, Macromolecules 1996,29, 7661; [11b] D. A. Shipp, K. Matyjaszewski, Macro-molecules 1999, 32, 2948; [11c] M. Souaille, H. Fischer,Macromolecules 2001, 34, 2830.
[12] [12a] A. Goto, T. Fukuda, Macromolecules 1997, 30, 4272;[12b] T. Fukuda, T. Terauchi, A. Goto, K. Ohno, Y. Tsujii, T.Miyamoto, S. Kobatake, B. Yamada, Macromolecules 1996,29, 6393.
[13] M. Zhang, W. H. Ray, Ind. Eng. Chem. Res. 2001, 40, 4336.[14] [14a] M. K. Georges, R. A. Kee, R. P. N. Veregin, G. K.
Hamer, P. M. Kazmaier, J. Phys. Org. Chem. 1995, 8, 301;[14b] W. Devonport, L. Michalak, E. Malmstrom, M. Mate,B. Kurdi, C. J. Hawker, G. G. Barclay, R. Sinta,Macromolecules 1997, 30, 1929; [14c] K. Ohno, Y. Tsujii,T. Fukuda, Macromolecules 1997, 30, 2503.
[15] [15a] M. Buback, S. Klingbeil, J. Sandmann, M.-B. Sderra,H.-P. Vogele, H. Wackerbarth, L. Wittkowski, Z. Phys.Chem. 1999, 210, 199; [15b] M. Buback, J. Sandmann, Z.Phys. Chem. 2000, 214, 583.
[16] S. Marque, C. Le Mercier, P. Tordo, H. Fischer, Macro-molecules 2000, 33, 4403.
[17] M. K. Georges, G. Hamer, A. R. Szkurhan, A. Kazemedah, J.Li,Polym. Prepr. (Am. Chem. Soc., Div. Polym. Chem.) 2002,27, 191.
[18] [18a] R. P. N. Veregin, P. G. Odell, L. M. Michalak, M. K.Georges, Macromolecules 1996, 29, 4161; [18b] W. C.Buzanowski, J. D. Graham, D. B. Priddy, E. Shero, Polymer1992, 33, 3055.
[19] K. Tortosa, M. F. Cunningham, B. Keoshkerian, M. K.Georges, J. Polym. Sci., Part A: Polym. Chem. 2002, 40,2828.
[20] J. W. Ma, J.-A. Smith, M. F. Cunningham, K. McAuley, M.K. Georges, B. Keoshkerian,Chem.Eng. Sci. 2003, 58, 1163.
[21] M. Lin, M. F. Cunningham, B. Keoshkerian, Macromol.Symp. 2004, 206, 263.
A Semi-Batch Process for Nitroxide Mediated Radical Polymerization 241
Macromol. Mater. Eng. 2005, 290, 230–241 www.mme-journal.de � 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim