Вы находитесь на странице: 1из 6

Full Paper

Continuous Atom Transfer Radical


Polymerization in a Tubular Reactor

ller, Michael F. Cunningham,* Robin A. Hutchinson*


Matthias Mu

The use of a tubular reactor for conducting living radical polymerizations by atom transfer
radical polymerization (ATRP) was investigated. Solution polymerization experiments were
performed with styrene and butyl acrylate to elucidate the influence of a continuous reaction
process on conversion, molecular weight, and
polydispersity compared to batch polymeri-
zation experiments. The continuous polymeri-
zations were well controlled. Initial conversion
was found to be slightly higher in the tubular
reactor than in a batch polymerization run at
similar conditions, while number average mo-
lecular weight and polydispersity are compar-
able between the continuous and batch
processes. Residence time distribution studies
showed the reactor exhibits nearly plug flow
behavior.

Introduction Recently, Noda et al.[3] reported the preparation of MMA


homo- and block co-polymers via ATRP using a microreactor
Living/controlled radical polymerization (L/CRP) offers the consisting of a poly(tetrafluoroethylene) (PTFE) tube of 10
potential for major advances in the manufacture of poly- mL volume. Schork and Smulder[4] have investigated
meric materials through its control of polymer micro- theoretical aspects of using a continuous stirred tank
structure - narrowing of molecular weight distributions reactor for ATRP, NMP, and RAFT polymerizations. They
(MWD), controlled composition distribution along the predicted a polydispersity index (PDI) of two compared to
chain, and targeted placement of functional groups and the theoretical ideal polydispersity of one for a batch
branch points. Most L/CRP studies have been done using reactor. Enright et al.[5] investigated nitroxide mediated L/
bulk or solution polymerization in batch reactors. There CRP in miniemulsion using a tubular reactor. RAFT
are only a few references on living radical polymerization polymerization in miniemulsion has been studied
in a continuous reactor. Homogeneous bulk atom transfer by Smulders et al.[6] using a train of CSTRs and Russum
radical polymerization (ATRP) of methyl methacrylate in a et al.[79] using a tubular reactor.
continuous packed bed of silica supported catalyst has There are several potential benefits of using a contin-
been demonstrated successfully by Zhu and co-workers.[1,2] uous tubular reactor for ATRP polymerizations. Synthesiz-
ing polymers with controlled microstructure (e.g., block
copolymers) can be simplified by using feed streams
M. Mu ller, M. F. Cunningham, R. A. Hutchinson
Department of Chemical Engineering, Dupuis Hall, Queens
situated along the reactor. The conversion at which the
University, 19 Division St., Kingston, ON, Canada K7L 3N6 additional monomer is added can be easily adjusted by
Fax: (1) 613 533 6637; varying the flow rate. A tubular reactor permits better
E-mail: michael.cunningham@chee.queensu.ca, control of the temperature profile, especially during
robin.hutchinson@chee.queensu.ca the heating period, and temperature changes during

Macromol. React. Eng. 2008, 2, 3136


2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200700029 31
M. Mu
ller, M. F. Cunningham, R. A. Hutchinson

polymerization, which in the end leads to better controlled Analyses


polymer properties. Furthermore, operating a tubular
The molecular weight and PDI were measured by gel permeation
reactor under pressure, which is needed close to or above chromatography (GPC). The samples were first passed through a
the boiling point of monomer or solvent, is much easier column packed with basic alumina to remove the residual copper.
than running a stirred tank reactor under the same The GPC instrument was equipped with a Waters 2960 module
pressure, for both laboratory and industrial scale. This containing four Styragel columns with pore sizes of 100, 500, 103,
paper describes initial results using a continuous tubular and 104 A , coupled with a Waters 410 differential refractive index
reactor to produce homopolymers of butyl acrylate and (RI) detector (480 nm) and a Wyatt Technology DAWN EOS
styrene via ATRP. photometer multi-angle light scattering (LS) detector (690 nm,
30 mW Ga-As laser). THF was used as the eluent and the flow rate
was set to 1 mL  min1. The LS detector was calibrated with toluene,
Experimental Part normalized with 30 000 g  mol1 narrow polystyrene standard, and
the data were processed using Astra (Version 4.90.08) software. The
The reactor setup is based on a similar setup used by Enright et al.[5] system was calibrated with 11 narrow polystyrene standards.
to investigate nitroxide mediated polymerization in miniemulsion. Conversion was analyzed gravimetrically by drying small samples
The tubular reactor setup consists of stainless steel tubing (150 m at room temperature for 2 d and by gas chromatography using
length, 3.2 mm outer diameter, 2.1 mm inner diameter, reactor a Varian CP 3800 system fitted with a 50 m  0.25 mm inner
volume 0.52 L) immersed in an oil bath. The flow was controlled diameter fused-silica CP-Sil wall coated open tubular capillary
by a Masterflex peristaltic pump situated at the outlet of the reactor column (WCOT). Sample injection was controlled by a Varian model
tube. Between the pump and oil bath, a heat exchanger is used CP-8410 injector. Analytes were detected using a flame ionization
to cool the reaction mixture down. The actual mass flow was detector. The GC was controlled by Varian Star software.
measured by a balance (Mettler Toledo PG 5002s) at the outlet. A
scheme of the reactor is shown in Figure 1. Two storage tanks were
used in order to be able to refill one storage tank while the other is
used for feeding the reactor. Therefore, operation time is not limited Polymerization Procedure
by the volume of the storage tanks. Monomer, solvent, CuBr, and PMDETA were premixed for at least
Batch reactions for comparison to the continuous reactor were 30 min in an Erlenmeyer flask while purging the solution with
done utilizing a 1-L glass autoclave built into a Mettler-Toledo nitrogen. If there was no insoluble copper at the bottom after the
LabMax system. initial mixing period, the initiator was added and the solution was
poured into the two storage tanks of the tubular reactor setup. In
the case where insoluble copper was observed, the mixing period
Materials was extended by 30 min, followed by 20 min to let the insoluble
parts settle to prevent transfer to the storage tanks. In most cases
Styrene (99%, Aldrich), butyl acrylate (99%, Aldrich), toluene (99%,
there was no insoluble catalyst. After purging with nitrogen for an
Aldrich), benzonitrile (99%, Aldrich), acetonitrile (99%, Aldrich),
additional 10 min the storage tanks were closed and the nitrogen
Cu(I)Br (98%, Aldrich), N,N,N0 ,N00 ,N0 - pentamethyldiethylenetri-
valve was regulated to provide a pressure of 2 bar.
amine (PMDETA, 99%, Aldrich), and methyl 2-bromopropionate
To begin the continuous reactions, the reactor tube was quickly
(MBrP, 98%, Aldrich) were used as received.
filled from storage tank #2 which took 56
min. The pump head of the peristaltic pump
was closed and the pump started with the
desired flow rate. Samples were taken at
intervals at the outlet of the reactor, which
results in samples with successively longer
reaction times and therefore higher conver-
sion until the full residence time of the tubular
reactor is reached. In principle it is possible to
take samples in the middle of the reactor
using T-junctions, but this would interrupt the
flow pattern and therefore was not done in
this study.

Residence Time Distributions (RTDs)


The RTD of the tubular reactor was deter-
mined using water. After applying a pulse of
NaCl solution, the conductivity was measured
Figure 1. Continuous tubular reactor apparatus for ATRP polymerization. using a conductivity meter placed between

Macromol. React. Eng. 2008, 2, 3136


32 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200700029
Continuous Atom Transfer Radical Polymerization in a Tubular Reactor

the heat exchanger and the peristaltic pump. Control experiments


without the reactor were done to determine the time the salt
solution needs to reach the reactor from the injection point and to
reach the conductivity meter after leaving the tubular reactor.

Results

The rate of polymerization for styrene and butyl acrylate


using ATRP is significantly lower than the rate of the
corresponding normal free radical polymerization. There-
fore, longer times are needed to achieve the same level
of conversion. Using a tubular reactor setup there are
basically two different design options within the limita-
tion of a reasonable flow rate range: a short tube with large
inner diameter or a long tube with small inner diameter.
For this setup the latter was chosen in order to provide a Figure 3. Cumulative distribution function measured by applying
a pulse of NaCl solution in water. The vertical line is the theo-
reasonable average velocity along the tube. Higher velocity
retical expectation for an ideal tubular reactor. The temperature
should help to prevent fouling at the walls, which is of the oil bath was 21 8C.
especially important at high conversion and molecular
weight.

Reynolds number of water under these conditions is


smaller than ten due to the very low flow rate. Therefore,
Residence Time Distribution Studies
axial mixing can play an important role as well. When
A typical result of the residence time distribution is shown Russum et al.[8] investigated RAFT miniemulsion poly-
in Figure 2. Three replicate experiments are shown using merization in a tubular reactor under low flow rates, they
the same flow rate. Figure 3 shows the cumulative distri- also found residence time distributions close to plug flow
bution function for one of the detected signals given in and attributed this to a high axial mixing.
Figure 2. For comparison, the theoretical graphs expected The results of all residence time distribution experiments
for laminar and plug flow are also shown. The experi- are given in Table 1. The full width at half maximum of the
mental RTD is close to plug flow. This is most probably signals (like the ones shown in Figure 2) becomes smaller
caused by the fact that the tubular reactor is coiled to fit with increasing flow rate and increasing temperature. At
in the oil bath. Due to the coiled tubing secondary flow
patterns are introduced leading to better mixing and Table 1. Results of the residence time distribution experiments
therefore a narrow residence time distribution. The using water as solvent. The conductivity was measured after
applying a pulse of NaCl solution. For each temperature and flow
rate, three experiments were done giving the mean residence
time and full width at half maximum of the NaCl pulse.

Temperature Mass Mean Full width


flow residence at half
time maximum

-C g  minS1 s s

21 2.13 16 068 751


21 2.11 15 859 754
21 2.07 15 621 763
21 3.11 10 708 600
21 3.11 10 698 600
21 3.14 10 639 591
80 3.17 10 262 360
Figure 2. Results of three replicate residence time distribution
experiments at a flow rate of 3.1 g  min1. The temperature of the
80 3.17 10 240 354
oil bath was 21 8C. 80 3.19 10 192 350

Macromol. React. Eng. 2008, 2, 3136


2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mre-journal.de 33
M. Mu
ller, M. F. Cunningham, R. A. Hutchinson

Table 2. Formulations used in each tubular and batch polymerization experiment of the Figure 4 and 5. The masses and moles of monomer,
initiator, catalyst, and ligand are given.

BA batch BA tubular STY batch STY tubular

m n m n m n m n

g mol g mol g mol g mol

Monomer 362.57 2.83 900.83 7.03 302.58 2.91 1 142.92 10.97


CuBr 1.53 10.7  10S3 3.79 26.4  10S3 4.16 29.0  10S3 15.73 109.7  10S3
PMDETA 6.06 35.0  10S3 15.21 87.8  10S3 10.03 57.9  10S3 37.96 219.0  10S3
S3 S3 S3
MBrP 11.94 71.5  10 29.65 177.5  10 9.67 57.9  10 36.53 218.7  10S3
Temperature [-C] 80 80 110 110
Solvent [g] 155.44 (acetonitrile) 386.85 (acetonitrile) 129.13 (toluene) 488.99 (toluene)

80 8C, the full width at half maximum is 355 s and the mean part of the graph. There is a slight difference in con-
residence time is 10 240 s, only slightly different (3.5%) version between the tubular and batch experiment. The
than the expected value. Since the polymerizations are conversion-time profiles shown in the left part of the
carried out at 80 8C and higher, one can assume that figures show a higher conversion for the tubular experi-
the effect of the residence time distribution on the PDI will ment, even if the time to fill the tubular reactor is taken
be small. into account. On the other hand, the evolution of molar
mass with respect to the conversion is the same for the two
reactor types. This indicates that there is no loss of growing
Polymerizations chains in the batch reactor compared to the tubular
reactor. The differences between the theoretical and
Solution homopolymerizations of butyl acrylate and
experimental molecular weights are small, indicating
styrene were investigated using both the continuous
high initiation efficiency and a nearly constant number of
tubular reactor and the batch reactor. First experiments
living chains during polymerization. The evolution of
were tried with a copper to initiator ratio of 1:1. Since
ln(M0/M) with time depicted by the open symbols in the
the copper bromide was not fully soluble under these
left part is quite linear. According to the equation ln(M0/
conditions the amount of copper was decreased until
M) kpRt (where M is the monomer concentration, M0
the solution became homogeneous, while the amount of
the monomer concentration at time zero, kp the overall
ligand was held constant. The formulations are described
propagation rate coefficient, R the radical concentration,
in Table 2. The amount of initiator was chosen to reach a
and t is the time) this is an evidence of no noticeable loss of
molecular weight of about 5 000 g  mol1 at full conver-
active radicals during polymerization.
sion. This value was chosen as it is typical of that used in
The PDI except for the first samples taken from the
automotive coatings. The temperatures of 80 8C for butyl
tubular reactor is less than 1.15, which is indicative of
acrylate and 110 8C for styrene were chosen to be the same
excellent control of the polymerization.
used by Fu working on heterogeneous ATRP.[10,11]
A possible reason of the difference in conversion might
be the mode of operation. Russum et al.[9] also found an
increase in conversion using a tubular reactor compared to
Homopolymerization of Styrene
the batch control experiment in their investigation of
The results of the experiments with styrene using the
formulation in Table 2 are given in Table 3 and Figure 4.
Table 3. Summary of end results of experiments using the for-
Circles and squares indicate results from batch and tubular mulations given in Table 2. Listed are the conversion, molecular
experiments, respectively. The batch experiments show a weight, and polydispersity of the last sample taken.
continuous increase in conversion, as expected. In case of
the tubular runs, the conversion stays constant after the BA BA STY STY
operating time exceeds approximately one mean residence batch tubular batch tubular
time of the reactor, since the reaction time is now limited
Conversion [%] 82 71 75 50
by the residence time and not the operating time any
more. The value of the mean residence determined at 80 8C Mn 4 797 4 160 4 426 2 807
in water is indicated by the vertical line in the upper PDI 1.09 1.09 1.13 1.14

Macromol. React. Eng. 2008, 2, 3136


34 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200700029
Continuous Atom Transfer Radical Polymerization in a Tubular Reactor

control of the polymerization, and


therefore we investigated the use
of a different solvent. Due to its
low boiling point of 82 8C, acet-
onitrile was deemed unsuitable,
and benzonitrile (bp 189 8C)
was chosen instead.
Initial experiments with ben-
zonitrile showed no precipitation
in a batch experiment but when
used in the tubular reactor did
lead to a blocked tube after 2 h
of operation. This time the block-
age occurred in or following
the heat exchanger, when the
reaction mixture was cooling
down. Opening a three-way valve
attached between the reactor
Figure 4. Results of styrene homopolymerization. Tubular and batch experimental results are and the heat exchanger for test-
depicted by squares and circles, respectively. Open symbols correspond to the right axis and ing purposes revealed that the
closed symbols to the left. The vertical line in the upper left part indicates the mean residence hot reaction mixture was homo-
time of the high temperature experiment in Table 1. geneous. Further tests are cur-
rently being done to optimize the
RAFT polymerization in miniemulsion. In case of the cooling unit to prevent blockage as a result of catalyst
batch experiments, the initiator is added after the reaction precipitation during the cooling step.
mixture is heated up, to be as close as possible to the
tubular reactor, in which the reaction mixture is heated up
nearly instantaneously entering the reactor. Adding the
Homopolymerization of Butyl Acrylate
initiator in the batch experiments usually takes between
40 and 50 s and additional time to mix. Therefore, the The results of the experiments with butyl acrylate are
initiation period may be a little longer in the case of the shown in Table 3 and Figure 5. As with styrene there is a
batch experiments. slight difference in conversion between the batch and the
After styrene polymerizations in the tubular reactor tubular experimental results, which can be explained
were completed, at the beginning of the cleaning proce- following the same arguments given for styrene.
dure, it was observed that small particles were washed out One can observe a slight curvature in the ln(M0/M)
of the reactor. On one occasion the reactor became blocked versus time profile for the batch experiment. This indicates
during the experiment. This is most probably due to a decrease in the active radical concentration during
precipitation of the Cu(II) catalyst complex. The solubility polymerization. There are two possible reasons, either a
of the Cu(II) catalyst is lower than that of Cu(I).[10,11] During loss of radicals by termination or side reactions. Radicals
polymerization there may be gradual accumulation of may be lost by reactions with the ligand PMDETA as pro-
Cu(II) due to a small but finite termination rate. While this posed by Sharma et al.[13] On the other hand, the deviation
is not a serious problem in a stirred tank batch reactor, it between the theoretical molar mass and the experimental
can cause serious operability issues with a tubular reactor. values is small, which indicates that there is no significant
There are several possible approaches to solve this loss of radicals during polymerization and that the
problem. The easiest would be to choose a different ligand initiation efficiency is close to unity. Another possibility
such as 4,4-di-(5-nonyl)-2,2-bipyridine (dNbpy) which is the backbiting reaction which leads to a mid-chain
leads to a homogeneous reaction mixture.[12] However, radical which is more stable and therefore propagates
this ligand is not commercially available in large volumes slower than the normal radical. At higher conversion the
and is therefore too expensive for industrial applications. backbiting reaction becomes more favored.[14] The impli-
Alternative approaches to make the reaction mixture cations of these side reactions on ATRP will be studied in
homogeneous are to reduce the amount of copper in the future work.
solution or switch to another solvent which is a better The agreement between the results for evolution of
solvent for the catalyst complex. The former choice would molar mass and polydispersity with conversion of the
lead to a lower conversion and likely negatively impact batch and tubular experiment is remarkable. Except for

Macromol. React. Eng. 2008, 2, 3136


2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mre-journal.de 35
M. Mu
ller, M. F. Cunningham, R. A. Hutchinson

polymerization system. The for-


mulation has to be homogeneous
throughout the full conversion
and temperature range used in
the reactor setup. Even small
amounts of precipitated catalyst
particles can lead to problems
with long-term operation, which
is not a concern for batch reactors.
This is an initial work done to
determine the feasibility of using
a tubular reactor. Further details
will be published in a future paper.

Acknowledgements: Financial sup-


port from the German Academic
Exchange Service (DAAD) and from
the Natural Sciences and Engineering
Figure 5. Results of butyl acrylate homopolymerization. Tubular and batch experimental results Research Council of Canada is grate-
are depicted by squares and circles, respectively. Open symbols correspond to the right axis and fully acknowledged.
closed symbols to the left. The vertical line in the upper left part indicates the mean residence
time of the high temperature experiment in Table 1. Received: July 27, 2007; Revised:
October 5, 2007; Accepted: October
the first two data points, the differences between the 18, 2007; DOI: 10.1002/mren.200700029
two experimental setups lie within the analytical uncer- Keywords: ATRP; butyl acrylate; continuous; kinetics; styrene;
tainties. tubular reactor
In contrast to the styrene experiments, the tubular
reactor with butyl acrylate was run for 6 h without any
problem. Cleaning of the reactor showed no evidence of
solid particles washed out with the cleaning solvents
water and acetone. Therefore, longer operation should be [1] Y. Shen, S. Zhu, R. Pelton, Macromol. Rapid Commun. 2000,
feasible. Higher solubility of the catalyst complex (espe- 21, 956.
[2] Y. Shen, S. Zhu, F. Zeng, R. H. Pelton, Macromolecules 2000,
cially the Cu(II) species) in BA compared to styrene explains
33, 5427.
the absence of fouling, and underscores the importance of [3] T. Noda, A. J. Grice, M. E. Levere, D. M. Haddleton, Eur.
the catalyst solubility in the polymerization to the overall Polym. J. 2007, 43, 2321.
operability of a tubular reactor. [4] F. J. Schork, W. Smulders, J. Appl. Polym. Sci. 2004, 92, 539.
[5] T. E. Enright, M. C. Cunningham, B. Keoshkerian, Macromol.
Rapid Commun. 2005, 26, 221.
[6] W. W. Smulders, C. W. Jones, F. J. Schork, Macromolecules
Conclusion 2004, 37, 9345.
[7] J. P. Russum, C. W. Jones, F. J. Schork, Macromol. Rapid
Living radical polymerization of styrene and butyl acrylate Commun. 2004, 25, 1064.
using ATRP was demonstrated in a continuous tubular [8] J. P. Russum, C. W. Jones, F. J. Schork, AIChE J. 2006, 52, 1566.
reactor, resulting in polymer with low PDI. The reaction [9] J. P. Russum, C. W. Jones, F. J. Schork, Ind. Eng. Chem. Res.
2005, 44, 2484.
kinetics was similar for batch and continuous reactors. The [10] Y. Fu, Ph. D. Thesis, Queens University, Kingston, ON 2007.
conversion in the continuous reactor is slightly higher [11] Y. Fu, A. Mirzaei, M. F. Cunningham, R. A. Hutchinson,
than in the batch control experiments. The molecular Macromol. React. Eng. 2007, 1, 425.
weight evolution with conversion and the polydispersity [12] K. Matyjaszewski, T. E. Patten, J. Xia, J. Am. Chem. Soc. 1997,
is very close to the ones obtained during batch control 119, 674.
[13] R. Sharma, A. Goyal, J. M. Caruthers, Y.-Y. Won, Macromol-
experiments. ecules 2006, 39, 4680.
Using a tubular instead of a batch reactor for ATRP [14] R. X. E. Willemse, A. M. van Herk, E. Panchenko, T. Junkers,
polymerization imposes an additional constraint on the M. Buback, Macromolecules 2005, 38, 5098.

Macromol. React. Eng. 2008, 2, 3136


36 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mren.200700029

Вам также может понравиться