You are on page 1of 134

Polymerization Processes 1

Polymerization Processes
Archie E. Hamielec, Institute for Polymer Production Technology, Department of Chemical Engineering,
McMaster University, Hamilton, Ontario, L8S 4L7, Canada
Hidetaka Tobita, Department of Materials Science and Engineering, Fukui University, Fukui, 910, Japan

1. Introduction–Trends in Poly- 3.3.1.2. Molecular Masses, Long-Chain


mer Reaction Engineering . . . 7 Branching, and Cross-Linking . . 50
2. Polymerization Mechanisms 3.3.2. Examples of Free-Radical Poly-
and Kinetics . . . . . . . . . . . . . 8 merization . . . . . . . . . . . . . . 50
2.1. Step-Growth Polymerization . . 9 3.3.2.1. Homopolymerization – Linear
2.1.1. Linear Polymerization . . . . . . . 9 Chains . . . . . . . . . . . . . . . . . 50
2.1.2. Interfacial Polymerization . . . . . 12 3.3.2.2. Copolymerization – Linear
2.1.3. Nonlinear Polymerization . . . . . 12 Chains . . . . . . . . . . . . . . . . . 54
2.2. Chain-Growth Polymerization . 14 3.3.2.3. Copolymerization – Long-Chain
2.2.1. Free-Radical Polymerization . . . 15 Branching . . . . . . . . . . . . . . . 55
2.2.1.1. Initiation . . . . . . . . . . . . . . . 16 3.3.3. Polymerization Processes . . . . . 55
2.2.1.2. Propagation . . . . . . . . . . . . . . 18 3.3.3.1. Solution Polymerization . . . . . . 55
2.2.1.3. Termination . . . . . . . . . . . . . 19 3.3.3.1.1. Polymer Soluble in Monomer . . 55
2.2.1.4. Chain Transfer to Small 3.3.3.1.2. Addition of a Solvent in which
Molecules . . . . . . . . . . . . . . . 21 both Monomer and Polymer are
2.2.1.5. Kinetics of Linear Polymerization 22 Miscible . . . . . . . . . . . . . . . . 55
2.2.1.6. Effect of Temperature . . . . . . . 25 3.3.3.1.3. Polymer – Polymer Demixing dur-
2.2.1.7. Branching Reactions . . . . . . . . 26 ing Polymerization . . . . . . . . . 56
2.2.2. Ionic Polymerization . . . . . . . . 28 3.3.3.2. Precipitation Polymerization . . . 57
2.2.2.1. Cationic Polymerization . . . . . . 29 3.3.3.2.1. Polymer Insoluble in its Monomer 57
2.2.2.2. Anionic Polymerization . . . . . . 30 3.3.3.2.2. Monomer Functioning as Solvent
2.2.2.3. Ziegler – Natta Polymerization . . 32 for the Polymer . . . . . . . . . . . 60
2.3. Copolymerization . . . . . . . . . 34 3.3.3.3. Suspension Polymerization . . . . 63
2.3.1. Copolymer Composition . . . . . 35 3.3.3.3.1. Qualitative Description . . . . . . 64
2.3.2. Kinetics of Copolymerization . . 37 3.3.3.3.2. Dispersants . . . . . . . . . . . . . . 66
2.3.3. Copolymerization of Vinyl and 3.3.3.3.3. Mechanism of Particle Formation 67
Divinyl Monomers . . . . . . . . . 38
3.3.3.3.4. Industrial Applications . . . . . . . 70
3. Polymerization Processes and
3.3.3.4. Emulsion Polymerization . . . . . 73
Reactor Modeling . . . . . . . . . 40
3.3.3.4.1. Theories of Emulsion Polymeriza-
3.1. Introduction . . . . . . . . . . . . . 40
tion . . . . . . . . . . . . . . . . . . . 74
3.2. Processes and Reactor Modeling
3.3.3.4.2. Physicochemical Parameters of
for Step-Growth Polymerization 41
Dispersions . . . . . . . . . . . . . . 85
3.2.1. Types of Reactors and Reactor
Modeling . . . . . . . . . . . . . . . 41 3.3.3.4.3. Inverse Emulsion Polymerization 88
3.2.2. Specific Processes . . . . . . . . . 44 3.3.3.4.4. Semi-Batch Emulsion Polymeri-
zation . . . . . . . . . . . . . . . . . 89
3.3. Processes and Reactor Modeling
for Chain-Growth Polymeriza- 3.3.3.4.5. Continuous Emulsion Polymeri-
tion . . . . . . . . . . . . . . . . . . . 47 zation . . . . . . . . . . . . . . . . . 90
3.3.1. Material Balance Equations for 3.3.4. Miscellaneous Processes . . . . . 93
Batch, Semi-Batch, and Continu- 3.3.5. Ionic Polymerization Modeling . 96
ous Reactors . . . . . . . . . . . . . 47 3.3.5.1. Introduction . . . . . . . . . . . . . 96
3.3.1.1. Rates of Reaction and Copolymer 3.3.5.2. Heterogeneous Coordination Po-
Composition . . . . . . . . . . . . . 48 lymerization . . . . . . . . . . . . . 96

c 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a21 305
2 Polymerization Processes

3.3.6. Process Variables, Reactor Dy- 3.3.6.1.2. Monomer Coupling Without Ter-
namics/ Stability, On-Line Moni- mination Plug Flow and Batch Re-
toring and Control . . . . . . . . . 97 actors (CPFR/BR) . . . . . . . . . 103
3.3.6.1. Influence of Reactor Type and 3.3.6.1.3. Polymer Coupling . . . . . . . . . 104
Configuration on Molecular Mass 3.3.6.1.4. Copolymerization . . . . . . . . . . 107
and Copolymer Composition Dis-
3.3.6.1.5. Long-Chain Branching and Cross-
tributions, and on Long-Chain
Linking . . . . . . . . . . . . . . . . 109
Branching and Cross-Linking . . 97
3.3.6.1.1. Monomer Coupling with Bimo- 3.3.6.2. Reactor Dynamics and Stability . 111
lecular Termination Plug Flow and 3.3.6.3. On-Line Monitoring and Control 112
Batch Reactors (CPFR/BR) . . . . 99 4. References . . . . . . . . . . . . . . 114

Sections 3.3.3.1 – 3.3.3.5 and 3.3.6.1 were D stirrer diameter


based on the article Polymerisationstechnik in Dop mean diffusion coefficient for
Ullmann’s, 4th ed. written by Heinz Gerrens. oligomeric radicals and latex particles
DMT dimethyl terephthalate
List of symbols Ed activation energy for initiator decompo-
sition
A chemical species; vacant adsorption site
E/E 0 mass fraction of material passing out of
[A] concentration of species A
reactor with a residence time t to t + dt
[A]0 initial concentration of species A
Ef activation energy for chain-transfer re-
A1 , A2 , A3 adjustable parameters
action
ABS acrylonitrile – butadiene – styrene
EL activation energy for average chain
rubber-modified copolymer
lengths
ACA aminocaproic acid
EN activation energy for polymer particle
A (h) energy required to separate to a distance
nucleation
h=∞, two drops of diameter d=1 ini-
Ep activation energy for propagation
tially separated by a distance h0
ER activation energy for polymerization
Am surface area of micelles
E (t) residence-time distribution for a flow
Ap surface area of polymer particles
reactor at steady state
B chemical species
Et activation energy for bimolecular termi-
BHET bis-hydroxyethyl terephthalate
nation
BR batch reactor
Eu modified power number
C pi dimensionless moments of polymer dis-
EG ethylene glycol
tribution for chain transfer to polymer
EGDMA ethylene glycol dimethacrylate
[= K fp Qi /(K p [M])]
EPS expandable polystyrene
CS surfactant concentration
ESR electron spin resonance spectroscopy
CCD chemical composition distribution
f initiator efficiency; functionality of
CMC critical micelle concentration
monomer
CPFR continuous plug flow reactor
fj mole fraction of monomer of type j
CSTR continuous stirred-tank reactor with an
Fi, in molar flow rate of monomer of type i
ideal residence-time distribution
into the reactor
CTA chain-transfer agent
F in total molar flow rate (of all monomer
d particle diameter
types) into the reactor
d̄ average particle diameter
F Ii, in molar flow rate of initiator of type i into
d 32 Sauter mean diameter of a spherical-
the reactor
particle suspension
Fj mole fraction of monomer of type j,
d 50 diameter at which 50 wt % of particles
chemically bound in polymer produced
pass through a sieve
instantaneously
d min minimum particle diameter
d max maximum particle diameter
Polymerization Processes 3

F̄j mole fraction of monomer of type j K p∗ rate constant for polymeric radicals
chemically bound in accumulated poly- adding to pendant double bonds on
mer polymer chains
F̄ 1 mole fraction of monomer 1 (containing K p j , Kij propagation rate constant for
i
an abstractable atom) in accumulated monomer of type j adding to polymeric
polymer active center of type i
F̄ 2 mole fraction of monomer 2 (containing K p ij rate constant for polymeric radical of
a reactive carbon – carbon bond type i adding to a double bond on
F pi, in molar flow rate of monomer of type i a monomer unit of type j chemically
chemically bound in polymer into the bound in the polymer
reactor K p ijk propagation rate constant for monomer
Fr Froude number of type k adding to a polymeric active
F T, in molar flow rate of chain-transfer agent center of type ij
T into the reactor Kt total bimolecular termination constant
G+ counterion (K tc + K td )
GPC gel permeation chromatography K t0 total bimolecular termination constant
HCSTR homogeneous CSTR at zero conversion of monomer
HDPE high-density polyethylene K tc rate constant for bimolecular termina-
H – H Hui – Hamielec styrene polymerization tion by combination
model K tc ij termination by combination rate con-
HIPS high-impact polystyrene stant for polymeric radicals of types i
I initiator or catalyst and j (chemical control)
[I] concentration of initiator or catalyst K̄ tcN number-average bimolecular termina-
K chemical rate constant; equilibrium tion constant by combination
constant K td rate constant for bimolecular termina-
Ka absorption constant for oligomeric rad- tion by disproportionation
icals entering polymer particles K td ij termination by disproportionation rate
KA adsorption rate constant constant for polymeric radicals of types
Kd initiator decomposition constant i and j (chemical control)
K dp depropagation constant K̄ tdN number-average bimolecular termina-
KD desorption rate constant tion constant by disproportionation
Kfj rate constant for polymeric radical K̄ tN total number-average bimolecular ter-
i
of type i abstracting an atom from mination constant
monomer of type j chemically bound K tp termination rate constant in polymer
in polymer particles
K fm transfer to monomer rate constant K t (r, s) total bivariate distribution for
K fp rate constant for chain transfer to poly- diffusion-controlled bimolecular ter-
mer mination of polymeric species of chain
K fT rate constant for chain transfer to CTA lengths r and s
K fT i rate constant for chain transfer from K tw termination rate constant in aqueous
polymeric radical of type i to CTA phase
K fX transfer to small molecule X rate con- K̄ tW , K̄ tZ total weight- and z-average bimolec-
stant ular termination constants
Ki rate constant for monomer adding to a L characteristic length of energy-con-
primary radical taining large eddies
Kp propagation rate constant L length of path traversed by a growing
K ′p propagation rate constant for transfer radical from its point of origin to the
radical point where it precipitates
K− p propagation rate constant for free ion L reactor length
± LALLS low-angle laser light scattering
Kp propagation rate constant for ion pair
LCB long-chain branching
4 Polymerization Processes

LDPE low-density polyethylene NM number of monomer units; number


LLDPE linear low-density polyethylene of micelles; number of monomer
mi number of moles of monomer i in ter- molecules consumed
polymer (Eq. 3.101) Nn number of polymer particles containing
Mc average molecular mass between cross- n radicals
links Np number of polymer particles per unit
Mi monomer of type i volume
Mm aggregation number for emulsifier NT moles of CTA in reactor
molecules in micelles NBR nitrile – butadiene rubber
M mi molecular mass of monomer of type i NIRS near infrared spectroscopy
M̄ N , M̄ W , M̄Z , M̄Z +1 number-, weight-, Z and p conversion of functional groups
Z + 1-average molar mass (molecular pc critical threshold
mass, respectively) P growing polymer particle; polymer
[M] total monomer concentration Pc conversion of functional groups at gela-
[M]0 initial monomer concentration; tion point
monomer concentration in feed Pcr critical chain length for precipitation
[M]c equilibrium concentration of monomer Pi moles of monomer of type i chemically
at the ceiling temperature bound in polymer in the reactor
[Mi ] concentration of monomer of type i Pij polymer containing i units of monomer
[M]p concentration of monomer in the poly- of type 1 and j units of monomer of type
mer particles 2
M – H Marten – Hamielec polymerization [Pm ] concentration of polymer with chain
model length m
MMA methyl methacrylate Pm,n dead polymer chain containing m units
MWD molecular mass distribution (molar of monomer 1 and n units of monomer
mass distribution) 2
n number of monomer types PN number-average chain length of poly-
n order of reaction mer produced instantaneously
n̄ average number of radicals per particle P̄N number-average chain length of accu-
N 0 , N number of functional groups at time mulated polymer
zero and t P̄sol
N number-average chain length of sol
N total number of moles in the reactor; molecules
stirrer speed Pr polymer molecule of chain length r
NA number of moles of A-functional PW weight-average chain length of polymer
groups; Avogadro number produced instantaneously
N A0 initial number of moles of A-functional P̄W weight-average chain length of accu-
groups mulated polymer
NB number of moles of B-functional P̄sol
W weight-average chain length of sol
groups molecules
N B0 initial number of moles of B-functional PDI polydispersity index of polymer pro-
groups duced instantaneously
Ni moles of monomer of type i in the reac- PDI polydispersity index of accumulated
tor polymer
NI number of moles of initiator in the re- PE polyethylene
actor; number of growing chains PEK polyetherketone
N I0 initial number of moles of initiator in PES polyethersulfone
the reactor PETP poly(ethylene terephthalate)
N Ii moles of initiator of type i in the reactor PFR plug-flow reactor
N (r) number chain length distribution PMMA poly(methyl methacrylate)
(number-fraction of polymer molecules PP polypropylene
of chain length r) PPS poly(phenylene sulfide)
PS polystyrene
Polymerization Processes 5

PSD particle size distribution Rp ij consumption rate of monomer of type j


PVAL poly(vinyl alcohol), partially hydro- by propagation with polymeric radicals
lyzed of type i
PVC poly(vinyl chloride) Rp,o initial polymerization rate

P∗ polymeric active center Rr polymeric radical of chain length r

[P∗] concentration of polymeric active cen- [Rr ] concentration of polymeric radicals of
ters (ionic or radical type) chain length r
P∗i polymeric active center with active cen- Rt rate of bimolecular termination
ter located on monomer of type i chem- Rtc rate of bimolecular termination by com-
ically bound in the polymer chain bination
P∗ij polymeric active center with active cen- Rtd rate of bimolecular termination by dis-
ter located on monomer of type j which proportionation
is adjacent to monomer of type i chem- RTD residence time distribution

ically bound in the polymer chain [R ]w concentration of radicals in the aqueous
P∗m,n,i polymer chain containing m units of phase
monomer 1, n units of monomer 2, with ∆S 0 entropy change of polymerization at the
active center on monomer i standard state
Qi i-th moment of the dead polymer distri- S surface area of polymer particle
bution S surfactant
r polymer chain length SAN styrene – acrylonitrile copolymer
r polymer particle radius SBR styrene – butadiene rubber
rM micelle radius SCSTR segregated CSTR
rp polymer particle radius SSH stationary-state hypothesis
r 1 , r 2 reactivity ratios t time
R gas constant t 1/2 half-life of initiator

R polymeric radical t1 polymer particle nucleation time

[R ] concentration of polymeric radicals ts time when polymer particle nucleation
Re Reynolds number ceases
Rf rate of chain transfer T temperature
RFP (r) production rate of polymer molecules T chain-transfer agent

with chain length r T CTA transfer radical
Ri rate of radical entry into polymer parti- Tc ceiling temperature
cles Tg glass transition temperature
R∗in initiator/catalyst fragment with an ac- TPA terephthalic acid

tive center TREF temperature rise elution fractionation
R̂in initiator radical with a peroxide end U MWD nonuniformity index
group v volume of polymer particle

Rin initiator or primary radical v volumetric flow rate into and out of re-

[Rin ] concentration of primary radicals actor
Ri,w radical generation rate in the aqueous vc capture rate of radicals by polymer par-
phase via initiator decomposition ticles
RI rate of initiation (rate of generation of vf flocculation rate of precipitated (pri-
polymeric radicals with chain length mary) polymer particles
unity via initiation) V volume of reacting mixture in the reac-
RIM reaction injection molding tor
Rp rate of polymerization (monomer con- V0 intial volume of reacting mixture in the
sumption rate via propagation reac- reactor
tions) V in total volumetric flow rate into the reac-
Rp i consumption rate of monomer of type i tor
via propagation reactions V out total volumetric flow rate out of the re-
actor
6 Polymerization Processes

Vm specific volume of monomer σ SW interfacial tension between solid and


Vp specific volume of polymer water
Vs volume of solvent in the reactor σ WG interfacial tension between water and
V s, in volumetric flow rate of solvent into the gas
reactor σ WM interfacial tension between water and
VCM vinyl chloride monomer monomer
We Weber number τ kinetic parameter (dimensionless)
W̄ g weight fraction of gel τ mean residence time
W (r) weight chain length distribution ϕ phase volume ratio
(weight fraction of polymer of chain ϕ, ϕ′ probability of propagation
length r) ϕm volume fraction of monomer
W (r), W (r, t) “instantaneous” weight chain ϕp volume fraction of polymer

length distribution ϕi number fraction of polymeric radicals
W̄ (r), W̄ (r, t) weight chain length distribution of type i (terminal model)
of accumulated polymer ϕ∗i number fraction of active or live poly-
W 1 , W 2 weight fractions of homopolymers 1 mer molecules of type i (radical or
and 2 ionic, terminal model)

x monomer conversion ϕij number fraction of polymer radicals of
X• small molecule with a labile atom type ij (penultimate model)
X transfer radical Φ kinetic parameter (dimensionless)
z exponent indicating dependence of N p χ Flory – Huggins polymer – solvent in-
on emulsifier and initiator concentra- teraction parameter
tions ψ (r) number fraction of polymeric radicals
α stoichiometric imbalance of chain length r
∆α1 , ∆α2 differences between thermal expan-
sion coefficients above and below T g
for homopolymers 1 and 2
β kinetic parameter (dimensionless) 1. Introduction–Trends in Polymer
γ kinetic parameter (dimensionless) Reaction Engineering
∗γ prec volume fraction of precipitant
δ kinetic parameter (dimensionless) The worldwide production of synthetic poly-
ε characterizes the radical capture effi- mers, estimated at ca. 100 × 106 t/a in 1990 [1]
ciency of latex particles relative to mi- and at ca. 170 × 106 t/a in 2000 [957], continues
celles; energy-dissipation rate to grow in spite of criticism from environmental-
ε mean rate of energy dissipation per unit ists. Polymer waste has become an urgent topic
mass for industry, providing new and challenging ar-
η moles of monomer consumed per active eas of research and development on recycling,
site (≡ P̄N ) reuse, and degradation. The technical principles
ηc viscosity of continuous phase of polymer reaction engineering will no doubt
ηd viscosity of disperse phase play a significant role in the solution of some
̺c density of continuous phase of these problems. With the increase in produc-
̺d density of disperse phase tion volumes of commodity polymers (LLDPE,
̺el elastic cross-link density HDPE, PP, PVC, and PS copolymers), large-
̺m i density of monomer i reactor technology (suspension PVC) and con-
̺p density of polymer tinuous processes (production of LLDPE in con-
σ standard deviation tinuous fluidized-bed reactors, e.g., UNIPOL
σ2 statistical variance process) are being developed [1], [2]. In the early
σ interfacial tension days of the polymer industry, polymers were
σ SG interfacial tension between solid and specialty materials, produced in batch reactors
gas by using faithfully followed recipes scaled up
σ SM interfacial tension between solid and from the chemist’s beaker. The process engi-
monomer neer, although versed in the principles of chem-
Polymerization Processes 7

ical reaction engineering, had little background the models, groups of rate constants often appear
in polymer chemistry, polymerization kinetics, when calculating rates and polymer properties.
and polymer characterization techniques. This A knowledge of the Arrhenius equation (over-
has changed dramatically in the last two decades, all frequency factor and activation energy) is
as evidenced by the rapid growth of the field of usually sufficient. Another factor which should
polymer reaction engineering within the chemi- be noted is that process models, no matter how
cal reaction engineering discipline. Process pa- detailed, cannot track polymerization rates and
rameters, such as residence-time distribution, polymer properties in real time without feed-
micromixing, and segregated flow, whose in- back from online sensors. The variability in trace
fluence on productivity and selectivity of small impurity levels cannot be accounted for with-
molecule reactions has been studied for many out periodically adjusting kinetic parameters in
years, appear to be far more important for poly- a process model. The great effort made by chem-
merization reactors in that they influence poly- ical kineticists to measure individual elemen-
mer properties dramatically [1–7]. tary rate constants are not in vain. Elementary
The development of engineering and spe- rate constants can be related to the structure of
cialty polymers with a better balance of prop- the reactants, but more importantly for process
erties or with a particular unique property has modelers, elementary rate constants can be used
been growing rapidly. In this regard, it has been to discriminate kinetic models (for example, the
found to be often more economic to produce terminal and penultimate models in copolymer-
a new polymer from existing commodity poly- ization). At this point it is appropriate to em-
mers rather than to start with a new monomer phasize the need for on-line sensors to monitor
and produce polymer in the usual manner. Tech- polymer properties so that process models can
niques such as polymer alloying and blending be used more effectively in state estimation and
are particularly attractive. These and other tech- control.
niques use chemical modifications of existing
polymers by chain scission, long-chain branch-
ing, cross-linking and grafting. These chemical 2. Polymerization Mechanisms and
modifications are usually carried out with poly- Kinetics
mer melts in an extruder reactor [1]. This pro-
cess is often called reactive polymer processing. Polymerization reactions can be classified as ei-
This is a new and commercially promising area ther step-growth or chain-growth reactions. It
where the principles of polymer reaction engi- has been proposed that these mechanisms should
neering could be profitably exploited. be termed random and sequential polymeriza-
Since 1980, modeling of polymerization re- tions [18], [19] since these terms have more sig-
actors has become more comprehensive. Inter- nificance statistically and are devoid of infer-
est has focussed on the prediction of polymer ence concerning the chemistry of the reactions
properties (chemical composition and molecular involved in the polymerizations. In this article,
mass distribution, long-chain branching, cross- however, the conventional terms step-growth
link density, polymer particle size distribution, and chain-growth polymerization are used. It is
and particle morphology). To develop a pre- important to note that this is a classification of
dictive model, account must be taken of the reaction mechanisms, not of the structure of the
chemistry and physics of all of the relevant mi- repeating unit, since many polymers can be syn-
croscopic processes which occur in the poly- thesized either by step-growth or chain-growth
merization process. Detailed physical property polymerization. Generally, however, polymer
and thermodynamic data on the partitioning of physical properties can differ significantly de-
species among phases is required to quantita- pending on the polymerization mechanism, and
tively calculate the concentrations of reactants this is often due to the difference in molecu-
at the loci of polymerization. Valid kinetic rate lar masses, i.e., polymers synthesized by chain-
constants (frequency factors and activation en- growth polymerization often have higher molec-
ergies) are also required. In this regard, one ular masses.
should note that the values for individual ele- These two types of growth reaction differ
mentary rate constants are often not required. In basically in terms of the time-scale of vari-
8 Polymerization Processes

ous reaction events, namely, the size of poly- hours. Chain-growth polymerizations require an
mer molecules increases at a relatively slow active center, which may be a free radical, cation,
rate over a much longer period of time in step- or anion. Once an active center is created, a poly-
growth polymerization. With step-growth poly- mer chain grows extremely rapidly, and when the
merization, the reactions that link monomers, growing chain is deactivated by a termination re-
oligomers, and polymers involve the same reac- action, the polymer chain is dead and no longer
tion mechanism, and any two molecular species takes part as a reactant. With free-radical poly-
(monomer, oligomer, or polymer) can be cou- merization, however, the so-called dead polymer
pled. The growth of a polymer chain proceeds chain is not always truly dead because under cer-
slowly from monomer to dimer, trimer, tetramer, tain circumstances it may itself react with rad-
and so on, until full-sized polymer molecules are icals. The active center may initiate the growth
formed at high monomer conversions. Polymer of many polymer chains.
chains continue to grow from both ends through-
out the polymerization and, therefore, both chain
lifetimes and polymerization times are usually
of the order of hours.

Figure 2. Number chain length distribution in linear step-


growth polymerization

2.1. Step-Growth Polymerization


2.1.1. Linear Polymerization
Figure 1. Linear polymers produced via step-growth poly- Figure 1 shows some representative linear poly-
merization mer chains produced commercially by step-
On the other hand, in chain-growth polymer- growth polymerization. In step-growth polymer-
ization, polymer molecules generally grow to ization, there is generally only one type of chem-
full size in a time-scale which is much smaller ical reaction which links molecules of all sizes.
than the time required for high conversion of Some of the typical chemical reactions are es-
monomer to polymer. The lifetime of a growing terification, amidation, the formation of ure-
polymer molecule may be less than a few sec- thanes, and aromatic substitution. The growth
onds for a free-radical polymerization, which is reaction in step-growth polymerization can be
a typical example of chain-growth polymeriza- represented by the general reaction
tion, while a typical polymerization time to ob- m mer + n mer −→ (m + n) mer (2.1)
tain high monomer conversion may be several
Polymerization Processes 9

The kinetic study of such reactions would be above example), namely, the number-average
extremely difficult if the rate constant for the chain length P̄N and the conversion of func-
coupling reaction depended on the size of both tional groups p for linear step-growth polymer-
species. Fortunately, various kinetic studies have ization was first derived by Carothers [20]. P̄N
shown that the rate constant is effectively in- is simply given as the total number of monomer
dependent of chain length except perhaps for molecules initially present divided by the total
oligomers. This is often referred to as the con- number of molecules present at time t.
cept of equal reactivity of functional groups.
P̄N =NA0 /NA = 1/ (1 −p) (2.6)
Consider the example of step-growth poly-
merization shown below.

nA−A+nB−B→—
[ A − A − B − B—

] n (2.2)

In the case of polyesterification of a diol and a


diacid, A may be a hydroxyl group and B may
be a carboxyl group, although the low molecular
mass condensation byproduct is not shown. As
will be shown later, an almost exact equivalence
in the number of functional groups is necessary
to obtain polymers with high molecular mass,
although a nonstoichiometric condition may be
used to control molecular mass. In the case of ex-
act stoichiometric ratio of the two types of func-
tional groups, i.e., [A] = [B], the polymerization
rate or the rate of disappearance of functional
Figure 3. Weight chain distribution in linear step-growth
groups is given by polymerization
1 d (V [A])
− =K [A]2 (2.3) Equation (2.6) shows that very high conver-
V dt
sions are necessary to obtain large chain lengths.
except for self-catalyzed polymerization, in For example, P̄N = 100 requires a conversion of
which case the rate is third order in monomer 99 %. Equation (2.6) assumes a stoichiomet-
(the self-catalyzed polymerization may not be a ric ratio of unity. If a slight excess of one bi-
useful reaction from the practical point of view functional monomer is used, all chain ends will
of productivity). Neglecting the volume change eventually consist of the group present in excess.
during polymerization, integration of Equation When N A0 < N B0 , the stoichiometric imbalance
(2.3) gives α is given by α = N A0 /N B0 . The total number
of monomer molecules initially present is given
1/ (1 −p) = 1 +K [A]0 t (2.4)
by (N A0 + N B0 )/2 = N A0 (1 + 1/α)/2. Now con-
where [A]0 is the initial (at t = 0) concentration sider the situation at conversion p (p is usually
of A groups, and p is the conversion of functional defined with respect to the deficient group, so
groups, which is defined as that p is defined for A groups in this exam-
 ple). Since each chain end is an unreacted func-
p = NA0 −NA /NA0 (2.5) tional group, the total number of end groups is
[N A0 (1 − p) + N B0 (1 − αp)]. Each molecule
where N A0 and N A are the total number of moles
possesses two end groups, so that the total
of A groups at t = 0 and at any later time t, respec-
number of polymer (plus monomer) molecules
tively. Equation (2.4) has been verified by sev-
is [N A0 (1 − p) + N B0 (1 − αp)]/2. Therefore,
eral kinetic studies. As shown here, the rate ex-
Equation (2.6) can be modified as follows
pression for a step-growth polymerization is the
same for monomer molecules, oligomers, and NA0 (1 + 1/α) /2
P̄N = 
polymers. NA0 (1 −p) +NB0 (1 −αp) /2
The relationship between the average num- 1 +α
= (2.7)
ber of structural units (A – A and B – B in the 1 +α − 2αp
10 Polymerization Processes

As conversion p approaches unity, P̄N ap- reasons why bulk polymerization is quite often
proaches (1 + α)/(1 − α). Thus if α = 0.99, the used commercially for the production of poly-
maximum number-average chain length is only esters and polyamides.
199. This example illustrates the importance of The molecular mass distribution can most
precise control of the stoichiometric ratio to ob- easily be derived by using statistical methods for
tain a desired chain length. a stoichiometric ratio of unity [21]. The conver-
In general, in order to produce high molecular sion p can be interpreted as the probability that
mass polymer by step-growth polymerization, a functional group selected at random has re-
the system must satisfy the following require- acted. Consider the probability that a randomly
ments: selected molecule consists of r monomer units
(this quantity is equal to the number chain length
1) Very accurate control of the stoichiometric distribution). This polymer molecule possesses
ratio of functional groups r − 1 reacted functional groups, and one unre-
2) Absence of side reactions acted functional group. Therefore, the number
3) Availability of high-purity monomers chain length distribution, N (r) is given by
4) Reasonably high polymerization rate
5) Little tendency towards cyclization reactions N (r) =p(r−1) (1 −p) (2.8)

The weight chain length distribution W (r) is


given by

r N (r) =r p(r−1) (1 −p)2 (2.9)
X
W (r) =r N (r) /
r=1

The number and weight chain length distribu-


tions are shown schematically in Figures 2 and
3, respectively. The weight-average chain length
is given by

X
P̄W = r W (r) = (1 +p) / (1 −p) (2.10)
r=1

Since the number-average chain length is


given by Equation (2.6), the polydispersity in-
dex, PDI = P̄W /P̄N , is given by (1 + p), and there-
fore the PDI approaches two as complete con-
version is approached.
Various statistical treatments other than that
shown above have been developed to calculate
the molecular mass distribution for linear step-
growth polymerization [22–26]. Although these
statistical methods appear to work well, kinetic
approaches based on the use of material balances
may have greater generality [27–33]. For an A–
Figure 4. Nonlinear (network) polymers produced via step-
growth polymerization B type monomer in a batch reactor, Equations
(2.8) and (2.9) can also be derived from the fol-
Since high molecular mass polymer is not lowing infinite set of differential equations.
produced until nearly complete conversion of
monomer has occurred, the viscosity is relatively d [P1 ] /dt = −2 K [P1 ] [P] (2.11)
low throughout most of the conversion range. m−1
Thermal control and mixing is not overly dif- d [Pm ] /dt =K
X
[Pr ] [Pm−r ]
ficult, which is opposite to the case for chain- r=1
growth polymerization. These are some of the − 2 K [Pm ] [P] (m ≥ 2) (2.12)
Polymerization Processes 11

where [Pm ] is the concentration of polymer 2.1.3. Nonlinear Polymerization


molecules with chain length m, and [P] is the
total concentration of polymer and monomer. Another important class of polymers pro-
For example, it is straightforward to derive duced by step-growth polymerization are non-
the molecular mass distribution for the cases linear polymers formed by polymerization of
in which a slight amount of monofunctional monomers with more than two functional groups
reagent is used. Kinetic approaches would be per molecule. Some of the nonlinear polymers
easier to apply to reactors other than batch re- produced commercially by step-growth poly-
actors, such as semi-batch and continuous flow merization are shown in Figure 4.
reactors, although a statistical derivation for a In the course of network formation, a polymer
stirred-tank reactor has been reported [34]. molecule of effectively infinite molecular mass
may be formed. At this point, termed the gel
point, the visible formation of a gel or insoluble
2.1.2. Interfacial Polymerization polymer fraction is observed. The gel molecule
is insoluble in a good solvent even at elevated
Interfacial polymerization may provide a temperatures under conditions at which degrada-
method to produce very high molecular mass tion does not occur. Various physical properties
polymers by step-growth polymerization [35], of the system change abruptly at the gel point.
[36]. In interfacial polymerization, polymers are Gelation should be understood as a critical phe-
formed at or in the vicinity of the phase bound- nomenon having similarities with other critical
ary of two immiscible monomer solutions. This phenomena such as vapor – liquid condensation,
technique requires an extremely fast polymer- nuclear chain reactions, and ferromagnetism.
ization. The best reaction type for step-growth It was Carothers [20], who first derived an
polymerization would be Schotten – Baumann equation for the extent of reaction at the gel
reactions involving acid chlorides. For example, point. He defined a gel molecule as one with infi-
polyamidation is performed at room tempera- nite molecular mass. His criterion that gelation
ture by placing an aqueous solution of diamine occurs when the number-average chain length
over an organic phase containing the diacid P̄N goes to infinity is not acceptable, since poly-
chloride. The polymer formed at the interface mer molecules larger than P̄N are always present
can be pulled off as a continuous film or fil- and will become gel molecules earlier than this
ament. The amine – acid chloride reaction rate hypothetical gel point. However, the concept
is so fast that the polymerization becomes dif- of the “infinitely large molecule” was fully es-
fusion controlled. Once the polymer molecules tablished by Flory [37–39] using a statistical
begin to grow and monomer molecules start to approach. His criterion for the onset of gela-
add to polymer chain ends, incoming monomer tion is that it occurs when the weight-average
molecules tend to react with polymer chain ends chain length P̄W goes to infinity. Since a gel
before they can penetrate through the polymer molecule is the largest molecule in the reaction
film to start the growth of new chains. Thus, system, higher-order moments of the molecu-
polymers with much higher molecular masses lar mass distribution could also be used to de-
are formed. Since the reaction is diffusion con- termine the gel point. Fortunately, the second-
trolled, there is no need to start with an exact and higher-order moments approach infinity si-
balance of the two monomers. The lower tem- multaneously, at least for batch polymerizations
peratures used reduce the relative rates of side [40], [41], and the criterion of infinite P̄W is ac-
reactions, and, therefore, the purity of monomers ceptable.
is not as important as with most other step- Flory devised a simple tree-like model, as
growth polymerizations. In spite of the advan- shown in Figure 5, and used the following sim-
tages that interfacial polymerization offers, this plifying assumptions:
process has not attracted wide industrial use,
mainly because of the high cost of the required 1) All functional groups of the same type are
reactive monomers and the large amount of sol- equally reactive
vent which must be removed and recovered. 2) All functional groups react independently of
one another
12 Polymerization Processes

3) No intramolecular reactions occur in finite a molecule with many functional groups, the gel
species once formed acts like a giant sponge, rapidly
consuming sol polymer molecules.
This tree-like concept was generalized by
Gordon et al. [45], [46] based on the theory of
stochastic branching processes, which is consid-
ered to be a part of Graph Theory [47], [48]. This
technique involves abstract mathematics and re-
quires the derivation of the probability generat-
ing functions. The method is general but rather
difficult to use for real problems. To avoid the use
of probability generating functions, other prob-
abilistic methods have been proposed [49–52].
Figure 5. A schematic drawing of Flory’s tree-like model Among them the Macosko – Miller model [50–
(functionality f = 3) 52] using conditional probabilities is becoming
The tree-like model is called the Bethe Lattice or Cayley popular due to its simplicity. All the models
tree by physicists mentioned above are fully equivalent, that is,
only the mathematical language is slightly dif-
ferent. These statistical models, which are some-
times called the classical theories, have a long
history and have proven their power of refin-
ability to accommodate highly system-specific
effects such as unequal reactivity [53], [54], sub-
stitution effects [55], [56], and intramolecular
reactions [18], [19], [57–59], which are impor-
tant in real systems.
One drawback of the statistical theories men-
tioned above is that they assume an equilib-
rium system (i.e., the size distribution is calcu-
lated anew at each time) and they do not con-
sider the kinetic buildup of the system. There-
fore, the classical theories may not be appli-
Figure 6. Molecular mass change and gel growth during cable for kinetically-controlled systems. It has
network formation (functionality f = 3) been shown that although there is no difference
between equilibrium and kinetically-controlled
His basic proposal was that the gel point systems under Flory’s simplifying assump-
is reached when the expectancy of finding the tions, the difference becomes significant as the
next generation in a particular existing molecule conditions deviate further from Flory’s as-
is unity. For the tri-functional monomer units sumptions [60], [61], so that in real systems the
shown in Figure 5, the conversion at the gel kinetic features may be dominant. It has been
point is given by pc = 1/( f − 1) = 0.5, where f is argued that the kinetic buildup can also be ac-
the functionality of a monomer unit. His model counted for by using a statistical approach [62].
was a brilliant development and it provides the Another disadvantage of statistical ap-
starting point for most theories of polymer net- proaches may be the excessive modifications re-
work formation. A few years later Stockmayer quired to generalize them for different reactor
[42–44] further developed Flory’s idea based types (e.g., continuous reactors). The kinetics
on the most-probable size distribution, and their approach was originally shown in the appendix
theory is usually called the Flory – Stockmayer of a paper by Stockmayer [42]. Based on the
theory. Examples of the calculated development chemical kinetics, the reaction rate would be
of the number- and weight-average chain lengths proportional to the product of the number of un-
of the sol fraction and of the weight fraction reacted functional groups in the respective reac-
of gel are shown in Figure 6. Since the gel is tion partner, so that an infinite set of differential
Polymerization Processes 13

equations similar to Equations (2.11) and (2.12) ity. De Gennes wrote in his book [80] that “it
can be set up. This idea has been applied to poly- took more than thirty years to convince exper-
meric systems [60], [63–70]. imentalists that mean-field theory was wrong”.
However, at present the percolation models are
far from simulating actual network formation
quantitatively, because the bonds are too rigid,
the movement of molecules is too suppressed,
and necessary chemical rules of bond formation
are ignored. The percolation theory is essentially
devoted to describing the behavior near the criti-
cal threshold pcr , where the system-specific fea-
tures are not important.
Network polymers are increasingly used as
engineering materials because of their excellent
stability toward elevated temperature and physi-
cal stress. Since the three-dimensional polymers
are neither soluble nor fusible once made, the fi-
nal stage of polymerization is usually carried out
in a mold of the desired shape.

Figure 7. Example of percolation at the gel point in a square 2.2. Chain-Growth Polymerization
lattice (pc = 0.5) [78]

All the theories mentioned above belong to a Chain-growth polymerization is initiated by a


mean-field theory. On the other hand, the perco- reactive species, R∗in , produced from an initiator
lation theory [71–77], which is considered to be or catalyst I.
equivalent to a non-mean-field theory, has been
I −→ n R∗in (2.13)
applied to polymeric gelation [78], [79]. The per-
colation theory is usually associated with a lat- Depending on the type of active center, chain-
tice model to describe network structure. One growth polymerization can be divided into free-
of the simplest examples is the two-dimensional radical, anionic, and cationic polymerization.
lattice shown in Figure 7. In this figure, each The reactive species R∗in adds to a monomer
bond which has been formed is shown as a to form a new active center, and monomer
short line connecting two monomers, though the molecules are added to the active center suc-
monomers are not shown. In the random (stan- cessively. This process is called the propagation
dard) percolation theory each site of a very large reaction:
lattice is occupied randomly with probability
p, independent of its neighbors. Some nearly
“infinite” molecules can be seen in Figure 7,
where “infinite” means that they span the whole where M represents a monomer molecule, and
sample. Mathematical methods to calculate this P∗r is an active polymer molecule with chain
threshold exactly are restricted so far to two di- length r. In general, the propagation reaction is
mensions [77], and therefore, for practical cal- represented by
culations the Monte Carlo simulation is usually
used. It is easy to understand why gelation is
a critical phenomenon from the lattice model,
because in the vicinity of the gel point only a
In chain-growth polymerization, only
few additional bonds are necessary to form a
molecules with an active center can propa-
molecule which spans the whole sample. The
gate, so that polymer molecules once formed
percolation theory emphasizes the universality
may be considered dead polymer for linear
of critical phenomena and space dimensional-
chain-growth polymerization. Dead polymer
14 Polymerization Processes

molecules do not take part as reactants there-


after. The active center is always on the chain
end when linear chains are being produced ex-
clusively. Polymer chain growth is terminated at
some point by unimolecular and/or bimolecular
termination. Bimolecular termination of active The stoichiometric coefficient n is two for
centers occurs only in free-radical polymeriza- thermal decomposition of initiators. A free-

tion. radical Rin derived from the initiator is called
Carbon – carbon double bonds and the a primary or initiator radical.
carbon – oxygen double bond in aldehydes and 2) Propagation reactions, which are responsible
ketones are the two main types of func- for the growth of polymer chains by addition
tional groups which undergo chain-growth of monomer to a radical center.
polymerization. The polymerization of the
carbon – carbon double bond is much more im-
portant, as most commercial monomers with
carbon – carbon double bonds readily undergo
3) Bimolecular termination reactions between
free-radical polymerization (an important ex-
two radical centers, which give a net con-
ception is propene). The carbonyl bond is not
sumption of radicals. These consist of dis-
generally susceptible to polymerization by rad-
proportionation (Eq. 2.19) and combination
ical initiators due to its highly polarized struc-
(Eq. 2.20).
ture. Another reason is that most of the carbonyl
monomers (except formaldehyde) possess very
low ceiling temperatures [81], [82] (the tempera-
ture above which active polymer chains depoly-
merize rather than grow).
Most of the commercial vinyl monomers
(CH2 =CHX and CH2 =CXY, and monomers in where Pr is a polymer molecule of chain
which fluorine is substituted for hydrogen) can length r and does not have a radical center,
be polymerized with free radicals. Whether a while a polymer radical (or macroradical) of

vinyl monomer can be polymerized by anionic chain length r has the symbol Rr .
or cationic mechanisms strongly depends on 4) Chain transfer to small molecules which
the type of monomer. Monomers with electron- causes the cessation of growth of poly-
donating groups attached to the doubly bonded mer radicals while generating small trans-
carbon atoms form stable carbenium ions and fer radicals simultaneously. Chain-transfer
polymerize best with cationic initiators. Con- reactions do not give a net consumption of
versely, monomers with electron-withdrawing radicals, and if the transfer radicals are as
substituents form stable carbanions and require reactive as polymer radical (or more reac-
anionic initiators. It should be noted that ions tive) these reactions should not affect the
of low stability would be expected to react with polymerization rate or monomer consump-
carbon – carbon double bonds; however, in many tion rate when the bimolecular termination
cases they cannot be formed or else are easily reactions are chemically controlled. Chain-
consumed by side reactions. transfer reactions to small molecules re-
duce the size of polymer radicals and there-
fore would increase bimolecular termination
2.2.1. Free-Radical Polymerization rates when these reactions are diffusion con-
trolled (bimolecular termination rates may
Generally, free-radical polymerization consists be chain-length dependent under these con-
of four types of elementary reaction. ditions).

1) Initiation reactions, which continuously gen-


erate radicals during polymerization.
Polymerization Processes 15

where K d is a thermal decomposition rate con-


stant with units of inverse time, most often s−1 .
X may be monomer, a solvent molecule, or For a batch reactor, the change in the number of
a chain-transfer agent. When X is a poly- moles of initiator N I is given by
mer molecule, polymer molecules with long-
dNI /dt = −Kd NI (2.24)
chain branches are formed. Long-branch for-
mation is discussed in Section 2.2.1.7. For isothermal decomposition (isothermal poly-
merization) Equation (2.24) can be integrated
The sequence of elementary reactions, in analytically to obtain
Equations (2.16) – (2.22) results in total radical
concentrations of the order 10−9 – 10−5 mol/L NI =NI0 exp (−Kd t) (2.25)
for most commercial polymerizations. Since where N I0 is the number of moles of initiator at
polymer molecules with high molecular masses time t = 0.
are produced from the very start of polymeriza- Thus the half-life of an initiator is given by
tion, the reacting solution can be quite viscous
t1/2 = − ln (0.5) /Kd = 0.693/Kd (2.26)
over most of the monomer conversion range. The
high viscosities not only cause problems in mix- Knowledge of K d for an initiator therefore per-
ing and heat removal, but also can affect reaction mits calculation of the initiator half-life t 1/2 .
rates (reactions such as bimolecular termination Since K d has an Arrhenius temperature depen-
of polymer radicals). This topic is discussed in dence, K d and t 1/2 both depend on temperature.
Section 2.2.1.3. Activation energies for peroxides and azo ini-
Free-radical polymerization is the most com- tiators are ca. 120 kJ/mol, so the decomposition
monly used method for the synthesis of poly- rate is highly temperature dependent, and the
mers from vinyl and divinyl monomers. Some useful temperature range is quite small (decom-
typical monomers which readily undergo free- position rate is either too fast or too slow outside
radical polymerization are ethylene, styrene, the useful temperature range, which normally
vinyl chloride, vinylidene chloride, acryloni- spans about 30 ◦ C).
trile, vinyl acetate, methyl methacrylate, methyl To complete the initiation step, the initiator

acrylate, acrylamide, etc. Of all chain-growth radicals (Rin ) must add to the double bond of a
polymerization processes, it is the most widely monomer molecule to generate a polymer radi-

studied and best understood. cal of unit chain length Rl . In most polymeriza-
tions, this step (Eq. 2.17) is much faster than the
rate of initiator decomposition (Eq. 2.16). The
2.2.1.1. Initiation homolysis of the initiator is the rate-determining
step in the initiation sequence, and the initiation
Free radicals may be generated in a monomer in rate RI is given by
a number of ways. The most often used method is
RI = 2Kd f [I] (2.27)
to add chemical initiators, such as azo and perox-
ide compounds, to the monomer in low concen- where f is the initiator efficiency. The initiator
trations (usually < 1 wt % based on monomer). efficiency is defined as the fraction of radicals
When heated, the initiator decomposes, gener- produced by initiator decomposition that initiate
ating radicals which act as active centers for polymer radicals. Note that not every initiator
monomer addition. For example, organic per- molecule, which in principle can produce two
oxides (ROOR′ ) decompose thermally by O – O polymer radicals, does so. Some primary rad-
bond cleavage to produce two initiator radicals icals may react with themselves or with other
as follows (other side reactions may of course molecules to form stable species which do not
occur). form either polymer radicals or molecules. It
should be noted that chain transfer to the initia-
tor does not decrease f . The initiator efficiency
16 Polymerization Processes

usually has values in the range 0.2 – 1.0 at low Even when the K d ’s for both peroxide groups
monomer conversions, where polymer concen- are the same, both groups on the same peroxide
trations are low. The major cause of low ini- molecule do not decompose at the same time;
tiator efficiency is recombination of the radical thus a significant fraction of the polymer chains
pairs before they diffuse apart, which is called will have a peroxide end group. These terminal
the cage effect [83], [84]. When an initiator de- peroxide groups will later decompose, generat-

composes, the primary radicals Rin are near- ing polymer radicals with an initiator fragment
−10 −9 in the backbone. The practical benefits include
est neighbors for about 10 – 10 s. During
this interval they are surrounded by a “cage” of higher molecular masses at the same tempera-
solvent and monomer molecules through which ture or comparable molecular masses at higher
they must diffuse to escape from the cage. Since temperatures. Polymerization at higher temper-
reactions between radicals are extremely fast, atures results in higher productivity. These ben-
there is reasonable probability that reaction bet- efits will only accrue when most of the poly-
ween primary radicals occurs. Direct recombi- mer chains are formed by bimolecular termina-
nation may simply regenerate the original initia- tion of polymer radicals. When chain transfer
tor molecules, but other reactions can also occur to small molecules produces most of the poly-
that consume initiator radicals without forming mer chains, these benefits will no longer exist,
polymer chains. In particular, since azo initiators and since bifunctional initiators are more expen-
decompose with the elimination of a nitrogen sive than monofunctional initiators it is recom-
molecule, recombination of the primary radicals mended that the latter be used.
results in the formation of a stable molecule that The decomposition rates of peroxy and azo
cannot generate radicals, and thus there may be compounds can be increased by irradiation with
a significant decrease in initiator efficiency. Ef- ultraviolet and visible light. Unlike thermal
ficiency decreases with increasing viscosity of decomposition, the activation energy for pho-
the reaction medium [85], [86]. Thus f decreases tochemical initiation is approximately zero, so
during the course of polymerization and may ap- polymerization can be initiated at much lower
proach zero at very high polymer concentrations temperatures. Compounds such as benzoin and
where the diffusion coefficient of primary radi- disulfides, whose bonds are too strong to un-
cals in the “cage” is very small [87], [88]. dergo thermal homolysis, are effective radical
When selecting an initiator type, in general initiators under ultraviolet irradiation. Photo-
one needs to consider the decomposition rate chemical polymerization has been applied in
constant, water and oil solubility, stability of ini- coatings and inks for metal, paper, wood and
tiator fragments on chain ends, and other factors. plastics, in photo-imaging, printing circuits, and
Another important point is the activity of the ini- adhesives, although its use is limited by low pen-
tiator radical center towards the abstraction of etration into the polymerizing mass.
atoms (e.g., hydrogen atoms) from the polymer Another method of lowering the activation
backbone. This can lead to chain scission, long- energy of the peroxide decomposition reaction
chain branching, and possibly cross-linking. is to use redox initiation systems. The addition
Up to this point, only monofunctional ini- of a reducing agent results in radical formation
tiators (initiators with one peroxide or one from an oxidation – reduction reaction between
azo group per molecule) have been consid- the two components. Generally, the reaction is
ered. There are commercially available bifunc- illustrated as follows.
tional initiators (with two peroxide groups per •
molecule) with some potentially useful applica- An+ + ROOR′ −→ A(n+1)+ + R′ O + RO− (2.29)
tions [89–92]. Their function can be illustrated
as follows: where A is a reducing agent and ROOR′ is a per-
oxide. The integer n is two for Fe2+ and zero for
N,N-dimethylaniline. The fate of the free radical
depends on the relative concentration of reduc-
ing agent and monomer. Since the redox initiaton
reactions do not produce a pair of radicals, the
cage effect is not operative. At high monomer
Polymerization Processes 17

concentrations most of the free radicals initi- may be formed, leading to a very complex reac-
ate polymerization, and ordinary second-order tion mixture with the formation of many byprod-
kinetics are obeyed. Redox initiation is usually ucts as well as the formation of long branches
used in the temperature range 0 – 50 ◦ C. and possibly cross-linkages. Since photon ener-
All methods mentioned above employ initia- gies for UV are lower, UV radiation generally
tors. There are, however, other means to initiate gives cleaner polymerizations with the forma-
polymerization. Styrene (and some substituted tion of linear chains, although monomers which
styrenes such as p-methylstyrene) and methyl undergo photolysis by UV radiation are limited.
methacrylate polymerize at elevated tempera-
tures in the absence of a free-radical initiat-
ing system. The accepted mechanism for ther- 2.2.1.2. Propagation
mal initiation of styrene is the Mayo mecha-
nism [93], [94], which involves the formation The propagation reaction (Eq. 2.18) controls
of a Diels – Alder dimer intermediate which re- both the rate of growth and the structure
acts with styrene to produce radicals. The Mayo of the polymer chain. Monomers which un-
mechanism is consistent with an observed initi- dergo free-radical polymerization are com-
ation rate which can vary from second to third monly monosubstituted or 1,1-disubstituted eth-
order in monomer during the course of polymer- ylenes, CH2 =CHX or CH2 =CXY. With 1,1-di-
ization [95–97], and some confirming evidence substituted ethylenes both substituents should
has been reported [98–100]. A serious disadvan- not be large, since propagation would be ster-
tage of the use of thermal initiation for styrene is ically hindered. 1,2-Disubstituted ethylenes are
the formation of undesirable byproducts (cyclic normally considered very difficult to polymer-
dimers and trimers) which are difficult to remove ize since the approach of the propagating rad-
to give a high-quality polystyrene. icals to a monomer is sterically hindered. 1,2-
Irradiation with UV, high-energy electrons, Disubstituted ethylenes can, however, often be
and γ-rays can initiate polymerization with incorporated into copolymers.
or without the presence of initiators. Radia- Due to steric and resonance effects, vinyl
tion initiation has been used almost exclusively monomers predominantly undergo head-to-tail
for polymer modification (chain scission, long- addition:
chain branching, cross-linking, and grafting).
These radiation processes are characterized by
a zero activation energy for radical generation
and as a consequence a low activation energy
for polymerization. Therefore, they are effective
In certain cases when the substituents are
at both low and high temperatures. With radi-
small and do not have large resonance stabiliz-
ation initiation, polymer molecular masses in-
ing effects, head-to-head propagation may oc-
crease with increasing temperature, which is op-
cur. For example, approximately 16 % head-to-
posite to that for chemically initiated free-radical
head placement has been reported for poly(vinyl
polymerizations (the high activation energy for
fluoride) [101].
initiator decomposition is responsible for this).
In free-radical polymerization, chain mi-
Both UV and electrons have small penetration
crostructure is largely independent of initiation
depths and are therefore used for polymeriza-
mechanism and initiator type. Polymers pro-
tions in thin layers. Gamma rays have high
duced by free-radical polymerization are largely
penetration depths but require expensive safety
atactic, since the terminal carbon – carbon bond
installations. Radiation polymerization may be
can rotate freely during chain growth. The con-
initiated by radicals, cations, or anions. The ef-
figuration of a monomer unit in the chain is
fectiveness of a radical center depends on the
not determined during its addition to the rad-
chemistry of the monomer and the polymeriza-
ical center but only when the next monomer
tion conditions. Most radiation polymerizations
molecule adds to it. The slight preference for
are free-radical except at very low temperatures
syndiotactic over isotactic placement is caused
where ionic species are sufficiently stable. With
by steric and/or electrical repulsion between
radiation initiation various active intermediates
18 Polymerization Processes

substituents in the chain, although at high tem- It has long been recognized that some bulk
peratures their effects are progressively dimin- polymerizations stop at well below 100 % con-
ished. For example, the fraction of syndiotactic version [110]. This phenomenon has success-
diads of poly(vinyl chloride) changes from 0.67 fully been explained as due to a glassy-state tran-
to 0.51 as the synthesis temperature increases sition of the polymerizing mass [111]. Although
from − 78 ◦ C to 120 ◦ C [102]. For methyl meth- the polymerization can proceed very slowly in
acrylate, it is 0.86 at − 40 ◦ C and 0.64 at 250 ◦ C the glassy state [109], [112], for practical pur-
[103], [104]. poses it can be assumed that polymerization
For most chain-growth polymerizations stops when the system changes from a viscous
(free-radical or ionic) the propagation reactions liquid to a solid glass. It has been proposed that
are reversible at elevated temperature and the the initiator efficiency f approaches zero when
rate of depropagation is significant [105], [106]. the system reaches a glassy state [87], [113] and
that this is mainly responsible for the cessation
of polymerization.

where K dp is the rate constant for depolymeriza- 2.2.1.3. Termination


tion (depropagation). The ceiling temperature
T c , which is the temperature above which active An active center on a growing polymer radical
polymer chains depolymerize rather than grow, may be destroyed by a variety of processes, in-
is reached when the propagation and depropa- cluding termination by added substances. The
gation rates are equal. Based on thermodynamic latter reactions are called inhibition and retarda-
arguments, the ceiling temperature can be re- tion processes, and are not considered here. This
lated to the equilibrium monomer concentration section discusses bimolecular termination reac-
[M]c as follows: tions between polymer radicals. Although one
of the radicals involved in bimolecular termina-
Tc = ∆H/ ∆S 0 +R ln [M]c

(2.32)
tion may be an initiator radical, under normal
where ∆H is the heat of polymerization, ∆S 0 polymerization conditions such reactions may
the entropy change of polymerization at unit be negligible since the concentration of initiator
monomer concentration, and R the gas constant. radicals is much smaller than that of polymer
The ceiling temperature T c is not a singu- radicals.
lar value but is a function of monomer con- Bimolecular termination of two polymer rad-
centration. At any temperature, a concentra- icals can occur by combination or coupling:
tion of monomer exists at which the reaction
in Equation (2.31) is at equilibrium. The exis-
tence of this equilibrium concentration prevents
monomer conversion reaching 100 %.
Normally, this equilibrium monomer concen-
tration is too low to detect. A notable exception is
or by disproportionation, in which case a hy-
α-methylstyrene whose T c for 100 % monomer
drogen radical is transferred from one polymer
concentration is 61 ◦ C, and the equilibrium
chain to the other. The result is the formation
monomer concentration at 25 ◦ C is 2.2 mol/L
of two polymer molecules, one of which has a
[107].
terminal double bond.
Conventionally, it has been assumed that the
propagation rate constant K p is independent of
chain length. Experimental results have shown
that K p is independent of chain length at least
for chain lengths > 16 for styrene and > 62 for
methyl methacrylate [108]. The propagation rate
constant K p is relatively insensitive to the vis- Termination by combination and dispropor-
cosity of the system except at very high polymer tionation can occur simultaneously, and the rel-
concentrations [88], [109]. ative importance of these two modes of termi-
Polymerization Processes 19

nation depends on monomer type and polymer- In general, the termination rate constant
ization temperature. Experimental data are not should depend on the size of the polymeric rad-
available for many monomers; however, radicals ical reactants, the concentration and molecular
which undergo termination by combination ap- mass distribution of the accumulated polymer,
pear exclusively to have the structure (1) [114]. solvent type, and temperature. As the concen-
A well-known example is styrene, which experi- tration of entanglement points increases during
ences termination by combination almost exclu- the course of polymerization, the functional de-
sively over a wide range of temperatures [114], pendence of the termination rate constant on
[115]. On the other hand, radicals which undergo the chain lengths of the polymeric radical reac-
disproportionation and combination may have tants should change dramatically. At some high
the structure (2). monomer conversion (high polymer concentra-
tion) when the polymeric radicals are trapped
(diffusion of the center of mass of polymer
chains is essentially zero), radical centers may
continue to move (due to monomer addition
by propagation) and undergo bimolecular ter-
For methyl methacrylate, combination and dis- mination [118]. With this form of termination,
proportionation are both important at low tem- the termination rate constant should be inde-
perature, with disproportionation becoming the pendent of the chain lengths of the polymeric
dominant mode at high temperatures [116–118]. radical reactants. Originally, upon examination
Since bimolecular termination reactions are of the autoacceleration of the conversion – time
intrinsically very fast, these reactions are likely curve, it was believed that the termination reac-
to be diffusion controlled when they involve tion only became diffusion controlled at some
radical centers on polymeric reactants. The monomer conversion greater than zero and that
autoacceleration of polymerization rate that this occurred when the polymer chains were suf-
is a consequence of diffusion-controlled ter- ficiently entangled (with a sufficient number of
mination is usually called the gel effect or physical entanglement points). It has been rec-
Trommsdorff – Norrish effect. Figure 8 illus- ognized that the bimolecular termination reac-
trates the autoacceleration in rate for the poly- tions may be diffusion controlled even at zero
merization of methyl methacrylate [119]. The monomer conversion (zero polymer concentra-
interpretation proposed was that the increase in tion) [114], [120–122]. At low monomer con-
rate is a consequence of a decrease in the rate versions, where polymer chains in a good sol-
of termination, due to the large increase in vis- vent are isolated coils, translational diffusion of
cosity of the reacting medium, thus giving an the center of mass of the chains is sufficiently
increase in radical concentration. rapid, and the limiting step is the so-called seg-
mental diffusion rate of the radical center in the
coil. With limited interpenetration of the coils, a
finite and significant time is required for the rad-
ical centers on the partially penetrating coils to
meet in a suitable reaction volume where bimo-
lecular termination may occur. The probability
of finding the radical centers in a suitable reac-
tion volume near the coil surfaces decreases as
the coil sizes increase, and, therefore, the termi-
nation constant should increase with decreasing
polymer molecular mass, with decreasing good-
Figure 8. Conversion – time histories for the polymeriza- ness of the solvent (which depends on the nature
tion of methyl methacrylate in benzene initiated by benzoyl and concentration of monomer, inert solvent,
peroxide at 50 ◦ C and polymer), and with increasing polymer con-
The different curves are for various concentrations of
monomer in solvent [119] centration (increasing monomer conversion). It
is well known that in dilute solution the poly-
20 Polymerization Processes

mer coil size decreases with increasing polymer where x is the monomer conversion, K t0 is
concentration. the termination rate constant at zero monomer
At somewhat higher conversions, when there conversion (x = 0), and A1 , A2 , A3 are ad-
are a sufficient number of chain entanglement justable parameters. An application is shown
points, the translational diffusion rate of the cen- in Figure 9. The adjustable parameters A1 , A2 ,
ter of mass of polymer coils decreases dramati- A3 are usually estimated by fitting isothermal
cally and bimolecular termination rates become conversion – time curves. The adjustable param-
translationally diffusion controlled. Of course, eters should be functions of temperature and
shorter polymer chains will experience transla- possibly initiator concentration (radical initia-
tional diffusion-controlled termination at higher tion rate). Note that K t estimated by fitting poly-
monomer conversions (higher polymer concen- merization rate and number-average molecular
trations) than longer chains, and clearly the masses is a number-average termination rate
bimolecular termination rate constant will be constant. Although, this termination rate con-
chain-length dependent and the bivariate distri- stant may predict rates of polymerization and
bution K t (r, s) will change its shape dramat- number-average molecular masses adequately,
ically with increasing polymer concentration. its use to calculate higher averages will under-
Finally, when the polymer coils are trapped, estimate weight-average, and Z-average molec-
K t should become independent of chain length ular mass [88].
[118].

Figure 9. Polymerization of acrylamide – rate and molec- Figure 10. Effect of inhibitors and retarders
ular mass development (T = 60 ◦ C, [M]0 = 3.4 mol/L, initi- a) No retarder or inhibitor ; b) With retarder; c) With in-
ated by potassium peroxosulfate [I]0 = 5.2 × 10−4 mol/L) hibitor
– – – = Constant K t ; —- = Use of Equation (2.35) [126]

The mechanisms of diffusion-controlled re-


actions in polymer systems are being clari- 2.2.1.4. Chain Transfer to Small Molecules
fied both theoretically and experimentally (see
During free-radical polymerization, chain trans-
for example [122]), but at present developing
fer to small molecules X may occur. The small
a general formulation for K t for the whole
molecule may be initiator, monomer, chain-
course of polymerization is a formidable task.
transfer agent, solvent, inhibitor, or impurity. In
Considering the complexity of the mechanisms
general, these chain-transfer reactions can be re-
of diffusion-controlled termination reactions, in
presented by Equations (2.21) and (2.22).
some cases it may be necessary to use an empir-
ical approach for reactor calculations [96], [97],
[123–126]. For example, K t may be approxi-
mated by

Kt =Kt0 exp − A1 x + A2 x2 + A3 x3
 
(2.35)
Polymerization Processes 21

When K ′p is approximately zero (i.e., X is a
stable radical) X is called an inhibitor. If K ′p is
smaller than the propagation rate constant K p ,
X is called a retarder. Idealized behavior of in- Termination by disproportionation
hibitors and retarders is shown schematically in
Figure 10. The kinetics of inhibition and retar-
dation can be found elsewhere [127].
For an added agent to be a chain-transfer Termination by combination
agent, K ′p must be approximately equal to K p
(or K p < K ′p , if chain length is large enough).
Therefore, the chain-transfer agent reduces the
molecular masses but does not affect rates of
polymerization. To derive the kinetic rate equations, the fol-
The chain-transfer rate constants for most lowing assumptions are usually made:
monomers are about 104 –105 times smaller than 1) All rate constants are independent of chain
the propagation rate constant (K fm /K p = 10−5 – length.
10−4 ). 2) Chain lengths are sufficiently large that the
The presence of monomer molecules is in- total rate of monomer consumption may be
evitable, so that the value of K fm /K p places an equated to the rate of monomer consump-
upper limit on the polymer molecular mass that tion by the propagation reactions alone [this
can be obtained with a given monomer. Larger is often called the long-chain approximation
K fm values are observed when the propagating (LCA)].
radicals have very high energies (high reactivi- 3) Radicals generated in chain-transfer reac-
ties), such as in the case of ethylene, vinyl ac- tions propagate with monomer rapidly and
etate, and vinyl chloride. thus do not affect the polymerization rate.
4) The stationary-state hypothesis (SSH) is
valid for radical reactions. One can therefore
2.2.1.5. Kinetics of Linear Polymerization assume that both the rates of radical genera-
The elementary reactions involved in linear free- tion and consumption are much greater than
radical polymerization (chains produced are lin- the rate of change of radical concentration
ear with no branches or cross-links) are as fol- with respect to time [128], [129].
lows: Let us first derive an expression for the poly-
Initiation merization rate Rp , applying the above assump-
tions. The balanced equation for polymer radi-
cals with chain length r is given by
  ∞
1 d V R·l X
= RI +Kfm [M] [R·r ]
V dt r=2
Propagation +Kp′ [T· ] [M] −KfT [R·1 ] [T] −Kp [R·1 ] [M]
− (Ktc +Ktd ) [R·1 ] [R· ] (2.36)

1 d (V [R·r ])  
=Kp R·r−1 [M] −Kp [R·r ] [M]
Chain transfer to monomer V dt
−Kfm [R·r ] [M] −KfT [R·r ] [T]
− (Ktc +Ktd ) [R·r ] [R· ] (r ≥ 2) (2.37)

where RI is the initiation rate (RI = 2 K d f [I])


Chain transfer to small molecule (T) ∞
P
and [R· ] = [R·r ], which is the total polymer
r=1
radical concentration.

The transfer radical con-
centration [T ] is given by
22 Polymerization Processes

1 d (V [T· ]) latter being a direct consequence of the bimo-


=KfT [R· ] [T] −Kp′ [T· ] [M] (2.38)
V dt lecular nature of the termination reaction.
Now consider the weight chain length dis-
Applying the stationary-state hypothesis gives
tribution W (r). Application of the stationary-
KfT [R· ] [T] =Kp′ [T· ] [M] (2.39) state hypothesis for polymer radicals with chain
length r (Eq. 2.36 and 2.37) gives.
Summation of Equations (2.36) and (2.37) over
all chain lengths (1 to infinity) and substituting [R·l ] (2.46)
Equation (2.39) into the sum gives RI +Kfm [M] [R· ] +KfT [T] [R· ]
=
∞ Kp [M] +Kfm [M] +KfT [T] + (Ktc +Ktd ) [R· ]
1 d (V [R· ]) X 1 d (V [R·r ])
=
V dt V dt [Rr· ] (2.47)
r=1 h i
· 2 Kp [M] R·r−1
=RI +Kp′ [T· ] [M] −KfT [R· ] [T] − (Ktc +Ktd ) [R ]
=
=RI − (Ktc +Ktd ) [R· ]2 (2.40) Kp [M] +Kfm [M] +KfT [T] + (Ktc +Ktd ) [R· ]

Application of the stationary-state hypothesis Let us introduce the following dimensionless



for the total polymer radical concentration [R ], groups:
gives Rtd +Rf Ktd [R· ] +Kfm [M] +KfT [T]
τ= = (2.48)
Rp Kp [M]
RI =Kt [R· ]2 (2.41)
Rtc Ktc [R· ]
where K t = K tc + K td . In the above formalism, β= = (2.49)
Rp Kp [M]
the termination rate, Rt is given by
where
Rt =Kt [R· ]2 (2.42) •


Rp = K p [R ] [M]; propagation rate

It is worth noting here that Rt = 2 K t [R ]2 is of- Rtd = K td [R ]2 ; rate of termination by dis-

ten used in the literature, although Rt = K t [R ]2 proportionation

is more widely used in free-radical polymeriza- Rtc = K tc [R ]2 ; rate of termination by com-
tion (e.g., the compilation of kinetic rate con- bination • •
stants in [130]). One must distinguish carefully Rf = K fm [R ] [M] + K fT [R ] [T]; rate of
which type of termination rate constant is being chain transfer.
used when consulting the literature on polymer-
ization kinetics. Since RI = Rtd + Rtc , Equations (2.46) and
From Equation (2.41), the total polymer rad- (2.47) can be simplified as follows:
ical concentration is given by τ +β
[R·1 ] = [R· ] (2.50)
0.5 1 +τ +β
[R ] = (Rl /Kt )
·
(2.43)
1  · 
[Rr· ] = Rr−1 (2.51)
Based on the long-chain approximation, the 1 +τ +β
polymerization rate Rp is given by
Therefore,
1 d (V [M])
Rp = − =Kp [R· ] [M]
V dt [Rr· ] = [R· ] (τ +β) Φr (2.52)
 
Kp
= 0.5
RI0.5 [M] (2.44) where
Kt

Since RI = 2Kd f [I] , Φ = 1/(1 + τ + β)



Kp
 Now consider the production rate of polymer
Rp = (2Kd f [I])0.5 [M] (2.45) molecules with chain length r, RFP (r), which is
Kt0.5
given by
Equation (2.45) predicts a first-order depen-
denceon monomer concentration and a square-
root dependence on initiator concentration, the
Polymerization Processes 23

1 d (V [Pr ]) ∞
RFP (r) =
X τ (2 +τ +β) +β (3 +τ +β)
PW = rW (r) =
V dt (τ +β)2
r=1
= (Kfm [M] +KfT [T] +Ktd [R· ]) [R·r ]
2τ + 3β
r−1 ≈ (2.58)
1 X   (τ +β)2
+ Ktc [R·s ] R·r−s (2.53)
2 s=1
The instantaneous number-average chain length

Substituting for [Rr ] using Equation (2.52) gives PN is given by
1 (1 +τ +β) 1
RFP (r) (2.54) PN = ∞ = ≈ (2.59)
P (τ +β/2) (τ +β/2)

β
 W (r) /r
=Kp [R· ] [M] (τ +β) τ + (τ +β) (r − 1) Φr r=1
2
The polydispersity index PDI for polymer pro-
The instantaneous weight chain length distribu- duced instantaneously is given by
tion W (r) is therefore given by
PW (2τ + 3β) (τ +β/2)
rRFP (r) PDI = ≈ (2.60)
W (r) = ∞
PN (τ +β)2
P
rRFP (r)
r=1 If β = 0, i.e., termination by combination does
not occur, the polydispersity index takes on the
n o
(τ +β) τ + β2 (τ +β) (r − 1) rΦr
= maximum value, PDI = 2. On the other hand, if
1 +τ +β
  τ = 0, i.e., chain termination is solely by bimo-
β
= (τ +β) τ + (τ +β) (r − 1) rΦr+1 (2.55) lecular termination through combination, PDI
2 takes on the minimum value, 1.5. Figure 11
If β ≪ τ , that is most polymer chains are formed shows the PDI as a function of the fraction of
by chain transfer and/or termination by dispro- chain termination by bimolecular termination by
portionation, Equation (2.55) reduces to combination, β/(τ + β).
 r−1  2
1 τ
W (r) =τ 2 rΦr+1 =r (2.56)
1 +τ 1 +τ

Here, 1/(1 + τ ) is the probability of growth for


a given polymer radical, and τ /(1 + τ ) is the
probability that a polymer radical stops grow-
ing. Therefore, Equation (2.56) is essentially the
same as the distribution derived for linear step-
growth polymerization (Eq. 2.9), based on a sta-
tistical argument.
It is sometimes more convenient to describe
the chain length distribution as a continuous
function rather than a discrete function, and this
can often be done with small error as r is usually
very large. Therefore, the following approxima-
tion may be useful:
 
β
W (r) ≈ (τ +β) τ + (τ +β) (r − 1) r Figure 11. Effect of the type of chain termination on the
2
polydispersity index
· exp {− (τ +β) r} (2.57)

Note that (τ + β) has the value of ca. 10−6 – The W (r), PW , PN , and PDI derived here
10−2 for usual free-radical polymerization give the instantaneous properties. In linear free-
[(τ + β) ≪ 1]. radical polymerization the polymer molecules
The weight-average chain length PW for once formed are inert and do not react further.
polymer produced instantaneously is given by
24 Polymerization Processes

In general, since the concentrations of monomer, the effect of temperature can be estimated by
initiator, and chain-transfer agent change with the change in the ratio of three rate constants,
time, the chain length distribution of the ac- K p (K d /K t )0.5 . Since each kinetic rate constant
cumulated polymer is always broader than the is considered to follow the Arrhenius equation,
instantaneous distribution. Particularly, when the activation energy of polymerization E R is
bimolecular termination is strongly diffusion given by
controlled and most of the polymer chains are
Ed Et
produced by bimolecular termination, the accu- ER =Ep + − (2.65)
mulated distribution broadens significantly with 2 2
increasing conversion. The polydispersity index where E p , E d , E t are activation energies for
PDI for commercial polymers is usually larger propagation, initiator decomposition, and bimo-
than two and this is the result of a drift in mo- lecular termination. Typical values for E p , E d ,
lecular mass averages of the instantaneous dis- E t are 30, 120, and 15 kJ/mol, respectively.
tribution. The accumulated distribution and its Therefore, E R is ca. 80 kJ/mol and this is largely
averages can be calculated as follows: due to the very high activation energy for ini-
tiator decomposition. The high activation en-
Zx
1 ergy for polymerization means that the rate of
W̄ (r) = W (r) dx (2.61)
x polymerization increases strongly with increas-
0
Zx
ing temperature. With redox initiation, E d is
1 ca. 40 kJ/mol and therefore E R is considerably
P̄W = PW dx (2.62)
x smaller at ca. 40 kJ/mol. With radiation initia-
0
x tion, E d is close to zero and E R is ca. 20 kJ/mol.
P̄N = (2.63)
Rx 1
Now consider the effect of temperature on the
dx
0
PN molecular mass of polymer obtained. For a sim-
P̄W ple example, consider the case where τ ≪ β, that
PDI = (2.64) is, termination by combination produces most of
P̄N
the polymer chains. Equation (2.58) gives
Superscript bars denote accumulated properties.
These integrals may be replaced with equiva- 1
PW ∝
lent ordinary differential equations which then β
can be solved by using readily available differ- Kp [M] Kp
= ∝ (2.66)
ential equation solvers. The variation of the ki- Ktc (2Kd f [I] /Ktc )0.5 (Ktc Kd )0.5
netic parameters of W (r) with respect to con-
version must be known to calculate the accumu- The activation energy for average chain lengths
lated properties. In a similar manner, accumu- E L is given by
lated molecular mass properties may be calcu- Ed Et
lated for semi-batch and continuous reactors. In EL =Ep − − (2.67)
2 2
a well mixed flow reactor with an ideal residence
time distribution (CSTR), the chain length dis- With initiators E L is about − 40 kJ/mol. The av-
tribution produced is the instantaneous distribu- erage chain lengths decrease significantly with
tion. In batch reactors the distribution is there- increasing temperature when initiators are used
fore broader [131], [132]. Other mathematical and bimolecular termination controls molecular
techniques to derive the distribution functions mass development (most of the polymer chains
can be found elsewhere [29], [32], [33], [133– are produced by bimolecular termination). With
135]. radiation initiation, E L is about 20 kJ/mol and
molecular masses increase moderately with tem-
perature.
2.2.1.6. Effect of Temperature When unimolecular termination (chain trans-
fer to small molecules) dominates in the produc-
In free-radical polymerization initiated by the tion of polymer chains, E L is given by
thermal decomposition of an initiator, the poly-
merization rate is given by Equation (2.45), and EL =Ep −Ef (2.68)
Polymerization Processes 25

where E f is the activation energy for the chain- even near zero conversion and therefore long-
transfer reaction. E f depends on the type of chain branching reactions are more significant
chain-transfer agent, but usually E p − E f < 0, in emulsion than in solution or bulk polymer-
and therefore molecular masses usually decrease ization. Examples of monomers for which long-
with increasing temperature when unimolecular chain branching via chain transfer to polymer is
termination (chain transfer to small molecules) important include ethylene and vinyl acetate.
controls molecular mass development. When long-chain branching is important the
calculation of the full molecular mass distribu-
tion requires excessive computation. However,
2.2.1.7. Branching Reactions the leading moments and molecular mass av-
erages can be readily calculated by using the
For the elementary reactions shown in Sec- method of moments for an infinite set of dif-
tion 2.2.1.4, radical centers are always located ferential equations which describe the chemical
on chain ends, and the dead polymer chains are kinetics [61], [134], [139–147]. The i-th order
chemically inert and are all linear. Branched moment of the polymer distribution can be de-
or cross-linked polymer can be obtained by fined by
chemical treatment of linear chains using pro- ∞
cesses such as vulcanization, radiation, and melt Qi =
X
r i [Pr ] (2.70)
processing with peroxides. This section, how- t=1
ever, is mainly concerned with branching reac-
tions which may occur during polymerization. Therefore, the number- and weight-average
Branched and cross-linked polymers are of sig- chain lengths of the accumulated polymer are
nificant commercial interest, but quantitative in- given by P̄N = Q1 /Q0 and P̄W = Q2 /Q1 , respec-
terpretation has been limited due to a lack of an- tively. For the elementary reactions shown in
alytical techniques for the comprehensive char- Section 2.2.1.4 plus Equation (2.69), some of the
acterization of branched polymers. Some of the lower order moments for batch polymerization
newer analytical techniques are summarized in are given by [61], [145–147]:
[136–138]. 1 d (V Q0 )
=τ +β/2 (2.71)
V0 dx
Chain Transfer to Polymer. Chain transfer
to polymer involves the abstraction of an atom
from the backbone of a polymer chain and re- 1 d (V Q1 )
= 1 (2.72)
sults in the formation of a backbone radical cen- V0 dx
ter. Monomer addition to this radical center pro-
duces a tri-branching point and a long-chain
branch whose average length is equal to that of 1 d (V Q2 ) 2 (1 +Cp2 ) β (1 +Cp2 )2
= + (2.73)
the primary chains produced at the same instant. V0 dx τ +β +Cp1 (τ +β +Cp1 )2

with initial conditions: Q0 = Q1 = . . . . . . = 0 at


x = 0, where
Kfp Qi
Cpi =
Kp [M]
Ktd [R· ] +Kfm [M] +KfT [T]
τ=
Kp [M]
Chain transfer to polymer may be neglected Ktc [R· ]
β=
at low monomer conversions as polymer concen- Kp [M]
trations are low. However, under more practical
and V 0 is the initial volume of the reacting mix-
situations where polymerizations are carried to
ture. Figure 12 shows calculated weight-average
very high conversions, it may be important. In
chain lengths for a range of the kinetic parame-
emulsion polymerization the polymer concen-
ters, keeping τ and β constant with conversion.
tration in polymer particles is relatively high
Note that reactions involving the addition of a
26 Polymerization Processes

polymer radical center to a double bond on the mixture always has a relatively high polymer
end of a polymer chain are not accounted for in concentration. Comparing long-chain branching
the moment equations above (Eq. 2.71 and 2.73). frequencies in batch and CSTR reactors at the
same monomer conversion levels shows that the
residence time distribution of the CSTR favors
branching [150], [151].

Figure 13. Effect of temperature on the number-average


chain length of polyisobutylene and isobutene – isopropene
copolymer [161]
Figure 12. Weight-average chain length development dur-
a) Polyisobutylene with AlEtCl2 as initiator; b) Polyiso-
ing polymerization with chain transfer to polymer [147]
butylene with BF3 as initiator; c) Polyisobutylene with
AlCl3 as initiator; d) Isobutene – isopropene with BF3 as
initiator; e) Isobutene – isopropene with AlEtCl2 as initia-
It is of interest to know whether chain trans- tor; f) Isobutene – isopropene with AlCl3 as initiator
fer to polymer can lead to gelation. Flory [148],
[149] predicted that this reaction type would not
cause gelation without the assistance of other in- When the reactivity of polymer radicals
terlinking processes. By application of Equation is high (as with polyethylene), intramolecu-
(2.73) it has been clearly shown that chain trans- lar chain transfer (transfer to a location on its
fer to polymer plus termination by dispropor- own chain) or backbiting may occur. In this
tionation (i.e., there is no termination by com- case polymer chains with short branches are
bination) can never cause gelation. With termi- formed. In the synthesis of low-density poly-
nation by combination, however, gelation can ethylene (LDPE) by high-pressure free-radical
occur under certain circumstances [61], [145], polymerization, short-chain branching frequen-
[147]. cies are high (typically 20 – 40 branches per
In general, when long-chain branching oc- 1000 backbone carbon atoms). For polyethyl-
curs, the molecular mass distribution broadens. ene, the amount of short-chain branching is
This is especially true for polymerization in a about ten times greater than long-chain branch-
continuous stirred-tank reactor as the reacting
Polymerization Processes 27

ing. For polyethylene the backbiting reaction ganic packagings used in size exclusion chro-
can be represented as follows matography are lightly cross-linked polymeric
networks. The kinetics of copolymerization of
vinyl and divinyl monomers are discussed in
more detail in Section 2.3.3.

2.2.2. Ionic Polymerization

The energy required to form a pair of ions from


a neutral molecule is large, and therefore these
with n = 3 and 4 most probable [152]. very unstable ions must be stabilized by sol-
vation at low temperature before polymeriza-
Reactions with Double Bonds in Polymer tion will occur. Polar solvents cannot be used
Chains. When a radical center on the end of to solvate ions because they are overly reactive
a polymer chain adds to a double bond on the and destroy the ionic initiators. Ionic polymer-
end of a polymer chain, a tri-branching point is izations are usually carried out at low tempera-
formed. ture in solvents of low polarity. These solvents
may give ion pairs as well as free ions. Thus,
a propagating ionic chain may have a counter-
ion close to the active center during its growth.
The proximity of the ion on the growing chain
to its counterion depends on the type of coun-
terion, which is determined by the initiator type,
and the solvation power of the solvent. There-
Terminal double bonds on the ends of poly- fore, unlike free-radical polymerization the type
mer chains may be formed by chain transfer of initiator and the nature of the solvent have a
to monomer, termination by disproportionation, large effect on monomer addition during chain
and by β-scission at backbone radical centers. growth. The propagation rate constant, there-
Reaction with double bonds located on the fore, depends not only on temperature but also
polymer backbone (internal or pendant double on the type of initiator and the type and amount
bonds) leads to formation of tetra-branching of solvent.
centers. For ionic chain-growth polymerizations in
solvents with high solvating power where the
distance between the propagating active cen-
ter and the counterion is large, the factors gov-
erning the stereochemistry are similar to those
for free-radical polymerization. However, in sol-
vents with poor solvating power, there may be
Pendant double bonds are obtained when extensive coordination between initiator, propa-
monomers containing two or more reactive dou- gating chain end, and monomer, which results in
ble bonds are polymerized. For example, the isotactic (or syndiotactic) placements almost ex-
homopolymerization of a diene and the co- clusively, i.e., stereospecific polymerization oc-
polymerization of vinyl and divinyl monomers curs.
produces polymer chains with pendant double Unlike radical polymerization, bimolecular
bonds. The photopolymerization of multifunc- termination between active centers does not oc-
tional monomers is used in surface coatings and cur in ionic polymerization. Termination of an
in the replication of optical discs [153–155]. For active center on a polymer chain occurs by reac-
densely cross-linked systems the buildup of rad- tion with the counterion, solvent, monomer, or
ical concentration is so rapid that the stationary- other species. Often, the initiation reactions are
state hypothesis may not be valid. Superab- very fast, and the initiator is consumed in the
sorbent polymers used in diapers and porous or-
28 Polymerization Processes

early stages of polymerization before the poly- polar impurities and this often precludes the es-
mer chains have grown much beyond oligomeric tablishment of a stationary state. It is also diffi-
size. In the absence of unimolecular termina- cult to establish what proportion of the initiator
tion a population of polymer chains having the produces growing polymer chains.
same molecular mass can grow. The concentra- Cationic polymerization of vinyl monomers
tion of ionic reactive centers is usually much is essentially limited to those with electron-
larger than that of radical centers. Ionic poly- donating substituents such as 1,1-dialkyl,
merization kinetics are not as well understood alkoxy, phenyl or vinyl groups. Cationic poly-
as those radical-based polymerizations because merization involves initiation, propagation, ter-
of the requirements of extreme purity for the mination, and chain transfer to small molecules
components of the reacting mixture. much as in free-radical polymerization.
Elastomers such as butyl rubber and poly- To initiate cationic polymerization, protonic
isoprene; high-density polyethylene; polypro- acids such as H2 SO4 , HClO4 , and H3 PO4 or
pylene and its copolymers are widely produced Lewis acids such as BF3 , AlCl3 , TiCl4 , and
by ionic polymerization. Polypropylene of high SnCl4 are used. Lewis acids are by far the most
molecular mass cannot be produced by radical important initiators for industrial cationic poly-
polymerization. In addition to carbon – carbon merizations. Initiation with Lewis acids requires
double bonds, carbonyl double bonds [81], [82], the presence of a trace of proton donor such as
alkynes [156], and carbon – nitrogen double water, alcohol, and organic acid or a cation donor
bonds [157] can be polymerized via ionic mech- such as alkyl halide.
anisms. However, the discussion here is limited
to carbon – carbon double bonds.

2.2.2.1. Cationic Polymerization

Cationic polymerization proceeds through at-


tack on the monomer by an electrophilic species,
resulting in heterolytic splitting of the double In general,
bond to produce a carbenium ion [158], [159].

Most polymerizations exhibit a maximum


rate at some ratio of initiator and coinitiator con-
centrations. This optimum ratio varies widely
from one initiator system to another and in some
instances the solvent has an effect on the ratio.
The most important commercial high poly- (Note that it has also been proposed that the pro-
mers produced by cationic polymerization are togen or catinogen be referred to as the initiator,
polyisobutylenes and butyl rubber (a copoly- and the Lewis acid as coinitiator [159], although
mer of isobutylene and a 1,3-diene, usually iso- conventional terminology is exactly the reverse.)
prene). A typical polymerization is carried out The propagation reaction involves the succes-
at about − 100 ◦ C in chlorinated solvents such sive insertion of monomers into the partial bond
as chloromethane and is initiated by AlCl3 . The between the propagating species and its coun-
polymerization is very fast. terion.
In the following section, interesting features
of cationic polymerization are discussed, but no
attempt is made to develop quantitative rate ex-
pressions. Carbenium ions are very sensitive to
Polymerization Processes 29

The apparent propagation rate constant K p can produce polymer with sufficient molecular
depends on the type of initiator and solvent used. mass only at very low temperatures. Figure 13
In general, K p becomes larger as the acidity of shows the Arrhenius plot for the number-average
the initiator and/or the dielectric constant of the chain length [161]. There is a change in the slope
solvent increases. This is due to the fact that the of this plot around − 100 ◦ C. This has been at-
reactivity of free ions in monomer addition is tributed to a change in the chain termination step
much greater than that of ion pairs. from chain transfer to monomer below − 100 ◦ C
A wide variety of reactions may lead to ter- to chain transfer to solvent above − 100 ◦ C.
mination of chain growth in cationic polymer-
Table 1. Effect of counterion on the propagation rate constants (K −
p
ization, but it is usually ambiguous and difficult and K ± ◦
p ) in the anionic polymerization of styrene in THF at 25 C
to distinguish termination reactions from chain-
transfer reactions. Chain transfer to monomer is Counterion K±
p K−
p K × 107
most often responsible for the formation of dead Li+ 160 6.5 × 104 2.2
polymer: Na+ 80 6.5 × 104 1.5
K+ 60 – 80 6.5 × 104 0.8
Rb+ 50 – 80 6.5 × 104 0.1
Cs+ 22 6.5 × 104 0.02

If the counterion is sufficiently nucleophilic,


termination by combination may occur:
2.2.2.2. Anionic Polymerization
Anionic polymerizations show many of the
same characteristics as cationic polymeriza-
However, this reaction is not very common.
tions. However, since the nature of carbanions
In fact, true termination in which the activity of
is different from carbenium ions, there are dis-
a chain carrier is lost without regeneration of
tinct differences. In contrast to cationic poly-
an active center is very rare if no impurities ca-
merization, neither termination nor chain trans-
pable of destroying an active center are present.
fer occur in many anionic polymerizations (liv-
However, since the carbenium ion is highly reac-
ing polymerization) especially when polar sub-
tive, it would be a formidable task to remove all
stances are absent. Anionic active centers are
these impurities from the reaction components.
usually much more stable than cationic active
Under commercial conditions it is likely that ac-
centers. Although anionic polymerizations pro-
tive centers are consumed to a significant extent
ceed rapidly at low temperatures, they are not
by impurities. Termination reactions in cationic
usually as temperature sensitive as cationic poly-
polymerizations are unimolecular, as they are
merizations, and polymerizations usually pro-
in anionic and in anionic-coordination polymer-
ceed well at ambient temperature and higher.
ization. Generally, for cationic systems the poly-
A variety of basic initiators have been used
merization rate is given by
to initiate anionic polymerizations [162], [163].
Rp =K [M]1−3 [I] (2.85) The initiation involves the addition of an anion
(base) to the double bond of the monomer.
The first-order rate with respect to initiator con-
centration is a consequence of the fact that ter-
mination reactions are unimolecular. The overall
activation energy for polymerization E R may be
ca. − 40 to 60 kJ/mol [160]. For many polymer-
ization systems E R is negative and the rather un- where the C− · · · G+ bond can have character
usual phenomenon of increasing polymerization ranging from partially covalent to completely
rate with decreasing temperature is observed. ionic. Alkyllithium initiators have been most
The activation energy for the degree of poly- widely used in the polymerization of butadiene
merization is always negative, and therefore the and isoprene, since they are easy to prepare and
average chain length decreases with increasing are soluble in hydrocarbon solvents. A compli-
temperature. Usually, cationic polymerizations cation which arises when alkyllithiums are used
30 Polymerization Processes

in nonpolar solvents such as benzene, toluene, As shown here, although the concentration of
cyclohexane, and n-hexane is association of var- free ions is only 1.2 % of the total propagating
ious organolithium species. This phenomenon species, approximately 90 % of the monomer is
is important, since the associated species are es- consumed by the free ion.
sentially unreactive in propagation [162], [164]. Since carbanions are relatively stable, anionic
Propagation occurs by the successive inser- polymerization with carefully purified reagents
tion of monomers into the partial bond between may lead to systems in which chain termination
the propagating anion and its cationic counter- is absent. Such polymers are referred to as living
ion. polymers [167], [168].
Let us now develop an expression for the
chain length distribution for living polymer. As-
suming that initiation is instantaneous, the to-
tal number of growing chains N I is constant
throughout the polymerization. Now a monomer
molecule adds to a polymer chain as shown in
Figure 14. The probability that it adds to poly-
mer chain (1) is 1/N I , and that it does not add to
The polymerization rate is fundamentally ex- polymer chain (1) is 1 − 1/N I . Therefore, when
pressed by Rp = K p [M] [M− ], where [M− ] is N M monomer units have been consumed and
the concentration of active species. When non- bound into polymer chains the probability that
polar solvents such as dioxane are used, all active a randomly selected polymer chain possesses r
species may be ion pairs. However, in polar sol- monomer units is given by a binomial distribu-
vents such as THF, the effect of dissociation to tion:
free ions on polymerization rate cannot be ne- 
NM

1
r 
1 NM −r

glected. Therefore, the propagation rate is given N (r) = 1− (2.92)
r NI NI
by the sum of the rates for free propagating
anions (R− ) and for ion pairs, R− (G+ ). N (r) is also the number chain length distribu-
tion. Since 1/N I ≪ 1 and r is large, the binomial
Rp =Kp− R− [M] +Kp± R− G+ [M]
   
(2.88) distribution reduces to the Poisson distribution:

The two propagating species are in equilibrium, e−η η r


N (r) = (2.93)
r!
where η = N M /N I , which is equal to the number-
average chain length, P̄N .
where K is the dissociation constant and is given
by

K = R− G+ / R− G+
    
(2.90)

Some experimental values for K ± p and K p


are listed in Table 1 [165], [166]. Consider


the simple example of the anionic polymeriza-
tion of styrene in THF at 25 ◦ C, initiated by
sodium naphthalide with an initial concentration
[I] = 1 × 10−3 mol/L. Assuming the initiation
reaction is instantaneous, [R− ] = 1.22 × 10−5
and [R− (G+ )] = 9.88 × 10−4 mol/L. There-
Figure 14. Schematic drawing for the derivation of chain
fore, length distribution in living polymerization

Kp− R− [M] 6.5 × 104 1.22 × 10−5


   
=
The weight chain length distribution W (r)
Kp± [R− (G+ )] [M] (80) 9.88 × 10−4

and the weight-average chain length P̄W are
= 10.0 (2.91) given by
Polymerization Processes 31

rN (r) e−η η r−1 total exclusion of the others by an appropriate


W (r) = ∞ = (2.94) choice of initiator system, as shown in Table 2
P (r − 1) !
rN (r)
r=1 [169].

The ZN initiators are the only ones which can
be used to polymerize α-olefins such as propene
X
P̄W = rW (r) =η + 1 (2.95)
r=1 and 1-butene. Although the phenomenon of
stereoisomerism is not applicable to the symmet-
Therefore, the polydispersity index PDI is given rical ethylene, the use of ZN initiators produces a
by polyethylene with much less long-chain branch-
PDI =P̄W /P̄N = 1 + 1/P̄N (2.96)
ing than that obtained by free-radical polymer-
ization.
With sufficiently large average chain length, the The ZN initiators consist of a combination of
polydispersity should approach unity. To obtain alkyls or hydrides of group 1 – 3 metals with salts
a narrow molecular mass distribution (MWD), of group 4 – 8 metals. Some of the components
the rate of initiation should be greater than the of ZN initiators are as follows:
rate of propagation. Polystyrenes having poly- Group 1 – 3 metal Transition metal
dispersity as low as 1.01 have been synthesized (C2 H5 )3 Al, (C2 H5 )2 AlCl TiCl4 , TiCl3 , VCl3
by using sodium naphthalide as initiator, and (C2 H5 )AlCl2 , (C2 H5 )2 Be VOCl3 , Ti(OC4 H9 )4
these are widely used in the calibration of gel C5 H11 Na CrCl3 , MnCl3

permeation chromatography. It is of interest to


note that narrow MWD can be obtained even The titanium – aluminum initiator systems,
when there are two reactive site types (free ion especially (C2 H5 )3 Al – TiCl3 , have been most
and ion pair). This is a result of the equilibrium thoroughly studied. Many important ZN initia-
between the site types, and growing chains will tors are solids and during polymerization are
have both site types attached to their ends for the suspended in liquid or gaseous media. Hetero-
same fraction of time during their growth. geneous initiator systems appear to be neces-
Sequential addition of monomers to a liv- sary for the production of isotactic polyolefins,
ing anionic polymerization system is at present although soluble initiators are used for the syn-
the most useful method for synthesizing well- thesis of syndiotactic polypropylene.
defined block copolymers. Discussions on the mechanism of ZN poly-
merization can be found elsewhere [169–173].
At present, however, none of the proposed mech-
2.2.2.3. Ziegler – Natta Polymerization anisms have been comprehensively verified and
the observed kinetics are usually quite complex.
Due to the interactions between the propa- Some examples of the initial stages of heteroge-
gating chain end, counterion, and incoming neous ZN polymerization are shown in Figure 15
monomer molecule, ionic polymerizations tend [174]. The particles of the transition metal com-
to give stereospecific polymers. Polar monomers ponent usually consist of aggregates of smaller
such as methacrylates and vinyl ethers undergo crystals, so that when the particle size of the
stereospecific polymerization initiated by con- transition metal component is relatively larger
ventional ionic initiators under certain condi- (curves c and d), the mechanical pressure of the
tions. However, the coordinating power of the growing polymer chains cleaves the particles.
Ziegler – Natta initiators is much stronger than This increases both the surface area of the initia-
the usual ionic initiators, so it is appropriate tor and the number of active sites. After this ini-
to treat Ziegler – Natta (ZN) systems separately. tial period, a steady-state rate may be observed.
The use of ZN initiators for diene polymeriza- The time required to reach this steady-state can
tion has yielded remarkable results that far sur- be reduced by initially using smaller particles.
pass the stereospecificity exhibited by organo- When the transition metal particles are ground
lithium initiators. With butadiene, four different or crushed to a very small size just before the
stereospecific polymer structures, namely, cis- polymerization, the rate of polymerization in-
1,4, trans-1,4, syndiotactic-1,2, and isotactic- creases rapidly and exceeds the steady-state val-
1,2 can each separately be obtained to almost ues (curves a and b). This phenomenon may in-
32 Polymerization Processes
Table 2. Stereospecific polymerization of butadiene [169]

dicate the existence of highly active but short- In this section a very simple example is con-
lived sites. This kind of behavior can be avoided sidered, too approximate to accurately describe
by aging the initiator system prior to addition of real systems, but useful to illustrate a model
monomer. building process. Consider the following poly-
merization scheme [175].
1) Adsorption of monomer (M) on the active
site

where P∗l is a growing chain with unit chain


length and A is a vacant active site.
2) Propagation

Figure 15. Effect of previous physical treatment on a sample


of α–TiCl3 on the propene polymerization rate (in grams
C3 H6 per gram TiCl3 per hour) at constant pressure and where P∗r is the growing polymer chain with
temperature length r
a), b) Ground α–TiCl3 (sizes ≤ 2 µm); c), d) Unground α– 3) Desorption of polymer chain from the active
TiCl3 (sizes 1 – 10 µm) site
Many ZN polymerizations exhibit a contin-
uous decrease in rate rather than reaching a
steady-state in which the rate curve reaches a
maximum and then continues to decline. The where Pr is a dead polymer molecule with
rate of decline varies with the polymerization chain length r.
system used. The decline in polymerization rate Assuming that the polymer chain is suffi-
has been attributed to (1) a decrease in the num- ciently long (long-chain assumption), the poly-
ber of active centers, (2) a lowering of activ- merization rate Rp is given by
ity of individual active centers due to structural
Rp =Kp [P∗ ] [M] (2.100)
changes, and (3) a lowering of activity of in-
dividual active centers due to encapsulation by where [P∗] is the total concentration of active

polymer. sites, namely, [P ∗ ] =
P
[P∗r ] .
r=1
Polymerization Processes 33

If the stationary-state hypothesis is applied to (1 − Φ′ ) is the probability that a growing chain


[P∗], stops growing.
Therefore, Equations (2.110) and (2.111) are
KA [M] [A] =KD [P∗ ] (2.101) essentially the same as Equations (2.8) and (2.9)
or Equation (2.56). The polydispersity index
Each active site is either occupied by growing (PDI) for instantaneously formed polymer is
polymer chains or is vacant. Denoting the total two.
concentration of adsorption sites by [A]0 , An interesting feature of heterogeneous ZN
initiators is the very broad MWD of the polymers
[A]0 = [A] + [P∗ ] (2.102)
produced, although the above simple model pre-
Combining Equations (2.101) and (2.102) gives dicts PDI = 2 for instantaneously formed poly-
mer. The PDI of accumulated polymer may be
[P∗ ] =K [A]0 [M] / (1 +K [M]) (2.103) 10 or higher. In the case of copolymerization, the
compositional heterogeneity is also large even
where K = K A /K D . Therefore, the polymeriza- when monomer ratios are kept constant during
tion rate is given by polymerization. Presently, there are two main
theories which try to explain these large disper-
Rp =Kp K [A]0 [M] / (1 +K [M]) (2.104) sities, namely, the presence of a distribution of
activities for the active sites or diffusional ef-
Next consider the instantaneous molecular mass
fects which limit the transport of reactants to the
distribution for this process. The population bal-
active sites [176].
ance equations for growing polymer chains are

1 d V P∗1
 
=KA [M] [A] 2.3. Copolymerization
V dt
−Kp [M] [P∗1 ] −KD [P∗1 ] (2.105)
Copolymerization permits the synthesis of an al-
1 d (V [P∗r ]) most unlimited number of polymer types and is
=Kp [M] P∗r−1 −Kp [M] [P∗r ]
 
V dt therefore often used to obtain a better balance of
−KD [P∗r ] (r ≥ 2) (2.106) properties for commercial applications. Copoly-
mers may be synthesized by chain-growth
By application of the SSH for [P∗r ],
and step-growth polymerization. In step-growth
1 polymerization, different monomers with the
[P∗l ] = [A] (2.107)
1 +τ ′ same type of functional group generally show
1  ∗  only minor differences in reactivity. As a result,
[Pr∗ ] = Pr−1 (2.108) most copolymers prepared by step-growth poly-
1 +τ ′
merization contain essentially random place-
where τ ′ = K D /(K p [M]). Therefore, ments of repeat units, with the composition of
r the copolymers essentially the same as those of
[Pr ] =KD [P∗r ] =KD [A] Φ′ (2.109) the original monomer mixture.
′ ′ In contrast, strong selective effects often oc-
where Φ = 1/(1 + τ ).
cur in chain-growth copolymerizations, and the
The number and weight chain length distri-
composition of the copolymer formed may dif-
butions are given by
fer greatly from the composition of the origi-
[Pr ] r−1 nal monomer mixture. This section deals ex-
= 1 −Φ′ Φ′

N (r) = (2.110)

P clusively with chain-growth copolymerization.
[Pr ]
r=1 Chain-growth copolymerization can be carried
r [Pr ] 2 ′ r−1 out with various types of active centers including
W (r) = =r 1 −Φ′ Φ (2.111)

P free-radical, cationic, and anionic species. Free-
r [Pr ]
r=1 radical copolymerization is most commonly
used due to its higher alternating tendencies.
where Φ′ is the probability that a growing poly-
mer chain adds another monomer molecule, and
34 Polymerization Processes

2.3.1. Copolymer Composition nal model. The specific polymerization rates of


monomers 1 and 2 are given by
The composition of copolymers cannot be de-
termined from a knowledge of the homopoly- 1 d (V [M1 ])
− =K11 [P∗1 ] [M1 ] +K21 [P∗2 ] [M1 ]
merization rates of each monomer. In 1944, V dt
the instantaneous copolymer composition equa- (2.116)
tion was proposed independently by several re-
searchers by assuming that the chemical activity 1 d (V [M2 ])
− =K12 [P∗1 ] [M2 ] +K22 [P∗2 ] [M2 ]
of a propagating chain depends solely on the ter- V dt
(2.117)
minal monomer unit on which the active center
is located [177–180]. This model is called the where [P∗1 ] and [P∗2 ] are the total concentrations
terminal model, and the copolymer chain can of active centers of types 1 and 2 with
be considered as a first-order Markov chain. For

∞ X
binary systems, the following four propagation X
[P∗1 ] =
 ∗ 
Pm,n,1 (2.118)
reactions are possible. m=1 n=1
∞ X
X ∞
[P∗2 ] =
 ∗ 
Pm,n,2 (2.119)
m=1 n=1

Application of the SSH gives

K21 [P∗2 ] [M1 ] =K12 [P∗1 ] [M2 ] (2.120)

Dividing Equation (2.116) by Equation (2.117)


and using Equation (2.120) gives the instanta-
where P∗m,n,1 is a live copolymer chain with m neous copolymer composition equation:
units of monomer 1 (M1 ) and n units of monomer F1 − d (V [M1 ]) [M1 ] (r1 [M1 ] + [M2 ])
2 (M2 ) bound in the polymer chain and with the = =
F2 − d (V [M2 ]) [M2 ] ([M1 ] +r2 [M2 ])
active center located on terminal monomer unit (2.121a)
1.
The reactivity of the propagating species may or
be affected by the penultimate monomer unit. In
such cases, the model is referred to as the penul- r1 f 21 +f1 f2
F1 = 2 + 2f f +r f 2
(2.121b)
timate model or a second-order Markov chain r1 f 1 1 2 2 2

[181], [182] and propagation consists of eight


where F 1 and F 2 are the mole fractions of
reactions. Further expansion is possible by con-
monomers 1 and 2 in the copolymer produced
sidering the effects of remote units preceding the
instantaneously, and f 1 and f 2 are the mole frac-
penultimate unit, such as the pen-penultimate
tions of unreacted monomer. The reactivity ra-
model [183–185].
tios are defined by
It is customarily assumed that the propa-
gation rate constants are independent of chain K11
r1 = (2.122)
length and that the chains are sufficiently large K12
( long-chain assumption, LCA). The LCA in- K22
cludes the approximation that monomer con- r2 = (2.123)
K21
sumed in reactions other than propagation is neg-
ligible and that the SSH is valid for each type It is straightforward to extend this method
of active center; i.e., the rate of formation of to multicomponent polymerization [188–190].
any type of active center is equal to its rate of Note that Equation (2.121 a or b) is strictly valid
consumption. Statistically, these conditions are only for infinitely long polymer chains and its
equivalent to the statistical stationary condition application to short chains may introduce sig-
[185–187]. nificant error [191–193].
Now consider a binary copolymerization When applying Equation (2.121 a or b), it is
whose propagation reactions follow the termi- usually assumed that all of the copolymer chains
Polymerization Processes 35

have the same composition. However, since the In this instance, the Stockmayer instantaneous
chain length of a copolymer is finite, the compo- bivariate distribution of chain length and com-
sition and chain lengths of the individual poly- position will be obtained. Another method of
mer molecules cannot all be identical. Therefore, composition control is to use semi-batch oper-
for copolymer chains produced instantaneously ation in which monomers are fed to maintain a
there is a bivariate distribution of composition constant ratio of monomer concentrations in the
and chain length [194–200]. The variance of the reactor [209], [210].
copolymer composition distribution is approxi-
mately inversely proportional to the chain length
and, therefore, for sufficiently long chains it may
be reasonable to neglect the composition distri-
bution. In order to treat oligomeric molecules,
application of discrete mathematics such as the
finite Markov chain theory [184], [201] is nec-
essary.
In batch copolymerization, there is a compo-
sitional drift due to the change in the compo-
sition of the unreacted monomer mixture with
time. The total monomer conversion x is related
to the mole fraction f 1 of unreacted monomer 1
by the following equation [202]:
d f1 f1 −F1
= (2.124)
dx 1 −x
Figure 16. Batch copolymerization of styrene (M1 ) and
f1 =f10 , F1 =F10 at x = 0. methyl methacrylate (M2 ) at 60 ◦ C for f 10 = 0.8, f 20 = 0.2,
r1 = 0.53, r2 = 0.56
Therefore, if the relationship between f 1 and F 1 and F 2 are the accumulated mole fraction of monomer
1 and 2 bound in the copolymers.
F 1 is known, Equation (2.124) can be solved
numerically. If the terminal model is applicable In the terminal model, monomer sequence-
to a binary system (Eq. 2.121 a or b), Equation length distribution depends only on the product
(2.124) can be integrated analytically to obtain of reactivity ratios r 1 r 2 [211], [212]. In free-
the following equation [203–208]: radical copolymerizations r 1 r 2 is generally less
 α  β  γ than unity, indicating a higher alternating ten-
f1 f2 f10 −δ
x= 1− (2.125) dency. The styrene – maleic anhydride system is
f10 f20 f1 −δ
an example of a very highly alternating one with
where both r 1 and r 2 very close to zero. For ionic co-
α = r 2 /(1 − r 2 ), polymerizations, there is a general lack of any
β = r 1 /(1 − r 1 ), tendency towards alternation, and r 1 r 2 is usually
γ = (1 − r 1 r 2 )/{(1 − r 1 ) (1 − r 2 )} close to or greater than unity. Generally, in free-
δ = (1 − r 2 )/(2 − r 1 − r 2 ). radical copolymerization, the reactivity ratios
are relatively insensitive to the reaction medium
Except for copolymerizations at the and temperature. However, in ionic copolymer-
azeotropic point, there will be some composi- ization, temperature and reaction medium can
tional drift. (At the azeotropic point, the compo- significantly affect reactivity ratios.
sition of the instantaneously produced copoly- Traditional methods for estimating reactivity
mer and of the unreacted monomers is the same; ratios [177], [213–215] are based on the trans-
hence in a batch reactor there is no compositional formation of the instantaneous copolymer com-
drift at this point.) An example of compositional position equation into a form that is linear in the
drift during batch polymerization is shown in parameters r 1 and r 2 . While these linearizations
Figure 16 [208]. One method to avoid compo- provide simple techniques for parameter estima-
sitional drift is to use a continuous stirred-tank tion, they are generally statistically invalid be-
reactor with an ideal residence time distribution.
36 Polymerization Processes

cause the independent variable has experimental Propagation


error and the dependent variable does not have a
constant variance [216–223]. Both of the latter
assumptions are necessary for the linear least
squares method to be a statistically valid es-
timation method. Although it has been shown
that proper experimental design can allow the
use of linear least squares analysis [222], [224],
the estimation of the reactivity ratios is a typical Transfer to monomer
example of a problem in nonlinear estimation.
Reactivity ratios are now generally estimated
by application of procedures based on the sta-
tistically valid error-in-variables-model (EVM)
[220], [221], [225–230]. These methods allow
all the sources of experimental error to be ac-
counted for. Transfer to small molecule
In spite of the fact that reactivity ratios ob-
tained using invalid estimation procedures have
often been used, the copolymer composition
equation based on the terminal model has given
useful predictions. It is generally ineffective to Termination by disproportionation
use the copolymer composition equation for
comparing models. Information on the copoly-
mer microstructure (e.g., monomer sequence
length distribution) is needed to compare the va-
lidity of terminal and penultimate models [179],
[231], [232]. Advancements in NMR techniques
have provided suitable information on sequence Termination by combination
length distribution. However, cases have been
reported in which sequence length distribution
information does not permit model discrimina-
tion [233].
As will be discussed in Section 2.3.2 the com-
position equation includes the rate constants of •
where Rm, n, 1 is a polymer radical with m units
elementary reactions only in ratios, while the of monomer 1 (M1 ) and n units of monomer 2
rate equation is dependent on their absolute val- (M2 ) bound in the polymer chain with active
ues, and therefore success of the former equation center located on monomer unit 1. Pm ,n is a
does not insure success of the latter [234]. polymer molecule with m units of monomer 1
and n units of monomer 2. The polymerization
rate, Rp is given by
2.3.2. Kinetics of Copolymerization
Rp =K11 [R·1 ] [M1 ] +K12 [R·1 ] [M2 ]
Consider the kinetics of free-radical copolymer- +K21 [R·2 ] [M1 ] +K22 [R·2 ] [M2 ]
ization of monomers M1 and M2 , assuming the =Kp [R· ] [M] (2.126)
terminal model is applicable. Important elemen- • •
tary reactions are: where [R1 ] and [R2 ] are given by
Equations (2.118) and (2.119) and
Initiation
[R· ] = [R·1 ] + [R·2 ] , [M] = [M1 ] + [M2 ] , and
K p is defined by
Kp = (K11 f1 +K12 f2 ) ϕ·1 + (K21 f1 +K22 f2 ) ϕ·2
(2.127)
Polymerization Processes 37
• • •
where f 1 = [M1 ]/[M], and ϕ1 = [R1 ]/[R ]. In the early development of the kinetics of
Equation (2.127) is an example of a pseudo- copolymerization, chemically controlled bimo-
kinetic rate constant [145–147], [209], [210], lecular termination was assumed to be operable
[235–238]. For an N-component system, the [239–241]. However, since these reactions have
pseudo-kinetic rate constants can be defined as been shown to be diffusion controlled [120–
follows 122], [242], Equations (2.131) and (2.132) must
Propagation be modified by using appropriate models which
account for this. In this context, the use of the
N
X cross-termination factor, ϕ = K t12 /(K 11 K 12 )0.5 ,
Kp = Kij ϕ·i fj (2.128)
i=1
is not acceptable. Several models for diffusion-
j=1 controlled termination of binary copolymeriza-
Chain transfer to monomer tion have been proposed. Atherton and North
[243], [244] proposed the following for the ter-
N
X mination constant.
Kfm = Kfij ϕ·i fj (2.129)
i=1 Kt =Kt11 F1 +Kt22 F2 (2.133)
j=1

The mole fractions of monomers bound in poly-


Chain transfer to small molecule mer chains F 1 and F 2 can be calculated by us-
N
X
ing the copolymer composition equation. On the
KfT = KfTi ϕ·i (2.130) other hand, Russo and Munari [245] proposed
i=1 the use of penultimate effect for the termination
reactions since segmental diffusion is highly de-
Termination by disproportionation
pendent on the last two portions of the chain.
N
X Until the mid 1980s, a prevailing view of co-
Ktd = Ktdij ϕ·i ϕ·j (2.131) polymerization kinetics was that the propagation
i=1
j=1
process is, in most cases, correctly described by
the terminal model, whereas the termination pro-
Termination by combination cess involves complexity still to be elucidated.
However, it has also been speculated that it is the
N
Ktc =
X
Ktcij ϕ·i ϕ·j (2.132) propagation step that needs further study, while
i=1 the termination step is well described by simple
j=1
models such as Equation (2.133), and the penul-
Even though these pseudo-kinetic rate constants timate effect on propagation reactions is being
depend on the monomer mole fraction and examined from the point of view of copolymer-
change with time in a batch reactor, the simpli- ization rates [233], [246–252].
fication achieved when dealing with multicom-
ponent polymerizations is very great. Further-
more, since the mole fraction of each polymer 2.3.3. Copolymerization of Vinyl and
• Divinyl Monomers
radical type ϕi is independent of chain length
for sufficiently long chains, these same pseudo- The use of the terms vinyl and divinyl here is
kinetic rate constants can be used to calculate not according to their strict definitions. Here, a
molecular mass distribution as well as poly- vinyl monomer is defined as a monomer with
merization rate [61],[235] . Therefore, by us- a single reactive double bond (a double bond
ing pseudo-kinetic rate constants, a multicompo- which will readily add to a radical center) and a
nent polymerization reduces to a homopolymer- divinyl monomer is a monomer which has two
ization, and therefore Equations (2.55) – (2.64) such double bonds.
are all applicable for copolymerization. With The free-radical copolymerization of vinyl
appropriate definitions for pseudo-kinetic rate and divinyl monomers is important in the manu-
constants, this method may also be usefully ap- facture of ion-exchange resins, chromatographic
plied when copolymerization kinetics follow the packings, superabsorbent polymers (with a
penultimate model [235]. rapidly growing market in baby diapers and
38 Polymerization Processes

Figure 17. A schematic drawing of a cross-linked polymer network synthesized by free-radical copolymerization of vinyl and
divinyl monomer

promise of the same in several other market vinyl monomer increases, the heterogeneity of
areas), cross-linked latex polymers, and other the polymer network (i.e., regions having very
products. Figure 17 shows a schematic of a poly- different M c ) becomes significant [265–268].
mer network synthesized by copolymerization Reactivity ratios for the copolymerization of
of vinyl and divinyl monomers. There may be vinyl and divinyl monomers are rather difficult
many radical centers on the polymer network to determine. If the reactivities of both double
during polymerization since the mobility of rad- bonds on the divinyl monomer are the same and
ical centers chemically bound to the network independent, and cyclization does not occur, the
can be highly restricted. Strong autoacceleration conventional copolymer composition equation
in polymerization rate has been reported during (Eq. 2.121 a or b) is still valid when monomer
network formation by free-radical copolymeri- concentration is replaced by double bond con-
zation [253–260]. The live double bonds located centration and the reactivity ratios defined with
on polymer chains are called pendant double respect to each type of double bond are used
bonds. Kinetic behavior of radical centers on [269]. However, difficulties generally arise due
polymer chains and of pendant double bonds are to the complicated behavior of pendant double
the most important factors influencing the ki- bonds, which may react inter- and intramolec-
netics of network formation during free-radical ularly, and whose reactivity may differ from
polymerization. From the point of view of the that of monomer double bonds. At the moment
mobility of chains, network polymers may be when a divinyl monomer is chemically bound
considered as a heterogeneous reaction system, in a polymer chain, another double bond on the
and the formation of microgels before the gel just-reacted divinyl monomer may be the nearest
point is reached may be a general feature of net- neighbour of the active center, and therefore cy-
work formation in free-radical polymerization clopolymerization may occur under certain con-
[257], [261–265]. The number-average molec- ditions. For example, the free-radical polymer-
ular mass between cross-links M c is important ization of diallyl quaternary ammonium salts
from the point of view of the elastic properties gives soluble, not cross-linked polymers with lit-
of a gel molecule. As the mole fraction of di- tle or no residual unsaturation [270], [271].
Polymerization Processes 39

nite molecules, the predictions of kinetic and


equilibrium models are the same [61], [147],
[280]. However, none of the above idealized as-
sumptions are strictly applicable to a real sys-
tem [265], [281–287]. Kinetic models for net-
work formation by free-radical polymerization
are being developed [61], [62], [145–147], [288–
295]. Figure 18 shows the change in the av-
erage chain length and weight fraction of gel
during network formation in the copolymeriza-
tion of methyl methacrylate and ethylene glycol
In general, cyclization may also involve two dimethacrylate [147].
or more monomer units. Since cyclization reac-
tions are controlled by the conformational statis-
tics of the sequence of bonds and do not follow
the conventional rate law, larger cycles, formed
when an active center adds to a pendant double
bond on its own chain or to two double bonds
on another chain, may also affect the copoly-
mer composition even when the terminal model
is applicable. Various copolymer composition
equations which account for cyclization reac-
tions have been proposed [272–276]; however,
it may be difficult to apply these equations to
higher mole fractions of divinyl monomer or to
high monomer conversions. Strictly speaking, in
order to know the copolymer composition, it is
necessary to know the kinetic behavior of pen-
dant double bonds completely; that is, a knowl-
edge of the reactivity ratios r 1 and r 2 is insuffi-
cient to estimate the change in composition dur-
ing copolymerization. However, r 1 and r 2 are
usually obtained without taking into account the Figure 18. Development of average chain length within sol
monomer consumed by active centers located on fraction and weight fraction of gel during batch copolymer-
just-reacted pendant double bonds. Such param- ization of methyl methacrylate and ethylene glycol dimeth-
eters may be better regarded as empirical param- acrylate (0.25 mol %) at 70 ◦ C initiated by AIBN [147]
• = Experimental data
eters which do not reflect the chemical reactivity
of radical centers with double bonds, and they
may change with monomer conversion.
As for the kinetics of network formation, 3. Polymerization Processes and
fundamental models which assume an equilib- Reactor Modeling
rium system (see Section 2.1.3) have been exten-
sively applied to chain-growth polymerization 3.1. Introduction
[39], [43], [51], [52], [148], [277–279]. How-
ever, chain-growth polymerizations are kineti- The polymer reactor model is now becoming ac-
cally controlled, so that the application of the cepted as a valuable tool whose use contributes
conventional approaches may be in error. It has significantly to all aspects of process technology
been shown that under Flory’s simplifying as- for polymer manufacture. This includes process
sumptions, namely that (1) the reactivities of all design, optimization, state estimation, and con-
types of double bonds are equal, (2) all double trol. Through process design, polymers with a
bonds react independently of one another, and unique and desirable combination of properties
(3) there are no intramolecular reactions in fi- can often be obtained. Process parameters such
40 Polymerization Processes

as residence time distribution (RTD) are usually 3.2. Processes and Reactor Modeling for
not considered by polymer synthesis chemists, Step-Growth Polymerization
although RTD can influence chemical compo-
sition distribution (CCD), molecular mass dis- 3.2.1. Types of Reactors and Reactor
tribution (MWD), long-chain branching (LCD) Modeling
and gel/sol ratios. In the early days of the poly-
mer industry, the chemist played the major role In step-growth polymerization, high molecular
in product and process development and scale- mass polymers are usually not produced until the
up. This has changed, with the process engineer final stage of reactions, so that thermal control
now playing a significant role in all phases of and mixing of the reaction mixture do not present
commercialization of new and improved poly- serious problems in the earlier stages. However,
mer products. His broad experience with pro- since the final stage of polymerization is very im-
cess fundamentals and computer modeling are portant for the production of polymers with high
essential to obtain high-quality products, safely molecular mass, handling of very high viscosi-
and economically. ties and temperatures and high interfacial area
Dynamic reactor models can be used in a to remove small molecules are required. Various
variety of ways. Stability and control of poly- polymerization processes and reactor types, both
mer reactors should be considered at the de- for batch and continuous production, have been
sign stage and control problems minimized then, proposed. Examples of reactors for high viscosi-
rather than take corrective action after the plant is ties are shown in Figures 19 and 20. Careful se-
built. Complex interactions which are involved lection of the polymerization reactor is very im-
in polymerization (highly nonlinear temperature portant to produce high-quality polymers [296],
and concentration effects) preclude optimal de- [297].
sign based on experimentation alone because the
cost would be prohibitive. Models can be used to
identify potential sources of product variability
and strategies to minimize their effects. Models
can be used to store information on process tech-
nology in a concise and readily retrievable and
modifiable form.
Process models can be used to train chemists,
chemical engineers, and plant operators and give
them a feel for the dynamics of the polymeriza-
tion process.
The most expensive aspect of model devel-
opment is experimental estimation of model
parameters; highly instrumented bench-, pilot-
scale, and plant-scale reactors are required. Sta-
tistically designed experiments should be per-
formed to permit efficient parameter estimation
and model development. Modeling is an iterative
process and the very act of developing a deter-
ministic model permits a greater understanding
of the relevant microscopic processes which oc-
cur during polymerization or polymer modifica-
tion. As additional data (plant, pilot-plant, and
bench-scale) become available, model structure
and parameters can be updated.
This section considers recent developments
in polymerization/polymer modification pro- Figure 19. Vertical cone ribbon blade reactor (Mitsubishi
cesses and discusses advances in polymer re- Heavy Industries)
actor modeling, state estimation, and control.
Polymerization Processes 41

Molecular mass distribution of linear poly-


mers produced by step-growth polymerization in
a batch or a PFR basically follows the most prob-
able distribution [303]. (Note that batch reactor
and PFR are, in principle, equivalent). The mo-
lecular mass distribution may be controlled by
varying its reaction path if the reaction system is
in a nonequilibrium state. Assuming irreversible
step-growth polymerization without interchange
reactions, the effect of reactor types, such as ho-
mogeneous CSTR, segregated CSTR and PFR
with a recycle loop, on molecular mass distribu-
tion have been considered [301], [302], [304],
[305]. An important feature of step-growth poly-
merization is that the variance of the molecu-
lar mass distribution is smallest in a batch re-
actor or PFR and is largest in a homogeneous
CSTR, which is quite contrary to that for chain-
growth polymerization. This result may be dis-
Figure 20. Horizontal high-viscosity reactor appointing, since it is, in principle, impossible
to produce polymers whose polydispersity in-
dex M W /M N is smaller than two in step-growth
The batch reactor is the most versatile reac- polymerization at sufficiently high conversions.
tor type and is used extensively for specialty The polydispersity index of polymers produced
polymers at low production volumes. Some in a batch reactor is given by 1 + p, as shown
examples of step-growth polymerizations car- in Section 2.1.1., where p is the conversion of
ried out in such reactors are nylon 6, phe- the functional group. However, in commercial
nol – formaldehyde, urea – formaldehyde, and polymeric materials, polymers with narrower
melamine – formaldehyde. In polycondensation distributions are not always superior to those
reactions, it may be necessary to remove con- with broader distributions, since various levels
densation products to attain sufficient conver- of properties are required at the same time. The
sion. When the volume of the reaction mass de- use of a cascade of CSTRs and/or PFRs with
creases continuously with time, such reactors recycle loops may be one method to obtain a
are called semi-batch reactors. (A tank-type re- molecular mass distribution with a polydisper-
actor which does not operate at steady state is sity index larger than two. However, in practice,
defined as a semi-batch reactor). For example, these methods may have shortcomings because
in the production of poly(ethylene terephthal- they need a long start-up period and, therefore,
ate), PETP, since methanol or ethylene glycol problems may occur with the stability of the re-
evaporates during polymerization, the batchwise action system. A method in which additional
production of PETP is considered a semi-batch monomers are fed intermediately to a batch re-
operation. actor or a PFR has been proposed [306]. The-
On the other hand, newer high-capacity oretical analysis of this intermediate monomer
plants often use continuous processes. The first feed method has also been carried out [307], and
approximation for a continuous process is a it has been shown that the polydispersity index
model that consists of plug flow reactors (PFR) can be easily controlled over a wide range with
and continuous stirred-tank reactors (CSTR) in values greater than 2.
various combinations (Fig. 21), although vari- In a batch reactor, the reverse reactions and
ous nonideal effects such as flow pattern in the the interchange reactions (redistribution reac-
reactor, mass- and heat-transfer limitations, and tions) do not change the MWD from the most
residence time distribution must be considered probable distribution [303], [308–310]. How-
for a detailed analysis and design of real reactors ever, these reactions do change the MWD of
[298–302]. polymers produced in CSTR, PFR with a recycle
42 Polymerization Processes

Figure 21. Representative models for polymerization reactors


A) CSTR + PFR; B) CSTR + CSTR + PFR; C) PFRs; D) CSTRs

loop, and intermediate monomer feed method. neously with those for chemical reactions in a
Some consideration of these reactions in a CSTR rational model for these complex situations.
is given in [311]. Qualitatively, these effects
lower the polydispersity and make the MWD
approach the most probable distribution. This
result seems reasonable, since any MWD ap-
proaches the most probable distribution with
a polydispersity index of two when polymer
chains are severed randomly [312], [313].
Other than the common reactor types dis-
cussed above, other special types of reactor sys-
tem may be applied. For example, in the poly-
merization of urethanes, the reaction rates are
so high that reaction takes place even when the
monomers are being mixed and pumped into
molds. In situ polymerization to form the de-
sired articles directly from monomeric liquids
is known as reaction injection molding (RIM)
[314], [315]. A schematic of the RIM process
is shown in Figure 22. The RIM processing of
polyesters, epoxy resins, polyamides, and dicy-
clopentadienes has also been introduced, al-
though more than 95 % of the total produced by
RIM is polyurethane [315]. In the RIM process,
the reaction is almost complete by the time the Figure 22. Schematic drawing of the RIM process
material fills the mold, and therefore the mix- a) Monomer A; b) Monomer B; c) Polymerizing mixture;
ing and flow equations must be solved simulta- d) Mold; e) Mixer
Polymerization Processes 43

In the finishing stage of nylon 6, nylon 66, and 1) Ring opening of ε-caprolactam by water
PETP polymerizations, higher molecular mass (Eq. 3.1), which produces aminocaproic acid
polymers may be obtained by solid-state poly- (ACA)
merization in which polymerization occurs by 2) Polycondensation of ACA (Eq. 3.2)
heating chips or flakes of a material below its 3) Acid-catalyzed polyaddition (ring-opening
melting point in a stream of hot gases in a flu- polymerization) by nucleophilic attack of the
idized bed or in a drier operated under vacuum amine nitrogen on the lactam (Eq. 3.3)
[316–320]. The monomer, condensation prod-
ucts, and various byproducts diffuse out, and
further reaction takes place inside the solid. The
progress of these types of reaction is affected
significantly by the diffusion of the condensa-
tion products and the morphology of the solid.
Although step-growth polymerization has a
very long history, a systematic kinetic treatment
like that available for free-radical polymeriza-
tion does not exist because of limitations due to
system-specific side reactions and the scarcity
of reliable kinetic data. However, this synthetic
route is becoming more important due to the
development of materials synthesized by step-
growth polymerization such as aramides, PPS, Step-growth polymerization of the amino
PEK, and PES. The production technology of acid (Eq. 3.2) accounts for only a few percent
step-growth polymers seems to have revived as of total polymerization of ε-caprolactam. How-
an attractive research area and is enjoying a Re- ever, step-growth polymerization is important
naissance period. since it usually determines the final degree of
polymerization at equilibrium. The molecular
mass distribution is essentially the most prob-
3.2.2. Specific Processes able distribution, except for the presence of
monomer and cyclic oligomers. Since low mo-
Polyamides (see also → Polyamides) are lecular mass substances lower the polymer qual-
manufactured by two basic routes. One of ity, they are usually removed by leaching or vac-
these is synthesis from cyclic monomers such uum treatment of the polymer melt. The forma-
as lactams. Polymerization of these substances tion of cyclic oligomers is an important side re-
requires ring opening and subsequent chain action [327–329]. For the simulation and opti-
growth. Another class of synthetic polyamides mization of polymerization reactors, only small
is formed from diamines and diacids. ring formation or overall ring formation is con-
The most common types of polyamides are sidered to make the analysis easier [330–333].
nylon 6 and nylon 66. The term nylon is of- Various types of polymerization reactors
ten used for synthetic aliphatic polyamides. One have been proposed both for batch and continu-
number indicates that the product was prepared ous processes. A commonly used industrial reac-
from a single monomer and represents the num- tor for a continuous process is a tubular reactor
ber of carbon atoms in the repeating unit. Two such as the conventional VK column (Verein-
numbers refer to the number of carbon atoms in facht Kontinuierliches Rohr) [321], [322], [334],
the diamine and that in the diacid, respectively. which consists of a vertical tube operating at at-
Nylon 6 is typically produced by the hy- mospheric pressure. The feed enters the top of
drolytic polymerization of ε-caprolactam [321– the column and is heated to ca. 220 – 270 ◦ C.
325], although polymers with higher molecular The simplest model for this type of reactor is a
mass can be produced by ionic polymerization PFR. However, according to impulse response
[321], [323], [326]. The major reactions in the experiments, the flow is approximately laminar
hydrolytic polymerization are rather than plug flow [322], [335], and the reac-
tor should be modeled as a CSTR followed by
44 Polymerization Processes

a tubular reactor when a large quantity of wa- be carried out solely in batch processes. How-
ter is used since a significant convection current ever, complete elimination of oxygen has made it
and mixing is provided by the evaporating water possible to carry out continuous polymerization.
[322]. An example of a continuous melt polymeriza-
Nylon 66 is manufactured by polycondensa- tion process is shown in Figure 23. The aqueous
tion of hexamethylenediamine and adipic acid nylon salt solution is heated to above 200 ◦ C at
[336], usually in a multistage process. First, ny- > 17 bar in an oxygen-free atmosphere. There-
lon salt (hexamethylenediammonium adipate) is after, the pressure is reduced to atmospheric and
prepared from stoichiometric quantities of hexa- vapor is separated from polymer to promote
methylenediamine and adipic acid in water. The polymerization to the desired high molecular
salt can easily be separated by precipitation with mass. It has also become possible to polymerize
methanol. molten hexamethylenediamine and adipic acid
directly [330], [336]. Polymerization can also
be completed in the solid state.
Several kinetic studies on the synthesis of ny-
lon 66 have been published [337–341]. How-
ever, more information is necessary for detailed
simulation and optimization of nylon 66 reac-
tors.

Polyesters (see also → Polyesters). The pro-


duction of high molecular mass polyesters dif-
fers somewhat from that of polyamides. In the
case of nylons, the chemical equilibrium favors
the polyamide under polymerization conditions.
With polyester formation, however, the equilib-
rium is much less favorable. In order to drive
the reaction in the forward direction, the con-
densation product must be removed continu-
ously, usually by application of high vacuum.
For polyester reactors, a high vacuum, a high
temperature, and a high interfacial area with suf-
ficient surface renewal are required, especially
at high conversions.
Both saturated and unsaturated polyesters
are produced. Among the saturated polyesters,
poly(ethylene terephthalate), PETP, is produced
in the largest quantity, and is used for production
of fibers, films, molding plastics, and beverage
containers. In this section, the engineering as-
pects of PETP formation are illustrated as an
Figure 23. Continuous melt polymerization of nylon 66 example of a polyester production process.
There are two major routes to synthesize
The use of nylon salt guarantees the presence PETP industrially, although the objective in each
of equimolar amounts of – NH2 and – COOH case is to obtain an intermediate product – i.e.,
groups. Close control of diamine – diacid bal- bis(hydroxyethyl)terephthalate (BHET).
ance is important to control the final polymer Two major routes to synthesize BHET are es-
molecular mass and reactive end groups. ter interchange of dimethyl terephthalate (DMT)
Nylon 66 is fairly unstable at high tempera- and direct esterification of terephthalic acid.
tures in the presence of oxygen. Not only degra- Figure 24 shows an example of the PETP pro-
dation but also cross-linking may occur. Be- duction process via the ester interchange route.
cause of this instability, polymerization used to The ester interchange reaction
Polymerization Processes 45

the reaction mixture molten and polymerization


fast.

is operated in the temperature range


150 – 210 ◦ C at atmospheric pressure. The use
of a catalyst is common [342], [343]. The meth-
anol and ethylene glycol (EG) emerging from
the reactor are passed through a rectifying col- For PETP production, a dual catalyst system in
umn and EG is fed back to the reactor. It is very which one component is specially active for es-
difficult to force the ester interchange reaction to ter interchange and the other for polymerization
completion, and therefore after a particular con- is quite often used [351]. The production of high
version (usually 90 – 95 %), the reaction mixture molecular mass polymer requires the complete
is passed on to the polycondensation stage. The removal of ethylene glycol due to the unfa-
reaction mixture consists of oligomers of various vorable equilibrium, and therefore a vacuum
types. Oligomers with degrees of polymeriza- is applied. Especially in the final stage of the
tion as high as three may be formed [342], [344], polycondensation reaction, a very high vacuum
[345]. Several reactor models for both batch and is required since the reaction system becomes
continuous processes have been proposed [342– highly viscous. Consideration of the limitations
350]. An optimization study showed that the of mass and heat transfer is very important.
ester interchange reactor should be operated Various types of reactors such as rotating disc
initially at high temperature to obtain high con- contactors, wiped film reactors, partially filled
version; the temperature should be lowered to screw extruders have been developed as fin-
reduce side reactions [343], [348]. ishers for the polycondensation reaction [342],
[343], [350]. Details of fluid mechanics, mix-
ing, and mass and heat transfer characteristics
are required for a rational analysis and design
of such high-viscosity reactors. In addition to
polycondensation reactions, various side reac-
tions must also be considered since a very high
temperature is used.
Melt polycondensation of PETP is not gen-
erally carried out beyond a particular extent of
polymerization since the degradation reactions
dominate the process and the product quality
may suffer from various undesirable byproducts.
To attain higher molecular masses, the products
may be subjected to solid-state polymerization
[316–320]. Newer processes, especially for bev-
erage and food containers, prefer to stop melt
polymerization at lower conversion, and solid-
state polymerization is extensively applied.
Direct esterification of terephthalic acid
Figure 24. Continuous polymerization process of PETP via (TPA) and ethylene glycol was generally not
ester interchange route
preferred earlier because of the difficulties in
In the polycondensation stage, the reaction the purification of TPA due to its low solubil-
temperature is raised to 265 – 285 ◦ C to keep ity and high melting point. However, with im-
provements in technology, the direct esterifica-
46 Polymerization Processes

Figure 25. Continuous polymerization process of PETP via direct esterification route

tion method has been gaining in importance. The comonomer systems under a variety of operat-
process is claimed to give polyesters with supe- ing conditions. Bulk (suspension) and solution
rior quality due to their low content of carboxyl polymerizations are considered first; extensions
end groups and diglycol linkages [351]. In the required for multiphase systems (emulsion, in-
modeling of this process, aside from the diffi- verse emulsion, suspension, dispersion, and gas-
culties caused by the various reactions and mass phase processes) may be found elsewhere [355].
balances involved, it is necessary to take account Special cases where spatial variations in temper-
of the heterogenity of the reactions due to the low ature and concentrations are important (e.g., in
solubility of TPA in EG. Simulation and control tubular reactors or packed beds) are also consid-
of the direct esterification reactors is reported ered.
in [343], [350], [352–354]. Figure 25 shows a
flow diagram of a continuous process for PETP Bulk, solution, and suspension polymer-
production by direct esterification. ization systems are characterized by the fact that
all of the reactions proceed in a single phase
with no spatial variations in temperature and
3.3. Processes and Reactor Modeling for concentration. A model for a reactor carrying
Chain-Growth Polymerization out such polymerizations would consist of a set
of material balances giving the rates of accu-
3.3.1. Material Balance Equations for Batch, mulation, inflow, outflow and a reaction source
Semi-Batch, and Continuous Reactors (sink) term for the various monomers, initiators,
chain-transfer agents, and polymer in the reac-
The following material balance equations ap- tor. These balance equations are now given in
ply for multicomponent polymerization, accom- general form
modate operation of a well-stirred reactor (no Monomer balances:
spatial variations in temperature and concentra-
tions), and may be used to simulate different dNi /dt =Fi, in − (Ni /V ) Vout −Rpi V (3.7)
Polymerization Processes 47

where Additional Ingredient Balances. In order to


calculate Rpi for free-radical systems, the to-
Ni is the number of moles of monomer i in
tal polymer radical concentration and, therefore,
the reactor
initiator concentration are required. In addition,
Fi, in are molar flow rate of monomer i into the
balances for the chain-transfer agent (for mo-
reactor
lecular mass calculations) and for the solvent
V is the reaction volume in the reactor
in solution polymerizations are required. These
V out is the total volumetric flow rate of all
balances follow:
species out of the reactor
Rp i is the net rate of disappearance of dNIi /dt =FIi, in − (NIi /V ) Vout −RIi V (3.10)
monomer i by reaction
Reaction Volume. Since the density of a poly- dVs /dt =Vs, in − (Vs /V ) Vout (3.11)
mer is usually significantly greater than that of its
monomer, the reaction volume V decreases with dNT /dt =FT, in − (NT /V ) Vout −RT V (3.12)
conversion for isothermal polymerization in a where N Ii is the number of moles of initiator
batch reactor. This shrinkage must also be taken of type i in the reactor, F Ii, in is the molar flow
into account in semi-batch and continuous op- rate of initiator i into the reactor, RIi is the con-
erations. Neglecting volume change on mixing sumption rate of initiator i by reaction, V s is the
polymer and monomers (thermodynamic data volume of inert solvent in the reactor, N T is the
are most often not available and deviation from number of moles of chain-transfer agent (CTA)
ideality is often not great) the change in reaction in the reactor, F T, in is the molar flow rate of CTA
volume (V ) may be calculated by using (assum- into the reactor, and RT is the consumption rate
ing V is the equilibrium volume): of CTA by reaction.
It is convenient to sum the monomer bal-
dV /dt =Vin −Vout − shrinkage rate
ance equation (Eq. 3.7) over n, the number of
n
=Vs, in +
X
Fi, in Mmi /̺mi monomer types to give
i=1
n   dNM /dt =Fin − (NM /V ) Vout −Rp V (3.7a)
X 1 1
− Rpi Mmi − V − Vout (3.8)
i=1
̺ mi ̺p where N M is the total number of moles of
monomer in the reactor, F in is the total molar
where n is the number of monomer types, V s, in flow rate of monomer to the reactor, Rp is the
is the volumetric flow rate of inert solvent total molar consumption rate of monomer by re-
into the reactor, M mi is the molecular mass of action.
monomer i, ̺mi is the density of monomer i, ̺p
is the density of polymer produced instanta-
neously. 3.3.1.1. Rates of Reaction and Copolymer
Polymer Balances. With a batch reactor, Composition
where there is no inflow and outflow of poly-
mer from the reactor, the total amount of poly- With application of the pseudo-kinetic rate con-
mer formed and its composition can be obtained stant method, Rp can be expressed as
directly from the monomer balances. However,
with semi-batch and continuous operation, ad- Rp =Kp [M] [P∗ ] (3.13)
ditional balances are required and these follow
where
dPi /dt =Fpi, in − (Pi /V ) Vout +Rpi V (3.9) n
X
[M] =NM /V = [Mi ] (3.13a)
where Pi is the number of moles of monomer i i=1
chemically bound in the polymer “in the reac-
tor”, F pi, in is the molar flow rate of monomer i and [Mi ] is the concentration of monomer of
bound in the polymer flowing into the reactor. type i in the reactor.
When the terminal model [356] for copoly-
merization is valid
48 Polymerization Processes

and K̄ tN is the number-average bimolecular ter-


n
X mination constant. When termination by com-
[P∗ ] = [P∗i ] (3.13b) bination and disproportionation are both signif-
i=1 icant
where [P∗i ] is the concentration of active centers
K̄tN =K̄tcN +K̄tdN (3.14c)
of type i in the reactor and
n X
X n For chemically controlled bimolecular termina-
Kp = Kpij ϕ∗i fj (3.13c) tion, pseudo-kinetic rate constants for K̄ tcN and
i=1 j=1 K̄ tdN may be found in [355–357]. Details con-
cerning the definitions and use of K̄ tN , K̄ tW ,
where ϕ∗i is the number fraction of active cen-
K̄ tZ , may be found elsewhere [358], [359]. The
ters of type i in the reactor, and fj is the mole
definition of K̄ tN follows:
fraction of monomer of type j in the reactor.
When the penultimate model [356] for co- ∞ X
X ∞

polymerization is valid K̄tN = Kt (r, s) ψ (r) ψ (s) (3.14d)


r=1 s=1
n
where K t (r, s) is an arbitrary bimolecular termi-
X
[P∗ ] =
 ∗
Pij (3.13d)
ij nation constant distribution for reaction of poly-
meric radicals of chain length r and s and ψ (r)
and and ψ (s) are the number fractions of polymeric
n
X
radicals of chain length r and s. Note that K̄ tN
Kp = Kpijk ϕ∗ij fk (3.13e) can be used to calculate Rp and M̄ N but if used
ijk to calculate higher molecular mass averages they
would be underestimated if a significant number
The fraction of active centers of type i (ϕ∗i ) can
of polymer chains are formed by bimolecular
be found by using the stationary state hypoth-
termination reactions [358].
esis (SSH). Model development will continue
Realistic calculations of [R] using Equation
assuming that the terminal model for copoly-
(3.14 b) requires that the SSH be valid and that
merization is valid. The ϕ∗i may be found by
diffusion-controlled bimolecular termination be
using
accounted for. The first direct experimental test
n
X n
X by ESR of the validity of the SSH for bulk
Rpji = Rpij (3.14a) polymerization of methyl methacrylate (MMA)
j=1 j=1
j6=i j6=i (where linear polymer chains are produced)
and of MMA – ethylene glycol dimethacrylate
with i = 1, . . . n and where Rpij is the consump- (EGDMA) (where chains with long branches
tion rate of monomer j adding to the active center and polymeric networks are formed) has shown
i. that SSH is valid for bulk MMA polymeriza-
For anionic, cationic, and anionic coordina- tion, but is not valid when substantial cross-
tion polymerization, the estimation of the total linking occurs [360]. Coyle et al [361] have
number of active centers is not always straight confirmed the validity of SSH for bulk poly-
forward, and this quantity is sometimes used as merization of MMA using numerical solutions
an adjustable parameter. For free-radical poly- of the full set of kinetic equations. Attempts to
merization, a balance between radical genera- model diffusion-controlled bimolecular termi-
tion rate and bimolecular termination rates pro- nation is considered later when discussing the
vides the following with use of the SSH polymerization of MMA (see page 53) which
1/2 is considered the model system because of its
[R· ] = RI /K̄tN (3.14b) extreme Trommsdorff – Norrish effect. At very
• high monomer conversions, when the polymer-
where [R ] is the total concentration of poly- ization temperature is below the glass transition
meric radicals in the reactor, RI is the generation temperature of the polymer being synthesized,
rate of polymeric radicals of chain length unity, the initiator efficiency and propagation constant
Polymerization Processes 49

both begin to fall dramatically, and these ef- Long-Chain Branching and Cross-
fects should be properly accounted for. These ef- Linking. The production of long branches and
fects will also be considered in detail later when cross-links requires that so-called dead polymer
MMA polymerization is discussed. The mod- chains take part in branching reactions to pro-
eling of diffusion-controlled bimolecular termi- duce tri- and tetrafunctional branch points via
nation, propagation, and the cage effect on ini- transfer to polymer and by addition of polymer
tiator efficiency is not entirely clear, and much radical centers to polymer double bonds. The
research must be done before these topics may mechanisms of anionic, cationic, and anionic
be considered standard engineering practice. coordination polymerization invariably produce
The polymerization rates for individual linear polymer chains. There may be a few ex-
monomer type j can be calculated from ceptions, but these will not be considered.
Pseudo-kinetic rate constants for transfer to
n
! "
Rpj =
X
Kpij ϕ∗i fj [M] [P∗ ] (3.14e)
polymer and polymer radical addition to pendant
i=1
double bonds may be defined as [355], [357],
[363–365]
with j = 1, . . . n.
n
The mole fraction of monomer of type j in X
Kfp = Kfpij ϕi F̄j (3.15)
copolymer produced instantaneously is given by ij
n
X
Fj =Rpj /Rp (3.14f) Kp∗ = Kp∗ij ϕi F̄j (3.16)
ij
Monomer sequence length distributions may be
calculated by using equations given in [356]. when the terminal model is valid; where F̄j
is the mole fraction of monomer j chemically
bound in the accumulated polymer. When a sig-
3.3.1.2. Molecular Masses, Long-Chain nificant number of labile atoms have been ab-
Branching, and Cross-Linking stracted and when a significant number of dou-
ble bonds have been consumed, Equations (3.15)
Linear Copolymer Chains. For chain and (3.16) should be modified to account for this
lengths greater than about 50 and when the [363], [364].
terminal model for copolymerization is valid, When polymer chains that are normally inert
Stockmayer’s bivariate weight chain length to further reaction undergo long-chain branch-
distribution may be used to calculate the com- ing reactions, the instantaneous MWD is no
position and chain length of binary copolymer longer a permanent quantity. Therefore, the
produced instantaneously [356], [362]. The bi- method of moments should be used to calculate
variate distribution of the accumulated polymer the molecular mass averages. For details on the
is readily calculated by integrating the instan- use of the method of moments for the calcula-
taneous distribution using weighting factors tion of sol molecular mass averages before and
based on instantaneous rates of polymer pro- after the gelation point and other methods for
duction. Unfortunately, an analytical function calculating sol – gel fraction and cross-linking
for a multidimensional distribution (involving density, see [363–365].
three or more monomer types) has yet to be
derived. However, part of this problem has al-
ready been solved. In the limit of large chain 3.3.2. Examples of Free-Radical
lengths, the copolymer composition of chains Polymerization
produced instantaneously may be assumed to
3.3.2.1. Homopolymerization – Linear
be independent of chain length, and therefore,
Chains
the instantaneous bivariate weight chain length
distribution is that given by the same expres- The modeling techniques described in Section
sion as for homopolymerization with all of the 3.3.1 are illustrated with actual monomer sys-
chains having the same composition Fi . For tems and experimental kinetic data. The sim-
more details, see Section 2.3.1 and [356]. plest modeling examples (homopolymerization
50 Polymerization Processes

producing linear chains) are considered first, be- a wide range of reactor type and operational con-
ginning with the thermal bulk polymerization of ditions. These include:
styrene, which is relatively easy to model and the
modeling has been most successful in a variety 1) Wu et al. [374] used the model for a theo-
of applications. retical/experimental optimization study em-
ploying temperature programming in a batch
reactor with highly successful results.
Bulk Thermal Polymerization of Styrene 2) Kirchner et al. [375] applied the model to
(T > 100 ◦ C). This system is comparatively a CSTR and obtained accurate predictions.
easy to model for the following reasons: 3) Tien et al. [375] applied the model to a tubu-
lar reactor with internal mixers.
1) The Mayo mechanism for thermal initiation 4) DIERS [376], [377] (Design Institute for
of radicals is valid [366–368] Emergency Relief Systems) a consortium of
2) The polystyrene chains are linear chemical and insurance companies financed
3) The Trommsdorff – Norrish effect (see Sec- the design and construction of a unique
tion 2.2.1.2), although significantly affecting adiabatic reactor system (see Fig. 26) for
the polymerization rate, has at most a minor the measurement of temperature and pres-
effect on molecular mass development be- sure/time variations during adiabatic run-
cause most of the polymer chains are pro- away exothermic reactions. They measured
duced by chain transfer to the Diels – Alder responses for the adiabatic thermal poly-
intermediate [366]. merization of styrene (Fig. 27 and 28) and
compared them with those predicted by
The size of the polymeric radicals (radical
the Hui – Hamielec polystyrene model and
centers are exclusively on chain ends) depends
found excellent agreement.
on temperature and monomer concentration (or
conversion), and the molecular mass distribu-
tion of the accumulated polymer differs little
from that of the polymeric radicals. In general,
the self-diffusion coefficients of polymer radi-
cals should depend on the size of the macrorad-
ical, the concentration and MWD of the accu-
mulated polymer, and temperature. The size of
the macroradical is of greater importance than
the MWD of the accumulated polymer [369],
[370].
An empirical correlation of the bimolecular
termination constant with polymerization tem-
perature and monomer conversion should pro-
vide a model which is applicable over a range
of process conditions (batch and continuous re-
actor operation, temperature programming, etc).
Another important factor, revealed in the work of
Kirchner and Riederle [367], is that reactions
involved in thermal initiation do not become
diffusion controlled at monomer conversions as
high as 97 %. The first effective model for the
bulk thermal polymerization of styrene was de-
veloped by Hui and Hamielec [371] and later Figure 26. DIERS VSP (adiabatic batch reactor)–monitors
extended to higher temperatures (T < 230 ◦ C) temperature and pressure changes during runaway of highly
exothermic reactions [377]
[372] and shown to be valid for a continuous a) Containment vessel (ca. 4 L); b) Test cell; c) Outer can;
stirred-tank reactor up to 280 ◦ C [373]. This d) Guard heater; e) Inner heater; f) Insulation; g) Exhaust
model has been evaluated by many workers over and supply; h) Bypass; i) Fill
Polymerization Processes 51

Figure 27. VSP self-heat rate response for the adiabatic ther-
mal polymerization of styrene (80 wt % styrene and 20 wt % Figure 29. Thermal polymerization of p-methylstyrene:
ethylbenzene) compared with the predictions of the H–H molecular mass development followed by GPC and light
polystyrene model [377]
scattering (LALLS). Deviation for M W between GPC and
– – – Constant heat of reaction (− 700 kJ/kg); —- Temper-
LALLS suggests that LCB frequency is increasing with
ature-dependent heat of reaction; • Experimental data
monomer conversion. Solid curves are model predictions
neglecting chain transfer to polymer [378].

A monomer which polymerizes thermally


and almost identically to styrene is 4-
methylstyrene [378]. With the latter monomer
some long-chain branching occurs due to chain
transfer to methyl hydrogens in the polymer
backbone. An effective experimental technique
to confirm the presence of long branches [378] is
to measure M̄ W by light scattering and by GPC
at different monomer conversions in isothermal
batch polymerization. The molecular mass cali-
bration curve for GPC should be constructed us-
Figure 28. VSP Pressure – time response for the adiabatic ing linear poly(4-methylstyrene). At low conver-
thermal polymerization of styrene (80 wt % styrene and sions, where chains are almost all linear, M̄ W by
20 wt % ethylbenzene) compared with the predictions of the light scattering and GPC should agree. However,
H–H polystyrene model with temperature-dependent heat of at high conversions where long-chain branching
reaction [377]
(kPa = psi × 6.89) should be significant (if it is present at all), M̄ W
a) Prediction: total pressure (maximum 360 psig); by light scattering should be greater than that by
b) Prediction: partial vapor pressure (maximum 312 psig) GPC and the difference should increase with in-
• Experimental data creasing monomer conversion. Figure 29 shows
The pressure calculation required the use of an example of this behavior. Long-chain branch-
the Flory – Huggins equation (Eq. 3.17). These ing in poly(4-methylstyrene) has been satisfac-
evaluations under extreme conditions confirm torily modeled by accounting for chain trans-
that the H–H model for the thermal polymer- fer to polymer and using light scattering M̄ W ’s
ization of styrene is indeed useful in a vari- which should be valid for branched homopoly-
ety of applications. These kinds of evaluations, mer [378]. Another effective procedure to verify
which, however, have not been carried out for the presence or absence of long-chain branching
any other polymerization model, should be con- is to add a previously synthesized narrow-MWD
sidered standard tests for polymerization models sample of the polymer that is to be synthesized
before they are accepted for general engineering to the monomer solution and then polymerize.
use. The newly synthesized polymer is then com-
52 Polymerization Processes

pletely separated from the old polymer by GPC. length-dependent model for bimolecular termi-
If the GPC response for the added polymer shifts nation. This work has been summarized in a
towards larger radii of gyration (hydrodynamic recent article by Adams et al. [393]. The first
volume) this is evidence for long-chain branch- important models [381–383] for MMA poly-
ing [379]. merization were carefully evaluated [384] using
comprehensive rate and molecular mass data in-
Bulk polymerization of methyl methacry- volving three different initiator types measured
late (MMA) with free-radical initiator is the by Röhm. The Marten – Hamielec model was
model monomer system for the investigation found to better satisfy the specifications for a
of diffusion-controlled bimolecular termination polymer reactor model that can be used to op-
and propagation and the decrease in initiator ef- timize commercial production systems. It was
ficiency at high monomer conversions (cage ef- pointed out however, that further work should
fect). It has the largest Trommsdorff – Norrish be done to investigate the applicability of the
effect because most of the polymer chains are model to systems that have been prepolymerized
produced by bimolecular termination and the and also polymerized nonisothermally (perhaps
higher average molecular masses (M̄ W , M̄ Z , with temperature programming or adiabatically)
M̄ Z+1 ) increase dramatically in the absence of a and for systems with mixed initiators. Panke
chain-transfer agent. The MWD shifts to higher [394] has more recently shown that when us-
molecular masses and sometimes becomes bi- ing prepolymer, simulations are better when the
modal [380]. GPC detector responses, multi- parameter n is changed from 1.75 to 0.5 in the
plied by monomer conversion so that the total M–H model. From a fundamental point of view
area under the detector response is proportional the M–H model has several weaknesses which
to the amount of polymer in the batch reactor, are should be pointed out. Chain-length dependence
shown in Figure 30. At the monomer conversion of bimolecular termination is accounted for in an
where the number of physical chain entangle- overly simplified manner. The weight-average
ment points becomes significant a spike of high molecular mass of the accumulated polymer
molecular mass polymer is produced. There- does reflect the change in the size of the macro-
after, polymer with lower molecular masses is no radicals but in a dampened fashion. For example,
longer produced. The instantaneous MWD has a sudden change in the size of the macroradi-
clearly shifted towards higher molecular mass cals would not be felt soon enough, particularly
(higher molecular mass polymer has a lower when a substantial amount of dead polymer has
GPC retention volume or retention time). MMA accumulated. This model effectively uses one
is generally polymerized below the glass transi- termination constant (the number-average ter-
tion temperature of PMMA (ca. 110 ◦ C) and as mination constant [358]). It can therefore predict
a consequence the initiator efficiency and prop- polymerization rate and number-average molec-
agation constant decrease dramatically at high ular mass, but invariably underestimates M̄ W
monomer conversions (∼ 80 %) due to restric- and higher molecular mass averages. The M–
tions on diffusion rates during the glassy-state H model as well as most others have assumed
transition because of appreciable loss in free vol- that K p becomes diffusion controlled while the
ume. There is also evidence that during the rapid initiator efficiency f remains constant. It has
autoacceleration in polymerization rate (due to long been known that the initiator efficiency falls
large increase in polymer radical concentration at high conversions [395–398]. To separate the
[360]) radicals become frozen in the glassy-state 1
product f 2 K p which appears in the rate expres-
[358]. Radical pairs which form in the cage sion, accurate molecular mass data at high con-
may become frozen after a few monomer addi- versions are required. This has not been possible
tions because of monomer starvation. This ho- to date, because of the difficulty of measuring
mopolymerization is difficult to model. Never- M̄ N and M̄ W for the very high molecular mass
theless, there have been many serious attempts PMMA produced during the autoacceleration of
to model this system [381–392]. More recent reaction rate. For all of the models for which
research has focused on predicting the fall in f was taken to be independent of monomer con-
initiator efficiency and propagation constant at version, the observed decrease in K p is actually
high conversion and on the use of a simple chain-
Polymerization Processes 53
1
the decrease in f 2 K p . The change in K p at very copolymers synthesized with vinyl monomers
high monomer conversions can be measured by having a wide variety of structures and range
ESR [360], [399] and of course this would per- of reactivities. The terminal model has not been
1
mit the estimate f from the product f 2 K p . An- as successful in the prediction of comonomer
other factor which has been completely over- sequence length distributions [405], [406] and
looked is that during the autoacceleration when the propagation rate [407–413]. Model discrim-
approaching the glassy-state transition, the ac- ination has been based on the use of mea-
tual volume of the polymerizing mixture may surement techniques which provide estimates
be significantly greater than the local equilib- of the elementary overall (or pseudo-kinetic)
rium volume. This is largely responsible for the rate constants K p and K̄ tN . These include ro-
fact that the limiting conversion depends on the tating sector [407–409], spatially intermittent
initiator concentration at high initiator and rad- polymerization (SIP) reactor [414], and pulsed-
ical concentration levels [400–402]. laser techniques [411–413]. ESR measurements
[410] have also been used to show the ex-
istence of penultimate effects. Most of these
studies considered the binary copolymerization
of styrene and methyl methacrylate; however,
other monomer pairs have been investigated in-
cluding: styrene – acrylonitrile [408] and styrene
with a series of alkyl methacrylates [408], [415].
The use of the pulsed-laser technique has a se-
rious limitation when applied to copolymeriza-
tion. One must measure the MWD by GPC and
the potential errors are several and potentially
serious. For example, there is inevitable statis-
tical broadening of composition distribution for
short chains and the uncertainty that the con-
cept of universal molecular mass calibration is
valid. Nonuniform composition of chains in the
detector cell makes the conversion of detector
Figure 30. GPC chromatograms showing molecular mass response to polymer solute concentration uncer-
development due to the Trommsdorff–Norrish effect for the tain and in addition the errors due to deviation
isothermal bulk free-radical polymerization of methyl meth- from the universal calibration curve may be sig-
acrylate in a batch reactor
Parameter x denotes monomer conversion (T = 70 ◦ C, nificant.
[AIBN] = 0.3 wt %) [380] Most of the serious attempts to model binary
copolymerizations from an industrial perspec-
During the autoacceleration in polymer- tive (accounting for diffusion-controlled bimo-
ization rate which accompanies diffusion- lecular termination and propagation and for
controlled bimolecular termination, it is dif- the fall in initiator efficiency at high conver-
ficult to maintain isothermal conditions. The sions) have employed the terminal model and
use of large ampoule reactors with inadequate the pseudo-kinetic rate constant method [355],
surface-to-volume ratios for heat transfer has [357], [416–423]. There have been some recent
been recently criticized [403], [404]. The use attempts to employ the penultimate model in this
of ampoule reactors (3 – 5 mm diameter) is rec- regard [424], [425]. It should be stressed that
ommended for kinetic investigations of the whether terminal or penultimate models are used
Trommsdorf – Norrish effect [404]. appropriate pseudo-kinetic rate constants can be
defined for calculating rates of polymerization
and molecular mass distribution [356].
3.3.2.2. Copolymerization – Linear Chains

The terminal model has been very useful for


predicting the average compositions of binary
54 Polymerization Processes

3.3.2.3. Copolymerization – Long-Chain ratios for reactors are usually < 2 to achieve
Branching acceptable mixing. When jacket cooling is no
longer sufficient to maintain isothermal poly-
The few serious attempts to model polymer- merization, additional modes of heat transfer
ization rate and molecular mass development must be used, as shown in Figure 31. Internal
have employed the pseudo-kinetic rate con- cooling coils are often not practical because they
stant method with the method of moments tend to interfere with stirring. External tubular
[355], [357], [363–365], [418], [419], [422]. The coolers can, in principle, provide a very large
method of instantaneous MWD, which is such a heat-transfer area, but they may have very large
powerful method for calculating the full MWD pumping requirements in the case of highly
for linear multicomponent polymers is not use- viscous solutions. Reflux cooling removes the
ful when dead polymer chains can be reactivated heat of polymerization by evaporation of solvent
during formation of long-chain branches. The and/or monomer; the condensed vapor is recy-
instantaneous MWD is no longer a permanent cled to the reacting mass. Condensers may be as
quantity but loses some of its chains, which be- large as necessary, with the limiting factor usu-
come branched during the course of polymer- ally the amount of vapor that can be treated with-
ization. It should be noted here that the pseudo- out causing intense foaming or spattering of the
kinetic rate constant method is equally valid for polymer solution. Remixing of the condensed
the modeling of linear and branched copoly- liquid with the more viscous reacting mass may
mer chain synthesis. The most common ana- also be difficult.
lytical technique for the measurement of long-
chain branching frequency (average number of
branches per polymer molecule or per 1000 3.3.3.1.2. Addition of a Solvent in which
backbone carbon atoms) is GPC with a dual de- both Monomer and Polymer are Miscible
tector system (mass concentration detector plus
either a viscometer or light scattering photome- The addition of a solvent in which both monomer
ter as detector in series) or GPC with off-line and polymer are miscible lowers the viscosity
viscometry [426]. of the reacting mass, thereby improving its flow
and heat-transfer characteristics. As a result, and
depending upon the nature and con-centration
3.3.3. Polymerization Processes of the solvent, the Trommsdorff – Norrish effect
can be either completely suppressed or signifi-
3.3.3.1. Solution Polymerization cantly reduced. Use of solvent can be especially
beneficial when evaporative cooling is used. In
3.3.3.1.1. Polymer Soluble in Monomer choosing a solvent, it is important to take into ac-
count the possibility of chain transfer to solvent
Kinetics. A polymerization in which the with a concomitant reduction in polymer mo-
polymer being synthesized is soluble in its lecular mass. Removal of solvent and residual
monomer may be called a bulk or mass as well as (unreacted) monomer from the highly viscous
a solution polymerization. A full description of polymer solution requires very high surface ar-
the polymerization kinetics and modeling tech- eas to permit rapid devolatilization at the mod-
niques can be found in Section 2.2.1 and Section erate temperatures required to minimize poly-
3.3.2. merization and degradation of chains during de-
volatilization to reduce off-spec polymer. Fig-
Polymerization Processes. The main tech- ure 32 illustrates possible designs of devolatiliz-
nical problems associated with solution poly- ers. One of these, the vacuum degasser, incorpo-
merization are heat removal, recovery of resid- rating a type of spray device [427], operates adia-
ual monomer and solvent, and the manipula- batically, which means that the heat for evapo-
tion of highly viscous solutions and melts. It is ration is supplied by the solution itself. In con-
well known that the ratio of cooling surface area trast, a degasser in series with a tubular heat ex-
to volume of reacting mixture decreases as the changer causes some of the monomer/solvent to
reactor volume increases. Length-to-diameter
Polymerization Processes 55

Figure 31. Alternative methods of heat removal for polymerization conducted in a stirred reactor
A) Internal cooling coils; B) External cooler; C) Reflux cooling

evaporate during passage through the heat ex- [435]: dilution with solvent and, in most cases,
changer. This evaporation accelerates the flow of an increase in temperature increases the compat-
product, thereby increasing the heat-transfer rate ibility; increasing the molecular masses of the
[428]. Twin-screw extruders with one or more polymers has the opposite effect. Except in the
vapor outlets that can be connected to a vac- case of high dilution, incompatibility in solution
uum source are also utilized for the removal of is the rule for pairs of polymers, even when the
monomer and solvent [429], [430]. Intermesh- solvent is a good solvent for both polymer types.
ing and self-cleaning screws provide continuous Systems such as polymer A – polymer B –
renewal of the evaporating surface. The use of block or graft copolymer AB can be re-
thin-film evaporators with rotating wiper blades garded as a polymeric oil-in-oil emulsion in
has also been suggested [431]. which the copolymer functions as an emul-
Given stable operation of the devolatilizer, sifying agent [436–439]. Such systems arise,
solution polymerization can be carried out in for example, in the manufacture of high-
a continuous stirred-tank reactor [432], [433]. impact polystyrene (HIPS), or in the prepara-
Conversion is normally lower than with tower tion of ABS (acrylonitrile – butadiene – styrene)
processes, however, requiring large amounts of by solution polymerization. In the simplest
monomer/solvent to be recovered and recycled case, ca. 5 – 10 % polybutadiene is dissolved in
to the reactor. Special attention must therefore be monomeric styrene to give a homogeneous so-
paid to the buildup of impurities (e.g., inhibitors, lution suitable for polymerization. The poly-
chain-transfer agents) over time by using appro- styrene synthesized is incompatible with the
priate purge streams. Problems with processes polybutadiene present, causing phase separa-
involving step-growth polymerization are dis- tion even at very low monomer conversion
cussed in Section 3.2. and producing a polybutadiene – styrene con-
tinuous phase and a polystyrene – styrene dis-
perse phase. Simultaneously, graft polymeriza-
3.3.3.1.3. Polymer – Polymer Demixing tion produces polystyrene branches on polybu-
during Polymerization tadiene backbone. The graft copolymer serves
Polymer – polymer demixing is especially im- as an emulsifier, accumulating at the interface
portant in the production of thermoplastics and stabilizing the oil-in-oil emulsion.
whose application characteristics are enhanced With increasing conversion, the volume frac-
by the presence of dispersed domains containing tion of the polystyrene phase (which is initially
an elastomer. Generally, the thermodynamics small) increases considerably due to formation
are unfavorable for complete miscibility bet- of additional polystyrene which absorbs styrene
ween different types of polymers. For a system monomer. Finally, often at a phase-volume ra-
polymer A – polymer B – solvent, the follow- tio of about unity, a phase reversal occurs, with
ing generalizations are usually applicable [434], the rubber phase now the disperse phase, and the
56 Polymerization Processes

Figure 32. Options for removing residual monomer and solvent


A) Strand degasser; B) Tubular evaporator; C) Degassing extruder; D) Thin-film evaporator

polystyrene phase the continuous phase. Agita- tures, since the other morphologies shown gen-
tion is essential for the completion of phase re- erally provide lower impact strength.
versal, since rapid approach to equilibrium with
respect to transfer of monomer/polymer must be
attained in a highly viscous medium. A common 3.3.3.2. Precipitation Polymerization
result of inadequate agitation is the interpenetra-
tion of two continuous phases [438], [440–442]. The term precipitation polymerization refers to
Phase reversal can also lead to various processes in which the initial ingredients of
types of emulsions; e.g., polystyrene – styrene a recipe are soluble, giving a homogeneous
droplets can be occluded within the rubber par- solution, but the synthesized polymer precipi-
ticles of the disperse phase. This is actually de- tates during the course of polymerization. The
sirable as it increases the volume fraction of the precipitated polymer is generally swollen with
disperse rubber phase. A reinforcing effect is monomer and with nonsolvent if present. To
observed, with the resulting HIPS having higher ensure convenient handling the disperse phase
impact strength for a given mass of rubber in must be finely divided. This is achieved by ef-
comparison with materials prepared by emul- fective agitation, and a protective colloid (steric
sion grafting, in which occluded polystyrene is stabilizer) is often used. The associated prob-
not formed [443], [444]. Rather than rely on in lems are similar to those with HIPS manufac-
situ formation of grafted rubber, it is also possi- ture.
ble from the outset to utilize block copolymers
of styrene and butadiene as an additive in the
recipe. The desired particle size distribution and 3.3.3.2.1. Polymer Insoluble in its Monomer
morphology of the rubber particles in HIPS is
then achieved by varying the intensity of agita- Polymers that are insoluble in their monomers
tion, the viscosities of the disperse and continu- include poly(vinyl chloride) and other polymers
ous phases, the graft activity of the primary rad- derived from halo or pseudohalo-substituted
icals of the initiator, the molecular mass of the ethylenes such as vinyl bromide, vinylidene
continuous phase, and the mass fraction and mo- chloride, trifluoroethylene, and acrylonitrile.
lecular mass of the blocks in the diblock copoly- Polyethylene also falls into this category, at least
mer [445–454]. when produced under moderately high pressure.
Figure 33 provides examples of the morphol- For the system vinyl chloride – poly(vinyl chlo-
ogy of these rubber-modified polystyrene sys- ride), for example, virtually no polymer dis-
tems. Morphologies of commercial significance solves in the monomer in the temperature range
are essentially limited to cell and capsule struc- 30 – 60 ◦ C. Solubility of the monomer in the
polymer is described well by the simplified
Polymerization Processes 57

Figure 33. Possible morphologies for rubber particles in impact resistant polystyrene [454]
Length of the scale bar: 1 µm

Flory – Huggins equation (3.17) [435, p. 511], grows at the expense of the monomer-rich phase
[455–457]: with increasing monomer conversion. This ex-
  plains the autoacceleration in rate as monomer
p
ln = ln (1 −ϕP ) +ϕP +χϕ2p (3.17) conversion increases.
p0

where
p0 is the vapor pressure of pure vinyl chloride
p is the partial pressure of vinyl chloride
ϕp is the volume fraction of poly(vinyl chlo-
ride)
χ is the polymer – solvent interaction param-
eter.
According to [457], deviations from Equa-
tion (3.17) are expected at low vapor pressures.
The conversion – time curves for the bulk poly-
merization of such monomer – polymer systems
all display rate increases with increasing con-
version. This phenomenon has been especially Figure 34. Time – conversion curve for the bulk polymer-
thoroughly investigated for the bulk polymer- ization of vinyl chloride [459]
ization of vinyl chloride. Its cause may be in- Polymerization temperature 50 ◦ C, initiator lauryl perox-
terpreted as follows: even at very low conver- ide, theoretical curves dashed
[I]: • 0.78 × 10−3 ;  1.57 × 10−3 ; △ 3.38 ×10−3 ; ×
sion, poly(vinyl chloride) precipitates, forming 5.50 × 10−3 mol/mol vinyl chloride
a monomer-swollen polymer-rich phase. Initia-
tor is partitioned between the phases, and thus Talamini has suggested that such a reac-
radical generation and polymerization occur in tion may be described in terms of two homo-
both phases. The bimolecular termination rate geneous polymerizations occurring in parallel,
is diffusion controlled in the polymer-rich phase with rapid monomer transfer from the monomer-
and hence the concentration of radicals is higher rich to the polymer-rich phase [458], [459]. As-
in this phase. Even though the monomer con- suming that the monomer/polymer mass ratio in
centration is lower in the polymer-rich phase, the polymer-rich phase is independent of con-
the specific polymerization rate Rp is higher. In version or time, and that the ratio of the spe-
addition, the volume of the polymer-rich phase cific polymerization rates in the two phases re-
58 Polymerization Processes

mains constant (Talamini suggests a ratio of ca. to which a continuous supply of monomer flows
19 at 50 ◦ C), then the observed conversion – time from the gaseous phase. Commercial examples
curve can be described well up to ca. 70 % con- of gas-phase processes include the Union Car-
version, as shown in Figure 34. However, Ta- bide Unipol process [473–476] and the BASF
lamini’s model fails at higher conversions be- process [477–481] for low-pressure polymeriza-
cause the monomer-rich phase disappears, and tion of ethylene and propylene (→ Polyolefins,
polymerization now occurs only in the remain- Chap. 1.5.3., → Polyolefins, Chap. 2.5.6.). Fig-
ing polymer-rich phase, with a monomer con- ure 35 shows a schematic of the Unipol process
centration which decreases with increasing re- in which polyethylene powder is produced us-
action time. Several authors have offered exten- ing ethylene as a fluidizing gas in a fluidized-
sions and modifications of the original Talamini bed reactor that contains a modified chromium
model [460–468]. catalyst. A high-density polyethylene (HDPE)
which has a very broad molecular mass distri-
bution and linear chains is produced. Copoly-
merization of ethylene with propene, 1-butene,
and 1-hexene gives products with a controlled
amount of short-chain branching and lower
polymer density (LLDPE) [482]. The densities
of LLDPE are between those of HDPE and
LDPE made in the high-pressure free-radical
processes.

Polymer – monomer – precipitant systems


employ a nonsolvent for the polymer, while
the precipitant is miscible with the monomer.
The most important example is probably the
so-called slurry process for the manufacture
of high-density polyethylene (HDPE), iso-
tactic polypropylene, and their copolymers
Figure 35. Schematic diagram of the Union Carbide gas-
(→ Polyolefins, Chap. 1.5.2., → Polyolefins,
phase process for manufacturing HDPE [474] Chap. 2.5.1.) Ziegler – Natta (transition-metal
a) Fluidized-bed reactor; b) Catalyst transfer tanks; c) Cata- catalysts) are used in these processes. Polymeri-
lyst feeders; d) Product discharge tanks; e) Multiclone dust zation is often conducted continuously in a train
separator; f) Air coolers; g) Compressor; h) Product de-
gassing tank; i) Filter; j) Ethylene tank; k) Pneumatic trans-
of well-mixed reactors in the presence of a non-
port system solvent hydrocarbon, usually a C6−7 hydrocar-
bon, at ca. 50 – 100 ◦ C and 5 – 30 bar. Polymer
An industrial example of precipitation poly- chains grow on suspended, very fine catalyst par-
merization is the Pechiney – Saint Gobain two- ticles. Soluble, homogeneous catalysts of the
stage bulk polymerization process [469–472], Ziegler – Natta type are also known and have
or the more popular suspension polymerization gained commerical interest. The productivity of
where poly(vinyl chloride) precipitates in the modern catalysts is so high (greater than 100 kg
monomer droplets (→ Poly(Vinyl Chloride)). polyethylene per gram transition metal) that sub-
Precipitation polymerization also encompasses sequent removal of the catalyst from the polymer
what has been called “gas-phase” polymeriza- product is not required. An overview of earlier
tion, a process in which polymer particles form developments with respect to these catalysts is
within a monomer vapor. Such polymerization provided in [483].
does not actually occur in the gas phase, how- Stirred-tank reactors are generally used for
ever, because the catalyst resides either within this process, often linked in series to give
or on the surface of existing polymer particles, a cascade or train [484], [485]. A simplified
and a significant amount of monomer is dis- schematic of the Hoechst process is shown
solved in the polymer. The actual site of poly- in Figure 36. Similar slurry processes also ex-
merization is thus within the polymer particle, ist for polypropylene [486] and copolymers
Polymerization Processes 59

Figure 36. Schematic of the Hoechst slurry-process for HPDE [484]


a) Mixing vessels; b) Reactor; c) Postreactor; d) Centrifuge; e) Stripper; f) Dryer; g) Silo for HDPE powder

based on ethylene – propene (EPM) and ethyl- A similar process is used by Exxon for the
ene – propene – diene (EPDM) mixtures. Acry- synthesis of butyl rubber [489]. In this case, iso-
lonitrile, sometimes together with a comonomer, prene serves as the comonomer and methyl chlo-
can be polymerized in aqueous phase using ride takes the place of ethylene. Polymerization
a redox system such as K2 S2 O8 –Na2 S2 O5 or is carried out in a continuous stirred-tank reac-
H2 O2 –Fe2+ [487]. The polymer precipitates as tor. Residual monomer and diluent are removed
a fine powder, which is then filtered, washed, from the resulting polymer suspension by strip-
and dried. Dissolution of the polymer powder ping with hot water.
in dimethylformamide produces a spinning so-
lution that can be used to produce polyacryloni-
trile fibers. This process has been conducted in a 3.3.3.2.2. Monomer Functioning as Solvent
continuous manner [488]. One of the first com- for the Polymer
mercial precipitation polymerizations was the
belt conveyor process for polyisobutylene devel- The Polymer – Monomer – Precipitant
oped by BASF (Fig. 37; see also → Polyolefins, System. If the polymer is soluble in its
Chap. 4.2.4.). Equal amounts of isobutene and monomer, precipitation polymerization requires
ethylene are first mixed and then deposited in the addition of a precipitant that is miscible with
the liquid state on a circulating steel belt. A sec- the monomer. Polymerization then begins in a
ond inlet is used to add a solution of BF3 in solvent – nonsolvent mixture and it ends in a
ethylene. Cationic polymerization occurs very phase consisting only of pure precipitant once
rapidly, and the heat of polymerization is dissi- conversion of the monomer is complete. Thus,
pated by the evaporation of ethylene. Thus, eth- the solubility relationships change during the
ylene functions here not only as a precipitant but course of the polymerization, and polymeriza-
also as a medium for evaporative cooling. The tion may initially occur in a homogeneous man-
resulting gaseous ethylene is purified, liquefied, ner prior to the onset of precipitation, induced by
and recycled. the enrichment in precipitant that accompanies
the consumption of monomer.
60 Polymerization Processes

Figure 37. Schematic diagram of the BASF process for manufacturing polyisobutylene [433]
a) Storage vessel for liquid isobutene; b) Storage vessel for liquid ethylene; c) Refrigerating condenser; d) Storage vessel
for ethylene containing 0.03 % boron trifluoride; e) Conveyor belt reactor; f) Heated degassing screw; g) Gaseous ethylene;
h) Purification with calcium oxide ; i) Gasometer; j) Compressor


γprec =
volume of precipitant
(3.18)
volume of precipitant + volume of solvent

is plotted in Figure 38 as a function of conversion


for various initial concentrations of monomer
[M]0 . The dashed lines represent extrapola-
tions from experimental cloud-point titration
data, which give the ∗γ prec values at which the
first polymer fractions precipitate at 65 ◦ C. For
an initial concentration [M]0 < 4.1 mol/L, ex-
tremely small conversions suffice to cause pre-
cipitation. At high monomer concentrations the
polymer initially remains in solution because the
volume fraction of methanol required for poly-
mer precipitation has not yet been reached. In
such cases, the critical value of ∗γ prec is ex-
ceeded only at higher conversion: the greater
[M]0 , the higher the conversion required.
In precipitation polymerization of this type it
Figure 38. Volume-fraction ∗γ prec (—-) of methanol as a is especially important to ensure that the poly-
function of conversion for various initial concentrations of
monomer [M]0 [491] mer precipitates in finely divided form. Among
Azeotropic precipitation copolymerization of other things this ensures that the particles do
styrene – acrylonitrile in methanol (– – –), precipita- not overheat during polymerization. The pro-
tion point determined using cloud-point titration for two cess has been employed on large scale by BASF
different degrees of polymerization
[490–492], with batch polymerization occur-
Figure 38 illustrates these relationships, us- ring in stirred reactors under reflux at 65 ◦ C.
ing as an example the copolymerization (in Azobisisobutyronitrile was employed as inita-
methanol as precipitant) of 40 mol % acryloni- tor, and poly(vinyl pyrrolidone) or poly(vinyl
trile and 60 mol % styrene (at the azeotropic ether) as a protective colloid. The resulting
composition). The volume fraction ∗γ prec of styrene – acrylonitrile copolymer was separated
methanol in the reaction mixture, defined as by centrifugation and the recovered methanol
was distilled and recycled to the next batch.
Polymerization Processes 61

Polymer dispersions in nonaqueous media Figure 39 A is a schematic representation of


with particle size in the range 0.01 – 10 µm (non- the adsorption of di-block, multiple-block, and
aqueous dispersions, colloids, organosols) have graft copolymers on the surface of growing poly-
been thoroughly investigated by Barret and mer particles. Soluble and insoluble portions of
coworkers [493]. Barret characterizes “disper- the dispersant molecules must be kept carefully
sion polymerization” as a process in which an in balance. If the insoluble part is too small, or
insoluble – and thus dispersed – polymer is pre- if it interacts too weakly with polymer particles,
pared from a monomer dissolved in an organic then adequate adsorption will occur only when
diluent to which has been added an amphiphatic the dispersant concentration in the continuous
block or graft copolymer to serve as a dispersant. phase is very high. If the soluble portion is too
However, here this is regarded as a special case large, the dispersant will be present largely as
of precipitation polymerization in which com- aggregates or micelles with little tendency to
plete coagulation of polymer particles is pre- dissociate and be adsorbed on the interface. Fig-
vented and the particle size is controlled. ure 39 B depicts the equilibrium situation. Fi-
The key to controlling particle size is the se- nally, it is important to note that multifunctional
lection of dispersant type. Among the most ef- amphipathic molecules like those shown in Fig-
fective dispersing agents are the so-called am- ure 39 C can also function as weak flocculating
phipathic molecules: block and graft copoly- agents.
mers made up of two polymeric components, In a well-stabilized dispersion, each particle
only one of which is insoluble in the continu- is covered by a layer of freely moving poly-
ous, diluent-containing phase. Graft copolymers mer chains, which are in turn dissolved in the
of this type often form during the polymerization continuous phase. These layers prevent the fre-
as a result of grafting on the dissolved polymer, quently colliding particles from approaching so
but it is not absolutely essential that the insoluble closely that Van der Waal’s attractive forces be-
portion of the dispersant be identical to or solu- come dominant. A simplified model (Fig. 40)
ble in the disperse phase. In many cases its insol- suggests that the mechanism of steric stabiliza-
ubility in the diluent is sufficient to ensure ade- tion involves an increase in the local concentra-
quate adsorption on the particle surface. These tion of polymer chains or segments as a result of
amphipathic dispersing agents act as steric sta- overlapping and mutual chain interpenetration
bilizers [494]. [496] as two polymer particles approach. This
induces an osmotic pressure and increase in the
free energy ∆G R . To compensate for this effect,
solvent flows into the regions of higher polymer
concentration and drives the particles apart. A
positive value of ∆G R = ∆H R − T ∆S R can be
due either to enthalpic effects (∆H R ) or to en-
tropic effects (T ∆S R ). It is therefore possible
to divide the contributions to steric stabilization
into three categories [498]:
1) Enthalpic stabilization
∆H R and ∆S R are both positive
∆H R > T ∆S R
The dispersion flocculates on warming
Figure 39. Steric stabilization of precipitating poly- 2) Entropic stabilization
mer/monomer particles with the aid of amphipathic block ∆H R and ∆S R are both negative
and graft copolymers [495]
A) Schematic representation of adsorption of amphipathic T ∆S R > ∆H R
molecules on the polymer particles; —- insoluble groups, The dispersion flocculates on cooling
· · · · · soluble groups; B) Equilibrium established in the 3) Combined enthalpic – entropic stabilization
course of a precipitation polymerization P = (growing) poly- ∆H R is positive
mer particle; C) Schematic representation of flocculation
due to multifunctional amphipathic molecules ∆S R is negative
62 Polymerization Processes

The dispersion is stable over a wide range of With protective colloid:


temperature. Temperature 80 ◦ C
Reaction time 8h
Styrene 100 parts (by mass)
Water 200 parts (by mass)
Benzoyl peroxide 0.4 parts (by mass)
Poly(vinyl alcohol) 0.5 parts (by mass)
[partially hydrolyzed poly(vinyl acetate)]

With pickering emulsifier:


Temperature 80 ◦ C
Reaction time 10 h
Styrene 100 parts (by mass)
Water 200 parts (by mass)
Figure 40. Model for steric repulsion caused by the overlap Benzoyl peroxide 0.2 parts (by mass)
of two spheres containing dissolved molecular chains [496], Barium sulfate 1.0 part (by mass)
[497] C18 SO3 Na 0.002 parts (by mass)
C = Concentration of the polymer chains in the adsorption
layers

More complete discussions of steric stabi-


lization may be found elsewhere [495], [498– The term “suspension polymerization” is per-
503]. A survey of precipitation polymerization haps inappropriate, because precipitation and
is available [504]. emulsion polymerizations also produce suspen-
sions of polymer particles in a continuous phase.
The distinction from precipitation polymeriza-
3.3.3.3. Suspension Polymerization tion is that it is initiated in a homogeneous
mixture, while suspension polymerization takes
Definition. The term “suspension polymeriza- place in an emulsion. The beads or powder par-
tion” includes a series of processes, all of which ticles produced in a suspension polymerization
involve emulsifying monomers to droplets by are roughly of the same size as the original
stirring them in a suspending medium in which monomer droplets with diameters on the or-
they are insoluble in the presence of a free- der of 10−3 to 0.5 cm. Emulsion polymeriza-
radical initiator, usually one that is soluble in tion also starts with a monomer emulsion, but
the monomer. When the polymer formed is the initiator is usually one that is soluble in the
soluble in the monomer, nonporous spherical continuous suspending phase rather than in the
“beads” are formed, hence the term “suspension monomer. Moreover, the resulting latex particles
bead polymerization”. If, however, the polymer are very much smaller (diameter range 5 × 10−6
precipitates during polymerization, the result- to 3 × 10−5 cm or 0.05 to 0.3 µm) than the orig-
ing polymer particles are composed of many inal monomer droplets. Borderline cases which
smaller primary particles. They are opaque, usu- might be called emulsion or suspension poly-
ally possess an irregular surface, and may have merization are discussed later in Section 3.3.3.4.
substantial internal porosity. This type of poly- In the vast majority of cases, the suspend-
merization has been called “suspension pow- ing medium for suspension polymerization is
der polymerization”. The dispersants (protective water, although inverse-suspension polymeriza-
colloids) are either macromolecules that are in- tions are also known and used commercially
soluble in the suspending medium or insoluble, to produce very high molecular weight poly-
usually inorganic, powders, the so-called Pick- mers and copolymers based on the comonomer
ering emulsifiers [505]. Their function is first acrylamide. Here a water-soluble monomer is
to assist in the formation of the initial monomer dispersed in a hydrophobic organic suspending
emulsion and then to stabilize the resulting poly- medium, usually in the presence of water in the
mer particle suspension. The following recipes disperse phase.
are examples for suspension polymerization in-
volving protective colloid and inorganic powder
dispersants.
Polymerization Processes 63

3.3.3.3.1. Qualitative Description in the tacky intermediate stage, individual poly-


mer particles tend to form incompletely fused
Generally, dispersants are employed at a con- clumps. Coagulation at this critical stage of con-
centration (relative to the aqueous phase) of version is somewhat inhibited by the action of
0.1 – 5 wt % in the case of protective colloids the dispersant, but other effective measures to re-
and 0.1 – 2 wt % for Pickering emulsifiers. A duce coagulation may also be taken, including
typical initiator concentration is 0.1 – 1 wt % adjusting the densities of the two phases to make
relative to the monomer. The volume ratio them more similar, or by increasing the viscos-
monomer/aqueous phase is usually between ity of the aqueous continuous phase. Rapid poly-
25 : 75 and 50 : 50, and the stereometric limit, merization during the sticky stage minimizes the
which cannot be exceeded with spheres of uni- number of collisions among polymer particles
form size, is 74 : 26. and thus should reduce coagulation.
The reactor vessel is usually a stirred tank. An experiment with dye-labeled monomer is
The monomer is subjected either to turbulent also applicable when polymerization occurs in
pressure fluctuations or viscous shear forces, the disperse phase [506–508]. In this case dye-
which break it into small droplets that assume containing polymer beads generated in a paral-
a spherical shape under the influence of inter- lel polymerization are added to the suspension
facial tension. These droplets undergo constant during polymerization. It is observed that be-
collision (collision rate ≥ 1 s−1 ), with some of yond a certain conversion (which depends on
the collisions resulting in coalescence. In the ab- the reaction conditions) coagulation and parti-
sence of stabilizers, a dynamic equilibrium is cle breakup cease entirely. This is known as the
eventually established, leading to a stationary “particle identity point” or the limit of dynamic
mean particle size. Individual drops do not re- equilibrium. Increasing the dispersant concen-
tain their unique identity, but instead undergo tration (i.e., reducing the size of particles) dis-
continuous breakup and coalescence. This phe- places the identity point towards lower conver-
nomenon can easily be demonstrated by the ad- sion [509], [510].
dition of a small amount of a monomer that has The schematic diagram in Figure 41 shows
been labeled with a water-insoluble dye. The dye these fundamental relationships. The upper part
is rapidly distributed uniformly in the disperse of the diagram corresponds to dynamic equilib-
monomer phase. rium in a monomer emulsion that contains no
In some cases, an appropriate dispersant can dispersant, or to an emulsion undergoing poly-
be used to induce the formation of a protective merization that has not yet reached the identity
film on the droplet surface. As a result, pairs or point. The monomer phase is broken up into long
clusters of drops that tend to coalesce are bro- strands by the stirrer, and these in turn fragment
ken up by action of the stirrer before the crit- into spherical droplets due to interfacial tension
ical coalescence period elapses. A stable state forces. The droplets may then agglomerate into
is ultimately reached in which individual drops larger aggregates, finally coalescing into larger
maintain their identities over prolonged periods drops, which either break apart again due to stir-
of time. In this case, addition of dye-bearing ring or collect as an extended monomer phase.
monomer does not result in migration of the dye The lower portion of Figure 41 shows the var-
into other droplets. Such a system is described ious stages that lie between the identity point
as a turbulence-stabilized emulsion. and the end of the polymerization process. In
In the simple case of a polymer that is a turbulence-stabilized emulsion, one in which
miscible in all proportions with its monomer monomer drops maintain their identity, the prob-
(e.g., styrene and methyl methacrylate), various ability is high that individual drops will poly-
viscosity states of the disperse phase are tra- merize directly to primary beads. Nevertheless,
versed during the course of polymerization. The even in a suspension bead polymerization, ir-
initially nonviscous, liquid monomer is trans- regular particles sometimes appear, composed
formed gradually into an increasingly viscous of many individual smaller polymer particles, a
solution of polymer in monomer, and as conver- consequence of some clusters surviving past the
sion proceeds the disperse phase acquires the identity point.
characteristics of a solid polymer. Particularly
64 Polymerization Processes

Figure 41. Schematic diagram of dispersion and polymerization in a suspension bead polymerization
A) Monomer emulsion in the absence of dispersant, or a polymerization mixture that has not yet reached the identity point;
B) Polymerizing mixture after passing the identity point

A polymerizing bead with a diameter For what has already been said about the
d = 10−2 cm contains ca. 108 growing polymeric course of suspension polymerization, together
radicals. The effects of subdivision due to parti- with various results from the literature [507],
cle breakup in suspension polymerization are not [510], [513], [518], [519], it is possible to estab-
anticipated to occur at the levels found in emul- lish a number of special factors – apart from
sion polymerization (polymer particles are sub- those common to all free-radical polymeriza-
micron in size) where radical contents of 0.5 or tions – that exert an important influence on par-
lower per particle occur [511], [512]. Each bead ticle size and size distribution:
may be regarded as a small, isolated reactor. For 1) Geometric factors of the reactor: profile, type
this reason, the observed polymerization kinet- of stirrer, stirrer diameter D relative to the
ics correspond directly to those for bulk poly- reactor dimensions, bottom clearance of the
merization [506], [513], [514]. Here again the stirrer, and internal fittings
Trommsdorff – Norrish and glass effects must be 2) Operating parameters: stirrer velocity N,
taken into account, as must the effects associated stirring and polymerization time, phase vol-
with demixing when the polymer is insoluble in ume ratio ϕ, fill level of reactor, and temper-
its monomer [459], [460], [462], [515]. Indeed, ature T
the process is sometimes referred to as a water- 3) Substance parameters: dynamic viscosities
cooled bulk polymerization. To ensure that con- η c and η d and densities ̺c and ̺d of the con-
version is as complete as possible, it is common tinuous and disperse phases, and interfacial
to employ mixtures of initiator types with dif- the tension σ.
ferent half-lifes, and to allow the polymeriza-
tion temperature to increase in the final stages During monomer fragmentation, in colli-
of conversion [516], [517]. With many industri- sions leading to agglomeration, and in de-
ally important suspension polymerizations, sim- agglomeration of clusters, the most important
ply preparing a polymeric material is not suffi- consideration is the energy introduced into the
cient. Often a particular particle-size distribu- reaction mixture per unit time. This can be ex-
tion and morphology must be achieved, as in pressed in terms of the mean rate of energy dis-
the manufacture of expandable polystyrene, or sipation per unit mass ε. In the case of a stirred
poly(vinyl chloride), whose particles must later reactor containing baffles and operated with a
absorb substantial quantities of plasticizer for high Reynolds number, the following equation
some applications. is applicable [520]:
ε̄ =KN 3 D2 cm2 s−3 or W kg−1
 
(3.19)
Polymerization Processes 65

where K is a dimensionless constant that de- tially hydrolyzed poly(vinyl alcohol), the resid-
pends on the type of stirrer. Increasing the stirrer ual acetate groups act as hydrophobic points of
speed N causes a decrease in particle size. attachment, while the OH groups are directed
The concentration of dispersant [S] deter- into the continuous aqueous phase. The most
mines the maximum particle surface area that important factor in dispersant effectiveness is
can be stabilized and influences the interfacial an appropriate balance between hydrophilic and
tension. Increasing [S] leads to a decrease in hydrophobic groups [521]; molecular mass is of
particle diameter, as does lowering the interfa- considerably less significance. The wettability
cial tension. Increased disperse phase viscosity of a polymer can be varied by the addition of
reduces particle breakup and thus leads to larger trace amounts of a low molecular mass surfac-
particles. tant.
Higher concentrations of protective colloid
may cause a bead polymer to be contami-
nated by a small amount of a much more
finely divided emulsion polymer. This should be
avoided because it results in polymer loss dur-
ing workup. The chance of encountering such
a problem is enhanced by increased solubility
of the monomer and initiator in the continu-
ous phase. However, emulsion polymerization
via homogeneous nucleation in the continuous
phase can be suppressed by addition of a water-
soluble inhibitor such as NH4 SCN or a copper
salt [513], [522].

Powdered Dispersants (Pickering Emulsi-


Figure 42. Schematic representation of the adsorption of fiers). Finely divided, usually inorganic, insol-
partially hydrolyzed poly(vinyl alcohol) on the surface of a
dispersed particle [510]
uble solids may also be employed as disper-
All − OH groups (omitted for clarity) are directed toward sants in suspension polymerization [505], [513],
the aqueous phase, and residual acetate groups toward the [523]. Common choices include barium sul-
disperse phase fate, talc, aluminum hydroxide, hydroxyapatite,
tricalcium phosphate, calcium oxalate, magne-
sium carbonate, and calcium carbonate. It is ad-
3.3.3.3.2. Dispersants vantageous to dissolve out the dispersant after
the polymerization (e.g., in dilute acid), thereby
Protective Colloid Dispersants. Organic minimizing polymer contamination. Monomer
protective colloids include natural products emulsions based on systems of this type are re-
such as alginates, tragacanth, agar, and starch ferred to as three-phase emulsions, a term intro-
as well as modified natural polymers such as duced by Wenning [507]. In the case of a poly-
carboxymethylcellulose (sodium salt), hydroxy- mer that is insoluble in its monomer, a fourth
ethylcellulose, and methylcellulose. Among the phase occurs during the course of polymeriza-
effective synthetic polymers are styrene – maleic tion.
anhydride copolymer, poly(methacrylic acid), Solid dispersants must be wet by two immis-
poly(vinyl pyrrolidone), poly(vinyl alcohol), cible liquids, and they must also exhibit a certain
and partially hydrolyzed poly(vinyl acetate). degree of self-adhesion. The wettability can be
The important feature of all these materials modified by adsorption of low molecular mass
is their amphipathic character, which explains surfactants. This technique is referred to as mod-
their ability to lower the interfacial tension and ulation and it has much in common with tech-
to concentrate at the monomer – water interface. niques used in mineral flotation processes.
As shown in Figure 42, protective colloids are The wettability of a solid (S) by a liquid such
probably adsorbed in such a way that they form as water (W) in the presence of a gas (G) de-
loops near the particle surface. In the case of par- pends upon the wetting angle α, which results
66 Polymerization Processes

from equilibrium between three interfacial ten- water-in-oil emulsion and the polymer forms a
sions: block that contains water-filled voids [507]. It is
clear that the limiting cases C and G are inap-
σSW +σWG · cosα =σSG (3.20) propriate for suspension polymerization. A good
Pickering dispersant must possess amphiphatic
characteristics, thus leading to one of the cases
D – F. Many pure inorganic substances are com-
pletely wetted by water, the chief reason why
modulation is so important in practice as a means
of inducing the transition from C to D.

3.3.3.3.3. Mechanism of Particle Formation

Suspension polymerization always begins with


dispersion of a monomer in water. It thus seems
reasonable to first consider the formation of a
monomer emulsion – or more generally, the for-
mation of an emulsion of an organic liquid in
water – and then examine the effects of poly-
merization in the disperse phase.
Figure 43. Wetting of a Pickering dispersant by monomer
and water, modulated by addition of a surfactant Droplet Size in an Emulsion Subject to
S solid; G gas; M monomer; W water Turbulent Mixing. Especially simple relation-
For further explanation, see text ships should govern the suspension polymeriza-
Figure 43 A illustrates the significance of this tion of a turbulence-stabilized emulsion in which
relationship. An angle α = 0◦ corresponds to single drops maintain their separate identities.
total wetting. For α= 90◦ , σ SW = σ SG , and at However, such an emulsion must fulfill a series
α = 180◦ no wetting occurs. In a three-phase of conditions [524–526]:
emulsion, two immiscible liquids compete for 1) Stirring must be sufficiently intensive to sep-
wetting the solid S: monomer M and water W. arate any droplet pairs or clusters that begin
As indicated in Figure 43 B, there is again an to form. The adhesive forces between two
equilibrium condition: drops increase roughly linearly with the drop
diameter d, while the forces exerted on the
σSW +σWM · cosα =σSM (3.21)
drops by stirring show a higher order depen-
Taking the example of barium sulfate as the dis- dence on drop diameter. The probability of
persant in a styrene – water bead polymerization, separation therefore increases with diameter;
the barium sulfate is at first wetted more effec- given a particular stirring intensity, the diam-
tively by water than by styrene, so σ SW < σ SM . eter must remain above a specific minimum
Water wets the barium sulfate with a small con- d min :
tact angle α and cos α is positive (Fig. 43 D).
Addition of a surfactant whose polar groups are −3/8
dmin =C1′ ̺c A (h)−3/8 ε̄−1/4 (3.22)
adsorbed on the barium sulfate makes the surface
more hydrophobic (or lypophilic). As a result, where C ′1 is an empirical constant, ̺c is the
σ SW increases and σ SM decreases, leading to the density of the continuous phase and A (h)
states shown in Figure 43 E and F. In the limiting is the energy required to separate to a dis-
case (Fig. 43 G), the particle is fully immersed in tance h = ∞ two drops of diameter d = 1 ini-
the monomer phase and loses its effectiveness as tially separated by a distance h0 . The value
a dispersant. In the course of passing from D to F of A (h) is strongly dependent upon the thick-
it is not uncommon for a phase reversal to occur: ness and characteristics of the adsorbed pro-
the oil-in-water emulsion is transformed into a tective film. Equation (3.19) may be used
Polymerization Processes 67

to convert Equation (3.22) to the following where C ′3 and C 3 are empirical constants,
form: g is the acceleration due to gravity, ϕ
is the volume ratio (disperse phase vol-
−3/8 ume/continuous phase volume), and f (ϕ) is
dmin =C1′ ̺c A (h)−3/8 N −3/4 D−1/2 (3.23)
an empirical function.
If d < d min , drops will coalesce until a diam- According to [524–526], turbulence stabi-
eter d min has been reached. If energy dissipa- lization is only possible if the drop diameter is
tion in the stirred reactor is not uniform (e.g., larger than d min as given by Equation (3.23),
if only near the stirrer blades is it sufficient to and smaller than both d max from Equation (3.25)
break up drop clusters), then material circu- and d max from Equation (3.27). It is reasonable
lation within the vessel must be sufficiently to assume that the Sauter mean diameter given
rapid to ensure that each cluster approaches by
the stirrer blades at least once during the crit-
ical coalescence time. Σ ηi d3i
d32 = (3.28)
2) The dissipated stirring energy must not be Σ ηi d2i
so large as to cause significant breakup of
individual drops, that is, the critical Weber is proportional to the maximum diameter d max .
number W e (crit) or a maximum drop diam- Figure 44 illustrates these relationships. Stabil-
eter d max must not be exceeded: ity is achieved only in the shaded region bet-
ween the lines defined by Equations (3.23),
(3.25), and (3.27). If ε < εmin , then d min (co-
5/3
W e (crit) =C2′ dmax ̺c σ −1 ε̄2/3 (3.24) alescence) > d max (demixing) and the emulsion
will separate into two phases. On the other hand,
which is transformed by using Equation
if ε > εmax , then d min (coalescence) > d max
(3.19) into:
(breakup) and the emulsion will be unstable
because the energy necessary for separating a
−3/5 3/5
dmax =C2 ̺c σ N −6/5 D−4/5 (3.25) droplet pair is sufficient to breakup the droplets
themselves.
where C ′2 and C 2 are empirical constants and
σ is the interfacial tension. Equation (3.25)
is only applicable if the viscosity ratio η d /η c
is not too large. Otherwise, the stable drop
diameter increases with increasing values of
η d /η c , because turbulent oscillations in the
disperse phase are damped (viscosity stabi-
lization of drop size).
3) A density difference between the phases re-
sults in a tendency towards settling, and this
increases with increasing drop size. Stirring
energy must be sufficient to counteract this
tendency. The maximum drop diameter d max
that can be sustained in an emulsion at a par-
ticular rate of energy dissipation ε is given
Figure 44. Theoretical relationship governing the frag-
by: mentation (Eqs. 3.22 and 3.23), coalescence (Eqs. 3.24 and
3.25), and demixing (Eqs. 3.26 and 3.27) of droplets in a
 3 stirred reactor. Only in the shaded region can a turbulence-
̺c 1
dmax =C3′ f (ϕ) ε̄2 (3.26) stabilized emulsion form [525]
̺d −̺c g3
A prerequisite for the validity of Equa-
and using Equation (3.19) one obtains tions (3.22) – (3.27) is the applicability of Kol-
mogoroff ’s theory of local isotropy [527–529].
3
The characteristic length L of the energy-con-

̺c 1
dmax =C3 f (ϕ) N 6 D4 (3.27)
̺d −̺c g3 taining large eddies (L ≈ 0.08 D, according to
68 Polymerization Processes

[244]) must be larger, and the characteristic in stirring speed has no influence on particle
length l of the energy-dissipating small ed- size. Apparently, a turbulence-stabilized state
dies [l = (γ 3c /ε)1/4 ], where γ c is the kinematic is reached at the very onset of polymerization.
viscosity of the continuous phase] must be Evaluation with respect to particle size is partic-
much smaller than the drop diameter d (i.e., ularly straightforward in this case, since changes
L ≫ d ≫ l). Equations (3.24) and (3.25) [531– in the above-mentioned parameters during poly-
534] as well as (3.22) and (3.23) [525], [526], merization are irrelevant.
[535] have been verified experimentally. Nev- A series of experiments in geometrically sim-
ertheless, they apply only for very small phase ilar reactors (see Fig. 45) coupled with dimen-
ratios ϕ < 0.015, which are unrealistic for indus- sional analysis, permitted derivation of the fol-
trial suspension polymerizations. The polymer- lowing dimensionless equation in the case of this
izing system is also subject to nonnegligible ef- especially simple system:
fects of the viscosity ratio η d /η c . In [536], for ex-  0.1
ample, this has been taken into account by intro- d50 ηd
=kRe0.5 W e−0.9 F r −0.1 ϕ0 (3.30)
duction of the viscosity group η d (̺d σ d)−1/2 , D ηc
which depends only on the physical properties where d 50 is the bead diameter below which
of the disperse phase. At higher phase ratios, 50 wt % of the particles pass through the sieve
both d 32 and d max increase with increasing dis- (in other words 50 wt % of the particles have a
tance from the stirrer [534], [537], [538]. Various diameter less than d 50 ), D is the stirrer diameter,
corrections [531], [534], [539], [540] have been which throughout the experiments was kept in a
suggested for Equation (3.25), all of which take constant ratio to the reactor diameter, and k is a
the form: numerical constant.
d32
=A (1 +Bϕ) W e−3/5 (3.29)
D
where A and B are numerical constants. For
phase relationships such as those applicable on
an industrial scale it is therefore not possible to
assume a homogeneous energy-dissipation rate
ε; instead, it must be anticipated that ε is large
in the vicinity of the stirrer, becoming smaller
in more remote circulation zones. A simulation
program has been described [541–543] that ap-
proaches this problem by dividing the reactor
into various stirring and circulation zones and
then applying Monte Carlo methods.

The Size of the Polymer Particles. Con-


version of a monomer into polymer increases
the viscosity η d and density ̺d of the disperse
phase, and lowers the volume phase ratio ϕ. The
interfacial tension σ also changes, and grafting
reactions between polymeric radicals on the or-
ganic protective colloid and the monomer may
have to be accounted for. As has been demon-
strated by Hopff et al. [544–547] in the case
of a suspension bead polymerization of methyl Figure 45. Dependence of the bead diameter d 50 on the
methacrylate with a relatively high concentra- stirrer velocity N [545]
Experiments conducted in four geometrically similar stirred
tion of a protective colloid [poly(vinyl alcohol), reactors with stirrer diameter D
Mowiol N 70 (88)], systems exist in which the Curve parameter: protective colloid concentration in
final particle size is already established at very g/100 cm3
low conversion. Polymerizing droplets main-
The dimensionless groups are:
tain their identity, and subsequent reduction
Polymerization Processes 69

D 2 N ̺c
 
d
Re = = ratio of inertial/viscous forces =kW e−0.6 Eu−1.0 (3.31)
ηc D
DN 2 or
 
Fr= =
g
ratio of inertial/gravitational forces (3.30a) d ∼D−0.8 N −1.2 (3.32)

D 3 N 2 ̺c has not yet been established experimentally (Eu


 
W e= =
σ is here a modified power number in power per
ratio of inertial/interfacial forces unit volume). A not further defined suspension
polymerization of methyl methacrylate in 1 %
It follows that d 50 = kN −1.5 polyacrylamide solution was transformed with
A relationship similar to Equation (3.30) has constant particle size on the basis of
been found for the suspension polymerization
of vinyl chloride [548], [549]. In this case, how- 
D2
2/3
ever, a large particle diameter region was identi- N1 =N2 (3.33)
D1
fied with d = kN −1.9 at low stirrer speeds (with
correspondingly large particle diameters) as well from a laboratory reactor (1) with D = 10.1 cm
as a fine-particle size region with d = kN −0.6 . to a half-tonne scale (2) with D = 81.2 cm [555].
An analogous effect has also been reported for
styrene [550].
In contrast to the very early establishment of 3.3.3.3.4. Industrial Applications
particle diameter in the experiments of Hopff et
al., dye-labeling experiments have shown that in Suspension polymerization is used for produc-
other systems [506–508], [550], [551] the par- ing a wide variety of polymer types, the most
ticles reach the so-called identity point only at important of which are mentioned briefly. Stan-
high conversion. With these systems it is appar- dard polystyrene for use in injection molding
ently necessary to take into account the time and is manufactured by suspension bead polymer-
conversion dependence of the parameters enu- ization, as is poly(methyl methacrylate) and its
merated above. Another important observation copolymers containing small amounts of acry-
[550], [552] is the fact that in certain systems the late esters. Clear transparent polymers are of-
stirring time in which the unpolymerized emul- ten required, so formulations involving Picker-
sion reaches its final droplet diameter is longer ing dispersant (e.g., MgCO3 ) that can be dis-
than that taken by the polymerizing system to solved out of the polymer with dilute acid after
reach 50 % conversion. A faster polymerization polymerization are particularly advantageous.
would therefore give larger polymer particles In the case of styrene – acrylonitrile copoly-
due to viscosity stabilization. mers, the method of choice for batch suspen-
The problems associated with scaling up sion polymerization is normally that involving
from small to industrial scale, particularly for the azeotropic composition to minimize compo-
particle size, are an important concern in sus- sitional drift. Nevertheless, complications often
pension polymerization. In principle, Equation arise, because considerably more acrylonitrile
(3.30) should be applicable, suggesting that all than styrene dissolves in the aqueous contin-
that is required for maintaining constant d/D uous phase. As conversion proceeds, acryloni-
when scaling up is to keep Re, W e, Fr, and η d /η c trile diffuses into the polymer particles and the
constant. However, Equations (3.30 a) show that monomer ratio in the beads changes, causing
this is not possible without changing the com- the composition of the copolymer to change as
position of the system. Furthermore, it is d, not well [556]. The same phenomenon can accom-
d/D, that must be kept constant in an industrial pany copolymerization in heterogeneous sys-
scaleup. To what extent it is possible at high tems generally [557].
Reynolds numbers to avoid an exact similarity High-impact polystyrene and ABS are often
transformation and scale the process up on the prepared in a combined process. This begins
basis, for example, of [553], [554] with a solution of polybutadiene in styrene or
70 Polymerization Processes

styrene/acrylonitrile and permits bulk polymer- tained. This point marks the beginning of the
ization to occur under stirring until phase rever- so-called “sticky-stage” at which point the co-
sal or inversion has occurred. Water and disper- alescence rate begins to increase, causing the
sant are then added and the polymerization is beads to grow from a mean particle diameter
completed in suspension [558], [559]. of about 0.2 mm to the desired size, depending
Expandable polystyrene is prepared by sus- on the original stabilizer formulation, for the fi-
pension polymerization in the presence of a nal application of the resin product. During this
blowing agent, such as pentane [560], [561]. It stage of the process, particle size is monitored
is also possible to introduce the blowing agent in by periodic sampling of particle size growth
a second step after polymerization, allowing it rate. Additional stabilizer may be added if the
to diffuse into the beads [562], [563]. Warming growth rate is too large. At 65 – 68 % conver-
to 80 – 110 ◦ C, generally with steam, causes the sion of monomer, the identity point is reached.
beads to expand by foaming, their volume in- At this point the particle viscosity is sufficiently
creasing by a factor of ca. 30 – 50. Particle size large that collisions between particles are elas-
and size distributions both play important roles. tic and particle size growth ceases. In addition,
For example, if thin-walled objects are to be the density of dispersed and continuous phases
made from expandable polystyrene, especially are almost identical, and the suspension is very
small polymer particles are required. The un- stable. Autoacceleration of the polymerization
foamed product is fractionated by sieving prior rate becomes appreciable at the identity point
to drying. Figure 46 shows a schematic of a typ- and increases until ca. 95 % conversion where a
ical EPS production facility. glassy-state transition occurs and the beads be-
come hard. Once the beads are hard the reac-
tion mixture is heated to a temperature above
the glass transition temperature of the polysty-
rene (T g ≈ 100 ◦ C). During heating, the reactor
is pressurized with a blowing agent (5 – 8 % with
respect to polymer), a low-boiling hydrocarbon
(C4 – C7 ). The reactor is then pressurized with
nitrogen at 700 – 950 kPa and the so-called im-
pregnation stage starts and proceeds for 3 – 8 h.
During impregnation the blowing agent diffuses
into the beads. At the same time, the free vol-
ume increases, and the finishing initiator rapidly
Figure 46. Schematic representation of the manufacture of
Styropor by batch suspension polymerization generates radicals, causing a relatively rapid in-
a) Mixing tank; b) Stirred reactor; c) Puffer tank; d) Cen- crease to a monomer conversion of ca. 99.9 %.
trifuge; e) Sieving; f) Drying; g) Silo; h) Packaging The impregnation time should be sufficient to
allow the blowing agent to reach the core of the
The required amount of deionized water is particle and to cause the breakup of hydrocarbon
loaded into the reactor at ambient temperature domains within the polymer matrix, giving a uni-
and agitation is started. Styrene and the initiator form distribution. At the end of the impregnation
pair, AIBN or benzoyl peroxide plus a finishing stage the suspension is cooled to 20 – 30 ◦ C, de-
initiator such as di-tert-butyl peroxide or tert- pending on the blowing agent type, to freeze in
butyl peroxybenzoate are pumped into the reac- the ingredients and prevent bead expansion dur-
tor at a constant rate, while the stabilizer, either ing handling. Some typical properties of an EPS
an inorganic, insoluble, finely divided powder or grade are:
a polymeric steric stabilizer is added. The reac-
tor is closed and the heating cycle starts. Droplet Monomer conversion > 99.9 %
size development occurs during the heating cy- M̄ W × 10−3 200 – 300
Polydispersity 2.2 – 2.4
cle. When the polymerization temperature is Mean particle diameter d̄ 0.4 – 1.5 mm
reached, between 75 ◦ C and 95 ◦ C depending on PSD breadth σ/d̄ 0.20 – 0.25
the initiator type, the polymerization proceeds Blowing agent concentration 5.0 – 8.0 %
until a conversion in the range 32 – 35 % is ob-
Polymerization Processes 71

Table 3 lists required properties of the bead observed that no further breakup occurs beyond
for different applications. 50 % monomer conversion when the particle vis-
cosity is about 104 cP. Addition of stabilizer is
Table 3. EPS bead properties and applications
used to obtain the desired d̄ at the identity point.
d̄, mm σ/d̄ Density, kg/m3 Use At the identity point both breakup and coales-
cence rates are zero. At the end of stage II, the
1.5 0.20 8 – 16 construction
(insulation) PSD has been established except for some minor
1.0 0.15 12 – 20 packaging shrinkage in the third stage due to density dif-
0.4 0.10 50 – 70 coffee cups ferences between monomer and polymer. Fig-
ure 47 shows a typical growth path for EPS
beads with d̄ = 0.425 mm. Since d̄ continually
increases during stage II, the variance of the PSD
also increases significantly. Figure 48 shows the
broadening of the PSD with conversion observed
by Konno et al. [565].

Figure 47. Growth path for EPS beads (mean diameter d̄


versus monomer conversion) during suspension polymer-
ization of styrene [564]
Stage I: breakup and coalescence rates are about equal Figure 48. Transient drop size distributions for styrene sus-
Stage II: coalescence rate exeeds breakup rate pension polymerization [565]
Stage III: conversions exceed identity point (no further ◦ 3 % monomer conversion, 10 min
breakup and coalescence) 2 20 % monomer conversion, 120 min
△ 45 % monomer conversion, 240 min
Suspension polymerization processes give a
rather broad PSD. In some EPS applications,
such as for coffee cups, a very narrow PSD is A potentially practical method of narrowing
required and for this grade the suspension poly- the PSD is to speed up the polymerization rate,
merization process is commonly carried out in by, for example, employing bifunctional initia-
two stages to minimize off-spec resin. In the first tors [566] to reduce the time for coalescence in
stage, styrene is polymerized to give the nar- stage II. This approach is based on the interac-
rowest PSD possible and then the suspension is tion of polymerization variables with the fluid
removed from the reactor and the polymer par- dynamical variables to achieve a desired PSD.
ticles are classified by sieving to give the re- Suspension polymerization is the most im-
quired narrow PSD. The particles are then re- portant commercial process for the manufacture
suspended in the reactor to permit impregnation of poly(vinyl chloride) [→ Poly(Vinyl Chlo-
with a blowing agent in the second stage. Sus- ride), Chap. 4.1.]. Porous beads with a rough sur-
pension polymerization for EPS production is a face are advantageous for rapid incorporation of
three stage process. In the first stage, the parti- plasticizers. This effect can be achieved by suit-
cle viscosities are low and an equilibrium par- able choice of dispersant [515], [567–569] or
ticle size is obtained (rates of droplet breakup by the addition of a few percent n-butane [570],
and coalescence are equal). In the second stage, [571]. A review provides a comprehensive sur-
where the viscosity of the particles is sufficiently vey of polymer particle morphology develop-
high, the particles grow (coalescence rate ex- ment during suspension polymerization of vinyl
ceeds breakup rate). For EPS, Villalobos [564] chloride [572]. A serious attempt to predict the
size distribution of primary polymer particles
72 Polymerization Processes

and bead porosity has been made [573]. Ion- sizes up to 200 m3 . Their construction presents
exchange resins are almost always made by sus- numerous engineering problems [554], [584–
pension polymerization giving spherical beads 588]. The most effective stirring in large reac-
in the size range ca. 0.3 – 1.2 mm. Most prod- tors is achieved with impellers driven from be-
ucts use polystyrene cross-linked with divinyl- low. The patent literature describes a wide va-
benzene to which functional groups are subse- riety of approaches for continuous suspension
quently attached (e.g., – SO3 groups by treat- polymerization. Apparently to date there are no
ment with sulfuric acid, chlorosulfonic acid, or continuous suspension polymerizations carried
SO3 ). Weakly acidic cation exchangers are pre- out commercially. The main problems relate to
pared by copolymerization of divinylbenzene the fact that, on the one hand, it is important to
with methacrylic acid, acrylate esters of lower assure the most uniform shear gradients possible
alcohols, or acrylonitrile. In this case the func- (i.e., total back-mixing) to establish the desired
tional groups are incorporated during polymer- particle size distribution and to prevent coagula-
ization, or they result from subsequent hydroly- tion and clumping in dead spaces. On the other
sis. Macroporosity is achieved by addition of an hand, maximum conversion is also desirable,
inert liquid which is a solvent for the monomer and this is only possible in a continuous process
but a nonsolvent for the polymer and which can with a narrow residence time distribution. Diffu-
readily be removed after polymerization is com- sion of monomer through the continuous phase
plete [574]. may not be sufficiently rapid to ensure that con-
With water-soluble monomers, especially ac- version is the same in all particles, independent
rylamide and comonomers (e.g., acrylic acid, of their residence time in the reactor. A particle
dimethylaminoethyl acrylate) it is common to that has a long residence may have a very low
use “inverse microsuspension polymerization” monomer concentration with concomitant high
in which a concentrated aqueous solution of the level of branching, cross-linking, and gel.
monomer is emulsified in an oily continuous
phase, usually a hydrocarbon, and then polymer-
ized with either an oil- or water-soluble initiator. 3.3.3.4. Emulsion Polymerization
When an oil-soluble initiator is used with an aro-
matic continuous phase, the kinetics have been Emulsion polymerization is probably the most
shown to resemble those of emulsion polymer- versatile of the polymerization techniques, ap-
ization [575–577]. The resulting polymer parti- plicable with many monomer types in batch,
cles are much smaller than the original monomer semi-batch (semicontinuous), and continuous
droplets, and the number of radicals per polymer processes. The product is a finely divided (par-
particle is small (< 1). However, when paraffinic ticle diameters ca. 0.05 – 0.3 µm) aqueous poly-
oil continuous phases are used, as is most com- mer dispersion (latex) containing up to 60 wt %
mon commercially, the locus of polymerization solids, which can either be used in latex form
is in the monomer droplets. This has been ver- or first coagulated and dried. Particle size dis-
ified by dynamic light scattering measurements tribution may play an important role in the fi-
which failed to detect inverse micelles and indi- nal application, as with emulsion PVC paste
cated a constant particle morphology with in- products. However, with some polymers (e.g.,
creasing conversion [578–580]. The polymer- styrene – butadiene rubber), the latex particles
ization therefore physically and kinetically re- are coagulated during work up to rubber bales
sembles a suspension polymerization and the and so their size is irrelevant, except perhaps in
prefix “micro” is added because average poly- the coagulation stage of the process.
mer particle diameter is nominally 1 µm, well Suspension and emulsion polymerization
below the usual size range for suspension poly- are both begun in an aqueous emulsion of a
merization. Suitable dispersants include Picker- monomer with low water solubility. In emul-
ing emulsifiers, polymers bearing hydrophilic sion polymerization, however, the locus of poly-
groups, or block or graft copolymers whose merization is in latex particles and not monomer
components differ in solubility [581–583]. droplets, the latex particles being much smaller
Stirred batch reactors are by far the most com- and having a much larger total interfacial area
mon for suspension polymerization, reaching [589]. Various theories exist for the mechanism
Polymerization Processes 73

of latex particle formation, differing primar- ous phase. Initiator decomposes in the water
ily with respect to the effect of the degree of phase to generate primary radicals, which prop-
monomer solubility in water. The polymeriza- agate with monomer dissolved in water to form
tion process in the latex particles is also quite dif- oligomeric radicals. When an oligomeric radical
ferent from that in monomer – polymer droplets enters a micelle it propagates rapidly with sol-
in suspension polymerization, because the pri- ubilized monomer to form a polymer particle.
mary radicals generally form in the aqueous con- In a typical emulsion polymerization there are
tinuous phase and migrate from there into the about 1013 monomer droplets per liter of emul-
latex particles. The most obvious difference in sion, with an average droplet size of about 3 µm.
the kinetics compared to solution or suspension This compares with ca. 1018 micelles, each con-
polymerization is the fact that polymerization sisting of ca. 100 emulsifier molecules with a
rate and molecular mass of the polymer pro- diameter of about 5 – 10 nm. The total interfa-
duced may be increased simultaneously in emul- cial area of the micelles is about three orders-
sion polymerization. Various explanations have of-magnitude larger than that of the monomer
been offered for the observed kinetic behavior. droplets. Consequently, oligomeric radicals in
The differences relate primarily to the question the aqueous phase are much more likely to dif-
of whether or not radicals, once they have en- fuse into a micelle swollen with monomer than
tered a latex particle, are then capable of leav- into a monomer droplet. Polymerization thus oc-
ing again and entering other particles. The de- curs almost exclusively in the micelles and poly-
cisive factors are the prevalence of chain trans- mer particles which are later formed, consum-
fer to small molecules (e.g., monomer) and the ing monomer that arrives by diffusion through
water solubility of the monomer. Finally, the the aqueous phase from the monomer droplets.
Trommsdorff – Norrish effect also plays an im- Micelles are thus gradually transformed into
portant role but with some modified features. polymer (latex) particles with a diameter of ca.
It should also be noted that polymer particles 0.1 µm and a concentration of ca. 1017 particles
in emulsion polymerization have a rather high per liter. As polymerization proceeds a form of
polymer concentration even at their birthtime. subdivision occurs, with monomer being trans-
Several reviews discuss kinetics and mecha- ferred from large monomer droplets with a con-
nisms [590–602] as well as mathematical mod- comitant increase in total interfacial area. As a
els [357], [603] applicable to emulsion polymer- consequence, micelles are consumed by being
ization. Proceedings from various symposia on “stung” with an oligomeric radical from the wa-
emulsion polymerization and related topics have ter phase and by being adsorbed on new inter-
also been published [604–611]. facial area which is continuously being formed.
When all the micelles are consumed and the con-
centration of emulsifier in the aqueous phase is
3.3.3.4.1. Theories of Emulsion just about to fall below the CMC, polymer parti-
Polymerization cle nucleation (via micellar nucleation) ceases.
The interval from the start of the generation of
Qualitative Theory. The qualitative theory oligomeric radicals in the aqueous phase to the
of batch emulsion polymerization is due pri- point where micelles have been consumed is
marily to the groups of Fikentscher [589], called Stage I in the emulsion polymerization
[612–614] and Harkins [615–618]. It is based process. At the end of Stage I, as illustrated in
on a system consisting of water, a monomer Figure 50, there is a rapid drop in the free emul-
with low water solubility, an emulsifier, and a sifier concentration (which throughout Stage I
water-soluble initiator that decomposes to pro- is equal to the CMC due to equilibrium with
duce radicals in the aqueous phase (see Fig. 49). micelles) and the surface tension (head-space
The emulsifier concentration is above the criti- gas/latex) rises rapidly from its previously sta-
cal micelle concentration (CMC) and thus mi- tionary value. Because the fractional coverage
celles form. The hydrophobic interior of the of the surface of polymer particles falls after the
micelles contains solubilized monomer, which end of Stage I problems with particle stability
is apportioned by diffusion out of the emul- and coagulation may occur.
sified monomer drops and through the aque-
74 Polymerization Processes

Stage II is known as the polymer particle


growth stage, during which the number of par-
ticles remains constant (in the absence of co-
agulation), as does the monomer concentra-
tion [M]p in the latex particles as a result
of monomer diffusion from the reservoir of
monomer droplets. Because of the extremely
high interfacial areas (polymer particle/water
and monomer droplet/water) and associated very
rapid mass transfer of monomer, there is an
equilibrium with respect to monomer transfer
from monomer droplets to polymer particles (the
chemical potential of monomer is the same in all
three phases, monomer droplet/water/polymer
particle). The interfacial energy per unit volume
for the small polymer particles contributes sig-
nificantly to the free energy and thus must be
accounted for as shown in Equation (3.34):
2V1 σ
= − ln (1 −ϕp ) +ϕp +χϕ2p

(3.34)
rRT
where V 1 is the molar volume of monomer, σ
is the interfacial tension (latex particle/aqueous
phase), r is the radius of the latex particle; ϕp
is the volume fraction of polymer in the la-
tex particle (ϕm = 1 − ϕp , with ϕm the volume
Figure 49. Schematic representation of an emulsion poly-
merization
fraction of monomer); χ is the Flory – Huggins
A) Particle nucleation stage; B) Particle growth stage; C) polymer – solvent interaction parameter [435,
Monomer finishing stage p. 522], [455].
a) Monomer droplet; b) Micelle; c) Emulsifier molecule; Swelling and monomer concentration [M]p
d) Latex particle; e) Water; f) Radical; g) Monomer
molecule
both increase with decreasing interfacial tension
σ, increasing particle radius r, and with a de-
crease in the parameter χ (which is equivalent
to the monomer being a better solvent for the
polymer; only when χ < 0.5 are monomer and
polymer completely miscible).
Since σ and r increase simultaneously dur-
ing Stage II, both ϕm and [M]p remain rela-
tively constant [619], [620] provided monomer
droplets are present. The constancy of N p (the
total number of polymer particles per liter) and
[M]p usually results in a constant polymeriza-
tion rate Rp during Stage II.
Stage III, known as the depletion or monomer
finishing stage, begins with the disappearance
of all monomer droplets. The only reservoir of
monomer for the polymerization in the latex par-
ticles is the aqueous phase. This is hardly suffi-
cient and in Stage III the monomer concentration
Figure 50. Overall rate of reaction Rp and surface tension
σ as a function of conversion during the three phases of
[M]p falls with time and conversion. The viscos-
emulsion polymerization (schematic) ity of the latex particles increases dramatically
due to increase in the number of physical chain
Polymerization Processes 75

entanglement points as the polymer concentra- K0 S Ri


tion increases. The self-diffusion coefficients of ≫ (3.36)
v Np
polymeric radicals fall and the Trommsdorff–
Norrish effect, which was active in both Stages I In this case the average number of radicals per
and II, increases in intensity with monomer con- particle which is given by
version [621], [622]. This can lead to a heat-kick P
nNn
when temperature control of the polymeriza- n̄ = P ≪ 1 (3.37)
Nn
tion is inadequate. Another phenomenon which
may occur is a glassy-state transition in the la- is much less than unity.
tex particles. This occurs when the polymeriza- Case 2) Rate of termination > rate of en-
tion temperature is lower than the glass transi- try ≫ rate of exit
tion temperature of the polymer being synthe- Ktp Ri K0 S
sized. The monomer acts as a plasticizer for > ≫ (3.38)
v Np v
the polymer, and propagation essentially ceases
when the polymer – monomer solution under- This case leads to n̄ = 1/2 on the basis of the
goes glass transition at the polymerization tem- following simple considerations [511]: when an
perature. oligomeric radical enters a polymer particle con-
taining no radicals polymerization is rapid and a
Overall Rate of Polymerization with a particle containing one radical is formed. When
Given Number of Particles. To calculate the a second radical enters this particle, termina-
rate of polymerization it is necessary to know the tion is instantaneous and two radicals are anni-
number of polymer particles containing n radi- hilated (two radicals cannot coexist for any sig-
cals (or the average number of radicals per latex nificant time). Instantaneous termination occurs
particle). Assuming a stationary state, the fol- for small latex particles when the termination
lowing particle population balance may be writ- constant K tp is sufficiently large. Since the exit
ten [623]: rate is negligibly small, a given latex particle
will alternately contain either one or no radical.
Thus averaged over time n̄ = 1/2. One can also
obtain this result with a simple kinetic analysis:
application of the stationary-state hypothesis for
radicals:
 
N1 N1
Ri = 2Ri and n̄ = = 1/2 (3.39)
Np Np

where N p = N 0 + N 1 .
Since water-phase termination affects Ri by
the same factor (on both sides of the equation)
its magnitude is irrelevant. Radical scavengers in
the water phase will for the same reasons have
The formation rate of particles with n radicals no effect on n̄. Radical scavengers in the poly-
equals the disappearance rate of particles with n mer particles would, however, lower N 1 and n̄
radicals. to values below 1/2.
where Ri is the rate of radical entry (radical Case 3) Rate of exit ≪ rate of entry ≥ rate of
cm−3 s−1 ); K 0 is the rate constant for radical exit termination
(cm/s); S is the particle surface area (cm2 ); K tp is
the rate constant for bimolecular termination in K0 S Ri Ktp
≪ ≥ (3.40)
the polymer particles (cm3 radical−1 s−1 ); and v Np v
v is the particle volume. Here the latex particles are effectively flooded
Smith and Ewart have derived expressions with radicals, so that n̄ ≫ 1, and it can be shown
for several limiting cases [623]: that the kinetics in the polymer particles corre-
Case 1) Rate of radical exit ≫ rate of radical spond to those in bulk polymerization. During
entry the buildup of radicals in the particles with time,
76 Polymerization Processes

radical entry rate will exceed the bimolecular ter- Gardon has described a series of investi-
mination rate; however, a stationary-state with gations in which the assumption dN n /dt = 0 in
respect to radical concentration will be reached Equation (3.35) was not made [620], [628]. Re-
as with bulk polymerization. alistic parameters when used in the nonstation-
A general solution to Equation (3.35) has ary form of Equation (3.35) led to the conclu-
been provided by Stockmayer [624] with mi- sion that negligible errors were introduced in
nor corrections by O’Toole [625]. After multi- Equation (3.35) by neglecting the transient ac-
plication by v/K tp and the substitutions cumulation term [597, pp. 566, 602 ff], [629].
vRi K0 S One possible exception is a system subject to
α= , m= , a = (8α)1/2 a very large Trommsdorff–Norrish effect and
Np Ktp Ktp
where K tp ≫ K p is no longer valid. In princi-
Equation (3.35) is transformed into: ple, Equations (3.41) and (3.42) may be used
Nn+2 (n + 2) (n + 1) +Nn+1 (n + 1) m +Nn−1 α = to estimate the magnitude of the Trommsdorff–
Nn [n (n − 1) +m n +α]
Norrish effect in emulsion polymerization.
Using an appropriate relationship for the ter-
Solutions to this equation follow: mination constant K tp and its dependence on
a I0 (a) polymer concentration or monomer conversion
n̄ = for m = 0 (3.41a)
4 I1 (a) in Stage III, one can calculate n̄ versus conver-
a Im (a) sion as shown in Figure 52 for styrene emul-
n̄ = for 0<m < 1 (3.41b) sion polymerization. Figure 53 shows calculated
4 Im−1 (a)
(m − 1) a Im−2 (a)
conversion – time curves for several monomer
n̄ = + for m ≥ 1 (3.41c) systems using appropriate expressions for K tp
2 4 Im−1 (a)
as a function of conversion. It is clear that the
Here the expressions Im (a) = i−m Jm (ia) are Trommsdorff–Norrish effect is of greater impor-
first-order Bessel functions, tabulated in [626], tance in emulsion polymerization when Case 2
[627]. An approximate solution to the Bessel kinetics are obeyed and when the polymer par-
functions is available in [597, pp. 556, 602 ff]. ticles are larger (vinyl acetate is clearly a Case
Figure 51 shows the dependence of n̄ on a and 1 system).
m. Knowledge of n̄ allows calculation of Rp us-
ing the relationship:
Rp =Kp [M]p n̄Np /NA (3.42)
where N A is Avogadro’s number.

Figure 52. Theoretical course of the average number of rad-


icals n̄ per latex particle with increasing conversion [630]
Emulsion polymerization of styrene, calculated on the basis
Figure 51. Average number of radials n̄ in latex particles as
of Equation (3.41)
a function of the parameters a and m Equation (3.41) [625]
Polymerization Processes 77

posed of a monomer-rich shell and a polymer-


rich core [631–633]. The monomer concentra-
tion in the shell, which is the site of polymeriza-
tion, should then remain constant even in Stage
III. These kinetic arguments have been criticized
by several other authors [630], [634], [635] and
the observed facts can also be largely accounted
for by considering the Trommsdorff – Norrish
effect in which a simultaneous decrease in [M]p
and K tp effectively compensate for each other.
However, experiments involving the polymer-
ization of styrene on a seed latex where either
the monomer or the seed latex is labeled [632]
clearly suggest that in Stage II the monomer does
indeed polymerize preferentially in the outer
shell. The contradiction with respect to Equation
(3.34), which has also been verified experimen-
tally [620], [636], [637], has not been adequately
explained. Neutron scattering experiments indi-
cate that the polystyrene latex particles are uni-
formly swollen by styrene.
If the exit rate of radicals is negligible (m = 0),
then a sudden reduction in the rate of initiation
in the water phase to Ri = 0 should result in a
Figure 53. Time – conversion curves for emulsion polymer-
constant Rp . In fact, however, experiments in-
ization volving intermittent γ-Co [638] or UV radiation
A) Methyl methacrylate (60 ◦ C); B) Styrene (60 ◦ C); C) [639] as initiator reveal that Rp decreases rapidly
Vinyl acetate (50 ◦ C) as soon as irradiation ceases. This suggests that
—- calculated with the Trommsdorff–Norrish effect
· · · · · calculated without the Trommsdorff–Norrish effect
radicals can be desorbed from one latex parti-
◦ experimental data cle and reabsorbed in another. There thus exists
an alternative termination process: interparticle
It seems remarkable that the empirical rela- termination that reduces the mean concentration
tionships for K tp as a function of conversion of radicals n̄. Equation (3.35) makes no provi-
found for bulk polymerization can be applied sion for possible reentry of a desorbed radical.
so effectively to emulsion polymerization, as According to Ugelstad [640–643], the rate of
shown in Figure 53. Part of the answer as to radical entry Ri is a function of both the rate of
why this agreement exists may be that when radical formation in the aquous phease by initia-
two radicals cannot coexist in a polymer parti- tor decomposition Ri, w and the rate of desorp-
cle (instantaneous termination), n̄ is independent tion, diminished by the rate of termination in the
of K tp . At higher conversions when n̄ is clearly aqueous phase:
> 0.5 for a Case 2 system, the size of polymeric

radicals in the latex particles should depend on
nNn − 2Ktw [R· ]2w
X
Ri =Ri, w +Kd (3.43)
the rates of termination and radical entry when n=1
chain transfer to monomer is negligible. How-
ever, if chain transfer to monomer is significant, In contrast to Eq. (3.35), this expression contains
the size of polymeric radicals depends only on the desorption constant K d = K 0 S/v [s−1 ], re-
polymerization temperature and not on the type moving the precondition that the rate of radical
of polymerization process. exit be proportional to the latex particle surface
To explain kinetic results obtained with area. Assuming that the radical absorption rate

styrene, for which Rp fails to decline on entering into polymer particles is proportional to [R ]W

Stage III, it has been suggested that one should and substituting for Ri = K a [R ]W and multipli-
consider uneven swelling of latex particles com-
78 Polymerization Processes
 
cation by KtpvNp and making the following soluble hydrocarbon diluents. It has been sug-
substitutions: gested that in some systems desorbed radicals
cross-terminate in the water phase and are not re-
Ri v Ri, w v absorbed. Nomura et al. [649] used the pseudo-
α= ; α′ = ;
Ktp Np Ktp Np kinetic rate constant method to develop expres-
Kd v 2Np Ktp KtW sions for n for binary copolymerizations that fol-
m= ; Y=
Ktp Ka2 v low the terminal model. Nomura et al. [650]
gives also observed that the extent of Stage I (particle
nucleation stage) could be lengthened by radical
α = α′ + m n̄ − Y α2 (3.44) desorption.
where Y is a dimensionless group. Mechanism of Particle Formation. As
Equations (3.41) and (3.44) permit calcula- emulsion polymerization always involves a cer-
tion of n̄ as a function of a′ , m, Y . The results tain amount of monomer in the aqueous phase,
for Y = 0 (i.e., negligible termination in the water and for this reason all the routes depicted in
phase) are shown in Figure 54. It is apparent that Figure 55 represent possible reactions of the
there is a rather large range of parameter values radicals or radical ions that form in the water
(α′ and m) for which n̄ < 0.5; this is discussed phase by initiator decomposition:
further in [640] and [597, p. 559 ff].
1) Entry into micelles, which are then trans-
formed into latex particles
2) Propagation within the aqueous phase until
a specific critical degree of polymerization
Pcr is reached, at which point the growing
macroradical precipitates to form a latex par-
ticle
3) Entry into an existing monomer droplet,
which is converted into a latex particle
4) Entry into an existing latex particle, which is
then subject once again to propagation (seed
polymerization)

Figure 54. Average number of radicals n̄ per latex particle,


assuming interparticle termination and a negligible amount
of termination in the aqueous phase (Y = 0), calculated by
using Equations (3.41 and 3.44) [640]

The number of radicals in a polymer par-


ticle clearly depends on rates of radical entry,
desorption, bimolecular termination, and chain
transfer. Although radical capture efficiencies by
polymer particles can be high [644], efficiencies
considerably below 100 % have been observed
[645]. Brooks [644] has presented general ex-
pressions for radical absorption rates which ac-
count for reabsorption and give capture efficien-
cies less than 100 %. Various experimental tech-
niques for the measurement of desorption rates
have been employed [645–647]. Lichti et al.
Figure 55. Possible radical reactions leading to particle for-
[648] obtained significantly enhanced desorp- mation during emulsion polymerization (schematic)
tion rates from polystyrene particles containing
Polymerization Processes 79

With all of these reactions, newly formed par- dNp


ticle surface area must somehow be stabilized, =Ri, w (3.45)
dt
either by emulsifier or through the charge orig-
inally present on the radical ion, which usually at time t = t 1 , where Ap = as · [S], it follows that
remains at the surface of the particle as the end  3/5
group of the resulting polymer. If the available 0.53 as [S]
t1 = (3.46)
degree of stabilization is insufficient, the inter- µ2/5 Ri, w
facial area will decrease by coagulation (or floc- Consequently, N p = Ri, w · t 1 and
culation) until a point is reached at which the
charge carriers present once again permit stabi- 
Ri, w
2/5
lization. Np = 0.53 (as [S])3/5 (3.47)
µ
Depending on the nature of the rate-
determining step, the rate of absorption of rad- This approximation leads to an upper limit for
icals into particles or micelles may be propor- N p , because even during Stage I some radicals
tional to the particle radius, surface area, or vol- enter existing polymer particles.
ume. Latex particles and micelles can also dif- Model 2) Latex particles and micelles absorb
fer in their capture efficiencies, and differences radicals at a rate proportional to their current in-
may even be observed between various latex par- terfacial areas Ap and Am . The rate of particle
ticles as a result of differences in size, surface formation is therefore
charge, and concentration of monomer and rad-  
dNp Ap
icals [643]. =Ri, w 1− (3.48)
dt as [S]
In principle, the various reaction paths in Fig-
ure 55 may compete with each other, in which and it follows that
case the system is extremely complex. It is of-
2/5
ten possible in practice to select reaction condi- Ri, w

Np = 0.37 (as [S])3/5 (3.49)
tions that cause one particular pathway to dom- µ
inate. Further elaboration of the theory is fre-
where µ is the volume growth rate of the latex
quently based on assumptions which led to one
particles.
mechanism being dominant in polymer parti-
cle formation. Smith and Ewart [623] fol- Table 4. Influence of various parameters on the number of particles
lowed the Fikentscher – Harkins theory in as- N p , the overall rate of reaction Rp , and degree of polymerization
PN [612]
suming that latex particles form only as a result
of the entry into an emulsifier micelle of a rad- Parameter Np Rp P∗
N

ical formed in the aqueous phase. Particle for- Surfactant ∼ [S]3/5 ∼ [S]3/5 ∼ [S]3/5
mation would then continue until the total sur- concentration
[S]
face area of the latex particles Ap corresponds
Initiator ∼ [I]2/5 ∼ [I]2/5 ∼ [I]−3/5
exactly to the interface-covering capacity of the concentration
emulsifier as · [S] minus the area of monomer [I]
droplets. A correction for emulsifier dissolved in Temperature E N = ER = 3/5 (E P − E d )
2/5 (E d − E P ) EP + EN
the water should also be made. In this expression
∗ Ignoring chain transfer
[S] is the emulsifier concentration and as is the
emulsifier covering capacity per unit of emulsi-
fier (per mole or gram). Based on this assump- Table 4 illustrates the interdependency of the
tion, two limiting cases can be described, in each number of particles N p , the rate of polymeriza-
of which it is further assumed that n̄ = constant tion Rp , and the degree of polymerization PN
and [M]p = constant, and that the polymer par- based on Equations (3.42) and (3.47) or (3.49)
ticles grow with a constant volumetric growth [612]. The relationships have been confirmed
rate, dv/ dt = µ = constant. experimentally over wide ranges of the rele-
Model 1) All of the radicals formed in the vant variables for such relatively water-insoluble
water phase by initiator decomposition enter mi- monomers as styrene, butadiene, isoprene, and
celles, so that the rate of formation of polymer chloroprene [651–658]. By contrast, other rela-
particles equals the rate of generation of radicals. tionships are found to apply in certain situations,
80 Polymerization Processes

especially in the case of more soluble monomers Smith – Ewart kinetics are applicable in the form
with large transfer-to-monomer rate constants, of Equation (3.47).
including vinyl acetate and vinyl chloride [640], On the other hand, in the case where
[642], [659–664]. Even in the case of styrene it ε N p /[S] ≫ 1, there is a somewhat differ-
is not possible to explain all the experimental ent dependency, with N p ∼ (Ri, w )2/7 and
results on the basis of the Smith – Ewart theory. N p ∼ [S]5/7 . To obtain agreement between cal-
If particle formation were to conform to Model culated and experimental N p , it is necessary to
1, then at the end of Stage I, each latex par- introduce the values ε = 1.3 × 105 for styrene
ticle would contain precisely one radical, i.e., [666] and ε = 1.2 × 107 for vinyl acetate [667].
n̄ = 1 [665]. In addition the molecular mass of the This is explained by assuming that radicals exit
polymer chain in a particle nucleated near t = 0 very rapidly from very small latex particles as
would be extremely large at the end of Stage I. well as micelles – so rapidly that they do not have
For Model 2, n̄ = 0.67 at the end of Stage I [628]. sufficient time to propagate to the extent required
Since n̄ = 0.5 during Stage II, Rp must peak at for insolubility in the water phase. Hansen and
the end of Stage I. However, no such maximum Ugelstad [597, p. 556, 602 ff], [668] have ap-
has been observed with styrene. Moreover, N p plied Danckwerts theory of diffusion with ho-
calculated by using Equations (3.47) or (3.49) mogeneous reaction [669] to this problem. The
often exceeds observed values by a factor of rate of absorption by latex particles that already
2 – 3. These discrepancies can be rationalized by contain a radical is significantly larger than that
assuming that latex particles are more efficient of latex particles and micelles containing no rad-
at capturing radicals than micelles [665], [666]. icals, because the rate of termination between
Some particle coagulation would also reduce the two radicals is much higher than the rate of prop-
discrepancy. If radicals formed in the aqueous agation. Termination by combination with a sec-
phase can enter either micelles or latex particles, ond radical normally leads to molecules that are
the condition for radical balance assures that no longer able to desorb due to having exceeded
d [R· ] the critical degree of polymerization Pcr .
=Ri, w −K1 [Mi ] [R· ] −K2 Np [R· ] (3.50) In the derivation of Equations (3.47) and
dt
where [Mi ] is the micelle concentration. Assum- (3.49) the volumetric growth rate µ of a latex
ing a stationary state is valid, one obtains particle, given by

Ri, w dv Kp ϕ m ̺m
[R· ] = (3.51) µ= = · ·n̄ (3.54)
K1 [Mi ] +K2 Np dt NA (1 +ϕm ) ̺p

Particle formation should only occur after the was considered constant since n̄ was taken to be
entry of a radical into a micelle, so a constant, independent of the latex particle age.
dNp This assumption is no longer valid when Case
=K1 [Mi ] [R· ] (3.52) 1 kinetics apply, and n̄ increases with particle
dt
growth. Reference [643] has therefore adopted
Substitution of Equation (3.51) into (3.52) gives
the following general relationship for the rate of
dNp K1 [Mi ] Ri, w Ri, w particle formation:
= =
dt K1 [Mi ] +K2 Np 1 + K2 Np
K [M ] 1 i dNp δNM rχ
M
Ri, w =Ri · (3.55)
= (3.53) dt δNM rχ χ
M +NM rp
εNp
1+ [S]
where N M is the number of micelles, r M is the
where [S]/[Mi ] = M m is the aggregation number micelle radius, and r p is the particle radius, χ
for emulsifier molecules in micelles (in many (= 1, 2, 3) is the exponent that characterizes the
cases, M m ≈ 100); K 2 N p /K 1 [M i ] = ε N p /[S] is dependence of the capture rate on the radii of the
the ratio of the rate of entry into latex particles micelles and polymer particles, and δ is an effi-
versus micelles, and ε = (K 2 /K 1 )M m character- ciency factor for the capture of radicals by mi-
izes the radical capture efficiency of latex parti- celles relative to polymer particles. This in turn
cles relative to micelles. For ε N p /[S] ≪ 1, Equa- is a function of the factor ε in Equation (3.56):
tion (3.53) is equivalent to Equation (3.45) and
Polymerization Processes 81

 χ to [S] or [I], while the order with respect to the


Mm rp
δ= (3.56) variable that is held constant decreases [643].
ε rM

Neglecting termination in the aqueous phase,


Equation (3.43) is transformed into

Ri =Ri, w +Kd n̄ Np (3.57)

and Equations (3.55) and (3.57) then lead to


dNp Ri,w + Kd n̄ Np
=h (3.58)
rp x
  i
dt N
1 + δ Np r M M

Assuming that only monomer transfer radicals


are capable of desorbing from latex particles,
Hansen and Ugelstad [643] constructed a
balance equation for the formation and disap-
pearance of latex particles containing a single
monomer transfer radical, and then proceeded
to solve it under the assumption of steady state. Figure 56. Number of particles N p as a function of A) the
One can then assume either that desorbed radi- initial emulsifier concentration [S]0 and B) the rate of initi-
cals do not add monomer in the aqueous phase, ation Ri, w for five different monomers [643] δ = 1.0; x = 1;
and instead simply wander from one particle Ri, w in A = 1016 L−1 s−1 ; [S]0 in B = 1.0 g/L
dotted lines: Smith – Ewart theory (Eqs. 3.47 and 3.49)
to another, or that such radicals do in fact add VAc: vinyl acetate; VC: vinyl chloride; MMA: methyl meth-
at least one monomer molecule in the aqueous acrylate; BMA: n-butyl methacrylate; S: styrene
phase. In the latter case, no monomer transfer
radicals absorb into the latex particles, and the In the case of monomers that are significantly
chain-transfer reaction becomes the sole source more soluble in water than styrene, it is also nec-
of monomer radicals in the particles. For further essary to take into account the fact that latex par-
details refer to the original article [643]. What ticles can be formed by homogeneous nucleation
finally results from these calculations is a set in the aqueous phase [655]. However, Equations
of curves representing the number of particles (3.47) and (3.49) based on the Smith – Ewart the-
N p as a function of the emulsifier concentration ory (which, incidentally, do not explicitly refer to
[S] and the rate of initiation Ri, w as shown in particle formation from micelles) remain valid
Figure 56 for various monomer types, assum- even when homogeneous nucleation is signifi-
ing there is no chain transfer to monomer in the cant [670]. The rate of particle formation then
aqueous phase. Other graphs of interest corre- becomes equal to either Ri, w (for Model 1 and
sponding to various values of δ and χ may be Eq. 3.47) or Ri, w minus the rate of radical cap-
found in [643]. These curves serve as a basis for ture by previously formed latex particles (Model
the rule that the sum of the orders with respect to 2 and Eq. 3.49). The volumetric growth rate µ is
[S] or [I] (or Ri, w ) for styrene, methyl methac- taken here to be constant, and particle nucle-
rylate, vinyl acetate, or vinyl chloride is always ation ceases when the total latex particle surface
in the vicinity of 1.0. Stated more generally area corresponds to as [S]. The concentration of
free emulsifier should at this point be somewhat
Np ∼ [S]z , Np ∼ [I]1−z (3.59) lower than the CMC.
Fitch and coworkers have developed the the-
where 0.6 < z < 1.0. If desorption and reabsorp- ory of homogeneous particle nucleation further
tion dominate (total radical entry rate ≫ Ri, w ) [606, pp. 73 – 102, 103 – 116], [671–675]. With
z = 1.0; if there is no desorption of radicals, [S] < CMC, a radical produced in the aqueous
z = 0.6 (usual value for styrene). The actual value phase by initiator decomposition continues to
of z may vary with [S] and [I]. If all other vari- grow in this phase until it is either absorbed by
ables are held constant, then an increase in [S] an existing latex particle or a specific critical de-
or [I] leads to an increase in the order z relative gree of polymerization Pcr has been exceeded,
82 Polymerization Processes

whereupon the macroradical precipitates and


forms a primary polymer particle. In this case
the rate of particle formation is described by
dNp
=Ri, w − vc (3.60)
dt
A steady-state (where Ri, w = vc ) is reached at
t = t s , which means that the number of polymer
particles nucleated is given by

Zts

Np = Ri, w − vc dt (3.61)
0

originally, the rate of capture vc was taken to be


[628], [673].

vc =Ri, w Np πr 2 L

(3.62)

where L is the length of the path traversed by


the growing radical from its point of origin to
Figure 57. Number of particles N p as a function of
the point where it precipitates as a primary par- conversion in the emulsion polymerization of methyl
ticle (collision theory). This proposal has been methacrylate with sodium dodecylsulfate as emulsifier
subjected to criticism [493, pp. 163 ff], [629] and [606, pp. 73 – 102]
the expression Concentrations in mol/L: [S2 O2− 8 ] = 7.35 × 10
−4 ,

[HSO− 3 ] = 1.14 × 10 −3 ; T = 30 ◦ C

vc = 4πDop CS Np r (3.63) In certain formulations (e.g., in the absence of


emulsifier, or where the emulsifier concentration
where Dop is the mean diffusion coefficient is very low) charge density on the polymer par-
for oligomeric radicals and latex particles in the ticle surface is insufficient to stabilize the result-
aqueous phase and C S is the concentration of ing particles. Moreover, with relatively soluble
oligomeric radicals in the same phase in fact pro- monomers such as methyl acrylate the interfa-
vides a better fit than Equation (3.62) to exper- cial tension is especially low, and emulsifiers are
imental data on particle nucleation in the pres- apparently less strongly adsorbed. The relation-
ence of seed particle [675]. The effectiveness ships here are further complicated by the fact
of seed particles in capturing oligomeric radi- that particles tend to coagulate until the surface
cals before they grow and precipitate to form area has been reduced (surface charge density
new particles corresponds not to N p r 2 but rather has increased) sufficiently to restore stability, a
to N p r. Hansen and Ugelstad [597, pp. 556, process for which experimental evidence also
602 ff ], [643] have proposed a model for radical exists [590, pp. 292 ff], [629], [678]. Figure 57
capture that takes into account not only the pos- shows experimental data for the production of
sibility of radical desorption, but also the electro- particles by homogeneous nucleation, in which
static repulsion between radicals and particles. methyl methacrylate was polymerized with var-
One practical application of these considera- ious rates of initiation Ri, w and sodium dodecyl
tions relates to stepwise seed polymerization as sulfate concentrations ranging from zero to the
a means of preparing uniformly large (monodis- CMC. With increasing conversion the number
perse) latex particles [676], [677], in which the of particles passes through a maximum before
key consideration is keeping vc as large as pos- falling to a constant value. The conversion at this
sible; i.e., preventing the formation of new poly- maximum as well as the final constant number of
mer particles. polymer particles is lower for lower Ri, w values
and lower emulsifier concentrations. Very sim-
Polymerization Processes 83

ilar results have been obtained with ethyl acry- Song and Poehlein [688–690] have devel-
late [679]. Electron micrographs of the resulting oped new approaches to modeling homogeneous
polyacrylate latex particles after etching with O2 nucleation for monomers having a range of water
clearly show that they were formed by coagula- solubility. They account for the variation of the
tion of a large number of smaller primary parti- critical chain length for precipitation with con-
cles. centration, and for sparingly soluble monomers
If limited coagulation (or flocculation) oc- suggest that oligomers generated in the water
curs, it is necessary to also take into account phase form micelles which are then stung by a
the rate of flocculation vf , transforming Equa- radical to generate polymer particles. They also
tion (3.60) into showed that the number of polymer particles N p
increases slowly with emulsifier concentration
dNp [S] below the CMC, where homogeneous nu-
=Ri, w − vc − vf (3.64)
dt cleation is dominant, but rapidly increases above
The Smoluchowski theory [680] can be used to the CMC due to micellar nucleation. Others have
calculate the flocculation rate vf [674]. concluded that micellar nucleation is dominant
If to an anionic or cationic emulsifier there is even for monomer with appreciable water solu-
added a small amount of a straight-chain alco- bility [691].
hol whose chain length corresponds at least to Radical scavengers (e.g., O2 and monomer
that of the emulsifier, then emulsification leads inhibitors such as tert-butyl catechol) can have
to an especially stable emulsion consisting of interesting effects on the number of polymer par-
many very small monomer droplets. Polymeri- ticles formed and n̄. Radical scavengers have
zation of such an emulsion with an oil- or water- been classified as water soluble and monomer
soluble initiator can, in an appropriate formula- (organic) soluble [692], [693]. Oxygen, which
tion, lead to almost total adsorption of the emul- dissolves in water, monomer droplets, micelles,
sifier on the surface of the monomer droplets. and polymer particles, mainly influences the ini-
Since the number of micelles is no longer large tiation of particle nucleation. Oxygen dissolved
compared to the number of monomer droplets, in the aqueous phase reacts rapidly with primary
polymerization occurs partially or even princi- radicals, preventing particle nucleation. This re-
pally within the monomer droplets. The conse- sults in an induction period in which no poly-
quence is a bimodal distribution of polymer par- merization occurs. The induction period ends
ticle sizes, where small particles formed in the when the oxygen content of the aqueous phase
aqueous phase are accompanied by large parti- has been reduced to an almost indetectable level.
cles whose origin is in the monomer droplets. In After the induction period, apart from usually a
the limiting case, the size distribution of the la- minor loss of initiator, the polymerization takes
tex particles corresponds almost exactly to that its normal course. However, in the presence of,
of the original monomer droplet size distribution say, tert-butyl catechol (an oil-soluble inhibitor)
[681], [682]. These relatively large latex parti- in the micelles and polymer particles, polymer
cles may contain considerably more than one particles are nucleated, but n̄ can be very low
radical and the observed polymerization kinetics due to radicals reacting with inhibitor. This re-
are similar to those for homogeneous bulk poly- duces the volumetric growth rate µ and hence
merization [683], [684]. This therefore repre- a larger number of polymer particles are nu-
sents a transition toward suspension bead poly- cleated (refer to Equations (3.47), (3.49), and
merization. (3.54). Once the organic-soluble inhibitor has
An emulsion polymerization in which poly- been consumed, the polymerization rate recov-
merization occurs in both latex particles (formed ers but to a rate higher than originally expected
from micellar or homogeneous nucleation) and in the absence of the inhibitor. This phenomenon
monomer droplets is called miniemulsion poly- can cause low-frequency oscillations in reactor
merization (a term coined by the Lehigh Uni- trains when inhibitor contents in monomer recy-
versity Group). For such systems bimodal PSD cle streams vary and potential safety problems
as well as polymerization rates have been both for all reactor types (heat generation rate may be
measured and predicted [685–687]. excessive).
84 Polymerization Processes

3.3.3.4.2. Physicochemical Parameters of masses depend on radical entry rate per parti-
Dispersions cle, which in turn is some unknown function of
particle size and number. For the special case
This section treats some properties of disper- of a monodisperse latex the radical entry rate
sions that can be unambiguously described in is known, however, and this case can be easily
terms of physiochemical quantities. In addition treated when termination is instantaneous (rate
to these, applications technology makes use of of radical entry is much smaller than the rate of
a number of other characteristics, primarily em- termination in the polymer particle; or in other
pirical quantities and tests, [694], [695]. words two radicals cannot coexist in a poly-
mer particle). This is true for Case 2 kinetics
Molecular Mass Distribution of Linear (n̄ = 1/2). An expression is now derived which
Homopolymer Chains. In emulsion polymer- describes molecular mass development for this
ization, chain transfer to monomer is more im- special case. Transfer to monomer and to chain-
portant than in homogeneous bulk polymeriza- transfer agent are included in the analysis.
tion because bimolecular termination rates for a Consider an oligomeric radical in a polymer
given polymerization rate are generally lower for particle. The probability that this radical adds r
emulsion polymerization. This is why it is pos- monomer molecules and then terminates either
sible to obtain high molecular mass at high poly- by chain transfer to monomer or chain-transfer
merization rates in these systems. When transfer agent T or by instantaneous termination with a
to monomer is dominant in controlling molec- new oligomeric radical which enters the particle
ular mass development, the process type is ir- from the aqueous phase is given by ϕr (1 − ϕ),
relevant and molecular masses of linear chains where
depend solely on polymerization temperature. Kp [M]p
Therefore, equations developed in Section 2.2 ϕ= R 
i, w
Kp [M]p + Kfm [M]p +KfT [T]p + NA
for bulk and solution polymerization in homoge- Np

neous media may be used to calculate molecular (3.65)


mass distributions. A useful experimental test
to determine whether chain-transfer reactions The instantaneous weight chain length distribu-
to small molecules (monomer or chain-transfer tion W (r, t) is therefore given by
agent) are dominant in producing polymer is to r (1 − ϕ) ϕr
change the concentration of initiator by several W (r, t) =
R0
fold. If molecular masses decrease with increas- r (1−ϕ) ϕr dr

ing initiator concentration then a significant frac- 2
= r (1−ϕ) exp (− (1−ϕ) r) (3.66)
tion of polymer chains are formed by bimolec-
ular termination. However, if polymer molecu- where K fm and K fT are rate constants for chain
lar mass is independent of initiator concentra- transfer to monomer and chain-transfer agent T.
tion, then chain-transfer reactions are dominant Equation (3.66) is the most probable distribu-
in producing polymer chains [696]. tion, with number- and weight-average degrees
When termination reactions are significant in of polymerization given by
molecular mass development, the calculation of
MWD and average molecular masses is gener- 1
PN = (3.67)
ally not straightforward. There are several com- (1 −ϕ)
plicating factors. Firstly, the MWD depends on
the polymer particle size distribution, which is PW /PN = 2 (3.68)
difficult to predict in commercial latex systems.
Even if the PSD were known one would still These expressions should be valid for systems
need to know the mechanism by which radicals such as styrene and methyl methacrylate when
enter particles. Attempts to predict the PSD have Case 2 kinetics apply and the PSD is quite nar-
been made by using population balance methods row. It is worth mentioning that in bulk polymer-
[697–700]. ization of styrene at low conversions the poly-
Experimental verification of these PSD pre- dispersity (PW /PN ) is 1.5 because bimolecular
dictions is lacking in most cases. The molecular termination by combination produces most of
Polymerization Processes 85

the polymer chains. A polydispersity of 2 in where K fp , K ∗p are chain transfer to polymer


emulsion polymerization is due to the fact that rate constant and rate constant for the addition
when termination is instantaneous, termination of polymeric radicals to terminal and pendant
by combination acts like termination by dis- double bonds, respectively; Q1 is the number of
proportionation. In fact the polydispersity (for moles per liter of monomer molecules chemi-
polymer produced instantaneously) obtained in cally bound to polymer chains (proportional to
emulsion polymerization of any monomer type grams per liter of polymer); Q0 is moles of poly-
will always be equal to or greater than that in mer per litre which have a terminal double bond;
homogeneous solution or bulk polymerization. F̄ 1 is the fraction of monomer molecules in the
Equations (3.66) – (3.68) apply to an essentially polymer chains which have an extractable atom;
monodisperse latex. For polydisperse PSD it is F̄ 2 is the fraction of monomer molecules in the
clear that instantaneous molecular mass proper- polymer chains which have a pendant double
ties will be broader as will the MWD of polymer bond. As these dimensionless groups increase
accumulated in the polymer particles over poly- in value, long-chain branching and cross-linking
merization times of several hours. For most com- (cross-linking density and gel/sol ratio) increase.
mercial recipes, chain-transfer agents are used to These groups increase with monomer conver-
control molecular mass development. In these sion and usually increase with polymerization
cases, the MWD is independent of the PSD, temperature. Long-chain branching and cross-
and molecular mass calculations for both poly- linking reactions are more important in emul-
mer produced instantaneously and accumulated sion polymerization than in homogeneous so-
polymer are straightforward. ϕ is then equal lution, bulk, and suspension polymerization be-
to K fT [T]p /K p [M]p , the instantaneous weight cause even at the birth of a latex polymer particle,
chain length distribution is again given by Equa- the concentration of polymer in the particle is
tion (3.66), and the MWD of the accumulated large [709]. Chain branching, cross-linking, and
polymer for a batch reactor is given by Equation gel formation are undesirable for the manufac-
(3.69) ture of cold SBR and NBR by continuous emul-
sion polymerization. To minimize long-chain
Rt
W (r, t) Rp dt branching, three important features of plant de-
0 sign are:
W̄ (r, t) = (3.69)
Rt
Rp dt
0 1) A large number of back-mixed reactors in se-
Several theoretical papers dealing with MWD ries (this gives a narrow residence-time dis-
development in emulsion homopolymerizations tribution which is desirable for low levels of
are available [701–705]. The problem has now branching)
been fully solved [706], [707] for the case when 2) The final conversion is kept rather low (ca.
two or more radicals can coexist in a polymer 63 % monomer conversion for cold SBR)
particle (zero – one – two, systems). When the 3) The polymerization temperature (ca. 5 ◦ C for
PSD is broad, the population balance equations cold SBR) is kept low to reduce K fp /K p and
for particle size and molecular mass are cou- K ∗p /K p
pled, requiring numerical solution of a large set When carrying out semi-batch emulsion
of partial differential equations [708]. polymerization, extending the addition time of
monomers lowers [M]p , and this can lead to
Polymer Chains with Long-Chain Branch- higher levels of long-chain branching and cross-
ing. Long-chain branching and cross-linking linking.
can occur as a result of chain transfer to polymer
and by addition of polymeric radicals to termi- Linear and Branched Copolymer Chains.
nal and pendant double bonds. The dimension- For such calculations, it is recommended that
less groups which are relevant in polymer chain the pseudo-kinetic rate constant method be em-
modification by these reactions are ployed. Emulsion copolymerizations can be suc-
F̄1 Kfp Q1 Kp∗ Q0 F̄2 Kp∗ Q1 cessfully modeled (calculations of polymeriza-
, and tion rate, chain microstructure, molecular mass,
Kp [M]p Kp [M]p Kp [M]p
86 Polymerization Processes

and long-chain branching and cross-linking) by Gianetti [699] in a theoretical study of the
application of the pseudo-kinetic rate constant evolution of PSD in Stage 2 has reviewed the
method [417], [649], [710] and properly ac- more important theoretical studies and discussed
counting for changes in chemical and physical their limitations. The results obtained represent
properties during the course of polymerization a generalization of previous theories in that
[711–713]. 1) A comprehensive description for monomer
Copolymer composition is usually control- systems which conform to zero – one kinet-
ling semi-batch operation [357], [405], [714], ics has been made
[715]. The composition of polymer particles de- 2) The contribution of latex particles having
pends on the thermodynamic compatibility of more than one growing radical has been
polymer chains and is not always spatially uni- properly accounted for
form. Core/shell morphology, which is often 3) Bimolecular termination has been explicitly
the planned particle morphology for specialty included in the model (at least where mutual
products, can be effected by grafting and cross- annihilation of two growing chains is not the
linking reactions as well as compositional drift main termination reaction)
[715–718]. Special end use properties can be ob- 4) General expressions for the time evolution
tained by controlling morphology [719], [720]. of the cumulants of the PSDs (both in the
The morphology of latex particles is also a func- transient and nontransient case) have been
tion of the mechanism of particle formation. determined
Limited coagulation (flocculation) leads to latex
particles composed of a number of not entirely Gianetti’s solutions not only revise theoret-
fused smaller primary particles. Such particles ical misinterpretations which appeared in the lit-
are often not smooth and spherical, displaying a erature, but also allow rate coefficients to be ex-
rather rough surface and an irregular lumpy or tracted with more precision from reliable experi-
“potato-like” shape [721]. mental data. In this connection it has been shown
that rate coefficients obtained from experimen-
Polymer Particle Size Distribution. In tal PSD data are in good agreement with val-
general, the PSD is narrower the lower the ues found by using dilatometry to measure rates
conversion at the end of Stage 1 is relative of polymerization. Recent interest in analyzing
to the final conversion at the end of the poly- the experimental time evolution of PSD is due
merization. In the limiting case of stepwise seed mainly to the possibility of determining kinetic
polymerization, no new polymer particles are rate constants with a small confidence interval
produced, even though a very high monomer [602], [723]. According to Gianetti’s solutions
conversion is reached, resulting in larger latex PSD and kinetic data are not independent of each
particles that are nearly monodisperse in PSD other at steady state (in other words, if a certain
[676], [677]. A semiquantitative calculation volume dependence for the entry rate constant is
[590, p. 275 ff] based on Smith – Ewart Case 2 assumed, from the analysis of an experimental
kinetics shows that PSD narrows with increasing PSD one must find the same volume dependence
monomer/water ratio, as well as with increasing for the exit rate constant). Apparently, PSDs for
temperature, increasing initiator concentration systems with n̄ ≤ 1/2 are insensitive to assump-
in the aqueous phase, and decreasing emul- tions about the mechanisms of radical entry and
sifier concentration [S]. Experiments involving exit. In this connection, a new approach based on
styrene have confirmed these predictions. Quan- the effect of PSD on MDW has been proposed
titative calculations are also in good agreement [724]. In fact, a similar approach was proposed
with experiment [606, pp. 153 – 161] for styrene much earlier by DeGraff and Poehlein [725].
emulsion polymerization. A more comprehen- A CSTR was used to obtain a broad PSD for
sive population balance model which accounts styrene emulsion polymerization, and assuming
for coagulation in the seeded emulsion poly- that radical capture was according to the colli-
merization of vinyl chloride has been developed sion theory (capture rate ∼ r 2 ), a polydispersity
[722]. The predicted bimodal distributions of of 4.84 was calculated for the MWD, DeGraff
particle size are in good agreement with experi- and Poehlein’s MWD measurements showed
mental data. that the molecular mass averages M̄ N and M̄ W
Polymerization Processes 87

were independent of mean residence time and a dispersion can be converted into an integral
that the polydispersity was ca. 3. The discrep- film on a surface by evaporation of the water.
ancy between theory and experiment may be due This normally corresponds approximately to the
to any number of possibilities, including: glass transition temperature T g of the polymer
and is one of the most important application pa-
1) Perhaps the collision model is not valid (try
rameters of a polymer dispersion. Copolymer-
the diffusion model)
ization of monomers whose homopolymers have
2) Larger latex particles may not obey Case 2
a range of glass transition temperatures makes
kinetics
it possible to achieve desired values for both
3) Transfer to monomer may not be insignifi-
the glass temperature and the minimum film-
cant
forming temperature. The glass transition tem-
Similar experimental data were found for perature [T g ] of the copolymer can be calculated
continuous emulsion polymerization of styrene by using the equation
in a CSTR [726]. Simultaneous growth of la-
W1 Tg1 +K W2 Tg2 ∆α1
tex particles by polymerization and coagulation Tg = with K = (3.70)
W1 +K W2 ∆α2
may occur at various stages in emulsion poly-
merization [688–690], [723], [727–730]. For the where W 1 and W 2 are mass fractions of
calculation of PSD in the emulsion polymeriza- monomers 1 and 2 in the copolymer, T g1 and T g2
tion of vinyl chloride, it may be necessary to are the glass transition temperatures of the cor-
account for coagulation [731], [732]. responding homopolymers, and ∆α1 and ∆α2
are the differences between the thermal expan-
The extent of coverage of the particle sur- sion coefficients above and below T g (melt and
faces by emulsifier is an important factor in the glass) for the two homopolymers.
stability of a dispersion. For example, this cover-
age should not be too large if the polymer is to be
later recovered by controlled coagulation. The 3.3.3.4.3. Inverse Emulsion Polymerization
surface coverage is evaluated in a detergent titra-
tion [733], [734], in which the amount of emul- In inverse emulsion polymerization, a hy-
sifier that must be added to give total monomo- drophilic monomer is dissolved in water, emulsi-
lecular coverage of the surface is measured. To- fied in a continuous hydrophobic oil phase with
tal coverage is indicated, for example, by the a water-in-oil emulsifier, and polymerized by us-
point where additional emulsifier no longer low- ing an oil- or water-soluble initiator. The product
ers the surface tension (surface tension remains is a latex consisting of very small polymer par-
constant after micelles are formed). If the sur- ticles, swollen with water and suspended in an
face coverage required for particle stability is oil continuous phase. The emulsified monomer
known, emulsifier can be added during Stage II droplets are also very small, and they may serve
to maintain stability but not form new micelles as the source of some of the latex particles
with the danger of nucleating a second family of (a form of miniemulsion polymerization). The
particles. More is involved here than simply the overall rate of polymerization is significantly
adsorption of an emulsifier on the particle sur- higher here than in the corresponding solution
faces, however. The majority of the hydrophilic polymerization. This observation strongly sup-
end groups on polymer molecules also tend to ports the idea that what is occurring is a true
migrate to the interface (e.g., – OH, – SO4 , – inverse emulsion polymerization rather than an
COOH), thereby contributing to stabilization of inverse suspension polymerization. When oil-
the polymer particles. A comparable effect is soluble initiators are used with aromatic con-
achieved by the addition of small amounts of tinuous phases the kinetics have been shown
hydrophilic comonomers (e.g., acrylic or meth- to resemble those of emulsion polymerization
acrylic acid), which polymerize preferentially at [576], [577], [735] with the locus of nucleation
the particle surface [721]. of particles in inverse micelles. However, when
paraffinic oil continuous phases are used, as is
The minimum film-forming temperature most common commercially, the locus of par-
of a polymer is the lowest temperature at which ticle nucleation and polymerization is in the
88 Polymerization Processes

monomer droplets and the process is properly cesses. A monomer-feed process based on a
called inverse suspension polymerization (the very large starter mix resembles more batch pro-
term microsuspension is often used because the cesses, whereas an emulsion semi-batch process
monomer droplets are nominally 1 µm in diam- involving a small initial charge with most of the
eter) [575]. This has been verified by dynamic monomers being fed in during the semi-batch
light scattering measurements which failed to period is more similar to continuous processes.
detect inverse micelles and indicated a con- These two semi-batch methods are often men-
stant particle size throughout the polymeriza- tioned in the patent literature [745]. Their advan-
tion [578–580]. Reviews of polymerization in tages include better control of polymerization
inverse emulsions and microemulsions are avail- and heat-generation rates by appropriate manip-
able [736–738]. Inverse microsuspension poly- ulation of feed rates. Moreover, the concentra-
merization has been used to study the kinetics tion of unreacted monomer can be minimized
of polymerization of acrylamide with cationic (offering a safer process which is less prone to
comonomers (quaternary ammonium cationic runaway) and cold feed can give higher produc-
monomers) [739]. tivity (less demand for jacket or condenser cool-
Water-in-oil emulsions of high molecular ing, for example). In the case of copolymeriza-
mass polyacrylamide and acrylamide copoly- tions, it is possible to manipulate monomer feeds
mers (anionic and cationic) are particularly use- to control copolymer composition [746–751]. In
ful as flocculating agents in sewage treatment particular, such methods are often exploited for
and as retention aids in papermaking. At the the purpose of reaching a specific degree of poly-
present time inverse emulsion and microsuspen- merization, particle size, or particle size distri-
sion polymerization is the only technology for bution without changing the overall composition
the large-scale production of very high molec- of the copolymer in the latex particles. Adjust-
ular mass polymers in a liquid-like form that ment of the process parameters can also have
affords a convenient suspension viscosity, high an influence, for example, on the stability of an
solids content, and easy phase inversion during acrylate dispersion [752–755].
application [740–742]. Kinetic studies [756], [757] have shown that,
provided the feed rate is not too large, both
types of semi-batch polymerization experience
3.3.3.4.4. Semi-Batch Emulsion a quasi-steady state in which the polymeriza-
Polymerization tion rate and feed rate of monomer remain com-
parable. During the semi-batch feeding period,
In semi-batch polymerizations, water, the concentration of monomer [M]p is rela-
monomers, initiators, and emulsifiers may be tively constant and it stays below the satura-
added to a reactor over a time period which tion concentration predicted by Equation (3.34).
is comparable to the total polymerization time The resulting conversion – time curves are lin-
(usually over several hours). Several feed poli- ear over a wide range of monomer feed times.
cies have been employed. For example, a starting Increasing the monomer feed rate causes an in-
mixture of water, emulsifier, and initiator is first crease in [M]p and according to Equation (3.42)
prepared, usually also containing a portion of this also leads to increases in both Rp and PN .
the monomer; the remaining monomer (often to- Slower feed rates prolong the steady state pe-
gether with initiator) is added during the course riod, and this should result in a narrower MWD
of polymerization (monomer feed) [743]. With for the polymeric product [757], [758]. Prolong-
another procedure, the polymerization is begun ing the monomer feed period lowers [M]p and
in a portion of the total batch, the rest being raises polymer concentration. For certain types
fed into the reactor gradually after the onset of of monomer this can lead to greater long-chain
polymerization in the form of a monomer emul- branching and cross-linking [355], [357].
sion (emulsion feed), although the composition When every ingredient except a reserve por-
of the feed may well differ somewhat from the tion of the monomer is initially charged to the
material used to initiate the process [744]. reactor, the phase ratio of monomer/water differs
Semi-batch processes are intermediate bet- from the corresponding batch process, but this
ween batch and continuous polymerization pro- has no effect on the number of particles nucle-
Polymerization Processes 89

ated. On the other hand, when polymerization is to slow monomer feed, for example) then the
initiated in a portion of the complete recipe, the rate of polymerization will equal the monomer
number of polymer particles nucleated per unit feed rate. The rate of polymerization and heat-
volume of emulsion may not be the same as for generation rate will increase with increasing
a batch process. Clearly the number of particles monomer feed rate. The polymerization temper-
nucleated in the initial charge will be much less. ature can therefore be controlled by manipulat-
If the onset of addition is delayed until some ing the monomer feed rate.
time after polymerization has begun, then Stage For the control of copolymer composition
I (the nucleation stage) for the initial charge may distribution, various monomer feed policies may
have been completed. If no further particles are be used and equations have been developed to
nucleated during the semi-batch feed period, the determine the appropriate time-dependent feed
feed will be entirely directed towards the growth rates required [356], [357], [405]. Consideration
and stabilization of the existing polymer parti- of molecular mass, long-chain branching, and
cles. The number of polymer particles is then cross-linking is also possible [357].
fewer and their size at the end of the polymer-
ization is larger. Based on the same total recipe,
the two semi-batch processes just described can 3.3.3.4.5. Continuous Emulsion
be compared and contrasted as follows: Polymerization
When the initial charge and the semi-batch
feed have the same recipe (composition), fewer Continuous emulsion polymerization is used
(hence larger) latex particles are produced than particularly where the latex is to be coagu-
for the batch process or the semi-batch process lated and the solid polymer recovered. Im-
with just monomer fed to the reactor over time. portant examples include the manufacture of
If particle formation is avoided during the semi- styrene – butadiene rubber (cold SBR) [760–
batch period, then the initial rate of polymeriza- 762], acrylonitrile – butadiene rubber (NBR)
tion Rp, o and the number of particles nucleated [763], polychloroprene [764], [765], SAN and
N p will be proportional to the size of the initial ABS polymers [766–769], and poly(vinyl chlo-
charge. ride) [733], [770]. For products used in latex
Assuming particle formation does not oc- form, most manufacturers prefer the flexibility
cur during the semi-batch feed period, emulsion associated with batch and semi-batch processes,
feed provides a narrower PSD than monomer especially with product lines that are subject to
feed or batch polymerization. If particle for- frequent change. Apparently, however, there is
mation does occur during emulsion feed, then also a certain amount of large-scale polymer dis-
the PSD becomes broader, and in extreme cases persion manufacture by the continuous process.
it can be even broader than with the other Systematic investigations have been carried
processes. Particle formation during semi-batch out on the kinetics and mechanisms of par-
feed can be suppressed by using a larger amount ticle formation in continuous stirred-tank or
of initial charge and prolonging the time period stirred-cascade reactors for styrene [771–781],
before semi-batch feed is initiated (i.e., more styrene – butadiene [782], [418], [419], ethylene
polymer particles are nucleated and grow larger [783], vinyl chloride [733], [784], [785], vinyl
before semi-batch feed is initiated). Taking the acetate [786–791], methyl methacrylate [786],
increasing surface area into account also makes and methyl acrylate [776]. A tubular reactor has
it possible to adjust the emulsifier feed so that also been analyzed [780] and review articles are
new polymer particles are not nucleated and no available [792], [793] on the general subject of
coagulation occurs [759]. continuous emulsion polymerization.
As long as the rate of addition of monomer is The most important elements of the theory
sufficiently large to keep the polymer particles of emulsion polymerization in a stirred reac-
saturated, then the rate and degree of polymer- tor or stirred reactor cascade have been pre-
ization for both semi-batch processes are inde- sented by Gerhsberg and Longfield [771].
pendent of feed rate. They began with the Smith – Ewart assumption
If the monomer concentration falls below its that latex particles originate with the entry of
saturation value in the polymer particles (due an oligomeric radical into a micelle, and that
90 Polymerization Processes

radicals enter both latex particles and micelles be towards older and larger polymer particles)
according to their surface areas (collision the- and n̄ = 0.5 and combination with Equations
ory, with no distinction between micelles and (3.72) and (3.74) then gives
polymer particles). The following equation then
Ri,w τ
describes the steady state with particle formation Np =    (3.76)
Kl Ri,w τ Kp τ [M]p
rate equal particle outflow rate (Case 2 kinetics 1+ a [s] l−K [M]
s 2 p
apply)
As can be seen the comparison of Equations
dNp Ri, w Am Np
= − = 0 (3.71) (3.74) and (3.76), the second term in the de-
dt Am +Ap τ
nominator of Equation (3.76) is analogous to the
where Am and Ap are the surface areas per unit quantity Ap /Am . At low concentrations of emul-
volume of latex for micelles and polymer parti- sifier [S] and high values of Ri, w and τ , it can
cles and τ is the mean residence time of a single be assumed that Ap /Am ≫ 1, so Equation (3.76)
CSTR. The residence-time distribution for latex takes on the simplified form
particles in a CSTR is given by !2/3
as [S] 1 −K2 [M]p
dNp 1 Np = (3.77)
E (t) dt = = exp (− t/τ ) dt (3.72) K1 (Kp τ )2/3 [M]p
Np τ

where E (t) dt is the fraction of latex particles in Substitution into Equation (3.42) then gives
the exit stream with age t. Because latex particles 1/3
!2/3
as [S] Kp 1 −K2 [M]p
are very small and have a density close to that of Rp =K3 [M]p (3.78)
τ 2/3 [M]p
water, the continuous phase, it is reasonable to
assume that the residence-time distribution of la- Here the constants K 1 –K 3 are functions of the
tex particles and the aqueous phase are the same specific volumes V m and V p of monomer and
(latex particles follow streamlines). polymer as well as of the molecular masses of
It is further assumed that the role of molecu- the monomers; K 1 and K 3 contain in addition the
larly dissolved emulsifier in the aqueous phase assumed constant n̄ = 0.5 within the latex parti-
can be neglected relative to the amount of emul- cles [771].
sifier in micelles and adsorbed on latex particles. In order to compare exprimental data ob-
Additional assumptions are that [S] ≫ CMC and tained over a wide range of conditions, it is con-
that venient to rewrite Equation (3.76) in dimension-
less form [772]:
as [S] NA =Am +Ap (3.73)
!2/3
Ri, w τ Ri, w τ τ Kp [M]p
 
The number of particles is then given by = 1 +K1 (3.79)
Np as [S] 1 −K2 [M]p
Ri, w τ
Np = (3.74) or
(1 +Ap /Am )

The particle surface area Ap for spherical lat- π1 = 1 +π2


exparticles with various ages and sizes thus be-
comes If N p is expressed in terms of Equation (3.42)
and introduced into Equation (3.79), what results
ZN is a second dimensionless equation for Rp :
Ap = 4.85 v 2/3 dNp (3.75)
!2/3
0 Ri,w τ Kp [M]p n̄ Ri,w τ τ Kp [M]p
= 1+K1
Substitution of v = f (K p , t, [M]p , n̄) into Equa- Rp N A as [S] 1−K2 [M]p
tion (3.75), where [M]p is constant (due to the (3.80)
very rapid diffusion of monomer from monomer or
droplets to polymer particles and, if monomer π3 = 1 +π2
droplets are absent, due to rapid diffusion of
monomer from polymer particle to polymer par- Comparisons of data supplied by three differ-
ticle; the direction of monomer diffusion should ent authors [771], [772], [776] are presented in
Polymerization Processes 91

Figures 58 A and 58 B. In Figure 58 A, essen- though it had earlier been predicted on theoret-
tially all of the points lie above the 45◦ line; ical grounds for autocatalytic processes in gen-
that is, fewer latex particles are formed than eral [795–806]. Autocatalytic acceleration here
predicted by theory. By contrast, the experi- is a consequence of the Trommsdorff –Norrish
mental values for Rp in Figure 58 B, which are effect. The mass balance for monomer at steady
more precise, correspond well to the theoretical state is given by
straight line. Overall, one may conclude that in
the case of styrene the theory is well substanti- v̇ [M]0 − v̇ [M] −Rp V = 0 (3.81)
ated. The Gershberg – Longfield theory has been where υ̇ is the volumetric flow rate to and from
extended and modified a number of times. Thus the reactor, [M]o is the monomer concentration
DeGraff and Poehlein dropped the assump- in the feed, [M] is the monomer concentration
tion that n̄ = 0.5 and instead calculated n̄ using in the reactor and exit stream, V is the volume of
the Stockmayer – O’Toole solution for different the reacting mass in the reactor, and Rp is the rate
particle sizes (Eq. 3.41) [772]. This accounts for of polymerization in the reactor. Furthermore,
the possibility of more than one radical per par-
ticle for the larger particles. Nomura et al. [773] 1" 
[M]0 − [M] =Rp (3.82)
also applied Equation (3.53), according to which τ
radicals are absorbed more rapidly into latex par- where τ = V /υ̇ is the mean residence time in the
ticles than into micelles, to continuous emulsion reactor.
polymerization and obtained a prediction that is Figure 59 shows a graphical solution of Equa-
in better agreement with experiment (as com- tion (3.82) and Rp versus monomer concen-
pared to that in Fig. 58 A). Brooks [794] dis- tration, in which data have been plotted as a
cusses various equations for the entry of rad- function of both monomer concentration [M]
icals into latex particles and micelles, assum- and conversion X. At low values of mean res-
ing among other things that micelles disintegrate idence time τ , there is one intersection (one so-
only slowly under subsaturation conditions, and lution) between experimental Rp = f ([M]) and
that even in the presence of micelles the surface the straight line (Eq. 3.82) which implies a single
of the latex particle may not be saturated with stable operating point at low conversion. By con-
emulsifier. trast, longer residence times lead to three points
of intersection, with stable operating points at
both low and high conversions and an unsta-
ble operating point at an intermediate conver-
sion. For a train or cascade of n stirred reactors
there may be as many as n + n (n + 1)/2 stable
operating points and n unstable operating points
for the individual reactors as well as n + 1 sta-
ble states for the entire cascade. Oscillations in
the number of particles, the polymerization rate,
and the interfacial tension have attracted consid-
erable theoretical attention. These are attributed
Figure 58. A) Number of particles according to Equation to bursts of particle formation that accompany
(3.79), and B) Overall rate of polymerization according to swings between states of supersaturation of the
Equation (3.80) for the continuous emulsion polymeriza- latex particle surfaces, indicating large amounts
tion of styrene (dimensionless representation) [772]
2 from [772] of free emulsifier (and micelles), and states char-
H from [771] acterized by undersaturation, in which particle
◦ from [776] formation almost ceases. This phenomenon has
• calculated from values for Rp with the assumption that long been recognized in industrial-scale con-
n̄ = 1/2
tinuous emulsion polymerization [733] and has
The existence of multiple steady states for now been thoroughly investigated in the case of
the isothermal operation of a CSTR was first styrene [771], [778], methyl methacrylate [786],
demonstrated in the case of continuous emulsion [807], [808], and vinyl acetate [786], [798],
polymerization by Gerrens et al. [776–778], al- [799], [809].
92 Polymerization Processes

mer particle seed generator in series with one or


more larger finishing reactors has been used to
eliminate oscillations for Case 1 systems (vinyl
acetate and vinyl chloride) [785], [646], [814],
[815].
To complete the discussion of emulsion poly-
merization, a few brief comments will be made
about an area which has almost been completely
neglected in the literature. This is the effect of
cross-linking in the polymer particle on emul-
sion polymerization kinetics and modeling (ef-
fect on particle nucleation and growth). The first
and very important effect of cross-linking is on
the equilibrium swelling of polymer particles by
monomer. Taking into account the elastic free
energy change due to the cross-linking network
Figure 59. Emulsion polymerization of styrene. Stationary
structure as well as the free energy contributions
operating points for a continuous stirred reactor at various of mixing and interfacial tension, it is possible to
residence times τ [778] derive an equation which gives the equilibrium
• Stable, ◦ unstable operating points; Rp in milligrams monomer concentration in polymer particles
styrene per gram latex per minute
2V1 σ
= − ln (1 −ϕp ) +ϕp +χϕ2p
 
rRT
 

1/3 ϕ p  ̺p
− ϕp − ̺¯el (3.83)
2 ̺M
where ̺el is the elastic cross-link density of the
accumulated polymer [816]. The first thing to
notice is that cross-linking reduces the monomer
concentration and hence the volumetric growth
rate of polymer particles. According to Equation
(3.47) or (3.49) the number of polymer parti-
cles nucleated in Stage I will be larger. In an
experimental study using styrene and divinyl
monomers, Nomura et al. have confirmed [817]
Figure 60. Oscillations in the continuous emulsion poly- that much higher numbers of particles are gener-
merization of vinyl acetate at three different initiator con-
centrations [I]. Comparison between experimental (◦ · ◦ · ◦) ated when initiated particles are cross-linked. An
[786] and calculated (– – –) values [787] obvious effect of cross-linking is the strength-
ening of the Trommsdorff – Norrish effect with
Model calculations are available for styrene concomitant earlier and larger decrease with
[800], vinyl acetate [787], [790], [791], and conversion of the termination constant and the
vinyl chloride [785]. Figure 60 illustrates the consequences thereof. When the molecular mass
close agreement between experiment and the- of primary chains produced in smaller polymer
ory. Radical desorption has a strong influence particles is larger, one might expect gelation
(with accompanying more rapid polymer parti- to occur earlier in smaller particles. There are
cle nucleation rates) on the oscillatory behavior clearly many interesting phenomena related to
of continuous emulsion polymerization in a cross-linking in latex particles which await in-
CSTR [810–812] and radical desorption rate vestigation.
constants have been measured using a CSTR
[810], [813]. In the absence of particle nucle-
ation, oscillations (at least those not due to 3.3.4. Miscellaneous Processes
thermal effects) do not occur. A CSTR (with
split flow) or tubular reactor operating as a poly- The modeling of high-pressure tubular and au-
toclave reactors for the manufacture of LDPE
Polymerization Processes 93

has received a great deal of attention [818–836]. It is particularly effective for chemical modi-
These models are notable for the detailed kinet- fication of high molecular mass, solvent-free,
ics, which include transfer to polymer, backbit- commodity polymers to tailor-made products
ing, and β-scission in an attempt to calculate to meet the growing diversity in polymer ap-
both short- and long-chain branching frequen- plications. The extruder reactor provides short
cies. The polymerizations are almost adiabatic, and controlled residence times; efficient mix-
with large temperature increases, and phase sep- ing, particularly for reactants at low concentra-
aration between ethylene and polyethylene may tions; and good heat transfer. Multistage reac-
occur in autoclave reactors. The tubular reactor tion, with sequential feeding of reactants and
is considered to be in plug flow, while mixing removal of byproducts, is possible. There are
in the autoclave is rather complex and various certain advantages of twin-screw over single-
mixing models have been considered. A com- screw extruders [839]. These include: intense
prehensive copolymerization model for ethylene shear mixing (with better surface renewal rates)
and comonomers in a high-pressure continuous giving molecular-level mixing (micromixing),
tubular reactor has been developed [836]. improved heat transfer, multistage feeding, vent-
A comprehensive tubular reactor model for ing of volatile byproducts, and screw speed can
the solution polymerization of vinyl acetate has be changed without changing the throughput.
been developed by Hamer and Ray [837]. The These attributes of the twin-screw extruders are
kinetics employed account for transfer to poly- ideal for chemical modification where a small
mer and reaction with terminal double bonds amount of the low molecular mass compound to
(trifunctional branching frequencies are calcu- be grafted on the polymer chains must be ho-
lated). The model solves the equations of mo- mogeneously mixed with a highly viscous base
tion for both axial and radial velocity profiles. polymer.
Experimental validation of the model has been Chemical modifications are often carried out
carried out. using free-radical reactions in the temperature
The bulk and solution polymerization of range 170 ◦ C – 350 ◦ C. When a polymer con-
styrene in tubular reactors has also been exten- taining a peroxide and a compound to be grafted
sively studied both theoretically and experimen- is melt processed, chain scission, long-chain
tally [375], [838]. Spatial distributions of veloc- branching, cross-linking, and grafting reactions
ity, temperature, and concentration were mod- may occur simultaneously. To optimize product
eled [838] and it was shown that for tube diame- properties, the rates of these reactions must be
ters > 2 cm, thermal runaway may occur. Inter- controlled.
nal mixers were used in larger diameter tubes Chain Scission. Polymer chain scission oc-
to give better radial mixing and this technique curs when chemical bonds (usually C – C) along
appears to give polystyrenes with high molecu- the backbone break. Since a higher molecu-
lar mass at a productivity of commercial interest lar mass chain has a greater number of such
[375]. References to other important studies on bonds, it experiences chain scission preferen-
styrene production in tubular reactors may be tially (for polypropylene, scission seems to be
found elsewhere [375], [838]. random with respect to the position of the bond
along the chain backbone [840], [841]). There-
Chemical Modification of Polymers in Ex- fore, the broad MWDs, usually found for poly-
truders The commercial incentive for the chem- olefins narrow as the average molecular mass
ical modification of polymers by chain scission, falls. With random scission occurring exclu-
long-chain branching, cross-linking, and graft- sively, the limiting polydispersity for all initial
ing is the enhancement of the physical and chem- MWDs is two and the limiting MWD is Flory’s
ical properties of polymers and polymer mix- most-probable distribution. A reduction in mo-
tures (alloys, blends, additives) and/or improve- lecular mass improves processability. The out-
ment of their processability. In many cases, the standing commercial example of the use of con-
extruder has been found to be an effective chem- trolled chain scission of polypropylene (PP) is in
ical reactor in which chemical modification of the production of controlled-rheology PP used
high molecular mass polymers in the melt can for fiber manufacture, film-extrusion, and for
be done economically for low-volume products. fast molding applications [842].
94 Polymerization Processes

Long-Chain Branching and Cross-Linking. sufficiently energetic to abstract backbone atoms


Long-chain branches (tetrafunctional) are and generate backbone radical centers.
formed when radical centers on the backbone of Other reaction types, including anionic,
polymer chains experience bimolecular termina- cationic, anionic coordination, and condensation
tion by combination. If this long-chain branch- can be carried out in an extruder reactor [839].
ing process continues, a three-dimensional net- Kowalski [848] describes an extruder process
work (cross-linked gel) is formed. The pro- for halogenation of butyl rubber. Saleem and
duction of higher molecular mass chains by Baker [849] describe a process in which poly-
long-chain branching produces a polymer that styrene and polyethylene along with their re-
exhibits increased die swell and melt strength, active counterparts (polystyrene with oxazoline
and improved strain-hardening properties [842]. groups, and polyethylene with carboxylic acid
Cross-linking of polymer chains results when groups) are reacted to form polymeric compata-
bimolecular coupling (via bimolecular termina- bilizers in situ.
tion by combination of polymer backbone rad- Kinetics and Extruder Reactor Modeling.
icals or the reaction of grafted chemical groups Reviews of the kinetics of chemical modifica-
such as those with silane functionality [843], tion of polymer by free-radical mechanisms are
[844]) of polymer chains to form tetrafunc- available [839], [850], [851]. Motha et al. [852]
tional branches occurs repeatedly. The cross- have developed a model for the grafting of unsat-
linked polymer is a mixture of polymeric gel, urated acids and silanes on polyethylene. This
of effectively infinite molecular mass, and sol model does not attempt to predict the gelation
which is made up of linear and branched chains. point nor the gel/sol ratio in the post-gel re-
Property enhancement due to cross-linking in- gion. A very general model which permits one
cludes: increased service temperature, solvent to follow the change in the MWD of polymer be-
resistance, flexural modulus, low-temperature ing modified by simultaneous random scission,
impact strength, environmental stress cracking long-chain branching, cross-linking, and graft-
resistance, and reduced creep [842]. ing has been developed [850], [851]. The model,
Grafting onto polymer chains can occur which is in the form of an integro-differential
when a mixture of the polymer, peroxide, and equation is based on a simpler version devel-
unsaturated or saturated compound to be grafted oped earlier by Saito [851], [852]. Saito found
are melt processed in an extruder. For exam- an analytical solution for the limiting case of
ple, acrylic acid or maleic anhydride can be random scission occurring exclusively. Kimura
grafted onto PP or PE to enhance compatabil- [853] obtained a limiting analytical solution for
ity with more polar blends or to improve adhe- the case in which random cross-linking is oc-
sion to metals, glass fibers, and to other polymer curring exclusively. The general case can be
types [845], [846]. Vinyl silanes can be grafted handled by using the method of moments to
to polyolefins to subsequently produce cross- give the molecular mass averages. The widely
links in the presence of moisture [843]. Func- used Charlesby – Pinner equation was obtained
tionalized additives (antioxidants, flame retar- by using an idealization which is questionable
dants, pigments) can be grafted onto the host [854]. Simultaneous scission and cross-linking
polymer to give the desired property enhance- are treated as two processes in series, with ran-
ment. These chemically bound additives are ef- dom scission occurring first and then cross-
fectively mixed on a molecular scale and do not linking. In addition, the initial MWD must be
diffuse out of the matrix during use [847]. Flory’s most probable distribution. Numeri-
Source of Radicals. During normal extrusion cal methods must be used to solve the integro-
of polyolefins, alkyl radicals may be formed differential equation for the case of simultaneous
thermomechanically, but at too low a rate to initi- scission and cross-linking. This is not a trivial
ate chemical modification to the extent normally problem and has not been successfully solved to
desired. To provide a suitable source of radicals, date.
organic peroxides are added to the polymer be- The design of extruders for reactive process-
fore melt processing. In these applications the ing is much more challenging than for purely
peroxides used must generate radicals which are physical processing [839]. Extruders must han-
dle polymer melts with continuously changing
Polymerization Processes 95

properties. Residence-time distribution (RTD) The gas-phase UNIPOL process has become
and micromixing can significantly affect the one of the most successful commercial processes
properties of the chemically modified polymer. for the production of polyolefins [858]. In this
One of the important advantages of twin-screw process, homo- and copolymerizations of ethyl-
over single-screw extruders is the greater control ene with α-olefin comonomers are carried out
they allow over RTD and mixing. The modeling in a fluidized-bed reactor using heterogeneous
of conventional or physical extrusion processes Ziegler – Natta or Philips supported metal oxide
has been well documented in the literature, par- catalysts [859].
ticularly for single-screw extruders. Details of Modeling of these processes can be classified
the modeling of extruders may be found in the re- into three categories [857], [860]:
view article by Tzoganakis [839]. Tzoganakis
et al. [855] developed a comprehensive rheoki- 1) Microscale kinetic phenomena: coordina-
netic model for the random scission of isotactic tion kinetic mechanism, multiple active site
polypropylene in a single-screw extruder. Finite- types, stereoregularity, copolymer composi-
element or finite-difference methods will play tion and molecular mass distributions, and
an important role in the solution of the three- catalyst deactivation kinetics
dimensional flow equations for both single and 2) Mesoscale particle phenomena: fracture of
twin-screw extruders [856]. catalyst particles, heat generation in parti-
cles, mass transfer, and growth and influence
of changing properties of reaction environ-
3.3.5. Ionic Polymerization Modeling ment
3) Macroscale reactor phenomena: overall
3.3.5.1. Introduction mass and energy balances, gas – liquid
mass transfer effects, influence of reaction
The most powerful processes for the control of medium, effects of particle motion, and heat
MWD of linear chains are living anionic and transfer limitations and influence of reactor
cationic polymerizations. These processes are type
interesting from the point of view of process en-
gineering because the MWD of the product can Similar modeling categories could be con-
be varied widely by changing the RTD of the sidered for the manufacture of PVC and EPS by
process or by oscillating the process in a con- free-radical suspension polymerization, where
trolled manner. These living ionic processes will particle morphology and foamed cell structure
be considered from this perspective in Section are important considerations.
3.3.6.1. One of the major challenges in modeling het-
erogeneous coordination polymerizations has
been to rationalize the very broad MWDs
3.3.5.2. Heterogeneous Coordination and copolymer composition distributions (CCD)
Polymerization that are often observed. There have been two
The commercially important anionic coordina- schools of thought. Ray and coworkers have
tion processes (heterogeneous ZN polymeriza- until recently attempted to explain the broad
tion) are now discussed. Seppala and Auer MWDs in terms of diffusional resistances for
have published an excellent review of much reactants, which increase as the catalyst parti-
of the relevant subject matter [857]. Third- cles are enveloped by a growing shell of poly-
and fourth-generation catalysts have already mer. An additional resistance may result from
been commercialized for slurry polymeriza- the crystallization of polymer chains. Others
tions. These newer catalysts possess high and have attempted to explain the broad MWDs and
stable activity, high stereospecificity indepen- CCDs as due to multiple sites of different ac-
dent of polymerization time, and increase in ac- tivity. These sites have different compositions.
tivity with increasing temperature up to about To model a binary copolymerization it can be
80 ◦ C. Electron donors may be used to regulate assumed that each site type produces a poly-
isotacticity. Catalysts may also be designed to mer with Stockmayer’s bivariate distribution
control the size and shape of polymer particles. [861], [862]. The more difficult problem is, of
96 Polymerization Processes

course, to determine the number of active site all polymers. It is thus important to know the
types and the concentration of each. Two active- distribution functions that correspond to vari-
site-type models have been used [862], [863] ous polymerization mechanisms and how these
as well as site-type distribution functions [864]. distributions change with reactor type and con-
Among the diffusion-limited models, the multi- figuration. Another way of expressing this is to
grain model is the most comprehensive. It is as- note that these molecular properties are strong
sumed that fragmentation of the original cata- functions of the residence-time distribution of
lyst particle is complete before appreciable poly- the polymerizing mass and that once an optimal
mer has been produced [865], [866]. Thus, the residence-time distribution has been established,
large macroparticle is composed of many cata- one can choose reactor types and manipulate re-
lyst fragments encapsulated by polymer [865], actor configuration to obtain this residence-time
[867–869]. Both macroparticle and microparti- distribution in a practical manner. Most devia-
cle mass transfer can be important for large cat- tions of predicted and measured molecular prop-
alyst particles of high activity [857]. Ray et al. erties are likely to be due to assumptions about
[868] found that the maximum possible temper- levels of micromixing and segregated flow and
ature rise in both micro and macroparticles is the comparison of predicted instantaneous mo-
negligible in slurry polymerization of propyl- lecular properties with measurements of these
ene. However, in gas-phase processes the tem- properties on accumulated polymer (e.g., instan-
perature inside the polymer particle could reach taneous properties can change with time in a
its melting point under some conditions [866]. batch reactor as, say, [M] decreases and hence
Near the beginning of polymerization, the poly- the distributions for molecular properties for
mer particles may heat up because of the high the accumulated polymer will have larger vari-
catalyst activity and polymerization rate. In sum- ances). From an industrial standpoint, however,
mary, it can be said that the very broad MWDs any process of interest involves high – if not
and CCDs can be explained by invoking mul- complete – conversion, and a wide variety of
tiple site types. However, under certain condi- batch and continuous reactors is utilized. The
tions, mass- and heat-transfer resistances may relationships between reactor types and config-
also play a significant role. The development urations and the various polymerization mecha-
of more realistic models for copolymerization nisms are often quite complex, and it will only
on heterogeneous coordination catalysts will re- be possible within the confines of this article to
quire the measurement of the bivariate distri- treat certain simple ideal reactors and reaction
bution of molecular mass and composition by types.
a combination of analytical techniques such as Figure 61 shows four ideal reactors, char-
TREF/GPC and TREF/NMR to observe the de- acterizing them on the basis of residence-time
tailed microstructure of the copolymer chains distribution and the temporal and spatial course
[870]. of chemical reaction. Batch reactors (BR), con-
tinuous plug flow reactors (CPFR), and homo-
geneous continuous stirred-tank reactors (HC-
3.3.6. Process Variables, Reactor Dynamics/ STR) are treated in the article Stirred –Tank and
Stability, On-Line Monitoring and Control Loop Reactors. With a homogeneous continu-
3.3.6.1. Influence of Reactor Type and ous stirred-tank reactor, perfect mixing down to
Configuration on Molecular Mass and the molecular level is assumed, and for a CSTR
Copolymer Composition Distributions, and with an ideal residence-time distribution, there
on Long-Chain Branching and are by definition no spatial variations in tempera-
Cross-Linking ture and concentrations. Mixing at the molecular
level is often called “micromixing”. This condi-
Survey of Idealized Reactor Types and tion is often not achieved in practice, however,
Simple Polymerization Reactions. Molecular and for this reason a fourth ideal reactor type
mass and copolymer composition distributions, is introduced, the segregated continuous stirred-
long-chain branching, and cross-linking can ex- tank reactor (SCSTR) [801–803]. Here the fluid
ert a significant influence on the properties of phase is regarded as subdivided into many small
Polymerization Processes 97

isolated compartments. Each compartment con-


tains a large number of molecules, which are
permanently confined within the limits of that
compartment; therefore, the individual compart-
ments function as miniature batch reactors with
different residence times in the flow reactor. The 3) Polymer coupling (e.g., polycondensation)
compartments themselves are taken to be ideally
mixed, leading to what is called “macromix-
ing” despite total segregation of molecules in
different compartments. Thus, the sum of all
the compartments in a segregated continuous In both cases of monomer coupling, many in-
stirred-tank reactor have the same residence- dividual monomer molecules add successively
time distibution as the contents of the homoge- to the growing chains. Monomer coupling with
neous stirred-tank reactor. A macroscopic mean bimolecular termination is characteristic of free-
taken over all the compartments in the efflu- radical polymerizations. Termination may occur
ent stream and in the reactor itself would show through the combination of two growing radi-
concentrations and temperature that are constant cals (degree of coupling K = 2). Termination by
both spatially and with respect to time. On the disproportionation is an example of a process
other hand, a probe capable of microscopic sam- in which K = 1 [806]. Monomer coupling with-
pling of individual compartments would reveal out termination is encountered, for example, in
concentrations that varied in a statistical man- living anionic or cationic polymerizations. Poly-
ner from one compartment to another. Given the mer coupling between macromolecules is char-
high viscosity of a typical polymerizing solution acteristic of polycondensation and polyaddition
it is quite likely that many solution polymeriza- reactions. The discussion that follows has been
tions actually occur in segregated systems [804]. restricted deliberately to the simplest reactions
Ideal reactors are discussed in greater detail in, possible, with precisely simultaneous initiation
for example, [795], [805]. of all active centers in the case of anionic poly-
Polymerization reactions can be divided ac- merization and exact stoichiometric balances in
cording to their kinetics into three classes: polycondensation. Chain-transfer reactions will
be neglected, as will exchange reactions (e.g.,
transesterification, transamidation). A few con-
1) Monomer coupling with bimolecular termi- sequences of these complex reactions are out-
nation (e.g., free-radical polymerization) lined in [871].
The three types of polymerization differ con-
siderably in the way the degree of polymer-
ization of the accumulated polymer P̄N (aver-
aged over the course of the conversion process)
changes as a function of conversion. As shown in
Figure 62, a high degree of polymerization P̄N is
reached at very small conversions in free-radical
polymerization. In a batch reactor this value sub-
sequently decreases with increasing conversion
as a consequence of decrease in the monomer
concentration [M] (assuming that the Tromms-
dorff effect is negligible and that the radical ini-
tiation rate is essentially constant). With living
polymerization in the absence of termination re-
actions, on the other hand, P̄N increases linearly
2) Monomer coupling without termination
with conversion, and in polycondensation a high
(e.g., living anionic polymerization)
degree of polymerization is achieved only at
very high conversion (e.g., 95 % conversion is
required to reach P̄N = 20).
98 Polymerization Processes

Figure 61. Schematic illustration of the course of reaction in various types of reactors
E/E 0 = mass fraction of the material throughput traversing the reactor with a residence time between t and t + dt; C A = concen-
tration of component A; t = time; τ = mean residence time; x = spatial coordinate; L = reactor length; n = order of the reaction

Batch and plug flow reactors provide equiva-


lent results, so they can conveniently be consid-
ered together. There are thus nine different com-
binations possible for the three types of poly-
merization and the three ideal reactor types, as
summarized in Table 5, which also includes a
qualitative description of the resulting molecu-
lar mass distributions.

3.3.6.1.1. Monomer Coupling with


Bimolecular Termination Plug Flow and
Batch Reactors (CPFR/BR)

Figure 62. Cumulative degree of polymerization P̄N as a Under the assumption that R1 = constant (con-
function of conversion for various polymerization reactions stant radical generation rate; half-life t 1/2 of
a) Monomer coupling with termination; b) Monomer cou- the initiator ≫ residence time τ of reactor con-
pling without termination; c) Polymer coupling tents) and [M] = constant (small conversion in-
crement), the instantaneous degree of polymer-
Polymerization Processes 99
Table 5. Molecular mass distribution for polymerization reactions in various types of reactors

ization PN , when termination is exclusively by conversion x (decreasing monomer concentra-


disproportionation, is given by tion [M]) is a set of Schulz – Flory distributions
with different PN for decreasing degrees of poly-
Kp [M] 1 merization, as shown in Figure 63.
PN = 1/2
= (3.84)
(Ktd RI ) (1 −ϕ)

and the instantaneous mass chain length distri-


bution is given by the Schulz – Flory (or most
probable) distribution [872], [873]:

W (r) = (1 −ϕ)2 rϕr−1


≈ (1 −ϕ)2 r exp = [− (1 −ϕ) r] (3.85)

where
kp [M]
ϕ=
kp [M] + (Ktd RI )1/2

The breadth of the distribution is described ei-


ther by the nonuniformity U or by the polydis-
persity index PDI.
PW
U= −1 (3.86)
PN
PW Q2 Q0
PDI = = (3.87)
PN Q21
Figure 63. Mass distribution W (r) of instantaneously
with PDI = 2 for k = 1 (bimolecular termina- formed polymer for monomer coupling with termination.
tion by disproportionation); PDI = 1.5 for k = 2 Batch reactor at small conversion increments, or homoge-
neous continuous stirred-tank reactor.
(bimolecular termination by combination). The PN, 0 = 1000, PN, x = PN, 0 (1 − x); degree of coupling
PDI (or U) can be expressed in terms of k=1
the moments of the frequency distribution of
chain lengths Q as shown in Equation (3.87). Note that PDI = 2 for all the molecular mass
Thus, if one examines the entire range of con- distributions (actually mass chain length distri-
versions (i.e., dropping the assumption that butions to be more precise) in Figure 63. By con-
[M] = constant) what results with increasing trast, the statistical variance σ 2 of the associated
100 Polymerization Processes

frequency distribution decreases with increasing chain-transfer agents) is negligible can cause
conversion. However, since σ 2 = P2N (PDI − 1), dramatic increases of molecular mass with con-
σ 2 continuously decreases as PN decreases with version. See, for example, data for the change of
conversion, except in the case of a monodisperse molecular mass distribution with conversion for
MWD with PDI = 1. For this reason, one of the the bulk homogeneous polymerization of methyl
quantities PDI or U is preferred for expressing methacrylate [380]. On the other hand, when
distribution breadth. chain transfer to monomer (as in vinyl chloride
polymerization) or to chain-transfer agent pro-
duces most of the polymer chains, the Tromms-
dorff effect has no influence on molecular mass
development. In particular, when chain trans-
fer to monomer produces most of the polymer
chains, molecular mass development will de-
pend on temperature and be independent of pro-
cess type and reactor type [889].

Figure 64. Cumulative mass distribution W (r) for polymer


formed throughout the conversion interval x = 0 to x = x for
monomer coupling with termination in a BR/CPFR
P̄N, x = 1000, k = 1 Figure 65. Radical polymerization of styrene in an HCSTR
at low viscosity. Comparison between observed (– – –) and
Integration of the distributions in Figure 63 calculated (—-) molecular mass distributions [891]
with appropriate weighting factors gives the mo- Reaction conditions:
lecular mass distribution of the final polymer Feed: 2.39 mol/L monomer, 7.44 mol/L solvent (benzene),
0.0152 mol/L initiator (AIBN); temperature 74.1 ◦ C; vis-
product at the end of the batch [874–877]. To cosity η = 1.5 mPa · s; mean residence time τ = 160 min
simplify comparison, the distributions in Fig-
ure 64 have been calculated in such a way that
for each conversion x, the degree of polymeriza-
tion of the accumulated polymer P̄N = 1000. It
is clear that distributions broaden with increas-
ing conversion. Representations of PDI and P̄N
as functions of conversion follow in Figures 68
and 69. Theoretical treatments of this case can
be found in [878–888] and in Section 2.2. Before
leaving CPFR/BR reactors, the following points
should be noted. Firstly, the Trommsdorff effect
can have a significant effect on molecular mass
development with increasing conversion. The Figure 66. Mass distribution W (r) of polymer formed by
significant lowering of either K tc or K td when monomer coupling with termination in an SCSTR [883]
chain transfer to small molecules (monomer or PN, 0 = 2000; k = 2
Polymerization Processes 101

In a CPFR/BR, the width of the copoly- Homogeneous Continuous Stirred-Tank


mer composition distribution can be substan- Reactor (HCSTR). Denbigh has shown that in
tial if one monomer type reacts much faster a homogeneous CSTR two opposing factors in-
than the other in propagation. For the same final fluence the molecular mass distribution [878–
conversion, the residence-time distribution for 881]. The constancy of [M] at steady state leads
CPFR/BR is ideal compared to the other reactor to a narrowing of the distribution relative to that
types when minimization of long-chain branch- obtained in a batch reactor, while the range of
ing, cross-linking and gel/sol ratios is required. residence times leads in principle to broadening
of the distribution. However, in radical polymer-
ization, the lifetime of a growing radical is ca.
1 s, very small compared to the mean residence
time of any industrial reactor. This first factor
(constant [M]) thus predominates and the result-
ing distribution is narrower in an HCSTR than in
a CPFR/BR. With polymerization reactions that
do not involve termination, the lifetime of the
growing chain is comparable to the mean resi-
dence time of the reactor, so that broadening of
the MWD by the residence-time distribution is
dominant.
In the steady state with [M] = constant, the
same conditions are present in the stationary
Figure 67. Mass distribution for radical polymerization in state as in a batch reactor at low conversion or at
three different reactors [885] some higher conversion over a small conversion
Conversion x = 0.6; PN,0 = 1000; k = 2 increment, resulting in the Schulz – Flory dis-
tribution described by Equations (3.85) – (3.87).
A comparison of observed and calculated mo-
lecular mass distributions for an HCSTR, as
shown in Figure 65, indicates very good agree-
ment in regions of low viscosity where the
Trommsdorff – Norrish effect plays no signifi-
cant role. Even when the Trommsdorff – Norrish
effect is appreciable, the Flory – Schulz distribu-
tion should still apply when diffusion-controlled
bimolecular termination is not chain-length de-
pendent [358]. Additional experimental studies
on continuous free-radical polymerization are
described in [367], [373], [890–895].
Experimental Theoretical
M̄ W = 20 990 M W = 20 440
M̄ N = 14 020 M N = 13 650
PDI = 1.50 PDI = 1.50

Segregated Continuous Stirred-Tank Re-


actor (SCSTR). Individual compartments, all
of which can be regarded as miniature batch re-
actors, display residence times or reaction times
Figure 68. Polydispersity index PDI = PW /PN as a function
according to the exit-age distribution function of
of conversion for radical polymerization in three different a CSTR which is
reactors 1
E (t) = exp (− t/τ ) (3.88)
τ
102 Polymerization Processes

The effect of segregation is a broadening of the or at incomplete conversion


MWD to a breadth greater than that for a batch
[M]0 − [M] [M]0 x
reactor, and the breadth increases with con- η= = (3.90)
[I]0 [I]0
version. Figure 66 provides several examples,
starting with the Schulz – Flory distribution at a If initiator molecules are counted as chain units
conversion x = 0. Figure 67 compares molecular
mass distributions for free-radical polymeriza- PN = 1 +η (3.91)
tion in the three reactor types. The correspond-
ing changes in the polydispersity index PDI are The assumption is made here that initiation is
illustrated in Figure 68, and Figure 69 shows the much faster than propagation, i.e., K i ≫ K p .
degree of polymerization for the accumulated At t = 0, all active molecules will have a de-
polymer as a function of conversion. Thus, the gree of polymerization PN = 1, and the num-
narrowest possible molecular mass distribution ber of propagation steps per active site (i.e.,
for free-radical polymerization is the Schulz – the kinetic chain length) is η = PN − 1. The case
Flory distribution and it is obtained in an HC- K i < K p leads to a broader distribution [898],
STR, and the broadest distribution is obtained in [899], but one that is still narrower than the
a SCSTR. For additional theoretical studies, see Schulz – Flory distribution. The molecular mass
[883], [885], [896], [897]. distribution corresponds to a Poisson distribu-
tion [435, pp. 346 – 339], [900] (see also Section
2.2.2.2):
e−η η r−1
W (r) = (3.92)
(1 +η) (r − 1) !

with a polydispersity
1
PDI =PW /PN = 1 + (3.93)
η

Figure 69. The course of the degrees of polymerization of


the accumulated polymer P̄N and P̄W as a function of con-
version during radical polymerization

3.3.6.1.2. Monomer Coupling Without


Termination Plug Flow and Batch Reactors
(CPFR/BR)

In the absence of termination reactions the ki-


netic chain length η corresponds to the ratio of
the initial concentrations of monomer and ini-
tiator Figure 70. Mass distribution W (r) of polymer formed by
monomer coupling without termination in a SCSTR [885]
[M]0 PN = 100; Schulz – Flory distribution for x = 0; Poisson dis-
η= (3.89) tribution for x = 1
[I]0
Polymerization Processes 103

Thus the distribution becomes narrower as large number of molecules with equal residence
the contribution from statistical broadening de- times. The molecular mass distribution is there-
creases with chain growth (increase in η) (cf. fore subject to less broadening relative to a batch
the bottom curve in Fig. 71). A comparison of reactor than in the case of the HCSTR. At very
Equations (3.87) and (3.93), or inspection of Fig- low conversion, it is possible to assume constant
ure 70 indicates that Poisson distribution is sig- monomer concentration, so the distributions in
nificantly narrower than the Schulz – Flory dis- the three reactor types are almost identical. At
tribution. high conversion the effect of the residence-time
distribution diminishes. At complete conversion
the lifetime of the active species must be signifi-
cantly shorter than the mean residence time, and
the molecular mass distribution becomes equiv-
alent to that in a batch reactor. Figure 70 shows
the transition from a Schulz – Flory to a Pois-
son distribution with increasing conversion. Fig-
ure 71 is a plot of polydispersity PDI as a func-
tion of PN and x. The bottom curve (for x = 1.0)
corresponds to Equation (3.93) for a batch reac-
tor.

3.3.6.1.3. Polymer Coupling

Only the formation of linear chains, either from


monomer of type A – B or from exact stoichio-
Figure 71. Polydispersity index PDI = PW /PN as a function metric mixtures of monomer types A – A and
of PN for polymerization by monomer coupling without ter- B – B is considered here. Ring formation is also
mination in a SCSTR [885] excluded.

Plug Flow and Batch Reactors


Homogeneous Continuous Stirred-Tank
(CPFR/BR). The functional group conversion
Reactor (HCSTR). If it is assumed that chain
p is defined as the fraction of functional groups
growth terminates abruptly as growing chains
that have reacted at a given time
leave the reactor, then the lifetime of an active
growing chain is equal to its residence time in N0 −N
p= (3.94)
the reactor. According to [879–881], the mo- N0
lecular mass distribution in an HCSTR should
be broader than for a CPFR/BR. Several au- For the degree of polymerization it follows that
thors have demonstrated that the Schulz – Flory N0 1
distribution is obtained for living polymeriza- PN = = (3.95)
N (1 −p)
tion (with instantaneous initiation) in an HCSTR
1 +p
[882], [883], [885], [901–904]. PW = (3.96)
1 −p

Segregated Continuous Stirred-Tank Re- The molecular mass distribution is given by


actor (SCSTR). The rate of living polymeriza-
tion is first-order with respect to [M], so Rp and x W (r) = (1 −p)2 r pr−1 (3.97)
are the same for both HCSTR and SCSTR [795,
pp. 309 – 331], [805]. Segregation inhibits total with
molecular mixing, whereas in an HCSTR all
PDI =PW /PN = 1 +p (3.98)
molecules are distributed according to Equation
(3.88), this is true for a SCSTR only with respect and is once again the Schulz – Flory distribu-
to all compartments, each of which contains a tion with PDI = 2 at high conversions [435,
104 Polymerization Processes

pp. 336 – 339], [873], [882], [883], [902], [905– polymerization the Schulz – Flory distribution
908]. always has a PDI = 2 because ϕ, the probabil-
ity of propagation, is almost always very close
to unity. This is not so for polycondensations,
where p can vary from 0 to 1, and thus PDI varies
from 1 to 2.

Figure 72. Molecular mass distribution W (r) of polymer


formed by polymer coupling in a CPFR/BR [873]

Figure 74. Polydispersity index PDI = PW /PN as a function


of PN for polymer coupling [885]

Homogeneous Continuous Stirred-Tank


Reactor (HCSTR). Monomer and polymer
molecules are capable of reacting with each
other throughout their entire residence time in
the reactor. Thus, according to the principle of
Denbigh [879–881], a broader molecular mass
distribution is to be expected in an HCSTR than
in a batch reactor. The broad distribution of res-
idence times means that material leaving the re-
actor always contains a relatively large amount
of monomer and polymer with a low degree of
polymerization. On the other hand, with increas-
ing molecular mass the probability of the forma-
tion of very large macromolecules by the cou-
pling of two smaller ones increases greatly. Cal-
culations give a very broad distribution (Fig. 73)
Figure 73. Molecular mass distribution W (r) for polymer- in which the number-average degree of polymer-
ization by polymer coupling in three different reactors [883]
ization PN is the same as in a batch reactor, but
Figure 72 shows the molecular mass distri- the weight-average PW is much higher [883],
bution at various conversions. For free-radical [885], [902], [907], [909], [910]:
Polymerization Processes 105

Figure 75. Composition of the copolymer produced in a batch reactor (BR) as a function of monomer composition of the
charge as well as conversion
First column: instantaneous composition; second column: instantaneous compositions, starting with monomer ratios
[M1 ] : [M2 ] = 1 : 3, 1 : 1, and 3 : 1; third column: cumulative compositions based on the same starting ratios

Segregated Continuous Stirred-Tank


1 Reactor (SCSTR). Segregation reduces the
PN = amount of polymer with very high or very low
1−p
molecular mass [883], [885]. Figure 73 com-
pares the molecular mass distributions expected
1 +p2 for the three reactor types and Figure 74 shows
PW = (3.99) the polydispersity index PDI as a function of
(1 −p)2
PN . In terms of both PN and PDI, the SCSTR
This corresponds also to a considerably higher lies between the CPFR/BR and the HCSTR.
PDI, which increases rapidly with increasing Note that the conclusions drawn in the pre-
conversion (Fig. 74): ceding two sections are a result of purely the-
oretical considerations, and they actually have
1 +p2 little to do with practice. On the one hand, Equa-
PDI = (3.100)
(1 −p) tion (3.95) (see also Fig. 62) indicates that at-
taining a reasonably high degree of polymer-
ization requires a very high conversion; HC-
106 Polymerization Processes

Figure 76. Ternary copolymerization in the system styrene (M 1 ), 2,5-dichlorostyrene (M2 ), and acrylonitrile (M3 ) [912]
A) Instantaneous copolymer composition (arrow point) as a function of composition of the monomer mixture (origin of the
arrow); B) Partial azeotropes, where arrows outside the diagram indicate the positions of binary azeotropes

STR and SCSTR systems would therefore ap- more than two comonomers. Figure 76 shows a
pear inappropriate (extreme viscosities would three-fold combination represented in terms of
be inappropriate for a CSTR). On the other triangular coordinates. The compositions of var-
hand, many polycondensations and polyaddi- ious comonomer mixtures are indicated by the
tions are accompanied by rearrangements at origins of arrows, with the corresponding arrow
the heteroatoms, including transesterifications points showing the instantaneous compositions
or transamidations. This means that polymer of the resulting polymers. The length of an ar-
with a Schulz – Flory distribution would be ob- row is thus a measure of the deviation between
tained even in an HCSTR [435, pp. 336 – 339], compositions of free monomer and copolymer.
[911]. Such an arrow degenerates to a point in the case
of an azeotropic composition. A phenomenon
known as “partial azeotropy” is of interest in the
3.3.6.1.4. Copolymerization context of multicomponent systems, with only
The theory of copolymerization kinetics is co- one of the monomers represented by an identical
vered in Section 2.3. However, the performance mole fraction in both copolymer and monomer
of the three reactor types (CPFR/BR, HCSTR, mixture. Thus, for ternary copolymerization
and SCSTR) were not compared and therefore m1 [M1 ]
this topic is considered here. = (3.101)
m1 +m2 +m3 [M1 ] + [M2 ] + [M3 ]

Copolymerization in CPFR/BR Reactors. A systematic computer search of the literature


Figure 75 gives several illustrations of composi- has produced copolymerization parameters for
tional drift for binary copolymerization for var- 37 ternary azeotropes, 4 quaternary azeotropes,
ious comonomer pairs in continuous plug flow and one 5-component azeotrope [912]. Fig-
and batch reactors. The first two columns are ure 77 shows copolymer composition distribu-
for the mole fraction of monomer 1 chemically tions (CCDs) for the monomer pairs and com-
bound in copolymer chains produced instanta- positions treated in Figure 75. Copolymer of
neously (over a small conversion or time in- uniform composition can be expected during
terval) as a function of mole fraction of free a batch polymerization only at the azeotropic
monomer 1 and of conversion. The third col- composition, or at extremely low conversions.
umn shows the mole fraction of monomer 1 in Strictly speaking, a copolymer is not uniform at
the accumulated copolymer chains versus con- all molecular masses even under azeotropic con-
version. Azeotropes also occur in systems with ditions. What is in fact produced is more nearly a
Polymerization Processes 107

Figure 77. Copolymer composition distributions (CCDs) for the monomer pairs of Figure 75. CCDs of completely polymer-
ized batches with molar ratios [M1 ] : [M2 ] = 1 : 3 (first column), 1 : 1 (second column), and 3 : 1 (third column)
The mean composition of the copolymer corresponds to that in the batch. The height of each block represents that fraction of
the overall copolymer that has a composition within the block width of 0.05
Experimental data and calculated values as a function of f 1 ; 60 ◦ C, [I] = 1 g/L azobisisobutyronitrile

statistical distribution about a mean [913], [914], composition. The only other situation in which
although at high degrees of polymerization this the composition of the copolymer corresponds
variation is minor in comparison to the hetero- precisely to the composition of the monomer
geneity that results from progressive changes in mixture in the feed is the technically unrealis-
the composition of the monomer mixture with tic limit of complete conversion, which would
conversion. require impractically long residence times. The
instantaneous copolymer equation, of course,
Copolymerization in an HCSTR. In an can always be used when monomer composi-
HCSTR operating at steady state the tempera- tion in the exit stream is used (as opposed to
ture and concentration are constant in both space inlet stream) [915], [916]. Figure 78 shows the
and time. The result is therefore a chemically change in steady-state copolymer composition
uniform copolymer. In the limiting case of very for the system styrene/acrylonitrile (40/60) as
low conversion, the copolymer composition cor- a function of conversion. Thus, a copolymer
responds to the predictions of Equation (2.121) with a particular composition can be made in
and Figure 75 when concentrations of monomer practice (at least approximately) either by estab-
in the inlet stream are used to calculate monomer lishing in advance a preferred composition for
108 Polymerization Processes

the monomer feed and then varying the extent of again producing a uniform CCD. The amount
conversion, or by varying the feed composition of homopolymer produced is, of course, greater
at a fixed conversion. at 73 % than at 35 % conversion.

Figure 79. Copolymerization of methyl methacrylate


Figure 78. Continuous copolymerization in the system (M1 ) – vinyl acetate (M2 ) in three types of reactor: BR,
styrene ( f 1 = 0.4) / acrylonitrile ( f 2 = 0.6) [915] HCSTR, and SCSTR [917]
Composition of accumulated copolymer F i as a function of A) Instantaneous composition, calculated with r 1 = 20,
conversion r 2 = 0.015; B) Copolymer composition distribution (CCD)
at f 1 = 0.4, 35 and 73 % conversion in a BR (curves 1 and
4), a HCSTR (2 and 5), and a SCSTR (3 and 6)
Copolymerization in a SCSTR. It was
noted earlier that true mixing is often less com- Curve: 1 2 3 4 5 6
Conversion x: 0.35 0.35 0.35 0.73 0.73 0.73
plete than envisioned by the HCSTR concept in
a polymerization reactor operating under condi-
tions of high viscosity. Most industrial reactors
probably provide an efficiency of mixing that 3.3.6.1.5. Long-Chain Branching and
falls between the values assumed for the HCSTR Cross-Linking
and SCSTR. By introducing certain simplify-
ing assumptions, O’Driscoll and Knorr [917] The amounts of long-chain branching and cross-
were able to calculate CCDs corresponding to linking (with gel formation) obtained in free-
the three reactor models CPFR/BR, HCSTR, and radical polymerizations depend strongly on the
SCSTR. Figure 79 A shows a representation of residence-time distribution of the reactor and the
a typical instantaneous copolymer composition conversion (or polymer concentration). Com-
diagram. Figure 61 B shows CCD curves for the pared to the HCSTR, the CPFR/BR gives
three ideal reactor types at two conversions. The less long-chain branching, cross-linking (cross-
feed composition is fixed at f 1 = 0.4 for all cases linking density), and gel for the same con-
shown. Open circles represent mean copolymer version. Apparenty, calculations on long-chain
compositions F 1 which are not very different in branching and cross-linking in a SCSTR have
the three cases. The differences in the CCDs are not been performed; however, one could specu-
greater, however. At 35 % conversion the CCD late that the extent of long-chain branching in a
is relatively narrow in a CPFR/BR (curve 1), SCSTR would lie between that for a CPFR/BR
absolutely uniform in an HCSTR (curve 2), and and HCSTR. Graessley et al. [918] have car-
already quite nonuniform in a SCSTR (curve ried out theoretical modeling and experiments
3). In the latter, more than 10 % of the poly- for CPFR/BR and HCSTR using vinyl acetate
mer is homopolymer of monomer 2. A SCSTR homopolymerization. Their experimental poly-
always provides several batch-reactor-like com- dispersity data are shown in Figure 80 for
partments with very long residence times, high CPFR/BR and HCSTR. Note that long-chain
conversions, and extreme changes in monomer branching is due to chain transfer to polymer and
and copolymer compositions. At 73 % conver- addition of polymeric radicals to terminal double
sion the CCD shows nearly as much nonuni- bonds on polymer chains and that such branch-
formity with a CPFR/Batch (curve 4) as with a ing leads to increases in the polydispersity in-
SCSTR (curve 6), only the HCSTR (curve 5) dex starting at 2.0 for linear chains produced at
Polymerization Processes 109

low conversion (in other words, the larger the


PDI, the higher the long-chain branching fre-
quency). A survey by Reichert and Moritz
[918] covers the effect of reactor type as well
as the effect of changes in the properties of
the reacting mixture on the polymerization pro-
cesses. For example, the reaction mixture vis-
cosity may increase by several orders of magni-
tude with conversion and this may affect kinetic
rates (e.g., Trommsdorff – Norrish effect, reduc-
tion in initiator efficiency, and propagation rate
constant), as well as processes of heat, mass, and
momentum transfer, the quality of mixing and
the residence-time distribution of the continu-
ous process. They also point out that the reactor
type (RTD and degree of segregation) can signif- Figure 81. Anionic (living) polymerization of butadiene
icantly affect long-branching and cross-linking showing measured versus model-predicted heterogeneity
index HI (polydispersity index PDI) [923]
levels in free-radical polymerization.

Figure 80. Polydispersity index PDI = PW /PN as a func-


tion of conversion for batch and CSTR bulk polymerization
of vinyl acetate
a) Continuous; b) Bulk Figure 82. Anionic (living) polymerization of butadiene
showing the effect of RTD on heterogeneity index or breadth
A semi-batch process has been investigated of MWD [923]
using a comprehensive kinetic model for long- a) Batch/plug flow reactor; b) 2 CSTRs
chain branching and cross-linking for the pur- The MWDs produced by living anionic poly-
poses of producing homogeneous polymeric merization, as shown earlier, are strong func-
networks [356], [363–365], [919], [920]. In the tions of RTD. Chang et al. [923] used changes
model, “the cross-linking/branching density dis- in RTD to modify MWD in butadiene poly-
tribution” was proposed. The use of the dis- merization in experiments involving a reactor
tribution enables one to calculate MWD irre- train, while Meira et al. [924–930] used peri-
spective of the reactor type [921]. This model odic operation (in residence time) to broaden
has been successfully applied to various batch the MWD as well as controlled addition of
copolymerizations involving methyl methacry- killing agent in semi-batch operation. Chang et
late – ethylene glycol dimethacrylate and acryl- al. [923] developed a comprehensive model for
amide –N,N ′ -methylene bisacrylamide. It has the anionic solution polymerization of butadiene
been shown that the variance of the cross-link with n-butyllithium initiator, tetramethylethyl-
density distribution is reduced with the use of enediamine modifier, and hexane solvent. The
a chain transfer agent. Boots [922] confirmed model successfully predicted monomer conver-
the effect of shorter primary chains qualitatively sion, vinyl structure, and polydispersity for both
using a Monte Carlo simulation. batch reactors and continuous reactor trains.
110 Polymerization Processes

Figure 81 shows predicted versus measured be valid for low-viscosity solution processes;
heterogeneity index (polydispersity index or however, in bulk polymerization, the viscosity
M W /M N ), covering the range from that obtained can increase by several orders of magnitude and,
in a PFR or batch reactor to that obtained in a therefore, all 3 assumptions are questionable.
single CSTR. Figure 82 shows the polydisper- Henderson again points out that a thermal run-
sity for 2 CSTRs of equal volume in series with away can occur when:
equal conversion of monomer in each, giving the
lowest value of polydispersity, 1.5. 1) The process side heat-transfer coefficient de-
creases significantly
2) The agitator provides insufficient mixing of
3.3.6.2. Reactor Dynamics and Stability the reaction mass, allowing hot spots to de-
velop
Reichert and Moritz [918] discuss reactor sta- 3) The agitator is putting more mechanical
bility and safety under the subtopics thermal and work into the system than designed for
concentration stability. The strong nonlinear de-
pendence of polymerization rate (or heat pro- Ray et al. [788–791], [936–945] have made
duction rate) on temperature is due to the Ar- the most significant contributions to the under-
rhenius dependence of rate on temperature and standing of polymer reactor dynamics and sta-
to the large activation energy and heat of poly- bility. Comprehensive reactor models were de-
merization. The dynamic behavior of polymer veloped for a range of polymerization processes
reactors is therefore greatly affected by temper- and reactor types: nonisothermal solution homo-
ature. The influence of the nonlinear dependence and copolymerization in a CSTR [931], [939],
of polymerization rate on monomer concentra- [940]; batch and continuous emulsion polymer-
tion, on reactor dynamics is much smaller than ization reactors [788–791], [941], [942]; hetero-
that for thermal effects. Both static and dynamic geneous coordination polymerization for both
instabilities may occur when stability limits are liquid and gas dispersion reactors [943–945].
crossed. The reactor is statically unstable if tem- The contributions by Ray et al. are notable not
perature and conversion rapidly move from one only for the comprehensive mathematical mod-
state to another as if alternately experiencing “ig- els but also for attempts at experimental verifi-
nition” and “extinction”. Dynamic instabilities cation. In this regard, the experimental investi-
give rise to periodic changes in temperature and gation of the effect of chain-transfer agents on
monomer conversion with certain phase shifts the stability of continuous latex reactors is very
observed [931], [932]. Thermal instabilities may revealing. Whereas under most operation con-
occur in various reactor types (CSTR, PFR, non- ditions the continuous emulsion polymerization
ideal reactors) [933], [934]. Henderson [935] of styrene is stable, the addition of a chain-trans-
points out that for all prior publications on mul- fer agent to the recipe can lead to sustained os-
tiple steady states and instability for free radical cillations. Some of Ray’s general conclusions
polymerization in a CSTR, three basic assump- include:
tions were made: 1) Bifurcation analysis reveals a parame-
1) The heat-transfer coefficient of the vessel ter space rich in dynamic phenomena. It
wall does not change with monomer conver- confirms limit cycle behavior (in which
sion monomer conversion does not smoothly ap-
2) The viscous dissipation term for the agitation proach a constant steady-state value, but
is negligible rather oscillates with constant frequency and
3) Perfect mixing (ideal RTD and micromixing) amplitude indefinitely) and analyzes the sta-
bility of periodic branches in detail providing
Mechanical work of the agitator can be a sig- useful guidelines for experimental design.
nificant factor in the energy balance for a pilot- 2) Full-scale reactors exhibit dynamic behavior
scale reactor (perhaps for solution and bulk poly- of high complexity.
merization, but not for suspension and emulsion 3) Semi-batch reactors can be operated under
polymerization), while it is almost negligible for conditions that insure a fast approach to a
commercial scale. Assumptions (2) and (3) may
Polymerization Processes 111

steady-state operating value, hence permit- time require measurements and parameter up-
ting almost perfect control. The possibility of dating to track and control the process. The need
oscillatory behavior was, however, demon- for monitoring the polymerization is clear and
strated. to be most effective in controlling the process,
4) Multiplicity (multiple steady states) oc- on-line monitoring is highly desirable. A survey
curs under isothermal conditions due to the of on-line sensors for polymerization reactors
Trommsdorff – Norrish effect and there ap- was made by Chien and Penlidis [947]. Al-
pears to be no relationship between phenom- though considerable advances have been made
ena causing multiplicity and reactor oscilla- in polymer reaction engineering over the past
tions. two decades, sensor technology for on-line mon-
5) The PSD plays a major role in latex reac- itoring is still in its infancy. The major problem
tor stability. Models which calculate only a has been the on-line measurement of polymer
mean diameter d (and not the full PSD) can- and particle properties (copolymer composition,
not predict oscillations under any operating molecular mass averages, long-chain branching,
conditions. as well as cross-linked gel content and poly-
6) Polymer particle multiplicity, which leads to mer particle size distribution in suspension and
polymer melting and sticking (in gas-phase emulsion polymerization). Sensors in the reac-
heterogeneous coordination polymerization, tion mixture are readily fouled by sticky polymer
such as UNIPOL) can cause a thermal run- and polymer particles. Another factor to explain
away of the reactor (UNIPOL is open-loop the lack of progress is the fact that sensor devel-
unstable). Multiplicity and Hopf bifurcation opment is a multidisciplinary task with exper-
of the fluidized-bed reactor can lead to run- tise required in statistics, mathematical model-
away and long periods of slowly damped os- ing, process understanding, reactor design, ad-
cillations. The reactor is prone to runaway vanced control concepts, electronics, and in-
even under feedback control because of cool- strumentation engineering [947]. Even the mea-
ing limitations and nonlinear sensitivity to surement of polymer properties off-line is diffi-
process disturbances. cult (e.g., measurement of molecular mass aver-
Lu and Brooks [946] have confirmed that la- ages and long-chain-branching frequencies for
tex reactor startup policy can have a significant heterogeneous copolymers [870]). Chien and
effect on reactor stability. Penlidis [947] classify sensors as those which
It is clear, based on these studies of reactor monitor reactor operation [by measuring tem-
stability, that a comprehensive dynamic model perature, pressure, flows and level (of reaction
of the polymerization reactor is required to prop- mixture in the reactor)], and those that moni-
erly design a continuous manufacturing process tor polymer properties (e.g., by measuring den-
(process stability, control, and startup proce- sity, viscosity, concentrations, copolymer com-
dures should be considered at the design stage). position, particle size, using densitometers, vis-
The interactions are too complex to permit an ef- cometers, gas and liquid chromatographs, IR and
fective design based on costly experimentation UV spectrophotometers, light scattering, ultra-
alone. sonics, fiber-optic techniques, robotics, reactor
energy balances, and state estimation/filtering
techniques). Most of the commercial applica-
3.3.6.3. On-Line Monitoring and Control tions of on-line monitoring involve those which
monitor reactor operation, although reactor en-
The goals in commercial polymerizations are ergy balances are commonly used to monitor
consistent polymer properties, high productiv- monomer conversion. The main problem with
ity, and safe operation achieved in the most eco- the successful use of reactor energy balances lies
nomic manner possible. Polymerizations (par- in the evaluation of the transient accumulation
ticularly ionic and free-radical) are very sensi- terms (time derivatives of temperature) and in
tive to temperature and impurities (radical scav- accounting for the propagation of measurement
engers and poisons at ppm levels) and, therefore, errors (temperature and flow rate) into the cal-
adjustments may have to be made during poly- culated heat release term. Acceptable results are
merization. Process models when applied in real
112 Polymerization Processes

often obtained through rapid sampling and then timating these properties on-line by using fun-
averaging or filtering all the measurements at damental mathematical models have been pro-
regular short intervals [949]. posed. State estimators such as the Kalman Filter
Two promising on-line monitoring tech- [952], [953] and recursive parameter estimators
niques are based on the use of ultrasound and [954] have been advocated to update model pa-
NIRS (near infrared spectroscopy) [948]. Ul- rameters and predict polymer properties.
trasonic velocity and attenuation measurements Feedback control over measured or inferred
have been used for monitoring high-temperature properties in continuous reactors has been ac-
composition in extruders and melt-transfer lines; complished by using empirical models [955],
high-temperature imaging studies in extruders; [956].
and for ultrasonic spectroscopy of dispersions. However, in spite of the many advances that
NIRS is an older technique which has been im- have been made in the methodology for the ad-
proved by newly available computing power. vanced control of polymerization reactors, there
It is the most promising process spectroscopy are still few industrial applications.
available at present because of its process com-
patability and ability to analyze multicomponent
mixtures. NIRS has been applied as an on-line
monitoring technique for both extruders used for
chemical modification and polymer reactors.
The on-line monitoring and control of poly-
mer and polymer particle properties should pro-
vide a major improvement in product consis-
tency and productivity.
There have been a number of excellent re-
views on control of polymerization reactors
[947], [949], [950]. Most industrial control prac-
tice to date has centered around batch reactor
sequencing and the control of process variables
such as temperature, pressure, and viscosity. Re- Figure 83. Continuous emulsion polymerization in a reac-
actor energy balances, as mentioned earlier, have tor train with a split feed (small polymer particle nucleating
also been used to infer polymer production rates reactor followed by one or more larger finishing reactors)
[956]
and monomer conversion. a) Small seeding reactor; b) First large reactor in train
Much of the academic literature has focused
on the computation of optimal operating poli- To again emphasize the point that control
cies for batch reactors by using nonlinear pro- problems should be considered at the design
gramming or the maximum principle [374]. stage of a polymer manufacturing process, ref-
These policies are open-loop in that they provide erence is made to the well-known stability prob-
optimal feed and temperature policies versus lems that are experienced with continuous emul-
time. Kozub and MacGregor [951] presented sion polymerization (in a single CSTR) involv-
a much simpler solution to the optimal trajec- ing monomer(s) that experience Case I kinetics
tory problems based on an “instantaneous prop- (e.g., vinyl acetate and vinyl chloride). Modern
erty” approach. This latter approach is applica- control theory did not provide suitable control
ble only to chain-growth polymerizations (see methodology for the single CSTR. It was de-
Section 2.2 for details of instantaneous property cided, therefore to go back to the design stage,
calculations), but it allows for the feedback im- redesign the process by using a dynamic reac-
plementation of these optimal policies via non- tor model and minimize control problems at the
linear control. design stage. The new design which is shown
Due to the lack of sufficient on-line sen- in Figure 83 has almost trivial control problems
sors to monitor polymer properties of interest or [949], [956]. Experimental verification of the in-
because some property measurements are only crease in stability of this new design may be seen
available infrequently from quality control lab- in Figure 84. The initial response (conversion
oratories, various methods for inferring or es-
Polymerization Processes 113

Figure 84. Conversion responses to CSTR reactor system shown in Figure 83 for continuous emulsion polymerization of vinyl
acetate
Part A: operation with a single CSTR (shows sustained oscillations)
Part B: switching over to seed reactor followed by one finishing reactor (responses for both reactors are shown, with very
stable response for downstream reactor and minor disturbance for seed reactor) [611]

versus time) is for a single CSTR. The follow- 9. R. W. Lenz: Organic Chemistry of Synthetic
ing two responses are for the small seed reactor High Polymers, Wiley-Interscience, New York
and for the larger finishing reactor. 1968.
10. F. W. Billmeyer: Te xbook of Polymer Science,
Wiley-Interscience, New York 1971.
4. References 11. D. H. Solomon (ed.): Step-Growth
Polymerization, Marcel Dekker, New York
1. J. V. Seppala, K. H. Reichert: “Trends in 1972.
Polymer Reaction Engineering,” Kem.-Kemi 12. F. A. Bovey, F. W. Winslow (eds.):
17 (1990) 346. Macromolecules, Academic Press, New York
2. W. H. Ray: “Current Problems in 1979.
Polymerization Reaction Engineering,” ACS 13. G. Odian: Principles of Polymerization,
Symp. Ser. 226 (1983) 101. Wiley-Interscience, New York 1981.
3. Ullmann, 4th ed., 19, 109. 14. A. Rudin: The Elements of Polymer Science
4. H. Gerrens, Chem. Ing. Tech. 52 (1980) 477; and Engineering, Academic Press, Orlando,
Chemical Engineering 4 (1981) 1. Fla. 1982.
5. H. Gerrens, Chem. Tech. (1982) June, 380; 15. P. C. Hiemenz: Polymer Chemistry, Marcel
(1982) July, 434. Dekker, New York 1984.
6. D. H. Sebastian, J. A. Biesenberger: 16. S. K. Gupta, A. Kumar: Reaction Engineering
Polymerization Engineering, Wiley, New York of Step Growth Polymerization, Plenum Press,
1983. New York 1987.
7. K. H. Reichert, H. U. Moritz: “Polymer 17. R. B. Seymour, C. E. Carraher: Polymer
Reaction Engineering,” Comprehensive Chemistry, Marcel Dekker, New York 1988.
Polymer Science, vol. 3, Pergamon Press, New
York 1989, p. 327. Specific References
18. J. L. Stanford, R. F. T. Stepto, J. Chem. Soc.,
General References Faraday Trans. 71 (1975) 1292.
8. P. J. Flory: Principles of Polymer Chemistry, 19. R. F. T. Stepto: “Intra-Molecular Reaction and
Cornell University Press, Ithaca, New York Gelation in Condensation or Random
1953. Polymerization,” in R. N. Haward (ed.):
114 Polymerization Processes

Developments in Polymerization-3, Applied 45. M. Gordon, Proc. R. Soc. London A 268


Science Publ., Barking 1982, p. 81. (1962) 240.
20. W. H. Carothers, Trans. Faraday Soc. 32 46. M. Gordon, S. B. Ross-Murphy, Pure Appl.
(1936) 39. Chem. 43 (1975) 1.
21. P. J. Flory: Principles of Polymer Chemistry, 47. M. Gordon, W. B. Temple: “The Graph-Like
Cornell University Press, Ithaca, N.Y. 1953, State of Matter and Polymer Science,” in A. T.
chap. 8. Balaban (ed.): Chemical Application of Graph
22. W. Burchard, Polymer 20 (1979) 589. Theory, Academic Press, New York 1976,
23. D. Durand, C. M. Bruneau, Makromol. Chem. p. 299.
180 (1979) 2947. 48. S. I. Kuchanov, S. V. Koralev, S. V. Panyukov,
24. T. M. Orlova, S.-S. A. Pavlova, L. V. Adv. Chem. Phys. 72 (1988) 115.
Dubrovina, J. Polym. Sci., Polym. Chem. Ed. 49. D. S. Pearson, W. W. Graessley,
17 (1979) 2209. Macromolecules 11 (1978) 528. O. Durand,
25. F. Lopez-Serrano, J. M. Castro, C. W. C. M. Bruneau, Makromol. Chem. 83 (1982)
Macosko, M. Tirrell, Polymer 21 (1980) 263. 1007, 1021.
26. E. Ozizmir, G. Odian, J. Polym. Sci., Polym. 50. D. R. Miller, C. W. Macosko, J. Polym. Sci.
Chem. Ed. 18 (1980) 2281. Polym. Phys. Ed. 26 (1988) 1.
27. W. H. Abraham, Ind. Eng. Chem. Fundam. 2 51. C. W. Macosko, D. R. Miller, Macromolecules
(1963) 221. 9 (1976) 199.
28. H. Kilkson, Ind. Eng. Chem. Fundam. 3 52. D. R. Miller, C. W. Macosko, Macromolecules
(1964) 281. 9 (1976) 206.
29. J. A. Biesenberger, AIChE J. 11 (1965) 369. 53. D. R. Miller, C. W. Macosko, Macromolecules
30. H. Kilkson, Ind. Eng. Chem. Fundam. 7 11 (1978) 656.
(1968) 354. 54. D. Durand, C. M. Bruneau, Polymer 24 (1983)
31. T. T. Szabo, J. F. Leathrum, J. Appl. Polym. Sci. 587.
13 (1969) 477. 55. M. Gordon, G. R. Scantlebury, Proc. R. Soc.,
32. W. H. Ray, J. Macromol. Sci. Rev. Macromol. London, A 292 (1966) 380. M. Gordon, G. N.
Chem. C 8 (1972) 1. Malcom, Proc. Roy. Soc, London, Ser. A 292
33. W. H. Ray: “Polymerization Reaction (1966) 29. D. Durand, C. M. Bruneau, Polymer
Engineering,” in L. Lapidus, N. R. Amundson 24 (1983) 592.
(eds.): Chemical Reactor Theory, 56. D. R. Miller, C. W. Macosko, Macromolecules
Prentice-Hall, Englewood Cliffs, N. J. 1977, 13 (1980) 1063.
p. 532. 57. R. W. Kilb, J. Phys. Chem. 62 (1958) 969.
34. N. A. Dotson, Macromolecules 22 (1989) 58. M. Gordon, G. R. Scantlebury, J. Polym. Sci.,
3690. Part C 16 (1968) 3933.
35. P. W. Morgan: Condensation Polymers by 59. W. B. Temple, Makromol. Chem. 160 (1972)
Interfacial and Solution Methods, 277.
Wiley-Interscience, New York 1965. 60. H. Galina, A. Szustalewicz, Macromolecules
36. F. Millich, C. E. Carraher, Jr., J. Preston (eds.): 22 (1989) 3124.
Interfacial Synthesis, vol. III, Marcel Dekker, 61. H. Tobita, Ph. D. Thesis, McMaster University,
New York 1982. Hamilton, Canada 1990.
37. P. J. Flory, J. Am. Chem. Soc. 63 (1941) 3083. 62. A. N. Dotson, R. Galvan, C. W. Macosko,
38. P. J. Flory, J. Am. Chem. Soc. 63 (1941) 3091. Macromolecules 21 (1988) 2560.
39. P. J. Flory, J. Am. Chem. Soc. 63 (1941) 3096. 63. O. Saito, J. Phys. Soc. Jpn. 13 (1958) 198.
40. C. Cozewith, W. W. Graessley, G. Ver Strate, 64. O. Saito: “Statistical Theory of Crosslinking”,
Chem. Eng. Sci. 34 (1979) 245. in M. Dole (ed.): The Radiation of
41. S. K: Gupta, A. Kumar: Reaction Engineering Macromolecules, Academic Press, New York
of Step Growth Polymerization, Plenum Press, 1972, p. 223.
New York 1987, chap. 4. 65. S. I. Kuchanov, L. M. Pis’men, Polym. Sci.
42. W. H. Stockmayer, J. Chem. Phys. 11 (1943) USSR (Engl. Transl.) 14 (1972) 985. M.
45. Gordon, G. N. Malcom, Proc. R. Soc. London
43. W. H. Stockmayer, J. Chem. Phys. 12 (1944) A, 292 (1966) 29. D. Durand, C. M. Bruneau,
125. Polymer 24 (1983) 592.
44. W. H. Stockmayer, J. Polym. Sci. 9 (1952) 69; 66. K. Dusek, Makromol. Chem. Suppl. 2 (1979)
J. Polym. Sci. 11 (1953) 11. 35.
Polymerization Processes 115

67. E. Donoghue, J. H. Gibbs, J. Chem. Phys. 70 89. S. S. Ivanchev, Polym. Sci. USSR, (Engl.
(1979) 2346. Transl.) 20 (1978) 2157.
68. S. I. Kuchanov, Ye. S. Povolotskaya, Polym. 90. V. R. Kamath, Mod. Plast. 58 (1981) Sept.,
Sci. USSR (Engl. Transl.) 24 (1982) 2499. 106.
69. S. I. Kuchanov, Ye. S. Povolotskaya, Polym. 91. K. Y. Choi, W. R. Liang, G. D. Lei, J. Appl.
Sci. USSR (Engl. Transl.) 24 (1982) 2512. Polym. Sci. 35 (1988) 1547.
70. J. Mikes, K. Dusek, Macromolecules 15 92. M. A. Villalobos, M. Eng. Thesis, McMaster
(1982) 93. University, Hamilton, Canada 1989.
71. S. R. Broadbent, J. M. Hammersley, Proc. 93. F. R. Mayo, J. Am. Chem. Soc. 90 (1968) 1289.
Cambridge Philos. Soc. 53 (1957) 629. 94. W. D. Graham, J. G. Green, W. A. Pryor, J.
72. J. M. Hammersley, Proc. Cambridge Philos. Org. Chem. 44 (1979) 907.
Soc. 53 (1957) 642. 95. F. R. Mayo, J. Am. Chem. Soc. 75 (1953) 6133.
73. H. L. Frisch, J. M. Hammersley, J. Soc. Ind. 96. A. W. Hui, A. E. Hamielec, J. Appl. Polym. Sci.
Appl. Math. 11 (1963) 894. 16 (1972) 749.
74. J. W. Essam: “Percolation and Cluster Size,” in 97. A. Husain, A. E. Hamielec, J. Appl. Polym.
C. Domb, M. Green (eds.): Phase Transitions Sci. 22 (1978) 1207.
and Critical Phenomena, Academic Press, 98. N. J. Barr, W. I. Bengough, G. Beveridge, G. B.
New York 1972, p. 197. Park, Eur. Polym. J. 14 (1978) 245.
75. S. Kirkpatrick, Rev. Mod. Phys. 45 (1973) 574. 99. H. F. Kaufmann, Makromol. Chem. 180
76. C. Domb, E. Stoll, T. Schneider, Contemp. (1979) 2649, 2665, 2681.
Phys. 21 (1980) 577. 100. Y. K. Chong, E. Rizzardo, D. H. Solomon, J.
77. D. Stauffer: Introduction to Percolation Am. Chem. Soc. 105 (1983) 7761.
Theory, Taylor and Francis, London 1985. 101. C. W. Wilson III, E. R. Santee, Jr., J. Polym.
78. D. Stauffer, A. Conoglio, M. Adam, Adv. Sci., Part C 8 (1965) 97.
Polym. Sci. 44 (1982) 103. 102. G. Talamini, G. Vidotto, Makromol. Chem.
79. D. Durand: “Network Formation – From Basic 100 (1967) 48.
Theories Towards More Realistic Models,” in 103. T. G. Fox, H. W. Schnecko, Polymer 3 (1962)
R. A. Pethrick (ed.): Polymer Yearbook 3, 575.
Harwood Academic Publishers, New York 104. T. Otsu, B. Yamada, M. Imoto, J. Macoromol.
1986, p. 229. Sci. Chem. 1 (1966) 61.
80. P. G. de Gennes: Scaling Concepts in Polymer 105. F. S. Dainton, K. J. Ivin, Nature 162 (1948)
Physics, Cornell University Press, Ithaca, N.Y. 705.
1979, p. 139. 106. H. Sawada: Thermodynamics of
81. O. Vogl: “Kinetics of Aldehyde Polymerization, Marcel Dekker, New York
Polymerization,” C. H. Bamford, C. F. H. 1976.
Tipper (eds.): Comprehensive Chemical 107. H. W. McCormick, J. Polym. Sci. 25 (1957)
Kinetics, vol. 15, Elsevier, Amsterdam 1976, 488.
chap. 5. 108. T. P. Davis, K. F. O’Driscoll, M. C. Piton, M. A.
82. O. Vogl: “Aldehyde Polymers,” in Winnik, Macromolecules 22 (1989) 2785.
Encyclopedia of Polymer Science and 109. M. J. Ballard et al., Macromolecules 19 (1986)
Engineering, vol. 1, Wiley-Interscience, New 1303.
York 1985, p. 623. 110. P. J. Flory: Principles of Polymer Chemistry,
83. R. N. Noyes: “Cage Effect,” in Encyclopedia Cornell University Press, Ithaca, N.Y. 1953,
of Polymer Science and Technology, vol. 2, p. 127.
Wiley-Interscience, New York 1965, p. 796. 111. K. Horie, I. Mita, H. Kambe, J. Polym. Sci.
84. G. Odian: Principles of Polymerization, Polym. Chem. Ed. 6 (1968) 2663.
Wiley-Interscience, New York 1981, p. 216. 112. M. Stickler, Makromol. Chem. 184 (1983)
85. H. Kiefer, T. G. Traylor, J. Am. Chem. Soc. 89 2563.
(1967) 6667. 113. K. Ito, Polym. J. (Tokyo) 16 (1984) 761.
86. E. Niki, Y. Kamiya, J. Am. Chem. Soc. 96 114. G. C. Eastmond: “The Kinetics of
(1974) 2129. Free-Radical Polymerization of Vinyl
87. G. T. Russell, D. H. Napper, R. G. Gilbert, Monomers in Homogeneous Solution,” in
Macromolecules 21 (1988) 2141. C. H. Bamford, C. F. H. Tipper (eds.):
88. S. Zhu, A. E. Hamielec, Macromolecules 22 Comprehensive Chemical Kinetics, vol. 14 A,
(1989) 3098. Elsevier, Amsterdam 1976, chap. 1.
116 Polymerization Processes

115. J. C. Bevington, H. W. Mellville, R. P. Taylor, Weight Distribution of Polymers,” in J.


J. Polym. Sci. 12 (1954) 449. Brandrup, E. H. Immergut (eds.): Polymer
116. G. V. Schulz, G. Henrici-Olive, S. Olive, Handbook, vol. II, Wiley-Interscience, New
Makromol. Chem. 31 (1959) 88. York 1975, p. 405.
117. C. H. Bamford, G. C. Eastmond, D. Whittle, 134. H. Tompa: “The Calculation of Mole-Weight
Polymer 10 (1969) 771. Distributions from Kinetic Schemes,” in C. H.
118. M. Stickler, D. Panke, A. E. Hamielec, J. Bamford, C. F. H. Tipper (eds.):
Polym. Sci., Polym. Chem. Ed. 22 (1984) 2243. Comprehensive Chemical Kinetics, Elsevier,
119. G. V. Schulz, G. Haborth, Makromol. Chem. 1 Amsterdam 1976, chap. 7.
(1948) 106. 135. M. Tirrell, R. Galvan, R. L. Laurence:
120. A. M. North: “The Influence of Chain “Polymerization Reactors,” in J. J. Carberry,
Structure on the Free-Radical Termination A. Varma (eds.): Chemical Reaction and
Reaction,” in A. D. Jenkins, A. Ledwith (eds.): Reactor Engineering, Marcel Dekker, New
Reactivity, Mechanism and Structure in York 1987, chap. 11.
Polymer Chemistry, Wiley-Interscience, New 136. D. J. P. Harrison, W. R. Yates, J. F. Johnson, J.
York 1974, chap. 5. Macromol. Sci. Rev. Macromol. Chem. C 25
121. K. F. O’Driscoll: “Kinetics of Bimolecular (1985) 481.
Termination,” in Comprehensive Polymer 137. M. Andreis, J. L. Koenig, Adv. Polym. Sci. 89
Science, vol. 3 Pergamon Press, London 1989, (1989) 69.
p. 161. 138. S. Shiga, Polym. Plast. Technol. Eng. 282
122. I. Mita, K. Horie, J. Macromol. Sci., Rev. (1989) 17.
Macromol. Chem. Phys. C 27 (1987) 91. 139. C. H. Bamford, H. Tompa, J. Polym. Sci. 10
123. A. W. Hui, A. E. Hamielec, J. Polym. Sci., Part (1953) 345.
C 25 (1968) 167. 140. C. H. Bamford, H. Tompa, Trans. Faraday
124. J. H. Duerksen, A. E. Hamielec, J. Polym. Sci. Soc. 50 (1954) 1097.
25 (1968) 155. 141. C. H. Bamford, W. G. Barb, A. D. Jenkins, P. F.
125. G. Z. A. Wu, L. A. Denton, R. L. Laurence, Onyon: The Kinetics of Vinyl Polymerization
Polym. Eng. Sci. 22 (1982) 1. by Free Radical Mechanism, Butterworth
126. C. J. Kim, A. E. Hamielec, Polymer 25 (1984) Publications, London 1958, chap. 7.
845. 142. W. W. Graessley, H. Mittelhauser, R.
127. G. C. Eastmond: “Chain Transfer, Inhibition Maramba, Makromol. Chem. 86 (1965) 129.
and Retardation,” in C. H. Bamford, C. F. H. 143. W. W. Graessley, H. Mittelhauser, J. Polym.
Tipper (eds.): Comprehensive Chemical Sci., Polym. Phys. Ed. 5 (1967) 431.
Kinetics, vol. 14 A, Elsevier, Amsterdam 1976, 144. O. Saito, K. Nagasubramanian, W. W.
chap. 2. Graessely, J. Polym. Sci., Polym. Phys. Ed. 7
128. C. H. Bamford, W. G. Barb, A. D. Jenkins, P. F. (1969) 1937.
Onyon: The Kinetics of Vinyl Polymerization 145. H. Tobita, A. E. Hamielec, Makromol. Chem.,
by Free Radical Mechanism, Butterworth Macromol. Symp. 20/21 (1988) 501.
Publications, London 1958. 146. H. Tobita, A. E. Hamielec, Macromolecules
129. C. H. Bamford: “Radical Polymerization,” in 22 (1989) 3098.
Encyclopedia of Polymer Science and 147. H. Tobita, A. E. Hamielec: “Cross-linking
Technology, vol. 13, Wiley-Interscience, New Kinetics in Free-Radical Copolymerization,”
York 1988, p. 708. in K.-H. Reichert, W. Geiseler (eds.): Polymer
130. J. Brandrup, E. H. Immergut (eds.): Polymer Reaction Engineering, VCH
Handbook, Wiley-Interscience, New York Verlagsgesellschaft, Weinheim 1989, p. 43.
1975. 148. P. J. Flory, J. Am. Chem. Soc. 69 (1947) 2893.
131. Z. Tadmor, J. A. Biesenberger, Ind. Eng. 149. P. J. Flory: Principles of Polymer Chemistry,
Chem. Fundam. 5 (1966) 336. A. E. Cornell University Press, Ithaca, N.Y. 1953,
Hamielec, J. W. Hodgins, K. Tebbens, AIChE p. 384.
J. 13 (1967) 1087. 150. K. Nagasubramanian, W. W. Graessley, Chem.
132. R. Cintron-Cordero, R. A. Mostello, J. A. Eng. Sci. 25 (1970) 1549.
Biesenberger, Canad. J. Chem. Eng. 46 151. K. Nagasubramanian, W. W. Graessley, Chem.
(1968) 434. Eng. Sci. 25 (1970) 1559.
133. L. H. Peebles, Jr.: “Rates of Polymerization, 152. F. A. Bovey, F. C. Schilling, F. L. McCrackin,
Average Molecular Weights, and Molecular H. J. Wagner, Macromolecules 9 (1976) 76.
Polymerization Processes 117

153. R. Dowbenko et al., Prog. Org. Coat. 11 173. G. Odian: Principles of Polymerization,
(1983) 71. Wiley-Interscience, New York 1981, chap. 8.
154. G. A. Senich, R. E. Florin, J. Macromol. Sci., 174. G. Natta, I. Pasquon, Adv. Catal. 11 (1959) 1.
Rev. Macromol. Chem. Phys. C 24 (1984) 239. 175. J. A. Biesenberger, D. H. Sebastian: Principles
155. J. G. Kloosterboer, Adv. Polym. Sci. 84 (1988) of Polymerization Engineering,
1. Wiley-Interscience, New York 1983, p. 79,
156. T. Masuda, T. Higashimura, Macromolecules 130.
7 (1974) 728. 176. A. B. de Carvalho, P. E. Gloor, A. E. Hamielec,
157. G. C. East, H. Furukawa, Polymer 20 (1979) Polymer 30 (1989) 280.
659. 177. F. R. Mayo, F. M. Lewis, J. Am. Chem. Soc. 66
158. G. A. Olah, J. Am. Chem. Soc. 94 (1972) 808. (1944) 1594.
159. G. Odian: Principles of Polymerization,
178. T. Alfrey, Jr., G. Goldfinger, J. Chem. Phys.
Wiley-Interscience, New York 1981,
12 (1944) 205.
p. 342 – 344.
179. F. T. Wall, J. Am. Chem. Soc. 66 (1944) 2050.
160. H. Sawada: Thermodynamics of
180. I. Sakurada: Kyojugo Hanno, Society of
Polymerization, Marcel Dekker, New York
Polymer Chemistry, Tokyo 1944 p. 35.
1976, chap. 4.
161. J. P. Kennedy, R. G. Squires, Polymer 6 (1965) 181. E. Merz, T. Alfrey, Jr., G. Goldfinger, J. Polym.
579. Sci. I. (1946) 75.
162. D. H. Richards: “Anionic Polymerization,” in 182. G. E. Ham: “Theory of Copolymerization,” in
R. N. Haward (ed.): Development in G. E. Ham (ed.): Copolymerization,
Polymerisation-1, Applied Science Wiley-Interscience, New York 1964, chap. 1.
Publishers, Essex 1979, chap. 1. 183. G. E. Ham, J. Polym. Sci. 45 (1960) 169.
163. S. Bywater: “Anionic Polymerization,” in 184. F. P. Price, J. Chem. Phys. 36 (1962) 209.
Encyclopedia of Polymer Science and 185. K. Ito, Y. Yamashita, J. Polym. Sci., Part A 3
Engineering, vol. 2, Wiley-Interscience, New (1965) 2165.
York 1985, p. 1. 186. B. D. Coleman, T. G. Fox, J. Polym. Sci., Part
164. M. Fontanille: “Carbanoic Polymerization: A 1 (1963) 3183.
General Aspects and Initiation,” in 187. C. W. Pyun, J. Polym. Sci. 8 (1970) 1111.
Comprehensive Polymer Science, vol. 3, 188. T. Alfrey, Jr., G. Goldfinger, J. Chem. Phys.
Pergamon Press, London 1989, p. 365. 12 (1944) 322.
165. D. N. Bhattacharyya, C. L. Lee, J. Smid, M. 189. C. Walling, E. R. Briggs, J. Am. Chem. Soc.
Szwarc, J. Phys. Chem. 69 (1965) 612. 67 (1945) 1774.
166. D. N. Bhattacharyya, J. Smid, M. Szwarc, J. 190. T. Alfrey, Jr., G. Goldfinger, J. Chem. Phys.
Phys. Chem. 69 (1965) 624. 14 (1946) 115.
167. M. Szwarc, M. Levy, R. Milkovich, J. Am. 191. T. Fueno, J. Furukawa, J. Polym. Sci., Part A 2
Chem. Soc. 78 (1956) 2656. (1964) 3681.
168. M. Szwarc: “Living Polymers,” in 192. M. N. Galbraith, G. Moad, D. H. Solomon,
Encyclopedia of Polymer Science and T. H. Spurling, Macromolecules 20 (1987)
Technology, vol. 8, Wiley-Interscience, New 675.
York 1968, p. 303. 193. K. F. O’Driscoll, T. P. Davis, Polym. Commun.
169. J. Boor, Jr., Macromol. Rev. 2 (1967) 115. 30 (1989) 317.
170. A. Ledwith, D. C. Sherrington: “Reactivity and 194. R. Simha, H. Branson, J. Chem. Phys. 12
Mechanism in Polymerization by Complex (1944) 253.
Organometallic Derivatives,” in A. D. Jenkins, 195. W. H. Stockmayer, J. Chem. Phys. 13 (1945)
A. Ledwith (eds.): Reactivity, Mechanism and 199.
Structure in Polymer Chemistry, 196. J. Stejskal, P. Kratochvil, D. Strakava,
Wiley-Interscience, New York 1974, p. 383. Macromolecules 14 (1981) 150.
171. W. Cooper: “Kinetics of Polymerization 197. C. Tosi, G. Catinella, Makromol. Chem. 137
Initiated by Ziegler – Natta Catalysts,” in C. H. (1970) 211.
Bamford, C. F. H. Tipper (eds.): 198. J. Stejskal, P. Kratochvil, Macromolecules 20
Comprehensive Chemical Kinetics, vol. 15, (1987) 2624.
Elsevier, Amsterdam 1976, chap. 3. 199. J. C. J. F. Tacx, H. N. Lissen, A. L. German, J.
172. J. Boor, Jr.: Ziegler – Natta Catalysts and
Polym. Sci., Polym. Chem. Ed. 26 (1988) 61.
Polymerizations, Academic Press, New York
1979.
118 Polymerization Processes

200. J. C. J. F. Tacx, A. L. German, Polymer 30 224. G. C. Laurier, K. F. O’Driscoll, P. M. Reilly, J.


(1989) 918. Polym. Sci., Polym. Symp. 72 (1985) 17.
201. F. P. Price: “Copolymer Composition and 225. M. J. Box, Technometrics 12 (1970) 219.
Tacticity,” in G. G. Lowry (ed.): Markov 226. H. I. Britt, R. H. Luecke, Technometrics 15
Chains and Monte Carlo Calculations in (1973) 233.
Polymer Science, Marcel Dekker, New York 227. T. L. Sutton, J. F. MacGregor, Canad. J. Chem.
1970, p. 187. Eng. 55 (1977) 602.
202. I. Skeist, J. Am. Chem. Soc. 68 (1946) 1781. 228. H. Patino-Leal, P. M. Reilly, K. F. O’Driscoll,
203. I. H. Spinner, B. C.-Y. Lu, W. F. Graydon, J. J. Polym. Sci., Polym. Lett. Ed. 18 (1980) 219.
Am. Chem. Soc. 77 (1955) 2198. 229. P. M. Reilly, H. Patino-Leal, Technometrics 23
204. V. E. Meyer, G. G. Lowry, J. Polym. Sci., Part (1981) 221.
A 3 (1965) 2843. 230. L. H. Garcia-Rubio, M. G. Lord, J. F.
205. V. E. Meyer, G. G. Lowry, J. Polym. Sci., MacGregor, A. E. Hamielec, Polymer 26
Polym. Chem. Ed. 5 (1967) 1289. (1985) 2001.
206. W. Ring, Makromol. Chem. 101 (1967) 145. 231. M. Berger, I. Kuntz, J. Polym. Sci., Part A 2
207. R. L. Kruce, J. Polym. Sci., Part B, 5 (1967) (1964) 1687.
437. 232. D. J. T. Hill, J. H. O’Donnell, P. W. O’Sullivan,
208. R. K. S. Chan, V. E. Meyer, J. Polym. Sci., Part Macromolecules 15 (1982) 960.
C 25 (1968) 11. 233. K. F. O’Driscoll, T. P. Davis, J. Polym. Sci.,
209. A. E. Hamielec, J. F. MacGregor: “Modelling Polym. Lett. Ed. 27 (1989) 417.
Copolymerization–Control of Composition, 234. T. Fukuda, Y.-D. Ma, H. Inagaki, Makromol.
Chain Microstructure, Molecular Weight Chem., Phys. Suppl. 12 (1985) 125.
Distribution, Long Chain Branching and 235. H. Tobita, A. E. Hamietec, Polymer 32 (1991)
Cross-linking,” in K.-H. Reichert, W. Geiseler 2641.
(eds.): Polymer Reaction Engineering, Hanser 236. M. Nomura, M. Kubo, K. Fujita, Research
Publishers, New York 1983, p. 21. Reports of the Faculty of Engineering, Fukui
210. A. E. Hamielec, J. F. MacGregor, A. Penlidis, University 29 (1981) 167.
Makromol. Chem., Macromol. Symp. 10/11 237. T. O. Broadhead, A. E. Hamielec and J. F.
(1987) 521. MacGregor, Makromol. Chem., Phys. Suppl.
211. S. Igarashi, J. Polym. Sci., Part B 1 (1963) 359. 10/11 (1985) 105.
212. C. Tosi, Makromol. Chem. 108 (1967) 307. 238. H. Tobita, A. E. Hamielec, ACS Symp. Ser.
213. M. Fineman, S. D. Ross, J. Polym. Sci. 5 404 (1989) 242. H. Tobita, A. E. Hamielec.
(1950) 259. Polymer 32 (1991) 2641.
214. F. R. Mayo, C. Walling, Chem. Rev. 46 (1950) 239. H. W. Melville, B. Noble, W. F. Watson, J.
191. Polym. Sci. 2 (1947) 229.
215. T. Kelen, F. Tudos, J. Macromol. Sci., Chem. 240. H. W. Melville, B. Noble, W. F. Watson, J.
A 9 (1975) 1. Polym. Sci. 4 (1949) 629.
216. D. W. Behnken, J. Polym. Sci., Part A 2 (1964) 241. C. Walling, J. Am. Chem. Soc. 71 (1949) 1930.
645. 242. P. Wittmer, Makromol. Chem. Suppl. 3 (1979)
217. P. W. Tidwell, G. A. Mortimer, J. Polym. Sci., 129.
Part A 3 (1965) 369. 243. J. N. Atherton, A. M. North, Trans. Faraday
218. P. W. Tidwell, G. A. Mortimer, J. Macromol. Soc. 58 (1962) 2049.
Sci., Rev. Macromol. Chem. C 4 (1970) 281. 244. A. M. North: The Kinetics of Free Radical
219. R. M. Joshi, J. Macromol. Sci., Chem. A 7 Polymerization, Pergamon Press, London
(1973) 1231. 1966, p. 94.
220. R. Van der Meer, H. N. Lissen, A. L. German, 245. S. Russo, S. Munari, J. Macromol. Sci., Chem.
J. Polym. Sci., Polym. Chem. Ed. 16 (1978) A 2 (1968) 1321.
2915. 246. T. Fukuda, Y.-D. Ma, H. Inagaki, Polym. J.
221. B. Yamada, M. Itahashi, T. Otsu, J. Polym. (Tokyo) 14 (1982) 705.
Sci., Polym. Chem. Ed. 16 (1978) 1719. 247. T. Fukuda, Y.-D. Ma, H. Inagaki,
222. R. C. McFarlane, P. M. Reilly, K. F. O’Driscoll, Macromolecules 18 (1985) 17.
J. Polym. Sci., Polym. Chem. Ed. 18 (1980) 248. Y.-D. Ma, T. Fukuda, H. Inagaki,
251. Macromolecules 18 (1985) 26.
223. K. F. O’Driscoll, P. M. Reilly, Makromol. 249. T. Fukuda, K. Kubo, Y.-D. Ma, H. Inagaki,
Chem., Macromol. Symp. 10/11 (1987) 355. Polym. J. (Tokyo) 19 (1987) 523.
Polymerization Processes 119

250. T. Fukuda, Y.-D. Ma, H. Inagaki, Makromol. 277. P. J. Flory, J. Am. Chem. Soc. 69 (1947) 30.
Chem., Rapid Commun. 8 (1987) 495. 278. D. Durand, C. M. Bruneau, Eur. Polym. J. 21
251. T. Fukuda, Y.-D. Ma, K. Kubo, A. Takada, (1985) 527.
Polym. J. (Tokyo) 12 (1989) 1003. 279. D. Durand, C. M. Bruneau, Eur. Polym. J. 21
252. T. P. Davis, K. F. O’Driscoll, Makromol Chem., (1985) 611.
Rapid Commun. 10 (1989) 509. 280. H. Tobita, A. E. Hamielec: “Network
253. M. Gordon, R.-J. Roe, J. Polym. Sci. 21 Formation in Free Radical Polymerization,” in
(1956) 27 – 90. P. J. Lemstra, L. A. Kleintjens (eds.):
254. B. T. Storey, J. Polym. Sci., Part A 3 (1965) Integration of Fundamental Polymer Science
265. and Technology, vol. IV, Elsevier Applied
255. K. Ulbrich, M. Ilavsky, K. Dusek, J. Kopecek, Science Publishers, London 1990, p. 33.
Eur. Polym. J. 13 (1977) 579. 281. R. Okasha, G. Hild, P. Rempp, Eur. Polym. J.
256. K. Ulbrich, K. Dusek, M. Ilavsky, J. Kopecek, 15 (1979) 975.
Eur. Polym. J. 14 (1978) 45. 282. R. S. Whitney, W. Burchard, Makromol. Chem.
257. A. C. Shah, I. W. Parsons, R. N. Haward, 181 (1980) 869.
Polymer 21 (1980) 825. 283. C. D. Frick, A. Rudin, R. H. Wiley, J.
258. J. N. Nieto et al., Eur. Polym. J. 23 (1987) 551. Macromol. Sci., Chem. A 16 (1981) 1275.
259. W.-H. Li, A. E. Hamielec, C. M. Crowe, 284. G. Hild, R. Okasha, Makromol. Chem. 186
Polymer 30 (1989) 1513. (1985) 93.
260. S. Zhu, Y. Tian, A. E. Hamielec, D. R. Eaton, 285. G. Hild, R. Okasha, Makromol. Chem. 186
Polymer 31 (1990) 154. (1985) 389.
261. K. Horie, A. Otagawa, M. Muaoka, I. Mita, J. 286. D. T. Landin, C. W. Macosko, Macromolecules
Polym. Sci., Polym. Chem. Ed. 13 (1975) 445. 21 (1988) 846.
262. H. Kast, W. Funke, Makromol. Chem. 180 287. H. Tobita, A. E. Hamielec, Polymer 31 (1990)
(1975) 1335. 1546.
263. J. Spevacek, K. Dusek., J. Polym. Sci., Polym. 288. S. I. Kuchanov, L. M. Pis’man, Polym. Sci.,
Phys. Ed. 18 (1980) 2027. USSR (Engl. Transl.) 13 (1971) 2288.
264. H. Galina et al., Eur. Polym. J. 16 (1980) 1043. 289. H. Tobita, A. E. Hamielec, Makromol. Chem.,
265. K. Dusek: “Network Formation by Chain Macromol. Symp. 35/36 (1990) 193.
Cross-linking (Co)polymerization,” in R. N. 290. H. J. Herrmann, D. Stauffer, D. P. Landau, J.
Haward (ed.): Developments in Phys., A: Math. Gen. 16 (1983) 1221.
Polymerisation, Applied Science Pub, London 291. R. Bansil, H. J. Herrmann, D. Stauffer,
1982, p. 143. Macromolecules 17 (1984) 998.
266. M. K. Gupta, R. Bansil, J. Polym. Sci., Polym. 292. R. Bansil, H. J. Herrmann, D. Stauffer, J.
Lett. 21 (1983) 969. Polym. Sci., Polym. Symp. 73 (1985) 175.
267. M. Nagy, Colloid Polym. Sci. 263 (1985) 245. 293. H. E. Stanley, F. Family, H. Gould, J. Polym.
268. J. Baselga, I. Hernandez-Fuentes, I. F. Pierola, Sci., Polym. Symp. 73 (1985) 19.
M. A. Llorente, Macromolecules 20 (1987) 294. H. M. J. Boots: “Simulation Model for Densely
3060. Cross-Linked Networks Formed by
269. W. E. Gibbs, J. Polym. Sci., Part A 2 (1964) Chain-Reactions,” in L. A. Kleintjens, P. J.
4809. Lemstra (eds.): Integration of Fundamental
270. G. B. Butler, R. J. Angelo, J. Am. Chem. Soc. Polymer Science and Technology, Elsevier
79 (1957) 3128. Applied Science, London 1986, p. 204.
271. C. L. McCormick, G. B. Butler, J. Macromol. 295. G. P. Simon et al., Macromolecules 22 (1989)
Sci., Rev. Macromol. Chem. C 8 (1972) 201. 3555.
272. J. Roovers, G. Smets, Makromol. Chem. 60 296. H. Gerrens, Chem. Tech. vol. 12, (1982) June,
(1963) 89. 380.
273. W. E. Gibbs, R. J. McHenry, J. Polym. Sci., 297. H. Gerrens, Chem. Tech. (1982) July, 434.
Part A 2 (1964) 5277. 298. O. Levenspiel: Chemical Reaction
274. D. Braun, W. Brendlein, Makromol. Chem. Engineering, John Wiley & Sons, New York
167 (1973) 203. 1972.
275. A. Matsumoto, T. Aso, S. Tanaka, M. Oiwa, J. 299. G. F. Froment, K. B. Bischoff: Chemical
Polym. Sci., Polym. Chem. Ed. 11 (1973) 2357. Reactor Analysis and Design, John
276. K. Dusek, J. Spevacek, Polymer 21 (1980)
Wiley & Sons, New York 1979.
750.
120 Polymerization Processes

300. J. M. Smith: Chemical Engineering Kinetics, 326. G. Odian: Principles of Polymerization,


McGraw-Hill, New York 1981. Wiley-Interscience, New York 1981, p. 536.
301. J. A. Biesenberger, D. H. Sebastian: Principles 327. P. H. Hermans, D. Heikens, P. F. van Velden, J.
of Polymerization Engineering, Polym. Sci. 30 (1958) 81.
Wiley-Interscience, New York 1983. 328. J. M. Andrews, F. R. Jones, J. A. Semlyen,
302. S. K. Gupta, A. Kumar: Reaction Engineering Polymer 15 (1974) 420.
of Step Growth Polymerization, Plenum Press, 329. J. A. Semlyen, Adv. Polym. Sci. 21 (1976) 41.
New York 1987. 330. S. K. Gupta, A. Kumar; Reaction Engineering
303. P. J. Flory: Principles of Polymer Chemistry, of Step-Growth Polymerization, Plenum Press,
Cornell University Press, Ithaca, N.Y. 1953, New York 1987, chap. 7.
chap. 8. 331. S. Mochizuki, N. Ito, Chem. Eng. Sci. 28
304. H. Kilkson, Ind. Eng. Chem., Fundam. 2 (1973) 1139.
(1964) 281. 332. S. Mochizuki, N. Ito, Chem. Eng. Sci. 33
305. Z. Tadmor, J. A. Biesenberger, Ind. Eng. (1978) 1401.
Chem., Fundam. 5 (1966) 336. 333. S. K. Gupta, C. D. Naik, P. Tandon, A. Kumar,
306. Kao Corporation, JP-Kokai 27 0704, 1990. J. Appl. Polym. Sci. 26 (1981) 2153.
307. H. Tobita, Y. Ohtani, Polymer 33 (1992) 801.
334. Ullmann, 4th ed. 19 118.
308. J. G. Watterson, J. W. Stafford, J. Macromol.
335. K. Tai, Y. Arai, T. Tagawa, J. Appl. Polym. Sci.
Sci., Chem. A 5 (1971) 679.
27 (1982) 731.
309. J. W. Stafford, J. Polym. Sci., Polym. Chem.
336. D. B. Jacobs, J. Zimmerman in C. E.
Ed. 19 (1981) 3219.
310. H.-G. Elias, J. Macromol. Sci., Chem. A 12 Schildknecht, I. Skeist (eds.): Polymerization
(1978) 183. Processes, Wiley-Interscience, New York
311. A. Kumar, R. K. Agarwal, S. Gupta, J. Appl. 1977, p. 424.
Polym. Sci. 27 (1982) 1759. 337. N. Ogata, Makromol. Chem. 42 (1960) 52.
312. O. Saito, J. Phys. Soc. Jpn. 13 (1958) 198. 338. N. Ogata, Makromol. Chem. 43 (1961) 117.
313. O. Saito in M. Dole (ed.): The Radiation 339. Y. Murakami: Fundamental Principles and
Chemistry of Macromolecules vol. 1, Analysis of Polymerization Reactors,
Academic Press, New York 1972, p. 223. Baifukan, Tokyo 1976, p. 127, in Japanese.
314. L. J. Lee, Rubber Chem. Technol. 53 (1980) 340. A. Kumar, S. Kuruville, A. R. Raman, S. K.
542. Gupta, Polymer 22 (1981) 387.
315. L. T. Manzione: “Reaction Injection Molding,” 341. A. Kumar, R. K. Agarwal, S. K. Gupta, J. Appl.
in Encyclopedia of Polymer Science and Polym. Sci. 27 (1982) 1759.
Engineering, vol. 14, Wiley-Interscience, New 342. M. Katz in C. E. Schildknecht, I. Skeist (eds.):
York 1988, p. 72. Polymerization Processes, Wiley-Interscience,
316. J. Zimmerman, J. Polym. Sci. B 2 (1964) 955. New York 1977, p. 468.
317. F. C. Chen, R. G. Griskey, G. H. Beyer, AIChE 343. S. K. Gupta, A. Kumar: Reaction Engineering
J. 15 (1969) 680. of Step-Growth Polymerization, Plenum Press,
318. T. M. Chang, Polym. Eng. Sci. 10 (1970) 364. New York 1987, chap. 8.
319. L. H. Buxbaum, J. Appl. Polym. Sci., Appl. 344. M. J. Barandiaran, J. M. Asua, Polymer 31
Polym. Symp. 35 (1979) 59. (1990) 1347.
320. M. Thakur: “Solid-State Polymerization,” 345. M. J. Barandiaran, J. M. Asua, Polymer 31
Encyclopedia of Polymer Science and (1990) 1352.
Technology, vol. 15, Wiley-Interscience, New 346. J. W. Ault, D. A. Mellichamp, Chem. Eng. Sci.
York 1989, p.362. 27 (1972) 2219.
321. H. K. Reimschessel, Macromol. Rev. 12 347. K. Ravindranath, R. A. Mashelker, J. Appl.
(1977) 65. Polym. Sci. 26 (1981) 3179.
322. K. Tai, T. Tagawa, Ind. Eng. Chem., Prod. Res. 348. A. Kumar, V. K. Sukthankar, C. P. Vaz, S. K.
Dev. 22 (1983) 192. Gupta, Polym. Eng. Sci. 24 (1984) 185.
323. J. Sebenda, Prog. Polym. Sci. 6 (1978) 123.
349. K. Ravindranath, R. A. Mashelker, J. Appl.
324. S. K. Gupta, A. Kumar, J. Macromol. Sci., Rev.
Polym. Sci. 27 (1982) 471.
Macromol. Chem. Phys. C 26 (1986) 183.
325. R. J. Welgos: “Polyamide, Plastics,” 350. K. Ravindranath, R. A. Mashelker, Chem. Eng.
Encyclopedia of Polymer Science and Sci. 41 (1986) 2969.
Engineering, vol. 11, Wiley-Interscience, New
York 1988, p. 445.
Polymerization Processes 121

351. I. Goodman: “Polyesters,” Encyclopedia of 372. A. Husain, A. E. Hamielec, J. Applied Polym.


Polymer Science and Technology, vol. 12, Sci. 22 (1978) 1207.
Wiley-Interscience, New York 1988, p. 1. 373. A. E. Hamielec, J. F. MacGregor, S. Webb, T.
352. K. Ravindranath, R. A. Mashelkar, Polym. Spychaj in K. H. Reichert, W. Geiseler (eds.):
Eng. Sci. 22 (1982) 610. Polymer Reaction Engineering,
353. T. Yamada, Y. Imamura, O. Makimura, Polym. Hüthig & Wepf Verlag, New York 1986, p. 185.
Eng. Sci. 25 (1985) 788. 374. G. Z. Wu, L. A. Denton, R. L. Laurence,
354. T. Yamada, Y. Imamura, O. Makimura, Polym. Polym. Eng. Sci. 22 (1982) 1.
Eng. Sci. 26 (1986) 708. 375. T. Rintelen, K. Riederle, K. Kirchner in K. H.
355. A. E. Hamielec, J. F. MacGregor, A. Penlidis: Reichert, W. Geiseler (eds.): Polymer Reaction
“Multicomponent Free-Radical Engineering, Hanser Publishers, New York
Polymerization in Batch, Semi-Batch and 1983, p. 269. N. Khac Tien, E. Flaschel, A.
Continuous Reactors,” Makromol. Chem., Renken in K. H. Reichert, W. Geiseler (eds.):
Macromol. Symp. 10/11 (1987) 521. Polymer Reaction Engineering, Hanser
356. H. Tobita, A. E. Hamielec: “The Kinetics of Publishers, New York 1983, p. 175.
Free-Radical Copolymerization–The 376. H. K. Fauske, J. C. Leung, Chem. Eng. Progr.
Pseudo-Kinetic Rate Constant Method,” 81 (1985) 39.
Polymer 32 (1991) 2641. 377. J. C. Leung, H. K. Fauske, Thermochim. Acta
357. A. E. Hamielec, J. F. MacGregor: “Modelling 104 (1986) 13.
Copolymerizations – Control of Chain 378. O. Chiantore, A. E. Hamielec, Polymer 26
Microstructure, Long Chain Branching, (1985) 608.
379. J. Braks, Ph. D. Thesis, University of
Cross-linking and Molecular Weight
Waterloo, Canada 1973.
Distributions,” in K. H. Reichert, W. Geiseler
380. S. T. Balke, A. E. Hamielec, J. Applied Polym.
(eds.): Polymer Reaction Engineering, Hanser
Sci. 17 (1973) 905.
Publishers, New York 1983, p. 21. 381. J. Cardenas, K. F. O.’Driscoll, J. Polym. Sci.,
358. S. Zhu, A. E. Hamielec, Macromolecules 22 Chem. Ed. 14 (1976) 883.
(1989) 3093. 382. J. Cardenas, K. F. O’Driscoll, J. Polym. Sci.,
359. H. M. J. Boots, J. Polym. Sci., Polym. Phys. Ed. Chem. Ed. 15 (1977) 1883, 2097.
20 (1982) 1695. 383. F. L. Marten, A. E. Hamielec, ACS Symp. Ser.
360. S. Zhu, Y. Tian, A. E. Hamielec, D. R. Eaton, 104 (1979) 43.
Polymer 31 (1990) 154. 384. M. Stickler, D. Panke, A. E. Hamielec, J.
361. D. J. Coyle, T. J. Tulig, M. Tirrell, Ind. Eng. Polym. Sci., Chem. Ed. 22 (1984) 2243.
Chem. Fundam. 24 (1985) 343. 385. W. Y. Chiu, G. M. Carratt, D. S. Soong,
362. W. H. Stockmayer, J. Chem. Phys. 13 (1945) Macromolecules 16 (1983) 348.
199. 386. S. K. Soh, D. C. Sundberg, J. Polym. Sci.,
363. H. Tobita, A. E. Hamielec, Makromol. Chem., Chem. Ed. 20 (1982) 1299, 1315, 1331, 1345.
Macromol. Symp. 20/21 (1988) 501. 387. K. Ito, Polym. J. (Tokyo) 12 (1980) 499.
364. H. Tobita, A. E. Hamielec, Macromolecules 388. H. M. J. Boots, J. Polym. Sci., Polym. Phys. Ed.
22 (1989) 3105. 20 (1982) 1695.
365. H. Tobita, A. E. Hamielec, Makromol. Chem., 389. T. J. Tulig, M. V. Tirrell, Macromolecules 14
Macromol. Symp. 35/36 (1990) 193. (1981) 1501.
366. W. A. Pryor: Free Radicals, McGraw-Hill, 390. T. J. Tulig, M. V. Tirrell, Macromolecules 15
New York 1966. (1982) 459.
367. K. Kirchner, K. Riederle, Angew. Makromol. 391. B. M. Louie, G. M. Carratt, D. S. Soong, J.
Chem. 111 (1983) 1. Appl. Polym. Sci. 30 (1985) 3985.
368. Y. K. Chong, E. Rizzardo, D. H. Solomon, J. 392. D. Achillias, C. Kiparissides, J. Applied
Am. Chem. Soc. 105 (1983) 7761. Polym. Sci. 35 (1990) 1303.
369. M. Tirrell, T. J. Tulig, in K. H. Reichert, W. 393. M. E. Adams et al., Makromol. Chem.,
Geiseler (eds.): Polymer Reaction Engineering, Macromol. Symp. 35/36 (1990) 1.
394. D. Panke, Makromol. Chem., Rapid Commun.
Hanser Publishers, New York 1983, p. 247.
7 (1986) 171.
370. C. T. Liao, D. C. Sundberg, Polymer 27 (1986)
395. J. C. Bevington, Trans. Faraday Soc. 51
265.
(1955) 1392.
371. A. W. Hui, A. E. Hamielec, J. Applied Polym.
396. F. DeSchrijver, G. Smets, J. Polym. Sci., Part
Sci. 16 (1972) 749.
A-1 4 (1966) 2201.
122 Polymerization Processes

397. J. H. Duerksen, A. E. Hamielec, J. Polym. Sci. 421. K. M. Jones, D. Bhattacharya, J. L. Brash,


C 25 (1968) 155. A. E. Hamielec, Polymer 27 (1986) 602.
398. A. W. Hui, A. E. Hamielec, J. Polym. Sci. C 25 422. D. Bhattacharya, A. E. Hamielec, Polymer 27
(1968) 167. (1986) 611.
399. M. J. Ballard et al., Macromolecules 19 (1986) 423. I. M. Yaraskavitch, J. L. Brash, A. E.
1303. Hamielec, Polymer 28 (1987) 489.
400. M. Stickler, Makromol. Chem. 184 (1983) 424. M. A. Dubé, A. Penlidis, K. F. O’Driscoll,
2563. Chem. Eng. Sci. 45 (1990) 2785.
401. J. G. Kloosterboer, Adv. Polym. Sci. 84 (1988) 425. K. F. O’Driscoll, J. Huang, Eur. Polym. J. 26
1. (1990) 643.
402. J. G. Kloosterboer, G. F. C. Lijten, H. M. J. 426. A. E. Hamielec, A. C. Ouano, L. L. Nebenzahl,
Boots, Makromol. Chem., Macromol. Symp. J. Liq. Chromatogr. 1 (1978) 527. A. E.
24 (1989) 223. Hamielec, H. Meyer: “Online Molecular
403. P. D. Armitage et al., Polymer 29 (1988) 2221. Weight and Long Chain Branching
404. S. Zhu, A. E. Hamielec: “Heat Effects for Measurements using SEC and Low-Angle
Free-Radical Polymerization in Glass Laser Light Scattering,” Developments in
Ampoule Reactors,” Polymer 32 (1991) 3021. Polymer Characterization–5, Elsevier Applied
405. A. E. Hamielec, J. F. MacGregor, A. Penlidis: Science Publishers, London 1986, p. 95 – 130.
“Copolymerization Kinetics,” Comprehensive M. Styring, J. Armonas, A. E. Hamielec,
Polymer Science, vol. 3, Pergamon Press, Polym. Mat. Sci. Eng. 54 (1986) 88. M.
Oxford 1989, p. 17 – 32. Styring, J. Armonas, A. E. Hamielec, J. Liq.
406. D. A. Tirrell: “Copolymer Composition,” Chromatogr. 10 (1987) 783. Ullmann, 4th.
Comprehensive Polymer Science, vol. 3, ed. 19, 107 – 165. H. Bieringer, K. Flatan, D.
Pergamon Press, Oxford 1989, p. 195 – 206. Reese, Angew. Makromol Chem. 123/124
407. T. Fukuda, Y. D. Ma, H. Inagaki, (1984) 307.
Macromolecules 18 (1985) 17. 427. Dow, US 2 530 409, 1948, US 2 727 884, 1953,
408. Y. D. Ma, T. Fukuda, H. Inagaki, US 2 849 430, 1955. Distrene, Br. Plast. 30
Macromolecules 18 (1985) 267. (1957) no. 1, 26 – 27.
409. T. Fukuda, Y. D. Ma, H. Inagaki, Makromol. 428. H. Ritter, Chem.-Ing.-Tech. 41 (1969)
Chem., Rapid Commun. 8 (1987) 495. 419 – 426.
410. H. Tanaka, K. Sasai, T. Sato, T. Ota, 429. R. Friedrich, M. Eisenstein, Chem.-Ing.-Tech.
Macromolecules 21 (1988) 3534. 50 (1978) 466 – 467.
411. O. F. Olaj, I. Bitai, G. Gleixner, Makromol. 430. K.-M. Hess, Chem.-Ing.-Tech. 51 (1979) 245.
Chem. 186 (1985) 2569. 431. F. Widmer, Adv. Chem. Ser. 128 (1973)
412. O. F. Olaj, I. Bitai, F. Hinkelmann, Makromol. 51 – 67.
Chem. 188 (1987) 1689. 432. G. Beckmann, E. F. Engel, Chem.-Ing.-Tech.
413. O. F. Olaj, P. Kremminger, I. Schnoll-Bitai, 38 (1966) 1025 – 1031.
Makromol. Chem., Rapid Comm. 9 (1988) 433. G. Daumiller, Chem.-Ing.-Tech. 40 (1968)
771. 673 – 682.
414. H. K. Mahabadi, K. F. O’Driscoll, J.
434. R. L. Scott, J. Chem. Phys. 17 (1949)
Macromol. Sci. Chem. A 11 (1977) 967.
279 – 284.
415. T. P. Davis, K. F. O’Driscoll, M. C. Piton, M. A.
435. P. J. Flory: Principles of Polymer Chemistry,
Winnik, Macromolecules 23 (1990) 2119.
Cornell University Press, Ithaca, N.Y. 1953,
416. J.-F. Kuo, C.-Y. Chen, C.-W. Chen, T.-C. Pan,
p. 554 ff.
Polym. Eng. Sci. 24 (1984) 22.
436. G. E. Molau, J. Polym. Sci. Part A 3 (1965)
417. G. Storti et al., Makromol. Chem., Macromol.
1267 – 1278, 4235 – 4242.
Symp. 35/36 (1990) 213.
437. G. E. Molau, J. Polym. Sci. Polym. Lett. Ed. 3
418. T. O. Broadhead, A. E. Hamielec, J. F.
(1965) 1007 – 1015.
MacGregor, Makromol. Chem., Suppl. 10/11
438. G. E. Molau, H. Keskkula, J. Polym. Sci.
(1985) 105.
Polym. Chem. Ed. 4 (1966) 1595 – 1607.
419. J. Kanetakis, F. Y. Wong, A. E. Hamielec, J. F.
439. G. E. Molau, W. M. Wittbrodt, V. E. Meyer, J.
MacGregor, Chem. Eng. Commun. 35 (1985)
Appl. Polym. Sci. 13 (1969) 2735 – 2736. G. E.
123.
420. L. H. Garcia, M. G. Lord, J. F. MacGregor, Molau, Kolloid Z. Z. Polym. 238 (1970)
A. E. Hamielec, Polymer 27 (1986) 602. 493 – 498.
Polymerization Processes 123

440. Dow, US 2 694 692, 1954. 468. J. Ugelstad, J. Macromol. Sci. Chem. A 11
441. J. L. Amos, Polym. Eng. Sci. 14 (1974) 1 – 11. (1977) 1281 – 1305.
442. B. W. Bender, J. Appl. Polym. Sci. 9 (1965) 469. J. Jaeger, Chem. Ztg. 90 (1966) 70 – 73.
2887 – 2894. 470. J.-C. Thomas, SPE J. 23 (1967) no. 10,
443. E. R. Wagner, L. M. Robeson, Rubber Chem. 61 – 65.
Technol. 43 (1970) 1129 – 1137. 471. J.-C. Thomas, Hydrocarbon Process. 47
444. T. O. Craig, J. Polym. Sci. Polym. Chem. Ed. (1968) no. 11, 192 – 196.
12 (1974) 2105 – 2109. 472. J. Chatelain, Br. Polym. J. 5 (1973) 457 – 465.
445. H. Willersinn, Makromol. Chem. 101 (1967) 473. Union Carbide, BE 723 775, 1968.
296 – 316. 474. D. M. Rasmussen, Chem. Eng. 79 (1972)
446. G. E. Molau, W. M. Wittbrodt, no. 21, 104 – 105.
Macromolecules 1 (1968) 260 – 264. 475. H. L. Batleman, Plast. Eng. 31 (1975) no. 4,
447. G. F. Freeguard, M. Karmarkar, J. Appl. Polym. 73 – 75.
Sci. 15 (1971) 1649 – 1655. 476. Union Carbide, Hydrocarbon Process. 56
448. G. F. Freeguard, M. Karmarkar, J. Appl. Polym.
(1977) no. 11, 212.
Sci. 15 (1971) 1657 – 1663.
477. BASF, US 3 634 282, 1968; BASF, US
449. G. F. Freeguard, M. Karmarkar, J. Appl. Polym.
3 639 377, 1967; BASF, US 3 652 527, 1968.
Sci. 16 (1972) 69 – 82.
478. BASF, Oil Gas J. 68 (1970) no. 47, 64.
450. G. F. Freeguard, Polymer 13 (1972) 366 – 370.
451. G. F. Freeguard, Br. Polym. J. 6 (1974) 479. K. Wisseroth, Angew. Makromol. Chem. 8
205 – 228. (1969) 41 – 60.
452. J. P. Fischer, Angew. Makromol. Chem. 33 480. K. Wisseroth, Kolloid Z. Z. Polym. 241 (1970)
(1973) 35 – 74. 943 – 954.
453. H. Cherdron, Chimia 28 (1974) 553 – 560. 481. K. Wisseroth, Chem. Ztg. 101 (1977)
454. A. Echte, Angew. Makromol. Chem. 58/59 271 – 284.
(1977) 175 – 198. 482. J. C. Davis, Chem. Eng. 85 (1978) no. 1,
455. M. L. Huggins: Chemistry of High Polymers, 25 – 27.
Wiley, New York 1958, p. 38 – 51. 483. B. Dietrich, Appl. Polym. Symp. 26 (1975)
456. H. Gerrens, W. Fink, E. Köhnlein, J. Polym. 1 – 11.
Sci. Polym. Symp. 16 (1967) 2781 – 2793. 484. H. Kreuter, B. Dietrich, Chem. Eng. 81 (1974)
457. A. R. Berens, Angew. Makromol. Chem. 47 no. 16, 62 – 63.
(1975) 97 – 110. 485. Hoechst, Hydrocarbon Process. 56 (1977)
458. G. Talamini, J. Polym. Sci. Polym. Phys. Ed. 4 no. 11, 208. A. Kageyama, A. Hirotani,
(1966) 535 – 537. Hydrocarbon Process. 51 (1972) no. 11,
459. A. Crosato-Arnaldi, P. Gasparini, G. Talamini, 97 – 98. E. Susa, Hydrocarbon Process. 51
Makromol. Chem. 117 (1968) 140 – 152. J. (1972) no. 7, 115 – 116. Montecatini Edison,
Swoboda, J.-C. Heilig, Kunststoff-Symposium Hydrocarbon Process. 50 (1971) no. 11, 195.
Moskau, 11 – 14 Oct, 1977. Snam Progetti, Hydrocarbon Process. 50
460. A. F. Hauss, J. Polym. Sci. Polym. Symp. 33 (1971) no. 11, 197.
(1971) 1 – 12. 486. Hoechst, Hydrocarbon Process. 56 (1977)
461. A. H. Abdel-Alim, A. E. Hamielec, J. Appl. no. 11, 213. Mitsui – Montedison,
Polym. Sci. 16 (1972) 1093 – 1101. Hydrocarbon Process. 56 (1977) no. 11, 214.
462. C. Hoheisel, Angew. Makromol. Chem. 34 487. F. S. Dainton, P. H. Seaman, D. G. L. James,
(1973) 19 – 33. R. S. Eaton, J. Polym. Sci. 34 (1959)
463. O. F. Olaj, Angew. Makromol. Chem. 47 209 – 228.
(1975) 1 – 14. 488. Hoechst, Hydrocarbon Process. 54 (1975)
464. O. F. Olaj, J. Makromol. Sci. C A 11 (1977) no. 11, 175. Ullmann, 3rd ed., 14,
1307 – 1317. pp. 276 – 280.
465. O. F. Olaj, J. W. Breitenbach, K. J. Parth, N. 489. K. Thomas, India Rubber World 130 (1954)
Philippovich, J. Macromol. Sci. Chem. A 11 203.
(1977) 1319 – 1331. 490. BASF, DT 961 309, 1954. BASF DT
466. J. Ugelstad, H. Lervik, B. Gardinovacki, E. 1 003 446, 1955.
Sund, Pure Appl. Chem. 25 (1971) 121 – 152. 491. H. Gerrens, H. Ohlinger, R. Fricker,
467. J. Ugelstad, H. Fløgstad, T. Hertzberg, E.
Makromol. Chem. 87 (1965) 209 – 227.
Sund, Makromol. Chem. 164 (1973)
171 – 181.
124 Polymerization Processes

492. H. Gerrens, D. J. Stein, Makromol. Chem. 87 514. H. J. Reinhardt, R. Thiele, Plaste Kautsch. 19
(1965) 228 – 247. (1972) 648 – 654.
493. K. E. J. Barrett (ed.): Dispersion 515. E. Farber, M. Koral, Polym. Eng. Sci. 8 (1968)
Polymerization in Organic Media, Wiley, 11 – 18.
London 1975. 516. Dow, BE 677 224, BE 677 173, 1965.
494. W. Heller, T. L. Pugh, J. Chem. Phys. 22 517. Br. Petroleum, US 3 726 848, 1970.
(1954) 1778. W. Heller, Pure Appl. Chem. 12 518. E. Trommsdorff, C. E. Schildknecht:
(1966) 249 – 274. “Polymerizations in Suspension,” in C. E.
495. D. J. Walbridge: The Design and Synthesis of Schildknecht (ed.): Polymer Processes,
Dispersants for Dispersion Polymerization in chap. 3, Interscience, New York 1956,
Organic Media, chap. 3 in [493], p. 45 – 114. p. 69 – 109. M. Munzer, E. Trommsdorff:
496. E. W. Fischer, Kolloid Z. 160 (1958) “Polymerizations in Suspension,” in C. E.
120 – 141. Schildknecht, I. Skeist (eds.): Polymerization
497. R. H. Ottewill, T. Walker, Kolloid Z. Z. Polym. Processes, chap. 5, Wiley, New York 1977,
227 (1968) 108 – 116. p. 106 – 142.
498. D. H. Napper, Ind. Eng. Chem. Prod. Res. Dev. 519. E. Farber: “Suspension Polymerization,” in
9 (1970) 467 – 477. H. F. Mark, N. G. Gaylord, N. M. Bikales
499. K. Jäckel, Kolloid-Z. 197 (1964) 143 – 151. (eds.): Encyclopedia of Polymer Science and
500. R. H. Ottewill: “Effect of Nonionic Surfactants Technology, vol. 13, Interscience, New York
on the Stability of Dispersion,” in M. J. Schick 1970, p. 552 –571.
(ed.): Nonionic Surfactants, vol. 1, chap. 19, 520. J. H. Rushton, E. W. Costich, H. J. Everett,
Dekker, New York 1967, p. 627 – 682. Chem. Eng. Prog. 46 (1950) 395 – 404,
501. D. H. Napper, Trans. Faraday Soc. 64 (1968) 467 – 476. F. Wolf, B. Hoffbauer, S. Eckert,
1701 – 1711. D. H. Napper, J. Colloid Interface Plaste Kautsch. 19 (1972) 26 – 28.
Sci. 29 (1969) 168 – 170. D. H. Napper, J. 521. F. H. Winslow, W. Matreyek, Ind. Eng. Chem.
Colloid Interface Sci. 32 (1970) 106 – 114. 43 (1951) 1108 – 1112.
502. D. H. Napper, A. Netschey, J. Colloid 522. BASF, DT 810 812, 1949.
Interface Sci. 37 (1971) 528 – 535. 523. Röhm & Haas, DT 735 284, 1935; DT
503. D. H. Napper, R. J. Hunter in M. Kerker (ed.): 747 596, 1935; US 2 171 765, 1934.
Surface Chemistry and Colloids, Butterworth, 524. R. Shinnar, J. M. Church, Ind. Eng. Chem. 52
London 1972, p. 279 – 292. D. H. Napper: (1960) 253 – 256.
Polymeric Stabilization of Calloidal 525. J. M. Church, R. Shinnar, Ind. Eng. Chem. 53
Dispersions, Academic Press, London 1983. (1961) 479 – 484.
504. A. Guyot: “Precipitation Polymerization,” in 526. R. Shinnar, J. Fluid Mech. 10 (1961)
Comprehensive Polymer Science, Pergamon 259 – 275.
Press, Oxford 1989, p. 261. 527. A. N. Kolmogoroff, C. R. Acad. Sci USSR 30
505. S. U. Pickering, J. Chem. Soc. 91 (1907) (1940) 301, 538.
2001 –2021. S. U. Pickering, Kolloid Z. 7 528. A. N. Kolmogoroff, J. Fluid Mech. 13 (1962)
(1910) 11 – 16. 82 –85.
529. R. H. Kraichnan, J. Fluid Mech. 62 (1974)
506. W. P. Hohenstein, H. Mark, J. Polym. Sci. 1
305 –330.
(1946) 127 – 145.
530. L. A. Cutter, Ph. D. Thesis, Columbia
507. H. Wenning, Makromol. Chem. 20 (1956)
University, New York 1960. H. G.
196 – 213.
Schwartzberg, R. E. Treybal, Ind. Eng. Chem.
508. E. H. Merz, J. Appl. Polym. Sci. 3 (1960) 374.
Fundam. 7 (1968) 1 – 6.
509. H. Wenning, Kolloid Z. Z. Polym. 182 (1962)
531. T. Vermeulen, G. M. Williams, G. E. Langlois,
60 –75.
Chem. Eng. Prog. 51 (1955) no. 2, 85 F – 94 F.
510. H. Wenning, Kunstst. Plast. 5 (1958)
532. F. B. Sprow, Chem. Eng. Sci. 22 (1967)
328 – 340.
435 – 442.
511. R. N. Haward, J. Polym. Sci. 4 (1949)
533. H. T. Chen, S. Middleman, AIChE J. 13
273 – 287.
(1967) 989 – 995.
512. S. Kaichi, Bull. Chem. Soc. Jpn. 29 (1956)
534. Y. Mlynek, W. Resnick, AIChE J. 18 (1972)
241 –245.
122 – 127.
513. E. Trommsdorff, Makromol. Chem. 13 (1954) 535. W. A. Rodger, V. G. Trice, J. H. Rushton,
76 –89. Chem. Eng. Prog. 52 (1956) 515 – 520.
Polymerization Processes 125

536. J. O. Hinze, AIChE J. 1 (1955) 289 – 295. 566. M. A. Villalobos, A. E. Hamielec, P. E. Wood,
537. F. B. Sprow, AIChE J. 13 (1967) 995 – 998. J. Appl. Polym. Sci. 42 (1991) 629.
538. H. D. Schindler, R. E. Treybal, AIChE J. 14 567. Edison, US 3 228 919, 1960.
(1968) 790 – 798. 568. Chem. Werke Hüls, DE-AS 2 528 950, 1975.
539. C. A. Coulaloglou, L. L. Tavlarides, AIChE J. 569. Firestone, US 4 000 355, 1975.
22 (1976) 289 – 297. 570. Hoechst, DE-AS 1 021 165, 1956.
571. M. Macoveanu, K. H. Reichert, Angew.
540. J. H. Vanderveen, M. S. Thesis, Univ. of
Makromol. Chem. 44 (1975) 141 – 149.
California, Berkeley UCRL 8 733, Berkeley
572. T. Y. Xie, A. E. Hamielec, P. E. Wood, D. R.
1960.
Woods, J. Vinyl Technol. 13 (1991) 2.
541. M. A. Zeitlin, L. L. Tavlarides, AIChE J. 18 573. C. Kiparissides, Makromol. Chem., Macromol.
(1972) 1268 – 1271. Symp. 35/36 (1990) 171.
542. M. A. Zeitlin, L. L. Tavlarides, Can. J. Chem. 574. Bayer, DT 1 045 102, 1957; DT 1 113 570,
Eng. 50 (1972) 207 – 215. 1957.
543. M. A. Zeitlin, L. L. Tavlarides, Proc. 5th 575. D. Hunkeler, A. E. Hamielec, W. Baade,
European/2nd Intern. Symp. on Chem. Polymer 30 (1989) 127.
Reaction Eng., Elsevier, Amsterdam 1972, 576. P. Trijasson, P. Pith, M. Lambla, Makromol.
p. B 1, 1 – 12. Chem. Macromol. Symp. 35/36 (1990) 141.
544. H. Hopff, E. Lutz, Kunstst. Plast. 5 (1958) 577. C. Graillot, C. Pichot, A. Guyot, M. S.
341 –344. El-Aasser, J. Polym. Sci., Polym. Chem. Ed.
545. H. Hopff, H. Lüssi, P. Gerspacher, Makromol. 24 (1986) 427.
Chem. 78 (1964) 24 – 36, 37 – 46. 578. W. Baade, K. H. Reichert, Makromol. Chem.,
546. H. Hopff, H. Lüssi, E. Hammer, Makromol. Rapid Commun. 7 (1986) 235.
Chem. 82 (1965) 175 – 183, 184 – 189. 579. F. Candau, Y. S. Leong, R. M. Fitch, J. Polym.
547. H. Hopff, H. Lüssi, E. Hammer, Makromol. Sci., Polym. Chem. Ed. 23 (1985) 193.
Chem. 84 (1965) 274 – 281, 282 – 285, 580. J. W. Vanderhof et al., J. Dispersion Sci.
286 – 289. Technol. 5 (1984) 323.
548. H. Hedden, Dissertation Univ. Stuttgart 1969. 581. Dow, DT 1 081 228, 1957.
582. Nalco, US 3 767 629, 1971.
549. H. Hedden, Chem.-Ing.-Tech. 42 (1970) 583. Röhm GmbH, DE-OS 2 009 218, 1970.
457 – 462. 584. G. Beckmann, Adv. Chem. Ser. 128 (1973)
550. F. Langner, H. U. Moritz, K. H. Reichert, 37 – 50.
Chem. Ing. Tech. 51 (1979) 746. 585. G. Beckmann, Chem. Technol. 3 (1973)
551. F. Wolf, S. Eckert, Plaste Kautsch. 18 (1971) 304 – 310.
580 –583. 586. G. Beckmann, Chemie-Techn. 5 (1976) no. 4,
552. H. Wenning, Kunstst. Plast. 5 (1958) 135 – 139.
328 – 340. 587. L. F. Albright, C. G. Bild, Chem. Eng. 82
553. B. A. Hills, Br. Chem. Eng. 6 (1961) no. 2, (1975) no. 19, 121 – 128.
104 –106. 588. B. Terwiesch, Hydrocarbon Process. 55
554. G. Beckmann, Chem. Ing. Tech. 36 (1964) (1976) no. 11, 117 – 121.
169 – 174. 589. H. Fikentscher, Angew. Chem. 51 (1938) 433.
555. D. B. Scully, J. Appl. Polym. Sci. 20 (1976) 590. H. Gerrens, Fortschr. Hochpolym. Forsch. 1
2299 – 2303. (1959) 234 – 328.
556. G. Mino, J. Polym. Sci. 22 (1956) 369 – 383. 591. E. W. Duck: “Emulsion Polymerization,” in
557. J. Brandrup, Faserforsch. Textiltechn. 12 H. F. Mark, N. G. Gaylord, N. M. Bikales
(1961) 133 – 140, 208 – 213. (eds.): Encyclopedia of Polymer Science and
558. Monsanto, US 2 862 906, 2 862 907, 1957. Technology, vol. 5, Interscience, New York
559. Dow, US 3 267 178, 1962. 1966, p. 801 – 859.
592. B. M. E. van der Hoff: “Emulsion
560. BASF, DT 845 264, 1950.
Polymerization,” in K. Shinoda (ed.): Solvent
561. BASF, DT 941 389, 1951.
Properties of Surfactant Solutions, Marcel
562. BASF, DT 951 299, 1953; 963 955, 1953.
Dekker, New York 1967, p. 285 – 340.
563. Koppers, GB 756 654, 1953.
593. J. W. Vanderhoff: “The Mechanism of
564. M. Villalobos: Foamed Polymer Course Notes,
Emulsion Polymerization,” in G. E. Ham (ed.):
McMaster University, Hamilton, Canada 1990.
Kinetics and Mechanisms of Polymerization,
565. M. Konno, K. Arai, S. Saito, J. Chem. Eng.
vol. 1, part II, Marcel Dekker, New York 1969,
Jpn. 15 (1982) 131.
p. 1 – 138.
126 Polymerization Processes

594. A. E. Alexander, D. H. Napper: “Emulsion 612. H. Fikentscher, H. Gerrens, H. Schuller,


Polymerization,” in A. D. Jenkins Angew. Chem. 72 (1960) 856 – 864.
(ed.): Progress in Polymer Sci., vol. 3, 613. H. Finkentscher, Kunststoffe 53 (1963)
Pergamon Press, Oxford 1971, p. 145 – 197. 734 – 740.
595. J. L. Gardon: “Mechanism of Emulsion 614. C. Heuck, Chem. Ztg. 94 (1970) 147 – 157.
Polymerization,” in J. K. Graven, R. W. Tess 615. W. D. Harkins, J. Chem. Phys. 13 (1945)
(eds.): Applied Polymer Science, Org. Coatings 381 – 382.
and Plastic Div., ACS, Washington, D. C. 616. W. D. Harkins, J. Chem. Phys. 14 (1946)
1975, p. 138 – 160. 47 – 48.
596. J. L. Gardon: “Emulsion Polymerization, 617. W. D. Harkins, J. Am. Chem. Soc. 69 (1947)
Theory,” in H. F. Mark, N. M. Bikales (eds.): 1428 – 1444.
Encyclopedia of Polymer Science and 618. W. D. Harkins, J. Polym. Sci. 5 (1950)
Technology, Supplement vol. 1, Interscience, 217 – 251.
New York 1976, p. 238 – 259. 619. B. M. E. van der Hoff, Adv. Chem. Ser. 34
597. J. Ugelstad, F. K. Hansen, Rubber Chem. (1962) 6 – 31.
Technol. 49 (1976) 536 – 609. 620. J. Gardon, J. Polym. Sci. Polym. Chem. Ed. 6
598. G. W. Poehlein: in R. Buscall, T. Corner, J. F. (1968) 2859 – 2879.
Stageman (eds.): Polymer Colloids, Elsevier 621. H. Gerrens, Z. Elektrochem. 60 (1956)
Applied Sci. Publishers, Barking 1985, 400 – 404.
p. 45 – 68. 622. H. Gerrens, Ber. Bunsenges. Phys. Chem. 67
599. A. Penlidis, J. F. MacGregor, A. E. Hamielec, (1963) 741 – 753.
AIChE J. 31 (1985) 881. 623. W. V. Smith, R. H. Ewart, J. Chem. Phys. 16
600. J. W. Vanderhoff, J. Polym. Sci., Polym. Symp. (1948) 592 – 599.
72 (1985) 161. 624. W. H. Stockmayer, J. Polym. Sci. 24 (1957)
601. B. W. Brooks: “Developments in Emulsion 314 –317.
Polymerization,” in K. H. Reichert, W. Geiseler 625. J. T. O’Toole, Appl. Polym. Sci. 9 (1965)
(eds.): Polymer Reaction Engineering, VCH 1291 – 1297.
Publishers, New York 1989, p. 3. 626. E. Jahnke, F. Emde: Funktionentafeln mit
602. D. H. Napper, R. G. Gilbert: “Polymerizations Formeln und Kurven, 2nd ed., B. G. Teubner,
in Emulsions,” Comprehensive Polymer Leipzig 1933.
Science, vol. 4. 627. M. Abramowitz, I. A. Stegun: Handbook of
603. K. W. Min, W. H. Ray, J. Macromol. Sci. Rev. Mathematical Functions, 10th ed., Dover,
Macromol. Chem. 11 (1974) no. 2, 177 – 255. New York 1972.
604. J. Polym. Sci. Polym. Symp. 27 (1969). 628. J. L. Gardon, J. Polym. Sci. Polym. Chem. Ed.
605. Br. Polym. J. 2 (1970) 1 – 66, 116 – 177. 6 (1968) a 623 – 641, b 643 – 664, c 665 – 685,
606. R. M. Fitch (ed.): Polymer Colloids, Plenum d 687 – 710, e 2853 – 2857.
629. J. Ugelstad, F. K. Hansen, K.
Press, New York 1971.
Herder-Kaggerud, Faserforsch. Textiltech. 28
607. J. Makromol. Sci. Chem. 7 (1973) 601 – 736.
(1977) 309 – 320.
608. ACS Symp. Ser. 24 (1976).
630. N. Friis, A. E. Hamielec, J. Polym. Sci. Polym.
609. “Emulsion Copolymerization Mechanisms and
Chem. Ed. 11 (1973) 3321 – 3325.
Processes; Relations between Colloid
631. M. R. Grancio, D. J. Williams, J. Polym. Sci.
Structure and Properties,” Die
Polym. Chem. Ed. 8 (1970) 2617 – 2629,
Makromolekulare Chemie, Macromolecular
2733 – 2745.
Chemistry and Physics, Suppl. 10/11, March
632. P. Keusch, J. Prince, D. J. Williams, J.
1985.
Macromol. Sci. Chem. 7 (1973) 623 – 646. P.
610. “Copolymerization and Copolymers in
Keusch, D. J. Williams, J. Polym. Sci. Polym.
Dispersed Media,” Die Makromolekulare
Chem. Ed. 11 (1973) 143 – 162.
Chemie, Macromolecular Chemistry and
633. P. Keusch, R. A. Graff, D. J. Williams,
Physics, Suppl. 35/36, May 1990.
Macromolecules 7 (1974) 304 – 310. D. J.
611. G. W. Poehlein, R. H. Ottewill, J. W. Goodwin
Williams, J. Polym. Sci. Polym. Chem. Ed. 11
(eds.): Science and Technology of Polymer
(1973) 301 – 303. D. J. Williams, J. Polym. Sci.
Colloids, vols. I and II, Nato ASI Series, Series
Polym. Chem. Ed. 12 (1974) 2123 – 2132.
E: Applied Sciences, no. 68, Martinus, Nijhoff
634. J. L. Gardon, J. Polym. Sci. Polym. Chem. Ed.
Publishers, 1983.
11 (1973) 241 – 251.
Polymerization Processes 127

635. J. L. Gardon, J. Polym. Sci. Polym. Chem. Ed. 660. M. K. Lindemann: “The Mechanism of Vinyl
12 (1974) 2133 – 2135. Acetate Polymerization,” in G. E. Ham (ed.): :
636. M. Morton, S. Kaizerman, M. W. Altier, J. Kinetics and Mechanisms of Polymerization,
Colloid Sci. 9 (1954) 300 – 312. vol. 1 part I, Marcel Dekker, New York 1967,
637. E. Vanzo, R. H. Marchessault, V. Stannet, J. p. 207 – 329, 259 – 290.
Colloid Sci. 20 (1965) 62 – 71. 661. R. Patsiga, M. Litt, V. Stannett, J. Phys. Chem.
638. D. Hummel, G. Ley, C. Schneider, Adv. Chem. 64 (1960) 801 – 804.
Ser. 34 (1962) 60 – 86. 662. M. Litt, R. Patsiga, V. Stannett, J. Polym. Sci.
639. G. V. Schulz, J. Romatowski, Makromol. Polym. Chem. Ed. 8 (1970) 3607 – 3649.
Chem. 85 (1965) 195 – 226. J. Romatowski, 663. N. Friis, L. Nyhagen, J. Appl. Polym. Sci. 17
G. V. Schulz, Makromol. Chem. 85 (1965) (1973) 2311-2327.
227 – 248. 664. M. Nomura et al., J. Chem. Eng. Jpn. 4 (1971)
640. J. Ugelstad, P. C. Mörk, J. O. Aasen, J. Polym. 160 – 166.
Sci. Polym. Chem. Ed. 5 (1967) 2281 – 2288. 665. A. G. Parts, D. E. Moore, J. G. Watterson,
641. J. Ugelstad, P. C. Mörk, P. Dahl, P. Rangnes, J. Makromol. Chem. 89 (1965) 156 – 164.
Polym. Sci. Polym. Symp. 27 (1969) 49 – 68. 666. M. Harada et al., J. Appl. Polym. Sci. 16
642. J. Ugelstad, P. C. Mörk, Br. Polym. J. 2 (1970) (1972) 811 – 833.
667. M. Nomura, M. Harada, W. Eguchi, S. Nagata,
31 – 39.
ACS Div. of Polymer Chemistry, Polymer
643. F. K. Hansen, J. Ugelstad, Makromol. Chem.
Preprints 16 (1975) no. 1, 217 – 222.
180 (1979) 2423 – 2434. Polym. Sci. C, Polym.
668. F. K. Hansen, J. Ugelstad, J. Polym. Sci.
Lett. 26 (1988) 385.
Polym. Chem. Ed. 16 (1978) 1953 – 1979.
644. B. W. Brooks, Colloid Polym. Sci. 265 (1987) 669. P. V. Danckwerts, Chem. Eng. Sci. 2 (1953)
58. 1 – 13.
645. B. S. Hawkett, D. H. Napper, R. G. Gilbert, J. 670. C. P. Roe, Ind. Eng. Chem. 60 (1968) 20 – 33.
Chem. Soc., Faraday Trans. 76 (1980) 1323. 671. R. M. Fitch, Off. Dig J. Paint Technol. Eng. 37
646. H. C. Lee and G. W. Poehlein, J. Dispersion (1965), no. 10, 32 – 48.
Sci. Technol. 5 (1984) 247. 672. R. M. Fitch, M. B. Prenosil, K. J. Sprick, J.
647. B. W. Brooks, B. O. Makanjuola, J. Chem. Polym. Sci. Polym. Symp. 27 (1969) 95 – 118.
Soc., Faraday Trans. 77 (1981) 2659. 673. R. M. Fitch, C. H. Tsai, J. Polym. Sci. Polym.
648. G. Lichti et al., J. Chem. Soc., Faraday Trans. Lett. Ed. 8 (1970) 703 – 710.
80 (1984) 2911. 674. R. M. Fitch, Br. Polym. J. 5 (1973) 467 – 483.
649. M. Nomura, M. Kubo, K. Fugita, J. Appl. 675. R. M. Fitch, Lih-Bin Shih, Progr. Colloid
Polym. Sci. 28 (1983) 2767. Polym. Sci. 56 (1975) 1 – 11.
650. M. Nomura, I. Horie, M. Kubo, K. Fujita, J. 676. J. W. Vanderhoff, J. F. Vitkuske, E. B.
Appl. Polym. Sci. 37 (1989) 1029. Bradford, T. Alfrey, J. Polym. Sci. 20 (1956)
651. W. V. Smith, J. Am. Chem. Soc. 70 (1948) 225 – 234.
2177 –2179; 71 (1949) 4077 – 4082. 677. N. Dezelić, J. J. Petres, G. J. Dezelić, Kolloid
652. M. Morton, P. P. Salatiello, H. Landfield, J. Z. Z. Polym. 242 (1970) 1142 – 1150. N.
Polym. Sci. 8 (1952) 111 – 121. Dezelić, J. J. Petres, G. J. Dezelić in J. E.
653. M. Morton, P. P. Salatiello, J. Polym. Sci. 8 Mulvaney (ed.): Macromolecular Syntheses,
(1952) 279 – 287. vol. 6, Wiley, New York 1977, p. 85 – 89.
654. I. M. Kolthoff et al.: Emulsion Polymerization, 678. A. R. Goodall, M. C. Wilkinson, J. Heran, J.
Interscience, New York 1955, p. 191 – 201. Colloid Interface Sci. 53 (1975) 327 – 331.
655. E. Bartholomé, H. Gerrens, R. Herbeck, H. M. 679. V. I. Yeliseyeva, A. V. Zuikov, ACS Symp. Ser.
Weitz, Z. Elektrochem. 60 (1956) 334 – 348. 24 (1976) 62 – 81.
656. H. Gerrens, E. Köhnlein, Z. Elektrochem. 64 680. M. V. Smoluchowski, Z. Phys. Chem. 92
(1960) 1199 – 1210. (1918) 129 – 168.
681. J. Ugelstad, M. S. El-Aasser, J. W. Vanderhoff,
657. Z. Maňyásek, A. Řežábek, J. Polym. Sci. 56
J. Polym. Sci. Polym. Lett. Ed. 11 (1973)
(1962) 47 – 56.
503 – 513.
658. K. P. Paoletti, F. W. Billmeyer, J. Polym. Sci.,
682. J. Ugelstad, H. Fløgstad, F. K. Hansen, T.
Part A 2 (1964) 2049 – 2062.
Ellingsen, J. Polym. Sci. Polym. Symp. 42
659. D. M. French, J. Polym. Sci. 32 (1958)
(1973) 473 – 485.
395 – 411.
683. J. Ugelstad, F. K. Hansen, S. Lange,
Makromol. Chem. 175 (1974) 507 – 521.
128 Polymerization Processes

684. A. R. M. Azad, J. Ugelstad, R. M. Fitch, F. K. 707. E. Giannetti, G. Storti, M. Morbidelli, J.


Hansen, ACS Symp. Ser. 24 (1976) 1 – 23. Polym. Sci., Part A: Chem. Ed. 26 (1988)
D. P. Durbin, M. S. El-Aasser, G. W. Poehlein, 1835, 2307.
J. W. Vanderhoff, J. Appl. Polym. Sci. 24 708. K. W. Min, W. H. Ray, J. Macromol. Sci. Rev.
(1979) 703 – 707. Macromol. Chem. C 11 (1974) 177.
685. H. Tang, P. L. Johnson, E. Gulari, Polymer 25 709. N. Friis, A. E. Hamielec, J. Appl. Polym. Sci.
(1984) 1357. 19 (1975) 97. H. Tobita, Macromolecules 25
686. G. I. Litvinenko, V. A. Kaminski, M. G. Slinko, (1992) 2671.
Khim. Promst. (Moscow) 8 (1984) 463. 710. T. O. Broadhead, A. E. Hamielec, J. F.
687. J. M. Asua, V. S. Rodriguez, C. A. Silebi, M. S. MacGregor, Makromol. Chem., Suppl. 10/11
El-Aasser, Makromol. Chem., Macromol. (1985) 105.
Symp. 35/36 (1990) 59. Pergamon Press, 711. X. Z. Kong, C. Pichot, J. Buillot, Eur. Polym. J.
Oxford 1989, p. 171. 24 (1988) 485.
688. Z. Song, G. W. Poehlein, J. Macromol. Sci. 712. J. Guillot in K. H. Reichert, W. Geiseler (eds.):
Chem. A 25 (1988) 403, 1587. Polymer Reaction Engineering,
689. Z. Song, G. W. Poehlein, J. Colloid Interface Hüthig & Wepf, Basel 1986, p. 147.
Sci. 128 (1989) 486, 501. 713. F. V. Loncar, M. S. El-Aasser, J. W. Vanderhoff,
690. Z. Song, G. W. Poehlein, Polym. Plast. Polym. Mater. Sci. Eng. 54 (1986) 453.
Technol. Eng. 29 (1990) 377. 714. M. H. Andrus, Polym. Mater. Sci. Eng. 54
691. M. Nomura, U. S. Satpathy, Y. Kouno, K. (1986) 603.
Fujita, J. Polym. Sci. 37 (1989) 1029. 715. J. W. Vanderhoff, NATO ASI Ser. E : 138
692. B. P. Huo et al., J. Appl. Polym. Sci. 35 (1988) (1987) 23.
2009. 716. V. Dimonie, M. S. El-Aasser, A. Klein, J. W.
693. A. Penlidis, J. F. MacGregor, A. E. Hamielec, Vanderhoff, J. Polym. Sci. Polym. Chem. Ed.
J. Appl. Polym. Sci. 35 (1988) 2023. 22 (1984) 2197.
694. Dispersionen synthetischer Hochpolymerer, 717. H. Schuller in K. H. Reichert, W. Geiseler
Springer, Berlin 1969, F. Hölscher: (eds.): Polymer Reaction Engineering,
“Eigenschaften, Herstellung und Prüfung,” Hüthig & Wepf, Basel 1986, p. 137.
part 1, H. Reinhard: “Anwendung,” part 2. 718. M. Okubo, A. Yamada, R. Matsumato, J.
695. H. Warson: The Applications of Synthetic Polym. Sci. Polym. Chem. Ed. 18 (1980) 3219.
Resin Emulsions, Ernest Benn, London 1972. 719. M. Okubo, Makromol. Chem., Macromol.
696. N. Friis, A. E. Hamielec, J. Appl. Polym. Sci. Symp. 35/36 (1990) 307.
19 (1975) 97. 720. A. Zosel, W. Heckelmann, G. Ley, W.
697. K. W. Min, W. H. Ray, J. Macromol. Sci. Rev. Mächtle, Makromol. Chem., Macromol. Symp.
Macromol. Chem. C 11 (1974) 177. 35/36 (1990) 423.
698. C. Kiparissides, S. R. Ponnuswamy, Chem. 721. D. Distler, G. Kanig, Colloid Polym. Sci. 256
Eng. Commun. 10 (1981) 283. (1978) 1052 – 1060.
699. E. Giannetti in K. H. Reichert, W. Geiseler 722. K. W. Min, H. I. Gostin, Ind. Eng. Chem. Prod.
(eds.): Polymer Reaction Engineering, VCH Res. Dev. 18 (1979) 272.
Publishers, Basel 1989, p. 343. 723. G. Lichti, R. G. Gilbert, D. H. Napper, J.
700. D. C. Sundberg, J. Appl. Polym. Sci. 23 (1979)
Polym. Sci. Polym. Chem. Ed. 21 (1983) 269.
2197.
724. M. Cociani et al. in K. H. Reichert, W. Geiseler
701. J. L. Gardon, J. Polym. Sci. A 6 (1968) 665.
(eds.): Polymer Reaction Engineering, VCH
702. S. Katz, R. Shinnar, G. M. Saidel, Adv. Chem.
Publishers, Basel 1989, p. 199.
Ser. 91 (1969) 145.
725. A. W. DeGraff, G. W. Poehlein, J. Polym. Sci.
703. K. W. Min, W. H. Ray, J. Appl. Polym. Sci. 22
A 2 9 (1971) 1955.
(1968) 89. K. W. Min, W. H. Ray, J. Macromol.
726. G. C. Lin, W. Y. Chin, L. C. Huang, J. Appl.
Sci. Rev. Macromol. Chem. C 11 (1974) 177.
Polym. Sci. 25 (1980) 567.
704. D. C. Sundberg, J. D. Eliassen: Polymer
727. P. J. Feeney, D. H. Napper, R. G. Gilbert,
Colloids, Plenum, New York 1971,
Macromolecules 17 (1984) 2520.
p. 121 – 137.
728. J. L. Guillaume, C. Pichot, A. Revillon,
705. C. C. Lin, W. Y. Chin, J. Appl. Polym. Sci. 23
Macromol. Chem. Suppl. 10/11 (1985) 69.
(1979) 2049.
706. G. Lichti, R. G. Gilbert, D. H. Napper, J. 729. A. S. Dunn, Macromol. Chem., Suppl. 10/11
Polym. Sci. Polym. Ed. 18 (1980) 1297. (1985) 1.
Polymerization Processes 129

730. J. Snuparek, Z. Kleckova, J. Appl. Polym. Sci. 755. V. I. Yeliseyeva et al., Polym. Sci. USSR (Engl.
29 (1984) 1. Transl.) 11 (1969) 1136 – 1144.
731. T. Hjertberg, E. M. Sorvik, J. Polym. Sci., Part 756. H. Gerrens, Kolloid Z. Z. Polym. 227 (1968)
A, Polym. Chem. Ed. 24 (1986) 1313. 92 – 107.
732. K. Tauer, I. Mueller, Chem. Abstracts 101 757. J. J. Krackeler, H. Naidus, J. Polym. Sci.
(1984) 7729 g. Polym. Symp. 27 (1969) 207 – 235.
733. B. Jacobi, Angew. Chem. 64 (1952) 539 – 543. 758. S. Jovanović, J. Romatowski, G. V. Schulz,
734. S. H. Maron et al., J. Colloid Sci. 9 (1954) Makromol. Chem. 85 (1965) 187 – 194.
89 – 103, 104 – 112, 263 – 268, 347 – 352, 759. J. E. Vandegaer, J. Appl. Polym. Sci. 9 (1965)
353 – 358, 382 – 384. 2929 – 2938.
735. W. Hübinger, K. H. Reichert in K. H. Reichert, 760. J. J. Owen, C. T. Steele, P. T. Parker, E. W. C.
W. Geiseler (eds.): Polymer Reaction Carrier, Ind. Eng. Chem. 39 (1947) 110 – 113.
Engineering, VCH Publishers, Basel 1989, 761. R. W. Laundrie, R. F. McCann, Ind. Eng.
p. 207. Chem. 41 (1949) 1568 – 1570.
736. A. S. Dunn: “Polymerization in Micelles and 762. P. Baumann, Chimia 12 (1958) 233 – 245.
Microemulsions,” in Comprehensive Polymer 763. W. Hofmann: Nitrilkautschuk, Berliner Union,
Science, Pergamon Press, Oxford 1989, p. 219. Stuttgart 1965.
737. F. Candau: “Polymerization in Inverse 764. Du Pont, US 2 384 277, 1939.
Microemulsions,” in Comprehensive Polymer 765. Du Pont, US 2 831 842, 1953.
Science, Pergamon Press, Oxford 1989, p. 225. 766. Union Carbide, US 2 984 703, 1956.
738. F. Candau in F. Candau, R. H. Ottewill (eds.): 767. Uniroyal, GB 1 168 760, 1967.
Scientific Methods for the Study of Polymer 768. Monsanto, GB 1 154 820, 1966.
Colloids and their Applications, vol. XIII, 769. Societá Italiana Resine, FR 2 074 118, 1970.
Kluwer Publishers, Dordrecht, Boston 1991, 770. R. D. Dunlop, F. E. Reese, Ind. Eng. Chem. 40
p. 73. (1948) 654 – 660.
739. D. Hunkeler, A. E. Hamielec, Polymer 32 771. D. B. Gershberg, J. E. Longfield, Symp.
(1991) 2626. Polymerization Kinetics and Catalyst Systems,
740. Dow, US 3 284 393, 1959. part 1, preprint no. 10, 54th Meeting A. I. Ch.
741. Nalco, US 3 624 019, 1970. E., New York, Dec. 2 – 7, 1961.
742. BASF, DAS 2 432 699, 1974. 772. A. W. DeGraff, G. W. Poehlein, J. Polym. Sci.
743. IG-Farbenind., DT 663 469, 1930. Polym. Phys. Ed. 9 (1971) 1955 – 1976.
744. IG-Farbenind., DT 900 019, 1937. 773. M. Nomura et al., J. Appl. Polym. Sci. 15
745. Houben-Weyl, 4. ed., “Makromolekulare (1971) 675 – 691.
Stoffe,” vol. XIV–1, part 1, p. 159, 186, 333, 774. T. Ueda, S. Omi, H. Kubota, J. Chem. Eng.
337, 874, 878, 912, 1051. Jpn. 4 (1971) 50 – 54.
746. IG-Farbenind., DT 629 220, 1933. 775. S. Omi, T. Ueda, H. Kubota, J. Chem. Eng.
747. K. Chujo, Y. Harada, S. Tokuhara, J. Polym. Jpn. 2 (1969) 193 – 198.
Sci. Polym. Symp. 27 (1969) 321 – 332. 776. H. Gerrens, K. Kuchner, Br. Polym. J. 2
748. J. Sňupárek, Angew. Makromol. Chem. 25 (1970) 18 – 24.
(1972) 105 – 112, 113 – 119. 777. H. Gerrens, K. Kuchner, G. Ley,
749. J. Šňupárek, Angew. Makromol. Chem. 37 Chem.-Ing.-Tech. 43 (1971) 693 – 698.
(1974) 1 – 9. 778. G. Ley, H. Gerrens, Makromol. Chem. 175
750. J. Šňpárek, F. Krška, J. Appl. Polym. Sci. 20 (1974) 563 – 581.
(1976) 1753 – 1764. 779. B. W. Brooks, H. W. Kropholler, S. N. Purt,
751. J. Šňupárek, F. Krška, J. Appl. Polym. Sci. 21 Polymer 19 (1978) 193 – 196.
(1977) 2253 – 2260. 780. M. Ghosh, T. H. Forsyth, ACS Symp. Ser. 24
752. V. I. Yeliseyeva, N. G. Zharkova, A. V. (1976) 367 – 378.
Chubarova, P. I. Zubov, Polym. Sci. USSR 781. J. D. Stevens, J. O. Funderburk, Ind. Eng.
(Engl. Transl.), 7 (1965) 171 – 178. Chem. Process. Des. Dev. 11 (1972) 360 – 369.
753. V. I. Yeliseyeva, P. I. Zubov, U. F. 782. G. Beckmann, H. Matis, Chem.-Ing.-Tech. 38
Malofeyevskaya, Polym. Sci. USSR (Engl. (1966) 209 – 214.
783. S. Senrui, A. Kodama, M. Takehisa, J. Polym.
Transl.) 7 (1965) 1496 – 1504.
Sci. Chem. Ed. 12 (1974) 2403 – 2417.
754. V. I. Yeliseyeva et al., Polym. Sci. USSR (Engl.
784. A. R. Berens, J. Appl. Polym. Sci. 18 (1974)
Transl.) 9 (1967) 813 – 822.
2379 – 2390.
130 Polymerization Processes

785. A. Penlidis, A. E. Hamielec, J. F. MacGregor, 808. F. J. Schork, W. H. Ray, J. Appl. Polym. Sci. 28
J. Vinyl Technol. 6 (1984) 134. (1983) 407.
786. R. K. Greene, R. A. Gonzalez, G. W. Poehlein, 809. M. S. Nomura et al., ACS Meeting Preprints,
ACS Symp. Ser. 24 (1976) 341 – 358. San Francisco 1980.
787. C. Kiparissides, J. F. MacGregor, A. E. 810. H. C. Lee, G. W. Poehlein, Chem. Eng. Sci. 41
Hamielec, J. Appl. Polym. Sci. 23 (1979) (1986) 1023.
401 – 418. 811. C. H. Lee, R. G. Mallinson, AIChE J. 34
788. J. B. Rawlings, W. H. Ray, AIChE J. 33 (1987) (1988) 840.
1663. 812. J. B. Rawlings, W. H. Ray, Chem. Eng. Sci. 42
789. J. B. Rawlings, J. C. Prindle, W. H. Ray, in (1987) 2767.
K. H. Reichert, W. Geiseler (eds.): Polymer 813. R. N. Mead, G. W. Poehlein, Ind. Eng. Chem.
Reaction Engineering, Hüthig & Wepf Verlag, 28 (1989) 51.
New York 1986, p. 1. 814. M. Nomura in K. H. Reichert, W. Geiseler
790. J. B. Rawlings, W. H. Ray, Polym. Eng. Sci. 42 (eds.): Polymer Reaction Engineering,
(1987) 2767. Hüthig & Wepf, Basel 1986, p. 42.
815. M. Pollock, J. F. MacGregor, A. E. Hamielec,
791. J. B. Rawlings, W. H. Ray, Polym. Eng. Sci. 28
ACS Symp. Ser. 197 (1982) 209.
(1988) 237, 257.
816. H. Tobita: “Crosslinking Kinetics in Emulsion
792. G. W. Poehlein, D. J. Dougherty, Rubber
Polymerization,” Polymer 25 (1992) 2671.
Chem. Technol. 50 (1977) 601 – 638.
817. M. Nomura, K. Fujita: “Particle Nucleation in
793. A. Penlidis, J. F. MacGregor, A. E. Hamielec:
Emulsion Copolymerization Containing
“Continuous Emulsion Polymerization:
Multifunctional Monomers,” Polym. Int.
Design and Control of CSTR Trains,” Chem.
(1992).
Eng. Sci. 44 (1989) 273. 818. M. Buback, Makromol. Chem. 181 (1980)
794. B. W. Brooks, Br. Polym. J. 5 (1973) 373.
199 – 211. 819. Y. Tatsukami, T. Takahashi, H. Yoshioka,
795. O. Levenspiel: Chemical Reaction Makromol. Chem. 181 (1980) 1108.
Engineering, Wiley, New York 1962, p. 145. 820. S. Seidl, G. Luft, J. Macromol. Sci. Chem.
796. T. Matsuura, M. Kato, Chem. Eng. Sci. 22 A 15 (1981) 1.
(1967) 171 – 183. 821. S. Goto, K. Yamamoto, S. Furui, M. Sugimoto,
797. R. S. Knorr, K. F. O’Driscoll, J. Appl. Polym. J. Appl. Polym. Sci., Appl. Polym. Symp. 36
Sci. 14 (1970) 2683 – 2696. (1981) 21.
798. M. Nomura, personal communication. 822. G. Donati et al., ACS Symp. Ser. 196 (1982)
799. R. K. Greene, Ph. D. Thesis, Dept. of Chem. 579.
Eng., Lehigh Univ., Bethlehem, 1976. 823. M. Buback, H. Lendle, Makromol. Chem. 184
800. R. F. Dickinson, C. E. Gall, 59th CIC (1983) 193.
Conference, London, Ontario, June 1976. 824. W. Hollar, P. Ehrlich, Chem. Eng. Commun.
801. P. V. Danckwerts, Chem. Eng. Sci. 8 (1958) 24 (1983) 57.
93 – 102. 825. Y. Ogo: “Polymerizations at High Pressures,”
802. Th. N. Zwietering, Chem. Eng. Sci. 8 (1958) J. Macromol. Sci. Revs. Macromol. Chem.
101. Phys. C 24 (1984) 1.
803. Th. N. Zwietering, Chem. Eng. Sci. 11 (1959) 826. P. Feucht, B. Tilger, G. Luft, Chem. Eng. Sci.
1 – 15. 40 (1985) 1935.
804. E. B. Naumann, J. Macromol. Sci. Rev. 827. H. Mavridis, C. Kiparissides, Polym. Process
Macromol. Chem. 10 (1974) 75 – 112. Eng. 3 (1985) 263.
805. E. Fitzer, W. Fritz: Technische Chemie. Eine 828. S. K. Gupta, A. Kumar, M. V. G.
Einführung in die Chemische Krishnamurthy, Polym. Sci. Eng. 25 (1985) 1.
Reaktionstechnik, Springer, Berlin 1975, 829. B. J. Yoon, H. K. Rhee, Chem. Eng. Commun.
p. 352 ff. 24 (1985) 253.
806. G. V. Schulz, Z. Phys. Chem. Abt. B 43 (1939) 830. P. P. Shirodkar, G. O. Tsien, Chem. Eng. Sci.
25 – 46. 41 (1986) 1031.
831. P. Postelnicescu, A. M. Dumitrescu, Materiale
807. F. J. Schork, W. H. Ray in D. R. Bassett, A. E.
Plastice 24 (1987) 2.
Hamielec (eds.): “Emulsion Polymers and
832. S. K. Gupta, Current Science 56 (1987) 19.
Emulsion Polymerization,” ACS Symp. Ser. 833. M. Tjahjadi, S. K. Gupta, M. Morbidelli, A.
165 (1981). Varma, Chem. Eng. Sci. 42 (1987) 2385.
Polymerization Processes 131

834. A. Brandolin, N. J. Capiati, J. N. Farber, E. M. 853. T. Kimura, J. Phys. Soc. Jpn. 17 (1962) 1884.
Valles, Ind. Eng. Chem. Res. 27 (1988) 784. 854. A. Charlesby and S. H. Pinner, Proc. R. Soc.
835. J. K. Beasley: “Polymerization at High London A 249 (1959) 367.
Pressure,” in Comprehensive Polymer Science, 855. C. Tzoganakis, J. Vlachopoulos, A. E.
vol. 3, Pergamon Press, Oxford 1989, p. 273. Hamielec, Intl. Polym. Proc. 3 (1988) 141.
836. R. C. M. Zabisky, W.-M. Chan, P. E. Gloor, 856. A. Hrymak, McMaster University, personal
A. E. Hamielec: “A Kinetic Model for Olefin comunication 1990.
Polymerization in High Pressure Tubular 857. J. V. Seppala, M. Auer: “Factors Affecting
Reactors–A Review and Update,” Polymer 33 Kinetics in Slurry Type Coordination
(1992) 2243. Polymerization,” Prog. Polym. Sci. 15 (1990)
837. J. W. Hamer, W. H. Ray, Chem. Eng. Sci. 41 147.
(1986) 3083. 858. I. D. Burdett, “The Union Carbide UNIPOL
838. S. S. Agarwall, C. Kleinstreuer, Chem. Eng. Process: Polymerization of Olefins in a
Sci. 41 (1986) 3101. Gas-Phase Fluidized Bed,” AIChE Annual
839. C. Tzoganakis: “Reactive Extrusion of Meeting, Washington, D.C., Nov./Dec., 1988.
Polymers, A Review,” Adv. Polym. Techn. 9 859. K. Y. Choi, W. H. Ray, J. Macromol. Sci., Rev.
(1989) 321. Macromol. Chem. 25 (1985) 57.
840. D. Suwanda, R. Lew, S. T. Balke, J. Appl. 860. W. H. Ray, Ber. Bundesanst. Ges. Phys. Chem.
Polym. Sci. 35 (1988) 1019, 1033. 90 (1986) 947.
841. C. Tzoganakis, J. Vlachopoulos, A. E. 861. B. de Carvalho, P. Gloor, A. E. Hamielec,
Hamielec, Polym. Eng. Sci. 28 (1988) 170. Polymer 30 (1989) 280.
842. G. Foster: “Processing and Product 862. K. B. McAuley, J. F. MacGregor, A. E.
Performance Control” in Polymer Reaction Hamielec, AIChE J. 36 (1990), 837.
Engineering Course Notes, McMaster 863. R. Galvan, M. Tirrell, Chem. Eng. Sci. 41
University, Hamilton, Canada 1990. (1986) 2385.
843. G. M. Gale: “Crosslinked Polyethylene 864. B. de Carvalho, P. Gloor, A. E. Hamielec,
Extrusions Using Direct Silane Injection,” Polymer 31 (1990) 1294.
ANTEC /84, Society of Plastics Engineers, 865. S. Floyd et al., J. Appl. Polym. Sci. 33 (1987)
1984, p. 2. 1021.
844. L. Panzer, “Advances in Silane Crosslinking of 866. R. L. Laurence, G. N. Chiovetta in K. H.
Polyethylene,” ANTEC /88, Society of Plastics Reichert, W. Geiseler (eds.): Polymer Reaction
Engineers, 1988, p. 1421. Engineering, Hanser Verlag, München 1983,
845. R. Greco et al., Polym. Proc. Eng. 4 (1986) p. 74.
253. 867. E. D. Nagel, A. V. Kirillov, W. H. Ray, Ind.
846. A. Hogt, ANTEC /88, Society of Plastics Eng. Chem. Prod. Res. Dev. 19 (1980) 372.
Engineers, 1988, p. 1478. 868. S. Floyd, K. Y. Choi, T. W. Taylor, W. H. Ray,
847. S. Al-Malaika in J. L. Benham, J. F. Kinstle
J. Appl. Polym. Sci. 32 (1986) 2231, 2935.
(eds.): Chemical Reactions on Polymers,
869. T. W. Taylor, K. Y. Choi, H. Yuan, W. H. Ray,
ACS, Washington 1988, chap. 29.
“Physicochemical Kinetics of Liquid Phase
848. R. C. Kowalski, Chem. Eng. Prog. 85 (1989)
Propylene Polymerization,” in Proceedings
no. 5, 67.
International Symposium on Transition Metal
849. M. Saleem, W. E. Baker, J. Appl. Polym. Sci.
Catalyzed Polymerization, Gordon and
39 (1990) 655.
Breach, New York 1983.
850. A. E. Hamielec, P. E. Gloor, S. Zhu, “Chemical
870. J. M. Vela Estrada, A. E. Hamielec:
Modification of High Molecular Weight
“Measurement of the Bivariate Distribution of
Polymers in Extruders – Experimentation and
composition and Molecular weight for Binary
Computer Modelling of the Kinetics of Chain
Copolymers – A Review,” Polym. Reaction
Scission, Long Chain Branching, Crosslinking
Engng. (1992).
and Grating” in 4th International Congress on
871. H. Gerrens, Proc. 4th Intern./6th European
Compatibilizers and Reactive Polymer
Symp. Chem. Reaction Engng. Heidelberg,
Alloying, New Orleans 1991.
vol. 2, April 6 –8, 1976, DECHEMA,
851. A. E. Hamielec, P. E. Gloor, S. Zhu, Canad. J.
Frankfurt/Main 1976, p. 585 – 614.
Chem. Eng. 69 (1991) 611.
852. K. Motha, J. Seppala, C. Bergstrom, Polym. 872. G. V. Schulz, Z. Phys. Chem. Abt. B 30 (1935)
Eng. Sci. 29 (1989) 1579. 379 –398.
132 Polymerization Processes

873. P. J. Flory, J. Am. Chem. Soc. 58 (1936) 897. M. Rätzsch, R. Nitzsche, Plaste Kautsch. 17
1877 – 1885. (1970) 6 – 8.
874. E. F. G. Herington, A. Robertson, Trans. 898. L. Gold, J. Chem. Phys. 28 (1958) 91 – 99.
Faraday Soc. 38 (1942) 490 – 501. 899. L. H. Peebles: Molecular Weight Distributions
875. G. Gee, H. M. Melville, Trans. Faraday Soc. in Polymers, Interscience, New York 1971,
40 (1944) 240 – 251. p. 137 ff.
876. L. Küchler: Polymerisationskinetik, Springer, 900. P. J. Flory, J. Am. Soc. 62 (1940) 1561 – 1565.
Berlin 1951, p. 74 ff. 901. W. H. Abraham, Ind. Eng. Chem. Fundam. 2
877. C. H. Bamford, W. G. Barb, A. D. Jenkins, P. F. (1963) 221 – 224.
Onyon: The Kinetics of Vinyl Polymerization 902. H. Kilkson, Ind. Eng. Chem. Fundam. 3
by Radical Mechanisms, Butterworth, London (1964) 281 – 293.
1958, p. 279 ff. 903. T. E. Corrigan, M. J. Dean, J. Makromol. Sci.
878. K. G. Denbigh, Trans. Faraday Soc. 40 (1944) Chem. 2 (1968) 645 – 662.
904. L. L. Böhm et al., Fortschr. Hochpolym.
352 –373.
Forsch. 9 (1972) 1 – 45.
879. K. G. Denbigh, Trans. Faraday Soc. 43 (1947)
905. H. Dosal, R. Raff, Z. Phys. Chem. Abt. B 32
648 –660.
(1936) 117 – 129.
880. K. G. Denbigh, J. Appl. Chem. 1 (1951) 906. P. J. Flory, Chem. Rev. 39 (1946) 137 – 197.
227 – 236. 907. J. A. Biesenberger, AIChE J. 11 (1965) 369,
881. K. G. Denbigh: Chemical Reactor Theory, 371 – 373.
Cambridge University Press, Cambridge 1965, 908. W. H. Ray, J. Macromol. Sci. Rev. Macromol.
p. 104 – 107. Chem. 8 (1972) 1 – 56.
882. J. A. Biesenberger, Z. Tadmor, J. Appl. Polym. 909. N. H. Smith, G. A. Sather, Chem. Eng. Sci. 20
Sci. 9 (1965) 3409 – 3416. (1965) 15 – 23.
883. J. A. Biesenberger, Z. Tadmor, Polym. Eng. 910. W. H. Abraham, Chem. Eng. Sci. 25 (1970)
Sci. 6 (1966) 299 – 305. 331 – 335.
884. Z. Tadmor, J. A. Biesenberger, J. Polym. Sci. 911. W. Kuhn, Ber. Dtsch. Chem. Ges. B 63 (1930)
Polym. Lett. Ed. 3 (1965) 753 – 759. 1503 – 1509.
885. Z. Tadmor, J. A. Biesenberger, Ind. Eng. 912. P. Wittmer, F. Hafner, H. Gerrens, Makromol.
Chem. Fundam. 5 (1966) 336 – 343. Chem. 104 (1967) 101 – 119.
886. S. Katz, G. M. Saidel, AIChE J. 13 (1967) 913. R. Simha, H. Branson, J. Chem. Phys. 12
319 – 326. (1944) 253 – 267.
887. R. Thiele, Chem. Tech. (Leipzig) 19 (1967) 914. W. H. Stockmayer, J. Chem. Phys. 13 (1945)
221 –225. 199 – 207.
888. R. Thiele, K.-D. Herbrich, J. Fischmann, 915. W. Ring, Dechema-Monogr. 49 (1964)
Chem. Tech. (Leipzig) 22 (1970) 445 – 460. 75 – 97.
889. N. Friis, A. E. Hamielec, J. Appl. Polym. Sci. 916. W. Ring, Makromol. Chem. 75 (1964)
19 (1975) 97. 203 – 207.
890. S. T. Balke, A. E. Hamielec, J. Appl. Polym. 917. K. F. O’Driscoll, R. Knorr, Macromolecules 2
Sci. 17 (1973) 905 – 949. (1969) 507 – 515.
891. J. H. Duerksen, A. E. Hamielec, J. W. Hodgins, 918. K. H. Reichert, H. U. Moritz: “Polymer
AIChE J. 13 (1967) 1081 – 1086. Reaction Engineering,” Comprehensive
892. J. H. Duerksen, A. E. Hamielec, J. Polym. Sci. Polymer Sience, vol. 3, Pergamon Press,
Polym. Symp. 25 (1968) 155 – 166. Oxford 1989, p. 327.
919. H. Tobita, A. E. Hamielec: “Control of
893. A. E. Hamielec, J. W. Hodgins, K. Tebbens,
Network Structure in Free-Radical
AIChE J. 13 (1967) 1087 – 1091.
Crosslinking Copolymerization,” Polymer
894. A. W. Hui, A. E. Hamielec, J. Polym. Sci.
(1992)
Polym. Symp. 25 (1968) 167 – 189.
920. H. Tobita, A. E. Hamielec, in “Computer
895. N. Khac Tien, E. Flaschel, A. Renken in K. H.
Applications in Applied Science II,” ACS
Reichert, W. Geiseler (eds.). Polymer Reaction
Symp. Ser. 404 (1989) 242.
Engineering, Hanser, New York 1983, p. 175. 921. H. Tobita, A. E. Hamielec in P. J. Lemstra,
C. Cozewith, J. Appl. Polym. Sci. 15 (1971) L. A. Kleintjens (eds.): Integration of
2855 – 2863. Fundamental Polymer Science and
896. R. W. Dunn, C. C. Hsu, Adv. Chem. Ser. 109 Technology–4, Elsevier Applied Science,
(1972) 85 – 86. London 1990, p. 33.
Polymerization Processes 133

922. H. M. J. Boots: “Simulation Model for Densely 942. J. C. Prindle, W. H. Ray: “Emulsion
Cross-linked Networks Formed by Polymerization Model Development for
Chain-Reactions,” in L. A. Kleintjens, P. J. Operation Below the CMC,” AIChE Meeting,
Lemstra (eds.): Integration of Fundamental New York, Nov. 15 – 20, 1987.
Polymer Science and Technology, Elsevier 943. R. A. Hutchinson, W. H. Ray, J. Appl. Polym.
Applied Science, London 1986, p. 204. Sci. 34 (1987) 657.
923. C. C. Chang, J. W. Miller, Jr., G. R. Schorr, J. 944. K. Y. Choi, W. H. Ray, Chem. Eng. Sci. 40
Appl. Polym. Sci. 39 (1990) 2395. (1985) 2261.
924. G. R. Meira, A. F. Johnson, Polym. Eng. Sci. 945. K. Y. Choi, W. H. Ray, Chem. Eng. Sci. 43
21 (1981) 415. (1988) 2587.
925. D. A. Couso, G. R. Meira, Polym. Eng. Sci. 24 946. Y. J. Lu, B. W. Brooks, Chem. Eng. Sci. 44
(1984) 391. (1989) 857.
926. D. A. Couso, L. M. Alassia, G. R. Meira, J. 947. D. C. H. Chien, A. Penlidis,: “On-Line Sensors
Appl. Polym. Sci. 30 (1985) 3249. for Polymerization Reactors,” J. Macromol.
927. G. L. Frontini, G. E. Elicabe, D. A. Couso,
Sci. Rev. Macromol. Chem. Phys. C 30 (1990)
G. R. Meira, J. Appl. Polym. Sci. 31 (1986)
1.
1019.
948. C. H. Sohl: “Process Sensors,” in Polymer
928. L. M. Alassia, D. A. Couso, G. R. Meira, J.
Reaction Engineering, Engineering
Appl. Polym. Sci. 36 (1988) 481.
929. G. L. Frontini, G. E. Elicabe, G. R. Meira, J. Foundation (AIChE), Santa Barbara 1991.
Appl. Polym. Sci. 33 (1987) 2165. 949. J. F. MacGregor, A. Penlidis, A. E. Hamielec:
930. L. M. Alassia, G. L. Frontini, J. R. Vega, G. R. “Control of Polymerization Reactors: A
Meira, J. Appl. Polym. Sci., Part C: Polym. Review,” Polym. Process Eng. 2 (1984) 179.
Lett. 26 (1988) 201. 950. G. E. Elicabe, G. R. Meira: “Estimation and
931. J. W. Hamer, T. A. Akramov, W. H. Ray, Chem. Control in Polymerization Reactors: A
Eng. Sci. 36 (1981) 1897. Review,” Polym. Eng. Sci. 28 (1988) 121.
932. A. D. Schmidt, W. H. Ray, Chem. Eng. Sci. 36 951. D. Kozub, J. F. MacGregor: “Feedback Control
(1981) 1401. of Polymer Quality in Semi-Batch
933. J. A. Biesenberger, D. H. Sebastian, Polym. Copolymerization Reactors,” Chem. Eng. Sci.
Eng. Sci. 16 (1976) 117. 47 (1992).
934. D. H. Sebastian, J. A. Biesenberger, Polym. 952. H. Schuler: “Automation,” IFAC PRP-4,
Eng. Sci. 19 (1979) 190. Ghent, Belgium, Pergamon Press, London
935. L. S. Henderson, Chem. Eng. Prog. 83 (1987) 1980, p. 369.
42. 953. L. Gagnon, J. F. MacGregor: “State Estimation
936. A. Uppal, W. H. Ray, A. B. Poore, Chem. Eng. for Continuous Emulsion Polymerization,”
Sci. 29 (1974) 967. Can J. Chem. Eng. 69 (1991) 648.
937. A. Uppal, W. H. Ray, A. B. Poore, Chem. Eng. 954. K. McAuley, J. F. MacGregor: “Online
Sci. 31 (1976) 205. Inference of Polymer Properties in an
938. J. B. Planeaux, K. F. Jensen, Chem. Eng. Sci. Industrial Polyethylene Reactor,” AIChE J. 37
41 (1986) 1497. (1991) 825.
939. A. D. Schmidt, A. B. Clinch, W. H. Ray, Chem.
955. J. Roffel, T. W. Hoffman, J. F. MacGregor,
Eng. Sci. 39 (1984) 419.
DYCORD 89, Maastricht, Netherlands 1989.
940. F. Teymour, W. H. Ray, Chem. Eng. Sci. 44
956. J. F. MacGregor, A. E. Hamielec, A. Penlidis,
(1989) 1967.
941. M. J. Barandiaran, J. C. Prindle, W. H. Ray: M. Pollock, IFAC PRP-5, Antwerp, Belgium,
“The Effects of Chain Transfer Agents on the Pergamon Press, London 1983, p. 291.
Stability of Continuous Polymerization 957. http://www.kvs.ch/bulletin2000/bulletin2 200-
Reactors,” AIChE Meeting, Washington, 0/Kap2.htm
Nov. 27 – Dec. 2, 1988.

Polymerization Technology → Polymerization Processes


Polymers, Electrically Conducting 1

Polymers, Electrically Conducting


Herbert Naarmann, BASF Aktiengesellschaft, Ludwigshafen, Federal Republic of Germany

1. Introduction . . . . . . . . . . . . . . . . . 1 5.4. Polythiophene . . . . . . . . . . . . . . . . 10


2. Synthetic Routes . . . . . . . . . . . . . . 2 5.5. Polyphenylene . . . . . . . . . . . . . . . 10
3. Principles of Electrical Conduction . . 3 5.6. Poly(Phenylene Sulfide) . . . . . . . . . 12
4. Orientation Processes . . . . . . . . . . . 4 5.7. Poly(Phenylene Vinylene) . . . . . . . . 13
5. Types of Electrically Conducting Or-
5.8. Polyaniline . . . . . . . . . . . . . . . . . . 13
ganic Materials . . . . . . . . . . . . . . . 5
5.1. Polyacetylene . . . . . . . . . . . . . . . . 5 5.9. Miscellaneous Polymers . . . . . . . . . 15
5.2. Polydiacetylenes . . . . . . . . . . . . . . 7 6. Uses . . . . . . . . . . . . . . . . . . . . . . 18
5.3. Polypyrrole . . . . . . . . . . . . . . . . . 8 7. References . . . . . . . . . . . . . . . . . . 18

1. Introduction polymers”, “hydrogen-bonded polymers”, and


“mixed polymers”. Highest conductivity values
Electrically conducting polymers (ECPs) are reached about 10−3 S/cm. In 1964 Little the-
materials with an extended system of C=C con- oretically evaluated the possibility of supercon-
jugated bonds. They are obtained by reduction ductivity in polymers and suggested a model,
or oxidation reactions (called doping), giving consisting of a polyene chain with cyanine, dye-
materials with electrical conductivities up to like substituents [3]. In the same year system-
105 S/cm. These materials differ from polymers atic studies were presented based on aromatic
filled with carbon black or metals because the and heterocyclic compounds exhibiting electri-
latter are only conductive if the individual con- cal conductivities of 0.5 S/cm [4], followed by
ductive particles are mutually in contact and studies correlating doping, pressure, irradiation,
form a coherent phase. and chain length to conductivity, with values up
This review concerns the synthesis routes, to 100 S/cm [5].
polymerization techniques, doping, orientation, Interest heightened and became acute
and development of well-defined, highly con- from 1975 when IBM scientists showed that
ducting polymeric materials. Their wide range poly(sulfur nitride), (SN)n , was superconduc-
of potential uses from electrodes in recharge- tive [6] and MacDiarmid’s group reported
able batteries to organic transistors is limited [7] the doping of polyacetylene films prepared
by their vulnerability to air and moisture due to by Shirakava [8] reaching conductivity values
their highly conjugated structures and the dop- of 38 S/cm. Since then many expectations have
ing agents. Electrically conducting materials are been raised, but scientific progress and practi-
compiled, their specific properties and potential cal applications have been limited; they depend
applications are described. on the reproducible production of well-defined
Numerous attempts have been made to syn- specimens, the determination of synthesis con-
thesize “conductive organic materials”. The ditions, and the laws relating these conditions to
first was the synthesis of polyaniline by product properties. Synthetic methods are im-
F. Goppelsroeder in 1891 [1]. After decades proving; more easily processible, soluble, flexi-
interest grew in organic polymers as insulators, ble materials are now being produced.
but not as electrical conductors.
In the late 1950s organic semiconductors be-
came the focus of investigations. Preliminary 2. Synthetic Routes
studies in this field up until the mid 1960s
are reviewed in [2]. The semiconducting poly- The synthesis of electrically conducting poly-
mers were termed “covalent organic polymers”, mers with conjugated −HC=CH− bonds re-
“charge-transfer complexes”, “organometallic quires the controlled coupling of a large number

c 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a21 429

You might also like