Mathematical Modeling and Computer Simulatordatabase For Emulsion Polymerizations

Mathematical modeling and computer simulator/database for
emulsion polymerizations
J. Gao
a
, A. Penlidis
b,
*
a
Polymer Process Group, Polymer Research Laboratory, BASF-AG, Ludwigshafen 68161, Germany
b
Department of Chemical Engineering, University of Waterloo, Waterloo, Ont., Canada N2L 3G1
Received 14 June 2001; accepted 11 September 2001
Abstract
Though emulsion polymerization has been commercialized for more than half a century, some important aspects
like particle nucleation, coagulation, etc. are still not well understood. To simulate this complicated process, a
general approach has been adopted. Information from classical modeling sources in the literature (e.g. [J Macro-
mol Sci, Rev Macromol Chem C11 (1974) 177; J Polym Sci Chem 19 (1981) 25; ACS Symp Ser 313 (1986) 219;
J Macromol Sci, Rev Macromol Chem 38 (1998) 207]; just to mention a few) has been carefully reviewed and
general mass (molar), energy, population (dead polymer molecules and radicals) and particle balances have been
written and evaluated with experimental data, with the nal aim to develop a model that is exible, reliable and
practical at the same time. q 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Emulsion polymerization; Mathematical modeling of emulsion polymerization; Latex production
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
1.1. Brief overview of the classical SmithEwart theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
1.2. Recent advances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
2. Development of emulsion polymerization simulation model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
2.1. Initiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
2.2. Particle nucleation and growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
2.2.1. Micellar nucleation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
2.2.2. Homogenous nucleation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
2.2.3. Mass balances for radicals in the water phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
Prog. Polym. Sci. 27 (2002) 403535
0079-6700/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S0079- 6700( 01) 00044- 2
www.elsevier.com/locate/ppolysci
* Corresponding author. Tel.: 11-519-888-4567; fax: 11-519-888-6179.
Abbreviations: AN, acrylonitrile; BA, butyl acrylate; CMC, critical micelle concentration; CSTR, continuous stirred tank
reactor; KPS, potassium persulfate; MMA, methyl methacrylate; PVC, polyvinyl chloride; Sty, styrene; VCL, vinyl chloride
2.2.4. Recent modeling efforts on particle nucleation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
2.2.5. Specic modeling considerations on particle nucleation . . . . . . . . . . . . . . . . . . . . . . . . 424
2.3. Monomer partitioning in emulsion homopolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
2.3.1. Empirical approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
2.3.2. Theoretical approach: thermodynamic equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
2.4. Desorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
2.5. Average number of radicals per particle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
2.6. Overall mass balances and molecular weight development . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
2.6.1. Mass balance equations for each species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
2.6.2. Molecular weight development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
2.7. Reaction kinetics at high conversion levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
3. Emulsion polymerization simulation package/database overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
3.1. General description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
3.1.1. Computational options screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
3.1.2. Output les screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
3.1.3. Flow options screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
3.2. Package database overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
3.2.1. Monomer systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
3.2.2. Other components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
3.3. Description of database items . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
4. Simulation of styrene emulsion homopolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 446
4.1. Model testing results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
4.1.1. Bataille et al. [86] and Bataille and Gonzalez [92] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
4.1.2. Canegallo et al. [94] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452
4.1.3. de La Rosa et al. [95] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452
4.1.4. Blackley and Sebastian [96,97] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
4.1.5. Mayer et al. [98] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
4.1.6. Harada et al. [13] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
4.1.7. Salazar et al. [99] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
5. Simulation of emulsion homopolymerization of vinyl acetate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
5.1.1. Nomura et al. [49] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
5.1.2. Penlidis et al. [103] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
6. Simulation of homopolymerization of acrylic monomers: methyl methacrylate, butyl acrylate and ethyl
acrylate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
6.1. Simulation of methyl methacrylate emulsion homopolymerization . . . . . . . . . . . . . . . . . . . . . . . 471
6.1.1. Model testing results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
6.2. Simulation of butyl acrylate emulsion homopolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
6.3. Simulation of ethyl acrylate emulsion homopolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
7. Extensions to emulsion copolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
7.1. Brief literature summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
7.2. Monomer partitioning in copolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
7.3. Desorption in copolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504
7.4. Simulation of copolymerization of styrene/MMA in emulsion . . . . . . . . . . . . . . . . . . . . . . . . . . 504
7.5.1. Nomura et al. [147] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506
7.5.2. Forcada and Asua [146,151] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
8. Case study: extending the nucleation stage in emulsion polymerization . . . . . . . . . . . . . . . . . . . . . . . 512
8.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
8.2. An interesting control problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
J. Gao, A. Penlidis / Prog. Polym. Sci. 27 (2002) 403535 404
8.3. A possible approach to controlling the PSD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514
8.3.1. Extension of nucleation stage 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516
8.3.2. Emulsier feed rate policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 518
9. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
Appendix A. Seeded emulsion polymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
Appendix B. Impurity effects on emulsion kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527
1. Introduction
Recently Van Herk and German [5] asked the question whether it will ever be possible to model
emulsion co-/terpolymerization after reviewing several difcult aspects in emulsion polymerization. It
has been long recognized that emulsion polymerization is a complex heterogeneous process involving
transport of monomers and other species and free radicals between aqueous and organic phases.
Compared to other heterogeneous polymerizations, like suspension or precipitation, emulsion polymer-
ization is likely the most complicated system. The rate of polymerization in the organic phase is not only
controlled by monomer partitioning but also affected by other phenomena like particle nucleation, and
radical absorption and desorption. Particle stability is affected by emulsier type, amount of emulsier
and ionic strength of the dispersion media. All these factors make modeling of this system very difcult.
Other difculties encountered in modeling emulsion polymerization include numerical intensity in
solving sets of nonlinear ordinary differential (or integro-differential) equations coupled with algebraic
equations and the lack of information on certain model parameters. Though emulsion polymerization has
been commercialized for more than half a century, some important aspects like particle nucleation,
coagulation, etc. are still not well understood. To simulate this complicated process, a general approach
has been adopted. Information from classical modeling sources in the literature (e.g. Refs. [14]; just to
mention a few) has been carefully reviewed and general mass (molar), energy, population (dead polymer
molecules and radicals) and particle balances have been written and evaluated with experimental data,
with the nal aim to develop a model that is exible, reliable and practical at the same time.
1.1. Brief overview of the classical SmithEwart theory
Advances in achieving a basic understanding of emulsion polymerization have been made by many
researchers. Because of the complexity of the system, contradictory conclusions often exist in the early
studies. It is not the scope of this paper to give a historical literature review on all aspects of emulsion
polymerization. Instead, emphasis was focused on more recent developments. The basic mechanism of
emulsion polymerization was rst postulated by Harkins [6]. A typical emulsion polymerization recipe
usually includes a dispersion medium (water), monomer, water-soluble initiator and emulsier. Fig. 1 is
an illustrative diagram of this heterogeneous system. Emulsier dissolves in water at very low concen-
tration, whereas at higher concentrations (above the critical micelle concentration, cmc), it will form
aggregates called micelles. Sodium dodecyl sulfate (SDS) is a typical emulsier, with a hydrophobic and
a hydrophilic end. When micelles are formed, the hydrophilic end of each emulsier molecule on the
surface points towards the aqueous phase while its hydrophobic end points inwards. Hence, the interior
Nomenclature
a partition coefcient for monomer radicals between aqueous and polymer phases, dened in
Eq. (61)
a root-mean-square end-to-end distance per square root of the number of monomer units
A parameter in free volume model
A
mic
total surface area of free micelles (dm
2
)
A
md
total surface area of monomer droplets (dm
2
)
A
p
total surface area of polymer particles (dm
2
)
B parameter in free volume model
B
n3
number of trifunctional branch points per polymer chain
B
n4
number of tetrafunctional branch points per polymer chain
C
p
heat capacity (cal/mol K)
CTA chain transfer agent
[CTA]
w
concentration of chain transfer agent dissolved in water (mol/l)
d
p
diameter of polymer particle (dm)
D
po
initial diffusivity of monomer radicals in polymer phase (dm
2
/s)
D
p
diffusivity of monomer radicals in polymer phase (dm
2
/s)
D
w
diffusivity of monomer radicals in water phase (dm
2
/s)
f initiator efciency
F radical capture efciency
F
in
i
inlet ow rate of species i (mol/min)
F
out
i
outlet ow rate of species i (mol/min)
F
S
inlet ow rate of emulsier (mol/min)
DH reaction enthalpy (cal/mol)
H enthalpy (cal/mol)
H
0
enthalpy at reference temperature of T
0
(cal/mol)
I initiator
[I]
w
concentration of initiator dissolved in water (mol/l)
j
c
monomer units per segment
j
cr
critical chain length at which water phase radical can be absorbed
DG
i
Gibbs free energy for species i
k
cp
rate constant of radicals captured by polymer particles (min
21
)
k
cm
rate constant of aqueous phase radicals captured by micelles (min
21
)
k
d
rate constant of initiator decomposition (min
21
)
k
des
rate constant of desorption (min
21
)
k
1
rate constant of reaction (6) (l/mol min)
k
2
rate constant of reaction (7) (l/mol min)
k
fcta
rate constant of transfer to chain transfer agent (l/mol min)
k
fm
rate constant of chain transfer to monomer (l/mol min)
k
p
rate constant for propagation in polymer particles (l/mol min)
k
p
p
rate constant for chain transfer to terminal double bond (l/mol min)
k
pp
p
rate constant for chain transfer to internal double bond (l/mol min)
k
pw
rate constant for propagation in aqueous phase (l/mol min)
k
t
rate constant for termination in polymer particles (l/mol min)
k
tw
rate constant for termination in aqueous phase (l/mol min)
k
3
free volume parameter
k
t
% percentage of termination reaction by combination
k
wp
partition coefcient dened in Eq. (48)
k
z
rate constant of chain transfer to inhibitor (l/mol min)
m
d
partition coefcient for monomer radicals between aqueous phase and polymer phase
M monomer
[M] monomer concentration (mol/l)
[M]
p
monomer concentration in polymer particles (mol/l)
[M]
w
monomer concentration in water phase (mol/l)
[M]
w, sat
saturated concentration in aqueous phase at saturation level in homopolymerization of the
monomer (mol/l)
MIC micelle
[MIC]
w
micelle concentration in water phase (mol/l)
M
n
instantaneous number average molecular weight (Da)
M
w
instantaneous weight average molecular weight (Da)
M
n
accumulated number average molecular weight (Da)
M
w
accumulated weight average molecular weight (Da)
MW
M
molecular weight of monomer
n number of radicals
n average number of radicals per particle
N
A
Avogadro's number (6.02 10
23
molecules/mol)
N
21
Fe
number of moles of Fe
21
ions
N
31
Fe
number of moles of Fe
31
ions
N
i
(V) number of particles containing i number of radicals of volume V
N
j
number of moles of species j
N
p homo
number of particles generated by homogenous nucleation
N
p mic
number of particles generated by micellar nucleation
N
p
total number of polymer particles
N
n
number of polymer particles containing n radicals
N
seed
total number of seeds
P polymer
P
r
polymer of chain length r
r
mic
radius of micelle (dm)
r
p
radius of particle (dm)
R gas constant (1.987 cal/mol K)
RA reducing agent
[RA] concentration of reducing agent (mol/l)
R
c
rate of radical capture
R
I
rate of initiation (mol/l min)
R
It
rate of thermal initiation (mol/l min)
R
j
z
radical with j monomer units
[R
1
z
]
w
concentration of free radicals of chain length 1 in the water phase (mol/l)
[R
j
z
]
w
concentration of free radicals of chain length j in the water phase (mol/l)
[R
o
z
]
w
concentration of primary free radicals in the water phase (mol/l)
[R
w
z
] total free radical concentration in the water phase (mol/l)
R
o
z
primary radical without monomer fragment
R
p
rate of polymerization
R
r
z
radical of chain length r
S
a
area covered by one molecule of emulsier (dm
2
/molecule)
S solvent
[S] solvent concentration (mol/l)
[S]
cmc
critical micelle concentration (mol/l)
[S]
t
total concentration of surfactant (mol/l)
t time
t
f
nucleation time, i.e. time of completion of stage 1
T temperature (K)
T
0
reference temperature (K)
T
gj
glass transition temperature of component j (K)
T
jacket
temperature of reactor jacket (K)
UA product of total heat transfer coefcient and area (cal/K)
V
d
total volume of monomer droplets (l)
V
j
volume of component j (l)
V
j
volume of component j in all phases (l)
V
w
j
volume of component j in aqueous phase (l)
V
p
j
volume of component j in polymer particles (l)
V
d
j
volume of component j in monomer droplets (l)
V
m
molar volume of monomer (l)
V
mseed
molar volume of monomer in the seed (l)
V
latex
total volume of latex (l)
v
p
average volume of a polymer particle (l)
V
p
total volume of polymer particles (l)
V
pseed
volume of polymer in the seed (l)
V
R
active reactor volume (l)
V
seed
total volume of seeds (l)
V
w
total volume of water (l)
V
wt
total volume of aqueous phase (l)
w
j
weight fraction of component j in the copolymer chain
w
p
weight fraction of polymer
x monomer conversion level
x
c
conversion level at the end of interval II in emulsion polymerization
x
seed
nal conversion in seeds
of each micelle is highly hydrophobic. Monomers can exist inside micelles (monomer solubilization), or
form large (compared to micelles) monomer droplets stabilized by emulsier.
Conventional emulsion polymerization starts in the aqueous phase, where water soluble initiators
decompose and generate primary radicals. These primary radicals will propagate in the aqueous phase
rst and then will enter surrounding micelles. Emulsion polymerization is considered to go through three
intervals. In the rst interval, polymer particles are generated. Polymer particles continue to grow in the
z critical chain length at which water phase radicals can be irreversibly absorbed
Z inhibitor
[Z]
w
concentration of inhibitor dissolved in water (mol/l)
Greek letters
a
j
free volume expansion coefcient of species j
b parameter used to calculate molecular weights
e ratio of radical absorption between micelles and particles
r
m
density of monomer (g/cm
3
)
r
p
density of polymer (g/cm
3
)
r
des
rate of desorption (mol/l min)
s surface tension (dyn/cm)
s Lennard-Jones parameter
t parameter used to calculate molecular weights
f
p
j
volume fraction of species j in polymer particles
f
d
j
volume fraction of species j in monomer droplets
f
w
j
volume fraction of species j in aqueous phase
f
j
volume fraction of species j
f
p
p
volume fraction of polymer in polymer particles
x
i;j
interaction coefcient between species i and j
Subscripts
j reacting species
i; j monomer type
m monomer
p polymer
p polymer particles
d monomer droplets
w aqueous phase
Superscripts
p polymer particles
d monomer droplets
second stage by absorbing more monomer from the monomer droplets, which serve as a monomer
reservoir. On the basis of thermodynamic equilibrium, the monomer concentration in the polymer
particles during the rst and second intervals remains relatively constant. The second interval ends
when all monomer droplets are consumed. In the third interval, polymerization is completed when all the
monomer left in the particles is consumed or a limiting conversion is reached. Fig. 2 displays the most
important phenomena in a typical emulsion reaction. Early studies on emulsion polymerization kinetics
were limited to aspects like the effect of initiator and emulsier on the number of particles and rate of
polymerization in a qualitative way. Smith and Ewart [7] were the rst group that quantitatively
expressed Harkins' postulation in an empirical formulation that related number of particles to the rate
of initiation and emulsier concentration as shown below
N
p
= 0:37(r=m)
2=5
(a
s
S)
3=5
(1)
where N
p
is the number of particles per liter of water, a
s
, the surface area occupied by unit weight of
emulsier, S, the weight concentration of surfactant (emulsier), r, the rate of aqueous radical genera-
tion by initiation and m is the particle volume growth rate.
Eq. (1) is widely referred to as the SmithEwart equation and indicates the important relationship that
the particle number depends on the 0.6 power of the emulsier concentration and on the 0.4 power of the
initiator concentration.
Models in the 1960s and 1970s were more or less extended/modied versions of the SmithEwart's
expression but applied to different monomer systems. Representative ones are the efforts by Gardon
[812] and Harada et al. [13].
In a way similar to bulk and solution polymerization (R
p
= k
p
[M][R
z
]); where [R
z
] is the total radical
Fig. 1. Emulsion polymerization diagram (stages I and II).
concentration, the rate of polymerization in emulsion polymerization, R
p
, can be written as
R
p
= k
p
[M]
p
N
p
n
N
A
V
p
(2)
where k
p
is the rate constant for propagation, [M]
p
, the monomer concentration in the polymer particles,
N
p
, the total number of particles in the reaction mixture, n; the average number of radicals per particle,
V
p
, the total volume of particles and N
A
is Avogadro's number. N
p
n=N
A
V
p
gives essentially the total
radical concentration. To calculate the rate of polymerization, it is necessary to know the values of all
four variables [M]
p
, N
p
, V
p
and n in Eq. (2).
According to the SmithEwart scheme, R
p
in interval I increases due to the increasing number of
newly formed particles, however R
p
will remain relatively constant during interval II, since no new
particles are formed and the monomer concentration inside the particles [M]
p
remains almost constant
based on thermodynamic equilibrium. The conversion versus time curve in this interval appears to be
linear. The SmithEwart equation has been applied successfully to monomers with very low water
solubility (like styrene). However, many monomers with higher water solubility deviate from the
SmithEwart classical kinetic scheme. The most important reasons for these discrepancies are listed
below:
1. If the monomer is water soluble like acrylonitrile, vinyl acetate, methyl acrylate, etc., there will be
additional water phase polymerization.
2. If there is signicant radical desorption present, the average number of radicals per particle n will be
lower than 0.5. Whenever n is less than 0.5, the response of rate of polymerization to changes in
emulsier concentration will not be as predicted by SmithEwart theory. Several commercially
important monomer systems fall into this class, which includes vinyl chloride, vinylidene chloride,
methyl acrylate, and vinyl acetate.
Fig. 2. Emulsion polymerization diagram.
3. The SmithEwart model can only be applied to intervals I and II. At high conversion levels, there
may be several radicals coexisting in a large particle, and n may well be above 0.5. Each particle is
like a mini-bulk reactor and bulk kinetics can be applied. The rate of polymerization is higher in stage
III than it is in stage II. Examples of this deviation can be observed in methyl methacrylate and butyl
acrylate emulsion polymerization.
A large body of experimental evidence conrms the above deviations, thus the SmithEwart scheme
is limited to styrenic systems at low to mid-range of conversion. Ugelstad and Hansen [14], Min and Ray
[1], Alexander and Napper [15] and Gilbert and Napper [16] gave extensive reviews on earlier kinetic
studies and further discussed deviations from the SmithEwart scheme in great depth.
1.2. Recent advances
Early research investigated only specic aspects in emulsion polymerization. With recent advances in
both emulsion kinetics and polymer characterization, more in-depth studies have been carried out on
important aspects in emulsion polymerization and better understanding has been achieved. A good
review on recent advances in emulsion kinetics is given by Gilbert [17]. Researchers have also expanded
their work from homopolymerization to copolymerization and even terpolymerization. Several models
have been developed by different research groups and these models have been extended to predict long
chain branching, particle size, sequence length and molecular weight characteristics.
There is a large body of models in the literature that describes certain aspects of emulsion polymer-
ization. This paper does not intend to give a thorough review on all models presented; rather, it focuses
on signicant research work in the last two decades. Table 1 lists the most comprehensive models
presented in recent years. Models that can predict molecular weight characteristics are marked as
MWD, and models that can deliver particle size characteristics are marked as PSD.
Though there are a number of models available in the literature, most of them deal only with specic
aspects in emulsion polymerization and are far from being general. Only models that are more or less
comprehensive in nature have been selected to be listed in Table 1. Overall, the models from Ray's
group are probably the most sophisticated and detailed among all available at the cost of a very complex
mathematical model structure. To keep these models tractable yet still rigorous, simplications have
been given by Saldivar et al. [4].
2. Development of emulsion polymerization simulation model
The objective here is to build a general model that covers the entire spectrum of important physical
chemical phenomena in emulsion polymerization. The comprehensive model consists of a set of differ-
ential equations as well as algebraic equations based on mass, energy and population balances. The
following sections cover each important aspect in emulsion polymerization.
The model was constructed based on a set of assumptions. These assumptions are:
1. Particle size is monodisperse. If particles are generated in stage I in a relatively short period of time,
this assumption holds. This assumption will not be valid if more emulsier is added into the system in
a semi-batch or CSTR operation.
2. Reactor is perfectly mixed.
3. The main loci of polymerization are the polymer particles. This assumption is usually justied unless
we deal with mini- or microemulsion cases.
2.1. Initiation
A water soluble initiator I, like potassium persulfate (KPS), is commonly used in emulsion polymer-
ization. When heated, it will decompose into two highly reactive free radicals R
o
z
, which then react with
Table 1
Models of emulsion homo-/copolymerization
Groups Remarks Systems
Ballard et al. [67] Copolymerization model for 01 systems
Broadhead et al. [131] Copolymerization model, CSTR operation, PSD, long
chain branching frequency
Sty/BD
Congalidis et al. [129] PSD, copolymerization model MMA, Sty
Dougherty [29,130] Comprehensive model, MWD, PSD, copolymerization Sty/MMA
Dube et al. [30] Long-chain branching, copolymerization model, PSD,
MWD
BD/AN
Forcada and Asua [146,151] Sequence length, PSD Sty/MMA
Giannetti et al. [162] Reviews on PSD
Giannetti [163] Copolymerization model
Guillot [47] Thermodynamic aspects in copolymerization Sty/AN
Hamielec et al. [78,132,164] Branching, crosslinking, PSD, MWD
Lichti et al. [165] Reviews on PSD, MWD
Lin et al. [71] Azeotropic composition, copolymerization model Sty/AN
Mead and Poehlein [134,135] Copolymerization model, PSD Sty/MA, Sty/AN
Min and Ray [1,62,166] Comprehensive model, PSD, MWD MMA
Nomura et al.
[66,147,149,150]
PSD, copolymerization model MMA/Sty
Penlidis et al. [167] CSTR reactor, PSD, MWD PVC
Penlidis et al. [103] Batch reactor, PSD, MWD VAc
Penlidis et al. [168] Dynamic and steady-state modeling of batch, semi-
batch and CSTR latex reactors, PSD, MWD
Penlidis et al. [3] Comprehensive review, copolymerization model, PSD,
MWD
PVC, VAc/PVC
Rawlings [169] Comprehensive model, PSD, MWD, CSTR stability MMA, Sty, VAc
Richards et al. [128] An updated version of the early model of Congalidis et
al. [129], copolymerization model, no micellar
nucleation assumed, MWD, PSD
MMA, MMA/Sty
Rawlings and Ray [45,138] An updated version of the early Min and Ray's model,
PSD, MWD, CSTR operation, studies on CSTR
oscillations
MMA
Saldivar et al. [4] Most recent version of Min and Ray's model, PSD,
MWD, copolymerization model
MMA/Sty, Sty/a-
methyl Sty, Sty/BD
Storti et al. [170] Copolymerization model AN/Sty/MMA
Storti et al. [171] Modeling of molecular weight distribution
Urretabizkaia et al. [172,173] Terpolymerization model MMA/BA/VAc
monomers available in the water phase
S
2
O
22
8
-
k
d
2SO
2z
4
(3)
SO
2z
4
1M -
k
p
R
z
1
(4)
where k
d
is the rate constant for initiator decomposition. The rate of initiation is:
R
I
= 2fk
d
[I]
w
(5)
Redox initiation can be used when emulsion polymerization is especially conducted at low temperatures.
A typical example of this is rubber production at 510 8C. A redox system includes a metal complex, a
reducing agent (RA) and an oxidizing agent (initiator). A typical initiation mechanism involving a redox
system as proposed by Anderson and Proctor [18] follows:
S
2
O
22
8
1Fe
21
-
k
1
SO
2z
4
1Fe
31
1SO
22
4
(6)
RA 1Fe
31
-
k
2
Fe
21
1x (7)
SO
2z
4
1M-
k
p
R
z
1
(8)
The mass balances for the reacting species involved in reactions (6)(8) are
dN
i
dt
= F
in
i
2k
1
[I]
w
N
Fe
21 2F
out
i
(9)
dN
RA
dt
= F
in
RA
2k
2
[RA]
w
N
Fe
31 2F
out
RA
(10)
dN
Fe
21
dt
= 2k
1
[I]
w
N
Fe
21 1k
2
[RA]
w
N
Fe
31 (11)
dN
Fe
31
dt
= k
1
[I]
w
N
Fe
21 2k
2
[RA]
w
N
Fe
31 (12)
dN
SO
2z
4
dt
= k
1
[I]
w
N
Fe
21 2k
p
[M]
w
N
SO
2z
4
(13)
where N stands for the number of moles of a particular species denoted by the subscript and F is the ow
rate of a particular reactant. The total mass of iron (both Fe
21
and Fe
31
) remains constant as can be seen
from reactions (6) and (7), and this leads to
dN
Fe
dt
=
dN
Fe
21
dt
1
dN
Fe
31
dt
(14)
and
N
Fe
= N
Fe
21 1N
Fe
31 (15)
If we assume that (dN
Fe
21
/dt) is negligible, through a mass balance on Fe
21
, the number of moles of Fe
21
can be obtained as:
N
Fe
21 =
k
2
N
Fe
[RA]
w
k
1
[I]
w
1k
2
[RA]
w
(16)
Finally, the total rate of initiation R
I
including both thermal decomposition of a water soluble initiator
and redox initiation is expressed as
R
I
= 2fk
d
[I]
w
1k
1
N
Fe
21
V
w
[I]
w
(17)
where V
w
is the total volume of water in liters.
2.2. Particle nucleation and growth
Particle nucleation by far is the most important phenomenon in emulsion polymerization. It has
generated a lot of research interest and remains the most discussed and sought-after subject in emulsion
kinetics. This is because not only the rate of polymerization is directly related to the total number of
particles but also particle size distribution is a key indicator of latex physical properties. Despite many
efforts made by various groups so far, the understanding of particle generation is poor, and the prediction
of particle number and size is still not very successful. There are several reasons for this.
1. The measurement of number and size of polymer particles which are in the region of a few hundred
angstroms presents extremely difcult experimental problems.
2. There are many complex microprocesses occurring simultaneously, e.g. radical absorption, precipita-
tion, coagulation, etc. and each microprocess itself is difcult to understand and model.
In the early studies, it was accepted that particles are generated by micelles absorbing radicals from
the water phase. This is the so-called micellar nucleation. Smith and Ewart [7] stated that the rate of
particle nucleation in the presence of micelles is controlled by the laws of diffusion. They then used
Eq. (1) to calculate the total particle number as a function of emulsier concentration and initiation rate.
A new particle nucleation mechanism was proposed by Hansen and Ugelstad [19] and Fitch and Tsai
[20,21] after the original SmithEwart's postulation. It was stated by these two groups that particles
could be generated by precipitated water phase oligomer radicals. This is the so-called homogenous
particle nucleation. It is now widely accepted that both phenomena coexist in most emulsion polymer-
izations. Both mechanisms are discussed below.
2.2.1. Micellar nucleation
There are basically two different theories that describe radical absorption by micelles or particles.
Smith and Ewart [7] postulated that this phenomenon is a diffusion process, however the Smith and
Ewart scheme (Eq. (1)) actually reects a collision process. Micellar nucleation can also be described as
a collision process [812,22]. The expression for the rate of micellar nucleation (R
c
) based on diffusion
and collision theories is given in Eqs. (18) and (19), respectively
R
c
= 4prD
w
[R
z
]
w
(18)
R
c
= 4pr
2
k
mp
[R
z
]
w
(19)
where D
w
in Eq. (18) is the diffusivity coefcient of the radicals in the water phase and k
mp
in Eq. (19) is
the mass transfer coefcient for water phase radical oligomers. In both equations, r is the radius of the
micelle/particle and [R
z
]
w
is the concentration of water phase radical oligomers. It is very obvious that
the only difference in the two equations is that the rate of absorption of radicals is proportional to the
micelle/particle radius according to diffusion theory, whereas according to collision theory, it is propor-
tional to the micelle/particle surface area.
Both diffusion and collision theories have been used to describe particle nucleation by a number of
authors. Hansen and Ugelstad [19,23] and Song and Poehlein [24,25] used the diffusion theory to
calculate the rate of radical absorption. Groups using collision theory include Fitch and Tsai
[20,21], Min and Ray [1] and Dickinson [26]. Generally speaking, it is difcult to distinguish
which theory is more advantageous over the other. Barrett [27] pointed out that radicals are
assumed to travel in a straight line in the collisional approach and this underestimates the
probability of collision with a particle. Fitch and Tsai [20,21] were the rst to use collision
theory to quantify the rate of radical capture, and in a later study, Fitch and Shih [28] measured the
radical capture rate for MMA seeded polymerization. Their experimental data support the conclusion
that R
c
is proportional to the rst power of particle radius, thus it should be considered as a diffusional
process. From a theoretical point of view, it is unlikely that the mass transfer coefcient k
mp
in Eq. (19) is
a constant, since its value depends on many factors like particle size, ionic strength of the medium,
radical chain length and surface charge of the particle. Therefore, Eq. (19) is a very simplied formula-
tion to describe the process. However, this does not imply that the diffusion approach is of any advantage
over the collision approach. As a matter of fact, the micellar particle nucleation process is neither a pure
diffusional nor a collisional process, but rather a combination of both. The approach adopted in this
paper is based on diffusion theory simply because there is more kinetic information available in the
literature.
Hansen and Ugelstad [19,23] were the only group that attempted to quantify effects of radical chain
length, particle size and surface charge density on the rate of particle absorption. They proposed the
concept of reversible/irreversible absorption, which postulated that an absorbed radical might escape out
of the particle many times before it is irreversibly absorbed. An improved version of Eq. (18) for the rate
of radical capture is:
R
c
= 4prD
w
[R
z
]
w
F (20)
F in Eq. (20) is an efciency factor for radical absorption. It takes the effects of radical solubility, particle
surface electrical potential and particle size into account. The detailed calculation of F is complicated
and requires many additional parameters that are difcult to obtain. The conclusions from Hansen and
Ugelstad's [23] study are: (1) when radicals are large and hydrophobic, absorption is irreversible, i.e.
F < 1; (2) when particles are small, the electrostatic effects are negligible and (3) if small radicals are
more hydrophilic, absorption is more likely to be reversible, i.e. F - 0: Though Hansen and Ugelstad's
[19,23] approach is more theoretically sound, the absorption efciency is nevertheless very difcult to
estimate, and its complexity makes it somewhat impractical to use. In this emulsion model, Eq. (18) is
used without implementing the capture efciency F. Instead, D
w
in Eq. (18) should be considered as a
lumped parameter which combines many factors rather than a pure diffusivity coefcient. This probably
is the reason why reported values of D
w
in the literature vary considerably.
Another approach based on collision theory, because it postulates that the rate of particle nucleation is
proportional to the total micelle surface area, would suggest that:
dN
p
dt
= N
A
k
cm
[R
z
tot
]
capt
r
mic
_ _
A
m
A
m
1eA
p
(21)
e in Eq. (21) represents the ratio of radical absorption between micelles and particles. A
p
is the total
surface area of particles, A
m
is the free micellar area, [R
tot
z
]
capt
is the concentration of water phase radicals
that can be absorbed (captured) by either micelles or particles, k
cm
is the rate constant for radical
absorption by micelles and r
mic
is the radius of a micelle. It is also assumed that each particle surface
is entirely covered by emulsier molecules. Several groups including Dougherty [29], Penlidis et al. [3]
and Dube et al. [30] used Eq. (21) to calculate N
p
. Parameters k
cm
and e have to be estimated from
experimental data. The free micellar area available for micelle formation (and hence, for particle
formation via micellar nucleation) can be calculated as
A
mic
= ([S]
t
2[S]
cmc
)S
a
V
w
N
A
2A
p
2A
md
(22)
where S
a
is the area covered by a single emulsier molecule. [S]
t
is the total concentration of emulsier,
and [S]
cmc
is the critical micelle concentration. The values of these three variables can be calculated for a
specic emulsier. A
md
is the total surface area of monomer droplets, and since A
md
is several orders of
magnitude smaller than the total surface area of particles/micelles, it can safely be omitted. The area of
particles A
p
can be calculated by the expression below, if a particle is considered spherical:
A
p
= (36pN
p
)
1=3
(V
p
)
2=3
(23)
2.2.2. Homogenous nucleation
Priest [31] rst observed that particles can still be formed without the presence of micelles (in this case
there is either no emulsier or its concentration is below the CMC). When radicals in the aqueous phase
propagate beyond their solubility (due to the continuous addition of monomer units), they become a
primary particle, also called a particle precursor. A primary particle is stabilized either by the initiator
charges (segments) at the chain ends or available emulsier in the system. Napper and Alexander [32]
later conrmed Priest's observation and described the homogenous particle nucleation qualitatively.
Fitch and Tsai [20,21] was the rst group that proposed a detailed mechanism for this self-nucleation
process and gave a quantitative calculation for the rate of homogenous particle nucleation. They
assumed that a water phase radical would travel a distance L before it becomes a primary particle.
The longer the distance, the more likely it will be absorbed by an existing particle. If it is absorbed by an
existing particle, no new particle is formed. These primary particles will subsequently undergo extensive
occulation. According to Fitch and Tsai's [20,21] postulation, the rate of particle nucleation is then
dN
p
dt
= R
I
2R
c
2R
f
(24)
where R
c
is the rate of absorption of radicals, R
I
is the rate of radical generation through initiation and R
f
is the rate of occulation. When there is sufcient emulsier in the system, primary particles are stable
and occulation is nearly negligible. This makes the rate of particle nucleation directly related to R
I
and
R
c
, and R
f
can be dropped from Eq. (24).
Fitch and Tsai [20] derived an expression for the rate of radical absorption (capture) as
R
c
= pr
2
p
LR
I
N
p
(25)
where L is the average distance of diffusion a water phase radical travels before it has grown to a size at
which it precipitates as primary particle and r
p
is the particle radius.
Eq. (25) suggests that the rate of radical absorption is proportional to the cross-section area of the
radical and the particle, therefore the radical absorption process can be described as a collision process.
A key parameter in Fitch and Tsai's model is the average distance L. Fitch and Tsai [20] demonstrated
that L is in the magnitude of 2.8 10
23
cm, and it is obvious that L is very sensitive to monomer
solubility in the water phase. In a later publication, Fitch [33] treated the radical absorption as a diffusion
process, and the overall rate of radical absorption was expressed as
R
c
= 4pr
p
D
w
N
p
[R
z
]
w
(26)
[R
z
]
w
is the radical concentration in the water phase and N
p
is the total number of particles. After the
original work on homogenous particle nucleation by Fitch and Tsai [20,21], a number of research groups
have attempted to quantify the homogenous particle nucleation process using different approaches. A
more rigorous model was proposed by Hansel and Ugelstad [19]. This group modeled the homogenous
nucleation by using a population balance on water phase oligomer radicals with chain length less than j
cr
,
which is a critical chain length beyond which an oligomer radical will precipitate and form a primary
particle. Using this approach, Hansen and Ugelstad [19] successfully calculated [R
z
]
w
, which is needed to
calculate the rate of radical absorption. In order to understand Hansen and Ugelstad's more rigorous
approach (see Section 2.2.3 for details), a visualization of all water phase phenomena is essential. Such
visualization will help one to establish a mass balance of oligomer radicals of various chain lengths.
Fig. 3 elucidates how a particle is formed either by homogenous particle nucleation or by micellar
absorption.
2.2.3. Mass balances for radicals in the water phase
The mass balance for radicals in the water phase is affected by radicals entering the water phase,
radicals leaving the water phase and reactions involving water phase radicals. A complete list of possible
reactions of radicals in the aqueous phase is given in Table 2.
(1) Radical generation by initiation. Chemical initiator decomposition is the major source for radicals
entering the water phase. As long as there is initiator present, there will always be radicals generated.
The mechanism and rate expressions have already been given in Section 2.1.
(2) Radical desorption from particles. Chain transfer reactions like transfer to monomer and to CTA
in a polymer particle will produce a monomeric or CTA radical. Such small radicals may desorb into the
water phase, especially when monomer or CTA radicals are more water soluble.
(3) Radicals captured by a micelle. If a micelle captures a radical, it then becomes a particle. This is
the so-called micellar nucleation discussed earlier. Most particles are generated this way.
(4) Radicals captured by a particle. Particles compete with micelles in absorbing water phase radicals.
The amount of radicals entering particles is proportional to the total particle surface area.
(5) Radical propagates with monomers. Primary radicals generated either by initiation or by chain
transfer reactions will propagate with monomer dissolved in the aqueous phase and grow into oligomers.
Oligomers produced in this step may form particles when they reach a critical size (as discussed earlier).
F
i
g
.
3
.
P
a
r
t
i
c
l
e
f
o
r
m
a
t
i
o
n
i
n
a
q
u
e
o
u
s
p
h
a
s
e
.
(6) Radical terminates with another radical. This reaction stops water phase radical growth and
produces oligomeric polymer. This reaction is usually considered unimportant by many groups.
(7) Radical reacts with impurities. Water-soluble impurities kill active radicals and generate a dead
(inert) molecule. The overall effect is the delay of the start of emulsion polymerization (induction time).
This reaction is usually neglected by most models in the literature. More details on impurity effects on
reaction kinetics can be found in Appendix B.
(8) Reactions involving small molecules. In general, chain transfer to small molecules has little effect
on the overall particle number and size. However, chain transfer to small molecules actually produces
emulsier-like oligomers which may have some effect in emulsier-free reactions.
(9) Radicals precipitate and form a particle. This is another source of particle nucleation (homo-
genous nucleation). The method used in this paper to model homogenous particle nucleation is analo-
gous to the postulation in Ref. [19], but more general by including reactions like chain transfer to CTA
and inhibitor, as shown in Fig. 3. One very important reaction which is not displayed in Fig. 3 is the
desorption process. The original Hansen and Ugelstad's [19] postulation is based on styrene emulsion
polymerization, a monomer with negligible desorption, therefore desorption was not included in their
model. To develop a general and yet comprehensive model, desorption was considered when construct-
ing the mass balances for radicals of various chain lengths, as illustrated below. The mass balances for
radicals with chain length from 1 to j
cr
21 are:
d[R
z
o
]
w
dt
= R
I
1
k
des
N
p
n
N
A
V
w
2k
pw
[M]
w
[R
z
o
]
w
2k
z
[R
z
o
]
w
[Z]
w
2k
tw
[R
z
o
]
w
[R
z
]
w
(27)
d[R
z
1
]
w
dt
= k
pw
[M]
w
[R
z
o
]
w
2k
pw
[M]
w
[R
z
1
]
w
2k
cp
[N
p
][R
z
1
]
w
2k
cm
[MIC][R
z
1
]
w
2k
z
[R
z
1
]
w
[Z]
w
2k
fcta
[R
z
1
]
w
[CTA]
w
2k
tw
[R
z
1
]
w
[R
z
]
w
(28)
Table 2
Water phase phenomena involving radicals
Process Results
Radicals entering the water phase
Initiation Major source of radical generation
Desorption Small radicals entering water phase
Radicals leaving the water phase
Radical captured by a micelle Micellar particle nucleation
Radical captured by a particle Does not change total particle number
Radical captured by a monomer droplet Negligible in conventional emulsion polymerization
Reactions involving radicals
Radical propagates with monomers No effect on total number of radicals
Radical termination Produces oligomeric polymer in water phase
Radical reacts with impurities Reduces total number of radicals in water phase
Radical reacts with small molecules Chain transfer reaction, produces monomeric radicals
Radicals form a particle Homogenous particle nucleation
d[R
z
j
]
w
dt
= k
pw
[M]
w
[R
z
j21
]
w
2k
pw
[M]
w
[R
z
j
]
w
2k
cp
[N
p
][R
z
j
]
w
2k
cm
[MIC][R
z
j
]
w
2k
z
[R
z
j
]
w
[Z]
w
2k
fcta
[R
z
j
]
w
[CTA]
w
2k
tw
[R
z
j
]
w
[R
z
]
w
(29)
d[R
z
j
cr
21
]
w
dt
= k
pw
[M]
w
[R
z
j
cr
22
]
w
2k
pw
[M]
w
[R
z
j
cr
21
]
w
k
cp
[N
p
][R
z
j
cr
21
]
w
2k
cm
[R
z
j
cr
21
]
w
[MIC]
2k
z
[R
z
j
cr
21
]
w
[Z]
w
2k
fcta
[R
z
j
cr
21
]
w
[CTA]
w
2k
tw
[R
z
j
cr
21
]
w
[R
z
]
w
(30)
In Eqs. (27)(30), k
pw
is the rate constant for propagation in the water phase (l/mol min), k
cp
, the rate
constant for radical capture by particles (l/mol min), k
cm
, the rate constant for radical capture by micelles
(l/mol min), k
des
, the rate constant for radical desorption (l/mol min), k
tw
, the rate constant for termination
in the water phase (l/mol min), k
fcta
, the rate constant for chain transfer to (water soluble) CTAs (l/mol min),
k
z
, the rate constant for reactions with inhibitor (l/mol min), [CTA]
w
, the CTA concentration in water
(mol/l), [M]
w
, the monomer concentration in water (mol/l), [MIC]
w
, the micelle concentration in water
(mol/l), [Z]
w
, the concentration of water soluble inhibitor (mol/l), [R
o
z
]
w
, the concentration of primary
radicals (without monomer unit) in the water phase (mol/l), [R
j
z
]
w
, the concentration of radicals of chain
length j in the water phase (mol/l), R
o,w
z
, the primary radical in the water phase, and R
j,w
z
is the radical of
chain length j in the water phase (j = 1 to j
cr
21).
In homogenous particle nucleation, a particle is formed by a precipitated radical whose chain length is
at the critical chain length j
cr
. The last propagation step involving a radical of chain length j
cr
21 and a
monomer can actually be considered as the particle formation step, thus the rate of homogenous particle
nucleation is expressed as:
d[N
p homo
]
dt
= k
pw
[M]
w
[R
z
j
cr
21
]
w
(31)
The concentration of all water phase radicals of various chain lengths can be expressed as:
[R
z
]
w
=
j
cr
21
j=0
[R
z
j
] (32)
In Eq. (32), R
z
is the decomposed initiator fragment when j = 0: Adding Eq. (27) to Eq. (30) gives:
d[R
z
]
w
dt
= R
I
1k
des
N
p
n
N
A
V
w
2k
pw
[M]
w
[R
z
j
cr
21
]
w
2k
cp
[N
p
][R
z
]
w
2k
cm
[MIC][R
z
]
w
2k
z
[Z]
w
[R
z
]
w
2k
fcta
[CTA]
w
[R
z
]
w
2k
tw
[R
z
]
2
w
(33)
If the steady-state hypothesis is applied to all radicals in the water phase, i.e. setting left-hand side of
Eqs. (27)(30) to zero, one obtains the following equations:
[R
z
o
]
w
=
R
I
1k
des
N
p
n=(N
A
V
w
)
k
pw
[M]
w
1k
z
[Z]
w
1k
tw
[R
z
]
w
(34)
[R
z
j
]
w
=
k
pw
[M]
w
k
pw
[M]
w
1k
cp
[N
p
] 1k
cm
[MIC] 1k
tw
[R
z
]
w
1k
z
[Z]
w
1k
fcta
[CTA]
w
[R
z
j21
]
w
(35)
Eq. (35) is valid for radicals with chain length of 1 to j
cr
21, hence Eqs. (34) and (35) can be back-
substituted into Eq. (31) and one can describe the rate of particle formation as follows:
d[N
homo
]
dt
=
(R
I
1k
des
N
p
n=(N
A
V
w
))k
p
[M]
w
k
pw
[M]
w
1k
tw
[R
z
]
w
1k
z
[Z]
w
_ _
j
cr
21
j=1
k
p
[M]
w
k
pw
[M]
w
1k
cp
[N
p
] 1k
cm
[MIC] 1k
z
[Z]
w
1k
fcta
[CTA]
w
1k
tw
[R
z
]
w
_ _
(36)
A close examination of the denominator of the rst term in Eq. (36) would lead to the assumption that
k
p
[M] qk
tw
[R
z
]
w
1k
z
[Z]
w
: This assumption can be justied based on the comparison displayed below:
k
pw
[M]
w
k
z
[Z]
w
k
tw
[R
z
]
w
10
2
10
3
10
22
10
23
10
22
10
23
10
23
10
25
10
7
10
9
10
210
10
212
If the denominator of Eq. (36) can be safely reduced to k
pw
[M]
w
, then Eq. (36) is simplied as:
d[N
homo
]
dt
= (R
I
1k
des
N
p
n=(N
A
V
w
))
j
cr
21
j=1
k
pw
[M]
w
k
pw
[M]
w
1k
cp
[N
p
] 1k
cm
[MIC] 1k
z
[Z]
w
1k
fcta
[CTA]
w
1k
tw
[R
z
]
w
_ _
(37)
Eq. (37) is the ultimate expression used to calculate homogenous particle nucleation. However, the total
radical concentration [R
z
]
w
still remains unknown. This can be obtained by applying the steady-state
hypothesis to Eq. (33), which yields
[R
z
]
w
=
2B 1
B
2
14(R
I
1k
des
N
p
n=(N
A
V
w
))k
tw
_
2k
tw
(38)
where
B = k
cp
N
p
=(N
A
V
w
) 1k
cm
[MIC] 1k
z
[Z]
w
1k
fcta
[CTA] (39)
Now, if we substitute [R
z
]
w
back into Eq. (37), the rate of homogenous particle nucleation can be readily
calculated. Eq. (37) is interesting from a statistical point of view, since the second term of Eq. (37)
represents the probability of a water soluble radical adding a monomer unit and is always less than 1.
Eq. (37) also reveals the fact that the more water soluble the radicals (which implies a higher j
cr
), the
lower the rate of homogenous particle nucleation. Similar conclusions were also conrmed in Ref. [24].
It should also be stated that in the calculation of the rate of homogenous particle nucleation, the rate
constants for radical absorption by micelles and particles, if based on diffusion theory, are
k
cm
= N
A
4pD
w
r
mic
(40)
k
cp
= N
A
4pD
p
r
p
(41)
r
mic
and r
p
are the radius for micelles and particles, and D
w
, D
p
are the diffusivities of radicals in the water
phase and inside the polymer particle, respectively. The values for both diffusivity coefcients vary
widely and are difcult to obtain. Values of D
w
for styrene and vinyl acetate can be found in
Refs. [24,34], respectively, yet caution should be exercised when citing these literature values. Song
and Poehlein [25] did an extensive investigation on homogenous particle nucleation for a number of
monomers including Sty, BA, MMA, VAc and VCl.
The calculation for the total concentration of micelles is now shown below. Each micelle is an
aggregation of a number of emulsier molecules. The concentration of micelles in mol/l is calculated by:
[MIC] =
A
mic
4pr
2
mic
N
A
V
w
(42)
A
mic
in Eq. (42) is the total surface area of all micelles which can be obtained from Eq. (22).
Other reacting species like CTAs and impurities also play a role in the particle nucleation process. The
overall effect of each individual species depends largely on their water solubility, which implies that
partitioning is very important.
Finally, in a conventional emulsion polymerization, i.e. when the emulsier concentration is above its
CMC, particles are generated by both homogenous and micellar nucleation, and the following expres-
sion calculates the rate of change of the total number of particles:
dN
p
dt
=
dN
p homo
dt
1
dN
p mic
dt
(43)
2.2.4. Recent modeling efforts on particle nucleation
Several investigations on particle formation were conducted by Gilbert's groups [16,35,36]. In one of
their most recent publications on particle nucleation, Maxwell et al. [36] stated that radical absorption by
particles is not a collisional process because: (1) particle surface charge density does not affect radical
absorption and (2) ionic strength of the medium exhibits no inuence on radical absorption. On the basis
of previous experimental evidence, Maxwell et al. [36] summarized their experimental ndings and
proposed a new model for particle formation. They claimed that the propagation of a water phase radical
to its critical chain length z is the rate-determining step in the radical absorption process, thus radical
absorption should be considered as a diffusional process. Gilbert's group also observed in their experi-
ments that water phase radicals with chain length less than the critical chain length z are reversibly
absorbed, i.e. small radicals can enter and escape a particle many times before they can be irreversibly
absorbed. This phenomenon directly indicates that the absorption efciency F is very small for radicals
with chain length less than z. Their statement is in agreement with Hansen and Ugelstad's [19] conclu-
sion that the radical absorption efciency is rather small for smaller radicals. It is now commonly
accepted that not all radicals of any size can be irreversibly adsorbed. This nding suggests some
modication of Eqs. (28) and (29). Radicals with short chain length (say, less than j
cr
/2) may not be
absorbed by micelles or particles. This implies that radicals which cannot be absorbed should not be
included in the calculation of [R
j
z
]
w
in terms such as k
cp
[N
p
][R
j
z
]
w
and k
cm
[MIC][R
j
z
]
w
in Eq. (29) ([R
1
z
]
w
will
not be absorbed in Eq. (28). Some groups have already implemented such a modication in their model
[30], assuming that radicals of chain length less than half of the critical chain length j
cr
do not prefer to be
captured. This assumption is an empirical approximation. Maxwell et al. [36] attempted to theoretically
calculate the minimum chain length z for which radicals can be captured. On the basis of thermo-
dynamics, Maxwell et al. [36] calculated this minimum z for styrene to be about 2. The commonly
accepted j
cr
for styrene is 35 (presented in Ref. [19]), therefore Maxwell's calculation is in agreement
with what Dube [30] have practiced in their modeling effort. Maxwell et al. [36] also calculated z for a
number of other monomers like BA, MMA, VAc and VCl. Their calculated z for MMA is 45 which is
much smaller than 1/2 of j
cr
for MMA (j
cr
for MMA has been reported up to 50). Though there are still
discrepancies in the actual value of z for a number of monomers, recent kinetic investigations reveal the
fact that not all radicals of any size can be irreversibly absorbed. It is thus reasonable to assume that z is
approximately equal to half the value of j
cr
for most monomers. Though this may seem to be a crude
measure, it is the best starting point at our current level of understanding of particle nucleation kinetics.
Kshirsager and Poehlein [37] attempted to validate Maxwell's et al. [36] postulation. They measured the
minimum chain length z for vinyl acetate using bombardment-mass spectroscopy. They reported a value
of 56 for z for VAc, which is very close to the theoretical prediction (78) given by Maxwell et al. [36]
based on thermodynamics. Kshirsager and Poehlein's [37] measurement of z indirectly suggested that
colloidal theory may not be applicable to describe the radical absorption process. In order for the
colloidal theory to be valid, the size of oligomer radicals must be quite large, and this is clearly not
supported by the experimental measurements from Kshirsager and Poehlein [37].
Table 3 is a summary of all models to date on particle nucleation.
2.2.5. Specic modeling considerations on particle nucleation
(1) Steady-state hypothesis on aqueous phase radicals. The diffusional approach this model adopted
was an improved version of Hansen and Ugelstad's [19] original postulation. After it was implemented
into the model, it was tested to reproduce Hansen and Ugelstad's calculations. The comparison of our
model's prediction (generated by using exactly the same parameter values and reaction conditions
indicated in Ref. [19]) with that from Ref. [19] is displayed in Fig. 4. It is very clear in Fig. 4 that
our model reproduced exactly the results given by Hansen and Ugelstad [19]. Fig. 4 also reveals the
effect of critical chain length on particle generation. More water-soluble monomers have a higher critical
chain length, which consequently results in a slower particle nucleation rate. This is simply because
more water soluble monomers (oligomers) are more likely to stay in the water phase for longer periods of
Table 3
Summary of particle nucleation models
Model Authors Formula
Collisional process Fitch and Tsai [20,21] R
c
= 4pr
2
p
k
cm
N
p
[R
z
]
w
Diffusional process Hansen and Ugelstad [19,23] R
c
= 4pr
p
D
w
N
p
[R
z
]
w
Diffusion/propagational Maxwell et al. [36] R
c
k
p
[M]
w
[R
z
]
w
Collisional/empirical Dougherty [29] and Penlidis et al. [3] R
c
total micelle surface area
Surface coverage Yeliseyeva and Zuikov [226] R
c
surface charge density
Colloidal Penboss et al. [180] DLVO-type colloidal
time before they are absorbed by micelles or existing particles. In Hansen and Ugelstad's original study,
the validity of the steady-state hypothesis was also carefully examined. Hansen and Ugelstad have
demonstrated that the prediction generated under the steady-state hypothesis differs from that generated
without the steady-state hypothesis only in the rst 10 s, whereas afterwards both curves become
indistinguishable. Therefore it is reasonable to state that the steady-state hypothesis can be safely
used. Asua et al. [38] also independently veried the validity of the steady-state hypothesis. More
detailed studies on both steady-state and nonsteady-state particle nucleation can be found in Refs.
[24,25].
(2) Model discrimination. More details were considered in modeling particle nucleation by Hansen
and Ugelstad [23]. Water phase radicals derived from initiator decomposition have an ionic initiator
fragment attached, which makes these radicals surface active. In contrast, radicals generated by chain
transfer to monomers and CTAs do not have an ionic chain end. Hansen and Ugelstad [23] treated these
two types of radicals differently by using different propagation rate constants for propagation and
absorption. Maxwell et al. [36] also assumed that radicals of any chain length without ionic initiator
attached can be irreversibly absorbed. It is obviously of theoretical advantage to take this effect into
modeling considerations, but it inevitably increases the model's complexity because many additional
rate constants must be estimated. Although Hansen and Ugelstad [23] presented a model by considering
the effect of ionic initiator at the radical chain end on absorption, desorption and propagation, they did
not present any estimated values for all the rate constants used in their model. Recently, de Arbina et al.
[39] compared three models of various levels of complexity, which are: (1) a simple model that does not
consider the effect of ionic initiator at the radical chain end; (2) a model that considers the effect of ionic
initiator at the radical chain end, but assumes no irreversible absorption for radicals of chain length less
Fig. 4. Model testing on homogenous particle nucleation.
than z and (3) a model that considers the effect of ionic initiator at the radical chain end, but assumes
irreversible absorption. They found that all three models tted the experimental data in a similar way
with little difference and they should be considered equivalent. They thus concluded that there is no
advantage gained by increasing the model's complexity.
2.3. Monomer partitioning in emulsion homopolymerization
Another important reason why emulsion polymerization is such a complex system is that all reaction
ingredients partition themselves in all phases. To keep track of concentrations in each phase throughout
the reaction is a difcult task simply because there is a lack of satisfactory theory that can fully describe
the partitioning phenomena. To make matters worse, monomer partitioning becomes even more compli-
cated in co-/terpolymerization when several monomers as well as polymer coexist in the same phase. An
extensive literature search reveals that most research groups focus on solving this problem empirically,
which implies that monomer concentration in both water and polymer particle phases is measured
experimentally. This approach was very popular in early studies and is still the most widely used method
nowadays. It is relatively easy to measure monomer concentrations in all phases. From a modeling point
of view, a partitioning coefcient is very easy to implement in a model. However, this method becomes
increasingly cumbersome and unreliable when more than one monomers are involved in the reaction.
Additionally, it is desirable for a model to be capable of predicting monomer concentration for a new
emulsion system without determining the partitioning coefcient each time. In such a case, a theoretical
approach is certainly advantageous. It was rst demonstrated by Morton et al. [40] that monomer
partitioning can be described by solving the thermodynamic equilibrium equations of participating
species in each phase. The principle behind it is the thermodynamic law which states that at equilibrium
state, the chemical potential of each species is the same in all phases under the assumption that
thermodynamic equilibrium can be quickly established without mass transfer limitations. Later on,
Guillot [41] presented a similar thermodynamic equilibrium equation for copolymers. It is obvious
that calculating monomer concentration based on thermodynamic equilibrium is more theoretically
sound but it certainly requires more parameters that are difcult to obtain. Both empirical and theoretical
approaches have their own merits and drawbacks. Table 4 gives a comparison of the two approaches.
Sections 2.3.1 and 2.3.2 describe each approach in greater detail. Because the empirical approach is
rather straightforward, emphasis will be placed on the theoretical approach.
2.3.1. Empirical approach
On the basis of a mass balance for monomer, it is easy to express the concentration of monomer in the
Table 4
Comparison of methods for monomer partitioning in emulsion polymerization
Empirical approach Theoretical approach
Advantages Easy to implement Theoretically sound, general, can be applied to
partitioning involving n number of monomers
Disadvantages Lack of generality, limited to
specic monomer systems, lack of
information for copolymer systems
Numerically difcult to solve sets of nonlinear
equilibrium equations; parameters are not available
for many systems and are difcult to estimate
polymer particle phase as
[M]
p
=
N
m
V
p
(44)
V
p
is the total volume of all polymer particles, which can be expressed as:
V
p
=
x
c
r
p
1
1 2x
c
r
m
_ _
N
m
o
MW
M
(45)
The number of moles of monomer can be calculated as N
m
= (1 2x)=N
mo
; where N
mo
is the initial
number of moles of monomer and x
c
is the so-called critical conversion at which monomer droplets
disappear. Substitute Eq. (45) and the one for N
m
back to Eq. (44) to obtain the expression for the
monomer concentration in the particles as:
[M]
p
=
(1 2x
c
)
1 2x
c
(1 2r
m
=r
p
)MW
M
(46)
In Eqs. (45) and (46), r
m
and r
p
are the density for monomer and polymer, respectively. Many physical
properties of the emulsion system will change drastically at the critical conversion, thus x
c
can be
experimentally determined by measuring the sudden change of vapor pressure, turbidity, conductivity,
etc. In stages I and II, the polymer particle composition remains relatively the same, so it is reasonable to
assume that the monomer concentration also remains the same, thus Eq. (46) provides a formula for
calculating the monomer concentration in stages I and II. Eq. (46) was used by Dougherty [29] in his
model for predicting monomer partitioning.
In interval III, x
c
,x ,1, and similarly:
[M]
p
=
(1 2x)
(1 2x 1xr
M
=r
p
)MW
M
(47)
Eq. (47) is the expression for the monomer concentration in the particles after the monomer droplets
disappear, and x is the overall conversion.
Another way of obtaining monomer concentration in the particle phase is the use of overall empirical
partitioning coefcients. A partition coefcient is dened as:
k
wp
=
[M]
w
[M]
p
(48)
k
wp
is the ratio of the monomer concentration in the water phase over monomer concentration in the
polymer phase. k
wp
values for certain monomer systems and for a very limited number of copolymer
systems are available in the literature.
Despite the simplicity, the above two methods are limited in certain ways. First of all, both methods
lead to a constant monomer concentration inside the particles throughout stages I and II. This is unlikely,
because particle size denitely affects monomer partitioning. Additionally, the empirical method
becomes more and more implausible when it is applied to multicomponent systems. Partitioning coef-
cients for each monomer must be estimated. These limitations of the empirical approach can be over-
come by using the theoretical approach, which is discussed in Section 2.3.2.
2.3.2. Theoretical approach: thermodynamic equilibrium
The theoretical approach is established based on two principles: (1) the Flory [42] and Huggins [43]
mixing model of small molecules and long polymer chains; and (2) the thermodynamic law of chemical
potential at equilibrium. Applying these two principles to emulsion homopolymerization, it is relatively
easy to write down the chemical potential of monomer in the polymer particle phase. Morton et al. [40]
rst proposed the following expression
DG
i;particle
=RT = m
i
= m
m;p
1m
s;p
= ln(1 2f
p
) 1f
p
1x
ip
f
2
p
1
2s

V
m
RTr
p
(49)
where f
p
is the volume fraction of polymer inside the particle, s, the interfacial tension, r
p
, the particle
radius, x
ip
, the interaction parameter between monomer i and its homopolymer, and

V
m
is the molar
volume of monomer i.
The chemical potential of monomer i in Eq. (49) contains two terms: the free energy due to mixing
m
m,p
, and the interfacial free energy due to the increasing surface area, m
s,p
. For monomer i in monomer
droplets, a similar term can be written as:
DG
i;droplet
=RT = m
i;d
= m
m;d
1m
s;d
= ln(1 2f
p
) 1f
p
1x
ip
f
2
p
1
2s

V
m
RTr
d
(50)
The chemical potential of monomer i in monomer droplets contains the same terms as in a polymer
particle. In emulsion homopolymerization, there is only one component in a monomer droplet (mono-
mer), therefore, f
p
is zero, and this would reduce the rst three terms to zero. The last term represents the
free energy of interface between monomer droplets and aqueous phase. The diameter r
d
of a monomer
droplet is in the range of 10
4
10
5
A
. This is rather large compared to the diameter of particles, which is

about 10
2
10
3
A
, hence, the contribution fromthe last term in Eq. (50) is very small, and this leads to the
free energy of monomer i in a monomer droplet to be near zero. As to monomer dissolved in the aqueous
phase, its free energy can be written as
DG
i;aqueous
RT
= m
i;w
= ln
[M]
w
[M]
w;sat
_ _
(51)
[M]
w,sat
is the solubility of monomer in the water phase at saturation. In stages I and II, the aqueous phase
is saturated with monomer, so [M]
w
is equal to [M]
w,sat
, and this makes DG
i,aqueous
/RT equal to zero.
According to thermodynamic equilibrium, the chemical potentials of monomer in each phase must be
equal, e.g.
DG
i
RT
_ _
particle
=
DG
i
RT
_ _
droplet
=
DG
i
RT
_ _
aqueous
(52)
Since both DG
i,droplet
/RT and DG
i,aqueous
/RT are zero, Eq. (49) can be rewritten as:
DG
i;particle
=RT = ln(1 2f
p
) 1f
p
1x
ip
f
2
p
1
2s

V
m
RTr
p
= 0 (53)
Eq. (53) is the well-known theoretical expression for monomer partitioning in stages I and II when there
is existence of monomer droplets. Eq. (53) is a nonlinear equation and the root f
p
must be solved for
iteratively. While in stage III, when all monomer droplets disappear, the aqueous phase is no longer
saturated with monomer, which implies that [M]
w
,[M]
w,sat
, and the thermodynamic equilibrium must
be rewritten as
DG
i
RT
_ _
particle
=
DG
i
RT
_ _
aqueous
(54)
subsequently
ln(1 2f
p
) 1f
p
1x
ip
f
2
p
1
2s

V
m
RTr
p
= ln
[M]
w
[M]
w;sat
_ _
(55)
There are two unknowns in Eq. (55), the polymer volume fraction f
p
and the monomer concentration in
the water phase [M]
w
. [M]
w
can be rewritten based on its denition
N
m
2V
p
(1 2f
p
)r
m
=MW
M
V
w
= [M]
w
(56)
In order to obtain the roots of f
p
and [M]
w
, Eqs. (55) and (56) must be solved simultaneously at each
iteration (time step).
Eq. (53) reects the contribution of mixing as well as of surface energy to the free energy of the
monomer. Compared to the empirical approach described in Section 2.3.1, Eq. (53) represents a more
sophisticated approach. It also indicates that the chemical potential of monomer i in the particle phase is
affected by three factors, i.e. (1) the miscibility of polymer and monomer i through x
ip
; (2) the interfacial
tension effect through s, where the value of s is inuenced by the choice of emulsier, monomer and
ionic strength; and (3) the size of particle through r
p
. Clearly Eq. (53) is more theoretically sound than its
empirical analog, however the additional sophistication is obtained at the expense of introducing para-
meters like s and x
ip
that are difcult to obtain. There is only a very limited number of sources that give
values of s and x
ip
. In early studies, Morton et al. [40] used a tensiometer to determine interfacial
tension for styrene. They reported a value of 4.5 dyn/cm for s for styrene. The value of the interaction
parameter x
ip
they reported for styrene monomer and its homopolymer is 0.43. Gardon [22] extended
Morton's original work to more monomer systems and several monomer swelling ratios (swelling ratio
is an indication of how much monomer can be dissolved in the polymer particle), and parameters x
ip
and
s, as well as solubility were determined in his work. He found that the values of x
ip
and s for styrene
and MMA are the most consistent, despite variation of particle size, emulsier level and temperature.
Gardon [22] reported exactly the same values for both x
ip
and s for styrene as those given by Morton
et al. [40]. For MMA, the value of x
ip
is 0.585 ^0.005, and s is 1.6 ^0.4 dyn/cm. Gardon [22]
considered the interaction coefcient x
ip
to be between 0.2 and 0.6 and s to range from 1 to 30 dyn/
cm. One interesting nding discussed in Ref. [22] is that s should be lower in stage I during which most
particles are small and completely covered by emulsier. When in stage II, particles grow and are less
covered by emulsier, and such partial coverage will result in a larger value of s. However, Gardon [22]
offered no solution on how to quantify the partial coverage effect. Perhaps the most important conclusion
in Ref. [22] is that monomer concentration in particles varies little with conversion during stages I and II
as long as thermodynamic equilibrium is maintained. This conclusion is in agreement with experimental
observations from many other groups.
Though the thermodynamics of monomer partitioning has been discussed by a number of groups
[4446], only very few actually applied it to predict monomer concentration changes in particles during
the entire course of the reaction. Guillot [41] developed a computer simulation program to describe
monomer partitioning in AN/Sty copolymerization in emulsion based on thermodynamics principles.
Guillot [41] presented expressions for monomer chemical potential in all phases for both homopoly-
merization and copolymerization. Compared to Eq. (53), Guillot's [41] expression for chemical potential
included more terms to take the elasticity effect and electric work into account. Such practice never-
theless makes solving the thermodynamic equations much more complicated and introduces additional
parameters. Unfortunately, Guillot [41] did not give values for the parameters used in his equation.
Additional results for AN/Sty and VAc/BA copolymerization were given in a later publication by the
same author [47].
It would be illustrative if Eq. (53) is plotted, as demonstrated in Fig. 5. In Fig. 5, the Gibbs free energy
of monomer (styrene is used as an example in this case) in stages I and II is plotted against f
p
for various
particle sizes. The interaction parameter x
i,p
for styrene and its homopolymer is 0.4. Sodium dodecyl
sulfate was used as emulsier and the surface tension s is 4.5 dyn/cm. Particle radius ranges from 25 to
400 A
. The root of Eq. (53) lies on the grid line at which the Gibbs free energy is zero. It is very clearly
demonstrated in Fig. 5 that the larger the particle size, the lower the volume fraction of polymer, which
indicates that larger particles are more swollen with monomer.
When in stage III, Eqs. (55) and (56) replace Eq. (53) and they must be solved simultaneously. The
Fig. 5. Gibbs free energy in stages I and II in styrene homopolymerization in emulsion.
Gibbs free energy expressed in Eq. (55) is plotted in Fig. 6 at four different levels of conversion, while
other conditions are the same as those in Fig. 5. In Fig. 6, the root of Eqs. (55) and (56) f
p
resides on the
line at which the Gibbs free energy is zero, and ranges from 30 to 90%. We can observe that all four
curves are nearly vertical in their respective root region, which indicates that even a very small variation
of the independent variable (f
p
in this case) will result in large changes in the Gibbs free energy.
Furthermore, the left-hand side of Eq. (56) must remain positive.
2.4. Desorption
As was mentioned in Section 2.2, absorbed radicals may desorb from polymer particles. The driving
forces behind desorption are likely to be the radical's solubility, mobility and several other factors. The
smaller the polymer particle, the easier the radical may desorb. Desorption is an important phenomenon
in emulsion polymerization. Deviations from the early SmithEwart model are largely caused by
desorption. It is commonly accepted nowadays that desorbed radicals are monomeric radicals which
are the product directly from chain transfer to monomer. These radicals (without an initiator fragment
attached) are small and mobile. Once they desorb out of a particle, they are likely to stay in the vicinity of
the particle and may be reabsorbed into the particle. In Section 2.2, an expression was already given for
the rate of desorption:
r
des
=
k
des
N
p
n
N
A
V
w
(57)
Fig. 6. Gibbs free energy in stage III in styrene emulsion homopolymerization.
The key thing in Eq. (57) is to evaluate the rate constant for desorption, k
des
. Expressions for k
des
have
been developed by a number of research groups including Harada et al. [48], Nomura et al. [49,50],
Ugelstad and Hansen [14], Nomura and Harada [51] and Asua et al. [52]. Harada et al. [48] used a
stochastic approach and derived an expression for k
des
given below
k
des
= n 1
k
p
[M]
p
m
d
d
2
p
12D
w
d
_ _
21
k
fm
k
p
1
k
fcta
[CTA]
p
k
p
[M]
p
1
R
I
(1 2 n)
N
p
k
p
[M]
p
n
_ _
k
p
[M]
p
(58)
where d is
d =
m
d
D
p
m
d
D
p
16D
w
(59)
m
d
in both Eqs. (58) and (59) is a partitioning coefcient, which is equivalent to k
wp
21
in Eq. (48), d
p
is the
average diameter of polymer particles, D
p
is the diffusivity of radical inside the polymer particles and D
w
is the diffusivity of radical in the water phase. Eq. (58) indicates that desorption should be considered
consisting of several complex processes including chain transfer to monomer, CTA and/or other small
molecules. Harada et al. [48] tested their expression for several monomers and achieved good agree-
ment. Nomura et al. [49] simplied Harada's equation by assuming the majority of desorbed radicals are
monomeric radicals derived from chain transfer to monomer, and if n p0:5; Eq. (58) is then reduced to:
k
des
=
12D
w
d
m
d
d
2
p
_ _
k
fm
k
p
(60)
Eq. (60) was later on veried by Nomura and Harada [51] using a deterministic approach and it was
successfully applied to vinyl acetate and vinyl chloride. When using Eq. (62), users must be cautious that
it was derived based on several assumptions, like no re-entry of desorbed radicals, instantaneous
termination inside polymer particles, negligible water phase termination and polymer particles contain-
ing at most one radical. These assumptions are subject to further justication at higher conversion levels
when the interior of a polymer particle becomes more viscous or for monomers that are water soluble
(i.e. where water phase termination cannot be neglected).
Ugelstad and Hansen [14] also investigated the desorption kinetics and proposed a similar expression
for k
des
k
des
=
k
fm
k
/
p
12D
w
(a 1D
w
=D
p
)d
2
p
_ _
(61)
where k
fm
is the chain transfer to monomer rate constant, k
/
p
is the propagation (re-initiation) rate
constant for radicals generated by the transfer reaction, and a is the partition coefcient for monomer
radicals between water phase and particles (a is identical to m
d
of Eq. (60)).
Both Nomura's and Ugelstad's expressions have been used by a number of authors for the emulsion
polymerization of vinyl acetate [5358]. Asua et al. [52] examined Nomura's model and found it
deviated fromexperimental ndings reported by Adams et al. [59]. Asua et al. [52] suspected the validity
of the assumptions of instantaneous termination inside polymer particles and negligible water phase
termination utilized in Nomura's model. It was considered in Asua's postulation that a reabsorbed
monomer radical is more likely to redesorb instead of undergoing instantaneous termination, and
additionally, a desorbed monomer radical like vinyl acetate may propagate in the water phase. Finally,
re-entry of desorbed radicals was also included in Asua's model. In their attempt to overcome previous
limitations, Asua et al. [52] derived the following expression
k
des
= k
fm
[M]
p
K
o
bK
o
1k
p
[M]
p
(62)
where
K
o
=
12D
w
=m
d
d
2
p
1 12D
w
=m
d
D
p
(63)
b =
k
pw
[M]
w
1k
tw
[R
z
]
w
k
pw
[M]
w
1k
tw
[R
z
]
w
1k
cp
N
p
=(N
A
V
w
) 1k
cm
[MIC] 1k
z
[Z]
w
(64)
b in Eq. (64) is the probability of a monomeric desorbed radical undergoing water phase propagation or
termination, and k
cp
is the rate constant for radical absorption by polymer particles. When b = 0; a
desorbed radical will be reabsorbed back to the particle. When b = 1; a desorbed radical will undergo
water phase termination or propagation instead of being reabsorbed. Asua et al. [52] further rewrote
Nomura's model as:
k
des
= k
fm
[M]
p
K
o
K
o
n 1k
p
[M]
p
(65)
A close comparison of Eq. (62) with Eq. (65) indicates that they are very close to each other. Eqs. (62)
and (65) become identical when both b and n become zero. This implies that both Asua's and Nomura's
models give the same prediction for monomers with very low water solubility (b - 0), however, in the
case of more water soluble monomers like vinyl acetate or methyl acrylate (b .0), Nomura's model
predicts a higher rate constant for desorption. Since b has a value between 0 and 1, the previous
Nomura's and Ugelstad's models can be considered as a special case of Asua's model when n is in
the same range of 01. Asua et al. [52] stated that their model is advantageous over Nomura's model
because it is more general. Adams et al. [59] used a g-radiolysis relaxation technique to directly
determine the desorption rate constant k
des
, and their measured k
des
is better predicted by Asua's
model. After evaluating all the models discussed earlier, Asua's model for desorption (Eq. (62)) was
selected for implementation due to its apparent advantages.
It is worth mentioning that all cases discussed earlier assumed a constant d (Eq. (59)) in their
expressions. This is unlikely the case, especially at the late stages of polymerization, because the interior
of a polymer particle becomes very viscous, and hence the desorption process becomes diffusion
controlled. To quantify the diffusion limitation on desorption, D
p
should be treated as a function of
viscosity. This subject has not been discussed so far in a quantitative way. Only Friis and Hamielec [60]
used an empirical expression to calculate the decrease in D
p
:
D
p
= D
po
0:0017x 1
1 2x
1 20:19x
_ _
2
_ _
(66)
Expression (66) was used for vinyl acetate emulsion polymerization. To develop a more general model,
the free-volume theory was adopted in our approach. In a way similar to the treatment of diffusion
controlled rate constant for propagation, D
p
is calculated as:
D
p
= D
po
exp 2B
1
V
F
2
1
V
Fpcrit
_ _ _ _
(67)
The modeling of desorption was recently reviewed by Gilbert's group in Refs. [61,62]. They
presented a detailed model for emulsion polymerization including radical desorption and reab-
sorption. It was assumed in their model that only monomeric radicals could desorb. This assump-
tion was justied with the fact that the solubilities and diffusivity of dimeric and trimeric radicals
are much lower than monomeric radicals. Casey et al. [61] treated initiator-derived radicals
differently from transfer reaction derived radicals, and this was done by using a separate rate
constant for absorption for the two different types of radicals. Casey et al. [61] claimed that a desorbed
monomeric radical in the aqueous phase is either being reabsorbed or terminates with other radicals,
while propagation of a monomeric radical in the water phase is negligible. Though Casey et al. [61] gave
a detailed discussion on the justication of the assumptions used in their model, their expression for the
rate constant for desorption is exactly the same as that given by Ugelstad and Hansen [14] and Nomura
and Harada [51]. Morrison et al. [62] applied Casey's et al. [61] theory to styrene, where the rate constant
for desorption was estimated at various particle sizes and monomer concentrations. It was found that the
rate constant for desorption was mostly affected by particle size. However, Morrison's et al. [62]
estimation must be treated with caution because their experimental (measuring) error is quite large
(50%).
2.5. Average number of radicals per particle
To calculate the average number of radicals per particle, n; a population balance must be written for
the number of particles with zero, one, two and more radicals per particle. The solution of these
population balance equations can be obtained under two steady-state hypotheses:
1. The rate of radical ow into the water phase (radical desorption from polymer particles plus initiation)
equals the rate of radical ow out of the water phase (radical absorption and termination).
2. The rate of formation of particles with n free radicals is equal to the rate of disappearance of particles
with n free radicals.
The rst hypothesis can be expressed as:
k
des
N
p
n=(N
A
V
w
) 1R
I
= k
cp
[R
z
]
w
N
p
=(N
A
V
w
) 1k
cm
[R
z
]
w
N
p
=(N
A
V
w
) 1k
tw
[R
z
]
2
w
1k
z
[R
z
]
w
[Z]
w
(68)
The second hypothesis is expressed as dN
n
=dt = 0; i.e. the formation and disappearance rates of particles
containing n radicals are the same. N
n
is the number of particles containing n radicals. There are four
sources that affect N
n
:
1. When a radical enters a particle that has n 21 radicals, the rate of formation is:
k
cp
[R
z
]
w
[N
n21
]
2. When a radical desorbs from a particle that has n 11 radicals, the rate of disappearance is:
k
des
(n 11)[N
n11
]
3. When two radicals terminate inside a particle that has n 12 radicals, the rate is:
k
t
(n 12)(n 11)[N
n12
]=v
p
where v
p
is the average volume of a single polymer particle.
4. When a radical reacts with inhibitors/impurities in a particle that has n 11 radicals, the rate is:
k
z
[Z]
p
n[N
n11
]
A recurrence formula can be subsequently set up to express the second hypothesis, which is:
k
cp
[R
z
]
w
[N
n
] 1k
des
(n)[N
n
] 1k
t
[(n)(n 21)=v
p
][N
n
] 1k
z
[Z]
p
(n)[N
n
]
= k
cp
[R
z
]
w
[N
n21
] 1k
des
(n 11)[N
n11
] 1k
t
[(n 11)(n 12)=v
p
][N
n12
] 1k
z
[Z]
p
(n 11)[N
n11
]
(69)
Applying the steady-state hypotheses to the particle balances gives a series of simultaneous algebraic
equations for particles with zero, one, two or more radicals. Assuming a monodisperse particle size
distribution, O'Toole [63] solved the series of algebraic equations. The solution of the recursive equation
(Eq. (69)) was proposed as a modied Bessel function to give an expression for n: Ugelstad et al. [64]
proposed the following expression as an approximate numerical solution
n =
a
m 1
2a
m 11 1
2a
m 12 1
2a
m 13 1
(70)
with
a =
k
cp
[R
z
]
w
V
p
N
A
N
p
k
t
(71)
and
m =
(k
des
1k
z
[Z]
p
)V
p
N
A
k
t
N
p
(72)
The above solution (Eq. (72)) given by Eq. (70) has been conrmed with experimental data in Ref. [65].
Nomura et al. [66] summarized the results of a number of experimental investigations fromthe literature,
compared them to the predictions of the Bessel function solution and concluded that this solution is
valid.
Several authors have also attempted to obtain an exact solution for the particle population balance
equations without using the steady-state hypothesis. Ugelstad and Hansen [14] claimed that their solu-
tion was not signicantly different from O'Toole's solution for the same problem under the state-state
hypothesis. Ballard et al. [67] made similar statements in their studies. More recently, Li and Brooks [68]
gave an explicit analytical solution to the population balance equations for the particles, and the authors
claimed that their solution represents an improvement over Eq. (70).
Another approach for obtaining n is to develop partial differential equations representing the number
of particles containing zero, one, two or more radicals as functions of both time and volume. This will
give a distribution of particles with different numbers of radicals. The average number of radicals per
particle in this case can be obtained by summing the elements of all particle size distributions using the
following formula
n =
1
i=0
iN
i
(V)
1
i=0
N
i
(V)
(73)
N
i
(V) is the number of particles of volume V containing i number of radicals. This approach has been
applied successfully by Min and Ray [69], Lichti et al. [70], Lin et al. [71] and Chen and Wu [72].
Nevertheless, Eq. (73) does not have apparent advantages over Eq. (70). Additionally, Eq. (73) is more
computationally expensive, therefore Eq. (70) was adopted in our emulsion model.
2.6. Overall mass balances and molecular weight development
The emulsion polymerization model developed herein is based on mass balances for each reacting
species. All these mass balance equations are in the form of ordinary differential equations (ODE) and
are solved simultaneously by an ODE solver. This section lists all the reactions involved in a homo-
polymerization in emulsion. Mass balances for each reactant, energy balances and molecular weight
development equations for both linear and branched systems are presented. The reactions that follow
take place inside polymer particles.
Propagation:
R
z
r
1M-
k
p
R
z
r11
(74)
Termination by combination:
R
z
n
1R
z
m
-
k
tc
P
n1m
(75)
Termination by disproportionation:
R
z
n
1R
z
m
-
k
td
P
n
1P
m
(76)
Chain transfer to monomer:
R
z
r
1M-
k
fm
P
r
1R
z
1
(77)
Chain transfer to polymer:
R
z
r
1P
s
-
k
fp
P
r
1R
z
s
(78)
Chain transfer to chain transfer agent (CTA):
R
z
r
1CTA -
k
fcta
P
r
1CTA
z
(79)
Chain transfer to inhibitor/impurities:
R
z
r
1Z-
k
z
P
r
1Z
z
(80)
Terminal double bond polymerization:
R
z
n
1P
s
-
k
p
p
R
z
n1s
(81)
Internal double bond polymerization:
R
z
n
1P
s
-
k
p p
p
R
z
n1s
(82)
The quantity of each species in the reaction mixture can be calculated through a differential equation
based on the species mass balance. Since this package also simulates homopolymerization in CSTR or
semi-batch reactors, input and output ow rates of any reactant species are taken into account in all the
kinetic, molecular weight development and energy balance differential equations. The following is a list
of all these equations. In these equations, N
j
is the number of moles of species j, F
j
in
is the inlet molar
owrate of species j, F
j
out
is the outlet molar owrate of species j, V
p
is the total volume of polymer
particles, and V
w
is the total volume of the aqueous phase.
2.6.1. Mass balance equations for each species
Monomer: although it is considered that the polymer particle phase is the main locus of polymeriza-
tion, to make the model complete, monomer consumption in the aqueous phase is also included
dN
M
dt
= F
in
M
2
k
p
[M]
p
N
p
n
N
A
V
p
2k
pw
[M]
w
[R
z
]
w
2F
out
M
(83)
Initiator:
dN
I
dt
= F
in
I
2k
d
[I]
w
V
w
2F
out
I
(84)
Polymer: polymer produced in both polymer particle and aqueous phases is included
dN
Polymer
dt
= F
in
M
1
k
p
[M]
p
N
p
n
N
A
V
p
1k
pw
[M]
w
[R
z
]
w
2F
out
M
(85)
Inhibitor: the mass balance below is the total consumption of inhibitor in both polymer particle and
aqueous phases. It has been noted that oil-soluble and water-soluble inhibitors affect the reaction kinetics
in a very different way. More details on this subject are given in Appendix B.
dN
z
dt
= F
in
z
2k
z
[Z]
p
N
p
n
N
A
V
p
2k
z
[Z]
w
[R
z
]
w
2F
out
z
(86)
CTA: Like other small molecules in emulsion polymerization, CTA may partition itself between the
polymer particle phase and the water phase depending on its solubility. The mass balance below is the
overall consumption rate of CTA in both phases:
dN
CTA
dt
= F
in
CTA
2
k
fcta
[CTA]
p
N
p
n
N
A
V
p
2k
fcta
[CTA]
w
[R
z
]
w
2F
out
CTA
(87)
Finally, the change of the total volume of all polymer particles must be followed in order to calculate
concentrations of reaction species. Note that polymer particles grow continuously in stages I and II when
there is constant monomer supply from monomer droplets, while in stage III, polymer particles shrink
due to the density difference between polymer and monomer. Two separate equations are given below,
during stages I and II
dV
p
dt
=
MW
M
k
p
[M]
p
N
p
n=N
A
1MW
M
k
pw
[M]
w
[R
z
]
w
f
p
r
p
(88)
whereas in stage III:
dV
p
dt
=
MW
M
k
pw
[M]
w
[R
z
]
w
r
p
2
MW
M
k
p
[M]
p
N
p
n
V
p
N
A
1
r
m
2
1
r
p
_ _
(89)
Energy balance:
d(V
latex
DH)
dt
= F
in
H
1
k
p
[M]
p
N
p
n
N
A
V
p
(2DH
p
) 1k
p
[M]
w
[R
z
]
w
(2DH
p
) 2UA(T
R
2T
jacket
) 2F
out
H
(90)
Isothermal case:
d(V
latex
DH)
dt
= 0 (91)
Enthalpy:
H = H
o
1C
p
(T
R
2T
o
) (92)
2.6.2. Molecular weight development
The average radical concentration inside polymer particles is dened as:
[R
z
]
p
=
N
p
n
N
A
V
p
(93)
Instantaneous number average molecular weight:
M
n
=
MW
M
(t 1b=2)
(94)
Instantaneous weight average molecular weight:
M
w
=
MW
M
(2t 13b)
(t 1b)
2
(95)
where
t =
k
td
(k
p
[M]
p
[R
z
]
p
)
(k
p
[M]
p
)
2
1
k
fm
k
p
1
k
fcta
[CTA]
p
k
p
[M]
p
1
k
z
[Z]
p
k
p
[M]
p
(96)
b =
k
tc
(k
p
[M]
p
[R
z
]
p
)
(k
p
[M]
p
)
2
(97)
To calculate the accumulated molecular weight averages for both linear and branched systems, the
method of moments is used. From the denition of moments, the ith moment of the polymer molecule
distribution is
m
i
=
1
r=1
r
i
[P
r
] (98)
when i = 0; m
0
= [P]; which is the total concentration of polymer molecules.
The accumulated number average and weight average molecular weights are dened by the following
two equations:
M
n
= MW
M
m
1
m
0
(99)
M
w
= MW
M
m
2
m
1
(100)
0th moment:
1
V
p
d(V
p
m
0
)
dt
= t 1
b
2
2
k
pp
p
m
1
k
p
[M]
p
2
k
p
p
m
0
k
p
[M]
p
_ _
k
p
[M]
p
[R
z
]
p
2m
0
V
out
(101)
1st moment:
1
V
p
d(V
p
m
1
)
dt
= k
p
[M]
p
[R
z
]
p
t 2
k
td
[R
z
]
p
k
p
[M]
p
_ _
2m
1
V
out
(102)
2nd moment:
1
V
p
d(V
p
m
2
)
dt
= g 12 1 1
k
p
p
m
1
1k
pp
p
m
2
k
p
[M]
p
_ _
j
l
1
k
tc
[R
z
]
p
k
p
[M]
p
j
2
l
2
_ _
k
p
[M]
p
[R
z
]
p
2m
2
V
out
(103)
with
g = 1 1
R
I
k
p
[M]
p
[R
z
]
p
1
k
fm
k
p
1
k
fcta
[CTA]
p
k
p
[M]
p
1
k
z
[Z]
p
k
p
[M]
p
(104)
l =
k
t
[R
z
]
p
k
p
[M]
p
1
k
fm
k
p
1
k
fcta
[CTA]
p
k
p
[M]
p
1
k
z
[Z]
p
k
p
[M]
p
1
k
fp
m
1
k
p
[M]
p
(105)
j = 1 1
k
fm
k
p
1
k
fcta
[CTA]
p
k
p
[M]
p
1
k
z
[Z]
p
k
p
[M]
p
1
k
p
p
m
1
k
p
[M]
p
1
R
I
k
p
[M]
p
[R
z
]
p
1
k
fp
k
p
1
k
pp
p
k
p
_ _
m
2
[M]
p
(106)
The branching frequencies for both trifunctional and tetrafunctional branching averages are calculated as
follows:
dV
p
m
0

B
n3
dt
= k
p
[R
z
]
p
V
p
k
fp
k
p
m
1
1
k
p
p
k
p
m
0
_ _
(107)
dV
p
m
0

B
n4
dt
= k
p
[R
z
]
p
V
p
k
pp
p
k
p
m
1
(108)
2.7. Reaction kinetics at high conversion levels
It is known that emulsion polymerization has its own unique kinetic characteristics compared to bulk/
solution polymerization. Friis and Hamielec [73] was the rst group that studied the gel effect on the
kinetics of styrene homopolymerization in emulsion. The existence of diffusion controlled termination
was suggested by the observation that n starts to increase rapidly beyond 30% conversion (the conver-
sion level at which monomer droplets start to disappear). This is due to the rapid decrease of k
t
. The same
group observed similar characteristics in MMA [74] and VAc homopolymerization [75]. In all cases,
they used an empirical expression for the diffusion controlled k
t
. In their expression, the diffusion
controlled k
t
was tted with a polynomial expression shown below
k
t
= k
t0
exp(2A
1
x 1A
2
x
2
1A
3
x
3
) (109)
where k
t0
is the chemically controlled rate constant for termination. Gel effect in emulsion polymeriza-
tion was also observed and treated in a quantitative way by Harris et al. [76]. They conducted seeded
emulsion polymerization experiments of styrene. By assuming n equal to 1/2, the rate constant for
propagation k
p
was backcalculated. It was found that k
p
decreased sharply after 85% conversion, and
that the diffusion controlled k
p
could be adequately described by free volume theory. Sundberg et al. [77]
used the approach to describe the diffusion controlled termination and propagation in styrene and MMA
homopolymerization in emulsion. Sundberg et al. [77] did an extensive study on gel effect phenomena in
emulsion polymerization of styrene and MMA. The effects of particle size, number of particles, initiator
concentration as well as temperature on k
t
and k
p
were studied in detail. It was concluded that gel effect
phenomena do exist in emulsion polymerization of some monomer systems, having a profound effect on
reaction kinetics. Both Harris et al. [76] and Sundberg et al. [77] used the same approach based on free
volume theory and successfully described the gel effect observed in their experiments. Other authors that
used the same approach include Hamielec and MacGregor [78] and Dube et al. [30]. In this paper, the
free volume theory was used to describe diffusion controlled kinetic phenomena. Detailed implementa-
tion has already been discussed in Ref. [79].
It has been known that the interior of polymer particles is very viscous. For many monomer systems,
there is usually at least 6070% polymer in the particles at the very start of the reaction. During stages I
and II, the viscosity of particles remains relatively constant due to the constant supply of monomer from
monomer droplets. Inside polymer particles, the termination reaction becomes diffusion controlled once
polymerization starts. The initial value of the rate constant for termination is low but relatively constant
in stages I and II. Such postulation was conrmed by other groups. Harris et al. [76] and Sundberg et al.
[77] claimed that molecular masses are typically controlled by transfer reactions at low conversion range
and are relatively constant over much of the conversion history. Once all monomer droplets are depleted,
more and more monomer is turning into polymer inside polymer particles and the interior of particles
becomes more viscous. Under such circumstances, the rate constant for termination will begin to drop
until it reaches its lower limit (the reaction diffusion controlled k
t
). As polymerization proceeds, latex
particles may turn glassy if the reaction is conducted below the polymer mixture glass transition
temperature, and consequently, propagation becomes diffusion controlled. For some monomer systems,
like EA and MA, it has already been demonstrated in Ref. [79] that the reaction kinetics in bulk
polymerization is greatly affected by reaction diffusion control. In emulsion polymerization of these
monomers, the dominance of reaction diffusion control phenomenon will be enhanced further.
Though the gel effect is signicant in some cases, not every monomer system exhibits the same
characteristics. The gel effect is not usually observed in polymerizations of monomers with high
desorption rate, like vinyl acetate or vinyl chloride. Molecular weight in this case will be controlled
by chain transfer reactions rather than termination. The gel effect phenomenon is more dominant when n
becomes higher than 1, i.e. when there are more than one radicals in one particle. In this case, the rate of
polymerization is independent of the overall particle number, and emulsion polymerization can be
considered as `pseudo-bulk' polymerization.
3. Emulsion polymerization simulation package/database overview
3.1. General description
Similar to the previously developed bulk and solution package [79,80], the emulsion simulation
package consists of two major programs: MASTER and DATABASE. A detailed description of the
bulk/solution package was given in Ref. [79]. This section intends to give a very brief description of
the emulsion package with emphasis on the characteristics that are unique to emulsion polymerization.
Fig. 7 displays the recipe design screen for emulsion polymerization. It will ask the user to dene the
reaction formulation, i.e. select the type and amount of monomer, initiator, emulsier, etc. along with
their own units. There is a prompt at the bottom center of the screen that will lead the user to other input/
output screens of the program.
3.1.1. Computational options screen
Fig. 8 displays the computational options screen. On this screen, users can set up the initial reactor
temperature, select reactor temperature prole options (isothermal, adiabatic, and user-dened tempera-
ture prole) and input the heat transfer parameter (UA is in cal/K min, U is the overall heat transfer
coefcient, A is the heat transfer area). Maximum simulation time and conversion level will stop the
simulation at the specied point. Numerical spacing and tolerance parameter determine the computation
speed and accuracy. Polymerization induction time is another option. It is a useful option when water-
soluble inhibitor is present in the emulsion polymerization system.
At the bottom half of this screen, different model options are listed. Each model option can be
activated or deactivated in a toggled mode. At present, diffusion controlled propagation and termination
according to Refs. [81,82], and `reaction diffusion' according to Refs. [8385] are options available.
Monomer partitioning can be solved either empirically (partition coefcients) or theoretically (via
thermodynamic equations, see Section 2.3). `SSH' stands for steady-state hypothesis for radicals.
Other options include variable initiator efciency, monomer soluble/water soluble inhibitors, seeded
polymerization, emulsier-free cases, etc. It should be stated here that there is no universal rule to
determine what options must be selected for a specic polymerization system. It is up to the user to
decide. The basic principle behind this is that a combination of options should be tried to achieve the best
simulation performance and results. As always, experience and the process data/understanding are the
guidelines.
Fig. 8. Computational options screen.
Fig. 7. Reaction formulation screen.
3.1.2. Output les screen
Fig. 9 displays the list of all output les. This table is also an indication of what prediction capability
this model has. In this screen, users can choose the output les to be generated in order to graphically
view how a specic variable changes with either conversion or time. Each output le can be viewed by
using the SHOW utility program supplied with the package. Users can also import output les to other
graphics software, like Excel, Quattro, Lotus, Grapher, etc. to view and plot the simulation results.
Fig. 9. Output les screen.
Fig. 10. Flow options screen.
3.1.3. Flow options screen
Fig. 10 displays the ow options screen. If the polymerization is carried out in the semi-batch mode,
the ow options screen will let users input the specic inow and outow rates as well as their starting
and ending ow time. For both CSTR and semi-batch cases, and for each reactant in the system, ow
rate, and starting and ending times must be specied. Users again have the option to choose different
units to meet simulation needs.
3.2. Package database overview
The DATABASE program organizes and updates all the information for each species in the package
database, such as physical properties, kinetic rate constants and other parameters. The database provides
all the necessary information the Master program needs to perform a simulation. Sections 3.2.1 and 3.2.2
will give a brief description for each reacting species.
3.2.1. Monomer systems
The database currently includes ve monomer systems as well as other reaction components such as
initiators, emulsiers, chain transfer agents (CTAs), etc. All monomers are listed in Table 5.
3.2.2. Other components
An emulsion polymerization typically involves monomer, initiator, emulsier, water, and CTA added
to produce polymer with desired molecular weight. Inhibitors are commonly used in monomer storage
and transportation to prevent self-polymerization. If a monomer is not puried before use or the reactor
is not properly deaerated before a run, the residual inhibitor or oxygen present in the reaction will kill
radicals. Thus, an induction period is often observed. Table 6 lists the initiators, emulsiers, CTAs and
inhibitors currently available.
3.3. Description of database items
The main body of the database set for each monomer system consists of its physical and chemical
properties. Physical properties for a monomer and its polymer are molecular weight, density, glass
transition temperature, heat capacity, etc. Rate constants for various events in emulsion, like desorption,
absorption, chain transfer reactions, and propagation and termination in both water and polymer phases
represent monomer chemical reactivities. All these characteristics can in principle be determined.
However, for most monomers, their kinetic rate constant values reported in the literature vary over a
Table 5
Monomer list
Symbol Chemical name
EA Ethyl acrylate
MMA Methyl methacrylate
Sty Styrene
Vac Vinyl acetate
BA Butyl acrylate
wide range, despite the fact that monomers like MMA, styrene and VAc have been the research focus for
many years. The lack of understanding of the polymerization mechanism, inconsistent parameter esti-
mation techniques, lack of design of experiments and different techniques employed in the measure-
ments make it very difcult to reach a commonly accepted single value of one particular rate constant.
The current database at least provides a set of starting values for monomer characteristics. A set of
monomer characteristics for styrene is listed below as an example.
Physical and chemical properties of styrene
Items Value Unit
MW
M
100.12 g/mol
k
p
4.703 10
11
exp(29805/RT) l/mol min
2DH 19.27 kcal/mol
T
gm
167.1 K
V
fcrit
3.5819 exp(22125.7/RT)
V
fm
0.025 10.001(T 2T
gm
)
r
m
0.949 20.00128(T 2273.15) g/cm
3
C
pm
430 cal/kg K
k
t
1.04619 10
10
exp(22950.45/RT) l/mol min
k
tc
(%) 100
B 1.0
d 0.001 l/g
A 6.2 A
s 5.85 A
j
c
120
k
fm
1.48678 10
12
exp(217 543/RT) l/mol min
k
fp
0 l/mol min
k
p
p
0 l/mol min
k
pp
p
0 l/mol min
r
p
1.11 g/cm
3
C
pp
437.5 cal/kg K
Table 6
Other database components
Initiators Emulsiers CTAs Inhibitors
Lauroyl peroxide Fenopon Co436M Carbon tetrachloride Oxygen
Potassium persulfate Sodium dodecyl
sulfate (SDS)
n-Dodecyl mercaptan Hydroquinone
Hydroxy peroxide Aerosol MA/AMA t-Dodecyl mercaptan DPPH
a
Ammoniun persulfate Aerosol OT/AOT p-Octyl mercaptan Methyl ethyl hydroquinone
Sodium persulfate GDMA
b
a
2,2-Diphenyl-1-picrylhydrazyl hydrate.
b
Glycol dimercapto acetate.
(continued)
Items Value Unit
T
gp
378 K
V
fp
0.025 10.00048(T 2T
gp
)
k
3
0.00153 exp(14 805/RT)
m 0.5
n 1.75
A 1.552
k
pw
4.703 10
11
exp(29805/RT) l/mol min
k
tw
1.04619 10
10
exp(22950.45/RT) l/mol min
D
w
1.76 10
29
dm
2
/min
D
p
1.76 10
212
dm
2
/min
j
cr
5
k
wp
1400
s 4.5 dyn/cm
l 0.4
[M]
w
0.005 mol/l of water
It must be mentioned that in the list above, the rate constants for propagation and termination in the
water phase could be different from their corresponding values in the polymer phase. This is true
especially for polar monomers whose reactivity is affected by reaction media considerably. In the
example of the styrene database, these values are identical. However, to account for media effect on
other monomers (like carboxylic acids, acrylonitrile, etc.), rate constants for propagation and termina-
tion in the water phase should have their own separate values.
Other components, like initiators, emulsiers, etc. have their own database. Tables 7 and 8 are
examples of a set of database items for potassium persulfate and sodium dodecyl sulfate, respectively.
4. Simulation of styrene emulsion homopolymerization
An emulsion polymerization model has been developed according to the kinetic scheme described in
Section 2. The database associated with the model was also developed in parallel. The next step is to test
the model with a set of monomers with different emulsion polymerization characteristics. In emulsion
polymerization, monomers are usually classied into three categories, i.e.
Case I: monomers with high water solubility and signicant desorption, like vinyl acetate, vinyl
chloride, etc.
Case II: monomers with low water solubility and negligible desorption, like styrene, butadiene, etc.
Case III: monomers that exhibit signicant gel effect like MMA.
In the following sections, monomers in each category are tested. Such tests will verify our model's
validity. In our model testing, all model predictions are generated by the same database. This means that
once the database for a particular polymer system is set, there will be no more parameter tuning, i.e. the
database becomes `read only'.
Styrene has been long considered the most thoroughly studied monomer in emulsion polymerization
because it ts into the SmithEwart case II kinetic scheme well. Gardon [811] in a series of papers
distinguished three intervals in styrene emulsion homopolymerization in batch. According to the Smith
Ewart case II postulation, the average number of radicals per particle remains constant at 0.5 throughout
stages I and II of the reaction. However, this is not always true. Betaille et al. [86] conducted a series of
experiments of styrene homopolymerization and observed that n is greater than 0.5 throughout the
reaction. They attributed this to gel effect. Sudol et al. [87,88] noticed in their studies on making
monodispersed polystyrene latex seeds that n is 1/2 at the beginning of the reaction but it increases to
as high as 10 at about 87% conversion. Styrene has very low water solubility and negligible desorption,
and the gel effect is pronounced in styrene emulsion polymerization. The polymer phase is rather viscous
at the beginning of the reaction due to the presence of a high percentage of polymer in the particles, and
hence it is reasonable to consider that the termination reaction is diffusion controlled from the very
beginning. This gel effect leads to a much lower initial value for the rate constant for termination. Friis
and Hamielec [65,73,75] were the rst group that studied the gel effect in emulsion homopolymerization
of several monomers including styrene. The gel effect phenomenon was modeled by using an empirical
expression for the rate constant of termination. A similar approach was also adopted by Morbidelli et al.
[89] in their modeling effort. Friis and Hamielec [73] showed that n starts to increase quite early in the
reaction. The gel effect affects not only the rate of polymerization but also the molecular weight. James
and Piirma [90] gave molecular weight trends for styrene as well as MMA throughout the entire reaction.
The trends show that the molecular weight of styrene starts to increase at the end of stage II, and that is
exactly when the gel effect starts to dominate. Similar statements were also made by Piirma et al. [91].
Many papers have been reviewed to study the kinetics of styrene emulsion homopolymerization and
collect kinetic as well as experimental data. Table 9 lists useful papers that focus on styrene emulsion
polymerization kinetics.
Table 7
Database items for potassium persulfate
Items Value Unit
MW
M
271.3 g/g mol
k
d
1.524 10
18
exp(233 320/RT) l/mol min
f 0.5
Table 8
Database items for sodium dodecyl sulfate
Items Value Unit
MW
M
288.38 g/g mol
CMC 0.008 mol/l
S
a
3 10
217
dm
2
Micelle radius 2550 A
4.1. Model testing results

4.1.1. Bataille et al. [86] and Bataille and Gonzalez [92]
Bataille et al. [86] conducted emulsion homopolymerization of styrene at 60 8C. Five runs were
performed, and the reported experimental conditions and recipes are summarized in Table 10.
All ve runs were simulated using the recipes/conditions listed in Table 10. Simulation results for
conversion are displayed in Figs. 1113. In all ve runs, there are large but consistent discrepancies
between our model predictions and reported conversion data. An investigative check on all runs suggests
there is a systematic trend that all experimental data lag behind model predictions. To identify the culprit
of the discrepancies, the reported experimental setup and procedure were closely examined. It was
reported by Bataille et al. [86] that all reactions were conducted in a 1 l glass reactor and samples
Table 9
Papers on styrene emulsion polymerization kinetics
Authors Remarks
Asua et al. [174] Estimation of rate constants for radical absorption and desorption
Bataille et al. [86,92,93]
a
Full conversion data, studies on redox system
Blackley and Sebastian [96,97]
a
Experimental data, redox system
Canegallo et al. [94]
a
On-line densitometer measurements
Campbell [175] Experimental data, modeling
de La Rosa et al. [95]
a
Calorimetry measurements
Giannetti [176] Particle size distribution, modeling, kinetic parameter information
Harada et al. [13]
a
Full conversion range experimental data, modeling
Hawkett et al. [35] Radical absorption efciency
James and Piirma [90] Molecular weight development, gel effect studies
Lichti et al. [2] Particle size distribution
Mayer et al. [98]
a
Particle nucleation
Miller et al. [177] Molecular weights
Morton et al. [178] Kinetic parameters
Morbidelli et al. [89] Modeling, studies on gel effect and ionic strength
Nomura et al. [179] CSTR operation
Penboss et al. [180] Seeded emulsion polymerization, particle nucleation
Piirma et al. [91] Molecular weight development and kinetic studies
Said [181] Molecular weights
Salazar et al. [99]
a
Molecular weight control policy studies
Sudol et al. [87,88] Seeded emulsion polymerization, monodispersed particle latex
a
Paper contains data that are used for model testing.
Table 10
Reaction recipes for Bataille et al. [86] (all numbers are in wt%, amount of water is xed at 100%)
Recipe Run 1 Run 2 Run 3 Run 4 Run 5
Potassium persulfate 0.24 0.24 1.00 0.24 1.00
Sodium dodecyl sulfate 0.67 0.67 0.67 2.00 2.00
Styrene 40 30 30 30 30
Fig. 11. Simulation of styrene emulsion homopolymerization at 60 8C (runs #1 and #4).
Fig. 12. Simulation of styrene emulsion homopolymerization at 60 8C (runs #2 and #5).
Fig. 13. Simulation of styrene emulsion homopolymerization at 60 8C (run #3).
Fig. 14. Simulation of styrene emulsion homopolymerization at 60 8C (model testing on particle generation).
were taken periodically by using a pipette. Though monomer was washed prior to the reaction to
remove inhibitor, glass reactors are known to have poor seals and oxygen leakage during the
reaction was very likely. Furthermore, when samples were taken using a pipette, the reactor
contents were exposed to the atmosphere, and large quantities of oxygen entered the reactor.
Finally, in Bataille's et al. [86] experiment setup, nitrogen was not pumped through the reactor to keep
oxygen away. It was very likely that there was severe oxygen leakage in the reaction. Since oxygen can
partition itself in both water and oil phases, its effect on reaction kinetics is an induction time (as
appeared in runs #3 and #5) as well as a hindered reaction rate (in all runs). A detailed
discussion of inhibitor effects of emulsion polymerization kinetics is given in Appendix B. Oxygen
has a much less pronounced effect on particle generation, therefore, a much better agreement between
model predictions and experimentally measured total particle number per liter of water is obtained
(Fig. 14).
The same group continued their research on styrene emulsion homopolymerization with focus on the
effect of various redox systems on reaction kinetics [92,93]. Nitrogen was pumped through the reaction
mixture to avoid oxygen. Simulation results from the runs from Ref. [92] are displayed in Figs. 15 and
16. Predictions for conversion are now in better agreement with reported conversion data. The better
agreement is largely due to the absence of oxygen in the vessel. Predictions for weight average mole-
cular weight are also in good agreement with reported experimental data.
Fig. 15. Simulation of styrene emulsion homopolymerization at 60 8C (water: 700 g, styrene: 300 g, SDS: 14.28 g, KPS:
7.07 g).
4.1.2. Canegallo et al. [94]
On-line densitometer results in emulsion polymerization were reported by Canegallo et al. [94].
Problems associated with the on-line measurements such as thermal stability of samples, gas bubbles
in the line, sample calibration, etc. were discussed in detail. Their experimental data (on-line measure-
ment of conversion) for styrene homopolymerization are presented in Fig. 17. Styrene homopolymer-
ization was carried out with and without inhibitor removal (TBC: t-butyl catechol). Apparently, the run
without inhibitor removal is slower than the one with puried monomer. An initial induction time about
1015 min was observed, this induction time was taken into account and the model delivered reliable
predictions for both runs. The agreement is satisfactory.
4.1.3. de La Rosa et al. [95]
Recently, calorimetric techniques are used to obtain on-line measurements of conversion in emulsion
polymerization. Since emulsion polymerization proceeds with a heat evolution, the measured heat
release from the reaction can be translated into rate of polymerization. de La Rosa et al. [95] reviewed
the use of calorimetry techniques and conducted a set of runs for styrene batch emulsion polymerization
at 50 8C at various levels of emulsiers and initiators. In their rst set of runs, the initiator concentration
was set at 0.0046 mol/l with varying emulsier levels between 0.05 and 0.07 mol/l. Model testing results
for these three runs are displayed in Figs. 1820. Good agreement was obtained in all cases.
The second series of styrene emulsion polymerization experiments were conducted at xed emulsier
concentration with varying initiator concentration, ranging from 0.0046 to 0.011 mol/l. de la Rosa et al.
[95] did not report conversion proles but used capillary hydrodynamic fractionation (CHDF) to
Fig. 16. Simulation of weight average molecular weight in styrene emulsion homopolymerization at 60 8C.
Fig. 17. Simulation of emulsion homopolymerization of styrene at 50 8C (monomer: 90 g, water: 603 g, KPS: 1.05 wt%, SDS:
3.753 g).
Fig. 18. Simulation of styrene emulsion homopolymerization at 50 8C (M: 196.53 g, water: 393.06 g, KPS: 0.0046 mol/l, SDS:
0.05 mol/l).
0.06 mol/l).
0.07 mol/l).
determine particle number. Reported total particle numbers per liter of water for four runs in the second
set of experiments are displayed in Figs. 2124. It should be noted that initiator concentration does not
have a very drastic effect on particle nucleation. That is the reason why the nal N
p
values in Figs. 2124
are relatively close to each other. Nevertheless, model predictions for all the experiments are in a close
range within experimental measurement error, considering the difculty associated with particle number
measurement determination, and hence model predictions should be considered satisfactory.
4.1.4. Blackley and Sebastian [96,97]
Blackley and Sebastian [96,97] studied the inuence of various emulsiers and inorganic electrolytes
on emulsion homopolymerization of styrene and copolymerization of styrene and acrylic acid. Two
homopolymerization runs were carried out under identical reaction condition. Fig. 25 displays their
reported conversion prole including a replicated run along with model predictions. The experimental
data reproducibility is good as can be seen in Fig. 25. The model successfully followed the conversion
points throughout the entire run. Although the gel effect is present from the very beginning, the
commonly observed S-shape conversion versus time curve in bulk/solution polymerization does not
appear here. This is possibly due to the relative constant particle monomerpolymer composition during
stages I and II.
4.1.5. Mayer et al. [98]
Mayer et al. [98] carried out styrene emulsion homopolymerization at changing temperatures. During
the rst 165 m of the reaction, the reactor temperature was kept at 50 8C, whereas after this point, the
temperature was raised to 75 8C for about 80 min. The measured conversion is shown in Fig. 26. It can be
seen that in the rst temperature region, the reaction reached 90% conversion within 1 h and a limiting
Fig. 21. Simulation of styrene emulsion homopolymerization at 50 8C. Prediction of total number of particles.
Fig. 25. Simulation of styrene emulsion homopolymerization at 45 8C (Sty: 100 g, water: 200 g, KPS: 0.5 g, SDS: 5 g).
conversion was present. In the next hour, polymerization was nearly halted due to the glassy state of the
latex particles. When temperature was raised to 75 8C, the elevated temperature brought the nal
conversion level higher up to 95%. In the model testing of this set of data, the programming temperature
option was activated. This option enables the model to simulate the same reaction using the reported
temperature prole. The model gives good prediction of conversion in both the initial temperature region
as well as in the elevated temperature region. The model also predicts the nal particle diameter of
67.2 nm which agrees very well with the actual (transmission electron microscopy was used) measure-
ment of 68 nm.
4.1.6. Harada et al. [13]
Harada et al. [13] performed detailed studies on styrene emulsion homopolymerization over a wide
range of reaction conditions. They also developed a mathematical model to simulate all the
experiments they conducted. Harada's et al. [13] experiment results conrmed the fact that styrene is
a typical case II monomer as Smith and Ewart's theory depicted. In all their runs, stage I ends at around
14% of conversion and stage II ends around 43%. This model predicts the end of stage II at a
conversion level of about 35%. Additionally, the monomer concentration inside polymer particles
was determined as 5.45 mol/l. This measurement has conrmed that our model's prediction of
5.8 mol/l is reasonable.
A set of ve runs were simulated and the results are presented in Fig. 27. The effect of emulsier
concentration on reaction kinetics is clearly shown. Higher emulsier concentration results in higher rate
of polymerization. In all ve runs, model predictions agree well with measured conversion data.
Fig. 26. Simulation of styrene emulsion homopolymerization (styrene: 4 mol/l, KPS: 0.0075 mol/l, SDS: 0.132 mol/l).
4.1.7. Salazar et al. [99]
It is known that polymer produced in emulsion polymerization has very high molecular weight
compared to its bulk/solution counterpart. In order to control physical properties of the nal product,
CTAs are commonly used in emulsion polymerization. Salazar et al. [99] performed some studies on
molecular weight control in emulsion polymerization of styrene using various types of CTAs. It is
known that in the mercaptan group, CTAs with more carbon units in the backbone have lower water
solubility. A CTA with low water solubility resides predominantly in the monomer droplets at the
beginning of the reaction. As the reaction proceeds, it must diffuse from the monomer droplets through
the aqueous phase into the polymer particles. A CTA with lower water solubility usually encounters
higher diffusion resistance, thus the effectiveness of a CTA depends not only on its chemical
nature but also on its diffusivity. Salazar et al. [99] listed the apparent chain transfer constant
(dened as the ratio of the rate constant for chain transfer to CTA over that for propagation) for
a number of CTAs with numbers of carbon atoms ranging from 9 to 13. It can be concluded that
the higher the number of carbon atoms in a CTAs backbone (this implies that this CTA has a
lower water solubility), the lower the chain transfer constant is. Salazar et al. [99] used t-nonyl
mercaptan (with 9 carbon atoms) and n-dodecyl mercaptan (with 12 carbon atoms) in their experiments.
The chain transfer constant for these two CTAs is 1.95 and 0.31, respectively. Though it is of theoretical
advantage to use the `true' chain transfer constant for the CTAs and then incorporate diffusion limita-
tions for each CTA into the model, such practice is difcult due to the lack of information of certain
parameters like diffusivity of CTAs in different phases. In contrast, the semi-empirical approach of using
the àpparent' chain transfer coefcient provides an easy and effective way of modeling emulsion
Fig. 27. Simulation of styrene emulsion homopolymerization at 50 8C (water: 1 l, monomer: 572 ml, [KPS]: 0.0046 mol/l of
water).
polymerization with CTAs, and its usefulness is conrmed in the simulation of experiments presented in
Ref. [99].
Two homopolymerization runs of styrene were carried out by Salazar et al. [99], with t-nonyl
mercaptan and n-dodecyl mercaptan. These two runs were carried out under the same conditions and
recipe, except for the use of different CTAs. Figs. 28 and 29 display model predictions for conversion
and unswollen polymer particle size (`dry' polymer particles without residual monomer) for the run with
n-dodecyl mercaptan added, while Figs. 30 and 31 are the predictions for the run with t-nonyl mercaptan
added. The agreement between model predictions and measured conversion and particle size is good.
The results from both runs are very similar to each other, indicating the fact that the effects of both CTAs
on the rate of polymerization and polymer particle size are very close to each other. Perhaps the biggest
difference for these two CTAs is the way they affect the molecular weight proles shown in Figs. 32
and 33. As mentioned before, t-nonyl mercaptan has higher water solubility and lower diffusion resis-
tance, therefore its concentration in the polymer particles is higher compared to n-dodecyl mercaptan.
The more predominant presence of t-nonyl mercaptan in the polymer particles results in a relatively
lower molecular weight (both number and weight average) in the early stages of polymerization. As the
reaction proceeds, more CTA is consumed and its concentration in the polymer particles gradually goes
down, and subsequently polymer of higher molecular weight is produced. This trend is clearly shown in
Figs. 32 and 33 for both number average and weight average molecular weights.
In contrast, n-dodecyl mercaptan has a very different way of affecting molecular weight of the
polymer produced. Due to its low solubility and higher diffusion resistance, n-dodecyl mercaptan
stays in the monomer droplets initially and gradually diffuses into polymer particles. The result of
Fig. 28. Simulation of styrene emulsion homopolymerization at 70 8C (styrene: 130.7 g, water: 515.6, KPS: 0.233 g, SDS:
2.62 g).
Fig. 29. Simulation of styrene emulsion homopolymerization at 70 8C. Unswollen particle diameter (styrene: 130.7 g, water:
515.6, KPS: 0.233 g, SDS: 2.62 g).
Fig. 30. Simulation of styrene emulsion homopolymerization at 70 8C (styrene: 130.7 g, water: 515.6, KPS: 0.233 g, SDS:
2.62 g).
such a characteristic is higher molecular weight averages in the beginning of the reaction, which
gradually decrease as the reaction proceeds. It is very satisfactory to see that the unique trends of
molecular weight averages inuenced by two types of CTAs are followed very well by the model.
This also proves that the use of an àpparent' chain transfer coefcient is very practical and reliable.
5. Simulation of emulsion homopolymerization of vinyl acetate
Vinyl acetate bulk/solution polymerization is characterized as a system with high degree of branching.
A detailed kinetic study and model testing of homo-/copolymerization for this monomer in bulk/solution
are summarized in Refs. [79,80]. In emulsion polymerization, vinyl acetate is a typical case I monomer
with high water solubility and signicant desorption. Kinetic studies on this monomer found in the
literature are primarily focused on the desorption phenomenon. In addition to the difculties in modeling
branching in vinyl acetate polymerization, desorption makes modeling of vinyl acetate emulsion poly-
merization more complicated compared to other monomers with low water solubility like styrene or
butadiene. The desorption process itself is very difcult to understand, despite the fact that it has been
studied by a number of researchers for many years. A very good review on many aspects of vinyl acetate
emulsion polymerization can be found in Ref. [100]. Chern and Poehlein [101] also reviewed this subject
and developed their own model to simulate vinyl acetate emulsion polymerization.
It is important to understand the characteristics associated with emulsion polymerization of vinyl
acetate before developing a mathematical model. These characteristics are:
1. Emulsion polymerization kinetics of VAc does not t the 01 SmithEwart kinetic scheme. Due to
Fig. 31. Simulation of styrene emulsion homopolymerization at 70 8C. Unswollen particle diameter (styrene: 130.7 g, water:
515.6, KPS: 0.233 g, SDS: 2.62 g).
F
i
g
.
3
2
.
S
i
m
u
l
a
t
i
o
n
o
f
n
u
m
b
e
r
a
v
e
r
a
g
e
m
o
l
e
c
u
l
a
r
w
e
i
g
h
t
i
n
s
t
y
r
e
n
e
e
m
u
l
s
i
o
n
h
o
m
o
p
o
l
y
m
e
r
i
z
a
t
i
o
n
a
t
7
0
8
C
(
s
t
y
r
e
n
e
:
1
3
0
.
7
g
,
w
a
t
e
r
:
5
1
5
.
6
,
K
P
S
:
0
.
2
3
3
g
,
S
D
S
:
2
.
6
2
g
)
.
desorption, the average number of radicals per particle n is much lower than 1/2 (typically n is in the
order of 10
22
10
23
) throughout the entire reaction in many cases.
2. The rate of polymerization stays relatively constant throughout a large part of the reaction, and the
gel effect is usually not observed. This is because desorption greatly reduces the probability of
coexistence of multiple radicals inside one particle.
3. Branching is important.
4. Termination is not as important as it is in bulk reactions. Rate of propagation and molecular weight
averages developed are less dependent on termination. In contrast, molecular weight is controlled by
chain transfer reactions.
5. Chain transfer to monomer is the rst step in the desorption process.
In the list above, desorption and chain transfer to monomer are the most important features. It has been
observed by a number of authors [54,60,75,102] that the molecular weight of polymer produced in vinyl
acetate emulsion polymerization is controlled by chain transfer reactions such as transfer to monomer/
polymer/CTAs rather than termination reactions. Friis and Hamielec [60,75] pointed out that termination
in vinyl acetate emulsion polymerization plays a minor role compared to emulsion polymerization of
other monomers like styrene or MMA, and this is because it is unlikely that multiple radicals can coexist
in one single particle due to the high rate of desorption. This conclusion was conrmed by the work of
Friis et al. [102]. Friis et al. [102] derived a model to calculate the leading moments of the molecular
weight distribution as a function of conversion. It can be concluded from their work that molecular
weight in vinyl acetate emulsion polymerization is controlled by chain transfer reactions and is inde-
pendent of particle number and initiator concentration. Furthermore, branching is more signicant in
Fig. 33. Simulation of weight average molecular weight in styrene emulsion homopolymerization at 70 8C (styrene: 130.7 g,
water: 515.6, KPS: 0.233 g, SDS: 2.62 g).
emulsion polymerization than observed in bulk/solution polymerization, and this can be attributed to the
higher polymer content inside the particles even at the beginning of the reaction. The same method used
by Friis et al. [102] was also used by Friis and Hamielec [60] in their modeling effort on emulsion
polymerization of vinyl acetate.
Friis and Nyhagen [54] investigated vinyl acetate emulsion kinetics both experimentally and theore-
tically. They observed in their experiments that the conversion versus time curve was linear between 15
and 85% conversion levels. They explained this phenomenon is partly due to a decrease in the desorption
rate of radicals from the polymer particles. To describe such a change mathematically, they derived an
expression that relates the radical diffusivity in polymer particles to the free volume of the system as
shown below
D
p
= D
p0
exp
2bxd
m
(1 2a)
(1 2x)d
p
1axd
m
(110)
where a is the ratio of free volume for polymer over free volume for monomer, V
fp
/V
fm
; b, the ratio of
critical free volume V
p
over free volume for monomer, V
p
/V
fm
; D
p0
, the initial radical diffusivity in
polymer particlesl; d
m
, the monomer density; d
p
, the polymer density; and x is the conversion.
The critical free volume V
p
in b in Eq. (110) is the free volume at which the system becomes diffusion
controlled. The only other group that adopted a similar approach is Chern and Poehlein [101]. They
proposed that the diffusivity of monomer D
m
is
D
m
= D
m0
exp[2V
p
m
(1=V
f
21=V
f0
)] (111)
where V
p
m
is an adjustable parameter and V
f
stands for the free volume of the system. Eqs. (110), (111)
and (67) are all very similar to each other. They are all based on the same principle (free-volume theory).
Friis and Nyhagen's equation directly links the diffusivity D
p
with conversion level. A diffusivity that
changes with conversion level is important in modeling vinyl acetate emulsion polymerization, since it
reects the interplay between the gel effect and the desorption process.
A detailed discussion on desorption has already been given in Section 2.4. Many papers mentioned in
Section 2.4 are good references for this monomer system and they will not be cited here again. Selective
papers are listed in Table 11, especially with respect to useful data sources for model testing.
5.1.1. Nomura et al. [49]
Nomura et al. [49] performed a kinetic investigation on many aspects of vinyl acetate emulsion
homopolymerization (effect of initiator levels, emulsier levels, etc.). Their experimental runs were
simulated and results are presented in Figs. 3437. Fig. 34 shows one run with 12.5 g/l of water SDS as
emulsier and 1.25 g/l of water KPS as initiator. It should be pointed out that in most of their runs there
was oxygen present. The run shown in Fig. 34 is no exception. The prediction of the overall trend of
conversion prole is good. The oxygen effect is more pronounced in runs with a lower initiator amount
used. This can be seen in Fig. 35. A total of ve runs were carried out with initiator concentration ranging
from 0.156 to 2.5 g/l of water. It is no surprise to see that the run with the lowest initiator concentration
exhibits the longest induction time, while the run with the highest initiator concentration started almost
immediately.
Nomura et al. [49] stated that the conversion prole in most of their experiments exhibits linearity,
therefore by assuming a constant R
p
throughout the reaction, the average number of radicals per particle
can be backcalculated if the total number of particles in the latex is known. It should be realized that the
rate of polymerization may not be strictly constant, therefore n derived from this backcalculation method
is only approximate. Nevertheless, it at least demonstrates the trend of how n changes during the
Table 11
Selective reference list for vinyl acetate emulsion polymerization
References Remarks
Badran et al. [182] Redox system initiation in VAc emulsion polymerization
Bataille et al. [183] Agitation effect on reaction kinetics, experimental data
Chern and Poehlein [101] Review on VAc emulsion polymerization, kinetic model
de La Rosa et al. [184] Kinetic parameters in VAc copolymerization
Dimitratos et al. [185] BAVAc copolymerization
Dunn and Taylor [186] Experimental data, effect of ionic strength
El-Aasser and Venderhoff [100] General review, many useful papers
Farber [187] CSTR operation of MMA-VAc
Friis and Nyhagen [54] General kinetic study of VAc polymerization, experimental data
Friis et al. [102] Molecular weight and branching development
Friis and Hamielec [60,75] Kinetic studies and molecular weight development
Friis and Hamielec [65] Studies on gel effect
Greene et al. [188] CSTR operation/stability of MMA, VAc polymerization
Hawkett et al. [35] Radical absorption efciency
Kiparissides et al. [5557] CSTR operation of VAc
Kiparissides et al. [189] Oscillation studies on emulsion polymerization of VAc
Kong et al. [123] VAcBA batch copolymerization
Kshirsager and Poehlein [37] Measurement of critical size of oligomer radicals before forming
particles in VAc emulsion polymerization
Lange et al. [190] Seeded emulsion polymerization of VAc
Lee and Mallinson [191] Molecular weight development for VAc
Lee and Mallinson [192] Surfactant effects on molecular weight and PDI
Lichti et al. [193] General kinetic studies
Lichti et al. [2] Chain transfer mechanism in VAc
Litt and Stannett [194] Seeded Vac homopolymerization experiments
Litt and Chang [195] Chain transfer effect
Meuldijk et al. [196] Polymerization of VAc in a pulsed packed column reactor
Misra et al. [197] Batch and semi-batch copolymerization of VAcBA
Moustafa et al. [198] Emulsier-free emulsion polymerization of VAc
Noel et al. [143] VAc/MA copolymer, monomer partitioning
Nomura et al. [49]
a
Experimental data
Nomura et al. [50] Experimental data and modeling
Nomura and Harada [51] Rate constant for desorption
Omi et al. [199] General modeling
Penlidis et al. [103]
a
Experimental data, modeling
Sarkar et al. [200] Thermal decomposition of KPS with VAc
Singh and Hamielec [201] Full conversion experimental data, emulsier-free reaction
Trivedi [202] Experimental data, monomer partitioning
Urquiola et al. [203] Emulsion polymerization of VAc using polymerizable surfactant
Zollars et al. [34] Experimental data, modeling
a
Experimental data tested.
Fig. 34. Simulation of vinyl acetate emulsion polymerization at 50 8C (KPS: 1.25 g/l of water, M: 0.5 g/ml water).
Fig. 35. Simulation of vinyl acetate emulsion polymerization at 50 8C (SDS: 6.25 g/l of water, M: 0.5 g/ml water).
Fig. 36. Comparison of model predicted and reported average number of radicals per particle (SDS: 6.25 g/l of water, KPS:
1.25 g/l of water, M: 0.5 g/ml water).
Fig. 37. Monomer partitioning in vinyl acetate emulsion homopolymerization at 50 8C (M: 0.5 g/ml water, SDS: 1.88 g/l of
water, KPS: 1.25 g/l of water).
reaction. Nomura et al. [49] reported n for some runs using the backcalculation method, and the reported
n for one of their runs is compared with model predictions in Fig. 36. It is evident in Fig. 36 that at the
beginning of the reaction, n is rather small and in the range of 10
23
. It gradually increases into the range
of 10
22
, and at the end of the reaction, n increases due to the decreased rate of desorption. The model
follows the trend of n in a satisfactory way as can be seen from Fig. 36.
Monomer partitioning plays an important role in emulsion polymerization. Nomura et al. [49]
measured weight fraction of monomer in polymer particles for several runs with various amounts of
initiator and emulsier. It was found that variation of these conditions had little effect on monomer
partitioning. Fig. 37 demonstrates how vinyl acetate partitions itself among different phases. In stages I
and II, the weight fraction of monomer remains nearly constant up to 2025% conversion which marks
the end of stage II, then it decreases steadily. The model prediction in Fig. 37 is given by using monomer
partitioning coefcients, and the agreement is obviously very good.
5.1.2. Penlidis et al. [103]
Penlidis et al. [103] also investigated vinyl acetate emulsion homopolymerization and they developed
a mathematical model for this monomer system. Many runs were performed under a variety of reaction
conditions and recipes. Conversion was measured by either off-line gravimetry or on-line densitometry.
Figs. 38 and 39 show two runs performed at two levels of initiator. It is obvious that the reported
conversion measurement is reliable because off-line gravimetry agrees with on-line densitometry
well. In all the runs performed by Penlidis et al. [103], some inhibitor was left in the monomer solution
in order to mimic real industrial practice. With various lengths of induction time, the model is able to
deliver reliable predictions on conversion data.
Fig. 38. Simulation of vinyl acetate emulsion homopolymerization at 50 8C (VAc: 1.1 l, water: 3 l, SDS: 33.6 g, KPS: 2.5 g).
Data reproducibility was very good in Ref. [103] runs. This is demonstrated in Fig. 40. There are three
runs conducted under the same reaction conditions, and the measured conversion points for all three runs
agree with each other well, and so does the model prediction. Fig. 41 illustrates the temperature/initiator
effect on the reaction. Three runs were carried out at similar reaction conditions but under different
temperatures/initiator loadings. Apparently, a higher temperature results in a faster reaction rate.
Furthermore, a higher temperature results in a shorter induction time. This is because impurities are
consumed at a faster rate while more radicals are generated in the same period of time.
Penlidis et al. [103] also measured molecular weight data as well as particle size in their experiments.
The average diameter of unswollen particles in several of their run was measured by off-line turbidity
spectra and hydrodynamic chromatography. Results are plotted in Fig. 42. The two runs differ from each
other in terms of level of emulsier. As expected, the run with the higher amount of emulsier generated
smaller particles because the same amount of monomer is distributed among more particles, and as a
consequence, the average particle size is smaller. The model prediction is in the right range with some
discrepancies. The simulation results are nevertheless acceptable considering the uncertainties involved
in the particle size measurement. There is usually approximately a 100200 A
error involved in the

measurement, therefore model predictions are within the correct range.
Weight average molecular weights for runs at different temperatures were measured by using off-line
low angle laser light scattering (LALLS) in Figs. 43 and 44, or LALLS-GPC in Fig. 45. All three gures
reveal a trend that weight average molecular weight gradually increases as reaction proceeds. This is
because at the later stages of the reaction, the desorption rate decreases and therefore n becomes larger.
Clearly, the model follows such a trend well and the predictions for the weight average molecular weight
are quite satisfactory. The temperature effect on molecular weight is also evident.
Fig. 39. Simulation of vinyl acetate emulsion homopolymerization at 50 8C (VAc: 1.1 l, water: 2.8 l, SDS: 33.6 g, KPS: 1.8 g).
6. Simulation of homopolymerization of acrylic monomers: methyl methacrylate, butyl acrylate
and ethyl acrylate
6.1. Simulation of methyl methacrylate emulsion homopolymerization
Methyl methacrylate is characterized as a monomer with moderate water solubility, and the average
number of radicals per particle is higher than 1/2. Methyl methacrylate is a case III monomer. The gel
effect is more pronounced in methyl methacrylate emulsion polymerization than styrene or vinyl acetate.
This phenomenon was observed by a number of research groups. Zimmt [104] performed emulsion
polymerization of both styrene and methyl methacrylate. He noticed that in methyl methacrylate emul-
sion polymerization, there was a strong acceleration in the rate of polymerization, while the same
behavior was not observed in styrene emulsion polymerization. The same characteristics were also
conrmed by Friis and Hamielec [74] and Sundberg et al. [77]. Additional evidence of the presence
of strong gel effect was presented by Chang et al. [105]. This group used ESR to measure the radical
concentration inside the particles in both seeded and unseeded polymerizations of methyl methacrylate.
In both cases, it was observed that there was a sharp increase in radical concentration at the late stages of
the reaction. The shape of the radical concentration prole is nearly the same as that in bulk polymer-
ization. A strong gel effect also affects the molecular weight of polymethyl methacrylate. James and
Piirma [90] stated that weight average molecular weights increased in stage III and they attributed such
an increase to autoacceleration. Though there is adequate evidence to state that methyl methacrylate
emulsion polymerization exhibits a strong gel effect, the treatment of this phenomenon still needs much
improvement. In the early studies, Friis and Hamielec [74] used an empirical expression for the rate
Fig. 40. Simulation of vinyl acetate emulsion homopolymerization at 50 8C (VAc: 1.15 l, water: 2.86 l, SDS: 28.8 g, KPS: 3 g).
Fig. 41. Simulation of vinyl acetate emulsion homopolymerization at various temperatures (M: 1150 ml, water: 2860 ml, SDS:
28.8 g).
Fig. 42. Average unswollen particle diameter in vinyl acetate emulsion homopolymerization at 50 8C (M: 1150 ml, water:
2860 ml, KPS: 3 g).
Fig. 43. Weight average molecular weight data in vinyl acetate emulsion homopolymerization at 50 8C (M: 1150 ml, water:
2860 ml, KPS: 3 g, SDS: 28.8 g).
Fig. 44. Weight average molecular weight data in vinyl acetate emulsion homopolymerization at 40 and 70 8C (M: 1150 ml,
water: 2860ml, SDS: 28.8 g).
constant for termination to describe the conversion data collected in their experiments. Ballard et al.
[106,107] also used an empirical expression for k
t
in their studies. Nomura and Fujita [108] stated that k
t
was constant through stages I and II, and then it started to decrease. Sundberg et al. [77] was the group
that attempted to model the rate constant for termination from a theoretical basis. The free-volume
theory was used to interpret the change of k
t
in their experiments.
It is important to note that the starting value of k
t
is many orders of magnitude smaller than its value in
bulk/solution polymerization due to highly viscous particle media. This characteristic has been veried
by a number of groups [11,104,106108]. Many authors also used various methods to estimate values of
k
t
as well as the rate constant for propagation k
p
, and their estimates are summarized in Table 12.
Among the groups listed, work from Ballard et al. [106,107,109] is worth mentioning. The k
t
value
they estimated is much higher than most other groups. What they estimated was an àpparent' k
t
which
also included the contribution from reaction diffusion controlled termination. It must be stated here that
reaction diffusion controlled termination plays an even more important role in emulsion polymerization
than in bulk/solution polymerization. The reason for this is simply that the reaction mixture in polymer
particles is very viscous even at the beginning of the reaction. This group also gave an empirical
expression for k
p
which is:
k
p
= k
o
p
exp[229:8(w
p
20:84)] (112)
Expression (112) is used when the weight fraction of polymer in the particles w
p
exceeds 84%. In a very
similar fashion, Nomura and Fujita [108] presented the following two expressions for k
t
and k
p
:
k
f
= 8:5 10
215
exp(212:3w
p
) (113)
Fig. 45. Weight average molecular weight data in vinyl acetate emulsion homopolymerization at 60 8C (M: 1150 ml, water:
2860 ml, KPS: 3 g, SDS: 28.8 g).
k
p
= 1:33 10
7
exp(213:6w
p
) (114)
Expression (114) is used only when w
p
.73%. Before 73% conversion, k
p
is set to 650 l/mol s.
Additional literature papers with regards to methyl methacrylate emulsion polymerization were
collected to gather kinetic information. Table 13 gives a list of papers that provide useful kinetic
information or experimental data for model testing.
6.1.1. Model testing results
6.1.1.1. Barton et al. [110]. Barton et al. [110] conducted experiments of seeded emulsion polymer-
ization of methyl methacrylate. Polybutyl acrylate latex seed was prepared in batch emulsion polymer-
ization at 60 8C. The nal conversion was measured as 95.6%. The average hydrodynamic diameter of
the seed particles was measured as 170 nm, and the number of particle seeds was thus backcalculated as
1.09 10
17
per liter of polymer latex. Four polymerization runs were performed using various recipes. In
two runs, retarder 2,2,6,6,-tetramethyl-4-octadecanoyloxypiperidinyl-1-oxyl (STMPO) was added to
investigate its effect on reaction kinetics. It was observed that the presence of STMPO signicantly
suppressed the reaction rate. Table 14 gives the recipes used for all their runs.
Run 2 was simulated using the reported reaction conditions and recipe. Fig. 46 displays the simulated
conversion prole. It is noted that in run 2, the ratio of MMA to PBA solid in the initial charge is 3.16,
and this implies an initial conversion of 25%. However, this is in contradiction to what is shown in Fig.
46, where the two conversion measurements at the early stages of the run are well below 20%. No
reasonable explanation was given by Barton et al. [110]. Model predictions at higher conversion levels
agree reasonably well with reported data. Fig. 47 displays the change of the size of polymer particles.
The model predicted that the initial particle diameter was at 160 nm and it gradually increased to the
230 nm level at which the limiting conversion occurred. Model predictions are in close agreement with
what was reported by Barton et al. [110].
Table 12
Rate constants for termination and propagation (MMA)
Temperature 40 8C 50 8C 60 8C
k
t
(l/mol s) 6000
a
19 000
b
90 000
c
7000
d
k
p
(l/mol s) 171
e
340
e
500
e
790
c
441
a
650
f
450
d
560
d
684
d
a
Ref. [204].
b
Ref. [77].
c
Refs. [107,109].
d
Value used by this model.
e
Ref. [205].
f
Ref. [108].
Runs 3 and 4 are not simulated due to the lack of kinetic information on STMPO. Simulation of run 1
was impossible due to several factors. First, the initial MMA/PBA ratio in the charge was 0.632, which
corresponds to an initial conversion level of about 62%. Apparently, this is not what was reported by
Barton et al. [110]. Secondly, there was very little monomer added in this run and it was reported there
were no monomer droplets formed. Such low level of monomer charge and the absence of monomer
droplets may ultimately change MMA partitioning characteristics.
6.1.1.2. Canegallo et al. [94]. Canegallo et al. [94] used calorimetry to measure on-line conversion data
Table 13
Literature references on MMA emulsion polymerization
Authors Remarks
Ballard et al. [106] Review on MMA emulsion polymerization, modeling
Ballard et al. [107] Measurement of rate constant for termination
Ballard et al. [109] Measurement of rate constant for propagation
Barton [206] MMA emulsion polymerization with oil soluble initiator: AIBN
Barton et al. [110]
a
Seeded emulsion polymerization of methyl methacrylate
Canegallo et al. [94]
a
On-line measurement of conversion using densitometry
Chang et al. [105] Measurement of radical concentration in MMA emulsion seeded polymerization
Dube [111]
a
MMA emulsion homopolymerization
Fontenot Schork [114]
a
MMA batch polymerization in mini/macroemulsions
Fontenot Schork [112,113]
a
MMA batch polymerization in mini/macroemulsions, mathematical modeling
Friis and Hamielec [74] Studies on termination in MMA emulsion polymerization
Gardon [11]
a
Experimental data
James and Piirma [90] Molecular weight development
Ley et al. [204] Half-life of propagating radicals in MMA emulsion polymerization
Louie et al. [159] Effect of oxygen in MMA emulsion polymerization, modeling
Nomura and Fujita [108] Many experiments for MMA emulsion polymerization under a wide range of
reaction conditions, simplied model developed
Schork and Ray [207] Multiple steady-state studies in MMA continuous emulsion polymerization
Schork and Ray [115]
a
On-line densitometry in MMA emulsion polymerization
Soh [205] Measurement of rate constant for propagation
Sundberg et al. [77] Studies on diffusion controlled kinetics, experimental data
Tanrisever et al. [208] Emulsier-free MMA emulsion polymerization
Wang and Chu [209] Mixed emulsiers
Zimmt [104] Experimental data on conversion and particle size
a
Data used for model testing.
Table 14
Recipes of seeded emulsion polymerization of MMA
Run no. PBA latex (ml) MMA (ml) Water (ml) STMPO
1 50 10 20 Not added
2 20 20 40 Not added
3 50 10 20 Yes
4 20 20 40 Yes
for several runs of methyl methacrylate at 50 8C. The reported conversion proles as well as model
predictions are presented in Fig. 48. The shapes of the conversion curves resemble those observed in
bulk/solution runs of the same monomer. Autoacceleration is present in all three runs and so is the
limiting conversion. The current model describes all these kinetic characteristics well.
6.1.1.3. Dube [111]. Dube [111] conducted one experiment of methyl methacrylate emulsion homo-
polymerization at 60 8C using mixed emulsiers of AMA-80 and AOT-75. Conversion was determined
by gravimetry. He also measured particle size as well as molecular weight. The simulation of conversion
data is shown in Fig. 49. Model predictions can follow the conversion data very well up to 40%, at which
point model predictions gradually lag behind the actual conversion measurements. Disagreement also
exists in the nal conversion. The model predicts a limiting conversion of about 90% while almost full
conversion was determined in Dube's run. The observed discrepancies are suspected to be caused by
nonisothermality. As emphasized before, this monomer system is characterized by a strong gel effect,
and the polymerization is highly exothermic. It is likely that at the peak of the gel effect, there is heat
accumulation. The possible nonisothermal polymerization may also be used to explain the observed full
conversion. Theoretically speaking, there should be a limiting conversion in this experiment as the
model predicts, because the reaction temperature is below the T
g
of polymethl methacrylate. Higher
temperatures at the end of the reaction would surely result in a higher nal conversion level. Another
possible cause of the discrepancies at higher conversion levels is of course experimental anomalies
during gravimetry, which were suspected in Ref. [111] for this specic run.
Fig. 46. Simulation of seeded methyl methacrylate homopolymerization in emulsion at 60 8C (0.294 g polybutyl acrylate seed
per ml latex, N
p
: 1.09
17
/l latex, initial seed diameter 170 nm, nal conversion in the seed: 95.6%, seed: 20 ml, MMA: 20 ml,
water: 40 ml).
Fig. 47. Simulation of particle size in seeded methyl methacrylate homopolymerization in emulsion at 60 8C (0.294 g polybutyl
acrylate seed per ml latex, N
p
: 1.09
17
/l latex, initial seed diameter 170 nm, nal conversion in the seeds: 95.6%, seed: 20 ml,
MMA: 20 ml, water: 40 ml).
Fig. 48. Simulation of methyl methacrylate emulsion homopolymerization at 50 8C using calorimetry.
Fig. 49. Simulation of methyl methacrylate emulsion homopolymerization at 60 8C (MMA: 1.5 kg, water: 2.401 kg, APS:
1.0506 g, AMA-80: 7.496 g, AOT-75: 8.325 g).
Fig. 50. Simulation of particle size for methyl methacrylate emulsion homopolymerization at 60 8C (MMA: 1.5 kg, water:
2.401 kg, APS: 1.0506 g, AMA-80: 7.496 g, AOT-75: 8.325 g).
Fig. 50 is the unswollen particle size for the run simulated. Figs. 51 and 52 show experimental as well
as simulation results for the number of particles and molecular weight averages. In all cases, the model
gives satisfactory predictions.
6.1.1.4. Fontenot and Schork [112,113]. Fontenot and Schork [112,113] is the rst group that presented
a mathematical model that can simulate both conventional emulsion as well as miniemulsion polymer-
ization. In miniemulsion, monomer droplets are stabilized through the use of cosurfactants or stabilizers
like long chain fatty alcohols or alkanes. Monomer droplets are typically much smaller than those in
conventional emulsion polymerization. Fontenot and Schork [112,113] conducted several conventional
and miniemulsion polymerizations to study the effect of initiator and emulsier on the reaction kinetics.
Their work provides a good source of data for model testing. Figs. 5359 summarize model testing
results using this group's data.
Fig. 53 shows the effect of initiator on the rate of polymerization. In these three runs, the concentration
of KPS ranges from 0.002 to 0.02 mol/l of water. Figs. 5457 show the effect of emulsier on the rate of
polymerization. A number of runs were carried out with a xed amount of KPS at various emulsier
levels ranging from 0.0093 to 0.1 mol/l of water. For the run with 0.02 mol/l of water of SDS, Fontenot
and Schork [112] also measured the nal particle diameter. Their reported diameter was in the range of
10001200 A
, which is very close to what our model predicted (Fig. 58). For all the experiments tested,
simulation results are in good agreement with actual experimental measurements regardless of the
concentration level of emulsier or initiator used in the recipe. This veries our model's reliability
once more.
The temperature effect on the rate of polymerization is best seen in Fig. 59. Three experiments were
Fig. 51. Simulation of number of particles in methyl methacrylate emulsion homopolymerization at 60 8C (MMA: 1.5 kg,
water: 2.401 kg, APS: 1.0506 g, AMA-80: 7.496 g, AOT-75: 8.325 g).
Fig. 52. Simulation of molecular weight in methyl methacrylate emulsion homopolymerization at 60 8C (MMA: 1.5 kg, water:
2.401 kg, APS: 1.0506 g, AMA-80: 7.496 g, AOT-75: 8.325 g).
Fig. 53. Simulation of conversion in methyl methacrylate emulsion homopolymerization at 50 8C (MMA: 300 g, water: 700 g,
SDS: 0.02 mol/l of water).
KPS: 0.005 mol/l of water).
Fig. 58. Simulation of particle size in methyl methacrylate emulsion homopolymerization at 50 8C (MMA: 300 g, water: 700 g).
Fig. 59. Simulation of conversion in methyl methacrylate emulsion homopolymerization (MMA: 300 g, water: 700 g, KPS:
0.005 mol/l of water: SDS: 0.02 mol/l of water).
conducted at 40, 60 and 70 8C. Model predictions are good for the 60 and 70 8C runs, but discrepancies
exist for the 40 8C. As often observed in many other examples, impurity effects are suspected to be much
more pronounced at low temperature levels. An additional 25 min of induction was used in the simula-
tion. However, the measured conversion data still lag behind model predictions, which indicate the
retardation effect of impurities.
6.1.1.5. Fontenot and Schork [114]. Fontenot and Schork [114] later conducted more experiments under
a slightly modied recipe (different monomer to water ratio). Measured conversion data are presented in
Figs. 60 and 61. The conversion curves displayed in Figs. 60 and 61 have the same shape as those
experiments shown in Figs. 5457 and 59.
The authors also reported the number of particles for their runs. Instead of directly measuring the
number of particles, they backcalculated the number of particles per liter of water using measured
particle diameter and conversion results. Particle diameter was measured by using transmission electron
microscopy. Model predicted numbers of particles for the three runs are plotted in Fig. 62 along with the
values from Ref. [114]. The agreement is satisfactory.
6.1.1.6. Gardon [11]. Gardon [11] carried out full conversion range experiments for a number of
monomers including methyl methacrylate. He noticed that the ratio of k
t
/k
p
was low, and he stated
that this would be the result of high viscosity within the particles. Gardon [11] performed one run of
methyl methacrylate emulsion polymerization at 55 8C. Model testing results for this run are displayed in
Fig. 63.
KPS: 0.005 mol/l).
KPS: 0.005 mol/l).
Fig. 62. Number of particles in methyl methacrylate emulsion homopolymerization at 50 8C (MMA: 220 g, water: 510 g, KPS:
0.005 mol/l).
6.1.1.7. Schork and Ray [115]. Schork and Ray [115] conducted methyl methacrylate emulsion
polymerization at 40 8C. Reaction conversion was monitored by an on-line densitometer. To verify
their on-line measurements, they also used gravimetry to determine conversion off-line. Both results
are displayed in Fig. 64. It is, however, noticed that in the mid-range of conversion in their experiment,
their on-line measurement gave lower conversion than that from off-line samples. Since at the time
there were more uncertainties associated with the on-line measurement, data obtained from their off-
line samples are more trustworthy. It is good to see that the model gives predictions that are closer
to their off-line measurements, which again indicates good reliability of our model's predictive
powers.
6.2. Simulation of butyl acrylate emulsion homopolymerization
Unlike monomers tested previously, butyl acrylate emulsion polymerization received little attention
though it is an important ingredient in latex paints and adhesives. There has been no effort in the
literature to cover detailed kinetic aspects of this monomer system. Maxwell et al. [116] report butyl
acrylate to have moderate water solubility (0.006 mol/l of water), which is between styrene and methyl
methacrylate. Maxwell et al. [116] also measured monomer concentration in polymer latex particles,
which was about 3.2 mol/l, and this leads to a partitioning coefcient between oil and water phases of
about 533. The rate constant for termination determined by this group is 750 l/mol s (at about 85%
conversion), which is very low. The low value of k
t
is probably due to the gel effect. Maxwell et al. [116]
Fig. 63. Simulation of conversion in methyl methacrylate emulsion homopolymerization at 55 8C (M/W: 40/60, SDS:
0.244 wt%, KPS: 0.165 wt%).
suggested that at the late stages of the reaction, termination is dominated by reaction diffusion control.
They estimated that the rate constant for desorption is 0.004 s
21
, and this indicates that desorption of this
monomer is not very signicant.
Capek and Fouassier [117] found that monomer concentration in polymer particles is 4.4 mol/l, and
estimated the rate constant for propagation at 1360 mol/l s at 60 8C. More information about k
p
and k
t
can
be found in Ref. [118]. Their estimate is 170 l/mol s for k
p
and 3400 for k
t
at no specied temperature.
Mallya and Plamthottan [119] investigated several aspects of butyl acrylate emulsion polymerization
kinetics. They found that monomer concentration in polymer latex particles is in the range of 3.1
3.25 mol/l, which is very close to Maxwell's et al. [116] estimation. This group also claimed that k
t
is in
the range of 10
3
10
4
l/mol s at 50 8C. The low value of k
t
leads to a high value for n; which was
estimated in the range of 530 at mid- to high-conversion range. This group also estimated the rate
constant for desorption as 0.0027 s
21
, which is similar to what Maxwell et al. [116] estimated. The low
desorption rate constant once again implies that desorption is not very signicant in BA emulsion
polymerization.
There have been no studies on molecular weight development for this monomer system. Lovell et al.
[120] studied chain transfer to polymer in butyl acrylate emulsion polymerization. They presented
direct evidence of chain transfer to polymer and formation of branched chains. The chain transfer
to polymer proceeds via abstraction of a hydrogen atom from a tertiary CH bond in a butyl
acrylate repeating unit. Lovell et al. [120] however did not estimate the rate constant for chain transfer to
polymer.
From all the kinetic information gathered from the literature with regards to butyl acrylate emulsion
Fig. 64. Simulation of conversion in methyl methacrylate emulsion homopolymerization using on-line densitometry (M/W:
43%, [SDS]: 0.02 mol/l of water, [APS]: 0.01 mol/l of water).
polymerization, it can be summarized that butyl acrylate has the following characteristics:
1. Moderate water solubility.
2. Insignicant desorption, high n:
3. Strong gel effect; reaction diffusion control mechanism dominates in late stages of reaction.
4. Presence of chain transfer to polymer, branching.
Additional papers that might be useful for kinetic studies or modeling for this monomer are listed in
Table 15.
6.2.1.1. Capek and Postik [121]. Capek and Postik [121] performed one experiment of butyl acrylate
emulsion homopolymerization at 60 8C in their studies of crosslinking polymerization involving divinyl
benzene. Measured conversion data are displayed in Fig. 65. It can be seen that conversion goes to
completion due to the very low T
g
(245 8C) of polybutyl acrylate. Despite the presence of some
induction time, the model follows the conversion versus time curve well.
6.2.1.2. Cruz et al. [122]. This group performed studies on butyl acrylate/styrene copolymerization in
emulsion. One homopolymerization run with butyl acrylate was carried out and is used for model testing.
Table 15
Literature review on butyl acrylate emulsion homopolymerization
References Remarks
Barandiaran et al. [118] Estimation of rate constant for desorption and absorption
Capek et al. [210] Initiation of oil soluble initiator (AIBN)
Capek [211] Effect of initiator type and concentration
Capek [212] Use of mixed emulsiers
Capek and Potisk [121]
a
Copolymerization of BA/AN
Capek and Fouassier [117] Photopolymerization of butyl acrylate
Chern and Chen [213] Effect of carboxylic acid on particle nucleation in butyl acrylate
emulsion polymerization
Chu and Lin [214] Stabilization with mixed surfactants
Cruz et al. [122]
a
Full conversion butyl acrylate homopolymerization
Dube [111]
a
Batch emulsion polymerization of butyl acrylate
El-Aasser et al. [215] Copolymerization of BA/VAc
Kong et al. [123]
a
Full conversion range data
Kukulj et al. [216] Catalytic chain transfer
Lovell et al. [120] Chain transfer to polymer
Mallya and Plamthottam [119] Rate constant for termination
Maxwell et al. [116] General kinetics
Maxwell et al. [217,218] Reaction kinetics at high conversion
Mcbain and Piirma [219] Effect of chain length on reaction kinetics
O'Neil and Hoigne [220] Effect of emulsier on reaction kinetics
a
Reported experimental data used for model testing.
Apparently, the reported experimental data are problematic, as displayed in Fig. 66. The conversion
curve reached a plateau after 80% conversion, indicating a reduced rate of polymerization. Full conver-
sion was never reached as theory predicts, and an oxygen leak was suspected. Prediction of conversion at
the early stages is in the right range.
6.2.1.3. Dube [111]. Dube [111] conducted two experimental runs of butyl acrylate emulsion homo-
polymerization using mixed emulsiers of AMA-80 and AOT-75. The reaction was performed in a 5 l
stainless steel reactor. The initial reaction temperature was 60 8C. Conversion results of the rst run are
displayed in Fig. 67. It is easy to see that there is a large discrepancy in the mid-range of the reaction. The
disagreement is caused by nonisothermality in this run. On the basis of Dube's own observations, severe
temperature excursions (as much as 30 8C) were observed in the beginning of the reaction. Such
nonisothermality lasted for about 30 min. This explains why between 10 and 40 min the reported
conversion data points are much higher than the model predictions. It is clear that a reactor jacket
alone is inadequate to remove the heat of reaction fast enough. To avoid heat accumulation, a cooling
coil was installed in the same reactor in the second run, and the initiator amount was reduced by half to
slow down the rate of polymerization. The reaction was then carried out at the same temperature. Better
agreement is reached between model predictions and experimental results in the second run as shown in
Fig. 68, though nonisothermality was not entirely eliminated.
Model predictions on particle diameter and number of particles for both runs are good. Simulation
results for diameter for both runs are presented in Figs. 69 and 70, while results for total particle number
Fig. 65. Simulation of conversion in butyl acrylate homopolymerization in emulsion at 608C (water: 100 g, BA: 6.66 g, SDS:
2 g, APS: 0.2 g).
Fig. 66. Simulation of conversion in butyl acrylate homopolymerization in emulsion at 608C (BA: 100 g, KPS: 4 g, water:
600 g, SDS: 0.1 g).
Fig. 67. Simulation of conversion in butyl acrylate emulsion homopolymerization under nonisothermal conditions (initial
temperature: 60 8C, BA: 1436 g, water: 2419 g, APS: 1.45 g AMA-80: 14.472 g, AOT-75: 14.325 g).
Fig. 68. Simulation of conversion in butyl acrylate emulsion homopolymerization at 60 8C (BA: 1436 g, water: 2419 g, APS:
1.45 g, AMA-80: 14.472 g, AOT-75: 14.325 g).
Fig. 69. Simulation of particle size in butyl acrylate emulsion homopolymerization under nonisothermal conditions (initial
temperature: 60 8C, BA: 1436 g, water: 2419 g, APS: 145 g, AMA-80: 14.472 g, AOT-75: 14.325).
are displayed in Figs. 71 and 72. The agreement indicates that temperature excursions have less of an
effect on particle size and total number of particles.
6.2.1.4. Kong et al. [123]. This group conducted copolymerization of BA/VAc in emulsion. One
homopolymerization run is used for model testing as shown in Fig. 73. The reaction temperature is
60 8C, far above PBAs T
g
, and no limiting conversion is expected. This is conrmed by the reported
conversion measurements. The nal conversion for the reaction is in the range of 95%. Overall model
predictions are good.
6.3. Simulation of ethyl acrylate emulsion homopolymerization
Ethyl acrylate is an important component in latex paints and adhesives due to its low polymer T
g
.
However, it has been rarely studied and is not well understood. Kinetic information with regards to this
monomer has been collected and listed in Table 16.
Clearly, only very limited kinetic information or experimental data are available in the literature.
Among the papers listed earlier, Adhikari et al. [124] provided useful information on the thermal
decomposition of KPS in ethyl acrylate emulsion polymerization. The rate constant for decomposition
was measured as 1.4 10
26
s
21
. Another useful piece of information from this group is the measured
monomer-partitioning coefcient between oil and water phases, which is about 16.
McCurdy and Laidler [125] investigated the emulsion polymerization of a number of acrylates
and methacrylates. They found that the rate of acrylates was consistently higher than methacrylates.
They contributed this behavior to steric and inductive effects. Shoaf and Poehlein [126] conducted
Fig. 70. Simulation of particle size in butyl acrylate emulsion homopolymerization at 60 8C (BA: 1436 g, water: 2419 g, APS:
1.45 g, AMA-80: 14.472 g, AOT-75: 14.325 g).
Fig. 72. Simulation of number of particles in butyl acrylate emulsion homopolymerization at 60 8C (BA: 1436 g, water: 2419 g,
APS: 1.45 g, AMA-80: 14.472 g, AOT-75: 14.325 g).
Fig. 71. Simulation of number of particles in butyl acrylate emulsion homopolymerization under nonisothermal conditions
(initial temperature: 60 8C, BA: 1436 g, water: 2419 g, APS: 1.45 g, AMA-80: 14.472 g, AOT-75: 14.325 g).
copolymerization of ethyl acrylate and methacrylic acid. The water solubility of ethyl acrylate was
2.5 wt%. Perhaps the most useful work is from Capek et al. [127]. This group attempted to study
copolymerization of ethyl acrylate with a number of divinyl monomers (used as crosslinkers). A set
of experiments of emulsion homopolymerization of ethyl acrylate were also carried out. Reported
conversion, particle number and size were the full conversion experimental data for model testing.
Simulation results are presented in Section 6.3.1.
Fig. 73. Simulation of conversion in butyl acrylate emulsion homopolymerization at 60 8C (water: 400 g, AOT: 3 g, AMA: 3 g,
BA: 200 g, APS: 0.6 g).
Table 16
Literature summary on ethyl acrylate emulsion homopolymerization
References Remarks
Adhikari et al. [124] Thermal decomposition of KPS in ethyl acrylate emulsion polymerization at 50 8C
Arzamendi et al. [221] Ethyl acrylate and methyl methacrylate copolymerization
Capek et al. [127]
a
Copolymerization of ethyl acrylate with crosslinker
Capek [222] Copolymerization of methyl methacrylate and ethyl acrylate
Fitch et al. [223] Kinetics of particle nucleation
Maw and Piirma [224] Use of nonionic emulsier
McCurdy and Laidler [125] Kinetic studies of acrylates and methacrylates
Shoaf and Poehlein [126] Copolymerization of methacrylic acid and ethyl acrylate
Snuparek [225] Emulsier effect on ethyl acrylate emulsion homopolymerization
Yeliseyeva and Zuikov [226] Emulsion polymerization of acrylic monomers
a
Experimental data used for model testing.
Experiments from Ref. [127] are the only available data for model testing. A set of runs was carried
out at various emulsier concentrations. Model testing results are displayed in Figs. 7476. The most
noticeable feature in these three runs was the fast rate of polymerization ethyl acrylate exhibited.
Polymerization was completed within about 15 min. The fast reaction observed in ethyl acrylate emul-
sion polymerization is not surprising, since its bulk/solution polymerization exhibits the same charac-
teristics. Comparatively, the gel effect in emulsion polymerization is even more pronounced than it is in
bulk. Fig. 74 shows the run with the highest emulsier concentration. There is clearly a very large
discrepancy between model predictions and measured conversion. The possible reason for such a large
disagreement could be attributed to temperature excursions. In Fig. 74, the entire reaction was completed
within 710 min. With such a high rate of polymerization, heat accumulation was very likely, especially
since the reaction was conducted in a glass reactor without a cooling coil. As emulsier levels were
gradually reduced, the temperature excursions in the next two runs (Figs. 75 and 76) were not as severe
as for the run in Fig. 74, and hence the data are closer to the model predictions.
The model predictions for the run with the lowest emulsier level (Fig. 76) agree with the experi-
mental data well, though temperature excursions may still not be entirely eliminated.
In summary, ethyl acrylate emulsion polymerization has the following characteristics:
1. Modest water solubility (close to that of methyl methacrylate).
2. High rate of polymerization, pronounced gel effect.
3. Insignicant desorption.
Fig. 74. Simulation of conversion in ethyl acrylate emulsion homopolymerization at 60 8C (EA: 92.4 g, water: 150 g, KPS:
0.1 g).
0.1 g).
0.1 g).
7. Extensions to emulsion copolymerization
7.1. Brief literature summary
After having developed a mathematical model for emulsion homopolymerization and achieved good
model testing results, the next natural step is to extend it to simulate copolymerization systems. As has
been demonstrated in the development of the bulk/solution model in Refs. [79,80], the same methodol-
ogy is used for the extension of the emulsion model.
If emulsion homopolymerization is considered as a very complex system, modeling of emulsion
copolymerization is an even more challenging task. There has been little effort in modeling of emulsion
polymerization before the 1980s. As already discussed in Section 1.2, most literature models are limited
to specic reaction conditions, and are far from being general in nature. A summary of models for
emulsion polymerization was given in Table 1. In this section, only efforts that presented a comprehen-
sive and general model are discussed.
Gilbert and coworkers [67,128,129] described their extension of homopolymerization to a copoly-
merization model using Hamielec's pseudo-homopolymerization approach. Their model neglected
micellar nucleation, because micelles were considered merely a reservoir of emulsier despite the
fact that there is a large body of evidence of micellar nucleation. Monomer partitioning was solved
empirically. Though many rate constants in their model are expressed as a mathematical average of
individual rate constants of homopolymers, the authors used one single value in their implementation as
a simplication.
Another copolymerization model available in the literature is the work of Dougherty [29,130]. In his
model, monomer partitioning was solved using the knowledge of x
c
. Though Dougherty [29,130] model
is comprehensive from a theoretical point of view, many simplications were used when the model was
implemented. Dougherty [29,130] used apparent rate constants and the rate constants for termination
were expressed empirically as a function of conversion to reect the gel effect. The model was then
tested to simulate the copolymer system of Sty/MMA.
Hamielec and coworkers [3,131,132] gave a comprehensive review of modeling of emulsion poly-
merization. They also presented an emulsion copolymerization model based on a population balance
approach. The pseudo-kinetic rate constant method was extensively used in their model. The gel effect
was treated using free-volume theory instead of using empirical expressions for the rate constants.
Particle size distribution was modeled by using the particle birth time as an internal coordinate in
their work.
The most recent advances are the efforts of Saldivar et al. [4]. Their model for emulsion copolymer-
ization is the continuation of previous work by Min and Ray [1] and is very sophisticated and compre-
hensive. An extensive review on modeling of emulsion copolymerization including particle size and
molecular weights is given in Ref. [4]. In the model presented, the pseudo-kinetic rate constant method is
extensively used. Saldivar et al. [4] stated that monomer partitioning in copolymerization can be solved
for either empirically through the use of partition coefcients or theoretically by using thermodynamic
equilibrium, however they did not describe the actual implementation of either method. Their model was
tested with three copolymer systems (Sty/MMA, Sty/a-methyl Sty, Sty/BD) with success.
It is worth noting that the use of the pseudo-kinetic rate constant method reduces a copolymerization
kinetic scheme to that of a homopolymerization. The mathematical structure of the copolymerization
model is similar to that of the homopolymerization model. However, in emulsion copolymerization,
certain phenomena, e.g. monomer partitioning and radical desorption, require additional treatment.
Model modications with regards to these aspects are detailed in Sections 7.2 and 7.3.
7.2. Monomer partitioning in copolymerization
Monomer partitioning in emulsion homopolymerization has already been discussed in Section 2.3.
Monomer partitioning in copolymerization is of critical importance, since it is directly related to the
calculation of the rate of polymerization as well as of copolymer composition. However, this is a very
difcult subject. Monomer partitioning in copolymerization is much more complicated than in homo-
polymerization because it involves two monomers and their copolymer. Monomer partitioning in
copolymerization can certainly be solved for by using empirical partition coefcients, however, these
coefcients must be determined for every copolymer system investigated. A model for a copolymer
system, when such partition coefcients are not available, will be unable to deliver adequate predictions.
Under such circumstances, monomer partitioning can only be solved for through thermodynamic
equilibrium. The use of partitioning coefcients in copolymerization is relatively straightforward, and
is very similar to what has been described in Section 2.3. In this section, emphasis will be focused on
how to solve for partitioning through more general thermodynamic equilibrium equations.
In multicomponent emulsion polymerization, the number of thermodynamic equations needed to fully
describe monomer partitioning increases drastically. In stages I and II when there are still monomer
droplets present, there is a total of three phases, namely, the polymer particle phase, the aqueous phase
and the monomer droplet phase. For polymerization involving n monomers, a total of 3n nonlinear
equations are needed. These 3n nonlinear equations must be solved for at each iteration simultaneously.
In case of copolymerization, a total of six equations must be solved compared to only one (Eq. (53)) that
has to be solved for in homopolymerization. The thermodynamic equations for the chemical potential for
monomers A and B in the three phases are listed below
DG
A;particle
=RT = ln f
p
A
1(1 2m
AB
)f
p
B
1f
p
p
1x
AB
(f
p
B
)
2
1x
Ap
(f
p
p
)
2
1f
p
B
f
p
p
(x
AB
1x
Ap
2x
Bp
m
AB
) 1
2s

V
A;m
RTr
p
(115)
DG
B;particle
=RT = ln f
p
B
1(1 2m
AB
)f
p
A
1f
p
p
1x
AB
(f
p
A
)
2
1x
Bp
(f
p
p
)
2
1f
p
A
f
p
p
(x
AB
1x
Bp
2x
Ap
m
AB
) 1
2s

V
B;m
RTr
p
(116)
DG
A;droplet
=RT = ln f
d
A
1(1 2m
AB
)f
d
B
1x
AB
(f
d
B
)
2
1
2s

V
A;m
RTr
d
(117)
DG
B;droplet
=RT = ln f
d
B
1(1 2m
AB
)f
d
A
1x
AB
(f
d
A
)
2
1
2s

V
B;m
RTr
d
(118)
DG
A;aqueous
=RT = ln
f
w
A
f
w
A;S
(119)
DG
B;aqueous
=RT = ln
f
w
B
f
w
B;S
(120)
where x
AB
is the interaction parameter between monomers A and B, x
ip
, the interaction parameter
between monomer i and polymer (i = A; B); m
A B
, the ratio of molar volume of monomer A and B,
i.e. m
AB
= V
A;m
=V
B;m
; f
p
i
; the volume fraction of monomer i in the polymer particle phase (i = A; B);
f
p
p
; the volume fraction of polymer in the polymer particle phase, f
d
i
; the volume fraction of monomer i
in monomer droplet (i = A; B); f
w
i
; the volume fraction of monomer i in aqueous phase (i = A; B); f
w
i;S
;
the volume fraction of monomer i in aqueous phase at saturation level in homopolymerization of
monomer i (i = A; B); r
d
, r
p
, the radii of monomer droplets and polymer particles, respectively, and
s is the surface tension of polymer particles or monomer droplets.
Expressions for the chemical potential of monomers A/B in the polymer particle phase, monomer
droplets and aqueous phase are expressed by Eqs. ((115)/(116)), ((117)/(118)) and ((119)/(120)), respec-
tively. The equilibrium equations in stages I and II are constructed as follow:
DG
A;particle
=RT = DG
A;droplet
=RT (121)
DG
A;particle
=RT = DG
A;aqueous
=RT (122)
DG
A;aqueous
=RT = DG
A;droplet
=RT (123)
DG
B;particle
=RT = DG
B;droplet
=RT (124)
DG
B;particle
=RT = DG
B;aqueous
=RT (125)
DG
B;aqueous
=RT = DG
B;droplet
=RT (126)
Eqs. (115)(126) are solved simultaneously for seven unknowns, i.e. the volume fraction of both
monomers in all three phases plus the volume fraction of the polymer in the polymer particle phase.
Compared to Eq. (53), Eqs. (115) and (116) involve three more interaction parameters, i.e. x
AB
, x
AP
and
x
BP
. These three parameters are usually difcult to obtain. Liu et al. [133] stated that it was very difcult
or nearly impossible to determine the interaction parameter of acrylonitrile and polyacrylonitrile in their
studies on swelling in styrene/acrylonitrile emulsion copolymerization, so they instead used solubility
information to estimate the interaction parameters.
In Eqs. (117) and (118), the last term on the right-hand side represents the contribution from surface
tension, and since monomer droplets are very large compared to polymer particles, this term can be
safely dropped. When in stage III, monomer droplets disappear, and the thermodynamic equilibrium
equations reduce to Eqs. (122) and (125).
A number of research groups solved for monomer partitioning using the thermodynamic approach.
Guillot [41,47] listed thermodynamic equilibrium equations for styrene/acrylonitrile copolymerization.
His model can predict monomer concentration in particles and copolymer composition. Mead and
Poehlein [134,135] developed a model to simulate styrene/methyl acrylate emulsion copolymerization
in a CSTR. Equilibrium equations similar to Eqs. (115)(126) were listed. Delgado et al. [136] applied
the same method to butyl acrylate/vinyl acetate copolymerization. Guo et al. [137] studied styrene
microemulsion polymerization and treated monomer partitioning in a similar way. Other groups that
used the thermodynamic approach include Saldivar et al. [4], Liu et al. [133] and Rawling and Ray
[45,138].
German's group performed detailed studies on monomer partitioning in both homopolymerization
and copolymerization. Solving the thermodynamic equilibrium equations in all three stages in homo-
polymerization was rst discussed by Maxwell et al. [139]. This group also studied the effect of particle
size, temperature and molecular weight on monomer partitioning. Monomerpolymer interaction para-
meters for several monomers (styrene, methyl acrylate, methyl methacrylate and vinyl acetate) were
estimated. Maxwell et al. [140] later extended their work to copolymerization of styrene/methyl acry-
late, and the thermodynamic equilibrium equations were simplied under several assumptions: (1)
interaction parameters x
AP
, x
BP
were assumed to be the same; (2) molar volumes for each monomer
are close to each other so m
AB
<1; (3) contributions of mixing of the two monomers in the monomer
droplets are small compared to the other terms in Eqs. (117) and (118). By using these assumptions, Eqs.
(115)(120) are greatly simplied and can be easily solved numerically. Maxwell et al. [140] investi-
gated several copolymer systems (styrene/butyl acrylate, styrene/methyl acrylate, methyl acrylate/butyl
acrylate) and experimentally determined the concentration of monomers in each phase. They concluded
that the ratio of volume fractions of the two monomers in polymer particles is very close to that in
monomer droplets. Additionally, the relationship between the concentration of monomer in the aqueous
phase and in the monomer droplet phase obeys Henry's law. Maxwell et al. [141] and Schoonbrood and
German [142] later gave more justication to the assumptions they used through parameter sensitivity
work. Noel et al. [143] used the same assumptions and studied monomer partitioning in styrene/methyl
acrylate copolymerization. Schoonbrood [144] extended the work to the terpolymer system of styrene/
methyl methacrylate/methyl acrylate using the same methodology. Though this simplied method from
German's group has been adequate in describing the specic copolymer system, its validity for other
copolymer systems has not been tested widely. It is still desirable that the full set of equilibrium Eqs.
(115)(120) be solved for numerically without being limited to certain systems.
Solving Eqs. (115)(120) is not an easy task. Conventional numerical methods for solving sets of
nonlinear equations like Gauss Elimination or LU decomposition are time consuming and they greatly
reduce the overall computation speed. Certainly an easy yet efcient method is still needed. Such a
method can be found in Ref. [145]. Armitage's et al. [145] method has several major advantages which
are easy implementation, quick convergence, and applicability to all three stages.
After implementing Armitage's algorithm, the model was tested for the validity of the algorithm. In
the model testing phase, a copolymerization of styrene/methyl methacrylate was employed using a
conventional recipe (Sty: 50 g, MMA: 50 g, SDS: 6.25 g, KPS: 1.25 g, water: 1 l, Temperature:
50 8C). Testing results are presented in Figs. 7780. Fig. 77 shows the change of monomer concentration
in the polymer particles. It is clear that in stages I and II, monomer concentration in the polymer particles
increases as conversion and particle size increase. After 30% conversion, the concentration for both
monomers in the polymer particles starts to decrease due to the depletion of monomer droplets. Fig. 78
displays the change of total volume of all three phases. The total volume of monomer droplets decreases
fromthe very start of the reaction and it gradually goes to zero around 30%, which marks the end of stage
II. The total volume of polymer particles increases before 30% due to particle growth in stages I and II. It
then decreases after 30% (in stage III) due to the density difference between monomer and polymer. The
total volume of the aqueous phase remains nearly constant throughout the entire reaction. Fig. 79 shows
the changes of the volume of each individual monomer in all three phases. It is satisfactory to observe
that both monomers disappear in monomer droplets at about the same time (around 30%). Finally, Fig. 80
Fig. 77. Calculation of monomer concentration in polymer particles in styrene and methyl methacrylate emulsion copolymer-
ization.
Fig. 78. Total volume of all three phases in styrene and methyl methacrylate emulsion copolymerization.
Fig. 80. Comparison of ratio of volume of styrene over methyl methacrylate in monomer droplets and polymer particles.
Fig. 79. Volume of individual monomers in different phases.
veries one important monomer partitioning characteristic in emulsion copolymerization, i.e. the ratio of
monomer volume (V
d
A
=V
d
B
) in polymer particles is nearly identical to the ratio of V
d
A
=V
d
B
in monomer
droplets. It is satisfactory to see that the current model is able to describe all important monomer
partitioning characteristics in emulsion copolymerization correctly.
7.3. Desorption in copolymerization
Desorption for emulsion homopolymerization has already been discussed in Section 2.4. Expressions
for the rate constant for desorption proposed by Asua et al. [52] and Nomura and Harada [51] have been
closely examined and compared. Asua's expression is chosen and implemented into the current model
for its advantages and generality. It must be stated that Asua's expression (Eq. (62)) is only applicable
for emulsion homopolymerization. It must be modied when applied to emulsion copolymerization.
Forcada and Asua [146] derived a new expression for the overall rate constant for desorption in emulsion
copolymerization as follows
k
des
= k
des;A
1k
des;B
(127)
k
des,A
and k
des,B
in Eq. (127) are the rate constants for desorption of monomeric radicals of type A and B,
respectively. They can be calculated as follows
k
des;i
= (k
fm;A
P
A
1k
fm;B
P
B
)[M
i
]
p
K
o;i
b
i
K
o;i
1k
p;A
[M
A
]
p
1k
p;B
[M
B
]
p
(128)
where (i = A or B)
P
A
=
k
p;BA
[M
A
]
p
(k
p;BA
[M
A
]
p
1k
p;AB
[M
B
]
p
)
(129)
P
B
= 1 2P
A
(130)
b
i
=
k
pw;iB
[M
B
]
w
1k
tw;iB
[R
z
]
w
k
pw;iB
[M
B
]
w
1k
tw;iB
[R
z
]
w
1k
cp;i
[N
p
] 1k
cm;i
[MIC] 1k
z;i
[Z]
w
(131)
k
cp,i
is the rate constant for radical absorption by polymer particles for monomer i, k
cm,i
, the rate constant
for radical absorption by micelles for monomer i, k
z,i
, the rate constant for reaction with inhibitor for
monomer i, and k
fm,i
is the rate constant for chain transfer to monomer for monomer i.
K
o,i
in Eq. (128) can be calculated using Eq. (63). It has been veried by Asua et al. [52] and Forcada
and Asua [146] that expression (127) can adequately calculate the overall rate constant for desorption in
emulsion copolymerization. This expression has been implemented into the current model.
7.4. Simulation of copolymerization of styrene/MMA in emulsion
A number of groups have studied this copolymer system in their kinetic studies of emulsion copo-
lymerization. Among them, Nomura's group has performed many experiments. This group also devel-
oped a model to simulate styrene/MMA emulsion copolymerization. Nomura et al. [147] rst quantied
desorption in emulsion copolymerization of this system due to the presence of MMA. Nomura et al.
[147] also investigated monomer partitioning in this copolymer system. Monomer concentration for both
monomers in polymer particles was expressed as a function of the weight fraction of MMA (w
md
) in
monomer droplets as shown below:
[M
Sty
]
p
= 22:4
1 2w
md
4 2w
md
(132)
[M
MMA
]
p
= 27:6
w
md
5 2w
md
(133)
Nomura et al. [148] further studied monomer partitioning of this copolymer system when the concen-
tration of emulsier was in the vicinity of CMC. The same group [66] proposed a method of solving for
n: Nomura et al. [149] and Nomura and Fujita [150] presented detailed model development and tested
their model with experimental data for conversion, copolymer composition and monomer concentration
in polymer particles. Research work from Nomura's group provides both kinetic information for model
development and experimental data for model testing.
Asua's group also developed a model to simulate emulsion copolymerization of styrene/MMA. They
carried out experiments to collect data for their model testing. Forcada and Asua [146] presented their
model development in detail. An expression for the overall rate constant for desorption was derived (Eq.
(127)). The same group [151] presented a method to calculate molecular weight averages for this
copolymer system. Experimental work from Asua's group is a good data source for model testing.
Model testing results using their reported data are presented in Section 7.5.
Other research groups also investigated the kinetics of emulsion copolymerization of this copolymer
Fig. 81. Simulation of styrene/methyl methacrylate emulsion copolymerization at 50 8C (styrene: 100 g/l of water, MMA:
100 g/l of water, SDS: 6.25 g/l of water).
system. Goldwasser and Rudin [152] provided some kinetic information for the rate constant of cross
chain transfer to monomer. They also presented their estimates of reactivity ratios. Rios et al. [153]
focused their work on how physical properties of this copolymer are affected by polymer composition.
Models that can simulate this copolymer system were also developed by Richards et al. [128] and
Dougherty [29,130]. Both groups tested their models with experimental data from Nomura's group.
Their research provides a lot of kinetic information for many key parameters.
7.5.1. Nomura et al. [147]
Nomura et al. [147] performed a number of experiments under various reaction conditions. Fig. 81
presents four runs with various amounts of initiator (KPS). The agreement between model predictions
and reported conversion data is good. Limiting conversion was observed in all four runs, and this is
expected since the reaction temperature is well belowthe T
g
of copolymer. The effect of emulsier on the
rate of polymerization is shown in Fig. 82. It is satisfactory to see that the current model delivers reliable
predictions on conversion versus time proles under a wide range of reaction conditions. Nomura et al.
[147] reported the total number of particles per liter of water for the runs shown in Fig. 82. Comparison
of predictions from the current model and the experimental data is shown in Fig. 83. It is obvious that
particle nucleation was completed within 510% conversion, and the total number of particles in all four
runs remained very stable in the subsequent course of the reaction. This particle nucleation characteristic
is adequately described by the model.
The effect of monomer feed composition on the rate of reaction was studied by Nomura et al. [147] in
Fig. 82. Simulation of styrene/methyl methacrylate emulsion copolymerization at 50 8C (styrene: 100 g/l of water, MMA:
100 g/l of water, KPS 1.25 g/l of water).
a number of runs. Simulation results are presented in Fig. 84. Copolymerizations with higher styrene
content exhibit lower rate of polymerization, as expected.
Copolymer composition is simulated by our model in Fig. 85. The good agreement between our model
predictions and reported measurements shown in Fig. 85 indicates the important fact that the estimation
of reactivity ratios from bulk/solution copolymerization can be used for the same copolymer system in
emulsion copolymerization provided monomer partitioning is solved satisfactorily. Model testing results
for monomer partitioning for this copolymer system are given in Figs. 86 and 87. Nomura et al. [147]
reported monomer concentration of styrene and MMA in polymer particles throughout the entire course
of the reaction. The solid lines in Figs. 86 and 87 are model's predictions by solving for the thermo-
dynamic equilibrium equations using Armitage's algorithm (described in Section 7.2). It is very satis-
factory to see that the model can adequately describe monomer concentration changes at all ve different
monomer feed ratios. The good agreement also further conrms the validity of Armitage's algorithm.
7.5.2. Forcada and Asua [146,151]
Another good data source for model testing is the work from Forcada and Asua [146]. This group
performed a set of ve runs at various monomer feed compositions. The molar ratio of styrene in the
monomer feed ranged from 10 to 90%. Comparisons of our model's predictions and reported conversion
proles are presented in Figs. 8891. In some of their runs (Figs. 8991), noticeable induction time was
observed at the beginning of the reaction. The model predictions for copolymer composition for all ve
runs are displayed in Fig. 92 along with reported data. The agreement is good. On the basis of model
testing results in Figs. 85 and 92, it can be stated that the MayoLewis equation derived for bulk/solution
polymerization can still be used in emulsion copolymerization as long as monomer partitioning is
adequately solved for.
Fig. 83. Prediction of number of particles in styrene/methyl methacrylate emulsion copolymerization at 50 8C (styrene: 100 g/l
of water, MMA: 100 g/l of water, KPS 1.25 g/l of water).
Fig. 84. Simulation of styrene/methyl methacrylate emulsion copolymerization at various monomer feeds (styrene: 100 g/l of
water, MMA: 100 g/l of water, SDS 6.25 g/l of water, KPS: 1.25 g/l of water).
Fig. 85. Calculation of copolymer composition in styrene and methyl methacrylate emulsion copolymerization at 50 8C.
Fig. 86. Prediction of styrene concentration in polymer particles by solving thermodynamic equilibrium equations.
Fig. 87. Prediction of methyl methacrylate concentration in polymer particles by solving thermodynamic equilibrium equations.
Fig. 88. Simulation of styrene and methyl methacrylate emulsion copolymerization at 50 8C (Sty 1MMA: 90 g, KPS: 0.18 g,
SDS: 2.25 g, water: 270 g).
SDS: 2.25 g, water: 270 g).
SDS: 2.25 g, water: 270 g).
SDS: 2.25 g, water: 270 g).
8. Case study: extending the nucleation stage in emulsion polymerization
8.1. Introduction
Semi-batch polymerization, a common industrial practice, involves the continuous or intermittent
addition of monomer or any other ingredient to the batch during the progress of the reaction. In emulsion
systems, the monomerswollen polymer particles can be kept starved for monomer, which leads to
higher polymerization rates and also, easier control of the reaction rate. Semi-batch systems are greatly
superior to batch systems for handling copolymerization reactions. The addition of monomer during
polymerization also has the benecial effect of providing extra cooling of the reaction mixture and thus,
permitting better temperature control. Finally, semi-batch polymerization allows the potential for better
particle size control of the product, as it will hopefully become obvious from the development of the
present paper.
8.2. An interesting control problem
To develop effective control strategies for polymer particle concentration and subsequently, for
polymer particle size, it is of considerable importance to establish the time scale for polymer particle
nucleation (i.e. of stage 1 of a typical emulsion polymerization). The effectiveness of any control scheme
would then depend on the nucleation time. For a typical case II system (e.g. styrene), the number of
SDS: 2.25 g, water: 270 g).
polymer particles, N
p
(t) generated up to any time t during stage 1, is generally given by:
N
p
(t) = R
I
N
A
t (134)
Therefore, the nally produced number of particles is
N
p
(final) = R
I
N
A
t
f
(135)
where t
f
is the nucleation time (i.e. time of completion (duration) of stage 1). If one substitutes typical
values in Eq. (135), e.g. K
2
S
2
O
8
initiator at 50 8C gives an R
I
of about 6 10
29
mol/l s, then, to nucleate
,10
18
polymer particles per liter of latex, the time t
f
required is in the order of magnitude of 5 min. For
case I kinetics, this time interval can be even less than 5 min [54,103].
From this rough result, it is clear that manipulation of initiator and/or emulsier concentrations or
owrates to control polymer particle nucleation would not be at all an easy task, under the above time
scales. In other words, the time that one has got available to interfere in the emulsion system during its
rst stage is very limited. What is more, Eq. (135) does not apply to case I systems like vinyl acetate,
since the total radical entry in these systems is much greater than the initiation rate, a fact that leads to
nucleation periods of even shorter duration.
The particle size distribution (PSD) of a fully converted latex depends on the conditions in stages 13
of the emulsion process. It depends on the original PSD at the end of stage 1 and the subsequent particle
growth rate of various sized particles in stages 2 and 3. The controlled addition of emulsier during
particle growth stabilizes particles without further particle nucleation. A second aspect in the control of
PSD is related to the particle sticky stage which often occurs at intermediate conversion levels. During
the sticky stage, wall fouling and formation of coagulum often occur. The use of semi-batch strategies is
more effective to eliminate the sticky stage as well as the problems that it leads to.
Min and Gostin [154] developed a mathematical model for the semi-batch emulsion polymerization of
vinyl chloride. The model was shown to predict the bimodal particle size distribution in seeded emulsion
polymerization and the results were in good agreement with pilot plant data. The following character-
istics and assumptions were built into their model: (1) all temperature uctuations of the reactor,
including the initial period of heat-up, were taken into account; (2) the polymer particles were con-
sidered homogenous with no internal structure; (3) polymer particles were considered to be stabilized
both by emulsier and by polymer chain ends on the particle surface; (4) polymer particle agglomeration
was empirically accounted for in the model; (5) the Trommsdorf effect was considered to be present
from the very early stages of the reaction; (6) shrinkage of reacting volume due to polymerization was
recognized, and (7) the reactor was considered well mixed and homogenous with regard to temperature
and reaction ingredients. The particle size distribution equations were transformed into a set of moment
equations for computational efciency and the effects of the following variables on the nal latex
particle size distribution were investigated: seed particle size, quantity of seed, solid content of seed
and initial amount of initiator. Some of their observations are the following:
1. A large amount of seed made micelles appear at a high conversion (,32%), resulting in a bimodal
PSD at higher conversions.
2. The predicted size of small particles was 0.07 mm smaller than that experimentally measured. This
was due to a much higher rate of agglomeration between small particles in the experiments than that
predicted by the model. Due to the same agglomeration, the concentration of small particles
decreased after micelles had disappeared.
3. Higher solids concentration in the seed led to a greater number of particles growing into large
particles. This required more emulsier coverage and left less emulsier available to form micelles.
Therefore, only a few second-generation particles were formed.
4. The model was nally used to theoretically develop techniques of metering surfactant and initiator to
control the growth of small particles.
In a similar study, Gordon and Weidner [155] developed an interesting way to control the particle size
distribution of emulsion PVC by varying the rate of emulsier addition during the polymerization. For
reproducibility, the emulsier solution should be metered according to a reaction dependent variable
such as percent conversion. By monitoring the heat liberated by the reaction, the rate of monomer
conversion, and therefore, the progress of the reaction could be followed. This liberated heat was
accurately quantied with precise conversion measurements. The common steady-state method of
measuring the heat given off was shown to be inadequate and to lead to an appreciable error in the
conversion measurements. An unsteady-state heat balance was shown to be better suited for these
calculations. With conversion thus accurately measured, they developed control algorithms which
would regulate the addition of emulsier as a function of conversion and therefore, would eventually
change the PSD of the nal product.
8.3. A possible approach to controlling the PSD
In the batch emulsion polymerization of vinyl acetate (VAc), a massive generation of polymer
particles takes place during stage 1, which often lasts less than 5 min. This is clearly shown in Fig. 93
which represents the simulation of a typical batch run with 0.0417 g mol/l of water emulsier (sodium
lauryl sulfate, SLS) and 0.00222 g mol/l of water initiator (K
2
S
2
O
8
). The ingredient concentrations for
the simulation were chosen based on typical experimentally veried levels [103]. As can be seen from
Fig. 93, the nucleation time or the duration of stage 1, i.e. the time elapsed from the beginning of the
polymerization reaction until the number of polymer particles N
p
(t) assumes a constant nal value, is
about 3 min. The initial dead time is due to typical water-soluble inhibitor levels, which give rise to
induction periods. Therefore, in a very short time one obtains the nal number of polymer particles,
which subsequently start to grow during stage 2. Since all polymer particles start to grow nearly at the
Fig. 93. Number of particles versus reaction time for a typical batch emulsion VAc run.
same time, their growth rate in stage 2 will be roughly the same and the nal particle size distribution
will be quite monodispersed. Thus, all properties of the nal product dependent on polymer particle
number and size are strictly dened from stages 1 and 2, which are both completed in roughly less than
15 min. This time interval of 1520 min was experimentally observed by several workers [54,156] and
is also correctly predicted by the batch model. This interval is, of course, not sufcient for feedback
control action to be taken according to the progress of the reaction, especially if one would like to apply
on-line feedback control based on latex property trajectories [157]. Furthermore, if the initial PSD is
narrow, the nal PSD will be quite narrow and one has to be satised with the nal product without
having the capability and exibility of interfering in the system and thus regulating the nal properties
according to various demands. Therefore, it would be of interest if one could nd a method which would
be able to extend the particle nucleation time (i.e. the duration of stage 1), and thereby broaden (or alter,
in general) the PSD and control the number of particles generated.
In the vinyl acetate case, the situation is complicated even more by the fact that the critical conversion
x
c
is ,20%, i.e. droplets disappear at low conversions, as compared to x
c
for vinyl chloride which is
,75%. Depending on the recipe, homogenous nucleation can be important for the early stages of
polymerization, since ,10
12
10
15
polymer particles can be generated by homogenous nucleation in
the absence of micelles (a typical homogenous batch run is shown in Fig. 94). The amount of emulsier
has to be chosen such that it would guarantee a particle surface coverage of ,70% during the whole
range of conversion, since the system is very sensitive to particle agglomeration. Finally, desorption of
radicals from polymer particles into the water phase is very important, causing at any instant, the total
radical entry into particles to be greater than the radical generation by initiator decomposition.
In the vinyl acetate case, the number of polymer particles generated is almost independent of initiator
concentration. On the other hand, the emulsier concentration has a pronounced effect on the number of
particles formed. Therefore, this latter variable might be used in the effort to extend the nucleation time.
The idea to be tested would be to feed the emulsier in pulses or use a very slow constant feed rate and
thus cause a series of consecutive particle generations, each one of them lasting until the free emulsier
level in the reaction mixture A
m
(t) becomes less than the critical micelle concentration (CMC). The
pulses or the slow emulsier feed rate will stop when N
p
(t) reaches a desired constant value. This point
would be the end of stage 1. It might also be possible by feeding the emulsier in pulses or by using a
very slow feed rate to generate bimodal or multimodal particle size distributions. (Equally, one could use
Fig. 94. Number of particles versus reaction time for a typical batch emulsion VAc run (homogenous nucleation).
the alternative approach of seeding with particles at prespecied times without causing any internal
particle generations.) In other words, the previously mentioned manipulation of emulsier feed rate in
the semi-batch mode would not only provide a time period sufcient to exploit feedback information
from the reacting system, but would also enable one to achieve effective control over the particle size of
the produced latex.
8.3.1. Extension of nucleation stage 1
A series of simulation runs was performed using arbitrarily chosen pulses of emulsier and
manipulating the emulsier feed rate or the total amount of emulsier in the reaction mixture. The
results showed once more that the system was sensitive to emulsier adjustments. Since developing feed
policies for emulsier directly appeared to be very difcult due to the sensitivity of the particle genera-
tion period to small changes in emulsier concentration, a more fundamental approach was taken
whereby A
m
(t), the free emulsier level in the reactor, was manipulated to give the desired PSD. The
emulsier feed rate policy needed to give this A
m
(t) was then solved for. If A
m
(t) is less than zero, then no
micelles exist in the system and the micellar contribution to particle generation is zero. If A
m
(t) is zero,
then the amount of total emulsier present in the system is such that it can bring the micellar concentra-
tion up to the CMC and cover the existing polymer particles. Finally, if A
m
(t) exceeds the zero level, then
the total emulsier present in the system is used partially to cover the surface of the existing polymer
particles, partially to bring the micellar concentration up to the CMC level, and the remainder to exceed
the CMC level and thus form micelles. These micelles will contribute to generating new polymer
particles, which in their turn will contribute to causing an increase in the total particle surface area,
which has to be covered by soap. Thus A
m
(t) will fall again and the previously described cyclical process
will occur again, if soap is fed at a high enough rate.
If N
p
(t) is the desired total number of polymer particles, then theoretically, if A
m
(t) is forced to exceed
the zero level N times by such an amount which would be capable of generating a number of (N
p
(t)/N)
polymer particles each time, N
p
(t) polymer particles will be nally generated in N consecutive particle
generations, and since each generation can be forced, at least in the simulations, to occur between
desired points in time, the above strategy would be able to extend the duration of stage 1 to some
desired time scale.
Fig. 95 shows the adjustment of A
m
(t) which would give ve successive particle generations. In
Fig. 95. Variation of A
m
(t) in order to obtain successive particle generations.
practice, the duration of the periods for which A
m
(t) exceeds the zero level must get progressively longer
for each new successive generation because of the larger particle surface area. The explanation for this is
evident from the well known functional relationship of Eq. (136) (see, for example, any book on
emulsion polymerization or Ref. [103]):
f (t) r(t)
A
m
(t)
A
m
(t) 1A
p
(t)
(136)
Eq. (136) states that the particle nucleation rate (particle net generation rate), f (t); is proportional to the
radical production rate into the water phase by initiator decomposition and by desorption, r(t); times the
ratio of A
m
(t) to (A
m
(t) 1A
p
(t)). In general, r(t) is a function of N
p
(t), and the ratio of surface areas on
the right-hand side of Eq. (136) depends upon A
p
(t) which increases with each new particle generation.
Fig. 96 gives the number of polymer particles as a function of time. The ve consecutive generations can
easily be seen and a comparison with Fig. 93 clearly shows the improvement as far as the duration of
stage 1 is concerned. In Fig. 96, the number of polymer particles assumes its nal constant value after
,40 min, which was the nal objective. In Fig. 97, the previous picture of N
p
(t) versus time is compared
with the new one, which was obtained by adjusting A
m
(t).
Fig. 96. Number of particles versus time corresponding to the A
m
(t) variation of Fig. 95.
Fig. 97. Comparison of N
p
(t) between the usual and the extended case.
8.3.2. Emulsier feed rate policy
Although A
m
(t) was used in the simulations as the adjusted variable to extend the particle
nucleation time, it may not be used in practice because it is difcult to follow it during the
course of the reaction. Additionally, even if it were possible to follow it during the reaction, its
level would be regulated from the emulsier feed rate. Therefore, it is necessary to see what sort of
emulsier feed rate policy the previous A
m
(t) versus time history shown in Fig. 95 would suggest. Before
doing that, it should be mentioned here that a series of simulation runs was performed using the identical
adjusted A
m
(t) policy of Fig. 95 and different initiator levels, all of which gave the same results
and the same nal N
p
(t) versus time picture, since N
p
(t) does not depend on initiator level for the
VAc (case I kinetics) case. The run with an initiator level of 0.005 g mol/l of water was chosen to be
presented here.
Fig. 98 shows the total particle surface area as a function of time for the adjusted A
m
(t) policy of Fig.
95, and an initiator level of 0.005 g mol/l of water. Fig. 99 shows the total emulsier level in the reaction
mixture as a function of time, as it was backcalculated from the A
m
(t) changes. The horizontal line on
Fig. 99 corresponds to the constant emulsier concentration of 0.0437 g mol/l of water of the usual batch
Fig. 98. Total polymer particle surface area corresponding to Fig. 95.
Fig. 99. Total emulsier level in the reactor corresponding to Fig. 95.
case for comparison reasons. Since
d[S]
t
(t)
dt
= S
F
(t) (137)
where S
F
(t) represents the emulsier feed rate in gram moles per unit time per unit volume, Fig. 99
suggests a constant emulsier feed rate given by the slope of the total emulsier level in the reactor
versus time curve.
9. Concluding remarks
This emulsion model is one of the very few that can simulate emulsion homo-/copolymerization under
a wide range of reaction and operation conditions (batch, semi-batch, seeded or unseeded, etc.). A
systematic approach has been adopted to develop a general model that can describe all important
physicochemical phenomena in emulsion polymerization both qualitatively and quantitatively. Great
effort has also been made to keep the model structure simple yet still rigorous. Problems encountered in
modeling particle nucleation, absorption and desorption of radicals, monomer partitioning and gel effect
in emulsion polymerization have all been addressed.
The extensiveness of the database of this simulation package is the most important feature that
makes this model distinct from most other models presented in the literature. The database
contains physical/chemical information for many monomers, initiators, emulsiers and other polymer-
ization ingredients. The richness of the database and its structure provide the emulsion model with great
exibility to simulate emulsion homo-/copolymerization of different recipes under a wide range of
conditions.
The literature review and model testing presented in this paper are also believed to be most extensive
so far. The emulsion model has been tested extensively for many monomer systems with experimental
data. In most cases, model predictions are satisfactory and this conrms the reliability of this simulation
package. Through the extensive model testing process, a large body of evidence has been gathered to
demonstrate that it is possible to model emulsion homo-/copolymerization in a satisfactory way [5]. The
limit of the modeling progress depends on ones understanding of the process.
Despite the recent advances in the understanding of emulsion polymerization kinetics and the
satisfactory simulation results presented in this paper, it must be nevertheless stated that there are
still areas that need further investigation, especially in the mechanism of particle nucleation and particle
coagulation. As for the future development of this emulsion model, now that the main mathematical
model frame has been established, it should be further tested with additional homopolymer and new
copolymer/terpolymer systems. Careful experimental design and polymer property measurements would
denitely provide valuable information for model development, parameter estimation and model testing.
During many years of effort in model developing and testing (bulk/solution and emulsion), many
difculties and problems encountered have made us ponder on the philosophy behind modeling of free-
radical polymerization in general. We thus ought to agree with the following comment: It is the mark
of an instructed mind to rest satised with the degree of precision which the nature of the subject permits
and not to seek an exactness where only an approximation of the truth is possible (Aristotle, many
centuries ago).
Acknowledgements
Financial support is gratefully acknowledged from the Natural Sciences and Engineering Research
Council (NSERC) of Canada, ICI, Worldwide, and Uniroyal, USA. Many thanks to Prof. J.M. Asua for
useful discussions.
Appendix A. Seeded emulsion polymerization
Seeded emulsion polymerization is a process widely used in both industry and research. In seeded
polymerization, polymer particles are preformed in a separate reaction and a subsequent polymerization
is then carried out. Major advantages of seeded emulsion polymerization are: (1) further complication of
reaction kinetics is avoided due to the absence of particle nucleation and (2) physical properties (particle
size, number of particles) of the preformed particles can be determined in advance. The kinetics of
seeded polymerization were studied by Napper et al. [158] and Ugelstad and Hansen [14]. In the recent
literature, almost 75% of the publications deal with seeded systems. Simulation of seeded emulsion
polymerization has been implemented in the emulsion model as a user option. When this option is
activated, the model will automatically initialize the mass balance equations for monomer, polymer and
particle number. This is described in the equations below.
The volume of the seeds can be calculated by the following
V
seed
= V
mseed
1V
pseed
=
N
seed
pd
3
pseed
6
(A1)
where d
pseed
is the diameter of the seeds, V
mseed
, the volume of monomer in the seeds, V
pseed
, the volume of
polymer in the seeds, and N
seed
is the total number of seeds.
When seeds are preformed, if there is still some monomer left in the seeds, the nal conversion level in
the seeds can be calculated as
x
seed
% =
V
pseed
r
p
V
mseed
r
m
1V
pseed
r
p
(A2)
where x
seed
% is the conversion level in the seeds.
Combining Eqs. (A1) and (A2) leads to the total volume of initial polymer in the seeds as:
V
pseed
=
N
seed
pd
3
pseed
=6
1 1
1 2x
seed
%
x
seed
%
_ _
r
p
r
m
(A3)
The initial volume of monomer in the seeds is:
V
pseed
= V
seed
2V
pseed
(A4)
The total number of seeds (particles), N
seed
, average seed diameter, d
pseed
, and the total volume of seeds,
V
seed
are usually reported, thus the volume of polymer and monomer in the seeds can be calculated by
using Eqs. (A3) and (A4). These two values are needed for initialization of the overall mass balances for
monomer and polymer in the subsequent polymerization.
Appendix B. Impurity effects on emulsion kinetics
Impurity or inhibitor plays an important role in emulsion polymerization, yet how it affects the
reaction kinetics has been rarely studied. Louie et al. [159] found that oxygen acts as inhibitor in methyl
methacrylate emulsion polymerization, however their work was limited to theory without experimental
verication. Perhaps Huo et al. [160] and Penlidis et al. [161] are one of the very few groups that
investigated impurity effects in emulsion polymerization both experimentally and theoretically. On the
basis of partitioning, different types of impurities can be classied as water-soluble or oil-soluble. It has
been demonstrated by Huo et al. [160] and Penlidis et al. [161] that these two types of impurities affect
reaction kinetics in a very different way.
Oxygen is considered as a water-soluble impurity. Huo et al. [160] discovered that oxygen acts mainly
by scavenging free radicals in the aqueous phase. This effect directly leads to induction times. After all
oxygen is consumed in the aqueous phase, polymerization starts and proceeds normally. It is expected
that other types of water-soluble impurities will behave in the same fashion. In the experiments carried
out by Huo et al. [160], various amounts of water-soluble inhibitor of hydroquinone were injected in the
batch emulsion polymerization of styrene. Different lengths of induction time were observed in their
runs. The conversion results from Huo et al. [160] are plotted in Fig. A1. The amount of hydroquinone
was between 20 and 100 ppm. Induction times of 2040 min were observed. It is important to notice that
the conversion curves for all four runs were nearly parallel to each other. This indicates that the rate of
polymerization for these runs is not affected by the presence of hydroquinone. All four runs reached a
nal conversion level of about 90%. The current model predicts the conversion prole for all four runs
well.
Fig. A1. Effect of water soluble inhibitor in styrene homopolymerization at 60 8C (styrene: 1053 g, water: 2 l, SDS: 63.18 g,
KPS: 3.263 g).
Monomer soluble impurity behaves in a much different way. Monomer soluble impurity hinders the
polymerization inside the particles, and this leads to a suppressed rate of polymerization. Furthermore,
since particles grow at a much slower rate, micelles disappear at a slower rate. The direct result of this is
a prolonged nucleation stage, which subsequently results in more particles generated. After the impurity
is consumed, the rate of polymerization can be higher than that in reactions with less or no impurity due
to the higher number of particles generated. Huo et al. [160] investigated styrene emulsion homo-
polymerization with t-butyl catechol as the monomer soluble inhibitor. Similar to their studies on
water-soluble impurities, a set of experiments was performed with different levels of t-butyl catechol.
Measured conversion data are plotted in Figs. A2A4. A small amount of t-butyl catechol will also
dissolve in the water phase and result in some induction time. The effect of t-butyl catechol on particle
size is demonstrated in Fig. A5. As expected, the average particle diameter for the run with 200 ppm
t-butyl catechol is smaller than that with 50 ppm t-butyl catechol due to a slower particle growth rate
and a much higher number of particles.
Penlidis et al. [161] further extended Huo's et al. [160] studies on impurities to vinyl acetate emulsion
polymerization. A total of ve runs were performed with various levels of t-butyl catechol. Fig. A6
shows the experiment without any t-butyl catechol added. Induction time was observed in this run due to
traces of oxygen in the aqueous phase. Penlidis et al. [161] added t-butyl catechol in the other four runs.
Figs. A7 and A8 compare our model predictions with reported conversion data. It is obvious that a higher
inhibitor level leads to a lower rate of polymerization. Oxygen was present in all runs and it resulted in
induction times of various lengths. The inhibitor effect on particle size is more pronounced in the runs
with 100 and 200 ppm t-butyl catechol. Simulation results for particle diameter for runs without inhibitor
and with 100 and 200 ppm inhibitor are presented in Figs. A9A11. The nal particle diameter for the
Fig. A2. Styrene emulsion homopolymerization without impurities at 55 8C (styrene: 100 g, water: 200 g, SDS: 6 g, KPS:
0.31 g).
Fig. A3. Styrene emulsion homopolymerization with monomer soluble impurities at 55 8C (styrene: 100 g, water: 200 g, SDS:
6 g, KPS: 0.31 g).
Fig. A4. Styrene emulsion homopolymerization with monomer soluble impurities at 55 8C (styrene: 100 g, water: 200 g, SDS:
6 g, KPS: 0.31 g).
Fig. A5. Particle size in styrene emulsion homopolymerization with t-butyl catechol at 55 8C (styrene: 100 g, water: 200 g,
SDS: 6 g, KPS: 0.31 g).
Fig. A6. Emulsion homopolymerization of vinyl acetate without monomer soluble inhibitor (VAc: 1.15 l, water: 2.86 l).
Fig. A7. Emulsion homopolymerization of vinyl acetate with monomer soluble inhibitor (VAc: 1.15 l, water: 2.86 l).
Fig. A9. Emulsion homopolymerization of vinyl acetate without monomer soluble inhibitor (VAc: 1.15 l, water: 2.86 l).
runs with inhibitor is below but close to 800 A
as shown in Figs. A10 and A11, compared to the particle

diameter for the run without inhibitor (1200 A
) shown in Fig. A9. The smaller particle size is the direct

result of the inhibitor effect on particle growth. It is satisfactory that the model is able to describe well the
effect of t-butyl catechol on both particle growth rate and the nal particle size.
References
[1] Min KW, Ray WH. On the mathematical modeling of emulsion polymerization reactors. J Macromol Sci, Rev Macromol
Chem 1974;C11:177255.
[2] Lichti G, Hawkett BS, Gilbert RG, Napper DH, Sangster DF. Styrene emulsion polymerization: particle-size distribu-
tions. J Polym Sci Chem 1981;19:92538.
[3] Penlidis A, MacGregor JF, Hamielec AE. Mathematical modeling of emulsion polymerization reactors: a population
balance approach and its applications. ACS Symp Ser 1986;313:21940.
[4] Saldivar E, Dafniotis P, Ray WH. Mathematical modeling of emulsion copolymerization reactors. I. Model formulation
and application to reactors operating with micellar nucleation. J Macromol Sci, Rev Macromol Chem 1998;38:207325.
[5] Van Herk AM, German AL. Modeling of emulsion co- and terpolymerizations: will it be ever possible? Macromol Theor
Simul 1998;7:55765.
[6] Harkins WD. J Am Chem Soc 1947;69:1428.
[7] Smith WV, Ewart RH. Kinetics of emulsion polymerization. J Chem Phys 1948;16:592.
[8] Gardon JL. Emulsion polymerization. I. Recalculation and extension of the SmithEwart theory. J Polym Sci Chem
1968;6:62341.
[9] Gardon JL. Emulsion polymerization. II. Review of experimental data in the context of the revised SmithEwart theory.
J Polym Sci Chem 1968;6:64364.
[10] Gardon JL. Emulsion polymerization. III. Theoretical prediction of the effects of slow termination rate within latex
particles. J Polym Sci Chem 1968;6:62341.
[11] Gardon JL. Emulsion polymerization. IV. Experimental verication of the theory based on slow termination rate within
latex particles. J Polym Sci Chem 1968;6:687710.
[12] Gardon JL. Emulsion polymerization. V. Lowest theoretical limits of the ratio k
t
/k
p
. J Polym Sci Chem 1968;6:28537.
[13] Harada M, Nomura M, Kojima H, Eguchi W, Nagata S. Rate of emulsion polymerization of styrene. J Appl Polym Sci
1972;16:81133.
[14] Ugelstad J, Hansen FK. Kinetics and mechanism of emulsion polymerization. Rubber Chem Technol 1976;49:536609.
[15] Alexander AE, Napper DH. Emulsion polymerization. In: Jensen DP, editor. Progress in polymer science. Oxford:
Pergamon Press, 1971. p. 14597.
[16] Gilbert RG, Napper DH. The direct determination of kinetic parameters in emulsion polymerization systems. J Macro-
mol Sci, Rev Macromol Chem Phys 1983;C23:12786.
[17] Gilbert RG. Emulsion polymerization: a mechanistic approach. New York: Academic Press, 1995.
[18] Anderson HM, Proctor SI. Redox kinetics of the peroxydisulfateironsulfoxylate system. J Polym Sci 1965;A3:2343.
[19] Hansen FK, Ugelstad J. Particle nucleation in emulsion polymerization. I. Atheory for homogeneous nucleation. J Polym
Sci Polym Chem 1978;16:195379.
[20] Fitch RM, Tsai CH. Particle formation in polymer colloids. III. Prediction of the number of particles by a homogeneous
nucleation theory. In: Fitch RM, editor. Polymer colloids. New York: Plenum Press, 1971. p. 73102.
[21] Fitch RM, Tsai CH. Homogeneous nucleation of polymer colloids. IV. The role of soluble oligomeric radicals. In: Fitch
RM, editor. Polymer colloids. New York: Plenum Press, 1971. p. 10316.
[22] Gardon JL. Emulsion polymerization. VI. Concentration of monomers in latex particles. J Polym Sci Chem
1968;6:285979.
[23] Hansen FK, Ugelstad J. Particle formation mechanism. In: Piirma I, editor. Emulsion polymerization. New York:
Academic Press, 1982. p. 5192.
[24] Song ZG, Poehlein GW. Particle formation in emulsion polymerization: transient particle concentration. J Macromol Sci
Chem 1988;A25:40343.
[25] Song, Poehlein. Particle formation in emulsion polymerization: particle number at steady-state. J Macromol Sci Chem
1988;A25:1587632.
[26] Dickinson RG. PhD Thesis. Department of Chemical Engineering, University of Waterloo, 1976.
[27] Barrett KEJ. Dispersion polymerization in organic media. New York: Wiley, 1975.
[28] Fitch RM, Shih LB. Emulsion polymerization: kinetics of radical capture by the particle. Prog Colloid Polym Sci
1975;56:111.
[29] Dougherty EP. The SCOPE dynamic model for emulsion polymerization. I. Theory. J Appl PolymSci 1986;32:305178.
[30] Dube MA, Penlidis A, Mutha RK, Cluett WR. Mathematical modeling of emulsion copolymerization of acrylonitrile/
butadiene. Ind Engng Chem Res 1996;35:443448.
[31] Priest WJ. Particle growth in the aqueous polymerization of vinyl acetate. J Phys Chem 1952;56:107783.
[32] Napper DH, Alexander AE. Polymerization of vinyl acetate in aqueous media. Part II. The kinetic behavior in the
presence of low concentration of added soaps. J Polym Sci 1962;61:127.
[33] Fitch RM. Latex particle nucleation and growth. In: Basset DR, Hamielec AE, editors. Emulsion polymers and emulsion
polymerization. Washington, DC: American Chemical Society, 1981. p. 129.
[34] Zollars RL, Chen CT, Jones DA. Kinetics of the emulsion polymerization of vinyl acetate. J Appl Polym Sci
1979;34:73342.
[35] Hawkett BS, Napper DH, Gilbert RG. Radical capture efciencies in emulsion polymerization. J Polym Sci Chem
1981;19:31739.
[36] Maxwell IA, Morrisson BR, Napper DH, Gilber RG. Entry of free radicals into latex particles in emulsion polymeriza-
tion. Macromolecules 1991;24:1629.
[37] Kshirsagar RS, Poehlein GW. Radical entry into particles during emulsion polymerization of vinyl acetate. J Appl Polym
Sci 1994;54:90921.
[38] Asua JM, Adams ME, Sudol ED. A new approach for the estimation of kinetic parameters in emulsion polymerization
systems. I. Homopolymerization under zeroone conditions. J Appl Polym Sci 1990;39:1183213.
[39] de Arbina LL, Barandiaran MJ, Gugliotta LM, Asua JM. Emulsion polymerization: particle growth kinetics. Polymer
1996;37:590716.
[40] Morton M, Kaizerman S, Altier MW. Swelling of latex particles. J Colloid Sci 1954;9:30012.
[41] Guillot J. Some thermodynamic aspects in emulsion copolymerization. Makromol Chem Suppl 1985;10/11:23564.
[42] Flory PJ. Principles of polymer chemistry. New York: Cornell University Press, 1953.
[43] Huggins ML. Physical chemistry of high polymers. New York: Wiley, 1958.
[44] Harvey, Leonard. Thermodynamics of equilibrium copolymerization in solution bulk copolymerization. Macro-
molecules 1972;5:698.
[45] Rawlings JB, Ray WH. The modeling of batch and continuous emulsion polymerization reactors. Part I. Model formula-
tion and sensitivity to parameters. Polym Engng Sci 1988;28:23756.
[46] Dimitratos YN. PhD Thesis. Department of Chemical Engineering, Lehigh University, 1989.
[47] Guillot J. In: Reichert KH, Geiseler W, editors. Emulsion polymerization high conversion polymerization polyconden-
sation. New York: Huthig & Wepf, 1986. p. 14764.
[48] Harada M, Nomura M, Eguchi W, Nagata S. Studies of the effect of polymer particles on emulsion polymerization.
J Chem Engng Jpn 1971;4:5460.
[49] Nomura M, Harada M, Nakagawara K, Eguchi W, Nagata S. The role of polymer particles in the emulsion polymeriza-
tion of vinyl acetate. J Chem Engng Jpn 1971;4:1606.
[50] Nomura M, Harada M, Eguchi W, Nagata S. In: Piirma I, Gardon JL, editors. Emulsion polymerization. New York:
Academic Press. 1976. p. 10221.
[51] Nomura M, Harada M. Rate coefcient for radical desorption in emulsion polymerization. J Appl Polym Sci
1981;26:1726.
[52] Asua JM, Sudol ED, El-Aasser MS. Radical desorption in emulsion polymerization. J Polym Sci Chem 1989;27:3903
13.
[53] Ugelstad J, Mork PC, Dahl P, Rangnes P. A kinetic investigation of the emulsion polymerization of vinyl chloride.
J Polym Sci C 1969;27:4968.
[54] Friis N, Nyhagen L. Akinetic study of the emulsion polymerization of vinyl acetate. J Appl PolymSci 1973;17:231127.
[55] Kiparissides C, MacGregor JF, Hamielec AE. Continuous emulsion polymerization of vinyl acetate. Part I. Experimental
studies. Can J Chem Engng 1980;58:4855.
[56] Kiparissides C, MacGregor JF, Hamielec AE. Continuous emulsion polymerization of vinyl acetate. Part II. Parameter
estimation and simulation. Can J Chem Engng 1980;58:5664.
[57] Kiparissides C, MacGregor JF, Hamielec AE. Continuous emulsion polymerization of vinyl acetate. Part III. Detection
of reactor performance by turbidity-spectra and liquid exclusion chromatography. Can J Chem Engng 1980;56:6571.
[58] Pollock MJ, MacGregor JF, Hamielec AE. Computer applications in applied polymer science. ACS Symp 1981:197.
[59] Adams ME, Napper DH, Gilbert RG, Sangster DF. Free radical exit from latex particles. J Chem Soc Faraday Trans
1986;82:1979.
[60] Friis N, Hamielec AE. Kinetics of vinyl chloride and vinyl acetate emulsion polymerization. J Appl Polym Sci
1975;19:97113.
[61] Casey BS, Morrison BR, Maxwell IA, Gilbert RG, Napper DH. Free radical exit in emulsion polymerization. I.
Theoretical model. J Polym Sci Chem 1994;32:60530.
[62] Morrison BR, Casey BS, Lacik I, Leslie GL, Sangster DF, Gilbert RG, Napper DH. Free radical exit in emulsion
polymerization. II. Model discrimination via experiment. J Polym Sci Chem 1994;32:63149.
[63] O'Toole JT. J Appl Polym Sci 1965;9:12917.
[64] Ugelstad J, Mork PC, Aasen JO. Kinetics of emulsion polymerization. J Polym Sci A-1 1967;5:22818.
[65] Friis N, Hamielec AE. Gel effect in emulsion of vinyl monomers. In: Piirma I, editor. Emulsion polymerization. New
York: Academic Press, 1982. p. 8291.
[66] Nomura M, Kubo M, Fujita K. Kinetics of emulsion copolymerization. III. Prediction of the average number of radicals
per particle in an emulsion copolymerization system. J Appl Polym Sci 1983;28:276776.
[67] Ballard MJ, Gilbert RG, Napper DH. Improved methods for solving the SmithEwart equations in the steady state.
J Polym Sci Lett 1981;19:5337.
[68] Li BG, Brooks BW. Prediction of average number of radicals per particle for emulsion polymerization. J Polym Sci
Chem 1993;31:2397402.
[69] Min KW, Ray WH. The computer simulation of batch emulsion polymerization reactors through a detailed mathematical
model. J Appl Polym Sci 1978;22:89112.
[70] Lichti G, Gilbert RG, Napper DH. The mechanism of latex particle formation and growth in the emulsion polymerization
of styrene using the surfactant sodium dodecyl sulfate. J Polym Sci Chem 1983;21:26991.
[71] Lin C, Ku HC, Chiu WY. Simulation model for the emulsion copolymerization of acrylonitrile and styrene in azeotropic
composition. J Appl Polym Sci 1981;26:1327.
[72] Chen SA, Wu KW. Emulsion polymerization: determination of the average number of free radicals per particle by use of
the number average volume of the particles. J Polym Sci Chem 1990;28:285766.
[73] Friis N, Hamielec AE. Kinetics of styrene emulsion polymerization. J Polym Sci 1973;11:33215.
[74] Friis N, Hamielec AE. Note on kinetics of methyl methacrylate emulsion polymerization. J Polym Sci Chem
1974;12:2514.
[75] Friis N, Hamielec AE. Gel-effect in emulsion polymerization of vinyl monomers. Polym Prepr 1975;16(1):1927.
[76] Harris B, Hamielec AE, Marten L. In: Basset DR, Hamielec AE, editors. Emulsion polymers and emulsion polymeriza-
tion, ACS Symp Ser, 195. ACS, 1981. p. 31526.
[77] Sundberg DC, Hsieh JY, Soh SK, Baldus RF. In: Basset DR, Hamielec AE, editors. Emulsion polymers and emulsion
polymerization. ACS Symp Ser 1981;195:32743.
[78] Hamielec AE, MacGregor JF. Latex reactor principles: design, operation, and control. In: Piirma I, editor. Emulsion
polymerization. New York: Academic Press, 1982. p. 31955.
[79] Gao J, Penlidis A. A comprehensive simulator/database package for reviewing free-radical homopolymerization. J
Macromol Sci Rev 1996;36(2):199404.
[80] Gao J, Penlidis A. A comprehensive simulator/database package for reviewing free-radical homopolymerization. J
Macromol Sci Rev 1998;38(4):651780.
[81] Marten FL, Hamielec AE. High conversion diffusion controlled polymerization. In: Henderson JN, Bouton TC, editors.
Polymer reactors and processes. ACS Symp Ser 1979;104:4370.
[82] Marten FL, Hamielec AE. High conversion diffusion controlled polymerization of styrene. I. J Appl Polym Sci
1982;27:489505.
[83] Russell GT, Napper DH, Gilbert RG. Termination in free-radical polymerizing systems at high conversion. Macro-
molecules 1988;21:213340.
[84] Stickler M, Panke D, Hamielec AE. Polymerization of methyl methacrylate up to high degrees of conversion: experi-
mental investigation of the diffusion-controlled polymerization. J Polym Sci Polym Chem 1984;22:224353.
[85] Allen PEM, Patrick CR. Kinetics and mechanisms of polymerization reactions. New York: Wiley, 1974.
[86] Bataille P, Van BT, Pham QB. Emulsion polymerization of styrene. I. Review of experimental data and digital simula-
tion. J Polym Sci Chem 1982;20:795810.
[87] Sudol ED, El-Aasser MS, Vanderhoff JW. Kinetics of successive seeding of monodisperse polystyrene latexes. I.
Initiation via potassium persulfate. J Polym Sci Chem 1986;24:3499513.
[88] Sudol ED, El-Aasser MS, Vanderhoff JW. Kinetics of successive seeding of monodisperse polystyrene latexes. II. Azo
initiators with and without inhibitors. J Polym Sci Chem 1986;24:351527.
[89] Morbidelli M, Storti G, Carra S. Role of micellar equilibria on modeling of batch emulsion polymerization reactors.
J Appl Polym Sci 1983;28:90119.
[90] James Jr. HL, Piirma I. Molecular weight development in styrene and methyl methacrylate emulsion polymerization.
Polym Prepr 1975;16:198204.
[91] Piirma I, Kamath VR, Morton M. The kinetics of styrene emulsion polymerization. Polym Prepr 1975;16:18791.
[92] Bataille P, Gonzalez A. Emulsion polymerization of styrene. III. Effect of Ti(III) and Cl(1) on the polymerization of
styrene initiated by potassium persulfate. J Polym Sci Chem 1984;22:140917.
[93] Bataille F, Bataille P, Fortin R. Emulsion polymerization of styrene. IV. Effect of Co(II), Ni(II), and Pb(II) on the
polymerization of styrene initiated by potassium persulfate. J Polym Sci Chem 1988;26:14717.
[94] Canegallo S, Storti G, Morbidelli M, Carra S. Densimetry for on-line conversion monitoring in emulsion homo- and
copolymerization. J Appl Polym Sci 1993;47:96179.
[95] de La Rosa LV, Sudol ED, El-Aasser MS, Klein A. Details of the emulsion polymerization of styrene using a reaction
calorimeter. J Polym Sci Chem 1996;34:46173.
[96] Blackley DC, Sebastian SARD. Inuence of surfactant type and purity upon emulsion copolymerization of styrene and
acrylic acid. Br Polym J 1987;19:2530.
[97] Blackley DC, Sebastian SARD. Effects of inorganic electrolytes upon emulsion polymerisation reactions. I. Effects upon
kinetics of styrene emulsion. Br Polym J 1989;31:31326.
[98] Mayer MJJ, van den Boomen FHAM, Paquet DA, Meuldijk J, Thoenes D. The polymerization rate of freshly nucleated
particles during emulsion polymerization with micellar nucleation. J Polym Sci Chem 1996;34:174751.
[99] Salazar A, Gugliotta LM, Vega JR, Meira GR. Molecular weight control in a starved emulsion polymerization of styrene.
Ind Engng Chem Res 1998;37:358291.
[100] El-Aasser MS, Vanderhoff JW. Emulsion polymerization of vinyl acetate. London: Applied Science Publishers,
1981.
[101] Chern CS, Poehlein GW. Reaction kinetics of vinyl acetate emulsion polymerization. J Appl Polym Sci 1987;33:2117
36.
[102] Friis N, Goosney D, Wright JD, Hamielec AE. Molecular weight and branching development in vinyl acetate emulsion
polymerization. J Appl Polym Sci 1974;18:124759.
[103] Penlidis A, MacGregor JF, Hamielec AE. A theoretical and experimental investigation of the batch emulsion polymer-
ization of vinyl acetate. Polym Proc Engng 1985;3:185218.
[104] Zimmt WS. The emulsion polymerization of methyl methacrylate. J Appl Polym Sci 1959;1:3238.
[105] Chang HR, Parker HY, Westmoreland DG. Continuous ESR measurement of propagating free-radical concentrations for
batch emulsion polymerization of methyl methacrylate. Macromolecules 1992;25:55578.
[106] Ballard MJ, Napper DH, Gilbert RG. Kinetics of emulsion polymerization of methyl methacrylate. J Polym Sci Chem
1984;22:322553.
[107] Ballard MJ, Napper DH, Gilbert RG, Sangster DF. Termination-rate coefcients in methyl methacrylate polymeriza-
tions. J Polym Sci Chem 1986;24:102741.
[108] Nomura M, Fujita K. Kinetics and mechanisms of unseeded emulsion polymerization of methyl methacrylate. Polym
React Engng 1994;2:31745.
[109] Ballard MJ, Gilbert RG, Napper DH, Pomery PJ, O'Sullivan PW, O'Donnel JH. Propagation rate coefcients from
electron spin resonance studies of the emulsion polymerization of methyl methacrylate. Macromolecules 1986;19:1303
8.
[110] Barton J, Hlouskova Z, Juranicova V. Partitioned polymerization. 4. Seeded emulsion polymerization of methyl metha-
crylate. Makromol Chem 1992;193:16777.
[111] Dube AM. PhD Thesis. Department of Chemical Engineering, University of Waterloo, 1994.
[112] Fontenot K, Schork FJ. Simulation of mini/macro emulsion polymerizations. I. Development of the model. Polym React
Engng J 1993;1:75109.
[113] Fontenot K, Schork FJ. Simulation of mini/macro emulsion polymerization. II. Sensitivities and experimental compar-
ison. Polym React Engng J 1993;1:289342.
[114] Fontenot K, Schork FJ. Batch polymerization of methyl methacrylate in mini/macroemulsions. J Appl Polym Sci
1993;49:63355.
[115] Schork FJ, Ray WH. On-line monitoring of emulsion polymerization reactor dynamics. In: Bassett DR, Hamielec AE,
editors. Emulsion polymers and emulsion polymerization. ACS Symp Ser 1981;165:50514.
[116] Maxwell IA, Napper DH, Gilbert RG. Emulsion polymerization of butyl acrylate. Kinetics of particle growth. J Chem
Soc Faraday Trans 1987;83:144967.
[117] Capek I, Fouassier JP. Kinetics of photopolymerization of butyl acrylate in direct micelles. Eur Polym J 1997;33:173
81.
[118] Barandiaran MJ, Adams ME, De La Cal JC, Sudol ED, Asua JM. New approach for the estimation of kinetic parameters
in emulsion polymerization systems. II. Homopolymerization under conditions where n .0.5. J Appl Polym Sci
1992;45:218797.
[119] Mallya P, Plamthottam SS. Termination rate constant in butyl acrylate batch emulsion polymerization. Polym Bull
1989;21:497504.
[120] Lovell PA, Shah TH, Heatley F. Chain transfer to polymer in emulsion polymerization of n-butyl acrylate studied by
13
C
NMR spectroscopy and GPC. Polym Commun 1991;32:98103.
[121] Capek I, Potisk P. Radical emulsion polymerization of n-butyl acrylate and its copolymerization with acrylonitrile in the
presence of crosslinker. Polym J 1992;24:103748.
[122] Cruz MA, Palacios J, Garcia-Rejon A, Ruiz LM, Rios Guerrero L. Copolymerisation en emulsion styreneacrylate de
butyl en reacteur ferme. Makromol Chem Suppl 1985;10/11:87103.
[123] Kong XZ, Pichot C, Guillot J. Kinetics of emulsion copolymerization of vinyl acetate with butyl acrylate. Eur Polym J
1988;24:48592.
[124] Adhikari MS, Sarkar S, Banerjee M, Konar RS. Thermal decomposition of potassium persulfate in aqueous solution at
50 8C in the presence of ethyl acrylate. J Appl Polym Sci 1987;34:10925.
[125] McCurdy KG, Laidler K. Rates of polymerization of acrylates and methacrylates in emulsion systems. Can J Chem
Engng 1964;42:8258.
[126] Shoaf GL, Poehlein GW. Batch and continuous emulsion copolymerization of ethyl acrylate and methacrylic acid.
Polym Plast Technol Engng 1989;28:289317.
[127] Capek I, Kostrubova J, Barton J. Effect of a bi-unsaturated monomer on the emulsion polymerization of ethyl acrylate.
Makromol Chem Macromol Symp 1990;31:21326.
[128] Richards JR, Congalidis JP, Gilbert RG. Mathematical modeling of emulsion copolymerization reactors. J Appl Polym
Sci 1989;37:272756.
[129] Congalidis JP, Richards JR, Bilbert RG. Mathematical modeling of emulsion copolymerization reactors. ACS Symp Ser
1989;404:36078.
[130] Dougherty EP. The SCOPE dynamic model for emulsion polymerization. II. Comparison with experiment and applica-
tions. J Appl Polym Sci 1986;32:307995.
[131] Broadhead TO, Hamielec AE, MacGregor JF. Dynamic modeling of the batch, semi-batch and continuous production of
styrene/butadiene copolymers by emulsion polymerization. Makromol Chem Suppl 1985;10/11:10528.
[132] Hamielec AE, MacGregor JF, Penlidis A. Multicomponent free-radical polymerization in batch, semi-batch and contin-
uous reactors. Makromol Chem Macromol Symp 1987;10/11:52170.
[133] Liu X, Nomura M, Fujita K. Thermodynamic correlation of partial and saturation swelling of styreneacrylonitrile
copolymer particles by styrene and acrylonitrile monomers. J Appl Polym Sci 1997;64:9319.
[134] Mead RN, Poehlein GW. Emulsion copolymerization of styrenemethyl acrylate and styreneacrylonitrile in contin-
uous stirred reactor tanks. 1. Int Engng Chem Res 1988;28:51.
[135] Mead RN, Poehlein GW. Free-radical transport from latex particles. J Appl Polym Sci 1989;38:10522.
[136] Delgado J, El-Aasser MS, Silebi CA, Vanderhoff JW, Guillot J. Miniemulsion copolymerization of vinyl acetate and
butyl acrylate. II. Mathematical model for the monomer transport. J Polym Sci Phys 1988;26:1495517.
[137] Guo JS, Sudol ED, Vanderhoff JW, El-Aasser MS. Modeling of the styrene microemulsion polymerization. J Polym Sci
Chem 1992;30:70312.
[138] Rawlings JB, Ray WH. The modeling of batch and continuous emulsion polymerization reactors. Part II: Comparison
with experimental data from continuous stirred tank reactors. Polym Engng Sci 1988;28:25774.
[139] Maxwell IA, Kurja J, Van Doremaele GHJV, German AL, Morrison BR. Partial swelling of latex particles with
monomers. Makromol Chem 1992;193:204963.
[140] Maxwell IA, Kurja J, Van Doremaele GHJ, German AL. Thermodynamics of swelling of latex particles with two
monomers. Makromol Chem 1992;193:206580.
[141] Maxwell IA, Noel LFJ, Schoonbrood HAS, German AL. Thermodynamics of swelling of latex particles with two
monomers: a sensitivity analysis. Makromol Chem, Theor Simul 1993;2:26974.
[142] Schoonbrood HAS, German AL. Thermodynamics of monomer partitioning in polymer lattices: effect of molar volume
of the monomers. Macromol Rapid Commun 1994;15:25964.
[143] Noel LFJ, Van Zon JMAM, Maxwell IA, German AL. Prediction of polymer composition in batch emulsion copoly-
merization. J Polym Sci Chem 1994;32:100926.
[144] Schoonbrood HAS. PhD Thesis. University of Eindhoven, 1994.
[145] Armitage PD, De La Cal JC, Asua JM. Improved methods for solving monomer partitioning in emulsion copolymer
systems. J Appl Polym Sci 1994;51:198590.
[146] Forcada J, Asua JM. Modeling of unseeded emulsion copolymerization of styrene and methyl methacrylate. J Polym Sci
Chem 1990;28:9871009.
[147] Nomura M, Yamamoto K, Horie I, Fujita K. Kinetics of emulsion copolymerization. II. Effect of free radical desorption
on the rate of emulsion copolymerization of styrene and methyl methacrylate. J Appl Polym Sci 1982;27:2483501.
[148] Nomura M, Fujita K, Kouno Y, Satpathy US. Investigation on the locus of particle formation in emulsion polymerization
containing partially water-soluble monomers. J Polym Sci C 1988;26:38590.
[149] Nomura M, Horie I, Kubo M. Kinetics and mechanism of emulsion copolymerization. IV. Kinetic modeling of emulsion
copolymerization of styrene and methyl methacrylate. J Appl Polym Sci 1989;37:102950.
[150] Nomura M, Fujita K. On the prediction of the rate of emulsion copolymerization and copolymer composition. Makromol
Chem Suppl 1985;10/11:2542.
[151] Forcada J, Asua JM. Emulsion copolymerization of styrene and methyl methacrylate. II. Molecular weights. J Polym Sci
Chem 1991;29:123142.
[152] Goldwasser JM, Rudin A. Emulsion copolymerization of styrene and methyl methacrylate. J Polym Sci Chem
1982;20:19932006.
[153] Rios L, Garcia-Rejon A, Cruz MA, Palacios J, Ruiz LM. Batch and semicontinuous emulsion copolymerization styrene/
methyl methacrylate: kinetics and viscoelastic properties. Makromol Chem Suppl 1985;10/11:47788.
[154] Min KW, Gostin HI. Simulation of semi-batch emulsion polymerization reactors for polyvinyl chloride (PVC) system.
Ind Engng Chem Prod Des Dev 1979;18(4):2728.
[155] Gordon JL, Weidner KR. Emulsion polymers and emulsion polymerization. ACS Symp 1981;165:51532.
[156] Keung CKJ. Masters Thesis. Department of Chemical Engineering, McMaster University, 1974.
[157] MacGregor JF, Penlidis A, Hamielec AE. Control of polymerization reactors: a review. Polym Proc Engng 1984;2(2/
3):179206.
[158] Napper DH, Netschey A, Alexander AE. Effects of nonionic surfactants on seeded polymerization. J Polym Sci A-1
1971;9:8189.
[159] Louie BM, Chiu WY, Soong DS. Control of free-radical emulsion polymerization of methyl methacrylate by oxygen
injection. I. Modeling study. J Appl Polym Sci 1985;30:3189223.
[160] Huo BP, Campbell JD, Penlidis A, MacGregor JF, Hamielec AE. Effect of impurities on emulsion polymerization. Case
II. Kinetics. J Appl Polym Sci 1988;35:200921.
[161] Penlidis A, MacGregor JF, Hamielec AE. Effect of impurities on emulsion polymerization. Case I. Kinetics. J Appl
Polym Sci 1988;35:2023.
[162] Giannetti E, Storti G, Morbidelli M. Emulsion polymerizations. II. Kinetics, molecular weight distributions, and polymer
microstructure of emulsion copolymers. J Polym Sci Chem 1988;26:230743.
[163] Giannetti E. Emulsion polymerizations. 3. Theory of emulsion copolymerization kinetics. Macromolecules
1989;22:2094102.
[164] Hamielec AE, MacGregor JF, Penlidis A. Modeling copolymerizationscontrol of composition, chain microstructure,
molecular weight distribution, long chain branching and crosslinking. In: Reichert KH, Geiseler W, editors. Polymer
reaction engineering. Berlin: Springer-Verlag, 1983. p. 217.
[165] Lichti G, Gilbert RG, Napper DH. Theoretical predictions of the particle size and molecular weight distributions in
emulsion polymerization. In: Piirma I, editor. Emulsion polymerization. New York: Academic Press, 1982. p. 94144.
[166] Min KW, Ray WH. Structure framework for modeling emulsion polymerization reactors. ACS Symp Ser 1976;24:359
66.
[167] Penlidis A, Hamielec AE, MacGregor AE. Dynamic modeling of the continuous emulsion polymerization of vinyl
chloride. J Vinyl Technol 1984;6:13442.
[168] Penlidis A, MacGregor JF, Hamielec AE. Dynamic modeling of emulsion polymerization reactors. AIChE J
1985;31:8819.
[169] Rawlings JB. PhD Thesis. Department of Chemical Engineering, University of Wisconsin, 1985.
[170] Storti G, Morbidelli M, Carra S. Comput Appl Appl Polym Sci II ACS Symp Ser 1989;404:79402.
[171] Storti G, Polotti G, Cociani M, Morbidelli M. Molecular weight distribution in emulsion polymerization. I. The homo-
polymer case. J Polym Sci Polym Chem 1992;30:73150.
[172] Urretabizkaia A, Asua JM. High solids content emulsion terpolymerization of vinyl acetate, methyl methacrylate, and
butyl acrylate. I. Kinetics. J Polym Sci Chem 1994;32:176178.
[173] Urretabizkaia A, Arzamendi G, Unzue MJ, Asua JM. High solids content emulsion terpolymerization of vinyl acetate,
methyl methacrylate, and butyl acrylate. II. Open loop composition control. J Polym Sci Chem 1994;32:177988.
[174] Asua JM, De La Cal JC. Entry and exit rate coefcients in emulsion polymerization of styrene. J Appl Polym Sci
1991;42:186977.
[175] Campbell JD. MASc Thesis. Department of Chemical Engineering, McMaster Univeristy, 1985.
[176] Giannetti E. Nucleation mechanisms and particle size distributions of polymer colloids. AIChE J 1993;39:121027.
[177] Miller CM, Clay PA, Gilbert RG, El-Aasser MS. An experimental investigation on the evolution of the molecular weight
distribution in styrene emulsion polymerization. J Polym Sci Chem 1997;35(6):9891006.
[178] Morton M, Salatiello PP, Landeld H. Absolute propagation rates in emulsion polymerization. III. Styrene and isoprene.
J Polym Sci 1952;VIII:27987.
[179] Nomura M, Kojima H, Harada M, Eguchi W, Nagata S. Continuous ow operation in emulsion polymerization of
styrene. J Appl Polym Sci 1971;15:67591.
[180] Penboss IA, Napper DH, Gilbert RG. Styrene emulsion polymerization. J Chem Soc Faraday Trans 1983;79:125771.
[181] Said ZFM. Molecular-weight distributions in the thermally initiated emulsion polymerization of styrene. Makromol
Chem 1991;192:40514.
[182] Badran AS, Moustafa AB, Yehia AA, Shendy SMM. Emulsion polymerization of vinyl acetate initiated by potassium
persulfatecyclohexanone sodium bisulte redox pair system. J Polym Sci Chem 1990;28:41124.
[183] Bataille F, Dalpe JF, Dubuc F, Lamoureux L. The effect of agitation on the conversion of vinyl acetate emulsion
polymerization. J Appl Polym Sci 1990;39:181520.
[184] de la Cal JC, Adams ME, Asua JM. Parameter estimation in emulsion copolymerization. Makromol Chem Macromol
Symp 1990;35/36:2340.
[185] Dimitratos J, El-Aasser MS, Georgakis C, Klein A. Pseudosteady states in semicontinuous emulsion copolymerization.
J Appl Polym Sci 1990;40:100521.
[186] Dunn AS, Taylor PA. The polymerization of vinyl acetate in aqueous solution initiated by potassium persulfate at 60 8C.
Makromol Chem 1965;83:20719.
[187] Farber JN. Steady state multiobjective optimization of continuous copolymerization reactors. Polym Engng Sci
1986;26:499507.
[188] Greene RK, Gonzalez RA, Poehlein GW. Emulsion polymerization. In: Piirma I, editor. ACS Symp Ser 1976;24:341
58.
[189] Kiparissides C, MacGregor JF, Hamielec AE. Continuous emulsion polymerization. Modeling oscillations in vinyl
acetate polymerization. J Appl Polym Sci 1979;23:40118.
[190] Lange DM, Poehlein GW, Hayashi S, Komatsu A, Hirai T. Kinetic analysis of seeded emulsion polymerization of vinyl
acetate. J Polym Sci Chem 1991;29:78592.
[191] Lee CH, Mallinson RG. Surfactant effects in the emulsion polymerization of vinyl acetate. J Appl Polym Sci
1990;39:220518.
[192] Lee CH, Mallinson RG. A model for molecular weights in vinyl acetate emulsion polymerization. AIChE J
1988;34:8408.
[193] Lichti G, Gilbert RG, Napper DH. The growth of polymer colloids. J Polym Sci Chem 1977;15:195771.
[194] Litt MH, Stannett V. Emulsion polymerization of vinyl acetate. II. J Polym Sci A-1 1970;8:360749.
[195] Litt MH, Chang KHS. In: Basset DR, Hamielec AE, editors. Emulsion polymers and emulsion polymerization, ACS
Symp Ser, 168. 1981. p. 45570.
[196] Meuldijk J, van Strien CJG, van Doraemale FAHC, Thoenes D. A novel reactor for continuous emulsion polymerization.
Chem Engng Sci 1992;47:26038.
[197] Misra SC, Pichot C, El-Aasser MS, Vanderhoff JW. Batch and semicontinuous emulsion copolymerization of vinyl
acetatebutyl acrylate. II. Morphological and mechanical properties of copolymer latex lms. J Polym Sci Chem
1983;21:238396.
[198] Moustafa AB, Abd El-Hakim AA, Mohamed GA. Some parameters affecting the emulsier-free emulsion polymeriza-
tion of vinyl acetate. J Appl Polym Sci 1997;63:23946.
[199] Omi S, Kushibiki K, Iso M. The computer modeling of multicomponent, semibatch emulsion copolymerization. Polym
Engng Sci 1987;27:47082.
[200] Sarkar S, Adhikari MS, Banerjee M, Konar RS. Mechanism of persulfate decomposition in aqueous solution at 50 8C in
the presence of vinyl acetate and nitrogen. J Appl Polym Sci 1990;39:106177.
[201] Singh S, Hamielec AE. Liquid exclusion chromotography. A technique for monitoring the growth of polymer particles in
emulsion polymerization. J Appl Polym Sci 1978;22:57784.
[202] Trivedi MK, Rajagopal KR, Joshi SN. Emulsion polymerization of vinyl acetate. J Polym Sci Chem 1983;21:20116.
[203] Urquiola MB, Sudol ED, Dimonie V, El-Aasser MS. Emulsion polymerization of vinyl acetate using a polymerizable
surfactant. III. Mathematical model. J Polym Sci Chem 1993;31:140315.
[204] Ley GJM, Schneider C, Hummel DO. Half-life measurements in emulsion polymerization. J Polym Sci, C 1969;27:119
37.
[205] Soh SK. Measurements of propagation rate constant from emulsion polymerization. J Appl Polym Sci 1980;25:29938.
[206] Barton J. On the initiation mechanism in emulsion polymerization systems. Makromol Chem Macromol Symp
1990;31:1123.
[207] Schork FJ, Ray WH. The dynamics of the continuous emulsion polymerization of methylmethacrylate. J Appl Polym Sci
1987;34:125976.
[208] Tanrisever T, Okay O, Sonmezoglu C. Kinetics of emulsier-free emulsion polymerization of methyl methacrylate. J
Appl Polym Sci 1996;61:48593.
[209] Wang HH, Chu HH. The stabilization effect of mixed-surfactants in the emulsion polymerization of methyl methacry-
late. Polym Bull 1990;24:20714.
[210] Capek I, Barton J, Karpatyova A. Emulsion polymerization of butyl methacrylate initiated by 2,2
/
-azoisobutyronitrile.
3. On the applicability of the modied SmithEwart model. Makromol Chem 1987;188:70310.
[211] Capek I. Emulsion polymerization of butyl acrylate. 2. Effect of the initiator type and concentration. Macromol Chem
Phys 1994;195:113746.
[212] Capek I. Photopolymerization of butyl acrylate in direct emulsier-mixed micelles. Makromol Rep 1996;33:20917.
[213] Chern CS, Chen YC. Semibatch emulsion polymerization of butyl acrylate stabilized by a polymerizable surfactant. Prog
Polym Sci 1996;28:62732.
[214] Chu HH, Lin CC. The stabilization effect of mixed-surfactants in the emulsion polymerization of n-butyl acrylate. Polym
Bull 1992;28:41926.
[215] El-Aasser MS, Makgawinata T, Vanderhoff JW. Batch and semicontinuous emulsion copolymerization of vinyl acetate
butyl acrylate. I. Bulk, surface, colloidal properties of copolymer latexes. J Polym Sci Chem 1983;21:236382.
[216] Kukulj D, Davis TP, Suddaby KG, Haddleton DM, Gilbert RG. Catalytic chain transfer for molecular weight control in
the emulsion homo- and copolymerizations of methyl methacrylate and butyl methacrylate. J Polym Sci Chem
1997;35:85978.
[217] Maxwell IA, Morrisson BR, Napper DH, Gilber RG. The effect of chain transfer agent on the entry of free radicals in
emulsion polymerization. Makromol Chem 1992;193:303.
[218] Maxwell IA, Verdurmen EMFJ, German AL. High-conversion emulsion polymerization. Makromol Chem
1992;193:267795.
[219] McBain CBD, Piirma I. Inuence of steric length in electrosteric surfactants on emulsion polymerization. J Appl Polym
Sci 1989;37:141522.
[220] O'Neill T, Hoigne J. g-Radiation-initiated emulsion polymerization of n-butyl acrylate and of methyl methacrylate. J
Polym Sci A-1 1972;10:58193.
[221] Arzamendi G, Leiza JR, Asua JM. Semicontinuous emulsion copolymerization of methyl methacrylate and ethyl
acrylate. J Polym Sci Chem 1991;29:154959.
[222] Capek I. On the kinetics of emulsion copolymerization of methyl methacrylate and ethyl acrylate in the presence of
hydrocarbons. Makromol Chem 1992;193:142338.
[223] Fitch RM, Palmgren TH, Aoyagi T, Zuikov A. Kinetics of particle nucleation and growth in the emulsion polymerization
of acrylic monomers. J Polym Sci Polym Symp 1985;72:2214.
[224] Maw T, Piirma I. Emulsion polymerization in the presence of nonionic surfactant. Polym Prepr 1983;24:712.
[225] Snuparek Jr. J. The effectiveness of some commercial emulsiers in emulsion polymerization. I. Die Angew Makromol
Chem 1980;88:618.
[226] Yeliseyeva VI, Zuikov AV. Interfacial phenomena in emulsion polymerization of polar monomers. In: Piirma I, Gardon
JL, editors. Emulsion polymerization. ACS Symp Ser 1982;24:6281.

Mathematical Modeling and Computer Simulatordatabase For Emulsion Polymerizations

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mathematical Modeling and Computer Simulatordatabase For Emulsion Polymerizations

Uploaded by

Copyright:

Available Formats

Mathematical modeling and computer simulator/database for

. This is rather large compared to the diameter of particles, which is

4.1. Model testing results

error involved in the

as shown in Figs. A10 and A11, compared to the particle

) shown in Fig. A9. The smaller particle size is the direct

You might also like