Polymerization in Nonuniform Latex

Particles: Distribution of Free Radicals

CHORNG-SHYAN CHERN and GARY W. POEHLEIN, S~h00lof
Chemical Engineering, Georgia Institute of Technology,Atlanta,
Georgia 30332-0100

Synopsis
Previous kinetic studies in emulsion polymerization have almost always involved an assumption
of uniform distribution of free radicals in the latex particle. Such an assumption is not likely to
reflect reality in many systems that employ water-soluble initiators because the hydrophilic
end-group of the oligomeric free radical will preferentially stay in the surface layer of the particle.
This constrained end-group location would result in nonuniform distribution of free radicals in
the polymerizing latex particles. A Monte Car10 simulation of the growth of a single polymer
chain within the latex particle supports this hypothesis. Such a nonuniform distribution of free
radicals in the latex particle is expected to have an influence on reaction kinetics and product
properties. The mechanism for transport of free radicals out of polymerizing latex particles is
reexamined based on the proposed concept, and a modified expression for the desorption rate
constant is presented.

INTRODUCTION
Emulsion polymerization is an important commercial technique used to
produce polymeric materials such as synthetic elastomers, coatings, adhesives,
and scientific and medical colloids. The process can be used to form high
molecular weight polymers at high rates in a low viscosity medium.
The main ingredients of an emulsion polymerization system include mono-
mer, dispersant, emulsifier, and initiator. Water is commonly used as the
dispersant. A water-insoluble monomer can be dispersed in water by means of
an oil-in-water emulsifier and polymerized with a water-soluble initiator. The
latex product is a colloidal dispersion of polymer particles that can often be
used without separation.
Harkins' postulated a reaction scheme in which emulsion polymerization
was divided into three distinct intervals: (1)nucleation of polymer particles in
the presence of both micelles and monomer droplets, (2) growth of polymer
particles in the absence of micelles, and (3) polymerization within the mono-
mer-swollen polymer particles in the absence of both micelles and monomer
droplets.
Smith and Ewart2 treated the Harkins theory in a quantitative manner.
They developed the well-known steady-state recursion equation for calcula-
tion of the average number of free radicals per particle. They did not obtain a
general solution. However, three limiting cases were considered. Case 2,
sometimes called ideal emulsion polymerization, is based on the assumption
that two growing free radicals cannot exist in one polymer particle at the
same time because of the very rapid termination reaction.

Journal of Polymer Science: Part A: Polymer Chemistry, Vol. 25, 617-635 (1987)
0 1987 John Wiley & Sons, Inc. CCC 0360-3676/87/020617-19$04.~
618 CHERN AND POEHLEIN

Stockmayer3 and O'Toole4 presented general solutions to the recursion

equation. The average number of free radicals per particle was expressed as a
function of free radical absorption and desorption rates. Ugelstad et al.5
further considered the effects of free radical desorption from latex particles,
reabsorption, and mutual termination in the aqueous phase. Ugelstad pre-
sented a solution to the recursion formula based on a mass balance of free
radicals in the reaction system. Poehlein et al.637expanded this work to take
into account the broad particle size distributions encountered in a steady-state
seed-fed continuous stirred-tank reactor (CSTR). Their kinetic model has
proved to be very successful in predicting the reaction rate, particle size
distribution (PSD), and molecular weight distribution (MWD) for the latex
product formed in a steady-state seed-fed CSTR.
The recursion equation of Smith and Ewart and all of the published
solutions involve a conventional assumption that free radicals are distributed
uniformly in the polymer particles in a stochastic sense. This is not a
reasonable assumption in many emulsion polymerization systems because the
hydrophilic initiator end-group of the oligomeric free radical acts as a surfac-
tant and tends to stay a t the surface of the polymer particle. Such a
constrained end-group location could result' in a nonuniform distribution of
free radicals in the polymer particles. Persulfate, for example, is a commonly
used water-soluble initiator. The oligomeric free radical initiated with per-
sulfate would have an SO; ion a t the end of the molecule. This hydrophilic
end-group would preferentially remain in contact with the aqueous phase a t
the polymer particle surface. In this manner, the movement of a growing free
radical by Brownian motion and/or by a propagation event would be re-
stricted near the surface of the polymer particle.
Considerable evidence can be found in the literature to support this hy-
pothesis, a t least in a qualitative sense. Williams et al.'?' studied emulsion
polymerization of styrene and proposed that the morphology of the latex
particle during polymerization was composed of a polymer-rich core sur-
rounded by a monomer-rich shell (core-shell morphology). Williams' initial
arguments were based on conversion-versus-time data. Their results showed a
constant polymerization rate well beyond the end of Interval I1 (up to about
60%conversion). They suggested that, this experimental observation could be
explained by constant values of monomer concentration a t the polymerization
site in the particles and average number of free radicals per particle. These
arguments supported the proposed core-shell morphology.
Napper" considered the ordinary laws of diffusion and pointed out that
diffusion rates of the species present in the polymer particle would not
support the concept of such large differences of polymer and monomer
concentrations between the core phase and the shell phase. Gardon" showed
that the diffusive mean free path of monomer molecules would be much larger
than the radius of the polymer particle, and, under this condition, a monomer
concentration gradient within a latex particle was unlikely to exist. He further
indicated that the results of the nearly constant polymerization rate of
Williams et al. could be attributed to the compensating effects of decreasing
monomer concentration in the polymer particles (owing to reaction conver-
sion) and increasing average number of free radicals per particle (owing to the
gel effect) during Interval 111.
 NONUNIFORM LATEX PARTICLES 619

Nevertheless, Williams' later workg utilized radioactive styrene that was

introduced into the reaction system a t specific times during the polymeriza-
tion. A structural analysis of the resulting latex particles clearly showed
that new polymer was formed in the outer shell of the polymer particles.
Vanderhoff l2 attempted to reconcile the different viewpoints by suggesting
that the new polymer was formed near the surface layer of the latex particle
owing to the hydrophilic end-groups (e.g., SO;). He demonstrated, via con-
ductometric titration, that a relatively large fraction (about 30-60%) of SO,
end-groups were found on the surface. This experimental observation provided
a plausible explanation for the nonuniform particle morphology observed by
Williams and coworkers. Recently Fisher et al.13 studied the emulsion poly-
merization of perdeuterated methyl methacrylate in the presence of poly-
methyl methacrylate seed latex. By using the small-angle neutron scattering
(SANS) analysis they found that the resultant polymer particles had a
core-shell morphology. Their work provides strong supporting evidence for
Williams' results.

CORE-SHELL EMULSION POLYMERIZATION

Since Williams et al.**' reported experimental evidence for nonuniform
particle morphology, many controversial arguments and uncertainties regard-
ing the core-shell theory have been presented. An attempt is made here to
establish a quantitative physical model to help understand the mechanisms
involved in core-shell emulsion polymerization.
Conventionally, it is presumed that monomer and polymer molecules
are uniformly distributed throughout the latex particle. On the contrary,
Williams and coworkers presented experimental evidence for nonuniform
particle morphology. Based on kinetic and thermodynamic considerations,
they proposed that, under a wide range of conditions, growing polymer
particles exhibit a distributed structure in which a monomer-rich shell en-
capsulates a polymer-rich core. These represent two opposing viewpoints
concerning this controversial but very interesting issue.
It is believed that monomer molecules can freely move in the polymer
particle owing to their relatively small size. Thus, monomer is most likely to
be uniformly distributed within the polymer particle. To verify this concept,
an equation for the calculation of monomer concentration profiles in a poly-
mer particle, derived by Klein,14 was employed in this study. Suppose that an
irreversible propagation reaction M + P,, -+ KP
P, + . is first order in each of
the reactants, so that, for isothermal conditions a t steady state, the rate of
diffusion into the element less the rate of diffusion out will equal the rate of
disappearance of reactants within the element (Figure 1).Hence, the monomer
mass balance may be written as,

If it is further assumed that the diffusivity of the monomer molecule, Dp, is

constant (a good assumption before diffusion controlled propagation occurs),
620 CHERN AND POEHLEIN

r 0
Fig. 1. Monomer concentration profile in a latex particle.

this difference equation becomes

with the following boundary conditions:

BC 1: r = 0,d[M],/dr= 0

BC 2: r = r,, [MI,= [MI,,

where M represents the monomer molecule, P, . and P,+l* are the free
+
radicals with n and n 1 monomer units, respectively, and K p is the
propagation rate constant. r is the radial distance from the center of the
particle, r, is the radius of the particle, A r is the shell thickness, [ M I , and
[ M ] , , are the monomer concentrations at r and r,, respectively, [ P . 3 is the
total free radical concentration in the particle, and Dp is the dh’usion
coefficient for the monomer molecule in the particle.
The solution to Equation (1) with the above boundary conditions can be
expressed as follows:
[M ]lp = sinh( +r’)/r’sinh( (P) (3)
where

[MI;= [Ml,/[Ml,s,
r’ = r/r,, and --/,r
(P =
 NONUNIFORM LATEX PARTICLES 621

_-
__.---

/ -

0.9 0.2 0.4 0.6 0.8 1.0

NORMRLIZED RRDIUS
Fig. 2. Monomer concentration profiles in a latex particle.

Equatio (3) was used to simulate the monomer concentration profiles i a

polymer particle for different values of cp (phi). The simulation results are
shown in Figure 2. The greater the value of cp, the larger is the monomer
concentration gradient in the polymer particle. For a very small value of @
(cp l), there is almost no monomer concentration gradient within the
polymer particle. Two homopolymer systems were examined in the context of
the solution given above for eq. (1). The first was styrene emulsion polymeri-
zation (a relatively water-insoluble monomer), and the second was vinyl
chloride emulsion polymerization (a relatively water-soluble monomer). Values
of the physical properties (rs and 0,) and the kinetic parameters ( K , and
[P - I,) in eq. (3) which were necessary for calculations were chosen from the
l i t e r a t ~ r e . ~ The
. ’ ~ results for both styrene and vinyl chloride emulsion poly-
merizations are presented in Tables I and 11,respectively.
The approximate value of 0,for both styrene and vinyl chloride monomers
in emulsion polymerization a t 60°C was estimated to be about 2.0 x lop6
cm2/sec?An examination of Figure 2 and Tables I and I1 clearly indicates
that a significant monomer concentration gradient is not possible for an
acceptable value of 0, and the corresponding cp. Therefore, it can be con-
cluded from this simple analysis that monomer will be uniformly distributed
within the particle volume for most emulsion polymerization systems.
Polymer molecules are considered to be incapable of moving freely inside
the latex particle because of long chain characteristics and chain entangle-
ments.16 Hence, a polymer chain is unlikely to translate without restriction.
The growing polymer molecule (free radical) can move by molecular motion or
by a propagation event. Based on this concept, and taking into account that
the end of the oligomeric free radical will preferentially remain in the surface
622 CHERN AND POEHLEIN
TABLE I
@ Values Calculated for Styrene Emulsion Polymerization

+ 0,(cm2/sec)

8.10 x 1 0 - ~ 2.00 x 10-6

1.00 1.32 x 1 0 - l ~
2.50 2.10 x 1 0 - l ~
5.00 5.26 X
10.00 1.32 X

T = 60°C.
K p = 1.76 X 10’ L/yl-sec.
[ P . 3, =-1.06 X 10- mol/L ( E = 0.5).
r, = 265 A.

TABLE I1
+ Values Calculated for Vinyl Chloride Emulsion Polymerization
+ ~
0,(cm2/sec)

1.96 x lor4 2.00 x 10-6

1.00 7.50 x 10-l~
2.50 1.20 x 10-l~
5.00 3.00 x 10-15
10.00 7.50 X

T = 50°C.
K p = 1.00 x lo4 L/mol-sec.
[ P . 3, =-3.00 X mol/L (E = 0.127).
rs = 505 A.

layer of the polymer particle, the movement of a growing polymer molecule

(free radical) is likely to be confined near the surface layer of the polymer
particle.
According to the above discussion, the nonuniform particle morphology now
can be interpreted as resulting from an “encapsulation” process. This hetero-
geneous morphology can be viewed as the result of “old” polymer molecules
being encapsulated with “new” polymer molecules during the course of
polymerization. If the old polymer molecules possess different physical proper-
ties from those of new polymer molecules, then the core-shell structure would
be expected. Vanderhoff et a1.I2 employed a conductometric titration tech-
nique to study the distribution of initiator end-groups in the polymer par-
ticles. His work showed that about 30-60% of SO, end-groups were found on
the particle surface. This implies that during the course of polymerization, old
polymer molecules would be encapsulated with new polymer molecules, and
some of the old polymer molecules would be buried inside the latex particles.
This experimental result gives credence to the postulated encapsulation mech-
anism. The following Monte Car10 simulation of the growth of a single
polymer chain within the latex particle also supports this hypothesis.
 NONUNIFORM LATEX PARTICLES 623

DISTRIBUTION OF FREE RADICALS IN LATEX PARTICLES

A cube that is subdivided into volume units equivalent to one scaled
monomer molecule is used to represent the latex particle. The cubic geometry
is chosen for the model because it is simpler than that of a sphere. The error
introduced by the edge and corner effects should be less with an increasing
number of volume units in the cube. Figure 3 illustrates a polymer particle
and the corresponding cubic lattice model. One goal of this research will be to
calculate the distribution function of the number of free radicals within the
latex particles, p(z)---z = 0 a t the particle surface and z = 1 a t the particle
center.
The preliminary calculations have involved the simulation of free radical
movement by monomer propagation within the cubic lattice. The first step is
t o place inactive polymer molecules a t a specified volume concentration in the
cube. This is done in a random manner, and thereafter each site in the lattice
contains either a scaled monomer molecule or a scaled polymer chain segment.
The second step then involves the random placement of a free radical on a
surface site that contains a monomer molecule. Thereafter the free radical is
allowed to move, in a random manner, to other adjacent sites that contain
monomer. This polymerization movement of the growing polymer chain (free
radical) is continued until a specified degree of polymerization is achieved or
until the free radical is completely surrounded by sites filled with dead
polymer segments.
The first calculations were based on simulation of a polymer particle with a
diameter of 4000 via a cube containing 25 lattice sites on each edge or (25)3
A
total sites. When a 4000 particle is simulated by a 25 X 25 X 25 cube, each

Distribution of Free
= Radicals i n Particle

Fig. 3. Latex particle and the corresponding cubic lattice model.

624 CHERN AND POEHLEIN

0
Diameter of Monomer Molecule = 14.8 A

0
Diameter of the Polymer Particle = 4 , 0 0 0 A

Volume of Cube Which Contains = ( 1 4 . 8 E - l 0 ? m3

One Monomer Molecule = 3.24E-27 m3

Volume of Polymer Particle = 1/6n(4E-7? m3

- 3.35E-20 m3
Volume Ratio = 3 . 3 5 E - 2 0 / 3 . 2 7 E - 2 7 = 1.024E7

(Approximate number of monomer molecules/Particle)

Assuming that molecular chain length is controlled
by monomer chain transfer
-
xn,styrene at 6 0 OC = Kp/Kfn = 2E4

A Scaled value of chain length for the 2 5 x 2 5 ~ 2 5cube

can be computed from the following equation
-
Number of Molecules in Particle
.................................. = ----_-------
Xn,styrene
-
Number of Scaled Molecules in Cube Xn,scaled

Fig. 4. Scaling calculations for the number average degree of polymerization in the cubic
lattice model.

lattice volume actually represents more than a single monomer molecule or

polymer chain link. Thus, the number average degree of polymerization (x,)
in the 25 X 25 X 25 lattice cube needs to be scaled to represent the actual X,
in a 4000 8, particle. Figure 4 shows the scaling calculations used to simulate a
styrene polymerization in which the end of chain propagation, and hence free
radical movement, is controlled by chain transfer to monomer. In this case,
the scaled value of xnin the 25 X 25 x 25 lattice cube is calculated to be 30.
Random walk calculations (Monte Carlo simulations) were carried out for a
number of free radicals, and these results were averaged to obtain p(z). Figure
5 shows the results of these Monte Carlo computations. Clearly, in this case,
free radicals are not uniformly distributed in the particle volume. The calcu-
lated p ( z ) is a decreasing exponential function. The averaged p ( z ) is less
scattered as the number of random walk simulations is increased from 10 to
28. Figure 6 shows similar calculations for a 40 x 40 x 40 cube in which the
scaled z, is 124. The solid line represents the least-square best fit of p ( z ) by
an exponential function [eq.(4)] based on 43 simulations.
In p ( z ) = -1.4259 - 4.37872 (4)
The free radical volume concentration profile in the latex particle, p’, can be
obtained directly from the results of these Monte Carlo calculations as shown
 NONUNIFORM LATEX PARTICLES 625

0.0 0.1 0.2 0.3 0.4

Z
Fig. 5. Distribution of free radicals calculated from a 25 X 25 X 25 cubic lattice model.

in Figure 7 for the 40 x 40 X 40 lattice cube. Again, the solid line represents
the least-square best fit of p'(z) by an exponential function [eq. (5)] based on
43 simulations.
In p'( Z) = - 2.0141 - 1.64012 (5)
Comparing Figures 6 and 7, it can be seen that the shape of p' is more flat
than that of p. This is simply because the number of lattice volume units of
the outer layer of the cube is greater than that of the inner layer.

0.0 0.1 0.2 0.3 0.4 0.5 0.E

2
Fig. 6. Distribution of free radicals calculated from a 40 X 40 X 40 cubic lattice model.
626 CHERN AND POEHLEIN

N
d- 0 43sDmATIa.ls
-Imsr-~BEsTFIT

-
0
d-

a
l
z-
-
t j w
Q 2-

z-
v

N
2-
0
2 I I I I I I

0.0 0.1 0.2 0.3 0.4 0.5 0.6

z
Fig. 7. Volume concentration profiles of free radicals calculated from a 40 X 40 X 40 cubic
lattice model.

As discussed previously, free radicals that enter polymer particles from the
aqueous phase contain end-groups that are usually hydrophilic. The distribu-
tion of free radicals in the latex particles, in a stochastic sense, will certainly
depend on how strongly these end-groups are held at the particle surface.
Other factors such as particle size, monomer-to-polymer ratio (conversion
levels), chain transfer reactions, and molecular diffusion phenomena will also
influence the distribution of free radicals in the polymer particles.
For simplicity, the present Monte Car10 calculations use a cubic model to
represent the spherical latex particle. The influence of geometry on the
calculation of the distribution of free radicals in the latex particle (i.e., edge or
corner effects) introduced by the cubi,cmodel can be reduced by increasing the
dimensions of the cubic lattice. The proposed model does not include a
stochastic distribution for the free radical chain length at which a chain
transfer reaction occurs. Furthermore, it does not take into account the
movement of monomer molecules and polymer segments by diffusion in the
cubic lattice. These considerations can be incorporated into the model for
future calculations. The model is based on uniform distributions (in a stochas-
tic sense) of monomer and polymer in the particle, but free radical penetration
is assumed to occur only via the propagation reaction, with segmental diffu-
sion presumed to be small during the life of a normal radical.
Determination of the distribution of free radicals in the polymer particles
quantitatively can be very important for understanding reaction kinetics and
product properties. One significant problem in which the distribution of free
radicals in the latex particles is important involves grafting reactions. An
example of a grafting reaction is the polymerization of styrene in the presence
 NONUNIFORM LATEX PARTICLES 627

of a polybutadiene seed latex. This process is important in the manufacture of

impact-resistant (toughened) polystyrene. Emulsion polymerization of styrene
is carried out with monomer-swollen polybutadiene seed particles for the
purpose of grafting polystyrene chains onto polybutadiene. The growing free
radicals must penetrate the polybutadiene core and must be in contact with
the residual double bonds of polybutadiene in order for the desired grafting
reaction to occur. Such penetration would be highly probable during the early
stage of polymerization because the styrene-swollen polybutadiene particles
would have a nearly uniform distribution of polybutadiene (PBD). As poly-
styrene (PS) chains form near the surface of the latex particle, however, the
distance from the surface where free radicals enter the latex particle to the
PBD core would become greater, and the penetration of free radicals into
the PBD core would become less. Eventually, grafting reactions could cease as
the process of penetration of free radicals into the PBD core stops.
The distribution of free radicals in the polymer particles could also play an
important role in influencing final product properties. Design and control of
the polymer particle morphology is an important aspect of many commercial
emulsion polymerization processes. The PS/PBD system is only one example
of such processes operated to obtain the desired polymer particle morphology.
The development of core-shell and other polymer particle structures is of
primary importance, and it could be strongly dependent on the distribution of
free radicals in the polymer particles. In addition, knowledge of the distri-
bution of free radicals in the polymer particles is also important in the study
of emulsion polymerization kinetics. Such distributions could be incorporated
into kinetic models to predict the reaction rate, PSD, and MWD in two-phase
emulsion polymerization systems. Knowledge of free radical distribution could
also lead to an improvement in understanding the chemical and physical
mechanisms associated with the phenomenon of transport of free radicals out
of the polymer particles into the aqueous phase. The results of the present
simulation will be utilized to illustrate the application of nonuniform radical
distributions to the calculation of radical desorption coefficients.

TRANSPORT OF FREE RADICALS OUT OF

LATEX PARTICLES
The nonuniform distribution of free radicals in latex particles could have an
impact on the transport of free radicals out of the particles into the aqueous
phase. This so called “desorption” phenomenon involves a chain transfer
reaction (to monomer or chain transfer agent) that generates a small, rather
mobile free radical followed by diffusion of that free radical out of the latex
particle into the aqueous phase. The desorbed free radical may react with
dissolved monomer molecules by propagation. This oligomeric free radical
could then precipitate to nucleate a new primary particle. It also could be
reabsorbed into a polymer particle or even terminated by another oligomeric
free radical in the continuous phase. The process of transport of free radicals
out of the latex particles into the aqueous phase will reduce the average
number of free radicals per particle and consequently lower the polymeriza-
tion rate. Moreover, if emulsion polymerization is carried out in a continuous
stirred-tank reactor, this free radical desorption phenomenon will also have an
628 CHERN AND POEHLEIN

influence on the particle size distribution of the latex product. This has been
demonstrated by Poehlein et al.6*7
Ugelstad and Hansen" considered that a monomer-unit free radical gener-
ated by a chain transfer reaction inside the latex particle could diffuse
through the particle to the interface and desorb into the aqueous phase. They
further assumed that the monomer-unit free radical concentration at the
interface was in equilibrium with that in the aqueous phase immediately
outside the latex particle. In addition, the monomer-unit free radical con-
centration in the aqueous phase at infinite distance from the latex particle
was assumed to be zero. Based on the consideration of diffusion with chemical
reaction in a spherical polymer particle, they derived an expression for the
desorption rate constant, KA as shown below;

K i m 3( 4 a/3)2'3D, Dp
KA= -
K; aDp+ D,

where K i m is the chain transfer reaction rate constant to monomer, K; is the

reinitiation rate constant for the monomer-unit free radical, D, and Dp are
the diffusion coefficients for the monomer-unit free radical in the aqueous
phase and latex particle, respectively, a is the distribution coefficient for the
monomer-unit free radicals between the latex particle and aqueous phase, and
r is the radius of the latex particle. Independently, Harada and N o m ~ r a ~ ~
used both stochastic and deterministic approaches to develop a theoretical
equation for the monomer-unit free radical demrption rate constant that is
equivalent to eq. (6). Neither of these groups considered the fact that free
radicals are not uniformly distributed in the latex particles. The theory for
the free radical transport phenomenon would need modification if such were
the case.
Figure 8 illustrates two possible monomer-unit free radical concentration
profiles (p,) in polymerizing latex particles. Curve a represents a nearly
uniform distribution of such free radicals in the particle that is used in
previous kinetic studies in the transport of free radicals out of the latex
particles. In this case,only one flux that is directed toward the aqueous phase
is established in the reaction system. On the other hand, curve b represents a
nonuniform distribution of such free radicals in the particle. For this reaction
system it is necessary to consider two fluxes (i.e. one is directed toward the
aqueous phase and the other is directed toward the center of the particle).
The distribution of monomer-unit free radicals in polymerizing latex particles
is dependent on the location in which such free radicals are formed and the
viscosity of the reaction system (ie., mobility of the monomer-unit free
radical). Monomer-unit free radicals formed by chain transfer reactions are
similar in size to monomer molecules and should have a relatively high
mobility as long as the viscosity of the reaction system is not too high. If the
mobility of the monomer-unit free radical is high enough, curve a represents a
reasonable approximation for the concentration profile of the monomer-unit
free radical in the latex particle. On the other hand, this concentration profile
is strongly dependent on the local concentration of long-chain free radicals
within the particle volume if the viscosity of the reaction system is very high
 NONUNIFORM LATEX PARTICLES 629

0 2 - 1 1 .o
Particle Particle
Center Surface

Fig. 8. Distribution of monomer-unit free radicals in latex particles. (a) Uniform distribution;
(b) nonuniform distribution.

(e.g., at high conversion). The reason for this is that, as discussed above, free
radicals are not uniformly distributed throughout the latex particle, and,
consequently, the concentration profile of the monomer-unit free radical
generated by chain transfer reaction of the free radical to a monomer molecule
is also most likely to be nonuniform in the latex particle. This nonuniform
distribution of monomer-unit free radicals in the latex particle will certainly
influence the desorption mechanism. The following derivation for the mono-
mer-unit free radical desorption rate constant is based on the physical picture
of curve b in Figure 8.
First, i t is assumed that the propagation reaction and chain transfer
reaction are irreversible and second order. Furthermore, the monomer-unit
free radical concentration at the interphase is assumed to be in equilibrium
with that in the aqueous phase just outside the particle. The monomer-unit
free radical concentration in the aqueous phase at infinite distance is also set
to zero. For isothermal conditions at steady state, the rate of diffusion into a
differential element less the rate of diffusion out will equal the rate of
disappearance of reactants within the element. Hence, the monomer-unit free
radical mass balance may be written as,
630 CHERN AND POEHLEIN

If it is further assumed that at a given conversion the diffusion coefficient of

monomer-unit free radical, D,, is constant, this difference equation becomes

with the following boundary conditions:

BC 1: r = 0, d[M.],/dr = 0

BC 2: r = rs,

where [ M * ]ks = [ M * ],,/a. Here, [P 1, designates the total free radical

concentration (monomer-unit free radical is not included) in the particle.
[ M . 1,: [ M . Ips, and [ M I;, represent the monomer-unit free radical con-
centrations in the particle, at r,, and at rs immediately outside the particle,
respectively; [P. lPavand [ M . lPavrepresent the average total free radical
concentration and monomer-unit free radical concentration in the particle,
respectively; and 0,represents the diffusion coefficient for the monomer-unit
free radical in the aqueous phase.
Equation (8), along with the above boundary conditions, can be used to
obtain the monomer-unit free radical concentration profiles in the latex
particle. With the knowledge of [ M . 1, as a function of r , one can derive an
expression for K J . A t this point, the question remaining is how to determine
the quantities [ P. I,, [ P. lPav,and [ M a ]pa,. The following approach seems
to give reasonable estimates for these quantities.
Determination of [P. lPavis straightforward, and the result can be ex-
pressed as follows:

[P ' l p a v = ~ / N . v o , ( l- KIJK,) (9)

where iE is the average number of free radicals per particle, NAv is Avogadro's
number, and up is the volume of a single latex particle. The term (1 - K,JK,)
in eq. (9) represents the fraction of free radicals in the particle that are larger
than the monomer-unit free radical.
In order to determine [ P* I,, one must know [P . lPavand the distribution
function of free radicals in the latex particle under a certain reaction condi-
tion. In the present work, the 40 X 40 x 40 lattice cube that is equivalent to a
4000 A latex particle is chosen as the reference to calculate [P. 3,. Equation
(5) can be rewritten as follows:

p'(z) = p;exp( -kz) (10)

where p: is the value of p'(z) at r,, and k is a constant. The values of p: and K
can be calculated directly by comparing eqs. (5) and (10). In this case, K has a
 NONUNIFORM LATEX PARTICLES 631

value of 1.64. The average of p’( z), ( p & ) then can be calculated according to
the following equation:

= 3p;[(l/k - 2 / k 2 + 2 / k 3 ) - 2exp( -k)/k3]

= P;~XP( - kzav ) (11)

zav is estimated to be 0.222 from eq. (1) and the value of k . Now, [ P . 1, can
be calculated from the following equations:

CP-I, - P:e.P(-kz)
(12)
P 1 * pav P ~ ~ X-
P kzav)
(

[ P a ] ,= 1.4392z/NAVup(1 - KfJK,)eXp[-k(l - r/R)] (13)

where R is equal to the radius of the reference latex particle, 2000 A.

The only remaining unknown parameter is [ M * Ipav, which is difficult to
estimate. For [M . lPavthe following relationship holds:

[ M * 1p a v (~/N.vup) ( K,JK, 1 (14)

If the process of transport of free radicals out of polymerizing latex particles is

insignificant, then the equality relationship in eq. (14) is valid. In this work,
eq. (8) and the indicated boundary conditions are transferred into finite
difference forms and solved with the aid of a computer to give [ M . 1,. The
parameters that are necessary for calculation are taken or estimated from the
literature. The solution procedure is summarized as follows:
1. Assume a value for [ M e Ipav.
2. Calculate [ M . 3,.
-
3. Calculate new [ M lPav.
4. If I[M * lpav,new - [ M * lpav,oldl < e, stop.
5. If not, go back to 1.
where e is the tolerance in error.
Several calculated monomer-unit free radical concentration profiles in the
latex particle are shown in Figure 9. [M . 3, only shows a small concentration
gradient in the latex particle when the mobility for the monomer-unit free
radical is high (0, =2 x cm2/sec), whereas [ M * 3, shows a significant
concentration gradient in the latex particle when the mobility for the mono-
mer-unit free radical in the latex particle is low ( D = 2 x lo-’ cm2/sec).
4
These results are in agreement with the above discussion. Figure 2, Tables I
and 11, and Figure 9 show that monomer-unit free radicals are more likely to
distribute themselves nonuniformly in the latex particle than monomer mole-
cules because the formation of monomer-unit free radicals is strongly depen-
dent on the nonuniform distribution of oligomeric free radicals within the
particle volume.
632

N I
9 0.0
Particle
Center

4.0 8.0

r ICMl
12.0

1
16.0

o-6
Fig. 9. Monomer-unit free radical concentration profiles in a latex particle.

The resultant monomer-unit free radical concentration profile can be fitted

into an exponential equation that has the following form:

W I p = [M.lp,exP[-k’(1 - r/.,)] (15)

where [M . Ips is the monomer-unit free radical concentration a t r,, and k’ is
a constant. Then the average monomer-unit free radical concentration in the
latex particle, [M . Ipav,can be computed according to the following equation:

= 2[M-],,[(l/k’- 2/k’2 + 2/kt3) - 2 e ~ p ( - k ’ ) / k ’ ~ ] (16)

The monomer-unit free radical flux that is directed toward the aqueous phase
in Figure 8 (curve b), N,, can be expressed as the following equation:
Ns = 4ar,D,[M el’,,
= 4ar,D,[M .]p,/a (17)
Combining eqs. (16) and (17), one obtains

N,
I
4arsD,[M . p a v
= (18)
3 a [ ( l / k ’ - 2/k’2 - 2/k’3) - 2exp(-k’)/kr3]
 NONUNIFORM LATEX PARTICLES 633

Rearranging eq. (18) gives an expression for the desorption rate constant Kdm
for the monomer-unit free radical as shown in eq. (19):

The desorption rate constant, K,, defined by (rate of desorption) = K,CnN,’8

can be written as

where N, is the concentration of latex particles containing n free radicals.

Finally, the desorption rate constant, K ; , defined by K d = K;(N/V,),18 has
the following form;

KA= -
Kfm (4n/3)2’30w
K ; a [ ( l / k ’ - 2/k’2 + 2/kt3) - 2e~p(-k’)/k’~] (21)

where N is the latex particle concentration, and V, is the total volume of latex
particles per unit volume of aqueous phase. Equation (21) represents the
modified expression for the desorption rate constant that takes into account
the fact that both of the long chain free radicals and monomer-unit free
radicals are not distributed uniformly in the latex particle. The modified
desorption rate constant has a form similar to that developed by Ugelstad and
Hansen” and Harada and Nomura.lg Calculated desorption rate constants
using different equations [i.e., eqs. (6) and (21)] can be compared under the
same reaction condition. The parameters that are necessary for calculation are
obtained from the literature. The results are summarized in Table 111. Both
equations predict comparable results for a Dp value of 2 x cm2/sec. The
calculated desorption rate constant using eq. (6) is about 2.5 times greater
than the modified desorption rate constant. However, at a lower value of Dp
(2 X lo-’ cm2/sec), in which [M * I p shows a significant concentration gradi-
ent within the particle volume as illustrated in Figure 9, the calculated

TABLE 111
Styrene Emulsion Polymerization

2x 2.3 x lo-’’ 6.0 x lo-’’

2 x 10-~ 2.6 X lo-’’ 6.9 x 10-13

T = 60°C.
K,,,,= 8.80 X L/mol-sec.
KL = K , = 1.76 x lo2 L/mol-sec.
0, = 2.00 x cm*/sec.
a = 1300.
634 CHERN AND POEHLEIN

dmrption rate constant using eq. (6) is smaller than the modified desorption
rate constant. This result is not surprising because the desorption rate
constant developed in this work has taken into consideration the nonuniform
distribution of monomer-unit free radicals in the latex particle that will result
in a greater concentration gradient and hence a greater driving force for
dmrption at the interface.

CONCLUSIONS
Previous kinetic studies in emulsion polymerization have almost always
involved an assumption that free radicals are uniformly distributed in the
latex particle. This assumption is not likely to be valid for many emulsion
polymerization systems because initiators generate free radicals that are
hydrophilic and often ionic. Thus, the free radicals that enter the monomer-
swollen polymer particles are likely to be low molecular weight oligomers with
hydrophilic end-goups. These end-groups will have a strong tendency to stay
on the surface of the latex particles and thus constrain the movement of the
growing free radicals into the particle interior. A very simple model using
Monte Car10 technique has been developed to calculate the distribution of free
radicals in particles with the constraint that the end-groups of these growing
chains remain on the surface. Simulations with this model predict that free
radical concentrations will be statistically greater near the surface of the
particles. These results are in qualitative agreement with some experimental
results from grafting studies and from morphology studies in polystyrene and
poly(methy1 methacrylate) latexes. Such a nonuniform distribution of free
radicals in the latex particle could have a significant influence on reaction
kinetics and product properties. Based on the concept of nonuniform distri-
bution of free radicals in the latex particle, the kinetics of transport of free
radicals out of polymerizing latex particles has been reexamined, and a
modified expression for the dwrption rate constant has been presented. This
concept will be used to compute grafting efficiencies in two-phase particles in
future papers.

Financial support from the National Science Foundation under Grant No. CBT-8413987, from
the Petroleum Research Fund of ACS under Grant No. 3426-AC6, and from Georgia Institute of
Technology is gratefully acknowledged.

References
1. W. D. Harkins, J. A m . C h m . Soc., 69,1428 (1947).
2. W. V. Smith and R. H. Ewart, J. Chem. Phys., 16,592 (1948).
3. W. H. Stockmayer, J. Polym. Sci., 21, 314 (1957).
4. J. T. O’Toole, J. Appl. Polym. Scz., 9,1291 (1965).
5. J. Ugelstad, P. C. Merk and J. 0. Aasen, J. Polym. Sci., Purt A-1, 5, 2281 (1967).
6. G. W. Poehlein, W. D. Dubner, and H. C. Lee, Br. Polym. J., 14,143 (1982).
7. H. C. Lee, PhD Dissertation, School of Chemical Engineering, Georgia Institute of Technol-
ogy, Atlanta, Georgia (1985).
8. M. R. Grancio and D. J. Williams, J. Polym. Sci., Purt A-1, 8, 2617 (1970).
9. P. Keusch, J. Price, and D. J. Williams, J. Mucromol. C h m . , Chem., A7, 623 (1973).
10. D. H. Napper, J. Polym. Sci., Purt A-1, 9,2089 (1971).
11. J. L. Gardon, J. Polym. Sci. Polym. Chem. Ed., 11, 241 (1973); ibid., 12,2133 (1974).
 NONUNIFORM LATEX PARTICLES 635

12. J. W. Vanderhoff, Notes, 14th Annual Emulsion Polymers Short Course, Lehigh University
(1983).
13. L. W. Fisher, private communication (1985).
14. A. Klein, Notes, 16th Annual Emulsion Polymers Short Course, Lehigh University (1985).
15. A. Berens, J . Appl. Polym. Sci., IS,2379 (1974).
16. M. Tirrell and T. J. Tulig, in Polymer Reaction Engineering, K. H. Reichert and W.
Geiseler, E d . , Hanser, New York, (1983), p. 247.
17. J. Ugelstad and P. C. Mark, Br. Polym. J., 2, 31 (1970).
18. J. Ugelstad and F. K. Hansen, Rubber Chem. & Technol., 49, 536 (1976).
19. M. Nomura, M. Harada, W. Eguchi and S. Nagata, J . Chem. Eng. Japan, 4, 54 (1971);
ibid., 4, 160 (1971); J . Polym. Sci., Chem. Ed., 20, 1261 (1982).

Received January 27,1986

Accepted April 25, 1986