Sie sind auf Seite 1von 16

Journal of Membrane Science 192 (2001) 55–70

Mathematical model for encapsulation by


interfacial polymerization
Jiang Ji a,∗ , Ronald F. Childs b , Mahesh Mehta a
a Koch Membrane Systems Inc., 850 Main Street, Wilmington, MA 01887-3388, USA
b Department of Chemistry, McMaster University, Hamilton, Ont., Canada L8S 4M1
Received 11 December 2000; accepted 25 April 2001

Abstract
A general model for encapsulation by diffusion- and reaction-controlled interfacial polymerization under nonsteady-state
boundary conditions has been developed. Special models for either diffusion-controlled or reaction-controlled interfacial
polymerization under both steady- and nonsteady-state boundary conditions can be obtained by simplifying the general
model. These special models agree with the similar models in the literature. The general model developed is consistent with
the experimental data reported in the literature. This work significantly extends existing theories for the encapsulation by
interfacial polymerization and provides an important guideline for the effective control of the thickness and porosity of the
capsule wall prepared by interfacial polymerization. © 2001 Elsevier Science B.V. All rights reserved.
Keywords: Microcapsules; Controlled release; Encapsulation; Interfacial polymerization; Modeling

1. Introduction very important to the effective control of capsule wall


structure (thickness and porosity) and release perfor-
Microcapsules consisting of a core (liquid or solid) mance. Although a large amount of work has been
and a permeable or non-permeable wall have been reported in this area, no satisfactory general model is
widely used in release and transfer control [1–3]. The available [1–9].
capsule core may contain active compounds such as Pearson and Williams [6] studied encapsulation
catalysts, drugs, anti-fouling compounds or toners, involving the interfacial polymerization of an isocy-
which are surrounded by a suitable wall material that nate with a diol. They developed a theoretical model
regulates release. The controlled release of the capsule that was solved numerically for both steady- and
contents strongly depends on capsule wall thickness nonsteady-state interfacial polymerizations involving
and porosity [4]. Interfacial polymerization is one of a consideration of monomer diffusion and the rate of
most important methods used to make microcapsules interfacial reaction. The model was consistent with
[1–3]. A clear understanding of how a variety of fac- the experimental data obtained in the early stages of
tors, such as monomer concentration, diffusion and the polymerization. The complexity of the numeri-
interfacial reaction, influence the capsule formation is cal method used has reduced the significance of the
model, as a result it is not commonly used.
∗ Corresponding author. Tel.: +1-978-694-7264;
Warburton [5] developed an alternative model for
fax: +1-978-694-7015. encapsulation involving multiple emulsions. This
E-mail address: jij@kochind.com (J. Ji). model assumes that the physical chemistry of curved

0376-7388/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 6 - 7 3 8 8 ( 0 1 ) 0 0 4 9 5 - 1
56 J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70

Nomenclature Greek symbols


A0 constant defined in Eq. (8) β ratio of water molecule to monomer
((m4 kg)/(kmol2 s)) A diffused together from the outer
B0 constant defined in Eq. (9) surface of the capsule wall to the
reaction zone as shown in Fig. 1
((m5 kg)/(kmol2 s))
δ thickness of the reaction zone,
C0 constant defined in Eq. (10)
as shown Fig. 1 (m)
((m7 kg)/(kmol2 s))
γ reaction ratio given in Eq. (34)
CA (r, t) concentration of monomer A in the
ϕ volume fraction
capsule wall at position r and time t,
ρP density of the capsule wall formed by
as shown in Fig. 1 (kmol/m3 )
the interfacial polymerization (kg/m3 )
CA (r0 , t) concentration of monomer A at the
outer surface of the capsule wall at
Subscripts
time t, as shown in Fig. 1 (kmol/m3 )
0 reaction- and diffusion-controlled
CB (Ri , t) concentration of monomer B in the
polymerization
capsule core at time t, as shown in
A monomer A
Fig. 1 (kmol/m3 )
B monomer B
D diffusion coefficient of monomer A
D diffusion-controlled polymerization
in the capsule wall (m2 /s)
max maximum value
D0 constant defined in Eq. (11)
min minimum value
((m8 kg)/(kmol2 s)) P average polymer
DM constant defined in Eq. (4) R reaction-controlled polymerization
(D M = D P M U /ρ P ) (m5 /(kmol s)) U repeat unit of the polymer
E0 constant defined in Eq. (12) V void
((m9 kg)/(kmol2 s2 )) W water
F0 constant defined in Eq. (13) Z a small molecule formed from
((m6 kg)/(kmol2 s2 )) each condensation reaction
k apparent reaction rate constant
(k = δk r ) (m4 /(kmol s))
surfaces can be related to that of planar surface. In this
kI constant defined in Eq. (2)
model, both the diffusion and interfacial reaction are
(m7 /(kmol2 s))
again considered. Instead of the thickness of capsule
kr second order reaction rate
wall, the interface rigidity, which is a function of the
constant (m3 /(kmol s)) degree of crosslinking, was used to characterize the
K partition coefficient of monomer A capsule formation. Although an analytical solution
between the capsule wall and the was obtained by Warburton, no direct correlation of
continuous aqueous phase the interface rigidity to the capsule wall thickness was
Mi molecular weight of component given. This limits the usefulness of the model in that
i (kg/kmol) the release rate of the active core material across the
P average degree of polymerization capsule wall is inversely proportional to the capsule
r radial coordinate normal to the wall thickness [4].
spherical surface, as shown in Janssen and te Nijenhuis [7–9] have described an
Fig. 1 (m) encapsulation technique involving the polyconden-
S membrane area (m2 ) sation of terephthaloyl dichloride with diethylenetri-
t polymerization time (s) amine at the interface of an oil-in-water emulsion.
vP volume of the capsule wall, They considered a spherical capsule as a plane and
zone II in Fig. 1 (m3 ) proposed a diffusion controlled model to describe
X thickness of the capsule wall (m) the encapsulation. This model was in agreement with
J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70 57

experimental results for short polymerization times monomer B) in an organic solvent, then immersed
with the thickness of the capsule wall linearly increas- in an continuous aqueous phase containing a second
ing with the square root of polymerization time. How- monomer, such as a diamine (here called monomer A)
ever, this model incorrectly predicts that the capsule [7–9]. Typically, the organic capsule core is insoluble
wall thickness approaches infinity when the polymer- in the continuous aqueous phase. Reaction between
ization time approaches infinity. This is a result of us- monomers A and B to form a polymer occurs at the
ing a planar geometry instead of a spherical one in that interface between isolated organic and continuous
a material balance on the core material is impossible. aqueous phases. The actual process of the encapsula-
A new general model is needed which can correctly tion by interfacial polymerization is complex. Thus,
handle the mass balance of the core material to some simplifying assumptions are made to facilitate
overcome the problems discussed above. More model development and solution. The assumptions
importantly, such a theoretical model should correctly are the followings.
describe and predict the encapsulation by interfa-
1. The system considered here is the encapsulation
cial polymerization not only in the early stages but
of a dispersed organic phase containing monomer
also in the entire course under either nonsteady- or
B by interfacial polymerization with monomer A
steady-state boundary conditions. Furthermore, the
in a continuous aqueous phase. The concentration
new model should be simple and easy to use.
of monomer A in the continuous aqueous phase is
Recently, we reported the fabrication of thin-film
uniform [7–9,11]. The concentration of monomer
composite polysulfonamide membranes by interfa-
B is uniform in the organic capsule core up to the
cial polymerization of an aliphatic diamine with an
reaction zone and constantly changes with time
aromatic disulfonyl chloride [10]. The thickness of
during the polymerization [7–9,11]. Reaction takes
the thin-film barrier layer formed by interfacial poly-
place in a reaction zone of constant thickness,
merization was found to first increase and then level
δ, in the organic phase adjacent to the interface
off with polymerization time. In order to describe
[11–19].
the formation of the flat sheet thin-film composite
2. The capsule wall formed by interfacial polymer-
membrane, a general model was developed which
izaton has a spherical geometry with an inside
considered both diffusion- and reaction-controlled in-
diameter, R, and outside diameter, r0 . The capsule
terfacial polymerization under nonsteady-state bound-
wall consists of micro voids and a dense polymer
ary conditions [11]. The model was in agreement
phase [7–9].
with the experimental data, and predicted a result that
3. Only monomer A, with associated water molecules,
would be expected at infinitive polymerization times.
is limited by diffusion from the outer surface of
In the current work, the same methodology developed
capsule wall to the reaction zone close to the inner
in our previous work [11] is used to develop a general
surface of the capsule wall [7–9]. All of monomer
model for the encapsulation by both diffusion- and
A arrived at the reaction zone reacts with monomer
reaction-controlled interfacial polymerization under
B to form a polymer. The diffusion coefficient of
nonsteady-state boundary conditions. The model
monomer A, with the associated water molecules,
developed is fitted with the experimental data reported
is assumed to be constant in the newly formed
by Janssen and te Nijenhuis [9].
polymeric capsule wall [7–9,11–13], and the diffu-
sion rate of monomer A in the micro voids is much
2. Theory faster than in the dense polymer phase [7–9].
4. If the flat sheet geometry in the previous work
2.1. A general model for both reaction- and [11] is replaced with the spherical geometry, the
diffusion-controlled interfacial polymerization assumptions (2, 3, 4, 7, 9, 10) made in the pre-
under nonsteady-state boundary conditions vious work regarding the model development are
still valid for the present work.
2.1.1. Model assumptions
In an encapsulation, the capsule core is first loaded A detail discussion on the model assumptions is
with a monomer, such as a diacid chloride (here called given in our previous work [11].
58 J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70

Fig. 1. A schematic representation of the model at time t.

2.1.2. Model development the density of the newly formed polymeric capsule
In terms of the model, some of the key terms and wall, MP the average molecular weight of the polymer,
a schematic presentation is given in Fig. 1. A full dv P /dt is the volume growth rate of the capsule wall,
list of symbols is given at the end of the paper. The kr the second order reaction rate constant, CA (R, t) is
coordinate r is fixed at r = 0 at the center of the cap- the concentration of monomer A in the reaction zone
sule core. As the polymerization proceeds, a spherical at time t, CB (R, t) the concentration of monomer B
thin-film shell forms, and the reaction zone which that reacts with monomer A in the organic phase, δ
is initially located at r = r0 , is found at r = R(t) the reaction zone thickness, S = 4π R 2 is the capsule
at time t. Monomer A first partitions into the newly area, and dvP = −S dR is the volume of the newly
formed polymeric capsule wall and then diffuses formed capsule wall. The minus sign is due to the fact
to the reaction zone where a second order reaction that the capsule wall grows towards the center in an
occurs. A typical concentration profile at time t, is opposite direction of the r-axis, because dv P is always
illustrated in Fig. 1. Since R(t) varies with time, this larger than zero. The average molecular weight, MP , is
becomes a moving boundary problem. defined in the same way as in our previous work [11].
The following equations can be formulated based on Substitution for dv P in Eq. (1) and rearrangement
the assumptions given above. At time t, the formation then gives
of polymeric capsule wall due to the second order
reaction is given by dR kI
− = CA (R, t)CB (R, t) (2)
PρP ϕP dvP dt ϕP
= kr CA (R, t)CB (R, t)δS (1)
MP dt where dR/dt is the growth rate of the capsule wall
where P is the average degree of polymerization, ρ P formed by interfacial polymerization. On the basis of
J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70 59

assumption 3, the growth rate must be related to the D0 = ϕP2 kI CB (r0 , 0)MU r04 − ϕP3 kI ρP r04 (11)
diffusion rate of A to the reaction zone, expressed as
E0 = DM CA (r0 , 0)kI r03 [ρP ϕP − CB (r0 , 0)MU ] (12)
PρP ϕP dvP
F0 = DM CA (r0 , 0)kI ϕP ρP (13)
MP dt
S[CA (R, t) − CA (r0 , t)] Substitution of Eqs. (2) and (7) into Eq. (4) and rear-
=− (3)
((R − r0 )ϕV )/DV + ((R − r0 )ϕP )/DP rangement give a growth rate of the capsule wall as

As discussed in our previous work [11], Eq. (3) is dR E0 − F0 R 3


= (14)
simplified to the following equation: dt A0 R 4 + B 0 R 3 + C 0 R + D 0
dR DM [CA (r0 , t) − CA (R, t)] Substitution of Eq. (14) into Eq. (2) and rearrangement
− = 2 (4) gives
dt ϕP R − r0
ϕP
Similarly, a volume fraction of polymer in the capsule CA (R, t) = −
kI CB (R, t)
wall is obtained as follows [11]:
E0 − F 0 R 3
M U ρW × (15)
ϕP = (5) A0 R 4 + B 0 R 3 + C 0 R + D 0
βMW ρP + MU ρW
Defining α = r0 /Rmin and ω = R/Rmin , integration of
and a volume fraction of voids in the capsule wall is Eq. (14) gives a relation between the reaction time, t,
given as follows, ϕV = 1 − ϕP . and the radius of the newly formed capsule wall, R, as
A material balance on monomer B in the capsule
A0 Rmin
5 B0 Rmin
4
core gives t= I4 + I3
E0 E0
4πR 3 4π(r03 − R 3 )ϕP ρP P C0 Rmin
2
D0 Rmin
CB (R, t) + + I2 + I1 (16)
3 3MP E0 E0
4πr0 CB (r0 , 0)
3
where
= (6)
3  ω y4 1
Under nonsteady-state conditions, Eq. (6) determines I4 = dy = (α 2 − ω2 )
α 1−y
3 2
CB (R, t), the concentration of monomer B at (R, t).  
1 (1 − α) /(1 + α + α 2 )
2
The first term on the left hand side of Eq. (6) is the + ln
amount of monomer B (kmol) left in the capsule core 6 (1 − ω)2 /(1 + ω + ω2 )
 
at time t, the second term is the amount of monomer B 1 2ω + 1
+ √ arctan √
(kmol) converted into the polymer. The right hand side 3 − 3
is the amount of monomer B (kmol) in the dispersed  
1 2α + 1
organic phase at time t = 0. − √ arctan √ (17)
3 − 3
Rearrangement of Eq. (6) gives
 ω
r3 (r 3 − R 3 ) ϕP ρP y3
CB (R, t) = 03 CB (r0 , 0) − 0 3 (7) I3 = dy = α − ω
R R MU α 1−y
3
 
In order to simplify the above equations, we define the 1 (1 − α)2 /(1 + α + α 2 )
+ ln
following constants: 6 (1 − ω)2 /(1 + ω + ω2 )
 √ 
1 ω 3
A0 = −ϕP3 kI ρP (8) − √ arctan
3 −2 −ω
B0 = ϕP3 kI ρP r0 + ϕP DM MU (9)  √ 
1 α 3
+ √ arctan (18)
C0 = ϕP3 kI ρP r03 − ϕP2 kI CB (r0 , 0)MU r03 (10) 3 −2 − α
60 J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70

 ω  
y 1 (1 − α)2 /(1 + α + α 2 ) 2.2.1.1. Reaction-controlled interfacial polymeriza-
I2 = dy= ln
α 1−y (1 − ω)2 /(1 + ω + ω2 )
3 6 tion. If the rate of interfacial polymerization is much
  slower than that of monomer diffusion (k I  D M ),
1 2ω + 1
− √ arctan √ the growth rate of the capsule wall is controlled
3 3 by the rate of interfacial reaction. In this situation,
 
1 2α + 1 CA (R, t) ≈ CA (r0 , t), and Eqs. (1)–(22) are still
+ √ arctan √ (19) valid. However, in Eq. (16) only a term related to the
3 3
interfacial reaction remains, thus we have A0 /E0 ≈ 0,
   C0 /E0 ≈ 0, D0 /E0 ≈ 0, and
ω
dy 1 (1 − α)2 /(1 + α + α 2 )
I1 = = ln B0 ϕ P MU BR
α 1−y
3 6 (1 − ω)2 /(1 + ω + ω2 ) = =
 √  E0 kI CA (r0 , 0)r0 [ϕP ρP − CB (r0 , 0)MU ]
3 E R
1 ω 3 (23)
− √ arctan
3 −2 − ω
It should be noted that Eqs. (7), (14)–(16), and (23)
 √ 
1 α 3 can be independently derived from Eqs. (2) and (6)
+ √ arctan (20) based on the assumptions specified in this section and
3 −2 −α
the general assumptions made in Section 2.1.1.
The thickness of the newly formed capsule wall is
2.2.1.2. Diffusion-controlled interfacial polymeriza-
given by X = r0 − R. When the reaction time, t,
tion. If the rate of interfacial reaction is much faster
approaches infinity, CB (R, ∞) approaches zero, the
than that of monomer diffusion (D M  k I ), the
capsule wall approaches the maximum thickness, as a
growth rate of the capsule wall is controlled by the
result of R approaching a minimum value, Rmin
rate of monomer diffusion to the reaction zone. Under
 1/3   this condition, C A (R, t) ≈ 0, and Eqs. (3)–(22) are
E0 CB (r0 , 0)MU 1/3
Rmin = = 1− r0 (21) still valid. However, in Eq. (16) only terms related
F0 ϕP ρ P
to monomer diffusion remain, and are changed as
Rmin can be obtained experimentally, and as such follows
Eq. (21) is important as it correlates Rmin with E0 ϕP3 ρP
A0
and F0 . And the maximum capsule wall thickness is =−
obtained as follows, Xmax = r0 − Rmin . E0 DM r03 CA (r0 , 0)[ϕP ρP − CB (r0 , 0)MU ]
The reduced thickness of the newly formed capsule AD
= (24)
wall is defined as ED
X r0 − R
= (22) B0 ϕP3 ρP
Xmax r0 − Rmin =
E0 DM r02 CA (r0 , 0)[ϕP ρP − CB (r0 , 0)MU ]
X/Xmax is also the reaction conversion of monomer B
BD
at reaction time t. = (25)
ED
2.2. Special models C0 ϕP2 CD
= = (26)
E0 DM CA (r0 , 0) ED
Several special models can be derived by simpli-
D0 ϕP2 r0 DD
fying the above general model. =− = (27)
E0 DM CA (r0 , 0) ED
2.2.1. Interfacial polymerization under It should be noted that Eqs. (7), (14)–(16) can be inde-
nonsteady-state boundary conditions pendently derived from Eqs. (4) and (6) based on the
Under nonsteady-state boundary conditions, the assumptions specified in this section and the general
following special models can be obtained. assumptions made in Section 2.1.1.
J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70 61

2.2.2. Interfacial polymerization under the condition ϕP (r0 − R) ϕP X


t= =
of constant concentration of monomer B in the kI CA (r0 , 0)CB (r0 , 0) kI CA (r0 , 0)CB (r0 , 0)
capsule core (32)
If the amount of monomer B converted to the poly-
mer is much less than that left in the capsule core, then Mathematically, this result is the same as the solid
the concentration of B in the capsule core is essen- spherical model developed by Neogi and Allan [21,22]
tially constant. Under this condition, the second term for the controlled release of pesticides by decomposi-
on the left hand side of Eq. (6) is negligible compared tion of the polymeric carrier. As such it is the reverse of
to the first term, thus CB (R, t) ≈ CB (r0 , 0). This the encapsulation. It is important to note that Eq. (32)
condition is approximately satisfied in the early stage is mathematically identical to Eq. (25) that we ob-
of the polymerization, even under nonsteady-state tained previously for a flat sheet model [11]. It should
boundary conditions. be emphasized that in this work Eq. (32) is obtained as
Although the thickness of the capsule wall formed a special case of a more general model derived above.
by interfacial reaction changes with time, this case
is often referred to as the steady-state model in the 2.2.2.3. Diffusion-controlled interfacial polymeriza-
literature [20]. tion. If the reaction rate is much larger than the
diffusion rate, i.e. k I
D M , then the reaction term on
2.2.2.1. Reaction- and diffusion-controlled interfacial the right hand side of Eq. (30) is negligible compared
polymerization. Assuming that both reaction and to the diffusion, thus Eq. (30) reduces to
monomer diffusion are rate controlling steps, Eq. (4)
still holds, but Eq. (2) is changed to ϕP2 (r0 − R)2 ϕP2 X 2
t= = (33)
dR kI 2DM CA (r0 , 0) 2DM CA (r0 , 0)
− = CA (R, t)CB (r0 , 0) (28)
dt ϕP Mathematically, Eq. (33) is the same as Eq. (26) that
Combining Eq. (4) with Eq. (28) gives the growth rate we obtained previously for a flat sheet model [11].
of the capsule wall as Eq. (33) is also similar to the model reported by Evans
[20] for the corrosion of a metal. Evans showed that
dR DM kI CA (r0 , 0)CB (r0 , 0) the model was consistent with experimental data on
− = (29)
dt DM ϕP + ϕP2 kI CB (r0 , 0)(r0 − R) the oxidation of copper. Eq. (33) is also essentially the
same as reported by Janssen and te Nijenhuis [7–9] for
Integration of Eq. (29) gives the polycondensation of terephthaloyl dichloride with
ϕP2 (r0 − R)2 ϕP (r0 − R) diethylenetriamine at the interface of an oil-in-water
t= + (30) emulsion. It should be noted that, in this work, Eq. (33)
2DM CA (r0 , 0) kI CB (r0 , 0)CA (r0 , 0)
is obtained as a special case of a more general model
The first term on the right hand side of Eq. (30) is derived above.
the time needed for monomer diffusion from the outer Although the above special models were derived
surface of the capsule wall to the reaction zone, and by simplifying the general model, it is important to
the second term is the reaction time. This can be seen point out that these special models can also be derived
clearly from the following two special models. independently from the general assumptions discussed
earlier and the assumptions specified in this section.
2.2.2.2. Reaction-controlled interfacial polymeriza- The spherical model under the nonsteady-state
tion. If the diffusion rate of monomer A is much boundary conditions developed in this work is dif-
faster than the reaction rate, i.e. D M
k I , Eq. (29) ferent from the flat sheet model developed earlier
reduces to due to the material balance associated with different
dR kI CA (r0 , 0)CB (r0 , 0) geometry. The spherical model under the steady-state
=− (31) boundary conditions is mathematically identical to
dt ϕP
the flat sheet model, as no material balance is required
and Eq. (30) reduces to for the steady-state models.
62 J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70

3. Results and discussion t, approaches infinity, the capsule wall approaches


the maximum thickness, thus R approaches Rmin . The
In this section, the validity of the model is first right hand side of Eq. (16) approaches infinity, and the
qualitatively assessed by examining the model under growth rate given by Eq. (14) approaches zero, and as
several limited conditions, then the model is quanti- expected, Eq. (7) gives C B (R, ∞) = 0.
tatively assessed (1) by comparing with the experi- Similar analyses can be applied to the special mod-
mental data reported in the literature [7–9], and (2) els to give the same conclusions as discussed above.
by systematically varying a variety of factors that The above analysis clearly indicates that the model is
influence interfacial polymerization. physically reasonable. The general model, Eq. (16),
provides an analytical solution for the formation of the
3.1. Qualitative assessment of the general model capsule wall.

The validity of the models developed above can be 3.2. Comparison of the general model with the
qualitatively assessed in terms of the physical meaning published experimental data
of the models at certain limits.
The general model, given by Eq. (16) along with In order to verify the validity of the general model,
Eqs. (7), (14), (15) and (22), correctly reduces to the a modified Marquart non-linear parameter estimation
expected values at various points in the time domain. routine, UWHAUS [23] was used to fit the general
At t = 0, Eq. (16) gives R = r0 , i.e. X = r0 − R = 0, model, Eq. (16), to the experimental data reported
the thickness of the capsule wall is zero. Based on by Janssen and te Nijenhuis [9] for the capsule wall
Eq. (14), at t = 0, the maximum growth rate of the thickness versus the polymerization time. The results
capsule wall is proportional to the initial monomer obtained are presented in Fig. 2.
concentration. At t = 0 and R = r0 , both Eqs. (7) In Janssen and te Nijenhuis’ work, the cap-
and (15) correctly reduce to the initial concentration sule wall was formed by interfacial polymeriza-
of B and A, respectively. When polymerization time, tion of terephthaloyl dichloride (TDC) contained in

Fig. 2. Comparison of the general model with the experimental data reported by Janssen and te Nijenhuis [9]. The solid line is produced
by the general model using Eq. (16) and the parameters in Tables 1 and 2.
J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70 63

Table 1 Table 2
Variables used in fitting the model to experimental data The results of data fitting with 95% confidence intervals
CA (r0 , 0) = 1.0 kmol/m3 Diffusion coefficient (1.00 ± 0.15) × 10−12 m2 /s
CB (r0 , 0) = 0.233 kmol/m3 Number of degrees of freedom 9
k = 1.0 × 10−3 m4 /(kmol s)
KA = 1.0
MA = MDETA = 103.17 kg/kmol reported that both methods were equally effective. As
MB = MTDC = 203.02 kg/kmol
MZ = MHCl = 36.45 kg/kmol
a result, the second method was used in the current
ρ P = 1200 kg/m3 work because of its simplicity.
ϕ P = 0.78 The partition coefficient, K, of DETA between the
aqueous phase and capsule wall was assumed to be
one. Based on Janssen and te Nijenhuis’ work [7–9],
dispersed dimethylphthalate droplets with diethyl- the rate constant, kr , of the polymerization of TDC
enetriamine (DETA) in a continuous aqueous phase. with DETA is in the order of 104 –106 m3 s/kmol, thus
Thus, monomer B corresponds to TDC, and monomer the rate of interfacial polymerization of TDC and
A to DETA in the general model discussed above. DETA is controlled by monomer (DETA) diffusion.
The general model can be correlated to the interfacial For the data fitting purpose, the actual value of kI
polymerization of TDC with DETA as follows: in Eq. (2) is not critical in this work because it is a
diffusion-controlled process.
TDC + γ DETA → Polymer + HCl (34)
Therefore, a single parameter fitting of Eq. (16) to
The effect of HCl produced from the condensation re- the experimental data was conducted using a modified
action is negligible due to the presence of 1 kmol/m3 Marquart non-linear parameter estimation routine,
of NaOH in the system used by Janssen and te Nijen- UWHAUS [23]. The solid points in Fig. 2 are the
huis [9]. The reaction ratio, γ , is in a range between experimental data reported by Janssen and te Nijen-
2/3 and 1, i.e. 2/3 ≤ γ ≤ 1. If γ = 1, a linear huis [9]. The solid line was produced by the general
polymer is formed. If 2/3 < γ < 1, a crosslinked model using Eq. (16) and parameters in Tables 1
network is formed. Therefore, in attempting to fit the and 2. The capsule wall thickness linearly increases
general model with the experimental data reported by with the square root of reaction time as shown in
Janssen and te Nijenhuis [9], the apparent reaction Fig. 2. The data fitting gives a diffusion coefficient of
rate constant kI , in Eq. (2) was replaced with kI γ , and (1.00 ± 0.15) × 10−12 m2 /s as shown in Table 2. This
the apparent diffusion coefficient, DP , in Eq. (3) was value is reasonable, because it is very close to the
replaced with DP /γ [7]. The exact numerical value of reported diffusion coefficient of 1,6-hexanediamine in
γ was not given by Janssen and te Nijenhuis [7–9]. polyamide, 8.0 × 10−12 m2 /s [25]. The above results
All parameters except K, k and ϕ P in Table 1 were clearly demonstrated that the general model devel-
taken from the work by Janssen and te Nijenhuis [7–9]. oped in this work is consistent with experimental data
The average polymer volume, ϕ P , in the capsule wall reported by Janssen and te Nijenhuis [7–9] with the
was measured from SEM’s of the capsule wall shown capsule wall thickness linearly increasing with the
in [7] using Sigma Scan graphic analyzer [24]. No square root of reaction time. It is important to point
numerical value of ϕ P was given in [7–9] although it out that the special model derived above for diffusion
was used by the authors. In Janssen and te Nijenhuis’ controlled interfacial polymerization, Eq. (33), can
work [9], two methods were used to calculate polymer also describe the data in Fig. 2.
fraction, ϕ P , in the capsule wall. In the first method,
the water flux was assumed to be a constant, thus the 3.3. Effect of water diffusion on the
polymer fraction, ϕ P , in the capsule wall is propor- thickness and porosity of the capsule wall
tional to the reaction time. In the second method, the
water flux was assumed to be proportional to the flux The general model developed in this work allows
of the DETA monomer, resulting in ϕ P being indepen- the formation of a porous capsule wall by considering
dent of reaction time. Janssen and te Nijenhuis [7–9] the simultaneous diffusion of water and monomer A.
64 J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70

Table 3 sebacoyl chloride in an organic solution, because the


The relationship between the polymer volume fraction, ϕ P , in presence of water resulted in a porous membrane.
the capsule wall and the molar ratio, β, of water to the amine
A typical impact of the polymer volume fraction
monomer transported together
on the thickness of capsule wall is displayed in Fig. 3.
β ϕP At a given polymerization time, the thickness of cap-
0 1.0 sule wall increases with decreasing volume fraction
2 0.87 of the polymer in the capsule wall. These simulations
4 0.78 indicate the importance of water transport from the
8 0.64
16 0.47
aqueous phase into the forming capsule wall on the
porosity and thickness of the capsule wall. It also pro-
vides an important guideline to control the porosity
The relationship between the polymer volume frac- and the thickness of the capsule wall. For example,
tion, ϕ P , in the capsule wall and the molar ratio of the diffusion of water should be eliminated or min-
water to diethyltriamine flux, β, given by Eq. (5), imized to obtain a dense capsule wall (membrane),
is presented in Table 3. A detail discussion on this and maximized to obtain a very porous capsule wall.
relationship was given in our previous work [11]. As
expected, the amount of water transported increases, 3.4. Kinetic behavior of capsule wall formation by
corresponding to increasing β, the polymer volume interfacial polymerization
fraction, ϕ P , decreases. If no water is transported with
diamine monomer, a dense capsule wall is formed Having established the model and approximated a
because ϕ P = 1 when β = 0. This result is consistent set of parameters, then the effects of different factors
with the experimental results reported by Morgan and on the kinetics of interfacial polymerization and on the
Kwoleck [16]. They could not make a dense reverse formation of capsule wall can be studied. In particular,
osmosis membrane by liquid–liquid interfacial poly- those parameters which are difficult to measure exper-
merization of hexamethylenediamine in water and imentally are of interest. In this section, the general

Fig. 3. The effect of polymer fraction in the capsule wall and polymerization time on the thickness of the capsule wall. The curves are
produced using Eq. (16) and the parameters in Tables 1 and 2.
J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70 65

Fig. 4. The effect of polymerization time on the thickness and the growth rate of the capsule wall. The curves are produced using Eqs. (16)
and (14) and the parameters in Tables 1 and 2.

model developed for encapsulation by reaction- and increases with polymerization time. However, at a
diffusion-controlled interfacial polymerization under higher initial concentration of B, the capsule wall
nonsteady-state boundary conditions, and the param- grows faster and approaches a higher maximum wall
eters in Tables 1 and 2, except where noted otherwise, thickness, but takes a longer time to reach the same
are used in the following simulation. reduced thickness, X/Xmax . These simulations indi-
The relationship between the growth rate of the cate that the use of a higher initial B concentration
thickness of the capsule wall with polymerization can significantly increase the growth rate of the cap-
time is displayed in Fig. 4. The growth rate decreases sule wall and shorten the time required for a certain
rapidly with time in the early stage of polymerization, wall thickness, thereby increasing the efficiency of
then gradually approaches zero. In contrast, the cap- the encapsulation process.
sule wall thickness increases dramatically with time The effect of the initial concentration of monomer
at the beginning, then slowly approaches the maxi- A on interfacial polymerization and capsule formation
mum value with time, even when the reaction time are illustrated in Fig. 6. As can be seen from Fig. 6,
approaches infinity. For the case shown in Fig. 4, the the capsule wall thickness increases more rapidly and
polymerization is completed in about 0.8 h, however, approaches the maximum value at shorter times when
the wall thickness exceeds 50% of the maximum the concentration of A, CA (r0 , 0), is increased. How-
wall thickness in less than 25% of the reaction time. ever, the maximum thickness of the capsule wall is
As would be expected the early stage of polymer- independent of the initial A concentration. The maxi-
ization plays a more important part in the formation mum thickness of the capsule wall is solely determined
of capsule wall than the later stages of the reaction. by the initial B concentration, Eqs. (21) and (22).
Prolonging the polymerization time is not effective in The rate constant for the polymerization is an impor-
attempting to increase the capsule wall thickness. tant parameter with respect to the kinetics of capsule
The impact of monomer B concentration on the wall formation by interfacial polymerization [13–19].
capsule formation is shown in Fig. 5. The capsule Experimentally it is difficult to study the effect of
wall thickness is calculated by Eq. (16). Regardless change in rate constant without making other changes
of initial B concentration, the capsule wall thickness in the system. The effect of the reaction rate constant
66 J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70

Fig. 5. The effect of monomer B concentration and polymerization time on the thickness of the capsule wall. The curves are produced
using Eq. (16) and the parameters in Tables 1 and 2.

Fig. 6. The effect of monomer A concentration and polymerization time on the thickness of capsule wall. The curves are produced using
Eq. (16) and the parameters in Tables 1 and 2.
J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70 67

Fig. 7. The effect of the apparent reaction rate constant, k, and polymerization time on the concentration profile of monomer A in the
reaction zone. The curves are produced using Eq. (15) and the parameters in Tables 1 and 2.

Fig. 8. The effect of the apparent reaction rate constant, k, and polymerization time on the concentration profile of monomer B in the
reaction zone. The curves are produced using Eq. (7) and the parameters in Tables 1 and 2.
68 J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70

Fig. 9. The effect of the apparent reaction rate constant, k, and polymerization time on the thickness of the capsule wall. The curves are
produced using Eq. (16) and the parameters in Tables 1 and 2.

and polymerization time on the concentrations of A polymerization of a diamine with a diacid chloride to
and B in the reaction zone, CA (R, t) (Eq. (15)) and form polyamide belongs to this type [7–9,12–19].
CB (R, t) (Eq. (7)), are displayed in Figs. 7 and 8, For cases involving intermediate reaction rate con-
respectively. Three types of interfacial polymeriza- stants, (e.g. 1 × 10−6 m4 /(kmol s)), the driving force
tion can be distinguished from Fig. 7 in terms of the for diffusion (C A (r0 , 0)−C A (R, t)), is initially small.
rate-controlling step. As time proceeds, the driving force increases, indicat-
When the apparent reaction rate constant is small ing the importance of diffusion on the growth rate of
(1 × 10−8 m4 /(kmol s)), no significant change in capsule wall, and finally decreases as the dicarbonyl
both CA (R, t) and CB (R, t) is observed after 0.7 h chloride is consumed and polymerization is controlled
reaction from Figs. 7 and 8, respectively. This is by the reaction kinetics. Thus, for this case both reac-
a typical reaction-controlled interfacial polymeriza- tion and diffusion are important for the formation of
tion. In this case, the diffusion of A has no effect capsule wall. The wall thickness gradually increases
on the capsule wall formation, because the driving with time and eventually approaches the maximum
force (concentration gradient) for diffusion of A value, Fig. 9.
is zero. The wall thickness increases linearly with Diffusion coefficient of monomer A has a simi-
time as shown in Fig. 9. It should be noted that lar impact on the kinetics of capsule wall formation
the same result can be obtained using the special as discussed above for reaction rate constant. Three
model, Eq. (32), for reaction-controlled interfacial types of encapsulation process can be found, namely:
polymerization. (1) reaction controlled, (2) diffusion controlled, and
When the reaction rate constant is large, 1 × (3) both reaction and diffusion controlled processes.
10−3 m4 /(kmol s), the diffusion of A is the rate con-
trolling step. As soon as the polymerization starts,
the concentration CA (R, t) falls to zero, and CB (R, t) 4. Conclusion
also quickly approaches zero with time. As a result,
the capsule wall thickness rapidly approaches the A general model for the encapsulation by both
maximum value (Fig. 9). In general, the interfacial diffusion- and reaction-controlled interfacial polymeri-
J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70 69

zation under nonsteady-state boundary conditions rate of the capsule wall, but not the maximum thick-
has been developed, which is consistent with the ness. The higher the concentration of B, the shorter
experimental data published in the literature [7–9]. the polymerization time needed to obtain a certain
The general model developed can be reduced to the capsule wall thickness. However, with a higher con-
following special models under given conditions: centration of B longer reaction time is required to
reach the same reaction conversion of B expressed by
1. nonsteady-state boundary conditions;
the reduced thickness, X/Xmax , of the capsule wall.
• reaction-controlled interfacial polymerization, The thickness of the capsule wall rapidly increases
• diffusion-controlled interfacial polymerization, with time during the early stage of polymerization,
then levels off with time as the polymerization pro-
2. constant concentration of monomer B in the
ceeds, while the growth rate of the capsule wall
capsule core;
decreases with polymerization time. After the initial
• both diffusion- and reaction-controlled interfa- rapid growth, the capsule wall thickness does not
cial polymerization, increase much. Thus, lengthening the polymeriza-
• reaction-controlled interfacial polymerization, tion time in an attempt to increase the capsule wall
• diffusion-controlled interfacial polymerization. thickness would be ineffective.
Diffusion of water, associated with the monomer A,
The analytical solutions to the second model of during the capsule wall formation has an important im-
case 2 agreed with the spherical model developed pact on the physical structure of capsule wall formed
by Neogi and Allan for the controlled release of by interfacial polymerization. A porous capsule wall is
pesticides by decomposition of the polymeric car- formed when water diffuses together with monomer A
rier [21,22]. And the analytical solutions to the third towards reaction zone. Elimination of water diffusion
model of case 2 agreed with the flat sheet models during capsule wall formation gives a dense capsule
published earlier [7–9,11]. Therefore, the general wall free of micro voids.
model developed in this work is not only supported Although the models discussed above are devel-
by the experimental data [7–9] but also by its special oped for encapsulation by interfacial polymerization,
models being mathematically identical to the mod- they are mathematically suitable for any kind of in-
els published in the literature [7–9,11,21,22]. All the terfacial reactions which satisfy the assumptions used
models developed above are simple, have easy to use to develop the models. The models developed provide
analytical solutions, have clear physical meaning, and an important guide for the effective control of the
predict results that would be expected for the encap- capsule wall thickness and porosity.
sulation by interfacial polymerization. This work has
significantly extended the existing theories of capsule
wall formation by interfacial polymerizations.
The effects of the reaction rate constant and the dif- References
fusion coefficient on capsule wall formation have been
[1] D.R. Karsa, R.A. Stephenson (Eds.), Encapsulation and
systematically studied. Both the reaction rate constant
Controlled Release, Royal Society of Chemistry, Bookcraft
and the diffusion coefficient strongly affect the cap- Ltd., Bath, 1993.
sule wall growth rate. Three cases of interfacial poly- [2] K. Hong, S. Park, Preparation of polyurea microcapsules
merizations have been predicted by the model in terms containing ovalbumin, Mat. Chem. Phys. 64 (2000) 20–24.
of the rate controlling step: (1) diffusion-controlled, [3] S.J. Risch, G.A. Reineccius (Eds.), Encapsulation and Contro-
(2) reaction-controlled, and (3) both diffusion- and lled Release of Food Ingredients, American Chemical Society,
reaction-controlled. Monomer concentration has an Washington, DC, 1995.
important effect on the capsule wall formed. Increas- [4] L.T. Fan, S.K. Singh, Controlled Release, A Quantitative
Treatment, Springer, Berlin, 1989, 37 pp.
ing the concentration of monomer B increases not
[5] B. Warburton, Microcapsules from multiple emulsions, in:
only the growth rate but also the maximum thickness D.R. Karsa, R.A. Stephenson (Eds.), Encapsulation and
of the capsule wall. However, increasing the con- Controlled Release, Royal Society of Chemistry, Bookcraft
centration of monomer A only increases the growth Ltd., Bath, 1993, pp. 35–51.
70 J. Ji et al. / Journal of Membrane Science 192 (2001) 55–70

[6] R.G. Pearson, E.L. Williams, Interfacial polymerization of [15] F. MacRitchie, Interface effects on chemical reaction rate, in:
an isocyanate and a diol, J. Polym. Sci., Polym. Chem. 23 F. Millich, C.E. Carraher Jr. (Eds.), Interfacial Synthesis, Vol.
(1985) 9–18. I, Fundamentals, Marcel Deker, New York, 1977, p. 103.
[7] L.J.J.M. Janssen, K. te Nijenhuis, Encapsulation by interfacial [16] P.W. Morgan, S.L. Kwolek, Interfacial polycondensation. II.
polycondensation. I. The capsule production and a model for Fundamentals of polymer formation at liquid interfaces, J.
wall growth, J. Membr. Sci. 65 (1992) 59–68. Polym. Sci. 40 (1959) 299–327.
[8] L.J.J.M. Janssen, K. te Nijenhuis, Encapsulation by interfacial [17] P.W. Morgan, Development of low temperature polyconden-
polycondensation. II. The membrane wall structure and the sation processes and aromatic polyamides, J. Polym. Sci.,
rate of wall growth, J. Membr. Sci. (1992) 69–75. Polym. Symp. 72 (1985) 27–37.
[9] L.J.J.M. Janssen, K. te Nijenhuis, Encapsulation by interfacial [18] P.W. Morgan, Condensation Polymers: By Interfacial and
polycondensation. III. Microencapsulation; the influence of Solution Methods, Wiley/Interscience, New York, 1965,
process conditions on wall permeability, J. Membr. Sci. 79 p. 19.
(1993) 11–26. [19] P.W. Morgan, Interfacial polymerization, in: H.F. Mark, N.M.
[10] J. Ji, B.J. Trushinski, R.F. Childs, J.M. Dickson, B.E. Bikales, C.G. Overberger, G. Menges, J.I. Kroschwits (Eds.),
McCarry, Fabrication of thin-film composite membranes with Encyclopedia of Polymer Science and Engineering, Vol. 8,
pendant, photoreactive diazoketone functionality, J. Appl. Wiley, New York, 1985, pp. 221–237.
Polym. Sci. 64 (1997) 2381–2398. [20] U.R. Evans, The relation between tarnishing and corrosion,
[11] J. Ji, J.M. Dickson, R.F. Childs, B.E. McCarry, Mathematical Trans. Am. Electrochem. Soc. 46 (1925) 247–282.
model for the formation of thin-film composite membranes [21] A.N. Neogi, G.G. Allan, Controlled release pesticides:
by interfacial polymeriation: porous and dense films, concepts and realization, Adv. Exp. Med. Biol. 47 (1974) 195.
Macromolecules 33 (2000) 624–633. [22] S.A. Pathwardhan, K.G. Das, in: K.G. Das (Ed.), Controlled
[12] V. Enkelmann, G. Wegner, Untersuchungen zum mechani- Release Technology, Wiley, New York, 1983, p. 52.
smus der grensflachen-polykondensation von Nylon-6.10, [23] D.A. Meeter, Nonlinear Least Square (UWHAUS), University
Makromol. Chem. 157 (1972) 303–306. of Wisconsin Computing Center, Medison, WI, 1965.
[13] V. Enkelmann, G. Wegner, Mechanism of interfacial polycon- [24] D.R. Parker, J. Norby, J. Bennington, Sigma Scan Scientific
densation and the direct synthesis of polyamide membranes, Measurement System, CA, 1988.
Appl. Polym. Symp. 26 (1975) 365–372. [25] V. Enkelmann, G. Wegner, Transport properties of aromatic
[14] F. MacRitchie, Mechanism of interfacial polymerization, polyamide membranes, J. Appl. Polym. Sci. 21 (1977) 997–
Trans. Faraday Soc. 65 (1969) 2503–2507. 1007.

Das könnte Ihnen auch gefallen