Journal of Controlled Release 92 (2003) 361 – 368

www.elsevier.com/locate/jconrel

Controlled release with coating layer of permeable particles

Ryusei Ito, Boris Golman, Kunio Shinohara *
Laboratory of Particulate Engineering, Division of Materials Science and Engineering,
Graduate School of Engineering, Hokkaido University, West 8, North 13, Kita-ku, Sapporo 060-8628, Japan
Received 15 May 2003; accepted 10 July 2003

Abstract

An enhanced method was proposed for controlled release of core material using a coating layer of fine permeable particles
dispersed in an impermeable wax prepared by dry-based process. A mathematical model was constructed to describe in detail the
core material release by diffusion through the connected permeable particles inside the coating layer. The effective diffusivity was
simulated by a random walk method taking into account the structure of the coating layer. The released characteristics were
measured for the urea core particle coated with the layer of the starch permeable particles dispersed in the paraffin wax. The
calculated results were in a good quantitative agreement with experimental data in all range of coating conditions. As a result, the
low release rate was proven to be obtained with thicker coating layer of lower volume fraction of permeable particles. Moreover,
the application of permeable particles instead of soluble ones [J. Chem. Eng. Jpn. 35 (2002) 40] resulted in significant decrease in
release rate.
D 2003 Elsevier B.V. All rights reaserved.

Keywords: Controlled release; Coating layer; Permeable particles; Mathematical model; Random walk simulation

1. Introduction The encapsulated particles are mainly produced by

Controlled release is an important technique uti-
lized in various fields for supplying a prescribed
amount of necessary material at desired time [2]. The
encapsulated particles are the most commonly used
functional particles for controlled release because the
active component occupies large volume fraction, and
the release rate of core material is of zero order, which
is governed by the diffusion of dissolved material in
the coating layer [3].
* Corresponding author. Tel.: +81-11-706-6590; fax: +81-11-
706-6593.
E-mail address: kushinoh@eng.hokudai.ac.jp (K. Shinohara).
spraying a low permeable polymer dissolved in organ-
ic solvent that causes environmental problems. Such
solvent has the carcinogenic properties and to be
regulated in the near future. Therefore, the water-based
polymer solution has been suggested for the produc-
tion of coating layer [4,5]. However, the polymer
coating obtained is fairly unstable during long storage,
and the coating process also requires much heat for
water evaporation. To overcome these problems, a
nearly dry-based process was proposed to coat active
core particles with soluble fine ones dispersed in an
impermeable wax layer by utilizing a high-shear mixer
[1]. This process requires much less heat for the
formation of the stable wax layer than the conventional
spraying method with polymer solution. The soluble

0168-3659/$ - see front matter D 2003 Elsevier B.V. All rights reaserved.
doi:10.1016/S0168-3659(03)00363-8
362 R. Ito et al. / Journal of Controlled Release 92 (2003) 361–368
particles in the coating layer will dissolve when the
coated particle is brought in contact with water. Thus,
the penetration paths will be formed in place of
connected soluble particles and the dissolved core
material will diffuse to the outer side of coating layer
through these paths.
The requirement for release time widely varies in
different application fields, i.e. from hours to days in
pharmacy [6] and in a scale of days or months in
agriculture [7,8]. However, the reduction of release
rate of dry-based coated particles described above is
limited by the size of soluble particles. The core
particle will be difficult to be coated with considerably
small soluble particles due to the preferable formation
of aggregates of small particles as a result of increased
influence of cohesion force. Therefore, the dispersion
of permeable fine particles instead of soluble ones in
the impermeable wax layer is proposed here to
achieve the prolonged release of the dry-based coated
particle.
The objectives of this paper are to study the release
characteristics of an urea core particle coated on dry
basis with the layer of fine permeable particles dis-
persed in the impermeable wax and to verify the
detailed mathematical model developed for the descrip-
tion of core material release by diffusion through the
connected permeable particles inside the coating layer.
2. Materials and methods
Spherical urea agglomerates (Mitsui Chemicals,
Japan) commonly utilized as fertilizer was sieved in
Fig. 1. Size distribution of starch particles.
Fig. 2. Flow chart of dry-based coating of particles.
the size range of 1400– 2000 Am to be used as the
model core particles. Starch particles (Kanto Kagaku,
Japan) under 63 Am were applied as permeable par-
ticles. The number based size distribution of starch
particles for maximum length illustrated in Fig. 1 was
measured with image analysis software, Semafore
(JEOL, Japan) on SEM photographs. The data were
fitted by the distribution curve of log-normal type with
an average size of 6.18 Am and a geometric standard
deviation of 1.55. The paraffin wax powder (PARAF-
LINT SPRAY-30, Sasol Wax, Germany) with an
average size of 8 Am was applied for the formation
of impermeable layer. A 1-wt.% ethanol solution of
ethylcellulose, EC (Kanto Kagaku, Japan), was adop-
ted as a binding liquid.
The coating of core particles with starch-permeable
particles and paraffin wax powder was carried out with
a high-shear type of mixer – granulator (NMG-1L,
Nara Machinery, Japan). It is equipped with a verti-
cally mounted large size impeller rotated near the
bottom of the bowl and a small size granulator
mounted horizontally on the side wall.
The flow chart of the dry-based coating procedure
is presented in Fig. 2. At first, the starch particles were
thoroughly mixed with paraffin wax powder in the
bowl. Then, the urea particles were added to the
mixture and they were altogether rotated to promote
the formation of the layer of starch and wax particles
over the urea core. The aggregates thus obtained were
relatively weak and to make them more stable, the
small amount of EC solution was sprayed above the
rotated particle bed. Typically, the weight ratio of EC
 R. Ito et al. / Journal of Controlled Release 92 (2003) 361–368 363

Table 1 surface of the coated particle through the penetration

Composition and layer thickness of coated particles paths.
Starch volume Layer Paraffin Starch Urea
fraction(
) thickness (Am) wax (g) (g) (g)
3.1. Evaluation of effective diffusivity
0.10 50 11.9 2.21 100
100 25.2 4.68
The diffusion rate of the dissolved core material in
200 56.3 10.4
0.20 100 22.4 9.35 the coating layer strongly depends on the layer struc-
0.30 19.6 14.0 tural characteristics [9]. The volume fraction and the
0.40 16.7 18.7 size distribution of permeable particles mainly govern
0.50 14.0 23.4 the structural features of the layer. The random walk
method [10] was adopted to simulate the structure of
the permeable coating layer and to estimate the effec-
solution to the powder mixture of starch and paraffin tive diffusivity. A three-dimensional cubic volume
wax was as 1– 6. The aggregates were then dried at consisting of more than 10,000 fine permeable par-
333 K to evaporate the ethanol solvent from the binder ticles is considered for the simulation of the molecular
solution. The reasonably strong coated agglomerates movement as a representative part of the coating layer.
were formed as the ethylcellulose binds the starch and The flow chart of the simulation is illustrated in Fig. 3.
wax particles in the layer. Finally, the agglomerates At first, the layer structure is generated by random
were heated at 373 K for 10 min to melt wax powder packing of spherical permeable particles into the cube
and form the coating layer. Thus, the coated urea without overlap. The periodic boundary conditions are
particles were obtained with connected permeable maintained during the simulation. The size of the
starch particles dispersed inside the impermeable particles is chosen in compliance with the desired
wax layer. The compositions and layer thicknesses of log-normal distribution using a Box –Muller algorithm
coated particles are summarized in Table 1. The [11]. The spherical particles are then encapsulated with
impeller and granulator rotational speed was kept a virtual shell [12], which can overlap neighbor par-
constant at 200 and 300 rpm, respectively.
To estimate the release rate, the coated particles in a
basket were suspended in a beaker with 300 ml of
distilled water kept at 303 K under constant stirring.
The urea concentration in 1-ml sample taken from the
solution at desired time was measured by high-perfor-
mance liquid chromatography. A chromatograph (LC-
Vp series, Shimadzu, Japan) is equipped with a col-
umn (ODS-A, YMC, Japan) kept at 313 K. Distilled
water was used as an eluent at constant flow rate of 1.5
ml/min. The amount of urea in the sample of 20 Al was
measured with a refractive index detector (RID-10A,
Shimadzu, Japan).

3. Theoretical

The moment the coated particle is placed in contact

with water, it starts penetration to the particle core
through paths formed with the connected permeable
particles inside the coating layer. The core particle
dissolves upon water reaches the outer core surface.
Then, the dissolved core material diffuses to the outer Fig. 3. Flow chart of diffusivity simulation by random walk model.
364 R. Ito et al. / Journal of Controlled Release 92 (2003) 361–368
ticles and shells. This shell was introduced to account
for the out-of-roundness of permeable particles used in
the experiments, which results in larger contact area
between the neighbor particles than the area between
spherical particles. The shell thickness was selected to
be proportional to the coated particle radius and its
value is obtained by fitting experimental release data.
Next, the availability of penetration paths extended
from the bottom to the top side is checked for the
whole cube volume, and the volume fraction, Vpath, is
calculated for these paths. Finally, the random move-
ments of test points are simulated inside the paths [13].
The direction of the movement is randomly chosen for
each step. If the test point is intended to go outside the
path, it stops for the time step and again retries the
movement in random direction. The effective diffusiv-
ity, Deff, is calculated as an averaged displacement for
10,000 points normalized for a large number of steps
[14]:
 
X2
Deff ¼ Vpath Dperm
DX 2 s
where X is the test point displacement, DX is the
constant step displacement, s is the dimensionless time
count and Dperm is the diffusivity of core material in
the void space of permeable particle. The value of
Dperm is the fraction of corresponding diffusivity value
in the bulk solvent phase, Dbulk.
3.2. Diffusion through coating layer
The concentration of core material in the coating
layer is described by mass balance over the spherical
shell in the layer as:
 2  concentration, Cout(t), due to the negligible resistance
BCðr; tÞ B Cðr; tÞ 2 BCðr; tÞ
e ¼ Deff þ ð2Þ
Bt Br2 r Br
where C(r,t) is the core material concentration inside
the layer, r is the radial coordinate, t is the diffusion
time, Deff is the effective diffusivity of core material
inside the layer and e is the volume fraction of
permeable particles. The notation of the release model
by diffusion through the coating layer is presented in
Fig. 4.
The release of core material through the coating
layer is considered to proceed in two consecutive
Fig. 4. Notation for release model of coated particle.
stages. During the first stage of release, the solid core
dissolves and shrinks in size but it still exists. Then, the
second stage starts when the core completely disap-
peared. In this stage, the concentration of the inner
solution that occupies the core space decreases with
time until it equilibrates with the solution in the
reservoir.
ð1Þ
3.2.1. During dissolution of core
Here, the water penetration through the coating
layer to the core particle and the subsequent dissolu-
tion of the core are assumed to progress significantly
faster than the diffusion of dissolved material from the
core to the outer reservoir. Therefore, if the solid core
still exists, the concentration, Cin(t), of the dissolved
core material is constant in the inner solution, which
occupies the space between the solid core and the inner
surface of the coating layer because it is equal to the
concentration at saturated conditions, Csat. The core
material concentration on the outer surface of the
coating layer is assumed to be equal to the reservoir
to the mass transfer from this surface to the bulk liquid
as a result of intensive mixing with agitator inside the
reservoir. Consequently, the boundary conditions for
Eq. (2) are given by:
Cðrin ; tÞ ¼ Cin ðtÞ ¼ Csat ; Cðrout ; tÞ ¼ Cout ðtÞ; ð3Þ
where rin and rout are the radii of inside and outside
surfaces of the coating layer, respectively.
The initial condition for Eq. (2) is as follows:
Cðr; 0Þ ¼ 0 at rin VrVrout ð4Þ
 R. Ito et al. / Journal of Controlled Release 92 (2003) 361–368 365

The release rate of the core material to the perfectly The application of Eq. (7) for the calculation of
mixed reservoir is given on the outer surface of the concentration gradient on rin results in the following
coating layer as follows: equation:

dCout ðtÞ BCðr; tÞ  2 dCin ðtÞ k
V ¼
nDeff 4prout ; ð5Þ ¼ ðCout ðtÞ
Cin ðtÞÞ; ð11Þ
dt Br r¼rout dt Vcore

where V is the reservoir volume and n is the number of where Vcore is the core volume. The corresponding
coated particles used for the release test. equation for the concentration of core material in the
Since the reservoir was filled with the pure solvent reservoir could be written as
prior to each release test, the initial condition for Eq.
(5) is given as dCout ðtÞ nk
¼
ðCout ðtÞ
Cin ðtÞÞ ð12Þ
dt V
Cout ð0Þ ¼ 0 ð6Þ
The initial conditions for Eqs. (11) and (12) are as
As confirmed by our preliminary numerical solu- follows:
tion of Eq. (2) [1], the steady state is instantly attained
in the course of diffusion of the core material through Cðrin ; tsolve Þ ¼ Cin ðtsolve Þ ¼ Csat ;
the thin coating layer following the change in the outer kn

concentration. Therefore, assuming constant Cout dur- Cðrout ; tsolve Þ ¼ Cout ðtsolve Þ ¼ Csat ð1
e
V tsolve Þ ð13Þ
ing the short time interval, the steady state solution of
Eq. (2) is as follows: The reservoir and inner solution concentrations are
given by solving the system of ordinary differential
equations, Eqs. (11) and (12), with initial conditions
rin Csat ðrout Cout
rin Csat Þðr
rin Þ by Eq. (13) as follows:
CðrÞ ¼ þ ð7Þ
r rðrout
rin Þ
   
nkt kðt
tsolve Þ nktsolve
nVcore þ V
nVcore exp


V exp

Then, the solution of Eq. (5) could be obtained Cout ðtÞ ¼ Csat
V Vcore V
nVcore þ V
from Eq. (7) in the following form:
ð14Þ
 
kn    
Cout ðtÞ ¼ Csat 1
e
V t ; ð8Þ nVcore þ V þ V exp

nkt kðt
tsolve Þ


V exp

nktsolve
V Vcore V
Cin ðtÞ ¼ Csat
nVcore þ V
where the coefficient k is defined as: ð15Þ
r in r out
k ¼ 4Deff ð9Þ The time for complete dissolution of the core,
r out
r in tsolve, could be obtained from the overall mass
balance for the release system. When the solid core
3.2.2. During dissipation of core solution disappears, the inner solution of concentration Csat
After the solid core has completely disappeared fills the entire volume of initial solid core and the
at time tsolve, the concentration of inner solution amount of dissolved core material, Min(tsolve), is
starts to decrease with time. The material balance on equal to CsatVcore. The amount of the core material
the inner surface of the coating layer could be released from one coated particle to the reservoir
written as for the time tsolve, Mrel(tsolve), could be calculated
as:

dCin ðtÞ Deff BCðr; tÞ  2 Cout ðtsolve ÞV
¼ 4rin ð10Þ
dt Vcore Br r¼rin Mrel ðtÞ ¼
n
ð16Þ
366 R. Ito et al. / Journal of Controlled Release 92 (2003) 361–368
Taking into account the negligible amount of the
dissolved material inside the thin coating layer, the
following mass balance equation could be setup:
Min ðtsolve Þ ¼ Mcore
Mrel ðtsolve Þ
where Mcore = Vcore(q/Mw) is the initial amount of the
solid core material, q is the density of core material and
Mw is the molecular weight of core material.
Finally, tsolve is obtained from Eq. (17) as:
  
V Vcore q
tsolve ¼
ln 1
n
1
kn V Mw Csat
Then, the released fraction, R(t), at time t is defined
as:
RðtÞ ¼ Cout ðtÞ=Cout ðlÞ
4. Results and discussion
4.1. Effect of starch content
Fig. 5 shows the measured release curves of urea
particles coated with a layer of 100 Am in thickness,
estimated from the total quantity of coating materials
and various starch amount. The measured time for the
complete dissolution of uncoated urea particle was
less than 1 min under the same condition. The
released fraction of coated particles increased more
Fig. 5. Effect of starch content on release fraction.
ð17Þ
ð18Þ
Fig. 6. Effective diffusivity at various starch contents.
slowly than uncoated ones with elapse of time since
ð19Þ the diffusion of the urea dissolved is inhibited by the
coating layer. The values of the effective diffusivity
shown in Fig. 6 were estimated by fitting experimen-
tal data with Eqs. (8) and (14) using a nonlinear least
square method. This model describes experimental
release data reasonably well, as illustrated in Fig. 5.
The small discrepancy between experimental and
calculated results for / = 0.30 could possibly be
attributed to the microcrack generation in the coating
layer during release. The release rate gets lower for
particles of lower volume fraction of starch due to
stronger diffusion resistance inside the coating layer.
Decrease in the volume of paths formed by connect-
ing starch particles corresponds to the lower proba-
bility of molecules to diffuse from a particle to
neighbor one through the contact area between the
particles [9].
4.2. Effect of coating layer thickness
Fig. 7 illustrates the effect of coating layer thick-
ness on the release rate for particles coated with 10
vol.% of starch. The release rate becomes low for the
thick coating layer since the molecules of the dis-
solved core material should diffuse in the long dis-
tance inside the layer to reach the outer reservoir.
Following Eq. (9), the coefficient k increases with
the thickness of coating layer, rout
rin. This results in
lower reservoir concentration, Cout, according to Eq.
(8) or Eq. (14). The lines in this figure were calculated
for the same value of effective diffusivity equal to
 R. Ito et al. / Journal of Controlled Release 92 (2003) 361–368
0.15  10
11 m2/s. The calculated curves are in good
agreement with the experimental data for particles of
all layer thicknesses.
4.3. Effect of permeable particles
The contribution of permeable particles to the
hindering of the release rate was assessed by the value
of diffusivity of dissolved core material inside the
permeable particle in comparison with the one in the
bulk phase.
The random walk method could be used for the
simulation of the ratio of effective diffusivity to that
inside permeable particles, provided that the value of
virtual shell thickness is specified. Here, the value of
the ratio of virtual shell thickness to permeable
particle radius, Rv, was estimated by fitting the exper-
imental data on effective diffusivity with the calculat-
ed one for various virtual shell thicknesses, as shown
in Fig. 8. The higher release rate could be obtained for
the larger Rv due to the larger value of Deff as the
result of larger contact area between connected par-
ticles. The value of Rv equal to 0.2 had the least
difference between the experimental and simulated
data. Then, the diffusivity inside the permeable par-
ticles could be calculated as the ratio of Deff estimated
by fitting experimental release data in Fig. 6 to the
value of Deff/Dperm simulated by the random walk
method. As a result, D perm was obtained as
3.86  10
11 m2/s. This value is 45 times lower than
that of urea diffusivity in bulk water of 1.37  10
9
Fig. 7. Effect of coating layer thickness on release characteristics.
367
Fig. 8. Evaluation of virtual shell thickness.
m2/s, according to the literature data [15]. Therefore,
it was possible to achieve significantly lower released
rate with permeable particles than with soluble one in
our previous study [1].
5. Conclusions
The dry-based coating of core particle with the
impermeable wax dispersing fine permeable par-
ticles was proposed to extend the release time of
the core material. The detailed mathematical model
was established to describe the core material release
by diffusion through the connected permeable par-
ticles inside the coating layer. The good agreement
between the experimental data and the calculated
results for all coating conditions confirms the
applicability of the developed model. As a result,
the dispersion of permeable particles instead of
soluble ones in the coating layer yielded in signif-
icant decrease in release rate. In addition, the
longer release time was achieved with thicker
coating layer of lower volume fraction of perme-
able particles.
Nomenclature
C concentration of core material solute
(kmol/m3)
Dbulk diffusivity of core material solute in bulk
phase (m2/s)
Deff effective diffusivity of core material solute
within coating layer (m2/s)
368 R. Ito et al. / Journal of Controlled Release 92 (2003) 361–368

Dperm diffusivity of core material solute inside References

permeable particle (m2/s)
f(x) frequency distribution of permeable particle [1] R. Ito, B. Golman, K. Shinohara, Controlled release of core
size (m
1) particle coated with soluble particles in impermeable layer,
J. Chem. Eng. Jpn. 35 (2002) 40 – 45.
L coating layer thickness (m) [2] Toray Research Center: Jyoho Gijutu to Yoto Tenkai, Asahi
M(t) mole amount of solid core material at time t Kosoku Printing, Tokyo, 1997, pp. 1 – 8.
(kmol) [3] C. Berkland, M. King, A. Cox, K. Kim, D.W. Pack, Precise
Mcore initial mole amount of solid core material control of PLG microsphere size provides enhanced control of
(kmol) drug release rate, J. Control. Release 82 (2002) 137 – 147.
[4] N. Shimono, T. Takatori, M. Ueda, M. Mori, Y. Higashi, Y.
Mrel(t) total mole amount of released core material Nakamura, Chitosan dispersed system for colon-specific drug
(kmol) delivery, Int. J. Pharm. 245 (2002) 45 – 54.
Mw molecular weight of core material (kg/kmol) [5] W.A. Chan, C.D. Boswell, Z. Zhang, Comparison of the re-
n number of coated particles in basket lease profiles of water soluble drug carried by eudragit-coated
capsules in different in-vitro dissolution liquids, Powder Tech-
R(t) release fraction
nol. 119 (2001) 26 – 32.
Rv ratio of virtual shell thickness to core particle [6] R.W. Baker, H.K. Lonsdale, in: A.C. Tanquary, R.E. Lacey
radius (Eds.), Controlled Release of Biologically Active Agents, Ple-
r radial coordinate in coated particle (m) num, New York, 1973, pp. 15 – 71.
t release time (s) [7] U. Shavit, M. Reiss, A. Shaviv, Wetting mechanisms of gel-
V volume of reservoir (m3) based controlled-release fertilizers, J. Control. Release 88
(2003) 71 – 83.
Vcore initial core volume (m3) [8] S. Shoji, Meister: Controlled Release Fertilizer: Properties and
Vpath volume fraction of percolation paths Utilization, Konno Printing, Sendai, Japan, 1999, pp. 16 – 20.
X distance from initial to final test points inside [9] R. Ito, B. Golman, K. Shinohara, Effect of porous structure in
coating layer (m) coating layer on release rate of core particle, J. Chem. Eng. Jpn.
DX step size of test point movement (m) 36 (2003) 676 – 682.
[10] Y. Nakano, J.W. Evans, Monte Carlo simulation of diffusion
x size of permeable particle (m) of gases in a porous solid: calculations for a new class of
xav average size of permeable particles (m) solids, J. Chem. Phys. 78 (1983) 2568 – 2572.
e voidage or volume fraction of permeable [11] R.Y. Rubinstein, Simulation and the Monte Carlo Method,
particles inside coating layer Wiley, New York, 1981, p. 111.
rg geometric standard deviation of permeable [12] L.M. Schwartz, E.J. Garboczi, D.P. Bentz, Interfacial transport
in porous media: application to DC electrical conductivity of
particle size mortars, J. Appl. Phys. 78 (1995) 5898 – 5908.
q density of core material (kg/m3) [13] K. Binder, D.W. Heermann, Monte Carlo Simulation in Statis-
s dimensionless diffusion time tical Physics, 2nd ed., Springer-Verlag, Berlin, Germany, 1992.
[14] S. Trinh, P. Arco, B.R. Locke, Effective diffusivities of point-
Subscript like molecules in isotropic porous media by Monte Carlo
simulation, Transp. Porous Media 38 (2000) 241 – 259.
in at inner boundary of coating layer [15] Ch.J. Geankoplis, M.R. Okos, E.A. Grulke, Diffusion of urea
out at outer boundary of coating layer and potassium chloride in albumin solution, J. Chem. Eng.
sat at the saturation condition Data 23 (1978) 40 – 44.
solve at time when the solid core dissolves
completely