You are on page 1of 13

Journal of Controlled Release 102 (2005) 569 581

www.elsevier.com/locate/jconrel

Mathematical modeling and in vitro study of controlled drug


release via a highly swellable and dissoluble polymer matrix:
polyethylene oxide with high molecular weights
Ning Wua,1, Li-Shan Wanga, Darren Cherng-Wen Tana,
Shabbir M. Moochhalab, Yi-Yan Yanga,*
a

Institute of Bioengineering and Nanotechnology, 31 Biopolis Way, The Nanos, #04-01, Singapore 138669, Singapore
b
DSO National Laboratory, 27 Medical Drive, #12-00, Singapore 117510, Singapore
Received 1 June 2004; accepted 1 November 2004
Available online 8 December 2004

Abstract
A mathematical model is developed to describe the transport phenomena of a water-soluble small molecular drug
(caffeine) from highly swellable and dissoluble polyethylene oxide (PEO) cylindrical tablets. Several important aspects in
drug release kinetics were taken into account simultaneously in this theoretical model: swelling of the hydrophilic matrix and
water penetration, three-dimensional and concentration-dependent diffusion of drug and water, and polymer dissolution. The
moving boundary conditions are explicitly derived, and the resulting coupled partial differential equations are solved
numerically. In vitro study of swelling, dissolution behavior of PEOs with different molecular weights and drug release are
also carried out. When compared with experimental results, this theoretical model agrees with the water uptake, dimensional
change and polymer dissolution profiles very well for pure PEO tablets with two different molecular weights. Drug release
profiles using this model are predicted with a very good agreement with experimental data at different initial loadings. The
overall drug release process is found to be highly dependent on the matrix swelling, drug and water diffusion, polymer
dissolution and initial dimensions of the tablets. Their influences on drug release kinetics from PEO with two different
molecular weights are also investigated.
D 2004 Elsevier B.V. All rights reserved.
Keywords: Swelling; Diffusion; Dissolution; Controlled release; Polyethylene oxide (PEO) matrix; Caffeine

1. Introduction
* Corresponding author. Tel.: +65 68247106; fax: +65
64789084.
E-mail address: yyyang@ibn.a-star.edu.sg (Y.-Y. Yang).
1
Current address: The Department of Chemical Engineering,
Princeton University, USA.
0168-3659/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jconrel.2004.11.002

Hydrophilic polymeric matrices have attracted


many researchers due to their wide applications in
controlled drug delivery. When the release medium
(i.e. water) is thermodynamically compatible with a

570

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

polymer, the polymer may undergo a relaxation


process so that the polymer chains become more
flexible and the matrix swells. This could allow the
encapsulated drug to diffuse more rapidly out of the
matrix. On the other hand, it would take more time for
drug to diffuse out of the matrix since the diffusion
path is lengthened by matrix swelling. Moreover, it
has been widely known that swelling and diffusion are
not the only factors that determine the rate of drug
release [1]. For dissolvable polymer matrix, polymer
dissolution is another important mechanism that can
modulate the drug delivery rate. While either swelling
or dissolution can be the predominant factor for a
specific type of polymer [2], in most cases drug
release kinetics is a result of a combination of these
two mechanisms [3,4].
Among a variety of hydrophilic polymers, polyethylene oxide (PEO) is one of the most important
materials used in the pharmaceutical industries mainly
because of its non-toxicity, high water-solubility and
swellability, insensitivity to the pH of the biological
medium and ease of production. Recently, the
swelling and dissolution behavior of PEO tablets [5
8] and hydrogels [9] as well as their influences on
drug release characteristics have been studied. It is
found that, compared with low molecular weight
PEO, the high molecular weight PEO tablet swells to
a greater extent and the swelling of the polymer rather
than the dissolution of the polymer is the governing
factor for drug release. The compression force applied
during the manufacturing process, pH of the release
medium and the stirring rate do not affect the drug
release behavior significantly. However, most of the
above studies of PEO matrices are mainly experimental and there is no comprehensive mathematical
modeling of controlled drug release via PEO matrices
with different molecular weights. The effect of initial
dimensions of tablets on drug release kinetics is also
not studied.
The mathematical modeling of drug release is of
great importance in pharmaceutical science and
engineering because the idealized but key transport
mechanisms can be studied in the mathematical
model and the model itself can be used to predict
the effects of the composition and geometry on drug
release profiles, which is very helpful to the design
of new drug delivery system. The mechanism study
of drug release via a swellable and dissoluble

hydrophilic polymer matrix is not as extensive as


for purely diffusion, swelling or polymer dissolutioncontrolled drug release systems since all these
processes are coupled, thus making the models more
intricate and difficult to solve. Both empirical [10,11]
and theoretical [12,13] mathematical models have
been reported for drug release via either cross-linked
[14] or uncross-linked [15] polymer matrix in
literature. However, most of the models consider
only certain physical characteristics and neglected
the others. For example, diffusion is not considered
[11]; a pseudosteady state assumption is made [16];
transport of active agent is considered only in onedimensional [1719]. Recently, a two-dimensional
approach is reported [20] for HPMC tablets. Ficks
second law of diffusion with concentration-dependent diffusivity is used to describe water and drug
transport. A surface-normalized rate constant is used
to model polymer dissolution and swelling is
considered using moving boundary conditions. However, the moving boundary conditions are not
explicitly expressed.
The model developed here can be viewed as a
two-dimensional extension of the model developed
by Vrentas et al. [21] for the impurity removal in
polymer films by the addition of a second solvent, in
which the diffusion coefficients of two solvents in a
polymersolventsolvent system are related to the
concentration and properties of all the components
(two solvents and polymer) in the polymer film by
using a comprehensive free-volume theory of transport. In our work, we extend that model into twodimensional form to account for the geometry of
common cylindrical tablet. Polymer swelling and
dissolution are included since PEO is a highly
swellable and dissoluble polymer and for simplicity
we use a simpler form of diffusion coefficient which
is expressed as a function of water concentration
only.

2. Mathematical model
A schematic for the model considered here is
plotted in Fig. 1. The matrix is cylindrical with initial
radius of r 0 and initial height of 2z 0. Only the upper
half part of the cylindrical matrix is considered in
mathematical modeling since it is symmetrical to the

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

571

z
z
n

z0

r0

t=0

t>0

Fig. 1. Schematic of the tablet considered in the model.

lower half part. Swelling is assumed to be in radial


and axial directions and the diffusion of water and
drug are assumed Fickian diffusions with water
concentration dependent mechanism. Effect of
stresses arising in the swelling process on diffusion
and the influence of the diffusion-induced convection
associated with volume changes in mixing are not
considered here since for most polymersolvent
systems these effects can be neglected [18,22]. The
swelling is assumed to be homogeneous and at any
time, the total volume of matrix is the sum of volume
of water, drug and polymer. Polymer dissolution is
assumed on the surface of the tablet. Thus, the
positions of swelling front are the same as the
dissolution front and they are denoted as m and n in
radial and axial directions.
The three-dimensional basic equation for diffusion
of water and drug through the matrix can be written
in cylindrical coordinate with neglecting concentration gradient of both water and drug in angular
direction




BCi
1 B
BCi
B
BCi
rDi
Di

r Br
Bz
Bt
Br
Bz

1
C 1 C1 =C1;e
r r=r0

where subscript i=1 for water and i=2 for drug, C i is


the mass concentration of component i within the
matrix.
The diffusivities of drug and water are assumed to be
concentration-dependent and they can be represented
by a bFujita-likeQ free-volume model [23]


Di Di;e exp t  bi 1  C1 =C1;e b

where D i,e is the diffusion coefficient of water or drug


in the fully swollen polymer matrix system, b i is a
constant which characterizes the concentration dependence of water or drug diffusivity and C 1,e is the
equilibrium water concentration in fully swollen
polymeric matrix. D 1,e and b 1 are to be determined
by fitting of experimental data in pure polymer
swelling and dissolution studies. Values of D 2,e and
b 2 can be obtained by fitting the drug release data.
Experimentally these parameters can also be obtained
by pulsed-field-gradient spin-echo (PFGSE) NMR
techniques [24] although not investigated in this work.
The Fujita model of diffusion is a simplified model that
describes the transport of small molecules in polymer
based on free-volume theory of Vrentas and Duda
[25,26], in which the diffusion coefficient for a
penetrant is related to the volumetric and free-volume
properties of itself and other components in the
polymeric matrix. It has been successfully applied in
a variety of models of controlled drug release at
different release mediums [20,2729].
It is convenient to introduce the following dimensionless variables

C 2 C2 =C2;0
z z=r0

D1 D1 =D1;e
s D1;e t=r02

D2 D2 =D1;e

3
so that we can write Eq. (1) in dimensionless form
B Ci
1 B

Bs
r Br


r Di

B Ci
Br

B
Bz


Di

B Ci
Bz

where C 2,0 is the initial concentration of drug.

572

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

The initial conditions are


s0

Differentiating Eq. (8) with respect to time and


using Leibnizs rule yields

0V r VR
0V z VZ
C 1 0:0
C 2 1:0

dn
dm
C1;e
dn C1;e
dm
2pmn

pm2
2pmn
dt
dt
dt
dt
q1
q
! 1
Z nZ m
1 BC1
1 BC2

2prdrdz
q
q
Bt
0
0
1
2 Bt

pm2

The boundary conditions are


sN0

0V r VR
0V z VZ

z Z
r R

C 1 1:0 C 2 0:0
C 1 1:0 C 2 0:0

0V r VR

z 0

0 V z VZ

r 0

sN0

B C1
0:0
Bz
B C1
0:0
Br

B C2
0:0
Bz
B C2
0:0
Br
7

where R=m/r 0 and Z=n/z 0.


Additional boundary conditions on swelling front
m (or R) in radial direction and n (or Z) in axial
direction are needed. They can be derived from a
volume balance.
Z nZ m
1
2
2pm n 2
C1 r; z; t 2prdrdz
q1 0 0
Z nZ m
1
2
C2 r; z; t 2prdrdz
q2 0 0


Z t
1
mp;0 
Kp As dt
8

qp
0
where q i is the density of water (i=1), drug (i=2) or
polymer (i=p).
Note that the third term on the right-hand side of
Eq. (8) represents the polymer dissolution process,
which is rather complex and involves polymer chain
disentanglement and reptation [30]. To simplify the
model analysis, we use a single parameter, dissolution
rate constant K p, to characterize the surface (heterogeneous) dissolution of polymer. It is a mass rate
constant and defined as (1/A s)(dm p/dt)=K p. A s is the
total surface area of polymer matrix in contact with
the release medium and m p,0 is the initial mass of
polymer. K p can also be viewed as a product of
dissolution/mass transfer coefficient and the equilibrium polymer concentration at the dissolution
(swelling) front [18,31].

1
Kp pm2 2pmn
qp

Substituting Eq. (1) in Eq. (9) and simplifying it we


can get





C1;e
C1;e
dm
2 dn
1
1
pm
2pmn
dt
dt
q1
q


Z n
1
BC1

1
BC2

2pm
D1
D2
dz
q1
Br
rm q2
Br
rm
0



Z m
1
BC1

1
BC2

D1
D2
rdr
2p

q
q
Bz
Bz

zn

zn

1
 Kp pm2 2pmn
qp

10

If we assume that the swelling in radial direction is


independent on the swelling in z direction (edge effect
is neglected), we can split Eq. (10) into two separate
equations, which describe the rate of swelling front
advancement in r and z direction, respectively. They
are

BC1

D1
dz
Br
rm
0

Z n
Kp
1
BC2

D2
dz  n
11

q
Br

dm
1

n 1  f 1
dt
q1

BC1

D1
2rdr
Bz
zn
0

Z m
Kp
1
BC2

D2
2rdr  m2

q
Bz
q

dn
1

m 1  f 1
dt
q1
2

rm

zn

12
where f 1 is the equilibrium volume fraction of water in
the full-swollen matrix.

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

Physically, the initial and boundary conditions


mean
At t=0, the matrix is dry and a homogeneous
mixture of drug and polymer;
(ii) After the matrix is immersed into release
medium, the concentration of drug is always
zero at the surface of the cylindrical matrix, i.e.
a perfect sink condition is assumed. While the
water concentration at the surface is a constant,
which is equal to the equilibrium concentration
at a full-swollen state with polymer;
(iii) At the two symmetric planes located in the center
of the cylinder, symmetry conditions are applied;
(iv) The rate of advancement of swelling front is
determined by the amount of water diffusing in,
drug diffusing out of and polymer dissolving
from the matrix.

The initial and boundary conditions can be rewritten as


s0

(i)

Since the mathematical model developed here is a


moving boundary problem, special technique is
required. We use the front-fixing method, which was
proposed by Landau [32] and first applied to a finitedifference scheme by Crank [33] to fix the swelling
boundary.
Let n=r/R and g=z/Z. By using the relationships

sN0

0VnV1
C 1 0:0
C 2 1:0
0VnV1
0VgV1

B Ci
Bs

!
r ;z

B Ci

Bs

!

n;g

1 B C i n BR2
2 Bn R2 Bs

1 B C i g BZ 2
13
2 Bg Z 2 Bs

to transform from C i (r,z ,s) to C i (n,g,s), the diffusion


Eq. (4) can be written




B Ci 1 B
B Ci 1
B
B Ci 1
n Di

D
i
n Bn
Bg
Bs
Bn R2
Bg Z 2


1 B C i n BR2 1 B C i g BZ 2

2 Bn R2 Bs
2 Bg Z 2 Bs
14

0VgV1

C 1 1:0
C 1 1:0

g1
n1

15

C 2 0:0
C 2 0:0
16

1VnV1

g0

0VgV1

n0

sN0

sN0

B C1
0:0
Bg
B C1
0:0
Bn

B C2
0:0
Bg
B C2
0:0
Bn
17

n1

0VgV1

Z 1
1
dR2
B C 1

f1
1  f1
D1
dg
2
ds
Bn
n1
0

Z 1
Kp
B C 2

D2
dg  R
f2;0

Bn n1
qp
0
18

sN0

g1

0VnV1

B C 1

2ndn
Bg
g1
0

Z 1
Kp
B C 2

f2;0
D2
2ndn  Z

Bg g1
qp
0

1
dZ 2
f1
1  f1
2
dt

B Ci
1 B Ci

R Bn
Br
B Ci
1 B Ci

Z Bg
Bz

573

D1

19

where K p=K pr 0/D 1,e and f 2,0 is the volume fraction of


drug at s=0.
Eq. (14) can be solved with the initial and
boundary conditions of Eqs. (1519) numerically for

the spatial and time variations of C 1 and C 2. An


explicit finite-difference scheme is applied for solving
the coupled equations at each time step.
To compare the fitness of the theoretical model and
experimental results, the coefficient of determination,
R 2, is calculated [34]. It is defined as
R2

regression sum of squares


total sum of squares
n
X
yi  y i 2
i1
1 X
n

2
yi  yi
i1

20

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

3. Materials and methods


Poly(ethylene oxide) (PEO) with M w of 8106
and 4106 (Aldrich Chem. Milwaukee, USA) were
used as the dissoluble matrix material, in which
caffeine powder (Sigma, St. Louis, USA) is mixed
thoroughly by manually grinding in a stone mortar.
The resultant powder mixture is then compressed with
a laboratory hydraulic press (Graseby Specac) under a
pressure of 38 MPa for 1 min using two tablet
punches in 10-mm diameter, which have convex
surfaces. The total mass of each 10-mm-diameter
tablet was maintained at 360 mg with different
loadings of caffeine. Apparatus 2 of the USP
dissolution test (VK 7000, VanKel Technology
Group, Weston Parkway, USA) is employed for
polymer swelling, dissolution and drug release
studies. Simulated intestinal fluid (SIF, without
pancreatin), which is prepared using monobasic
potassium phosphate and sodium hydroxide (Merck,
Darmstadt, Germany), is used as a release medium.
The dissolution tests are performed in 1000 mL of
receptor medium with an agitation rate of 110 rpm.
Samples of 1 mL, which are replaced by an equal
volume of fresh receptor medium, are periodically
taken. All samples are filtered with 0.2-Am syringe
filters (Whatman, Clifton, USA) before being analyzed with high performance liquid chromatography
(HPLC). The HPLC system consists of a 2690
separations module (Waters, USA), a 996 photodiode
array detector (Waters) and a dedicated personal
computer (Compaq, USA). The UV detector wavelength is set at 270 nm for detection of caffeine. The
sample separation is achieved by using an InsertsilR
C8 analytical column (4.6150 mm, 5 Am, GL
Sciences, Tokyo, Japan), as well as an InsertsilR C8
guard column E (4.010 mm, 5 Am, GL Sciences),
at room temperature. The mobile phase consists of
methanol and ultrapure water in a ratio of 30:70 by
volume at a flow rate of 1.0 mL/min. An injection
volume of 10 AL is used.

For polymer swelling and dissolution study,


tablets of 360 mg pure PEO are made by the same
method as previously described and immersed in
1000 mL of water at 37 8C. The weight of wet and
dry PEO is accurately monitored as a function of
time. Dimensional change of tablet is measured
immediately after each sample is taken. Weight loss
is the difference between the original weight of tablet
and the weight of dry PEO and the amount of water
uptake is the weight difference between wet tablet
and dry PEO.
120

(a)
100
water uptake (%)

where y i and y i are the experimental and theoretical y


co-ordinates at time point i, y is the arithmetic mean
of the experimental y co-ordinates of the whole
series. Generally, we find the fit acceptable if R 2 is
close to 1.

80
60
40
Model
20
0

Experiment
0

12

18
time (hr)

24

30

36

20

24

120

(b)
100
water uptake (%)

574

80
60
40
Model

20

Experiment
0

12
time (hr)

16

Fig. 2. Water uptake of pure PEO tablets with two different


molecular weights. Squares denote experimental results and solid
line is the fitting curve using the model developed in this study. The
total weight of polymer tablet is 360 mg. The initial radius and
height are 0.5 and 0.381 cm, respectively. (a) PEO with M w=8106
(R 2=0.990); (b) PEO with M w=4106 (R 2=0.980).

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

3.5

4. Results and discussion

Before studying and modeling the drug release


kinetics of PEO matrices, both water uptake and
polymer dissolution studies of pure PEO tablets are
carried out and fitted with the theoretical results in
order to obtain water diffusion and polymer dissolution parameters. Fig. 2 shows the fit of the model to
the experimentally determined relative amount of
water uptake by pure PEO tablets of two different
molecular weights. As it can be seen, compared with

dimensional change X/X0

(a)

4.1. Swelling and dissolution of pure PEO matrices

n / z0

2.5
m / r0
2

1.5

15

20

2.2

Model

100

dimensional change X/X0

relative dry weight of PEO (%)

10
time (hr)

(b)

(a)
Experiment
90
80
70

n / z0

1.8
m / r0

1.6
1.4
1.2

60

1
0

10

20
time (hr)

30

40

(b)
Model

100

10

15

time (hr)
Fig. 4. Relative dimensional change of pure PEO tablets during
swelling and dissolution. Squares denote the experimental measurements. (a) PEO with M w=8106. (b) PEO with M w=4106.

120
relative dry weight of PEO (%)

2.4

110

50

575

Experiment
80
60
40
20
0

12
time (hr)

18

24

Fig. 3. The profile of relative dry weight of pure PEO tablets as a


function of time after immersion in the medium (both theoretical
and experimental). (a) PEO with M w=8106 (R 2=0.980); (b) PEO
with M w=4106 (R 2=0.954).

PEO of lower molecular weight, PEO with


M w=8106 shows a larger tendency to be hydrated
and the hydration time is up to 22 h, then the
dissolution process prevails on the swelling. On the
other hand, PEO of M w=4106 is hydrated completely within first 8 h. This is in agreement with
previous report [8]. A good agreement between the
theory and experiment can also be seen in Fig. 2.
Thus, the Fujita model of concentration-dependent
diffusivity of water describes the swelling process of a
hydrophilic polymer matrix (PEO) very well. It is
found that the diffusion coefficient of water within the
fully swollen PEO tablet of lower molecular weight
(2.9106 cm2/s) is slightly higher than that in higher

576

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

molecular weight PEO (2.6106 cm2/s), although


the concentration dependence constant b 1 is 2.0 in
both polymers. It is reasonable since the weaker chain
entanglement within lower molecular weight PEO
matrix can make water molecules more easily diffuse
through. The commonly used glassyrubbery interface is not included in our model because small water
molecules are able to diffuse across this interface
[18,35], as indicated by the low concentration dependence constant of water. Mathematically, at a glassy
rubbery interface, there is a discontinuity in the slope
of concentration profile. In our model, the slope of the
concentration profile within the matrix is always
continuous. Nonetheless, if the concentration dependence constant is high enough (i.e. solvent molecules

relative drug release (%)

80
60
40
20
0

100
relative drug release (%)

100

(a)

12
time (hr)

18

40
20

100

80
60
40
20
0

60

24

(c)

12
time (hr)

18

24

(b)

80

0
0

relative drug release (%)

relative drug release (%)

100

with low compatibility with polymer), the diffusion


front (or concentration profile) can be rather steep
[28], thus the glassyrubbery interface still can be
simulated approximately in our model.
In this model, a dissolution rate constant, K p, is
considered characterizing the polymer mass loss
velocity normalized to the actual surface of the
system: (1/A s)(dm p/dt)=K p. This constant can be
obtained from the fitting of modeling dry polymer
weight to experimental results. From Fig. 3, it can be
seen that dissolution of higher molecular weight PEO
is rather slow. It takes more than 40 h for the polymer
to reduce its weight to half, while PEO of lower
molecular weight dissolves totally within 24 h. The
dissolution constant is found to be 1.1104 mg/cm2-s

12
time (hr)

18

24

(d)

80
60
40
20
0

12
time (hr)

18

24

Fig. 5. Fit of the model to the experimentally determined relative amount of caffeine released at different loadings from PEO of M w=8106.
Squares denote experimental data. (a) 8.33% w/w (R 2=0.998); (b) 16.67% w/w (R 2=0.997); (c) 33.33% w/w (R 2=0.998); (d) 50% w/w
(R 2=0.995).

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

for PEO (M w=8106) and 8.0104 mg/cm2 s for


PEO (M w=4106). This agrees with the finding that the
presence of PEO of higher molecular weight causes a
formation of stronger matrix, which is less liable to
dissolve, than PEO of lower molecular weight [8]. It is
also possible to develop scaling laws for predicting the
dissolution constant of different molecular weight
PEOs as what Ju et al. [36] did.
With the knowledge of water diffusion and
polymer dissolution parameters, we are able to predict
the dimensional changes of the PEO tablet during
swelling and dissolution process from the numerical
calculation. Prediction of the relative dimensional
change in radial and axial directions during swelling

relative drug release (%)

relative drug release (%)

(a)

80
60
40
20
0

10
time (hr)

15

60

40

20

100

(c)

80

60

40

20

10
time (hr)

15

20

10
time (hr)

10
time (hr)

15

20

(d)

80
60

40
20

0
0

(b)

80

20

relative drug release (%)

relative drug release (%)

100

and dissolution is shown in Fig. 4. The experimental


results of radial change as function of time are also
compared with the model predictions. We do not
measure the dimensional change of the tablets in axial
direction due to difficulty in handling the tablet. The
profiles of dimensional change are consistent with the
water uptake profiles (Fig. 2). PEO of M w=4106
swells fast initially and then decreases significantly in
dimensions. Therefore, swelling is dominant for PEO
of M w=8106 over 20 h while for PEO of M w=4106
polymer dissolution dominates over swelling after
initial 8 h. It also shows that the relative increase in
height is much faster (approximately two folds) than
that in radius during swelling for both PEOs. Since the

100

100

577

15

20

Fig. 6. Fit of the model to the experimentally determined relative amount of caffeine released at different loadings from PEO of M w=4106.
Squares denote experimental data. (a) 8.33% w/w (R 2=0.994); (b) 16.67% w/w (R 2=0.997); (c) 33.33% w/w (R 2=0.991); (d) 50% w/w
(R 2=0.969).

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

4.2. Drug release from PEO matrices


The results of caffeine release from PEO matrices
of two different molecular weights at different
caffeine loadings into simulated intestinal fluid are
shown in Figs. 5 and 6. The model is fitted to the
experimentally determined relative amount of drug
released vs. time. It can be seen that drug release
profiles using this model are predicted with a very
good agreement with experimental data at different
initial loadings. Drug release from PEO matrices
involves two mechanisms, diffusion through swelling polymer and release via polymer dissolution.
Thus the swelling and dissolution behaviors of
tablets made of pure polymer play important roles
in the overall drug release process. As shown in
Figs. 5 and 6, both lower and higher molecular
weight PEO tablets do not provide zero-order release
kinetics. This is in agreement with the polymer
swelling and dissolution experiments we have done
before, in which no synchronization of swelling and
dissolution is shown. It can also be seen that the
caffeine release is faster in lower molecular weight
PEO matrix compared to the release rate in PEO of
higher molecular weight. This is, probably, due to
the faster swelling and dissolution rate of lower
molecular weight polymer. The model developed
here can simultaneously calculate the amount of
water uptake and polymer dissolution as a function
of time in the drug release process, which is difficult
to evaluate in experiment. Therefore more information on the release kinetics may be revealed by the
mathematical modeling. In Fig. 7, the predicted
relative amount of remaining polymer during caffeine release at different initial drug loadings is
plotted vs. time. It is found that as initial drug
loading increases, polymer dissolution becomes
more and more obvious since relative amount of
polymer dissolved in the release medium also
increases. Within the range of loadings investigated
in this paper, swelling and diffusion govern the
release kinetics for higher molecular weight PEO
since most of the polymer does not dissolve even

(a)
1

relative dry weight of PEO

surface area in the axial direction (1.57 cm2) is higher


than that in radial direction (1.20 cm2), the majority of
water is diffusing in the matrix from the axial
direction.

0.9

0.8

0.7

8.33%w/w
16.67%w/w
33.33%w/w

0.6
0

50.00%w/w
6

12
time (hr)

18

24

(b)
8.33%w/w
relative dry weight of PEO

578

0.8

16.67%w/w
33.33%w/w
50.00%w/w

0.6

0.4

0.2

0
0

10
time (hr)

15

20

Fig. 7. Model prediction of relative dry weight of PEO polymers


during drug release process at different initial drug loadings (weight
percentage). (a) PEO of M w=8106; (b) PEO of M w=4106.

when almost all drugs are depleted. On the other


hand, polymer dissolution plays at least an equal
role with swelling during the caffeine release via a
lower molecular weight PEO matrix.
It is worthwhile to find out expressions which
describe relationship between the equilibrium drug
diffusion coefficient D 2,e , the concentration dependence factor b 2 and the initial drug loadings in order to
predict drug release profiles at other initial loadings
and reduce the amount of experimental work. It is
found that power functions can be used to correlate
the equilibrium diffusion coefficient and the concentration dependence factor with initial drug loadings
approximately, as shown in Fig. 8.

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

579

In all cases, the initial drug loading in high molecular


weight PEO (M w=8106) tablets is kept in a constant
of 16.67% w/w. The predictions show that relative
drug release rate is faster when the initial height
decreases with constant radius or the initial radius
increases with constant height. This is because the
ratio of surface area to volume is lager with
decreasing radius if the height is kept constant. On
the other hand, if the radius is a constant the ratio of
surface area to volume is smaller at larger initial
height. Thus, the ratio of surface area to volume (i.e.
relative surface area) is an important parameter in
drug release kinetics.

(a)
relative drug release

0.8

0.6

0.4
2z0 / r0 = 0.5
2z0 / r0 = 1.0

0.2

2z0 / r0 = 2.0
0
0
1

4.3. Influence of tablet dimensions on the drug release


kinetics
The influence of initial tablet dimensions (radius
and height) on drug release profile is also studied
since they are the factors that can be easily changed in
both modeling and experiment. Fig. 9a shows the
modeling predictive effect of initial height on release
profile with constant radius, and the effect of changing
initial radius with constant height is shown in Fig. 9b.

10
time (hr)

15

20

(b)

0.8
relative drug release

Fig. 8. Correlation of b 2 and D 2,e with different initial caffeine


loadings (w) for two different molecular weight PEOs. Square:
D 2,e 10 6 cm 2 /s; triangle: b 2 . (a) PEO of M w =8 10 6 ,
D 2,e =1.5008106(w)0.2464 cm2/s, b 2=2.6056106(w)0.069; (b)
PEO of M w =4 10 6 , D 2,e =1.5353 10 6 (w ) 0.2143 cm 2 /s,
b 2=3.945106(w)0.086.

0.6

0.4
r0 / 2z0 = 0.5
r0 / 2z0 = 1.0

0.2

r0 / 2z0 = 2.0
0
0

10
time (hr)

15

20

Fig. 9. The effect of tablet (PEO of M w=8106) size on drug release


profile. The initial drug loading is 25.0% w/w. (a) Constant radius
(0.5 cm) with different heights (0.25, 0.5 and 1.0 cm); (b) constant
heights (0.5 cm) with different radii (0.25, 0.5 and 1.0 cm).

580

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581

Fig. 10. The effect of tablet (PEO of M w=8106) aspect ratio on


drug release profile. The initial volume is 0.25 cm3 and drug loading
is 25.0% w/w. From top to bottom, the release profiles correspond to
the tablet of (a) r 0=0.2 cm, 2z 0=1.989 cm (A s=2.75 cm2); (b) r 0=0.3
cm, 2z 0=0.884 cm (A s=2.23 cm2); (c) r 0=0.8 cm, 2z 0=0.124 cm
(A s=4.64 cm2); and (d) r 0=0.5 cm, 2z 0=0.381 cm (A s=2.57 cm2).

We are also interested in the effect of aspect ratio on


drug release profile. Four types of tablets from the
shape of a flat disc to a slim cylinder are modeled with
constant initial volume (0.25 cm3) and drug loading
(25% w/w). As one can see in Fig. 10, the tablets with
shapes of a flat disc or a slim cylinder release drug
much faster than the other two bregularQ shape tablets.
The difference in relative surface area is one of the
reasons why different shape of tablet has so different
release profile. However, the large difference in
relative surface area between the flat disc and the
slim cylinder shape does not show any significant
influence in terms of releasing rate. This probably
indicates that the length of diffusion path is also
another important factor that can affect drug release
rate. In tablets with either vary large or very small
aspect ratio, one of the diffusion paths (either in axial
or radial direction) is extremely short, which can
expedite the diffusion of drug in release medium.

weights (M w=8106 and M w=4106). The swelling


of the hydrophilic matrix and water penetration, threedimensional and concentration-dependent diffusion of
drug and water, and polymer dissolution are taken into
account simultaneously in this model. The moving
boundary conditions for the swelling and dissolution
front are explicitly derived, and the resulting coupled
partial differential equations are solved numerically
by using a front-fixing method. The model agrees
with the water uptake, polymer dissolution and
dimensional change profiles very well for pure PEO
tablets. No synchronization of swelling and dissolution is found in experiments and modeling for both
PEOs. It is found that swelling is the dominant factor
in drug release kinetics for higher molecular weight of
PEO (M w=8106) while both swelling and dissolution are important to caffeine release for lower
molecular weight PEO (M w=4106). Drug release
profiles using this model are predicted with a very
good agreement with experimental data at different
initial loadings. It is also found that when initial drug
loading increases, polymer dissolution becomes more
and more important in the release process. Besides
swelling and dissolution properties of polymer, the
ratio of surface area to volume and the aspect ratio of
initial tablets are also found to be influential in the
overall release profile. For possible further improvement of the model, the more complicated expression
of diffusion coefficient, which is related to the
microstructure of diffusion components and polymer
[21], may be incorporated into the analysis.

Acknowledgements
This research was funded by Agency for Science,
Technology and Research, as well as Defence Science
and Technology Agency, Singapore.

References
5. Conclusions
In this work, a mathematical model is developed
and in vitro study is carried out for the controlled
release of caffeine from polyethylene oxide (PEO)
cylindrical tablets with two different molecular

[1] J. Sujja-areevath, D.L. Munday, P.J. Cox, K.A. Khan,


Relationship between swelling, erosion and drug release in
hydrophilic natural gum mini-matrix formulations, Eur. J.
Pharm. Sci. 6 (1998) 207 217.
[2] P.I. Lee, in: D.H. Lewis (Ed.), Controlled Release of
Pesticides and Pharmaceuticals, Plenum Press, New York,
1981, pp. 39 48.

N. Wu et al. / Journal of Controlled Release 102 (2005) 569581


[3] K. Tahara, K. Yamamoto, T. Nishihata, Overall mechanism
behind matrix sustained release (SR) tablets prepared with
hydroxypropyl methylcellulose 2910, J. Control. Release 35
(1995) 59 66.
[4] M. Efentakis, G. Buckton, The effect of erosion and swelling on
the dissolution of theophylline from low and high viscosity
sodium alginate matrices, Pharm. Dev. Technol. 7 (2002) 69 77.
[5] A. Apicella, B. Cappello, M.A.D. Nobile, M.I.L. Rotonda, G.
Mensitieri, L. Nicolais, Poly(Ethylene Oxide) (PEO) and
different molecular-weight PEO blends monolithic devices
for drug release, Biomaterials 14 (1993) 83 90.
[6] C.J. Kim, Drug release from compressed hydrophilic polyoxRWSR tablets, J. Pharm. Sci. 84 (1995) 303 306.
[7] C.J. Kim, Compressed donut-shaped tablets with zero-order
release kinetics, Pharm. Res. 12 (1995) 1045 1048.
[8] L. Maggi, L. Segale, M.L. Torre, E.O. Machiste, U. Conte,
Dissolution behavior of hydrophilic matrix tablets containing
two different polyethylene oxides (PEOs) for the controlled
release of a water-soluble drug. Dimensionality study, Biomaterials 23 (2002) 1113 1119.
[9] H. Savas, O. Gqven, Investigation of active substance release
from poly(ethylene oxide) hydrogels, Int. J. Pharm. 224 (2001)
151 158.
[10] D.O. Cooney, Effect of geometry on the dissolution of
pharmaceutical tablets and other solids: surface detachment
kinetics controlling, AIChE J. 18 (1972) 446 449.
[11] H.B. Hopfenberg, in: D.R. Paul, F.W. Harris (Eds.), Controlled
Release Polymeric Formulations, ACS Symp. Ser. No., vol. 33,
American Chemical Society, Washington, 1976, pp. 26 32.
[12] P.I. Lee, Diffusional release of a solute from a polymeric
matrix-approximate analytical solutions, J. Membr. Sci. 7
(1980) 255 275.
[13] R.W. Baker, Controlled Release of Biologically Active
Agents, John Wiley and Sons, New York, 1987.
[14] D. Hariharan, N.A. Peppas, Modeling of water transport and
solute release in physiologically sensitive gels, J. Control.
Release 23 (1993) 123 135.
[15] N. Shah, G. Zhang, V. Apelian, F. Zeng, M.H. Infeld, A.W.
Malick, Prediction of drug-release from hydroxypropyl methylcellulose (HPMC) matriceseffect of polymer concentration, Pharm. Res. 10 (1993) 1693 1695.
[16] R.W. Baker, H.K. Lonsdale, in: A.C. Tanquary, R.E. Lacey
(Eds.), Advances in Experimental Medicine and Biology, vol.
47, Plenum Press, New York, 1974, pp. 15 72.
[17] L.T. Fan, S.K. Singh, A generalized model for swellingcontrolled release systems, Biotechnol. Prog. 2 (1986)
145 156.
[18] R.T.C. Ju, P.R. Nixon, M.V. Patel, D.M. Tong, Drug release
from hydrophilic matrix: 2. A mathematical model based on
the polymer disentanglement concentration and the diffusion
layer, J. Pharm. Sci. 84 (1995) 1464 1477.
[19] F.H.L. Wang, J.L. Duda, J.S. Vrentas, Analysis of impurity
migration in plastic containers, Polym. Eng. Sci. 20 (1980)
120 127.

581

[20] J. Siepmann, N.A. Peppas, Hydrophilic matrices for controlled


drug delivery: an improved mathematical model to predict the
resulting drug release kinetics (the bsequential layerQ model),
Pharm. Res. 17 (2000) 1290 1298.
[21] J.S. Vrentas, J.L. Duda, H.C. Ling, Enhancement of impurity
removal from polymer films, J. Appl. Polym. Sci. 30 (1985)
4499 4516.
[22] S. Alsoy, J.L. Duda, Influence of swelling and diffusioninduced convection on polymer sorption processes, AIChE J.
48 (2002) 1849 1855.
[23] H. Fujita, Diffusion in polymerdiluent system, Fortschr.
Hochpolym.-Forsch. 3 (1961) 1 47.
[24] P. Gao, P.E. Fagerness, Diffusion in HPMC gels: I. Determination of drug and water diffusivity by pulsed-field-gradient
spin-echo NMR, Pharm. Res. 12 (1995) 955 964.
[25] J.S. Vrentas, J.L. Duda, Diffusion in polymersolvent systems:
I. Reexamination of the free-volume theory, J. Polym. Sci.,
Polym. Phys. Ed. 15 (1977) 403 416.
[26] J.S. Vrentas, J.L. Duda, Diffusion in polymersolvent
systems: II. A predictive theory for the dependence of
diffusion coefficients on temperature, concentration, and
molecular weight, J. Polym. Sci., Polym. Phys. Ed. 15
(1977) 417 439.
[27] R.W. Korsmeyer, S.R. Lustig, N.A. Peppas, Solute and
penetrant diffusion in swellable polymers: I. Mathematical
modeling, J. Polym. Sci., Polym. Phys. Ed. 24 (1986)
395 408.
[28] F.A. Radu, M. Bause, P. Knabner, G.W. Lee, W. Friess,
Modeling of drug release from collagen matrices, J. Pharm.
Sci. 91 (2002) 964 972.
[29] H.L. Frisch, Simultaneous nonlinear diffusion of a solvent and
organic penetrant in a polymer, J. Polym. Sci., Polym. Phys.
Ed. 16 (1978) 1651 1664.
[30] B. Narasimhan, N.A. Peppas, Disentanglement and reptation
during dissolution of rubbery polymers, J. Polym. Sci., Polym.
Phys. Ed. 34 (1998) 947 961.
[31] P.I. Lee, N.A. Peppas, Prediction of polymer dissolution in
swellable controlled-release systems, J. Control. Release 6
(1987) 207 215.
[32] H.G. Landau, Heat conduction in a melting solid, Q. Appl.
Math. 8 (1950) 81 94.
[33] J. Crank, Two methods for the numerical solution of movingboundary problems in diffusion and heat flow, Q. J. Mech.
Appl. Math. 10 (1957) 220 231.
[34] P. Costa, J. Manuel, S. Lobo, Modeling and comparison of
dissolution profiles, Eur. J. Pharm. Sci. 13 (2001) 123 133.
[35] S.H. Gehrke, D. Biren, J.J. Hopkins, Evidence for Fickian
water transport in initially glassy poly(2-hydroxyethyl methacrylate), J. Biomater. Sci., Polym. Ed. 6 (1994) 375 390.
[36] R.T.C. Ju, P.P. Nixon, M.V. Patel, Drug release from hydrophilic matrices: I. New :scaling laws for predicting polymer
and drug release based on the polymer disentanglement
concentration and the diffusion layer, J. Pharm. Sci. 84
(1995) 1455 1463.

You might also like