Você está na página 1de 10

journal of

controlled
ELSEVIER
release
Journal of Controlled Release 38 (1996) 219-228

Liquid-liquid emulsification by rotor/stator homogenization


Yuh-Fun Maa *, Chung Hsu
Department of Pharmaceutical Research and Development, Genenteeh, Inc., 460 Point San Bruno Boulevard,
South San Franciseo, CA 94080, USA

Received 29 March 1995; accepted 15 August 1995

Abstract

Liquid-liquid emulsification is a critical step in the double-emulsion microencapsulation process ( W / O / W or O / W / O ) . This


study focuses on the preparation of the primary emulsion ( W / O or O / W ) using rotor/stator homogenization. These homogenized
emulsions were characterized based on the emulsion droplet size. Emulsion droplet size was mainly determined by the shear
forces generated in the turbulent zone within the gap between the spinning rotor and the stationary stator. It was found that the
size of these droplets decreases with increasing homogenization intensity and duration. The dependence of emulsion droplet size
on viscosity, total volume size, volume ratio of the continuous phase to the dispersed phase, and the rotor/stator design was also
investigated. The scaleability of rotor/stator homogenization is limited by the internal circulation determined by the rotor/stator
assembly. Experiments using a large-scale rotor/stator homogenizer were conducted and the circulation rate was estimated using
a simple flow model. As an alternative approach, an externally circulated flow-through homogenization process was used and
compared to the batch operation. The results derived from this study can serve as a basis for the evaluation of large-scale liquid-
liquid emulsification in a microencapsulation process.

Keywords: Liquid-liquid emulsification; Microencapsulation; Rotor/stator homogenization; Emulsion; Droplet size

1. Introduction two-step emulsification process which uses a water-in-


oil-in-water ( W / O / W ) double emulsion. The major
The generation of fine liquid-liquid emulsions is a difference between the primary and the secondary
common practice in the food and pharmaceutical indus- emulsion lies in the size of the emulsion droplets. The
tries. To be more specific, liquid-liquid emulsification primary emulsion is much finer ( 1/zm or smaller) than
is critical to the multiple-emulsion solvent extraction the secondary emulsion (normally 20-100/~m). Until
technique used for the preparation of multi-phase now, the majority of microencapsulation studies have
microspheres for sustained-release applications [ 1,2]. focused on the secondary emulsification, including our
Encapsulating active agents such as water-soluble pro- previous investigations on the development and the
teins and polypeptides is currently drawing increasing scale-up of the secondary emulsification ( O / W or W/
attention from researchers in this field due to their ther- O) using stirred tank mixers and static mixers [4,5].
apeutic importance [3]. Incorporating these hydro- However, there is only limited information available
philic macromolecules in organic polymer matrices, on the effect of the primary emulsion on the properties
either biodegradable or non-biodegradable, requires a of the microspheres [6]. Also, the primary emulsion
was rarely characterized in most of the studies. This
* Corresponding author. Tel. (415)-225-6349. study, therefore, will address the primary emulsion

0168-3659/96/$15.00 1996 Elsevier Science B.V. All rights reserved


S S D 1 0 1 6 8 - 3 6 5 9 ( 9 5 ) 001 23-9
220 E-F. Maa, C. Hsu / Journal of Controlled Release 38 (1996) 21 ~ 2 2 8

understand the mechanism and optimize the operating


conditions for small-scale and large-scale emulsifica-
tion. To solve the problem of non-uniform mixing com-
monly occurring in large-scale emulsification, an
alternative homogenization system aided by an external
~ stationary
stator circulation was designed and was compared with the
batch system.
~]~4 Shearing
rotor

2. Materials and methods

2.1. Materials
Fig. I. Schematic representation of the configuration of a rotor/stator
homogenizer. Poly(methyl methacrylate) ( P M M A ) was pur-
chased from Aldrich Chemical Company. Its molecular
( W / O and O / W ) , particularly the characterization of weight was determined by the supplier to be 25 000.
emulsion droplet size. Analytical-grade methylene chloride (MeCI2)
Modern emulsions have been prepared on an indus- obtained from Baxter was used as the solvent for the
trial scale by a variety of emulsification equipment preparation of an oil phase. Polyvinyl alcohol (PVA)
based on a similar operating principle, agitation. Rotor/ was used as an emulsifier and was provided by Air
stator homogenization belongs to this category but its Products and Chemicals, Inc. under the trade name of
mechanism and parameter effects lack thorough inves- Airvol 205. Bovine serum albumin (BSA) was used
tigation. The rotor/stator assembly consists of a rotor as a model protein. It was provided in a powder form
of two or more blades and a stator with either vertical of Cohn fraction V and was obtained from Interegen
or slant slots around the wall of the homogenizer cell. Co. The reasons for using PMMA as a model polymer
The rotor is housed concentrically inside the stator are that PMMA and the commonly used biodegradable
(Fig. 1 ). As the rotor rotates, it generates a vacuum to polymer, poly(DL-lactic-co-glycolic acid), had similar
draw the liquid in and out of the assembly, thereby microencapsulation properties and PMMA is very
resulting in circulation. One of the two major forces inexpensive [4,5].
which can reduce the size of the dispersed droplets is
mechanical impingement against the wall due to high
fluid acceleration. Another force is the shear force tak- 2.2. Rotor/stator homogenizer
ing place principally within the gap between the rotor
and the stator. At high rotational speeds, the flow in A Virtishear homogenizer (Virtis, Templest IQ)
this region is highly turbulent and contains eddies of consisting of a digital display microprocessor control,
different scales. Obviously, homogenization intensity an overhead drive, and a homogenizing shaft was used.
(power) and the residence time that emulsion droplets The shaft came in two sizes, 1 and 2 cm. The shaft tip
stay in the shearing field are the primary parameters for is a rotor/stator assembly capable of generating ultra-
controlling emulsion droplet size. Other parameters fine and fine dispersions depending on the shaft size
that might affect the performance of rotor/stator (Fig. 1 ). The I-cm assembly (micro-tip) has a rotor
homogenization are the viscosity of the two liquids, with four blades and a stator with 12 vertical narrow
surfactant, rotor/stator design, volume size, and vol- rectangular slots around the wall. The 2-cm assembly
ume ratio of the two phases. (macro-tip) has two blades and the stator has eight
In this study, we investigated the effect of the oper- slant slits around the wall. The rotational speed can be
ating parameters associated with rotor/stator homog- steadily controlled in the range of 5000 and 25 000 rpm
enization on the size of the emulsion droplets to for both shafts.
Y.-F. Maa, C. Hsu /Journal of Controlled Release 38 (1996) 219-228 221

homogenizer
2.3. Batch homogenization by rotor/stator
I

The schematic representation of the homogenization flow-through


device was described previously [4,5]. The emulsion
vessel is a jacketed glass container with an inner cyl- outlet
inder of 6 cm height (H) X 2.5 cm inner diameter (ID)
and a septum-capped side-port. The inner tube was
filled with 10-15 ml of the continuous phase (PMMA/
MeC12 solution or aqueous BSA solution). The dis- shaft" [[ pump
tip inlet /"'x
persed phase (aqueous BSA solution or PMMA/
MeC12 solution) was injected into the continuous phase
through the side-port with a syringe while the rotor/
stator homogenizer was running. To homogenize an
injection of
emulsion 10 times the original volume of the continu- protein aqueous polymer solution
ous phase ( 100 ml), a scaled-up version of the glass solution reservoir

device ( 10 cm H X 4 cm ID) and the macro-tip rotor/ Fig. 2. Schematic representation for the flow-through dispersion head
homogenization method. A 2-cm homogenization macro-tip is
stator were used. For batches 50 times (500 ml) and
housed within a flow-through dispersion head. The system consists
100 times ( 1000 ml) the original size, a 11 glass beaker of a solution reservoir, the inlet and outlet ports, and a septum capped
(15 cm H 10 cm ID) was used, and the dispersed t e e for the injection of protein solution. The emulsion is circulated
phase was injected from the top of the beaker with a by a peristaltic pump.
long needle extending to the rotor/stator tip. The injec-
tion time was 10-60 s depending upon the volume of with a septum-cap arm was inserted for the injection of
the emulsion. Homogenization duration was timed aqueous protein solution. The homogenized solution
from the end of the injection (time zero). was returned to the reservoir through the outlet port
A small sample of the emulsion (approximately 0.2 and circulated for a pre-determined time period. Emul-
ml) was taken at predetermined time periods (0, 0.5, sion samples of approximately 0.2 ml were taken
1, 2, 3, 4 and 5 min) using a syringe through the side- through the septum-capped port using a syringe. Other
port at a distance of 0.3-0.5 cm from the rotor/stator conditions such as sampling time and total homogeni-
tip. Emulsion samples were diluted 200-fold with meth- zation time were similar to those used in the batch
ylene chloride immediately after homogenization. The homogenization process. The use of a flow-through
emulsion droplet size was determined by a laser-based homogenizer allows a continuous emulsification proc-
particle size analyzer as described below. The validity ess. This continuous process will be compared with the
of this size characterization method will be discussed batch process in terms of large scale emulsification.
later.
2.5. Particle size and emulsion droplet size analysis
2.4. Flow-through homogenization by rotor/stator
A laser-based particle size analyzer (Brinkmann,
A flow-through rotor/stator homogenization was Model 2010) was used to determine the size of the
conducted using the apparatus presented in Fig. 2. In emulsion droplets. The system was capable of detecting
this system the dispersion head (Virtis company) particles in size ranging from 0.5 to 150/zm. The mean
which was sealed with two Viton O-rings houses the diameter of the droplets was calculated from the vol-
tip of a 2-cm homogenization shaft. A Masterflex con- ume-moment average of the particle size distribution,
sole drive peristaltic pump (5-600 rpm) equipped with i.e. (~d4dn) / (~d3dn). Each reported value represents
an easy-load pump head was used. The tubing was also the average of at least two determinations. The proba-
made of Viton (Masterflex, 0.25 inch). After the bility volume density graph was used for identifying
homogenizer was turned on, the peristaltic pump cir- the particle size distribution.
culated polymer solution from the reservoir into the Samples for droplet size determination required dilu-
inlet port of the dispersion head. Prior to the inlet, a tee tion to prevent light scattering interference from other
222 Y.-F. Maa, C. Hsu / Journal of Controlled Release 38 (1996) 219-228

droplets. Surface solidification (due to protein precip- 3. Results and discussion


itation or water extraction) occurred during dilution.
Several experiments were conducted to demonstrate 3.1. Batch homogenization by rotor/stator
that the diluted droplets do not coalesce and that hard-
ening occurs only at the interface. It was found that Background
decreasing the volume ratio between MeC12 (the dilut- Intense agitation creates turbulent flow consisting of
ing solvent for the W / O emulsion) and the emulsion turbulent eddies with a spectrum of eddy sizes. Hydro-
from 200:1 to 40:1 does not affect the droplet size. The dynamic forces, mainly shear forces, arise through the
droplet size did not change when acetone, a water- motion of turbulent eddies [7]. Turbulent eddies
miscible solvent which was supposed to facilitate water smaller than the size of the dispersed liquid drops are
extraction, was used as the dilution solvent. In addition, most effective in exerting shear forces on the surface
the droplet size did not change significantly with dilu- of these liquid drops [7,8]. As the surface shear
tion time. These observations suggest that emulsion exceeds the cohesive forces of the drops, they fragment
droplets were stable and that there was no coalescence to smaller drops. Fragmentation continues until the bal-
occurring. In another experiment, a water-saturated ance between these forces is reached. Our previous
MeCI 2 solution was used for dilution. The resulting studies [4,5 ] using dimensional group analysis have
droplets had the same size as those hardened in the indicated that the droplet size is a function of the exerted
water-free MeC12 solution, suggesting that the droplets shear force, the interfacial tension and viscosity of the
became hardened primarily at the surface. Furthermore, emulsification system in both an agitated vessel and a
a small amount of the emulsified sample was diluted static mixer. The turbulence is expected to be much
and stabilized at least 100-fold in MeC12 containing 2% more intensive in the rotor/stator homogenizer than in
Tween 85 (emulsion stabilizer) prior to size analysis. an agitated vessel or a static mixer because emulsifi-
Particle sizes measured before and after dilution in the cation in the rotor/stator homogenizer is achieved by
presence of Tween were essentially identical, suggest- rapid flow circulation and high shear force within the
ing that the hardened droplets do not coalesce. For the narrow gap between the rotor and the stator (less than
O / W system, the diluent was pure water. 0.5 mm). Factors that could influence the circulation
rate and shear force are the design of the rotor/stator
assembly, the rotational speed of the rotor, the total
2.6. Viscosity determination volume and the viscosity of the emulsion. These para-
meters were investigated in this study.

The viscosities of both the continuous and dispersed


Homogenization intensity effect
phases (/~o and/z,~) were determined using Cannon-
Fig. 3 shows the time course of emulsion droplet size
Fenske capillary viscometers at 25C. Equations
(EDS) decreases during homogenization at 5000 to
describing the relationship between viscosity and con-
24 000 rpm using either a micro-tip or a macro-tip. In
centration for PMMA/MeC12 solution and aqueous each case, the emulsion consists of 10 ml of 0.4 g/ml
polyvinyl alcohol (PVA) solution (the continuous PMMA/MeC12 solution as the continuous phase and 2
phase used in the O / W emulsion) at 25C have been ml of 0.2 g/ml aqueous BSA solution as the dispersed
reported previously [4]. phase. As seen in Fig. 3, all time-course profiles show
an initial size reduction period (transition) and a lev-
2.7. Interfacial tension determination eling-off period (equilibrium). The transition period
for the micro tip system is prominent at low homoge-
nization intensity (5000 rpm) and narrows down to be
The interfacial tension between the aqueous and barely noticeable at 24 000 rpm. This also holds true
organic phases was determined by a Surface Tensiomat for the macro tip system.
(Fisher, Model 21) equipped with a 6 cm circumfer- When the relationship between the equilibrium
ence platinum-iridium ring. Each data point represents emulsion droplet size (from Fig. 3) and homogeniza-
the average of three measurements. tion intensity for both the macro-tip and micro-tip sys-
Y.-F. Maa, C. Hsu / Journal of Controlled Release 38 (1996) 219-228 223

O 5K rpm (micro-tip)
20 . . . . . . . . . . . . I ' ' ' I"1 10K rpm (micro-tip)
A 15K rpm (micro-tip)
24K rpm (micro-tip)
~, 5K rpm (macro-tip)
" 10K tpm (macro-tip)
15 ~ I5K rpm (macro-tip)

o
0 1 3 2 4 5 6
Homogenization time(min)
Fig. 3. Profiles of the homogenization time vs. the emulsion droplet size for the emulsion of a 0.4 g/ml PMMA/MeC12 solution and a 0.2 g/ml
aqueous BSA solution at a volume ratio of 10:2 (ml:ml) homogenized by a micro and a macro-tip at different homogenization intensity from
5000 rpm to 24 000 rpm.

tems is plotted, the equilibrium emulsion droplet size Curve fitting in the form of Dp = a (homogenization
decreases with respect to the increase in homogeniza- intensity) *' gives b = - 1.72 for the micro-tip system
tion intensity by an exponential decay function to its ( R = 0 . 9 7 8 ) and b = - 1.14 for the macro-tip system
minimum. W e define this minimum droplet size as the ( R = 0 . 9 5 0 ) . In a previous study [4] on the effect of
ultimate equilibrium emulsion droplet size ( U E E D S ) . the system parameters on the microsphere size using a
The occurrence of UEEDS probably can be explained baffled, mixed tank, b was found to be - 1.12 for a
from the hydrodynamic point view. Turbulent eddies double 6-blade flat Rashton impeller system. Interest-
are required to break the dispersed phase into droplets. ingly, the b value for the macro-tip system is essentially
The scale of the turbulent eddies decreases with the identical to the b value for the tank system. This sug-
increase of homogenization intensity until a critical gests that a macro-tip system, although different from
point where the eddies are so small that their energy is the mixed tank system in homogenization mechanism,
lost as heat through viscous dissipation [8]. At a is able to produce the same homogenization effect as
homogenization speed of 15 000 rpm in the micro-tip the tank system.
system and l0 000 rpm in the macro-tip system, the
turbulent eddies reach this critical scale. Consequently, Viscosity effect
homogenization at even higher speeds cannot generate Fig. 4 shows the effect of homogenization duration
turbulent eddies of a size smaller than the critical scale on emulsion droplet size at different P M M A and BSA
to further fragment the droplets. It is also interesting to concentrations with an organic to aqueous volume ratio
note that at the same angular velocity, the linear veloc- of 10:2 homogenized at 5000 rpm using the micro-tip
ity of the homogenization tip in the macro system is system. The viscosity of P M M A / M e C I 2 solution (the
twice as fast as that in the micro system. The macro-tip continuous phase) at 0.2, 0.4 and 0.6 g / m l was deter-
system reached the UEEDS at 10 000 rpm but the mined to be 9.2, 44 and 313 cP, respectively. For aque-
micro-tip system achieved the UEEDS at 15 000 rpm ous B S A solutions (the dispersed phase) at 0.2, 0.3
instead of 20 000 rpm perhaps due in part to the differ- and 0.4 g / m l , their viscosities are 4.3, 40 and 410 cP,
ence in rotor design (2-blade rotor for the micro-tip vs. respectively. The emulsion size decreases with time for
the 4-blade rotor for the micro-tip). each condition studied. A more viscous continuous
224 Y.-F. Maa, C. Hsu / Journal of Controlled Release 38 (1996) 219-228

35 - -0 - - PMMA/BSA (0.2/0.2 g/mL)


- PMMA/BSA (0.6/0.2 g/mL)
-- ~.- - PMMA/BSA (0.4/0.2 g/mL)
30 '%
- -X- -PMMA/BSA (0.4/0.4 g/mL)
':~' " . 2 0 - - -]-- - P M M A / B S A (0.6/0.4 g / m L )
%
25 %

'~ 20
o
-, 15
o "0- ~0- -- -- - - 0 m --0

fi 10

. , . ,+ . . . . . . . + . . . . . . . + . . . . . . . + . . . . . . . +

, , , , I , . , , I , , , , I , , , , I . . . . I , , , ,

0 1 2 3 4 5 6
Homogenization time(min)
Fig. 4. Profiles of homogenization time vs. the emulsion droplet size for emulsions homogenized at 5000 rpm by a micro-tip at an organic:aqueous
volume ratio of 10:2. Emulsions used are PMMA/MeCI2 and aqueous BSA solutions at different concentrations (PMMA in g/ml; BSA in g/
ml): 0.2; 0.2, 0.4; 0.2, 0.6; 0.2 0.4; 0.4, and 0.6; 0.4.

phase has a slightly shorter transition period and was found to be proportional to / . L d i s p1e r s e d and to O.I

-- 0 . 4 3
smaller equilibrium emulsion droplet size than a less /Xconti . . . . . . Our previous liquid-liquid emulsification
viscous continuous phase. study [4,5] for preparing microspheres concluded that
0.22 -- 0 . 4 2
The relationship between the equilibrium emulsion microsphere size is proportional to /.Ldispersed~LLcont i. . . . .

droplet size (EDS) (data obtained from Fig. 4) and in a baffled, mixed tank, but proportional to
0.004 O.56
the viscosity of the two phases is shown in Fig. 5. Based /XdisperseJXco,,i, .... in a static mixer These emulsifica-
on the curve fitting in Fig. 5, the emulsion droplet size tion methods showed the same trend of EDS decreasing

- - O - - Dispersion phase (PMMA/MeCI) I


25 phase (aqueous BSA)
[] - Dispersed
I
I []
20

"\=.~ J I

15
f ..... y = 61.803 * x ^ ( - 0 . 4 2 9 1 4 ) R= 0 . 9 9 6 9 5
g LO - y = 10.804 * x ^ ( 0 . 1 0 5 5 1 ) R = 0 . 9 8 0 0 7

= 10

'~ 5
5
, i . , I , , , , I . , , . I , i , , I , , . .

0 100 200 300 400 500


Viscosity (cps)
Fig. 5. Curve fittings for the emulsion droplet size (from Fig. 8) as a function of solution viscosity for continuous phase (PMMA/MeCI2) and
the dispersed phase (aqueous BSA solution).
E-F. Maa, (2. Hsu /Journal of Controlled Release 38 (1996) 219-228 225

24

I
22 10/1 (organic:aqueousvolumeratio) I
k [] 10/2 (organic:aqueousvolumeratio)

20
o)
I
~- 18

14

12 , . . , I , , . . I , ~ , . I . , . , I . . . . I . . . .

0 l 2 3 4 5
Time(min)
Fig. 6. Profiles of the homogenization time vs. the emulsion droplet size for emulsions of a 0.4 g/ml PMMA/MeCI, solution and a 0.2 g/ml
BSA aqueous solution homogenized at 5000 rpm by a micro-tip at an organic to aqueous volume ratio of 10/1, 10/2 and 10/5.

with increasing viscosity of the continuous phase and The effect of ]-Ldispersed o n the emulsion droplet size
with decreasing viscosity of the dispersed phase. The appears straightforward. The viscoelastic force of the
absolute value of the coefficient for/Xco,1,....... in all of dispersed phase is another force that can resist droplet
the systems is higher than that for/Xdi~per~eJ, indicating fragmentation. Solution viscosity is a direct measure of
that the viscosity of the continuous phase has a more the fluid viscoelastic force and the transfer of momen-
dominating effect than the viscosity of the dispersed tum during fluid flow. Causing a viscous solution to
phase. Interestingly, the coefficient for/x~o,,, ...... in the flow requires higher energy and so does fragmenting
micro-tip system is essentially identical to the coeffi- viscous liquid drops. Thus, droplet size increases as the
cient for/Z~o,,~...... in the baffled, mixed tank system, viscosity of the dispersed phase increases. There is no
suggesting the viscosity of the continuous phase might clear similarity among the micro-tip homogenizer,
play a similar role in affecting the EDS in these two mixed tank, and static mixer in the extent of the effect
systems. Intuitively, dispersing a more viscous solution of the dispersed phase viscosity on EDS. However,
under the same agitation conditions requires higher ]-dispersed consistently has less of an effect on the emul-
input power. This viscosity effect can be explained sion droplet size than/Xco,t~.....
from the hydrodynamic point of view. The turbulent
shear stress (~.~t>) carried by eddies on the surface of
the dispersed phase was expressed by Bird et al. [9] Emulsion volume and volume ratio effects
as:
The effect of the volume ratio between the continu-
ous phase (PMMA/MeC12 solution) and the dispersed
"r~'~ = #~'>dv/dl ( 1) phase (aqueous BSA solution) on EDS is shown in
Fig. 6. Emulsions of three different volume ratios, 10:1,
where/x C'~ is the eddy viscosity and dv/dl, i.e. shear 10:2 and 10:5, were prepared by homogenization at
rate, represents the velocity change over the turbulence 5000 rpm using a micro-tip assembly. For comparison,
of scale 1./Z '~ is a complex function of several molec- injection lasted about 10 s regardless of the actual vol-
ular parameters including a linear relationship with the ume of the BSA solution. As seen in Fig. 6, an identical
viscosity of the continuous phase. This means that the equilibrium emulsion droplet size (approximately 13
shear stress increases with increasing viscosity of the /zm) was obtained in all three cases. The transition
continuous phase, resulting in smaller emulsion period of the time course profile, however, increased
droplets. with decreasing volume ratio. This trend is consistent
226 Y.-F. Maa, C. Hsu / Journal of Controlled Release 38 (1996) 219 228

4 ' ' " ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' '

< O 10/2 (organic:aqueous) in m L , injection tO sec I


k I [] 100/20 (organic:aqueous) in mL, injection 3 0 sec

=3t B
I
O
X
5 0 0 / 1 0 0 (organic:aqueous) in m L , injection 3 0 sec
1 0 0 0 / 1 0 0 (organic:aqueous) in m L , injection 6 0 sec
I
c

~ 1

0 , , , , I , , , , I , , , , I , , , , I

0 1 2 3 4
Homogenization time(min)
Fig. 7. Profiles of the homogenization time vs. the emulsion droplet size for emulsions of a 0.4 g/ml PMMA/MeCI2 solution and a 0.2 g/ml
aqueous BSA solution homogenized at 15 000 rpm by a macro-tip at an organic to aqueous volume ratio of 10/2 (ml:ml) with an injection
duration of 10 s, 100/20 with an injection duration of 30 s, 500/100 with an injection duration of 30 s and 1000/100 with an injection duration
of 60 s.

with that found in preparation of the P M M A micros- Exact solutions of this fluid dynamic problem are com-
pheres using a P V A solution as the continuous phase plex and difficult to solve. Nevertheless, the difference
(data not shown). in circulation rates between the micro-tip and macro-
The effect of emulsion volume on EDS was inves- tip assemblies can be estimated. The total area of the
tigated for the macro-tip system by increasing the total opening (A) and the radial velocity (L,) at the opening
emulsion volume from 12 ml (10 ml of 0.4 g / m l as a result of the pressure difference ( A p ) within the
P M M A and 2 ml of 0.2 g / m l B S A solution) to 120 ml vortex must be obtained. The Ap can be approximated
(100:20), 600 ml (500:100), and 1100 ml by a simplified system which contains a cylindrical
(1000:100), i.e. 10, 50 and 100 fold increase in the container of radius R rotating at an angular velocity of
continuous phase, respectively. Injection duration var- g2. This is a typical vortex system in which the shape
ied from 10 to 60 s depending on the injection volume of the free surface of the fluid is parabolic (Ref. [9],
of the dispersed phase. The profiles of EDS vs. duration pp. 98-100). The pressure profile can be expressed as:
of homogenization operated at 15 000 rpm are pre-
sented in Fig. 7. The equilibrium emulsion droplet size A p = p - - p o = -- pg( z-- zo) + p12Zr2/2 (2)
is consistently around 1.0-1.5 p,m regardless of the
where p is the pressure within the fluid, z the height
difference in the total emulsion volume. However, the
from the bottom of the container, r the distance from
homogenization time required to yield the EDS (i.e.
the axial line, Po and Zo the pressure and height at the
the transition period) changes depending on the total
lowest point of the free surface, respectively, p the fluid
volume and how long the injection lasts. When the
density, and g the gravity constant. Apparently, Ap
injection duration is added to the zero time point, it
increases as the fluid is closer to the wall. This simpli-
shows a trend that the transition period increases with
fied system differs from the actual system within the
increasing emulsion volume probably because a larger
homogenizing cell (Eq. 2) mainly in two aspects. Our
volume needs a longer time to process.
system generates an upside-down vortex and has open-
Flow circulation model ings around the wall. The radial velocity is not zero in
Flow circulation through the tip of the rotor/stator the actual system.
homogenization system is contributed by the pressure A p in Eq. 2 changes with P, i.e., the ratio of pressure
difference due to the vortex and centrifugal forces. drops for the 2-cm (macro-tip) and 1-cm (micro-tip)
Y.-F. Maa, C. Hsu / Journal of Controlled Release 38 (1996) 219 228 227

7 . . . . I . . . . I . . . . I . . . . I . . . .

FTDI-I (500/100 in mL, 0.6 L/min circulation)


6 FFDH (1000/100 in mL, 0.6 L/rain circulation)
~ ~ . . ~ A ~ FrDtt (1000/100 in mL, 1 L/rain circulation) .
,_.= 5
"3 ~O
4
m_
o
E 3 U
.....

2
~a
1 v~ O O

0 . . , , I , . . , I , . , , I . . . . I , , , ,

0 1 2 3 4 5
Homogenizationtime(min)
Fig. 8. Profiles of the homogenization duration vs. the emulsion droplet size for emulsions of a 0.4 g/ml PMMA/MeCIe solution and a 0.2 g/
ml aqueous BSA solution homogenized at 15 000 rpm by a macro-tip at an organic to aqueous volume ratio of 1000/100 (ml:ml) by a batch
operation, 500/1000 by FI'DH homogenization method circulated at 0.6 l/min, 1000/100 by FTDH homogenization circulated at 0.6 and 1 1/
min.

equals 4:1. If Zip is completely converted to velocity, enization systems when large rotor/stator homogeni-
A L, is proportional to zip0.5 according to the energy zation tips are used.
balance equation [9]. Therefore, the ZIt, ratio between
the macro-tip and the micro-tip is proportional to the r 3.2. Continuous-flow homogenization by rotorZ~tator
ratio, i.e. 2:1. The total opening area is 5 cm 2 for the
stator in the macro-tip system and 1.2 cm 2 for the stator External-circulation-aided homogenization offers an
in the micro-tip system. Since the circulation rate ( Q ) alternative homogenization method (Fig. 2). The con-
is proportional to the product of A and t,, the ratio of tinuous phase ( P M M A solution) in the reservoir was
pumped into a flow-through dispersion head ( F T D H )
( Q . . . . . ) . . . . . and (Qmicro) vortex is approximately 10:1.
which housed a macro-tip assembly for emulsification.
This ratio might be greater if the centrifugal force
Aqueous BSA solution was injected through a tee prior
( C F ) is considered. Assuming another simplified con-
to the inlet port. The emulsion returned to the reservoir
dition where C F is mainly caused by the deflection of
to form a closed-loop. Changes in EDS as a function
the fluid in the radial direction by the legs of the rotor,
of the homogenization duration for emulsions homog-
the resulting circulation rate is the product of the radial
enized by the FTDH method are shown in Fig. 8. All
velocity and the area of the plane generating the radial
the emulsions have the same composition (0.4 g / m l of
velocity. The ratio of ( Q . . . . . )CF and (Qmicro)CF w a s PMMA/MeC12 solution and 0.2 g / m l of aqueous BSA
calculated to be 5:1 for the C F effect alone; therefore, solution). P M M A solution was homogenized with 100
( Q ...... )to,al:(Q,nicro)total is at least 50:1 when both the ml of BSA solution at a pumping rate of 0.6 l / m i n for
vortex and C F factors are combined. A separate circu- the 0.5 and 1 1 batch and 1 l / m i n for the 1 1 batch at
lation experiment indicated that the micro-tip could 15 000 rpm. For comparison, the profile for a 1 I batch
circulate 50 ml of 0.4 g / m l P M M A solution satisfac- homogenization using the macro-tip at 15 000 rpm
torily, but was unable to homogeneously emulsify the (from Fig. 7) is also included in Fig. 8. The equilib-
volume of 100 ml. However, the macro-tip assembly rium emulsion droplet sizes by the FTDH method are
could successfully homogenize 21 of the same solution 2, 3, and 4.6/xm for the 1 1 batch at a pumping rate of
(data not shown). This suggests that our simplified 1 l/min, 0.5 1 batch at 0.6 l/min, and 1 I batch at 0.6 1/
theoretical calculation can explain large-scale h o m o g - min, respectively. These sizes were consistently larger
228 IL-F. Maa, C. Hsu / Journal of Controlled Release 38 (1996) 219-228

than those prepared by the batch method which pro- significant effect on the equilibrium EDS. Homogeniz-
duced droplets of around 1 /zm. The size of emulsion ing emulsions with volumes up to 1 1 using a macro-
droplets prepared by the FTDH method was determined tip assembly was successful. A flow-through dispersion
by the circulation rate and the total emulsion volume. head ( F T D H ) method was evaluated to compare with
Emulsion droplets spent less time in the rotor/stator the batch operation and was shown to be feasible for
gap for emulsions of large volume regardless of the homogenizing large emulsion volumes if a fast external
circulation rate because the emulsification time is deter- circulation is provided. The results derived from this
mined by the ratio between total volume and the gap study can serve as a basis for the evaluation of large-
volume. Since emulsion droplets are stable within scale liquid-liquid emulsification in a microencapsu-
hours because PMMA is a good emulsifier, the coales- lation process.
cence of droplets in the hold tank during circulation is
not a concern. Also, the droplet size was primarily
determined at the early stage ( mainly in the first 1 min). Acknowledgements
Droplets were hardened due to solvent extraction at the
surface and became difficult to be further fragmented The authors are grateful to Dr. Tom Patapoff and
during the later stage of homogenization. Therefore, Mr. Andrew Walsh for their suggestions during the
smaller droplets were obtained for emulsion of a preparation of this manuscript.
smaller volume at a higher external circulation rate.
Since the macro-tip system was estimated to be able to
circulate the emulsion at a rate greater than 1 1/min References
internally as mentioned in the earlier section, it pro-
duced smaller emulsion droplets in the batch process [1] Y. Ogawa, M. Yamamoto, H. Okada, T. Yashiki and T.
than the FTDH method did using the circulation rate of Shimamoto, A new technique to efficientlyentrap leuprolide
1 l/rain. It is feasible to use the FTDH method for large- acetate into microcapsulesof polylacticacid or copoly(lactic/
glycolic) acid, Chem. Pharm. Bull. 36 (1988) 1095-1103.
scale homogenization but it is efficient only at fast
[2] D. Bodmeier,T. Kissel and E. Traechslin,Factors influencing
circulation rates. the release of peptides and proteins from biodegradable
parenteral depot systems, J. Control. Release 21 (1992) 129-
137.
4. Conclusions [3] W. Sad6e, A revolutionin therapy? Pharm. Res. 3 (1986) 3~5.
[4] Y.F. Maa and C. Hsu, A study on microencapsulationreactor
Process parameters that could affect the emulsion scale-up by dimensionalanalysis,J. Microencapsul.,in press.
[5] Y.F. Maa and C. Hsu, Liquid-liquidemulsificationby static
droplet size during rotor/stator homogenization were mixers for use in microencapsulation,J. Microencapsul., in
investigated. The change of emulsion droplet size press.
(EDS) was found to be kinetic, reducing initially and [6] N. Nihant,C. Schugens,C. Grandfils,R. Jerome and P. Teyssie,
then reaching a constant, equilibrium value. The emul- Polylactide picroparticles prepared by double emulsion/
sion droplet size decreased with increasing homoge- evaporation technique. I. Effect of primary emulsion stability,
Pharm. Res. 11 (1994) 1479-1484.
nization intensity using both the micro- and [7] M.S. Croughan, E.S. Sayre and D.I.C. Wang Viscousreduction
macro-tips. An ultimate equilibrium emulsion droplet of turbulentdamage in animalcells culture,Biotechnol.Bioeng.
size of around 1 /xm was obtained when the homoge- 33 (1987) 862-872.
nization intensity exceeded 10 000 rpm with the macro- [8] S. Lakhotiaand E.T. Papoutsakis,Agitationinducedcell injury
tip system. The EDS decreased with increasing in microcarriercultures.Protectiveeffect of viscosityis agitation
intensity dependent: Experiments and modeling, Biotechnol.
viscosity of the continuous phase and with decreasing Bioeng. 39 (1992) 95-107.
viscosity of the dispersed phase. Changes in the emul- [9] R.B. Bird, W.E. Stewart and E.N. Lightfoot, Transport
sion volume ratio between the two phases showed no Phenomena,John Wiley & Sons, Inc., New York, 1960.

Você também pode gostar