Sei sulla pagina 1di 6

128 The Journal of Supercritical Fluids, 1996, 9, 128-133

Shrinking-Core Leaching Model for Supercritical-


Fluid Extraction
Motonobu Goto,* Bhupesh C. Roy, and Tsutomu Hirose
Department of Applied Chemistry, Kumamoto University, Kumamoto 860, Japan
Received August 18, 1995; accepted in revised form December 28, 1995
Extraction or leaching of a solute from a solid material is a process involving mass transfer in the
solid matrix. When the solute content in the solid material is sufficiently large as compared to the solubil-
ity in fluid phase, the process is similar to that of irreversible desorption. The shrinking-core model was
applied to the modeling of the extraction process. The model including axial dispersion in the extraction
column was solved numerically. Quasi-steady-state solution without axial dispersion was derived, and
the accuracy was discussed in comparison with the numerical solutions. The model calculations gave a
good agreement with the experimental extraction curve reported in literature.
Keywords: supercritical extraction, leaching, shrinking core, modeling, mass transfer
INTRODUCTION
Supercritical-fluid extraction has been applied to
natural materials to separate the desired components from
a solid matrix in several industries such as the food, per-
fumery, pharmaceuticals, and the petroleum industry.
Most of the work has been devoted to experimental stud-
ies where the focus has been on the composition of the
extracts or application to new materials.
The extraction process from natural materials such
as plants, beans, or seeds involves the release of solutes
from the porous or cellular matrix into the solvent. It is
a mass-transfer process. The solutes fixed or trapped in a
matrix by physical or chemical forces must be released
and transferred to the supercritical fluid by dissolving.
Then, the dissolved solutes diffuse through the matrix to
the surface of a particle. Finally, they move across stag-
nant film around a particle to the bulk fluid-phase.
The solutes of interest vary from small to large
molecules, or low to high boiling compounds such as
flavor components, or essential oil to lipids or
carotenoids. Their content in a feed material may be
small (less than 1%) or large (40% for seed oil). Thus,
rate-limiting process for the extraction of these materials
may differ from material to material. The dissolution ki-
netics involved in a model may depend on the ratio of the
solute content to the saturation concentration in the fluid
phase, the bound state of the solute on a matrix, and the
association with coexisting materials.
It is important to model the extraction process when
the extraction operation is to be optimized for commercial
application. Unfortunately, the extraction process for
natural materials is not well understood and is not always
operated at the optimal conditions.
Several attempts have been devoted for the modeling
of the extraction process. 1*2 The extraction of oil from
seeds has been most intensively modeled. Since oil con-
tent of seeds is relatively high and oil consists of similar
compounds, mass-transfer model was often employed.
Ficks diffusion model has been applied.2,3 A simpler
model has been also applied where mass transfer was ex-
pressed by a linear driving-force mode1.4+5 The driving
force for the mass transfer may be the difference of the so-
lute concentration between saturated concentration in fluid
phase and local concentration in pores in the matrix of a
particle. It is evident that natural materials are different
from uniformly porous materials such as adsorbents but
are cellular structures with partial destruction taking place
during the crushing process. Sovova6v7 considered a parti-
cle consisting of crushed cells and uncrushed cells. The
solutes can be easily extracted from the crushed cell,
whereas the extraction from the uncrushed cell is much
more difficult.
Desorption models have been developed where local
dissolving process is considered to be desorption of so-
lutes adsorbed on a matrix. Recasens et al.* developed a
model accounting for local adsorption/desorption kinetics,
intraparticle diffusion, and external mass transfer. Goto et
a1.9 applied the desorption model to the extraction of es-
sential oils where local adsorption equilibrium was as-
sumed.
When there is a sharp boundary within a particle be-
tween extracted part and nonextracted part, the shrinking-
0896-8446/96/0902-0128$7.50/O 0 1996 PRA Press
The Journal of Supercritical Fluids, Vol. 9, No. 2, 1996 Shrinking-Core Leaching Model 129
core leaching model may be useful. The shrinking-core
model has been used in solid-fluid reaction,O adsorp-
tion, and ion exchange. t2 The shrinking-core leaching
model was used for a single particleI and for a packed
bed.14 King and Catchpolet4 derived quasi-steady-state an-
alytical solution assuming negligible axial dispersion in a
bed.
In this paper we derive a shrinking-core leaching
model accounting for intraparticle diffusion, external fluid
film mass transfer, and axial dispersion. The model is
further simplified to derive an analytical solution. The ef-
fect of parameters involved in the model is discussed. The
model was applied to the experimental data for oil extrac-
tion from seeds.
THEORETICAL DEVELOPMENT
Model Including Axial Dispersion. This
model describes the situation of the irreversible desorption
followed by diffusion in the porous solid through the
pores. When mass transfer rate of the solute in the
nonextracted inner part is much slower than that in the
outer part where most of the solute has been extracted, or
the solute concentration is much higher than the solubil-
ity of the solute in the solvent phase, a sharp boundary
may exit between the outer and the inner region. A core
of inner region shrinks with the progress of the extrac-
tion. These situations can be modeled by the shrinking-
core model. An example of these situations is the super-
critical-fluid extraction of vegetable oils.
The following assumptions have been made to de-
rive the fundamental equations. The solvent flows axially
with interstitial velocity, v, through a packed bed in a
cylindrical extractor of height, L. Pure solvent enters the
bed. The process is isothermal. Considering axial disper-
sion, the material balance on the bulk fluid-phase in the
extractor is
+XD a2C l-&3k, dC
at az Laz2 &
---R[C-Ci(R)]. (1)
Time variation of the solid-phase concentration
(average oil concentration in a particle) is equated with the
rate of mass transfer of solute within external film sur-
rounding the particle.
The diffusion in outer region is given by
(2)
Solid phase solute exists within the core, the average
value of solid-phase concentration 4 being given by
(4)
Boundary conditions are given as follows. At the
core boundary, the concentration in the fluid phase is at
its saturation value.
Ci = C,,, at r = r,
(5)
Diffusion flux at the outer surface of a particle is equal to
the mass transfer through external film.
( 1
Dac,
e -& _ = kf[c-cml
r-R
(6)
Initial conditions are given as follows:
rc=R at t=O
(7)
C=O at t=O.
(8)
Danckwerts boundary condition at the inlet of column and
the exit condition are given by
vC-D z=O at z=O
L az
(9)
&=O at z=L.
Following dimensionless groups are defined to derive di-
mensionless formulae of the fundamental equations: x =
C/C,, 9 Xi = CilCs,, v
5 = t-JR, Z = zlL, a = vR21D,L, 0 =
(D,lR2)t, y = qlqo, b = C,,,/g,, Pe = LvlDL, Bi = kfRID,.
With these dimensionless groups, eqs l-3 become
3X 2X
-+a5g=KdZ2
ae
a fi-53Bi[x-xi(l)]
g=3Bib[x-x,(l)]
+& p$ =o,
( 1
Dimensionless boundary and initial conditions are
xi = 1 at 5 = 5,
=Bi[x-xi(l)]
<=I
(11)
(12)
(13)
(14)
(15)
(16)
130 Goto et al. The Journal of Supercritical Fluids, Vol. 9, No. 2, 1996
{,=l at 8=0
x=0 at 8=0
(17)
(18)
x-J-$=0 at Z=O
$=O at Z=l.
(19)
The following equations were derived by rearranging these
equations.
ax dx a a2x l--E 3Bi(x - 1)
---
%+Z=KdZ2 E l-Bi(l-l/t,) (21)
ag,-
bBi( x - 1)
&I - cz[l--Bi(l-lit,)]
(22)
These differential equations coupled with boundary and
initial conditions were solved numerically by Crank
Nicholsons method.
The yield, that is, cumulative amount of extract up
to time 8 is given by
9
xde . (23)
Quasi-Steady-State Solution Without
Axial Dispersion. When axial dispersion is negligi-
ble, a quasi-steady-state solution can be obtained by as-
suming that changes in & during the residence time of
solvent in the column are negligible and that changes in
the axial direction of the column at any given time are
small. By introducing the dimensionless time B* = 8 -
Z/a, integration of eq 2 1 gives
3BiZ
1
l-Bi(l-115,)
(24)
By substitution of x into eq 22 and integration
1
4,. (25)
The symbolic integration was done with a computer alge-
bra system (MathematicaTM, Wolfram Research Inc.),
which gives analytical solution in terms of the exponen-
tial integral function. For example computations, numer-
ical values were obtained directly from the analytical solu-
tion.
When 5, is further assumed to be constant during
integration of eq 22 with eq 24, the folIowing simplified
solution can be obtained:
0.6
x
0.4
TABLE I
Parameters for Example Calculation
0,
L
I;;
&
%
C
sat
a
b
Bi
Pe
1.5 x lO-O m2 s-
70x 10.rn
1 X 1O-3 m s-l
1 X 10e3 m
0.4
250 kg m-3
qdlO0
100 (value calculated with
above parameters is 95)
0.01
100
1000
b = 0.01
Figure 1. Effect of parameter a on the extracted concen-
tration.
3bBXJ*
=l~~~~i(l~~/~,)exp
l--E 3BiZ
a& 1-Bi(l-115,) (26)
I
This solution is identical to that derived by Catchpole.14
The time required for complete equilibrium extrac-
tion is given by eqs 27 and 28 on an assumption of equi-
librium extraction, that is, rate processes do not interfere
with the extraction.
1-&
e, =-
ab&
(27)
(28)
PARAMETRIC ANALYSIS AND ACCURACY
OF SIMPLIFIED MODELS
Values of the Parameters Used in the
Model. The parameter values used in model calculation
were taken from the literature.13 Table I shows the pa-
rameters used in this work.
The Effect of Parameter a. One of the im-
portant operation parameters is a = (v/L)(R2/D,) which is
an inverse of dimensionless residence time. The value of
The Journal of Supercritical Fluids, Vol. 9, No. 2, 1996
0.2
0
b I 0.01
Pe-loo0 -
Bi-100
0 1 2 3 4 5
@B.
Figure 2. Effect of parameter a on the solid-phase con-
centration.
a-10
b- 0.01
Bi-100 -
x
0 0.5 1 1.5 2 2.5
wee
Figure 3. Effect of Peclet number (Pe) on the extracted
concentration.
1
a-70 :
0.8
0.8
X
0.4
0.2
ot- - 1
1
2 e/e,
3 4 5
Shrinking-Core Leaching Model 131
a increases with flow rate and particle size, and decreases
with extractor length and intraparticle diffusivity. Figure
1 shows variations of effluent concentration, x, at the exit
of the extractor as a function of dimensionless time, e/e,.
Figure 2 shows average solid-phase concentration, 7,
which is equivalent to <i, at three locations in the extrac-
tor column. These curves were calculated by the numeri-
cal solution, eqs 21 and 22 at three values of a (=
vR21D&L). Solid-phase concentration, 7, decreases more
rapidly for smaller value of a. The wave of sharp decrease
in jj propagates from inlet to exit of the extractor in case
of small a where extraction proceeds from inlet to outlet
in the extractor. On the other hand, extraction occurs
similarly for entire region in the extractor for larger a.
This behavior is reflected in the effluent concentration at
the exit of the column. Thus, the extraction is more ef-
fective for smaller a, that is, smaller particle size and
larger residence time.
The Effect of Peclet Number. Peclet
number, Pe, is an index of axial dispersion. Figure 3
shows effluent concentration at the exit of the extractor, x
calculated by numerical solution at various Peclet num-
bers. Since the influence of the axial dispersion is ob-
served only when there is a sharp distribution along the
extractor column in flow direction, the results are com-
pared at a = 10. For smaller Peclet number, the exit con-
centration becomes broader due to the axial dispersion.
When Peclet number is larger than 100, the effect of axial
dispersion becomes insignificant. For larger value of a,
axial dispersion becomes less influential.
a-200
- Numerical solution
--------- QSS solution
- - SQSS solution
1;
0.8 :
0.6
a=500
Figure 4. Comparison of models at various a (b = 0.01, Bi = 100).
132 Goto et al. The Journal of Supercritical Fluids, Vol. 9, No. 2, 1996
0.8
0.6
X
0.4
6
CO, flow rate: 0.2 m3/h (S
0.8
0.6
X
0.4
0.2
1 2 3 4 5
1;
0.8
Bi=l
0.6 :
X
0.4
0.2 :
o--- --- .--- -- ---..- -
1
2 O/8, 3
4 5
Figure 5. Comparison of models at various Biot number
(a = 100, b = 0.01).
Accuracy of Quasi-Steady-State Solu-
tions. Figure 4 shows the comparisons of dimension-
less effluent concentration, x at the extractor exit among
the numerical solution, quasi-steady-state (QSS) solution,
eq 24 and simplified quasi-steady-state (SQSS) solution,
eq 26. Comparison is made for four different values of a
in Figures 4 (a), (b), (c), and (d). The SQSS solution
gives always larger value of x. For larger value of a (a >
200), QSS solution agreed well with the numerical solu-
tion for the entire range of dimensionless time, 0/e,.
However, for smaller value of a, the QSS solution devi-
ates from the numerical solution at smaller time. In these
cases accuracy of the QSS solution is worse than the
SQSS solution at smaller time.
The Effect of Biot Number. Relative im-
portance of intraparticle diffusion to external fluid-solid
mass transfer is represented by Biot number, Bi = kfR/D,.
Figure 5 shows the effect of Bi on the accuracy of the so-
lutions. When Bi is small, external mass transfer signifi-
cantly affects to the extraction behavior. Thus, the exit
35 MPa, 316 K
I
- Model (eqs 21-23)
0.2 0.3 0.4 0.5
Gas flow [m3 (STP)]
Figure 6. Comparison of model calculation and experi-
mental extraction curve of oil from rape seeds with carbon
dioxide.
31
I I I I I
0
- I
35 MPa, 316 K
q 0 Experimental data
(Brunner. 19841
~2~+&\~
- Model (eqs 21-22)
%~ 1
0 0.1 0.2 0.3
Gas flow [m3 (STP)]
0.4 0.5
Figure 7. Comparison of model calculation and experi-
mental extraction rate of oil from rape seeds with carbon
dioxide.
concentration at the initial time is smaller than unity for
smaller Bi. The accuracy of quasi-steady-state solutions
increases with decrease in Bi. When intraparticle diffusion
controls the extraction process for large Bi, sudden change
of the concentration at the surface of a particle at the ini-
tial stage of the extraction results in inadequate situation
for quasi-steady-state assumption.
COMPARISON WITH EXPERIMENTAL
DATA
The shrinking-core model was applied to the exper-
imental data. The extraction data of oil from rape seed
with supercritical carbon dioxide measured by Brunner2
was used to compare with the model calculation. Radius
of the crashed rape seed was 0.25 mm and the extractor
was 17 mm dia. and 220 mm long. Figure 6 shows the
extraction curve, that is, cumulative amount extracted ver-
sus carbon dioxide flowed at two operating conditions.
Figure 7 shows the extraction rate, which corresponds to
The Journal of Supercritical Fluids, Vol. 9, No. 2, 1996 Shrinking-Core Leaching Model 133
the differentials of the cumulative extraction curve. All
the parameters necessary for model calculation were ob-
tained from the literature* except for effective intraparticle
diffusivity, D,, and axial dispersion coefficient, DL.
External mass-transfer coefficient estimated by Wakao and
Kagueis correlation15 was adopted as used by Brunner.*
NOTATION
Estimation method of axial dispersion in supercriti-
Cal-fluid extractor is not available in literature. Erkey and
Akgermani6 measured axial dispersion by the chromato-
graphic technique. They obtained particle Peclet number
(=2Rv/DL) in range from 0.058 to 0.244 for naphthalene-
alumina system. When these values are converted to
Peclet number (=Lv/DL), one obtains Pe = 25-107 for the
experimental condition. Thus, we adopted the value of 50
for Peclet number. Axial dispersion did not influence to
the extraction curve in this case.
Bi
c
ci
c
Sal
$
L
Pe
9
?
2
r
r,
t
V
x
xi
Y
L
z
&
E c
Biot number = k,RID,
concentration in bulk fluid-phase, mol mm3
concentration in pores, mol mm3
saturation concentration, mol rnA3
effective intraparticle diffusivity, m2 s-r
axial dispersion coefficient, m* s-r
external mass-transfer coefficient, m s-l
length of extractor, m
Peclet number, LvlD,
solid-phase concentration, mol mm3
average value of q, mol me3
initial solid-phase concentration, mol mm3
particle radius, m
radial coordinate
Intraparticle effective diffusivity is most important
parameter for the extraction process from solid materials.
However, accurate value of the intraparticle effective diffu-
sivity can not be estimated, because the value changes de-
pending on the material and environment. We adopted 1.5
x IO-i0 m* s-i as an initial value for intraparticle effective
diffusivity as given in Table I. By fitting the calculated
results with the experimental data, we optimized the value
to be 1.5 x lo-i0 m* s-i (20.5 MPa, 324.7 K) and 0.75 x
lo-i0 m* s-i (35 MPa, 316 K). Wilke and Lees correla-
tion17 gives the dependence of temperature and pressure on
diffusivity. Diffusivity is proportional to T[K13*lP[bar]
according to the correlation. From this relation the ratio
of diffusivity at 20.5 MPa and 324.7 K to diffusivity at
35 MPa and 316 K is calculated to be 1.8. This value is
in good agreement with the ratio of the optimized intra-
particle effective diffusivity, 1.5 x lo-lo/O.75 x lo-i0 =
2.0.
radius of unleached core, m
time, s
interstitial fluid velocity, m s-r
dimensionless concentration in bulk fluid-phase, C/C,,,
dimensionless concentration in pores, CilC,,
dimensionless solid-phase concentration, q/q0
average value of y
axial coordinate in extractor
bed voidage
dimensionless time, (D,lR*)t
dimensionless radial coordinate, r/R
dimensionless radius of unleached core, r,lR
REFERENCES
(1)
As shown in Figures 6 and 7, the results calculated
by the model (eqs 21-23) agreed well. Diffusion within a
particle was rate-determining step in the extraction pro-
cess. In these cases, parameters, a and b, were as follows:
Akgerman, A.; Madras, G. Supercritical Fluids
Fundamentals for Application; Kiran, E.; Sengers, J.
M. H. L., Eds.; Kluwer Academic Publishers: Dordrecht,
1994; p 669.
Brunner, G. Ber. Bunsenges. Phys. Chem. 1984,88,
887.
Reverchon, E.; Donsi, G.; Osseo, L. S. Ind. Eng.
Chem. Res. 1993,32, 2721.
Lee, A. K. K.; Bulley, N. R.; Fattori, M.; Meisen, A. .I.
Am. Oil Chem. Sot. 1986.63, 921.
Schaeffer, S. T.; Zalkow, L. H.; Teja, A. S. .I. Supercrit.
Fluids 1989, 2, 15.
Sovova, H. Chem. Eng. Sci. 1994,49, 409.
Sovova H.; Kucera, J.; Jez, J. Chem. Eng. Sci. 1994,
49, 415.
Recasens, F.; McCoy, B. J.; Smith, J. M. AIChE .I.
1989, 35, 951.
a = 7.1, b = 0.014 (20.5 MPa, 324.7 K)
Goto, M.; Sato, M.; Hirose, T. J. Chem. Eng. Jpn.
1993, 26, 401.
a = 4.9, b = 0.054 (35 MPa, 316 K)
Since parameter a is relatively small, simplified analytical
solutions (eqs 25 or 26) can not be used accurately as in-
dicated in Figure 4.
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
CONCLUSIONS
Shrinking-core model was developed for the extrac-
tion from solid materials. Numerical solution of the
model equations simulated extraction behavior. Quasi-
steady-state solution gave a good approximation for larger
value of parameter a. However, the QSS solution cannot
be utilized when parameter a is small and Bi is large.
Although example calculation was demonstrated for su-
percritical-fluid extraction, the model can be applied to the
other extraction or leaching processes.
(14)
(15)
(16)
(17)
Levenspiel, 0. Chemical Reaction Engineering; John
Wiley & Sons: New York, 1972; p 361.
Hall, K. R.; Eagleton, L. C.; Acrivos, A.; Vermeulen,
T. Ind. Eng. Chem. Fundam. 1966,5, 212.
Guria, C.; Chanda, M. Trans I Chem. E 1994, 72, 503.
Jones, M. C. Supercritical Fluid Technology -
Reviews in Modern Theory and Applications; Bruno,
T. J.; Ely, J. F., Eds.; CRC Press: Boca Raton, FL,
1991; p 365.
King, M. B.; Catchpole 0. Extraction of Natural
Products Using Near-Critical Solvents; King, M. B.;
Bott, T. R., Eds.; Blackie Academic & Professional:
Glasgow, 1993; p 184.
Wakao, N.; Kaguei, S. Heat and Mass Transfer in
Packed Beds; Gordon and Breach: New York, 1982; p
156.
Erkey, C.; Akgerman, A. AIChE J. 1990,36, 17 15.
Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The
Properties of Gases and Liquids, 4th ed.; McGraw Hill:
New York, 1987; p 587.

Potrebbero piacerti anche