Professional Documents
Culture Documents
Abstract: A mathematical model has been developed for lution process, mass-transfer limitations may exist, espe-
immobilized enzyme-catalyzed kinetic resolution of cially for a more reactive system. If diffusion transfer limi-
racemate in a fixed-bed reactor in which the enzyme-
catalyzed reaction (the irreversible uni–uni competitive
tation occurs in the resolution process, the enzyme effective
Michaelis–Menten kinetics is chosen as an example) was enantioselectivity will be lowered to a certain extent, and
coupled with intraparticle diffusion, external mass trans- will result in a decrease in the optical purity and yield of the
fer, and axial dispersion. The effects of mass-transfer desired product. Matson and Lopez (1990) first described
limitations, competitive inhibition of substrates, deacti-
this phenomenon theoretically for a membrane bioreactor.
vation on the enzyme effective enantioselectivity, and
the optical purity and yield of the desired product are Fixed-bed reactors have traditionally been used for most
examined quantitatively over a wide range of parameters large-scale catalytic reactors because of their efficiency, low
using the orthogonal collocation method. For a first- cost, and ease of construction, operation, and maintenance.
order reaction, an analytical solution is derived from the Efforts have been made to study the reaction kinetics, mass-
mathematical model for slab-, cylindrical-, and spherical-
enzyme supports. Based on the analytical solution for the transfer mechanism, and mathematical model of fixed-bed
steady-state resolution process, a new concise formula- reactors. Most researchers have focused on steady-state op-
tion is presented to predict quantitatively the mass- erations (Bailey and Ollis, 1986; Froment and Bischoff,
transfer limitations on enzyme effective enantioselectiv- 1990, Marrazzo et al., 1975; Smith, 1981). Recently, inter-
ity and optical purity and yield of the desired product for
est in the mathematical modeling of unsteady-state opera-
a continuous steady-state kinetic resolution process in a
fixed-bed reactor. © 2001 John Wiley & Sons, Inc. Biotechnol tion has increased significantly because of its importance in
Bioeng 74: 29–39, 2001. fixed-bed reactor design and process control (Zheng and
Keywords: immobilized enzyme; kinetic resolution; Gu, 1996). Therefore, kinetic resolution of racemate by im-
mass-transfer limitations; fixed-bed reactor; analytical mobilized enzyme in a fixed-bed reactor is a topic for in-
solution; orthogonal collocation method
vestigation.
In terms of design of immobilized enzyme fixed-bed re-
INTRODUCTION actors, three major processes should be considered in order
to achieve maximal efficiency and high-quality production
Enzyme-catalyzed resolution of racemate has stimulated (Indlekofer et al., 1996): (1) transport of the reactant (prod-
much interest with regard to production of drugs and agro- uct) molecules form (to) the bulk phase to (from) the vicin-
chemicals in the last decade. Lipase, a relatively inexpen- ity of the immobilized enzyme, in which axial dispersion,
sive enzyme, has been widely employed as a biocatalyst for external mass transfer, and intraparticle diffusion should be
such reactions. This enzyme is usually immobilized onto involved; (2) stereoselective transformation of the reactant
porous support material to improve its activity and thermo- molecules by chemical reaction catalyzed by immobilized
dynamic stability and to separate easily from the mixture for enzyme; and (3) enzyme deactivation. For the second and
reuse. For the immobilized enzyme-catalyzed kinetic reso- third processes, many fundamental experiments have been
carried out to determine the reaction mechanism, deactiva-
tion rate of enzyme, and enzyme enantioselectivity in batch
Correspondence to: G. H. Xiu and fixed-bed reactors (Garcia et al., 2000; Indlekofer et al.,
* Present address: LSRE Laboratory of Separation and Reaction Engi-
1993; Rakels et al., 1994; Straathof et al., 1992; Tsai et al.,
neering, Faculty of Engineering, University of Porto, Rua Dr. Roberto
Frias s/n, 4200-465 Porto, Portugal. Telephone: 351-22-5081618; fax: 351- 1997). However, quantitative investigation about mass-
22-5081674; e-mail: ghxiu@hotmail.com. transfer limitations and the characteristics of the resolution
Contract grant sponsor: China Postdoctoral Science Foundation. process in a fixed-bed reactor has been rare.
冉 冊 冉 冊
tors (Indlekofer et al., 1996; Matson and Lopez, 1990), it
⭸cj 1 ⭸ ⭸cj t
must also evaluate the impact of mass-transfer restrictions p = Deff p rp − p XE exp − vj (1)
on enzyme effective enantioselectivity for the immobilized ⭸t r ⭸r ⭸r t 1
enzyme-catalyzed kinetic resolution process.
where exp(−t/t1) is the rate of enzyme deactivation, t1 is the
Matson and Lopez (1990) reported a quantitative analysis
time constant of enzyme deactivation, vj is the special re-
taking into account the simultaneous reaction/intraparticle
action rate of enantiomers, and p is the shape factor of the
diffusion processes pertinent to competitively reacting en-
porous support (p ⳱ 2 for sphere, p ⳱ 1 for cylinder, and
antiomers in an immobilized enzyme system for first-order
p ⳱ 0 for slab).
reaction. Indlekofer et al. (1996) numerically investigated
The pore diffusion equation [Eq. (1)] is coupled with the
the effectiveness factors in single catalyst pellets at steady
external bulk concentrations in the stream flowing through
state for competitive Michaelis–Menten kinetics and mass-
the fixed-bed reactor through the following boundary con-
transfer limitations on the optical purity and yield of the
dition:
desired product for a fixed-bed reactor at specified operat-
ing conditions.
In a previous study (Xiu et al., 2000), the mass-transfer Deff 冉 冊
⭸cj
⭸r r =R
= kf 关Cj − 共cj兲r =R兴 (2)
limitations for an immobilized enzyme-catalyzed kinetic
resolution process in a batch reactor was studied. In this The second boundary condition is given by the symmetry
work, however, the objectives are: condition at the center of the porous support:
1. To analyze quantitatively the mass-transfer limitations
(taking into account the axial dispersion, external mass- 冉 冊⭸cj
⭸r r =0
=0 (3)
transfer resistance, and intraparticle diffusion resistance)
on the enzyme effective enantioselectivity and the opti- The mass balance equation for the substrates in the fluid
cal purity and yield of the desired product for the im- phase is:
mobilized enzyme-catalyzed kinetic resolution process
⭸Cj ⭸2Cj u0 ⭸Cj 共1 − b兲 共1 + p兲
in a fixed-bed reactor using a numerical method (for = Dax 2 − − kf 关Cj − 共cj兲r =R兴
Michaelis–Menten kinetics) and analytical solution (for ⭸t ⭸Z b ⭸Z b R
first-order reaction) over a wide range of model param- (4)
eters.
with the Danckwerts boundary conditions at the reactor inlet
2. Based on an analytical solution in an extreme situation,
and outlet:
to present a concise formula for predicting the effects of
mass-transfer limitations, inhibition of substrates, and en-
zyme deactivation on enzyme effective enantioselectiv- Z = 0, Dax 冉 冊 ⭸Cj
⭸Z Z=0
=−
u0
关C − 共Cj兲Z=0兴
b j0
(5)
冉 冊
ity and optical purity and yield of the desired product for
irreversible uni–uni competitive Michaelis–Menten and ⭸Cj
Z = L, =0 (6)
other complex reaction kinetics. ⭸Z Z=L
KmR KmS
The other assumptions employed in this work for the
heterogeneous kinetic resolution process are isothermal, The enantioselectivity or enantiomeric ratio, E, of the
constant effective diffusivity of the substrate species inside enzyme is defined as the ratio of the specificity constants
the porous support, and homogeneously distributed enzyme (Vm/Km) of the enzyme for the R- and S-enantiomer
E=
VmS Ⲑ KmS
(9)
冉 冊
⭸xj
⭸ =0
=0 (12)
VmR Ⲑ KmR
⭸yj 1 ⭸2yj ⭸yj
The number of parameters used in this mathematical = − − 共1 + p兲St关yj − 共xj兲=1兴 (13)
⭸ Pe ⭸2 ⭸
model can be reduced by nondimensionalization. For our
work, the following dimensionless variables and groups are
introduced: = 0, 冉 冊
⭸yj
⭸ =0
= −Pe关1 − 共yj兲=0兴 (14)
xj =
cj
,y =
Cj
Cj0 j Cj0
dimensionless concentrations = 1, 冉 冊
⭸yj
⭸ =1
=0 (15)
Z xj = 0, yj = 0, = 0 (16)
= dimensionless axial distance
L
where:
r
= dimensionless radial distance xj
R Rj = (17)
xR xS
1+ +
u0t R S
= dimensionless contact time
Lb
Owing to the nonlinearity of Rj, an analytical solution
kf R could not be obtained from Eqs. (10)–(17). The orthogonal
Bi = Biot number collocation method is used to transform the actual system of
Deff
Eqs. (10)–(17) into a system of ordinary differential equa-
u0 L Peclet number based on reactor tions of initial-value type. These equations could then be
Pe = integrated into the time domain using Gear’s stiff variable
Daxb length
step integration routine to calculate the changes of exit con-
冑
for Eqs. (10)–(17). In the following sections, we first derive
VmSpXE an analytical solution, and present a more general expres-
S共0兲 = R S in the case of 1 → ⬁ sion for the exit concentration at steady state from Danck-
KmSDeff
werts boundary conditions and a first-order reaction as-
S sumption based on the work of Marrazzo et al. (1975). We
R = Thiele modulus for R-enantiomer then propose a formula. The general formula and its derived
公E equations are helpful in calculating the enzyme effective
S共0兲 enantioselectivity and for predicting the optical purity and
R共0兲 = R in the case of 1 → ⬁ yield of the desired product for an immobilized enzyme-
公E catalyzed kinetic resolution process in a fixed-bed reactor.
Kmj dimensionless Michaelis–Menten
j =
Cj0 constant Analytical Solution for First-Order Reaction in a
Fixed-Bed Reactor
bDeff L
Dr = bed-length parameter For first-order reaction kinetics, Eq. (17) becomes:
PR2u0
Rj = xj (18)
BiDrP共1 − b兲
St = Stanton number This situation is valid only for the enzymatic kinetics with
b
xj Ⰶ j (or, in dimension form, cj Ⰶ Kmj), or for dilute
substrate concentration leading to xj Ⰶ j.
Then, by writing in dimensionless form, Eqs. (1)–(8) be-
For the purpose of deriving an analytical solution from
come:
the equations just given, the following assumptions should
1 ⭸xj 1 ⭸ p ⭸xj
=
Dr ⭸ p ⭸
⭸ 冉 冊
− 2j Rj (10)
be made:
1. 1 → ⬁; that is, the enzyme activity is time-independent.
冑公
yj共0,兲 = 1 (19)
关j共0兲兴4 + 2 Ⲑ Dr2 + 关j共0兲兴2
yj共 → ⬁,兲 = 0 (20) dj = (28)
2
冑公
which were used by Rosen (1952).
Raghavan and Ruthven (1983) pointed out that the dif- 关j共0兲兴4 + 2 Ⲑ Dr2 − 关j共0兲兴2
ference in the Danckwerts and Rosen boundary conditions ej = (29)
2
leads to a significant difference in the solutions only when
both the axial Peclet number, Pe, and the bed-length pa- For expression of j , Dj , and Ej , see Table I.
rameter, Dr, are small. From a practical point of view, it is For high values of Pe (DL → 0), we can disregard all
not possible to make both of them small. Therefore, the terms of the order 1/Pe, and then Eq. (21) becomes:
solution derived based on Eqs. (19) and (20) can represent
兰 exp关− 共1 + p兲Stq 兴sin兵 − 关 +
yj⬁ 1 ⬁
this based on Danckwerts boundary conditions. yj = +
Because both enantiomers are converted according to 2 0
j
冋 冉 冊册
tions: Bi
冉 冑公 冊
yj⬁ = exp − 共1 + p兲St 1 − 共Pe → ⬁兲
Bi + j
a2j + b2j + aj
兰 exp
yj⬁ 1 ⬁ Pe
yj = + − sin (31)
冉 冑公 冊
2 0 2 2
The analytical solution of Eqs. (21)–(29) has a general
a2j + b2j − aj d form that takes into account the first-order reaction from the
 − (21)
2  work of Rasmuson and Neretnieks (1980).
冋冉 冊册
The exit concentrations of substrates at steady state
Pe (yj⬁)⳱1 are equal to the zeroth moment of the unreacted
yj⬁ = exp − ␦j (22)
2 fractions of substrates (Suzuki and Smith, 1971). Fortu-
冑
nately, we can derive a more general expression of (yj⬁)⳱1
␦j =
Pe2
4
+ 共1 + p兲PeSt 1 − 冉Bi
j + Bi 冊 (23)
from the zeroth moment based on the Danckwerts boundary
conditions. The final result is given by:
冋 Pe
册 1 Pe
冉 冊
冉 冊
aj = Pe + 共1 + p兲Stqj (24) 共yj⬁ 兲=1 = exp
4 Pe ␦j 2
+ sinh ␦j + cosh ␦j
4␦j Pe
bj = Pe关 + 共1 + p兲Sthj兴 (25) (32)
BiDj It should be noted that Marrazzo et al. (1975) derived Eq.
qj = 1 − (26)
D2j + E2j (32) for p ⳱ 2 with a dimension form.
p j Dj Ej
1a I1关j共0兲兴 ⬁
共d2j − ej2兲共␥n + dj2 − e2j 兲 + 4dj2e2j
⬁
兺 兺 共␥
4␥ndjej
I0关j共0兲兴 2 + Bi
n=1 共 ␥n + dj2 − ej2兲2 + 4d2j ej2 n=1 n + dj − ej 兲 + 4dj ej
2 2 2 2 2
a
I1[j(0)]/I0[j(0)] ⳱ ⌺⬁n⳱1 2j(0)/(␥n + [j(0)]2) (Xiu and Li, 1999), in which I0 and I1 are the
modified Bessel functions of zero and first orders, and √␥n are the roots of Bessel functions of zero
order.
公
冋 1
−
1
tanhS共0兲 S共0兲 册
冋 册
Eeff = E 共for p = 2兲
tion process can be neglected, based on Eqs. (1)–(8), at 1 1
−
(pseudo-)steady state the mass balance equation for a fixed- tanhR共0兲 R共0兲
bed reactor with ideal plug flow can be written as: (42)
−
dyj
d
= 共1 − b兲
VmjpXEL
Kmju0 冉 冊
exp −
t
t1 j
R
Evidently, from Eq. (42), if j(0) → 0, then Eeff will de-
generate into E.
共for intrinsic reaction in a fixed-bed reactor) For immobilized enzyme-catalyzed kinetic resolution of
(38) racemate in a batch reactor, Xiu et al. (2000) derived the
following expression for Eeff:
Therefore, we have:
dyS RS yS GS
=E =E (39) Eeff = E (43)
dyR RR yR GR
Considering yS0 ⳱ yR0 ⳱ 1 for a racemic mixture, integra- in which the global effectiveness factor Gj is (Goto et al.,
tion of Eq. (39) yields: 1990):
冋 册冋 册
ln yS ln关共1 − 兲共1 − eeS兲兴
= 3 1 1
ln yR ln关共1 − 兲共1 + eeS兲兴 −
j共0兲 tanhj共0兲 j共0兲
=E 共for intrinsic reaction in a fixed-bed reactor)
冋 册冋 册
Gj = 共for p = 2兲
(40a) j共0兲 1 1
− +1
Bi tanhj共0兲 j共0兲
ln yS ln关1 − 共1 + eep兲兴 (44)
=
ln yR ln关1 − 共1 − eep兲兴
=E 共for intrinsic reaction in a fixed-bed reactor) For an immobilized enzyme-catalyzed kinetic resolution
(40b) process in a fixed-bed reactor, based on Eq. (32), we have:
冋
共yj⬁兲=1 = exp − 共1 + p兲St 1 − 冉 Bi
Bi + j 冊册 共Pe → ⬁兲
(47)
Then Eq. (46) could be degenerated into the following con-
cise mathematical expression:
Bi
1−
Bi + S S共Bi + R兲
Eeff = = 共Pe → ⬁兲 (48)
Bi R共Bi + S兲
1−
Bi + R
If p ⳱ 2, Eq. (48) is the formula proposed by Xiu et al.
(2000) [i.e., Eq. (43)].
Furthermore, under the assumption of Bi → ⬁ (absence of
the external mass-transfer resistance), the expression of
(yj⬁)⳱1 and Eeff can be simplified to:
冋
共yj⬁兲=1 = exp − 共1 + p兲 冉 冊
1 − B
B
pDrj 册
共Bi → ⬁ and Pe → ⬁兲 (49)
and:
S
Eeff = 共Bi → ⬁ and Pe → ⬁兲 (50)
R
If p ⳱ 2, Eq. (50) is the formula proposed by Matson and
Lopez (1990) [i.e., Eq. (42)].
The difference between Eq. (46) and its derived equations Figure 3. (a) Mass-transfer limitations on eeP and eeS as a function of the
[Eq. (48) and (50)] should be noted. Eeff is dependent on the racemate conversion, , at steady state for Bi ⳱ 100, Pe ⳱ 100, and
bed-length parameter, Dr, if Eq. (46) is adopted; Dr is a [S(0)]2 ⳱ 1. Solid lines: Eq. (37) (Chen et al., 1982); dashed lines:
major adjustable parameter for deciding the racemate con- Michaelis–Menten kinetics with j ⳱ 1; dotted lines: first-order reaction
version, , as shown in Figure 2. Based on Eq. (48) and Eq. [Eq. (41) with Eq. (46)]. (b) Mass-transfer limitations on eeP and eeS as a
function of the racemate conversion, , at steady state for Bi ⳱ 10, Pe ⳱
(50), Eeff is independent of the bed-length parameter, Dr; in 20, and [S(0)]2 ⳱ 10. Solid lines: Eq. (37) (Chen et al., 1982); dashed
this case, Eeff has the same form for different types of re- lines: Michaelis–Menten kinetics with j ⳱ 1; dotted lines: first-order
actors. reaction [Eq. (41) with Eq. (46)].
Figure 4. Effect of [S(0)]2 on enzyme effective enantioselectivity, Eeff , Figure 6. Effect of Pe on enzyme effective enantioselectivity, Eeff , at Bi
at Bi ⳱ 100, Pe ⳱ 100, and Dr ⳱ 0.6 for E ⳱ 5 and E ⳱ 100. Solid lines: ⳱ 100, Dr ⳱ 1, and [S(0)]2 ⳱ 10 for steady state with E ⳱ 5 and E ⳱
Michaelis–Menten kinetics with j ⳱ 1; dashed lines: first-order reaction 100. Solid lines: Michaelis–Menten kinetics with j ⳱ 1; dashed lines:
[Eq. (46)]. first-order reaction [Eq. (46)].