Chiral separation of β -blocker drug (nadolol) by five-zone simulated moving bed chromatography

Chemical Engineering Science 60 (2005) 1337 – 1347 www.elsevier.com/locate/ces

Chiral separation of
-blocker drug (nadolol) by five-zone simulated moving bed chromatography Xin Wanga,∗ , Chi Bun Chinga, b a Chemical and Process Engineering Centre (CPEC), Block E5 Basement 08, National University of Singapore, 4 Engineering Drive 4,

Singapore 117576, Singapore b Division of Chemical and Biomolecular Engineering, College of Engineering, Nanyang Technological University, Singapore 637722

Received 1 December 2003; received in revised form 5 October 2004; accepted 16 October 2004

Abstract Nadolol, a
-blocker drug used in the management of hypertension and angina pectoris, has three chiral centers and is currently marketed as an equal mixture of four stereoisomers. Resolution of three of the four stereoisomers of nadolol was obtained previously by HPLC, with a complete separation of the most active enantiomer (RSR)-nadolol, on a column packed with perphenyl carbamoylated
-cyclodextrin (
-CD) immobilized onto silica gel. Continuous separation of enantiomer (RSR)-nadolol from its racemate (which is a ternary mixture in the chromatographic system of this study) in both 2-raffinate and 2-extract configuration of five-zone SMB was studied. Same experimental setup was applied to both configurations by modifying SMB controlling program accordingly. Separation performances of the five-zone SMB were investigated for both 2-raffinate and 2-extract configurations and same safety factors were applied to investigate the effect of m3 –m2 (or m4 –m3 ) on the separation performance systematically. The desired enantiomer of nadolol can be produced with a high purity and yield in 2-raffinate configuration compared with that in 2-extract configuration. 䉷 2004 Elsevier Ltd. All rights reserved. Keywords: Chiral separation; Nadolol; 5-zone simulated moving bed chromatography; Ternary separation;
-cyclodextrin

1. Introduction Chirality has been a major concern in the pharmaceutical industries. Nowadays more research efforts have been concentrated on the production of optically pure products due to increasing demand that such drugs are administered in optically pure form (Decamp, 1989; Rekoske, 2001). Nadolol, 5-{3-[(1,1-dimethylethyl) amino]-2-hydroxypropoxy}-1,2, 3,4-tetrahydro-cis-2, 3-naphthalenediol is a
-blocker drug widely used in the management of hypertension and angina pectoris. Its chemical structure has three stereogenic centers which allows for eight possible stereoisomers. However, the two hydroxyl substituents on the cyclohexane ring are fixed in the cis-configuration which precludes four ∗ Corresponding author. Tel.: +65 6874 2196; fax: +65 6873 1994.

E-mail address: [email protected], [email protected] (X. Wang). 0009-2509/$ - see front matter 䉷 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2004.10.007

stereoisomers. Nadolol is currently marketed as an equal mixture of four stereoisomers, designated as diastereomers of “racemate A” and “racemate B” (for the molecular structures, refer to Wang and Ching, 2002). Racemate A is a mixture of the most active stereoisomer I ((RSR)-nadolol) and its enantiomer II ((SRS)-nadolol) in 1:1 molar ratio, whereas racemate B is a mixture of stereoisomer III ((RRS)-nadolol) and its enantiomer IV ((SSR)-nadolol) also in 1:1 molar ratio. For a safer and more effective use, it is better to separate the enantiomer (RSR)-nadolol before use. Simulated moving bed (SMB) process has been extensively applied to the separation of chiral drugs and intermediates (Lehoucq and Verheve, 2000; Francotte et al., 1998; Pais et al., 1997; Pedeferri et al., 1999) over the last decade. Due to continuous countercurrent contact between liquid and solid phases, SMB process allows the decrease of desorbent requirement and the improvement of productivity per unit time and unit mass of stationary phase. Furthermore, SMB

1338

X. Wang, C. Bun Ching / Chemical Engineering Science 60 (2005) 1337 – 1347

process is believed to be able to achieve high-purity separation performance even when the resolution exhibited by an individual column is not efficient for a batch preparative separation, which is often encountered in chiral separations. Although having the advantages of low solvent consumption, high-purity products and high productivity, classical four-zone SMB is only capable of separating binary mixtures or dividing multi-component mixtures into strong adsorbing and weak adsorbing fractions. For a ternary separation, the ternary mixture can be separated into their components by operating two SMBs in a row, which can be either separated or combined in a single device. Other concepts of keeping the four zones either by alternating two different adsorbents or having a variation of the working flow rates with respect to time within a switching period were also proposed (Kearney and Hieb, 1992). Wooley et al. investigated a nine-zone system for glucose–xylose–sulfuric acid–acetic acid separations (Wooley et al., 1998). Five-zone SMB with a third fraction withdrawn from the system besides products of extract and raffinate was also discussed (Nicoud, 1999; Beste and Arlt, 2001; Nicolaos et al., 2003). Recently, a single cascade SMB system was investigated for ternary separation, which is especially suitable for system with little amount of the most strongly adsorbed component and a significant amount of the middle component (Kim et al., 2003). A five-zone SMB system with different strength solvents in different zones was also proposed for ternary separation of biomolecules (Abel et al., 2003). For the
-blocker drug nadolol, resolution of three of its four stereoisomers was obtained by HPLC in a previous study (Wang and Ching, 2002). A complete separation of the most active enantiomer (RSR)-nadolol was achieved on a column packed with perphenyl carbamoylated
-cyclodextrin (
-CD) immobilized onto silica gel. In this study, separation performance of the ternary mixture of nadolol by the five-zone SMB process was investigated. The study is consistent with the growing demands of multicomponent multi-fraction separations when the target drugs have more than one chiral centre.

2. Five-zone SMB separation of ternary mixture We consider a ternary mixture of component A, B and C and assume that component C is the least retained one, component B is the middle retained one and component A is the most retained one such that KA > KB > KC . Components C, B and A correspond to the first (component 1), second (component 2) and third eluted component (component 3) in chromatographic separation of the ternary mixture, respectively. For a better understanding, five-zone SMB can be regarded as a modification of conventional four-zone SMB, with a side stream introduced at the point of highest concentration of the middle component B. In particular, the introduction of the side-stream in Section 2 would result in the division of that section. The separator would then have five sections

Solid A B C Zone I

Zone II

Extract 1 (A, D)

Zone III

Extract 2 (A, B, D)

Zone IV

Feed (A, B, C,D)

Zone V

Raffinate (C, D)

Liquid Desorbent Make-up (D) Fig. 1. Liquid and solid streams involved in a ternary TCC process (2-extract configuration).

Solid A B C Zone II

Zone I

Extract (A, D)

Zone III

Feed (A, B, C, D)

Zone IV

Raffinate 2 (B,C,D)

Zone V

Raffinate 1 (C, D)

Liquid Desorbent Make-up (D) Fig. 2. Liquid and solid streams involved in a ternary TCC process (2-raffinate configuration).

in total and two extract streams, with the feed located between Sections 3 and 4 (see Fig. 1). Similarly, positioning the side-stream in the raffinate region would create the separator with two raffinate streams and a feed located between Sections 2 and 3 (see Fig. 2). For classical four-zone SMB processing a multi-component mixture of M compounds, it is possible to define a KEY component as follows: all components going from 1 to KEY are produced in the raffinate and the remaining compounds going from KEY+1 to M are produced in the extract (Mazzotti et al., 1994). Similarly, for a five-zone SMB separating a ternary mixture, the compounds 1–3 can be produced either in raffinate, Extracts 1 and 2 streams for KEY = 1, or in raffinates 1, 2 and extract streams for KEY = 2. The two configurations are designated as 2-raffinate and 2-extract five-zone SMB, respectively and the side stream is called extract 2 and raffinate 2 for the two configurations, respectively. For 2-raffinate configuration SMB, desired separation is achieved if each component migrates to its corresponding product outlet, as indicated by the arrows in Fig. 2. However, it should be noted that, although the amount of component C in raffinate 2 stream can be minimized, some C in this stream is inevitable since component C must pass through that point

X. Wang, C. Bun Ching / Chemical Engineering Science 60 (2005) 1337 – 1347 Table 1 Design criteria of flow rate ratios for the five-zone SMB Zone

5 4 3 2 1

Table 2 Different applications of two approaches in 2-raffinate five-zone TC chromatographic process

Flow rate ratio mj 2-raffinate configuration

2-extract configuration

m 5 < KC K B > m 4 > KC K A > m 3 > KB K A > m 2 > KB m 1 > KA

m 5 < KC K B > m 4 > KC K B > m 3 > KC K A > m2 > KB m 1 > KA

as it migrates towards the raffinate 1 outlet. Thus one cannot produce pure component B in raffinate 2 stream at steady state although theoretically 100% purity of A and C can be obtained in the extract and raffinate 1 stream, respectively. Therefore ideal separation result is to achieve an extract stream with a 100% yield of pure A, raffinate 1 stream with the greatest possible yield of pure C, and raffinate 2 stream with a 100% yield of B and least possible contamination with C. The conclusion is also true for the 2-extract configuration where one cannot produce pure component B in extract 2 stream although theoretically 100% purity of A (but at a lower yield due to losses in the extract 2 stream) and C could be obtained in the extract 1 and raffinate stream, respectively. Thus five-zone SMB process shown in Figs. 1 and 2 cannot be considered by itself if one wants to separate the middle retained component from a ternary mixture. The steady-state behavior of a five-zone true countercurrent (TCC) unit is determined only by zone flow rate ratios. For both 2-extract and 2-raffinate configurations, the system must fulfill certain requirements on the migration directions of each component of the ternary mixture to achieve the desired separation. The desired migration of the three components is shown in Figs. 1 and 2, respectively. As shown in Figs. 1 and 2, for a complete separation in j j j TCC unit, net flow rates (defined asfi = mj ci − qi ) of the strong components are negative so that they follow the solid flow while the net flow rates of the weak components must be positive to follow the fluid flow. These can be summarized to the following two cases regardless of the number of components and zones in the TCC process: The compound i follows the liquid flow: j fi


0 ⇒ mj
Hi (or mj
Ki ).

(1)

The compound i follows the solid flow: j

fi  0 ⇒ mj  Hi (or mj  Ki ).

1339

(2)

By applying these migration direction rules, the design criteria of the flow rate ratios, mj , in the five zones are summarized in Table 1. One of the key issues in operating SMB process is to determine zone flow rates and column switching time. Developed in the frame of equilibrium theory which neglects the effect of axial mixing and mass transfer resistances, safety margin

Definition of equations

Approach 1

Flow rate ratios

mj =

Approach 2

QTCC −QS
P j

QTCC j

vL
= QS vS (1−
) KA < m1 < ∞

mj =

QS (1−
P ) HA < m1 < ∞ HB < m2 < HA

K B < m 2 < KA

Complete HB < m3 < HA separation regions HC < m < HB 4 m5 < HC

K B < m 3 < KA

Retention time

tR = t0

in fixed bed Equivalence of

mj =

TCC and SMB


∗





+

K C < m 4 < KB m 5 < KB

1−
∗




 H

QSMB t ∗ − V
∗ j V (1 −
∗ )


 1−
tR =t0 1+ K




mj =

QSMB t∗ − V
j V (1 −
)

method (Khattabi et al., 2000) and triangle theory (Mazzotti et al., 1997; Storti et al., 1993) are currently widely applied SMB design approaches. In both methods, development of SMB is resort to its corresponding hypothetical true countercurrent (TCC) process and the conversion of TCC operation parameters to SMB unit, using geometric and kinematic equivalence between SMB and TCC process (Ruthven and Ching, 1989; Charton and Nicoud, 1995). Due to complex micropore structure of solid phase in SMB, two different approaches, known as pore diffusion model and solid diffusion model, are commonly applied in batch and countercurrent chromatographic process. It is worth noting that consistent application of these two approaches is important in the design and operation of SMB process. The necessary and sufficient conditions to achieve complete separation of ternary mixture for the two approaches are summarized in Table 2. It should also be mentioned that the discussion in Table 2 is based on 2-raffinate configuration and the results can be easily extended to 2-extract five-zone configuration. 3. Experimental 3.1. Chemicals HPLC-grade methanol was obtained from Fisher Scientific (Leics, UK). Glacial acetic acid and triethylamine were obtained from Merck (Germany). HPLC water was made in the laboratory using a Millipore ultra-pure water system. The racemate mixture of nadolol was purchased from Sigma (St. Louis, MO, USA). All purchased products are used without further purification. Empty column (25 cm×1 cm ID) assembly was purchased from Phenomenex (USA). The columns were packed with perphenyl carbamoylated beta-cyclodextrin bonded onto

1340

X. Wang, C. Bun Ching / Chemical Engineering Science 60 (2005) 1337 – 1347

silica gel using an Alltech pneumatic liquid pump (Alltech, USA) by slurry packing method. The silica gel was supplied by Hypersil (UK) with a particle size of 15–25 m. The eluent (desorbent) used was a binary mixture containing 80% aqueous buffer solution (1% TEAA, pH = 5.5) and 20% methanol. The feed solution was prepared by dissolving racemate of nadolol in the desorbent mixture at concentration of 0.043 mg/ml. The eluent and feed solution were degassed in a model LC 60H ultrasonic bath before running the experiment. 3.2. SMB separation system The SMB separation unit consists of 10 columns (25 cm × 1 cm ID) arranged in a 2–2–2–2–2 configuration, i.e., two columns per section. Through minor modifications of the system and controlling software, the same experimental setup was capable of carrying out operations for both configurations of five-zone SMB. For 2-raffinate five-zone SMB process, five external streams are present: the feed mixture to be separated, the desorbent, the extract stream enriched in the most retained species, A, the raffinate 1 stream enriched in the least retained species, C, and the raffinate 2 stream containing a mixture of the two less retained species, B and C. In the laboratory unit used in this work the fluid stream, coming out of the fifth section, is not recycled directly to Section 1, but is collected and recycled offline. This yields the so-called open-loop configuration, which is equivalent to the closed-loop configuration, where the desorbent is directly recycled to Section 1, provided that the stream coming out of Section 5 does not contain any of the components to be separated. The concentrations of the extract and raffinate streams were analyzed using Shimadzu SCL-10AVP chromatographic system. The samples of products were collected at the middle of the switch times at different cycle and switch times. An analytical column (25 cm × 0.46 cm ID) packed by perphenyl carbamoylated
-CD bonded onto 5 m silica gel was used to analyze the concentration of samples based on calibration lines obtained previously from external standard nadolol solutions. The absorbance wavelength was set at 280 nm. 3.3. Flow control system The two inlet streams, i.e., feed and eluent, as well as three of the four outlet streams, e.g., extract and the two raffinates, are controlled by five HPLC pumps and the recycled eluent stream is determined by overall material balance of the SMB unit. In particular, the feed solution is pumped in using a Shimadzu LC-10AT (Tokyo, Japan) pump and the eluent is pumped using a Perkin Elmer series 200 LC pump which mixes the binary mixture of buffer solution and methanol at the desired composition. An online vacuum degasser (SUPELCO) degasses all the liquid being pumped

into the system. The three outlet streams, i.e., the extract and two raffinate, are controlled by three Jasco PU-1587 pumps. For cross-checking of product flowrate, the vessels containing the collected products are weighed on electronic balances (Mettler AE240). The simulation of the counter current movement is obtained by switching six (10+1)-port multi-position valves (Valco Instruments, VICI EMTMA-CE), which are connected to each of the 10 columns. In particular, the feed or eluent solutions are fed to each of the 10 columns via two (10+1)-port multi-position rotary valves, respectively. The extract, two raffinates and the recycled eluent are withdrawn from the columns via four other (10+1)-port multi-position rotary valves. The switching of the rotary valves is controlled by software provided by the manufacturer. Furthermore, the flow direction is determined by 10 check valves (Upchurch company CV-3000, with 1.5 psi cracking pressure and 5000 psi holding pressure) located on the lines connecting each column to the following one. It is worth noting that in order to eliminate cross-contamination the check valve is inserted between the outlet connections of the previous column and the inlet connections of the following one. The setup was also applied to the 2-extract configuration SMB as long as the program, which controlled the switching of rotary valves, was modified accordingly and the extract, raffinates 1 and raffinate 2 streams were changed to extracts 1, 2 and raffinate stream for the 2-extract SMB, respectively.

4. Results and discussions 4.1. Equilibrium and kinetic parameters for ternary separation of nadolol Equilibrium and kinetic parameters were essential for design and modeling of SMB process. These parameters for the separation of nadolol have been evaluated on the perphenyl carbamoylated
-CD bonded silica gel with particle sizes of 15 m (Wang and Ching, 2002). However, more than one column are needed (normally 8–16 columns are used) in practical applications of SMB, thus a great deal of packing materials are required. Besides, due to pressure limit of the whole system, the maximum back pressure of each single column is restricted in order to achieve stable operation and accurate control of products flow-rates. So for economic and operation considerations, usually larger particle size (e.g., in the range of 10–50 m) of silica gel is used in the SMB operation. This is especially economical in the case when only specific application rather than general ones is required. In this study, perphenyl carbamoylated
-CD immobilized onto 15–25 m silica gel were synthesized and packed into 10 semi-preparative columns (25 cm ×1 cm ID). The bed voidage and axial mixing in the columns were determined by pulse chromatographic experiments using nonretained 1,3,5-tri-tert-butyl-benzene (TTBB) as the tracer. The average bed voidage and axial dispersion coefficients

X. Wang, C. Bun Ching / Chemical Engineering Science 60 (2005) 1337 – 1347

1341

Table 3 Definition of process performance parameters for 2 raffinate 5-zone SMB in complete separation region Performance parameter

Definition

Expressed by operation parameters

Extract Recovery (%)

Enrichment (%)

Productivity (g/h/l of solid) Desorbent requirement

Raffinate 2

Raffinate 1

Extract

Raffinate 2

Raffinate 1

EQ 100cA E FQ cA F

R2 Q 100cB R2 FQ cB F

R1 Q 100cC R1 FQ cC F

100

100

NA

E 100cA

R2 100cB

R1 100cC

m 3 − m2 m 1 − m2

m 3 − m2 m 3 − m4

NA

R1 Q cC R1 (1−
∗ )V N C

F (m −m ) CA 3 2 t ∗ NC

F (m − m ) CB 3 2

F (m − m ) CB 3 2

B cA

F cB

EQ cA E

(1 −
∗ )V N

F cC

R2 Q cB R2

C

(1 −
∗ )V N

C

(QD + QF )cD QF cTF

were found to be 0.47 and 0.0060 V, respectively. The adsorption isotherm was evaluated on an analytical column (25 cm × 0.46 cm ID) packed with the same CSP and the equilibrium constants were found to be 5.34, 6.80 and 11.20 for (SRS)- and (SSR)-nadolol (considered as one component), (RRS)-nadolol and (RSR)-nadolol, respectively. They correspond to the first, second and third peaks of the elution chromatogram and are represented by components 1, 2 and 3 or components C, B and A, respectively throughout the paper. Although it has been claimed that adsorption isotherms is independent of the particle size of adsorbent (Charton and Nicoud, 1995), our experiment results show that adsorption isotherms obtained on the 15–25 m silica gel are different from those obtained on the 15 m silica gel. This may be because that experimental conditions to produce these different size particles may not be identical, thus the surface chemistry may not be same and hence the isotherm (The difference is expected to be less noticeable if small particles are obtained by rushing and sieving of larger particles). As the consequence of different particle size distribution, chromatogram peaks are more diffused due to mass transfer and axialdispersion effect compared with those obtained from the same CSPs bonded onto 15 m regular silica gel. For design and operation of the five-zone SMB process, one can apply either design approach consistently and rely on the equivalence between SMB and TCC process. Thus proper values of m1 –m5 can be selected, and attempts should be made to increase production rate and enrichment, decrease desorbent consumption and at the same time maintaining the robustness of operation. The selection of the switching time is the result of a compromise: short switching times give a higher production rate and a shorter startup time (the time required to reach steady-state) but they also lead to higher flow-rates. Therefore, there is a lower limit for t ∗ , set according to the maximum pressure (or the maximum flow-rate) allowed. It should be mentioned that either t ∗ , or equivalently Q1 , the largest flow rate in the unit, should be determined by taking into account the upper limit of operation pressure in the unit. Finally, having de-

t ∗N


cD cTF

C

1+

m 1 − m5 m 3 − m2



t ∗N

C

−

R2 QR2 CC

(1 −
∗ )V N C

cided mi (i =1, . . . , 5) and t ∗ (or Q1 ), SMB and TCC equivalence equation in Table 2 is used to determine the liquid flow rate in the five sections of SMB and thus the inlet and outlet streams flow rates. The advantage of this approach is that the flow rate ratio is a dimensionless group bringing together information about column volume, V, unit flow rates, Qi , and switch time, t ∗ , and thus can be applied whatever the configuration, size and productivity of the SMB unit in both linear and non-linear systems. 4.2. Operation of 2-raffinate five-zone SMB Since component A of stereoisomer (RSR)-nadolol is the most potent and desirable enantiomer, the main purpose of the SMB separation in this study is to produce an extract stream with the highest purity and yield of this component. At the same time, separation performance such as purity, yield and productivity of component B and C were also examined in different operation conditions (the definitions of process performance parameters for 2-raffinate five-zone SMB are given in Table 3). One should bear in mind that 100% purity of component B cannot be produced in the side stream (whether it is raffinate 2 or extract 2 stream) due to the contamination with component C or A in the 2-raffinate or 2-extract configuration, respectively. Because 2-raffinate configuration could produce a 100% yield of pure A while 2-extract configuration is possible to produce 100% yield of pure C, we firstly study the separation performance of the five-zone SMB on the 2-raffinate configuration. Graphical representations of the separation conditions are shown in Fig. 3, based on design criteria summarized in Table 1. Since the 2-raffinate configuration has three separation sections (Sections 2–4), two diagrams are necessary to describe the operating conditions. Fig. 3(a) represents the conditions for Sections 2 and 3, which straddle the feed stream. This diagram is similar to that of a four-zone SMB without side-stream. Because of the introduction of feed stream, m3 is greater than m2 and therefore all the possible operating states must lie above the diagonal line (m3 = m2 ).

1342

X. Wang, C. Bun Ching / Chemical Engineering Science 60 (2005) 1337 – 1347

12.0 G

O

11.0

A B

C

10.0

D E F

m3

9.0

8.0

7.0

6.0

5.0 5.0

6.0

7.0

8.0

9.0

10.0

11.0

12.0

9.0

10.0

11.0

12.0

m2

(a) 12.0 G O 11.0

B C

A

D E F

10.0

m3

9.0

8.0

7.0

6.0

5.0 5.0

(b)

6.0

7.0

8.0 m4

Fig. 3. Operating diagrams for 5-zone SMBs with 2 raffinates (a) zone (m3 –m2 ) around feed; (b) zone (m3 –m4 ) around the second raffinate.

Similarly, Fig. 3 (b) represents the conditions around raffinate 2 stream. Since the output of raffinate 2 stream, m3 is greater than m4 and all the possible operating states must also lie above the diagonal line (m3 = m4 ). It is obvious that to produce the desired separation, the system must operate simultaneously within the valid operating areas in both diagrams. For the proper design and operation of 5-zone SMB, m1 and m5 are firstly selected to fulfill the inequalities of Table 1 to regenerate adsorbent and eluent in Sections 1 and 5, respectively. Then m2 , m3 and m4 are decided so that the corresponding operating points lie inside the separation regions, which are of triangle and rectangular shapes in the m3 –m2 and m3 –m4 diagrams, respectively. The points should be close to the theoretical optimal point in order to

achieve a high production rate, yet far away from it within the boundaries of the operating area to assure robustness of SMB operation. At the SMBs theoretical optimum operating state, indicated as point “O” in Fig. 3(a) and (b), the unit has the highest possible productivity and enrichment of products and the lowest desorbent consumption. This is because at this point the value of m3 –m2 and m3 –m4 has the highest and lowest values respectively within the operating regions. However, as mentioned early, the SMB performance at this condition is very sensitive to various kinds of disturbances. To avoid this, operation state away from the optimum operating point O should be chosen. Furthermore, instead of choosing m2 , m3 and m4 randomly inside the operation area, certain defined locus of the operating state was usually applied to study the effect of different flow rate ratios on the separation performance systematically (Beste and Arlt, 2001). Point A is located on the dashed line whose distance to the boundaries is a defined percentage of the width of the operating areas in both diagrams. The operating parameters and separation results are shown in Table 4. It was found that high purity of component A in extract product was obtained while purity of component C in raffinate 1 stream is around 94% due to undesired contamination with component B. This means the middle adsorbed component, B, moves in the direction of liquid flow in Section 4 rather than in solid flow as expected. It indicates that m4 could be larger than KB during the operation, suggesting the possible inaccuracies of equilibrium constants obtained (especially for KB due to the fact that the first and second peaks were not baseline separated) as well as less of robustness of the operations. A further analysis of the operating conditions reveals that distances from the dashed line to the boundaries of m4 = KB and m3 = KA in the (m4 , m3 ) diagram is proportional to the width of the corresponding operating region, represented by KB –KC and KA –KB , respectively. On the other hand, distances from the dashed line to the boundaries of m2 = KB and m3 = KA in the (m2 , m3 ) diagram is equal to each other since the widths of the triangle operating region are equal to each other (represented by KA –KB ). Obviously in most cases the equilibrium constants of the ternary mixture are not equally spaced and thus operating conditions is more likely to be disturbed in one direction (i.e., the distance of point A to boundary m4 = KB is only one third of that to boundary m3 = KA ), especially when the operation point is near to the optimal point. In order to guarantee robustness of the separation, proper safety factor were often used to calculate the operating values of flow rate ratios in four-zone SMB (Yun et al., 1997). The concept can be extended to five-zone SMB, and for simplicity the same safety factors can be applied. Accordingly, flow rate ratios can be expressed as follows: m1 = KA , (3) m2 = KB ,

(4)

m3 = KA /,

(5)

X. Wang, C. Bun Ching / Chemical Engineering Science 60 (2005) 1337 – 1347

1343

Table 4 Operating conditions of SMB experiments for 2-raffinate configuration Run

A B C D E F G

Flow rate ratios m1

m2

m3

m4

m5

12.10 12.10 12.10 12.10 12.10 12.10 12.10

7.18 7.12 7.28 7.52 7.76 8.00 6.98

10.82 10.70 10.46 10.12 9.81 9.51 10.91

6.67 6.49 6.35 6.15 5.96 5.78 6.62

4.94 4.94 4.94 4.94 4.94 4.94 4.94

Switch time t ∗ (min)

Flow rates (ml/min) Q1

QF

QR1

QR2

QE

E

R1

R2

18.00 18.00 22.85 18.71 14.72 10.86 18.00

7.54 7.54 6.15 7.51 9.54 12.94 7.54

2.11 2.08 0.80 0.80 0.80 0.80 2.28

1.00 0.90 0.65 0.67 0.72 0.81 0.97

2.41 2.45 1.88 2.22 2.74 3.59 2.49

2.86 2.89 2.21 2.56 3.08 3.95 2.97

99.8 99.9 100 100 100 100 99.4

94.5 99.4 99.4 99.2 99.1 98.8 66.1

38.2 39.2 39.6 38.1 36.3 35.0 34.0

100.0

100.0

Yield (%)

60.0

Ex R1 R2

80.0

Ex R1 R2

80.0 Purity (%)

Product purity (%)

40.0

60.0 40.0 20.0

20.0 0.0 1.0

1.5

2.0

2.5

3.0

0.0 1.0

3.5

1.5

2.0

3.0

3.5

2.5

3.0

4.0

Ex R1 R2

0.012

Ex R1 R2

Productivity (g/hr/l)

Enrichment (%)

60.0

40.0

2.5 m3 - m2

m3 - m2

20.0

0.008

0.004

0.000

0.0 1.0

1.5

2.0

2.5

3.0

3.5

m3 - m2

1.0

1.5

2.0

3.5

m3 - m2

Fig. 4. Effect of difference of m3 and m2 on the separation performance of 2-raffinate 5-zone SMB.

m4 = KB /,

(6)

m5 = KC /.

(7)

From Eqs. (4)–(6) one can obtain m2 m3 = KA KB ,

(8)

m3 = m4 (KA /KB ).

(9)

Eq. (8) corresponds to the portion of hyperbola going through the√theoretical optimal point (KB , KA ) and point √ ( KA KB , KA KB ) in the (m2 , m3 ) diagram, while Eq. (9) corresponds to the portion of straight line with the slope of KA /KB and passing the optimal point (KB , KA ) in the (m4 , m3 ) diagram. Experimental runs were carried out to find possible optimal operation conditions by changing the difference of m3 and m2 (and thus m3 and m4 ) along the locus described by Eqs. (8) and (9). Theoretical studies show that operation performance (i.e., productivity and desorbent requirement) can be improved by increasing the difference of m3 and m2 (note that difference of m3 and m4 also increases during the change) at the sacrifice of robustness of

operation. At condition G, less of operation robustness could be the reason that the separation behavior is easily to be disturbed. This may also explain that the purity of raffinate 1 stream decreased although extract stream purity was satisfactory. Based on the experimental results, point B could be the possible optimal operation condition among the studied conditions. It was noted that the separation requirement of flow rate ratios in Sections 2 and 3 were always met during various experiments, while the requirement in Section 4 was not always satisfied. This may be explained that axial dispersion and mass transfer effect could make the effective separation region narrower in both (m2 , m3 ) and (m4 , m3 ) diagrams in practical operations, among which SMB operation was much easier to be disturbed in the m3 –m4 diagram (fourth region of SMB)because the difference of equilibrium constant KA and KB was obviously larger than that of KB and KC . To study the effect of different operation conditions (m2 , m3 , m4 ) on the SMB performance, points C–F were selected whose feed flow rate were kept constant. In Fig. 4, it was found that the enrichment of extract and raffinates increased significantly when the difference of m3 and m2

X. Wang, C. Bun Ching / Chemical Engineering Science 60 (2005) 1337 – 1347 12.0

11.0

10.0

9.0

m3

was increased. In other words, solvent consumption was decreased. Productivity and yield of products increased slightly with increasing of the difference of m3 and m2 while purity of the streams did not change significantly. It was found from these runs that the desired component A was produced in the extract stream with a high purity and yield. Low purity of component B was obtained in the raffinate 2 stream due to the inevitable contamination with component C in this stream although its yield was high which suggested that almost all component B migrated in the liquid flow direction in Section 4 and discharged all the way from raffinate 2 stream. As for component C, it was obtained in high purity in raffinate 1 stream, but its yield decreased significantly compared with components A and B due to the discharge of component C in raffinate 2 stream with component B together. On the other hand, robustness of operation can be improved by decreasing the difference of m3 and m2 (or decreasing the difference of m3 and m4 ) when the operation condition was changed from point C to point F. Since feed flow rates were kept constant during the course of changing flow rate ratios, decreasing the difference of m3 and m2 could make shorter switch time and thus increase the eluent flow rate in the whole system, especially in the first section which has the highest flow rate in the SMB unit. As a consequence, the pressure of system increased continuously until stable operation was difficult to be maintained, which prevented good separation results to be achieved. This suggests that although decreasing the difference of m3 and m2 could improve the robustness of operation (at the expense of sacrificing operation performance), further decreasing of it could make SMB operation unstable and frustrate achieving good separation results.

8.0

7.0 I J

6.0 H

K

Q

5.0 5.0

6.0

7.0

8.0

9.0

10.0

11.0

12.0

m2

(a) 7.0 6.8 6.6 6.4 6.2

m3

1344

6.0 5.8

K J

5.6

I H

5.4

Q

5.2 5.0 5.0

4.3. Operation of 2-extract five-zone SMB In this part, separation performance of the five-zone SMB on the 2-extract configuration was examined. The design criteria for the five-zone 2-extract SMB was similar to that of the 2-raffinate configuration (as shown in Table 1), with the exception that in Section 3 KB > m3 > KC is required rather than KA > m3 > KB . Like the 2-raffinate SMB, two diagrams are needed to describe the operating conditions, which represent the conditions for Sections 3 and 4 (which straddles the feed stream) and Sections 2 and 3 (around the extract 2 stream), respectively. It is worth noting that m3 < m2 and m3 < m4 due to the output of extract 2 stream and introduction of feed stream, respectively. Thus all the possible operating states in Sections 2 and 3 must lie below the diagonal line (m3 = m2 ), as shown in Fig. 5(a). Similarly, operating states in Sections 3 and 4 must also lie below the diagonal line (m3 = m4 ), as shown in Fig. 5(b). Again, the system must operate simultaneously within the valid operating areas in both diagrams to obtain the desired separation.

(b)

5.5

6.0

6.5

7.0

m4

Fig. 5. Operating diagrams for 5-zone SMBs with 2 extracts (a) zone around the feed (m3 –m2 ); (b) zone around the second raffinate (m3 –m4 ).

At point Q in Fig. 5(a) and (b), the difference of m2 –m3 and m4 –m3 has the lowest and highest value compared with any other possible conditions within the operating region, giving the possibility that the unit has the highest productivity and product enrichment and the lowest desorbent consumption at this state. However, like that of 2-raffinate SMB, this theoretical optimal state is lack of robustness. Applied the same safety factor to the 2-extract SMB, flow rate ratios of m1 , m2 , m4 and m5 have the same form as those of Eqs. (3), (4), (6) and (7), respectively, with the exception of m3 which is now expressed as m 3 = K C .

(10)

Accordingly, following equations can be obtained to describe the locus of operation conditions, which limits the

X. Wang, C. Bun Ching / Chemical Engineering Science 60 (2005) 1337 – 1347

1345

Table 5 Operating conditions of SMB experiments for 2-extract configuration Run

H I J K

Flow rate ratios m1

m2

m3

m4

m5

12.10 12.10 12.10 12.10

7.06 7.17 7.28 7.40

5.54 5.63 5.72 5.81

6.55 6.45 6.35 6.25

4.94 4.94 4.94 4.94

Switch time t ∗ (min)

Flow rates (ml/min)

Product purity (%)

Q1

QF

QE1

QE2

QR

E1

E2

R

23.24 18.88 14.53 10.13

5.84 7.19 9.34 13.41

0.25 0.25 0.25 0.25

2.27 2.73 3.47 4.85

0.68 0.85 1.13 1.64

0.72 0.83 1.01 1.35

99.9 99.7 99.8 99.7

52.7 53.1 52.4 48.1

99.6 99.7 99.8 99.7

100.0 100.0

Ex 1

80.0

Ex 2

60.0

Yield (%)

Purity (%)

80.0

Ex 1 Raf

40.0

Ex 2 Raf

60.0 40.0 20.0

20.0

0.0

0.0 0.0

0.5

0.0

1.0

0.5

1.0

m4 - m 3

m4 - m3 35.0 30.0

Ex 1

25.0 20.0

Raf

Productivity (g/hr/l)

Enrichment (%)

40.0 Ex 2

15.0 10.0 5.0 0.0

0.005 Ex 1

0.004

Ex 2

0.003

Raf

0.002 0.001 0.000

0.0

0.5

1.0

m4 - m3

0.0

0.5

1.0

m4 - m3

Fig. 6. Effect of difference of m4 and m3 on the separation performance of 2-extract 5-zone SMB.

choice of operating points on one-dimensional curve in the (m2 , m3 ) and (m4 , m3 ) regions, respectively: m3 m4 = KB KC ,

(11)

m3 = m2 (KC /KB ).

(12)

Experimental runs were carried out to study the effect of different operation conditions on the SMB performance by changing the difference of m3 and m4 (and thus m3 and m2 ) along the locus described by Eqs. (11) and (12). In Table 5, points H, I, J, K were selected whose feed flow rate were kept constant at 0.25 ml/min. It is shown in Fig. 6 that enrichment of both extracts and raffinate increased significantly with increasing the difference of m4 and m3 . Productivity of the two extracts and raffinate increased slightly with the increase of m4 –m3 while purity and yield of the streams does not change significantly. It was found that the desired component A was produced in the extract 1 stream with a high purity but its yield is lowered considerably compared with that of 2-raffinate configuration since component A was also discharged from extract 2 stream. Meanwhile, low purity of component B was obtained in extract 2 stream due to the inevitable contamina-

tion with component A and unexpected component C in this stream although its yield was high suggesting that almost all B migrated in the liquid flow direction in Section 2 and in the solid flow direction in Section 3 and discharged all the way from extract 2 stream. As for component C, it was obtained in a high the purity in the raffinate stream since it was the only component contained in this stream, but its yield was decreased due to the discharge in extract 2 stream which was undesirable. This suggests that the flow rate ratio restriction for component C could be violated in Section 3. Axial dispersion and mass transfer effect could make the effective separation region narrower in both (m4 , m3 ) and (m2 , m3 ) complete separation regions. Besides, the smaller difference of KB and KC can also make the separation more difficult and easy to be disturbed. It is known that the robustness of operation can be improved by decreasing the difference of m4 and m3 (or increasing the difference of m2 and m3 ). Since feed flow rates were kept constant during the course of changing flow rate ratios, decreasing the difference of m3 and m4 could make shorter switch time and thus increase the eluent flow rate in the whole system, until it was difficult to maintain stable operation and therefore achieve good separation results. As

1346

X. Wang, C. Bun Ching / Chemical Engineering Science 60 (2005) 1337 – 1347

a guideline, unstable operation was always avoided in the practical operation of SMB.

t∗ V

switching time in SMB process (min) column volume

Greek letters 5. Conclusion Continuous chromatographic separation of enantiomer (RSR)-nadolol from its racemic mixture (which is a ternary mixture in the chromatographic system) in both 2-raffinate and 2-extract five-zone SMB has been studied. The same experimental setup was applied to both five-zone SMB configurations provided the controlling software was modified accordingly. Separation performances of the five-zone SMB were investigated for both 2-raffinate and 2-extract configurations and same safety factors were applied to investigate the effect of m3 –m2 (or m4 –m3 ) on the separation performance systematically. It was found that desired enantiomer of nadolol can be produced with a high purity and yield in the 2-raffinate configuration while a high purity but considerably decreased yield of this component was obtained in 2-extract configuration. In the SMB operations, perphenyl carbamoylated
-CD was bonded onto larger silica gel of 15–25 m for economic and operation considerations (i.e., due to pressure limit of the whole system), which resulted in lower column efficiencies. Although resolution of nadolol in individual column was decreased, experimental results showed that SMB processes had the ability for difficult chiral separation, especially for that of 2-chiral center drugs, which could be difficult to fulfill in batch process if the same efficiency of column was employed.



P
∗

safety factor in safety margin method bed voidage intra-particle porosity total porosity of column

Subscripts and superscripts A B C D E F i j L R1 R2 S SMB TCC

the third eluted component of nadolol racemic mixture (component 3) the second eluted component of nadolol racemic mixture (component 2) the first eluted component of nadolol racemic mixture (component 1) desorbent (eluent) extract product of 2-raffinate five-zone SMB SMB feed stream component i of ternary mixture (i = 1–3) j section of TCC and SMB (j = 1–5) liquid phase raffinate 1 product of 2-raffinate five-zone SMB raffinate 2 product of 2-raffinate five-zone SMB solid phase simulated moving bed chromatography true counter-current chromatography

References Notation ci H K mj mj qi Qj Qs t0 tR

mobile phase concentration based on fluid volume (mg/ml) equilibrium constant (dimensionless), as defined in Table 2 equilibrium constant (dimensionless), as defined in Table 2 fluid phase flow rate over sold phase flow rate in j section of the TCC and SMB unit, as defined in Table 2 net fluid phase flow rate over sold phase flow rate in j section of the TCC and SMB unit, as defined in Table 2 concentration of component i on stationary phase (mg/ml), as defined in Table 2 liquid phase flow rate in j section of TCC or SMB process solid phase flow rate in TCC and SMB process mean retention time of an unretained compound (min) retention time of a component in the ternary mixture (min)

Abel, S., Paredes, G., Stadler, J., Mazzotti, M., Morbidelli, M., 2003. 3FSMB operation for the separation of biomolecules, A.I.C.H.E. Annual meeting San Francisco, USA. Beste, Y.A., Arlt, W., 2001. Simulated countercurrent chromatography with sidestream discharge. Chemie-Ingenieur Technik 73, 1567–1572. Charton, F., Nicoud, R.M., 1995. Complete design of a simulated moving bed. Journal of Chromatography A 702, 97–112. Decamp, W.H., 1989. The FDA perspective on the development of stereoisomers. Chirality 1, 2–6. Francotte, E., Richert, P., Mazzotti, M., Morbidelli, M., 1998. Simulated moving bed chromatographic resolution of a chiral antitussive. Journal of Chromatography A 796, 239–248. Kearney, M., Hieb, K.L., 1992. US Patent 5 102 553. Khattabi, S., Cherrak, D.E., Fischer, J., Jandera, P., Guiochon, G., 2000. Study of the adsorption behavior of the enantiomers of 1-phenyl1-propanol on a cellulose-based chiral stationary phase. Journal of Chromatography A 877, 95–107. Kim, J.K., Zang, Y.F., Wankat, P.C., 2003. Single-cascade simulated moving bed systems for the separation of ternary mixtures. Industrial and Engineering Chemistry Research 42, 4849–4860. Lehoucq, S., Verheve, D., 2000. SMB enantioseparation: process development, modeling and operating conditions. The American Institute of Chemical Engineers Journal 46, 247–256. Mazzotti, M., Storti, G., Morbidelli, M., 1994. Robust design of countercurrent adsorption separation processes: 2. Multicomponent system. The American Institute of Chemical Engineers Journal 40, 1825–1842.

X. Wang, C. Bun Ching / Chemical Engineering Science 60 (2005) 1337 – 1347 Mazzotti, M., Storti, G., Morbidelli, M., 1997. Optimal operation of simulated moving-bed units for nonlinear chromatographic separations. Journal of Chromatography A 769, 3–24. Nicolaos, A., Muhr, L., Gotteland, P., Nicoud, R.M., Bailly, M., 2003. Application of equilibrium theory to ternary moving bed configurations (four plus four, five plus four, eight and nine zones) I. Linear case. Journal of Chromatography A 908, 71–86. Nicoud, R.M., 1999. Simulated Moving Bed (SMB)—Some Possible Applications for Biotechnology in Bioseparation and Bioprocessing Handbook, Wiley–VCH, New York. Pais, S.L., Loureiro, M.J., Rodrigues, A.E., 1997. Separation of 1,1’-bi2-naphthol enantiomers by continuous chromatography in simulated moving bed. Chemical Engineering Science 52, 245–257. Pedeferri, M., Zenoni, G., Mazzotti, M., Morbidelli, M., 1999. Experimental analysis of a chiral separation through simulated moving bed chromatography. Chemical Engineering Science 54, 3735–3748. Rekoske, J.E., 2001. Chiral separations. The American Institute of Chemical Engineers Journal 47, 2–5.

1347

Ruthven, D.M., Ching, C.B., 1989. Counter-current and simulated countercurrent adsorption separation process. Chemical Engineering Science 44, 1011–1038. Storti, G., Mazzotti, M., Morbidelli, M., Carra, S., 1993. Robust design of binary countercurrent adsorption separation processes. The American Institute of Chemical Engineers Journal 39, 471–492. Wang, X., Ching, C.B., 2002. Kinetic and equilibrium study of the enantioseparation of a three chiral center drug (nadolol) on a perphenyl carbamoylated
-cyclodextrin bonded chiral stationary phase by HPLC. Separation Science and Technology 37, 2567–2586. Wooley, R., Ma, Z., Wang, N.-H.L., 1998. A nine-zone simulating moving bed for the recovery of glucose and xylose from biomass hydrolyzate. Industrial and Engineering Chemistry Research 37, 3699–3709. Yun, T., Zhong, G., Guiochon, G., 1997. Experimental study of the influence of the flow rates in SMB chromatography. The American Institute of Chemical Engineers Journal 43, 2970–2983.