Intermittent simulated moving bed chromatography: 3. Separation of Tröger's base enantiomers under nonlinear conditions

doi:10.1016/j.chroma.2011.10.073

Journal of Chromatography A

Volume 1218, Issue 52, 30 December 2011, Pages 9345-9352

https://doi.org/10.1016/j.chroma.2011.10.073 Get rights and content

Abstract

One of the modified simulated moving bed (SMB) processes, the intermittent SMB (I-SMB) process, has been recently analyzed theoretically [1] and its superior performance compared to the conventional SMB process has been demonstrated at a rather low total feed concentration through experiments and simulations [2]. This work shows that the I-SMB process outperforms the conventional SMB process also at high feed concentration where the species are clearly subject to a nonlinear adsorption isotherm. In the case of the separation of the Tröger's base's enantiomers in ethanol on ChiralPak AD, the two processes operated in a six-column 1-2-2-1 configuration (one column in sections 1 and 4 and two columns in sections 2 and 3) and in a four-column 1-1-1-1 configuration (one column in each section) are compared at high feed concentration through both experiments and simulations. Even under nonlinear conditions the four column I-SMB process can successfully separate the two enantiomers achieving purity levels as high as the two six column processes and exhibiting better productivity.

Highlights

► We prove intermittent-SMB multicolumn chromatography under nonlinear conditions. ► The Troger's base's enantiomers have been separated on ChiralPak AD in ethanol. ► The I-SMB process outperforms the conventional SMB even at high feed concentration. ► The 4-column I-SMB process can achieve both high purity and high productivity.

Introduction

Intermittent simulated moving bed (I-SMB) chromatography is an extension of the conventional SMB process, which has been invented and patented by Nippon Rensui Corporation [3]. In the previous papers of this series we have theoretically analyzed this process [1] and demonstrated its good performance at rather low total feed concentration when compared with that of the conventional SMB process [2].

The conventional SMB process simulates a continuous countercurrent movement of the fluid phase and of the adsorbent by periodically switching the inlet and outlet ports to a set of identical chromatographic columns in the same direction of the fluid flow; the time period between two switches is the switch time, t^*. Typically a SMB unit is divided in four sections by two inlets and two outlets, in such a way that the feed and the mobile phase (desorbent) are introduced between sections 2 and 3 and between sections 4 and 1, respectively, whereas the extract (where the more retained species, labeled A in the following, is withdrawn) is collected between sections 1 and 2 and the raffinate (containing the less retained component B) is collected between sections 3 and 4. The outlet of section 4 is recycled to section 1 in the case of the rather common closed-loop configuration, otherwise it is collected and possibly recycled off-line in the case of the open-loop configuration.

On the contrary, the I-SMB process is operated in two different modes during one switch period t^*, which is in fact divided into two sub-intervals. During the first step I, of duration αt^* (0 < α < 1), the unit is operated as a three-section conventional SMB, where the feed and desorbent are normally introduced, the extract and the raffinate are normally collected, but there is no flow in section 4. During the second step II, the ports of the inlets and outlets are closed and the fluid is just circulated through all sections of the unit to adjust the relative position of the concentration profiles with respect to the positions of the inlet and outlet ports [1].

Its key features have been discussed in the frame of the equilibrium theory, and it has been shown that the synchronous partial feed and partial withdrawn operation of the I-SMB process makes it possible to achieve high product purity and good separation performance with four columns only [1]. The separation of the Tröger's base's enantiomers on ChiralPak AD stationary phases using ethanol as mobile phase has been carried out in an existing laboratory multi-column chromatography unit operated in both the I-SMB and the conventional SMB mode at rather low feed concentration where the species to be separated are subject to an adsorption isotherm exhibiting a mild nonlinearity. It has been demonstrated that indeed the four column (1-1-1-1 configuration, i.e. one column in each section) I-SMB process outperforms the conventional SMB process [2].

However, in most cases of practical interest in order to maximize productivity one has to deal with a separation carried out under nonlinear conditions, i.e. where the species are clearly subject to a nonlinear adsorption isotherm. In this study, the potential of the I-SMB process at high feed concentration, i.e. under nonlinear conditions, will be explored through experiments as well as simulations using the same model system consisting of the Tröger's base enantiomers on ChiralPak AD as stationary phase and in ethanol as mobile phase. The separation performance of the four-column and six-column (1-2-2-1 configuration, i.e. two columns in sections 2 and 3) I-SMB and conventional SMB processes at a total feed concentration from 5 g/L to 15 g/L will be analyzed and compared, in order to assess the potential at the I-SMB technology at high feed concentration.

Section snippets

Background

In this section we summarize background information from the previous papers of this series that are needed for the comprehension of the new material presented in this work [2].

Choice of the operating conditions for nonlinear I-SMB processes

For both the conventional SMB and the I-SMB process under linear chromatographic conditions, the choice of the flow rate ratios to achieve the complete separation of the two species to be separated is based on the same explicit criteria, i.e. those given by triangle theory, namely m₁ ≥ H_A ≥ m₃ ≥ m₂ ≥ H_B ≥ m₄ ≥ − ϵ^*/(1 − ϵ^*), where H_A and H_B are the Henry's constants of the two species in the feed mixture. This is based on a rigorous demonstration that is indeed possible in the linear case [1].

For the

Experimental results and discussion

In order to confirm the conclusions of the simulation study, about 20 experiments have been carried out. All four operating modes considered so far, i.e. six-column and four-column SMB and I-SMB processes, have been tested with a feed mixture of 5 g/L total concentration, whereas at the higher feed concentrations of 10 and 15 g/L only the four-column processes have been studied. The operating conditions and the separation performance in terms of purity, productivity and solvent consumption, of

Nomenclature

a_i,k, b_i,k

parameters in bi-Langmuir isotherm for component i

c_i

fluid phase concentration of component i

c_{F}^{T}

total feed concentration

c_P,i

concentration of component i at port P

{\bar{c}}_{P, i}

average concentration of component i at port P

D_i

axial dispersion coefficient of component i

H_i

Henry's constant of component i

k_s,ia_v

product of mass transfer coefficient and specific surface of component i

column length

m_j

flow rate ratio in section j of conventional SMB and I-SMB

n_{i}^{*}

adsorbed phase concentration of component i

References (23)

KatsuoS. et al.
J. Chromatogr. A
(2010)
KatsuoS. et al.
J. Chromatogr. A
(2010)
AbelS. et al.
J. Chromatogr. A
(2004)
KatsuoS. et al.
J. Chromatogr. A
(2009)
PedeferriM. et al.
Chem. Eng. Sci.
(1999)
Seidel-MorgensternA. et al.
J. Chromatogr.
(1993)
Seidel-MorgensternA. et al.
Chem. Eng. Sci.
(1993)
MihlbachlerK. et al.
J. Chromatogr. A
(2002)
MihlbachlerK. et al.
J. Chromatogr. A
(2006)
GuiochonG.
J. Chromatogr. A
(2002)

MazzottiM. et al.

J. Chromatogr. A

(1997)

Cited by (15)

General optimization strategy of simulated moving bed units through design of experiments and response surface methodologies
2016, Computers and Chemical Engineering
Citation Excerpt :
In the last decades simulated moving bed (SMB) chromatography has been increasingly used in fine chemistry and the separation of enantiomers may be cited as a successful application (Juza et al., 2000; Rajendran et al., 2009; Aniceto and Silva 2015a,b). Some recent examples are the separation of guaifenesin (Gong et al., 2014), flurbiprofen and ketoprofen (Ribeiro et al., 2011a,b), and Troger’s base enantiomers (Katsuo and Mazzotti, 2010; Katsuo et al., 2011). The SMB process is a separation technique derived from the earlier true moving bed (TMB).
The optimization of simulated moving bed (SMB) units is often performed through detailed phenomenological models and require extensive computation time. Hence several optimization methods like the Triangle Theory, and the concept of separation volume have been proposed. However, they do not provide accurate results, when mass transfer limitations are significant, or require a large number of simulations.
In this work, a combined Design of Experiments and Response Surface Methodology (DoE-RSM) approach is proposed for SMB optimization, aimed at providing good results with simplicity and reduced number of simulations. The separation of trans-stilbene oxide enantiomers is selected as case study in order to compare DoE-RSM with previous approaches. In the whole, accurate results are obtained with a few number of simulations, allowing for purities above 99.60% for both enantiomers, and productivity of 65.41 kg/(m³_adsorbent day). The versatility of DoE-RSM tool is also discussed, emphasising their advantages and general applicability.
Three column intermittent simulated moving bed chromatography: 3. Cascade operation for center-cut separations
2015, Journal of Chromatography A
A general design methodology for chromatographic three fraction separation by application of the three column intermittent simulated moving bed (3C-ISMB) cascade is proposed and experimentally validated by studying the purification of an intermediately retained stereoisomer of nadolol, from an equimolar mixture of its four stereoisomers. The theoretical part shows that the 3C-ISMB cascade can be easily designed by applying Triangle Theory. Moreover, a re-scaling approach for the second stage is proposed so as to account for the fact that the feed flow rates to stage 2 are generally higher as compared to stage 1 due to dilution in the latter. Scaling the columns of the second stage accordingly enables to run both stages under optimal conditions with respect to switching time and step ratio, which is an important advantage as compared to integrated ternary processes. The experimental part starts with studying the linear adsorption behavior of nadolol in heptane/ethanol/DEA on Chiralpak AD for varying ratios of heptane and ethanol. Based on that, a solvent composition of Hept/EtOH/DEA 30/70/0.3 (v/v/v) is selected and the competitive multi-component Langmuir isotherm of the quaternary mixture is determined by frontal analysis. The resulting isotherm parameters are used to design several first stage experiments aiming at removal of the most retained component. The resulting ternary intermediate product is reprocessed in several second stage experiments studying various configurations. Finally, the dilution of the intermediate product with Hept/DEA yielding a solvent composition of Hept/EtOH/DEA 60/40/0.3 (v/v/v) is examined showing that the resulting increase in retention is beneficial for final product purities. Moreover, the reduction in viscosity compensates for the dilution as it enables higher flow rates. Dilution of the intermediate product is hence the best option, yielding highest overall cascade productivity (2.10 gl⁻¹ h⁻¹) and highest product purity (97.8%) requiring a specific solvent consumption of 12 l/g of product.
Three-column intermittent simulated moving bed chromatography: 2. Experimental implementation for the separation of Tröger's Base
2014, Journal of Chromatography A
Citation Excerpt :
The series of experiments in run F gives evidence that the experimentally achievable extract purity is slightly lower than the corresponding raffinate purity, i.e. an upper limit in extract purity of about 99.5% cannot be overcome by changing the operating point. This asymmetry in the experimentally achievable purities can also be observed in runs A to E for both 3C-ISMB and I-SMB, as well as for I-SMB and SMB in our previous studies [9,10]. Therefore, we conclude that this limitation is not specific to the new 3C-ISMB process, but it is more general and might be related to a set-up specific unidentified cause, e.g. some sort of very mild cross-contamination.
The three-column intermittent simulated moving bed (3C-ISMB) process has been presented in the first part of this series [1]. A theoretical comparative analysis of the new process to intermittent simulated moving bed (I-SMB) demonstrated successfully the potential of the 3C-ISMB technology. Particularly, 3C-ISMB was shown to substantially outperform I-SMB in terms of productivity whilst maintaining high purity specifications and without significantly sacrificing solvent consumption. Moreover, we demonstrated the applicability of Triangle Theory for 3C-ISMB design, which is an important advantage compared to other modified SMB schemes. In the present work we report on the experimental implementation of the 3C-ISMB technology demonstrating the simplicity of retrofitting an existing SMB or I-SMB plant. Moreover, we perform an experimental comparative analysis studying the well-known separation of Tröger's Base enantiomers in pure ethanol on Chiralpak AD™. In a comprehensive series of experimental runs, each examining three different modes of operation, applying total feed concentrations ranging from 5 g/l to 17.4 g/l, we assess and compare the separation performances of both conventional I-SMB and 3C-ISMB. This series of experiments successfully demonstrates that 3C-ISMB delivers the same high purity levels as conventional I-SMB whilst significantly outperforming the conventional process in terms of productivity; in fact an increase of more than 80% is achieved.
Three column intermittent simulated moving bed chromatography: 1. Process description and comparative assessment
2014, Journal of Chromatography A
Citation Excerpt :
The regions of complete separation in the operating parameter plane for the standard SMB separation of a system subject to a Bi-Langmuir isotherm are readily calculated by applying the method developed by Migliorini et al. [22]. Although, the boundaries of these regions were derived for SMB and are based on the equilibrium theory model, i.e. a simplified model which neglects mass transfer resistance and axial mixing, Katsuo et al. showed that the vertex of the real region of complete separation for the I-SMB process is very well predicted by Triangle Theory [14,15]. This was shown both experimentally and through simulations where a small region around the vertex of the theoretical triangle was screened applying an equilibrium-dispersive model for the same system being studied in this work.
The three column intermittent simulated moving bed (3C-ISMB) process is a new type of multi-column chromatographic process for binary separations and can be regarded as a modification of the I-SMB process commercialized by Nippon Rensui Corporation. In contrast to conventional I-SMB, this enables the use of only three instead of four columns without compromising product purity and throughput. The novel mode of operation is characterized by intermittent feeding and product withdrawal as well as by partial recycling of the weakly retained component from section III to section I.
Due to the smaller number of columns with respect to conventional I-SMB, higher internal flow rates can be applied without violating pressure drop constraints. Therefore, the application of 3C-ISMB allows for a higher throughput whilst using a smaller number of columns. As a result, we expect that the productivity given in terms of throughput per unit time and unit volume of stationary phase can be significantly increased.
In this contribution, we describe the new process concept in detail and analyze its cyclic steady state behavior through an extensive simulation study. The latter shows that 3C-ISMB can be easily designed by Triangle Theory even under highly non-linear conditions. The simple process design is an important advantage to other advanced SMB-like processes. Moreover, the simulation study demonstrates the superior performance of 3C-ISMB, namely productivity increases by roughly 60% with respect to conventional I-SMB without significantly sacrificing solvent consumption.
Multi-column chromatographic process development using simulated moving bed superstructure and simultaneous optimization - Model correction framework
2014, Chemical Engineering Science
In this work, we demonstrate an improved framework for simulated moving bed (SMB) chromatographic process development using a superstructure formulation. Various optimal column configurations and operations are obtained by solving a multi-objective optimization problem representing the superstructure by maximizing feed throughput and minimizing desorbent usage. In order to resolve the model mismatch with experimental data, here we utilize the simultaneous optimization – model correction (SOMC) method in which process optimization is carried out in tandem with model correction using data obtained from experimental evaluations.
The potential of the superstructure–SOMC framework has been demonstrated by separating glucose and fructose using columns packed with a cation exchange resin with water as the mobile phase. The optimal operation modes found using the superstructure included standard SMB, three zone (3Z) operation, intermittent SMB (I-SMB) and a newly found outlet stream swing (OSS) – partial feed (PF) hybrid operation. All configurations were experimentally implemented using a Semba ${Octave}^{™}$ 100 Chromatography System.
Model-based identification of optimal operating conditions for amino acid simulated moving bed enantioseparation using a macrocyclic glycopeptide stationary phase
2014, Journal of Chromatography A
Teicoplanin aglycone columns allow efficient separation of amino acid enantiomers in aqueous mobile phases and enable robust and predictable simulated moving bed (SMB) separation of racemic methionine despite a dependency of the adsorption behavior on the column history (memory effect). In this work we systematically investigated the influence of the mobile phase (methanol content) and temperature on SMB performance using a model-based optimization approach that accounts for methionine solubility, adsorption behavior and back pressure. Adsorption isotherms became more favorable with increasing methanol content but methionine solubility was decreased and back pressure increased. Numerical optimization suggested a moderate methanol content (25–35%) for most efficient operation. Higher temperature had a positive effect on specific productivity and desorbent requirement due to higher methionine solubility, lower back pressure and virtually invariant selectivity at high loadings of racemic methionine. However, process robustness (defined as a difference in flow rate ratios) decreased strongly with increasing temperature to the extent that any significant increase in temperature over 32 °C will likely result in operating points that cannot be realized technically even with the lab-scale piston pump SMB system employed in this study.

View all citing articles on Scopus

^☆: The first and second part of this series were published by Katsuo and Mazzotti [1], [2].

¹: Permanent address: Mitsubishi Chemical Group Science and Technology Research Center, Inc., 1000, Kamoshida-cho, Aoba-ku, Yokohama 227-8502, Japan.

²: Current address: Bain & Company Switzerland, Inc., Rotbuchstrasse 46, 8037 Zurich, Switzerland.

³: Current address: MINES/ENSTA ParisTech, CEP/SCPI, 32 Bd Victor, 75015 Paris, France.

View full text

Intermittent simulated moving bed chromatography: 3. Separation of Tröger's base enantiomers under nonlinear conditions☆

Abstract

Highlights

Introduction

Section snippets

Background

Choice of the operating conditions for nonlinear I-SMB processes

Experimental results and discussion

Nomenclature

J. Chromatogr. A

J. Chromatogr. A

J. Chromatogr. A

J. Chromatogr. A

Chem. Eng. Sci.

J. Chromatogr.

Chem. Eng. Sci.

J. Chromatogr. A

J. Chromatogr. A

J. Chromatogr. A

J. Chromatogr. A