In pursuit of purity

Sometimes not even the most established processes can deliver a high enough level of purity for evaluation or commercialisation of new intermediates and potential development candidates, argues Dr Patrick Kaiser, r&d senior scientist in custom synthesis at Sigma-Aldrich in Switzerland

In evaluation, as well as commercialisation processes, it is essential that compounds are available in high purity; consequently purification methods applied in process development and scale-up are of the utmost importance.However, in cases where established processes such as crystallisation, extraction and distillation are not successful, batch chromatographic methods are valuable tools to purify compounds.

A broad choice of commercially available solid phase materials for preparative chromatographic devices, like HPLC, allows the development of suitable methods to solve almost any given separation problem. The principal drawbacks of these conventional techniques are the high costs for the stationary phase and the eluant, especially on a larger scale. Only discrete amounts of a mixture can be separated in a 'discontinuous' way. A continuous process, however, would increase productivity and reduce costs.

Continuous separations on an industrial scale via a simulated moving bed (SMB) process have been known since the 1960s.¹ Since then, mixtures of hydrocarbons, sugars and racemates have been separated using SMB devices on a multi-ton scale per year. The characteristic feature of SMB technology is the simulation of a counter-current between the liquid and solid phases in a loop of chromatographic columns (beds). This allows a continuous injection of a binary substrate mixture and a continuous withdrawal of product streams.

technical problems

Besides simulated bed chromatography, true moving bed (TMB) processes are also known. The application of TMB technology, however, is hindered by technical problems in realising the flow of the solid phase.

In a conventional preparative chromatographic separation, a defined sample of the mixture is injected into the column (figure 1a). The components of the mixture will exhibit different migration velocities due to their different adsorption interactions with the solid phase. Therefore, a separation during the elution process will take place.

In a counter-current chromatographic process, like TMB, the solid phase flows in the opposite direction to the eluant (moving bed). With a proper choice of velocities, the components migrate in different directions and will be separated completely (figure 1b).

This concept is realised in the true moving bed (TMB) process. The setup of a true moving bed device consists of four sections forming a circuit (figure 2): two separation sections (II, III) and two desorbing sections (I, IV). Each section is comprised of one or more chromatographic columns.

The substrate mixture, or feed, is introduced continuously into the device between sections II and III. The components are separated in the sections II and III and are withdrawn by the raffinate and the extract streams. The solvent loss is compensated by an eluant inlet between section I and IV. Due to differences in the net flow between sections I and II, and sections III and IV respectively, the components are desorbed totally from the solid phase material transported from section I to IV.

The total flow in each section can be adjusted to reach a dynamic steady state, so the concentration profiles do not change with time. The components are separated continuously; high feed concentration and high overloading of the solid phase leads to high productivity. A general limitation of this method is that only two components (or fractions) can be separated at a time.

continuous movement

As mentioned above, practical application of TMB is limited by the transport of the solid phase. To resolve this problem, the SMB process simulates the counter-current stream by switching the positions of the inlet and outlet valves one column position further in the direction of the solvent flow (figure 3). In an SMB set-up of n columns, after n switches (tacts) one cycle is completed and the valves are back in their starting positions.

For an infinite number of columns, the switching time becomes infinitely small and the movement continuous, and the SMB process is equal to a TMB process. In standard laboratory SMB devices four, eight, 12 or 16 columns are used. The flows are controlled by HPLC pumps. Figure 4 shows an SMB unit for small-scale production. As in a TMB process, the components are separated continuously with high productivity and low solvent consumption.

In an SMB set-up, conventional solid phase material for the preparative HPLC is used. After optimisation of an isocratic HPLC separation method on an analytical scale, thermodynamic parameters (isotherms) of the compounds involved are calculated from the HPLC data. With these parameters, the distribution of the compounds between the solid and the liquid phase can be calculated and the SMB separation can be simulated with computer software. The optimal flows, feed concentration and switching time to design a stable and efficient separation are also obtained by the simulation.

The SMB run is then started using these parameters and monitored by analytical HPLC or other chromatographic methods. If necessary, small corrections during the first cycles are made to obtain optimal performance. Once a stable SMB process has been established, the separation can be carried out continuously for many days or even weeks.

Compared with a separation on a preparative HPLC, the method development for an SMB process is more time-consuming. If a suitable analytical HPLC method is available, the method development and set up of the SMB can be done in approximately one week. Due to the higher productivity of an SMB, this is compensated by the reduced production time.

wide choice

As laboratory scale SMB devices are usually equipped with conventional preparative HPLC columns, a wide choice of solid phase materials is available for the separation design. Every mixture that can be separated on an analytical HPLC is also theoretically suitable for an SMB separation. A crucial limitation is that only two components or two fractions can be separated at a time. If the mixture consists of more than two components and/or if the substance of interest does not eluate as first or last, two separation cycles are necessary.

Recently, SMB instruments have become increasingly standard equipment for preparative chromatography in the pharmaceutical and fine chemical industry. A very attractive application of SMB is the separation of enantiomers on chiral phases, due to the high value of the materials isolated.

To achieve a satisfactory resolution of racemates with non-chromatographic methods (e.g. by kinetic enzymatic derivatisation or crystallisation with chiral auxiliaries) multiple resolution cycles are often required, lowering the recovery rate and increasing the costs. SMB technology offers recovery rates usually higher than 80% and purities above 98%. The separation of racemic DOLE (table 1) is discussed below for comparison with SMB and preparative HPLC separations.²

simple scale-up

Scalability is another advantage of SMB. Unlike chemical processes, the SMB separation process is minimally influenced by non-linear scale-up effects (e.g. column volume). Once a separation method has been developed for a small device, it can be adapted easily to an SMB with a far higher capacity. Scalability is an important issue for the pharmaceutical industry in particular, as regulatory problems due to process changes can be circumvented. Separations can be conducted with a capacity of a few grams up to multi-kilograms per day, also under cGMP regulations. These amounts are usually enough to cover the demand for pharmaceutical active substances or building blocks from preclinical trials to commercialisation.

Other SMB applications include the separation of components with very similar physical and chromatographic properties. UOP commercialised the SMB isolation of paraxylene from other C-8 hydrocarbons with its Sorbex process in 1964.¹ The separation of glucose and fructose mixtures is performed in many industrial plants on ton scale using SMB technology.

The reasons for the commercial success of these processes are the high productivity, purities and recovery rates. Equally important, the ongoing costs of separations on larger scale are minimal, as the life time of the columns is very high and the solvent consumption low.

low solvent use

SMB has many advantages over conventional batch column chromatography (e.g. HPLC). Daicel investigated the separation of racemic DOLE (Table 1), a precursor for the cholesterol reducing agent NK104 Ca.³

One of the most important benefits was the low solvent consumption, which was only a quarter of that of the HPLC separation. At the same time, a higher feed concentration and a higher loading for the SMB resulted in more than doubled productivity per kg solid phase material.

The superiority of the SMB separation is also reflected in the higher recovery rate achieved. In the SMB process, the solid phase is heavily overloaded with compounds, and compared with preparative HPLC, the plate numbers of the columns have a smaller effect on the separation. Therefore cheaper solid phase material with a bigger particle size can be selected, resulting in only a slight loss in performance.

The major disadvantage of the SMB is the limitation of separating only binary mixtures. Multiple separation cycles are necessary if three or more components need to be isolated in pure form. In these cases, HPLC separation might be a better choice. As the method development for an SMB separation is more time-consuming, the non-repetitive isolation of small amounts of material, e.g. for screening purposes, can usually be done faster by preparative HPLC.

In conclusion, SMB cannot substitute for preparative HPLC and vice versa. Both methods have their advantages and their drawbacks, and the choice of the separation method depends largely on the application problem.

overcoming limitations

Many attempts have been made to overcome the limitation of separating only binary mixtures. A couple of modified SMB processes have been developed recently. For example, by upgrading the array of valves and pumps, separation of tertiary or quaternary mixtures can be realised.

Another development is the integration of a reaction and the separation of the corresponding products. In a simulated moving bed reactor, one product can be spatially separated from the other product(s), allowing equilibrium reactions to be driven completion (e.g. condensations, esterifications).

In the future, further development should broaden the range of applications of SMB technology. To date the improvements of SMB have already established it as an attractive alternative to batch separation processes.