Simulated Moving Bed Chromatography: A Powerful Unit Operation

When it comes to controlling the quality of a final active pharmaceutical ingredient, high-performance liquid chromatography (HPLC) is one of the techniques used by quality control departments to analyze the final product. HPLC has gained popularity because of its efficient separation performance. Unfortunately, this separating power is rarely considered as a potential purification process at the commercial scale. The reason is that chromatography is always perceived as an expensive technique because of poor throughput and high solvent consumption. Although production-scale HPLC may have been expensive in the past because of glass columns, it is not today, thanks to major improvements in the technology.

AUTHORS

Highly efficient packing materials, also called stationary phases, with narrow particle-size distribution are available and can be used at high pressure. As a result, high throughput can be achieved with high purity and high recovery compared with traditional "column" chromatography. Furthermore, additional improvements to the technique allow for counter-current continuous operation that further reduces the need for solvent and significantly improves process throughput. This technique, called simulated moving bed (SMB), is currently in use in many industrial applications at large scales (see Figure 1). For example, the purification of sugar is conducted using large diameter SMB units with a throughput of several tons of purified fructose or glucose per day.

Figure 1: Commercial-scale simulated moving bed unit, 5 columns of 1 m in diameter. This unit can process several hundred tons of racemic solution per year. (PHOTO: AMPAC FINE CHEMICALS)

Description of the process

SMB has been presented in many different ways in the scientific literature. Using a merry-go-round of 5–8 columns and a set of valves per column, a simple batch separation process can be converted into a continuous operation. This is achieved by changing the relative velocities of each compound in the columns. By "moving" the columns at a constant speed against the eluent flow, the most-retained component appears to move with the columns while the less-retained component flows with the eluent. As a result, a gap between the two fractions is created and can be filled continuously. To avoid accumulation, outlet streams must be collected on either side of the point of entry of the feed. This is truly a counter-current process and the packing material is used more efficiently. The only limitations are that SMB separations are conducted in an isocratic mode and that only two product streams can be isolated. This makes SMB an excellent candidate for binary separations such as the separation of enantiomers.

Method development and chiral applications

The number of drugs entering the market that have one or more chiral centers has grown significantly in the past decade. Knowing that one enantiomer can potentially carry side effects has brought regulatory authorities such as the US Food and Drug Administration to favor the single enantiomer version of a drug whenever possible. Specific crystallization, enzymatic or chemical resolution, or chromatography can be added as a unit operation to the existing synthesis route to obtain the single enantiomer. Or a completely new asymmetric route can be designed to obtain the desired single enantiomer. As a result, pharmaceutical companies have invested substantial resources in trying to find the right chiral catalyst or the right enzyme that will provide the single enantiomer at a satisfactory enantiomeric purity.

The development of a chromatographic chiral separation is a straightforward process that requires a few steps, described as follows.

Find a separation. There are a handful of chiral stationary phases (CSP) that provide greater than 80% probabilities of finding a good separation. If these phases are not working, there are a few more that may be suitable. The selection of solvents for the separation traditionally has been limited to acetonitrile, alcohols, and mixtures of alcohols with heptane. However, with the development of more robust CSPs that can handle solvents such as acetone, MTBE, toluene, and ethers, more possibilities exist. These solvents allow for high solubility and low viscosities, which are key elements to achieve high throughput. The screening process typically takes 1–2 weeks. From this study a handful of conditions can be identified as potential candidates.

Determine the amount that can be separated. The "loading study" step is aimed at determining the maximum loading capacity of the CSP. The study consists of injecting on an analytical column (4.6-mm diameter X 250-mm long) packed with preparative CSP (typically 16 or 20 μm) an increasing volume of a concentrated solution of the racemic mixture until the separation is lost (see Figure 2). From these data valuable information about the behavior of the compound as a function of the concentration is collected. This information is entered into a computer model which in turn calculates the parameters required to operate an SMB unit. At this point an estimation of the production rate is obtained and the size of the SMB unit required to produce the desired amount of the single enantiomer can be calculated.

Figure 2: Example of a loading study. By increasing the volume injected, the peaks change shape according to their equilibrium isotherm. This information is used to model the SMB process and calculate the throughput. (FIGURE: AMPAC FINE CHEMICALS)

Demonstrate proof of concept. Simulations are not always perfect, and it is usually a good practice to demonstrate the separation on a small-scale unit. On such a unit, equipped with columns of 4.6 or 10 mm in diameter, 20–200 g of material can be processed (see Figure 3). This amount is enough to achieve steady state and obtain a representative sample of the product. The columns can be packed in house or purchased from the packing manufacturer.

Figure 3: Bench-top simulated moving bed unit, 8 columns of 4.6 mm or 1 cm in diameter. This unit is used for proof of concept. (PHOTO: AMPAC FINE CHEMICALS)

This step can take 2–4 weeks, which is long enough to evaluate the productivity as well as the robustness and the stability of the separation. At this stage, chemists can start developing the isolation process post-SMB separation (e.g, crystallization, solvent exchange, drying).

Scale-up for clinical quantities. The scale-up from the demonstration run is straightforward. Chromatography processes are scaled up on the basis of the linear velocity at the particle level (i.e., the flow rate is multiplied by the ratio of the column diameters squared.) For example, if 50 g of racemic feed can be processed per day on a 10-mm diameter column, then a 50-mm SMB unit can process 1250 g per day (25X). This is a very easy process to scale-up. As a matter of fact, performances often improve with the diameter of the columns because of the favorable ratio of column to piping volume.

Following the demonstration step, it is frequent that 5–25 kg are separated using the 8 X 50-mm unit to supply material for Phase I. This material can be made according to current good manufacturing practices and purified in a matter of weeks. This work is still conducted on lab-scale equipment, and the overall cost is very reasonable.

Conduct production at pilot scale. Typically, a campaign involving the separation of 200–500 kg of racemic feed follows the scale-up phase. Normally, this is to supply material for Phase II or Phase III clinical trials. At this stage, the separation process is well characterized. The effort is more concentrated in the handling of solvent (recycling, composition adjustment etc.) and the product recovery (solvent exchange, crystallization, drying, etc.). All these steps are conducted in equipment that is similar to commercial-scale units. This provides a complete and accurate picture of the final process at commercial scale.

Conduct commercial-scale production. Provided that the feed material is available, a process can be developed from method screening to commercial quantities in a mater of months. SMB as a unit operation to produce an API has been approved by FDA. For example, Escitalopram, an antidepressant developed by Lundbeck (Copenhagen, Denmark), was one of the first drugs to be approved using an SMB step. Since then, additional drugs have been approved, and a lot more are in the pipeline.

At this stage, this process, like any other process, must be well defined and ready for validation. The definition of the batch for a continuous process is based on time in the unit. This time can be the time required to process enough material to fill a dryer or fill the reactor for the next chemistry step. This must be defined clearly so that the pedigree of the product can be traced.

Other considerations

The stability and robustness as well as the cost of the chiral stationary phase are always a concern when a new SMB process is discussed. AMPAC Fine Chemicals (AFC) has been producing several hundred metric tons of an API with SMB as the penultimate step of manufacturing using the same packing material for the past eight years. Unpacking the columns every 12 or 18 months is sometimes required to remove the fines that accumulate as a result of slow attrition of the packing. Thus, for a commercial-scale project involving the separation of hundreds of metric tons per year over several years, the cost of the packing material amortized during three years is in the range of $5–10 per kilogram of product manufactured.

The cost of operating an SMB unit is not as expensive as it is commonly believed. Most of the solvent used for the process is recycled at three levels. The first level is internal to the SMB. The second level is at the falling films used on each outlet to preconcentrate the product. At this point 75–85% of the solvent is recycled. Finally, the solvent removed from the drying step can be recycled to achieve a total greater than 99%. AFC received an award from the Department of Toxic Substance Control (DTSC) and the Chemical Industry Council of California (CICC) for the implementation of the solvent recycling on the large-scale manufacturing using SMB. Several thousands of gallons of solvent are recycled instead of being wasted, thereby making SMB a green technology. Currently, the overall solvent usage at AFC for commercial-scale units (6 X 800 mm and 5 X 1000 mm) is less than a few drums of solvents per month.

The labor required for the operation of an SMB unit is also relatively low. A single operator can monitor the unit, and from time to time, a second operator is required for preparing the feed or for recovering the product from the dryer.

The cost of equipment installation is still relatively high mostly because of the cost of the unit and the amount of ancillary tanks required to operate the system. Nonetheless, this cost will eventually decrease as more units are sold around the world.

It is quite possible that for a separation with a decent productivity (2 kg of feed/day/kg of CSP or better) and relatively high annual volumes, the price of the product falls into the range of $75–150/kg — all included.

Nonchiral application

SMB is not limited to chiral separations. Any separation that can be turned into a binary separation is potentially an excellent candidate for SMB. Removal of troublesome impurities by SMB is a good example. Troublesome impurities are closely related to the main product or are dimers or trimers of the product and are very difficult to remove. Sometimes multiple crystallizations are required to meet the final product specifications.

As an example, at AFC we purify a natural product by SMB. There are numerous impurities in the natural product, but one in particular that is very difficult to remove by traditional purification processes. Removal of this impurity by crystallization would result in a substantial loss of the expensive natural product to the mother liquor, which in turn makes crystallization a very expensive technique. Work was undertaken at AFC to develop a commercially viable SMB process to achieve high purity and high recovery of the natural product. The chromatogram of the feed analyzed under the SMB conditions shows only two peaks, but each peak represents the combination of several species. The SMB not only removes the troublesome impurity but it also removes a significant portion of the other impurities, thereby bringing the product total purity from ~75% to >96%.

Conclusion

Because it provides high purity, high yields, and potentially high throughput, chromatography is the ultimate purification technique. With recent improvements in technology and stationary phases, the cost of using chromatography has significantly decreased, making this unit operation a very cost effective and commercially viable technology.

For a related article, see In the Loop: Continuous Chromatography for Chiral (and Other) Separations.

Olivier Dapremont, PhD, is director of chromatography business development at AMPAC Fine Chemicals, PO Box 1718, MS 1007, Rancho Cordova, CA 95741-1718, tel. 916.357.6242, fax 916.353.3523, olivier.dapremont@apfc.com.