Introduction

Simulated moving-bed (SMB) technology is a continuous separation technique that improves upon tra-ditional batch chromatography. It continuously separates liquid products or purifies feed streams using very much the same chromatography mechanisms familiar to most practicing chromatographers, including adsorption, ion exchange, size exclusion, hydrogen bonds, complexation, or a combination of these mech-anisms. The first recorded SMB chromatographic technique was developed in the petrochemical industry in the late 1940s [1] and has been widely used industrially in petrochemical refineries since the 1970s, in high-fructose corn-syrup processing since the 1980s, and enantiomer separations in the pharmaceutical industries since the 1990s. This chapter provides an introduction with an overview of the principles of SMB, its essential design tools, and several examples of potential SMB applications in biorefineries.

7.1.1 Principles of separations in batch chromatography and SMB

In conventional batch preparative chromatography, a feed mixture is first loaded into a column (or a series of columns) filled with a sorbent or stationary phase (such as activated carbon), and the feed pulse is then eluted with a desorbent (the mobile phase), isocratically (no change in the desorbent composition) or otherwise. A solute in the feed, which has a high affinity for the sorbent, has a high partition coefficient.

This means that a larger fraction of the solute exists in the sorbent phase than in the mobile phase. Since only solutes in the mobile phase migrate downstream in the column, the average migration velocity of a solute is proportional to its fraction in the mobile phase. For this reason, a higher affinity solute migrates more slowly than a low-affinity solute, resulting in separation of the various feed components in the column.

Separation and Purification Technologies in Biorefineries, First Edition.

Edited by Shri Ramaswamy, Hua-Jiang Huang, and Bandaru V. Ramarao.

c 2013 John Wiley & Sons, Ltd. Published 2013 by John Wiley & Sons, Ltd.

The exit stream from the column is either collected as waste or diverted as product. After product collection is completed, the column is regenerated, reequilibrated, and reused. Batch chromatography is most useful as a polishing step, where either the solute of interest (product) or a small amount of impurity is retained in a high-capacity, highly selective stationary phase. However, in most industrial separations, some impurity components in the feed have a slightly higher affinity and some have a slightly lower affinity than the product. Consequently, a high degree of product separation from both groups of impurities is needed to achieve high purity. In these instances, batch chromatography suffers from low yield, high solvent consumption, high product dilution, and poor column utilization. It is generally difficult to achieve high product purity and high yield simultaneously in batch chromatography.

Simulated moving-bed chromatography overcomes many of the drawbacks of conventional batch chro-matography. It employs a series of linked columns, often in an unbroken loop, with inlet and outlet ports between the columns, as shown in Figure 7.1 for a standard system with eight columns. The loop of columns is divided by four inlet or outlet ports into four zones, and each zone has a different mobile phase velocity or flow rate. The ports move periodically and synchronously along the loop to follow the migrating

Feed

Desorbent Extract Raffinate

Feed

Desorbent Extract Raffinate

Step n

Step n + 1 1

1 2 3 4 5 6 7 8

2 3 4 5 6 7 8

Zone I Zone II Zone III Zone IV

Zone IV Zone I Zone II Zone III

Figure 7.1 Principle of binary separation in a conventional four-zone SMB with two columns in each zone.

The black-filled circles represent the high-affinity (slow-moving) solutes, and the white circles represent the low-affinity (fast-moving) solutes. In step n+1, the desorbent, extract, feed, and raffinate ports all move one column downstream from step n

solute bands. In step n, for example, desorbent and feed, which has both the high-affinity (slow-moving) and the low-affinity (fast-moving) solutes, are continuously introduced into the loop through the desorbent port (inlet of Column 1) and the feed port (inlet of Column 5), respectively; some fast-migrating solute is continuously removed from the raffinate port (end of Column 6), which is located downstream from the feed port; and some slow-migrating solute is continuously removed from the extract port (end of Col-umn 2), which is located upstream from the feed port. In step n+1, the desorbent and feed ports move one column downstream to the inlets of columns 2 and 6, respectively. Similarly, the extract and raffinate ports move one column downstream to the end of columns 3 and 7, respectively.

The average port velocity and the zone flow rates are designed so that feed is added continuously into the mixed region in the loop, while the two products are removed continuously from the pure component regions on both ends of the mixed region, thus achieving high purity and high yield for both products. In essence the ports, on average, travel faster than the adsorption front of the slow-moving solute (black-filled circles) in zone III, which is the region between the feed and raffinate ports. As a result, the slow solute never reaches the raffinate port, allowing the raffinate to recover only the fast-moving solute. Similarly, the ports on average move slower than the desorption tail of the fast-moving solute in zone II (between the extract and feed ports). The extract port never reaches the tail of the fast-moving solute (white circles).

The extract thus collects only the slow-moving solute (black circles). Zone IV (between the raffinate and desorbent ports) allows for solute-free desorbent to travel back into zone I. The slow-moving solute is eluted from the first column of zone I before the column is moved to zone IV in the next step. The periodic port movement achieves a simulated countercurrent movement of the stationary phase (or bed) relative to the fluid phase and maintains high purity and high yield for both products. If a great many very small columns are connected, the ports will appear to move continuously in the loop. This limiting example is called a continuous moving-bed (CMB) or a true moving-bed (TMB).

7.1.2 The advantages of SMB

Simulated moving-bed technology provides significant advantages over conventional batch chromatography because (i) it requires only partial band separation; (ii) the two products are continuously withdrawn from the pure component ends of the moving solute band; and (iii) the solvent is automatically recycled.

Simulated moving-bed technology can thus achieve both high product purity and high yield while improving stationary phase productivity and desorbent utilization. These benefits result in significant cost savings over batch chromatography [2–6]. Furthermore, the lowered desorbent consumption reduces the environmental impact of the process. A recent comparison of different chromatographic techniques for separation of a racemic mixture found that the SMB technique has the lowest solvent consumption and the best column utilization at production scale [7]. Simulated moving-bed technology also reduces floor space, equipment size, and manpower, reportedly as much as four times for an industrial chiral separation process [8].

7.1.3 A brief history of SMB and its applications

Simulated moving-bed technology, the use of fixed beds and moving ports to simulate countercurrent move-ment of solids and liquids, was first used in the 1840s in England as the Shank’s system for leaching [9].

Eagle and Scott of California Research Corp. (Richmond, California, U.S.) in 1949 reported SMB as the Cyclic Adsorption process used for the commercial recovery of aromatics and olefins from petroleum [1, 10, 11]. UOP LLC (Des Plaines, Illinois, U.S.), now part of Honeywell (Morristown, New Jersey, U.S.), developed a family of SMB processes for the petrochemical industry. UOP’s Sorbex family of SMB

processes includes Molex for the recovery of linear paraffins, Parex for the recovery of para-xylene, Ebex for the recovery of ethyl benzene from mixed C8 aromatic isomers, and Olex for the recovery of olefins from paraffins. UOP has received more than 300 U.S. patents for SMB-related operations, equipment and applications in petrochemical derivatives, carbohydrates, fatty acids, biochemicals, and pharmaceuticals [2, 3, 12–14].

Most, if not all, of the industrially relevant SMB processes were in the petrochemical industry until the late 1970s. The next industrial adoption occurred in the sugar industry, predominantly in the corn-derived sugar industry, in the 1980s. The Sarex process was developed by UOP for sugar purification [2, 14].

Many other similar SMB sugar-purification processes were patented and adopted industrially: see U.S.

Patent No. 4 157 267 assigned to Toray Industries, Inc. (Tokyo, Japan) for SMB recovery of fructose on zeolite from a fructose-glucose feed mixture [15]; No. 4 182 633 assigned to Mitsubishi Chemical Industries Ltd. (Tokyo, Japan) for operational improvements on fructose-glucose SMB separation on acidic cation exchange resin [16]; and No. 4 412 866 assigned to the Amalgamated Sugar Co. (Ogden, Utah, U.S.) for backwashing the bed in a fructose-glucose SMB separation [17]; among many others. Examples of SMB sugar separations in the literature have also been reported [2, 3, 18–22]. We will later illustrate in detail the development of an SMB process for the purification of sugars from biomass hydrolysate.

The 1990s saw the development of selective but high-priced chiral stationary phases for enantiomer separation. Simulated moving-bed technology, with its significant improvement in stationary phase utiliza-tion, became the technique of choice for enantiomeric separation. Miller and colleagues repeatedly showed that SMB is superior to other process-scale enantiomer separations [5, 7, 8]. Lists of published SMB enantiomer separations and further discussion on the adoption of SMB in the pharmaceutical industry can be found in the literature [6, 23, 24].

The literature provides a long list of potential SMB applications, including waste removal, purifications of fine chemicals, organic acids [25–28], and pharmaceutics [4], which include enzymes [29], monoclonal antibodies [30], paclitaxel (a chemotherapeutic drug) [31–33], ascomycin derivative (an anti-inflammatory drug) [34], clarithromycin (an antibiotic) [35], cyclosporin (an immunosuppressive drug) [36], escitalopram (an antidepressant) [37], biosynthetic human insulin [38–45], and many others.

All of the above commercial applications were binary separations (i.e., a feed stream is split into two product streams), although some of the feeds contain more than two components, such as those in the UOP’s petrochemical applications. This is also true with the lab-scale SMB processes reported in the literature, except for a few examples where ternary or higher order separations were required for feed streams with three or more components. For example, in the biosynthetic human insulin studies conducted by the chapter authors [38–45], the insulin product was the middle component with a fast-moving impurity (high-molecular-weight proteins) and a slow-moving impurity (zinc chloride). The authors designed a tandem SMB process using the knowledge-driven design process, explained later, that increases yield by 10%, reduces solvent consumption by two-thirds, and improves bed throughput threefold over the existing commercial batch process.

As an active research field, SMB has yet to generate a single comprehensive book written for practicing engineers. Much of the knowledge is buried in chapters of highly technical books of chromatography, spread out in technical journals, and the often intentionally obtuse published patents and patent applications, as obvious from the list of references herein. It has, however, garnered sufficient attention to be published in recent editions of the oft-cited Perry’s Chemical Engineering Handbook, which now includes a brief account of SMB. Recent technical reviews can be found in Chin and Wang (2004), who discussed a variety of SMB process configurations and associated equipment [46]; Seidel-Morgenstern et al. (2008), who provided a review of new general developments in SMB [47]; and Rajendran et al. (2009) who provided a review of the SMB design method (called the triangle design) and details on new variations of SMB operations and chiral SMB separations [24].

7.1.4 Barriers to SMB applications

An SMB process, however, requires substantially more complex design than batch chromatography. For the standard four-zone SMB illustrated in Figure 7.1, for example, given the desorbent and stationary phase, four zone lengths, four zone flow rates, and the average port movement velocity (or switching time), a total of nine design parameters, must be specified. Trial and error in the nine-parameter space is challenging and time consuming. Even though the technique has been applied industrially for more than half a century, it is still considered a research-level process design and consequently has a much smaller trained expert base than traditional chromatography.

The equipment and operation of SMB are significantly more complex than those of batch chromatogra-phy. Although considerable work has been reported in the past few decades, process optimization, operation error identification and correction, and process robustness are still being actively researched.

The continuous recycle operation in SMB, for example, results in much longer residence times for the solutes [39]. By contrast, the solute residence times in batch chromatography are easily found, and they are always shorter than the batch-cycle times. The long solute residence times in SMB can be a critical issue for biologically sensitive products.

Furthermore, the feed port in SMB is located between the two product ports. For this reason, the residence times of fast-moving solutes are different from those of slow-moving solutes [39]. The residence time of a given component also depends on when it enters the SMB during a feed-injection step. A product collected at a given time consists of molecules from different feed injections. All of these characteristics of SMB result in complex residence time distribution for each component. An innovative feed strategy is needed to control the integrity of a feed batch [44], which is critical in pharmaceutical and other regulated industries.

More important, almost all feeds from biological sources have three or more major components. Very few examples of successful separations of multicomponent mixtures have been reported in the litera-ture. The design method and splitting strategies for these mixtures are not widely known, and the SMB equipment that can split and obtain a relatively pure product from complex feed mixtures is not readily available commercially.

The rest of this chapter is focused on (i) the design methods to achieve high product purity and high yield for multicomponent separations, (ii) simulation tools to reduce the actual number of experiments for process development, (iii) SMB equipment designs for multicomponent separations, (iv) a knowledge-driven design method, which is based on intrinsic adsorption and mass-transfer parameters, and (v) examples of SMB biorefinery applications.