Sie sind auf Seite 1von 53

847

Chapter 11

Laboratory-Scale Preparative Chromatography


11.1. Introduction 11.2. Thin-Layer Chromatography 11.3. Column Liquid Chromatography 11.3.1. Flash Chromatography 11.3.2. Dry-Column and Vacuum Chromatography 11.3.3. Low-and Medium-Pressure Chromatography 11.3.4. High-Pressure Chromatography 11.3.4.1. Instrumentation 11.3.4.2. Columns 11.3.4.3. Scale-Up 11.3.4.4. Column Overload Conditions 11.3.5. Displacement Chromatography 11.3.6. Simulated Moving Bed Chromatography 11.3.7. Sorbents for Biopolymer Separations 11.3.7.1. Natural and Synthetic Soft and Semi-Rigid Gels . . . . 11.3.7.2. Diffusion-Convection Sorbents 11.3.8. Affinity Chromatography 11.3.8.1. Reactive Dyes (Biomimetic Ligands) 11.3.8.2. Immobilized Metal Ion Affinity Chromatography . . . 11.4. Supercritical Fluid Chromatography 11.5. Gas Chromatography 11.6. Countercurrent Chromatography 11.7. References 848 848 850 851 855 857 860 862 863 865 866 870 871 874 874 877 879 882 883 884 886 889 893

848

The Essence of Chromatography

11.1 INTRODUCTION

The purpose of preparative-scale chromatography is the isolation of compounds with a specified purity, in amounts suitable for their intended application [1-6]. This might include the isolation of materials for structural elucidation, for biological or sensory evaluation, for use as analytical standards, for organic synthesis, or for commerce. The scale of the process varies considerably, and includes laboratory, pilot plant, and process-scale operations. For structural elucidation and initial biological activity screening, 30-50 mg of pure substance is usually adequate; for use as an analytical standard, > 100 mg is required; for synthesis, gram quantities will likely be needed; while, at the production scale, kilograms and upwards are possibly desirable. For the first three applications, separations can be handled in the laboratory, while for pilot plant and process-scale separations, special purpose equipment and facilities are usually required. In addition, process-scale separations must be performed to satisfy economic factors, often of less significance for laboratory-scale separations. For these reasons, process-scale separations are not treated explicitly in this chapter. Process-scale chromatography is of considerable importance for the manufacture of high-value added products, such as pharmaceuticals, flavors and fragrances, and biotechnology products, where the rather high unit costs of purification are not considered prohibitive [2,6-11]. The general layout of this chapter is to proceed from simple to more sophisticated techniques based on liquid chromatography, and then discusses other separation approaches. Liquid chromatography is the laboratory-scale technique of choice for the isolation and purification of materials that cannot be handled by crystallization or simple distillation. An exception is thermally stable and volatile mixtures, for which gas chromatography is the preferred method. The advantages of other methods are indicated at the point they are introduced.

11.2 THIN-LAYER CHROMATOGRAPHY Thin-layer chromatography (TLC) is a simple and undemanding technique, suitable for the separation and isolation of compounds of low volatility in amounts up to about 1 g [4,6,12,13]. Micropreparative separations (^2-10 mg) are possible at higher resolution on high-performance layers. Limited zone capacity restricts the number of components that can be separated to about 2-5 using conventional development techniques. Scale up to larger sample loads is achieved by increasing the thickness of the layer, and the length of the plate edge along which the sample is applied. The sample capacity increases roughly with the square root of the layer thickness, but resolution is generally less for thicker layers. Laboratory-made preparative plates, in sizes up to 20 x 100 cm with sorbent layers up to 1 cm thick are possible, but these dimensions are exceptional. More typical are 20 x 20 or 20 x 40 cm plates with sorbent layers 0.5 to 2 mm thick. Layers of these dimensions, prepared from silica gel, alumina or chemically bonded sorbents with an average particle size of about 25 |xm, are commercially available.

Laboratory-Scale Preparative Chromatography

849

Their broad particle size range, however, is one factor that contributes to their low efficiency. Tapered layers, prepared with a gradual increase in layer thickness from 0.3 mm at the bottom to 1.7 mm at the top, represent one approach to improving resolution [14]. The tapered layer results in the formation of a negative velocity gradient in the direction of mobile phase migration. As a result, the lower portion of a zone moves faster than the upper portion, keeping each component focused as a narrow band. Multiple development techniques on conventional layers provide another approach to improve separation performance (section 6.6.3). Conventional and tapered layers with concentrating zones are available for convenient sample application. Sample application is a critical step in preparative thin-layer chromatography [14,15]. The sample, usually as a 5-10% (w/v) solution in a volatile organic solvent, is applied as a band along one edge of the layer. The maximum sample load for a silica layer 1.0 mm thick is about 5 mg / cm^ (lower for cellulose and reversed-phase layers). Any of the automated band applicators for analytical thin-layer chromatography are equally suitable for preparative-scale thin-layer chromatography, and are the method of choice for sample application (section 6.5). A margin of 1-2 cm is frequently left at either vertical edge of the layer to minimize uneven development. Manual sample application by syringe or pipette should be performed carefully, to avoid damaging the separation layer, resulting in irregularly shaped zones after development. Sample zones can be focused to some extent by a short development ( ^ 1 cm) with a strong solvent. Some of the problems with manual sample application can be circumvented using layers with a concentrating zone. The concentrating zone is a narrow strip of inert silica of low retention that buts up to the separation layer. The sample solution can be applied to the concentrating zone without much skill. The soluble sample components are migrated to the boundary between the concentrating zone and separation layer with a strong solvent, where they are focused into a narrow band. Preparative-scale plates are usually developed in rectangular glass tanks (e.g. 21 x 21 X 9 cm) lined with thick filter paper on all sides. The chamber is charged with sufficient mobile phase for the development step, and to soak the filter paper liner. Equilibration of the vapor phase typically requires 1-2 h. Saturated developing chambers are preferred to minimize the formation of irregular solvent fronts and developed sample bands. The plates are usually inserted in a rack that holds them in a vertical position, and allows several plates to be developed simultaneously. Ascending development typically requires 1-2 h for a solvent-front migration distance of 18 cm. Most layers for preparative-scale thin-layer chromatography contain a fluorescent indicator, which simplifies locating separated UV absorbing bands. Non-UV absorbing compounds can be located by physical or chemical techniques. Spraying with water, or exposing the layer to iodine vapor, provide a reversible indication of zone positions. Spraying one edge of the layer with a chemical reagent (a small vertical section is usually detached for this purpose), or using detectable reference compounds separated along with the sample, provide alternative approaches for locating sample zones of interest. Separated zones are marked for removal, and scraped off the plate with a spatula or similar tool. A number of devices based on the vacuum suction principal

850

The Essence of Chromatography

Table 11.1 Comparison of separation techniques for preparative thin-layer chromatography Parameter Layer thickness (mm) Separation length (cm) Separation mode Isolation method Typical sample size (mg) Number of separated bands Method of mobile phase migration Forced flow Capillary flow 0.5-2 0.5-2 18 18 Linear Linear Off-line On-line 50-300 50-150 2-7 2-5 Centrifugal 1-4 12 Circular On-Hne 50-500 2-12

are available for removing marked zones from the layer. These devices work like a vacuum cleaner, and collect the desired sample bands in a Soxhlet thimble or a glass chamber with a fritted base. The sample is separated from the adsorbent by solvent extraction or elution. Ethanol (but not methanol due to the relatively high solubility of silica in methanol) or acetone, are commonly used for direct extraction. In some cases, addition of a small quantity of water to dampen the silica gel, before shaking for several minutes with several portions of an organic solvent, improves sample recovery. An alternative approach is to add enough water to cover the adsorbent material, and to extract the aqueous suspension several times with an immiscible organic solvent. Before solvent evaporation, colloidal silica is removed by filtration through a membrane filter. Common contaminants, such as phthalate esters and poly(esters), can be removed by chromatography on Sephedex LH-20 (section 11.3.3). Rotation planar chromatography (centrifugal TLC) and forced-flow development provide higher resolution and shorter separation times for preparative-scale thin-layer chromatography (section 6.6.5), Table 11.1 [4,6,12,13,16]. These methods allow on-line detection for automated fraction collection as sample bands leave the layer. Fine-particle layers can be used in forced flow separations, to further enhance resolution with some loss in sample capacity. Although for complex mixtures, preparative-scale high-pressure liquid chromatography is generally preferred.

11.3 COLUMN LIQUID CHROMATOGRAPHY Column liquid chromatography offers a wide range of options for preparative-scale separations that differ in resolving power, sample capacity, equipment requirements, and operating costs, Table 11.2 [1,4-6]. A distinction is possible in terms of the typical operating pressure. Classic or open column systems rely on capillary forces for transport of the mobile phase through the column. Since these forces are weak, the particle size must be large, and low efficiency, modest column lengths, limits the separation power. In addition, the separation time is compromised by the low flow rates achieved. Among the virtues of these columns are low costs, minimal equipment requirements, and high sample capacity. Stationary phase reuse is usually unnecessary, since it has a small influence on the cost of purification. For those cases that require

Laboratory-Scale Preparative Chromatography Table 11.2 Typical operational parameters for preparative-scale column liquid chromatography Type Open column Flash chromatography Low pressure Medium pressure High pressure Particle size (|jLm) 63-200 40-63 40-63 15-40 5-30 Inlet pressure (atm) 1 1-2 1-5 5-20 > 20 Flow rate (ml/min) Ts 2-10 1-4 3-16 2-20

851

Sample loading (g) 0.01-100 0.01-100 1-5 0.05-100 0.01-1

increased resolution, shorter separation times, or increased automation, the particle size must be reduced, and the column inlet pressure simultaneously increased, to maintain the optimum range of mobile phase velocities. These improvements can be obtained incrementally, as indicated in Table 11.2. Although a distinction is made between flash chromatography, low pressure, medium pressure, and high pressure operation, these are no more than regions of an operating continuum. The conditions indicated in Table 11.2 are considered typical and the boundaries flexible. One distinction that affects method selection is operating cost and equipment requirements. These increase in descending order in Table 11.2, and restrict availability in different laboratories. 11.3.1 Flash Chromatography Flash chromatography is widely used for laboratory-scale fractionation of mixtures from organic synthesis, or for analysis, when only a modest increase in resolution over open column systems is required [4-6,17-19]. These techniques employ short columns packed with particles of an intermediate size (40 - 63 ixm) combined with accelerated solvent flow achieved through modest pressure or suction. Compared with open column chromatography: separations are obtained in less time; isolated compounds are often purer, because resolution between bands is increased; and compounds that are degraded or altered during chromatography, are recovered in higher purity, because of the shorter contact time with the separation system. The main applications of flash chromatography are purification of synthetic products, isolation of target compounds from natural products, the simplification of mixtures prior to high resolution preparative (usually) liquid chromatography, and the fractionation of complex mixtures into simpler groups for analysis. Its primary virtue is low cost, since virtually no special equipment is required, and the stationary and mobile phases are inexpensive enough to be discarded after a single use, or can be recycled. Resolution is less than that obtained by medium and high-pressure liquid chromatography, but the operational costs and equipment needs are greater for these techniques. Flash chromatography is often employed as a preseparation technique to remove particle matter and sample components that are either weakly or strongly retained on the separation column in medium and highpressure liquid chromatography. This allows higher sample loads to be separated under more selective separation conditions, and avoids column contamination and

852

The Essence of Chromatography

regeneration problems. The production costs of the isolated products are thus rendered more favorable. Columns for flash chromatography are prepared as follows. A glass column of suitable length containing a small glass wool plug and a layer of acid washed sand, or with a glass frit at its base, is partially filled with sorbent using the dry packing or slurry packing technique. Incremental addition of the sorbent followed by tapping of the column with a hard object generally gives better results for dry packing than bulk filling of the column. After packing, the column is freed from trapped air and further consolidated by forcing several column volumes of a weak solvent through the column bed, until no further air bubbles are seen exiting the column and the bed is stable. It is difficult to pack wide diameter columns (> 5 cm) by dry-packing procedures, and in this case slurry packing is nearly always used. The column is partially filled with a small volume of weak solvent. A dilute suspension of the sorbent in the same solvent is added slowly, in increments, with excess solvent intermittently drained away. Periodic pressurization of the sorbent bed is used to aid consolidation. Tapping the side of the column with a hard object during packing is not generally recommended. The sample is added to the column in a small volume of solvent or adsorbed to a small amount of packing material. Finally, a thin layer of glass beads, acid-washed sand, or other inert material, is added to the top of the column to prevent disturbance of the column bed by solvent added for elution. The amount of free space above the packed bed must be sufficient to hold a volume of solvent equivalent to the fraction size collected, or a solvent reservoir must be inserted between the column and the pressure regulation valve [20]. The flow rate is adjusted to about 5 cm/min by application of pressure from a compressed gas cylinder or air pump, and controlled by the regulation valve. Pressures employed are typically less than 1 to 2 atmospheres with the various parts of the apparatus. Figure 11.1, held in place by springs, clamps, or screw-thread connectors. It is a wise precaution to use plastic coated glass columns, or a safety shield, to minimize the possibility of accidents. The column should not be allowed to run dry during the elution sequence. Radial compression columns have been used with large-scale flash chromatography systems [21]. Because of the limited operating pressure, columns are rarely longer than 30 cm, and 10 to 15 cm is recommended, unless longer columns are required to provide additional resolution. At a mobile phase velocity of about 5 cm/min, well-packed columns are expected to provide about 5 - 2 0 theoretical plates per cm of bed height, depending on the column packing density, and the quality of the sorbent material. Silica, and to a lesser extent alumina, are the most common stationary phases used for the separation of low molecular mass organic compounds. Chemically bonded silica sorbents are used for the separation of polar organic compounds in the normalphase and reversed-phase modes. Wide-pore, chemically bonded sorbents, are used for the separation of biopolymers [18,22]. Some separations require specially prepared stationary phases, such as silica gel impregnated with silver nitrate for the isolation of unsaturated compounds capable of forming charge transfer complexes with silver [23] (section 10.6.1), or silica and chemically bonded phases coated with cellulose tris(3,5-

Laboratory-Scale Preparative Chromatography

853

Air inlet adapter

Solvent

Glass frit

Collection vessel Figure 11.1. Apparatus for flash chromatography.

dimethylphenyl carbamate) to isolate single enantiomers from racemic mixtures [24] (section 10.4.2). The sample is usually added to the column in a small volume of a weak solvent, and the solution forced into the sorbent bed forming a narrow sample band. For samples of low solubility in weak solvents, the sample is taken up in a strong solvent and added to a small amount of column packing or other inert support. The solvent is then stripped from the slurry under vacuum to produce a dry free-flowing powder (1-2 g sample / g sorbent) that can be added to the top of the column. It is important that the sample is completely dry (high vacuum used to remove last traces of solvent), and free of lumps, to obtain symmetrical separated bands. If the sample layer is relatively long compared

854

The Essence of Chromatography

Table 11.3 Approximate sample-loading conditions for flash chromatography (density of silica ^ 0.45 g/ml) Column diameter (cm) Amount of silica gel (g) Sample loading for aI particular TLC resolution (g) A R F > 0.2 ARF > 0 . 1 0.04 0.16 0.36 0.6 1.0 Typical fraction volume (ml) 5 10 20 30 50

(a) Isocratic elution (bed height = 15 cm)1 1 5 0.1 20 0.4 2 45 3 0.9 80 1.6 4 2.5 130 5 (b) Stepwise gradient Column Diameter (cm) 3 4 6 8 10 14 elution (bed height Amount of siUca gel (g) 30 55 125 250 350 700 = 10 cm) Sample loading (g) 1-3 3-8 8-35 35-60 60-80 80-150

Typical fraction volume (ml) 50 - 100 100 - 200 200 - 300 200 - 300 300 - 500 300 - 500

to the column bed length, then a stepwise solvent gradient must be used for elution to minimize band broadening. There are no simple relationships between the sample amount that can be separated, the dimensions of the sorbent bed, and the volume and number of collected fractions. The sample capacity depends on the ease of separation of neighboring zones, the sorption capacity of the sorbent, and the method of elution. Wider columns and sorbents of high specific surface area are used to increase the sample capacity. Table 11.3. For stepwise gradient elution, it was assumed that the sample is separated into fractions of different polarity to estimate sample loads. Even for difficult samples, it is often more productive to use column overload conditions, combining fractions containing pure materials, and recycle those containing mixtures. Flash chromatography may lack the resolving power needed to separate the components of interest. In this case, a higher resolution technique, such as medium or high-pressure liquid chromatography is a better choice, perhaps using flash chromatography to isolate fractions containing the components of interest from other sample components. Thin-layer chromatography is widely used to optimize separation conditions for silica gel flash chromatography. For isocratic separations, a mobile phase that provides a Rp ^ 0.35 for the zone of interest is chosen. If several zones are to be separated, then the solvent strength is adjusted such that the center zone has a Rp ^ 0.35. If all zones of interest are well separated from each other and from impurities (ARp > 0.2), then the solvent strength is adjusted so that the most retained zone of interest has a Rp ^ 0.35. For fractionation and large sample loads, it is critical that the most selective solvent composition for the separation is used. This can be quickly

Laboratory-Scale Preparative Chromatography

855

identified from the Prisma model, a guided trial and error procedure using thin-layer chromatography (section 6.7.1). The same process can be used to identify solvent compositions suitable for recovery of sample zones in order of increasing polarity by stepwise gradient elution. For samples of wide polarity, a useful gradient is to start from a weak base solvent, such as hexane, and add to this various volume increments of a strong dipolar solvent (such as ethyl acetate, dichloromethane, chloroform, or acetone), terminating with the strong dipolar solvent. Then continue adding volume increments of a strong hydrogen-bond solvent (methanol, ethanol, 2-propanol) to the dipolar solvent, terminating with the strong hydrogen-bond solvent. Screening of the separation by thinlayer chromatography allows the solvent gradient to be trimmed and optimized to the requirements of individual separations. Predicting the number of fractions required at each step remains quite arbitrary, and is best conducted by monitoring the composition of each fraction as it is collected. When incrementing the composition of strong solvent in a binary mobile phase for silica gel, it is important to note that the solvent strength for the mixture has a steep curved profile (section 4.3.6.2). For compositions containing low volume fractions of strong solvent, the volume fraction of strong solvent should be incremented by small changes resulting in relatively large changes in retention, for example, 1 % , 3 % , 5 % , 1 0 % (v/v). At higher volume fractions of strong solvent, changes should be larger to produce a significant change in retention, for example, 30 %, 40 %, 60 %, 80 %, 100 % (v/v). Water-soluble compounds, including biopolymers and easily ionized compounds, are generally isolated by reversed-phase chromatography. Optimization of solvent composition by thin-layer chromatography is possible, but predictions may be unreliable owing to differences in the sorption properties of the sorbents used to prepare the layers and those used for flash chromatography. A better solution is to pack a short (10 cm) metal column with the sorbent for flash chromatography and use high-pressure liquid chromatography to optimize separation conditions. Ideally, for isocratic elution a solvent composition should be chosen that provides a retention factor of 2 - 3 for the component of interest, or the most difficult to separate components of a mixture. For mixtures of wide polarity, stepwise solvent gradients are easily constructed and optimized by the same approach. Separations by flash chromatography can be monitored on-line using UV-visible, refractive index, or evaporative light-scattering detection. Off-line monitoring, however, by collecting fractions that are subsequently combined based on the similarity of their composition is more common [4,21,25]. Thin-layer, gas or liquid chromatography, electrophoresis, bioassays, immunoassays, and spectroscopy (e.g. infrared and nuclear magnetic resonance) are suitable techniques to identify fractions that can be combined. Most of all, thin-layer chromatography is used because it is quick, portable, inexpensive and generally adequate for the task. 11.3.2 Dry-Column and Vacuum Chromatography Dry-column chromatography is a variant of preparative-scale thin-layer chromatography with similar resolution but a higher sample capacity. A glass column or Nylon tube

856

The Essence of Chromatography

is packed with a thin-layer sorbent, usually silica gel, to a height of 10 to 15 cm. Sample is added as a concentrated solution or preadsorbed onto a small amount of sorbent. Separation is achieved by developing with a suitable volume of solvent to reach the lower end of the bed. Suction at the bottom of the column and/or slight overpressure at the top may be required to supplement capillary forces in moving the mobile phase down the column. Separated bands are removed by extrusion, slicing (if a Nylon column is used), or by digging out, and the products freed from the sorbent by solvent extraction. The separation is fast, requires little solvent, and provides higher resolution than classical column techniques, owing to the use of sorbents with a smaller average particle size. It is suitable for the recovery of small quantities of material, since the loading capacity is only about 0.2 to 1.0 % w/w of the sorbent used depending on the difficulty of separating the bands of interest. Thin-layer chromatography provides a suitable technique for method development in most cases, although significant differences in separations can arise for mixed solvents, particularly when the solvent components differ in polarity and/or volatility. These differences result from the absence of a vapor phase in the dry-column technique. Nylon columns can be more difficult to pack than glass columns, particularly when longer lengths are used, but Nylon columns are easier to section, and allow colorless bands to be observed with a UV lamp. Glass columns, built up of segments connected by ground glass joints, simplify the extrusion process. Dry-column chromatography is not a widely used today. Preparative-scale thin-layer chromatography or flash chromatography is generally preferred. Although separations are fast, the recovery of separated zones is slow and labor intensive compared with elution methods. Vacuum chromatography can be taken to mean the operation of a short column under suction to accelerate solvent migration. Either a short column, or a Buchner filter funnel fitted with a glass frit, is dry-packed with sorbent [26]. The sorbent bed is consolidated by tapping the side of the column during filling, and pressing the top layer of the sorbent bed with a flat object, such as a stopper, while suction is applied at the other end. Consolidation is completed by releasing the vacuum and pouring a solvent of low polarity over the surface of the bed followed by restitution of the vacuum. If the column is packed correctly, the solvent front will descend the column in a horizontal line, otherwise the column should be sucked dry, repacked, and tested again. When all the solvent has passed through the column, residual solvent trapped between particles is removed by suction. A solution of the sample in a suitable (weak) solvent or preadsorbed onto a small amount of sorbent or inert material, such as Celite, is applied to the top of the column. The sample solvent, if used, is sucked gently into the column packing. A piece of filter paper with the same diameter as the inside diameter of the column, or funnel, is placed on top of the packed bed to prevent disruption of the bed during solvent addition. The column is eluted with appropriate solvent mixtures of gradually increasing solvent strength. Between solvent applications, the column is sucked dry, and the eluent collected in test tubes or round bottom flasks. A multiport manifold allows sequential fraction collection without having to disassemble the apparatus after each fraction is collected.

Laboratory-Scale Preparative Chromatography

857

Pre-column

Column

^ ^ ^ ^ ^
Detector Recorder

Solvents

'-OOP injector Waste

Fraction collector

Figure 11.2. Apparatus for low- and medium-pressure preparative-scale liquid chromatography. (From ref. [6]; Academic Press).

Vacuum chromatography is simple, rapid, and convenient. Optimum sample loads are similar to flash chromatography. However, it is not unusual to use sample overload conditions to separate simple mixtures by stepwise gradient elution, or to simplify mixtures for further separation. Under these conditions the sample loads may reach 10 % (w/w), or even higher, of the bed mass. 11.3.3 Low- and Medium-Pressure Chromatography Low-pressure liquid chromatography uses similar particle-size sorbents to flash chromatography, and its operation is only a little more complex. Column inlet pressures are similar, but a pump is used to maintain a constant mobile phase velocity, and sample introduction is usually by injection. On-line detection is frequently used together with an automated fraction collector. Medium-pressure liquid chromatography uses a similar experimental arrangement, Figure n . 2 . It affords higher resolution and shorter separation times by using sorbents of a smaller average particle size, and higher mobile phase velocities, by operating at higher inlet pressures (see Table 11.1). These techniques are suitable for use with soft gels employed in the purification of biopolymers (section 11.3.7). As well as some natural and synthetic polymer gels, that provide different selectivity to conventional inorganic oxide-based sorbents, for the purification of low molecular mass organic compounds [3-6,27]. Columns for low- and medium-pressure liquid chromatography are made from strengthened glass with a plastic protective coating. The wall thickness and column internal diameter determine the maximum operating pressure. Narrow-bore columns (e.g. 1 cm) can be used at pressures up to about 50 atm., while wide-bore columns (> 10 cm) are restricted to pressures of a few atm. Columns are available in lengths from about 10 cm to 1 m (20-25,30-35 and 44-50 cm are the most popular) with internal diameters from 1 to 10 cm (1-1.5, 2.5-4, 3.5-6.5 cm are the most popular). The wider

858

The Essence of Chromatography

bore columns contain several kilograms of sorbent, and can separate sample loads of 15-100 g, depending on the ease of separation under concentration overload conditions. For narrow-bore columns, sample loads < 3 g are common, corresponding to a sampleto-sorbent mass ratio of about 1:25 for difficult separations, and lower ratios for easy separations. Columns can be coupled in series to create greater lengths, when needed, to improve resolution. For low-pressure separations, inorganic oxides and chemically bonded sorbents of 40-63 |xm, and for soft gels 25-100 |xm, are typically used. For medium-pressure separations inorganic oxides, chemically bonded phases, and rigid or semi-rigid porous polymer sorbents with an average particle size of 15-25 or 25-40 jxm are typically used. Larger particle sizes are appropriate for long or series-coupled columns, operated at modest pressures. Packed columns can be purchased, or prepared in the laboratory by dry or slurry packing, without the need for special equipment [4,6,27-30]. Dry packing is generally used for inorganic oxide sorbents. The tap-and-fill method (section 4.5.2.1) is suitable for inorganic oxides with an average particle size > 20 |xm. Packing under nitrogen pressure is required for 15-|xm particles. The column with a frit at its base is clamped vertically, and a reservoir attached at the top. The sorbent is introduced into the column by means of a funnel until the column is completely filled and the reservoir contains excess sorbent equivalent to about 10 % of the column volume. Figure 11.3. The column is not vibrated or tapped during the filling process. The system is connected to a nitrogen cylinder and 10-atm pressure applied (with the column outlet open) until the packing height reaches a constant value. The cylinder valve is closed, and the pressure allowed to bleed away. The reservoir is then removed, and the filled column conditioned by passing mobile phase through it until no more bubbles appear in the eluent, and the pressure stabilizes. Slurry packing techniques are used to pack chemically bonded phases (silica gel can be packed in this way as well). The slurry is prepared by suspending the sorbent in a solvent that is a good wetting agent. The slurry is slowly poured into the column to completely fill the column and partially fill a short precolumn attached to the separation column. The sorbent is allowed to consolidate for about 1 h and then connected to the eluent pump at the precolumn end. Solvent is pumped through the column, increasing the flow rate and pressure in steps, until the column operating pressure is reached, and the mobile phase flow rate stabilizes. The precolumn is then removed, and the column inlet fitting attached to the column. The column is then conditioned by passing mobile phase through the column to replace the packing solvent. Soft gels are first swollen in the mobile phase, and the slurry formed slowly poured into the column with stirring or tapping. Excess solvent is drained away, and the process repeated until the column is completely filled. The column is connected to the pump, and the bed consolidated by increasing the flow rate and pressure in increments, until the operating conditions are reached. Voids, which may appear at the inlet end of the column, are filled by adjusting the inlet fitting to the level of the column bed. Alternatively, further amounts of packing (or glass beads) are added to the column to restore the bed height to the desired position.

Laboratory-Scale Preparative Chromatography

859

trr:^

Filling

Compacting

Figure 11.3. General procedure for dry packing medium-pressure, preparative-scale, liquid chromatography columns. (Fromref. [6]; Academic Press).

The operating pressure (< 50 atm) and flow rates (5-180 ml/min) for the full range of columns used in medium-pressure liquid chromatography can be generated by low cost reciprocating-piston pumps with exchangeable pump heads. All connections between the pump, injector, column, and detector are made with Teflon tubes, or similar materials, having specially-designed plastic end fitting to simplify assembly of the equipment. The sample is usually injected on-stream using a valve injector or, for large sample volumes, through the pump. Loop injectors are suitable for injecting sample volumes up to about 100 ml. For low-pressure columns, the sample can be added to the head of the column by syringe or pipette, and the solution allowed to drain to the top of the column bed. For wide-bore columns, the design of the column inlet and end fittings is crucial to ensure an even distribution of the sample across the column diameter. This can be achieved with a fritted disc and plunger assembly (adjusting the plunger position allows column voids to be filled). Figure 11.4. A fraction collector is normally used instead of or in conjunction with a flow-through detector (a short path length UV detector or refractive index detector) for sample collection. Fractions are collected in test tubes (240 x 20 ml) or larger vessels (120 x 50 or 48 x 250 ml) on a time, volume, or peak threshold basis. In the absence of on-line monitoring, thin-layer chromatography can be used to group similar fractions.

860
PTFE tubing

The Essence of Chromatography

Threaded spindle Fixing screw

Flow adaptor (adjustable length)

7P

Screw coupling with 0-ring seal

PTFE plug seal Glass body (4 mm wall thicliness) PTFE sintered plate (porosity: 12, 46, or 100 pm)

Figure 11.4. Expanded view of a column end fitting and conical shaped glass column for low- and medium-pressure, preparative-scale, liquid chromatography.

Separations should be performed with the most selective mobile phase. Suitable mobile phases are usually identified by thin-layer chromatography or analytical highpressure liquid chromatography [31]. The surface area of thin-layer sorbents is higher than typical column sorbents ( ^ 2 x). Mobile phases of suitable strength for preparativescale column separations, therefore, will have an Rp < 0.3 by thin-layer chromatography. Thin-layer chromatography can be used to select gradient elution conditions according to the location of zones in a standardized separation system [32]. Transfer of conditions from analytical high-pressure liquid chromatography to preparative-scale column chromatography is usually straightforward unless the difference in properties of the two column sorbents are significant. In general, normal-phase separation systems are preferred, since volatile organic solvents are easier to evaporate than aqueous solutions to recover the isolated compounds. Adequate solvent purity is an important consideration, since low level non-volatile impurities become concentrated during the evaporation step, and will contaminate the isolated compounds. 11.3.4 High-Pressure Chromatography Higher operating pressures allow preparative-scale separations with improved resolution in a shorter time using smaller diameter particles, set against higher operating costs

Laboratory-Scale Preparative Chromatography Table 11.4 Some common terms used in preparative-scale chromatography Parameter Cut points Cycle time Loading factor Overload Phase ratio

861

Production rate Recovery yield Recycle Saturation capacity

Specific production Throughput

Description Times, mobile phase volumes, or threshold values of the concentration when fraction collection begins or ends. Time separating two consecutive injections. Ratio of the sample size to the column saturation capacity. Operation of the column under nonlinear isocratic conditions resulting from the separation of large sample sizes. The ratio of the fraction of the column volume occupied by the stationary and mobile phases (F = [1 - e] / e, where 8 is the total porosity of the column). This definition is different from that used in analytical chromatography (section 1.4). Amount of product with a specified purity obtained per unit time. Ratio of the amount of the desired compound collected in the product fraction to the amount of sample injected on the column. A chromatographic process in which the fraction of the mixed zone is re-injected on-line, without intermediate collection and pooling. Amount of component required to saturate the stationary phase. In adsorption, amount of component required to form a monolayer on the surface of a unit amount of adsorbent. In ion exchange, amount required to exchange all the ion sites on the stationary phase. Volume of mobile phase consumed per unit amount of product produced. Amount of sample introduced into the column per unit time.

and the need for more complex instrumentation [1-6,11,33-36]. Instruments for analytical separations can be adapted for use with semipreparative columns with internal diameters < 2.5 cm containing about 25-65 g of silica-based sorbent. These columns are operated at flow rates between 15 and 30 ml/min and are suitable for the purification of 50-100 mg of sample per injection. Special purpose instruments are required for operation at the higher flow rates required by larger diameter columns. These are usually only found in laboratories that specialize in preparative-scale separations, or have a reoccurring need for purification of difficult samples. A 30-cm column with an internal diameter of 10 cm, contains about a kilogram of sorbent, and can separate 10-lOOg of sample per injection. The actual sample load is strongly dependent on the ease of separation, and can be an order of magnitude higher for simple separations (a > 5), or lower for difficult separations (a < 1.3), when using sample overload conditions. Optimum flow rates are between 250 and 600 ml/min. Some common terms used in preparative-scale liquid chromatography are summarized in Table 11.4. The production rate, specific production, or the recovery yield provide suitable objective functions to judge the relative success of individual methods. For efficient use of the separation system, the production rate and the recovery yield should be maximized. Invariably, this results in operating the column in an overloaded condition. Unfortunately, column operation under nonlinear conditions is complex, and optimum conditions are not as easy to predict as the less demanding, although less powerful, scale-up approach. To scale up an analytical separation, the same column packing, column length, and mobile phase velocity are used, and the column diameter increased

862

The Essence of Chromatography

to achieve the desired sample capacity. The analytical separation should be optimized to maximize the separation factor between critical peaks to allow higher column loadings before the bands overlap. 11.3.4,1 Instrumentation Wide bore columns operated at high production rates place different demands on the specification of instrument performance compared with analytical separations on narrow bore columns. For some applications, analytical instruments are easily adapted to preparative-scale operation, but in general, for all but occasional use and relatively small sample sizes, different equipment is required. Pumps for mobile phase delivery are required to generate isocratic or gradient mobile phase compositions at high flow rates and modest pressures, compared with analytical separations. Analytical pumps with a maximum flow rate of 10 ml/min are suitable for use with semipreparative columns, although the separation time may be compromised by the less than desirable flow rates. Some analytical pumps have exchangeable pump heads that extend the flow rate range to higher values, and are a better choice for dual analytical/preparative applications. For laboratory-scale preparative chromatography with a wide range of column types, a pump with a flow rate limit of at least 100 ml/min and a pressure limit of 200-250 atm is a reasonable compromise. Process-scale columns (I.D. > 10 cm) require flow rates of at least 2000 ml/min at operating pressures < 150 atm. Conventional rotary injection valves with external loops are suitable for the relatively small sample sizes typically used with semipreparative columns. For sample volumes of several milliliters, it is easier to pump the sample onto the column using the eluent pump, or preferably, a dedicated injection pump. Wide bore columns require a header device to distribute the sample and mobile phase over the cross-section of the column at the inlet, and as a collector at the exit [30,37-40]. The header consists of a frit and a distributor, and is usually part of the column inlet and exit fittings. The frit serves to prevent movement of the packed bed out of the column, and conceivably plays some role in the flow distribution. The distributor is typically a flat porous disc or a non-penetrable disc with branching channels, usually with a void space above it, to promote rapid radial flow. The distributor is designed to ensure that all streamhnes from the point of injection to the column packing (or from the collection region to the column exit) have identical transit times and a flat radial velocity flow profile. Path length and flow velocity heterogeneity at the head or bottom of wide bore columns can be a significant source of band broadening. Significant viscosity differences between the sample solution and the mobile phase can affect the shape of separated bands through viscous fingering. This results from the penetration (fingering) of the less viscous solution into the more viscous solution, as the two regions penetrate the packed bed. A wide linear response range and high flow rate capability, are desirable features in a detector for preparative-scale chromatography. Problems with temperature stability (refractive index) and pressure pulses (UV detector) are less important because of the high sample concentrations present in the mobile phase. Some analytical detectors have interchangeable flow cells, which allow replacement of the analytical cell with one of

Laboratory-Scale Preparative Chromatography

863

shorter path length to reduce sensitivity. Preparative flow cells usually have wider bore inlet and outlet connections to accommodate high mobile phase flow rates. For the same reason, wide bore capillary tubing (0.4-0.6 mm I. D.) is used for column connections, instead of the narrow bore tubing used with analytical columns. It is often convenient to operate analytical detectors with a low-dead volume flow splitter, so that only a few percent of the total column eluent passes through the detector. This is essential if a mass spectrometer is used for mass-selective detection of target compounds for automated fraction collection [11,41-44]. A mass spectrometer is an expensive detector for preparative-scale chromatography, but has a number of useful features in fully automated systems, where UV detection may lack the specificity to identify selected products for collection. With mass-directed fraction collection, no post-purification screening and pooling is required to identify product fractions, and tighter control is maintained over fraction size and number of collected fractions. The later is critical for unattended operation, if excessively large fraction beds are to be avoided. 11.3.4.2 Columns Wide bore columns must be heavily walled compared with analytical columns if they are to be used with small diameter particles, or alternatively, they must be operated at significantly lower inlet pressures, reflecting their lower bursting pressure [2,45,46]. Some stainless steel columns with an internal diameter of 5 cm, for example, have a maximum operating pressure of about 100 atm. Compression fittings may not seal adequately to wide diameter columns at high pressures, and flanged end fittings are generally used. They are either bolted or fastened with special collars, which can withstand high pressures. Dry or slurry packing of large diameter columns is more difficult than those of conventional diameters owing to the scale of the packing process, and the high flow rates and pressures required to consolidate the packed bed [36,45-48]. Anecdotal evidence indicates that columns of large internal diameter are less stable than analytical columns. This is attributed to the packing density being too low at the beginning of column operation. These problems are associated with the application of insufficiently intense vibration of the bed during dry packing, or an insufficiently high flow velocity, and hence packing pressure, during slurry packing. During dry packing, the coarse particles tend to segregate close to the wall region, resulting in a higher local-mobile phase velocity at the wall. The packing density for slurry packed columns is higher at the wall, while the core region is more permeable. Inefficient dissipation of frictional heat generated by the flow of mobile phase in wide bore columns can be a problem, since the stainless steel mantle and packed bed have different thermal expansion properties. Through repeated thermal expansion and contraction, the packed bed structure changes, and a void is produced at the top and along the walls of the column. The preferred solution to these problems is to continuously adapt the column volume to the changing bed volume by radial or axial compression. Figure 11.5 [45-48]. Axial compression can be exerted from above with an adjustable column head, or from the bottom with a fixed or floating piston. Radial compression is achieved by applying

864

The Essence of Chromatography

^'.

4m S^ Um
^msh

4<^

c^pooc I<^ OQOQ^ K<^\ voooc

Figure 11.5. Schematic diagram of continuous compression columns for high-pressure Preparative-scale Hquid chromatography. The arrows represent the force directions. A = radial compression, B = dynamic axial compression, and C = annular expansion.

hydraulic pressure between a flexible mantle and rigid container. A combination of axial and radial compression is possible by inserting a wedge or plunger in the center of the column top, which is screwed down, as the column conditions require. For columns with internal diameters < 2.5 cm, continuous bed chromatographic columns with a biporous structure (monolithic columns, section 4.2.7) represent a recent and promising development [49-51]. Dynamic axial compression columns consists of a cylinder, which is approximately three times longer than the column length desired, and a sliding piston with appropriate fittings to prevent leaks between the piston and the barrel [2,45,46,48,52]. Column bodies with internal diameters of 2.5, 5 and 7.5 cm are available for laboratory use, and in various diameters up to 80 cm, for process-scale applications. To prepare a column, the cylinder is first filled with a slurry of the packing material, the injection head bolted to the cylinder, and the piston slowly raised forcing the solvent out of the cylinder. A frit in the injection head fitting retains the packing during the compression phase. When the process of compression is complete, the cylinder is locked in position to prevent it from being pushed backwards, if the mobile phase inlet pressure exceeds the piston pressure used to consolidate the bed. An alternative design employs a floating piston, which uses the mobile-phase inlet pressure to continuously consolidate the column packing. After the separation, the injection head is unbolted, and the packing pushed out by the piston. Alternatively, a spring compression system that allows the column to be removed from the piston, and the system reused to pack another column, can be used [52]. The key parameters that affect the production of homogeneously packed and stable column are the packing material, the choice of slurry solvent, the slurry concentration, and the packing pressure [48,53-57]. There are no universal packing methods, and general approaches are adjusted to the requirements of individual packing materials.

Laboratory-Scale Preparative Chromatography

865

Radial compression columns consist of a plastic cartridge closed at both ends by a frit and a stream distributor [58,59]. The cartridge is dry packed with packing material and inserted into a steel cylinder forming a seal at both ends. A hydraulic fluid is introduced under pressure between the plastic cartridge and the steel cylinder, producing compression of the bed. For optimum performance, the hydraulic pressure should be about 150 % of the inlet pressure required to obtained the desired mobile phase velocity for the separation. For coarse particles, the separation performance is similar to slurry packed analytical columns prepared from the same packing material. In current practice, dynamic axial compression columns are generally preferred to radial compression columns. 113.4.3 Scale-Up The scale-up approach to preparative high-pressure liquid chromatography is quite straightforward, since the optimum separation conditions for the preparative-scale separation are extrapolated from a separation developed on an analytical column [1,33,34,60-63]. The analytical separation is optimized to maximize the separation factor between product peaks and product peaks and impurities. In addition, the retention factor for the product peaks is made as small as practical (k < 3) to conserve mobile phase, and to reduce the cycle time to facilitate multiple injections in the preparative-scale separation. Where possible, the mobile phase should contain components that are easily removed from the products to minimize the work-up of the collected fractions. In most cases, evaporation is used to separate products from the mobile phase. Consequently, the presence of low volatility additives and buffers are undesirable, and should be replaced by volatile components, where possible. Solubility of the sample in the mobile phase is an important consideration, since this may limit sample throughput. The higher solubility of organic compounds in organic solvents often results in normal-phase separations being preferred. It is notable that the preferred conditions for preparative-scale separations are not necessarily the same, as the conditions preferred for analysis. Once separation conditions are established, the maximum sample load is determined. To minimize sample waste this is determined for the analytical column. The sample size (volume or mass) is increased in increments until the product bands are just touching, but still separated from impurity peaks [60,64-66]. This corresponds to the upper region of the linear portion of the sorption isotherms for the products, or to a slight overload condition. For difficult separations, partially overlapped peaks may have to be accepted to obtain acceptable sample throughput, in which case, product purity is maintained by collecting the outside portions of the product peaks and recycling the middle portion. Figure 11.6. To scale-up the separation conditions to preparative chromatography, a preparativescale column of the same length, but with a wider bore than the analytical column, and packed with the same (or similar) stationary phase material, is selected. The operating conditions for the preparative-scale column are estimated from the scaling factor, which is determined by the ratio of the volume of stationary phase in the two columns. This

866

The Essence of Chromatography

Cut Point A B C

= 0.6

^ 100% Compound 1 88% Compound 1 12% Compound 2

Volume

100% Compound 2 88% Compound 2 12% Compound 1

Figure 11.6. Peak shaving technique for collection of pure product when the separation selectivity is inadequate for acceptable sample throughput.

is equivalent to the ratio of the square of the column diameters. The sample load (both mass and volume) and mobile phase velocity required for equivalent separation conditions on the preparative-scale column are obtained by multiplying the analytical column conditions by the scaling factor, Figure 11.7. The irregular shaped peaks in the preparative-scale separation are due to detector saturation and not column overload. In those cases where the available instrumentation does not allow use of the calculated flow rate, the highest available flow rate is used. In this case, a reduced production rate owing to lower column efficiency, and longer separation times has to be accepted. 11.3.4.4 Column Overload Conditions The scale-up approach to preparative chromatography is based on the theory for linear chromatography. Since there is a limit to the column size that can be prepared with the same efficiency as analytical columns, the only practical way to further increase sample throughput is to overload the column. A column is considered overloaded when retention factors obtained at low sample concentrations, change by more than 10% as the sample size is increased. A column might be overloaded either by increasing the sample concentration while maintaining a constant injection volume (concentration overload), or by increasing the injection volume while maintaining a constant sample concentration (volume overload) [64-66]. Under volume overload conditions, the sample concentra-

Laboratory-Scale Preparative Chromatography 0.15

867

ANALYTICAL SEPARATiON ,o
0.10

PREPARATiVE SEMRATION

ffi

<

ao5

^ C D44

< 2 H

UL. y^
2 4 6 8 K) MINUTES SAMPLE: FLOW: COLUMN: 40 pg 1.5 mL/min. 4.6X250 m m 2

\^
4 SAMPLE: FLOW: COLUMN:

\ ^ \
6 6 80 mg 20 mL/mia 16X250 mm

M
K) MINUTES

Figure 11.7. Application of the scale-up approach for the isolation of bilirubin isomers by high-pressure column liquid chromatography.

tion is kept constant and is confined to the linear range of the adsorption isotherm, but the volume injected is very large. For rectangular injection pulses of constant height (concentration) and increasing width (volume), the elution band becomes higher and wider. Ultimately, it becomes flat topped but remains symmetrical. Figure 11.8 [65]. For concentration overload, a small sample volume is injected, but its concentration exceeds the linear range of the adsorption isotherm. Accordingly, the band profile broadens and becomes assymetric; in the case of Langmuir type isotherms, the profile becomes close to a triangle, with an almost vertical front and a slanted tail. The production rate in the elution mode increases with increasing volume and/or concentration of the sample injected on the column [66,67]. Since the bandwidth of all components increases in both cases, there is a limit to the sample size that can be used effectively. When the sample size increases from the very low levels typically used in analytical applications, the recovery yield remains constant, and the production rate increases linearly with increasing sample size. When the band of the compound of interest touches its neighbor, the recovery yield starts to decline with increasing sample size, since the wings of the elution band must be clipped to eliminate contamination. Eventually, a maximum value for the production rate is reached, but since the separation is carried out under nonlinear conditions, optimum separation conditions are no longer predictable from data obtained under analytical conditions.

868

The Essence of Chromatography

ANALYTICAL SAMPLE
^^

A
VOLUME OVERLOAD

CONCENTRATION OVERLOAD
1

lri

1 / 1/

1 /

1 1

\J

Figure 11.8. Development of peak profiles during migration along the column for analytical and overload samples. (From ref. [65]; Elsevier).

The properties of chromatographic bands under nonUnear conditions are difficult to predict and to model compared with linear chromatography [65-77]. At high sample concentrations equilibrium isotherms are nonlinear, the concentration in the stationary phase increasing either faster or more slowly than the concentration in the mobile phase. Accordingly, solutes at different concentrations tend to move along the column at different velocities, and either the front or rear of the band will become steeper. Peaks often have sharp fronts and long, diffuse rear boundaries. The position of the high concentration front recedes with increasing sample size, while the end of the peak tail holds the same position, identical to the peak position observed in linear chromatography. Solute bands compete with each other for access to the adsorption sites on the stationary phase, resulting in interactions between bands and profile changes during their migration along the column. The amount of one component adsorbed at equilibrium depends on the concentration of all system components [66,78-81]. Compared with the results for a single component, displacement and tag-along effects

Laboratory-Scale Preparative Chromatography

869

are observed. These effects control the shape of the profiles for individual bands of a mixture when these bands remain unresolved for at least part of their time on the column. The intensities of the displacement and tag-along effects depend essentially on the sample size, the sample composition, and the parameters of the competitive equilibrium isotherm for the components involved. The intensity of the tag-along effect is measured by the length of the concentration plateau of the second component left behind by the first component. This plateau results from the fact that the velocity associated with a certain concentration of the second component is a decreasing function of the local concentration of the first component. An increase in the ratio of the columnloading factor for the second component, relative to the first, will lead to an increase in the intensity of the tag-along effect. The opposite is true for the displacement effect, which affects the first eluting peak of the pair, and results in its elution as a characteristic concentration plateau. To describe the peak shapes of a separation under overload conditions a clear understanding of how the competitive phase equilibria, the finite rate of mass transfer, and dispersion phenomena combine to affect band profiles is required [11,66,42,75,76]. The general solution to this problem requires a set of mass conservation equations; appropriate initial and boundary conditions that describe the exact process implemented; the multicomponent isotherms; and a suitable model for mass transfer kinetics. As an example, the most widely used mass conservation equation is the equilibrium-dispersive model

(dc I at) + F(aq / at) + u(ac / az) = DapCa^c / az^)

(i1.1)

where C is the solute concentration in the mobile phase, q the concentration in the stationary phase, t time, z distance along the column, F phase ratio (see Table 11.4), u the mobile phase velocity, and Dap the apparent dispersion coefficient, which accounts for the finite column efficiency (Dap = uL / 2N where L is the column length and N the column plate count for linear conditions). The isotherm model provides the connection between the concentration of each component in the mobile and stationary phases (section 1.5.5), and the specific initial and boundary conditions close the system of partial differential equations. There is no algebraic solution to the system of partial differential equations obtained, and numerical solutions are necessary. Prominent models for estimating peak profiles carry out a differentiation of the equilibrium isotherm with approximations for the mass transfer contribution. The equilibrium-dispersive model, above, assumes that all contributions due to nonequilibrium can be lumped into an apparent axial dispersion term. It further assumes that the apparent dispersion coefficient of the solutes remain constant, independent of the concentration of the sample components. For small particles, these approximations are reasonable for many applications. The ideal model assumes that the column efficiency is infinite. There is no axial diffusion, and the two phases are constantly at equilibrium. The band profiles obtained as solutions are in good agreement with experimental chromatograms for columns with N > 1000 having high loading factors. On the other

870

The Essence of Chromatography

hand, the agreement with experimental results is poor for small values of N with low loading factors. In this case, the contributions from mass transfer kinetics and axial dispersion to the band profiles become significant compared with the influence of the nonlinear isotherm. The lumped kinetic models combine the mass balance equation with a kinetic equation, relating the concentration change of each component in the stationary phase to its concentration in both phases, and to the equilibrium concentration in the stationary phase. Results obtained are generally similar to the equilibrium-dispersive model, with some exceptions noted [64]. All the above models require extensive computational time, and include experimental data that is difficult and time consuming to acquire. In general, the agreement between the models and experimental band profiles is reasonable. Due to the complex interactions among the large number of parameters that affect the operation of a column under overload conditions, it is difficult to establish a general guide to the selection of suitable operating conditions for different samples [11,46,64,65,80-82]. The following general observations are applicable to most cases. The column plate count should be as high as possible, and separations carried out using concentration overload conditions. The production rate of a purified fraction from the same sample increases with (a^ - 1), where a is the relative retention of the desired compound and its neighboring impurity (i.e. the ratio of the origin slopes of their equilibrium isotherms). Preparative columns should be operated at higher mobile phase velocities than is typical for analytical columns. There is an optimum value of dp^ / L (dp is the particle diameter and L the column length) for which the production rate is a maximum. This ratio is not easy to calculate for a given set of separation conditions, but is more important than specifying values of the individual parameters themselves. 11.3.5 Displacement Chromatography Displacement chromatography offers an alternative to elution chromatography for preparative-scale separations under nonlinear conditions [10,66,82,83]. It has found limited success for the purification of biopolymers by reversed-phase [10,83,84] and ion-exchange chromatography [85-88], but is not widely used for the purification of small molecules [89-91]. It has the potential for greater use, but remains a minor technique compared with elution chromatography. In displacement chromatography, the substances are resolved into consecutive zones (displacement train) of the respective pure substances, which leave the column in the order of their affinity for the stationary phase, rather than into peaks, like elution chromatography. Advantages of the displacement mode for purification are that the separated products are obtained in higher concentration than in the elution mode, solvent consumption is less, and band tailing is reduced as a result of the self-sharpening boundaries obtained. Separations are generally faster than for elution, but the cycle time is often greater, owing to the need to regenerate the column (rinse away the displacer and equilibrate with the initial mobile phase) between each injection. In displacement chromatography the column is first equilibrated with mobile phase (referred to as the carrier), chosen for its low solvent strength in the separation system,

Laboratory-Scale Preparative Chromatography

871

and as a good solvent for the sample. After equilibration, the sample is introduced into the column, usually dissolved in the mobile phase, where it is concentrated at the head of the column. Thus, it is important that the stationary phase exhibits strong retention of the sample components with the carrier solvent as eluent. In the next step, a concentrated solution of the displacer (usually in the carrier) is continuously pumped through the column. The displacer is a substance with a higher affinity for the stationary phase than any of the sample components, and causes the sample components to move down the column at speeds determined by the displacer front velocity. Stronger adsorbing sample components displace from the stationary phase surface those having weaker retention until the separation is achieved. The mixture separates into a displacement train, and once fully developed, the train is composed of adjacent rectangular bands of near uniform concentration, all moving at the same velocity. Following the breakthrough of the displacer front, the column needs to be regenerated and conditioned for further use. Critical parameters for a separation are the choice of the displacer and its concentration, the loading factor, and the column efficiency. The choice of displacer is probably the most critical step, and is usually identified in an empirical screening procedure, often by evaluating compounds similar to the sample, or polyelectrolytes in the case of ionexchange. For correct development of the displacement train, the adsorption isotherm of the displacer must lie above those of all the sample components. The concentration of the displacer controls the separation time and the concentration of the sample zones (the length [but not the height] of the separated zones is proportional to the sample amount). The displacer concentration is optimized for each system by a screening experiment. The choice of carrier solvent is less critical. It must be a good solvent for the displacer and the sample, as well as providing high retention factors for the sample in the separation system. To minimize the effects of axial dispersion, the optimum flow rate is usually low, compared elution chromatography. The column must be long enough for the development train to be fully formed, but need not be longer than this minimum length for a given sample size. Since the length of the zones in the displacement train increase with increasing sample amounts, longer columns are required to accommodate larger samples. Typically, the column length is fixed and the maximum sample size is established in a screening experiment. Injection of too large a sample for the column results in lower product purity owing to incomplete formation of the displacement train. 11.3.6 Simulated Moving Bed Chromatography Simulated moving bed chromatography overcomes the limitations of batch processes (e.g. elution and displacement chromatography), namely, discontinuous operation, inefficient use of the stationary phase, and high product dilution. Simulated moving bed chromatography was developed as a process-scale operation, where its economic benefits have been clearly demonstrated. In recent years, it has been downscaled to a pilot-plant and laboratory-scale processes, and is receiving increasing attention in the fine chemicals and pharmaceutical industries for the purification of high value added products (e.g. enantiomers) and in biotechnology. Simulated moving

The Essence of Chromatography

Eluent

Section 4

Raffinate B

Figure 11.9. Schematic diagram of a simulated moving bed chromatography for the separation of a binary mixture (A + B), where B is the component with the lower affinity for the stationary phase. (From ref. [94]; Elsevier).

bed chromatography is a complex process, and laboratory-scale operations are often performed to assist in scale-up operations. It is unclear, however, whether the favorable production economies of simulated moving bed chromatography are sufficient to warrant the extensive effort required for optimization in laboratories that need to purify modest amounts of a range of different mixtures. The simulated moving bed process is a way of implementing a true countercurrent chromatographic process in which the stationary and mobile phases are considered continuous phases moving in opposite directions. A true moving bed process is mechanically difficult to implement with liquid and solid phases. The simulated moving bed system consists of several identical columns (in most cases 6-12), which are connected in series. Figure 11.9 [92-95]. Each column is connected through a number of valves (depending on design) which can be switched to an open or closed position. A recycling pump inside the column circle delivers the mobile phase through all columns. Two additional pumps constantly inject the sample stream (feed) and fresh mobile phase, and two (or more) additional pumps withdraw the product streams (raffinate and extract flows). With multiple pumps, the internal recycle and all external flow streams are controlled and monitored. The countercurrent movement of the stationary and mobile phases is simulated by simultaneously switching the external and recycle flow streams, at discrete time intervals, and in the same direction. The periodic switching simulates movement of the solid phase by steps, and effectively the countercurrent movement of the stationary phase. In the "VARICOL" process, the flow streams are shifted asynchronously, changing the column allocation in the time domain between the external flow streams [96]. In both cases, the less retained sample component must move in the same direction as the mobile phase in all columns, and

Laboratory-Scale Preparative Chromatography

873

the more retained component in the opposite direction. Although simulated moving bed technology provides a number of advantages leading to cleaner, smaller, and faster processes, the main disadvantage is the limitation to either the separation of binary mixtures, or recovery of one component from a multicomponent mixture. The purification of ternary mixtures is possible, but the systems are significantly more complex to design and operate [97,98]. In general, simulated moving bed processes are operated under isocratic conditions. Separation performance, however, can be further improved by gradient operation in order to optimize the retention behavior of the mixture components in each section of the unit [99,100]. In the simulated moving bed chromatograph, the columns are considered distributed among four sections based on the location of the external flow streams. Figure 11.9. Section 1 is located between the mobile phase inlet (eluent) and the product stream containing the product with a higher affinity for the stationary phase (extract). Its function is to desorb the more strongly retained product. Section 2 is located between the extract line and the sample inlet stream (feed). Its function is to desorb the product with a lower affinity for the stationary phase. Section 3 is located between the feed and the product stream containing the component with a lower affinity for the stationary phase (raffinate). Its function is to adsorb the product with a higher affinity for the stationary phase. Section 4 is located between the raffinate and eluent streams. Its function is to adsorb the product with the lower affinity for the stationary phase. In the case of a binary mixture, the sample stream carries the two components into section 3. The component with the lower affinity for the stationary phase is eluted from the column(s) in the raffinate before switching the external flow streams. The component with higher affinity for the stationary phase remains bound to the stationary phase. Because of flow switching, it remains in the column until the column reaches section 1, where it is eluted in the extract stream. With proper working conditions, the more retained product is not eluted from the column(s) of section 3, and the less retained component does not adsorb in the column(s) of section 2. In section 4, the least retained product is adsorbed on the column(s), so the mobile phase is cleaned, and can be recycled to section 1. Totally empirical optimization of a simulated moving bed process is difficult and time consuming. Computer simulations are generally used for the design and optimization under linear [92,94,101-103] and nonlinear [95,102-105] operating conditions. Efficient operation requires control and knowledge of a large number of parameters. The most critical are: the column length and internal diameter; number of columns per section; stationary phase type and particle size; eluent composition; eluent, feed, raffinate, extract, and recycle flow rates; column pressure drop; and switching time intervals. For simulation of the separation process, a set of mass balance, kinetic, and isotherm equations are required for each component of the feed, and fluid dynamics equations for the columns. The model coefficients for these equations are obtained by exploratory experiments on a single column. Thus, it is important that the individual columns in the separation unit have (near) identical properties [104]. Simulated moving bed chromatography is a process operating design that is equally applicable to supercritical fluid [106,107] and gas [108-110] chromatography, as for liquid chromatography.

874

The Essence of Chromatography

11.3.7 Sorbents for Biopolymer Separations Sorbents used for preparative-scale chromatography can be placed into two categories. The larger group contains sorbents of similar morphology, surface chemistry, and mechanical strength to those employed for analytical separations (section 4.2). For preparative-scale applications, the particle size might be larger, but all other properties are virtually identical. For analytical applications the stationary phase is reused many times, but this is not always possible for preparative-scale applications where, in addition, the quantity of stationary phase employed for purification can be orders of magnitude higher. Economic considerations dictate that sorbents of a lower cost (and lower specification) are useful for the initial isolation of semi-pure products from crude samples. In addition, sorbents with a configuration that concurrently optimizes separation speed and sample capacity are desirable for preparative-scale chromatography, but are of less interest for analysis. Most sorbents in this second category evolved from the need for alternative approaches for the purification of biotechnology products and the isolation of biopolymers. Sorbents for preparative-scale chromatography can be categorized according to composition, structure and mechanical strength. Those termed soft gels, are suitable for use at low pressures (< 3-5 atm) because of dimensional instability at higher pressures. Semi-rigid gels can be used at low to medium pressures (< 15-30 atm), and provide higher performance and shorter separation times. These gels require solvation by an appropriate solvent to create a defined pore structure, suitable chromatographic properties, and a favorable sample capacity. 11.3.7.1 Natural and Synthetic Soft and Semi-Rigid Gels Starch, agarose, and cellulose are examples of natural polymers which spontaneously form pore networks when precipitated from a suitable solvent [6,111-116]. Dextran also forms a defined pore network in solution, but only after crosslinking [6,113117]. Agarose and crosslinked dextran, and to a lesser extent cellulose, are the most important natural gels for preparative-scale chromatography based on size exclusion, Table 11.5. These sorbents have a high concentration of surface hydroxyl groups providing biocompatibility, as well as facilitating chemical modification of the sorbent surface. Common applications are the separation of biopolymers by size exclusion or ion-exchange chromatography (after the introduction of ionic functional groups), and as a support for the preparation of affinity sorbents (section 11.3.8). Agarose is a linear copolymer of P-D-galactopyranose and 3,6-anhydro-a-L-galactopyranose with a low concentration of sulfate groups prepared from agar, isolated from marine algae. Non-crosslinked agarose beads (2-100 |xm) are produced by a simple emulsion-gelation procedure. Reaction with 2,3-dibromopropanol, epichlorohydrin or divinyl sulfone under strongly alkaline conditions forms crosslinks within the bundles of agarose chains, stabilizing the bead structure, while preserving the porosity. The pore size ( 5 - 1 0 0 nm) is controlled by the agarose concentration in the solution from which it is gelled. Agarose is stable in water and salt solutions at pH 4-9. Phenyl and octyl

Laboratory-Scale Preparative Chromatography Table 11.5 Natural soft and semi-:rigid gels for the separation of biopolymers (A/D agarose-dextran copolymer, D/bA dextran-methlene bisacrylamide c opolymer) Type Sepharose 6B 4B 2B CL-6B CL-4B CL-2B Superose 12 6 Sephedex G-25 G-50 G-75 G-lOO G-150 G-200 Superdex 75 200 Sephacryl S-100 HR S-200 HR S-300 HR S-400 HR S-500 HR Cellufine GCL-90 GCL-300 GCL-700 GCL-1000 GCL-2000 Polymer Agarose Properties Particle size (|xm) 45-165 60-140 60-200 45-165 60-140 60-200 34 34 1 0 - 4 0 or 4 0 - 1 2 0 1 0 - 4 0 or 4 0 - 1 2 0 1 0 - 4 0 or 4 0 - 1 2 0 1 0 - 4 0 or 4 0 - 1 2 0 1 0 - 4 0 or 4 0 - 1 2 0 1 0 - 4 0 or 4 0 - 1 2 0 22-44 22-44 2 0 - 7 5 or 4 0 - 1 0 5 2 0 - 7 5 or 4 0 - 1 0 5 2 0 - 7 5 or 4 0 - 1 0 5 20 - 75 or 40 - 105 20 - 75 or 40 - 105 45 _ 84 or 45 - 105 4 5 - 8 4 or 4 5 - 1 0 5 45 _ 84 or 4 5 - 1 0 5 45 _ 84 or 4 5 - 1 0 5 45 _ 84 or 4 5 - 1 0 5

875

Crosslinked Agarose Crosslinked Agarose Crosslinked Dextran

A/D A/D D/bA D/bA D/bA D/bA D/bA Cellulose

Fractionation range 1x104-4x10^ 6 X 10^ - 2 X 10^ 7 X 10^ - 4 X 10'^ 1 X 10^ - 4 X 10^ 6x10^-2x10^ 7 X 10^ - 4 X 10^ 1 X 10^ - 2 X 10^ 5x10^-4x10'^ 1 X 10^ - 5 X 10^ 1.5x10^-3x104 3 X 10^ - 7 X 10^ 4 X 10^ - 1 X 10^ 5 X 10^ - 3 X 10^ 5x10^-6x10^ 3x10^-7x104 1 x 104 - 1 X 10^ 1 X 10^ - 1 X 10^ 5 X 10^ - 2.5 X 10^ 1x104-1.5x10^ 2 X 104 - 8 X 10^ 2 X 104 - 3 x 10^ 5x10^-4x104 1 X 10^ - 1 X 10^ 2 X 104 - 3 x 10^ 2 X 10^ - 6 X 10^ 2 X 10^ - 3 X 10^

derivatives of crosslinked agarose are used for separations by hydrophobic interaction chromatography (section 4.3.5). Dextran, a branched homopolymer of D-glucose, is produced in bead form from a water-oil suspension to which a suitable crosslinking reagent (see agarose) is added. Point crosslinking of the dextran chains stabilizes the bead structure, forming soft gels with a pore size determined by the relative concentration of dextran and the crosslinking reagent. A low concentration of reactants produces materials with a high porosity at the expense of rigidity. Stronger materials are available by reaction of allyldextran with methylenebis(acrylamide) as the crosslinking reagent. Crosslinked dextrans are stable in salt solutions, organic solvents, and alkaline and weakly acidic conditions. Hydroxypropylation of a crosslinked dextran with a narrow pore size produces useful sorbents for the purification of low molecular mass organic compounds by sizeexclusion and adsorption chromatography [4,27]. The most popular sorbent for this type of application, Sephadex LH-20, is solvated by weak and medium polarity organic

876

The Essence of Chromatography

solvents. The maximum sample load is about 300 mg / g of dry sorbent, and the maximum exclusion limit about 4,000 Da. Conventionally produced cellulose powders (microcrystalline cellulose) consist of irregularly shaped fibrous particles of limited use for column chromatography. Beaded cellulose is prepared by dissolution of cellulose powder in a suitable solvent, followed by droplet formation in a suspension medium, and subsequent solvent extraction or crosslinking. Cellulose triacetate and tricarbamate derivatives are useful as low-cost sorbents for the process-scale separation of enantiomers (section 10.4.2). Synthetic organic gels in a range of particle and pore sizes are prepared by suspension polymerization for size-exclusion chromatography (section 4.2.4) [113116,118,119]. Gels with a range of solvent compatibility and swelling properties are obtained by varying the identity of the monomers and their relative concentration. Classic poly(styrene-divinylbenzene) gels are suitable for the purification of synthetic polymers and fractionation of low molecular mass organic compounds. Poly(acrylamide)-type gels are generally superior for the purification of biopolymers, and include poly(acrylamide), poly(methyl methacrylate), poly(hydroxymethyl methacrylate), poly(tris[hydroxymethyl] methacrylate), poly(ethylene glycol dimethacrylate), poly(vinyl acetate, and poly(vinyl alcohol). Poly(acrylamide)-type gels are stable over the pH range 4-13. The natural and synthetic polymer gels, described above, are suitable for the preparation of ion exchangers [6,7,113-115]. Simple coupling chemistry allows the introduction of sulfate, sulfonate (e.g. sulfoethyl, sulfopropyl and sulfobutyl), phosphoesters, and carboxymethyl groups. Cation exchangers are represented by tertiary amino groups (e.g. dimethylaminoethyl, diethylaminoethyl) and quaternary amino groups (e.g. tetramethylaminoethyl, tetraethylaminoethyl). The functional groups are attached to the polymer backbone through short, medium, or long spacer arms, by means of ether, alkylamine, or amide bonds. The diverse chemical nature of the support matrices, and the coupling chemistry employed results in ion-exchangers that differ significantly in protein retention and recovery [ 120-125]. Selection of a particular gel for a given application remains a difficult problem. The large number of sorbents now available, however, ensures a high level of success, if only after a series of screening experiments to eliminate gels with less desirable properties. Amongst the genre of gel-based ion exchangers, are two types designed for improved performance in preparative-scale chromatography. Hyper-diffusive particles are described later (section 11.3.7.2). Tentacle ion-exchangers consist of linear polymeric chains anchored to the support matrix. Figure 11.10 [122,125,126-128]. The average length of the polymer chains is 15 to 50 monomer units, each monomer unit carrying a charge center. A fully extended tentacle chain is about 10 nm long. To provide uninhibited access of proteins to the interior of the gel matrix, large pore diameters (100 to 500 nm) are required. The flexibility of tentacle ion-exchangers allows the charge distribution to adapt to the protein structure, and is less Hkely to cause unfolding or denaturation of unstable proteins compared with conventional ion exchangers with a rigid array of binding sites. In addition, the binding capacity of

Laboratory-Scale Preparative Chromatography

877

THROUOHPORE

Figure 11.10. Schematic diagram of a hyper-diffusive (gel-filled gigaporous) particle (left), tentacle ion-exchanger interacting with a globular protein (middle), and a perfusive (diffusion-convection) particle (right).

tentacle ion exchangers is often greater than for conventional polymer gels. This is a result of the larger number of ionic sites on the tentacles, and the flexibility of the tentacle chains, which allows greater access to sites on the protein, especially for large proteins. In specific cases, tentacle ion exchangers provide a reduction of non-specific interactions with the gel matrix, and a general improvement in selectivity and band shape. The concept of tentacle media also finds use in the development of materials for hydrophobic interaction and affinity chromatography. Hydroxyapatite is a crystalline form of hydroxylated calcium phosphate [Caio(P04)6 (OH)2] used as a powder, porous ceramic, or microcrystals embedded in an agarose matrix for the purification of proteins and DNA [129-131]. Hydroxyapatite provides complementary selectivity to conventional ion exchangers, and is often used for the further purification of fractions isolated by conventional ion exchangers. Separations on hydroxyapatite are due to a charged-based, mixed-mode retention mechanism. The surface of hydroxyapatite consists of positive calcium ions, referred to as C-sites, and clusters of negatively charged phosphate groups, referred to as P-sites, arranged in a crystalline structure. Carboxylic acid and phosphate functional groups are adsorbed at C-sites by coordination complexation. Ionized amine groups are adsorbed at P-sites. Continuous or stepwise concentration gradients of sodium or potassium phosphate at close to neutral pH, are used for separations. The phosphate ion competing with the protein or nucleic acid for the two types of adsorption sites. Adsorption of metal ions, particularly manganese, iron, and aluminum, from process materials results in discoloration of the hydroxyapatite and possible loss of separation properties [131]. 11.3.7.2 Diffusion-Convection Sorbents Diffusion-convection sorbents are biporous materials with a network of wide pores providing convective transport connected to smaller diffusion pores that posses most of the surface area, and are responsible for retention [132-136]. An optimized arrangement of diffusion and convection pores results in enhanced mass transfer properties and shorter separation times for macromolecules. Sorbents in this category include superporous agarose beads [137,138], biporous monolithic or continuous bed

878

The Essence of Chromatography

columns (section 4.2.7) [49-51], and perfusiveparticles [139-143]. Superporous agarose beads are prepared by a double emulsification procedure forming large flow pores (superpores) with a diameter 0.25 to 0.05 of the particle diameter (106-180 |xm) connected to normal diffusion pores. Perfusive particles are prepared by suspension polymerization, as rigid beads of highly crosslinked poly(styrene-divinylbenzene) made up of agglomerated microspheres, Figure 11.10. The void spaces between the microspheres and agglomerates define the pore size distribution, with the wide pores as regions separating the agglomerates. Particles of 20 or 50 |xm diameter with convection pores of 0.6-0.8 |xm and diffusion pores 0.08-0.15 |xm are typically used for purification of biopolymers. Poly(styrene-divinylbenzene) particles can be operated at pressure (< 200 atm) without compression. However, these matrices cannot be used for protein purification because of strong and irreversible interactions of proteins with the hydrophobic sorbent surface. A more suitable hydrophilic surface, for example, can be achieved by coating the surface of the poly(styrene-divinylbenzene) beads with poly(vinyl alcohol), and subsequently crosslinking the poly(vinyl alcohol) with gluteraldehyde [141]. In practice, a wide range of functionalized coatings and biospecific ligands (antibodies, proteins, peptides, nucleic acids, carbohydrates, etc.) can be immobilized on the interior particle surface for separations by reversed phase, ion exchange, hydrophobic interaction, and affinity chromatography. Transport of molecules through a homogeneous porous particle occurs by diffusion. For macromolecules, this is a slow process producing characteristic broad bands. At high flow rates diffusion-convection media allow convective mass transfer into the particle interior. Since convective transport of biopolymers is several orders of magnitude faster than diffusion, the interior binding sites within the pore network are contacted more rapidly than for packing materials relying solely on diffusive transport. Only a small percentage of flow transecting the particles is required in order to achieve a dramatic increase in mass transfer rates. The ratio of particle permeability to packedbed permeability, or the split ratio, is an important parameter determining column performance. The column plate height and binding capacity is only weakly dependent on flow rate. The reduced residence time improves recovery of bioactive compounds and reduces the cost of large-scale purification processes. Diffusion-convection particles have performance advantages over particles with standard pore sizes for the separation of macromolecules, mainly for applications that do not require high resolving power to accomplish fast separations. The low surface area of the particles, however, restricts the ultimate production rate. Polymeric gels have high loading capacities, but a weak mechanical structure, which restricts the optimum use of pressure and the production rate. For ion exchange purification of biopolymers, hyper-diffusive particles (also known as gel-in-a-shell or gel-filled gigaporous particles) overcome these limitations. This is done by encapsulating an ionic polymer gel in a rigid, porous silica-polymer composite shell. Figure 11.10 [10,123,141,144-148]. The pore volume of a silica substrate with surface reactive groups is filled with the desired ionic monomers, which are then polymerized within and throughout the pore volume. The gel structure is optimized with respect to the concentration of ionic groups, polymer

Laboratory-Scale Preparative Chromatography

879

immobilized ligand

Product to be purified

Product adsorbed on immobilized ligand

Product to be eluted

Figure 11.11. Principles of affinity chromatography. (Fromref. [178]; Elsevier).

flexibility, and diffusive properties of the polymeric network. The shell is optimized with respect to particle size, pore size distribution, and mechanical strength. The high efficiency of the hyper-diffusive particles is attributed to the favorable permeability of the polymer gel and modest particle size (40-60 (xm). The higher production rate results from the greater loading capacity of the hyper-diffusive particles compared with diffusion-convection particles. 11.3.8 Affinity Chromatography Affinity chromatography is a well-established technique for the isolation and purification of macromolecules, generally proteins and polynucleotides, based on highly specific three-dimensional molecular recognition [6,10,113-116,149-152]. There are virtually no direct applications to the purification of low molecular mass organic compounds with the exception of a few micropreparative applications employing molecularly imprinted polymers, usually for the separation of enantiomers [153-158]. Even in this case, these materials are generally used for analysis. They are somewhat limited for preparative-scale chromatography by the heterogeneity of binding sites, poor efficiency, and low loading capacity. Improvements in synthetic design may eventually reveal their true potential for large-scale purification. In addition, the well-developed, small-scale applications of affinity chromatography to clinical diagnostics, protein biding studies, and solid-phase extraction, etc., are not considered in this section [159-162]. In affinity chromatography, a ligand (molecule having specific recognition capability) is immobilized on a suitable insoluble support (matrix), which is usually a natural or synthetic polymer gel in bead or membrane form. The analyte (or target) is adsorbed selectively (captured) by the immobilized ligand, by simply passing a solution of the analyte through the column under favorable conditions for analyte-ligand binding. Figure 11.11. Weak and non-specifically bound impurities are removed by rinsing the column with the buffer used for sample application, or other composition that does not displace the analyte. The analyte is recovered as a concentrated solution by elution under conditions that favor dissociation of the analyte-ligand complex. For example, by adjusting the pH, ionic strength, or temperature of the sample application buffer, or by using specific solvents, or competitive free ligands added to the sample buffer.

880

The Essence of Chromatography

Table 11.6 Typical affinity ligands for protein separations Ligand Metal ions Reactive dyes Lectins Phenylboronic acids Carbohydrates Nucleic acids Amino acids Heparin Nucleotides, cofactors. Proteins A and G Hormones, drugs Antibodies Antigens Analytes Proteins with an abundance of histidine, tryptophan, and cysteine residues Most proteins, particularly nucleotide-binding proteins Glycoproteins, cells Glycoproteins Lectins Nuceleases, polymerases, and other DNA and RNA binding proteins Proteases Plasma proteins, such as clotting factors Enzymes, substrates, and inhibitors Immunoglobulins Receptor proteins Antigens Antibodies

Binding of the analyte is a monolayer adsorption process characterized by an equiUbrium isotherm, typically of the Langmuir of Fruendlich type (section 1.5.5). The saturation capacity of the adsorbent can be estimated through these relationships. The capacity of the adsorbent is fully utilized when the column is saturated, corresponding to the breakthrough point for sample loading. Thus, the general conditions for elution chromatography are of limited application for modeling the separation process. Plate theory does not really apply, and slow kinetics of the adsorption/desorption process is generally more important. Almost any compound may be used as an affinity ligand, if it binds selectively and reversibly to the analyte in question, and can be immobilized on a solid support without impairing the analyte-ligand-binding interaction. Table 11.6. For affinity chromatography the optimum dissociation constant for the ligand-analyte complex is about lO""* - 10"^ M. Larger binding constants have the potential to deliver higher purification, but may prevent elution of the bound protein in its native state. Similarly, weaker binding constants are disadvantageous, since proteins are merely retarded and quickly breakthrough the column. Typical ligands used for preparative-scale chromatography are either group-specific or highly selective, in their binding properties. Group-specific Ugands retain a range of analytes with similar binding properties. Highly selective ligands retain a single or small number of analytes. For every protein purification problem there is always an affinity solution, but cost and safety considerations may render these solutions impractical. As an example, antibodies are widely used for analysis, where only relatively small amounts are usually required, but their production and purification on a large scale for preparative-scale chromatography may be difficult to justify economically. In some cases, Hhybridoma technology may be able to address this problem. Even if production costs are acceptable, the immobilized antibodies may be unstable over the sequence of sample application, elution, and sanitation required for multiple use of the affinity adsorbent. For these reasons, while biological ligands (antibodies, enzymes, receptors, lectin,

Laboratory-Scale Preparative Chromatography

881

nucleic acids, etc.) continue to be used, their wider application is hampered by the above factors [149,150,163-166]. These considerations have lead to increasing interest in low molecular mass synthetic ligands, such as reactive dyes (section 11.3.8.1), metal chelates (section 11.3.8.2), thiophilic ligands [167], and boronic acids [6,166,168] for preparative-scale affinity chromatography. Synthetic ligands are often group-specific, and improving the selectivity of separations is of great importance. Possibilities include using competitive ligands or displacers, or by designing new ligands tailor-made for the analyte of interest. The support is expected to function as a scaffold covered with an optimum ligand concentration, in such a way that adequate access and orientation of the ligand for binding with the analyte is preserved. Desirable properties in an ideal support include: an absence of non-selective binding interactions associated with charged and hydrophobic surface groups; an abundance of surface functional groups (hydroxyl, carboxylic, amide, etc.) for ligand immobilization; adequate mechanical and chemical stability under a wide range of conditions, such as high and low pH, and temperature, and in situations that require organic solvents, detergents, etc. The use of modest pressures and the requirement of high surface areas to maximize sample capacity, favor the use of spherical porous particles with narrow particle size ranges between 10 to 500 |xm and pore sizes between 10-500 nm. Crosslinked and beaded poly (saccharides), particularly agarose and to a lesser extent cellulose, synthetic macroporous polymers, particularly poly(acrylamide)-based gels, and inorganic oxides (coated silica, porous glass), as well as hyper-diffusive and convection-diffusion particles (section 11.3.7.2) are widely used [113-116]. Crosslinked beaded agarose with an exclusion limit of about 10^ Da is the most popular support for laboratory-scale applications. Reagents that facilitate either direct reaction between the ligand and support, or introduce a different reactive functional group at the support surface, which is subsequently reacted with the ligand, are widely employed for immobilization of affinity ligands [6,113,169-173]. The majority of reactions for ligand immobilization employ activation of hydroxyl groups by reagents containing halogen, epoxy or carbonyl functional groups. Table 11.7. These supports are then reacted with ligands containing nucleophilic groups, such as amine, hydroxyl, or thiol groups. Since virtually all proteins and the majority of synthetic ligands contain at least one of these functional groups (or these functional groups are easily introduced into suitable ligands), a wide range of coupling chemistries is available for optimizing the attachment of the ligand to the support. Other reactions, employing carboxylic acid groups, for example, are available. The large number of chemically activated supports now available simplifies the production of affinity adsorbents in the laboratory. To ensure minimum interference in the binding of the analyte by the immobilized ligand the coupling reaction between the ligand and support should occur through a region of the affinity ligand remote from the binding site. In addition, the binding site of biopolymers is often located deep within the three-dimensional structure of the molecule, and steric hindrance prevents interaction of the ligand with the analyte. In this case, spacer arms, usually short chains 6-12 units (CH2 and/or O) long, are imposed

882

The Essence of Chromatography

Table 11.7 Coupling reactions for supports containing surface hydroxyl groups Activating agent Cyanogen bromide Tresyl chloride Epichlorohydrin 1,4-Butanediol diglycidyl ether l,r-Carbonyldiimidazole Cyanuric chloride Divinylsulfone Sodium periodate Glutaraldehde Reactive group on ligand Primary amines Primary amines, thiols Primary amines, thiols, hydroxyls Primary amines, thiols, hydroxyls Primary amines, hydroxyls Primary amines, hydroxyls, thiols Primary amines, hydroxyls Primary amines Primary amines

between the support surface and the ligand to ensure hgand accessibihty to the binding site. In addition, it is generally recognized that there is an optimum concentration of immobilized ligand for purification of a specific protein. This concentration, however, varies with the protein in question, and the optimum loading has to be ascertained by experiment. Affinity chromatography can be used as a single stage purification and concentration technique. More typically, though, it is used for one or more stages of a purification cascade that possibly includes ion exchange and hydrophobic interaction chromatography, as well as traditional isolation techniques, such as precipitation. Affinity chromatography is uniquely suited to the recovery of proteins from dilute solutions. Its high selectivity leads to its use for the final purification of isolates containing small amounts of impurities with similar properties to the desired product. How affinity chromatography is used in a purification cascade, however, will largely be dictated by economic considerations, since affinity adsorbents tend to be more expensive than other materials. 11.3.8.1 Reactive Dyes (Biomimetic Ligands) While biological ligands display high selectivity, they suffer from low binding capacity, limited reusability, and low scale-up potential. Synthetic ligands offer a number of advantages for the purification of proteins. Affinity adsorbents prepared from synthetic ligands are less expensive, scalable, durable, and reusable over multiple cycles. The exceptional stability of synthetic adsorbents allows harsh elution, cleaning, and sterilization protocols to be used. Synthetic dyes are the most important alternative to natural ligands for preparative-scale affinity chromatography [6,174-179]. These dyes are structural mimics for naturally occurring nucleotides and flavins. The most important group of synthetic dyes contain reactive, usually chlorotriazine, groups that facilitate bonding to different supports. These dyes also possess an extended aromatic backbone with a variety of functional groups that allow simultaneous interactions with protein binding sites, mimicking natural ligands. The often-poor selectivity, variable composition, leakage, and toxicity of the triazine dyes, which are inexpensive commodity chemicals, limit their potential applications. As a template for the development of more specific ligands for protein purification, however, they

Laboratory-Scale Preparative Chromatography

883

have proven uniquely useful. From molecular modeling techniques, it is possible to redesign the parent dye, or design de novo a new dye, that mimics naturally occurring ligands ("biomimetic" dye). The design strategy consists of several defined stages. Molecular modeling is used to formulate a preliminary design of a complementary ligand using a suitable binding region of the target protein as a template. Solidphase synthesis of a limited near-neighbor combinatorial library allows optimization of the ligand-protein interaction. The most promising ligands are then synthesized on a larger scale to provide sufficient material to characterize key properties and to permit chromatographic evaluation. For the latter purpose, the ligand is immobilized on a suitable support, and the chromatographic conditions optimized for purification of the target protein from the sample matrix. It is the possibility of combining information from X-ray crystallographic or NMR protein structures with combinatorial synthesis and advanced computational tools, that make the rational design of affinity ligands feasible. Similar design strategies are used to prepare synthetic ligands based on peptides, oligonucleotides, or small proteins. 11.3.8.2 Immobilized Metal Ion Affinity Chromatography Immobilized metal ion affinity chromatography (IMAC) relies on the ability of certain amino acid side chains to form coordinate bonds with immobilized metal ion complexes [6,181-186]. The adsorption of proteins mainly takes place through interactions with the imidazole ring of histidines. For other amino acids with electron donor atoms in their side chains, binding tends to be weak. Cysteines in natural proteins, are rarely available in an appropriate reduced form for binding to chelated metal ions. Immobilized metal ion affinity chromatography provides group-specific separations in general, as well as a highly selective purification tool for target proteins from complex biological samples. It is one of the most popular methods for the purification of recombinant proteins modified with oligo-histidine affinity tags at the N- or C-terminus [187-189]. Since oligo-histidine residues are uncommon among natural proteins, such affinity handles allow a one-step isolation of tagged proteins in high purity. In most cases, the insertion of a small tag composed of (usually) six histidine residues hardly modifies the activity or stability of the protein. This method is proving increasingly popular for the low cost production of large quantities of pure enzymes for industrial applications, and for use in molecular biology. In addition, substantial progress has been made in the selective purification of phosphorylated proteins using immobilized iron (III) chelates [182,190]. The main practical attributes of immobilized metal ions for affinity chromatography are their stability to a wide range of operating conditions, high capacity, mild elution conditions, ease of regeneration, and low cost. The adsorption of proteins in IMAC is based on the coordination between an immobilized metal ion and electron donor groups on the protein surface. Soft metal ions [e.g. Cu (II), Co (II), Zn (II), Ni (II)] show a preference for coordination with Ncontaining functional groups, such as the imidazole ring of histidine. Hard metal ions [e.g. Al (III), Ca (II), Fe (III)] show a preference for oxygen-containing groups, such as carboxylate or phosphate. The metal ions are immobilized on the support by interaction

884

The Essence of Chromatography

with a chelating ligand, such as iminodiacetic acid (IDA), nitrilotriacetic acid (NTA), carboxymethylated aspartic acid (CM-Asp) or tris(carboxymethyl)ethylenedianiine (TED). The binding of proteins to the affinity adsorbent depends on the choice of metal ion and the structure of the ligand (chelate group), as well as the length of any spacer arm, and the surface concentration of the ligand. All of which presupposes that the range of support pore sizes does not interfere with access of the protein to the metal binding site. Metal ion transfer must be avoided (the chelating ligand must bind the metal ion tightly enough to prevent its removal by sample proteins). Protein adsorption by the affinity adsorbent requires a pH that favors chelation with the protein (e.g. the nitrogen atoms of the imidazole ring of histidine are unprotonated), normally in neutral or slightly basic medium. Relatively high ionic strength buffers (e.g. containing 0.1 to 1.0 M NaCl) to reduce nonspecific electrostatic interactions between the protein and the metal chelate complex. Once the sample is loaded on the affinity adsorbent, weakly bound proteins are removed by rinsing with several bed volumes of buffer, or more efficiently, by a low concentration of eluting agent. A rinse step with a low ionic strength buffer allows removal of proteins retained by hydrophobic interactions with the support. The target protein is recovered by elution with an appropriate pH buffer, ligand exchange, or extraction of the metal ion by a stronger chelating agent, such as EDTA. Elution buffers with a lower pH, or declining pH gradient, are widely used for eluting target proteins bound to soft metal ions. A further option is to use a displacer, such as imidazole. With hard metal ions, elution requires an increasing pH gradient, or addition of a competing reagent to the mobile phase, such as organic acid or phosphate. Immobilized metal ion affinity chromatography columns have excellent regeneration properties.

11.4 SUPERCRITICAL FLUID CHROMATOGRAPHY Packed column supercritical fluid chromatography is emerging as a complementary technique to normal-phase liquid chromatography for the purification of low to intermediate molecular mass organic compounds. Favorable attributes include faster separations, less solvent waste, and rapid equilibration simplifying the use of composition gradients. There are no applications to biopolymers and other polar compounds with low solubility in supercritical fluid carbon dioxide and its mixtures with organic solvents. These mobile phases, sometimes containing small amounts of additives to improve peak shapes, are used exclusively in current applications. In addition, virtually all applications are limited to slurry packed semipreparative columns with internal diameters < 25 mm containing small diameter particles (< 20 |xm) [191-198]. This is not due to any fundamental restrictions, but reflects the limited flow range of hybrid instruments designed for both analytical and preparative use, found in most laboratories. These instruments meet the demands of laboratories that require access to a preparative capability on an occasional basis, and are interested in purifying a modest amount of sample.

Laboratory-Scale Preparative Chromatography

885

Instruments for all purposes from the laboratory-scale through pilot plant to processscale, including simulated moving bed systems, are available [6,106,107,194,195,199201]. Large-scale systems are designed with different engineering and economic considerations in mind compared with laboratory-scale systems, and are not discussed in this section. In the last few years, fully automated systems for high-throughput purification of combinatorial libraries with UV or mass-directed fraction collection have been described, completing the circle of laboratory-scale purification processes [202,203]. Method development is similar to the scale-up approach for high-pressure liquid chromatography (section 11.3.4.3). Preliminary analytical separations are used to optimize the selectivity factor and separation time. The sample concentration or volume is then increased to the maximum sample load for linear chromatography, or to overload conditions for increased productivity (section 11.3.4.4). Band broadening under column overload conditions is generally governed by saturation of the mobile phase [191]. Separations under overload conditions require injection of large sample volumes, unless the sample can be highly concentrated in a suitable solvent. For samples in weak solvents, injection volumes up to about 2 ml can be made on semipreparative columns. However, severe peak distortions and splitting can occur with even small injection volumes when the elution strength of the sample solvent is greater than the eution strength of the mobile phase (section 7.6.2.1). Strong solvents may be required to achieve adequate sample solubility. Solvent-elimination injection techniques are used to circumvent this problem [194,205,206]. An automated solvent-elimination injection system, uses two six-port valves separated by a small packed column trap. The sample is loaded onto the trap in the usual way and excess solvent evaporated by a stream of (warm) gas. The solvent-free sample is then transferred from the trap to the separation column by dissolution in the mobile phase. Otherwise, the sample is extracted from the trap with a mobile phase suitable for focusing the sample at the head of the separation column. Solventless injection is also useful for handling aqueous samples (although elimination of water is time consuming), and for concentrating dilute samples. For modest sample volumes, the evaporation step can be performed in an empty capillary tube, so long as there is a sufficient vapor pressure difference between the solvent and sample (see section 7.6.2.1). Products are collected after detection and/or on a time basis by automated fraction collection using different approaches [194,202,203,207-210]. Decompression of the mobile phase occurs with cooling, phase separation of mobile phase mixtures, and expansion of the fluid phase to a gas. Possible problems for sample collection are low recovery due to precipitation of the sample in transfer lines, and formation of aerosol particles that escape the collection vessel. Simple collection devices like an empty vessel at atmospheric pressure, or collection in solvent by bubbling the expanded gas through the solvent, result in significant sample losses at the high flow rates associated with semipreparative columns. For carbon dioxide as the mobile phase, cyclone or solid-phase adsorption traps are suitable trapping devices. Cyclone traps are designed for isolating single products or sections from a chromatogram, and are unsuitable for collecting several fractions from repetitive injections. The capacity of sorbent traps is

886

The Essence of Chromatography

limited by the amount and retentive properties of the sorbent, and recovery of the sample requires off-line solvent desorption. For preparative-scale applications using mixtures of carbon dioxide and an organic solvent, high-pressure trapping and solvent trapping are the preferred methods. Highpressure trapping involves transfer of fractions to a pressurized vessel, and condensation of the sample at low temperature. Typically, the temperature is low enough to solidify the mobile phase. Subsequently, the pressure is slowly released, and the carbon dioxide evaporated, leaving the sample in modifier as solvent. For solvent trapping, an additional solvent flow of 5 to 40% of the mobile phase flow is added upstream from the pressure control valve and after the detector. The solvent flow maintains the sample in solution after depressurization of the mobile phase. The solvent enhanced mobile phase is collected in a vessel partially filled with a small amount of solvent. The product stream is introduced below the solvent level, and the carbon dioxide allowed to escape. The collection vessel is cooled to increase the collection efficiency.

11.5 GAS CHROMATOGRAPHY The most common types of preparative-scale gas chromatographic instruments are based on packed column technology [211-217]. For simple separations, a short, wide packed column is generally used (e.g. 1-3 m x 6-10 cm I.D.) For difficult separations, higher column efficiencies are required, and long, narrow packed columns (e.g. 10-30 m X 0.5-1.5 cm I.D.), or multidimensional chromatographic techniques (section 3.8.1) [218], are used. Wide bore packed columns cannot be coiled, and unless a purpose designed preparative-scale gas chromatograph is available, the choice may be limited to long, narrow, coiled columns. Other critical considerations include the method of injection, detection, and sample collection. Syringe injection is possible for small volumes of volatile samples (e.g. 100 |xl) using slow injection. Typical sample volumes are 0.1 to 10.0 ml, which are too large for injection by conventional techniques developed for analytical packed columns. In this case, an automated injector employing pneumatic transfer from a reservoir through a capillary restrictor, a pneumatic piston pump, or syringe pump, is used. The injection process is controlled by time; a necessity owing to the limited thermal capacity of injection block heaters, and their inability to flash vaporize large sample volumes. To assist solvent vaporization, an evaporation device between the injector and column, is usually installed. This is often a tube, heated separately from the column oven, and packed with glass beads or metal spheres to increase its thermal capacity while simultaneously reducing dead volume effects. The evaporation unit is usually heated to a temperature 50C above the boiling point of the least volatile sample component. Vaporization problems can be circumvented by slow, on-column injection without carrier gasflow.This is often the most practical solution for injecting large sample volumes using analytical instruments. The principal disadvantage is the possibility of stripping the stationary phase from the head of the column. For on-column injection of liquid samples, a substantial pressure change may result from

Laboratory-Scale Preparative Chromatography

887

M
>uK
B
Cold w a l l ^ Hot wall Glass Rod Helix Annular Space B Insulation

50 W Heater

L . Coolant

Figure 11.12. Traps for sample recovery from a gas chromatographic effluent. A, U-tube; B, simple trap; C, multiple temperature gradient trap; D, Volman trap.

sample vaporization. Backflushing of sample into the carrier gas lines can be a problem, solved by positioning a backflush or needle valve in the carrier gas line. Separations are usually performed isothermally. Temperature programming is impossible with wide bore columns owing to the uneven radial temperature gradients generated across the column diameter. Temperature programming is only possible with long, narrow bore columns, and then only slowly. Separated components are detected by a thermal conductivity or flame ionization detector connected to the column via a splitter, so that only a few percent of the total column flow is diverted to the detector. Collection of sample vapors in the carrier gas effluent is performed automatically or manually, initiated by the detector signal. In purpose-designed instruments, injection and sample collection are automated for unattended operation. The exit from the detector splitter is usually led out through a side wall of the oven and thermostatted to avoid condensation of the sample. Several methods are used to trap samples, including packed and unpacked cold traps, solution and entrainment traps, total effluent and adsorption traps, Volman traps, and electrostatic precipitators. Figure 11.12 [219-223].

888

The Essence of Chromatography

The minimum requirement for efficient sample collection is that the partial pressure of the solute entering the trap must be greater than its equilibrium vapor pressure with the condensed phase at the trap temperature. The equilibrium vapor pressure of the solute can be reduced by lowering the temperature, by dissolving the solute in a solvent, or by adsorbing it onto a material of high surface area. However, the most common approach is to reduce the temperature of the effluent, either in a trap filled with a material such as glass beads or column packing, or in a simple open tube trap. In practice, the residence time of the sample in the trap may be too short to reach equilibrium. In which case, the sample will tend to pass directly through the trap, either as a supersaturated vapor or as a fog composed of small liquid droplets. Cold trapping of samples with boiling points greater than about 150C is almost inevitably accompanied by the formation of aerosols, which are then swept through the trap. Cooling the effluent through a gradual temperature gradient rather than subjecting it to an abrupt temperature change can minimize this effect. In addition, turbulent flow within the trap facilitates the collision of liquid droplets with the wall, thereby inducing their retention. The Volman trap consists of a double-walled vessel within which turbulence is created by maintaining the walls at different temperatures. The Volman trap is particularly useful for recovering gram-scale quantities. Electrostatic precipitators are also used to trap large fractions. As aerosols contain large numbers of charged droplets, the trapping efficiency is improved by passing the cooled effluent through a large electrostatic field. For occasional sample trapping, a simple packed or empty glass U-tube cooled in a dry ice-acetone or liquid nitrogen coolant, usually suffices. The stationary phase for the separation is identified from preliminary analytical separations. The support material is selected with the column length in mind. It is normally coarse with a narrow size distribution (e.g. 35-40 or 80-100 mesh). At typical carrier gas flow rates of 100 to 1000 ml/min, depending on the column internal diameter, coarse packing is needed to allow operation at reasonable inlet pressures. Since the stationary phase loading is generally higher and column length longer than is typical for analytical separations, operation at higher temperatures is also common. Reproducibly packing columns with reasonable efficiency is more difficult for wider bore columns [212,224-226]. This inferior performance is due to the uneven radial packing density resulting from particle size segregation, with the larger particles being closer to the wall. In addition, the packing in this region is less dense, due to the physical constraint of the wall. To minimize this problem, the shake, turn, and pressure method is recommended [224]. The column is shaken in the radial direction and rotated along its long axis while being packed, and periodically pressurized. Packing under suction with vertical tamping, affords a faster approach [226]. Since column permeability is an important consideration in preparative-scale gas chromatography, it is essential to minimize the production of fines during the packing process. Open tubular columns have a lower sample capacity than packed columns, but are useful for isolating a few milligrams of components from complex mixtures under high-resolution conditions [227-232]. Automated sequencing of injection and fraction collection to accumulate sufficient sample is generally used. Short lengths of wide bore

Laboratory-Scale Preparative Chromatography

889

capillary tubes, coated with a thick film of stationary phase, are efficient traps for room temperature sample collection. Either solvent or thermal desorption is used for sample recovery. Because of narrow peak widths, rapid switching between traps and/or waste is required to avoid cross contamination.

1 1 . 6 COUNTERCURRENT CHROMATOGRAPHY
Countercurrent chromatography is a preparative-scale technique, using two mutually saturated immiscible liquid phases in an open column for separations governed by the difference in partition coefficients for sample components [6,233-237]. One phase (the mobile phase) continuously passes through the other (the stationary phase), which is permanently retained in the column by an effective combination of column configuration and centrifugal force. The main advantages of countercurrent chromatography compared with column liquid chromatography are the absence of a solid support, and a higher loading capacity for columns of equal volume. Laboratoryscale instruments with a column volume of 100-400 ml can handle sample sizes of 0.5-10 g per injection. Process-scale machines are capable of handling larger sample sizes, but there is little information available concerning the design and operation of these instruments [238,239]. The major use of countercurrent chromatography is for the isolation of labile natural products, and to a lesser extent, the extraction and purification of pharmaceutical compounds, biopolymers, metal ions, and industrial chemicals [36,234-237,240]. All countercurrent chromatographs are either hydrostatic or hydrodynamic equilibrium systems [233,235,241,242]. Several early versions of these chromatographs became essentially redundant with the development of the high-speed countercurrent chromatograph [6,235,236], and the centrifugal partition chromatograph [234,243]. The high-speed countercurrent chromatograph (HSCCC) is an example of a hydrodynamic equilibrium system. Figure 11.13. A cylindrical column holder is equipped with a gear that is coupled to an identical stationary gear mounted at the central axis of a centrifuge. This gear arrangement produces a synchronous planetary motion of the column holder (i.e. revolution around the central axis of the centrifuge and rotation about its own axis), both at the same angular velocity, and in the same direction. This planetary motion prevents the tubes introducing and receiving the mobile phase from twisting, allowing continuous elution through the rotating column without the use of a rotary seal device. The column is directly wound around the holder to form single or multiple coiled layers. The sample is introduced from either end of the column, or at its midpoint. The motion of the column generates an Archimedian screw force. Independent of density, the contents of the column are driven towards one end called the head (the other end is referred to as the tail) [244]. The forces acting on the coiled column cause vigorous mixing of the two phases with formation of a succession of mixing and demixing zones in each successive turn of the coil. When hydrodynamic equilibrium is established, the phase ratio is constant, and only the mobile phase moves through the

890

The Essence of Chromatography

Axis of Rotation

- Axis of Revolution

Figure 11.13. Design principle of the multilayer coil planet centrifuge for high-speed countercurrent chromatography. (From ref. [235]. John Wiley & Sons)

column. The velocity of the mobile phase is a linear combination of the flow rate and the rotation speed [245]. The stationary phase is stable only as long as the centrifugal field is applied [246]. There is a minimum flow rate at which the stationary phase is forced out of the column. For the same operating conditions, flow rates about 5% of this value are desirable for stable operation and ease of sample introduction. For flow rates greater than 10% of the minimum displacement flow rate, the stationary phase is only weakly retained, and sample injection may result in stationary phase loss. Centrifugal partition chromatographs employ a complex arrangements of interconnected channels and narrow ducts in a centrifugal force field to promote mixing and separation of the two liquid phases. Figure 11.14. The column consists of a number of serially connected channels engraved in a plastic plate. Several plates are put together to form a cartridge attached to the rotor of a centrifuge. The column length is variable, and depends on the number of connected cartridges. The liquid stationary phase is retained in the channels and the mobile phase, which can be either the denser or lighter solvent phase, passes through it and collects in the ducts. At present, it is uncertain how the mobile phase actually flows through the stationary phase. While in the channels, the mobile phase may be continuous, like a film against one of the walls, flowing very quickly, then accumulating in a pool, before transfer to the next channel through a duct. Alternatively, it may be discontinuous, falling as droplets in the gravitational field, which are transferred immediately to the duct upon arriving at the channel outlet. A significant pressure drop is generated over the column during normal operation of the chromatograph. The mobile phase enters and leaves the column via rotary seals, which limit the maximum operating pressure to about 60 atm. at typical rotational speeds of 500-2000 rpm. Pressure constraints restrict the number of usable solvent systems, because flow rates that are too slow result in low sample throughput and poor separation

Laboratory-Scale Preparative Chromatography

891

mobile phase with mixture of components

rotor rotary seal cartridge

_0_T^\F^mobile phase ^stationary phase -channel duct


Figure 11.14. Design principle of the centrifugal partition chromatograph. (From ref. [234]. Marcel Dekker).

efficiency. Visualization of the flow patterns in the centrifugal partition chromatograph indicate that better phase mixing leads to improved separation efficiency, and confirm that the key parameters acting on mixing are flow rate and rotational speed [243]. Comparing the two types of countercurrent chromatograph, those working largely on hydrostatic principles can retain almost any biphasic system, but promote less efficient mixing resulting in low separation performance (wide peaks). Hydrodynamic systems provide better mixing and typical column plate numbers of 300-1000, but stationary phase retention is not as good. Some solvent systems used in the centrifugal partition chromatograph may prove unstable in the high-speed countercurrent chromatograph. Preparative-scale separations are optimized on an analytical scale initially to minimize sample waste, and then scaled-up to a large volume column. A series of screening experiments are employed to identify the most selective solvent system for the separation in which the sample has favorable solubility [6,237,245,247]. The solubility criterion is important for preparative-scale separations, since the sample throughput is limited by the phase in which the sample is least soluble. In addition, the solvent system should provide a suitable range of partition coefficients for the target compounds, and

892

The Essence of Chromatography

Table 11.8 Two-phase solvent systems for countercurrent chromatography Solvent system Hexane-acetonitrile Hexane-methanol Hexane-acetonitrile-chloroform Hexane-ethanol-water Hexane-ethyl acetate-acetonitrile-methanol Hexane-ethyl acetate-methanol-water Chloroform-methanol-water Toluene-acetonitrile-water-ethanol Chloroform-methanol-0.2 M acetic acid Ethyl acetate-ethanol-water n-Butanol-water n-Butanol-acetic acid-water n-Butanol-ethyl acetate-water Composition (volume basis) 1:1 (wide range of compositions possible) 1:1 (wide range of compositions possible) 5:5:1 6:5:1 5:2:5:4 3:7:5:5 or 1:1:1:1 7:13:8 or 1:1:1 or 13:7:2 or 4:4:3 or 5:5:3 3:4:3:2 1:1:1 2:1:2 1:1 (wide range of compositions possible) 4:1:5 4:1:4

the solvent mixture selected as the stationary phase should be adequately retained in the column under the chosen operating conditions, Table 11.8. The partition coefficients are easily estimated from the ratio of the sample concentration in each phase of a small volume of the biphasic solvent system at equilibrium using UV absorption or thin-layer chromatography. In general, the most suitable range of partition coefficients, K, is 1 < K < 2 for the centrifugal partition chromatograph and 0.5 < K < 1 for the highspeed countercurrent chromatograph. Once the partition coefficient is established, the retention volume for each solute in the separation system is given by
VR = VMP + K ( V T - V M P )

(11.2)

where VR is the solute retention volume, VMP the volume of mobile phase in the column (estimated as the retention volume of an unretained solute) and VT the total column volume. The partition coefficient is defined as the ratio of the solute concentration in the stationary phase divided by its concentration in the mobile phase at equilibrium. Since either phase of a two-phase system can be used as the mobile or stationary phase, the choice of operating conditions affects the value of the retention volume in Eq. (11.2). To increase the loading capacity of ionizable compounds, pH-zone refining countercurrent chromatography is used [6,235,248,249]. This technique produces a train of highly concentrated rectangular zones, similar to those obtained in displacement chromatography (section 11.3.5). The method requires the presence of a retainer acid or base in the stationary phase and a suitable counterion in the mobile phase. An example of a typical system consists of methyl tert-butyl ether containing 0.05 M trifluoroacetic acid (retainer acid) as the stationary phase and water containing 0.015 M ammonia (pH = 10) as the mobile phase for the separation of organic acids. Zone sharpening is a result of the recycling of the organic acids between the two phases, where they are alternately ionized and neutralized. The result of these exchanges is the formation of a series of solute zones with sharp boundaries moving through the column at the same speed. Each zone

Laboratory-Scale Preparative Chromatography

893

consists of a single compound with its own specific pH. The pKa and hydrophobicity of each compound determine the elution order of the zones. Charged minor components are concentrated at the boundaries of the major zones and elute as sharp peaks. The method is also applicable to the preparative-scale separation of enantiomers using an ionic chiral selector in the stationary phase. The advantages of pH-zone refining for preparative-scale applications are an increased sample loading capacity for a given column (ten fold or more), the production of highly concentrated fractions, the concentration of minor components making them easier to detect, and the possibility of detecting compounds without a chromophore by monitoring pH. For compounds with similar hydrophobicity a minimum difference of 0.2 pKa units is required. The method is not suitable for trace analysis, and each sample component should be present at > 0.1 mM.

11.7 REFERENCES
[1] B. A. Bidlingmeyer (Ed.), Preparative Liquid Chromatography, Elsevier, Amsterdam, 1987. [2] G. Ganetsos and P. E. Barker, Preparative and Production Scale Chromatography, Dekker, New York, NY, 1993. [3] R. J. P. Cannell (Ed.), Natural Product Isolation, Humana Press, Totawa, NJ, 1998. [4] K. Hostettmann, A. Marston and M. Hostettmann, Preparative Chromatography Techniques. Applications in Natural Product Isolation, Springer-Verlag, Berlin, 1998. [5] F. W. Collins, J. Clin. Ligand Assay 23 (2000) 273. [6] I. D. Wilson, E. R. Adlard, M. Cooke and C. E Poole (Eds.), Encyclopedia of Separation Science, Academic Press, London, 2000. [7] G. Subramaniann (Ed), Process Scale Liquid Chromatography, VCH, Weinheim, 1995. [8] G. Sofer and L. Hagel, Handbook of Process Chromatography, Academic Press, San Diego, CA,1997. [9] B. Pynnonen, J. Chromatogr. A 827 (1998) 143. [10] G. Subramanian (Ed.), Bioseparation and Bioprocessing. A Handbook, Wiley-VCH, Weinheim, vol. 1 and 2, 1998. [11] K. Valko (Ed.), Separation Methods in Drug Synthesis and Purification, Elsevier, Amsterdam, 2000. [12] J. Sherma and B. Fried (Ed), Handbook of Thin-Layer Chromatography, Dekker, New York, NY, 1995. [13] Sz. Nyiredy (Ed), Planar Chromatography. A Retrospective View for the Third Millennium, Springer Scientific Publisher, Budapest, Hungary, 2001. [14] Sz. Nyiredy, Anal. Chim. Acta 238 (1990) 83. [15] L. Botz, Sz. Nyiredy and O. Sticher, J. Planar Chromatogr. 3 (1990) 10. [16] Sz. Nyiredy, J. AOAC Int. 84 (2001) 1219. [17] W. C. Still, M. Kahn and A. Mitra, J. Org. Chem. 43, (1978) 2923. [18] I. Chappell and R E. Baines, Biochromatography 10 (1991) 236. [19] R Claeson, F. Tuchinda and V. Reutrakul, J. Sci. Soc. Thailand 19 (1993) 73. [20] G. A. Potter, J. Chromatogr. A 675 (1994) 237. [21] C. Edwards, L. A. Lawton, S. M. Coyle and R Ross, J. Chromatogr. A 734 (1996) 163. [22] M.-L. Milat and J.-R Blein, J. Chromatogr. A 699 (1995) 277. [23] T.-S. Li, J.-T. Li and H.-Z. Li, J. Chromatogr. A 715 (1995) 372. [24] S. J. Grieb, S. A. Matlin and A. M. Belenguer, J. Chromatogr. A 728 (1996) 195. [25] S. G. Westerbuhr and K. L. Rowlen, J. Chromatogr. A 886 (2000) 9. [26] S. W. Pelletier, H. R Chokshi and H. K. Desai, J. Nat. Prod. 49 (1986) 892. [27] H. Henke, Preparative Gel Chromatography on Sephadex LH-20, Huethig, Heidelberg, 1995.

894 [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70]

The Essence of Chromatography T. Leutert and E. von Arx, J. Chromatogr., 292 (1984) 333. G. C. Zogg, Sz. Nyiredy, and O. Sticher, J. Liq. Chromatogr., 12 (1989) 2031 and 2049. J. C. Parle, K. Raghavan and S. J. Gibbs, J. Chromatogr. 945 (2002) 65. T. Wennberg, J. P. Rauha and H. Vuorela, Chromatographia 53 (2001) S-240. P Renold, E. Madero and T. Maetzke, J. Chromatogr. A 908 (2001) 43. B. Porsch, J. Chromatogr. A 658 (1994) 179. L. R. Snyder, J. J. Kirldand and J. L. Glajch, Practical HPLC Method Development, Wiley, New York, NY, 1997. P. R. Brown and R. A. Hartwick (Eds.), High Performance Liquid Chromatography, Wiley, New York, NY, 1989. M. Verzele, Anal. Chem. 62 (1990) 265A. E G. Lode, A. Rosenfeld, Q. S. Yuan, T. W. Root and E. N. Lightfoot, J. Chromatogr. A 796 (1998) 3. Q. S. Yuan, A. Rosenfeld, T. W. Root, D. J. Klingenberg and E. N. Lightfoot, J. Chromatogr. A 831 (1999) 149. R. A. Shalliker, B. S. Broyles and G. Guiochon, J. Chromatogr. A 865 (1999) 73 and 83. F. Dardoize, G. Clodie and J. Chevalet, Anal. Bioanal. Chem. 372 (2002) 372. L. Zeng, L. Burton, K. Yung, B. Shushan and D. B. Kassel, J. Chromatogr. A 794 (1998) 3. W. M. A. Niessen, J. Lin and G. Bondoux, J. Chromatogr. A 2002 (in press) H. Cai, J. P. Kiplinger, W. K. Goetzinger, R. O. Cole, K. A. Laws, M. Foster and A. Schrock, Rapid Commun. Mass Spectrom. 16 (2002) 544. M. Bauser, J. Chromatogr. Sci. 40 (2002) 292. M. Verzele, M. de Coninck, J. Vindevogel, and C. Dewaele, J. Chromatogr., 450 (1988) 47. G. Guiochon and A. Katti, Chromatographia, 24 (1987) 165. K. Jones, Chromatographia 25 (1988) 547. G. Guiochon, T. Farkas, H. Guan-Sajonz, J. H. Koh, M. Sarker, B. J. Stanley and T. Yun, J. Chromatogr. A 762(1997)83. M. Schulte, D. Lubda, A. Delp and J. Dingenen, J. High Resolut. Chromatogr. 23 (2000) 100. M. Schulte and J. Dingenen, J. Chromatogr. A 923 (2001) 17. S. Ghose and S. M. Cramer, J. Chromatogr. A 928 (2001) 13. C. Orihuela, R. Fronek, L. Miller, D. Hondo and J. Murphy, J. Chromatogr. A 827 (1998) 193. D. R. Wu and K. Lohse, J. Chromatogr. A 658 (1994) 381. J. H. Koh, B. S. Broyles, H. Guan-Sajonz, M. Z. C. Hu and G. Guiochon, J. Chromatogr. A 813 (1998) 223. D. E. Cherrak, M. Al-Bokari, E. C. Drumm and G. Guiochon, J. Chromatogr. A 943 (2002) 15. G. Guiochon, E. Drumm and D. E. Cherrak, J. Chromatogr. A 835 (1999) 41. D. E. Cherrak and G. Guiochon, J. Chromatogr. A 911 (2001) 147. D. P Gervais, W S. Laughinghouse and G. Carta, J. Chromatogr. A 708 (1995) 41. M. Sarker, T. Yun and G. Guiochon, J. Chromatogr. A 728 (1996) 3. U. D. Neue, HPLC Columns. Theory, Technology, and Practice, Wiley-VCH, New York, NY, 1997. L. Miller, C. Orihuela, R. Fronek, D. Honda and O. Dapremont, J. Chromatogr. A 849 (1999) 309. E. R. Francotte, J. Chromatogr. A 906 (2001) 379. G. Piquet, M. Gatti, L. Barbero, S. Traversa, P Caccia and P Esposito, J. Chromatogr. A 944 (2002) 141. L. R. Snyder, G. B. Cox and P E. Antle, Chromatographia 24 (1987) 82. J. H. Knox and H. M. Pyper, J. Chromatogr. 363 (1986) 1. G. Guiochon, S. Golshan-Shirazi and A. M. Katti, Fundamentals of Preparative and Nonlinear Chromatography, Academic Press, San Diego, CA, 1994. S. H. Ngiam, Y. H. Zhou, M. K. Turner and N. J. Titchener-Hooker, J. Chromatogr. A 937 (2001) 1. L. R. Snyder, J. W. Dolan, and G. B. Cox, J. Chromatogr., 483 (1989) 63, 85, and 95. G. B. Cox, L. R. Snyder, and J. W Dolan, J. Chromatogr., 484 (1989) 409, 425, and 437. M. Czok and G. Guichon, Anal. Chem., 62 (1990) 189.

Laboratory-Scale Preparative Chromatography [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [109] [110] [Ill]

895

S. Golshan-Shirazi and G. Guiochon, J. Chromatogr., 506 (1990) 495. A. M. Katti and G. A. Guiochon, Adv. Chrom. 31 (1992) 1. R. M. Nicoud and A. Seidel-Morgenstem, Isolation Purif. 2 (1996) 165. O. Lisec, P. Hugo and A. Seidel-Morgenstem, J. Chromatogr. A 908 (2001) 19. T. Fomstedt and G. Guiochon, Anal. Chem. 73 (2001) 609A. K. Mihlbacher, K. Kaczmarski, A. Seidel-Morgenstem and G. Guiochon, J. Chromatogr. A 955 (2002) 35. S. Boclcer, M. Mazzotti, M. MorbideUi, C. Fleury and B. SchenJcel, Sep. Sci. Technol. 37 (2002) 1725. S. Golshan-Shirazi and G. Guiochon, Anal. Chem., 62 (1990) 217. Z. Ma and G. Guiochon, Anal. Chem., 62 (1990) 2330. J. Newburger and G. Guiochon, J. Chromatogr., 523 (1990) 63. S. Golshan-Shirazi and G. Guiochon, J. Chromatogr. 517 (1990) 229. A. Felinger and G. Guiochon, J. Chromatogr. A 796 (1998) 59. A. Felinger and G. Guiochon, Biotechn. Bioeng. 41 (1993) 134. S. Ramanan and A. Velayudhan, J. Chromatogr. A 830 (1999) 91. N. Tugcu, S. S. Bae, J. A. Moore and S. M. Cramer, J. Chromatogr. A 954 (2002) 127. V. Natarajan, B. W. Bequette and S. M. Cramer, J. Chromatogr. A 876 (2000) 51 and 63. B. Schmidt, C. Wandrey and R. Freitag, J. Chromatogr. A 944 (2002) 149. R. R. Deshmukh, W. E. Leitch and D. L. Cole, J. Chromatogr. A 806 (1998) 77. Y. Qi and J. Huang, J. Chromatogr. A 959 (2002) 85. A. S. Rathore and Cs. Horvath, J. Chromatogr. A 787 (1997) 1. L. H. Irgens, G. Farkas and Gy. Vigh, J. Chromatogr. 666 (1994) 603. G. Zhong and G. Guiochon, Adv. Chromatogr. 39 (1998) 351. S. Nagamatsu, K. Murazumi and S. MaJdno, J. Chromatogr. A 832 (1999) 55. G. Biressi, O. Ludemann-Hombourger, M. Mazzotti, R. M. Nicoud and M. MorbideUi, J. Chromatogr. A 876 (2000) 3. M. Schulte and J. Stmbe, J. Chromatogr. A 906 (2001) 399. O. Ludemann-Hombourger, G. Pigorini, R. M. Nicoud, D. S. Ross and G. Terfloth, J. Chromatogr. A 947 (2002) 59. A. Nicolaos, L. Muhr, R Gotteland, R.-M. Nicoud and M. Bailly, J. Chromatogr. A 908 (2001) 71 and 87. V. G. Mata and A. E. Rodrigues, J. Chromatogr. A 939 (2001) 23. S. Abel, M. Mazzotti and M. MorbideUi, J. Chromatogr. A 944 (2002) 23. D. Antos and A. Seidel-Morgenstem, J. Chromatogr. A 944 (2002) 77. J. Houwing, S. H. van Hateren, U. A. H. BiUiet, L. A. M. Lvan der Wielen, J. Chromatogr. A 952 (2002) 85. C. Mighorini, M. Mazzotti and M. MorbideUi, J. Chromatogr. A 827 (1998) 161. S. Khattabi, D. E. Cherralc, K. Mihlbachler and G. Guiochon, J. Chromatogr. A 893 (2000) 307. K. Mihlbacher, A. Jupke, A. Seidel-Morgenstem, H. Schmidt-Traub and G. Guiochon, J. Chromatogr. A 944 (2002) 3. A. Jupke, A. Epping and H. Schmidt-Traub, J. Chromatogr. A 944 (2002) 93. O. Di Giovanni, M. Mazzotti, M. MorbideUi, F. Denet, W. Hauck and R. M. Nicoud, J. Chromatogr. A 919 (2001) 1. F. Denet, W. Hauck, R. M. Nicoud, O. Di Giovanni, M. Mazzotti, J. N. Jaubert and M. MorbideUi, Ind. Eng. Chem. Res. 40 (2001) 4603. M. Juza, O. Di Giovanni, G. Biressi, V. Schurig, M. Mozzotti and M. MorbideUi, J. Chromatogr. A 813 (1998)333. G. Biressi, F. Quattrini, M. Juza, M. Mazzotti, V. Schurig and M. MorbideUi, Chem. Eng. Sci. 55 (2000)4537. G. Biressi, M. Mazzotti and M. MorbideUi, J. Chromatogr. A 957 (2002) 211. C. J. O. R. Morris and P. Morris, Separation Methods in Biochemistry, Wiley, New York, NY, 1976.

896

The Essence of Chromatography

[112] P. L. Dubin (Ed.), Aqueous Size-Exclusion Chromatography, Elsevier, Amsterdam, 1988. [113] K. K. Unger (Ed.), Packings and Stationary Phases in Chromatographic Techniques, Dekker, New York, NY, 1990. [114] R. Arshady, J. Chromatogr. 586 (1991) 181 and 199. [115] E. Boschetti, J. Chromatogr. A 658 (1994) 207. [116] M. Leonard, J. Chromatogr. B 699 (1997) 3. [117] L. Fisher, Laboratory Techniques in Gel Filtration Chromatography, Elsevier, Amsterdam, 1980. [118] J. Coupek and I. Vins, J. Chromatogr. A 658 (1994)391. [119] L. L. Lloyd, M. I. Millichip and J. M. Watkins, J. Chromatogr. A 944 (2002) 169. [120] P. R. Levison, S. E. Badger, D. W. Toome, M. Streater and J. C. Cox, J. Chromatogr. A 658 (1994) 419. [121] R R. Levison, C. Mumford, M. Streater, A. Brandt-Nielsen, N. D. Pathirana and S. E. Badger, J. Chromatogr. A 760 (1997) 151. [122] R DePhillips and A. M. Lenhoff, J. Chromatogr. A 883 (2000) 39. [123] A. Staby, I. H. Jensen and I. MoUerup, J. Chromatogr. A 897 (2000) 99. [124] A. Staby and I. H. Jensen, J. Chromatogr. A 908 (2001) 149. [125] R DePhillips and A. M. Lenoff, J. Chromatogr. A 933 (2001) 57. [126] M. Kaufmann, J. Chromatogr. B 699 (1997) 347. [127] R. Janzen, K. K. Unger, W. MuUer and M. T. W. Heam, J. Chromatogr. 522 (1990) 77. [128] W. MuUer, J. Chromatogr. 510 (1990) 133. [129] T. Kawasaki, J. Chromatogr. 544 (1991) 147. [130] Y. Shibusawa, J. Chromatogr. B 699 (1997) 419. [131] S. R. Shepard, C. Brickman-Stone, J. L. Schrimsher and G. Koch, J. Chromatogr. A 891 (2000) 93. [132] A. E. Rodrigues, J. Chromatogr. B 699 (1997) 47. [133] M. C. Garcia, M. L. Marina and M. Torre, J. Chromatogr. A 880 (2000) 169. [134] J. J. Meyers, O. K. Grosser and A. I. Liapis, J. Chromatogr. A 908 (2001) 35. [135] A. Geng and K.-C. Loh, J. Chromatogr. A. 918 (2001) 37. [136] S. K. Paliwal, Methods Enzymol, 270 (1996) 133. [137] R-E. Gustavsson, A. Axelsson and P-O. Larsson, J. Chromatogr. A 795 (1998) 199. [138] R-E. Gustavsson, A. Axelsson and P-O. Larsson, J. Chromatogr. A 830 (1999) 275. [139] R. Freitag, D. Frey and Cs. Horvath, J. Chromatogr. A 686 (1994) 165. [140] M. McCoy, K. Kalghatgi, F E. Regnier and N. Afeyan, J. Chromatogr. A 743 (1996) 221. [141] D. C. Nash and H. A. Chase, J. Chromatogr. A 776 (1997) 65. [142] D. Whitney, M. McCoy, N. Gordon and N. Afeyan, J. Chromatogr. A 807 (1998) 165. [143] D. C. Nash and H. A. Chase, J. Chromatogr. A 807 (1998) 185. [144] J. Horvath, E. Boschetti, L. Guerrier and N. Cooke, J. Chromatogr. A 679 (1994) 11. [145] D. C. Nash and H. A. Chase, J. Chromatogr. A 807 (1998) 185. [146] E. Boschetti, P. Girot, L. Guerrier and P. Santambien, J. Biochem. Biophys. Methods 31 (1996) 15. [147] S. Bouhallab, G. Henry and E. Boschetti, J. Chromatogr. A 724 (1996) 137. [148] D. Faman, D. D. Frey and Cs. Horvath, J. Chromatogr. A 959 (2002) 65. [149] M. S. Verrali (Ed.), Downstream Processing of Natural Products, Wiley, New York, NY, 1996. [150] S. R. Narayanan, J. Chromatogr. A 658 (1994) 237. [151] T. M. Phillips, Adv. Chromatogr. 29 (1989) 133. [152] A. Lyddiatt, Curr. Opinion Biotechnol. 13 (2002) 95. [153] H. S. Andersson, J. G. Karlsson, A.-C. Koch-Schmidt, K. Mosbach and I. A. Nichols, J. Chromatogr. A 848 (1999) 39. [154] T. Takeuchi and J. Haginaka, J. Chromatogr. B 728 (1999) 1. [155] R K. Owens, L. Karlsson, E. S. M. Lutz and L. I. Andersson, Trends Anal. Chem. 18 (1999) 146. [156] L. I. Andersson, J. Chromatogr. B 745 (2000) 3. [157] D. Tong, C. Hetenyl, Z. Bikadi, J. R Gao and S. Hjerten, Chromatographia 54 (2001) 7. [158] K. Haupt, Analyst 126 (2001) 747.

Laboratory-Scale Preparative Chromatography [159] [160] [161] [162] [163] [164] [165] [166] [167] [168] [169] [170] [171] [172] [173] [174] [175] [176] [177] [178] [179] [180] [181] [182] [183] [184] [185] [186] [187] [188] [189] [190] [191] [192] [193] [194] [195] [196] [197] [198] [199] [200]

897

M. G. Weller, Fresenius J. Anal. Chem. 366 (2000) 635. D. S. Hage, J. Chromatogr. B 768 (2002) 3 D. S. Hage, Clin. Chem. 45 (1999) 593. V. Pichon, M. Bouzige, C. Miege and M.-C. Hennion, Trends Anal. Chem. 18 (1999) 219. D. Josic and Y. P. Lim, Food Technol. Biotechnol. 39 (2001) 215. T. Bumouf and M. Radosevich, J. Biochem. Biophys. Methods 49 (2001) 575. H. Gadgil, S. A. Oak and H. V. Jarret, J. Biochem. Biophys. Methods 49 (2001) 607. T. T. Ngo (Ed.), Molecular Interactions in Bioseparations, Plenum Press, New Yorlc, NY, 1993. E. Boschetti, J. Biochem. Biophys. Methods 49 (2001) 361. X.-C. Liu and W. H. Scouten, J. Chromatogr. 687 (1994) 61. G. T. Hermanson, A. K. Mallia and P. K. Smith (Ed.), Immobilized Affinity Ligand Techniques, Academic Press, New York, NY, 1992. J. Turkova, J. Chromatogr. B 722 (1999) 11. M. Saleemuddin, Appl. Biochem. Biotechnol. 64 (1999) 203. M. Nisnevitch and M. A. Firer, J. Biochem. Biophys. Methods 49 (2001) 467. J. B. Wheatley and D. E. Schmidt, J. Chromatogr. A 849 (1999) 1. Y D. Clonis, N. E. Labrou, V. Ph. Kotsira, C. Mazitsos, S. Melissis and G. Gogolas, J. Chromatogr. A 891(2000)33. C. R. Lowe, S. J. Burton, N. P. Burton, W. K. Alderton, J. M. Pitts and J. A. Thomas, Trends Biotechnol. 10 (1992) 442. N. E. Labrou, E. Eliopoulos and Y. D. Clonis, Biotechnol. Bioeng. 63 (1999) 321. C. R. Lowe, A. R. Lowe and G. Gupta, J. Biochem. Biophys. Methods 49 (2001) 561. A. Denizli and E. Piskin, J. Biochem. Biophys. Methods 49 (2001) 391. S. Zhang and Y. Sun, J. Chromatogr. A 957 (2002) 89. P. R. Morrill, G. Gupta, K. Sproule, D. Winzor, J. Christensen, L MoUerup and C. R. Lowe, J. Chromatogr. B 774 (2002) 1. G. S. Chaga, J. Biochem. Biophys. Methods 49 (2001) 313. V. Gaberc-Porekar and V. Menart, J. Biochem. Biophys. Methods 49 (2001) 335. S. A. Lopatin and V. P. Varlamov, Appl. Biochem. Microbiol. 31 (1995) 221. R. D. Johnson and F. H. Arnold, Biotechnol. Bioeng. 48 (1995) 437. T. T. Yip and T. W. Hutchins, Mol. Biotechnol. 1 (1994) 151. R H. Arnold, Biotechnol. 9 (1991) 151. J. Crowe, H. Dobeh, R. Gentz, E. Hochuh, D. Stuber and K. Henco, Methods Mol. Biol. 31 (1994) 371. P. Armisen, C. Matero, E. Cortes, J. L. Barredo, F. Salto, B. Diez, L. Rodes, J. L. Garcia, R. FemandezLafuente and J. M. Gulsan, J. Chromatogr. A 848 (1999) 61. W. S. Choe, R. H. Clemmitt, H. A. Chase and A. P J. Middelberg, J. Chromatogr. A 953 (2002) 111. L. D. Holmes and M. R. Schiller, J. Liq. Chromatogr. fe Rel. Technol. 20 (1997) 123. C. Berger and M. Perrut, J. Chromatogr. 505 (1990) 37. K. D. Bartle, C. D. Bevan, A. A. Clifford, S. A. Jafar, N. Malak and M. S. Verrall, J. Chromatogr. A 697 (1995) 579. M. Hanson, Chromatographia 40 (1995) 139. K. Anton and C. Berger, Supercritical Fluid Chromatography with Packed Columns. Techniques and Applications, Dekker, New York, NY, 1998. C. M. Caude and D. Thiebault (Eds.), Practical Supercritical Fluid Chromatography and Extraction, Harwood, Amsterdam, 1999. F. Geiser, M. Schultz, L. Betz, M. Shaimi, J. Lee and W. Champion, J. Chromatogr. A 865 (1999) 227. K. Coleman, R. Boutant and F. Verillon, Isolation Purification 3 (1999) 9. G. Pettinello, A. Bertucco, P. Pallado and A. Stassi, J. Supercrit. Fluids 19 (2000) 51. E. Reverchon, J. Supercrit. Fluids 10 (1997) 1. J. W. King, Trends Anal. Chem. 14 (1995) 474.

898

The Essence of Chromatography

[201] M. Perrut, J. Chromatogr. A 658 (1994) 293. [202] T. A. Berger, K. Fogelman, T. Staats, P. Bente, I. Crocket, W. Farrell and M. Osonubi, J. Biochem. Biophys. Methods 43 (2000) 87. [203] T. Wang, M. Barber, I. Hardt and D. B. Kassel, Rapid Commun. Mass Spectrom. 15 (2001) 2067. [204] G. Cretier, J. Naffati and J. L. Rocca, J. Chromatogr. A 670 (1994) 173. [205] Y. Hirata, Y. Kawaguchi and K. Kitano, Chromatographia 40 (1995) 42. [206] G. Cretier, R. Majdalani, J. Neffati and J. L. Rocca, Chromatographia 38 (1994) 330. [207] D. M. Heaton, K. D. Bartle, P Myers and A. A. CUfford, J. Chromatogr. A 753 (1996) 306. [208] M. D. Burford, A. A. Clifford, K. D. Bartle, C. M. Cowey and N. G. Smart, J. Chromatogr. A 738 (1996)241. [209] J. W. King and Z. Zhang, Chromatographia 51 (2000) 467. [210] C. Turner, C. S. Eskilsson and E. Bjorkland, J. Chromatogr. A 947 (2002) 1. [211] R. Bonmati, G. Chapelt-Letoumeux, and G. Guiochon, Sep. Sci. Technol., 19 (1984) 113. [212] A. Zlatkis and V. Pretorius (Eds.), Preparative Gas Chromatography, Wiley, New York, NY, 1971. [213] E. Grushka (Ed.), Preparative-Scale Chromatography, Dekker, New York, NY, 1989. [214] G. W. A. Rijnders, Adv. Chromatogr., 3 (1966) 215. [215] G. Fuchs and M. Perrut, J. Chromatogr A 658 (1994) 437. [216] V. Schurig and H. Grosenick, J. Chromatogr. A 666 (1994) 617. [217] F. Quattrini, G. Bieressi, M. Juza, M. Mazzotti, C. Fuganti and M. Morbidelli, J. Chromatogr. A 865 (1999)201. [218] G. Schomburg, H. Kotter, D. Staffels, and W. Reissig, Chromatographia, 19 (1985) 382. [219] G. Magnusson, J. Chromatogr., 109 (1975) 393. [220] A. A. Casselman and R. A. B. Bannard, J. Chromatogr, 90 (1974) 185. [221] H. T. Badings, J. J. G. van der Pol, and J. G. Wassink, Chromatographia, 8 (1975) 440. [222] A. B. Attygale and E. D. Morgan, Anal. Chem., 55 (1983) 1379. [223] S. T. Adam, J. High Resolut. Chromatogr., 10 (1987) 369. [224] J. Albrecht and M. Verzele, J. Chromatogr. Sci., 8 (1970) 586. [225] J. Albrecht and M. Verzele, J. Chromatogr. Sci., 9 (1971) 745. [226] C. E. Reese and E. Grushka, Chromatographia, 8 (1975) 85. [227] J. Roeraade, S. Blomberg, and H. D. J. Pietersma, J. Chromatogr., 356 (1986) 271. [228] B. V. Burger and Z. Munro, J. Chromatogr., 262 (1983) 95. [229] R. T. Ghijsen and H. Poppe, J. High Resolut. Chromatogr., 11 (1988) 271. [230] S. Blomberg and J. Roeraade, J. Chromatogr., 394 (1987) 443. [231] B. V. Burger, M. Le Roux, and W J. G. Burger, J. High Resolut. Chromatogr., 13 (1990) 777. [232] T. I. Eglinton, L. I. Aluwihare, J. F. Bauer, E. R. M. Druffel and A. P McNichol, Anal. Chem. 68 (1996) 904. [233] W. D. Conway, Countercurrent Chromatography. Apparatus, Theory and Applications, VCH, New York, NY, 1990. [234] A. P. Foucault (Ed.), Centrifugal Partition Chromatography, Dekker, New York, NY, 1995. [235] Y Ito and W. D. Conway (Eds.), High-Speed Countercurrent Chromatography, Wiley, New York, NY, 1996. [236] J.-M. Menet and D. Thiebaut (eds.), Countercurrent Chromatography, Dekker, New York, NY, 1999. [237] A. Berthod and B. Billardello, Adv. Chromatogr 40 (2000) 503. [238] I. A. Sutherland, A. J. Booth, L. Brown, B. Kemp, H. Kidwell, D. Games, A. S. Graham, G. G. Guillon, D. Hawes, M. Hayes, L. Janaway, G. J. Lye, P. Massey, C. Preston, P Shering, T. Shoulder, C. Strawson and R Wood, J. Liq. Chromatogr. Rel. Technol. 24 (2001) 1533. [239] I. A. Sutherland, L. Brown, A. S. Graham, G. G. Guillon, D. Hawes, L. Janaway, R. Whitesides and P. Wood, J. Chromatogr. Sci. 39 (2001) 21. [240] K. Hostettmann and A. Marston, J. Liq. Chromatogr. Rel. Technol. 24 (2001) 1711. [241] N. B. Mandava and Y. Ito (Eds.), Countercurrent Chromatography, Theory and Practice, Dekker, New York, NY, 1988.

Laboratory-Scale Preparative Chromatography [242] [243] [244] [245] [246] [247] [248] [249]

899

Y. Ito, J. Chromatogr. 538 (1991) 3. L. Marchal, A. Foucault, G. Patissier, J. M. Rosant and J. Legrand, J. Chromatogr. A 869 (2000) 339. P. L. Wood, B. Jaber and I. A. Sutherland, J. Liq. Chromatogr. Rel. Technol. 24 (2001) 1629. A. Berthod and B. Billardello, J. Chromatogr. A 902 (2000) 323. I. A. Sutherland, Q. Du and P Wood, J. Liq. Chromatogr. Rel. Technol. 24 (2001) 1669. J. Goupy, Chemomet. Intell. Laborat. Syst. 59 (2001) 53. Y. Ito and Y Ma, J. Chromatogr. A 753 (1996) 1. R. Pennance, C. Viron, S. Blanchard and M. Lafosse, J. Liq. Chromatogr. Rel. Technol. 24 (2001) 1575.