646 BIOCHEMICAL AND SYNTHETIC POLYMER SEPARATIONS main stream side stream side stream 01 2 3 4 55 6 Column volumes 25 20 15 % acetonitrile Figure 13.45 Preparative separation of rh-insulin. Column, 10 μm Zorbax™ Process Grade C8 (150 × 9.4-mm i.d.); load, 153 mg rh-insulin from proinsulin process; gradient, 17–29% acetonitrile in 0.25M acetic acid in six column volumes; flow rate, 0.3 mL/min. Fractions from 3.3 to 4.3 column volumes were pooled (mainstream). Fractions 3.2–3.3 and 4.4–5.4, plus the protein eluted during column regeneration (not shown), were combined as the side stream. Adapted from [162]. and 4.4–5.4 C-Vs plus recovered solvent from column regeneration) contained an additional 15% of the product (for re-separation)—for an overall insulin recovery of 97%. The initial purity of the sample for RPC separation was 91.5%. This example illustrates a common property of preparative separations: in contrast to analytical chromatography, prep-LC chromatograms may suggest poor separation of product from impurities, but when fractions are collected and analyzed, the results often are acceptable. 13.9.2.9 Scale-Up Scale-up experiments were carried out next, using six, successively larger, axial-compression columns, with the bed-volume increased from 10 mL to 80L, as summarized in Tables 13.9 and 13.10. In each separation the weight of insulin applied to the column was 14 to 15 g per L of column-volume (C-V), the flow rate was 1.5 C-V/h, and the gradient slopes were 2%/C-V. Minor changes in gradient slope were necessary to maintain column performance, as measured by product purity and recovery. Flow rates were increased in proportion to the volume of the column as the process moved from lab to pilot plant to production. Purity was 98.5, 98.6, and 98.6% at lab scale, pilot scale, and production scale, respectively, while mainstream yields were 82, 79, and 83%. It may seem remarkable that mainstream purities, recoveries, and elution volumes remained consistent from lab scale to production scale over a 10,000-fold range of column volumes, but this should be true of scale-up if carried out properly (Section 15.1.2.1). 13.9 LARGE-SCALE PURIFICATION OF LARGE BIOMOLECULES 647 Table 13.9 Column Sizes Used for rh-Insulin Scale-up Studies Column size (mm) Internal volume (L) Type Insulin load (g) 150×9.4 0.01 Fixed 0.15 300×22 0.12 Fixed 1.7 500×50 0.59 Fixed 8.9 450×150 8 Axial 120 500×300 35 Axial 525 500×450 80 Axial 1200 Source: Data from [162]. Table 13.10 Summary of rh-Insulin Scale-up Studies Column Size (Volume) Scale Operating Conditions Product Flow Rate Gradient Load Purity Yield (C-V/h) a (%B/C-V) b (mg/mL) 150 × 9.4-mm (10 mL) Lab 1.6 2.2% 13 98.5% 82% 350 × 150-mm (6.2 L) Pilot plant 1.5 2.0% 15 98.6 79% 570 × 300-mm (40 L) Pilot plant 1.4 2.1% 15 98.6 83% Source: Data from [162]. a C-V is empty column-volume. b Change in %B per column-volume (C-V) of mobile phase; proportional to gradient steepness. 13.9.2.10 Production-Scale Purification The purification of rh-insulin on a production scale was next carried out for both two-chain insulin and proinsulin. Separation was more challenging for the two-chain process because of a 20%-higher concentration of structurally related impurities. The production-scale conditions were: 48 × 30-cm column, 500 g insulin sample, and a gradient of 17–30% acetonitrile over 6 C-V at a flow rate of 1.4 C-V/h (0.8 L/min). The purity of the charge was 80%, and the mainstream purity was 98.5%. For insulin derived from the proinsulin process, the purity of the feedstock was higher (91%), which resulted in purified product of higher purity (99.1%). The corresponding purification of a 1-kg sample with a 48 × 45-cm column yielded comparable results. The product fractions from high-performance RPC were subjected to an additional purification step by SEC. Two lots of rh-insulin from each process were then compared to insulin purified by conventional chromatography that did not use high-performance RPC. The results showed that purification by RPC results in higher purity levels, equivalent biopotency, and comparable low levels of contamination by 648 BIOCHEMICAL AND SYNTHETIC POLYMER SEPARATIONS endotoxin or host cell protein. No siloxanes (potential breakdown products from the stationary phase) were detected. 13.9.3 General Requirements for Prep-LC Separations of Proteins Targets for yield and purity are needed, as well as methods (usually RPC) to measure yield and purity. Estimates of product weight (e.g., tons/year) and targets for the expected delivery time and purification cost are also required. A defined starting material is necessary, as well as an expected range of purity for the starting material. A standard for the purified product is helpful, but the standard can be produced as part of process development. Mass spectrometry and other physical methods can be useful for confirming the identity and purity of purified product. Additional development requirements may be imposed if the separation is intended to produce pharmaceutical products under current Good Manufacturing Practices (cGMP; [165, 166]) or ISO 9000 guidelines [167]. For protein products, a primary requirement is product stability during sep- aration. While, in principle, proteins can be denatured during RPC and renatured afterward [168], this approach has generally not been favored for purifications by means of RPC. Not only must the product retain its biological activity, there should also be no detectable chemical changes such as oxidation, deamidation, or cleavage of peptide bonds. Mass spectrometry methods greatly simplify the task of detecting the latter modifications. Methods for assessing the stability of proteins under various conditions are well established. It is often preferable to measure the stability profile of a protein product before chromatography development, so that time is not wasted exploring modes or mobile phases that are incompatible with the product. 13.10 SYNTHETIC POLYMERS Separations of synthetic polymers are usually carried out for one of two purposes: (1) determination of the molecular-weight distribution of a sample (Section 13.10.3.1), and/or (2) determination of different compound types or classes in the sample (Section 13.10.3.2). These applications differ fundamentally from other HPLC separations covered in this book. For this and other reasons the present section represents only an introduction to separations of synthetic polymers. 13.10.1 Background Synthetic polymers are large, man-made molecules; in all cases they are formed from one or more different monomers, which occur in the molecule many times. If a single monomer is polymerized, the result is a homopolymer (Fig. 13.46a); short-chain members of such a sample are referred to as oligomers. An example is ethylene as monomer, with polyethylene as the resulting polymer (C 2 H 5 –[C 2 H 4 –] p−2 –C 2 H 5 , for the reaction of p ethylene molecules to form a polymer molecule). If two (or more) different monomers are used to create a synthetic polymer, we have a copolymer (Fig. 13.46f ). Homopolymers can differ in length, as in Figure 13.46a, c, e. Molecular length is expressed either as the degree of polymerization p (i.e., where p is the number of monomeric units) or the molecular weight of the polymer. 13.10 SYNTHETIC POLYMERS 649 (a) (b) (f ) (i) (k) (c) (e) (h) (d) (g) (j ) (l) Figure 13.46 Different structures of synthetic polymers. (a), (c), and (e), linear homopoly- mers of varying length; (b), (c), and (d) are linear homopolymers with different end-groups; (g), functional groups have been introduced along the chain; (h), a cyclic homopolymer; (f ), a random copolymer, (i), a block copolymer; (k), a graft copolymer; (j)and(l) are branched homopolymers, with short-chain branching and long-chain branching, respectively. Finally, homopolymers can differ in molecular shape or topology. Linear molecules exist (Fig. 13.46a–f ), as well as cyclic (small) oligomers (Fig. 13.46h)andrelated polymers. Depending on the synthetic process, branches may be deliberately or accidentally introduced (Fig. 13.46g, j–l). Individual molecules can differ in the number of branches, their length, and their position in the molecules. In the case of copolymers the sequence of the monomers is relevant. If this sequence is determined by a purely statistical process, random copolymers are formed. The opposite extreme is that of block copolymers (Fig. 13.46i), where long sequences of a single monomer occur within the molecule. Finally, there are graft copolymers (Fig. 13.46k), where chains formed from a different monomer are attached to the primary polymer backbone. Polymer properties are affected by structural features, which are therefore important. A higher molecular weight generally leads to a stronger polymer. End-groups and functional groups are critically important for polymers used in reactive formulations such as adhesives, sealants, and coatings. Branching usually affects the processing properties of polymers. Block copolymers can have very dif- ferent properties compared to random copolymers. One structural property that is not depicted in Figure 13.47 is the degree of stereoregularity of the chain, usually called tacticity. In an atactic polymer, the monomeric units are oriented in a random fashion; in an isotactic (or syndiotactic) polymer, all monomers are positioned in the same (or alternating) direction. Stereoregular polymers usually exhibit a much higher degree of crystallinity. As a result atactic polypropene (or ‘‘polypropylene,’’ as it used to be called) is a soft plastic, whereas isotactic polypropene is hard and strong. 650 BIOCHEMICAL AND SYNTHETIC POLYMER SEPARATIONS 4 3 2 1 0 –1 –2 ln k 0 50 100 %B p= 5 p = 15 25 100 0 5 10 15 20 25 30 35 (min) 023 (min) (min) 0 5 10 15 20 (a) (b) (c) (d) Figure 13.47 Polymer retention and separation in i-LC. (a) Illustration of the reten- tion behavior of a homopolymeric series; (b) hypothetical illustration of the elution of a low-molecular-weight homopolymer with a 35-minute gradient; (c) similar separation as in (b) for a gradient time of 10 minutes; (d) hypothetical illustration of the separation of a polymeric blend by interactive liquid chromatography (i-LC). Just as the properties of a synthetic polymer are affected by its molecular struc- ture, so is polymer chromatographic behavior. This allows us to separate polymers based on molecular weight, chemical composition (functionality), stereoregularity, degree of branching, and so forth, as summarized in Table 13.11 for different separation modes. Many physical properties of synthetic polymers are important in relation to their chromatographic behavior, none more so than their solubility. For any kind of material to be separated by liquid chromatography, it must be dissolved completely (any agglomerates or particles are detrimental to chromatographic separation). 13.10 SYNTHETIC POLYMERS 651 Table 13.11 Effect of the Molecular Structure of Synthetic Polymers on their Chromatographic Behavior Molecular End Chemical Stereo- Branching Section Weight Groups Composition Regularity Size-exclusion chromatography • – a ◦ – ◦ 13.8 Interactive liquid chromatography ◦◦ / • b • –– c 13.10.3.2 Temperature-gradient interaction chromatography •• ◦ d ––– Note: • ,majoreffect; ◦ , minor effect; —, no significant effect. a A significant adverse effect may be observed if end groups or functional groups show strong interactions with the stationary phase. b Effect is strong in isocratic (‘‘critical’’) chromatography of polymers. In gradient-elution LC it is usually overshadowed by the effects of molecular weight and, especially, chemical composition. c A significant effect may be observed if branching introduces different or additional functional groups. d Effect may be strong, but the technique is not usually applied for the separation of copolymers. For many polymers, complete dissolution can be difficult; polyolefins (polyethene, polypropene), for example, require elevated temperatures and a high-boiling solvent such as trichlorobenzene. Many polar polymers (polyesters, polyamides, polyke- tones) require special solvents such as hexafluoroisopropanol. Polymers can also require a long time to dissolve because of the strong interactions between chains (including ‘‘entanglements’’) and the slow diffusion of large molecules. Another important consideration is the HPLC detection of synthetic polymers. Many important types of polymers (poly-olefins, poly-acrylates, poly-alkoxides) lack UV chromophores. Consequently the RI detector is mainly used for isocratic sepa- rations of polymers (notably by SEC). Occasionally, infrared-absorption detectors (operating at a fixed wavelength) can be advantageous. In gradient separations, the evaporative light-scattering detector has been commonly used, but quantitation can be a problem; the charged-aerosol detector is a promising alternative. Although the characterization of (polar) polymers by mass spectrometry has greatly improved since the mid-1990s, the use of LC-MS for this purpose is still uncommon. Some common synthetic polymers, typical solvents used in size-exclusion chromatogra- phy, and the most commonly used detection methods are listed in Table 13.12. For additional information on possible detectors for polymer separation, see Chapter 3, Chapter 9 of [169], and [170–172]. 13.10.2 Techniques for Polymer Analysis Synthetic polymers are not amenable to the resolution of individual molecules, a difference that sets them apart from other samples for HPLC separation and analysis. Instead, polymer molecules come in a range of sizes or a molecular-weight distribution (MWD). While many techniques can be used to determine an average molecular weight for the sample, chromatographic methods such as size-exclusion Table 13.12 Common Synthetic Polymers, Typical Solvents Used in Size-Exclusion Chromatography, and Commonly Used Detection Methods Typical Typical Solvents Commonly Used Detectors Synthetic Water Methanol Tetrahydrofuran Hexafluoro Trichloro UV Refractive Evaporative Multi-angle Viscometry Polymer (THF) Isopropanol Benzene Index Light Light (HFIP) (TCB) Scattering Scattering Polyethene oxide (polyethylene oxide, polyethyleneglycol) ◦•◦ c,d •• Polypropene oxide (polypropylene oxide, polypropyleneglycol) •◦• c,d •• Polystyrene •• Poly (alkyl acrylate) and poly (alkyl methacrylate) a •◦•◦ c,d •• Aliphatic polyesters, aliphatic polyamides ••◦ c,d ◦◦ Aromatic polyesters, aromatic polyamides ◦ Polyethene (polyethylene), polypropene (polypropylene) • b ◦•• Note: • indicates applicability of a solvent or detector. a For example, poly (methyl acrylate) and poly (methyl methacrylate). b Requires operation at high temperatures (e.g., 150 ◦ C). c More realistic option for interactive liquid chromatography than for size-exclusion chromatography. d Charged-aerosol detection may be preferred alternative. 652 13.10 SYNTHETIC POLYMERS 653 chromatography (SEC) are able to determine the MWD, as well as number-average or weight-average molecular weights [169]. In addition to a MWD, copoly- mers exhibit a chemical-composition distribution. Spectroscopic techniques such as Fourier-transform infrared (FTIR) spectroscopy and (especially) nuclear-magnetic resonance (NMR) spectroscopy can provide detailed information on chemical com- position, while interactive liquid chromatography can separate different chemical types and provide a chemical-composition distribution (CCD). A nonexhaustive overview of a number of important techniques is provided in Table 13.13. It is seen that NMR is highly useful in the middle column (averages), whereas the right-side column (distributions) is dominated by chromatographic techniques. 13.10.3 Liquid-Chromatography Modes for Polymer Analysis 13.10.3.1 Size-Exclusion Chromatography Size-exclusion chromatography (SEC) is reviewed in Section 13.8 and in [169]; its application for the determination of molecular weight or molecular-weight distribution (MWD) is similar for both synthetic polymers and biopolymers. There are two main differences between these two applications of SEC. For synthetic polymers, SEC is used to determine a molecular-weight distribution [169], whereas for biopolymers the goal is the estimation of molecular weight for individual compounds. Likewise the solvent used as mobile phase is often different; usually aqueous mobile phases are used for biopolymers (gel filtration), and organic solvents for synthetic polymers (gel permeation). In the case of homopolymers, SEC can be coupled to other polymer- characterization methods, notably light-scattering and viscometry (for copolymers it is difficult to accurately correlate the resulting data [170]). Static light-scattering can be used to obtain accurate information on the (weight-average) molecular weight of polymer in the SEC effluent, provided that (1) we know how refractive index varies as a function of polymer concentration, and (2) the detector is properly calibrated. Also the concentration of the polymer in the effluent fraction must be accurately known, for example, by using a RI detector in conjunction with light-scattering. 13.10.3.2 Interactive Liquid Chromatography In SEC, conditions are selected to suppress interactions between the analyte and the stationary phase as much as possible. In interactive liquid chromatography (i-LC), these interactions are used to separate molecules by chemical type or functionality. While i-LC separations of polymers are, in many ways, similar to the separation of small molecules by HPLC, there are two overriding differences: (1) the molecular-weight range of polymers (large number of individual species that differ in molecular weight), and (2) a systematic change in analyte retention as the size of the solute molecule increases. High-molecular-weight analytes typically exhibit larger changes in k for a given change in %B, as seem in the examples of Figure 13.11 for several peptides and proteins. A similar example for synthetic polymers is illustrated in Figure 13.47a, which illustrates schematically how retention varies with composition for oligomers and polymers that differ in their size or degree of polymerization p (number of monomers). For the oligomers of Figure 13.47a,p equals 5–15; for the polymers, p equals 25 and 100. The curves for larger molecules (larger p) are increasingly steep, to the extent that for large polymers there is only a 654 BIOCHEMICAL AND SYNTHETIC POLYMER SEPARATIONS Table 13.13 Summary of Techniques for Determining Average Molecular Structures and Molecular Distributions of Synthetic Polymers Polymer Techniques for Techniques for Determining Property Determining Averages Complete Distributions Molecular weight Osmometry light scattering Size-exclusion chromatography Hydrodynamic chromatography Sedimentation Ultracentrigugation Chemical composition NMR FTIR pyrolysis GC-MS Interactive liquid chromatography (mainly gradient elution) Functionality (end groups or functional groups) NMR titration Interactive liquid chromatography (mainly isocratic) Chain regularity NMR Temperature-rising elution Fractionation Degree of branching NMR a Molecular-topology fractionation a Appropriate for polymers with a relatively low molecular weight. Also NMR is generally considered appro- priate only for determining short-chain branching, not long-chain branching. very narrow range of mobile-phase composition (%B) for which the polymer can be eluted isocratically. For this reason gradient elution is usually the method of choice for separations by i-LC. Gradient elution is usually carried out with linear gradients, corresponding to (roughly) constant values of gradient retention k ∗ (≡ k; see Section 9.1.3) for different polymeric species. Under these conditions retention times for each peak in a polymer sample will correspond to the intersection of plots as in Figure 13.47a with a horizontal line that corresponds to a given value of k or k ∗ . For higher values of k ∗ (corresponding to a longer gradient; Eq. 9.5), there are larger differences in retention time for adjacent peaks (and therefore better resolution), compared to separations with a shorter gradient. Finally, for a sufficiently fast gradient (and small enough value of k ∗ ), all solutes leave the column with the same retention time as a single peak. The latter behavior for the separation of a low-molecular-weight polymer with long and short gradients is illustrated in Figure 13.47b, c, respectively. In the long gradient of Figure 13.47b, the retention of individual oligomers differs enough so that there is a partial separation of the sample. In the short gradient of Figure 13.47c, this is no longer true, so a single peak is observed—corresponding to elution of all peaks at about the same time (e.g., k ≈ 1 in Fig. 13.47a). Separation as in Figure 13.47c or in Figure 13.47a for k = 1 is sometimes referred to as pseudocritical chromatography, as opposed to (isocratic) chromatog- raphy under critical conditions, as described in Section 13.10.3.4. Pseudocritical i-LC (i.e., with gradient-elution) is particularly useful for the separation of polymers according to chemical composition. This is illustrated in Figure 13.47d, which shows the separation of two different kinds of polymers. Retention is seen to be a function of chemical composition in this example, but not of molecular weight. Such pseud- ocritical conditions can be approached more closely for (1) higher molecular-weight polymers and (2) shorter gradient times. As a result gradient-elution i-LC is well suited for the determination of chemical-composition distributions. Figure 13.48b 13.10 SYNTHETIC POLYMERS 655 140 120 100 80 60 100 150 200 250 i-LC (min) SEC i-LC SEC (sec) (a) (b) (c) Figure 13.48 Comprehensive two-dimensional liquid chromatography (LC × SEC) of a copolymeric binder produced by two-stage emulsion polymerization of styrene and methyl methacrylate. (a) Second-dimension separation: SEC; (b) first-dimension separation, RPC lin- ear gradient from 65 to 100% in 170 minutes; (c) 2D separation. UV detection at 214 nm. For further details, see [170]. provides a practical illustration where a high-molecular-weight copolymer of styrene and methyl methacrylate is very well separated by chemical type (note three peaks in the chromatogram, for three chemical types in the sample). For a further discussion of Figure 13.38, see Section 13.10.4. 13.10.3.3 Liquid Chromatography under Critical Conditions Figure 13.47a suggests that there is a ‘‘critical’’ mobile-phase composition for which all oligomers co-elute. The potential benefit of working at or near critical conditions is that the effect of the homopolymeric chain on retention can be minimized, so that polymers with different structural elements (e.g., end-groups) can be separated as in Figure 13.47d—regardless of their MWD. This implies that critical conditions can be used to determine differences in polymer functionality [174]. NPC separations are especially suited for this kind of separation because of the large effect that a (polar) functional group can have on retention. 13.10.3.4 Other Techniques Some other HPLC procedures for polymer separation are noted in Table 13.13. Very large polymers can be separated by field-flow fractionation (FFF) and by hydrodynamic chromatography (HDC), techniques that are outside the scope of the present book. . polymers are important in relation to their chromatographic behavior, none more so than their solubility. For any kind of material to be separated by liquid chromatography, it must be dissolved completely (any. (distributions) is dominated by chromatographic techniques. 13.10.3 Liquid- Chromatography Modes for Polymer Analysis 13.10.3.1 Size-Exclusion Chromatography Size-exclusion chromatography (SEC) is reviewed. Interactive Liquid Chromatography In SEC, conditions are selected to suppress interactions between the analyte and the stationary phase as much as possible. In interactive liquid chromatography (i-LC),