Ion Exchange Frederick J. Dechow 1.0 INTRODUCTION In 1850 Thompson['] reported the first ion exchange applications which used naturally occurring clays. However, ion exchange resins have only been used in biochemical and fermentation product recovery since the 1930'~.[~1[~] In these early studies, biochemicals such as adenosine triphos- phate,i4] alcohols,[5] alkaloids,[6] amino acids,['] growth regulators,[*] hor- mone~,[~] penicillin[10] and vitamin B12["1 were purified using ion exchange resins. Ion exchange applications intensified following the work of Moore and Stein,[12] which showed that very complex mixtures of biochemicals, in this case, amino acids and amino acid residues could be isolated from each other using the ion exchange resin as a column chromatographic separator. In biotechnology applications today, ion exchangers are important in preparing water of the necessary quality to enhance the desired microorganism activity during fermentation. Downstream of the fermentation, ion exchange resins may be used to convert, isolate, purify or concentrate the desired product or by-products. This chapter discusses ion exchange resins and their use in commercial fermentation and protein purification operations. 382 Ion Exchange 383 1.1 Ion Exchange Processes Processes involving ion exchange resins usually make use of ion interchange with the resin. Examples ofthese processes are demineralization, conversion, purification and concentration. Chromatographic processes with ion exchange resins merely make use of the ionic environment that the resins provide in separating solutes. Demineralization is the process in which the salts in the feed stream are removed by passing the stream through a cation exchange column in the hydrogen ion form, followed by an anion exchange column in the hydroxide or ¡°free-base¡± form. Water is the most common feed stream in demineraliza- tion. It may also be necessary to remove the salts from a feed stream before fermentation. High metallic ion concentrations and high total salt content in the carbohydrate feed has been found to decrease the yield in citric acid fermentati~n.[¡¯~] These ions can be removed by passing the carbohydrate solution through cation and anion exchange resin beds. The salts required for optimum microorganism activity can be added in the desired concentration prior to fermentation. Conversion or metathesis is a process in which salts of acids are converted to the corresponding free acids by reaction with the hydrogen form of a strong acid cation resin. One such example would be the conversion of calcium citrate to citric acid. The terms may also be used to describe a process in which the acid salt is converted to a different salt of that acid by interaction with a ion exchange resin regenerated to the desired ionic form. Many fermentation products may be purified by adsorbing them on ion exchange resins to separate them from the rest of the fermentation broth. Once the resin is loaded, the product is eluted from the column for further purification or crystallization. Adsorbing lysine on ion exchange resin is probably the most widely used industrial method of purifying lysine. The fermented broth is adjusted to pH 2.0 with hydrochloric acid and then passed through a column of strong acid cation resin in the NH; form. Dilute aqueous ammonia may be used to elute the lysine from the resin.[14] Gordienko[151 has reported that treating the resin with a citrate buffer solution ofpH 3.2 and rinsing with distilled water before elution results in an 83-90% yield of lysine, with a purity of 93-96%. 384 Fermentation and Biochemical Engineering Handbook Ion exchange can be used to concentrate valuable or toxic products of fermentation reactions in a manner similar to purification. The difference between the two processes is in the lower concentration ofthe desired product in the feed solution of concentration processes. Shirato[l6] reported the concentration process for the antibiotic tubercidan produced from fermented rice grain using the microorganism, Streptomyces tubercidicus. Macroporous strong acid cation resin was used to concentrate the antibiotic from 700 pg/d in the fermentation broth to 13 mg/d when eluted with 0.25 N HCl. The yield of the antibiotic was about 83%. 1.2 Chromatographic Separation In most ion exchange operations, an ion in solution is replaced with an ion from the resin and the former solution ion remains with the resin. In contrast, ion exchange chromatography uses the ion exchange resin as an adsorption or separation media, which provides an ionic environment, allowing two or more solutes in the feed stream to be separated. The feed solution is added to the chromatographic column filled with the separation beads and is eluted with solvent, often water in the case of fermentation products. The resin beads selectively slow some solutes while others are eluted down the column (Fig. 1). As the solutes move down the column, they separate and their individual purity increases. Eventually, the solutes appear at different times at the column outlet where each can be drawn off separately. Chromatographic separations can be classed according to four types depending on the type of materials being separated: affinity difference, ion exclusion, size exclusion and ion retardation chromatography. These types of separations may be described in terms of the distribution of the materials to be separated between the phases involved. Figure 2 shows a representation ofthe resin-solvent-solute components of a column chromatographic system. The column is filled with resin beads ofthe solid stationary phase packed together with the voids between the beads filled with solvent. The phases of interest are (i) the liquid phase between the resin beads, (ii) the liquid phase held within the resin beads and (iii) the solid phase of the polymeric matrix of the resin beads. When the feed solution is placed in contact with the hydrated resin in the chromatographic column, the solutes distribute themselves between the liquid inside the resin and that between the resin beads. The distribution for component i is defined by the distribution coefficient, Kd,: Ion Exchange 385 where Cri is the concentration of component i in the liquid within the resin bead and C,, is the concentration of component i in the interstitial liquid. The distribution coefficient for a given ion or molecule will depend upon that component¡¯s structure and concentration, the type and ionic form of the resin and the other components in the feed solution. The distribution coefficients for several organic compounds are given in Table 1 .[171 Solutes Addad 4 Solute { Mixture Desorbent Added - 4 4- sol Utes Added To col URll Separation s1 ow Occurs Fast Colrponent Component Removed FlWl col uan Removed Fron COluSn Figure 1. The steps ofchromatographic separation are: addition ofthe mixed solutes to the column, elution to effect separations, and removal of the separated solutes. 386 Fermentation and Biochemical Engineering Handbook -Resin Bead -Interstitial Llquld /iquid In @sin (v,, (Vq) Figure 2. Representation of the three phases involved in chromatographic separation. The ratio of individual distribution coefficients is often used as a measure of the possibility of separating two solutes and is called the separation factor, a, or relative retention factor. From Table 1 , the separation Mors for acetone-formaldehyde separabil- ity are 0.49, 0.98 and 1.54 for Dowex 50WX8 (H'), Dowex 1X8(C1-) and Dowex 1X8(SOi2) resins, respectively. For comparison purposes, it may be necessary to use the inverse of a, so that the values would be 2.03 and 1.02 for Dowex SOWX8(H') and Dowex 1X8(Cl-), respectively. When a is less than 1 , the solute in the numerator will exit the column first. When a is greater than 1, the solute in the denominator will exit the column first. Ion Exchange 387 Table 1 Distribution Coefficients["1 Solute Resin Kd Ethylene Glycol Sucrose d-Glucose Glycerine Triethylene Glycol Phenol Acetic Acid Acetone Formaldehyde Methanol Formaldehyde Acetone Glycerine Methanol Phenol Formaldehyde Acetone Xylose Glycerine Pentaerythntol Ethylene Glycol Diethylene Glycol Triethylene Glycol Ethylene Diamine Diethylene Triamine Triethylene Tetramine Dowex 50-X8, H' .67 Dowex 50-X8, H' .24 Dowex 50-X8, H' .22 Dowex 50-X8, H' .49 Dowex 50-X8, H' .74 Dowex 50-X8, H' 1.20 Dowex 50-X8, H' .59 Dowex 50-X8, H' 3.08 Dowex 50-X8, H' .71 Dowex 50-X8, H' .6 1 Dowex 1-X7.5, C1- 1.06 Dowex 1-X7.5, C1- 1.08 Dowex 1-X7.5, C1- 1.12 Dowex 1-X7.5, C1- .61 Dowex 1-X7.5, Cl- 17.70 Dowex 1-X8, SO4=, 50-100 .66 Dowex 50-X8, Na' -56 Dowex 50-X8, Na' .63 Dowex 50-X8, Na' .67 Dowex 50-X8, Na' .61 Dowex 1-X8, SO,=, 50-100 1.02 Dowex 50-X8, Na' .45 Dowex 50-X8, Na' .39 Dowex 50-X8, Na' .57 Dowex 50-X8, Na' .57 Dowex 50-X8, Na' .64 Tetraethylene Pentamine Dowex 50-X8, Na' .66 The acetone-formaldehyde separation would be an example of affinity difference chromatography in which molecules of similar molecular weight or isomers of compounds are separated on the basis of differing attractions or distribution coefficients for the resin. The largest industrial chromatogra- phy application of this type is the separation of fructose from glucose to produce 55% or 90% fructose corn sweetener. 388 Fermentation and Biochemical Engineering Handbook Ion exclusion chromatography involves the separation of an ionic component from a nonionic component. The ionic component is excluded from the resin beads by ionic repulsion, while the nonionic component will be distributed into the liquid phase inside the resin beads. Since the ionic solute travels only in the interstitial volume, it will reach the end ofthe column before the nonionic solute which must travel a more tortuous path through the ion exchange beads. A major industrial chromatography application of this type is the recovery of sucrose from the ionic components of molasses. In size exclusion chromatography, the resin beads act as molecular sieves, allowing the smaller molecules to enter the beads while the larger molecules are excluded. Figure 31181 shows the effect ofmolecular size on the elution volume required for a given resin. The ion exclusion technique has been used for the separation of monosodium glutamate from other neutral amino acid~.I¡¯~] 1Do 90 ;. 9¡± Figure 3. compounds. [*I Effect of molecular weight on the elution volume required for glycol Ion retardation chromatography involves the separation of two ionic solutes with a common counter ion. Unless a specific complexing resin is used, the resin must be placed in the form of the common counter ion. The other solute ions are separated on the basis of different affinities for the resin. Ion retardation chromatography is starting to see use in the recovery of acids from waste salts following the regeneration of ion exchange columns. Ion Exchange 389 2.0 THEORY The important features of ion exchange reactions are that they are stoichiometric, reversible and possible with any ionizable compound. The reaction that occurs in a specific length of time depends on the selectivity of the resin for the ions or molecules involved and the kinetics of that reaction. The stoichiometric nature of the reaction allows resin requirements to be predicted and equipment to be sized. The reversible nature of the reaction, illustrated as follows: Eq. (3) R - H' + Na'Cl- C= R - Na' + H'Cl- allows for the repeated reuse of the resin since there is no substantial change in its structure. The equilibrium constant, K, for Eq. (l), is defined for such mono- valent exchange by the equation: IR-Na+] [H+Cl-l Eq. (4) [R-H+] [Na+Cl-] K= In general, if K is a large number, the reverse reaction is much less efficient and requires a large excess of regenerant chemical, HCl in this instanc,e, for moderate regeneration levels. With proper processing and regenerants, the ion exchange resins may be selectively and repeatedly converted from one ionic form to another. The definition of the proper processing requirements is based upon the selectivity and kinetic theories of ion exchange reactions. 2.1 Selectivity When ion B, which is initially in the resin, is exchanged for ion A in solution, the selectivity is represented by: where Zi is the charge and y. is the partial volume of ion i. The selectivity which a resin has for various ions is affected by many factors. The factors include the valence and size of the exchange ion, the ionic form of the resin, 390 Fermentation and Biochemical Engineering Handbook the total ionic strength of the solution, the cross-linkage of the resin, the type of functional group and the nature of the nonexchanging ions. The ionic hydration theory has been used to explain the effect of some ofthese factors on selectivity.[20] According tothis theory, the ions in aqueous solution are hydrated and the degree of hydration for cations increases with increasing charge and decreasing crystallographic radius, as shown in Table 2.r2l] It is the high dielectric constant of water molecules that is responsible for the hydration of ions in aqueous solutions. The hydration potential of an ion depends on the intensity of the change on its surface. The degree of hydration of an ion increases as its valence increases and decreases as its hydrated radius increases. Therefore, it is expected that the selectivity of a resin for an ion is inversely proportional tothe ratio ofthe valencehonic radius for ions of a given radius. In dilute solution, the following selectivity series are followed: Li <Na < K < Rb < Cs Mg < Ca < Sr < Ba F < C1< Br < I Table 2. Ionic Size of Cations[21] Crystallographic Hydrated Ionization Ion Radius (A) Radius (A) Potential Li Na K NH4 Rb cs Mg Ca Sr Ba 0.68 0.98 1.33 1.43 1.49 1.65 0.89 1.17 1.34 1.49 10.00 7.90 5.30 5.37 5.09 5.05 10.80 9.60 9.60 8.80 1.30 1 .oo 0.75 0.67 0.61 2.60 1.90 1.60 1.40 - The selectivity of resins in the hydrogen ion or hydroxide ion form, however, depends on the strength of the acid or base formed between the functional group and the ion. The stronger the acid or base formed, the lower is the selectivity coefficient. It should be noted that these series are not followed in nonaqueous solutions, at high solute concentrations or at high temperature. Ion Exchange 391 The dependence of selectivity on the ionic strength of the solution has been related through the mean activity coefficient to be inversely proportional to the Debye-Huckel parameter, where 'y* is the mean activity coefficient, A and B are constants, and ,u is the ionic strength of the solution. The mean activity coefficient in this instance represents the standard free energy of formation (-W) for the salt formed by the ion exchange resin and the exchanged ion. Figure 4[231 shows this dependence as the ionic concentration of the solution is changed. As the concentration increases, the differences in the selectivity of the resin for ions of different valence decreases and, beyond certain concentrations, the affinity is seen to be greater for the lower valence ion. f I I I I 0 0,4 008 la2 106 2,o MOLARITY OF SOLUTION Figure 4. Dependence of the activity coefficient on the ionic concentration of aqueous 392 Fermentation and Biochemical Engineering Handbook The selectivity of an ion exchange resin will also depend on its cross- linking. The polymer structure of the ion exchange resin can be thought of as collections of coiled springs which can swell or contract during the exchange of The cross-linking of the polymer limits the extent to which the resin may swell-the higher the degree of cross-linking, the lower the extent to which the resin can be hydrated. This limit on resin hydration determines the relative equivalent volumes of hydrated ions which the cross- linked polymer network can accommodate. This is shown in Table 3, [251 As the resin cross-linking or the fixed ion concentration is lowered, the selectivity of the resin decreases. Table 3. Selectivity and Hydration of Cation Resins With Different Degrees of Cros~linking[~~] 4% DVB 8% DVB 16% DVB Cation KH KH K H Li H Na m4 K cs T1 Ag 1.00 418 1.30 431 1.49 372 1.75 360 2.09 341 2.37 342 4.00 289 5.20 229 1.00 211 1.26 200 1.88 183 2.22 172 2.63 163 2.91 159 7.36 163 9.66 113 1.00 130 1.45 136 2.23 113 3.07 106 4.15 106 4.15 102 19.4 102 22.2 85 K = Selectivity compared to Li H = Hydration (g H,O/eq resin) DVB = divinylbenzene Ion Exchange 393 The degree of cross-linking can affect the equilibrium level obtained, particularly as the molecular weight of the organic ion becomes large. With highly cross-linked resins and large organic ions, the concentration of the organic ions in the outer layers of the resin particles is much higher than in the center of the particle. The selectivity of the resin for a given ion is also influenced by the dissociation constants of the functional group covalently attached to the resin (the fixed ion) and ofthe counter-ions in solutions. Since the charge per unit volume within the resin particle is high, a significant percentage of the hctional groups may not be ionized. This is particularly true if the hctional group is a weak acid or base. For cation exchange, the degree of dissociation for the functional group increases as the pH is increased; however, the degree of dissociation for the ions in solution decreases with increasing pH. Therefore, if a cation resin had weak acid functionality, it would exhibit little afiinity at any pH for a weak base solute. Similarly, an anion resin with weak base functionality exhibits little affinity at any pH for a weak acid solute. The influence of pH on the dissociation constants for resin with agiven functionality can be obtained by titration in the presence of an electrolyte. Typical titration curves are shown in Fig. 5 for cation resins and in Fig. 6 for anion resins.[26] For sulfonic acid functional groups, the hydrogen ion is a very weak replacing ion and is similar tothe lithium ion in its replacing power. However, for resin with carboxylic acid functionality, the hydrogen ion exhibits the highest exchanging power. Table 4[271[281 summarizes the effect different anion exchange resin functionalities have on the equilibrium ex- change constants for a wide series of organic and inorganic anions. The selectivity can also be influenced by the non-exchanging ions (co- ions) in solution even though these ions are not directly involved in the exchange reaction. An example of this influence would be the exchange of calcium ascorbate with an anion resin in the citrate form. Although calcium does not take part in the exchange reaction, sequestering of citrate will provide an additional driving force for the exchange. This effect, of course, would have been diminished had a portion of the ascorbate been added as the sodium ascorbate rather than the calcium ascorbate. For nonpolar organic solutes, association into aggregates, perhaps even micelles, may depress solution activity. These associations may be influenced by the co-ions present. 394 Fermentation and Biochemical Engineering Handbook PH 14 12 8 6 - 10 - 8- - 2- 0 I I I I I I 0 2 4 6 8 10 12 L I 2 6 8 MEQ NAOH PER GRAM RESIN Figure 5. Titration curves of typical cation exchange resins.[26] PH I I I I I 0 2 4 6 8 10 12 MEQ HCL PER GRAM RESIN Figure 6. Titration curves of typical anion exchange resins.[26] Ion Exchange 395 Table 4. Selectivity Coefficients for Strongly Basic Anion Re~in[~'1[~*1 Type I Anion Type I1 Anion Anion Salicylate I- HSO, NO,- Br- CN- HSO,- NO,- c1- HCO,- HCOO- C,H,O- H2p04 CH3COO- H2NCH2C 00- OH- F- KXcl 32.2 8.7 5.2 4.1 3.8 2.8 1.6 1.3 1.2 1 .oo 0.32 0.25 0.22 0.17 0.10 0.09 0.09 Anion KXcl Salicylate I- HSO, NO3- Br- CN- HSO,- NO; c1- OH- HCO,- HCOO- CH,COO- F- H2NCH2COO- C,H,O- H2PO.4 28 8.7 7.3 6.1 3.3 2.3 1.3 1.3 1.3 1 .oo 0.65 0.53 0.34 0.22 0.18 0.13 0.10 2.2 Kinetics The overall exchange process may be divided into five sequential steps: 1. The diffusion of ions through the solution to the surface of 2. The diffusion of these ions through the ion exchange 3. The exchange of these ions with the ions attached to the 4. The diffusion of these displaced ions through the particle 5. The diffusion of these displaced ions through the solution the ion exchange particles particle hctional group 396 Fermentation and Biochemical Engineering Handbook Each step of the diffusion, whether in the resin or solution phase, must be accompanied by an ion of the opposite charge to satisfy the law of electroneutrality . Kinetics of ion exchange is usually considered to be controlled by mass transfer in ion exchange particles or in the immediately surrounding liquid phase. The theory used to describe mass transfer in the particle is based on the Nernst-Planck equations developed by Helfferi~h[~~] which accounted for the effect of the electric field generated by ionic diffusion, but excluded convection. It is recognized that the Nernst-Planck theory fails to take into account the effect of swelling and particle size changes which accompany ion exchange or to take into account the slow relaxation of the resin network which causes the diffusion coefficient to vary with time. However, the approximations which these equations provide are a reasonable starting point and will most likely be found to be sufficient for the biotechnology engineer. Any further refinements would lead rapidly to diminished returns. Likewise, the mass transfer in the liquid phase is usually described according to the Nernst film concept using a version[30] of the Nernst-Planck equation or Glueckauf d311 simpler linear driving force approximation. There are five models[32] which can be used to represent the kinetics in ion exchange systems which involve liquid exchange phase mass transfer, solid phase mass transfer, and chemical reaction at the exchange group. Model 1. The liquid phase mass transfer with a linear driving force is the controlling element. This model assumes that there are no concentration gradients in the particle, that there is a quasi-stationary state of liquid phase mass transfer, that there is a linear driving force and that there is a constant separation factor at a given solution concentration. Model 2. The rate-controlling step is diffusion within the ion exchange particles. This model assumes that there are no concentrationgradients in the liquid phase and that there is no convection, either through solvent uptake or release, in the solid phase. Model 3. This model is controlled by the exchange reaction at the fixed ionic groups. This model assumes that the slowness ofthe exchange reaction allows for sufficient time for mass transfer to establish and maintain equilibrium so that no concentration gradient exists in either the ion exchange particles or in the liquid phase. Model 4. This is a variation of Model 3 in which the counter-ion from the solution does not permeate beyond the portion of particle which has been converted to the exchanging ionic form. The boundary of the unreacted core reduces the time such that this is called the shrinhng core model. It is this Ion Exchange 397 sharp boundary between the reacted and unreacted portion of the particles that distinguishes Model 4 from Model 3. Model 5. The rate controlling step is the diffusion of the counter ion across the converted portion of the particle. Since the exchange groups undergo a fast and essentially irreversible reaction with the counter ions, their type of reaction affects the rate of reaction and the geometry of the diffusing zone. Table 5f3*1 summarizes the effect of operating parameters (particle size, solution concentration, separation factor, stirring rate, resin exchange capacity, and temperature) on ion exchange kinetics described by these different models in batch reactors. Table 5. Dependence of Ion Exchange Rates on Experimental Conditions[32] Factor Model 1 Model 2 Model 3 Model 4 Model 5* Particle size cc '/r cc I/? independent a llr I/? (4 Solution OCC independent a c ac a cl (concentration) (c> Separation independent independent$ independent independent independent1 factor up to a (a> specific time when 21;cca when a<< 1. Stirring rate sensitive independent independent independent independent Resin aI/c independent independent independent cc '/c exchange capacity (c) Temperature 4%/"K 216%/"K function of function of =6%/"K (T 1 E*ct EAct Applicable to forward exchange only. Provided partition coeficient is independent of solution concentration. For complete conversion and constant solution composition. 398 Fermentation and Biochemical Engineering Handbook For the cases of interest, the rate of ion exchange is usually controlled by diffusion, either through a hydrostatic boundary layer, called film df@sion control or through the pores of the resin matrix, called particle dv@sion control. In the case of film diffusion control, the rate of ion exchange is determined by the effective thickness ofthe film and by the diffusivity of ions through the film. When resin particle size is small, the feedstream is dilute or when a batch system has mild stirring, the kinetics of exchange are controlled by film diffusion. In the case of particle diffusion control, the rate of ion exchange depends on the charge, spacing and size of the diffusing ion and on the micropore environment. When the resin particle size is large, the feedstream is concentrated, or when a batch system has vigorous stirring, the kinetics are controlled by particle diffusion. The limits at which one or the other type of diffusion is controlling have been determined by T~ai.[~~] When Kk26 > 50, the rate is controlled by film diffusion. When Kk26 < 0.005, the rate is controlled by particle diffusion. In these relationships, K is the distribution coefficient, k2 is the difisivity ratio (DJDf), 6 is the relative film thickness on a resin particle with a radius of a. Between these two limits, the kinetic description of ion exchange processes must include both phenomenon. The characteristic Nernst parameter 6, the thickness of the film around the ion exchange particle, may be converted to the mass transfer coefficient and dimensionless numbers (Reynolds, Schmidt and Sherwood) that engi- neers normally h terms of the solute concentration in the liquid, between 0.1 to 0.0 1 mom, the rate limiting factor is the transport to the ion exchange bead. Above this concentration, the rate limiting factor is the transport inside the resin beads.[35] During the loading phase of the operating cycle, the solute concentration is in the low range. During regeneration however, in which the equilibrium is forced back by addition of a large excess of regenerant ions, the solute is above the 0.1 moa limit. One of the important factors in the kinetic modeling of organic ions is their slow diffusion into the ion exchange resin. The mean diffusion time is listed in Table 6 as a function of resin particle size for different size classifications of substances.[36] With the larger organic ions, the contact time for the feed solution and the resin must be increased to have the ion exchange take place as a welldefined process such as occurs with the rapidly diffusing ions of mineral salts. Ion Exchange 399 Table 6. Characteristic Diffusion into Spherical Resin Particles for Various Substances[36] Coefficient of Diffusion (Order of Magnitude) Type of Sorbed Particle Intraparticle (cm2/sec) Substance (Ion) Radius (cm) Diffusion Mean Time of 10" Ions of mineral salts 0.05 3 min bearing a single charge 0.01 7 sec 0.005 1.8 sec 10-7 Ions of mineral salts 0.05 30 rnin amino acids 0.005 18 sec bearing several charges, 0.01 1.2 min 10-8 Tetraalkylammonium 0.05 5hr ions, antibiotic ions 0.01 12 min on macroporous resins 0.005 3 min 10-9 Dyes, alkaloids, 0.05 over 2 days antibiotics in standard 0.01 2hr ion exchange resins 0.005 0.5 hr 10-10 Some dyes, polypeptides 0.05 over 20 days and proteins 0.01 over 20 hr 0.005 over 5 hr In principle, fluidized ion exchange beds are similar to stirred tank chemical reactors. The general equations of kinetics and mass transfer can be applied to the individual fluidized units in an identical manner to those for chemical reactors. The primary difference lies in accounting for the behavior of suspended particles in the turbulent fluid.[37] The operation of these fluidized ion exchange beds is identical to that of the fixed beds, with the exception that the resin of each stage is confined by perforated plates and maintained in a fluidized suspension using liquid flow or impellers. 400 Fermentation and Biochemical Engineering Handbook The critical design parameter for fluidized beds is the loss or leakage of the solute through a given stage. The design equation for a single stage bed has been described by Marchello and Davis.[38] 2.3 Chromatographic Theory Mathematical theories for ion exchange chromatography were devel- oped in the 1940's by Wilson,[39] DeVault[401 and Gl~eckauf.[~'][~*] These theoretical developments were based on adsorption considerations and are useful in calculating adsorption isotherms from column elution data. Of more interest for understanding preparative chromatography is the theory of column processes originally proposed by Martin and Syr~ge[~~] and aug- mented by Mayer and Thompkin~,[~~I which was developed analogous to fractional distillation so that plate theory could be applied. One of the equations developed merely expressed mathematically that the least adsorbed solute would be eluted first and that if data on the resin and the column dimensions were known, the solvent volume required to elute the peak solute concentration could be calculated. Simpson and Wheat~n[~~] expressed this equation as: where V- is the volume of liquid that has passed through the column when the concentration of the solute is maximum (the midpoint of the elution of the solute). Kd, defined in Eq. 1, is the distribution coefficient ofthe solute in a plate of the column; Vrl is the volume of liquid solution inside the resin and V, is the volume of interstitial liquid. The mathematical derivation of Eq. 7 assumes that complete equilib- rium has been achieved and that no forward mixing occurs. Gl~eckaufI~~] pointed out that equilibrium is practically obtained only with very small diameter resin beads and low flow rates. Such restricting conditions may be acceptable for analytical applications, but would severely limit preparative and industrial chromatography. However, column processing conditions and solute purity requirements are often such that any deviations from these assumptions are slight enough that the equation still serves as an adequate first approximation for scaled-up chromatography applications. Theoretical Plate Height. A second important equation for chroma- tography processes is that used for the calculation ofthe number oftheoretical plates, i.e., the length of column required for equilibration between the solute Ion Exchange 401 in the resin liquid and the solute in the interstitial liquid. If the elution curve approximates a Gaussian distribution curve, the equation may be written as: 2c (c + 1) W* P= where P is the number oftheoretical plates; c ( = Kdcl/q) is the equilibrium constant; Wis the half-width of the elution curve at an ordinate value of l/e ofthe maximum solute concentration. For a Gaussian distribution, W = 40, where o is the standard deviation of the Gaussian distribution. The equilibrium constant is sometimes called the partition ratio. An alternate form of this equation is: Here Wis measured in the same units as V-. This form of the equation is probably the easiest to calculate from experimental data. Once the number of theoretical plates has been calculated, the height equivalent to one theoretical plate (H.E.T.P.) can be obtained by dividing the resin bed height by the value of P. The column height required for a specific separation oftwo solutes can be approximated where His the height ofthe column, P is the number of plates per unit of resin bed height and c is the equilibrium constant defined above. Note that the number of plates in a column will be different for each solute. While this equationmay be used to calculate the column height needed to separate 99.9% of solute 1 from 99.9% of solute 2, industrial and preparative chromatogra- phy applications typically make more efficient use of the separation resin by selectively removing a narrow portion of the eluted solutes, as illustrated in Fig. 7.[481 402 Fermentation and Biochemical Engineering Wan dbook 2 IO 118 0 - - - c -4 .6 .8 LO 1.2 1.4 1,6 1.3 2.3 2,2 2A Figure 7. Distribution of eluate into fractions for product, recycle, and waste for NaCl and glycol Table 7 shows how the theoretical plate number for a chromatographic system may be calculated from various combinations of experimental data. The band variance, t$, is calculated from the experimental data and combined with the retention time, & , for a given solute. Figure 8 shows the different experimental values which may be used to calculate q. Zone Spreading. The net forward progress of each solute is an average value with a normal dispersion about the mean value. The increased band or zone width which results from a series of molecular diffusion and non-equilibrium factors is known as zone spreading. The plate height as a hnction of the mobile phase velocity may be written as a linear combination of contributions from eddy diffusion, mass transfer and a coupling term: Ion Exchange 403 Table 7. Calculation of Plate Number from Chromatogram Measurements Coversion to Variance Plate number N = (tR / Ut)2 ______ tR and UT tR and baseline width wb ut=wbl4 N = l6(t~ 1 wb)2 tR and width at half height WO~ ut = w0.5 1 N = 5.54(tRIwo.5)¡¯ tR and width at inflection points ut = wi 12 N = 4(tR /wi)2 (0.607 h) Wl tR and band area A and height h ut=Alhfi N =2n(tRh/A)2 Figure 8. Identification of chromatographic peak segments for the calculation of column performance. 404 Fermentation and Biochemical Engineering Handbook A plot of Eq. 11 for any type of linear elution chromatography describes a hyperbola, as shown in Fig. 9.[491 There is an optimum velocity of the mobile phase for carrying out a separation at which the plate height is a minimum, and thus, the chromatographic separation is most efficient: where DM is the diffusion coefficient of the solute molecule in the mobile phase, D, is the diffusion coefficient in the stationary phase, dp is the diameter of the resin bead and R, = L/vt, where L is the distance the zone has migrated in time t. 1.0 t- E u f .O 0.5 2 I ¡¯ Molecular .P- diffusion - 1¡¯1 Resistonce to mass tranrfer ___----- ------------- I Eddy diffusion 1 II I 1 , I 1 I I I 11 ,I Carrier velocity, cm/sec 0 2 4 6 8 10 12 14 Figure 9. Relationship between late height and velocity of the mobile phase.[49] Resolution. A variation on calculating the required column height is to calculate the resolution or degree of separation of two components. Resolution is the ratio of peak separation to average peak width: Ion Exchange 405 The numerator of Eq. 13 is the separation of the two solutes¡¯ peak concen- trations and the denominator is the average band width ofthe two peaks. This form of the equation is evaluating the resolution when the peaks are separated by four standard deviations, 0. If R = 1 and the two solutes have the same peak concentration, this means that the adjacent tail of each peak beyond 20 fi-om the V,, would overlap with the other solute peak. In this instance there would be 2% contamination of each solute in the other. Resolution can also be represented[50] by: Resolution can be seen to depend on the number of plates for solute 2, the separation factor for the two solutes and the equilibrium constant for solute 2. In general, the larger the number of plates, the better the resolution. There are practical limits to the column lengths that are economically feasible in industrial and preparative chromatography. It is possible to changeP also by altering the flow rate, the mean resin bead size or the bead size distribution since P is determined by the rate processes occurring during separation. As the separation factor increases, resolution becomes greater since the peak-to- peak separation is becoming larger. Increases in the equilibrium constant will usually improve the resolution since the ratio c2/( 1 + c2) will increase. It should be noted that this is actually only true when c2 is small since the ratio approaches unity asymptotically as c2 gets larger. The separation factor and the equilibrium factor can be adjusted for temperature changes or other changes which would alter the equilibrium properties of the column operations. Equation 14 is only applicable when the two solutes are of equal concentration. When that is not the case, a correction factor must be used (A: + A:)/2A1A, where A, and A, are the areas under the elution curve for solutes 1 and 2, respectively. Figure 10 shows the relationship between product purity (q), the separation ratio and the number of theoretical plates. This graph can be used to estimate the number oftheoretical plates required to attain the desired purity of the products. 406 Fermentation and Biochemical Engineering Handbook For example, when the product purity must be 98.0%, then q = Am/m = 0.01, when the amount of the two solutes is equal. If the retention ratio, a, is equal to 1.2, then the number of theoretical plates from Fig. 10 is about 650. With a plate height of 0.1 cm, the minimum bed height would be 65 cm. In practice, a longer column is used to account for any deviation from equilibrium conditions. 10-l' 10-8 10-5 10-3 .01 .06 .1 10-l~ IO-( 10-4 3x10-3 .03 Figure 10. Relationshipbetween relative retention ration, number oftheoretical plates, and product purity. [461 Ion Exchange 407 3.0 ION EXCHANGE MATERIALS AND THEIR PROPERTIES Ion exchange materials are a special class of polyelectrolytes. The chemical and physical properties of an ion exchange material play a more important role in determining its suitability for a biochemical application than for other types of applications. The chemical properties to be considered are the matrix and the ionic functionality attached to the matrix. The important physical properties are the pore size, the pore volume, the surface area, the density and the particle size. A list of commercial producers of granular or bead ion exchange materials is given in Table 8. Table 8. Producers of Synthetic Ion Exchange Resins Company Country Tradename Bayer Chemolimfex Dow Ionac Mitsubishi Montecatini-Edison Ostion Permutit Permutit, AG Resindion Rohm & Haas Germany HwwY United States United States Japan Italy Czechoslovakia United Kingdom Germany United States Russia Italy Lewatit Varion Dowex Ionac Diaion Kastel Ostion' Zeocarb, Deacidite, Zerolit Orzelith, Permutit Relite Amberlite, Duolite AW-, AV-, KB-, KU- 3.1 Ion Exchange Matrix These materials can be broadly categorized into those which are totally inorganic in nature and those that are synthetic organic resins. 408 Fermentation and Biochemical Engineering Handbook Inorganic ion exchangers[51] include both naturally occurring materi- als such as mineral zeolites (sodalite and clinoptilolite), the greensands, and clays (themontmorillonite group) and synthetic materials such as gel zeolites, the hydrous oxides of polyvalent metal (hydrated zirconium oxide) and the insoluble salts of polybasic acids with polyvalent metals (zirconium phos- phate). The synthetic organic resins consist of cross-linked polymer matrix which is functionalized to provide their ion exchange capacity. The matrix usually must undergo additional reactions to provide the strong acid cation, strong base anion, weak acid cation or weak base anion functionality. Cross-linked polystyrene, epoxy-polyamine, phenol-formaldehyde, and cross-linked acrylic methacrylic acid resins are the most commonly used ion exchanges in industrial applications and have been used in biochemical applications, such as protein purifications and enzyme immobilizations. However, the hydrophobic matrices have the disadvantages that they might denature the desired biological material or that the high charge density may give such strong binding that only a fraction of the absorbed material might be recovered. Resins with cellulosic matrices are much more hydrophilic and these do not tend to denature proteins. Cellulosic resins have been used extensively in the laboratory analyses of biological materials, enzyme immobilizations and small scale preparations. The low capacity and poor flow characteristics have limited the usefulness of these matrices for larger applications. Recently, diethylaminoethyl (DEAE) silica gel was shown[52] to be an improvement over typical cellulosic-matrices resins for the separation of acidic and neutral lipids from complex ganglioside mixtures. The specific advantages claimed were: 1. An increase in flow rate was possible through the DEAE- 2. The DEAE-silica gel was able to be equilibrated much 3. The DEAE-silica gel was more easily regenerated. 4. The DEAE-silica gel was less susceptible to microbial attack. 5. The preparation of DEAE-silica gel from inexpensive silica gel was described as a simple method that could be carried out in any laboratory. silica gel. more rapidly with the starter buffer. Ion Exchange 409 3.2 Functional Groups The strong acid cation exchange resins are made by the sulfonation of the matrix copolymer. Strong acid cation resins are characterized by their ability to exchange cations or split neutral salts. They will fbnction throughout the entire pH range. The synthesis of weak acid cation resins has been described above. The ability ofthis type of resin to split neutral salts is very limited. The resin has the greatest affinity for alkaline earth metal ions in the presence of alkalinity. Only limited capacities for the alkali metals are obtained when alkalinity other than hydroxide is present. Effective use is limited to solutions above pH 4.0. The anion exchange resins require the synthesis of an active interme- diate. This is usually performed in the process called chloromethyhtion. The subsequent intermediate is reactive with a wide variety of amines which form different functional groups. The Type I resin is a quaternized amine resin resulting from the reaction oftrimethylamine with the chloromethylated copolymer. This functionalized resin has the most strongly basic fbnctional group available and has the greatest affinity for weak acids. However, the efficiency of regenerating the resin to the hydroxide form is somewhat lower than Type I1 resins, particu- larly when the resin is exhausted with monovalent anions. The Type I1 resin results when dimethylethanolamine is reacted with the chloromethylated copolymer. This quaternary amine has lower basicity than that ofthe Type I resin, yet it is high enough to remove the anions of weak acids in most applications. While the caustic regeneration efficiency is significantly greater with Type I1 resins, their thermal and chemical stability is not as good as Type I resins. Weak base resins may be formed by reacting primary or secondary amines or ammonia with the chloromethylated copolymer. Dimethylamine is commonly used. The ability of the weak base resins to absorb acids depends on the basicity of the resin and the pK of the acid involved. These resins are capable of absorbing strong acids in good capacity, but are limited by kinetics. The kinetics may be improved by incorporating about 10% strong base capacity. While strong base anion resins function throughout the entire pH range, weak base resins are limited to solutions below pH 7. The desired functionality on the selected matrix will be determined by the nature ofthe biochemical solute which is to be removed from solution. Its isoelectric point, the pH restrictions on the separation and the ease of 410 Fermentation and Biochemical Engineering Handbook eventually eluting the absorbed species from the resin play important roles in the selection process. Some resins have been developed with hctional groups specifically to absorb certain types of ions. The resins shown in Table 9 are commercially available. Table 9. Commercial Resins with Special Functional Groups Functionality Structure Iminodiacetate R-CH2N(CH2COOH)Z Thiol R-SH Aminophosphate R-CH2NHCHZPO3Hz Amidoxime R-C=N-OH Polyethylene Polyamine R-(NC2H4)mH I NH2 Phosphate R-PO3H The selectivity of these resins depends more on the complex that is formed rather than on the size or charge of the ions. Generally they are effective in polar and nonpolar solvents. However, the capacity for various ions is pH sensitive so that adsorption and elution can be accomplished by pH changes in the solution. These chelating resins have found most of their use in metal ion recovery processes in the chemical and waste recovery industries. They may find use in fermentation applications where the cultured organism requires the use of metal ion cofactors. Specific ion exchange resins have also been used in laboratory applications that may find eventual use in biotechnology product recovery applications.[53] A review of selective ion exchange resins has been compiled by Warshaw~ky.[~~] A diaminotetratacetic polymer developed by Mit~ubishi[~~] was developed for the purification of amino acid feed solutions. The conversions of chloromethylpolystyrene into thiolated derivatives for peptide synthesis have been described by Warshawsky and coworkers Ion Exchange 411 3.3 Porosity and Surface Area The porosity of an ion exchange resin determines the size of the molecules or ions that may enter an ion exchange particle and determines their rate ofdiffusion and exchange. Porosity is inversely related to the cross- linking of the resin. However, for gel-type or microporous resins, the ion exchange particle has no appreciableporosity until it is swollen in a solvating medium such as water. The pore size for microporous resins is determined by the distances between polymer chains or cross-linking subunits. If it is assumed that the cross-linking is uniform throughout each ion exchange particle, the average pore diameter of these resins can be approximated from the water contained in the fully swollen resin. The moisture content of cation resins as a function of the degree of cross-linking is shown in Figure 1 1 and, of anion resins, in Fig. 12. The calculated average pore size for sulfonic cation resins ranges from 16 to 20 8, as the concentration of the resin cross-linking agent (divinylbenzene) decreases from 20 to 2%. The calculated average pore size ofthe anion resins ranges from 18 to 14 A as the cross-linking is decreased from 12 to 2%. Even at low cross-linking and full hydration, microporous resin have average pore diameters of less than 20 A. The dependence of pore size on the percent of the cross-linking is shown in Table 10 for swollen microporous resins of ~tyrene-divinylbenzene.[~~] Sulfonic acid and carboxylic acid resins have also been equilibrated with a series of quaternary ammonium ions of different molecular weights to measure the average pore size when these resins have increasing degrees of cross-linking. These results are shown in Figs. 13 and 14. Figure 15 shows the change in the ionic diffusion coefficients of tetra- alkylammonium ions in strong acid cation resins as a function of mean effective pore diameter of the resins.[58] As the pore diameter is increased, the penetrability of the resins with respect to the large ions also increased. If an inert diluent (porogen) is incorporated into the monomer mixture before the copolymer is formed, it is possible to form a structure containing varying degrees of true porosity or void vol~me.[~~1[~~] Varia- tions in the amount of divinylbenzene cross-linking and diluent allow for a range of particle strengths and porosity to be made. Subsequent reactions with the appropriate chemicals result in the introduction of the same functional groups as discussed above. These are called macroporous or macroreticular resins. 412 Fermentation and Biochemical Engineering Handbook la 20 0 PERCENT DIVINVLIWIZW Figure 11. Moisture content of strong acid cation resins as a function of divinylbenzene content. Figure 12. Moisture content of strong acid anion resins as a function of divinylbenzene content. Ion Exchange 413 Table 10. Average Swollen Diameter of Cross-linked Polystyrene Beads in Tetrahydrofuran[57] Divinyl benzene Concentration (Cross-linking) Swollen Pore Diameter (¡±/) (¡±/) 1 2 4 8 16 77 54 37 14 13 Figure 13. Average pore diameter of sulfonic acid cation exchange resin as a function of degree of cross-linking. (Ref: 20, page 46) 414 Fermentation and Biochemical Engineering Handbook WATER DATA 0 LARGE ION ADSORPTION DATA 0 1111111 0 4 8 12 16 PERCENT DIVINYLBENZENE Figure 14. Average pore diameter of carboxylic acid cation exchange resin as a function of degree of cross-linking. (R& 20, page 47) Ion Exchange 415 0 I I RESIN PORE DIAMETER (A) Figure 15. Ion diffusion coefficients in macroporous sulfonated cation exchange resins.[s8] (1) Tetramethyl; (2) tetraethyl; (3) tetrabutyl ammonium ions. Since each bead of a given external diameter that is made by the inert diluent process will contain some void volume, there is actually less polymer available per unit volume for the introduction of functional groups. There- fore, these macroporous resins are inherently of lower total exchange capacity than gel-type resins of the same composition. Macroporous resins are most useful when extremely rigorous osmotic shock conditions are encountered, when the very high porosity is desirable from the stand point of the molecular weight of the material being treated or when nonpolar media are involved. The drawbacks of using macroporous resins are poorer regeneration efficiencies, lower total exchange capacities and higher regeneration costs. 416 Fermentation and Biochemical Engineering Handbook Until the advent of macroporous resins, the synthetic organic ion exchange resins were of such low porosity that large proteins and other macromolecules would be adsorbed or interact only with the exterior exchange sites on the resins. Therefore, although the microporous resins may have higher total exchange capacities than macroporous resins, the effective capacity ofmacroporous resins for protein or macromolecule adsorption may often times be greater than that of microporous resins. Typical macroporous ion exchange resins may have average pore diameters ranging from 100 8, to 4000 A. Table 11 shows the pore sizes of several-resins of different matrices that have been used in enzyme Pore volumes for macroporous resins may range from 0.1 to 2.0 mug. Table 11. Physical Properties and Capacities for Ion Exchange Resins[61] Adsorption Resin Pore Surface Resin Capacity Matrix Functionality Size Area Capacity for Enzyme phenolic 3O polyethylene 250 8, 68.1 m2/g 4.38 meqlg 3.78meqlg polyamine phenolic partially 3' 290 8, 95.3 m2/g 4.24 meq/g 3.57 meq/g polyethylene polyamine polystyrene polyethylene 330 8, 4.6 m2/g 4.20 meq/g 3.92 meq/g polyamine polystyrene polyethylene 560 8, 5.1 m2/g 4.75 meq/g 4.32 meq/g polyamine polyvinyl polyethylene 1400 A 15.1 m2/g 4.12 meq/g 3.72 meq/g chloride polyamine Ion Exchange 417 Normally, as the mean pore diameter increases, the surface area of the resin decreases. These surface areas can be as low as 2 m2/g to as high as 300 m2/g. Table 1 1 also points out that the total exchange capacity is not utilized in these biochemical fluid processes. Whereas, in water treatment applica- tions, one can expect to utilize 95% of the total exchange capacity, in biotechnology applications it is often possible to use only 10 to 20% of the total exchange capacity of gel resins. Macroporous resins have increased the utilization to close to 90% for the immobilization of enzymes, but biochemi- cal fluid processing applications where the fluid flows through an ion exchange resin bed still are limited to about 35% utilization even with macroporous resins. Table 12 shows the molecular size of some biological macromolecules for comparison to the mean pore size of the resins. When selecting the pore size of a resin for the recovery or immobilization of a specific protein, a general rule is that the optimum resin pore diameter should be about 4 to 5 times the length of the major axis of the protein. Increasing the pore size of the resin beyond that point will result in decreases in the amount of protein adsorbed because the surface area available for adsorption is being decreased as the pore size is increased. An example ofthis optimal adsorption ofglucose oxidase, as defined by enzyme activity, is shown in Fig. 16.[62] Enzyme activity is a measure of the amount of enzyme adsorbed and accessible to substrate. Table 12. Molecular Size of Biopolymers Maximum Length Biopolymer Molecular Weight of Biopolymer ~ Catalase 250,000 183 8, Glucose Isomerase 100,000-250,000 75-100 8, Glucose Oxidase 15,000 84 A Lysozyme 14,000 40 8, Papain 21,000 42A 418 Fermentation and Biochemical Engineering Handbook ENzYl€ ACTIVITY (Wm) 200 400 800 PORE DIAMETER (A) Figure 16. Effect of resin pore diameter on the enzyme activity of glucose oxidase.[6z] 3.4 Particle Density The typical resin densities may range from 0.6 g/cc to 1.3 g/cc for organic polymers. Silicate materials may be more dense up to 6 g/cc. Since the fermentation broth or other biochemical fluid may be more dense than water, the slow flow rates that are usually involved may require resins that have a greater density than water. A minimum flow rate may be necessary to maintain a packed bed when a fluid denser than water is being processed by a medium density resin. If this is not possible, an up-flow operation or batch process may be necessary. This is discussed in more detail in Sec. 6. Ion Exchange 419 The lower density resins are usually associated with a highly porous structurewhich has less mechanical strengththanthe typicalgel or macroporous resins. Wen the mean pore diameter ofa resin is greater than 2000 A, the resin would be subject to attrition in a stirred tank or may collapse in a tall column. 3.5 Particle Size Many ofthe resins used in the early biochemical separations were quite small (75-300 microns). With the development of macroporous resins, protein purifications were performed with resins ofthe 400- 1000 micron size since the macroporous structure allowed sufficient surface area for adsorp- tion almost independent of particle size. 4.0 LABORATORY EVALUATION OF RESIN The total exchange capacity, the porosity, the operating capacity and the efficiency of regeneration need to be evaluated in the laboratory when comparing resins for a given application. The total exchange capacity is usually determined by titrating the resin with a solution of acid or base to a specific end point. This type of information is readily available from the manufacturers of commercial ion exchange resins. The pore size of a microporous resin can be determined using water soluble standards, such as those listed in Table 13 .[63] Ifthe resin is made with an inert, extractable diluent to generate the macroporous structure, it is easier to determine the mean pore size and pore size distribution. Care must be taken so that the pores are not collapsed during the removal of the water from the resin. Martinola and Me~er[~~] have devised a method of preparing a macroporous resin for BET surface analysis or pore size analysis by mercury porosimetry . 1. Convert the ion exchange resin to the desired ionic form. 2. Add 500 ml of water-moist resin to a round bottom flask with an aspirator. Add one liter of anhydrous isopropyl alcohol and boil under reflux at atmospheric pressure for one hour, then remove the liquid. Repeat the isopropyl addition, boiling and aspirating four times. After this procedure the resin will contain less than 0.1% water. 3. After drying to constant weight at 10¡± torr and 5OoC, the resin sample is ready for pore size analysis. 420 Fermentation and Biochemical Engineering Handbook Table 13. Water Soluble Standard Samples for Pore Measurements[64] Sample Mean Pore Diameter (A) D2O Ribose Xylose Lactose Raffinose Stachyose TA8 T-10 T-40 T-70 T-500 T-2000 3.5 8 9 10.5 15 19 51 140 270 415 830 1500 aThe T-Standards are Dextrans from Pharmacia The design of an ion exchange unit requires knowledge of the capacity of the resin bed and the efficiency of the exchange process. The ¡°theoretical¡± capacity of a resin is the number of ionic groups (equivalent number of exchangeable ions) contained per unit weight or unit volume of resin. This capacity may be expressed as milli-equivalents (meq) per ml or per gram of resin. When deciding which resin to use for a given operation, batch testing in a small beaker or flask will allow resin selection and an approximation of its loading capacity. A useful procedure is to measure out 1, 3, 10 and 30 milliliter volumes of resin and add them to a specific volume of the feedstream. These volumes were chosen to have even spacing on a subse- quent log-log plot of the data. After the resin bed feed solution has been mixed for at least one half hour, the resin is separated from the liquid phase. The solute concentration remaining in the solution is then determined. The residual concentration is subtracted from the original concentration and the difference is divided by the volume ofthe resin. These numbers and the residual concentration are plotted on log-log paper and frequently give a straight line. Ion Exchange 421 A vertical line drawn at the feed concentration intersects at a point extrapolated from the data points to give an estimate of the loading of the solute on the resin. Figure 17 shows such a plot for glutamic acid absorbed on an 8% cross-linked strong acid cation microporous resin in a fermentation broth with 11 mg/ml of amino acid. The resin was placed in a beaker with 250 ml of broth. The extrapolation ofthe line for the 1,3, 10, and 30 ml resin adsorption data indicates that the loading of glutamic acid on this resin is expected to be 60 giliter resin. 011 1,o 0.01 RESIDUAL SOLUTE C~NCEWTRAT~~ Figure 17. Plot of glutamic acid adsorbed on an 8% cross-linked strong cation resin. After several resins have been tested in this manner, the resin is selected for column evaluation which has a high loading per ml of resin or a low residual with larger resin quantities. 422 Fermentation and Biochemical Engineering Handbook In practice, the ion exchange resin is generally operated at a level considerably below its theoretical capacity. Since the ion exchange reactions are equilibrium reactions, an impracticably large quantity of regenerant would be required to drive the reaction to completion. The ¡°operating¡± capacity of a resin is the number of ionic groups actually utilized per unit weight or volume of resin under a given set of operating conditions. The operating capacity of a resin is not directly proportional to the amount of regenerant used. ¡°Efficiency¡± is the concept used to designate the degree of utilization of the regenerant. Column efficiency is the ratio of the operating exchange capacity of a unit to the exchange that theoretically could be derived from a specific weight of applied regenerant. It is recommendedthat operating capacity and column efficiency be run initially on a small, laboratory scale to determine if the reaction desired can be made to proceed in the desired direction and manner. The column should be at least 2.5 cm in diameter to minimize wall effects. The preferential flow in a resin column is along the wall of the column. The percentage of the total flow along the wall of the column decreases as the column diameter increases and as the resin particle size decreases. The bed depth should be at least 0.5 m and the flow rate should be about 0.5 bed volumes per hour for the initial trial. These conditions are good starting points since it is desirable that the transition zone not exceed the length of the column. Using much larger columns would require quantities of the feedstream which are larger than may be readily available. A suggested operating procedure is outlined below. 1. Soak the resin before adding it to the column to allow it to reach its hydrated volume. 2. After the resin has been added to the column, backwash the resin with distilled water and allow the resin to settle. 3. Rinse the column of resin with distilled water for ten minutes at a flow rate of 50 ml/min. 4. Start the treatment cycle. Monitor the effluent to develop a breakthrough curve, such as shown in Fig. 18, until the ion concentration in the effluent reaches the concentration in the feed solution. 5. Backwash the resin with distilled water to 50-100% bed expansion for 5 to 10 minutes. 6. Regenerate the resin at a flow rate that allows at least forty-five minutes of contact time. Measure the ion concentration of the spent regenerate to determine the Ion Exchange 423 elution curve (Fig. 19) and the amount of regenerant actually used. 7. Rinse column with distilled water until the effluent has reached pH 7. h Bed Volum Figure 18. Concentration of adsorbed species in column effluent during column loading. Bed Volumes Figure 19. Elution curve showing concentration of adsorbed species eluted during resin regeneration. 424 Fermentation and Biochemical Engineering Handbook Feed concentrations, flow rates, and regenerant dosages may be varied to develop the relationship between resin utilization and regenerant efficiency so that the optimum operating conditions can be selected for the system. The first portion ofthe breakthrough curve in Fig. 18 shows the quality of product that can be obtained under the processing conditions. An integration of the area up to the breakthrough point provides an estimate for commercial column capacities for the space velocities used in the experiment. The velocity at which the mass transfer zone is moving through the column is given by dividing the length of the column by the time it takes to detect the solute in the column effluent. The difference between that time and the time at which the selected breakthrough concentration appears in the effluent, when multiplied by the velocity of the mass transfer zone, results in an approximation of the mass transfer zone. For simple molecules with large differences in distribution coefficients, a single eluting solution may be used to develop the chromatogram. However, more complex materials, such as peptides and proteins, require a shift in the ionic strength of the eluent. This can be done step-wise or as a gradient. Semem~a[~~] has proposed the following rules for the proper choice of eluent: 1. Use cationic buffers (Tris-HCI, piperazine-HClO,, etc.) with anion resins and anionic buffers (phosphate, acetate, etc.) with cation exchange resins. 2. With anion resins use decreasing pH gradients and with cation resins use rising pH gradients. 3. Avoid using buffers whose pH lies near the pK of the adsorbent. If the chromatographed solutes are to be isolated by solvent evapora- tion, the use of volatile buffers, such as carbonic acid, carbonates, acetates and formates of ammonium should be used. If better resolution is required, it may be obtained by changing the type of gradient applied. A convex gradient may be useful in improving the resolution during the last portion of a chromatogram or to speedup separation when the first peaks are well separated and the last few are not adequately spaced. A concave gradient can be used if it is necessary to improve resolution in the first part of the chromatogram or to shorten the separation time when peaks in the latter portion are more than adequately spaced. During the column test, the starting volume and the final volume of the resin should be measured. If there is a change of more than 5%, progressive volume changes as the resin is operated through several cycles should be Ion Exchange 425 recorded. These changes may be significant enough to affect the placement of laterals or distributors in the design of commercial equipment. For instance, carboxylic resins may expand by 90% when going from the hydrogen form to the sodium form. This type of volume change may dictate how the resin must be regenerated to prevent the breakage of glass columns due to the pressure from the swelling resin. Gassing, the formation of air pockets, within the resin bed is to be avoided. Gassing may occur because of heat released during the exchange reaction. It will also occur if a cold solution is placed in a warm bed or if the liquid level falls below the resin level. Keeping the feed solution 5°C warmer than the column temperature should prevent the gassing due to thermal differences, It is necessary to configure the experimental apparatus to insure that the feedstream moves through the column at a steady rate to maintain a well- defined mass transfer zone. Possible methods of maintaining constant flow are shown in Fig. 20. Once it is determined that the action will proceed as desired, subsequent optimization of the system in the laboratory calls for setting a packed resin column of approximately the bed depth to be used in the final equipment, typically one to three meters. LARGE VOLUME CONSTANT HEAD DEVICE PUMP OPERATIOW Figure 20. Equipment for laboratory evaluation of ion exchange resins.