Carbohydrate Separations Using Zeolite Molecular Sieves John D. Sherman and Chien C. Chao Molecular Sieve Department, Union Carbide Corporation, Tarrytown Technical Center, Tarrytown, New York 10591, U.S.A.
Recent studies of separations of carbohydrates have led to the development of new sorbents with improved performance for separation of fructose from glucose, and also the development of other new separations of both simple sugars, including: mannose from glucose; separation of lactose, galactose, and glucose; lactulose from lactose; and arabinose from other sugars; and a number of sugar alcohols, including: inositol from sorbitol; inositol or sorbitol from fructose and other sugars; mannitol from sorbitol and galactitol; and galactitol from sorbitol. Potential commercial uses are noted. The dominant adsorption mechanism involves "ligand exchange" of hydroxyl groups on polyol molecules for H20 molecules in the hydration spheres of cations attached to the walls of the zeolite cavities. The strength of sorption of different polyols on CaY zeolite increases in proportion to the relative strength of the corresponding polyol-cation complex formed in aqueous solution, as deduced from literature data on such complexes. However, such data are quite sparse, allowing few predictions to be made. In addition, significant discrepancies between such predictions and actual separations performance indicate that the sorption is also controlled by geometric constraints imposed by the zeolite pore geometries and cation positions. INTRODUCTION Our recent studies have led to the development of many new carbohydrate separations of present or future commercial potential, particularly separations of sugar alcohols and sugars. A number of these separations are described below, together with some of the preferred sorbents for each. The separation of fructose from glucose using zeolite molecular sieves has been reported previously by scientists in Japan [II, the united States [21, and The Netherlands [31, and is already in commercial use for the manufacture of high fructose corn syrup (HFCS). Improved sorbents for such separations are also described. EXPERIMENTAL Zeolite sorbents were prepared, packed into columns, and tested for their ability to separate carbohydrates by the pulse elution technique using water as the eluent. Typical conditions: Sorbent particle size: 2-20 micrometer zeolite powders or 420-840 micrometer (20x40 U.S. Screen Mesh) or 290-580 micrometer (30x50 U.S. Screen Mesh). Column dimensions: 40 or 160 em long, 0.77 cm inside diameter. Temperature: 70 C or room temperature. Eluent flow rate: 1.0 mL/min. Pulse concentration: 0.1 or 0.2 g/mL. Pulse size: 1.0 or 2.0 mL (5.0 microliters in some tests). Detectors: UltraViolet and Index of refraction. 1025
1026 (AP-6-l) Test Procedures: leolite mesh was dry packed into columns with mechanical vibration to obtain uniform packing of the column. Zeolite powders were packed into 40 cm long columns by slurry technique. The peak elution volume was measured from the introduction of the pulse to the elution of the peak maximum. The void volume was determined by measuring the elution volume for a non-sorbed tracer component, generally inulin, a water soluble polymer of fructose, with a molecular weight of 5000 -- much too large a molecule to enter the zeolite. Resolution factors or separation factors were determined using pulses containing a mixture of two or more components. Calculations: Separation Factor Resolution Factor
( Va - Vo ) / ( Vb - Vo ) 2 ( Va - Vb ) / ( Wa + Wb
where Va and Vb are the observed elution volumes of components a and b. Vo is the void volume. Wa and Wb are the chromatographic peak widths for components a and b at half the height of the peak maximum for that component. RESULTS Separations of sugars and sugar alcohols and some of the preferred sorbents which have been developed for each are described below, together with brief descriptions of existing or potential commercial applications. 1.
Separation of Fructose from Glucose 14]: High fructose corn syrup (HFCS) products containing 55-60~ fructose are equal to common table sugar (sucrose) in sweetness and sold as lower cost sweeteners. CaY zeolites or Ca++ resin sorbents are used in chromatographic-type selective adsorption processes to separate fructose from a 42~ fructose solution glucose (made from starch by enzymatic conversions) to provide products containing 90~ fructose which are sold at that concentration or blended with the 42~ fructose syrup to prepare a 55~ fructose syrup. We conducted a program to develop superior sorbents. Pulse test results on 160 cm columns at conditions of commercial interest (70 C, 0.5 BV/hr eluent rate, using pulses of commercial 42~ fructose syrup) are summarized below: Sorbent (290-580 micrometer unless indicated)
Pulse Concentration (Wt. percent OS *)
Pulse Size (bed volumes)
Resolution Factor (fructose/glucose)
----------------------- ------------------- -------------- -----------------50 ~ OS 0.17 BV 0.47 **
CaY zeolite *** New Zeolite 1 N_ Zeolite 2 Notes:
60 50 50 50 60
0.21 0.17 0.17 0.17 0.21
0.41 0.49 0.59 0.64 0.60
OS = dissolved solids Ca++ resin (in commercial HFCS use) (280 micrometer particles) CaY zeolite: Union Carbide Corporation sorbent employed in early commercial fructose separations
As these results show, the CaY zeolite provides separation performance superior to that of the resin sorbent even though the average particle size of the CaY was approximately 1.5 times that of tae resin. In addition, both of the new zeolite sorbents provide significant further improvements over that of the early commercial CaY sorbent; both have been prepared at large pilot scale and are currently undergoing evaluation tests for potential commercial use. The improved resolution factors provided by these new zeolite sorbents should allow higher fructose recoveries and improved process efficiency to be achieved in the production of 9~ fructose syrups.
J.D. Sherman and C.C. Chao
Separation of Mannose from Glucose (5): D-mannose is the most efficient raw material for the manufacture of the sugar alcohol mannitol, by direct hydrogenation of the mannose, with approximately 100% yield of mannitol. In addition, L-mannose has been identified as one sugar in a series of reactions designed to produce L-sucrose, a possible non-nutritive sweetener (6). There 'are presently two major sources of mannose: epimerization of glucose (7), yielding a mannose/glucose mixture, and hydrolysis of hemicellulose or plant tissue (8), yielding mixtures of many sugars. 2.
It is known to use a cationic exchange resin (i.e., the Ca++ form of Rohm and Haas' Amberlite XE200) to separate mannose from glucose (9). However, this method seems to be inefficient, and a better sorbent would appear to be desirable to make the method of separation by adsorption practical. Recovery of mannose from plant tissue hydrolyzate is substantially more difficult since the mixture contains many different sugars. For example, sodium based sulfite liquor (a plant tissue hydrolyzate) contains: sodium lignosulfonate 61.5%, xylose 3.5%, arabinose 1.5%, mannose 14.2%, glucose 5.5% and galactose 3.8%. We have discovered that certain cation forms of zeolites X and Y have excellent selectivity and kinetic properties for mannose separation (5). separation factors are summarized below:
SEPARATION FACTORS FOR MANNOSE SEPARATION FROM OTHER SUGARS Zeolite Powder:
0.85 0.75 1.0 1.0 1.0 0.5 1.6
1.09 1.5 1.04 1.8 1.0 2.1 1.4
---------- ---------- ---------- ---------KX
NaX NaY CaY SrY BaX BaY Note:
* ** * 20 ** 160
1.05 1.0 1.6 2.4 1.8 2.7 2.6
---------- ----------0.95 1.5 1.6 4.1 1.8 1.5 4.2
2.9 >3.0 2.7
x 40 mesh granules cm long column of 30 x 50 mesh granules
BaX, BaY, SrY, CaY, and NaY can be used to separate mannose from glucose. Since BaY sorbs mannose more strongly than arabinose, galactose, glucose, xylose, and cellobiose, it is particularly suitable for recovering mannose from the hydrolyzate of hemicellulose. The superior mannose/ga1actose selectivity of the BaX might also be employed in a two-stage process with the BaY. Separation of Lactose. Galactose and Glucose (10): Approximately 0.6 kg (dry basis) of whey containing -73% lactose is produced in the manufacture of 1 kg of cheese. Since, for example, the U.S. alone produces 2 billion lbs/yr of whey solids, enormous quantities of lactose are available at low cost as a by-product of the cheese-making industry. 3.
Whey also contains -13% protein and -14% inorganic salts. Ultrafiltration can be used to remove the protein and processes such as ion exclusion can be used to remove the salts, leaving a lactose solution in water. Lactose hydrolysis over an enzyme catalyst can then be used to generate a mixture of galactose, glucose, and unconverted lactose.
We have found that certain zeolite adsorbents may be used to separate these sugars from each other llOl. For example, the data show that BaY provides a selectivity factor of 1.9 for the separation of galactose from glucose. BaY also provides a SF (glucose/lactose) = 2.9, indicating that this sorbent could also be used to separate lactose from both glucose and galactose. We believe adsorption separation processes based upon BaY and other zeolites offer opportunities for recovery of useful by-products from whey. 4.
Separation of lactulose from lactose [111: alternative use of lactose is in the manufacture of lactulose, by isomerization over an enzyme catalyst. Lactulose is a di-saccharide sugar constituted of galactose and fructose which has properties of considerable interest in the medical and food industries. An
Odawara [121 teaches the use of an X- or Y-type zeolite substituted with alkaline earth metal ions (preferably Ca, Sr or Ba) to separate the galactose and lactulose using selective adsorption. However, Bax zeolite adsorbs neither lactulose nor lactose significantly. CaY does not adsorb lactulose particularly strongly. As a result, CaY is not particularly effective in separating the two sugars, since much of the lactulose contains quantities of lactose, rendering it impure. In the case of BaY zeolite, the rate of approach to adsorption equilibrium is very slow, requiring a low process flow rate (-2.9 i/m2min.). At such a low flow rate the adsorption/desorption cycle time would be very long. 'rhus, the capital investment for the process would not be efficiently utilized. We have discovered that barium-exchanged zeolites with framework structures similar to that of the Y-type zeolite, but with much lower ion concentrations than either X- or Y-type zeolites (here designated "BaSY") not only provides higher capacities than conventional BaY zeolites, but also unexpectedly exhibited greatly improved rates of adsorption and desorption and much higher lactulose/lactose resolution factors [111. The RF(lactulose/lactose) = 0.38 for KSY, 0.4 for BaY, and 1.04 for BaSY. 5.
Separation of L-arabinose from other sugars [131: Carbohydrate chemistry of the human body centers around sugars with "D" configurations. No human enzyme can synthesize or digest sugars of "L" configurations. On the other hand, the non-enzymatic chemistry and general properties of L-sugars should be essentially identical to their D-counterparts. It is this combination which is expected to make L-counterparts of such common sugars as L-fructose, L-glucose and L-sucrose ideal diet (i.e. non-nutritive) sweeteners, because they should taste like D-sugars and should be safe, yet are expected not to be metabolized by human enzymes. L-fructose, L-glucose and L-sucrose do not occur naturally, but naturallyoccurring L-arabinose can be used to make L-glucose which, in turn, can be isomerized to L-fructose which, in turn, can react with L-glucose to make L-sucrose [61. L-arabinose is a 5-carbon sugar, which can react with cyanide or nitromethane to extend the carbon chain link to 6 and, in further reactions, remove nitrogen to produce a mixture of L-glucose and L-mannose. Both glucose and mannose are not good sweeteners; L-fructose is a good sweetener. The mixture of sugars has to be separated and further transformed into sweeter sugars. L-mannose can be isomerized to L-glucose and L-glucose can be isomerized to L-fructose.
J.D. Sherman and C.C. Chao
L-arabinose can be obtained by hydrolysis of beet pulp, which gives a mixture ot L-arabinose, D-galactose and sucrose. If stronger hydrolysis conditions are used, the product mixture will also contain glucose and fructose. If wood is used as a raw material, the product mixture will contain mannose and xylose. In order to realize the potential of L-sugars as diet sweeteners, the separation problem must be solved. First, the L-arabinose has to be separated from the other sugars in the hydrolyzate. Second, L-glucose has to be separated from L-mannose. (As discussed above, this mannose/glucose separation can be carried out using BaY and other zeolite adsorbents.) ,We have discovered that BaK zeolite sorbs L-arabinose much more strongly than it does other sugars expected to be present in mixtures with the L-arabinose, with L-arabinose separation factors (vs. other sugars) as follows: 2.1 (mannose), 2.9 (fructose), 3.1 (xylose), 4.2 (galactose), 5.6 (glucose), 42 (cellobiose), and 84 (sucrose) . 6.
Separation of inositol from sorbitol. fructose, glucose, sucrose : Inositol and sorbitol are sugar alcohols which have higher market value than do the common sugars. Inositol is a generic name for a family of cyclohexanehexols C6H6(OH)6, in which there are nine possible stereoisomers. One isomer, myo-inositol, predominates in nature, and is a commercially important compound. It is a member of the Vitamin-B complex, in which role it possesses activity as a growth factor for certain animals and microorganisms. It is also a factor in the regulation of fat and cholesterol metabolism in higher animals. For convenience, myo-inositol will be referred to simply as inositol. No commercial process for the synthetic production of inositol is known to us. However, because it widely exists in living cells, it is possible to extract inositol from plants. Inositol exists in the free form in many fruits; e.g., about 8 wt.% of the soluble carbohydrates in almond hulls is inositol. We have discovered that cation-exchanged type-X or Y zeolites can be used for the separation of inositol and/or sorbitol, as well as other sugars commonly found in mixtures with inositol. The separation factor data on a number of zeolites are summarized below. As may be seen, NaX and other zeolites may be used to easily separate inositol from sorbitol and other sugars commonly found in almond hulls and other sources in mixtures with inositol. SEPARATION FACTORS FOR INOSITOL SEPARATION FROM OTHER CARBOHYDRATES zeolite Powder (Si/A12 ratio)
-------------NaX NaY NaY CaX CaY BaX BaY
(2.5) (3.7) (5.0) (2.5) (5.0) (2.5) (5.0)
3.2 2.4 1.7 17.3 3.9 9.6 5.6
2.8 2.7 1.1 2.7 0.9 4.7 4.4
---------- ---------- ---------- ---------- ----------9.7 20. 1.8 >26. 33.5 115. 13.4
4.0 >20. 1.5 0.4 0.2 1.6 2.3
4.0 3.1 1.6 0.4 0.3 3.3 1.9
30 x 50 mesh granules
Detailed design and experimental studies by another company have shown this separation process  to be technically feasible and economically attractive for commercial use.
1030 (AP-6-l) 7.
Separation of mannitol, sorbitol and galactitol : Mannitol, sorbitol and galactitol are polyhydric alcohols made by reduction of sugars. Very often, the reduction reaction products are mixtures of these polyhydric alcohols. Mannitol and sorbitol are especially important commercial polyhydric alcohols in their pure forms. Mannitol may be produced from invert . sugar, fructose or from mannose. Production from invert sugar results in a mixture of sorbitol and mannitol in a ratio of approximately 3:1, requiring their separation in order to obtain a pure product. In general, this is done by successive recrystallizations, an expensive and time-consuming process. Another method to separate mannitol from sorbitol involves the use of the calcium exchanged form of a sulfonated polystyrene cation exchange resin which is cross-linked with divinylbenzene. We have discovered  that improved separations of sorbitol from mannitol and of mannitol from galactitol can be achieved using the barium-exchanged form of the Type X zeolite. BaX provides separation factors of 1.6 for the sorbitol/ mannitol separation, and 1.6 for the mannitol/galactitol separation also. Use of Bax and other zeolite sorbents for these separations (and for the separation of mannose, as discussed earlier) offers potential new low-cost processes to produce mannitol and other wood sugar derivatives by the use of zeolite sorbents to recover valuable by-products from the wastes of the wood pulping industry. MECHANISM Different cationic forms of the same zeolite exhibit very different selectivities, supporting the simple conclusion that sorbate-zeolitic cation interactions are of vital importance. However, the patterns of selectivity differ greatly between the Type X and Y zeolites even when they are in the same cationic form. This shows that features of the. zeolite structure beyond cation type alone are also of great importance in controlling sorption selectivities. It is also noteworthy that some zeolites and resin sorbents in the same cation form provide similar selectivities for some specific separations. The most prominent of these is the fructose/glucose separation, where both CaY zeolite and Ca++ resin sorbents are both in large-scale commercial use. A review of the literature on sugar-cation complexes reveals that weak-tomoderately-strong complexes are formed in aqueous solutions between Ca++ ions and many of the common sugars and sugar alcohols. For example, a review by S.J. Angyal  of the topic of sugar-cation complexes describes tridentate complexes formed between Ca++ ions and three adjacent (syn-axial) hydroxyl groups (which form an almost equilateral triangle) on either of the two (equivalent) chair forms of cis-inositol. It has been shown that the electrophoretic mobility of the neutral sugar molecules in Ca++ solutions is due their being dragged along by the Ca++ ions to which they are attached and that it is reasonable to assume that the relative mobilities of different sugar molecules is a good measure of the relative stability constants of the sugar-cation complexes formed. It has also been pointed out by Angyal [17,18) that the chromatographic separation of sugars and sugar alcohols on Ca++ resin columns is caused by the differing extent of complex-formation between the polyols and the Ca++ ion. Polyols that form such complexes are retained and those that do not, emerge rapidly. Qualitatively, the sequence of emergence of the compounds on Ca++ resin columns can be predicted from their relative mobilities from paper electrophoresis in Ca++ solution .
J.D. Sherman and C.C. Chao
The basic mechanism of separation of sugars and sugar alcohols on Ca++ zeolites would seem likely to be similar to that on Ca++ resins, i.e., the Ca++ ion is coordinated with and held strongly by the sorbent and the polyol molecules are complexed by the Ca++ ion. The stronger the complex formed between the Ca++ ion and the polyol, the greater the retention volume of the polyol, thus providing the chromatographic separation. The stability of some cation-sugar complexes has been determined by NMR measurements and it has been shown that relative electrophoretic mobilities of sugars in salt solutions are proportional to the relative stabilities of the cation-sugar complexes. The .relative mobilities of various sugars and sugar alcohols have been measured relative to cis-inositol (relative mobility = 1.00); e.g., in 0.2 H Ca acetate plus 0.2 H acetic acid at room temperature, the relative mobility of fructose is 0.07 and that of glucose is 0.02. Since about 67% of the cis-inositol present in the Ca++ solution is complexed, therefore, the corresponding percentages of each sugar complexed in solution would be approximately 4.7% for fructose and 1.3% for glucose.
Figure 1 - RETENTION VOLUMES ON CaY POWDER COLUMN vs. PERCENT COMPLEXED IN AQUEOUS SOLUTION
0.90 0.80 0.70 0.IIIIE
1! ~ a:
0.20 0.15 0.10 0.05 0
o Monosaccharides •
Electrophoretic Mobility, MI In 0.2M Ca++ Acetate + 0.2 M Acetic Acid Relative to CIS-Inositol (MI '" 1.000} at Room Temp.
1032 (AP-6-l) If the Ca++ ions in the zeolite are highly exposed to entering polyol molecules, such that Ca++/polyol complexes may be formed in the zeolite similar to those formed in solution, without serious steric hindrances, then, as a first approximation, the relative strength of adsorption of different polyols should be the same as their relative degree of complexing in solution (or, their relative electrophoretic mobilities). In addition, quantitative agreement found between tne two should provide clear proof that this is the primary mechanism controlling the separation.
Such a comparison is made in Figure 1 for data on CaY zeolite. As may be seen, fairly good quantitative agreement is found for most of the polyols studied, indicating that the relative strength of the cation/polyol complexes are of primary importance in controlling the separations. Thus the primary mechanism of sorption of a polyol on a zeolite involves a ligand-exchange type of adsorption of the polyol onto the zeoli tic cation, with exchange of the water molecules attached to a cation with a multi-dentate complex of groups of adjacent pairs or triplets of hydroxyl groups on the polyol attached to the cation. A polyol forming a tri-dentate complex will be more strongly sorbed than one forming a bi-dentate complex, and a polyol forming complexes with more than one cation will be more strongly sorbed than one with a single cation. However, in many cases, the relative order of elution on the CaY column is not the same as that predicted from the relative electrophoretic mobilities. Figure 1 shows that both mannitol and ribose are sorbed much differently than would have been predicted. Also, electrophoresis predicts the sugar alcohols will sorb in the order: i