Gas-Liquid Chromatography of Carbohydrate Derivatives

GAS-LIQUID CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

BY C. T. BISHOP Division of Biosciences, National Research Council, Ottawa, Ontario, Canada

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. General Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

95 96 96 ........................... 100 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 . 111. Classes of Carbohydrate Derivatives.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 1 . Methyl Ethers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 2. Acetates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 3. Acetals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 4. (Trimethylsilyl) Ethers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 5. Miscellaneous. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 IV. Structure and Mobility. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 V. Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Structural Chemistry of Polysaccharides and Related Compounds.. . . . . . . 129 2. Analysis of Mixtures of Monosaccharides 3. Analysis of Reaction Products.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 VI. Liquid Phases... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

I. INTRODUCTION In their first report on liquid-liquid partition chromatography, Martin and Synge’ made the following statement: “The mobile phase need not be a liquid but may be a vapor. Very refined separations of volatile substances should therefore be possible in a column in which permanent gas is made to flow over gel impregnated with a non-volatile solvent in which the substances to be separated approximately obey Raoult’s law.” This prediction was substantiated eleven years later by the successful extension of the partition method to gas-liquid systems.2 Since that time, gas-liquid chromatography has been developed and refined at a faster rate, and applied more extensively, than any other analytical method in the history of chemistry. It has been used with practically all classes of organic compounds, and the phenomenal growth in the literature on the subject shows no signs of diminishing. The latest bibliography, which includes all papers (1) A. J. P. Martin and R. L. M. Synge, Biochem. J . , 36, 1358 (1941). (2) A. T.James and A. J. P. Martin, Biocheni. J . , 60, 679 (1952). 95

96

C. T.

BISHOP

in which gas chromatography was used up to October 1961, runs to 4338 entries.s It is, therefore, impractical to cite even most of the applications in a single article. The first report on gas-liquid chromatography of carbohydrate derivatives was published‘ in 1958. This article was followed by a rapid development of the method and by an increasing number of applications to problems in this field. A later article6 emphasized practical aspects of the subject, and was written primarily to encourage wider use of the new technique in carbohydrate chemistry. A number of important papers have appeared since that time, and the utility of gas-liquid chromatography in this area of organic chemistry is now well established. The present Chapter includes a description of the general methods and basic principles of gas-liquid chromatography. The development of the method as applied t o various classes of carbohydrate derivatives is surveyed, and this treatment is followed by a discussion of mobilities of compounds i n relation to their structure. Applications of the method to specific problems are described, and the liquid phases used in gas-liquid chromatography of carbohydrate derivatives are summarized. Finally, an attempt is made to assess the relative merits of gas-liquid chromatography and other chromatographic methods as applied to carbohydrate derivatives.

11. GENERALMETHODS 1. Theory

Gas-liquid chromatography can be regarded as being the repeated distribution of components between a mobile gas and a stationary liquid, with each effective distribution being a theoretical plate. The separation achieved with any column will, therefore, depend on the separation per theoretical plate and on the total number of plates. The separation per plate (separation factor, a) will be determined primarily by the nature of the liquid phase and of the components to be separated. For two components, A and B, the separation factor is the ratio of their partition co~ k ~ ) ,for the gas and liquid phases. In practice, the efficients, k A / k B ( k > partition coefficients are directly proportional to retention volumes, defined in Fig. 1, so that the separation factor is given by equation 1. a =

VRA/VRB

(1)

(3) S. T. Preston, in “Grw Chromatography,” N. Brenner, J. E. Callen, and M. D. Weiss, eds., Academic Press hc., New York, N. Y., 1962, p. 571. (4) A. G. McInnes, D. H. Ball, F. P. Cooper, and C. T. Bishop, J . Chromatog., 1, 556 (1958). (5) C. T. Bishop, Method8 Biochem. Anal., 10, 1 (1962).

GAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

97

TIME

FIG.1 .--Quantities in Gas-Liquid Chromatography. (A, peak representing component A ; B, peak representing component B; I, point of injection of sample; VRA,retention volume of peak A, also equals d in expression (2); VRB,retention volume of peak B. All horizontal measurements are made along the base line.)

Since retention volumes are measured a t the centers, or maxima, of the zones A and B, any two components for which a differs from unity should be separable. However, zones broaden as they pass through a column, leading to overlapping, and the separation of two components for which a is close to unity requires columns of impractical lengths. In these circumstances, another, more selective, liquid phase must be found that gives a more favorable value for a. Guides to choosing liquid phases and a summary of those that have been used to separate carbohydrate derivatives are given in Section VI (see p. 141). The number (n) of theoretical plates is reflected in the sharpness of the peaks on the chromatograms, and may be calculated from the following expression.

n = [4 ( d / w ) l2 where d and w are expressed in the same units and are the retention volume at the base line and the width of the peak, respectively (see Fig. 1). The height equivalent to a theoretical plate (HETP) is obtained by dividing the column length by the number of plates. This expression provides a useful means of evaluating the efficiency of a column, but does not indicate what factors contribute to the efficiency nor the optimum conditions for operating a column. The number of theoretical plates in a column depends on several parameters, such as: the rates of diffusion in the two phases, the nature and flow rate of the gas, the uniformity of the packing, the amount of the liquid phase, and the temperature. These parameters have been related t o column efficiency by the rate theory of van Deemter and co-

98

C. T. BISHOP

workers6as expressed by the following equation.

where A is a quantity characteristic of the packing; d, is the averagc particle diameter; y is a correction factor for the tortuosity of the interparticle spaces; D,,, is the diffusion coefficient in the gas phase; u is the linear velocity of the gas; k’ is the fraction of the sample in the liquid phase divided by the fraction in the vapor phase; d, is the thickness of the liquid film; and Dli, is the diffusion coefficient in the liquid phase. Obviously, it would be a considerable task to find the optiniuni combination of all these variables, and no-one has yet attempted so complete an analysis of any system. Fortunately, most separations by gas-liquid chromatography can be achieved with little consideration for optimal efficiency. However, experiments designed to test the theory have, in general, supported the van Deemter equation (3) and have provided some useful rules-of-thumb for the practical operation of gas-liquid columns. Equation (3) can be written in the following simplified form. HETP

=

A

+ R / u + PU

(4)

where the three constants A , B , and C represent eddy diffusion, molecular diffusion, and resistance to mass transfer, respectively. Any diminution in these three terms will reduce the height equivalent to a theoretical plate and, hence, improve the efficiency of the column. The A term is made up of two factors, A and d, (sce equation 3) , representing packing irregularities and particle diameter. Uniformity of particle size and careful packing will reduce A, and reduction in particle size will reduce dp. The B/u term represents molecular diffusion of the sample molecules in the carrier gas; this will be greatly diminished at high flow-rates because of shortness of elapsed time. This term also decreases with increasing molecular weight of the carrier gas, so that nitrogen, argon, or carbon dioxide are preferable to hydrogen or helium for attaining greater column efficiency. However, the last two gases are still widely employed because their use results in good sensitivity of thermal-conductivity detectors. The Cu term, representing resistance to mass transfer, involves several parameters (see equation 3 ) . Thus, a liquid phase having good solvent properties for the compounds being separated will provide a higher value for k’ and increased efficiency. The effective thickness of the liquid film, d,, also appears in this term and is an important factor. Thinner liquid films result in increased efficiency down to a point below which k’ is decreased, thus counteracting the im(6) J. J. van Deemter, F. J. Zuiderweg, and A. Klinkenherg, Cheni. Eng. Sn‘., 6, 271 (1956).

QAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

99

provement. Furthermore, a severe reduction in the amount of liquid phase may simply cause incomplete coverage of the surface of the solid support rather than complete coverage by a thinner layer. The liquid diffusion coefficient (Dli,), which also appears in this term, becomes larger with decreasing viscosity of the liquid phase, resulting in greater efficiency. The flow rate of the carrier gas (u) occurs in two terms of the van Deemter equation and, from the form of the relation, there will be an optimal flowrate for the most efficient operation of the column. Thus, a t flow rates below the optimum, the molecular term will make a larger contribution to the HETP; a t flow rates above the optimum, the resistance to mass transfer becomes an important factor. The effect of temperature on efficiency enters equation (3) through the diffusion terms (Dgasand Dli,) and k‘. As the temperature increases, D,, and D I ~both , increase, with opposite effects on the HETP. The value of k’ will decrease with higher temperatures, as more sample will be forced into the gas phase, and the efficiency will be decreased. The net effect of all these changes is hard to predict for a given system, but, in general, lower temperatures will give improved efficiency. Factors not covered by the van Deemter theory for determining column efficiencies are the size of the sample and its introduction into the column. For these factors to have no effect on the efficiency, the sample would have to be introduced into that portion of the column that contained only a single theoretical plate. This is impossible to achieve in practice, but it is apparent that small samples will lead to greater efficiency. To avoid diffusion, the sample is best introduced as a plug of vapor occupying the smallest possible space consistent with the sample size. With some kinds of equipment, this is accomplished by injection through a silicone-rubber septum by means of a syringe; in others, the sample is introduced directly onto the column packing, where it is vaporized immediately. The preceding discussion was provided to give a basic understanding of the principles of gas-liquid chromatography and of the various factors that contribute to the separations achieved on packed columns. Much of the to discussion was drawn from a number of textbooks or (7) A. I. M. Keulemans, in “Gas Chromatography,” C. G. Verver, ed., Reinhold Publishing Corp., New York, N . Y., 1957. (8) C. Phillips, “Gas Chromatography,” Butterwortha Scientific Publications, London, 1956. (9) “Vapour Phase Chromatography,” D. H. Desty and C. L. A. Harbourn, eds., Butterworth Scientific Publications, London, 1957. (10) “Principles and Practice of Gas Chromatography,” R. L. Pecsok, ed., John Wiley and Sons, Inc., New York, N. Y., 1959. (11) “Gas Chromatography,” V. J. Coates, H. J. Noebels, and I. S. Fagerson, eds., Academic Press Inc., New York, N. Y., 1958. (12) “Gas Chromatography,”N. Brenner, J . E. Callen, and M. D. Weiss, eds., Academic Press Inc., New York, N . Y., 1962.

100

C.

T. BISHOP

which the reader is referred for further details. These references provide by no means a complete list of relevant textbooks; five other books, not yet available to the writer, have been favorably reviewed.13 A similar discussion of the theory of capillary c 0 1 u m n s ~ ~isJ ~regarded as beyond the scope of the present article, because most of the work in this field is still done with packed columns. To date, the extremely high resolving power of capillary columns has not been required for most of the problems involving separation of carbohydrate derivatives. Furthermore, the extremely small sample-requirements of these columns preclude any possibility of recovering the separated components from the effluent gas-stream, a necessary procedure for their conclusive identification. 2 . Apparatus

The process of gas-liquid chromatography involves the introduction of a sample into a column which contains the liquid phase on an inert support and through which the carrier gas is passed. The effluent is then passed through a sensitive detecting device which measures the variation in some property of the gas, and the signal from the detector is amplified and recorded. The equipment, therefore, consists of the following basic units : an injection mechanism, the column, a detector, a gas-flow systcm, an amplifier, and a recorder. The number of manufacturers of gas chromatographs has increased as the method has grown in popularity, and 110 specific recommendations can be made here. However, to be useful, the equipment should provide for adequate control of temperature and gas flow, should permit convenient and rapid change of columns, and should have a dependable and serviceable amplifying and recording system. In addition to these requirements, probably the most important feature of design is a minimum of dead space between the injection site and the outlet port. If there is much empty space between the point of injection and the column packing, the sample will diffuse and will enter the column as a broad zone; this results in impaired separations. The effluent gas should pass directly from the column into the detector; any empty space will (13) H. W. Habgood, Science, 199, 579 (1963); J. H. Knox, “Gas Chromatography,” Methuen and Co. Ltd., London, John Wiley and Sons, Inc., New York, N . Y., 1962; A. B. Littlewood, “Gas Chromatography,” Academic Press Inc., New York, N. Y., 1962; S. Dal Nogare and R. 6. Juvet, “Gas-Liquid Chromatography,” Interscience Publishers, New York, N. Y., 1962; H. Purnell, “Gas Chromatography,’,John Wiley and Sons, Inc., New York, N. Y., 1962; “Gas Chromatography Abstracts, 1961,” C. E. H. Knapman and C. G. Scott, eds., Butterworths Scientific Publications, Washington, D. C., 1962. (14) M. J. E. Golay, in “Gaa Chromatography,” D. H. Desty, ed., Butterworths Scientific Publications, London, 1958, p. 36. (15) Reference 11, p. 1.

GAS-LIQUID CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

101

again allow diffusion, and the detector response will not be a true representation of the separation actually achieved. With some kinds of equipment, the total effluent passes through the detector before reaching the outlet port; in others, a split-stream technique is used, with only part of the effluent going to the detector, the remainder being vented through an outlet port. In both cases, the amount of dead space from the end of the column to the outlet port should be as small as possible, if separated samples are to be collected from the effluent gas. Careful and independent control of the temperatures of the column, injection site, and detector should be possible. The column must be heated uniformly over its entire length, with a variation probably not greater than 0.1"C. Some units now being sold are adapted for programming of the temperature, that is, provision of a gradual increase in the temperature of the column at a predetermined rate. Units having this adaptation have a modified gas-flow system that compensates for the temperature-induced changes in flow rate and pressure which would, otherwise, cause a variation in detector response. Temperature programming is a useful technique for separating mixtures of compounds having widely differing boiling points. However, the column should still be heated uniformly over its whole length. If the sample is not introduced directly onto the column packing, it is injected into a vaporizing block which should be maintained a t a temperature higher than that of the column. Some detectors are sensitive to changes in temperature and therefore require very careful control. In most kinds of such equipment, the detector is thermostated separately from the column and is kept a t a higher temperature than the column, to prevent condensation of eluants. A convenient arrangement has the detector and vaporizing block in the same heating unit, which is kept at a temperature some 60-70" higher than the maximum operating temperature of the column. For most work with carbohydrate derivatives, column temperatures will range from 150" to 225", and the detector and injection block are kept a t 285". Columns may be of practically any shape and size and may be made from metal or glass tubing; the only limittltions are those imposed by the capacity of the heating oven and the need for perfect seals where the column fits into the gas-flow system. For maximum versatility, columns should be of a type that can be prepared in the laboratory and should be readily interchangeable. In practice, columns having internal diameters between 4 and 8 mm. and lengths of 4 to 20 ft. are convenient and give good results. Columns having diameters much larger than this are less efficient, and excessive length introduces a problem in achieving and controlling a practical gasflow. The detector is wed for indicating the presence of samples in the effluent

102

C. T. BISHOP

gas from the column. A complete discussion of the many ways in which this can be accomplished is beyond the scope of this article. The older, integral detectors that respond to accumulative change in the gas have been replaced to a great extent by differential detectors, because of a greater sensitivity and an easier interpretation of the results. Electroncapture detectors are of the greatest utility for compounds having high electron-affinities, but they offer little advantage for work with carbohydrate derivatives. Gas-density balances are difficult to manufacture and are not used as detectors on most of the commercial equipment. The two types of detector in most general use are differential detectors which measure changes in ( a ) thermal conductivity or ( b ) ionization. The thermal-conductiwity detector consists of two sensing elements (hotwire filaments or thermistors) incorporated into some form of a Wheatstonebridge circuit. One sensing element is kept in a stream of pure gas, and the other in the effluent gas-stream from the column. Emergence of a compound from the column then causes a change in the thermal conductivity of the effluent gas; this throws the bridge circuit out of balance. The difference in current is then amplified and is used to drive the pen of a recorder, thus producing the chromatogram. The performance of this detector will be best if there is a wide difference between the thermal conductivities of the carrier gas and the samples. Hydrogen and helium are best in this regard, the latter being preferred because of the potential explosion hazard with hydrogen. Nitrogen can also be used, but its thermal conductivity is much closer to that of many organic compounds, and there is a consequent loss in detector sensitivity. Thermal-conductivi ty detcctors are sensitive to minor variations in temperature and in gas-flow rate, both of which must therefore be carefully controlled. The sensitivity of this type of detector is difficult to define, because it depends on many variables, such as the molecular weight and structure of the compound being detected, the temperature, the gas-flow rate, the nature of the carrier gas, and the quality of the electrical circuitry. The thermal-conductivity detector can be used with relatively large samples, particularly if thc signal can be attenuated, as on most of the commercial apparatus. This is a distinct advantage if the separated components are to be collected from the effluent gas for identification by conventional methods. As a rough approximation, thermal-conductivity detectors are used with sample sizes in the rangc of 0.1-30 mg. Ionization detectors measure variations in saturation current of a gas passing bctwecn two electrodes across which a constant voltage is applied. The variation in current is caused by ionized, organic vapors as they emerge from the column, and is amplified and recorded as with the thermalconductivity detector. The organic vapors are ionized either by a source of

QAS-LIQUID

CHROMATOQRAPHY OF CARBOHYDRATE DERIVATIVES

103

ionizing radiation or by burning in a hydrogen flame. I n the former116 argon is used as the carrier gas and this is raised to its metastable state, but is not ionized, by a beta-radiation source of about 80 millicuries; the source may be radium or Btrontium-90. An organic vapor entering the detector is ionized by transfer of energy from the metastable, argon atoms, and it then produces a current between the two electrodes. In the hydrogen-flame detector,” the thermal ionization caused by combustion of the organic vapor is measured as a difference in the electrical conductivity between two flames, one in the column effluent and the other in the reference gas. Both of these detectors are relatively insensitive to minor fluctuations in temperature and gas-flow rate, and these operational parameters need not be controlled so closely as for thermal-conductivity detectors. Both types of ionization detector are much more sensitive than those based on thermal conductivity, and both are used for detecting samples in the microgram range. In fact, the difficulty of injecting such small samples can be a limiting factor with these detectors, and this is generally overcome by splitting the gas stream, either before it enters the column or before it enters the detector. To collect products from the effluent gas, using equipment provided with an ionization detector, it is necessary to split the effluent, so that part of it goes to the detector and the rest to an outlet port. 3. Qualitative and Quantitative Analysis

The retention volume of a compound, determined as shown in Fig. 1 (see p. 97), is a characteristic property when all of the conditions of the separation can be duplicated exactly. To avoid slight differences that may be caused by minor variations in the operating conditions, retention volumes are usually made relative to that of a standard compound which is assigned a value of unity. An unknown may therefore be identified by comparison of its retention volume with that of an authentic sample, both made relative to the same standard and obtained under the same conditions. The condition most difficult to duplicate is the column packing, and this situation makes it unlikely that relative retention volumes reported can be reproduced with numerical exactitude. However, the relative orders of a series of such values should be reproducible, provided that the same liquid phase is used. It should be emphasized that identification by the comparison of relative retention volumes does not obviate the need for confirming the identification by conventional methods. Wherever possible, the separated components should be collected from the effluent gas-stream and be identified by infrared spectroscopy and by the preparation of (16) J. E.Lovelock, J . Chromalog., 1, 35 (1958). (17) I. G. McWilliam and R. A. Dewar, Ref. 14, p. 142.

104

C.

T. BISHOP

crystalline derivatives. A number of workers have described systems for repeated, automatic, sample injection and fraction collection in gas chromatography,1a-22 and such systems are available commercially. When available quantities are too small to permit positive identification, gas-liquid chromatography, preferably on different liquid phases, offers a better criterion of tentative identification than other chromatographic methods because of its greater resolving power. However, when compounds are thus tentatively identified, the precautions discussed later (see Section VII, p. 145) should be observed, in order to make sure that the product had not been modified during the chromatography. Detectors vary in their response to the compounds being sensed, with respect both to the amount and to the class of compound. Since the thermal conductivity of a mixture of two gases (the carrier gas and the sample) is not necessarily a linear function of composition, absolute quantitative measurements with thermal-conductivity detectors can only be obtained after calibration with standard compounds. However, response to structurally similar compounds is quite uniform, and reliable, relative, quantitative data can be obtained with this detector. Ionization detectors give a signal that is related directly to the mass or the concentration of the components, and the response is uniform over a broad range of operating conditions; these detectors are, therefore, ideally suited for quantitative work. Quantitative estimations are made by measuring the areas under the peaks given by the separated components. This measurement can be accomplished by using an automatic integrator or a planimeter, by triangulation of the peaks (compare Fig. 1), from the product of peak height and retention time,10p23 or by tracing the peaks onto paper of uniform density from which they can be cut and weighed. The last three of these methods have been compared in the analysis of a mixture of eight fatty acid methyl esters.24 The results were in excellent agreement, and the minor variations probably represent the experimental error in quantitative, gas-liquid chromatography. The choice of the method for measuring areas of peaks is, therefore, a matter of convenience and personal preference. Some caution should be exercised in cutting out peaks for weighing. Considerable error can be introduced in cutting out sharp peaks, and the (18) (19) (20) (21) (22) (23) (24)

D. Ambross and R. R. Collerson, Nature, 177, 84 (1956). E. P. Atkinson and G . A. P. Tuey, Ref. 14, p. 270. J. Hooimeijer, A. Kwantes, and F. van de Craate, Ref. 14, p. 288. E. Heilbronner, E. Kovats, and W. Simon, Hetv. Chim. A c h , 40, 2410 (1957). S. Sideman and J. Gilladi, Ref. 12, p. 339. J. C. Bartlet and D. M. Smith, Can. J . Chem., 38, 2057 (1960). K. K. Carroll, Nature, 191, 377 (1961).

QAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

105

paper used must be of uniform density, a requirement not fulfilled by chart paper.

111. CLASSESOF CARBOHYDRATE DERIVATIVES 1. Methyl Ethers a. Fully Methylated Methyl Glycosides.-Compounds that can be distilled or sublimed, even under diminished pressure, are generally amenable to separation by gas-liquid chromatography. Before the advent of paper chromatography, methylated monosaccharides were separated from each other by fractional distillation; it is, therefore, not surprising that these were the first carbohydrate derivatives on which gas-liquid chromatography was tested. The first report of gas-liquid chromatography of carbohydrate derivatives described the separation of fully methylated methyl glycopyranosides of D-xylose, carabinose, D-glucose, D-mannose, and D-galactose.' Fig. 2 shows the separation and the conditions under which it was obtained. Single anomers of each monosaccharide were used, so that starting materials could be well characterized and to avoid confusion in identifying the separated components. The fully methylated methyl

A

I I

I

2

3

B 4

TIME

6

(mid

FIQ. 2.Separation of Fully Methylated Methyl Glycopyranosides.4 (Peaks in order of appearance: (1) methyl 2,3,4-tri-0-methyl-@-~-xylopyranoside; (2) methyl 2,3,Ptri-Omethyl-&r.,-arabinopyranoside; (3) quinoline (standard); (4) methyl 2,3,4,6-tetra-Omethyl-a-D-mannopyranoside; (5) methyl 2,3,4,6-tetra-0-methyl-a-~-glucopyranoside; (6) methyl 2,3,4,6-tetra-O-methyla-~-galactopyranoside. Conditions: column (8 ft. x 7 mm.) of Apieeon M on Celite 545 (1:4 w/w); 170"; 130 ml. of helium/min.; thermal conductivity detector.)

106

C. T. BISHOP

2

a z

FIQ.3.Separation of Fully Methylated Methyl a- and fl-D-Glycopyranosides.*s (Xyl = xylose; Ara = arabinose; G = glucose; Man = mannose; Gal = galactose. Vertical dashes are at 10-min. intervals. Conditions: column (6 ft. x t in.) of fully methylated 0-(2-hydroxyethyl)cellulose on Chromoeorb (3:7 w/w); 190"; 34 ml. of helium/min.; inlet pressure, 10 p.8.i. Thermal-conductivity detector.)

glycosides could be recovered quantitatively and unchanged from the effluent gas-stream, showing that no anomerization, hydrolysis, or decomposition had occurred during the separation. These results established

TIME (rnins)

FIG.4.-Separation of Fully Methylated Methyl Glycosides of Sugars Present in Maple-sap Arabinoga1actan.P' (Key: (I) methyl 2,3,4tri-O-rnethyl~-~-rhamnopyranoside; (I1 and 111) methyl 2,3,5tri-0-methyl-a,~--carabinofuranosides; (IV) methyl 2,3,4-tri-O-methyl-a,~--carabinopyranosides; (V) methyl 2,3,5,6-tetra-O-methyl-a,8-~galactofuranosides; (VI) methyl 2,3,4,6-tetr~-methyl-a,8-~-g~actop~~osides. Conditions: column (4 ft. x 4 mm.) of butanediol succinate polyester on Celite (2:8 w/w); 150"; 150 ml. of argon/min.; 8-ionization detector.)

GAS-LIQUID

107

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

the utility of gas-liquid chromatography for this class of carbohydrate derivatives. The resolving power of the method was demonstrated by the successful separation of anomeric methyl glycosides as their fully methylated derivatives. Thus, as shown in Fig. 3, Kircher26was able to separate the anomeric methyl pyranosides of 2,3,4-tri-0-methyl-~-xylose; of 2,3,4,6tetra-0-methyl-D-glucose; and of 2,3,4,6-tetra-0-methyl-~-mannose. As in

I

40

I

30

1

20

Time (minutes)

1

10

I

0

FIQ. 5.-Separation of Fully Methylated Methyl D-Glucosides and Methyl D-Fructo(2)methyl 1,3,4,6sides.28 (Key: (1)methyl 2,3,4,6-tetra-0-methyl-8-~-glucopyranoside; tetra-0-methyl-a-D-fructofuranoside; (3) methyl 1,3,4,6-tetra-0-methy~-~-~-fructofuranoside; (4)methyl 2,3,4,6-tetra-0-methyl-a-~-glucopyranos~de; (5) methyl 1,3,4,5tetra-0-methyl-a-D-fructopyranoside; (6) methyl 2,3,5,6-tetra-0-methyl-(?)-~-glucofuranoside; (7) methyl 2,3,5,6-tetra-0-methyl-(?)-~-glucofuranoside; (8) methyl 1,3,4,5tetra-0-methyl-8-D-fructopyranoside.Conditions: column (5 ft. x 2 mm.) of 5% diethyleneglycol succinate polyester on Chromosorb W; 136'; 16 ml. of helium/min.; thermal-conductivity detector.)

the earlier work14the anomeric methyl glycosides of 2,3,4-tri-O-methyl-~were not separated. arabinose and of 2,3,4,6-tetra-0-methyl-~-galactose The separation of fully methylated furanosides from pyranosides in the Larabinose and D-galactose series, and the separation of a fully methylated methyl rhamnoside from these components, are shown26in Fig. 4; again, anomeric pairs were not separated. However, complete resolution of the four methyl glycosides of D-xylose and of D-arabinose, as their fully methyl(25) H.W.Kircher, Anal. Chem., 32, 1103 (1960). (26) G.A. Adams and C. T. Bishop, Can.J . Chem., 38, 2380 (1960).

108

C. T. BISHOP

ated derivatives, has been r e p ~ r t e d ,and ~ ' similar results have been obtained by Gee and Walker28in the n-glucose and n-fructose series, as shown in Fig. 5. The latter authors also reported the first separation of thc anomeric methyl pyranosides of 2,3,4,6-tetra-0-methyl-n-galactoseby gas-liquid chromatography. Most of the common sugars can, therefore, be separated from each other by gas-liquid chromatography of their fully methylated glycosides, and, under appropriate conditions, the four methyl glycosides of a single sugar can be resolved. Applications of these results to specific problems is considered in Section V, 2 and 3 (see pp. 137, 140). The separation of fully methylated methyl glycosides has been extended TABLEI Retention Volumes of Fully Melhylated Disaccharides Relative to Octa-0-methylsucros$ Retention volumes* Fully niethylated disaccharide Sucrose Trehalose Cellobiose Maltose Melibiose Lactose

3-O- Arabinopyranosyl-arabinopyranose

3-0-Arabinopyranosyl-arabinofuranose a

Methyl 8-glycoside

1.75 2.02 1.86 1.87 0.75 0.58

Methyl a-glycoside

1.00 1.44

1.92 2.30 2.21 2.03 0.86 0.68

Column, 20% w/w of Apiezon M grease on Celite 545; 220";300 ml. of argon/mnin.

to the derivatives of di- and tri-saccharides. Fully methylated sucrose was shown to be sufficiently volatile for application of gas-liquid chromatography,2bbut its separation from other methylated disaccharides was not attempted. Subsequently, it was shown that mixtures of fully methylated disaccharides can be resolved6 on a nonpolar, liquid phase, Apiezon M. Retention volumes of some fully methylated disaccharides are given in Table I, and the results show that two anomers, as well as isomers, can be resolved. Similar results have been obtained on columns containing a liquid phase of 5% neopentylglycol succinate polyester; this was also used for demonstrating separation of the fully methylated derivatives of the trisaccharides melezitose and raffinose.28 (27) C. T. Bishop and F. P. Cooper, Can. J . Chem., 41, 2743 (1963). (28) M. Gee and H. G. Walker, Anal. Chem., 34, 650 (1962).

GAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

109

TABLEI1 Retention Volumes of the Anomers of Methyl Tetra- and Tri-0-methyl-D-gluwpyranosidecP

Compound

On butanediol succinate polyestev

On Apiezon Ma

0.67

0.82

2,3,4,6-Tetra-, j3a anomer 2,3,4-Tri-, j3a anomer 2,4,6-Tri-, j32,3,6-Tri-, j33,4,6-T&, j3a anomer 2,3,6-Tri-, a2,4,6-Tri-, a-

1.000 0.98 1.12 1.09 1.12 1.13 1.13 1.26 1.38

1.000

1.78 2.19 2.34 2.48 2.62 2.65 3.56 3.56

, Conditions: Column (4 ft. X 4 mm.) of 20% by weight of liquid phase, on alkaliwaahed Celite 545; 150'; 60 ml. of argon/min.

L

a

2,3,4 I

I

I

B L I

a

2,386 I

I

I

I

I

D

a

r

L

J

2,486 I

I

I

I

I

a, 3,4,6

0

10

20TIME IN30 MINUTES40

50

Fro. 6 . S e p a r a t i o n of Methyl Tri-0-methyl-a- and fi-D-Glucopyranosides.ao(Conditions: column (10 ft. x 5 in.) of 20% by weight of butanediol succinate polyester on firebrick; 200"; 77 ml. of helium/min.)

CL

+

0

TABLE I11 Relative R e t e n t h Volumes of Methyl 0-Methyl-D-glucopyramsides on Polyester Columns at Different Temperatures81

Methyl 0-methyl-Dglucoloyramside 2,3,4,6-TetrA-methylfl anomer 2,3,4-Tri-O-methyl-~ 19anomer 2,3,6-Tri-O-methyl-afl anomer 2,4,6-Tri-O-methyl-afl anomer 3,4,6-Tri-o-methyl-~2,3-Di-O-me thyl-afl anomer 4,6-Di-O-methyl-afl anomer 3-Di-O-methy la?-Di-O-meth yla?-Di-0-methyl-a-

1

1.00

0.70 2.11 1.80 3.52 2.50 3.52 2.38 2.61 -

-

-

-

150"

170"

220"

Columna

Column.

Column.

8

1.00

0.69 2.34 1.98 3.90 2.70 3.90 2.66 2.89

-

-

-

-

-

-

-

-

-

3

1.00

0.75 2.18 1.68 3.68 2.19 3.68 2.12 2.66 -

-

1

1.00

0.73 1.95 1.75 2.92 2.34 2.92 2.25 2.32 -

-

-

-

9

1.00

0.71 2.52 1.55 3.87 2.84 3.87 2.76 3.53 5.49 -

3

1.00

0.65 2 .08 1.40 2.96 1.98 2.96 2.02 2.47 -

-

7.41

-

5.43 -

1

1.oo

0.79 1.91 1.75 2.64 2.02 2.64 2.02 2.23

-

4.58 3.93 -

-

9

1.00

3

1.00

0.77 1.69

0.72 1.84

3.18 2.52 3.18 2.44 2.29

2.59 2.18 2.59 2.18 6.66 5.30 4.92 6.10 3.90 7.90 11.oo

-

-

4.90

-

4.02

-

-

a All packinga were commercial products: 1, 1,4butanediol succinate polyester; 2, diethylene glycol succinate polyester; 3, ðylene glycol succinate polyester that had been used extensively for analysis of fatty acid esters.

9

=

1

s

GAS-LIQUID CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

11 1

b. Partially Methylated Methyl G1ycosides.-Much of the early work on the separation of partially methylated methyl glycosides was concerned with the methyl ethers of D-glucose. Thus, all of the possible trimethyl ethers of methyl a- and p-D-glucopyranoside were found to be separable by gas-liquid chromatography, using first a polar and then a nonpolar liquid phase.29The retention volumes of these compounds, relative to methyl 2.3.4,G-tetra-0-methyl-a-~-glucopyranos~de, are given in Table 11. A striking example of the separation of anomeric pairs in the tri-0-methyl-Dglucoside series was recorded by Kircher30and is shown in Fig. 6. Further separations of methyl tri-0-methyl-D-glucosides were reported by Klein and Barter,31who investigated the effects of temperature and of different polyester liquid phases on the separations. Their results, in Table 111, showed that better separations are obtained at lower temperatures; also, this study embodied the first systematic attempt to separate isomeric methyl di-0-methylhexosides. The anomeric methyl glycosides of 2,3- and 4,G-di-0-methyl-D-glucose were resolved, and other methyl di-O-methylD-glucosides were well separated but could not then be identified because of a lack of standards. The latter separation was obtained on a diethylene glycol succinate column that had been used for fatty acid ester analysis and was assumed to have lost a considerable proportion of its liquid phase. The results suggested that columns having a low concentration of liquid phase would give better separations of methyl di-0-methylhexosides. Prior to this investigation, it had been shown that the methyl glycosides of di-0methyl-D-glucoses could be separated from the tri- and tetra-methyl ethcrs, but that individual isomers are poorly resolved.2gKircheP had also shown that methyl di-0-methylhexosides are amenable to gas-liquid chromatography, by separating methyl 2,4-di-O-methy1-a- and -,B-Dglucopyranosides from the other methanolysis products of a methylated dextran (see Fig. 11, p. 131), and methyl 2,3-di-O-methyl-a,/3-~-mannopyranoside from the methanolysis products of methylated guaran (see Fig. 9, p. 130). Aspi11alP examined a wide range of methylated and partially methylated methyl glycosides by gas-liquid chromatography on two liquid phases. The retention volumes of these compounds, relative to are shown in Table IV methyl 2,3,4,G-tetra-O-methyl-~-~-glucopyranoside, and should be of considerable assistance to other workers in applying this technique. All of the sugars examined had been characterized as crystalline dcrivatives and were then converted to methyl glycosides by refluxing (29) C. T. Bishop and F. P. Cooper, Can. J . Chem., 38, 385 (1960). (30) H. W. Kircher, in “Methods in Carbohydrate Chemistry,” R. L. Whistler and M. L. Wolfrom, eds., Academic Press Inc., New York, N. Y., Vol. 1, 1962, p. 13. (31) E. Klein and C. J. Barter, Textile Res. J., 31, 486 (1961). (32) G. 0. Aspinall, J . Chem. Soc., 1676 (1963).

TABLE IV Relative Retention V o l u m of Methyl Glyco&" Methyl glycoside of 2,3,4-Tri-O-methyl-earabmose 2,3,5-Tri-O-methyl-barabmose 2,3-Di-O-methyl-tarabinose 2,4-Di-O-methyl-earabinose 2,5-Di-O-methyl-~-arabinose

3,4-Di-O-methyl-~arabinose 3,5-Di-O-methyl-carabmose 24-Methyl-barabmose 34-Methyl-carabmose

2,4Di-O-methyl-~-xylose 3,4-Di-O-methyl-~-xylose 20-Methyl-D-xylose 34-Methyl-D-xylose PO-Methyl-D-xylose 2,3,4-Tri-O-methyl-erhamnose 3,4Di-O-methyl-~-rhamnose 34-Methyl-crhamnose 2,3,4-Tri-O-methyl-cfucose 2-0-Methyl-cfucose

CL

c

1.04 0.56 (8)

{EE) 2.26 (sh) 1.89 (8) 2.15 1.08

3.42 (w) j6.l 6.95 (m) 0.46 (m) 1.50 (m)

{

1.79 (8) 1.49 (m) 1.36 (8) 4.11 3.55 (8) 4.35 0.46 0.73 (w) 3.66 0.72 4.25 1 .OO (m) 2.59 (m) 3.52 (m) 3.31 (m)

E3

Column a

Column l b -

0.72 (m) 1.76 (w) -

2.37 3.47 (w) -

2.55 -

4.47 (8) -

0.57 ( 8 ) 1.65 (w) -

1.97 (8) 1.63 (m) 6.23 5.57 (m) -

1.01 (8) -

0.83 0.47 (8)

{E2)

1.09 (sh) 0.70 (8) 0.99 0.60

cpi;;I

(0.45(m)

r.,,

(m) 0.73 (m) 0.71 (8) 1.01 0.94 (8) 1.09 0.46 0.61 1.01

-

{!1.00 : (m) %(%)

1.43 (8) 3.70 (8) 4.78 (8) 4.88 (8)

1.35 (m) 1.71 (m) 1.64 (m)

-

0.59 (m) 0.82 (w)

-

1.13 1.03 (w) 1.52 0.84 1.45 1.26 (8) 1.58 (m) 0.54 (8)

0.76 (m) 1.34 1.15 (m)

-

1.15 (m) 1.54 (w) 1.32 (8) 1.83 (8) 2.18 (8) 2.24 (8)

P

9

-

3.73 (m) 3.93 (vw) 4.70 (m) 4.70 ( 8 ) -

-

-

1.42 3.11 5.08 3.08 -

-

{z; ::; 1.29 1.65 2.35

1.71

2.53 (m) 8 . 4 (m) 7.18

4.12 (m)

-

3.24 9.3

(8) (8)

3.37 2.26 1.01 (8) 1.10 (w) 1.71 (m) 2.39 (m) 1.52 (m) 1.72 (a) 3.02 (w) 1.77 (m) 2.47 (m) 3.90 ( 8 )

3.22 ( 8 ) 3.22 (a) 1.60 2.89 (a) 2.07 (vw) 2.49 (m) 2.38 ( 8 ) 3.19 (vw) 4.20 (m) 4.40 ( 8 ) 3.21 (m) 1.16 (m) 1.31 (s) 2.05 (vw) -

1.79 ( 8 ) (m)

2.23 2.21 3.12 4.26

(a) (a) (m)

113

The relative intensities of peaks are indicated aa strong (s), medium (m), weak (w), and very weak (vw). Incomplete resolution is indicated by shoulder (sh). b Conditions: Column (120 X 0.5 cm.), 15% by weight of 1,4butanediol succinate polyester on acid-washed Celite (80-100 mesh); 175";80-100 ml. of argon/min.; ,%ionization detector. Conditions: Column (120 X 0.5 cm.), 10% by weight of polyphenyl ether [m-bis-(m-phenoxyphenoxy)benzene]on acid-washed Celite (80-100 mesh); 200";80-100 ml. of argon/min.; &ionization detector. As the methyl ester.

DERIVATIVES

2,3,4-Tri-O-methyl-~-glucuronic acidd 2,3-Di-O-methyl-~-glucuronic acidd 2,3,4-Tn-0-methyl-~-galacturonicacidd

(8)

{ZI(w)s' 3.05 3.72 (m)

3.94 (m) 2.74 -

2.46 (m) 2.30 (m) 1.52 (sh) 2.62 (m) 1.61 ( 8 ) 2.23 (w)

GAS-LIQWID C~ROMATOGRAPHY OF CARBOHYDRATE

3.12 ( 8 ) 1.80 7.5 3.21 ( 8 ) 4.30 (w) 4.17 (m)

114

C. T. BISHOP

with 3% methanolic hydrogen chloride for six hours. This procedure resulted in the formation of two, and from certain sugars four, methyl glycosides. Although specific glycosides from a single sugar could not be identified, the proportions and retention volumes of the individual components were found to be characteristic of the parent sugar. Gas-liquid chromatography of the methyl glycosides from some sugars showed only one peak, due either to lack of resolution under the conditions used or to the formation of a single, pure anomer. One sugar, 1,3,4-tri-O-methyl-~fructose, showed five peaks on gas-liquid chromatography after glycosidation, possibly because of the formation of some of the dimethyl acetal. Occurrence of more than two peaks from a single sugar could be accounted for by formation of both furanosides and pyranosides, because all such sugars were substituted in such a way that both ring forms could exist. The retention volumes recorded in Table IV were reproducible to within 2%, or better, on a single column, and did not vary by more than 5% on different columns containing the same liquid phase. A difference of 5y0 or more in retention volumes of components in a mixture provided sufficient resolution to indicate the presence of each component. The results described in this Section show that gas-liquid chromatography is a highly selective method for separating a wide range of methylatcd and partially methylated methyl glycosides. The only systematic attempts to separate all of the anomeric and isomeric methyl glycosides of a single which sugar were those with the methyl tri-0-methyl-~-glucosides,~~-~~ were performed in order to illustrate the powers of this technique. For practical applications, such separations are largely acadcrnic, because mixtures containing this many isomers are seldom encountered. c. Partially Methylated Polyhydric Alcohols.-Gas-liquid chromatog4' APIEZON M 200°C

,

, 20 Ole

ON C E L I T E 5 4 5

6 0 ml. ARGON/MIN. I

TIME (MINUTES)

FIQ.'I.--Separation of Penta-0-acetyl mono-0-methyl-D-glucitols."(Key: (1) 6-0methyl-D-glucitol pentaacetate; (2) 2-O-methyl-n-glucito1 pentaacetate; (3) W-methylD-glucitol pentaacetate; (4) 4-O-methyl-~-glucitolpentascetate.)

GAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

115

raphy of mono-0-methylhexoses as their methyl glycosides has not yet been reported. These compounds are suEciently volatile to pass through columns at high temperatures and at high flow-rates, but are poorly resolved under these conditions. Thus, in the only separation of mono-0methylhexose derivatives that has been reported,2ethe sugars were converted into the more volatile mono-0-methylhexitol pentaacetates by successive reduction and acetylation. Fig. 7 shows the separation of the acetates of the 0-methyl-D-glucitols derived from the four possible 0methyl-D-glucopyranoses in this way. Although this procedure is satisfactory for resolving a mixture of mono-0-methylhexoses, the more volatile di-, tri-, and tetra-0-methylhexitol acetates would, if present, probably be poorly resolved at the high temperatures required. With new techniques of temperature programming, more-stable liquid phases, and columns having low concentrations of liquid phase, the complete analysis of a mixture containing mono-, di-, tri-, and tetra-0-methylhexoses as either their methyl glycosides or corresponding hexitol acetates should now be possible. 2. Acetates Gunner, Jones, and Perry33were the first to report the successful separation of a mixture of fully acetylated alditols. These authors utilized a novel technique of packing columns with a mixture of liquid phases; and, of the eleven 0-acetylalditols examined, only D-glucitol hexaacetate and galactitol hexaacetate were not separated. This work was later extended to a wider range of 0-acetylalditols and to the separation of glycose acetates.34 An example of the separations obtained is shown in Fig. 8; Table V lists the retention volumes of a number of alditol acetates on four different column packings. These packings were all mixtures containing a polar (polyester) and nonpolar (hydrocarbon) liquid phase. On columns containing only the polar liquid phase, the alditol acetates were well resolved but had large retention volumes; the nonpolar phase, alone, gave only fair resolution, but the compounds emerged quickly and gave symmetrical peaks. It was, therefore, considered that mixtures of polar and nonpolar liquid phases might combine the advantages of each, and give columns on which the acetate derivatives would have reasonable retention volumes and yet be well separated. The derivatives of high molecular weight were severely retarded on the columns containing 20% of liquid phase, but emerged within reasonable times on the packing of 0.3% of grease on glass beads. This packing gave a rapid separation of the acetates of the Cq to Cs alditols, in the order of increasing molecular weight, but isomeric alditol (33) S. W. Gunner, J. K. N. Jones, and M. B. Perry, Chem. Znd. (London), 255 (1961). (34) S. W. Gunner, J. K. N. Jones, and M. B. Perry, Cun. J . Chem., 3B, 1892 (1961).

116

C. T. BISHOP

acetates were not resolved. The mixed columns, containing 20% of liquid phase, also gave good separations of glycose acetates, as shown in Table VI, and anomers, as well as isomers, were resolved. These results have been confirmed in other r e p o r t ~ on ~ ~the s ~ separation ~ of glycose acetates. One of these36 described the use of a fluoroalkyl silicone polymer, QF-1, as a BC

COLUMN PACKING

0

5

I0

c,

I5

~ I S ' C , 2 0 0 ARGON/MIN ~

20

25

35

TIME (MINUTES)

FIG. S.--Separation of Alditol acetate^.^' (Key: (A) tetra-0-acetylerythritol; (B) penta-0-acetylribitol; (C) penta-0-acetylxylitol; (D) hexa-0-acetylallitol: (E) hexa-0acetyh-iditol ; (F) hepta-0-acetyl-meso-glycero-allo-heptitol; (G) hepta-0-acetyl-Dglycero-D-manno-heptitol; (H) hepta-0-acetyl-bglycero-D-gluco-heptitol;and (I) octa-0acetyl-D-e7ythro-Lguluc~~-octitol.Conditions: column (120 x 0.5 cm.) of a 1 :1 (v/v) mixture of the two following packinga: (1) 1 :1 (w/w) mixtures of (a) 20% w/w butanediol succinate on Chromosorb W, 6&80 mesh, and (b) 20% w/w Apiezon M grease on silver-coated Chromosorb W, 60-80 mesh, with (2) 0.3% w/w Apiezon M grease on silver-coated glass-beads, 60-plus mesh; 213"; 200 ml. of argon/min.; @-ionization detector.)

liquid phase for separating polyacetates, and showed that it was highly selective for these compounds. The use of thin-film columns (1% of liquid phase) reduced the retention volumes without sacrificing resolution and permitted the separation of disaccharide acetates. A large number of disaccharide acetates have been separated on thin-film columns using glass (35) W. J. A. Vanden Heuvel and E. C . Horning, Biochem. Biophys. Reg. Commun., 4, 399 (1961). (36) R. J. Ferrier, Chem. Ind. (London), 831 (1961).

117

GAS-LIQUID CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

TABLE V Retention Volumess' of Fully Acetylated Alditols Relative to Penta-O-ucetyl-Larabinitol at dls"

Acetate of Glycerol Erythritol D-Threitol Ribitol t Arabinitol Xylitol Allitol D-Takol D-Mannitol D-Glucitol Galactitol D-Iditol 6-Deoxy-ctalitol 6-Deoxy-tmannitol (trhamnitol) 6-Deoxy-~-galactitol (D-Fucitol) [Pentaerythritol meso-glycer0-allo-Heptitol D-glycero-D-ultro-Heptitol D-glycero-D-manno-Hepti to1 meso-glycero-gulo-Heptitol D-glyczro-D-gahclo-Heptitol D-glycero-D-gluco-Heptitol cglycero-D-ghco-Heptitol meso-glycero-ido-Heptitol

A 0.120 0.378 0.403 0.918 1.000 1.18 1.78 2.07 2.13 2.39 2.42 2.74 0.707 0.715 0.750 0.807

Column packing. B C 0.118 0.358 0.429 0.895 1.000 1.18 1.89 2.16 2.25 2.59 2.59 2.97 0.712 0.733 0.778 0.798

5.60

D-eTythTO-Lga~llCtO-oCtitOl

D-threo-cgalacto-Octitol 5-&3-D-Xylopyranosyl-barabinitol

0.m 0.393 0.433 0.944 1.000 1.20 1.90 2.03 2.16 2.73 2.75 2.85 0.698 0.730 0.750 0.790 3.56 4.06 4.73 4.90 5.05 5.10 6.11 6.60 11.50 12.61 -

D

1.000

-

3.92 6.52 22.3

Column packings: (A) a 1:1 (w/w) mixture of (1) 20% w/w 1,4-butanediol succinate polyester on Chromosorb W, 60-80 mesh, with (2) 20% w/w Apiezon M grease on silver-coated Chromosorb W, 60-80 mesh; (B) a 1:l (w/w) mixture of (1) 20% w/w 1,4butanediol succinate polyester on Chromosorb W, 60-80 mesh, with (2) 20y0 w/w Dow-Corning grease on Chromosorb W, 60-80 mesh; (C) a 1:1 (v/v) mixture of packings A and D ; and (D) 0.3% (w/w) of Apiezon M grease on silver-coated glass beads, 60plus mesh.

beads as an inert support.37Table VII shows the retention volumes of these derivatives and the column packings used for their separation. Acetates of reducing disaccharides gave broad peaks under these conditions, and (37) H. G. Jones and M. B. Perry, Can. J . Chm., 40, 1339 (1962).

118

C. T. BISHOP

TABLEVI Retention Volumesa4of Glywae Acetates Relative to Penta-O-acelyl-barabinilol a1 213" Column packing" Ester Tetra-0-acetyl-a-D-xy lopyranose p anomer Penta-O-acetyl-a-D-galactopyranose p anomer Penta-O-acety1-a-D-glucopyranose j3 anomer Penta-O-acetyl-a-D-mannopyranose p anomer Tetra-0-acetyl-a-D-ribopyranose Tetra-0-acetyl-a-D-ribofuranose Penta-O-acetyl-a-D-a1tropyranose Tetra-O-acety1-0-4yxopyranose Hexa-0-acetyh - g l y cero-a-D-manno-heptopy ranose Hexa-0-acetyl-D-glycero-a-D-ghw-heptopyranose ~

A

B

C

0.715 0.910 2.04 2.70 2.14 2.57 2.22 2.74 1.00 1.12 2.52 0.842 4.96 5.60

0.700 0.891 2.05 2.258 2.25 2.57 2.35 2.73 1.07 1.21 2.69 0.802 4.93 5.63

0.605 0.778 1.92 2.55 2.21 2.41 2.26 2.64 0.98'3 1.13 2.50 0.800 4.93 5.60

~~~~

4

Packinga A, B, and C are the same &B in Table V.

anomers were not well separated. The best results were obtained with acetates of nonreducing disaccharides and with 0-glycosylglycitol derivatives prepared by successive reduction and acetylation of reducing disaccharides. Fully acetylated methyl glycosides have been separated by gas-liquid chromatography on mixed liquid phases3' and on a thin-film column of fluoroalkyl silicone polymer.36 However, single anomers of diff went sugars were used in each of these separations, and selectivity for this class of derivatives was not demonstrated. A mixture of glucose, mannose, and xylose can be analyzed by gas-liquid chromatography of their acetylated glycosides.38 The sugars were converted to methyl glycosides by acidic methanol, and these products were then acetylated by pyridine and acetic anhydride. Gas-liquid chromatography of the reaction mixture showed one peak for each sugar, indicating that anomers were not resolved; the conditions of glycosidation probably precluded the presence of any furanosides in the mixture. The only other carbohydrate acetates that have been examined by gasliquid chromatography are the fully acetylated 2-amino-2-deoxy-~-glucose (38) H. W. Kircher, Tappi, 46, 143 (1962).

GAS-LIQUID CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

119

TAB^ VII

Retention Volumes of Fully Acetubled Disaccharide8 Relative to Octa-O-acetyl~ucrose~~ Column packing" Acetate of

a-D-xylopyranosyl a-D-xylopyranoside 5-O-&D-Xylopyranosyl-carabinitol a,a-Trehalose Sucrose 3-O-p-~-Mannopyranosyl-~-mannitol Maltitol Lactitol Cellobiito1 Melibiitol Gentiobiitol

A*

B=

0.21 0.47 0.88

0.34 0.62 1.22 1 .me 1 .oo 1.38 1.35 1.72 1.50 1.60

1.Wd 1.09 1.30

-

1.46 1.46 1.74

a Column packings: (A) a 1:2 w/w intimate mixture of (1) 1% w/w polyphenyl ether on glass beads and (2)1% w/w D-C silicone oil 710 on glass beads, GO-plus mesh; (B) 1 % w/w methylsilicone rubber gum SE-30 on glass beads, 60-plus mesh. * At 238". At 236'. d A t 12.5 min. At 4 mini

and 2-amino-2-deoxy-~-galactose derivative^.^^ Good resolution of anomers and of the acetylated reduction products of these two amino sugars waa obtained on a stacked column containing two liquid phases; the retention volumes are shown in Table VIII. However, it appears that caution should TABLE VIII Retention Volumes of Acetamido Derivatives Relative to Penla-O-acetyl-Larabitol~ 2-Acetamido-2-deoxy derivative of

Retention volume

1.82 6.92 7.93 7.95 8.12 8.14 -

Column (117 X 0.5 cm.) containing 40 cm. of 5% w/w 2,2-dimethyl-l,3-propanediol sebacate on Chromosorb W, 60-80 mesh, on top of a 74-cm. column of 1%w/w SE-30 methylsilicone polymer on glass beads, 60-plus mesh; 214'; &ionization detector, 0

(39) H.G. Jones, J. K. N. Jones, and M. B. Perry, Can. J . Chem., 40, 1559 (1962).

120

C. T. BISHOP

be observed in the interpretation of gas-liquid chromatograms of acetylated amino sugar derivatives because of possible decomposition of the products on the column. It has been shown that 2-acetamido-2-deoxy-~glucose tetraacetate (a or @ anomer) yields two volatile products on gasliquid chromatography, neither of which is the original c0mpound.4~One of these products was shown to contain an oxazoline ring and double bonds which were presumed to have been formed by elimination of the elements of acetic acid. The effect of the N-acetyl group on this elimination was not clear, but a similar degradation has been observed with the acetylated amino alditols, 2-acetamido-2-deoxy-~-glucitolpentaacetate and 2-acetamido-2-deoxy-~-galactitol pentaacetate; mixtures of these two compounds gave, on gas-liquid chromatography, highly variable, quantitative results that could be accounted for only by a random degradation of the products on the column.4' 3. Acetals

One of the earliest publicationsz6 on gas-liquid chromatography of carbohydrate derivatives showed that 1,2,3-tri-O-acety1-4,6-O-ethylidenea,@-D-glucopyranoseand 1,2:5,6-di-0-isopropylidene-a-~-glucof uranose are TABLEIX Retention Volumes of Carbohydrate Acetals Relative to 1 ,d :6,6-Di-O~sopropy~idene-cr-~-g~uw~urano~~ Acetal

Retention volumerp

l,2:4,5-Di-O-isopropylideneribitol SDeoxy-l,2-0-isopropylidene-~-arabinose 1,2-O-Isopropylidene-3-O-methyl-~-fructose Methyl 4,6-0-isopropyl~dene-2,3-di-0-methyl-~-~-glucopyranoside 1,3:2,4-Di-O-methyleneribitol 1,2-O-Isopropylidene-cr-n-glucofuranose 5,Gcarbonate Methyl 4,6-0-ethylidene-2,3-d-O-methyl-&D-mannopyranoside Methyl 2,3:4,6-di-O-isopropylidene-t~-~-mannopyran08ide 3-O-Acetyl-l,2 :4,5-di-O-bopropylidene-~-fructose 2,3-O-Isopropylidene-~rhamnose 1,2:4,5Di-O-isopropylidene-~-frucfose Methyl 4,6-0-ethylidene-2,3-di-O-methyl-~-~-galactopyranoside 1,2-O-Isopropy~idene-~-threo-pentdose 1,2:5,6-Di-O-isopropylidene-~-glucose 2,3 :4,5-Di-O-isopropylidene-~-fructoae 1,6-Anhydro-3,4-O-bopropylidene-/3-~-galtctopyr~nose

0.32 0.42 0.47 0.51 0.58 0.58 0.65 0.66

0.72 0.79 0.82 0.84 0.97 1.00 1.04 1.12

(40)C.T. Bishop, F. P. Cooper, and R.K.Murray, Can. J . Cheni., 41, 2'245 (1963). (41) M.B. Perry, Can. J . Bioclbem., 42, 451 (1984).

GAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

121

TABLE IX-Continu& Acetal

2,3:4,5-Di-O-isopropylidenegalactitol 1,2:4,5-Di-O-isopropy~idene-~-gdactitol 1,2:3,4:5,6-Tri-O-isopropylidene-~-mannitol 2,3:5,6-Di-O-isopropylidene-~-mannose 1,2:5,6-Di-O-isopropylidene-~-mannitol 1,2:4,5-L)i-O-kopropyhdene-~-mannitol bS-Ethyl-l,2-0-isopropylidene-5-thio-~-xylose 4,6-O-Ethylidene-l,2-0-isopropylidene-~-galactose 2,3:5,6-Di-O-isopropylidene-~-mannono-1,4-lactone 4,6-U-Ethylidene-2,3-O-isopropylidene-~-mannose 3,6-Di-O-acetyl1,2-0-isopropylidene-5-O-rnethyl-a-~-glucofuranose 5-S-Ethyl1,2-O-isopropylidene-5-thio-~arabinose Methyl 3,4-O-isopropylidene-2-0-(methylsulfonyl)-~-~arabinopyranoside 2,3:4,5-Di-O-isopropylidene-1,6-di-0-(methylsulfonyl)galactitol 1,2-O-Isopropylidene-~-fructose

3,4:5,6-Di-O-kopropylidene-~-glucose dimethyl acetal

Retatiun volumesn

1.13 1.16 1.20 1.20 1.40 1.40 1.63 1.67 1.81 1.86 1.91 2.00

2.24 2.27 2.70 3.13 4.05 5.40 5.82 6.54 6.60 6.64 7.46 8.88

2,3-O-Isopropylidene-~-ribono-l,4lactone 3-0-Benzyl1,2 :4,5-di-O-isopropylidene-~-fructose Methyl 4,6-0-benzylide11e-2,3-di-O-methyl-a-~-glucopyr~noside 1,4:4,8-l)i-O-isopropylidene-3-0-(methylsulfony1)-wfructose 1,4-O-Isopropylidene-D-ghcofuranose 1,2:3,4-L)i-O-isopropylidene-6-O-(methylsulfonyl)-~-galactose 4-O-Uenzyl-3,1 :5,6-di-O-isopropylidene-~-mannose dimethyl acetal 1,3:2,4-l)i-0-methylene-~-glucitol 2,3:5,6-1~ ~ ~ - ~ s o p r o p y ~ ~ d e r i e - 4 - ~ - p - t o ~ y ~ s u ~ f o n y ~ - ~ - g ~ u c o ~ e

dimethyl acetal 8.92 12.8 3,5-L)i-O-acety1-2,4-O-benzylidene-~-xylose dimethyl acetal 16.7 4,3:4,5-~~~-~sopropy~~dene-l-~-p-to~y~su~ony~-~-fructose 1,3,5-Tri-O-acetyl-2,4-0-bensylidenexylitol 17.8 Methyl 3,4-0-kopropyhdene-2-0-p-tolylsuIfonyl-p-~18.0 arabinopyranoside 1,2:5,G-l~~-O-~sopropyl~dene-~-p-to~y~su~fony~-~-g~ucose 25.6 Methyl 4,6-0-isopropylidene-2,3-d~-0-(methylsulfonyl)-a-~40.1 glucopyranoside 42.7 Methyl 2,3-di-0-benzoyl-4,6-O-~opropylidene-a-~-glucos~de 4,6-O-Ethylidene-l,2-O-~opropylidene-3-O-p-tolylsulfonyl-~54.0 galactose 3,4-O-Ethylidene-l,2-O-kopropylidene-6-O-p-toly~sulfony~-~56.8 galactose Column (117 X 0.5 cm.)containing 40 cm.of (1) a 1:l v/v intimate mixture of 20% w/w Apiezon M grease on Chromosorb W, 60-80 mesh, and 20% w/w 1,4butanediol succinate polyester on Chromosorb W, 60-80 mesh, on top of a 77-cm.column of (2) 1% w/w SE-30methylsilicone polymer on glass beads,60-plusmesh;206".

122

C.

T. BISHOP

sufficiently volatile t o pass through the columns. An attempt was made to resolve a mixture of the 3-methyl, 3-ethyl, and 3-vinyl ethers of 1,2:5,0di-0-isopropylidene-a-D-glucofuranose,but the three components were not separated. Separation of a mixture of 5-deoxy-l,2-0-isopropylidene-~xylo-hexofuranose and ~-deoxy-l,2-O-isopropylidene-a-~-glucofuranose 011 a preparative scale has been described.42 A large number of acetals have been examined by gas-liquid chromatography39; their retention volumes are recorded in Table IX. This list includes a number of derivatives in which the free hydroxyl groups are substituted with benzoyl, benzyl, carbonate, methylsulfonyl, methyl, and p-tolylsulfonyl groups. Although most of the retention volumes are undoubtedly those of the compounds listed, there was no evidence that the compounds were stable under the conditions used. It has been shown that acetals can undergo rearrangements during gas-liquid c h r ~ m a t o g r a p h y .Thus, ~ ~ a mixture of O-benzylideneglycerols, from the acid-catalyzed condensation of glycerol with benzaldehyde, showed four peaks on gas-liquid chromatography, corresponding to the cis and trans isomers of 1,2-and 1,3-0-benzylideneglycerol.When the four components were collected from the effluent gas and rechromato-

-

cis-l,2- 0 Bemylideneglycerol

H

trans-l,2-0Bemylidenegly cerol

A0p k0fl HO

OH

0

H

-

~is-1~3-0 Benzylideneglycerol

0

-

trans - 1 , 3 -0 Benzylideneglycerol

(42) E. J. Hedgley, 0. Meresz, W. C. Overend, and R. Rennie, Chem. Ind. (London),

938 (1960).

GAS-LIQUID CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

123

graphed, three of the single components each gave the same four peaks found in the original sample. The fourth component could be rechromatographed unchanged, and was identified as the stable cis-1,3-0-benzylideneand 1,2:5,6-diglycerol. Similarly, 1,2 :3,4-di-0-isopropylidene-~-mannitol 0-isopropylidene-D-mannitol gave single peaks but with similar retention volumes (see Table IX). However, collection of both products from the effluent gas stream gave mixtures having depressed melting points. The diacetates of each di-0-isopropylidene-D-mannitol showed single peaks with widely differing retention volumes, and the products were recovered unchanged from the effluent gas. The mixtures having depressed melting points, obtained by gas-liquid chromatography of the two di-o-isopropylidene-D-mannitols, were therefore acetylated and examined separately. The results showed that the 1,2:3,4-di-0-isopropylidene-~-mannitol yielded 2% of the 1,2:5,6-di-O-isopropylidene isomer and two unknown products (1.5% and 0.5%) on gas-liquid chromatography; 96% of the initial product was recovered unchanged. The 1,2:5,6-di-O-isopropylidencD-mannitol yielded 13% of the 1,2:3,4-di-O-isopropylidene isomer and the same two unknown products (3% and 8%) ; 76% of the initial product was recovered unchanged. It therefore appears that thermal rearrangement of acetals of carbohydrates can occur during gas-liquid chromatography, but that this can be prevented by prior substitution of all available hydroxyl groups by substituents that are stable under the conditions used. 4. (Trimethylsilyl) Ethers The preparation of a number of (trimethylsilyl) ethers of carbohydrates, and their behavior on gas-liquid chromatography were studied by Hedgley and O~erend.'~ The liquid phases used were silicone elastomer and Apiezon M on coarse Celite, 20-50% w/w. Separations were best at high temperatures ( >220°), and with flow rates of about 100 ml. of hydrogen/minute; Apiezon M gave the better separations, but its useful life was short at such high temperatures. Individual methyl tetra-0- (trimethylsilyl)hexosides gave single peaks on gas-liquid chromatography, but resolution of isomers was poor and anomeric pairs were not separated. Methyl tetra-0- (trimethylsilyl) -a-D-glucopyranoside was separated from methyl 4,6-0-benzylidene-2,bdi-0- (trimethylsilyl)-a-D-glucopyranoside and 1,2-O-jsopropylidene-2,3,6tri-0-(tnmethyls~lyl)-a-~-glucofuranose.Similarly, methyl tetra-0-( trimethylsilyl)hexopyranosides were separable from methyl tri-0(trimethylsilyl)pentopyranosides, the latter having lower retention volumes. The temperatures used for these separations seemed excessively high in view of the boiling points of most of the compounds examined, and (43) E. J. Hedgley and W. G. Overend, Chem. Znd. (London), 378 (1960).

124

C.

T. BISHOP

TABLEX Retention V o ~ u m of e 0-Trimethyleilyl Derivatives of D-Pentopyranoses Relative lo Trimthylsilyl8,S,~-~r~-~-(t~methylsilyl)-~-~-xylopyranosi~~~

Derivative of D-Ribose D-Arabinose D-Xylose D-Lyxose

a

B a

B a

B a

B

Trimelhylsilyl tri-0(trinwthylsilyl)pyranosidea

Methyl tri-0(trimethylsilyl)pyranosia!e4

0.48 0.52 0.49 0.41 0.73 1.oo 0.39 0.53

0.39 0.43 0.40 0.38 0.69 0.77 0.37 0.43

Column, 20% Apiezon L grease on fire brick; 180"; 45 ml. of helium/min.

this circumstance was probably the reason for the poor separations. However, the work was a major contribution in showing that (trimethylsilyl) ethers of carbohydrates are easily preparable, sufficiently volatile for gasliquid chromatography, and readily hydrolyzed. The last point was most significant, because it showed that compounds which are separated as their TABLE XI Retention Volumes of 0-(Trimethykilyl) Derivatives of Sugars and Plant G l y c ~ s i d e s ~ ~ Retention volumesa Derivative of a-D-Xylose a-D-Glucose B anomer D-glycero-D-gulo-Heptose Sucrose a-Maltose 0 anomer Raffinose Stachyose Esculin Phloridzin

14006 0.41 1.ood 1.55 2.39 -

8lO"b

860""

1.OOa

1.22 1.62 10.4 11.6 13.1 99.0

-

1.00' -

6.4 52.4 3.2 7.3

Column, 3% SE-52 on Chromosorb W. * Relative to fully (trimethylsily1)ated a-D-glucopyranose. Relative to fully (trimethyleiy1)ated sucrose. At 20 mine. a At 1.2 mim. f At 2.3 mine.

GAS-LIQUID CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

125

(trimethylsilyl) ethers can be regenerated unchanged by a very mild hydrolysis (50% aqueous methanol at reflux temperature for 1 to 2 hours), The separation of (trimethylsilyl) ethers was greatly improved by . ~ ~ retciition volumes operating the columns at a lower t e m ~ e r a t u r e The given in Table X show the resolution of the isomeric tetra-0-(trimethylsilyl)pentopyranosides and the methyl tri-0- (trimethylsilyl)pentopyranosides. Further improvements in the preparation of (trimethylsilyl) ethers of carbohydrates and in their separation by gas-liquid chromatography have been reported, and the method has been extended to oligosaccharides up to the tetramer and to complex gly~osides.~~ To prepare the (trimethylsilyl) ethers, the compounds were dissolved or suspended in pyridine and and chloroshaken for a few minutes with 1,1,1,3,3,3-hexamethyldisilazane trimethylsilane (as catalyst). The reaction mixtures were then injected directly into the chromatograph. Retention volumes of some of these derivatives are shown in Table XI. The results show that anomers are very well separated at lower temperatures and that anomers of disaccharides are resolved. The molecular weight of fully (trimethylsilyl)ated stachyose is 1676, and this appears to be the highest molecular weight yet recorded for a substance that has yielded to analysis by gas-liquid chromatography. By temperature-programming of the column from 125" to 250", separation of (trimethylsilyl) ethers ranging from erythrose through pentoses, hexoses, and heptoses, to di- and tri-saccharides was accomplished within 75 minutes.

5. Miscellaneous Unsubstituted methyl glycosides are considerably less volatile than many other derivatives in which some of the hydroxyl groups of the sugar ring are substituted by ester or ether groups, and their separation by gasliquid chromatography might be regarded as impracticable. However, the separation of the anomeric methyl threofuranosides at 205" on a liquid phase of ethylene glycol succinate polyester has been reported,46 and a large number of methyl glycosides have been resolved on columns containing mixed liquid Table XI1 shows the retention volumes of these compounds; they were separated without apparent decomposition or rearrangement. The methyl pentosides could be recovered unchanged from the effluent gas stream, but methyl D-glucosides and methyl D-galactosides appeared to be contaminated with their l16-anhydrides (0). It was considered that 1,ganhydride formation occurred in the heated part of the (44) R. J. Ferrier, Tetrahedron, 18, 1149 (1962). (45) R. Bentley, C. C. Sweeley, M. Makita, and W. W. Wells, Biochem. Biophys. Res. Commun., 11, 14 (1963); J . Am. Chem. Soc., 86, 2497 (1963). (46) J. N. Baxter and A. S. Perlin, Can. J . Chem., 38, 2217 (1960).

126

C. T. BISHOP

TABLEXI1 Retention Volumes of Methyl Glycosides Relative to Methyl @-Larabinopyranoside* Retention volumes Glycoside Methyl a-Lfucopyranoside Methyl kbarabinopyranoside Methyl a-L-rhamnopyranoside Methyl @-D-xylopyranoside Methyl @-D-ribofuranoside Methyl a-carabinofuranoside Methyl a-D-arabinopyranoside [ 1,(i-Anhydro-@-D-galactopyranose Methyl a-D-glucopyranoside Methyl (methyl @-D-glucopyranosid)uronate Methyl a-D-galactopyranoside Methyl a-D-altropyranoside Methyl 8-D-glucopyranoside Methyl a-D-mannofuranoside Methyl 8-D-ghcofuranoside Methyl a-D-mannopyranoside Methyl P-D-galactopyranoside Methyl D-glycero-8-D-gulo-heptopyranoside [Phenyl 8-D-glucopyranoside

Column 10 (2090)

0.77

1.o o c

1.06 1.20 1.70 1.73 2.20 2.30 5.50 5.60 5.70 6.40 7.11 7.40 8.30 8.52 31.8 39.0

Column 9 (210°)

0.79 1.00d

1.00 1.25 1.75 1.56 1 .a01 5.40 2.53 5.34 6.13 7.00 6.90 7.21 7.10 7.74 -

-1

0 Column 1: 40 cm. of 1:1 v/v mixture of 20% Apiezon M grease on Chromosorb W, 60-80 mesh, and 20% w/w 1,4-butanediol succinate polyester on Chromosorb W, 60-80 mesh; on top of a 77-cm. column of silicone rubber gum, SE-30, 1% w/w on glass beads, 60-plus mesh. * Column 2: a 1:1 v/v intimate mixture of (1) 20% w/w Apiezon M grease on Chrornosorb W, 60-80 mesh, with (2) 20% w/w 1,Cbutanediol succinate polyester on Chromosorb W, 60-80 mesh, and (3) 0.1% w/w Apieaon M grease on glass beads, 60-80 mesh, 0 At 2.5 mins. A t 3 mins.

collection device and not on the column, because 1,6anhydro-~-~-galactopyranose, methyl a-, and methyl P-D-galactopyranoside each gave only one peak, with different retention volumes on the chromatogram. The application of gas-liquid chromatography t d the analysis of nucleosides has been described," and the retention volumes of some nucleoside derivatives are given in Table XIII.Although the free nucleosides are not suf€iciently volatile, derivatives in which the hydroxyl or amino groups had been blocked by combinations of acetylation, methylation, or isopropylidene acetal formation could be eluted successfully. The peaks obtained were well defined and reproducible, although antisymmetrical. (47) H. T. Miles and H.

M.Falea, Anal. Chem., 34, 860 (1962).

127

GAS-LIQUID CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

TULE XI11 Retention Volumes of Nucleoside Deri~atives'~ Retention volumep Compound

200"

230°

3CMethyluridine 5.8 Uridine 2,3,btriacetate 30.4 3'-Methyluridine 2,3,5-triacetate 2,3-O-Isopropylideneuridine 6.5 2-Deoxyuridine 3,bdiacetate 3'-Met hylthy midine 4.9 Thymidine 3,bdiacetate 3'-Methylthymidine 3,bdiacetate 1-(~,~,4,6~etra~-acety~-~-~-g~ucopyranosy~)urac~~ 1-(2,3,4,6Tetra-0-acetyl-j3-~-glucopyranosy1)-3methyluracil 1-(2,3,4,6Tetra~-acetyl-~-~-glucopyranosyl)~methoxy2(1H)-pyrimidone 5.4 1-(2,3,4,6-Tetra-0-acetyl-/3-~-glucopyranosy1) thymine 1-(2,3,4,6Tetra&-acetyl-~-~-g~ucopyranosy~)-3methylthymine 28.7 Adenosine 2,3,5triacetate N-Acetyladenosine 2,3,S-triacetate 2,3-O-Isopropylideneadenosine 4.0 2-Deoxyadenosine 3,bdiacetate N-Acetyl-2-deoxyadenosine3,bdiacetate 1'-Methylinosine 2,3,btriacetate 26.3 6Methoxy-9-(2,3,5-tri-O-acetyl-&~-ribofuranosyl)purine

2.1 5.4 2.9 1.5 3.4 1.4 3.6 1.7 10.4

266'

6.7 9.1 8.5 5.3 5.7 15.7 1.o

2.6 8.7 60.0 4.8

a Column (6 ft. X 4 mm.) of fluorinated alkyl silicone polymer (QF-1), 1% w/w on Gas-Chrom P, 100-140 mesh.

Other carbohydrate derivatives that have been examined by gas-liquid chromatography include anhydro sugars of both the e p o ~ i d e ~ and ~ s ' ~3,6anhydro'O type, and isomeric methyl (ethy1thio)furanoside d i a e e t a t e ~ ~ ~ and their desulfurized derivatives.60

IV. STRUCTURE AND MOBILITY Since many isomeric carbohydrate derivatives can be resolved by gasliquid chromatography, it is apparent that separations depend on steric factors, as well as on differences in molecular weight or degree of substitution. However, generalizations about structure and mobility must be (48) C. D. Anderson, L. Goodman, and B. R. Baker, J . Am. Chem. Soc., 81,898 (1959). (49) H. Newman, Chem. Znd. (London), 372 (1963). (50) G. Casini and L. Goodman, J. Am. Chem. Soc., 86, 235 (1903).

128

C. T. BISHOP

made with caution, because the order in which isomers are eluted depends on both the liquid phase and the derivatives being used. Thus, a general relationship between structures and relative retention volumes of methyl glycosides could not be established, because it was found that, in some cases, the order of elution was reversed when the liquid phases were changed." The order of elution of the pentitols as their pentaacetatcs was reversed when the penta-0- (trimethylsilyl)pentitols were examined on the same liquid phase.44This observation was attributed to the large differences in polarities between the two substituent groupings. It therefore appears that relations between structure and mobility require comparisons within one class of derivatives, separated on liquid phases of similar polarity. For example, the order of elution of the methyl tri-0-methyl-Dxylosides is @-pyranoside,a-pyranoside, 8-furanoside, and a-furanosidel6' whereas, in the (trimethylsilyl) ether series, the @-D-pyranosidehas the largest retention volume and the furanosides appear to be eluted fir~t.4~ However, the latter separation was achieved on a nonpolar liquid phase (Apiezon L grease) , whereas the former was obtained on (polar) Carbowax 6000. Although these differences in order of elution between the methyl and (trimethylsilyl) ethers of D-xylose may be maintained when the same liquid phase is used for both, it is, as yet, difficult to be certain about this point. There is one generalization in the methyl 0-methylglycopyranoside series that appears to be valid; it involves the order in which anomeric glycosides are eluted. In all methyl 0-methylglycopyranosides examined and on all liquid phases, that anomer in which the glycosidic methoxyl group is in cis relationship to the methoxyl group at C-2 has the higher retention volume. When the G 2 hydroxyl group is unsubstituted, the order in which the two anomers are eluted is reversed.6 There are as yet insufficient data to show whether this generalization is also valid for methyl 0-methylglycofuranosides. There also appear to be systematic differences in the orders of elutioii for pyranoside and furanoside derivatives. For the fully methylatcd glycosides, furanosides precede pyranosides in the a r a b i n ~ s egalactose,2E ,~~~~~ and fructose28series; the order is reversed for these derivatives of xylose6' and The soundest generalization between structure and mobility on gasliquid chromatograms exists in the 0-acetylalditol series. Here it has been shown conclusively that isomers having the largest number of ester groups on the same side of the planar, zig-zag carbon chain have the largest retention volumes; if two isomers have the same number of ester groups on the same side of the molecule then the one with those groupings closest (51) C.

T.Biahop and F. P. Cooper, Con. J . Chem., 40, 224 (1962).

GAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

129

together will have the larger retention volume.34 Similar relationships in the 0-acetylpyranose series are not quite so clear, although attempts have been made to establish generalizations36 and to relate mobilities of the 0acetyl and 0-(trimethylsilyl) derivatives." From the retention volumes of the pentose tetraacetates, Ferrier3e has observed that replacement of an equatorial by an axial substituent in an all-equatorial conformation increases the relative retention volume by 0.09 when any ring carbon atom other than C-1 is involved. Conversely, the change from the equatorial to the axial orientation of the 1-0-acetyl group causes a decrease in relative retention volume of 0.15. From these values, it was possible to calculate predicted retention volumes based on the accepted chair conformations for the different pentoses. Tetra-0-acetyl-a-D-ribopyranosewas the only derivative for which the results disagreed, and some doubt that it has a chair conformation was indicated. Exactly opposite effects were observed in the (trimethylsilyl) tri-0- (trimethylsilyl) pentopyrano~ides4~;introduction of an axial group at any position other than C-1 causes decreased retention volume. However, flipping at C-1, from equatorial to axial, had the same effect in the 0-(trimethylsilyl) series as in the acetates, namely, a decrease in retention volume. It was considered that the effect (of flipping at C-1) on retention volumes was due to two factors: (A) that resulting from the axial orientation (equal to that of other axial substituents) , and (B) a modifying factor caused by proximity of the ring oxygen atom. In the acetatm, factor B opposes and is greater than factor A; in the 0-(trimethylsilyl) series, electronic effects would be expected to be less, and B again opposes, but is less than, A. The net effect of two axial groups in a pyranose ring was difficult to assess, and in the acetates, it depends on the anomeric orientation. With the 0-(trimethylsilyl) derivatives, two axial groups cause a greater decrease in retention volume than one. It is apparent that factors influencing the mobilities of carbohydrate ring compounds are extremely complex. Mobilities depend not only on steric and electronic differences in the compounds but also on the shape, polarity, and size of the molecules making up the liquid phase. It must also be considered that the energy difference between chair conformations is less than 1 Kcal./mole for many pyranose derivatives, and that changes in conformation may well occur under the conditions of gas-liquid chromatography.

V. APPLICATIONS 1. Structural Chemistry of Polysaccharides and Related Compounds

Methylation analysis has always been of the utmost importance in the structural investigation of polysaccharides, and it is still a necessary corn-

130

C. T. BISHOP

plement to methods of partial hydrolysis and glycol cleavage.saOne limiting factor in methylation analysis has been the problem of resolving the mixtures of methylated sugars obtained by hydrolysis of the methylated polysaccharide. The introduction of chromatographics3-66 and electrophoretics6 techniqucs relieved some of the difficulties inherent in the separation of such products by fractional distillation of their methyl glycosides. However, the isolation of pure components from complex mixtures of methylated sugars can sometimes be a long and difficult procedure requiring paper chromatography in several solvent systems and, sometimes, paper electrophoresis. Under such circumstances, the quantitative data necessary for precise interpretation may not be reliable. With the greater resolving power, faster operation, and reliable quantitative data provided by gas-liquid chromatography, it was anticipated4 that the method would be useful in this area of carbohydrate chemistry, and, indeed, that is where it has found its widest application. To test this application of gas-liquid chromatography, the methanolysis products of a number of well-characterized, methylated polysaccharides were examined.26Figures 9, 10, and 11 show the chromatograms obtained from the methanolysis products of methylated guar gum, methylated starch, and methylated dextran, respectively. The results, both qualitative and quantitative, were in good agreement with those obtained previously Methyl 2,3,4,6- TETRA-OMethyl- D Golactoside

-

0

M E T H YL Z.3,6 - T R I - 0 METHYL-D- MANNOSIDE

-

M E T H Y L 2,3-DI 0 METHYL-0-MANNOSIDE

I

I

I

I

I

1

FIG. 9.-Separation of Methanolysis Products from Methylated Guar Gum.**[Column (6 ft. x t in.) of fully methylated O-(2-hydroxyethyl)cellulose on Chromosorb (3:7 w/w);220'; 75 ml. of helium/min. Vertical dashes are at 10-min. intervals.]

(52) H. 0.Bouveng and B. Lindberg, Aduan. Carbohydrate Chem., 16, 53 (1960). (53) G.N. Kowkabany, Aduan. Carbohydrate Chem., 9,303 (1954). (54) W.W.Binkley, Aduan. Carbohydrate Chem., 10, 55 (1955). (55) L. Hough, Methods Biochem. Anal., 1, 205 (1954). (56) A. B.Foster, Advan. Carbohydrate Chem., 12, 81 (1957).

GAS-LIQUID CHROMATOOHAPHY OF CAF~BOHYDRATE DERIVATIVES

P

131

M E T H Y L Z,3,6T R I - 0 - M E T H Y LO-GLUCOSID E

M E T H Y L 2,3,4,6TETRA-O-METHYL0-GLUCOSI

DE

1 FIQ. 10.-Separation of Methanolysia Products from Methylated Starch.26 (Column and conditions as for Fig. 9. Vertical dashes are at 10-min. intervals.)

for these polysaccharides by other chromatographic methods, and they demonstrated the reliability of gas-liquid chromatography in this application. The method has been applied to original investigations of the structures

27 '

< ( . 0

X

;

Methyl 2,4-DI-OMethyl D Glucoside

- -

FIQ. 11.4eparation of Methanolysia Products from Methylated Dextran." (Column, aa for Fig. 9, 220"; 70 ml. of helium/min. Vertical dashes are at 10-min. intervals.)

132

C. T. BISHOP

of a glucomannan from Jack-pine w00dlK7a mannan from a pathogenic a galactoyeast Candida albicans,68an arabinogalactan from maple ,~Q from Trichomannan from Tm'chophyton g r a n ~ Z o ~ ~ am galactomannan phyton interdigitalelE0and a glucan from Microsporum quinclceanum.6' Separations obtained in three of these investigations are shown in Figures

4 11. Apiezon M , 2 0 % on Celite 5 4 5 Temp. 150

O C ,

Argon 140 ml./min.

/ FIQ.12.4eparation of Methanolysis Products from Methylated Mannan of Candidu aZbicana.'* (Key: (A) methyl 2,3,4,6-tetra-O-methyl-a-~-mannopyranoside; (B) unknown; (C) methyl 3,4,6-tri-O-methyl-a-~-mannopyranoside; (I)) methyl 2,3,6-tri-O-methyla-D-mannopyranoside; (E) methyl 3,4-di-O-methyl-a-~-mannopyranoside.) 12, 13, and 14; they give some idea of the sensitivity of the method. In each of the separations shown, there was at least one unidentified component whose presence wa.9 not detected by paper chromatography of the reducing sugars. Indeed, the 2,5-di-O-rnethyl-~arabinose in the products from the methylated arabinogalactan (see Fig. 13) may well have been missed on paper chromatograms because of the difficulty in separating it from 2,4,6(57) C. T. Bishop and F. P. Cooper, Can. J . Chem., 38, 793 (1960). (58) C. T. Bishop, F. Blank, and P. E. Gardner, Can. J . Chem.,38, 869 (1960). (59) C. T. Bishop, F. Blank, and M. Hrankavljevic-Jakovljevic, Can. J . Chem., 40, 1816 (1962). (60) M. B. Perry and F. Blank, Can. J . Chem., in press. (61) H. Alfes, C. T. Bishop, and F. Blank, Can. J . Chem., 41, 2621 (1963).

GAS-LIQUID

CHROMATOQRAPHY OF CARBOHYDRATE DERIVATIVES

133

tri-0-methyl-D-galactose by that method. Similarly, gas-liquid chromatography has revealed the presence of a 2,3-di-O-methylarabinose and a 2,4,6-tri-0-methylgalactose as hydrolysis products from methylated Acacia pycnantha gum and gum ghatti, re~pectively.~~ Paper chromatography had not revealed these compounds as cleavage products of the methylated gums. Most methods for detecting methyl ethers on paper chromatograms depend on reaction with the reducing form of the sugar, but nonreducing components that may be missed by that method are

TIME (mins.)

FIG. 13.-Separation of Methanolysie Products from Methylated Arabinogalactan of Maple Sap.*6 (Key: (1) methyl 2,3,5-tri-O-methyl-c,p+arabinofuranoside; (2) methyl 2,5-di-O-methyl-a,p-~-arabinofuranoside;(3,4,5, and 6) not positively identified; (7) (8) methyl 2,4,6-tri-O-methyl-a-~methyl 2,4,6tri-O-methyl-p~-galactopyranoside; galactopyranoside; (9) methyl 2,4di-O-methyl-~-~-galactopyranoside; (10) methyl 2,4di-0-methyl-a-D-galactopyranoside. Conditions: 20% Apiezon M on Celite 545 (60-80 mesh); 150"; 170 ml. of argon/min.)

detectable by gas-liquid chromatography. For example, the latter method has revealed nonreducing compounds that are artifacts produced during the hydrolysis of methylated polysaccharides by formic acid : the l16-anhydro2,3,4-tri-0-methyl-~-~-g~ucopyranose from methylated dextranZ6 (see Fig. l l ) , and a formate ester or anhydro sugar from methylated arabinogalactan26(see peak 6, Fig. 14). It may be expected, therefore, that gasliquid chromatography will reveal cleavage products whose presence had not been indicated previously. It has been found that many methylated sugars, apparently homogeneous from the results of paper chromatography,

134

C. T. BISHOP

show appreciable proportions of impurities when their methyl glycosides are examined by gas-liquid chromatography.6J2 Probably the most common cause of impurities in methylated sugars is in their separation by liquid-liquid chromatography following incomplete hydrolysis of a mixture of methyl glycosides; a methyl tri-0-methylhexoside and a tetra-o-methyl-

8 FT. COLUMN, 10% NEOPENTYL GLYCOL SUCCINATE ON GAS-CHROM 2 , 100-120 MESH. 60 ML. HELIUM/MIN. 225.C

4

1

2

7

18

IS

12

9

6

3

TIME (MINS.)

FIG. 14.Separation of Methanolysis Products from Methylated Glucan of Microsporum quinclceanum.61 (Key: (1) methyl 2,3,4,6-tetra-0-methyl-&n-glucopyranoside; (2) methyl 2,3,4,6-tetra-O-methyl-cu-~-glucopyranoside; (3) unknown; (4) methyl 1,3,4tri-O-methyl-8-D-glucopyranoside;(5) methyl 2,4,6-tri-O-methyl-~-~-glucopyranoside; (6) methyl 2,3,4-tri-0-methyl-~t-~-gluoopyranoside; (7) methyl 2,4,6-tri-O-methyl-a-nglucopyranoside; (8) methyl 2,4-di-0-methyl-&~-gluoopyranoside;and (9) methyl 2 , 4 di-0-methyl-a-D-glucopyranoside.)

hexose have approximately the same mobilities on paper chromatograms, but only the latter is detected by the spray reagents. Gas-liquid chromatography has provided supporting evidence for the identification of methylated sugars as cleavage products from the following, fully methylated materials : pectic substances from lucerne,e2barley-husk and rye-flour arabinoxylans, a galactoglucomannan and an arabinogalac(62) G. 0. Aspinall and R. 5.Fanshawe, J . Chern. Soc., 4215 (1961). (63) G . 0. Aspinall and K. M. Rosa, J . Chem. Soc., 1681 (1963).

GAS-LIQUID

CHROMATOQRAPW

OF CARBOHYDRATE DERIVATIVES

135

tan from Scots pine,64gum arabiclB6gum tragacanth,66*67the xylan from perennial rye grass,68brain cerebroside sulfuric ester and “ceramide dihexoside” of erythrocyte^,^^ and the main globoside of human erythroc y t e ~The . ~ ~technique has also been used to characterize, as their acetates, the following polyhydric alcohols as components in synthetic resins : ethylene glycol, propylene glycol, diethylene glycol, glycerol, 1,lJ-tris (hydroxymethyl) ethane, l,l,l-tris (hydroxymethyl)propane, pentaerythritol, D-mannitol, and D-glucitol.” However, in none of this ~ o r k ~ ~ - 6was ~ - 7advantage ’ taken of the unique powers of gas-liquid chromatography for quantitative analysis. As discussed in section II,3 (p. 103) , absolute quantitative results raquire calibration of the detector with standard compounds, but relative quantitative results can be very reliable, particularly if the components in the mixture have similar structures. Analysis of a mixture of methyl 0-methylglycosides or of acetates should, therefore, provide reliable quantitative data, and experience has shown this to be true. The quantitative analysis, by gas-liquid chromatography, of methylated sugars from methylated guar gum, starch, and dextran gave the same results as found previously by other methodsz5; and examination of the distribution of substituents in partially methylated cellulose by gas-liquid chromatography gave results in agreement with those obtained by paper and column chromat~graphy.~~ In original investigations of the structures of polysaccharide^,^^^^^-^^ the quantitative data obtained by gas-liquid chromatography of methyl 0-methylglycosides fitted the theoretical requirements (branch points accounted for by end groups) and agreed with the results of periodate oxidation. In one of these investigations,69the amounts of 0methylmannose and 0-methylgalactose derivatives, as determined by gasliquid chromatography, gave a ratio of galactose to mannose that agreed with the ratio of these two sugars in the original polysaccharide as determined by two other methods. The relative amounts of 0-methylglucoses irom a methylated glucan were estimated by both gas-liquid chromatography and preparative, paper chromatography, with excellent agreement in the results.61On the other hand, the ratio of di- to tri- to tetra-o-methyl(64) G. 0.Aspinall and T. M. Wood, J . Chem. Soc., 1686 (1963). (65) G . 0. Aspinall, A. J. Charlson, E. L. Hirst, and R. Young, J . Chem. SOC.,1696 (1963). (66) G. 0. Aspinall and J. Baillie, J . Chem. Soc., 1702 (1963). (67) G. 0. Aspinall and J. Baillie, J . Chem. Soc., 1714 (1963). (68) G. 0. Aspinall, I. M. Cairncrose, and K. M. Ross, J . Chem. SOC.,1721 (1963). (69) T. Yarnakawa, N . Kiso, S. Handa, A. Makita, and S. Yokoyama, J . Biochem. (Tokyo), 02, 226 (1962). (70) T. Yarnakawa, S.Yokoyama, and N . Kiso, J . Biochem. (Tokyo), 62, 228 (1962). (71) G. G. Esposito and M. H. Swann, Anal. Chem., 33, 1854 (1961). (72) W. B. Neely, J. Nott, and C. B. Roberta, Anal. Chem., 34, 1423 (1962).

136

C. T. BISHOP

hexoses obtained from a cellulose column disagreed with the analysis of the same mixture by gas-liquid chromatography of their methyl glycosides. However, the results from the cellulose column did not fit the theoretical requirements or the monosaccharide composition of the original polysaccharide and were judged to be in error, probably because of adsorption of di- and tri-0-methyl sugars.KB The foregoing evidence indicates that gasliquid chromatography can be used safely for the quantitative analysis of mixtures of methyl 0-methylglycosides derived from polysaccharides. The results were obtained with thermal conductivity and ionization detectors, both of which appeared to give a uniform response within the series examined, although no calibrations were performed. An ionization detector has been calibrated with methyl tetra-, tri-, and di-0-methyl-D-mannosides, to which it responded in a uniform mannere0;any variations in response caused by the differences in degree of substitution of the methyl O-methylmannosides must, therefore, have been within the limits of experimental error, probably f2%. In the analysis of a mixture of methyl 0-methylglycosides by gas-liquid chromatography, two precautions must be observed: (a) that no component is lost preferentially before analysis, and (b) that all components in the mixture are sufficiently volatile to pass through the columns and be detected. The first precaution requires that extreme care be taken in preparing methyl glycosides of the methylated sugars. Evaporation of any solution containing fully methylated methyl glycosides must be done at atmospheric pressure, usually by a stream of air, to prevent losses of these very volatile products. The second point can easily be checked by paper chromatography of a separate sample of the methylated sugars to see if any mono-0-methyl or unsubstituted sugars are present. An example of the latter has been reported, and the amount of mono-0-methylhexose in the mixture was determined by paper chromatography."' In combination with the method for rapid and complete methylation of carbohydrates developed by Kuhn,13 gas-liquid chromatography is very useful in structural investigations of oligosaccharides. It has been proved that no oxidation of the reducing group in mono- or oligo-saccharides occurs during methylation by the Kuhn procedure, so these compounds can be methylated directly,?' methanolyzed, and the methanolysis products examined by gas-liquid chromatography. A series of oligosaccharides from a glucomannan was examined in this way, and it was found that as little as 0.5 mg. of oligosaccharides could be used.14 Further demonstrations of the value of this application of gas-liquid chromatography were the examinations of oligosaccharides obtained from rye-flour and barley-husk xylans (73) R. Kuhn, H. Triachmann, and I. Low, Angew. Chem., 67, 32 (1955). (74) 0. Perila and C. T. Bishop, Can. J . Chem., 39, 815 (1961).

GAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

137

by successive periodate oxidation, reduction, and mild h y d r o l y ~ i s from ~~; a xylan of perennial rye grass by partial acid hydrolysis6"; and from partial hydrolysis of an arabinogalactan from tamarack ~ o o d . 7 ~ Another useful application of gas-liquid chromatography in structural investigations of polysaccharides is in the analysis of products from periodate oxidation. Applied to a polysaccharide or oligosaccharide, the sequence of reactions consisting of periodate oxidation, reduction, and hydrolysis yields products whose amounts and identities can give considerable inforniatiori about the glycosidic linkages in the original compound. 70,77 Depending on the linkages prcsent, the products obtained may be glycerol, erythritol, threitol, and reducing sugars, a mixture of which can be arialyzed by gas-liquid chromatography of their acetates. This application of gas-liquid chromatography has been used in the investigation of a glucomaiinan from Jack-pine woodb7and of a glucan from Microsporum quinckeanuns .61 The series of reactions consisting of periodate oxidation, reduction, methylation, and h y d r o l y s i ~ ~should ~ - ~ ~also yield products amenable to analysis by gas-liquid chromatography of their acetates or (trimethylsilyl) ethers, but this application has not yet been reported.

2. Analysis of Mixtures of Monosaccharides Gas-liquid chromatography has not been applied extensively to the aiialysis of mixtures of monosaccharides. To be successful, the method requires that a volatile derivative be preparable in quantitative yield from each monosaccharide, and that the mixture of these derivatives being analyzed can be rcsolved completely. Possible derivatives that fulfil these requirements are the fully methylated or acetylated methyl glycosides, acetylated monosaccharides or alditols, and (trimethylsilyl) ethers of methyl glycosidcs, glycoses, or alditols. The monosaccharide compositions of an arabinogalactan from maple sap and of a glucomannan from tamarack wood have been determined by successive hydrolysis, methanolysis, methylation, arid gas-liquid chromatography.26J'1 The former analysis, shown in Fig. 4 (see p. lOS), illustrates the main difficulty in working with products of methanolysis, that is, the formation of as many as four glycosides from each monosaccharide; this increases the riumbcr of compounds to be identi(75) 8.Haq and G. A. Adams, Can. J . Chem., 39, 1563 (1961). (76) G . Jayme, M. SBtre, and S. Maria, Nalurwissenschujlen, 29, 768 (1941). (77) M. Abdel-Akher, J. K. Hamilton, R. Montgomery, and F. Smith, J . Am. Chem. Soc., 74, 4970 (1952). (78) J. K. Hamilton, G. W. Huffman, and F. Smith, J . Am. Chem. SOC.,81,2173 (1959). (79) J. K. Hamilton, G. W. Huffman, and F. Smith, J . Am. Chem. SOC.,81,2176 (1959). (80) I. J. Goldstein, J. K. Hamilton, and F. Smith, J . Am. Chem. Soc., 81,6252 (1959). (81) P. Kooiman and G. A. Adams, Can. J . Chem., 39, 889 (1961).

138

C.

T. BISHOP

fied on the chromatogram. This problem may be circumvented by finding conditions of gas-liquid chromatography under which the derivatives from different monosaccharides are separated but anomers and the two ring forms from a single sugar are not resolved. Thus, a mixture of xylose, glucose, and mannose has been analyzed by successive methanolysis, acetylation, and gas-liquid chromatography; only onc peak was obtained from each sugar.38Similar problems arise in using the (trimethylsilyl) ethers of methyl glycosides or glycoses where, again, anomers of furanosidcs and pyranosides are formed.44It has been shown, however, that mixtures of 2-acetamido-2-deoxy-~-glucose and 2-acetamido-2-deoxy-~-galactose can be analyzed by (trimethylsilyl)ation, and gas-liquid ~hromatography~l; this has been the only derivative of amino sugars to give reliable quantitative results by this technique. The separation of acetylated glyc0sed~-~6 may offer better possibilities for the quantitative analysis of monosaccharides, because it is likely that one ring form will predominate and mixtures of anomeric acetates can be isomerized to a single anomer. However, separation of alditol derivatives offers the best possibility for analysis of a mixture of monosaccharides because there are no ring forms and, hence, no anomers. The sugars can be reduced by borohydride to give quantitative yields of the corresponding alditolsJmJ3which can then bc separated by gas-liquid chromatography of a suitable, volatile derivative. The acetates have been used for this and permit the analysis of a wide TABLE XIV Molar Ralios of Two Melhyl 0-Melhylglycosides, as Determined by Gas-Liquid Chromatography% In synthetic mixture A

60.7 50.6 33.8 20.3 0.9

Found from peak areas

B

A

B

39.3 49.4 66.2 79.7 99.1

60.5 50.0 33.1 20.6 1.3

39.5 50.0 66.9 79.4 98.7

Mean Deviation

=

10.5

0 A, Methyl 2,3,4-tri-0-methyl-~--~-xylopyranoside; B, methyl 2,3,4,6-tetra-Omethyl-a-D-glucopyranoside.

(82) M. Abdel-Akher, J.K. Hamilton, and F. Smith, J . Am. Chem. Soc., 73,4691 (1951). (83) C.T. Bishop, Can. J . Chem., 88, 1636 (1960). (84)J. A. Hauae, J. A. Hubioki, and G . G. Haeen, Anal. Chem.,34, 1567 (1962).

GAS-LIQUID

139

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

TABLE XV Molar Ratios of Four Monosacchari&s, as Determined by Gas-Liquid Chromatography of Their Derived Alditol Acetates33 Calculateda

A 100 50 0

Foundb

A

B

A

100 100 100 50

100 50 10 100

100 100 100 50

B 102

49.8 10.2 101

c

D

E

A

c

D

E

100 100

50 100

100 100

101 50.1

100 100

50.2 100

99.5 103

A, D-Xylose; B, D-mannose; C, erythritol (standard); D, Larabinose; E, D-galactose.

As the derived alditol acetates.

range of sugars with excellent quantitative results. However, D-glucitol and galactitol hexaacetates were not separated under the conditions tested. These two alditol acetates were well separated on another liquid phase.36 Two other derivatives, the methyl and (trimethylsilyl) ethers, offer good possibilities for the separation of alditols by gas-liquid chromatography, but conditions for the separation of isomers have not yet been adequately explored. Erythritol and mixtures of ribitol and D-mannitol have been analyzed quantitatively by gas-liquid chromatography of their (trimethylsilyl) etherss6; however, the general applicability of this procedure will depend on the resolution of most of the isomeric alditols derivable from monosaccharides, and this has not yet been reported. The accuracy of the quantitative results obtained by gas-liquid chromatography is illustrated by the data in Tables XIV and XV. The synthetic mixtures in Table XIV were prepared by mixing different volumes of standard solutions of the two methyl O-methylglyc~sides.~~ The data in Table XV were obtained by gas-liquid chromatography of alditol acetates derived from standard mixtures of glycoses by reduction and acetylation; erythritol was used as an internal standard.** Mixtures of mono-, di-, and tri-pentaerythritol have been analyzed as their acetates by temperature-programmed, gas-liquid chromatography.86 Good, quantita(85) B. Smith and 0. Carlsson, A d a Chem. Scand., 17, 455 (1963). (46) D. S. Wiersma, R. E. Hoyle, and H. Rempie, Anal. Chm., 84, 1533 (1962).

140

C. T. BISHOP

tive results required calibration of the thermal-conductivity detector (which varied in its response to each compound). This variation was probably caused by the variation in temperature during the analysis and by the rather large differences in molecular weights of the compounds.

3. Analysis of Reaction Products When reaction products are sufficiently volatile or can be easily converted into a volatile derivative, gas-liquid chromatography is a convenient criterion of purity or a method for analyzing mixtures. In the synthesis of 2-deoxy-~-~-erythro-pentofuranosides,this method was used to check the and purity of the starting product , methyl 2,3-anhydro-/3-~-ribofuranoside, to analyze mixtures containing methyl 2 ,5-di-O-acetyl-3-S-ethyl-3-thiofl-bxylof uranoside and met h y1 3 ,5-di-o-acet y l-2-S-et hy l-2-t hio-p-D-arabinofuran~side.~~ Gas-liquid Chromatography has also been used to analyze the products from the debenzylation of ethyl 2,3-di-O-benzyl5-O-methyl-a-~-arabinofuranoside, to follow the hydrolysis of ethyl 5-O-methyl-a-~-arabinofuranoside, and to check the purity of final products in the syntheses of 4-0-and 5-O-niethyl-~-arabinose.8738Analysis of methyl glycosides as their fully methylated or acetylated derivatives has provided a means for studying the kinetics of glycoside formation and the equilibrium compositions of glycosides in methanolic hydrogen ~ h l o r i d e . ~ ~ ~ ~ ~ Gas-liquid chromatography has also been used to study the complete methylation of reducing carbohydrates. Various monosaccharides were methylated by the KuhnT3procedure, and analysis of the products showed that fully methylated, anomeric methyl furanosides and pyranosides are formed in different proportions from each sugar.8eThe composition of a mixture of sucrose monoesters has been examined by methylation of the mixture and saponification of the ester groups; the partially methylated mixture was then subjected to methanolysis, and the methyl 0-methyl-Dglucosides and -D-fructosides were analyzed by gas-liquid chromatography.28A similar approach was adopted to determine the composition of sucrose monomyristate prepared by transesterification,80 and the rate and equilibrium constant of the transesterificatioii reaction were obtained by gas-liquid chromatographic analysis for methyl myristate, using octadecane as a quantitative, internal ~tandard.~' In a study of 0-isopropylidene acetals of Garabinitol and ribitol, gas-liquid chromatography was used to show the (87) (88) (89) (90) (91)

I. R. Siddiqui and C. T. Bishop, Can. J . Chem., 40, 233 (1962). I. R. Siddiqui, C. T. Bishop, and G. A. Adam, Can. J . Chem., 39, 1595 (1961). H. G. Walker, M. Gee, and R. M. McCready, J . Org. Chem., 27, 2100 (1962). R. U. Lemieux and A. G. McInnes, Can. J . Chem., 40, 2394 (1962). R. U. Lemieux and A. G. McInnes, Can. J . Chem., 40, 2376 (1962).

GAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

141

presence of two di-0-isopropylideneribitols and to identify alditols as their acetates.g2

VI. LIQUIDPHASES The most important factor in gas-liquid chromatography is undoubtedly the column packing, particularly the nature of the liquid phase. The liquid phase may be a solid at low temperatures, but it must be liquid and virtually nonvolatile at the temperatures used during the separation. Unfortunately, not enough is known about the physical chemistry of solution and adsorption to permit predictions as to the most suitable liquid phase for a given class of derivatives; the choice must, therefore, be intuitive and empirical. Liquid phases that have been used with success for separating carbohydrate derivatives are listed in Table XVI; most of these are available commercially. In some of the early work on this technique, a number of carbohydrate liquid phases were tested because it was felt that similarities in structure between liquid phase and the components to be separated might result in enhanced resolution. The carbohydrate derivatives tested as liquid phases were methylated 0-(2-hydroxyethyl) cellulose, methylated starch, methylated guar gum, methylated dextran, starch triacetate, 0(2-hydroxyethyl) cellulose triacetate,26D-mannitol, D-mannitol hexastearate, partially benzylated raffinose, and partially methylated cellulose29;only the first of these gave satisfactory results. The successful separations on the liquid phases listed in Table XVI indicate that the degree of polarity of the liquid phase is more important than its structure. The groups of liquid phases are arranged in order of increasing degree of polarity and, all other things being equal, the more-polar liquid phases give better separations. However, derivatives having more than one free hydroxyl group are severely retarded on the polar liquids and have lower retention volumes on the nonpolar hydrocarbon or silicone polymers, methylated 0-(2hydroxyethyl) cellulose, or poly (phenyl ether). The separation of compounds that have low volatility (either because of high molecular weight or the presence of a number of free hydroxyl groups) requires high operatingtemperatures which limit the choice of liquid phase. The hydrocarbons (Apiezon L) and silicone polymers (SE-30, SE-52) of high molecular weight are stable at high temperatures and have given useful separations of some of the less volatile carbohydrate derivatives. The polyesters are polar liquid phases that are highly selective for many carbohydrate derivatives ; considerable variation is available in these liquid phases, which can be formed from practically any glycol and any dicarboxylic acid. However, polyesters (92) M. A. Bukhari, A. B. Foster, J. Lehmann, J. M. Webber, and J. H. Westwood, J . Chem. Soc., 2291 (1963).

TABLEXVI Liguid Phases for Gas-Liquid Chromatography of Carbohydrate Derivatives

Maz. temp."

Liquid phase Hydrocarbon Polymers (1) Apiezon M

(2) Apiezon L Silicone Polymers (3) Silicone Oil (DC-710)

(degrees)

275

300

Derivatives separaiedb

methyl ethers acetates (trimethylsilyl) ethers acetates (trimethylsilyl) ethers

References 4, 29, 57, 58, 64, 65, 67, 68, 74 29 43,85 36944 44 d

anhydrogly cosides acetals disaccharide acetates [pentaerythritol acetates] (trimethylsilyl) ethers (trimethylsiyl) ethers acetates acetak and acetates of nucleosides

45 27, 35, 84 47

-

methyl ethers

25, 74

Aryl Polymers (8) Polyphenyl ether (5 rings)

250

methyl ethers

32, 64-68, 74, 87, 88

Polyesters (9) 1,4-Butanediol succinate

225

methyl ethers

26, 29-32, 57, 63-68, 74, 88,91 38

225

(4) Methylsiicone (SE-30)

400

(5) Methylphenylsiicone (SE-52) (6) Fluorinated alkyl silicone (QF-1)

400

Yethylated Polysaecharides (7) Methylated O-(Zhydroxyethyl)cellulose

225

acetates

48

92

37 [861

85

j

(10) 2,2-Dimethyl-1,3-propanediolsuccinate

225

(11) Diethylene glycol succinate

225

(12) Diethylene glycol adipate (13) 2,2-Dimethyl-1,3-propanediol (neopentyl glycol) sebacate

225

(14) Ethylene glycol isophthalate (15) 1,2-Propanediol adipate (16) Ethylene glycol succinate (17) Diethylene glycol-pentwrythritol adipate and phosphoric acid

225

Polyglycols (18) carbowax6oO0 (19) Carbowax 20M

Mixed liquid phases (20) Silicone grease (DC11), (l), and (9) (21) Silicone grease (DC 11) and (9) (22) (1) and (9)

(23) (3)and (8) (24) (4)and (13) (25) (11,(41,and (9)

methyl ethers methylated aldonolactones anhydrogly cosides methyl ethers anhydrogly cosides methyl ethers

28,61 40 40 28, 31 49 69, 70

methyl ethers (trimethylsilyl) ethers acetates metals (trimethylsilyl) ethers

60 41 27 42 45

-

methyl ethers

72

175 250

methyl ethen methyl ethers acetals acetates

27, 51, 59 59 40 71

acetates acetates acetates glycosides acetates aminodeoxyglycoseacetaka acetals

33 34 34 37 37 39 39

225 -

225

-

-

The liquid phase is unstable above this temperature. specified.

* These refer to derivatives of monosaccharides or their alditols; exceptions are

0

r

E td

3

U

? 2

U M

E 5 2

9m

r I&

w

144

C. T. BISHOP

are the least stable of the liquid phases listed, and tend to decompose and be eluted from the columns. This causes changes in the characteristics of the column through depletion of liquid phase, interferes with the collection of separated components from the effluent gas, and contaminates detectors. In addition, the possibility of transesterification of hydroxylated compounds that are being separated must be considered. The polyester liquid phases should, therefore, be used at temperatures some 20" lower than the maxima cited. The polyglycols (Carbowax 6000 and 20M) are also useful polar-liquid phases that give separations very similar to those obtained on polyesters. These polyglycols, particularly those of high molecular weight (Carbowax 20M) , are more stable than the polyesters and are therefore more versatile. Column packings generally contain 10-20y0 of the liquid phase evenly dispersed on an inert support having fairly uniform particle-size. However, retention volumes of the less volatile derivatives can be diminished by the thin-film technique, in which only 1-5% of the liquid phase is used. Examples of this are the separations of disaccharide acetates using thin-film columns of fluoroalkylsilicone polymer (QF-1) of methylsilicone polymer,37and of a mixture of poly(pheny1 ether) and silicone oil (DC-710) .57 Separations of derivatives that are sufficiently volatile to be analyzed on thick-film columns may be improved by using the thin-film technique; the components being separated spend less time in the liquid phase, and the chromatograph can be operated at a lower temperature, thus enhancing resolution. Thus, it was found that a thin-film (5%) of a polyester (2,2-dimethyl-l,3-propanediolsuccinate) gave improved separations of anomeric and isomeric methyl O-methylglycosides and made possible the chromatography of fully methylated di- and trisaccharides2*at temperatures below 250". The mixed, liquid phases afford considerable versatility by retaining the selectivity of polar phases and the diminished retention volumes given by the nonpolar phases. The solid support for the liquid phase should be chemically inert and stable at the operating temperatures, and should have a large surface area. The importance of uniform particle-size to facilitate even packing has already been emphasized (see Section II,l, p. 96). Materials in most common use as inert supports are diatomaceous earth and glass beads, both of which are often pretreated chemically to minimize adsorption effects. Preparations of diatomaceous earth are available commercially, in narrow ranges of particle size, pretreated with acid, alkali, or dichlorodimethylsilane. A coating of silver has been used for diminishing adsorption on the inert s ~ p p o r t ~it ~probably - ~ ~ ; enhances heat transfer in the columns, as well. (93)

E. C. Orrnerod and R.P. W. Scott, J . Chromalog., 2, 1 (1960).

GAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

145

VII. CONCLUSIONS The number and variety of applications discussed in Section V (p. 129) indicate the general utility of gas-liquid chromatography in carbohydrate chemistry. The greater resolving power and more rapid separations are distinct advantages of gas-liquid chromatography over other chromatographic techniques. Since separations are usually complete within 45 minutes and the samples required are small, the method is ideally suited to exploratory work, checking the purity of reaction products, or rate studies. An additional advantage is the detection of components by measurement of a physical property instead of by reaction of a functional group. It is, therefore, possible to analyze mixtures of nonreducing derivatives or of those which do not contain glycol groups, and none of the sample is sacrificed for detection. However, probably the most important advantage of gas-liquid chromatography over other analytical methods is that precise, quantitative data can be obtained. Reaction mixtures can be injected directly into the apparatus, so that no errors are incurred through loss of products by evaporation, filtration, or adsorption on a precipitate. Analysis by gas-liquid chromatography also avoids the errors inherent in paperchromatographic methods involving elution of components from the paper and their estimation by chemical or colorimetric methods. The overriding limitation of gas-liquid chromatography is the requirement of volatility and thermal stability of the compounds to be separated. Although no systematic survey of these limits has yet been made, it appears that any monosaccharide derivative containing more than two free hydroxyl groups may not be sufficiently volatile for practical application of gas-liquid chromatography. The volatility of such compounds can be increased by substitution of the hydroxyl function with methyl, acetyl, or (trimethylsilyl) groups. The necessity for thermal stability is more difficult to provide, and yet it is of extreme importance. Although much attention has been given to the search for easily prepared carbohydrate derivatives of sufficient volatility for gas-liquid chromatography, the stabilities of such derivatives under the conditions used have not been critically assessed. Column temperatures may range from 150" to 250" during gas-liquid chromatography of carbohydrate derivatives, and the injection site is usually maintained at a temperature some 60" higher than that of the column. Oxidative and hydrolytic processes are probably not operative, or are held to a minimum, in the atmosphere of inert gas (helium, argon, or nitrogen) used as the mobile phase. However, many carbohydrate derivatives are noted for their lability at high temperatures and for undergoing complex rearrangements under a variety of conditions; the possibility of the occurrence of modification during gas-liquid chromatography should

146

C. T. BISHOP

therefore, not be overlooked. It has been shown that the following reactions occur during gas-liquid chromatography of carbohydrate derivatives4: (a) deamidation of 0-methylaldonamides, (b) conversion of the sugar ring from pyranoside to furanoside in methyl 3,6-anhydro-O-methyl-~mannosides, (c) rearrangement of acetal groups of alditol acetals, and (d) degradative rearrangement of acetylated amino sugars, with production, at least in part, of an unsaturated derivative containing an oxazoline ring. Two of these modifications (a and b) can be anticipated, and two (b and c) can be prevented by complete protection of hydroxyl groups, so that rearrangement is impossible. However, the degradation of acetylated amino sugars appears to occur by elimination of the elements of acetic acid, with formation of a double bond. This may explain an earlier observation that acetates must be injected directly onto the column packing to prevent decomposition of the samples and that discontinuities in the column packings should be avoided if satisfactory results are to be ~btained.~' Obviously, it is essential that compounds be stable under the conditions used, if erroneous results are to be avoided. A simple test that will show whether or not any compound is suitable for gas-liquid chromatography is its recovery, unchanged, from the effluent gas-stream. This has been performed a number of times with methyl O-methylglycosides14~25~2E~28~29~s7-K9~E1 and the isolation of (trimethylsilyl) ethers of Dxylose from gas-liquid chromatograms indicates that these derivatives, also, are stable.g4The possibility that some products on a chromatogram arc artifacts arising from another component can be checked by collection and re-injection of each separated component in the mixture. The occurrence of partial decomposition of a compound can be detected by quantitative estimations of its mixtures with a compound known to be stable under the conditions used. Gas-liquid chromatography can be used safely to separate any derivatives that meet these checks satisfactorily. The results of gas-liquid chromatography may be presented as tabulated retention-times, relative to a standard, or by reproduction of the recorder tracing (the chromatogram). One disadvantage of the former method of presentation is the lack of information about the degree of resolution obtained; peak widths or overlapping of components are not indicated. On the other hand, reproduction of chromatograms is expensive and, regrettably, may be refused by some Journals. A discussion of the relative merits and disadvantages of these methods of presentation has been published.BKWhen a separation is secondary to the main theme of a project and is used merely as a criterion of purity or to distinguish between a (94) R.J. Ferrier and M.F. Singleton, Tetrahedron, 18, 1143 (1962). (96) E. von Rudloff, Can. J . Chem., SO, 1190 (1961).

GAS-LIQUID

CHROMATOGRAPHY OF CARBOHYDRATE DERIVATIVES

147

limited number of reaction products, the presentation of retention volumes is probably sufficient. However, in some problems, the separation and quantitative analysis of a number of components in a mixture is highly pertinent; for example, in the analysis of methanolysis products from methylated polysaccharides. In such cases, the reproduction of the chromatogram is desirable in order to indicate the degree of resolution, the accuracy of quantitative results, and the presence and significance of any unidentified components. Since gas-liquid chromatography introduces a new order of sensitivity into analytical methods for carbohydrates, the detection of components whose presence had not hitherto been indicated is to be expected. This has already been demonstrated by the identification of 2,3-di-O-methylarabinose and 2,4,6-tri-O-rnethylgalactoseas cleavage products from methylated Acacia pycnantha gum and methylated gum ghatti, re~pectively.~~ The introduction of any new analytical technique seems to inspire its indiscriminate application, without use of proper precautions, to a variety of problems, many of which could be solved more quickly by other methods. It should be emphasized that gas-liquid chromatography is not a panacea for all problems in the analysis of carbohydrate derivatives. Fast, efficient resolution of any given mixture of derivatives requires judicious selection from among the methods of countercurrent partition, liquid-liquid chromatography on columns or paper, adsorption chromatography on columns or thin layers, electrophoresis, and gas-liquid chromatography. The last method is, under the appropriate conditions, a powerful addition to the list because of its high efficiency and rapid operation, and the reliability of both the qualitative and the quantitative results.