Studies on The Mechanism of Separation of Diastereoisomeric Esters by Gas-Liquid Chromatography Effect of Bulk Dissymmetry and Distance Between Optical Centers HERBERT C. ROSE, ROBERT L. STERN, and BARRY L. KARGER Department of Chemistry, Northeastern University, Boston, Mass.
b The relative volatilities and free energy differences in gas-liquid partition equilibria, A(AGo), of diastereoisomeric esters of acetylated lactic acid have been measured as a function of systematic variation in alcohol structure in order to gain insight into the factors which cause separation of these esters. Specifically, the effects of bulk dissymmetry at the alcoholic asymmetric carbon atom and distance between optical centers have been investigated. An interpretation based on steric factors is suggested as part of the mechanism of separation of these esters.
I
there has been considerable interest in the separation of diastereoisomers by gas chromatography. For example Weygand et al. ( 1 2 ) have examined volatile derivatives of dipeptides. Gil-A4v( 3 , 7 ) and his coworkers have studied in detail separations of diastereoisomeric acylated ahydroxy esters of secondary alcohols. Also Halpern and Westley ( 8 ) , among others , have separated diastereoisomeric esters of amino acids. Other representative papers can be found in the bibliography (4,10, 11). Xost of these investigations have not been concerned with elucidating the mechanism of the various separations. In one of the only investigations, Gault and Felkin (6) studied the mechanism of separation of erythroalcohols from threoalcohols. We have been examining the degree of separation of diastereoisomeric esters of acetylated lactic acid (a- acetoxypropionic acid) as a function of systematic variation in alcohol structure in order to gain insight into the factors responsible for separation of these esters. I n this paper we wish to report studies of the effect of bulk dissymmetry a t the alcoholic asymmetric carbon atom and of the effect of the distance between optical centers on separation. The differences of the diastereoisomers have been measured in terms of relative volatility and from this the free energy differences in gas-liquid partition equilibria have been determined. N RECENT YEARS
027 15
Quantitative Determination of Differences in Partition Behavior. Previ-
ous studies of diastereoisomers by gas-liquid chromatography have used the ratio of retention times from injection as the measure of separation. While this measure is certainly useful from a qualitative standpoint, the ratio does suffer in certain respects when quantitative differences between the diastereoisomers in their partition behavior are desired. I n the first place the ratio cannot be related to thermodynamic quantities such as free energy differences of gas-liquid partition equilibria. Secondly the ratio is dependent on column length, carrier gas velocity, and whether packed or capillary columns are used. X much better measure of diastereoisomeric differences on which to base mechanistic studies is the relative volatility, a ( 5 ) :
where t ~ and , tR2 are the retention times for the first and second components respectively, t, is the inert gas retention time, and K1 and K z are the respective gas-liquid partition coefficients. By subtraction of the retention of a nonsorbed gas from the retention times of the isomers, the column length and carrier gas velocity no longer affect diastereoisomeric differences, and intercomparisons of packed and capillary column behavior can be made. -41~0as seen in equation ( l ) , the isothermal determination of a allows one to compute the standard free energy differences between the diastereoisomeric pairs with respect to their gas-liquid partition equilibria. Thus : AGO = - R T In K
(2)
and AGz'
-
AGIO = A(AG") = - R T l n
a
(3)
I n the determination of a, care must be exercised to operate as closely to partition equilibrium as possible. There-
fore one must operate with heavily loaded columns so that adsorption effects on the solid support and the gasliquid interface are negligible. Also it is advisable to use column temperatures well below the boiling points of the diastereoisomers and sample sizes small enough that retention is concentration independent4.e. , on the linear portion of the isotherm. EXPERIMENTAL
A Barber-Coleman Model 5000 gas chromatograph with flame ionization detector was used for this work. The solid support was 80-100 mesh AW Chromosorb P, DNCS, obtained from F & ?tl Corp., and the two liquid phases were 1, 2, 3, tris-(Zcyanoethoxy) propane (F&M) and silicone oil D.C. 710 (Dow Corning Corp.). For each liquid phase a 10 ft. x 1/4-inch column with 20% liquid loading was made up in the conventional manner. For precise temperature measurement, an iron-constantan thermocouple was placed in the oven chamber. Temperature control was precise to 1 0 . 2 ' C. The retention time of methane was used as the inert gas time, t,. Since the capacity factors of the diastereoisomers were large, the slight differences in retention times of methane and air had a negligible effect on the determination of a. Ester Synthesis. The ionization constant for lactic acid a t 25' C. is 1.4 x and therefore self-catalytic esterification was employed in the synthesis of the diastereoisomers. I n a typical synthesis 100 mmoles of racemic alcohol, 110 mmoles of racemic lactic acid and ca. 50 ml. of benzene were refluxed for 10-12 hours. The theoretical amount of evolved water was collected with a Dean Stark trap. The cooled reaction mixture was washed with saturated aqueous sodium bicarbonate and then with water, followed by drying over anhydrous sodium sulfate. Benzene was removed a t atmospheric pressure, and then the hydroxyester was distilled on a semimicro Nester Faust spinning band column a t ca. 0.5 mm pressure. To obtain the acetylated hydroxyester, the hydroxyester was refluxed with an equimolar amount of acetic anhydride in dibutyl ether. The acetic acid and VOL. 36, NO. 3, MARCH 1966
469
cn3
0
H
II
I It
H
O
I - E-0-c-c-o c--R I I
-
No.
I o
R Et
I b I C
n-Pr n-8u
I d I c
i-Pr
Compound
t-Bu
Figure 1 . Diastereoisomeric esters of secondary alcohols studied by gas chromatography
Id ( R : i - P r )
Figure 2.
dibutyl ether were removed a t atmospheric pressure, and the acetylated hydroxyester was purified on the spinning band column a t ca. 0.5 mm. pressure. RESULTS AND DISCUSSION
Bulk Dissymmetry. The first series of diastereoisomers examined were compounds I.-I, shown in Figure 1. This study involves a systematic substitution a t the alcoholic asymmetric carbon atom of groups having increasing steric requirements. The results for this series of diastereoisomeric esters with both a relatively polar, 1, 2, 3, tris-(2-cyanoethoxy) propane, and nonpolar, silicone oil D.C. 710, liquid phase are shown in Tables I and 11. Comparisons of the data in Tables I and I1 reveals that for both columns, the A(AGO) values and thus the diastereoisomeric differences increase as R is systematically changed from an ethyl to a tert-butyl group. It should be
Ie(Rx t- Bu)
Chromatogramsof acetylated a-hydroxypropionate esters
further noted that for any individual diastereoisomeric pair, the free energy differences are greater on the polar column than on the nonpolar column. Gas chromatographic separation may be due to vapor pressure differences and/or differential solubilities of the solutes in the stationary liquid phase. Since it has been reported that the acetylated lactate ester of 2-butanol can be partially separated by fractional distillation ( I ) , it is probable that vapor pressure differences of the diastereoisomeric esters play a role in gas chromatography. However, if separation were due solely to vapor pressure differences, then the values of A(AG") would be independent of the stationary liquid phase. Comparison of the free energy differences of two columns of widely different polarity (Tables I and 11) reveals a significant increase in free energy differences for the polar stationary phase. Thus it seems clear that a t least on the polar column both
vapor pressure and solubility differences contribute to the observed separation. The small differences in a or in the A(AG") values have quite a marked effect on the degree of separation of the diastereoisomers. Figure 2 shows chromatograms of three pairs of diastereoisomers where R is ethyl, iso-propyl, and tert-butyl on the polar column. I t can be seen in this figure that while a increases only slightly from 1.057 to 1.107 (673, there is a large difference in the degree of separation. This result can be understood by examining the resolution equation, assuming equally spaced peaks ( 5 ) :
R
=
(G)("-)+ 1
kz
(4) where kP is the capacity factor for the final component and N Pis the number of theoretical plates for the final component. Since kz is large in all cases and since iVPwill be roughly the same for all compounds, the separation differences
(" :9 The relative volatility term has
depend almost entirely on the Table I, Separation Data for Diastereoisomeric Esters on 20% (2-cyanoethoxy) propane Column
Compound
t~~(rnin.)
I. Ib
53.18 68.75 95.45 69.95 75.42
1, Id
I* T
=
tR1
(min.)
50.39 64.65 88.00 64.30 68.25
1 , 2, 3 Tris' A(AG")
cal./mole
Q
1.059 1.065 1.086 1,089 1.107
- 44 - 50 - 65
- 69 - 80
125.0 f 0.2" C., inlet pressure = 28 p.s.i., outlet velocity = 2.40 cm./sec.
Table II.
Separation Data for Diastereoisomeric Esters on Oil Column
Compound
tRg
(min.)
tR1
(min.)
20% D.C. 710 Silicone A(AG")
Q
cal./mole
46.95 1.016 - 13 1.027 -20 72.04 115.71 1.047 -37 65.91 1.064 48 81.75 1.079 - 60 I. T = 125.0 & 0.2" C., outlet velocity = 2.32 cm./sec. inlet pressure = 28 p.s.i. 1, Ib O I Id
470
0
47.84 74.00 121.23 69.91 87.75
ANALYTICAL CHEMISTRY
-
(N2)lI2
-
term. its greatest effect on resolution when a is close to one which is the situation for these diastereoisomers. Thus small changes in a should greatly affect the degree of separation. Tables I and I1 further reveal that gas chromatography can be a most useful technique for measuring small free energy differences in molecules. Experimentally a was found to be reproducible to 1 0 . 2 % , which results in the A(AG") values having a precision of k 2 cal./ mole. I n most free energy difference measurements, one is taking differences in large numbers; however, in the gas chromatographic method the peak-topeak separation, which is easily measurable, can be directly related to the free energy differences. The method outlined is rapid, simple, and for the diastereoisomers, the thermodynamic data can be obtained on totally racemic material. Thus starting with racemic
0
H
II
1 I1
H
O
CH3 - C w O w C - C - 0 -
H
I-c
C- CHp-CH2-CH2-CH3
I
0
Separation
I
I
Yes
H
O
no CH3
0
H
CHf
H
O
- 0-CI -CI1- 0 -CHZ - H CI -CH
11 C H3- C
C 2-
2-
C H3
I11
no
I
I
C"3
Figure 3. Diastereoisomeric esters employed for the study of the effect of distance between optical centers
alcohol and racemic acid, two peaks representing two enantiomeric pairs, DDLL and DL-LD, are eluted from the gas chromatographic column. In Tables I and 11, CY has been determined for all compounds a t one temperature even though the molecular weights and thus the boiling points differ for the various diastereoisomers. I n gas chromatography CY is often found to be temperature dependent, frequently increasing as the two solutes are chromatographed further below their boiling points (9). However, in the case of these diastereoisomers, was found t o be temperature independent from 110'130" C. (It is not surprising that CY is temperature independent, for the A( A H ' ) values would be expected to be small.) The retention time of the last component of the lowest molecular weight diastereoisomer a t 110' C. was roughly equivalent to that of the highest molecular weight species a t 130' C. Thus differences in A(AG') values for these diastereoisomers a t 125' C. result from factors other than differences in molecular weights. (Y
Distance between Optical Centers.
X second structural investigation involved a systematic movement of the asymmetric centers away from each other one carbon a t a time, while keeping the molecular weight constant. For this study compounds IC, 11, and I11 (see Fig. 3) were synthesized and chromatographed. On a seiies of polar and nonpolar stationary liquid phases a t a variety of temperatures, it was found that compounds I1 and I11 could not be separated into the disastereoiqomeric pairs; however, in all cases compound IC produced qeparation (for example, see Tables I and 11). These results indicate that the distance between the two optical centers
must be an important variable in determining separation for these esters by gas chromatography. Some of the possible reasons for this result will be discussed in the next section on suggested mechanism. Suggested Mechanism. While the limited amount of data presented here makes a n unequivocal interpretation iniposoible, it is felt t h a t the results are suggestive of certain steric effects contributing t o separation. Examination of Fisher-Hirschfelder models reveals significant hindrance to free rotation about the carbon-oxygen single bond (bond X in Figure 4) with respect to the adjacent carbonyl group. Thus on a time average basis a conformation in which the methyl and hydrogen lie skewed to the carbonyl should be preferred as shown in Figure 4. While concentrations used in KMR are normally larger than those employed in a gas chromatographic column, it is still interesting to note that preferred conformations of the type in Figure 4 have also been suggested by Bowman et al. (2) in YAIR studies on i-propyl u-hydroxyesters and their derivatives. The energetic preference for the suggested conformation in Figure 4 should vary for the compounds studied (Figures 1 and 3). Thus moving the methyl group down the chain (away from the carbonyl group) results in increased conformational mobility of the groups attached to the asymmetric center. Also increasing the steric requirement of the R group in the esters of Figure 1 results in a decrease in conformational mobility due to an increase in the barrier to free rotation around the carbon-oxygen single bond. I n Tables I and I1 and in Figure 3, it can be seen that as conformational mobility around the carbon-oxygen
0
H
II
Figure 4. Preferred conformational distribution at the alcoholic asymmetric carbon atom of diastereoisomeric esters of secondary alcohols
single bond decreases, the chromatographic differences in the diastereoisomers, as reflected in the A(AG') values, increases. Thus the greater the fraction of time spent in the preferred conformation, the better the separation. As the methyl and hydrogen on the asynmetric alcoholic carbon atom become more fixed on a time average basis, the environment surrounding the lactate ester linkage becomes more asymmetric. The increased A(4GO) values on the polar column us. the nonpolar column, imply that polar regions on the ester molecule play an important role in determining diastereoisomeric differences. Gil-dv and Nurok ( 7 )have examined the effect of variations in acylating agent on relative retentions, and have found no change in the relative retention times for a series of a-alkanoyloxypropionates of 2-butanol. This result would tend to indicate that the ester linkage between the two asymmetric centers is the more important of the two in terms of differential interactions. As the asymmetry of the environment around this lactate ester linkage increaseq, it is expected that the A(AG') values qhould increase. Separation of racemates of organic compounds by gas chromatography has not yet been authenticated in the literature, and thus it is clear that the existence of two optical centers on the molecule is important for isomer separation. The above discussion has only considered one of the asymmetric centers on the ester molecule and therefore the hypothesis can only represent a first step in understanding the mechanism of separation. Many model compounds must be synthesized and their gas chromatographic behavior studied to check the hypothesis and examine the many questions that still remain unanswered. For example, what role do electronic effects play in these separations? How can the known absolute configurations (7) of the esters eluted from the column be related to the mechanism of separation? What part does the second asymmetric center have in separation? Experiments are in progress in our laboratory to answer these and other questions, and the results will be reported at a later date. VOL. 38, NO. 3, MARCH 1966
471
LITERATURE CITED
(1) Bailey, M. E., Hass, H. B., J . Am. Chem. SOC.63, 1969 (1941).
(2) Bowman, M. S., Rice, D. E., Switzer, B. R., Ibid., 87,4477 (1965). (3) R‘l G’l Gil-Avl Israel J . Chem. 1, 234 (1963). (4) Corey, E. J., Cassanova, J., Chem. and Ind. (London)1961, 1664.
(5) Dal Nogare, S., Juvet, R., “Gas-
Liquid Chromatography,” pp. 70-86, Interscience, New York, 1962. (6) Gault, Y., Felkin, J., Bull. SOC.Chim. France, 1965,742. (7) Gil-Av, E., Nurok, D., Proc. Chem. SOC.1962, 146. (8) Halpern, E., Westley, J. W., Chem. Commun. 12, 246 (1965). (9) Karger, B. L., Cooke, W. D., ANAL. CHEM.36, 985 (1964). (10) Pollock, G. E., Oyama, V. I., Johnson, R. D., J . Gas Chromatog. 5 , 174 (1965).
(11) Stern, R., Atkinson, E. R., Jennings, F. C., Chem. and Ind. (London) 1962,
1758. (12) Weygand, F., Rox, A., Schmidhammer, L., Konig, W., Angew. Chem. Int. Ed. Engl., 2, 183 (1963).
RECEIVED for review December 6, 1965. Accepted January 12, 1966. Presented at the 150th meeting of the ACS, Atlantic City, N. J., September 1965. Work supported by theBasic ResearchFund, Northwestern University.
Analysis of Alkylene Oxide Polymers by Nuclear Magnetic Resonance Spectrometry and by Gas-Liquid Chromatography ALAN MATHIAS and NORMAN MELLOR Imperial Chemical lndusfries Ltd. (Dyesfuffs Division), Hexagon House, Blackley, Manchester, The proportions of oxyethylene and oxypropylene in alkylene oxide polymers (polyethers) can be determined either by nuclear magnetic resonance spectrometry or by splitting the polyether with HBr and analyzing the bromination product by gas chromatography. The initiating polyol incorporated in such polyethers can also b e identified and determined by gasliquid chromatographic (GLC) examination of the bromination product under appropriate conditions; in this way, polyethers based on glycerol, 2,2di(hydroxy methyl)-1 -propanol (trirnethylol ethane), 2,2-di(hydroxy methyl)-1 -butanol (trimethylol propane), 1,2,6-hexane triol, pentaerythritol, sorbitol or mannitol, and triethanolamine have been analyzed.
E
THYLENE OXIDE/PROPYLENE
OXIDE
adducts of polyfunctional hydroxy compounds-polyethers-are used extensively as the polymeric alcohols for reacting with diisocyanates to make polyurethane foams. The physical properties of the foam depend to some extent on the chemical structure of these polyethers, and it is therefore important that a method of analysis be established for controlling their manufacture. Such a method would also be useful for elucidating the composition of unknown polyethers. Nadeau and Seumann (2) analyzed polyoxyethylene and polyoxypropylene compounds by pyrolysis a t 360’ C. and gas chromatographic analysis (GLC) of a sample of the gaseous product; Graham and Williams ( 1 ) split the polyethers with phosphoric acid to give acetaldehyde and propionaldehyde, which were determined colorimetrically. Nadeau and Waszeciak (5) reacted the 472
ANALYTICAL CHEMISTRY
polyethers with acetyl chloride, in the presence of ferric chloride, to give chloroethylacetate (from the oxyethylene groups) and a mixture of chloroisopropyl acetate and isopropyl diacetate (from oxypropylene groups) , and in this way determined the ratio oxyethylene/ oxypropylene in the range 15-50% oxyethylene. Analysis of the polyether sample for oxyethylene/oxypropylene ratio, without prior chemical conversion, can be carried out by nuclear magnetic resonance spectrometry (NMR), as detailed below : however, the apparatus is expensive and an alternative method, more suited to general laboratory practice, is desirable. Pyrolysis/GLC analysis is not readily controlled, so that reproducibility of the order required (say i 1 unit a t 20% oxyethylene content) is difficult to obtain. Chemical splitting of the ether linkages can be carried out in a number of ways, and the method of choice will obviously be that which gives complete breakdown of polyether t o give only one product from each different grouping in the polyether molecule. Such a method, involving splitting the polyether with HBr and GLC analysis of the products, is described below. DETERMINATION OF OXYETHYLENE CONTENT
The purpose of this work is to ascertain what proportion by weight of the polyether chain consists of oxyethylene and oxypropylene groups; for con-
r
9, England
venience, this has been expressed as the percentage by weight of oxyethylene calculated on the sum of oxyethylene and oxypropylene, and this has been
EO
+
abbreviated to %EO PO’ Determination by NMR Spectrometry. N M R spectrometry provides a quick and easy method for t h e analysis of oxyethylene/oxypropylene copolymers. Figure 1 shows the typical spectrum at 60 Mc./sec. obtained from such a sample examined as a 10% solution in carbon tetrachloride, with tetramethylsilane (TMS) as internal reference compound. dl1 the N M R results quoted in this paper were obtained using a Varian A60 spectrometer; the chemical shifts were expressed in 6 units-Le., p.p.m. downfield from the ThIS reference signal. The spectrum contains only two resonances, (9) the doublet centered at 1.08 6 due to the methyl groups of the oxypropylene units, and ( B ) a composite band from 3.2 to 3.8 6 due to the CHzO groups of the oxyethylene and oxypropylene units and also the CHO of the oxypropylene units. The resonance due to the hydroxyl group protons, which terminate the polyether, also occurs in band ( B ) . However, if a small amount of trifluoroacetic acid is added to the sample, exchange processes cause the hydroxyl resonance to move to low field. The composition of the copolymer is now obtained readily from the relative areas to bands (A) and ( B ) . Thus for a copolymer
