Interpretation and Correlation of Bulkiness Chirality and Separation Coefficients in the Resolution of Diastereoisomers by Gas-Liquid Partition Chromatography Binyamin Feibush Department of Chemistry, The Weizmann Institute of Science, Rehovot, Israel

RECENTLY, WORK DONE in our laboratory ( I ) on the resolution of enantiomers of N-trifluoroacetyl(TFA)-( + ) -a-amino acid esters on a n active “ureide” phase (2) led to a new approach to the analysis of the interaction between solute and solvents, which could be applied also in the separation of diastereomers.

In the paper (I), the separation factor, a, of homologous series of N-TFA-(+)-a-amino acid esters on the asymmetric “ureide” phase were reported. The separation factors of different homologous series with a common structural variant, when described graphically, produced families of parallel lines. For a given family, of homologous series a certain structural change produces the same shift of the separation factor, thus the a values for all the members of a certain series could be calculated, if the value of a t least one of them and the character of its family are known. The separation factors of CF3CONHC*HRCOIR‘ depend on the relative size of the substituents of the asymmetric carbon (H, R , and COIR’). While the hydrogen is always the smallest sub- stituent, the relative size of R and COIR’ depends on the compound considered: when their effective size is equal, no separation is observed; when R is the biggest substituent, the D-enantiomers interact stronger with the active phase and when the C 0 2 R ’ moiety is the largest group the L-enantiomers emerge last from the gas chromatographic column.

A similar phenomenon is observed in the separation of diastereomers. Karger et a/. (3) studied systematically the influence of structural and polar variation on the separation factor, but did not point out that separation factors of similar homologous series produce a family of parallel lines when described graphically or as could be seen numer- ically in Table I (a rearranged form of Table I of Ref. 3) that a certain structural change produces the same increment on the separation factors.

For such representation, Table IV of Ref. 3 is not complete and some values should be refined, but also here the same phenomenon is observed. One important point, that was not mentioned by the authors, is that the order of emergence of the diastereomers from the GC column is not held constant.

Gil-Av et al. (4 ) studied the separation of a-acetoxypro- pionates of n-alkan-2-01 and proved that for this homologous series the LD (or DL) diastereomer has the longer retention time of the [(LD, DL)-(LL, DD)] pairs. Figure 1 describes the LD isomer, viewing the molecule along its long axis, from the alcohol to the lactate moiety. It is seen, that both asym- metric carbons have the substituents in clockwise arrange- ment going from the large (L) to the medium (M) to the small (S) group. Thus, the diastereomer emerging last from the column, LD, has the same direction by which the sub-

(1) B. Feibush, E. Gil-Av, and T. Tamari, Israel J . Chern., 8, 50

(2) B. Feibush and E. Gil-Av, J . Gas Chromatogr., 1967,257. (3) R. L. Stern, B. L. Karger, W. J. Keane, and H. C. Rose, J .

(4) E. Gil-Av, R. Charles-Sigler, G. Fischer, and D. Nurok, J .

(1970).

Chromatogr., 39, 17 (1969).

Gas Chromatogr., 1966, 51.

Table I.a Difference in Separation Factors of Two Series of Compounds of Methylalkylcarbinyl a-Halopropionates

and a-Halobutyrates X O H I 1 1 I

I I R-C-C-0-C-R ’

CH 3 H x = CI X = Br ~- ~

Et i-Pr 1-Bu Et i-Pr I-Bu

R = Me 1.022 1.039 1.065 1.034 1.064 1.090

R’ =

(1 ,043p R = Et 1.008 1.025 1.048 1 028 1.049 1.075 A 0.014 0.014 0.017 (0.015) 0.015 0.015

(I The separation factors, a, were retrieved from Table I of ref. 3. * For (Y = 1.034, the A = 0.006. The experimental error for

a -1.03, using efficient capillary columns, is small enough to per- mit us to assume that inversion of numbers occurred. Using 1.043 as the correct a-value produces also a certain difference between the 4th column and the other columns. i.e. (5th-4th) 1.064 - 1.043 = 0.021, 1.049 - 1.028 = 0.021.

Figure 1. m-isomer (L-lactic acid, ~-n-alkan-2-01)

stituents of both asymmetric carbons diminish in their size, while for the (DD, LL) form they are oppositely directed.

It is seen (3) that for separation of CH3C*HOAcCO~- C*HRR’ the effective size of the methyl equals that of the vinyl (-CH=CHI) group for R and R’, respectively, and that of the ethyl is iso-steric with the isopropenyl (-CCH,=CHz) group. As could be seen, compound I1 (in Figure 2) is sep- arated with a coefficient of only 1.007 as a result of diminishing degree of chirality as expressed by difference in bulk. On the other hand, I, where the methyl group is smaller than the iso- propenyl group, separates with a factor of 1.056, and 111, where the isopropyl is larger than the isopropenyl group, separates with a factor of 1.053. This apparent irregular behavior can be readily understood, if it is assumed that reversal of the order of emergence of the diastereomer occurs and the gradual increase of the size of the saturated alkyl group is followed by . - .

a gradual change of the separation factor, a (g), from

0.947 for I, 1.007 for 11, to 1.053 for 111. The diastereomer emerging last from the column has the same direction of bulkiness chirality (L + M -+ S) as does the a-acetoxy- propionyl moiety (see Figure 36), Le., S-S for I but S-R for 111.

1098 ANALYTICAL CHEMISTRY, VOL. 43, NO. 8, JULY 1971

H CH2-CH3 H CHICH312 7 FH3 CH3-C -C02-C-H CH3-C-CO2 -C-H CH3-C-C02-C-H

I I I I I I O A C C - C H ~ OAc C-CHz OAc C-CHp

I I I CH3 CH3 CH3

I I1 I11 Figure 2. l-(l-Alkyl-2-methylprop-2-ene) a-acetoxypro- pionates

H CH=CHz I I I I

For the compounds: CH8-C-C02-C-H , where

OAc R R = methyl, the two substituents of the alcoholic asym- metric carbon are iso-steric and no separation is observed. But when R = Et, n-Pr, and n-Bu, the alkyl group is bigger than the vinyl substituent, followed with separation which

( a 1 0 - 2 - Alkonol

R

in Fischer project ion

OH

CH3

S- enontiorner

clockwise ( L - - M - S )

OH

R CH3

does gradually increase with the enlargement of the alkyl chain (see Table IV of ref. 3). For these alcohols the S- enantiomer is of the anticlockwise (L -f M - S) handedness (see Figure 3c), thus the R-S diasteromer has the larger reten- tion volume.

Table I1 summarizes partially the resulls represented in Table IV of ref. 3, with the appropriate corrections. The following numbers are obtained by the difference between any two rows or columns of Table 11.

1st column - 2nd column - 3rd column 3rd column 1st row - 2nd row

(0) - (-40) = 40 0 - (-40) = 40 (0) - 43 = -43 43 - 5 38 41 - 5 = 36 0 - 41 = -41 75 - 38 = 37 54 - 14 = 40 (-40) - 5 = -45 1st row - 3rd row 1st row - 4th row 1st row - 5th row (0) - 41 0 - 46

S- enontiomer in R - S nomencioture

OH

H ( S ) I

CH3

S- enont lomer

= -47 0 - 54 = -54 (0) - 15 = -75 = -46 (-40) - 38 = -78 (-40) - 14 = -54

(MI CH3

c l o c k w i s e ( L - M + S )

OH

S- enant iomer

onticlockwise ( L - M d S I

OH H (SI

S- enontiomer ont lclock w i so (L-M-S 1

Figure 3. Space projection of the S-enantiomers of different alkanols

ANALYTICAL CHEMISTRY, VOL. 43, NO. 8, JULY 1971 1099

Table 11. Difference in Free Energy of Solvation, Ai&, for Diastereorneric Saturated and Unsaturated Alkylcarbinyl a- Acetox ypropionates

H H I I I I

Apo (CH3-C-COrC-R’) of solvation in cal/mola

OAc R R = Meb -CH=CH$c -C(CH3)=CHzc R’= Meb -CH=CHzC -C(CH+CH2c

R’ = Me (Old 0 -40 n-Bu 54 14 Et 43 41 5 i-Pr 1 5 38 n-Pr 41 46

a Condition specified in Table IV of ref. 3. Po (RS) - Po (SS). P o (SS) - Po (RS).

d The data were retrieved from Table IV of ref. 3, except for the sign of the -40 value.

As could be seen, there is almost a constant difference between any two rows or columns, i.e., a given structural change produces a constant difference in the free energy of solvation of the isomers.

It would be of great interest to see if such generalities hold also when polar substituents like: halogen, nitrogen deriva- 1971.

tives, etc, replace a n alkyl group and not only when the re- placement is done between two alkyl groups.

RECEIVED for review January 21, 1971. Accepted March 10,

Spectrophotometric Determination of Cyanide, Sulfide, and Sulfite with Mercuric Chloranilate

R a y E. Humphrey and Willie Hinze Department of Chemistry, Sam Houston State University, Huntsville, Texas 77340

THE APPLICATION of various metal chloranilates for the spec- trophotometric determination of a number of anions using either the ultraviolet maximum at 330 nm or the visible peak at 525 nm is well known ( 1 ) . In some instances, the insoluble metal salt of the anion is formed releasing the chloranilate ion while other reactions involve the formation of soluble, slightly dissociated metal compounds with the chloranilate ion going into solution. Probably the most common example of the latter is the determination of chloride with mercuric chlor- anilate which involves the formation of soluble, undissociated mercuric chloride (2 , 3). A visual comparative procedure has been used for the estimation of cyanide ion with mercuric chloranilate, in which soluble, slightly dissociated, mercuric cyanide is apparently formed, and for sulfide ion, where the very insoluble mercuric sulfide precipitates (4). The useful concentration range for the visible absorption method for these two anions was not indicated. Apparently, neither cyanide ion nor sulfide ion has been determined by the chlor- anilate method using the ultraviolet peak at 330 nm. The molar absorptivity of the chloranilate ion is much higher at this wavelength than in the visible region (3). We have found that sulfite ion can be determined by reaction with mercuric chloranilate, presumably to form the soluble, nondissociated

(1 ) L. S. Bark, Itid. Chem., 40 (3), 153 (1964). (2) J. E. Barney I1 and R. J. Bertolacini, ANAL. CHEM., 29, 1187

(3) R. J. Bertolacini and J. E. Barney 11, ibid., 30, 202 (1958). (4) E. Hoffman, Z . Anal. Chem., 185,372 (1962).

(1957).

mercuric sulfite as shown in Equation 1. The sensitivity for sulfite at both the UV and visible peaks is considerably higher than that for chloride ion.

HgCh + so32- 4 HgS03 + Ch2- (1)

Also, the determination of cyanide ion and sulfide ion with mercuric chloranilate employing the measurement of absorp- tion at 525 nm has been investigated and data on the sensitiv- ity of the visible absorption for these anions is reported.

EXPERIMENTAL

Apparatus. Absorption measurements were made with a Beckman DB-G spectrophotometer and with a Beckman DK-2A spectrophotometer.

Mercuric chloranilate was an Eastman Reagent chemical. For part of this work the compound was washed with ethanol several times and dried in order to remove any chloranilic acid present and hence lower the blank (5). Chloranilic acid was also a n Eastman Reagent Chemical and was used as purchased. The compounds KCN, Nd2S03, and Na2S. 9H20 were all Baker Analyzed reagent chemicals and were used as received. All other chemicals and solvents were used without purification.

Procedure. A mixed solvent, prepared by using equal volumes of ethanol and water, was employed in obtaining all of the Beer’s law and recovery data. The solubility of mercuric chloranilate was low in this solvent and hence the

( 5 ) C. F. Hammer and J. H. Craig, ANAL. CHEM., 42, 1588 (1970).

Reagents.

1100 ANALYTICAL CHEMISTRY, VOL. 43, NO. 8, JULY 1971