Anal. Chem. 1990, 62,214-217
214
tional Bureau of Standards Mongraph 145; US. Government Printing Office: Washington, DC, 1975.
RECEIVED for review May 9,1989. Accepted October 24,1989. This study was performed as part of a doctoral dissertation by the senior author at the California Institute of Technology
(Caltech). Work was supported by the U.S. Department of Energy through BES-Material Sciences Contract W-31-109ENG-38 to Argonne National Laboratory and BES-Engineering and Geosciences Grant DE-FG03-88ER13851 to Caltech. Caltech Division of Geological and Planetary Sciences Contribution Number 4759 (675).
CORRESPONDENCE Reversing Enantioselectivity in Capillary Gas Chromatography with Polar and Nonpolar Cyclodextrin Derivative Phases Sir: In the 19809 a wide variety of chiral stationary phases (CSPs) were developed for the liquid chromatographic (LC) separation of enantiomers while relatively little was done in gas chromatography (GC). Some of these CSPs were based on naturally occurring chiral molecules such as protein (1-31, cellulose, and so on, while others were based on synthetic molecules ( 4 , 5 ) . One of the advantages of the synthetic approach in developing CSPs was that either enantiomeric modification could be used, thereby reversing retention order. This was useful for a number of reasons. For example, when determining optical purities in which one enantiomer is in excess, it is preferable to have the less concentrated isomer elute first. This is because the enantiomer in excess frequently produces a large, tailing peak. Often, the tailing can overlap with a smaller, late eluting peak. In addition, enantiomeric reversals can be useful in confirming separations and in mechanistic studies. Reversing enantioselectivity by using opposite configuration amino acid derivatives as GC stationary phases or different analyte derivatives was noted previously as well (6, 7). In this work, we report several new chiral GC stationary phases that consist of hydrophilic and hydrophobic derivatives of cyclodextrin. In addition to separating enantiomers that cannot be resolved by LC, many of these CSPs are interesting in that they show opposite enantioselectivity. Much of the early work on the use of native cyclodextrins as GC stationary phases was done by Smolkova-Keulemansova and co-workers (8-10) and Sybilska and associates (11). Because of their highly crystalline nature, native cyclodextrins (CD's) sometimes were difficult to use as stationary phases and generally were inefficient. However, this work demonstrated that CD's were highly selective, formed inclusion complexes with vaporized solutes and were worthy of continued study. Recently, Konig and co-workers produced lipophilic alkyl and alkyl-acyl derivatives of cyclodextrins that were liquids (12, 13). When coated on glass capillaries, a number of enantioselective GC separations were accomplished. Also, Schurig et al. have dissolved native and permethylated cyclodextrins in various GC stationary-phase liquids thereby obtaining viable CSP's (14). In an earlier work, we discussed requirements for obtaining liquid cyclodextrin derivatives as well as the use of nonpolar dialkyl CY-, 0-, and y-CD stationary phases on fused silica capillaries (15). Early kinetic work on cyclodextrin-catalyzed hydrolysis of racemic oxazolones indicated that a-CD and P-CD may have different enantioselectivities (16). However, these have never been confirmed or utilized in separations. Also, to our knowledge, there has never been a report showing that different derivatives of the same cyclodextrin show the opposite
enantioselectivity. In this communication, we show that the polar permethyl-O-((S)-2-hydroxypropyl)-CD often has the opposite enantioselectivity of the nonpolar alkyl derivatives. In a few cases, analogous CY-,P-, and y-cyclodextrin derivatives show opposite enantioselectivities. Consequently, it is no longer required that opposite molecular antipodes be "in hand" in order to have a generally useful chromatographic method in which enantioselectivities can be reversed. This approach is particularly useful for CSPs which utilize naturally occurring molecules that exist in a single enantiomeric form.
EXPERIMENTAL SECTION Fused silica capillary tubing (0.25 mm i.d.) was obtained from Alltech. (2,6-Di-O-alkyl)cyclodextrins were made as previously reported (17). The heptakis(2,6-di-O-pentyl)cyclodextrinswere made by reacting excess 1-bromopentanewith 3.0 g of the desired cyclodextrin in 30 mL of dimethyl sulfoxide (DMSO) at 50 "C for 2 h. The product was isolated by precipitation with water. The waxy precipitate was washed and dissolved in chloroform (CHCI,). The CHCl, solution was washed with water and the CHCl, was evaporated to leave the product, which was subsequently vacuum dried. This material then can be trifluoroacetylated by dissolving it in tetrahydrofuran (THF) with an excess of trifluoroacetic anhydride (TFA). The mixture is boiled for 2 h, then poured over ice to precipitate the product. The precipitate was washed with cold water and dissolved in CHC13. The CHCl, solution was extracted 3 times with 5% aqueous NaHCO, and 3 times with water. The CHCl, layer was dried with anhydrous Na2S04and evaporated. This viscous liquid was dried under vacuum overnight. Permethyl derivatives of 0-((S)-2-hydroxypropyl)cyclodextrin mixtures were made in two steps. First, the respective cyclodextrin was dissolved in aqueous NaOH (5% (w/w)) and the solution cooled in ice bath, then (8-propylene oxide was slowly added while stirring. After about 6 h in an ice bath the reaction was allowed to proceed for a day at room temperature, neutralized, and dialyzed briefly in order to remove the contaminating salts. The re-formed solution was filtered and the product obtained by freeze-drying. Permethylation was achieved by a reaction with methyl iodide after the dissolution of cyclodextrin derivative in a solution of NaH in DMSO (18, 19). Additional data on the make-up and properties of this compound are to be published subsequently. The capillaries were coated via the static method as previously reported (20). Amines and alcohols were derivatized with TFA or acetic anhydride. In each case, approximately 1.0 mg of the racemic analyte was dissolved in 0.5 mL of methylene chloride and 200 pL of the desired anhydride added. After reaction, dry N2 was bubbled through the solution to remove excess reagent. Sugars were trifluoroacetylated by the above procedure except that THF was used as the solvent. Also, because this reaction was somewhat
0003-2700/90/0362-0214$02.50/00 1990 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 62, NO. 2, JANUARY 15, 1990
215
Table I. GC Enantioselectivity Retention Data for Permethyl-(S)-hydroxypropyl (PMHP)Derivatized Cyclodextrin and Dipentyl-Derivatized Cyclodextrin structureb
racemic compound' 1,2,3,4-tetrahydro-l-naphthol
mandelic acid methyl ester
OH
1-cyclohexylethylamine
YNH2
a
elution order
stationary phase'
temp, "C
1.03 1.03
RS
S,R
PMHP-a-CD dipentyl-a-CD
100 100
1.04 1.01
RS
S,R
PMHP-a-CD dipentyl-8-CD
120
1.05 1.03
S,R RS
PMHP-P-CD dipentyl-a-CD
100 45
100
SP
1.09
RS
PMHP-CY-CD dipentyl-8-CD
150 150
2-amino-1-propanol
1.05 1.16
S,R
PMHP-8-CD DPTFA-P-CDd
120
RS
2-chloropropionic acid methyl ester
1.13 2.14
RS
erythrose
1.03 1.07
LJ'
arabinose
1.03
1-(1-naphthy1)ethylamine
@
HO
110
PMHP-$CD DPTFA-P-CD~
50
D,L
PMHP-&CD dipentyl-8-CD
80 80
1.04 1.20
D& LJJ
PMHP-0-CD dipentyl-P-CD
90
1.10
D,L
1.10
LJ'
PMHP-P-CD dipentyl-P-CD
90 90
1.08 1.07
DL
PMHP-8-CD dipentyl-P-CD
100 80
1.05 1.05
D,L
PMHP-8-CD dipentyl-p-CD
80 70
1.05 1.05
D,L
PMHP-@CD dipentyl-P-CD
80
L,D
PMHP-0-CD dipentyl-0-CD
100
S,R
60
70
OH
l-O-methyl-P-D,L-arabinopyranoside
&J
0 OCH,
HO
OH HWH2 0
ribose
Lh
6; @hH *w
xylose
HO
OH
lyxose
L,D
L,D
80
HO
sorbose
OH
1.12
LJ'
1.04
DL
90.
CH,OH
Sugars are shown in the a All compounds are trifluoroacetyl derivatives except 1-cyclohexylethylamine which was the acetyl derivative. D-configuration, the L-enantiomers have the opposite configuration at all stereogenic centers. 'All columns are 10 m long except for that dipentyl-P-CD column used to resolve mandelic acid methyl ester (which was 30 m). dThis stationary phase consists of trifluoroacetylated he~takis~2.6-di-0-~entvl)-B-cvclodextrin. ~~
~~
slower and the TFA was volatile, three additional aliquots of TFA were added at 7-min intervals. Both Hewlett-Packard (5710A) and Varian (3700) gas chromatographs were used for all separations. Split injection and flame ionization detection were utilized. The injection port temperature was 200 "C and N, was used as the carrier gas. RESULTS AND DISCUSSION Permethyl derivatives of 0-((S)-2-hydroxypropyl)cyclodextrin mixtures (PMHP-CD) are liquids at room temperature and can be used to coat undeactivated fused silica capillaries. They are nonvolatile and appear to be thermally stable at
temperatures up to 300 "C in the absence of oxygen. Also, these cyclodextrin derivatives have an affinity for water and seem to be much more polar than previously described alkyl-derivatized cyclodextrins (12,13). Many racemic solutes can be resolved on both PMHP-CD and alkyl-CD stationary phases. Most of these compounds (for which standards are available) have their enantiomeric elution order reversed on the two stationary phases. This is true for a variety of structural types of molecules including alcohols, amines, carboxylic acid esters, and sugars (Table I). Enantiomeric reversals also can occur between PMHP-CD stationary phases
216
ANALYTICAL CHEMISTRY, VOL. 62,NO. 2,JANUARY 15, 1990
Table 11. GC Enantioselective Separation Data for Analogous Derivatives of compound
separation factor, cy
1,2,3,4-tetrahydro-l-naphthol 1-(1-naphthy1)ethylamine
limonene oxide"
elution order
1.03
S,R
1.07
RS
1.06 1.09 1.06,1.20 1.02, 1.03
S,R RS
a-,8-, and
y-Cyclodextrin
stationary phase
column length, m
temp, "C
PMHP-(u-CD PMHP-P-CD dipentyl-P-CD dipentyl-y-CD PMHP-(u-CD PMHP-0-CD
10 20 10 10
100 120 150 150 90 100
+,-,-,+ -,+,-,+
10 20
This compound has two pairs of enantiomers. The reversal occurs only for the first pair of isomers.
1
-
1
I
2
-
TIME,
I
:-
1
!?
0.7
MI^
Flgure 1. Separation of o,L-arabinose on (A) a 10 M permethyl-0(S)-2-hydroxypropyC(3-cyclodextrin column (PMHP-(3-CD)and (B) a 10 M dipentyl-(3-cyclodextrin column (DP-(3-CD). Both separations were done at 80 O C with N, carrier gas. Note that the elution order of the D and L-enantiomers is reversed on these CSPs. The structure shown is of &-arabinose.
/
-
;-i
0.3
20
2.2
2.4
2.5
~ / T > K x103
I1
Plots of the log of the capacity factors ( k ' ) of ( R ) - and (S)-l-(1-naphthy1)ethylamineversus the reciprocal of the absolute Flgure 3.
temperature (K). No discontinuities or temperature-induced inversion of enantiomers was observed.
I
c
I
1
I
1
10
20
33
42
TIE, MlN
Enantiomeric separation of mandelic acid methyl ester on (A) a 10 M permethyl-O-(S)-2-hydroxypropyl-a-cyclodextrin column (PMHP-aCD) and (B) a 10 M dipentyC(3cyclodextrin column (DP-flCD). Both separations were done at 120 O C with N, carrier gas. Note that the elution order of the R - and S-enantiomers is reversed on these CSPs. This mixture was made so that the concentration of the R enantiomer was 50% more than the S-enantiomer. Flgure 2.
and trifluoroacetylated DP-CD stationary phase. See, for example, the data on 2-amino-1-propanol (Table I). Figures 1 and 2 show the chromatographically observed reversal in retention for D,L-arabinose and the methyl ester of mandelic acid. Although it appeared to be less common, enantioselective reversals also can occur among like-derivatized e-,p-, and y-cyclodextrins. Examples of this phenomenon are given in Table 11. Peak reversals have been observed between a- and (3-CD and between (3- and y-CD, but not between a-and yCD. Because the cyclodextrins are "size-selective", this is not surprising (21-23). According to the Gibbs-Helmholtz equation, it may be possible (in certain cases) to obtain a temperature-dependent inversion of the enantiomeric elution order. Indeed, there have been examples of this reported in the recent literature (24, 25). Retention versus temperature studies were done for many of the enantiomers separated in this work. A typical example for (R,S)-l-(1-naphthy1)ethylamine is shown in Figure 3. In
no case was a temperature-dependent inversion of the elution order observed for these solutes on these derivatized cyclodextrin stationary phases. As temperature is increased, the separation factor (a)decreases. Eventually a temperature is reached at which the enantiomers coelute. Further increases in temperature causes the retention of the coeluting isomers to decrease until they elute at the dead volume of the column. Currently, it is not known if this behavior is typical of all cyclodextrin-based GC stationary phases or if it is unique to the solutes and CD derivatives used in this study. In addition to being of great practical importance, this work raises several questions as to the nature of chiral recognition in these cyclodextrin derivatives. The definite size selectivity effects between the hydrophobic alkyl derivatives of a-, 0-, and y-CD seem to indicate that an inclusion complex is formed at these elevated temperatures. However, comparable size selectivities are not observed for the more polar PMHPcyclodextrins (15). The role played by the different substituents on the CD is most intriguing and difficult to explain. Undoubtedly there are a combination of factors (including polarity, steric bulk, orientation, degree of substitution, direct interactions, etc.) that affect chiral recognition. Currently, we are attempting to evaluate these by gas-phase calorimetry, computer modeling, and energy minimizaiton studies.
ACKNOWLEDGMENT The preparative help of Drs. C. T. Rao and Yan Xia is gratefully acknowledged. LITERATURE CITED (1) Hermansson, J. J . Chfomtogr. 1983, 269, 71-80. (2) Allenmark, S.;Bomgren. B.; Boren, H. J. Chromatogf. 1983, 269, 63-68. (3) Lindner, K. R.; Mannschrek, A. J . Chfomfogr. 1980, 193, 308-310. (4) Pirkle. W. H.; House, D. W.; Fin, J. M. J . Chfornatogr. 1980, 103, 143-1 58.
Anal. Chem. 1990, 62, 217-220 (5) Pirkle, W. H.; Pochapsky, T. C. J. Am. Chem. Soc. 1988, 7986, 708, 5627-28. (6) Feibush, B.; Oil-Av, E.; Tamari, T. perkin Tf8nS. 2 1972, 1197-1203. (7) Lochmuller, C. H.; Souter, R. W. J. Chromatogr. 1975, 88, 41-54. (8) Smolkova-Keulemansova, E. J. Chromatogr. 1882, 257, 17-34. (9) Smolkova-Keulemansova, E.; Kralova, H.; Krysl. S.; Feitl, L. J. Chrotn8tOgr. 1982, 247, 3-8. (10) Smolkova-Keulemansova, E.; Feltl, L.; Krysl, S. J. Inclusion Phenom. 1985. 3. 183-196. (11) Koscielskl, T.; Sybllska, D.; Jurczak, J. J. Chromatogr. 1883, 280, (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22)
13 1- 134. Konig, W. A.; Lutz, S.; Mischnick-Lubbecke, P.; Brassat, B.; Wenz, G. J. Chromatogr. 1888, 447, 193-197. Khlg, W. A.; Lutz, S.; Wenz, 0.; von der Bey, E. HRC CC,J . High Resolut . Chromatogr Chromatogr Commun 1988, 7 7 , 506. Schurig, V.; Nowotny, H.-P. J. Chromatogr. 1988, 447, 155-163. Armstrong. D. W. PMsburgh Conference Abshect Book; 1989, 001. Daffe, V.; Fastrez, J. J. am. Chem. SOC. 1880. 702, 3601-3607. Croff, A. P.; Bartsch, R. A. Tetrahedron 1983, 39, 1417-1474. Ptha, J.: Ptha, J. J. Ph8rm. Sci. 1985, 74, 987-990. Plha, J.; Rao, C. T.; Lindberg, B.; Seffers, P., submitted for publication in Carbohydr Res. Bouche, J.; Verzele, M. J. Gas Chromatogr. 1888, 8 , 501-505. Hinze. W. L. Sep. Purif. Methods 1981, 70, 159-237. Armstrong, D. W.: DeMond, W. J. chfOm8togf. Sci. 1884, 22, 411-415.
.
.
.
.
217
(23) Armstrong, D. W.; Ward, T. J.; Armstrong, R. D.; Beesley, T. E. Science 1988, 232,1132-1134. (24) Watabe, K.; Charles, R.; GII-Av, E. Angew. Chem. 1989, 707, 195-197. (25) Schurig, V.; Ossig, A.; Link, R. Angew. Chem. 1989, 701, 197-200.
Daniel W. Armstrong* Weiyong Li Department of Chemistry University of Missouri-Rolla Rolla, Missouri 65401
Josef Pitha National Institutes of Health NIA/GRC Baltimore, Maryland 21224 RECEIVED for review July 13,1989. Accepted October 30,1989. Support of this work by the Department of Energy, Office of Basic Sciences (DE FG02 88ER13819) is gratefuly acknowledged.
Effect of a Difference of the Column Saturation Capacities for the Two Components of a Mixture on the Relative Intensities of the Displacement and Tag-Along Effects in Nonlinear Chromatography Sir: The displacement and tag-along effects have been predicted by computations based on the use of the semiideal model of nonlinear chromatography (1). They are often observed in preparative applications of liquid chromatography, when the column is overloaded, and have been reported in many recent contributions (2-4). These effects are due to the fact that the velocity associated to a certain concentration of one of the components (5)depends also on the concentration of the other components locally present (6). The intensity of the displacement and the tag-along effects controls the shape of the profiles of the individual component bands of a mixture when these bands are not completely resolved. It is important to note that the displacement effect also controls the profile of an elution band after it has been separated from the bands of the compounds eluted after it. The profile of this band may never recover from the consequences of its interaction with the later eluted bands (6, 7). It is therefore important to understand what are the parameters which determine the intensity of these two effects and their relative intensity. The intensities of the displacement and of the tag-along effects depend essentially on the sample size, the composition of the feed and the parameters of the competitive equilibrium isotherm of the components involved. Most work carried out so far has been mainly concerned with the relative composition of the feed (1-4). The analytical solution of the ideal model has been derived in the case of a binary mixture, when the two components have competitive Langmuir equilibrium isotherms (6). This solution shows that the factor which controls the intensities of the displacement and the tag-along effects is not the mere relative composition of the feed or ratio of the concentrations of the two components (Co,2/Co,1)but is rather the ratio of the individual loading factors for the two components (Lf,Z/Lf,l= q s , l ~ O , P / q a , z ~ O , 1where , L,l and Lf,, are the individual loading factors of the two components, qs,land qs,2their column saturation capacities, and Co,land C0,, their concentrations in the feed). The loading factor of the column for a given compound is the ratio of the actual amount injected 0003-2700/90/0362-02 17$02.50/0
with the sample to the column saturation capacity for this compound (Lf,i= N i / ( l - c)SLqs,i,where Ni is the amount injected, in moles, qs,i is the column packing saturation capacity in mol/mL, t is the column packing porosity, and S and L are the column cross-section area and length, in cm2 and cm, respectively).
INTENSITY OF THE DISPLACEMENT EFFECT The intensity of the displacement effect can be measured by the ratio of the concentrationsof the first eluted component in the front (CIN) and the rear (CIM) sides of the second shock. If the second component does not displace the first one, there is no rear shock for the first component band. If the second component displaces strongly the first one, there will be an important rear shock for the first band. The ratio of the concentrations of the first component on both sides of the shock is given by eq 54 of ref 6
-C1,A'-
-1+-
C1,M
b2 4 r 1
In eq 1, bl and b2 are the second coefficients of the binary Langmuir isotherms of the two components and CY = a2/al, the ratio of their f i s t coefficients, is also the analytical relative retention. rl in eq 1 is the root of eq 22 of ref 6, which is in almost all cases nearly identical with Co,l/Co,z. The competitive Langmuir isotherms are written a:C: 6
qi = 1
,
+ blCl + bzC2
where i = 1, 2. With a Langmuir isotherm, the column saturation capacity, qs,i,is equal to the ratio ai/ bi. Introducing into eq 1 the value of rl = Co,l/Co,zand the relationships between the coefficients of the competitive Langmuir isotherm, a , and the column saturation capacities, we obtain
0 1990 American Chemical Society