J. Org. Chem. Table 11. Hydrogen Atom Positional Parameters (X103)and Isotropic Displacement Parameters (v, x 10) atom X I A (u) YIB (6) ZIC ( u ) Ui, 114 (2) 179 (2) -18 (2) 35 (3) -71 (2) 64 (2) 172 (2) 197 (2) 355 (2) 245 (2) 331 (2) 219 (2) 464 (2) 611 (2) 509 (2) 396 (2) 493 (2) 528 (2) 610 (2) 675 (2) 576 (2) 650 (2) 511 (2) 856 (3) 845 (3) 785 (3) 202 (2) 292 (2) 403 (2) 347 (3)
784 (2)
861 (2) 899 (2) 905 (2) 665 (2) 568 (1) 624 (1) 529 (2) 639 (1) 674 (2) 742 (2) 785 (1) 703 (2) 683 (2) 611 (2) 539 (1) 505 (2) 417 (2) 403 (2) 554 (2) 669 (2) 714 (2) 748 (2) 574 (2) 583 (2) 656 (2) 423 (2) 487 (2) 363 (2) 394 (2)
647 (3) 712 (3) 706 (3) 858 (3) 903 (2) 959 (2) 1038 (3) 797 (2) 915 (2) 680 (2) 728 (2) 907 (2) 588 (3) 574 (2) 532 (3) 689 (2) 929 (2) 700 (3) 816 (3) 816 (2) 915 (2) 789 (3) 812 (3) 707 (4) 548 (3) 618 (3) 946 (3) 1036 (3) 934 (3) 804 (3)
45 45 47 47 42 38 38
36 30 32 32 34 42 42 42 31
35 42 42 38 41 41 41 65 65 65 42 42 45 45
parameters for all non-hydrogen atoms were refined by full-matrix least-squares analysis using the reflections for which the observed intensity was greater than twice the corresponding standard deviation. The hydrogen atom positions were located in electron
1987,52, 2273-2276
2273
density difference maps, and they were refined isotropically. Weights used were Using observed data, the final reliability indexes werer R = 4.8, R, = 7.7%, S = 2, and (6/u)av = 0.10. Atomic scattering factors were taken from the I n t e r n a t i o n a l Tables f o r X-ray C r y ~ t a l l o g r a p h yfor ~ ~ neutral atoms and anomalous-dispersion corrections for non-hydrogen atoms were taken from Cromer and Liberman.33 All calculations except for MULTAN were performed with XRAY 76.34 The final coordinates and isotropic atomic displacement for all the atoms are given in Tables I and 11. Tables of bond lengths and angles and molecular packing patterns are available from W.L.D. upon request.
Acknowledgment. This work was supported, in part, by contract N01-HD-1-2810 from the National Institute of Child Health and Human Development, by Training Grant GM-555 from the National Institutes of Health General Medical Sciences, N.I.H., and by Research Grant 26546 from the National Institute of Arthritis, Diabetes, and Digestive and Kidney Diseases, N.I.H.
(31) Stout, G. H.; Jensen, L. H. X-Ray Structure Determination; Macmillan: New York, 1968; p 457. (32) International Tables for X-ray Crystallography; Kynoch: Birmingham, 1974; Vol. IV., pp 71-147. (33) Cromer, D.T.; Liberman, D.J. Chem. Phosphorus 1970, 53, 1891-1898. (34) Stewart, J. M., “The XRAY System”,Technical Report TR-446, 1976; Computer Science Center, University of Maryland, College Park, MD. (35) Dum, W. L.; Rohrer, D. C.; Strong, P. D.Acta Crystallogr., Sect. A 1971, 368. (36) Kirk, D.N.; Hartshorn, M. P. Steroid Reaction Mechanisms; Elsevier: New York, 1968; p 391.
Determination of Enantiomeric Purity of Polar Substrates with Chiral Lanthanide NMR Shift Reagents in Polar Solvents’ Linda M. Sweeting* Department of Chemistry, Towson S t a t e University, Baltimore, Maryland 21204
Debbie C. Crans and George M. Whitesides* ‘
Department of Chemistry, Harvard University, Cambridge, Massachusetts 02138 Received September 17,1986
The proton NMR spectra of 19 1,2- and 1,3-dioxygenatedcompounds were studied in deuteriated acetonitrile, acetone, and chloroform in the presence of chiral lanthanide NMR shift reagents tris(3-((heptafluoropropyl)hydroxymethylene)-d-camphorato)europium(III) (1) and tris(((trifluoromethyl)hydroxymethylene)-d-camphorato)europium(III) (2). Enantiotopic OH, CH, and CH, NMR resonances were best resolved in acetonitrile; line broadening obscured the scalar coupling. Enantiomeric excesses as high as 98% can be determined for 16 of these compounds in this polar solvent.
Chiral synthons are often smaller polar molecules, and it is useful to have a simple method of determining their optical purity. Methods based on chiral lanthanide shift reagents often fail with these substances2for several reasons. They may be contaminated with water and sparingly (1) Supported by the National Science Foundation, Grants CHE 85-08702 and NIH GM 30367. (2) For references to non lanthanide based methods, see: Luchinat, C.; Roelans, S. J. Am. Chem. SOC.1986, 108,4873-4878.
0022-3263/87/1952-2273$01.50/0
soluble in the nonpolar solvents normally used with shift reagents (CDC13,CD2C12).In these solvents, their strong binding to the lanthanide ion (often a reflection of chelation by multiple polar functions) gives broad resonances and, as a consequence, poor resolution between enantiotopic resonances in the NMR ~ p e c t r u m . ~We have re(3) Sullivan, G. R. Top. Stereochem. 1979,10,287-329. Fraser, F R . Asymmetric Synth. 1983, 1, 173-196. McCreary, M. D.; Lewis, D. W.; Wernick, D. L.; Whitesides, G. M. J.Am. Chem. SOC.1974,96,1038-1054.
0 1987 American Chemical Society
2274 J . Org. Chem., Vol. 52, No. 11, 1987
Sweeting et al.
OH
j
CH
OH
m
Figure 2. Proton NMR spectra of 0.2 M butane-1,2,4-triolin the presence of 0.2 M shift reagent 1 in acetonitrile-d,: “anhydrous”(upper) and in the presence of D 2 0 (lower;6 molar equiv of D20/mol of substrate added).
- , ,
P
, . . , ,
, . .
SIPPml
Figure 1. Proton NMR spectra of 0.2 M 3-(methylthio)propane-1,2-diolin the presence of 0.2 M shift reagenk 1 and 2 in CDCl,, CD,CN, and (CDJ2CO. cently reported that the enantiomeric excess of one representative molecule of this type, 3-chloropropane-1,2-diol, could be determined by the nonequivdence of enantiotopic OH resonances by using tris(3-((heptafluoropropyl)hydroxymeth1ene)-d-camphorato)europium(III) (1) in the polar, coordinating solvent acetonitrile? This observation was interesting both for its suggestion of broad utility for chiral shift reagents in polar solvents and for the spectroscopic observation of enantiotopic hydroxyl groups. We have examined the generality of this observation and find that a wide variety of 1,2- and 1,3-dioxygenatedcompounds exhibit well-resolved enantiotopic protons (OH, CH, and C H 3 ) in the presence of 1 or tris(((trifluoromethy1)hydroxymethylene)-d-camphorato)europium(III) (2) in acetonitrile or acetone. Acetonitrile consistently permits resolution of enantiotopic protons of 1,2-and l,&diols more completely than either chloroform or acetone; the lesa polar carbonyl compounds are resolved equally well in chloroform.
Results and Discussion Sensitivity of Enantiotopic Shifts to Structure. We summarize the spectral results obtained with a number of polar substances with chiral europium shift reagents in acetonitrile and acetone in Table I and compare those results with similar experiments conducted with chloroform as a solvent. Figures 1-3 give representative spectra. (4) Cram, D. C.;Whitesides, G. M. J. Am. Chem. SOC.1985, 107, 7019-7927.
_ _ ~
-~-----‘ ,
20
Sippm)
,
,
-0
Figure 3. Proton NMR spectra of 0.2 M propane-1,2-diolin the presence of 0.2 M shift reagent 2 in acetonitrile-d,. (a) R isomer, prepared from glucose by fermentation using Clostridium thermosaccharolyticum (ATCC 31960). (b) R isomer, prepared by reduction of hydroxyacetone by fermenting Saccromyces cereuisae. This sample has traces of impurities. (c) A mixture of 90% R from b and 10% racemic, prepared by volume with use of micropipettes. (d) Racemic. The small differences in chemical shift for the enantiomers between samples is caused primarily by small differences in concentration. A shift-reagent peak overlaps the CH,signal in each case. Table I summarizes qualitatively the ease of discrimination of enantiotopic protons. We estimate that if the valley between two enantiotopic resonances in the racemic mixture is less than 5%,an enantiomeric excess of 98% or less can be determined; such “base-line” resolution for at least one pair of enantiotopic resonances is indicated by “++”. “Useful” resolution corresponds to a other 1,2-dioxygen-containingspecies 2 1,3-dioxygen-containing species. The poor results for 3-methoxypropane-1,2-diol, excellent results for glycidol, and the similarity of bu( 5 ) Jakovac, I. J.; Jones, J. B. J. Org. Chem. 1979, 44, 2165-2168. Johnson, P. Y.; Jacobs, I.; Kerkman, D. J. J. Org. Chem. 1975, 40, 2710-2720. (6) Monofunctional substrates form both 1:l and 2:l complexes; it appears that the dioxygenated compounds occupy two sites on the octacoordinate europium(II1). (7) Sweeting, L. M.; Anet, F. A. L. Org. Magn. Reson. 1984, 22, 539-452. (8) Carbohydrate derivatives examined in acetone included 1,2-O-isopropylidine-a-D-ghcofuranose, 1,2:5,6-di-O-isopropylidene-a-~-allofuranose,and 4,6-O-benzylidene-l-0-methyl-~-~-glucopyranose, which are insufficiently soluble in acetonitrile to give useful spectra. Unfortunately, glyceraldehyde did not dissolve sufficiently in acetone to yield useful spectra.
2276 J. Org. Chem., Vol. 52, No. 11, 1987
tane-l,&diol and 3-methoxybutan-1-01 indicate that hydroxy, epoxy, and methoxy are very similar in binding strength carbonyls bind more poorly and thiols and thioethers negligibl~.~Our sample of glycerol 1-acetate was not chemically pure, but its spectra in the presence of 1 and 2 in acetonitrile showed useful resolution of enantiotopic protons; 2-nitropropan-1-01gave useful resolution with 2 in acetonitrile. Amino alcohols produced potentially useful results only for 2 in acetonitrile.1° Most carboxylic acids precipitated the ligands of the shift reagents; they can be successfully studied in water with other lanthanide sa1ts.l' The Influence of Water. The presence of water (often an impurity) in samples containing chiral europium shift reagents in chloroform results in formation of troublesome precipitates of europium oxide and in severe line broadening. In spectra taken in acetonitrile or acetone, small quantities of water cause few problems and, in fact, often result in an increase in spectral resolution (Figure 2). Most of the samples described in Table I contain small amounts of water (for example, see Figure 1). In the presence of diols the water is so weakly bound that its resonance is among those of the diol CH's or the shift reagent CH's; the chemical shift of water protons only reaches that of the diol OH protons of 3-chloropropane-1,2-diol when L/S 3 2. Changes in water content usually affect the line widths more than the shift differences; the resulting sensitivity of resolution to water content provides some uncertainty in the results but also provides an adjustable parameter with which to optimize the resolution. Although solutions of 2 in acetonitrile readily precipitate in the presence of water if substrate is absent, they do not precipitate when added to moist substrate. Acetone solutions of both 1 and 2 tolerate much more water contamination without precipitation. Comparison of 1 and 2. In general, we obtained the best enantiotopic discrimination with shift reagent 2 in acetonitrile solution. Relative to 1,2 is more likely to form precipitates in the presence of water; however, it does not have the small but potentially confusing peaks in the spectrum between 5 and 7 ppm that does 1. Examples. Figure 3 illustrates the application of chiral europium shift reagents in polar solvents to a representative problem. We wished to determine the enantiomeric purity of propane-1,2-diol prepared by two routes: fermentation using Clostridium thermosaccharolyticum and reduction of hydroxyacetonewith fermenting Saccromyces cerevisae (baker's yeast).12 Both procedures give the R stereochemistry. The enantiomeric purity of propane1,2-diol could not be determined with chiral shift reagents by the usual methods in nonpolar solvents; examination with 2 in acetonitrile was easily capable of detecting a 5% enantiomeric impurity. We also tested the commercial chiral reagents we used. (S)-Propane-l,Zdiol and (R)- and (S)-methyl3-hydroxy2-methylpropionate were found to be enantiomerically pure (>98% eel. (R)-and (S)-methyl3-hydroxybutyrate consisted of 90% of the indicated isomer (80% ee), and (R)-and (S)-butane-1,3-diol consisted of 87% and 91% of (9) Propane-l,2-dithiol showed neither enantiotopic discrimination nor significant shifts in acetonitrile or acetone in the presence of 1 or 2; 3-mercaptobutan-1-01 showed nonequivalence of CH3 and CH. (10) 3-Aminopropane-1,2-diol,2-aminopropan-1-01, and l-aminopropan-2-01. Although acetonitrile gave the best results, two liquid phases were formed in the presence of 1 and 2. (11) Kabuto, K.; Sasaki, Y. J. Chem. SOC.,Chem. Commun. 1984, 316-318. Reuben, J. J . Am. Chem. SOC.1980,102,2232-2237. (12) Cameron, D. C.; Cooney, C. L. BiolTechnology 1986,4,651-654.
Sweeting et al. the indicated isomer (75% and 82% ee), respectively.
Experimental Section Analysis of NMR Spectra. NMR spectra were obtained on a Bruker WM-300 (or AM-300) spectrometer using a 12-kHz spectral width, 6 5 O pulse, and 2.7-s repetition rate, summing 16 scans; the spectra shown in this paper have been subjected to a 1-Hz exponential line broadening. We prepared solutions of four compounds by weight to simulate partially resolved substrates (ca. 50% ee) and compared the known compositions with those determined by NMR. We did not use the integral routine of the spectrometer, since it requires a flat baseline between the resonances; instead we used signal heights and triangulated areas (height times the width at half-height). With use of resonances separated by at least a 15% valley, the standard deviation in the determination of enantiomeric purity using different resonances in the same substrates was 1% with use of height or area; the NMR value was within the estimated error of the weighing (5%) in each case. We found a bigger discrepancy between the percentage calculated from the NMR spectrum and that measured volumetrically; we ascribe the error to difficulties in measuring small volumes. Line widths for the enantiotopicresonances should be different because of the different lifetimes of the diastereomeric complexes of the enantiomen with the paramagnetic shift reagent. The observed differences in line width between enantiomeric resonances were negligibly small, and enantiomeric purities could be determined by peak heights.13 The line widths increased with chemical shift in all cases. Sample Preparation. To ensure that the nonequivalence of hydroxyl protons would not be obscured by chemical exchange, surfaces coming in contact with the solutions were cleaned by a method that permitted observation of coupling with the OH and adjacent CH for the alcohol^.'^ To minimize the sources of water, the fresh shift reagents were dried with phosphorus pentoxide in an Abderhalden drying pistol; commercial shift reagents often contain >0.1% water, ca. 1mol/mol of europium. Substrates were obtained c~mmerically'~ and were dried over molecular sieves if necessary. Solutions of shift reagent were prepared under nitrogen by using fresh sealed ampules of deuteriated solvent (usually containing Me4Si)and stored in vials equipped with a valve and rubber septum. These precautions reduced, but did not completely remove, water from the samples. An appropriate quantity (typically0.5 mL) of the ca.0.2 M shift reagent solution was added to ca. 0.1 mmol of substrate in an NMR tube and the spectrum obtained the same day. Except for the epoxides (which apparently polymerize), all solutions in acetonitrile and acetone were stable for days; chloroform solutions deteriorated more quickly. Solutions were not filtered; the occasional paramagnetic particles were allowed to settle in the vial or adhere to the upper walls of the NMR tube. Filtration of chloroform solutions of 2 had no effect on the line width (see Figure 1).
Acknowledgment. This work was completed while L.M.S. was on sabbatical at Harvard University and was funded in part by the NSF under the ROA program. The NMR spectrometers were purchased in part with Grant NSF 8008891 and NIH BRS Shared Instrumentation Grant 1-510 RR01748-01A1. (13) Ethyl 2,3-epoxybutyrate was exceptional; it had a staggering 2.7-ppm shift difference and a large difference in line width between enantiomeric CITs in chloroform. (14) All containers except plastic syringes and stainless steel needles were soaked for at least 1 h in concentrated hydrochloric acid, rinsed thoroughly with water and acetone, and dried in the oven until used. (15) All substrates and shift reagents except butane-l,a-diol, 1phenylethane-l,2-diol, and 3-mercaptopropane-l,2-diolwere obtained from Aldrich; acetonitrile-d3 was obtained from Aldrich, and acetone-d6 and chloroform-d were obtained from Merck. All substrates are hygroscopic; only ethyl 2,3-epoxybutyrate and 4-hydroxy-3-methylbutan-2-one were found to be free of water by NMR. 3-Hydroxy-2-butanone exists as a dimer in the solid state and in solution in acetonitrile; its NMR spectrum shows evidence of several diastereomers of the dimer. In the presence of chiral lanthanide shift reagent, the NMR spectrum shows only one isomer, which we assume is the monomer. Similar behavior has been observed from dihydroxyacetone,which exists as a monomer in polar solvents (Davis, L. Bioorg. Chem. 1973, 2, 197-201).
