NMR studies of chiral discrimination relevant to the liquid chromatographic enantioseparation by a cellulose phenylcarbamate derivative

Article abstract of DOI:10.1021/ja960050x

Chromatographic enantioseparation of 1,1'-bi-2-naphthol (2) and its mono- and di-O-methylated derivatives (3a, 3b), 2,2'-dihydroxy-6,6'-dimethylbiphenyl (4), and 10,10'-dihydroxy-9,9'-biphenanthryl (5) has been performed on cellulose tris(5-fluoro-2-methylphenylcarbamate) (1) as a chiral stationary phase for high-performance liquid chromatography (HPLC). The complete base-line separation of 2 and 4 was achieved with the elution order of enantiomers such that the (R)-isomers eluted first followed by the (S)-isomers. The resolution of 5 and the O-methylated 3 was difficult on 1. The cellulose derivative 1 dissolved in chloroform also exhibited a chiral discrimination for 2 and 4 in ¹H and ¹³C NMR spectroscopies as well as in HPLC. The hydroxy and some aromatic protons and carbon resonances of 2 and 4 were clearly separated into a pair of peaks due to enantiomers in the presence of 1. The binding geometry and dynamics between 1 and the enantiomers of 2 were investigated on the basis of spin-lattice relaxation time, ¹H NMR titrations, and intermolecular NOEs in the presence of 1. These NMR data are fully consistent with the chromatographic elution order. These results, combined with molecular modeling, reveal the chiral discrimination rationale.

Full text of DOI:10.1021/ja960050x

4036
J. Am. Chem. Soc. 1996, 118, 4036-4048
NMR Studies of Chiral Discrimination Relevant to the Liquid
Chromatographic Enantioseparation by a Cellulose
Phenylcarbamate Derivative
Eiji Yashima, Chiyo Yamamoto, and Yoshio Okamoto*
Contribution from the Department of Applied Chemistry, School of Engineering,
Nagoya UniVersity, Chikusa-ku, Nagoya 464-01, Japan
ReceiVed January 5, 1996X
Abstract: Chromatographic enantioseparation of 1,1′-bi-2-naphthol (2) and its mono- and di-O-methylated derivatives
(3a, 3b), 2,2′-dihydroxy-6,6′-dimethylbiphenyl (4), and 10,10′-dihydroxy-9,9′-biphenanthryl (5) has been performed
on cellulose tris(5-fluoro-2-methylphenylcarbamate) (1) as a chiral stationary phase for high-performance liquid
chromatography (HPLC). The complete base-line separation of 2 and 4 was achieved with the elution order of
enantiomers such that the (R)-isomers eluted first followed by the (S)-isomers. The resolution of 5 and the
O-methylated 3 was difficult on 1. The cellulose derivative 1 dissolved in chloroform also exhibited a chiral
discrimination for 2 and 4 in 1H and 13C NMR spectroscopies as well as in HPLC. The hydroxy and some aromatic
protons and carbon resonances of 2 and 4 were clearly separated into a pair of peaks due to enantiomers in the
presence of 1. The binding geometry and dynamics between 1 and the enantiomers of 2 were investigated on the
1
basis of spin-lattice relaxation time, H NMR titrations, and intermolecular NOEs in the presence of 1. These
NMR data are fully consistent with the chromatographic elution order. These results, combined with molecular
modeling, reveal the chiral discrimination rationale.
Introduction
as a porous gel or with silica gel. A great number of former
CSPs have been prepared, and the clarification of the chiral
discrimination mechanism on the CSPs have been attempted
using spectroscopic5 and computational methods.6 The CSPs
that have been most intensively studied in this respect are
cyclodextrin-based CSPs and Pirkle-type CSPs.5,6 The rational
models of interactions between the CSPs and enantiomers have
been proposed on the basis of the structural data of a crystalline
1:1 complex of a chiral selector and an enantiomer by X-ray7
and the solution NMR experiments including intermolecular
Since biosystems consist of optically active chiral polymers,
such as proteins, nucleic acids, and polysaccharides, living
organisms often show quite different physiological behaviors
toward one of a pair of enantiomers, especially chiral drugs.1
Therefore, detailed investigations of the pharmacokinetics,
physiological, toxicological, and metabolic activities of both
enantiomers of the chiral drugs have become essential for the
development of chiral drugs in the pharmaceutical industry.2,3
Moreover, optically active compounds have recently aroused
wide interest in many fields dealing with natural products,
agrochemicals, and ferroelectric liquid crystals; therefore, their
preparation and analysis are of increasing importance.
Chromatographic enantioseparations, particularly resolution
by high-performance liquid chromatography (HPLC), have
considerably advanced during the past decade not only for
determining their optical purity but also for obtaining optical
isomers, and in the pharmaceutical industry, chiral HPLC has
become essential for the research and development of chiral
drugs.2,3 The preparation of a chiral stationary phase (CSP)
capable of effective chiral recognition is the key to this
separation technique. Therefore, many CSPs for HPLC have
been prepared, and about 110 CSPs have already been on the
market.3,4
(4) Reviews: (a) Allenmark, S. G. Chromatographic Enantioseparation;
Ellis Horwood: Chichester, 1988. (b) A Practical Approach to Chiral
Separations by Liquid Chromatography; Subramanian, G., Ed.; VCH: New
York, 1994. (c) Armstrong, D. W. Anal. Chem. 1987, 59, 84-91A. (d)
Okamoto, Y. Chemtech 1987, 176-181. (e) Pirkle, W. H.; Pochapsky, T.
C. Chem. ReV. 1989, 89, 347-362. (f) Taylor, D. R.; Maher, K. J.
Chromatogr. Sci. 1992, 30, 67-85. (g) Chiral Separations. Fundamental
Aspects and Applications; Lindner, W., Ed.; J. Chromatogr. A 1994, 666.
(5) For NMR studies of chiral recognition with the aid of chromatography
using optically active small molecule CSPs, see: (a) Feibush, B.; Figueroa,
A.; Charles, R.; Onan, K. D.; Feibush, P.; Karger, B. L. J. Am. Chem. Soc.
1986, 108, 3310-3318. (b) Pirkle, W. H.; Pochapsky, T. C. J. Am. Chem.
Soc. 1986, 108, 5627-5628. (c) Pirkle, W. H.; Pochapsky, T. C. J. Am.
Chem. Soc. 1987, 109, 5975-5982. (d) Uccello-Barretta, G.; Rosini, C.;
Pini, D.; Salvadori, P. J. Am. Chem. Soc. 1990, 112, 2707-2710. (e)
Lipkowitz, K. B.; Raghothama, S.; Yang, J. J. Am. Chem. Soc. 1992, 114,
1554-1562. (f) Lohmiller, K.; Bayer, E.; Koppenhoefer, B. J. Chromatogr.
A 1993, 634, 65-77. (g) Oi, S.; Ono, H.; Tanaka, H.; Matsuzaka, Y.;
Miyano, S. J. Chromatogr. A 1994, 659, 75-86. (h) Uccello-Barretta, G.;
Pini, D.; Rosini, C.; Salvadori, P. J. Chromatogr. A 1994, 666, 541-548.
(i) Pirkle, W. H.; Welch, C. J. J. Chromatogr. A 1994, 683, 347-353. (j)
Pirkle, W. H.; Selness, S. R. J. Org. Chem. 1995, 60, 3252-3256. (k)
Kuroda, Y.; Suzuki, Y.; He, J.; Kawabata, T.; Shibukawa, A.; Wada, H.;
Fujima, H.; Go-oh, Y.; Imai, E.; Nakagawa, T, J. Chem. Soc., Perkin Trans.
2 1995, 1749-1759.
There are basically two types of CSPs. One consists of a
chiral small molecule that is usually bound to a support silica
gel, and the second is derived from a chiral polymer being used
X Abstract published in AdVance ACS Abstracts, April 15, 1996.
(1) (a) Waldeck, B. Chirality 1993, 5, 350-355. (b) Millership, J. S.;
Fitzpatrick, A. Chirality 1993, 5, 573-576. (c) White, C. A. In A Practical
Approach to Chiral Separations by Liquid Chromatography; Subramanian,
G., Ed.; VCH: New York, 1994; Chapter 1. (d) Cardwell, J. J. Chromatogr.
A 1995, 694, 39-48.
(2) Mutton, I. M. In A Practical Approach to Chiral Separations by
Liquid Chromatography; Subramanian, G., Ed.; VCH: New York, 1994;
Chapter 11.
(6) For reviews of computational studies of chiral recognition with the
aid of chromatography using optically active small molecule CSPs, see:
(a) Lipkowitz, K. B. In A Practical Approach to Chiral Separations by
Liquid Chromatography; Subramanian, G., Ed.; VCH: New York, 1994;
Chapter 2. (b) Lipkowitz, K. B.; Anderson, A. G. In Computational
Approaches in Supramolecular Chemistry; Wipff, G., Ed.; Kluwer: Dor-
drecht, 1994; pp 183-198. (c) Lipkowitz, K. B. J. Chromatogr. A 1995,
694, 15-37. (d) Topiol, S.; Sabio, M.; Moroz, J.; Caldwell, W. B. J. Am.
Chem. Soc. 1990, 110, 8367-8376.
(3) (a) Stinson, S. C. Chem. Eng. News 1994, Sep 19, 38-72. (b) Stinson,
S. C. Chem. Eng. News 1995, Oct 9, 44-74.
S0002-7863(96)00050-9 CCC: $12.00 © 1996 American Chemical Society

NMR Studies of Chiral Discrimination
J. Am. Chem. Soc., Vol. 118, No. 17, 1996 4037
NOE studies5 which have been proved to be very powerful tools
for understanding the nature of chiral discrimination occurring
in solution between small molecules as exemplified by host-
guest complexation.8
mechanistic studies on chiral discrimination at a molecular level
have been done particularly by spectroscopic methods on
polymeric CSPs.15 Chiral polymers usually have a number of
different binding sites with a different affinity to enantiomers
and the determination of their exact structures in both solid and
solution is laborious. This makes it difficult to evaluate a precise
chiral recognition mechanism. The only exception may be some
special protein-ligand and DNA-drug complexes, whose
structures have been determined by either X-ray or NMR
analysis in solution.16
The CSPs composed of chiral polymers such as polyacryl-
amides,9 one-handed helical polymethacrylates,10 polyamides,11
proteins,12 and polysaccharide derivatives13,14 have also been
extensively studied, many of which are now commercially
available. In contrast to the small-molecule CSPs, very few
(7) (a) Pirkle, W. H.; Burke, J. A., III; Wilson, S. R. J. Am. Chem. Soc.
1989, 111, 9222-9223. (b) Francotte, E.; Rihs, G. Chirality 1989, 1, 80-
85.
We have found that phenylcarbamate derivatives of cellulose
and amylose show high resolving power as CSPs when coated
on macroporous silica gel and can separate a wide range of
racemates including many drugs.14 Some of the derivatives have
been commercialized and used as very popular CSPs.17 How-
ever, the chiral recognition mechanism on the phenylcarbamate
derivatives at a molecular level remains elusive, although a
qualitative explanation has been given on the basis of chro-
matographic enantioseparation14 and computational studies.18
An NMR spectroscopic study on the chiral recognition
mechanism of polymer systems often fails because most
polymers are soluble only in the solvents which prevent
enantiomer discrimination. Most phenylcarbamate derivatives
of the polysaccharides with high chiral resolving power as CSPs
are typical cases. They are soluble only in polar solvents, such
as tetrahydrofuran (THF), acetone, and pyridine. In such polar
solvents, the phenylcarbamate derivatives show poor chiral
recognition for enantiomers because the solvents preferentially
interact with the polar carbamate residues which are the most
important binding site for chiral discrimination.
(8) (a) Cram, D. J. Angew. Chem., Int. Ed. Engl. 1988, 27, 1009-1020.
(b) Lehn, J. M. Angew. Chem., Int. Ed. Engl. 1988, 27, 89-112. (c)
Dharanipragada, R.; Ferguson, S. B.; Diederich, F. J. Am. Chem. Soc. 1988,
110, 1679-1690. (d) Echavarren, A.; Gala´n, A.; Lehn, J. M.; Mendoza, de
J. J. Am. Chem. Soc. 1989, 111, 4994-4995. (e) Spisni, A.; Corradini, R.;
Marchelli, R.; Dossena, A. J. Org. Chem. 1989, 54, 684-688. (f) Sanderson,
P. E. J.; Kilburn, J. D.; Still, W. C. J. Am. Chem. Soc. 1989, 111, 8314-
8315. (g) Liu, R.; Sanderson, P. E. J.; Still, W. C. J. Org. Chem. 1990, 55,
5184-5186. (h) Rebek, J., Jr. Angew. Chem., Int. Ed. Engl. 1990, 29, 245-
255. (i) Webb, T. H.; Suh, H.; Wilcox, C. S. J. Am. Chem. Soc. 1991, 113,
8554-8555. (j) Seel, C.; Vo¨gtle, F. Angew. Chem., Int. Ed. Engl. 1992,
31, 528-549. (k) Cotero´n, J. M.; Vicent, C.; Bosso, C.; Penade´s, S. J. Am.
Chem. Soc. 1993, 115, 10066-10076. (l) Webb, T. H.; Wilcox, C. S. Chem.
Soc. ReV. 1993, 383-395. (m) Bo¨hmer, V. Angew. Chem., Int. Ed. Engl.
1995, 34, 713-745.
(9) (a) Blaschke, G. Angew. Chem., Int. Ed. Engl. 1980, 19, 13-24. (b)
Blaschke, G. J. Liq. Chromatogr. 1986, 9, 341-368.
(10) (a) Yuki, H.; Okamoto, Y.; Okamoto, I. J. Am. Chem. Soc. 1980,
102, 6356-6358. (b) Okamoto, Y.; Okamoto, I.; Yuki, H.; Murata, S.;
Noyori, R.; Takaya, H. J. Am. Chem. Soc. 1981, 103, 6971-6973. (c)
Okamoto, Y.; Hatada, K. J. Liq. Chromatogr. 1986, 9, 369-384. (d)
Okamoto, Y.; Yashima, E. Prog. Polym. Sci. 1990, 15, 263-298. (e)
Chance, J. M.; Geiger, J. H.; Okamoto, Y.; Aburatani, R.; Mislow, K. J.
Am. Chem. Soc. 1990, 112, 3540-3547. (f) Okamoto, Y.; Nakano, T. Chem.
ReV. 1994, 94, 349-372.
However, we recently found that cellulose tris(4-(trimethyl-
silyl)phenylcarbamate) (CTSP)19 is soluble in chloroform and
(11) Saigo, K. Prog. Polym. Sci. 1992, 17, 35-86.
1
shows chiral discrimination in H NMR spectroscopy as well
(12) For reviews, see ref 4a,b,g. (a) Hermansson, J.; Eriksson, M. J. Liq.
Chromatogr. 1986, 9, 621-639. (b) Allenmark, S. G. J. Liq. Chromatogr.
1986, 9, 425-442. (c) Miwa, T.; Miyakawa, T. J. Chromatogr. 1987, 408,
316-322. (d) Erlandsson, P.; Marle, I.; Hansson, L.; Isaksson, R.; Pettersson,
C.; Pettersson, G. J. Am. Chem. Soc. 1990, 112, 4573-4574. (e) Wainer,
I. W.; Jadaud, P.; Schombaum, G. R.; Kadadker, S. V.; Henry, M. P.
Chromatographia 1988, 25, 903-907. (f) Noctor, T.; Felix, G.; Wainer, I.
W. Chromatographia 1991, 31, 55-59. (g) Haginaka, J.; Murashima, T.;
Seyama, C. J. Chromatogr. A 1994, 666, 203-210.
as in HPLC. For instance, the methine proton of trans-2,3-
diphenyloxirane was split into two singlet resonances in the
presence of CTSP,19 indicating that CTSP can discriminate the
enantiomers even in solution. The elution order of trans-2,3-
diphenyloxirane on CTSP correlated well with the downfield
1
shift of the (-)-isomer observed in the H NMR. CTSP can
work as the chiral shift reagent and discriminates the enanti-
omers of the Tro¨ger base, benzoin, mandelic acid, and several
sec-alcohols, such as 2-butanol and 2-octanol in CDCl3.20
However, the differences in the chemical shifts of the enanti-
omers in the presence of CTSP were too small to obtain reliable
(13) For cellulose esters as CSPs, see: (a) Hesse, G.; Hagel, R.
Chromatographia 1973, 6, 277-280. (b) Becher, G.; Mannschreck, A.
Chem. Ber. 1981, 114, 2365-2368. (c) Okamoto, Y.; Kawashima, M.;
Yamamoto, K.; Hatada, K. Chem. Lett. 1984, 739-742. (d) Ichida, A.;
Shibata, T.; Okamoto, I.; Yuki, Y.; Namikoshi, H.; Toda, Y. Chro-
matographia 1984, 19, 280-284. (e) Francotte, E.; Wolf, R. M.; Lohmann,
D. J. Chromatogr. 1985, 347, 25-37. (f) Okamoto, Y.; Aburatani, R.;
Hatada, K. J. Chromatogr. 1987, 389, 95-102. (g) Wainer, I. W.; Stiffin,
R. M.; Shibata, T. J. Chromatogr. 1987, 139, 139-151. (h) Francotte, E.;
Wolf, R. M. Chirality 1990, 2, 16-31. (i) Francotte, E. J. Chromatogr. A
1994, 666, 565-601.
(14) For reviews of phenylcarbamates of polysaccharides as CSPs, see:
(a) Shibata, T.; Okamoto, I.; Ishii, K. J. Liq. Chromatogr. 1986, 9, 313-
340. (b) Okamoto, Y.; Aburatani, R.; Hatano, K.; Hatada, K. J. Liq.
Chromatogr. 1988, 11, 2147-2163. (c) Okamoto, Y.; Kaida, Y. J. High
Resolut. Chromatogr. 1990, 13, 708-712. (d) Okamoto, Y.; Kaida, Y.;
Aburatani, R.; Hatada, K. In Chiral Separations by Liquid Chromatography;
Ahuja, S., Ed.; ACS Symposium Series 471; American Chemical Society:
Washington, DC, 1991; pp 101-113. (e) Okamoto, Y.; Kaida, Y. J.
Chromatogr. A 1994, 666, 403-419. (f) Dingene, J. In A Practical Approach
to Chiral Separations by Liquid Chromatography; Subramanian, G., Ed.;
VCH: New York, 1994; Chapter 6. (g) Oguni, K.; Oda, H.; Ichida, A. J.
Chromatogr. A 1995, 694, 91-100. (h) Yashima, E.; Okamoto, Y. Bull.
Chem. Soc. Jpn. 1995, 68, 3289-3307. For other leading references, see:
(i) Okamoto, Y.; Kawashima, M.; Hatada, K. J. Am. Chem. Soc. 1984, 106,
5357-5359. (j) Okamoto, Y.; Kawashima, M.; Hatada, K. J. Chromatogr.
1986, 363, 173-186. (k) Okamoto, Y.; Aburatani, R.; Hatada, K. Bull.
Chem. Soc. Jpn. 1990, 63, 955-957. (l) Okamoto, Y.; Ohashi, T.; Kaida,
Y.; Yashima, E. Chirality 1993, 5, 616-621. (m) Ishikawa, A.; Shibata, T.
J. Liq. Chromatogr. 1993, 16, 859-878. (n) Chankvetadze, B.; Yashima,
E.; Okamoto, Y. J. Chromatogr. A 1994, 670, 39-49. (o) Yashima, E.;
Fukaya, H.; Okamoto, Y. J. Chromatogr. A 1994, 677, 11-19. (p) Maeda,
K; Okamoto, Y; Morlender, N; Eventova, I; Biali, S. E; Rappoport, Z. J.
Am. Chem. Soc. 1995, 117, 9686-9689.
(15) For NMR studies of chiral recognition on polymeric CSPs, see: (a)
Oguni, K.; Matsumoto, A.; Isokawa, A. Polym. J. 1994, 26, 1257-1261.
(b) Pinkerton, T. C.; Howe, W. J.; Ulrich, E. L.; Comiskey, J. P.; Haginaka,
J.; Murashima, T.; Walkenhorst, W. F.; Westler, W. M.; Markley, J. L.
Anal. Chem. 1995, 67, 2354-2367. For computational studies of chiral
recognition on polymeric CSPs, see: (c) Wolf, R. M.; Francotte, E.;
Lohmann, D. J. Chem. Soc., Perkin Trans. 2 1988, 893-901.
(16) For reviews: (a) Schulz, G. E.; Schirmer, R. H. Principles of Protein
Structure; Springer-Verlag: New York, 1979; Chapter 10. (b) Saenger, W.
Principles of Nucleic Acid Structure; Springer-Verlag: New York, 1984;
Chapter 16. (c) Wu¨thrich, K. NMR of Proteins and Nucleic Acids; Wiley:
New York, 1986; Chapter 15 and references cited therein. (d) Lee, M.;
Shea, R. G.; Hartley, J. A.; Kissinger, K.; Pon, R. T.; Vesnaver, G.;
Breslauer, K. J.; Dabrowiak, J. C.; Lown, J. W. J. Am. Chem. Soc. 1989,
111, 345-354. (e) Zhang, X.; Patel, D. J. Biochemistry 1990, 29, 9451-
9466. (f) Eriksson, M.; Leijon, M.; Hiort, C.; Norde´n, B.; Gra¨ulund, A. J.
Am. Chem. Soc. 1992, 114, 4933-4934. (g) Paloma, L. G.; Smith, J. A.;
Chazin, W. J.; Nicolaou, K. C. J. Am. Chem. Soc. 1994, 116, 3697-3708.
(h) Blasko´, A.; Browne, K. A.; Bruice, T. C. J. Am. Chem. Soc. 1994, 116,
3726-3737.
(17) Commercialized as CHIRALCEL and CHIRALPAK by Daicel
Chemical Industries, Ltd., Tokyo, Japan.
(18) Yashima, E.; Yamada, M.; Kaida, Y.; Okamoto, Y. J. Chromatogr.
A 1995, 694, 347-354.
(19) Yashima, E.; Yamada, M.; Okamoto, Y. Chem. Lett. 1994, 579-
582.
(20) Okamoto, Y.; Yashima, E. Macromol. Symp. 1995, 99, 15-23.

4038 J. Am. Chem. Soc., Vol. 118, No. 17, 1996
Yashima et al.
Chart 1
Figure 1. Chromatogram of the enantioseparation of (RS)-1,1′-bi-2-
naphthol (2) on CSP 1 with hexane-2-propanol (90/10) as the eluent
at 20 °C. Column, 25 × 0.46 cm i.d.; flow rate, 1.0 mL min-1
.
Table 1. Capacity Factor (k1′), Separation Factor (R), and the
Difference in Free Energy (∆∆G) in the Enantioseparation of 2-5
on the CSP 1 at 23 °Ca
data on the chiral recognition rationale through 1H NMR
titrations and NOE measurements because of weak binding.
Very recently, we have found that cellulose tris(5-fluoro-2-
methylphenylcarbamate) (1) (Chart 1) is also soluble in chlo-
roform21 and is capable of discriminating enantiomers of 1,1′-
bi-2-naphthol (2) and 2,2′-dihydroxy-6,6′-dimethylbiphenyl (4)
racemate
k1′ b
R
∆∆Gc (kcal‚mol-1
)
2
2.31 ((R)-(+))
0.85 ((R)-(+))d
0.73 ((R)-(+))
0.46
1.39 ((R)-(+))
1.31 (-)
4.23
3.22d
ca. 1
1.00
3.22
ca. 1
0.85
0.69d
ca. 0
0
0.69
ca. 0
3ad
3bd
4
1
in H and 13C NMR with large chemical shift changes of the
enantiomers as well as in HPLC accompanying the large
separation factors (R > 3). These results suggest that the system
consisting of 1 and 2 or 4 will be suitable for elucidating the
chiral discrimination mechanism on cellulose-based CSPs at a
molecular level through NMR studies.
5
a Hexane-2-propanol (90/10) was used as the eluent at a flow rate
of 1.0 mL min-1 b The sign in parentheses represents the absolute
.
configuration and/or optical rotation of the first-eluting enantiomer.
c ∆∆G can be estimated by using eq 1 (see text). d The eluent was
hexane-2-propanol (70/30).
Results and Discussion
value (3.22), and the (S)-isomer was more retained as well as
in the resolution of 2. On the other hand, 10,10′-dihydroxy-
9,9′-biphenanthryl (5) was hardly separated under identical
conditions. Bulky aromatic groups around the OH groups may
disturb the efficient hydrogen bonding. From the separation
factor (R), the difference in free energy (∆∆G°) at a given
temperature can be calculated by using eq 1 (Table 1), where R
Chromatographic Enantioseparation. Figure 1 shows a
chromatogram of the resolution of racemic 1,1′-bi-2-naphthol
(2) on a column packed with cellulose tris(5-fluoro-2-meth-
ylphenylcarbamate) (1). The peaks were detected by a UV
detector and identified by a polarimetric detector. The (R)-
(+)-2 and (S)-(-)-2 enantiomers eluted at retention times
of t1 and t2, respectively, showing complete separation.
Chromatographic parameters, capacity factors, k1′ ) (t1 - t0)/
t0 and k2′ ) (t2 - t0)/t0, and separation factor (R ) k2′/k1′) were
found to be 2.31, 9.77, and 4.23, respectively. The dead time
t0 was estimated by using 1,3,5-tri-tert-butylbenzene as a
nonretained compound.22 This indicates that (S)-(-)-2 interacts
more strongly with the stationary phase consisting of 1 than
(R)-(+)-2.
To investigate the role of hydrogen bonding interaction
between the CSP 1 and 2 in chiral recognition, one or both of
the hydroxy groups of 2 were methylated (3a and 3b). The
chromatographic results of enantioseparation of 2 and four
analogues (3a, 3b, 4, and 5) on the same column 1 are given in
Table 1. For 3a and 3b, a remarkable decrease of interaction
in retention (k1′) and enantioselectivity (R) toward the CSP 1
was observed. These results indicate that hydrogen bonding
interaction through the hydroxy groups of 2 is the main force
for its retention and separation, and the simultaneous hydrogen
bonding through the two hydroxy groups may be essential for
effective resolution.
∆∆G° ) -RT ln R
(1)
is the gas constant of 1.987 cal mol-1 K-1 and T is the absolute
temperature in K. These values can also be estimated by the
NMR titration method in solution which will be discussed later.
One-Dimensional NMR Studies. The phenylcarbamate 1
can resolve many enantiomers in addition to 2 and 4 as a CSP
in HPLC.21 Since 1 is soluble in chloroform, one can investigate
1
the chiral discrimination of 1 in solution by H and 13C NMR
spectroscopies.
Figure 2 shows the 500 MHz H NMR spectra of (RS)-2 in
1
the absence (A) and presence (B) of 1 in CDCl3. The peak
assignments were done on the basis of 2D COSY and NOESY
experiments. The assignments for 2 were identical to those in
dimethyl sulfoxide-d6.23 Each of the hydroxy and naphthyl
protons (H4 and H6) of the enantiomers of 2 were significantly
separated into two peaks in the presence of 1. Other naphthyl
protons were also split with relatively small chemical shift
differences (Table 2). This clearly indicates that 1 can recognize
the enantiomers even in solution. The chemical shift differences
(∆∆δ) are sufficiently large enough to determine the ratio of
the enantiomers, indicating that 1 can be used as a chiral shift
An analogous biaryl compound, 2,2′-dihydroxy-6,6′-dimeth-
ylbiphenyl (4), was also resolved completely with a large R
(21) Yashima, E.; Yamamoto, C.; Okamoto, Y. Polym. J. 1995, 27, 856-
861.
(22) (a) Koller, H.; Rimbo¨ck, K.-H.; Mannschreck, A. J. Chromatogr.
1983, 282, 89-94. (b) Pirkle, W. H.; Welch, C. J. J. Liq. Chromatogr.
1991, 14, 1-8.
(23) Marin-Puga, G.; Hora´k, V.; Svoronos, P. Collect. Czech. Chem.
Commun. 1993, 58, 77-81.

NMR Studies of Chiral Discrimination
J. Am. Chem. Soc., Vol. 118, No. 17, 1996 4039
Figure 2. 1H NMR spectra of selected region of (RS)-2 (19.4 mM) in the absence (A) and presence (B) of 1 (36.1 mM glucose units) in CDCl3
at 23 °C. The assignments were performed with 2D COSY and NOESY experiments.
Table 2. 1H and 13C NMR Chemical Shifts (ppm) of 2 in the Absence and the Presence of 1 in CDCl3 at 23 °C
1H NMR
1-(R)-2b
13C NMR
1-(S)-2e
position
freea
1-(S)-2b
∆∆δc
freed
1-(R)-2e
∆∆δc
OH
1
2
3
4
5
6
7
8
5.036
5.435
5.234
0.201
110.74
152.70
117.71
131.40
129.40
128.38
124.02
127.47
124.16
133.35
111.05
152.59
117.63
131.42
129.37
128.31
123.94
127.42
124.16
133.52
110.94
152.68
117.72
131.34
129.34
128.36
123.97
127.40
124.21
133.41
0.11
-0.09
-0.09
0.08
7.391
7.985
7.358
7.917
7.385
7.967
-0.027
-0.050
0.03
7.900
7.378
7.312
7.156
7.819
7.337
7.297
7.137
7.881
7.372
7.304
7.145
-0.062
-0.035
-0.007
-0.008
-0.05
-0.03
0.02
-0.05
0.11
9
10
a [(RS)-2] ) 19.4 mM. b The chemical shifts of OH, H4, H6, and H9 proton resonances were based on the spectrum of (RS)-2 (19.4 mM) in the
presence of 1 (36.1 mM) (see Figure 2). Because of overlapping of the peaks, the other protons chemical shifts were measured separately for (S)-
and (R)-2 (each 9.7 mM) with 1 (18.1 mM). c Obtained by subtracting the 1-(R)-2 values from the 1-(S)-2 ones. d [(RS)-2] ) 116 mM. e The
chemical shifts of C1, C2, C3, C4, and C10 carbon resonances were based on the spectrum of (RS)-2 (116 mM) in the presence of 1 (54.2 mM)
(see Figure 3). Because of overlapping of the peaks, the other carbons chemical shifts were measured separately for (S)- and (R)-2 (each 58.2 mM)
with 1 (54.2 mM).
reagent. Although a large number of designed chiral hosts,
chiral solvating agents, and chiral lanthanide shift reagents have
been reported for recognizing optical isomers in solution by
NMR,8,24 only a few of them have used optically active polymers
as a chiral selector.15,25 On the basis of the measurement with
enantiomerically pure (R)- and (S)-2, it became clear that the
hydroxy protons ((S)-2-OH) were more largely shifted to
downfield accompanying with line broadening than the corre-
sponding (R)-2-OH, whereas the (S)-2-H4 and -H6 resonances
exhibited upfield shift and broadening, indicating that (S)-2
interacts more strongly with 1. The downfield shift of the OH
resonances is ascribed to hydrogen bond effects,5a,26 and the
upfield shifts for the aromatic protons of 2 are probably due to
π-stacked or shielding effect by a neighboring aromatic ring of
1.27 The significant broadening of these proton resonances of
(S)-2 indicates that exchange rates between the free and bound
forms of 2 to 1 is slow compared with the NMR time scale,25,28
although a clear coalescence point could not be observed from
dynamic NMR experiments of the mixture at temperatures from
+60 to -40 °C in CDCl3. As the temperature was lowered,
the resonances of the (S)-OH, -H4 and -H6 protons were more
broadened. The larger chemical shift movement and broadening
1
of the (S)-2 resonances than (R)-2 observed in the H NMR is
associated with the chromatographic elution order of the
enantiomer.
Discrimination of enantiomers in 1H NMR was also observed
for (RS)-4 in the presence of 1, and the structural features of
binding were similar to that for 1-(RS)-2 complexation; the
OH proton resonances were split into two peaks with large
downfield shifts (∆δ ) 0.563 and 0.165 ppm for (S)- and (R)-
4, respectively). The lower-field, broad peak can be assigned
to the (S)-OH protons, and these observations are consistent
(24) For reviews, see: (a) Rinaldi, P. L. Prog. Nucl. Magn. Reson.
Spectrosc. 1982, 15, 291-352. (b) Weisman, G. R. In Asymmetric Synthesis;
Morrison, J. D., Ed.; Academic Press: New York, 1983; Chapter 8. (c)
Fraser, R. R. In Asymmetric Synthesis; Morrison, J. D., Ed.; Academic
Press: New York, 1983; Chapter 9.
(25) DNA as a chiral polymer (1H NMR): (a) Rehmann, J. P.; Barton,
J. K. Biochemistry 1990, 29, 1701-1709. Polypeptides as a chiral polymer
(2H NMR): (b) Lafontaine, E.; Bayle, J. P; Courtieu, J. J. Am. Chem. Soc.
1989, 111, 8294-8296. (c) Canet, J.-L.; Fadel, A.; Salau¨n, J.; Canet-Fresse,
I.; Courtieu, J. Tetrahedron Asymmetry 1993, 4, 31-34. (d) Meddour, A.;
Canet, I.; Loewenstein, A.; Pe´chine´, J. M.; Courtieu, J. J. Am. Chem. Soc.
1994, 116, 9652-9656. (e) Canet, I.; Courtieu, J. Tetrahedron Asymmetry
1995, 6, 333-336. Cellulose esters as a chiral polymer (13C NMR): see
ref 15a.
(26) (a) Dobashi, Y.; Dobashi, A.; Ochiai, H.; Hara, S. J. Am. Chem.
Soc. 1990, 112, 6121-6123. (b) Nishiyama, H.; Tajima, T.; Takayama,
M.; Itoh, K. Tetrahedron Asymmetry 1993, 4, 1461-1464.
(27) Zimmerman, S. C.; Wu, W.; Zeng, Z. J. Am. Chem. Soc. 1991,
113, 196-201.
(28) Collins, J. G.; Shields, T. P.; Barton, J. K. J. Am. Chem. Soc. 1994,
116, 9840-9846.

4040 J. Am. Chem. Soc., Vol. 118, No. 17, 1996
Yashima et al.
Figure 3. 13C NMR spectrum of (RS)-2 (116 mM) in the presence of 1 (54.2 mM) in CDCl3 at 23 °C.
Table 3. 13C NMR Chemical Shifts (ppm) of 4 in the Absence
formation of polymers and its interaction with another molecule
at a molecular level.16
and the Presence of 1 in CDCl3 at 23 °C
position
freea
1-(S)-4b
1-(R)-4b
∆∆δc
In the NOESY spectrum of the free polysaccharide derivative
1, a number of NOE cross peaks were observed in the region
of glucose-glucose, aromatic-aromatic, and aromatic-methyl
proton resonances (see the supporting information). The
chemical shifts of the glucose protons (H1-H6) of 1 were
identical within 0.1 ppm to those of cellulose triacetate (CTA)
in CDCl3 reported by Buchanan et al.,29 and the NOE cross
peak pattern of the glucose proton resonances was also similar,
indicating that the glucose unit of 1 may have a conformation
similar to that of CTA. Buchanan et al. proposed a 5/4 helical
structure for CTA based on the calculated interproton distances
of the glucose protons, e.g., H1-H4′ (a prime represents a nuclei
of the adjacent glucose residue), by measuring peak volumes
of cross and diagonal peaks at a different short mixing times
(60-100 ms).29 They did not provide sufficient data for
determining the torsion angle about the glycoside bond defined
by two dihedral angles (H1-C1-O-C4′-H4′), although a
certain torsion angle was determined using molecular models
of CTA postulated by X-ray analysis. We attempted to estimate
the peak volumes of cross peaks in the glucose proton
resonances of 1 by means of the NOESY method acquired at
different short mixing times (60-150 ms) to determine the
torsion angle of the glycoside bond of 1 according to the
Buchanan’s method. Although it was difficult owing to the
broadening of the peaks and the overlapping in some glucose
protons, we could use the structural data of a cellulose derivative,
cellulose tris(phenylcarbamate) (CTPC), to construct a molecular
model of a CTPC derivative 1; the structure of crystalline CTPC
was determined to be a left-handed 3/2 helix by X-ray analysis.30
Analogous cellulose derivatives, cellulose tribenzoate31 and
cellulose tris(4-chlorophenylcarbamate),32 have been reported
to have the same left-handed 3/2 helical structure by means of
X-ray analysis. Moreover, it was reported from light scattering
and viscometry measurements that CTPC may hold a similar
helical structure even in solution.33,34 On the basis of these
reported structure data, we are able to build up a model polymer
as described later.
2
3
4
5
6
153.81
113.16
130.12
122.59
138.94
153.72
113.06
130.06
122.58
139.07
153.78
113.12
129.99
122.51
138.95
-0.06
-0.06
0.07
0.07
0.12
a [(RS)-4] ) 156 mM. b [(R)- or (S)-4] ) 156 mM, [4]/[1] ) 2.88.
c Obtained by subtracting the 1-(R)-4 values from the 1-(S)-4 ones.
with the chromatographic enantioseparation results (Table 1).
However, for 3a and 3b, no splitting due to the enantiomers
was observed in the presence of 1. These NMR results also
support the importance of the hydrogen bonding of the two
hydroxy groups for chiral discrimination as seen in the chiral
1
HPLC. The racemate 5 was hardly separated in H NMR as
well as in HPLC. These 1H NMR data are fully consistent with
the chromatographic data with respect to the enantioselectivity
and elution order.
Similar splittings of 2 and 4 into enantiomers in the presence
of 1 were also observed in 13C NMR spectroscopy. Figure 3
shows the 125 MHz 13C NMR spectrum of (RS)-2 in the
presence of 1 in CDCl3. The carbon resonances of free 2 were
unambiguously assigned from the H-13C COSY experiment.
1
In Figure 3, it is apparent that the C1-C4 and C10 resonances
are enantiomerically separated and the (S)-2 carbon resonances
are markedly broadened as seen in its 1H NMR spectrum (Figure
2). In Table 2 are summarized the 1H and 13C NMR chemical
shifts of (R)- and (S)-2 in the presence and absence of 1.
Chemical shift differences between the corresponding proton
or carbon resonances of (R)- and (S)-2 were obtained by
subtracting the (R)-2 values from the (S)-2 ones. The carbons
(C1-C4 and C10) exhibiting large splits (∆∆δ) belong to ring
A (see Figure 3), indicating that ring A may be favorably located
close to the chiral glucose residue; in other words, (S)-2 may
insert from ring A into the chiral groove of 1 to form the
hydrogen bond involving the two hydroxy groups of (S)-2 and
the carbamate residues of 1 (see below for detailed discussion).
Some carbons of (RS)-4 were also split into enantiomers in
the presence of 1 (Table 3).
(29) Buchanan, C. M.; Hyatt, J. A.; Lowman, D. W. J. Am. Chem. Soc.
1989, 111, 7312-7319.
(30) (a) Zugenmaier, P.; Vogt, U. Makromol. Chem. 1983, 184, 1749-
1760. (b) Vogt, U.; Zugenmaier, P. Ber. Bunsenges. Phys. Chem. 1985,
89, 1217-1224.
Two-Dimensional NMR Studies. The recent development
of 2D NMR techniques has provided a powerful tool for
constructing the structures of biopolymers and synthetic poly-
mers and for elucidating the interaction occurring in host-guest
bimolecular systems such as biopolymers and drugs. In
particular, a 2D NOESY spectroscopy is very useful to obtain
interproton distances, which provide information on the con-
(31) Steinmeier, H.; Zugenmaier, P. Carbohydr. Res. 1987, 164, 97-
105.
(32) Riehl, K. Ph.D. Thesis, Technischen Universita¨t Clausthal, Germany,
1992.
(33) Danhelka, J.; Netopil´ık, M.; Bohdanecky, M. J. Polym. Sci. Part
B: Polym. Phys. 1987, 25, 1801-1815.

NMR Studies of Chiral Discrimination
J. Am. Chem. Soc., Vol. 118, No. 17, 1996 4041
Table 4. Spin-Lattice Relaxation Times (T1/s) of 2 in the
Absence and Presence of 1 in CDCl3 at 19 °Ca
position
(R,S)-2b
1-(S)-2c
1-(R)-2c
13C
1
2
3
4
5
6
7
8
9
11.93
7.68
1.38
1.30
6.11
1.29
1.18
1.31
1.28
6.69
7.47
4.20
0.90
0.82
3.07
0.97
0.90
0.86
0.92
2.98
8.82
5.31
1.07
0.99
4.03
1.02
0.93
1.01
1.00
4.15
10
a The measurements of T1s were carried out three times, and the
deviation was less than 3%. b [(R,S)-2] ) 116 mM. c [(S)- or (R)-2] )
58.2 mM, [2]/[1] ) 1.07.
the 2- and 3-positions has not yet been attained because of the
difficulty of regioselective substitution on the two hydroxy
groups at the 2- and 3-positions of a glucose.
On the left side of the contour plots in Figure 4 are shown
column slices taken through the tops of these three methyl peaks
on the phenyl groups of 1. In the spectrum of free 1 (A), many
cross peaks were observed. Most of them can be ascribed to
intramolecular NOEs between the methyl and aromatic protons
on the same phenyl ring, though some of them cannot. These
unassigned NOEs may be due to those between the NH proton
and methyl protons or between the aromatic protons and the
methyl protons on a different phenyl residue. Three NH proton
resonances should exist at around 6.5-7.5 ppm. However, these
resonances could not be clearly observed at 30 °C. At 60 °C,
one clear and two rather broad NH resonances appeared in this
region, and their possible chemical shifts at 30 °C are marked
by asterisks in Figure 4A. These NH protons may be positioned
closely to the methyl protons showing NOEs.
The 1:2 mixture of (R)-2 and 1 (Figure 4B) exhibited a
NOESY spectrum very similar to the spectrum of Figure 4A,
and no intermolecular NOE cross peaks between 1 and (R)-2
were observed, indicating that a weak interaction exists between
them. This observation is in accord with the results in the chiral
HPLC and 1D NMR experiments. On the other hand, some
clear intermolecular NOE cross peaks being represented by
arrows in Figure 4C were observed between the aromatic H4,
H6, and H7 protons of (S)-2 and the methyl protons of 1 under
identical conditions. These data indicate that (S)-2 binds or
interacts with 1 more strongly than (R)-2, and the naphthyl
protons of (S)-2 are located closely to the methyl protons of 1
within less than 5 Å. All the intra- and intermolecular NOE
enhancements were negative as shown in the column slices,
which are often seen in 2D NOESY spectra of biopolymer-
drug complexes, indicating that the 1-(S)-2 complex appears
to be in the slow motion regime (ω0τ0 > 1.1).5c,8k,16c The
strongest intermolecular NOE cross peaks observed are between
the 6-methyl protons and H4 and H6 of (S)-2 and the 2- or
3-methyl protons and H7 of (S)-2. These observations may be
useful in order to construct a structural model for the complex-
ation between (S)-2 and 1. In the aromatic-aromatic region
of the NOESY spectrum, however, there exist no clear
intermolecular cross peaks, even when the NOESY spectra were
acquired at different mixing times (60, 80, 150, 300, and 500
ms).
Figure 4. Expanded NOESY spectra (right hand side) at a mixing
time of 300 ms of the free 1 (A) and the mixtures of (R)-2 and 1 (B)
and (S)-2 and 1 (C) (molar ratio, 1:2) in the region between the aromatic
protons (1 and 2) and methyl protons on the phenyl moiety of 1 in
CDCl3 at 30 °C. The concentration of 2 was 9.7 mM. Asterisks in A
indicate the possible chemical shifts of the NH proton resonances of
1. The aromatic proton resonances of 1 are denoted by o (ortho), m
(meta), and p (para). The arrows (C) indicate intermolecular NOE cross
peaks observed between the aromatic H4, H6, and H7 protons of (S)-2
and the methyl protons of 1. On the left hand side are shown the column
slices taken through the tops of the three methyl peaks on the phenyl
groups introduced at the 2,3- (a and b) or 6-position (c) of a glucose
unit of 1.
Figure 4 shows the NOESY spectra of free 1 (A), 1-(R)-2
(B), and 1-(S)-2 (C) (molar ratio, 2:1) in the regions between
the aromatic protons (1 and 2) and the methyl protons on the
phenyl group of 1. The aromatic proton resonances [ortho (o),
meta (m), and para (p)] were assigned from the COSY spectrum,
but the 2-, 3-, or 6-position of a glucose unit could not be
identified. The assignment of the methyl proton resonances on
the phenyl groups introduced at the 2-, 3-, or 6-position of a
glucose unit was attained by comparing the NMR data of a
regioselectively carbamoylated model polymer, cellulose 6-(5-
fluoro-2-methylphenylcarbamate)2,3-bis(4-chlorophenylcarbam-
ate). A similar chemical shift difference of the methyl protons
on the phenyl groups depending on the position of a glucose
residue has been observed for other methylphenylcarbamates
of cellulose.35 The assignment of the two methyl protons at
(34) The monomeric length (h) along the contour of the CTPC chain
was estimated to be 0.51 nm based on the data taken from ref 33, which is
in agreement with the pitch per residue (0.51 nm) of the left-handed 3/2
helix for the crystalline CTPC.30 A similar observation has recently been
reported for cellulose tris(3,5-dimethylphenylcarbamate): Tsuboi, A.;
Yamasaki, M.; Norisue, T.; Teramoto, A. Polym. J. 1995, 27, 1219-1229.
However, these agreements do not necessarily indicate that the CTPC and
its derivatives maintain the same helical conformation in solution, because
the h value is close to the monomeric projection for other cellulose
derivatives with a different conformation.
Complexation and Dynamics. An indication of different
dynamics of the two enantiomers bound to 1 was obtained
through the examination of relaxation properties of the enan-
tiomers in the presence of 1 (Table 4). Spin-lattice relaxation
(35) Kaida, Y.; Okamoto, Y. Bull. Chem. Soc. Jpn. 1993, 66, 2225-
2232.

4042 J. Am. Chem. Soc., Vol. 118, No. 17, 1996
Yashima et al.
Figure 5. 1H NMR titrations of (S)-2 (1.94 mM) with 1 (0-63.2 mM) in CDCl3 at 23 °C. The changes in chemical shifts of OH (A) and H4 (B)
proton resonances of (S)-2 were followed. Negative value indicates an upfield shift.
time (T1) is sensitive to molecular motions, and therefore, is
useful to know how the mobility differs between the enantiomers
of 2 in the presence of 1. All carbons of (R)- and (S)-2 in the
presence of 1 showed shorter T1s than those of the free 2, and
those for (S)-2 were always shorter than the corresponding T1
of the (R)-2. Remarkable reduction of the T1 was observed for
the C1-C5 and C10 carbons attached to the ring A (see Figure
3). These results indicate that the mobility of the ring A bearing
hydroxy groups was unequivocally restricted probably due to
binding to the polymer 1 through the intermolecular hydrogen
bond between the OHs at the C2 carbons in the ring A and the
CdO of carbamate residues of polymer 1.
glucose unit)-guest stoichiometry and are in agreement with
the 1:1 complexation. Because of solubility limit of 1 and 2 in
CDCl3, the maximum [1]t/[2] was less than ca. 40 (t ) total).
Thus, the titration data were analyzed using the following two
equations to reduce errors and to evaluate correctly the binding
constant (KS)
1/∆δ ) 1/∆δmax + 1/(∆δmaxKS[1]t)
[1]t/∆δ ) ([1]t + [2]t - [2]t∆δ/∆δmax)/∆δmax
(3)
+
1/(∆δmaxKS) (4)
Measurements of 1H spin-lattice relaxation time also support
the above speculation. In the presence of 1, the T1s for the H4
and H6 of (S)-2 were 2.4 and 2.6 s, respectively, whereas those
for the (R)-2 were 2.8 and 2.9 s, respectively; those values are
shorter than the T1s of free 2 (2.9 and 3.0 s, respectively).
The standard titration experiments of (S)- and (R)-2 in the
presence of 1 at various temperatures were carried out by means
of 1H NMR spectroscopy in order to estimate the binding
constants (KS and KR) per glucose unit for (S)- and (R)-2 and
the thermodynamic parameters ∆G°, ∆H°, and ∆S° for the
complexation; ∆G° is the change in total free energy, and ∆H°
and ∆S° represent the association changes in enthalpy and
entropy, respectively. The ratio of each binding constant
obtained at different temperatures can lead to ∆∆G°, ∆∆H°,
and ∆∆S° values through standard relations (eq 2)
where ∆δ and ∆δmax are the observed and calculated maximum
chemical shift changes, respectively. All data were used for
calculating ∆δmax and KS in eq 4, while the concentrations of 1
and 2 were chosen so as to meet the Benesi-Hildebrand
conditions (eq 3; [1]t/[2]t g 10).36a The chemical shift changes
in OH and H4 resonances of (S)-2 gave KS ) 15.5 and 15.7
M-1, respectively, for eq 3, which are in reasonably consistent
agreement with KS ) 17.1 and 18.0 M-1, respectively, for eq
4. In the former case, plots of 1/∆δ versus 1/[1]t led to a linear
relation with a correlation coefficient r > 0.992, and the standard
deviations of the calculated values from the experimental ones
in the curve fitting using eq 4 were below 10%. The KS and
∆δmax values obtained at different temperatures are listed in
Table 5.
The 1:1 complexation was also confirmed by the continuous
variation plot (Job plot) for the 1-(S)-2 complex (Figure 6) in
which the total concentrations of 1 and (S)-2 were kept constant
at 25 mM. The maximal complex formation occurred at around
0.5 mol fraction of 1. This result implies that each glucose
unit of the polymer 1 has the same binding affinity to (S)-2,
which may be attributed to the regular structure of 1 even in a
solution. If the polymer had a random coil conformation in
solution, there might exist a number of binding sites interacting
with a guest in different ways. This would lead to a different
complex from the 1:1 complex. The regular structure of 1 must
be responsible for the efficient chiral recognition capability in
NMR as well as in HPLC.
∆∆G° ) -RT ln(KS/KR) ) ∆∆H° - T∆∆S°
(2)
where the terms ∆∆G°, ∆∆H°, and ∆∆S° represent the
differences in the total free energy, enthalpy, and entropy,
respectively.
The 1H NMR titrations were carried out under the conditions
of constant [2] with varying [1] at 23, 30, 40, 50, and 60 °C in
CDCl3. Addition of 1 to a solution of (S)-2 caused downfield
shifts of the OH resonances, while the aromatic H4 and H6
resonances showed upfield shifts. Downfield shifts are expected
for the proton involved in a hydrogen bonding, and upfield shifts
are often seen for the aromatic residues which are π-stacked.
Figure 5 shows the titration curves between (S)-2 and 1 at 23
°C in CDCl3, where negative values indicate an upfield shift.
The titration curves in the figure were analyzed by both linear
(eq 3: method A)36 and nonlinear (eq 4: method B)37 least-
square methods using a binding model with a 1:1 polymer (one
(37) (a) Diederich, F. Angew. Chem., Int. Ed. Engl. 1988, 27, 362-386.
(b) Takai, Y.; Okumura, Y.; Takahashi, S.; Sawada, M.; Kawamura, M.;
Uchiyama, T. J. Chem. Soc., Chem. Commun. 1993, 53-54. (c) Sawada,
M.; Okumura, Y.; Shizuma, M.; Takai, Y.; Hidaka, Y.; Yamada, H.; Tanaka,
T.; Kaneda, T.; Hirose, K.; Misumi, S.; Takahashi, S. J. Am. Chem. Soc.
1993, 115, 7381-7388. (d) Albert, J. S.; Goodman, M. S.; Hamilton, A.
D. J. Am. Chem. Soc. 1995, 117, 114-1144. (e) Sawada, M.; Takai, Y.;
Yamada, H,; Hirayama, S.; Kaneda, T.; Tanaka, T.; Kamada, K.; Mizooka,
T.; Takeuchi, S.; Ueno, K.; Hirose, K.; Tobe, Y.; Naemura, K. J. Am. Chem.
Soc. 1995, 117, 7726-7736.
(36) (a) Kobayashi, K.; Asakawa, Y.; Kato, Y.; Aoyama, Y. J. Am. Chem.
Soc. 1992, 114, 10307-10313. (b) Kelly, T. R.; Kim, M. H. J. Am. Chem.
Soc. 1994, 116, 7072-7080.

NMR Studies of Chiral Discrimination
J. Am. Chem. Soc., Vol. 118, No. 17, 1996 4043
Table 5. Binding Constants (KS) for the Complexation of 1 with
(S)-2 and Saturation Shifts (∆δmax) for the Aromatic H4 and OH of
(S)-2a
method Ab
method Bc
∆δmax (ppm)d
T (°C)
23
KS (M-1 ∆δmax (ppm)d KS (M-1
)
)
H4
OH
H4
OH
H4
OH
H4
OH
H4
OH
15.7
15.5
15.9
12.4
10.1
8.5
6.2
4.1
4.9
3.0
-0.257
1.32
-0.218
1.34
-0.200
1.28
-0.196
1.66
18.0
17.1
17.1
14.0
11.4
9.5
7.1
6.6
5.9
4.8
-0.241
1.27
-0.217
1.27
-0.190
1.21
-0.180
1.16
30
40
50
-0.164
1.59
-0.144
1.09
60
a [(S)-2] ) 1.94 mM and [1] ) 0-63.2 mM in CDCl3. b Estimated
by the Benesi-Hildebrand analysis (eq 3 in the text). c Estimated by
the nonlinear least-squares method (eq 4 in the text). d Negative value
indicates an upfield shift.
Figure 7. Changes of glucose protons resonances (H1-H6) of 1 (18.0
mM) in the absence (A) and presence of (S)-2 (2.5 (B), 5.0 (C), 15
(D), and 30 mg (E)) in CDCl3 at 23 °C. The marks (× and *) denote
the impurity and the OH proton resonances of (S)-2, respectively.
The plots for the Ks values obtained by methods A and B
for the OH proton resonances in Table 5 gave ∆H°S ) -9.2 (
0.8 and -6.9 ( 0.2 kcal mol-1 and ∆S°S ) -25.5 ( 2.6 and
-17.6 ( 0.6 cal mol-1 K-1, respectively. The enthalpy factor
attributed to hydrogen bonding and π-stacking and/or CH-π
interaction overcomes the entropy loss due to the restriction of
mobility through hydrogen bonding in the binding process.
Titrations of 1 with (S)- and (R)-2 were also performed so as
to obtain the information with respect to binding sites of 1 in
the complexation. Figure 7 shows the 1H NMR spectra (glucose
protons (H1-H6) region) of 1 in the absence and presence of
(S)-2. The H2 proton resonance was dramatically affected by
(S)-2 and shifted upfield with binding, while the other glucose
proton resonances slightly moved. The significant upfield shifts
of the H2 proton resonance indicates that a naphthyl ring of
(S)-2 may be closely located above the H2 proton so that it can
substantially affect the ring current effect. The chemical shifts
changes of 1 against the concentrations of (S)- and (R)-2 are
plotted (Figure 8). The chemical shift movement of the glucose
protons induced by (R)-2 was relatively small.
The 19F NMR chemical shifts of the fluoro groups on the
phenyl residues of 1 were analogously altered by the complex-
ation with (S)-2, while (R)-2 hardly changed the chemical shifts
(Figure 9). These different chemical shift changes of 1 against
(S)- and (R)-2 also support the attractive interaction between
(S)-2 and 1.
Comparison of NMR and HPLC Results. It is particularly
interesting to compare the enantioselectivity and thermodynamic
parameters measured in solution by the 1H NMR titrations
described above with those obtained by chiral HPLC for the
binding of 2 with 1. Recently, it has been reported that in achiral
host-guest complexation systems the retention enthalpies (∆H°)
determined by HPLC using a stationary phase consisting of a
Figure 6. Job plot of (S)-2 with 1 in CDCl3 at 23 °C. The total
concentration is 25 mM, and the OH proton resonance of (S)-2 is
analyzed.
The titration experiments were carried out for (R)-2 with 1
under the same conditions as above. As expected from the 1D
and 2D NMR experiments combined with dynamics data, the
binding constant of (R)-2 to 1 was very small (1.5 M-1) at 23
°C. At higher temperatures (g30 °C) the complexation-induced
shifts (∆δ) were so small (<0.032 ppm even at [1]t/[(R)-2] )
33) that we could not determine reliable KR values (e1 M-1
)
1
by H NMR titrations. The binding constant KR at 23 °C was
evaluated by the Benesi-Hildebrand analysis for only H4 shifts
of the guest (referring to eq 3) because the labile OH proton
resonance disappeared after a few hours upon addition of 1 at
23 °C and after a few minutes at higher temperatures (i.e., 50
°C) even with a freshly prepared sample. However, the OH
protons of (S)-2 apparently resonated in the presence of 1 even
at high [1]t/[2] and at higher temperature ranges (30-60 °C).
These results suggest that the OH protons of (S)-2 may not
exchange with H2O in a CDCl3 solution because of tight binding
to 1 through hydrogen bonding, while those of (R)-2 exchanged
rapidly.38
(38) We used dry CDCl3 and samples in all NMR measurements, but
H2O bound to the polymer 1 (so called “bound water”) could not be removed
even after 1 was dried over P2O5 in vacuo at 100 °C overnight. The polymer
1 (a repeated glucose unit) was found to contain an almost equimolar amount
of bound water from the integral ratio of the 1H NMR spectrum. Direct
evidence of the effect of water to binding affinity (K) has not yet been
obtained. The effects of water on hydrogen bond based molecular
recognition of neutral substrates in chloroform, see: Adrian, Jr. J. C.;
Wilcox, C. S. J. Am. Chem. Soc. 1992, 114, 1398-1403.
The thermodynamic parameters (∆H°S and ∆S°S) for the
complexation of 1-(S)-2 in solution were estimated from the
linear plot (van’t Hoff plot) of ln KS versus 1/T using eq 5.
ln KS ) -∆H°/RT + ∆S°/R
(5)

4044 J. Am. Chem. Soc., Vol. 118, No. 17, 1996
Yashima et al.
Figure 8. Plots of chemical shifts of glucose protons (H1-H6) and methyl proton resonances of 1 versus [(S)-2/1] (A) and [(R)-2/1] (B) in CDCl3
at 23 °C. The conditions are identical to those shown in Figure 7.
1
From the binding constants, KS and KR obtained by the H
NMR titrations, the enantioselectivity (R) and the difference in
free energy (∆∆G°) upon diastereomeric complexation in
solution at 23 °C are calculated to be ca. 10.6 and 1.39 kcal
mol-1, respectively, using eqs 1 and 2. These values can be
compared with those estimated by chiral HPLC; R ) k2′/k1′ )
4.23 and ∆∆G° ) 0.84 kcal mol-1 at 20 °C. The former R
value (10.6) is ca. 2.5 times larger than that estimated by the
chiral HPLC. The difference may be derived from the use of
different solvents5b,i,15b or unfavorable, nonenantioselective
interactions of 2 with the remaining silanol on the surface of
silica gel in the chiral HPLC.41 The chromatographic enanti-
oseparation was carried out using a mixture of hexane-2-
1
propanol, whereas CDCl3 was used for the H NMR experi-
ments. The importance of hydrogen bond association between
1 and 2 for chiral discrimination has been proved by the chiral
1
HPLC and H NMR experiments. Consequently, 2-propanol
appears to disturb such an association. Therefore, in the chiral
HPLC the separation factor (R) increased as a decrease in the
content of 2-propanol in the mobile phase; R ) 3.22 (mobile
phase, 30% 2-propanol in hexane), R ) 4.01 (20% 2-propanol),
and R ) 4.23 (10% 2-propanol). The influence of 2-propanol
on enantioselectivity in solution was also confirmed by the 1H
NMR titrations. When 2-propanol was added to the mixture
of 1 and (RS)-2 in CDCl3 under conditions identical to those in
Figure 2, ∆∆δ for the OH and other aromatic proton resonances
Figure 9. 19F NMR titrations of 1 (18.1 mM) with (S)-2 (A) and (R)-2
(B) (0 (a), 5.0 (b), 15 (c), and 25 mg (d)) in CDCl3 at 23 °C. R,R,R-
Trifluorotoluene (-64.0 ppm) was used as the internal standard.
significantly decreased as an increase in the amount of 2-pro-
(40) Pirkle et al. have extensively studied chiral recognition in small
bimolecular systems relevant to the chiral HPLC by 1H NMR. They
estimated the binding constant (K) and thermodynamic parameters (∆G°,
∆H°, and ∆S°) for the more stable complex by 1H NMR titrations, although
those for the less stable complex could not be determined, probably due to
very weak binding.5c Very recently, Pinkerton et al. compared the ∆∆G°
values estimated by chiral HPLC separation of a racemic drug (U-80413)
with the third domain turkey ovomucoid chemically bonded silica gel as
the CSP and by 1H NMR titrations of the enantiomers using the protein.15b
(41) (a) Pirkle, W. H.; Hyun, M. H. J. Chromatogr. 1985, 328, 1-9. (b)
Dobashi, Y.; Hara, S. J. Org. Chem. 1987, 52, 2490-2496. Other factors
such as loading amount of 1 on silica surface may influence the enanti-
oselectivity.
host chemically bonded to silica gel are analogous to the
complexation enthalpies measured in solution by 1H NMR
titrations.39 However, most previous studies concerning the
chiral recognition mechanism on chiral stationary phases did
not deal with the comparison of ∆∆G° obtained by HPLC and
NMR.40
(39) Zimmerman, S. C.; Saionz, K. W. J. Am. Chem. Soc. 1995, 117,
1175-1176.

NMR Studies of Chiral Discrimination
J. Am. Chem. Soc., Vol. 118, No. 17, 1996 4045
panol added (∆∆δOH ) 0.021 and ∆∆δH4 ) 0.020 ppm upon
addition of 90 µL of 2-propanol).42 This observation is
consistent with the HPLC data. In other words, if the
chromatographic experiment is carried out using chloroform as
an eluent component, the enantioselectivity may be improved.43
Since KR at higher temperature ranges could not be deter-
mined due to very weak binding of (R)-2 with 1 (KR < 1 M-1),
1
we could not estimate the ∆∆H° and ∆∆S° values by the H
NMR titrations. However, these parameters can be readily
obtained by the chiral HPLC method at a temperature range of
20 e T e 60 °C through van’t Hoff plots using eq 644
Figure 10. Stereoviews of the energy-minimized structure of (S)-2.
ln R ) -∆∆H°/RT + ∆∆S°/R + ln(mS,T/mR,T
)
(6)
minimization (see the Experimental Section). The energy-
minimized (S)-2 (Figure 10) was manually placed in the groove
of the main chain so that all of the NMR data including the
intermolecular NOEs and the titration results as well as
intermolecular hydrogen bonds are visually satisfied. The
complex was further energy minimized to relieve unfavorable
van der Waals contacts. The resulting complex had a negative
nonbonded energy, indicating the attractive interaction between
1 and (S)-2 and the absence of any unfavorable contacts.
Figure 11 shows the lowest energy structure of the 1-(S)-2
complex. The polymer possesses a left-handed 3/2 helical
conformation, and the glucose residues are regularly arranged
along the helical axis. A chiral helical groove, or ditch, with
polar carbamate residues exists along the main chain. The polar
carbamate groups are located inside, and hydrophobic aromatic
groups are placed outside the polymer chain so that polar
enantiomers can get in the groove to interact with the carbamate
residues Via hydrogen bonding formation. This interaction
seems to be important for efficient chiral discrimination.14,18
The location of (S)-2 is best illustrated in Figure 11B, where
two OH protons of (S)-2 form hydrogen bonding with the
carbonyl oxygens of the carbamate groups at the 2- and
3-positions on two different glucose units (marked by broken
line). The distances between the hydrogens and oxygens are
1.97 and 2.24 Å, which are short enough for hydrogen bonding.
As shown in Figure 11C, the naphthyl ring protons (H4, H6,
and H7) are located close to the methyl protons on the phenyl
groups of 1 (CH-π interaction); the distances are 3.51, 4.69,
and 3.22 Å, respectively. These results are consistent with the
intermolecular NOE data shown in Figure 4. The π-π
interaction between the naphthyl and phenyl rings at the
3-position of a glucose residue may also support the interaction
shown in Figure 11C where the distance between the rings is
ca. 3-3.5 Å. This π-π interaction appears to contribute to
the upfield shifts of H4 and H6 proton resonances of 2.
The interaction model also explains the upfield shifts of the
H2 proton of the glucose residue as illustrated in Figure 11D;
the naphthyl ring is favorably positioned above the H2 proton.
Moreover, the difficulty of chiral discrimination for 10,10′-
dihydroxy-9,9′-biphenanthryl (5) is reasonably explained by
using the model. Bulky aromatic rings at around the OH groups
of 5 may prevent the formation of hydrogen bond.
where mS,T and mR,T are the binding capacities for R and S
enantiomers at the temperature being examined. If the ratio
mS,T/mR,T is constant with temperature, a plot of ln R versus
1/T should be linear with a slope of -∆∆H°/R and an intercept
of ∆∆S°/R, but if the ratio changes with temperature, the plot
of ln R versus 1/T should be nonlinear.
The obtained thermodynamic parameters from the linear plots
are ∆∆H° ) -2.2 ( 0.1 kcal mol-1 and ∆∆S° ) -4.9 ( 0.3
cal mol-1 K-1 for 2 and ∆∆H° ) -1.3 ( 0.1 kcal mol-1 and
∆∆S° ) -2.3 ( 0.5 cal mol-1 K-1 for 4 with hexane-2-
propanol (90/10) as the eluent.45
Molecular Modeling of the 1-(S)-2 Complex. The HPLC
and NMR experiments demonstrate that (S)-2 comes in a chiral
groove of 1 directed toward the H2 proton of the glucose through
intermolecular hydrogen bond between the OH protons and
probably the carbonyl oxygens of 1. Because of the observation
of a few intermolecular NOEs for the 1-(S)-2 complex and
the uncertainty of the structure of 1, a precise model for the
complex cannot be drawn. However, a model of the complex
can be proposed by using the HPLC and NMR data combined
with the structural data of cellulose trisphenylcarbamate (CTPC)
determined by X-ray analysis30 as previously described.
The initial model of 1 was constructed using three-
dimensional periodic boundary conditions in CERIUS2 starting
from the CTPC, which is reported to have a left-handed 3-fold
(3/2) helical structure, and then was energy-minimized by a
Dreiding force field.46 The initial structure of (R)- and (S)-2
was taken from the crystal structure reported for (RS)-247 before
(42) The negative influence of 2-propanol on chiral discrimination in
1H NMR was also observed in a similar CDCl3-soluble CTSP-enantiomer
system; the splitting methine proton resonances of trans-2,3-diphenyloxirane
due to the enantiomers in CDCl3 in the presence of CTSP disappeared by
the addition of 15 µL of 2-propanol.19
(43) Chloroform cannot be used as the mobile phase on the present chiral
stationary phase consisting of 1 coated on silica gel because of solubility
problem of 1 in chloroform. The chemically bonded-type CSPs of 3,5-
dimethylphenylcarbamates of cellulose and amylose can be used with
chloroform as mobile phase component, and a few racemates which were
not or poorly resolved on the coated-type CSPs were more efficiently
resolved on the chemically bonded-type CSPs with chloroform as a mobile
phase component (i.e., hexane-chloroform (95/5)).14o The influence of
chloroform as a mobile phase component on the chromatographic enanti-
oseparation has also been compared with the 1H NMR data in chiral
discrimination using CDCl3 as solvent, see ref 5b,i.
(44) (a) Pirkle, W. H.; Readnour, R. S. Anal. Chem. 1991, 63, 16-20.
(b) Loun, B.; Hage, D. S. Anal. Chem. 1994, 66, 3814-3822.
(45) When the enantioseparation of 2 and 4 on 1 was performed at high
temperatures (30-60 °C), the resolving power of 1 decreased gradually.
The R values of 2 and 4 at 20 °C were reduced to 3.81 and 3.00, respectively.
However, van’t Hoff plots (eq 6) for 2 and 4 gave an almost straight line
(see the supporting information).
Although the chromatographic retention behavior for (R)-2
in HPLC and the NMR data, for instance, the downfield shift
of the OH proton resonances in the presence of 1, suggest the
existence of weak hydrogen bond interaction between (R)-2 and
1, it is difficult to propose an analogous model for the 1-(R)-2
complex because of lack of the NOE data.
(46) (a) Mayo, S. L.; Olafson, B. D.; Goddard, W. A., III. J. Phys. Chem.
1990, 94, 8897-8909. (b) Rappe´, A. K.; Goddard, W. A., III. J. Phys.
Chem. 1991,95, 3358-3363. (c) Castonguay, L. A.; Rappe´, A. K.; Casewit,
C. J. J. Am. Chem. Soc. 1991, 113, 7177-7183. (d) Wang, Q.; Stidham, H.
Spectrochim. Acta 1994, 50A, 421-433. (e) Smith, T. L.; Masilamani, D.;
Bui, L. K.; Khanna, Y. P.; Bray, R. G.; Hammond, W. B.; Curran, S.; Belles,
J. J., Jr.; Binder-Castelli, S. Macromolecules 1994, 27, 3147-3155.
When (R)-2 was placed into the groove of 1 in the same way
as (S)-2, the simulation calculation indicated that the (R)-2
(47) Mori, K.; Masuda, Y.; Kashino, S. Acta Crystallogr. 1993, C49,
1224-1227.

4046 J. Am. Chem. Soc., Vol. 118, No. 17, 1996
Yashima et al.
Figure 11. Computer-generated depiction of the complex of 1-(S)-2. The (S)-2 is shown in yellow. Dashed lines correspond to hydrogen bonds.
(A) Configuration: along the helix axis; (B) perpendicular to the chain axis. (C and D) Expanded region of the same structural model showing the
interactions of (S)-2 with 1.
molecule could not exist in such a way that both the two OH
groups can form simultaneous hydrogen bonding to the carbonyl
oxygens of the carbamate groups. Only one OH group can form
a hydrogen bond. This supports the experimental evidences
that (R)-2 forms weaker hydrogen bond with 1 than (S)-2 in
the HPLC and NMR experiments.
Conclusions. (S)-1,1′-Bi-2-naphthol (2) site-selectively binds
to the phenylcarbamoylated cellulose derivative 1 through
multiple interactions including intermolecular hydrogen bonding
and π-π and/or CH-π interactions to afford a 1:1 complex.
The molecular modeling on the basis of the chromatography
and NMR data reveals the chiral discrimination rationale for a
cellulose phenylcarbamate derivative. The enantioselectivities
(R) and the thermodynamic parameters ∆H°, ∆S°, and ∆G° for
the more stable complex of 1-(S)-2 and the difference in free
energy (∆∆G°) in chiral discrimination process were separately
1
determined by H NMR titrations in solution and by chiral
HPLC. These results should provide useful information both
for understanding the chiral discrimination mechanism of the
other polysaccharide-based CSPs and for designing the more
excellent CSPs.
Experimental Section
Materials. Cellulose (Avicel) was purchased from Merck. The
degree of polymerization of the cellulose was estimated to be ca. 200
as its tribenzoate ester by gel permeation chromatography using THF

NMR Studies of Chiral Discrimination
J. Am. Chem. Soc., Vol. 118, No. 17, 1996 4047
as the eluent. 4-Fluoro-2-nitrotoluene, triphosgene, and (3-aminopro-
pyl)triethoxysilane were of guaranteed reagent grade from Tokyo Kasei.
Pyridine-d5 (99 atom % D) was purchased from Aldrich. Porous
spherical silica gel (Daiso gel SP-1000) with a mean particle size of 7
µm and a mean pore diameter of 100 nm was kindly supplied by Daiso
Chemical. All solvents used in the preparation of CSPs were of
analytical reagent grade, carefully dried, and distilled before use.
Solvents used in chromatographic experiments were of HPLC grade.
CDCl3 (99.8 atom % D, Nacalai) was dried over molecular sieves 4A
(Nacalai) and stored under nitrogen.
compound (7.8-14.0 mM) was injected into the chromatographic
system (20-100 µL) using a Rheodyne Model 7125 injector with a
1.0 mL loop. One-dimensional 1H, 13C, and 19F NMR spectra and 2-D
1H-1H COSY and NOESY spectra were recorded on a Varian VXR-
1
500S spectrometer operating at 500 MHz for H, 125 MHz for 13C,
and 470 MHz for 19F. The 2-D 1H-13C COSY spectrum was obtained
on a Varian Gemini 200 spectrometer. All NMR spectra were measured
in CDCl3 unless specified otherwise. Chemical shifts were reported
in parts per million (ppm) with tetramethylsilane (TMS, 0 ppm), CDCl3
(77.0 ppm), and R,R,R-trifluorotoluene (-64.0 ppm) as the internal
1
standard for H,13C, and 19F NMR, respectively.
(()-1,1′-Bi-2-naphthol (2) and (S)-(-)-2 ([R]25D -32°, c 1.5 g dL-1
,
1H NMR Titration. The 1H NMR titration experiments were
performed under two conditions. The concentration of 1 was calculated
based on glucose units. Condition 1. (S)- or (R)-2 was maintained at
constant concentration in the presence of increasing concentrations of
1 to measure binding constants (KS and KR). Stock solutions (1.94
mM) of (S)- and (R)-2 in CDCl3 were prepared. A 0.9 mL aliquot of
the stock solution was added using a hypodermic syringe to eight
separate vials containing 2.5, 5.0, 10, 15, 20, 25, 30, and 35 mg of 1,
respectively, and the solutions were transferred to eight 5-mm NMR
THF) were purchased from Tokyo Kasei. (R)-(+)-2 ([R]25 +32°, c
D
1.5 g dL-1, THF) was obtained from Kankyo Kagaku Center.
2-Hydroxy-2′-methoxy-1,1′-binaphthyl (3a) and 2,2′-dimethoxy-1,1′-
binaphthyl (3b) were prepared by alkylation of 2 with methyl iodide
in acetone in the presence of K2CO3.48 Racemic and (R)-(+)-
2,2′-dihydroxy-6,6′-dimethylbiphenyl (4) ([R]18 +91.5°, c 1.0 g
D
dL-1, ethanol)49 were gifts from Dr. Kanoh of Kanazawa University.
10,10′-Dihydroxy-9,9′-biphenanthryl (5)50 was provided by Professor
Yamamoto of University of Osaka Prefecture. Ee (>99%) of the
enantiomers of 2 and 4 was checked by HPLC using a chiral column
consisting of 1 as a CSP. 5-Fluoro-2-methylaniline was prepared by
reduction of 4-fluoro-2-nitrotoluene with SnCl2‚2H2O and hydrochloric
acid (35%) in ethanol.
1
tubes. After the tubes were sealed, H NMR spectra were taken for
each tube at 23, 30, 40, 50, and 60 °C, and ∆δ values were calculated
by subtracting the chemical shifts in the spectrum of the mixture from
the corresponding resonance of pure 2. Then, binding constants were
determined by using either eqs 3 and 4. Satisfactory fits were observed
in all cases for a 1:1 complexation.
Synthesis of Cellulose Tris(5-fluoro-2-methylphenylcarbamate)
(1). 5-Fluoro-2-methylaniline (17.0 g, 0.14 mol) was allowed to react
with triphosgene (17.8 g, 60 mmol) in dry toluene (380 mL) in the
presence of a catalytic amount of dry pyridine (1.0 mL). After
evaporation of the solvent, the residue was distilled under reduced
pressure (bp 85 °C/28 mmHg) to give 5-fluoro-2-methylphenyl isocy-
anate (14.4 g, 70%). Cellulose tris(5-fluoro-2-methylphenylcarbamate)
(1) was prepared according to the previously described procedure14 by
the reaction of cellulose (1.0 g) with a large excess of 5-fluoro-2-
methylphenyl isocyanate (6.0 g, 40 mmol) in dry pyridine (20 mL) at
ca. 80 °C for 40 h.21 The phenylcarbamoylated cellulose derivative
obtained was isolated as a methanol-water (5:1, v/v) insoluble fraction.
Condition 2. The cellulose derivative 1 was maintained at constant
concentration in the presence of increasing concentrations of (S)- or
(R)-2 to obtain information with respect to binding sites of 1 in the
complexation. A 18.0 mM solution of 1 in CDCl3 was prepared in a
5-mm NMR tube, and the initial NMR spectrum was recorded. To
this was directly added (S)- or (R)-2 (2.5, 2.5, 10, and 15 mg,
respectively), and NMR spectra were taken for each addition of 2.
Job Plot. The stoichiometry of the complex between 1 and (S)-2
was determined by the continuous variation plot (Job plot).51 Stock
solutions of 1 and (S)-2 in CDCl3 were prepared (25 mM). In eight
NMR tubes, portions of the two solutions were added in such a way
that their ratio changed from 0 to 1, keeping the total volume to be 0.8
mL. The 1H NMR spectra for each tube were taken at 23 °C, and the
change in chemical shift of the OH proton resonance of (S)-2 was used
to calculate the complex concentration using ∆δmax ) 1.274 ppm (see
Table 5). The complex concentration was plotted against the mole
fraction of 1 to give the Job plot shown in Figure 6.
1
Elemental analysis and H NMR data showed that hydroxy groups of
cellulose were almost quantitatively converted into the carbamate
moieties. IR (KBr): 3442, 3330 (νNH), 1742 (νCdO). 1H NMR
(pyridine-d5, 80 °C, TMS): δ 2.02, 2.04, 2.14 (s, CH3, 9H), 3.80, 3.97,
4.58, 4.71, 4.85, 5.13, 5.50 (br, glucose protons, 7H), 6.52, 6.68-6.82,
6.91, 6.99, 7.49, 7.67, 7.73 (aromatic, 9H), 8.30, 8.72, 9.24 (br,
NH, 3H). [R]25
:
-15° (c 1.0 g dL-1, THF). Anal. Calcd for
D
Spin-Lattice Relaxation Time (T1). T1 values were measured
without degassing by the inversion-recovery method and calculated by
standard programs supplied by Varian. Twelve different τ delays
varying from 0.013 to 25.6 s between 180° and 90° pulses with 20 s
pulse delay and number of scans of 20 were used for 1H T1
measurements. T1s of carbon resonances were measured separately
for large (3 < T1 < 12 s) and small (T1 e 1 s) values; pulse delay of
55.2 s with eight different τ (0.625 e τ e 80.0) and 76 scans for the
long T1s, and 6.2 s (0.125 e τ e 8.0) and 640 times for the short T1s
of free 2, 36.8 s (0.275 e τ e 35.2) and 552 times for the long T1s,
and 4.2 s (0.094 e τ e 6.0) and 3288 times for the short T1s of (S)-
(-)-2 in the presence of 1, and 47.2 s (0.375 e τ e 48.0) and 268
times for the long T1s and 5.2 s (0.088 e τ e 5.6) and 708 times for
the short T1s of (R)-(+)-2 in the presence of 1.
2D NMR. NOESY experiments were recorded in the phase-sensitive
mode at 30 °C without degassing. The NOESY spectra were collected
into 1024 complex points for 256 t1 increments with spectral widths of
4000 and 5000 Hz for free 1 and (S)- or (R)-2 in the presence of 1,
respectively, in both dimensions at mixing times of 60-500 ms. The
NOESY spectrum of free 2 was recorded in a similar way with a
spectral width of 800 Hz at mixing time of 700 ms. The data matrix
was zero filled to 1024 and apodized with a Gaussian function and
Fourier transformed in both dimensions. 1H-13C COSY spectrum of
free 2 was collected by using a 512 × 128 data matrix size at room
temperature. Spectral widths were 2733 and 376.4 Hz for 13C and 1H,
respectively.
(C30H28O8N3F3)n: C, 58.54; H, 4.58; N, 6.83. Found: C, 58.15; H,
4.30; N, 6.90;
Preparation of the Chiral Stationary Phase. A column packing
material was prepared as described previously14 using macroporous
silica gel which had been treated with a large excess of (3-aminopropyl)-
triethoxysilane in dry benzene in the presence of a catalytic amount of
dry pyridine at 80 °C overnight. The cellulose derivative 1 (0.75 g)
was dissolved in THF (10 mL), and the silanized silica gel (3.0 g) was
wetted with the polymer solution as uniformly as possible. Then, the
solvent was evaporated under reduced pressure. The remaining polymer
solution was adsorbed on the silica gel using the same procedure. The
packing material thus obtained was packed into a stainless-steel tube
(25 × 0.46 cm i.d.) by a conventional high-pressure slurry packing
technique using a Model CCP-085 Econo packer pump (Chemco). The
plate number of the column was 5800 for benzene with hexane-2-
propanol (90/10) as the eluent at a flow rate of 0.5 mL min-1. 1,3,5-
Tri-tert-butylbenzene was used as a nonretained compound for esti-
mating the dead time (t0).22
Instruments. Chromatographic experiments were performed on a
JASCO PU-980 chromatograph equipped with a UV (JASCO 875-
UV) and a polarimetric (JASCO DIP-181C, Hg without filter) detectors.
The column temperature (20, 30, 40, 50, and 60 °C) was controlled
with a water jacket. A solution of a racemate or an optically active
(48) Galli, B.; Gasparrini, F.; Misiti, D.; Pierini, M.; Villani, C.; Bronzetti,
M. Chirality 1992, 4, 384-388.
(49) Kanoh, S.; Tamura, N.; Motoi, M.; Suda, H. Bull. Chem. Soc. Jpn.
1987, 60, 2307-2309.
Molecular Modeling. Molecular modeling and molecular mechan-
ics calculation were performed with the Dreiding force field (version
(50) Yamamoto, K.; Fukushima, H.; Okamoto, Y.; Hatada, K.; Nakazaki,
M. J. Chem. Soc., Chem. Commun. 1984, 1111-1112.
(51) Job, P. Ann. Chim. Ser. 1928, 9, 113-134.

4048 J. Am. Chem. Soc., Vol. 118, No. 17, 1996
Yashima et al.
2.11)46 as implemented in CERIUS2 software (version 1.5, Molecular
Simulations Inc., Burlington, MA) running on an Indigo2-Extreme
graphics workstation (Silicon Graphics). Charges on atoms of 1 and
2 were calculated using Gasteiger in QUANTA (version 4.0, Molecular
Simulations Inc.) and QEq46 in CERIUS2, respectively; total charges
of the molecules were zero. The polymer model of 1 was constructed
using the crystalline structure of cellulose trisphenylcarbamate (CTPC)30
according to the previously reported method with a modification.18,52
First, a full energy minimization of a repeating unit of 1 containing
CH3O groups at the 1- and 4-positions of a glucose unit was performed
by using Conformational Search in CERIUS2. The energy minimization
was accomplished first by Conjugate Gradient 200 (CG 200) and then
by Fletcher Powell (FP) until the root mean square (rms) value became
less than 0.01 kcal mol-1 Å-1, respectively. Next, the monomeric unit
of 1 was allowed to construct a trimer with a left-handed 3-fold (3/2)
helix by Polymer Builder in CERIUS2 according to the structure of
CTPC.30 The trimer was placed into a simulation cell (x ) 30, y )
30, and z ) 15.166 Å) using three-dimensional periodic boundary
conditions by Crystal Builder in CERIUS2. The unit cell volume was
expanded to the directions perpendicular to the polymer axis (z) to avoid
interactions of the periodic polymer with neighboring ones in other
cells. The energy minimization of the periodic structure was then
performed by CG 200 and FP until the rms value became less than
0.01 kcal mol-1 Å-1, respectively. The resulting optimized trimer in
the unit cell was connected to give a nanomer (9mer) as the model
polymer of 1 in Figure 11 before placing the (S)-2 or (R)-2 on the
interaction site.
H, and C1′-C2′-O2′-H for the crystal structure are 88.8, 8.8, and
-6.6°, respectively, and 97.9, -0.7, and -0.8° for the optimized
structure, respectively (see Figure 10). The optimized (S)-2 was
manually placed into the interaction site of 1 so that all of the NMR
1
data including intermolecular NOEs and H NMR titration results as
well as intermolecular hydrogen bonds were visually satisfied. The
complex was further energy minimized by CG200 and FP to relieve
unfavorable van der Waals contacts, while the geometry of 1 was fixed.
A similar procedure was done for (R)-2.
Acknowledgment. We thank Professor P. Zugenmaier
(Technischen Universita¨t Clausthal) for sending us a copy of
Riehl’s Ph.D. Thesis. We acknowledge Professor M. Sawada
and Dr. Y. Takai (Osaka University) for their kind cession of
1
their fitting programs for H NMR titrations. We are also
grateful to Professor T. Norisue and Dr. T. Sato (Osaka
University) for sending us a copy of their manuscript prior
publication and for fruitful discussion. In addition, we
thank Professor K. Yamamoto (University of Osaka Prefecture)
and Dr. S. Kanoh (Kanazawa University) for providing 5 and
4, respectively. This work was partially supported by
Grant-in-Aids for Scientific Research on Priority Areas No.
06242101 from the Ministry of Education, Science and Culture,
Japan.
Supporting Information Available: Copies of 2D NOESY
spectra of 1, 1-(S)-2, and 1-(R)-2, and van’t Hoff plots in the
enantioseparation of 2 and 4 at various temperature ranges (4
pages). This material is contained in many libraries on
microfiche, immediately follows this article in the microfilm
version of the journal, can be ordered from the ACS, and can
be downloaded from the Internet; see any current masthead page
for ordering information and Internet access instructions.
The initial coordinates of (S)- and (R)-2 were taken from the crystal
structure data of (RS)-247 in the Cambridge Structural Database 3D
Graphics Search System.53 The initial structure was further energy-
minimized with CG200 and FP using the Dreiding force field. The
typical geometric parameters of (RS)-2 before and after minimization
are as follows; the torsion angles of C2-C1-C1′-C2′, C1-C2-O1-
(52) Yashima, E.; Noguchi, J.; Okamoto, Y. Macromolecules 1995, 28,
8368-8374.
(53) Allen, F. H.; Kennard, O. Chem. Des. Autom. News 1993, 8, 31-
37.
JA960050X