Effect of multivalency on the performance of enantioselective separation media for chiral HPLC prepared by linking multiple selectors to a porous polymer support via aliphatic dendrons

Article abstract of DOI:10.1021/jo011005x

Chiral stationary phases (CSPs) containing L-proline indananilide chiral selectors attached through a multivalent dendritic linker to monodisperse macroporous poly(2-hydroxyethyl methacrylate-co-ethylene dimethacrylate) beads have been prepared using two different approaches. The convergent method involves the preparation of ligands in solution and their subsequent attachment to the support. The divergent approach is based on the stepwise "on-bead" formation of the linker using methods that are typical of solid-phase synthesis. While the convergent CSPs feature well-defined ligands, their loading is relatively low. In contrast, the divergent technique affords CSPs with higher loading but with more limited control over precise ligand architecture. Excellent enantioselectivities characterized by separation factors of up to 31 were achieved for the separation of racemic N-(3,5-dinitrobenzoyl)-α-amino acid alkyl amides with these new CSPs under normal-phase HPLC conditions.

Full text of DOI:10.1021/jo011005x

J . Org. Chem. 2002, 67, 1993-2002
1993
Effect of Mu ltiva len cy on th e P er for m a n ce of En a n tioselective
Sep a r a tion Med ia for Ch ir a l HP LC P r ep a r ed by Lin k in g Mu ltip le
Selector s to a P or ou s P olym er Su p p or t via Alip h a tic Den d r on s
Frank H. Ling, Victor Lu, Frantisek Svec, and J ean M. J . Fre´chet*
Department of Chemistry, University of California, Berkeley, California 94720-1460, and Division of
Materials Science, E. O. Lawrence Berkeley National Laboratories, Berkeley, California 94720
frechet@cchem.berkeley.edu
Received October 15, 2001
Chiral stationary phases (CSPs) containing L-proline indananilide chiral selectors attached through
a multivalent dendritic linker to monodisperse macroporous poly(2-hydroxyethyl methacrylate-co-
ethylene dimethacrylate) beads have been prepared using two different approaches. The convergent
method involves the preparation of ligands in solution and their subsequent attachment to the
support. The divergent approach is based on the stepwise “on-bead” formation of the linker using
methods that are typical of solid-phase synthesis. While the convergent CSPs feature well-defined
ligands, their loading is relatively low. In contrast, the divergent technique affords CSPs with higher
loading but with more limited control over precise ligand architecture. Excellent enantioselectivities
characterized by separation factors of up to 31 were achieved for the separation of racemic N-(3,5-
dinitrobenzoyl)-R-amino acid alkyl amides with these new CSPs under normal-phase HPLC
conditions.
In tr od u ction
process monitoring, quality control, as well as for the
“polishing” of optical purity of products prepared using
large scale asymmetric synthesis, current goals in chiral
chromatography include both highly efficient analyses
optimized for rapid, real time, monitoring and control,
as well as large scale downstream processing of both
racemic and enantiomerically enriched products.
Following safety and regulatory suggestions, the vast
majority of new chiral drugs are now administered as
pure single enantiomers. Therefore, the development and
manufacturing of single enantiomer drugs continues to
be a prime target of the pharmaceutical industry.1 It is
well-known today that the enantiomers of a single
compound may differ considerably in properties such as
pharmacodynamics, pharmacokinetics, potency, toxicity,
and metabolism.2 Consequently, the testing of both
enantiomers of new chiral drugs is made by drug
companies and required by regulatory agencies for drugs
administered as a racemate.3
Synthesis from chiral starting materials, asymmetric
synthesis, and chromatographic separations are currently
the methods most often used to obtain enantiomerically
pure compounds.3a,4 Liquid chromatography is typically
carried out in columns packed with a chiral stationary
phase (CSP) that can be used in the analytical, prepara-
tive, and even the production scales. For example, the
recently implemented simulated moving bed preparative
liquid chromatography process enables the annual pro-
duction of up to 50 metric tons of pure enantiomers.5
Since chromatographic separations are also used for
By and large, the separation of racemates into pure
enantiomers is a challenging task because the physical
and chemical properties of both optical isomers are
identical in an achiral environment. While the separation
of enantiomers using a solid support was first suggested
by Willsta¨tter almost a century ago,6 it was not until 1960
that the first chromatographic enantioseparations were
realized.7 The pioneering studies of Allenmark, Arm-
strong, Blaschke, Cram, Davankov, Hermansson, Oka-
moto, Pirkle, and others have led to a variety of station-
ary phases useful in the separation of enantiomers. A
considerable number of CSPs for liquid chromatography
have emerged during the last two decades and more than
a hundred of them are now being actively used in
industrial and academic laboratories.8
In addition to the intrinsic capability of the im-
mobilized chiral selector to differentiate between the two
enantiomers, the separation factor R or enantioselectivity
is affected by a number of other factors such as the
manner by which the selector is tethered to the support,
the length and chemistry of the tethering arm, the nature
of the underlying support, and the presence of variable
amounts of undesired nonspecific interacting sites. All
of these factors affect the overall chiral recognition
* To whom correspondence should be addressed. Phone: (510) 643-
3077. Fax: (510) 643-3079.
(1) Stinson, S. C. Chem. Eng. News 1998, 76(38), 83.
(2) Sheldon, R. A. Chirotechnology: Industrial Synthesis of Optically
Active Compounds; Marcel Dekker: New York, 1993.
(3) (a) Brown, J . R. In A Practical Approach to Chiral Separations
by Liquid Chromatography; Subramanian, G., Ed.; VCH: Weinheim,
1994. (b) Caldwell, J . J . Chromatogr. 1996, 719, 3. (c) Branch S. K. In
Chiral Separation Techniques. A Practical Approach; Subramanian,
G., Ed.; Wiley-VCH: New York, 2001.
(4) (a) J inno, K., Ed. Chromatographic Separations Based on Mo-
lecular Recognition; Wiley-VCH: New York, 1997. (b) Ahuja, S., Ed.
Chiral Separations: Applications and Technology; American Chemical
Society: Washington DC, 1997. (c) Gawley, R. E.; Aube´, J . Principles
of Asymmetric Synthesis; Pergamon: New York, 1996. (d) Aboul-Enein,
H. Y., Wainer, I. W., Eds. The Impact of Stereochemistry on Drug
Development and Use; Wiley-VCH: New York, 1997.
(5) McCoy, M. Chem. Eng. News 2000, 78(25), 17.
(6) Willsta¨tter, R. Chem. Ber. 1904, 37, 3758.
(7) Klemm, L. H.; Reed, D. J . Chromatogr. 1960, 3, 364.
(8) (a) Allenmark, S. Chromatographic Enantioseparations: Methods
and Applications; Ellis Horwood, New York, 1991. (b) Siret, L.;
Bargmann-Leyder, N.; Tambute, A.; Caude, M. Analusis. 1992, 20, 427.
(c) Beesley, T. E.; Scott, R. P. Chiral Chromatography; Wiley-VCH:
New York, 1998.
10.1021/jo011005x CCC: $22.00 © 2002 American Chemical Society
Published on Web 03/13/2002

1994 J . Org. Chem., Vol. 67, No. 7, 2002
Ling et al.
Sch em e 1. Syn th esis of Ch ir a l Selector a
process and, therefore, the performance of the chiral
separation media.
Current Pirkle-type CSPs rely mostly on the mono-
valent interactions of an analyte with a single isolated
ligand attached to a support. In contrast, many recogni-
tion processes in nature are facilitated and significantly
enhanced by multivalent ligands and multivalent recep-
tors. In fact, multivalency is likely to emerge as the new
frontier in the design and development of a broad range
of biologically active compounds.9 Although the source
of the enhanced properties derived from multivalency has
not yet been unambiguously elucidated, it is believed to
originate either from the synergetic effect of specifically
located multiple ligands or from an aggregation and
precipitation process.10 Regardless of the mechanism, a
substantial increase in binding strength should prove
valuable to the development of at least some specialized
enantioselective separation media.
In 1986, Pirkle found that the separation factor for
enantiomers of bivalent N-(3,5-dinitrobenzoyl)leucine bis-
amide analyte on monovalent CSP was higher by a factor
of almost 20 when compared to that of the monovalent
analyte.11 Presumably, such enhancement in selectivities
was due to the multiple binding interactions similar to
those observed for chelates. Unfortunately, this approach
is not very practical, as it would require the “dimeriza-
tion” of the enantiomers onto a divalent linker, their
separation on a column, and the subsequent cleavage of
the desired single enantiomers from the linker. Therefore,
other concepts leading to a considerable enhancement of
the selectivity are desirable. One of these approaches may
consist in the use of multivalent selectors. Indeed, we
have recently observed in a preliminary study that
attachment of a chiral selector to organic porous polymer
beads through a branched linker increases the specific
enantioselectivity over that of the equivalent separation
medium with a traditional linear tether.12
Dendrons are monodisperse polymers that have a well-
defined architecture and are made from branched repeat
units. As a result of their highly branched structure,
dendritic molecules are ideal for positioning a specific
number of functional groups in precisely defined loca-
tions.13 Unlike typical dendrimers, dendrons have two
different functionalities: one located at the focal point
and a multiplicity of functionalities of an orthogonal type
located at the periphery. These features make dendrons
very suitable for assembling multivalent ligands and
their attachment to a solid support through a single
reactive site.
a
Key: (a) EEDQ, CH2Cl2; (b) TFA, AcOH, CH2Cl2; (c) succinic
anhydride, DMAP, 1:1 THF/Et3N.
the efficacy of each strategy, spacer inclusion, dendron
generation, and intermediate endcapping were examined
for their effects on the selector loading and the perfor-
mance of these CSPs.
Resu lts a n d Discu ssion
P r ep a r a tion of Ch ir a l Selector . All chiral stationary
phases in this study contained a model chiral selector 4
derived from L-proline characterized by very high selec-
tivity toward the dinitrobenzoyl derivatives of leucine and
alanine that we have previously discovered using meth-
ods of combinatorial chemistry.14 This selector was de-
rivatized with succinic anhydride to afford compound 5
containing a linker with carboxylic acid functionality that
is suitable for coupling with hydroxyl functionalities via
ester linkage (Scheme 1).
P r ep a r a tion of Mon ova len t Ch ir a l Sta tion a r y
P h a se. In contrast to the CSP with same proline-based
selector linked to the support via carbamate that we have
developed earlier,14 the monovalent (nondendritic) CSP
G0 was prepared by direct esterification of hydroxyl
groups of the poly(2-hydroxyethyl methacrylate-co-eth-
ylene dimethacrylate) support beads with selector 5 using
DIC-coupling chemistry shown in Scheme 4. The chem-
istry of this CSP matched that of our CSPs with multi-
valent selectors and served as a benchmark for the
evaluation of enantioselectivity.
In this paper, we wish to describe in detail two general
strategies that afford CSPs with dendritic polyester
linkers connecting a model selector to the porous polymer
support: a divergent scheme where the linker is grown
from the surface of the support followed by functional-
ization with selector (ligand) and a convergent scheme
where a well-defined selector-functionalized polyester
dendron is anchored onto the solid support. In assessing
P r ep a r a tion of Mu ltiva len t Ch ir a l Sta tion a r y
P h a ses. The current synthesis of dendritic molecules15
employs one of two basic approaches, convergent16 or
divergent.17 In the convergent strategy, the synthesis
begins at what will eventually become the periphery of
(9) (a) Henry, C. W.; McCarroll, M. E.; Warner, I. M. J . Chromatogr.
2001, 905, 319. (b) Liapis, A. I.; Grimes, B. A. J . Coll. Interface Sci.
2000, 229, 540.
(10) Fung, Y. S.; Long, Y. H. J . Chromatogr. 2001, 907, 301.
(11) Pirkle, W. H.; Pochapsky, S. J . Chromatogr. 1986, 369, 175.
(12) Liu, Y.; J uneau, K. N.; Svec, F.; Fre´chet, J . M. J . Polym. Bull.
1998, 41, 183.
(13) (a) Archut, A.; Vo¨gtle, F. Chem. Soc. Rev. 1998, 27, 233. (b)
Vo¨gtle, F.; Gestermann, S.; Hesse, R.; Schwierz, H.; Windisch, B. Prog.
Polym. Sci. 2000, 25, 987.
(14) Murer, P.; Lewandowski, K.; Svec, F.; Fre´chet, J . M. J . Anal.
Chem. 1999, 71, 1278.
(15) (a) Newkome, G. R.; Moorefield, C. N.; Vo¨gtle, F. Dendritic
Molecules: Concepts, Synthesis, Perspectives; VCH: Weinheim, 1996.
(b) Chow, H.-F.; Mong, T. K.-K.; Nongrum, M. F.; Wan, C.-W.
Tetrahedron 1998, 54, 8543.
(16) Hawker, C. J .; Fre´chet, J . M. J . J . Am. Chem. Soc. 1990, 112,
7638.

Enantioselective Separation Media for Chiral HPLC
J . Org. Chem., Vol. 67, No. 7, 2002 1995
Sch em e 2. P r ep a r a tion of CSP s w ith Den d r itic Lin k er s via th e Con ver gen t Ap p r oa ch a
a
Key: (a) (i) 21, DMAP, CH2Cl2, (ii) 0.3 M H2SO4, 1:2 H2O/MeOH; (b) (i) 5, DIC, DPTS, CH2Cl2, (ii) H2, Pd/C, EtOAc; (c) (i) HEMA-
EDMA 20, DPTS (0.5 equiv with respect to the selector), DIC (2 equiv with respect to the selector), CH2Cl2, room temperature, 72 h, (ii)
acetic anhydride, DMAP, CH2Cl2.
the dendrimer and converge toward a core. Given a
monomer of the form A-B2, the “B” sites are protected
and the reactive “A” site, located at the focal point, is
used for further linkage to another A-B2 unit in which
the A functionality is protected. The resulting composite,
a dendron, is then deprotected at its focal point and
reacted with another “A” protected monomer. This se-
quence of deprotection and coupling is successively
repeated until the dendron of desired generation is
obtained, at which point, the peripheral “B” functional-
ities are deprotected for functionalization with ligands.
In contrast, the divergent approach begins with growth
from a core toward the periphery. Starting with monomer
protected only at the focal “A” site, the dendron is
extended by linkage of the active “B” functionalities with
monomers protected only at their peripheral “B” sites.
Higher generation dendrons are prepared by successive
deprotection of the periphery followed by linkage with
“B” protected monomers. Both of these two methods are
amenable toward the preparation of CSPs with branched
linkers.
was the preparation of enantioselective stationary phases,
we chose not to use the architecture based on the
established poly(benzyl ether) dendron chemistry as the
aromatic units may contribute to nonspecific π-π inter-
actions. Instead, we used dendrons with aliphatic ester
chemistry prepared from derivatives of 2,2-bis(2-meth-
oxy)propionic acid (bis-MPA) 6 and 21 (Figure 1) that
have been recently developed in our group.18 The periph-
eral 1,3-diol groups of the dendrons were protected with
isopropylidene groups while their carboxylic acid func-
tionality at the focal point was protected by a benzyl ester
group. These units can be selectively deprotected and
subjected to further functionalization: the peripheral
diols are liberated by treatment with dilute sulfuric acid
solution in methanol-water mixture while the focal point
is deprotected by hydrogenolysis under atmospheric
pressure using palladium catalyst in ethyl acetate.
For the synthesis of the bivalent (G1) selector, both
hydroxyl groups of 6 were esterified with the selector
(17) (a) Tomalia, D. A.; Baker, H.; Dewald, J .; Hall, M.; Kallos, G.;
Martin, S.; Roeck, J .; Ryder, J .; Smith, P. Polym. J . 1985, 17, 117. (b)
Newkome, G. R.; Yao, Z.; Baker, G. R.; Gupta, V. K. J . Org. Chem.
1985, 50, 2003.
(18) Ihre, H.; Hult, A.; Fre´chet, J . M. J .; Gitsov, I. Macromolecules
1998, 31, 4061.
P r ep a r a tion by Con ver gen t Ap p r oa ch . Scheme 2
shows the reaction sequence used for the preparation of
multivalent CSPs in the convergent way. Since our target

1996 J . Org. Chem., Vol. 67, No. 7, 2002
Ling et al.
tionalities located at the pore surface of the beads by a
factor of 2. The multivalent selectors 8-10 were then
immobilized onto these pretreated beads to afford CSPs
C2-G1, C2-G2, and C2-G3. In the last series, tetraeth-
ylene glycol (TEG)19 spacer was inserted between the
hydroxyethyl site at the solid surface and the dendritic
moieties in order to provide the selector with enhanced
mobility and ease the steric constraints encountered
during the immobilization of these bulky branched selec-
tors. Our initial attempt to exploit tetraethylene glycol
as a spacer by attaching the functionalized dendrons to
the HEMA beads through its TEG hydroxyl terminus
failed. In contrast, a TEG spacer with a nucleophilic
amine end group shown in Scheme 3 facilitated the
coupling. For example, dendritic selectors 17 (G1, Scheme
3), 18 (G2), and 19 (G3) were prepared from the TEG
derivative 15 that was coupled with 5, 8, and 9, respec-
tively, followed by Boc-deprotection of the terminal amine
functionality. These selectors were then attached to
HEMA beads activated with 4-nitrophenyl chloroformate
to afford CSPs C3-G1, C3-G2, and C3-G3, respectively.
The unreacted 4-nitrophenyl groups were then displaced
by diethylamine. This approach enabled a certain in-
crease in the loading levels at higher generations of
dendritic selectors (Table 1).
F igu r e 1. Structures of the bis-MPA derivatives used as
building blocks for dendritic linkage.
Sch em e 3. Syn th esis of Den d r itic Selector w ith
Tetr a eth ylen e Glycol Lin k er a
Overall, our efforts to increase the selector loading
were not very successful. This is probably due to a
combination of factors that include the bulkiness of the
dendritic selector as well as the difficulty in coupling the
focal point of the dendron to the pendant hydroxyl groups
of the resin located at the rigid pore surface.
P r ep a r a tion by Diver gen t Ap p r oa ch . While the
preparation of dendrons is generally carried out in
solution, the preparation of dendrimers and dendritic
ligands on resin has also gained some attention re-
cently.20 Since the divergent solid-phase supported ap-
proach to dendritic structures obviates the need for
tedious repetitive purification steps and allows the use
of excess reagents to drive the reaction to completion, we
used this approach to prepare CSPs with dendritic
linkers. This preparation requires an iterative three-step
procedure: growth of the linker by addition of protected
bis-MPA anhydride 21, capping of unreacted hydroxyl
groups with acetic acid anhydride, and selective depro-
tection of the isopropylidene groups with dilute sulfuric
acid (Scheme 5)21 For example, the first generation CSP
D1-G1 is prepared by reacting a slurry of HEMA beads
20 with an excess of anhydride 21 in the presence of a
catalytic amount of DMAP. After the mixture reacts for
24 h at room temperature, acetic anhydride is added to
the mixture to cap the unreacted hydroxyl groups. The
capping ensures that the selector moieties are attached
to the linker and not to the hydroxyl functionalities
originating from the support. Following capping, the
isopropylidene protecting group are removed by treat-
ment with dilute sulfuric acid solution in methanol-
water mixture to liberate the peripheral hydroxyl groups.
a
Key: (a) (i) H2 (g), Pd/C, di-tert-butyl dicarbonate, EtOAc, (ii)
13, DMAP, CH2Cl2, (iii) H2 (g), Pd/C, EtOAc; (b) (i) 5, DCC,
DPTS,CH2Cl2, (ii) TFA, HOAc, CH2Cl2.
Sch em e 4. P r ep a r a tion of Mon ova len t Ch ir a l
Sta tion a r y P h a sea
a
Key: (a) 5, DIC, DPTS, CH2Cl2.
moieties 5 to afford intermediate 7, which is subsequently
subjected to hydrogenolysis to deprotect the carboxylic
acid, giving 8 (Scheme 2). Similarly, second- and third-
generation dendritic selectors 9 and 10 were prepared
by deprotection of peripheral diols followed by their
functionalization with 5 and completed by liberation of
the focal point functionality.
Three series of CSPs (C1, C2, and C3) were prepared
by reaction of multivalent selectors of the first, second,
and third generations (G1, G2, and G3), respectively,
with HEMA beads. In the C1 series, the dendritic
selectors 8-10 were attached to the solid support directly
using DIC coupling chemistry to afford CSPs C1-G1, C1-
G2, and C1-G3 (Scheme 2). In an attempt to increase
the surface coverage, the HEMA beads were first treated
with isopropylidene-2,2-bis(methoxy)propionic anhydride
21 followed by deprotection of the diol functionalities
using mild acid conditions. Presumably, this approach
should increase the number of reactive hydroxyl func-
(19) Bertozzi, C. R.; Bednarski, M. D. J . Org. Chem. 1991, 56, 4326.
(20) (a) Arya, P.; Rao, N. V.; Singkhonrat, J .; Alper, H.; Bourque,
S. C.; Manzer, L. E. J . Org. Chem. 2000, 65, 1881. (b) Basso, A.; Evans,
B.; Pegg, N.; Bradley, M. Tetrahedron Lett. 2000, 41, 3763. (c) Marsh,
A.; Carlisle, S. J .; Smith, S. C. Tetrahedron Lett. 2001, 42, 493. (d)
Ulrich, K. E.; Boegeman, S.; Fre´chet, J . M. J .; Turner, S. R. Polym.
Bull. 1991, 25, 551.
(21) (a) Lewandowski, K.; Svec, F.; Fre´chet, J . M. J . Chem. Mater.
1998, 10, 385. (b) Petro, M.; Svec, F.; Fre´chet, J . M. J . Anal. Chem.
1997, 69, 3131.

Enantioselective Separation Media for Chiral HPLC
J . Org. Chem., Vol. 67, No. 7, 2002 1997
Ta ble 1. Loa d in gs a n d Sep a r a tion F a ctor s r Obta in ed for En a n tioselective Sep a r a tion of Ra cem ic Com p ou n d s on CSP
G0 a n d Ch ir a l Sta tion a r y P h a ses P r ep a r ed by th e Con ver gen t Ap p r oa ch a
CSP
G0
C1-G1
0.27
C1-G2
0.12
C1-G3
0.13
C2-G1
C2-G2
C2-G3
0.12
C3-G1
0.14
C3-G2
0.14
C3-G3
0.23
selector loading,b mmol/g
analytec
DNB-Leu-diallyl
DNB-Leu-diethyl
DNB-Ala-diallyl
DNB-Ala-diethyl
0.27
0.28
0.08
separation factor, R
21.1
17.5
17.9
15.5
14.9
7.4
6.9
6.7
5.9
9.1
8.6
8.5
7.4
21.9
22.3
20.1
20.0
9.0
8.2
7.2
8.7
12.9
14.1
12.4
15.8
12.4
10.6
11.4
9.4
12.2
12.6
16.2
12.0
16.6
15.7
15.6
14.7
20.6
18.6
17.4
a
Chromatographic conditions: column, 150 × 4.6 mm i.d.; mobile phase, 20% hexane in CH2Cl2; flow rate, 1.0 mL min-1; void marker,
1,3,5-tri-tert-butylbenzene; UV detection, 254 nm. Determined by elemental analysis of nitrogen. c For structures of the analytes, see
b
Figure 2.
Sch em e 5. P r ep a r a tion of CSP s w ith Den d r itic Lin k er s via th e Diver gen t Ap p r oa ch a
a
Key: (a) (i) 21, DMAP, CH2Cl2, room temperature, 24 h, (ii) add acetic anhydride, room temperature,12 h; (b) (i) 5, DIC, DPTS,
CH2Cl2, room temperature, 24 h, (ii) acetic anhydride, DMAP, CH2Cl2.
Ta ble 2. Loa d in gs a n d Sep a r a tion F a ctor s r Obta in ed for En a n tioselective Sep a r a tion of Ra cem ic Com p ou n d s on
Ch ir a l Sta tion a r y P h a ses P r ep a r ed by th e Diver gen t Ap p r oa ch a
CSP
D1-G1
0.56
D1-G2
0.73
D1-G3
0.61
D1-G4
D2-G1
D2-G2
0.84
D2-G3
0.75
D2-G4
0.64
selector loading,b mmol/g
analytec
0.46
0.61
separation factor, R
DNB-Leu-diallyl
DNB-Leu-diethyl
DNB-Ala-diallyl
DNB-Ala-diethyl
27.8
30.7
25.8
26.7
23.3
25.8
20.4
21.3
19.4
21.7
20.6
21.6
18.3
20.2
18.4
19.1
25.7
28.1
23.6
24.4
24.0
28.0
24.4
26.0
23.4
28.2
23.6
26.2
22.6
26.3
23.6
26.8
a
Chromatographic conditions: column, 150 × 4.6 mm i.d.; mobile phase, 20% hexane in CH2Cl2; flow rate, 1.0 mL min-1; void marker,
1,3,5-tri-tert-butylbenzene; UV detection, 254 nm. Determined by elemental analysis of nitrogen. c For structures of the analytes, see
b
Figure 2.
Finally, the selector is linked to the modified beads.
Reacting the deprotected hydroxyl groups with anhydride
21 instead of selectors leads to branched linkers of higher
generations.
To demonstrate the effect of endcapping of unreacted
hydroxyl groups, we also prepared a second series of
CSPs D2 by omitting the capping between the extension
and deprotection steps. Although this approach affords
CSPs with somewhat higher selector loading (Table 2),
the exclusivity of their peripheral location is not guar-
anteed.
Eva lu a tion of CSP s. For chromatographic evaluation,
all CSPs were slurry packed into 150 × 4.6 mm i.d.
columns and used for the separation of N-(3,5,-dini-
trobenzoyl)-R-amino acid derivatives shown in Figure 2
in isocratic mode using 20% hexane/dichloromethane as

1998 J . Org. Chem., Vol. 67, No. 7, 2002
Ling et al.
series. Clearly, CSP C2-G3 retains L-enantiomer more
strongly (retention time 8 min) than either CSP C2-G1
or C2-G2 for which retention times of only 5.5 min were
observed. Although the retention times of the D-enanti-
omer do not vary much from CSP to CSP under our
experimental conditions, its modest increase on the C2-
G3 results in separation factors that are similar to that
of C2-G2, but much lower than that for the first genera-
tion C2-G1.
Clearly, the enantioselectivity depends on the selector
loading and since it is difficult to prepare CSPs of
different generation linkage that have identical loading,
we compared our CSPs using a “normalized” value of
specific selectivity R′ defined as the ratio of separation
factor R per unit of the selector loading that we intro-
duced in our previous studies.22 The specific selectivity
enables better assessment of the effects of branched
linkers on selectivity. For example, the R′ value for all
studied racemates is in the range of 60 to 80 for CSP
G0. CSPs C2-G2 exhibits R′ values between 90 and 110
while R′ values of 100 to 130 were observed for C2-G3.
This demonstrates that the selectivity per selector moiety
is enhanced for CSPs at higher generations. Although
this may be an indication of desired synergistic interac-
tions of selectors that are confined in a close proximity,
it can also be attributed to the lengthening of the distance
between the individual selectors and the surface of the
matrix.
F igu r e 2. Structures of the N-(3,5-dinitrobenzoyl)-R-amino
acid dialkyl amide analytes used in the study.
To improve the reaction of the sterically bulky dendron
with the support for increased loading, we also prepared
CSPs containing branched selectors with a tetraethylene
glycol unit linked to the focal point of the dendron
through an ester bond on one side and terminated with
the more nucleophilic primary amine functionality on the
other side (C3 series). Table 1 shows that the apparent
separation factors are higher than those of the C1 and
C2 series with G1 and G2 linkers. This increase appears
to result from their higher loading since their specific
selectivities (65 to 90) do not reveal any significant
enhancement due to the branching. However, the evalu-
ation of these CSPs cannot be fairly compared to those
achieved with CSPs of C1 and C2 series because the
selectors are attached to the support through a carbam-
ate group rather than an ester group.
F igu r e 3. HPLC chromatogram of the enantioseparation of
N-(3,5-dinitrobenzoyl)leucine diallyl amide via CSPs: (1) C2-
G1, (2) C2-G2, and (3) C2-G3. Mobile phase: 20% hexane in
CH2Cl2. Flow rate: 1.0 mL/min.
the mobile phase. Because of very high selectivities of
all of our CSPs, the racemates were separated with
excellent resolution.
Eva lu a tion of Diver gen t CSP s. Two series of CSPs
were prepared using the divergent methodology: (i) D1
with endcapping of unreacted hydroxyl groups after each
extension step and (ii) D2 in which the endcapping step
was omitted and all hydroxyl groups were allowed to
compete in further reaction steps. Provided that the
capping steps used in D1 series completely eliminates
all undesired hydroxyl groups, it ensures that all ligand
moieties are anchored only at the periphery of the
dendritic linker. In contrast, the individual selectors for
the D2 series may lie at varying distances from the
reactive site of the bead since the “internal” hydroxyl
groups that were not previously capped could participate
in the selector loading. Data from Table 2 support this
hypothesis since the loading levels for the D2 series are
higher for each dendritic generation. Obviously, capping
with anhydride 21 after linker extension diminishes the
amount of selectors that can be attached to CSPs of the
D1 series as compared to the D2 series, but this also
First, the enantioselectivity of monovalent CSP G0
with a loading of 0.26 mmol/g was examined as a
benchmark. The separation factors attained for all four
probe racemates are large with values ranging from 17.4
to 21.1.
Eva lu a tion of Con ver gen t CSP s. Selector loadings
and results of the separations for both monovalent CSP
G0 and the CSPs assembled in the convergent way are
shown in Table 1. The loading of CSPs with branched
linkers of both second and third generations attached in
a single step rapidly decreases and typically reaches
values in the range of 0.12-0.14 mmol/g.
These CSPs exhibited high selectivities in the range
of 5.9-22.3 toward our model analytes. As shown in
Table 1, most of these CSPs with branched linkers have
lower apparent selectivities compared to the monovalent
CSP G0. The enantioselectivity of these CSPs clearly
depends on the loading. The higher the selector loading,
the better the separation observed within each series.
Figure 3 demonstrates the separations of racemic N-(3,5-
dinitrobenzoyl)-leucine diallyl amide on CSPs of the C2
(22) Lewandowski, K.; Murer, P.; Svec, F.; Fre´chet, J . M. J . Anal.
Chem. 1998, 70, 1629.

Enantioselective Separation Media for Chiral HPLC
J . Org. Chem., Vol. 67, No. 7, 2002 1999
implies that the structure of the dendrimer grown from
the solid surface contains defects.
The divergent approach enables preparation of CSPs
with much higher loading levels (Table 2). For example,
the amount of selector is 0.56 mmol/g for D1-G1 and 0.61
mmol/g for D2-G1, both first generation ligands. This is
more than twice as much compared to G0 and several
times higher than the loadings of CSPs in the C series.
In theory, each increase in generation of the dendritic
linker should result in doubled loading. Although G2
exhibits a higher loading than G1, this increase repre-
sents only 30% for D1-G2 and 38% for D2-G2 confirming
that the peripheral hydroxyl groups were partially func-
tionalized. Moreover, a simple calculation shows intrinsic
limitations for the growth of perfect dendrimers on solid
support. Assuming that 65% of the hydroxyl groups
originating from the HEMA (3.0 mmol/g) are available
for coupling,21a then 0.24 g (0.20 mL) of bis-MPA are
immobilized in the first step. Clearly, ideal growth
beyond the first generation is unlikely, as this would
imply that the volume taken up by the linker exceeds
that available within the pores.
F igu r e 4. Chiral separation of N-(3,5-dinitrobenzoyl)leucine
diallyl amide on CSP D1-G1 under various overloading
conditions: Mobile phase: 20% hexane in CH2Cl2. Flow rate:
1.0 mL/min. Injection: (1) 0.10 mg, (2) 1.0 mg, (3) 5.0 mg, and
(4) 10.0 mg.
The higher loadings of these CSPs also endow them
with selectivities higher than those obtained for the
convergently prepared CSPs. Table 2 also summarizes
results of chromatographic evaluations of CSPs of both
series D1 and D2. The separations factors are in the
range of 18-31. For example, an R value of 30.7 was
obtained for the separation of racemic dinitrobenzoylleu-
cine diethylamide on CSP D1-G1. This selectivity is
about 50% higher than that observed for CSP G0 and
also significantly higher than the R values that we
reported previously for the separation of these race-
mates.14 However, the higher loading does not necessarily
lead to higher selectivity. While the separation factors
remain rather high for CSPs with higher generation
dendrons (G2 through G4), the R values diminish with
the increase in generation of the linker. The separation
factors of CSP D1-G1 are higher than those of CSP D2-
G1 even with a lower selector loading of the former. This
indicates that the selectivity increases with the distance
between the selector moiety and the solid support surface.
Surprisingly, the separation factors do not follow the
trend of the increased loading even within each series.
The R values decrease despite the higher content of
selector moieties. Despite both high loading and selectiv-
ity, the actual R′ values for these divergent CSPs are only
in the range of 30-50, considerably lower than those of
CSP G0 and CSPs of the convergent series. We deduce
that this could result from an incomplete random sub-
stitution of the periphery of the dendritic linker that is
not favorable for synergistic interactions.
En a n tiosep a r a tion s u n d er Over loa d Con d ition s.
Although an increase in throughput of chromatographic
separations can be achieved by simply using larger
columns scaled to the desired size operating under the
conditions typical of analytical scale separations, this
approach is resource intensive. This linear scale-up does
not require any specifically developed CSPs as it typically
involves a method that has already been developed on
the analytical scale. In contrast, processes that use
smaller columns and CSPs tailored for high enantiose-
lectivity and high loading of a specific analyte can be
operated under overload conditions, and thus, enable
high throughput operations at lower capital costs. Since
the preparation of CSPs using branched linkers affords
F igu r e 5. Separation of N-(3,5-dinitrobenzoyl)-leucine diallyl
amide under various overloading conditions on CSP D1-G1
visualized under chiral detector: Mobile phase: 20% hexane
in CH2Cl2. Flow rate: 1.0 mL/min. Injection: (1) 0.10 mg, (2)
1.0 mg, (3) 5.0 mg, and (4) 10.0 mg.
high loading of selector moieties, these stationary phases
are also attractive for the preparation of CSPs to be used
for large-scale chromatography. Figure 4 shows a series
of enantioseparations of 0.1 to 10 mg of racemic dini-
trobenzoylleucine diallylamide using an analytical 150
× 4.6 mm i.d. column packed with CSP D1-G1 and
isocratically eluted with a mixture of 20% hexane in
dichloromethane as the mobile phase at 1.0 mL/min.
While a valley between eluted bands of both enantiomers
is observed for injections of up to 5.0 mg, the absorption
exceeds the limits of the UV detector and no apparent
separation is observed for injection of 10 mg. In contrast,
the less sensitive polarimetric chiral detector always
monitors well-separated peaks of both enantiomers and
enables a more accurate cut between the fractions needed
to collect single enantiomers in the highest degree of
purity (Figure 5). In addition, this detector confirms that
the peaks contain opposite enantiomers. The peak areas
are equal within the range of experimental error indicat-
ing that no other chiral compound is coeluted.
The identical chemical shifts in NMR spectra of the
materials recovered by collecting two fractions of the
eluent cut in the valley between the peaks monitored by
the chiral detector also confirm the purity of both
enantiomers. Another check of purity of the separated
enantiomers was obtained by re-injecting them sepa-
rately in the same column (Figure 6). Only very small
peaks of the undesired enantiomers representing 0.8%
and 0.5% of the total peak areas are visible in the trace
together with large peaks containing 99.2% and 99.5%
of the desired D and L enantiomers, respectively. The

2000 J . Org. Chem., Vol. 67, No. 7, 2002
Ling et al.
under N2 from sodium benzophenone ketyl. Diisopropylcarbo-
diimide (DIC), dicyclohexylcarbodiimide (DCC), 2-ethoxy-1-
(ethoxycarbonyl)-1,2-dihydroquinoline (EEDQ), 2,2-bis(methoxy)-
propionic acid (bis-MPA), and 4-(N,N-dimethylamino)pyridine
(DMAP) were purchased from Aldrich and all other reagents
and solvents were obtained from commercial suppliers and
were used without further purification. Bis-MPA based build-
ing blocks and dendrons were prepared according to procedures
described elsewhere.18,24 All moisture or air-sensitive reactions
were carried out under nitrogen in a dry solvent. The 7 mm
monodisperse porous HEMA-EDMA beads consisting of 40 wt
% 2-hydroxyethyl methacrylate (HEMA) and 60 wt % ethylene
dimethacrylate (EDMA) were prepared using templated staged
suspension polymerization that was developed by our group.21a
These beads have a median pore size of 11 nm, a pore volume
of 0.3 mL/g, and a specific surface area of 140 m2/g.
In str u m en ta tion . 1H and 13C NMR spectra were obtained
with Fourier transform spectrometers Bruker AMX-300, Bruk-
er AMX-400, Bruker AM-400, and Bruker DRX-500. Spectral
data are reported as chemical shifts (multiplicity, coupling
constant in Hz, number of protons, identity if possible) in ppm
(δ units). Infrared spectra were recorded on a Nicolet Mattson
Genesis II FT-IR spectrophotometer using KBr technique.
Syn th esis of Selector w ith Ca r boxylic Acid Ha n d le
(5): (i) l-2-(In d a n -5-ylca r ba m oyl)p yr r olid in e-1-ca r boxy-
lic Acid ter t-Bu tyl Ester (3).14 To a solution of N-t-Boc-L-
proline (1; 85 g, 0.40 mol) and EEDQ (100 g, 0.41 mol) in
CH2Cl2 (200 mL) was added 5-aminoindan (2; 54 g, 0.41 mol)
at 0 °C (Scheme 1). The reaction mixture was stirred at 0 °C
for 2 h and then allowed to react overnight at room temper-
ature. The reaction mixture was diluted with 100 mL of CH2-
Cl2 and washed with 1 M HCl (3 × 50 mL), saturated aqueous
NaHCO3 solution (2 × 50 mL), H2O (50 mL), and brine (2 ×
50 mL). The organic phase was dried over MgSO4, filtered,
and concentrated to give a viscous oil. The crude product was
then redissolved in a 1:1 mixture of CH2Cl2 and EtOAc. Cooling
of the mixture to -10 °C affords 3 (116 g, 90%) as an off-white
crystalline solid.
(ii) Selector . L-P yr r olid in e-2-ca r boxylic a cid in d a n -5-
yla m id e (4). To a solution of 3 (20.2 g, 0.0611 mol) dissolved
in CH2Cl2 (100 mL) at 0 °C was added 90 mL of a 1:1 TFA/
HOAc mixture. The reaction mixture was stirred overnight
at room temperature. Progress of the deprotection was followed
by TLC (2:1 hexane/EtOAc). Upon complete deprotection, the
volatile compounds were removed in vacuo and the product
was diluted with H2O (50 mL) and CH2Cl2 (50 mL). Then, 2
M KOH (aq) was added at 0 °C until the pH of the aqueous
phase was between 9 and 10. The aqueous phase was then
extracted with CH2Cl2 (3 × 100 mL), and the combined organic
layers were washed with H2O (100 mL) and brine (100 mL),
dried over MgSO4, filtered, and concentrated to give selector
4 (14.0 g, 100%) as a waxy brown solid.
(iii) L-4-[2-(In d a n -5-ylca r ba m oyl)p yr r olid in -1-yl]-4-ox-
obu tyr ic Acid (5). Compound 4 (1.0 g, 4.3 mmol) and succinic
anhydride (0.6 g, 6 mmol) were dissolved in 1:1 THF/Et3N (10
mL) solution, and DMAP (0.05 g, 0.4 mmol) was added. The
reaction mixture was heated under reflux at 70 °C overnight.
Upon completion, the crude product was diluted with 1 M HCl
(100 mL) and extracted with EtOAc (4 × 50 mL). The organic
phase was dried over Na2SO4 and the solvent evaporated to
give 5 (1.3 g, 82%) as a light beige solid.
Syn th esis of Biva len t Selector a n d a Gen er a l P r oce-
d u r e for Syn th esis of Den d r itic Selector s: (Selector )2-
[G1]-CO2H (8). To a mixture of benzyl-2,2-bis(methoxy)-
propionate18 6 (0.50 g, 2.23 mmol), 5 (1.53 g, 4.62 mmol), and
DPTS (0.37 g, 1.24 mmol) in CH2Cl2 (ca. 30 mL) was added
DCC (1.38 g, 6.69 mmol) (Scheme 2). The reaction mixture was
stirred under inert atmosphere at room temperature overnight,
F igu r e 6. Reinjection of fractions from separated peaks
collected from the 10.0 mg overload separation of N-(3,5-
dinitrobenzoyl)leucine diallyl amide. Separation conditions:
column CSP D1-G1; mobile phase, 20% hexane in CH2Cl2; flow
rate, 1.0 mL/min.
separation in column packed with unbranched CSP G0
under identical conditions afforded peak purities of only
92.9% and 99.4%. This again demonstrates the positive
effect of the branched linker. A simple calculation reveals
that even with an unoptimized system, 1.0 kg per day of
the racemate could be separated using 1.2 kg of the
branched linker containing CSP D1-G1 packed into a
column and carried out under standard normal-phase
chromatographic mode.
Con clu sion s
Our results indicate that dendritic linkers are a
beneficial feature while engineering a chiral stationary
phase. In contrast to expectations arising from Pirkle’s
observation of a vast increase in selectivity factors for
bivalent analytes,11 we do not observe an increase in
selectivity of such magnitude for our separations in the
“reciprocal” mode. This could result from the differences
in kinetics of the selector/analyte interactions since in
our case the polyvalent counterpart is immobilized onto
a rigid matrix while the monovalent target molecules are
in the solution. While in Pirkle’s case each end of the
bis-analyte could independently and simultaneously in-
teract with two different chiral selectors on the CSP, in
our system, the analytes cannot associate with more than
one selector in an identical configuration. Despite this
limitation, very high separation factors of up to R ) 31
have been achieved with our CSPs containing branched
linkers. These are considerably higher than those seen
for the G0 counterpart containing only monovalent
selector moieties. This validates the positive synergistic
effect of multiple selectors located in well-defined posi-
tions within in the dendritic CSP. This synergism may
arise from the intercalation of the analyte between two
selectors located in precise positions. The preparation of
dendrons bearing chiral peripheral moieties can also be
easily combined with the screening of combinatorial
libraries of potential selectors for chiral recognition in
order to accelerate the discovery of selectors targeted
toward a specific racemate. Although much remains to
be done, this study opens a new avenue for the prepara-
tion and application of tailor-made CSPs in large-scale
high throughput enantioseparations.
(23) Hyun, M. H.; Choi, S. Y.; Um, B. H.; Han, S. C. J . Chromatogr.
A 2001, 922, 119.
(24) (a) Freeman, A. W.; Fre´chet, J . M. J . Org. Lett. 1999, 1, 685.
(b) Liu, M.; Fre´chet, J . M. J . Pharmaceut. Sci. Technol. Today 1999,
2, 393. (c) Grayson, S. M.; J ayaraman, M.; Fre´chet, J . M. J . Chem.
Commun. 1999, 1329.
Exp er im en ta l Section
Ma ter ia ls. Methylene chloride was distilled from CaH2
under inert atmosphere. Tetrahydrofuran (THF) was distilled

Enantioselective Separation Media for Chiral HPLC
J . Org. Chem., Vol. 67, No. 7, 2002 2001
filtered, and concentrated. The crude reaction mixture was
purified using flash chromatography to yield (selector)2-[G1]-
CO2CH2C6H5 7 (1.90 g, 100% yield) as a beige crystalline solid
that was then subjected to catalytic hydrogenolysis.
A mixture of 7 (1.00 g, 1.18 mmol) and Pd/C (0.096 g) in
EtOAc (ca. 20 mL) was stirred under H2 (g) overnight. The
catalyst was removed by filtration and the filtrate was
evaporated to afford 8 (0.89 g, 99.7%) as a white crystalline
solid (Scheme 2). Second (9) and third (10) generation dendritic
selectors were synthesized by methods similar to those afford-
ing 8.
Con ver gen t P r ep a r a tion of CSP s. CSPs C1-G1, C1-G2,
and C1-G3 were prepared according to Scheme 3 by adding
dendritic selectors 8 (G1), 9 (G2), and 10 (G3), respectively, to
a slurry of DPTS (0.5 equiv with respect to the selector) and
DIC (2 equiv) and beads 20 in 30 mL of CH2Cl2 (Scheme 2).
This reaction mixture was stirred at room temperature for 72
h. The modified beads were then filtered and washed with THF
(1 × 20 mL), H2O (2 × 20 mL), MeOH (2 × 20 mL), and CH2-
Cl2 (ca. 20 mL). The resulting beads were extracted for 24 h
with each CH2Cl2 and THF. The residual hydroxyl groups were
capped again by reaction with acetic anhydride catalyzed by
DMAP in 1:1 pyridine/THF mixture. These beads were then
filtered, washed with THF (60 mL) and triethylamine (40 mL)
and H2O (50 mL), and extracted with THF and CH2Cl2.
CSPs C2-G1, C2-G2, and C2-G3 were prepared by heating
a slurry of beads 20, isopropylidene-2,2-bis(methoxy)propionic
anhydride 21 (5 equiv), and DMAP (0.16 equiv) in THF/
pyridine at 60 °C for 24 h. The modified beads were filtered
and washed with THF and CH2Cl2. These beads were then
treated with 0.4 M H2SO4 in MeOH to remove the isopropyli-
dene protective group and regenerate the hydroxyl function-
alities. The selectors 8 (G1), 9 (G2), and 10 (G3) were
respectively immobilized onto these beads using methods
similar to those described above for the preparation of CSP
C1 series.
CSPs C3-G1, C3-G2, and C3-G3 were prepared by adding
17, 18, and 19, respectively, to a slurry of HEMA-EDMA beads
22 activated with 4-nitrophenylchloroformate and DMAP in
THF/pyridine. The reaction mixture was heated at 60 °C for
24 h. Then diethylamine was added to displace the unreacted
4-nitrophenyl groups. After the reaction was completed, the
beads were washed repeatedly with THF, CH2Cl2, DMF, EtOH,
H2O, MeOH, and Et2O. Extraction of these beads with THF
and CH2Cl2, followed by drying in vacuo afforded CSPs C3-
G1, C3-G2, and C3-G3.
Diver gen t P r ep a r a tion of CSP s. This preparation of
divergent CSPs was carried out in three major phases: (i)
solid-phase synthesis of dendritic linker followed by (ii) capping
with acetic anhydride and (iii) functionalization with selector
(Scheme 5). To a suspension of HEMA beads 20 (6.0 g) in
pyridine (24 mL) were added 21 (4.0 g, 12 mmol) and DMAP
(2.0 g, 16 mmol). After the mixture was stirred at room
temperature for 24 h, acetic anhydride (8.0 mL) was added
and the reaction was continued for an additional 12 h to
complete the capping of residual hydroxyl groups. The beads
were then filtered and washed with several portions of CH2-
Cl2 and MeOH. Compared to the original support, IR spectra
of these beads showed significantly smaller broad band
centered around 3500-1 cm. Removal of the isopropylidene
protecting group by an overnight treatment with 0.3 M H2-
SO4 in 1:2 H2O/MeOH solution afforded the deprotected beads
23 (G1) which were then collected by filtration and washed
with 0.3 M H2SO4, MeOH, THF, and CH2Cl2 and dried under
vacuum. The deprotection was accompanied by the reappear-
ance of the hydroxyl band in the IR spectrum. The preparation
of beads with higher generation dendritic linkers 24 (G2), 25
(G3), and 26 (G4) was carried out by recursively repeating
these procedures using reagents in similar proportions (Scheme
4).
Syn th esis of Biva len t Selector w ith TEG Lin k er a n d
a Gen er a l P r oced u r e for Syn th esis of Den d r itic Selec-
tor s w ith TEG Lin k er : (i) (OH)2-[G1]-[TEG]-NHBoc (15).
To a suspension of Pd/C (10%, 240 mg) in EtOAc (20 mL) under
H2 (g) was added an EtOAc (10 mL) solution of monoazide
derivate of tetraethylene glycol 11 (TEG)19 (2.00 g, 9.12 mmol)
and di-tert-butyl dicarbonate (1.2 equiv, 2.40 g, 11.0 mmol)
(Scheme 3). The reaction mixture was stirred overnight. The
catalyst was removed by filtration and the solvent evaporated
to give crude product. Flash chromatography was carried out
in two steps: 3:1 hexanes/EtOAc was used to remove residual
di-tert-butyl dicarbonate and 1:20 MeOH/CH2Cl2 to afford
[TEG]-NHBoc 12 as a colorless oil (1.40 g, 52%).
To a mixture of 12 (1.40 g, 4.76 mmol) in THF (ca. 20 mL)
and pyridine (ca. 20 mL) were added DMAP (0.29 g, 2.37
mmol) and benzylidene-2,2-bis(methoxy)propionic anhydride
13 (4.10 g, 9.61 mmol). The reaction mixture was stirred under
inert atmosphere at room temperature overnight before 5 mL
of H2O was added to quench unreacted 13. The solution was
extracted with NaHSO4 (3 × 50 mL), Na2CO3 (saturated 3 ×
20 mL), and brine (ca. 50 mL) and dried over MgSO4. After 1
h of drying, the solution was filtered, concentrated, and
redissolved in CH2Cl2 (ca. 4 mL). Addition of hexanes (ca. 20
mL) to the CH2Cl2 solution resulted instantly in a white
precipitate. The solid was removed, and the filtrate was
concentrated in vacuo to afford benzylidine-[G1]-[TEG]-NHBoc
14 (2.30 g, 97%) as a colorless oil. Compound 14 was depro-
tected using the hydrogenolysis procedure described above to
yield 15 (1.80 g, 95%) as a colorless oil.
(ii) (Selector )2-[G1]-[TEG]-NH2 (17). To a mixture of 15
(1.92 g, 4.69 mmol), 5 (3.40 g, 10.3 mmol), and DPTS (0.60 g,
2.04 mmol) in CH2Cl2 (ca. 60 mL) at 0 °C was added DCC (3.08
g, 14.9 mmol). The reaction mixture was allowed to slowly
warm to room temperature and stirred overnight. The di-
cyclohexylurea was removed by filtration, and the filtrate was
concentrated to give an orange crude. After column chroma-
tography (10% MeOH in CH2Cl2), (selector)2-[G1]-[TEG]-NH-
Boc 16 (5.00 g, quantitative) was obtained as a light orange
solid. Deprotection of 16 was achieved again using hydro-
genolysis to give 17 (4.50 g, quantitative) as a yellow solid.
Similarly, second- (18) and third-generation (19) dendritic
selectors with tetraethylene glycol spacers coupled to the focal
point were prepared by linking 15 to 8 and to 9, respectively,
using DCC coupling chemistry followed by removal of the Boc-
protecting group.
P r ep a r a tion of Ch ir a l Sta tion a r y P h a ses. Since two
different methods, convergent (C) and divergent (D), were used
to prepare CSP with dendritic linkers of various generations
(G), the CSPs used in this study are denoted with the letter
indicating the method of preparation, followed by the enu-
meration of the CSP, and a numeric value of the dendron
generation. For example, CSP C3-G2 stands for chiral station-
ary-phase number three prepared by the convergent approach
with a dendritic linker of the second generation.
Alternatively, we also prepared beads that were not end-
capped after growth of the dendritic linker. The addition of
acetic anhydride was omitted from the preparation to give
beads 27-30 with G1, G2, G3, and G4 linkers, respectively.
Once the dendritic linkers were grown on the beads, they
were subjected to coupling with selector. To a mixture of
deprotected beads 23 (ca. 1.5 g), selector building block 5 (0.5
g, 1.5 mmol), and DPTS (0.5 g, 1.5 mmol) in CH2Cl2 (10 mL)
was added DIC (1.2 equiv, 0.23 g, 1.8 mmol). After the mixture
was stirred at room temperature for 24 h, the beads were
filtered and washed with CH2Cl2, MeOH, and THF and air-
dried. The functionalization was accompanied by appearance
of amide N-H bending at 1547 cm-1. Residual hydroxyl groups
were capped by reaction of the beads with acetic anhydride (2
mL) and DMAP (0.5 g, 4.0 mmol) in CH2Cl2 (10 mL). After
stirring overnight, the beads were filtered and washed with
P r ep a r a tion of Mon ova len t CSP G0. To a slurry of
HEMA-EDMA beads 20 (1.3 g) in CH2Cl2 (ca. 10 mL) were
added 5 (0.5 g, 1.5 mmol), DPTS (1 equiv, 0.5 g, 1.5 mmol),
and DIC (1.25 equiv, 0.25 g, 1.8 mmol) (Scheme 4). After the
mixture was gently stirred overnight using rotation, the beads
were filtered and washed with CH2Cl2 (20 mL), THF (20 mL),
MeOH (2 × 20 mL), CH2Cl2 (20 mL), and Et2O (20 mL). The
residual hydroxyl groups were capped by reaction with acetic
anhydride catalyzed by DMAP in 1:1 pyridine/THF mixture
to give CSP G0.

2002 J . Org. Chem., Vol. 67, No. 7, 2002
Ling et al.
CH2Cl2, MeOH, 0.2 M H2SO4 solution in MeOH/H2O, and THF
and then dried under high vacuum to afford CSP D1-G1. IR
analysis showed the desired disappearance of the hydroxyl
peak. Identical coupling procedures were applied to beads 24-
30 to prepare CSPs D1-G2, D1-G3, D1-G4, D2-G1, D2-G2,
D2-G3, and D2-G4, respectively.
Ch r om a togr a p h ic Eva lu a tion . All CSPs were treated
with acetic anhydride before their packing into columns in
order to cap all residual hydroxyl groups that could be the
source of nonspecific interactions. After this treatment, the
CSPs were dispersed in 80:20 CH2Cl2/hexane and packed at a
constant pressure of 20 MPa into 150 × 4.6 mm i.d. stainless
steel columns. A Waters Alliance (model 2690 XE) instrument
with a 486 UV detector and a J ASCO OR-990 chiral detector,
or a system consisting of two Waters 510 HPLC pumps, a
717plus autosampler, and a 486 UV detector, were used
alternatively for the chromatographic experiments. Both
systems were controlled and data acquired by Waters Millen-
nium software.
Ack n ow led gm en t. Support of this research by a
grant from the National Institute of General Medical
Sciences, National Institutes of Health (GM-44885), and
the Division of Materials Sciences of the U.S. Depart-
ment of Energy under Contract No. DE-AC03-76SF00098
is gratefully acknowledged.
Su p p or tin g In for m a tion Ava ila ble: FT-IR spectra, NMR
spectra, and elemental analysis of 3, 4, 5, and dendritic
selectors and their intermediates. This material is available
free of charge via the Internet at http://pubs.acs.org.
J O011005X