NMR Structure Determination of Ion Pairs Derived from Quinine: A Model for Templating in Asymmetric Phase-Transfer Reductions by BH4- with Implications for Rational Design of Phase-Transfer Catalysts

Article abstract of DOI:10.1021/jo990530h

The solution structures of ion pairs formed by quarternary ammonium ions derived from quinine alkaloid with small hard anions (BH^-₄ or Cl^-) in CDCl₃ have been characterized by nuclear magnetic resonance methods. Structural observations have been correlated with the sense of asymmetric induction observed in the phase-transfer reduction of 9-anthryl trifluoromethyl ketone by borohydride (BH^-₄) when catalyzed by the quaternary N-benzylquinine ammonium ion. From interionic nuclear Overhauser effects (NOEs), it appears that the BH^-₄ ion occupies two of the four trigonal pyramidal sites formed by substituents of the quarternary nitrogen of the catalyst cation. One of these sites is in close proximity to the cation's hydroxyl group that is strictly required for asymmetric induction in the model reaction, while the other site is near the vinyl group on the cation. The vinyl group does not appear to be important for determining the sense or extent of asymmetric induction. Using energy-minimized structures derived from NMR data, it was predicted that the N-(9-methyleneanthryl)quinine - quarternary ammonium catalyst would give improved asymmetric induction in the model reaction due to a preferred anion occupancy at the site near the hydroxyl group. An improvement in enantiomeric excess (ee) is observed using the anthryl-modified catalyst, and NMR studies on the modified catalyst confirm the predicted change in anion binding site occupancies. The changes in site occupancies determined by NMR can be fitted to a simple kinetic model that correctly predicts the extent of change in ee.

Full text of DOI:10.1021/jo990530h

8
794
J . Org. Chem. 1999, 64, 8794-8800
NMR Str u ctu r e Deter m in a tion of Ion P a ir s Der ived fr om Qu in in e:
A Mod el for Tem p la tin g in Asym m etr ic P h a se-Tr a n sfer Red u ction s
-
by BH
4
w ith Im p lica tion s for Ra tion a l Design of
P h a se-Tr a n sfer Ca ta lysts
†
Christine Hofstetter, Patricia Stone Wilkinson, and Thomas C. Pochapsky*
Department of Chemistry, Brandeis University, Waltham, Massachusetts 02454-9110
Received March 25, 1999
The solution structures of ion pairs formed by quarternary ammonium ions derived from quinine
-
-
alkaloid with small hard anions (BH
4 3
or Cl ) in CDCl have been characterized by nuclear magnetic
resonance methods. Structural observations have been correlated with the sense of asymmetric
induction observed in the phase-transfer reduction of 9-anthryl trifluoromethyl ketone by borohy-
-
dride (BH
4
) when catalyzed by the quaternary N-benzylquinine ammonium ion. From interionic
-
nuclear Overhauser effects (NOEs), it appears that the BH
4
ion occupies two of the four trigonal
pyramidal sites formed by substituents of the quarternary nitrogen of the catalyst cation. One of
these sites is in close proximity to the cation’s hydroxyl group that is strictly required for asymmetric
induction in the model reaction, while the other site is near the vinyl group on the cation. The
vinyl group does not appear to be important for determining the sense or extent of asymmetric
induction. Using energy-minimized structures derived from NMR data, it was predicted that the
N-(9-methyleneanthryl)quinine-quarternary ammonium catalyst would give improved asymmetric
induction in the model reaction due to a preferred anion occupancy at the site near the hydroxyl
group. An improvement in enantiomeric excess (ee) is observed using the anthryl-modified catalyst,
and NMR studies on the modified catalyst confirm the predicted change in anion binding site
occupancies. The changes in site occupancies determined by NMR can be fitted to a simple kinetic
model that correctly predicts the extent of change in ee.
2
In tr od u ction
Phase-transfer catalysis (PTC) often provides an at-
an indanone (94%) reported in 1987 by a group at Merck.
3
Since then, several other alkylation reactions and an
epoxidation have been achieved with high ee using chiral
4
tractive alternative for organic synthesis, as PTC reac-
tions generally proceed under relatively mild environ-
PTC. The common factor in these reactions is an inter-
mediate enolate anion capable of forming an intimate ion
pair with the catalyst cation. In this case, the catalyst is
no longer just an ion shuttle, but acts as a template for
directing the approach of the alkylating agent to one face
of the prochiral enolate. In all of these reactions, the
chiral phase-transfer catalysts have been quarternary
ammonium salts derived from Cinchona alkaloids. The
Cinchona alkaloids have been used successfully as a
source of chirality in a wide variety of synthetic trans-
formations.5
1
mentally friendly conditions. PTC was initially developed
to allow nucleophilic substitution upon an organic sub-
strate using small, relatively hard anions as nucleophiles
in biphasic mixtures. Typically, lipophilic cations such
as quarternary ammonium or phosphonium ions have
been used as catalysts in such reactions. The role of the
catalyst in such reactions is primarily that of an ion
shuttle. The cation forms a tight ion pair with the reagent
anion, permitting the anion to be transferred into the
organic phase, where it reacts with the organic substrate.
PTC has evolved considerably over the past 30 years.
The role of the catalyst has also changed, especially with
respect to asymmetric synthesis. In such cases, a chiral
nonracemic catalyst is used to introduce asymmetry into
the transition state of the reaction, thereby generating
preference for one enantiomer of the product. The use of
PTC for asymmetric syntheses has an added benefit in
that the chiral catalysts used are generally synthesized
from readily available, inexpensive starting materials
derived from the chiral pool.
In 1990, we began an extensive investigation into the
solution structure and dynamics of ion pairs in nonpolar
solvents with an eye toward the rational design of PTCs,
including chiral PTCs.6 After noting that the phase-
-
transfer reduction of prochiral ketones by BH
4
in the
presence of catalytic quantities of the quinine-derived
PTC 8R-9(R)-hydroxy-1-(phenylmethyl)cinchonanium (“N-
benzylquininium” or NBQ 1, see Figure 1) chloride
(2) Hughes, D. L.; Dolling, U.-H.; Ryan, K. M.; Schoenwaltd, E. F.;
Grabowski, E. J . J . J . Org. Chem. 1987, 52, 4745-4752.
Although there are many papers in the literature
concerning asymmetric PTC, most report low enantio-
meric excesses (ee). The first report of significant ee
under phase-transfer conditions was the methylation of
(
3) (a) O’Donnell, M. J .; Wu, S.; Huffman, J . C. Tetrahedron 1994,
0, 4507-4518, (b) Corey, E. J .; Xu, F.; Noe, M. C. J . Am. Chem. Soc.
1997, 119, 12414-12415. (c) Lygo, B.; Wainwright, P. G. Tetrahedron
5
Lett. 1997, 38, 8595-8598.
(
4) (a) Lygo, B.; Wainwright, P. G. Tetrahedron Lett. 1998, 39, 1599-
1
602. (b) Arai, S.; Tsuge, H.; Shiori, T. 1998, 39, 7563-7566.
*
To whom correspondence should be addressed. E-mail: pochap-
(5) Shiori, T.; Ando, A.; Masui, M.; Miura, T.; Tatematsu, T.;
sky@brandeis.edu. Website: http://tucano.chem.brandeis.edu.
Bohsako, A.; Higashiyama, M.; Asakura, C. In Phase-Transfer Cataly-
sis Mechanisms and Synthesis; Halpern, M. E., Ed.; American Chemi-
cal Society: Washington, DC, 1996; Chapter 11 (and references
therein).
†
Current address: Bruker Instruments, Inc. Billerica, MA.
1) Starks, C. M.; Liotta, C. L.; Halpern, M. Phase-Transfer Cataly-
(
sis; Chapman and Hall, New York, 1994; Chapter 1.
1
0.1021/jo990530h CCC: $18.00 © 1999 American Chemical Society
Published on Web 11/03/1999

Ion Pairs Derived from Quinine
J . Org. Chem., Vol. 64, No. 24, 1999 8795
Sch em e 1
features of the cation. One site is in the vicinity of the
vinyl substituent of the quinuclidine and is called the V
site, while the other is near the hydroxyl group located
on the carbon R to the quinoline ring and is called the
OH site.8 It has been established that the hydroxyl
group near the OH site is essential for asymmetric
b,9
-
10
induction in PTC-catalyzed BH
4
reductions, and our
hypothesis was that only anions bound at the OH site
contribute significantly to the observed ee (Scheme 1). If
this hypothesis is correct, then if one wished to improve
the ee, it would be beneficial to block the V site, thereby
8
b,9
increasing the relative occupancy of the OH site.
Attempts to block the V-site exclusively (e.g., by replacing
23 on the methylene carbon of the benzyl group with a
H
sterically bulkier substituent, see Figure 1) have so far
been unsuccessful. However, molecular modeling sug-
gested that the V and OH anion binding sites on NBQ
are not symmetric with respect to the plane of the benzyl
group.8 The V-site is roughly coplanar with and adja-
cent to the benzyl group, while the anion in the OH site
sits slightly off-plane (see Figure 2). These observations
suggested to us that the replacement of the benzyl group
with a larger substituent might at least partially ac-
complish this blocking.8b Although the 1-methylenenaph-
thyl catalyst 2a gave only marginally better results in
the PTC reduction of 9-anthryl trifluoromethyl ketone
than did the benzyl catalyst (see Scheme 1 and Table 1),
the 9-methyleneanthracenyl-derived catalyst 3a , (8R-
b,9
F igu r e 1. Ion pairs formed by NBQ 1, NNQ 2, and NAQ 3 as
described previously (refs 8b and 9) and in the current work.
The proton numbering scheme shown for NBQ 1 and NAQ 3
is that referred to for discussion of the NMR resonance
assignments.
9
(R)-hydroxy-1-(anthrylmethyl)cinchonanium chloride, or
NAQCl (Figure 1)), gave significant improvement in ee
Table 1).8b Recently, other groups have discovered
results in modest asymmetric induction in some cases,7
we characterized the tight ion pair formed by NBQ and
(
independently that the benefits of replacing the benzyl
group by the 9-methyleneanthracenyl group in cinchona-
derived catalysts extend to asymmetric alkylations and
epoxidations as well.3
-
the tetrahydridoborate (BH
4
) ions in order to understand
8
the origins of asymmetry in this type of reaction. We
hoped that the solution structure of the ion pair might
yield insights into the origin of the asymmetry induced
in the products and that this information might be used
to design improved chiral catalysts. On the basis of the
,4
We now report the results of our examination of the
solution structures of the NBQ and NAQ cations and the
ion pairs they form with chloride and borohydride in
nonpolar solution, to obtain insight into the origins of
asymmetric induction in reductions catalyzed by these
cations. The reduction used as a model reaction in the
present work (Scheme 1) was chosen because of catalyst
stability and ease of monitoring the results by chiral
HPLC. Although no intermediate substrate anion is
present to interact with the catalyst cation (hence, the
modest ee relative to some asymmetric alkylations), there
are also no strong bases present, so complications in-
curred by catalyst degradation observed in alkylation
observation of specific interionic NOEs, it was determined
-
that the BH
4
anion binds at only two of the four trigonal
pyramidal sites provided by the quaternary nitrogen of
the NBQ cation in chloroform solution.8a One of the
unoccupied sites is inside the fused quinuclidine ring
system and is thus inaccessible. The other unoccupied
site appears to be occluded by the benzyl group intro-
duced upon alkylation of quinine. The two anion-occupied
sites are described by their proximity to structural
(
6) (a) Pochapsky, T. C.; Stone, P. M. J . Am. Chem. Soc. 1990, 112,
6
714-6715. (b) Stone, P. M.; Pochapsky, T. C.; Callegari, E. Chem.
11
reactions were not present to interfere with the NMR
Commun. 1992, 2, 178-179. (c) Pochapsky, T. C.; Wang, A.; Stone, P.
M. J . Am. Chem. Soc. 1993, 115, 11084-11091. (d) Mo, H.; Pochapsky,
T. C. J . Phys. Chem. B 1997, 101, 4485-4486. (e) Mo, H.; Wang, A.;
Wilkinson, P. S.; Pochapsky, T. C. J . Am. Chem. Soc. 1997, 119, 11666-
experiments described here.
1
1673.
7) J ulia, S.; Ginebreda, A.; Guixer, J .; Masana, J .; Tomas, A.;
Colonna, S. J . Chem. Soc., Perkins Trans. 1 1984, 574.
8) (a) Pochapsky, T. C.; Stone, P. M. J . Am. Chem. Soc. 1991, 113,
460-1462. (b) Stone, P. M. Ph.D. Thesis, Brandeis University, 1993.
(9) Mo, H.; Pochapsky, T. C. Prog. NMR Spectrosc. 1997, 30, 1-38.
(10) Colonna, S.; Fornasier, R. J . Chem. Soc., Perkin Trans. 1 1978,
371-373.
(11) O’Donnell, M. J .; Wu, S.; Huffman, J . C. Tetrahedron 1994, 50,
4507-4518.
(
(
1

8
796 J . Org. Chem., Vol. 64, No. 24, 1999
Hofstetter et al.
yellow solid was suction filtered after the mixture had cooled.
The precipitate was then dissolved in ethanol and filtered, and
the clear filtrate was evaporated under reduced pressure. The
orange crystalline solid remaining was dissolved in a minimal
2 2
amount of CH Cl . Addition of an excess of diethyl ether
precipitated solid 3a from solution. The solid was separated
from solution by filtration and air-dried: mp 156-160 °C dec;
1
H NMR assignments (italicized number corresponds to proton
numbering in Figure 1, 298 K, CDCl
), 8.51 (s, 29), 8.40 (d, 25), 8.06 (d, 3), 8.04 (d, 30), 7.99 (d,
OH), 7.98 (d, 1), 7.95 (d, 28), 7.67 (m, 5 and 32), 7.58 (m, 26),
.49 (m, 27 and 31), 7.38 (dd, 4), 7.14 (d, 23), 7.12 (d, 8), 6.07
d, 24), 5.50 (m, 20), 5.29 (m, 16), 4.98 (d, 22), 4.90 (d, 21),
3
) δ 9.27 (d, 33), 8.7 (d,
2
7
(
4
1
1
1
1
1
3
.20 (m, 9), 3.97 (s, OCH ), 3.30 (m, 18), 2.99 (m, 15), 2.66 (dd,
9), 2.34 (m, 11), 2.32 (m,14), 2.20 (m, 17), 1.91 (broad, 12),
.51 (m, 13), 1.48 (m, 10); 13C NMR δ 157.9, 147.3, 144.2, 143.5,
36.5, 133.1, 132.8, 131.7 (2 C), 130.9, 130.7, 129.5, 128.6,
28.3, 127.7, 126.5, 125.4, 124.9, 124.3, 120.9 (2 C), 118.0,
17.6, 112.3, 70.1, 66.5, 60.9, 56.4, 52.1, 38.3, 25.8, 25.4, 22.5.
P r ep a r a tion of 8R-9(R)-Hyd r oxy-1-(a n th r ylm eth yl)-
cin ch on a n iu m Bor oh yd r id e (3b). A 0.2 g portion of 3a was
dissolved in 3 mL of methylene chloride, and 1.3 g of NaBH
4
was dissolved in 10 mL of water to which 3 drops of 3 M NaOH
had been added. Half of the aqueous solution was shaken with
the CH Cl solution of 3a in a separatory funnel for 2 min,
2 2
and the aqueous layer was removed. The remaining aqueous
solution was added to the organic layer and the procedure
repeated. The organic layer was dried over Na
2 4
SO and
evaporated under reduced pressure, leaving a yellowish crys-
1
F igu r e 2. Representation of BH -
talline solid: mp 144-160 °C dec; H NMR (298K, CDCl
3
) δ
4
binding sites with respect
8
7
7
7
2
.97 (d, 33), 8.66 (d, 2), 8.51 (s, 29), 8.19 (d, 25), 8.01 (d, 30),
to the plane of the benzyl ring in the NBQ cation in ion pair
b as determined by interionic NOEs and minimization of the
.98 (d, 3), 7.92 (d, 28), 7.85 (d, 1), 7.69 (m, 32), 7.62 (d, 5),
.55 (m, 26), 7.51 (m, 31), 7.45 (m,27), 7.29 (dd, 4), 7.26 (d, 8),
.05 (d, 23), 5.80 (d, 24), 5.41, (m, 20), 5.19 (m, 16), 4.99 (d,
1
resulting structures. The perspective shown is down the bond
between the quaternary nitrogen (shown in gray, center of the
figure) and the carbon bearing H (see Figure 1). The carbon
9
is eclipsed by the nitrogen. The benzyl group is edge-on to the
viewer. The 8-hydroxyl oxygen and the quinoline ring are both
indicated by text, as is the vinyl substituent of the quinuclidine
3
2), 4.87 (d, 21), 4.05 (m, 9), 3.85, (s, OCH ), 3.21 (m, 18), 2.73
(
dd, 19), 2.34 (m, 11), 2.33 (m, 14), 1.87 (brd, 12), 1.48 (m, 13),
1
1
1
.40 (m, 10); 13C NMR δ 158.0, 147.5, 144.2, 143.3, 136.5, 133.0,
32.7, 132.1, 131.9, 131.0, 130.8, 129.9, 128.6, 128.1, 126.2,
25.6, 125.0, 123.7, 120.9, 120.8, 117.7, 117.3, 101.7, 70.6, 66.2,
fused ring. The large circles represent the approximate
-
61.4, 56.8, 56.0, 52.5, 38.3, 25.8, 25.3, 22.0.
positioning of the BH
4
ion in the two occupied anionic binding
Con d ition s for P TC Rea ction s. A 0.137 g (0.5 mmol)
portion of 9-anthryl trifluromethyl ketone and the desired
amount of catalyst (see Table 1) was dissolved in 6.0 mL of
sites in 1b. The circle marked OH is near the hydroxyl group,
and is mostly above the plane of the benzyl ring in the
perspective of the figure. The V site (closer to the vinyl group)
is in the plane of benzyl ring.
3 4
CHCl . The aqueous layer contained 0.0114 g of NaBH (0.3
mmol) in 10 mL of distilled water. The two solutions were
combined and stirred rapidly for 24 h at 4 °C. The enantiomeric
excess of the resulting 2,2,2-trifluoro-1-(9-anthryl)ethanol, 5,
was determined by chiral HPLC on a Pirkle covalent D-
phenylglycine column (Regis Technologies) using 10% isopro-
pyl alcohol in hexane as the eluent.
Ta ble 1. En a n tiom er ic Excess (% ee) a s a F u n ction of
Ca ta lyst a n d Ca ta lyst Con cen tr a tion for th e Red u ction
of Keton e 4 u n d er P h a se-Tr a n sfer Con d ition s (Sch em e 1)
(
All Rea ction s Rep or ted in Th is Ta ble Wer e P er for m ed
a s Descr ibed in th e Exp er im en ta l Section Usin g 0.08 M
(
0.5 m m ol) Keton e 4 a n d 0.03 M (0.3 m m ol)
Aqu eou s Na BH4)
NMR. NMR experiments were performed on a Bruker
AMX500 500 MHz NMR spectrometer using an inverse-
detection probe equipped with three-axis gradients. Samples
for NOE experiments were freeze-thaw-degassed three times.
catalyst
catalyst concn (M)
% ee (S)-5
1
1
2
3
3
3
3
3
3
a
a
a
a
a
a
a
a
a
0.008a
0.004
0.008
21.4 ((0.1)
20.8 ((0.2)
22.8 ((0.2)
29.3 ((1.0)
30.1 ((1.2)
29.5 ((0.5)
29.5 ((0.4)
29.8 ((0.3)
30.1 ((0.6)
1
H resonance assignments are based primarily on phase-
sensitive double-quantum filtered COSY and phase-sensitive
b
1
2
two-dimensional rotating frame NOE (ROESY) spectra. All
NOE data described in the text were obtained from ROESY
spectra. Phase sensitivity for both experiments was obtained
using time-proportional phase incrementation (TPPI). DQF-
0.08
0.04
0.02
0.01
0.008
0.005
a
COSY spectra were acquired with 4K data points in t
observe) and 400 TPPI points in t . ROESY spectra were
acquired with 4K data points in t (direct observe) and 512 or
00 TPPI points in t . A compensated ROESY pulse sequence
2
(direct
1
2
a
Reproduced from ref 8b. b Taken from ref 8b.
6
1
was used to correct for cross-relaxation rate dependence on
frequency offset.13 A cw spin-lock field of 2.5 kHz was applied
in each experiment during a mixing time of 200 ms. Experi-
ments were performed with different transmitter offsets in
order to distinguish ROESY cross-peaks from HOHAHA-
Exp er im en ta l Section
All reagents were used as purchased without further
purification, except for CDCl , which was deacidified by
passage through activated dry neutral alumina for use with
3
-
BH -derived ion pairs.
P r ep a r a tion of 8R-9(R)-Hyd r oxy-1-(a n th r ylm eth yl)-
cin ch on a n iu m Ch lor id e (3a ). Quinine (3 mmol, Acros) and
-chloromethylanthracene (3 mmol, Aldrich) were added to 15
mL of benzene. The mixture was refluxed for 5 h, and the
4
(
12) Bax, A.; Grzesiek, S. In Encyclopedia of Nuclear Magnetic
Resonance; Grant, D. M., Harris, R. K., Eds.; J ohn Wiley & Sons: New
York, 1996; Vol. 7.
9
(
13) Griesinger, C.; Ernst, R. R. J . Magn. Reson. 1987, 75, 261-
271.

Ion Pairs Derived from Quinine
J . Org. Chem., Vol. 64, No. 24, 1999 8797
induced artifacts. Chemical shift assignments were measured
from 1D H spectra (16K data points) when possible; otherwise,
the data were obtained from the centers of COSY cross-peaks
Ta ble 2. Su m m a r y of In tegr a ted 2D NOE Volu m es
Mea su r ed betw een BH4 P r oton s a n d P r oton s on th e
NBQ 1b a n d NAQ 3b Ca tion s (Un its Ar e Ar bitr a r y a n d
Com p a r a ble On ly w ith in a Colu m n )
1
-
(4K data points in the directly detected dimension).
NMR diffusion measurements were made using a stimu-
NOE from
BH4 to H
NBQBH4 1b
NOE volume
NAQBH4 3b
NOE volume
lated echo sequence incorporating pulsed field gradients.14
7
A
1
0 ms diffusion delay was set between the two 5 ms gradient
3
2
2
2
1
1
8
9
3 (OH)
5 (V)
3 (OH)
4 (V)
6 (OH)
8 (V)
(OH)
(V)
a
a
3.2 ((0.8)
1.3 ((0.7)
pulses, with a 5 s relaxation delay between pulse trains. Each
series of eight experiments started with a gradient strength
of 0.5 G/cm and increased to 16.5 G/cm. A final experiment
was performed at a strength of 0.5 G/cm in order to compensate
for any sample degradation during the experiment. Diffusion
24 ((2)
23 ((2)
33 ((5)
27 ((7)
28 ((8)
30 ((11)
0.96 ((0.34)
0.47 ((0.15)
0.63 ((0.13)
0.34 ((0.11)
2.1 ((0.4)
1
4
coefficients were calculated as described previously.
1
3
1
D
C spectra were obtained on a Varian Inova 400 MHz
0.98 ((0.22)
NMR and were processed with a 5 Hz line broadening using
the software on the spectrometer.
a
Overlapped signals. Volume integral is 35 ((9).
Bruker NMR data were processed on a Silicon Graphics O2
workstation using Felix97 (MSI). A 0.5-3 Hz line broadening
was applied in the direct observe dimensions of all 2D
experiments, and a 75-90° shifted sinebell was applied to the
indirectly detected dimensions, which were then zero filled to
binding sites. On the basis of the relative intensities of
NOEs between BH4 and protons 25, 8, 9, 16, and 18 in
NBQBH 1b (Table 2), it was established that two sites,
4
-
V and OH (vide supra), are occupied equally on a time
2
K points. A zero- or first-order polynomial baseline correction
average.8
b,9
was required for processing of ROESY data in the indirect
detect dimension.
We wished to ascertain whether the observed site
occupancies have any influence upon the sense and
extent of asymmetric induction in phase-transfer reduc-
Diffusion experiments (16K points) were processed without
window functions. Baseline (polynomial) corrections were
performed as needed. Echo attenuation was measured from
absolute integral values, the log of which was plotted as a
function of gradient strength (see ref 14).
-
tions of prochiral ketones by BH4 catalyzed by NBQCl
1
4
a . We proposed a model for the complex between ketone
and the NBQ cation 1 that rationalized the observed
Com p u ta tion a l. Structures of individual cations were
minimized using MOPAC (PM3) or MM2 provided with Chem
sense of asymmetric induction in light of what is known
concerning structural requirements of the PTC (Figure
3
D Ultra (Cambridgesoft). Structures for the NBQ and NAQ
8
b
3
). It has been shown that the hydroxyl group R to the
cations were built in ChemDraw (Cambridgesoft) starting from
the crystallographic structure of quinine (Cambridge Crystal-
lographic database). Minimization calculations were performed
starting with the aromatic substituent of the quaternary
nitrogen coplanar with the quinoline ring, while the remainder
of the structure was left unperturbed relative to the crystal
structure of quinine. A grid search was used to confirm that
this conformation of the aromatic substituent was in the
neighborhood of the global minimum for the structure. No
counterions were used in the minimizations of individual
cations. Final structures were all found to be consistent with
observed NOEs.
quinoline ring (adjacent to the OH site) is important in
1
0
the generation of asymmetry in these reductions. Alkyl-
ation of the hydroxyl group greatly reduces the extent of
asymmetric induction, giving nearly racemic product in
15
Scheme 1. In our model, the hydroxyl group acts as a
proton donor to the carbonyl oxygen of the substrate,
while stacking interactions between the quinoline ring
of the catalyst and anthryl ring on the substrate orient
-
the Si face close to the BH
4
for delivery of the hydride
to the carbonyl carbon of the substrate, resulting in the
Energy minimizations of the ion pairs formed by NBQBH
4
formation of the (S)-enantiomer of 5 (Figure 3). Only
-
1
b used to establish the time-average positions of BH
4
in OH
-
reaction with BH
4
bound at the OH site is expected to
and V sites (Figure 2) were performed using a steepest-
result in significant ee in the product alcohol 5.
descents minimization routine in QUANTA/CHARMm (MSI).
-
-
If BH
4
bound to the OH site is responsible for the
An NBQ cation and BH
4
anion, the structures of which had
observed asymmetric induction, then blocking the V site
both been minimized separately, were placed relative to each
other in positions determined by experimentally observed
interionic NOEs. Minimization was carried out until no further
relative motion was observed. No obvious conformational
changes occurred in the NBQ cation as a result of the presence
of the anion.
-
(at which bound BH
4
is assumed to give rise to racemic
product) might be expected to improve the observed ee
-
by increasing the relative concentration of BH
4
in the
OH site. We made several attempts to block the V site
by N-alkylating quinine with R-substituted benzyl groups
(i.e., 1-chloro-1-phenylethane or diphenylchloromethane),
Resu lts
so that the proton adjacent to the V site would be
replaced by a sterically larger group. To date, these
attempts have been unsuccessful (vide supra). However,
inspection of an energy-minimized model for the struc-
Solu tion Str u ctu r e of th e NBQBH
a Mod el for Asym m et r ic In d u ct ion in t h e BH
4
Ion P a ir a n d
-
4
Red u ction of 9-An th r yl Tr iflu or om eth yl Keton e.
Our preliminary investigations of interionic NOEs in ion
4
ture of the NBQBH ion pair 1b suggested a possible
means of at least partially blocking the V site. As
described above, we observed that the V and OH sites
are not symmetrically disposed around the quaternary
nitrogen (Figure 2). We predicted that introduction of an
aryl group larger than benzyl should block the V site to
pair NBQBH
are moderately stable in CDCl
frame NOEs between specific protons on the cation and
4
1b showed that borohydride salts of NBQ
3
, and interionic rotating
-
the BH
4
hydride protons permit a clear identification
8
a
of anionic binding sites on the chiral cation. Further-
more, the ratios of volume integrals of the NOEs associ-
ated with particular ion binding modes permit an esti-
mate to be made of relative occupancies of the anionic
a greater extent than the OH site, and lead to preferred
-
BH
4
occupancy at the OH site. The alkylation of quinine
with 1-chloromethylnaphthalene produced catalyst NNQCl
2
a ; this salt gave only marginally higher ee than 1a in
(
14) Pochapsky, S. S.; Mo, H.; Pochapsky, T. C. J . Chem. Soc., Chem.
Commun. 1995, 2513-2514.
(15) Author’s unpublished results.

8
798 J . Org. Chem., Vol. 64, No. 24, 1999
Hofstetter et al.
F igu r e 3. Stereoview of MM2-minimized model for the NAQ cation 3-ketone 4 complex that rationalizes the observed enantiomeric
excess for the reduction of ketone 4 under phase-transfer conditions (Scheme 1). Carbon atoms are black, oxygen atoms are
-
-
white, and nitrogen, fluorine, and all BH
4
atoms are gray. Only the hydrogens of the NAQ hydroxyl and of the BH
4
anion are
shown; all other hydrogens are hidden for clarity. The anthryl group of the NAQ cation 3 in the foreground and the quinoline ring
-
in the lower background. The BH
4
is placed in the OH binding site (see Figure 2 and text). Ketone 4 (right side of figure) is
arranged so as permit hydrogen bonding between the carbonyl oxygen of 4 and the NAQ 3 hydroxyl proton (indicated by a dotted
-
4
preferential access
hydrogen to the carbonyl carbon of 4),
line) while stacking the anthryl ring of 4 over the quinoline ring of 3. The arrangement shown gives the BH
-
to the si face of the ketone for delivery of hydride (indicated by an arrow from one BH
4
yielding the (S)-alcohol. This figure was prepared using Chem3D (Cambridgesoft, Inc.).
the model reaction (Table 1). In retrospect, this result is
not surprising, since only one site can be blocked at a
time by the naphthyl ring, and there is no obvious
structural reason the V site should be preferentially
blocked relative to the OH site. However, catalyst 3a
derived from 9-chloromethylanthracene gave a significant
increase in ee relative to catalyst 1a (from 21% to 30%
ee), suggesting that our structure-based analysis was
correct.8b
Com p a r ison of Solu tion Con for m a tion s of NAQ
a n d NBQ a n d Ion -P a ir Str u ctu r es of NAQBH
NBQBH . We next characterized the solution structure
of the NAQBH ion pair 3b in order to determine whether
4
a n d
4
4
the predicted changes in relative anion occupancies of
the OH and V sites in fact took place. A comparison of
the energy-minimized structures of the NBQ 1 and NAQ
3
cations show that they are very similar (Figure 4). In
both cases, the quinoline ring is twisted out of the plane
defined by the aromatic substituent of the quaternary
nitrogen. Observed NOEs support the conformations
shown for both cations. NOEs obtained from ROESY
F igu r e 4. Stereoviews of MOPAC (PM3)-minimized struc-
tures for the NBQ 1 and NAQ 3 cations (protons not shown).
A counterion is shown occupying the V site (see Figure 2), but
was not present during the MOPAC calculation (see text). This
figure was prepared using Chem3D (Cambridgesoft, Inc.).
spectra of NAQBH
4
3b indicate the anthryl ring is
positioned with respect to the quinuclidine so that the
anthryl protons 25 and 33 are equidistant on a time
average from the quinuclidine protons 18 and 16, respec-
tively, and also to methylene protons 24 and 23, respec-
tively. Consider the plane defined by the anthryl ring
situated in the above manner with respect to the quinu-
clidine ring: The quinoline ring must be rotated out of
plane to explain the presence of NOEs between proton 5
on the quinoline ring and protons 8, 9, and 24 (Figure
occupancy of the OH site relative to the V site is larger
in NAQBH
the model in Figure 2. These NOES are shown in Figure
and summarized in Table 2. As noted above, relative
NOE intensities indicate approximately equal occupan-
4 4
3b than in NBQBH 1b, as predicted from
5
cies for both the OH and V sites in NBQBH
in the NAQBH
4
1b. However,
-
4
ion pair 3b, NOEs between BH
4
and
cationic protons 33, 23, 16, and 8 (all of which indicate
anion occupancy of the OH site) are approximately double
4
). No NOEs are observed for either NBQ or NAQ that
suggest significant populations of any other conformation
for the quinoline ring than that shown (see Figure 4).
A comparison of the interionic NOEs observed in ion
the intensity of those NOEs that indicate the presence
-
of BH
4
at the V site (those being interionic NOEs
-
between BH
2). On the basis of these observations, the BH
4
and protons 25, 24, 18, and 9; see Table
-
pairs NBQBH
4
1b and NAQBH
4
3b shows clearly that
4
is twice

Ion Pairs Derived from Quinine
J . Org. Chem., Vol. 64, No. 24, 1999 8799
-
4 4
F igu r e 5. Portion of a contour plot of a 500 MHz ROESY spectrum of NAQBH 3b showing the NOE cross-peaks between BH
protons and protons on NAQ cation (CDCl
3
, 0.08 M, 298 K). The spectrum was obtained using a 200 ms 2.5 kHz spin-locking
-
pulse. The horizontal axis (labeled D2) shows the region of the spectrum containing the resonance of the BH
4
protons, which is
1
11
split into a 1:1:1:1 quartet by a 90 Hz J BH coupling to the quadrupolar B (I ) 3/2, four spin states) Numbering of resonances
along the D1 axis corresponds to proton numbering in Figure 1.
Sch em e 2
as likely to occupy the OH site than the V site in the
NAQ cation 3.
Although it is difficult to compare absolute NOE
intensities between different experiments, interionic
NOEs in NAQBH
observed for NBQBH
4
were in general weaker than those
. At first, this was taken as
4
evidence that the ion pairs formed by NAQ were looser
than those formed NBQ. However, NMR diffusion mea-
the two anion-binding sites differ between NBQBH
4
1b
surements for NBQBH
4
and NAQBH
4
indicate that both
and NAQBH 3b. In the case of 1b, the OH and V sites
4
cation and anion in a given ion pair diffuse at the same
rate. This shows that the ion pair is diffusing as a single
species and, hence, is a tight ion pair.16 A comparison of
are populated approximately 1:1 based on relative inte-
grations of interionic NOEs characteristic of the two sites.
-
In the case of NAQBH
4
3b, the BH
4
population of the
the diffusion coefficients for 0.008 M solutions of NBQBH
4
OH site is twice that of the V site. Together, these
observations can be fit to a kinetic model that successfully
predicts the extent of change in asymmetric induction
observed as a function of relative populations of the two
1
1
b and NAQBH
4
3b show, as expected, that NBQBH
4
6
-
b diffuses at a slightly greater rate, (6.75 ( 0.06) × 10
2
-6
2
cm /s, than does NAQBH
It is possible that the generally lower intensities of the
interionic NOEs in NAQBH relative to NBQBH are the
4
3b (6.47 ( 0.15) × 10 cm /s.
anion-binding sites in the catalyst (see Scheme 2). In this
-
model, the BH
4
bound in OH site is responsible for all
4
4
result of the expected overall increase in the steric
barriers to close approach of the anion to the cation
caused by the introduction of the anthryl group.
The other major difference between the NBQ 1 and
NAQ 3 is that the ring flip for the phenyl group of the
NBQ cation is fast on the chemical shift time scale.
Protons 25 and 29 are equivalent, as are protons 26 and
asymmetric induction, consistent with experimental evi-
dence.1
one enantiomer of 5, (S)-5, is produced at the OH site
i.e., that this site contributes exclusively to the produc-
0,15
We make the further simplification that only
(
tion of the alcohol enantiomer found in excess), and that
the production of (S)-5 at the OH site is governed by a
rate constant k . Ketone reduction at the V site is
1
assumed to result in equal amounts of both enantiomers
of 5, and the rate constant governing the production of
2
each enantiomer (obviously the same for both) is k . The
differential rate expressions for the formation of each
enantiomer are then given by
2
8 at 298 K for 1a and 1b. In NAQ 3, all of the anthryl
protons are nonequivalent. A 4 s saturation of proton 33,
which is well separated from other proton resonances,
showed no effect in the signal intensity of proton 25 in
the anthryl ring, indicating that no significant saturation
transfer is occurring on the time scale of the experiment.
Thus, the anthryl ring does not reorient rapidly on either
d(S)
dt
)
k [OH] + k [V]
(1)
1
2
the chemical shift or T
1
time scales.
d(R)
dt
Discu ssion
) k [V]
(2)
2
We now attempt to rationalize our observations con-
cerning the model reaction shown in Scheme 1 and the
NMR structures of the ion pairs formed by NBQBH
and NAQBH 3b. To summarize experimental data, we
have found that upon replacement of catalyst 1a with
-
where [V] is the concentration of BH
OH] the concentration of BH
concentrations are in arbitrary units). Although absolute
rates have not been measured for the reaction in Scheme
, it is possible to obtain relative rates of formation of
the two enantiomers of 5: Because the reduction is
irreversible, the mole fraction of each enantiomer in the
product is proportional to the rate at which that enan-
tiomer is formed. In the case of the reaction catalyzed
by NBQCl 1a , the relative rates of formation of (S)-5 to
4
in the V site and
in the OH site (all
-
[
4
4
1b
4
1
3
3
1
a in the model reaction, the ee increased from ∼21% to
0%, with the (S)-enantiomer of alcohol 5 in excess (Table
-
). We also note that the relative populations of BH
4
in
(
16) Mo, H.; Pochapsky, T. C. J . Phys. Chem. B 1997, 101, 4485-
4
486.

8
800 J . Org. Chem., Vol. 64, No. 24, 1999
Hofstetter et al.
(
R)-5 (S/R) are ∼3:2 (20% ee). Substitution of these
has pointed out that the same kinetics can be explained
relative rates into eqs 1 and 2, along with equal values
of [V] and [OH] (reflecting the ∼1:1 ratios of NOE
intensities used to establish occupancies of anion binding
sites in 1b) gives
by a rapid disproportionation of the mixed alkoxy hy-
-
droborates to regenerate BH
4
, such that the actual
reducing agent is always BH
4
.18 We have found that the
-
observed ee for the reaction shown in Scheme 1 does not
change over the course of the reaction (which takes
several hours to go to completion under the conditions
described here), supporting the notion that mixed alkoxy
hydroborates (which should change in concentration with
the extent of reaction) are either unimportant or else do
d(S)
)
3 ) k [1] + k [1]
(3)
(4)
1
2
dt
d(R)
dt
)
2 ) k [1]
2
not change the enantioselectivity appreciably. Similarly,
-
neither the BH
4
:ketone ratio or the catalyst concentra-
which can be easily solved to give k
1
) 1, k
2
) 2.
tion have any appreciable effect on the stereochemical
If the population of the OH site is doubled relative to
the V site, as is observed for 3b, i.e., [OH] ) 2[V], using
the same rate expressions and rate constants, one obtains
outcome of the reduction. These observations all indicate
-
that BH
4
is the only reducing agent present in signifi-
cant concentrations under the reaction conditions of
Scheme 1.
d(S)
dt
One conclusion that can be drawn from the kinetic
analysis presented here is that, at least in theory, much
higher ees are possible than what have been observed to
date for phase-transfer reductions of prochiral ketones
)
4 ) k [2] + k [1]
(5)
(6)
1
2
d(R)
dt
)
2 ) k [1]
-
2
with BH
4
if improved blocking of the V site can be
accomplished. We are continuing to attempt to synthesize
quinine-based phase-transfer catalysts with the V site
blocked more or less completely. However, even the 30%
ee for the reduction of 4 obtained using catalyst 3b is
equal to the best reported ee for such phase-transfer
reductions in the literature.19 We chose the reaction in
Scheme 1 primarily for ease of monitoring ee (the
enantiomers of 5 are well resolved on Pirkle-type chiral
stationary phases). We have not as yet investigated how
effective catalyst 3b is for the asymmetric reduction of
other ketones.
The model thus predicts relative rates of formation of
S/R of 4:2 using 3a as a catalyst for the reaction in
Scheme 1. This corresponds to an ee of 33%, only slightly
larger than the 30% actually observed using 3a as a
catalyst (Table 1).
A number of other assumptions are implicit in this
kinetic model. First, it is assumed that ion pairs NBQBH
b and NAQBH 3b represent the reactive species in the
phase-transfer reductions catalyzed by the chloride salts
a and 3a , respectively. Because 1a and 3a are added
4
1
4
1
-
The current work demonstrates that structural infor-
mation concerning the active species in PTC (that is, the
ion pair) can be used in the design of more effective
phase-transfer catalysts. Although the asymmetric in-
duction observed for the reaction shown in Scheme 1 does
not reach the level of a useful synthetic tool, the net
improvement in the asymmetric induction and the agree-
ment between the predicted and observed outcomes of
the catalyst modifications in terms of ion site occupancy
accentuates the importance of structural information for
rational catalyst design. Furthermore, the success of
anthrylated chinchona PTCs for other applications sug-
gests that catalysts designed for one type of reaction may
only in catalytic amounts to the reaction (Table 1), BH
is present in large excess to Cl , so this is a reasonable
assumption. Second, it is assumed that BH
4
-
-
4
remains
paired with the catalytic cation throughout the reaction
-
process (i.e., BH
4
3
free in CHCl solution is not an
important contributor to the reaction) and that the
hydride-transfer step is fast relative to the rate at which
anions exchange from one site to another. This also is
reasonable, since the hydride transfer is a bond-making/
breaking step (on the order of a bond vibrational lifetime),
while the lifetime of a discrete ion-binding mode in a
closed shell ion pair of this type is many orders of
magnitude longer.6 It is also assumed that BH
c,e
-
4
is the
3
b,c
have unexepected benefits for other reactions as well.
only reducing species; that is, mixed alkoxy hydroborates
are not important reducing agents in this reaction. This
seems unlikely at first in light of kinetic evidence
suggesting that transfer of the first hydride is the rate-
Ack n ow led gm en t. This work was supported by a
grant from the NSF (CHE9729318).
-
17a,b
J O990530H
determining step in BH
4
reductions.
However, House
(
18) House, H. O. Modern Synthetic Reactions, 2nd ed.; W. A.
(
17) (a) Brown, H. C.; Mead, E. J .; Shoaf, C. J . J . Am. Chem. Soc.
Benjamin, Inc.: Reading, MA, 1972; p 52.
(19) Dehmlow, E. V.; Wagner, S.; Muller, A. Tetrahedron 1999, 55,
6335-6346.
1
1
956, 78, 3616. (b) Rickborn, B.; Wuesthoff, M. T. J . Am. Chem. Soc.
970, 92, 6894.