2-Aryl-4,4,8-trimethyl-2-phosphabicyclo[3.3.0]octanes: Reactive chiral phosphine catalysts for enantioselective acylation

Full text of DOI:10.1021/ja990133o

J. Am. Chem. Soc. 1999, 121, 5813-5814
5813
be alkylated, a finding that may have other synthetic applications.
2-Aryl-4,4,8-trimethyl-2-phosphabicyclo[3.3.0]octanes:
Reactive Chiral Phosphine Catalysts for
Enantioselective Acylation
Edwin Vedejs* and Olafs Daugulis
Chemistry Department
UniVersity of Wisconsin
Madison, Wisconsin 53706
ReceiVed January 14, 1999
Exploratory studies stimulated by the discovery of chiral
phosphine catalysis using 1 for enantioselective acylation1 have
encountered exceptional catalysts based on the 2-phosphabicyclo-
[3.3.0]octane (“PBO”) skeleton. The first member of this catalyst
family (2) was found to activate m-chlorobenzoic anhydride for
aroyl transfer in the kinetic resolution of phenyl tert-butyl carbinol.
While the enantioselectivity was promising, (s ) 14 at rt), we
were intrigued to find that the reaction was >100-fold faster than
with the original catalyst 1, and that 2 was sufficiently active to
induce acyl transfer with other anhydrides. However, the key
cyclization step in the synthesis gave only 7% of 2 from the bis-
mesylate 4 and PhPH2/BuLi, together with 53% of the unreactive
exo-phenyl isomer 3. We therefore considered modifications of
Diastereoselective reduction of 8b to 12 followed by treatment
with CH3SO2Cl/Et3N afforded the bis-mesylate 6, and reaction
with PhPH2/BuLi gave an encouraging 3:1 ratio in favor of the
desired 5a (36% yield). When the sequence was performed via
the precedented2 cyclic sulfate 13 in place of 6, followed by BH3
complexation, the ratio of 14a:15a increased to 36:1 (85%
isolated, 99.7% ee, using recrystallized 13), and a similar
experiment with 3,5-di[tert-Bu]PhPH2/BuLi afforded the corre-
sponding diastereomers 14b:15b with a remarkable isomer ratio
of 91:1 (85%). The stable borane adduct 14a (Ar ) Ph) was fully
characterized, including confirmation of relative and absolute
configuration by X-ray crystallography. Thus, 8 is formed with
inversion of lactate configuration in the alkylation step.
Phosphines 5a or 5b were released by warming 14 with
pyrrolidine and were evaluated as catalysts in the kinetic
resolutions of alcohols 16 (Table 1).3 The match of anhydride,
catalyst, and substrate proved important. In the unique case of
phenyl tert-butyl carbinol 16f, catalyst 5a was best, and the
combination with benzoic anhydride gave the highest enantiose-
lectivity s.4 For a variety of other aryl alkyl carbinols, catalyst
5b in combination with (i-PrCO)2O in heptane was most effec-
tive.5
A striking improvement in enantioselectivity became possible
when it was observed that acylations catalyzed by 5b have an
unusual temperature/reactivity profile (Table 1).6 A temperature
drop of 60 °C (rt to -40 °C) decreased typical acylation rates by
less than a factor of 10! The spectacular enantioselectivities
summarized in Table 1 result from the extraordinary range of
accessible reaction temperatures in combination with relatively
catalyst structure that might improve phosphorus diastereoselec-
tivity in the cyclization, and that might also be convenient for
enantiocontrolled synthesis of the bicyclic phosphine.
Catalyst 5 was selected as the target on the basis of the
following considerations. Cyclization from 6 should reflect the
relative stability of transition states similar to monoalkylated
lithiophosphide rotamers 7A and 7B. If the methyl branch point
destabilizes rotamer 7A by 1,3-interactions vs 7B, then the correct
phosphorus configuration could be favored. Ester 8, a logical
precursor of 7, should be accessible in nonracemic form if the
alkylation of 2,2-dimethylcyclopentanone enolate 9 with the
lactate triflate 10 can be controlled to give the correct diastere-
omer. This requires enantiofacial selectivity in the enolate
alkylation, and stereospecific displacement of the triflate.
Alkylation of 9 with 10a gave temperature-dependent mixtures
of 8a and the undesired 11a (1:1 at rt; 1:10 at -60 °C) in THF.
Similar product ratios, but poor conversions, were obtained in
toluene. In an attempt to favor a more highly ordered transition
state and to increase reactivity in toluene by incorporating greater
lithium ion affinity in the ester, the methoxyethoxyethyl lactate
triflate 10b was tested in the alkylation. This experiment gave
an inverted 9:1 ratio of 8b:11b on millimole scale (7-8:1 ratio
on multigram scale). The complementary selectivity of 10a and
10b toward the enolate 9 allows either enantiotopic face of 9 to
(2) Burk, M. J.; Feaster, J. E.; Nugent, W. A.; Harlow, R. L. J. Am. Chem.
Soc. 1993, 115, 10125.
(3) (a) Ruble, J. C.; Tweddell, J.; Fu, G. C. J. Org. Chem. 1998, 63, 2794
(b) Copeland, G. T.; Jarvo, E. R.; Miller, S. J. J. Org. Chem. 1998, 63, 6784.
(c) Oriyama, T.; Hori, Y.; Imai, K.; Sasaki, R. Tetrahedron Lett. 1996, 37,
8543. (d) Kawabata, T.; Nagato, M.; Takasu, K.; Fuji, K. J. Am. Chem. Soc.
1997, 119, 3169.
(4) Kagan, H. B.; Fiaud, J. C. Top. Stereochem. 1988, 18, 249.
(5) For less soluble alcohols, acylations with 5b can be performed in toluene
with a 2-3-fold decrease in rate and some loss in enantioselectivity, or less
well in oxygenated or chlorinated solvents.
(6) One possible explanation is that lower temperature increases the
equilibrium concentration of 17, but other possibilities will need to be evaluated
by kinetic studies.
(1) Vedejs, E.; Daugulis, O.; Diver, S. T. J. Org. Chem. 1996, 61, 430.
10.1021/ja990133o CCC: $18.00 © 1999 American Chemical Society
Published on Web 06/02/1999

5814 J. Am. Chem. Soc., Vol. 121, No. 24, 1999
Communications to the Editor
Table 1. Catalytic Kinetic Resolution of 16 Using 5/(i-PrCO)2Oa
this small level of catalyst contamination can result in substantially
underestimated values for s when selectivities are high.9 To test
this proposition, 5b was repurified (99.7% ee material recrystal-
lized twice). The ee for the resulting 5b could not be measured
with precision, but a repeat of entry 14 gave (R)-18g with 95.2%
ee at 50.7% conversion, and (S)-16g was recovered with 98.0%
ee, values clearly improved over the original results and corre-
sponding to s ) 188! The value predicted9 if 5b has been upgraded
from 99.7% ee to .99.9% ee is s ) 186. The corrected number
is highly sensitive to assay error in catalyst ee while the
experimental s is also dependent on assay precision, and on
predictable kinetic behavior in the resolution. Therefore, reliance
on corrected s values would not be advisable unless experimental
data are available for the purified catalyst. Furthermore, the exact
value of s is not important when selectivities are so high.
Nevertheless, the mathematical correction is relevant now that
there are examples where chiral phosphine catalysts can be
compared favorably with enzymatic catalysts.10 The latter presum-
ably have .99.9% ee, so that comparisons with the nonenzymatic
catalyst having 99.7% ee would be misleading. Of course, yields
and ee values near 50% conversion provide a more meaningful
way to evaluate exceptionally enantioselective catalysts for
practical use.
en-
try
%
ee ee
Ar′, R′
C6H5, CH3
struct
s
temp
h
cat conv 18b 16c
1
2
3
4
5
6
7
8
9
16a 22
16a 42
16b 10
16b 49
16c 18
16c 57
16d 14
16d 31
16e 20
16e 100
rt
1
4
5
4
42.4 84.0 61.9
C6H5, CH3
-20 °Cd
2.5 29.2d 93.3 38.4
5.5 41.0 72.3 50.2
6.6 50.4 88.2 89.8
3.7 39.3 83.0 53.6
3.9 51.3 88.6 93.3
1-cC6H9,eCH3
1-cC6H9,eCH3
C6H5, n-C4H9
C6H5, n-C4H9
C6H5, i-C4H9
C6H5, i-C4H9
C6H5, i-C3H7
rt
-40 °C 14
rt
2
8
-40 °C
rt
2.5 3.9 46.4 76.1 65.9
-40 °Cd
7
10
3.5 41.9d 88.2 63.6
3.3 33.5 85.9 43.2
2.8 46.9 94.8 83.8
4.0 53.1 78.5 88.7
4.9 45.8 93.1 78.7
3.2 44.4 89.6 71.6
3.5 50.1 94.9 95.3
rt
10 C6H5, i-C3H7
11 C6H5, t-C4H9
12 C6H5, t-C4H9
13 o-CH3C6H4, CH3
14 o-CH3C6H4, CH3
-40 °C 42
16f
16f
16g 39
16g 145
24f,g rt
12
67f,g -40 °C 65
rt
1
4
-40 °C
15 o-CH3OC6H4, CH3 16h 38
rt
1.5 5.6 36.5 91.5 52.5
16 o-CH3OC6H4, CH3 16h 81g -40 °C 9.5 6.8 28.7 96.5 38.8
17 R-naphthyl, CH3
18 R-naphthyl, CH3
19 mesityl, CH3
16i
16i
16j
41
rt
1
7
10
2.7 42.0 90.9 65.8
3.9 29.8 97.0 41.2
3.8 40.2 79.2 53.4
99g -40 °C
15g,h rt
20 mesityl, CH3
16j 369g,h,i -40 °C 16 12.1 44.4 98.7 78.8
a All reactions with 99.7% ee 5b in heptane, 0.1 M substrate, unless
noted; rt reactions were done without positive temperature control; -40
°C experiments were controlled; absolute configuration by comparison
with the sign of optical rotation for the known alcohols; ee determination
by HPLC or GLC (entries 3, 4); empirical s values; see Supporting
Information for s values corrected for 99.7% ee catalyst. b Product ee
after saponification. c Unreacted substrate ee. d Reaction at 0.05 M
One added example 16j was studied using twice recrystallized
5b and isobutyric anhydride (entries 19,20). Despite the modest
s ) 15 measured in toluene at room temperature, the -40 °C
experiment gave s ) 369 and 389 in duplicate runs. A promising
value of s ) 21 was measured in heptane at room temperature,
but the corresponding experiment at -40 °C was not feasible
due to solubility limitations. Alcohol 16j also proved to be an
excellent substrate for kinetic resolution when 5b was used in
combination with acetic anhydride, and s ) 112 was measured
at -40 °C in toluene (51.8% conversion; 16j, 98.6% ee, product
18j, 91.6% ee after saponification).
f
substrate. e 1-Cyclohexenyl. Catalyst ) 5a; anhydride ) Bz2O. g Tolu-
ene solvent. h 5b with >99.9% ee was used. i s ) 389 in a duplicate
run.
typical temperature effects on ∆∆G*. Preliminary experiments
indicate that some of the unhindered substrates can be acylated
even at -78 °C. Especially rapid acylations were observed using
acetic anhydride, but enantioselectivities were significantly lower.7
Acylation rates are qualitatively proportional to the concentra-
tion of the anhydride as well as the catalyst, consistent with
reversible formation of transient 17 and rate-determining conver-
sion to 18. This information can be useful for designing
preparative scale experiments using <1% catalyst. However, the
data in Table 1 underestimate catalyst reactivity because no special
precautions were taken to exclude oxygen (nitrogen bypass
system; 3-6% 5b). In a typical experiment, this results in 15-
30% loss of catalyst due to phosphine oxide formation (NMR
assay). For extrapolation to preparative scale, a more rigorous
test experiment is recommended to establish % conversion vs
time. Thus, entry 14 was repeated on 0.1 mmol scale in
deoxygenated heptane using 0.8 ( 0.1 mol % 5b, resulting in
51.2% conversion to 18g after 14h. A gram-scale (7.5 mmol)
kinetic resolution of o-methylphenyl-1-ethanol, (rac)-16g, was
then performed under similar conditions (deoxygenated heptane),
but using 0.6 mol % catalyst (16 mg; 0.045 mmol). After 14 h at
-40 °C, isopropylamine was added to quench the anhydride, and
ester (R)-18g (48.5% conversion) was obtained with 95.7% ee
(48% isolated), while 46% of (S)-16g was recovered (90.2% ee;
s ) 142, preparative scale). According to NMR assay, 5b survived
acylation and workup (<10% phosphine oxide formation), but
catalyst recovery was not explored because 5b has been prepared
on multigram scale.8
Further studies are planned to explore the promising allylic
alcohol substrates (entry 4), to probe structure/enantioselectivity
correlations of Ar-PBO catalysts for a variety of applications,
and to better understand transition state preferences.11
Acknowledgment. This work was supported by the National Science
Foundation. The authors also thank Dr. D. R. Powell for the X-ray crystal
structure of 5a, and Mr. K. Gentile for the preparative scale synthesis of
5b.
Supporting Information Available: Experimental procedures, char-
acterization of new compounds, ee assay data, and X-ray data for 14b
(PDF). This material is available free of charge via the Internet at
http://pubs.acs.org.
JA990133O
(7) For example, s ) 10 for 16a using 5b/Ac2O in toluene at -40 °C.
(8) Upon request from North American research groups, 15 mg samples
of 14b can be provided on a limited basis to test new alcohol resolutions in
exchange for data.
(9) Ismagilov, R. F. J. Org. Chem. 1998, 63, 3772. The correction can
become quite large if s is large because the contaminating catalyst enantiomer
has high reactivity toward the (S)-alcohol, and because the (S):(R) ratio
increases as conversion approaches 50%.
(10) Sih, C. J.; Wu, S.-H. Top. Stereochem. 1989, 19, 63. Chen, C.-S.;
Sih, C. J. Angew. Chem., Int. Ed. Engl. 1989, 28, 695. Klibanov, A. M. Acc.
Chem. Res. 1990, 23, 114. Drueckhammer, D. G.; Hennen, W. J.; Pederson,
R. L.; Barbas, C. F., III; Gautheron, C. M.; Krach, T.; Wong, C.-H. Synthesis
1991, 499. Roberts, S. M. Chimia 1993, 47, 85.
(11) Early results suggest that cyclic analogues of 16 (Ar′ joined to R′) do
not match the transition-state requirements for good selectivity with 5b/(i-
PrCO)2O: s < 2 for R-indanol or R-tetralol at rt.
The data in Table 1 were obtained using 5b prepared from 13
with 99.7% ee. According to a recent mathematical analysis, even