A comparison of monocyclic and bicyclic phospholanes as acyl-transfer catalysts

Article abstract of DOI:10.1021/jo030007+

The synthesis and evaluation of chiral phosphines 11, 15a, 19a, 24a, and 28a as nucleophilic catalysts for anhydride activation and kinetic resolution of alcohols is described. The relative reactivity follows the order 11a > 11b > 15a > 1 in the monocyclic series, and 24a > 19a > 28a in the bicyclic series, with an overall rate advantage of ca. 2 orders of magnitude for the bicyclic phospholanes over the monocyclic analogues. The increased reactivity of the bicyclic phospholanes for the acylation of alcohols is attributed to conformational effects and ground-state destabilization in a highly associative mechanism. Kinetic resolution data demonstrate promising enantioselectivities for 24a.

Full text of DOI:10.1021/jo030007+

A Com p a r ison of Mon ocyclic a n d Bicyclic P h osp h ola n es a s
Acyl-Tr a n sfer Ca ta lysts
E. Vedejs,*,† O. Daugulis,‡ L. A. Harper,‡ J . A. MacKay,† and D. R. Powell‡
Chemistry Department, University of Wisconsin, Madison, Wisconsin 53706,
and Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109
edved@umich.edu
Received J anuary 9, 2003
The synthesis and evaluation of chiral phosphines 11, 15a , 19a , 24a , and 28a as nucleophilic
catalysts for anhydride activation and kinetic resolution of alcohols is described. The relative
reactivity follows the order 11a > 11b > 15a > 1 in the monocyclic series, and 24a > 19a > 28a
in the bicyclic series, with an overall rate advantage of ca. 2 orders of magnitude for the bicyclic
phospholanes over the monocyclic analogues. The increased reactivity of the bicyclic phospholanes
for the acylation of alcohols is attributed to conformational effects and ground-state destabilization
in a highly associative mechanism. Kinetic resolution data demonstrate promising enantioselec-
tivities for 24a .
SCHEME 1
In 1993, a report from our laboratory demonstrated
that tributylphosphine is comparable to p-(dimethylami-
no)pyridine as a nucleophilic catalyst for the acylation
of alcohols by anhydrides.1,2 The P-arylphosphines were
less effective than is Bu3P, but dialkylarylphosphines
retained sufficient reactivity to activate anhydrides at
room temperature. These observations stimulated a
survey of chiral dialkylarylphosphines as potential enan-
tioselective acylation catalysts, and initial screening
identified 2,5-dimethylphospholane 13 as a promising
nucleophilic catalyst (Scheme 1). Phospholane 1 activated
m-chlorobenzoic anhydride for chlorobenzoylation of al-
cohols via the transient ion pair intermediate 2. In the
case of racemic phenyl-tert-butylcarbinol (3), the chlo-
robenzoylation occurred with significant enantioselectiv-
ity (s ) kfast/kslow ) 13-15).4 However, the reaction was
very slow and required ca. two weeks to reach 25%
conversion to 4 at room temperature using 15% of the
catalyst. Furthermore, other dialkylphenylphosphines
such as 55 and 66 were unreactive under the same room
temperature conditions, while 7 and 85 were marginally
reactive but less selective compared to 1.
The initial results were encouraging because the
enantioselectivity for the conversion from 3 to 4 was
relatively high. After a long history of work in many
laboratories,7 this was the first example to demonstrate
s > 10 for a kinetic resolution based on acyl transfer
using a nonenzymatic catalyst.8 On the other hand,
catalyst reactivity was a major concern, and a large effort
had to be initiated to define the factors that might
† University of Michigan.
‡ University of Wisconsin.
(1) (a) Vedejs, E.; Diver, S. T. J . Am. Chem. Soc. 1993, 115, 3358.
(b) Vedejs, E.; Bennett, N. S.; Conn, L. M.; Diver, S. T.; Gingras, M.;
Lin, S.; Oliver, P. A.; Peterson, M. J . J . Org. Chem. 1993, 58, 7286.
(2) Ho¨fle, G.; Steglich, W.; Vorbru¨ggen, H. Angew. Chem., Int. Ed.
Engl. 1978, 17, 569. Scriven, E. F. V. Chem. Soc. Rev. 1984, 12, 129.
(3) (a) Vedejs, E.; Daugulis, O.; Diver, S. T. J . Org. Chem. 1996, 61,
430. (b) Burk, M. J .; Feaster, J . E.; Harlow, R. L. Tetrahedron:
Asymmetry 1991, 2, 569.
(4) Kagan, H. B.; Fiaud, J . C. Top. Stereochem. 1988, 18, 249.
(5) Vedejs, E.; Daugulis, O. Latvian J . Chem. 1999, 1, 31.
(6) Chen, Z.; Zhu, G.; J iang, Q.; Xiao, D.; Cao, P.; Zhang, X. J . Org.
Chem. 1998, 63, 5631. We are grateful to Prof. X. Zhang for providing
a generous sample of 6 as the borane complex.
(7) Wegler, R. Liebigs Ann. Chem. 1932, 498, 62. Wegler, R.; Ru¨ber,
A. Chem. Ber. 1935, 68, 1055. Bird, C. W. Tetrahedron 1962, 18, 1.
Weidmann, R.; Horeau, A. Bull. Soc. Chim. Fr. 1967, 117. Stegman,
W.; Uebelhart, P.; Heimgartner, H.; Schmid, H. Tetrahedron Lett. 1978,
19, 3091. Mukaiyama, T.; Tomioka, I.; Shimizu, M. Chem. Lett. 1984,
49. Ichikawa, J .; Asami, M.; Mukaiyama, T. Chem. Lett. 1984, 949.
Duhamel, L.; Herman, T. Tetrahedron Lett. 1985, 26, 3099. Weidert,
P. J .; Geyer, E.; Horner, L. Liebigs Ann. Chem. 1989, 533. Potapov, V.
M.; Dem’yanovich, V. M.; Klebnikov, V. A.; Korovina, T. G. Zh. Org.
Khim. 1986, 22, 1218 (1095 Engl). Evans, D. A.; Anderson, J . C.;
Taylor, M. K. Tetrahedron Lett. 1993, 34, 5563. Ishihara, K.; Kubota,
M.; Yamamoto, H. Synlett 1994, 611.
10.1021/jo030007+ CCC: $25.00 © 2003 American Chemical Society
Published on Web 05/29/2003
5020
J . Org. Chem. 2003, 68, 5020-5027

Monocyclic and Bicyclic Phospholanes as Catalysts
TABLE 1. m -Ch lor oben zoyla tion of 3 Ca ta lyzed by
SCHEME 2
Mon ocyclic P h osp h ola n esa
entry
catalyst
rel rate
s
1
2
3
4
1
1
36
4
13b
5
5.1
5.6
11a
11b
15a
2
a
All experiments at room temperature in dichloromethane
b
using 2.5 equiv of m-chlorobenzoic anhydride. Ref 3a.
phosphine boranes 12a and 12b (7.3:1 ratio). The borane
complexes were stable to chromatography, and the dia-
stereomers as well as the enantiomers could be separated
by HPLC. Enantiomerically enriched samples of the
phosphines 11a and 11b were then obtained by brief
warming with pyrrolidine or Et2NH. The major isomer
11a was confirmed to have the cis stereochemistry by
X-ray crystallographic analysis of the derived methiodide
salt.
With diastereomers 11a and 11b available, qualitative
reactivity comparisons were carried out using the m-
chlorobenzoylation of 3 as the test reaction. Significant
rate improvement was observed with both diastereomers
compared to 1 (trans-isomer 11b, 6-fold faster; cis-isomer
11a , ca. 36-fold faster according to comparisons of initial
rates), but enantioselectivity was decreased (s ) 5-5.1).
Furthermore, an attempt to improve enantioselectivity
by introducing additional steric constraints quickly en-
countered reactivity limitations. Thus, 2-tert-butylphos-
pholane 15a , available via a similar sequence from diol
13,13 gave a small increase in enantioselectivity (Table
1, entry 4; s ) 5.6). However, reactivity dropped consid-
erably compared to that of 11a .
increase the rate of acylation. During this search, im-
portant advances from other groups began to appear,9
and several intriguing nitrogen-based esterification cata-
lysts for acyl-transfer reactions are now known that are
capable of potentially practical levels of enantioselectivity
with s > 30, depending on the alcohol substitution
pattern.10
Syn th esis
a n d
Rea ctivity
of P h osp h ola n e
Der iva tives
In an effort to improve catalyst reactivity in the chiral
phosphine series, modifications of the lead structure 1
were considered that would improve access to the un-
shared electron pair at phosphorus. Removal of one of
the adjacent methyl substituents was the logical first
step, and our optimization studies began with the inves-
tigation of 2-monosubstituted phospholanes (Scheme 2).
Diastereomeric 2-methyl-1-phenylphospholanes 11a and
11b (previously reported as a mixture of isomers)11 were
prepared from the diol 9 via the known bismesylate 1012
by treatment with PhPH2/BuLi, and were isolated as the
Related structures were considered that might retain
the promising reactivity of 11a while allowing at least
some potential for structural modification. The bicyclic
phospholane 19a was one attractive possibility. The
electron pair at phosphorus would be more accessible
because the fused cyclopentane ring should increase the
P-CH-CH2 bond angle in 19a compared to the corre-
sponding P-CH-CH3 bond angle in 11a . Nucleophilic
reactivity might also be influenced by conformational
preferences resulting from repulsive interactions between
the fused cyclopentane ring and the P-phenyl substituent
in 19a . Compared to the situation in 11a , the additional
bulk on the lower face of 19a should favor P-phenyl
rotamers where the phenyl group is turned more toward
the adjacent bridgehead C-H bond and away from the
ring CH2 groups. This effect might destabilize the ground
state relative to the transition state for nucleophilic
catalysis, and could increase the reaction rate.
(8) Reviews: 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.
(9) Reviews: J arvo, E. R.; Miller, S. J . Tetrahedron 2002, 58, 2481.
Spivey, A. C.; Maddaford, A.; Redgrave, A. J . Org. Prep. Proced. Int.
2000, 32, 331.
(10) Ruble, J . C.; Latham, H. A.; Fu, G. C. J . Am. Chem. Soc. 1997,
119, 1492. Ruble, J . C.; Tweddell, J .; Fu, G. C. J . Org. Chem. 1998,
63, 2794. Bellemin-Laponnaz, S.; Tweddell, J .; Ruble, J . C.; Breitling,
F. M.; Fu, G. C. Chem. Commun. (Cambridge) 2000, 1009. (c)
Kawabata, T.; Nagato, M.; Takasu, K.; Fuji, K. J . Am. Chem. Soc. 1997,
119, 3169. (d) Spivey, A. C.; Fekner, T.; Spey, S. E. J . Org. Chem. 2000,
65, 3154. Spivey, A. C.; Maddaford, A.; Fekner, T.; Redgrave, A. J .;
Frampton, C. S. J . Chem. Soc., Perkin Trans. 1 2000, 3460. (e) Sano,
T.; Miyata, H.; Oriyama, T. Enantiomer 2000, 5, 119. Sano, T.; Imai,
K.; Ohashi, K.; Oriyama, T. Chem. Lett. 1999, 265. Oriyama, T.; Hori,
Y.; Imai, K.; Sasaki, R. Tetrahedron Lett. 1996, 37, 8543. (f) J arvo, E.
R.; Copeland, G. T.; Papaioannou, N.; Bonitatebus, P. J ., J r.; Miller,
S. J . J . Am. Chem. Soc. 1999, 121, 11638. Copeland, G. T.; J arvo, E.
R.; Miller, S. J . J . Org. Chem. 1998, 63, 6784.
The bicyclic phosphines 19a and 19b were made in the
usual way from the known diol 1714 via the bismesylate
18.14 However, in contrast to the behavior of the simpler
bismesylate 10, the ring-fused analogue 18 reacted with
PhPH2/BuLi to give a 1:6.7 mixture in favor of the
undesired exo-phenyl diastereomer 19b over 19a . The
diastereomers of the corresponding borane complexes 20a
(12) Kim, M.-J .; Lim, I. T.; Choi, G.-B.; Whang, S. J .; Ku, B.-C.; Choi,
J .-Y. Bioorg. Med. Chem. Lett. 1996, 6, 71.
(13) Loewen, P. C.; Makhubu, L. P.; Brown, R. K. Can. J . Chem.
1972, 50, 1502.
(11) Quin, L. D.; Somers, J . H.; Prince, R. H. J . Org. Chem. 1969,
34, 3700.
(14) Procha´zka, M.; Cˇ erny´, J . V.; Sm´ısˇek, M. Collect. Czech. Chem.
Commun. 1966, 31, 1315.
J . Org. Chem, Vol. 68, No. 13, 2003 5021

Vedejs et al.
SCHEME 3
and 20b were difficult to separate, and only 4% of racemic
20a was recovered from the mixture. Fortunately, this
provided sufficient material for deprotection to the phos-
phine 19a by brief warming with pyrrolidine, and allowed
evaluation of the rate of m-chlorobenzoylation of 3 using
racemic 19a . Remarkably, 19a was more reactive than
11a by at least 10-fold in the qualitative comparison.
Indeed, the reactivity was so much better in the case of
less hindered alcohols that it became difficult to make
comparisons under the usual room temperature condi-
tions with m-chlorobenzoic anhydride and the procedures
had to be modified as discussed later. On the other hand,
the synthesis and purification of 20a were difficult and
the HPLC separation of enantiomers was challenging in
the initial studies. The accumulation of preparative
obstacles raised concerns about pursuing this series of
structures. We therefore elected to investigate the more
highly hindered bicyclic analogue 24a in the hope that
reactivity would be retained and that the synthesis and
purification might be easier (Scheme 3).
The first attempts5 to prepare 24a from keto ester 2115
by the usual sequence via 2216 and the bismesylate 23
encountered the same problems as with 19a . Thus,
reaction of 23 with PhPH2/BuLi gave an unfavorable 1:7.6
mixture of 24a /24b. Fortunately, much better results
were obtained by changing the leaving groups. The key
improvement was to use a cyclic sulfate, 26, in place of
the bismesylate 23. Conversion of diol 22 into 26 was
performed following a 1,4-diol analogy by Burk et al.
(SOCl2, NaIO4/RuCl3),17 and the sequence afforded an
80% yield of crystalline 26. The subsequent reaction with
PhPH2/BuLi gave a mixture of 24a and 24b, and conver-
sion to the corresponding borane complexes resulted in
a 16:1 ratio of isomers 25a /25b. The desired major isomer
was isolated in 64% yield as a crystalline solid. Enan-
tiomerically enriched 25a was then obtained after HPLC
on a chiral support, and the relative stereochemistry and
the absolute configuration were established by X-ray
crystallography (Figure 1).
F IGURE 1. Conformers of 25a in the crystal lattice.
Compared to the analogous reaction of the bismesylate
23, the conversion from cyclic sulfate 26 to 24 occurred
with considerably higher selectivity (16:1), and with an
inverted preference for the formation of the more hin-
dered endo-phenyl diastereomer 24a . Even higher selec-
tivity has been encountered recently in a closely related
cyclization reaction using 3,5-disubstituted aryl phos-
phide reagents, as described in another paper from our
laboratory.18 The mechanistic details of this reaction are
poorly understood, but the initial event is likely to be the
SN2 displacement at the less hindered primary carbon
of 26. The dramatic change in diastereoselectivity at
phosphorus starting from the cyclic sulfate 26 compared
to the bismesylate 23 is therefore due to the behavior of
a monoalkylated phosphide intermediate, 27, in the
subsequent cyclization step. Formation of the endo-
phenyl diastereomer 24a probably requires intramolecu-
lar backside attack on the sulfate leaving group (OSO3Li)
in 27. On the basis of this assumption, the stereochem-
istry of 27 can be deduced from that of the product 24a .
However, we cannot explain why the difference in leaving
groups (MsOLi from 23, LiOSO3Li from 26) results in
the indicated phosphorus configurations. Presumably,
simple steric effects dominate transition-state preferences
starting from 23, and lead to the less hindered (undes-
ired) exo-phenyl diastereomer 24b. By implication, in-
teractions among the phosphorus unshared electron
pairs, monoalkyl sulfate oxygens, and two lithium ions
are responsible for the predominant formation of the
endo-phenyl diastereomer in the conversion from 26 to
27 to 24a , but the nature of the interactions is not
understood in detail. In any event, the much improved
diastereoselectivity profile in the cyclization reaction
starting from the cyclic sulfate 26 greatly simplifies the
(15) Blanc, M. G.; Haller, M. A. C. R. Hebd. Seances Acad. Sci. 1908,
146, 79.
(16) Diol 22 is formed with 25:1 selectivity for the trans-diastere-
omer.
(17) Burk, M. J .; Feaster, J . E.; Nugent, W. A.; Harlow, R. L. J .
Am. Chem. Soc. 1993, 115, 10125.
(18) Vedejs, E.; Daugulis, O. J . Am. Chem. Soc. 2003, 125, 4166.
5022 J . Org. Chem., Vol. 68, No. 13, 2003

Monocyclic and Bicyclic Phospholanes as Catalysts
SCHEME 4
SCHEME 5
complexes 33 obtained after treatment with THF-
borane. Furthermore, the phosphine derived from the
major borane complex proved to be unreactive as a
catalyst for anhydride activation under the standard
conditions, while the minor phosphine was reactive. In
all of the other bicyclic phosphines, the more reactive
diastereomer has proven to be the endo-phenyl isomer.
By that argument, the desired phosphine 28a must be
the minor cyclization product from the reaction of 31 with
PhPH2/BuLi, while the major isomer is the relatively
unreactive 28b. The cyclization diastereoselectivity fa-
voring the exo-phenyl isomer is therefore opposite by
comparison with the reaction of the analogous cyclic
sulfate 26 with PhPH2/BuLi, apparently due to differing
intramolecular interactions in the phosphide intermedi-
ate 32 compared to 27. Despite the disappointing dias-
tereoselectivity, the phosphine borane complexes 33a and
33b could be separated, purified, and subjected to HPLC
on a chiral support for resolution of enantiomers. De-
complexation by brief warming with pyrrolidine was then
conducted in the usual way, and provided sufficient
enantiomerically enriched 28a to evaluate reactivity and
enantioselectivity.
synthesis of 24a , and allows preparation of sufficient
material to evaluate the new phosphine.
Enantiomer separation of the borane complex 25a
using HPLC on a chiral support gave the individual
enantiomers with 63-85% ee, and crystallization of 25a
provided a further upgrade to afford a small amount of
material with >99% ee. The stereochemistry and the
absolute configuration of the crystallized sample were
established unambiguously using X-ray crystallography
(see the Supporting Information). The relative HPLC
retention time behavior of the individual enantiomers on
a chiral support was then used to develop an empirical
correlation with the enantiomers of the analogous phos-
phine 20a and of related phosphines to be discussed
shortly. Together with the enantioselectivity profiles of
the related bicyclic phosphines, this empirical data
allowed tentative assignments of absolute configuration
in all of the bicyclic phosphines.
Preliminary experiments demonstrated that 25a is
decomplexed by brief warming with pyrrolidine and that
the resulting 24a is more reactive than 19a in the
m-chlorobenzoylation of 3. We therefore carried out the
synthesis of one additional analogue in the bicyclic series
in the hope that further insight might be obtained
regarding the relationship between phosphine geometry
and catalytic reactivity. If the improved reactivity of
2-phosphabicyclo[3.3.0]octane derivative 19a or 24a com-
pared to the monocyclic phospholanes results from im-
proved access to phosphorus resulting from a small
increase in the P-CH-CH2 bond angle compared to the
situation in 11a or 15a , then a substantially faster rate
should be seen with the phosphetane analogue 28a where
the corresponding angle would be larger. To test this
proposal, the synthesis of 28a was carried out as de-
scribed in Scheme 4, using a small variation of the
approach that was successful for the preparation of 24a .
Conversion from the known keto ester 2919 to the cyclic
sulfate 31 was routine, but the subsequent phosphide
cyclization procedure gave a diastereomer ratio of only
3:1 according to NMR assay of the phosphine borane
Rela tive Rea ctivity Stu d ies w ith th e Bicyclic
P h osp h ola n e Ca ta lysts
As already mentioned, the reactivity of the bicyclic
catalysts was too high for convenient evaluation using
the prior conditions with m-chlorobenzoic anhydride,
especially for relatively nonhindered substrates such as
1-(1-naphthyl)ethanol (34) (Scheme 5). In principle, the
rate could be adjusted as needed by limiting the amount
of catalyst, but that approach gave less reproducible
reactivity data, perhaps due to competing oxidation of
the catalyst by residual oxygen in reagents or solvents.
More consistent rates were obtained by decreasing the
reactivity of the anhydrides, by maintaining relatively
(19) Cefelin, P.; Lochmann, L.; Stehlice, J . Collect. Czech. Chem.
Commun. 1973, 38, 1339. Raftery, M. J .; Bowie, J . H. Org. Mass
Spectrom. 1988, 23, 719.
J . Org. Chem, Vol. 68, No. 13, 2003 5023

Vedejs et al.
TABLE 2. Acyla tion s Ca ta lyzed by Bicyclic
P h osp h ola n esa
In principle, an equilibrium between 35 and 36 is also
plausible, although the P-acylphosphorane subunit of 36
lacks direct precedent.
entry
catalyst
alcohol
(RCO)2O
rel rate
s
A surprisingly small (less than 5-fold) decrease in
acylation rates was observed in representative examples
as the reaction temperature was lowered by 50-60 °C
using the bicyclic phospholane catalysts. This result
implies that activation enthalpy is relatively small, and
that the activation barrier is dominated by entropic
factors.21-23 Rates were therefore determined over a broad
temperature range to establish the activation parameters
for acylation of 34 using the most reactive bicyclic
catalyst 24a . A modified procedure was developed to
improve reproducibility and to maintain pseudo-first-
order conditions, including a larger excess of anhydride
(6 equiv over the alcohol), deoxygenation of solvents,
preparation and storage of catalyst solutions in an inert
atmosphere drybox, GLPC assay for percent conversion,
and quenching the anhydride as well as the catalyst prior
to assay. The latter precaution proved to be important
for the experiments monitored by GLPC assay.
Enantioselectivity data were also obtained (HPLC
assay on a chiral support, s corrected for catalyst ee24).
This allowed the determination of pseudo-first-order rate
constants for the competing acylations of both alcohol
enantiomers and calculation of the corresponding activa-
tion parameters (Table 3). As expected from the minimal
temperature dependence on acylation rates, the enthal-
pies of activation are small (1.4-4.5 kcal/mol), and the
activation entropies are large and negative (-64 to -74
eu). In each case, the less reactive alcohol enantiomer is
acylated via a pathway having significantly larger acti-
vation enthalpy, and marginally larger (less negative)
activation entropy. Therefore, the ∆∆G* term that re-
flects enantiomer discrimination is dominated by differ-
ences in activation enthalpy, resulting in a significant
temperature effect on enantioselectivity. The most dra-
matic example in this context is the benzoylation of 34
catalyzed by 24a , with s ) 10 at 35 °C and s ) 28 at -25
°C.
1
2
3
4
5
6
7
8
9
1
34
3
34
34
3
34
34
34
34
R ) Ph
R ) Ph
R ) iPr
R ) Ph
R ) Ph
R ) iPr
R ) Ph
R ) iPr
R ) Ph
1
11
76
300
18
110
1200
80
3.3
5.5
1.5
2.8
19a
19a
19a
24a
24a
24a
28a
28a
14
9, 15b
10,c 28b
1.2
130
4.8
a
2.5 equiv of anhydride, toluene solution, room temperature
unless noted; s values corrected for catalyst ee according to ref
b
24. -25 °C. c 35 °C.
high catalyst loading in the range of 10-20 mol %
relative to the alcohol, and by using the less hindered
alcohol 34 as the substrate for the comparisons. The
latter substrate is more representative of the arylalkyl-
carbinols featured in a related series of acylations using
an optimized catalyst,18 and was therefore employed for
most of the remaining experiments designed to probe
reactivity. However, the original substrate 3 was also
studied with two of the bicyclic catalysts using benzoic
anhydride as the acyl donor. In all cases, the phosphines
were generated by cleavage of the purified borane
complexes with pyrrolidine, and were used directly after
rapid filtration chromatography. Rate comparisons were
then performed under marginal pseudo-first-order condi-
tions using 2.5 equiv of the anhydride relative to the
alcohol reactant so that reactions could be monitored over
10-40% conversion, the range where NMR integration
was adequate for the purpose of determining the extent
of reaction. This qualitative technique allowed a ranking
of catalyst reactivity (Table 2) and demonstrated that the
more highly substituted bicyclic phospholane 24a is more
reactive than the phospholane 19a or the phosphetane
28a , and that benzoic anhydride is significantly more
reactive than isobutyric anhydride.
The rate comparisons involving the phosphetane cata-
lyst 28a were complicated by 15-20% catalyst decom-
position over the course of the acyl-transfer experiment,
as evidenced by the gradual appearance of four new
downfield signals in the 31P NMR spectrum (ca. 60-130
ppm). Because the extent of catalyst decomposition was
too small to explain the observed decrease in reactivity
compared to 24a , the details of the decomposition path-
ways have not been investigated. Furthermore, no such
complications were observed in any of the phospholane
experiments where the only side reaction detected was
oxidation to afford minor amounts of the phosphine oxide.
No 31P evidence was found to indicate the formation of a
detectable concentration of the P-acylphosphonium in-
termediate 35,20 or a pentavalent phosphorus species
such as 36, although the 31P signal characteristic of the
bicyclic phosphine catalysts was easily observed in the
range of 0 to -5 ppm throughout the experiments.
According to earlier work, 35 is likely to be formed as a
transient intermediate, but the concentration must be
too low for NMR detection under acyl-transfer conditions.
Ca ta lyst Con for m a tion
The X-ray crystal structure of borane complex 25a
provided an unexpected opportunity to explore confor-
mational preferences. Three distinct conformers of 25a
(I-A, I-B, I-C, Figure 1) were present in the unit cell,
suggesting that these geometries may be similar in
energy. Although this unusual situation undoubtedly
reflects crystal lattice packing effects, it provides three
convenient starting geometries for the identification of
the most likely solution-phase energy minima. More
importantly, it also offers a means to calibrate compu-
(21) For discussion of activation parameters for nucleophilic cataly-
sis in acyl-transfer reactions, see: (a) Oakenfull, D. G.; Riley, T.; Gold,
V. Chem. Commun. 1966, 385. (b) Milstien, J . B.; Fife, T. H. J . Am.
Chem. Soc. 1968, 90, 2164. (c) Schowen, R. L.; Behn, C. G. J . Am.
Chem. Soc. 1968, 90, 5839. (d) Butler, A. R.; Robertson, I. H. J . Chem.
Soc., Perkins Trans. 2 1975, 660.
(22) Koh, H. J .; Lee, J .-W.; Lee, H. W.; Lee, I. Can. J . Chem. 1998,
76, 710.
(23) Nucleophilic catalysis in phosphorylation analogies: Corriu, R.
J . P.; Lanneau, G. F.; LeClercq, D. Tetrahedron 1989, 45, 1959.
Chojnowski, J .; Cypryk, M.; Fortuniak, W. Heteroat. Chem. 1991, 2,
63.
(20) The adduct of acetyl chloride with tributylphosphine is char-
acterized by a 31P chemical shift in the region typical of phosphonium
salts, δ 28.8 ppm; see ref 1a, footnote 10.
(24) Ismagilov, R. F. J . Org. Chem. 1998, 63, 3772.
5024 J . Org. Chem., Vol. 68, No. 13, 2003

Monocyclic and Bicyclic Phospholanes as Catalysts
TABLE 3. Activa tion P a r a m eter s for Kin etic Resolu tion of 34 Ca ta lyzed by 24a
a,b
b
c
b
b
c
entry
solvent
(RCO)2O
∆G*fast
∆H*fast
∆S*fast
∆G*slow
∆H*slow
∆S*slow
1
2
3
toluene
MeCN
toluene
R ) Ph
R ) Ph
R ) iPr
20.8
20.8
22.1
1.90
1.72
1.39
-68.0
-68.7
-74.4
22.3
22.1
23.4
4.51
3.65
2.99
-64.1
-66.3
-73.3
a
Fast and slow refer to the two competing kinetic resolution pathways from fast-reacting and slow-reacting substrate enantiomers.
The ∆G values are calculated at 5 °C, the midpoint of the temperature range for entries 1 and 2 (-25 to +35 °C). Entry 3 was monitored
b
over the range from -25 to +22 °C, and enantioselectivities were determined as s ) 15 and s ) 9, respectively. Kilocalories per mole.
c Entropy units.
F IGURE 3. Energy minima for 24a (3-21G*).
F IGURE 2. Energy minima for 25a (3-21G*).
but both conformers experience a significant twist of the
P-phenyl group away from the C-B bond. The effect is
greater in the C-series of conformers (I-C and II-C),
presumably due to a larger steric effect from the pseudo-
axial endo-methyl group.
tational methods that might be useful for modeling
ground-state structures and for evaluating individual
geometric features that could be important in the transi-
tion state. This might help to understand why the bicyclic
phospholanes are 2 orders of magnitude more reactive
No energy minima were found that resemble the solid-
than the monocyclic analogues.
state conformer I-B. Conformers having the same com-
Initial modeling attempts focused on semiempirical
bination of phospholane and cyclopentane envelope or
methods (AM1, PM3) as implemented in Spartan. Local
twist envelope forms were processed, but could not be
energy minima were identified by systematically varying
P-phenyl dihedral angles and ring conformations. Neither
AM1 nor PM3 identified any of the three structures I-A,
I-B, and I-C as an energy minimum for 25a , suggesting
that these methods are unlikely to provide meaningful
relative energy information in the phospholane environ-
ment. However, the semiempirical study did provide a
broad range of ring conformers that could be optimized
evaluated without imposing artificial constraints on the
geometry. Unconstrained geometries invariably relaxed
to one of the other conformer series related to II-A and
II-C, suggesting that there is no energy minimum cor-
responding to I-B, and that the latter structure may be
the result of crystal lattice effects. Assuming that specific
solvation does not alter this picture, conformers II-A and
II-C are likely to be the principal equilibrium species in
using more sophisticated methods.
The next conformational search was carried out at the
solution. According to the 3-21G* comparison, these
conformers have essentially identical energies (within 0.1
3-21G* (ab initio) level, starting from a variety of ring
kcal/mol!). Although the precision of calculated energy
conformations and varying the P-phenyl dihedral angle
values is certainly not this high, the qualitative picture
in 3° increments. This search was far more time-intensive
fits well with the presence of both solid-state conformers
as expected, but the results were more in accord with
the notion that crystal lattice conformers may resemble
I-A and I-C in the unit cell, as well as with II-A and II-C
in solution, and suggests that the 3-21G* level is satis-
energy minima in solution.
factory for modeling other phospholanes.
Two energy minima were found in the 3-21G* search.
The same computational approach was therefore ex-
tended to the parent phosphine 24a . Results similar to
those of the borane complex series were obtained. Two
families of conformers were identified, and two energy
minima were found (III-A and III-C, Figure 3). Conformer
III-C is virtually identical to II-C, while III-A resembles
II-A, but has a more nearly planar phospholane subunit.
As in the borane complexes, the energies of the two
conformers are essentially identical (0.3 kcal/mol in
favor of III-A), and the phenyl ring in III-C is twisted
more from the plane of maximum unshared electron pair
One of these geometries (II-C, Figure 2) is nearly identi-
cal to I-C, while the other (II-A) closely resembles I-A.
The differences in the latter case involve subtle modifica-
tions between envelope and twist envelope geometries in
the five-membered rings. The former energy mimimum
(structure II-C) has the same bicyclic ring conformation
as the solid-state conformer I-C, but there is a small
difference in P-phenyl dihedral angles relative to the BH3
group (II-C, -31 ( 3°; I-C, -38.6°). The corresponding
dihedral angles differ somewhat more in the alternative
minimum energy conformers (II-A, -13 ( 3°; I-A, -25.0°),
J . Org. Chem, Vol. 68, No. 13, 2003 5025

Vedejs et al.
F IGURE 4. Energy minima for 37 and 38 (3-21G*).
density at phosphorus (ca. 42°) compared to that in III-A
(ca. 29°).
monocyclic and bicyclic phospholanes. Simple steric
considerations are sufficient to account for relative
reactivity in the monocyclic series, and increased access
to the unshared electron pair at phosphorus correlates
with the observed reactivity trends (11a > 11b > 15a >
1, Table 1). Matters are not so simple in the bicyclic series
(Table 2). Here, the trend appears to connect increased
reactivity with increased hindrance (for example, 24a >
19a > 37), but this is probably a coincidence. We believe
that the real factor is the role of unsymmetrical substitu-
tion on phospholane conformer and P-phenyl rotamer
preferences.
To gain further insight regarding conformational pref-
erences of P-phenylphospholane structures, the parent
monocyclic derivative 3725 was evaluated (3-21G*). The
most stable conformer found in this case was quite
different from any of the bicyclic phospholane minima.
The phenyl ring orientation is within 2° of bisecting the
C-P-C ring bond angle, and eclipsing the unshared
electron pair at phosphorus (Figure 4). The same mini-
mum energy geometry was found for the corresponding
borane complex 38. These results are consistent with the
conclusion stated earlier that the bicyclic ring system
enforces geometries having the P-phenyl ring turned 10-
30° toward the bridgehead hydrogens, and away from the
hindered endo face. They also suggest a possible explana-
tion for the exceptional reactivity of the bicyclic phos-
pholanes based on a simple steric effect related to
P-phenyl conformer preferences. Thus, geometries similar
to III-C have a more accessible unshared electron pair
at phosphorus compared to eclipsed conformers such as
that found in the monocyclic phospholane 37. If the
P-phenyl orientation is important as proposed, then 37
should be less reactive than the more highly substituted
bicyclic phospholane 24a . This proposition was tested by
using 37 as the catalyst for the isobutyroylation of 34
under the standard conditions used for Table 2. Mono-
cyclic 37 was found to be less reactive than any of the
bicyclic phospholanes, and 13-fold less reactive than 24a .
Our findings suggest that the bicyclic phospholanes
19a and 24a are more reactive than their monocyclic
analogues because the P-phenyl geometry corresponds
more closely to the preferred geometry of the transition
state for acyl transfer. The presence of a fused five-
membered carbocyclic ring probably destabilizes the
ground state relative to the transition state by favoring
geometries where the P-phenyl group is twisted to
minimize eclipsing interactions with the bond from
phosphorus to the carbonyl carbon derived from the
anhydride. This effect increases with increasing substitu-
tion next to the bridgehead, as in 24a .
The extrapolation from preferred phosphine borane or
phosphine geometries to the transition state for acyl
transfer is a large one. Pending further investigation of
more rigid structures, we cannot be certain which of the
two conformers (II-A or II-C) is the most reactive geom-
etry for nucleophilic catalysis with the catalyst 24a . Since
the energies of structures corresponding to the two
conformers II-A and II-C do not differ significantly in
either the phosphine 24a or the borane complex 25b, the
transition-state geometry for acyl transfer would be
controlled by new interactions between the catalyst and
Discu ssion
The reactivity of the phospholane family of nucleophilic
catalysts reveals surprising differences between the
(25) Mathey, F. C. R. Acad. Sci., Ser. C 1969, 269, 1066.
5026 J . Org. Chem., Vol. 68, No. 13, 2003

Monocyclic and Bicyclic Phospholanes as Catalysts
experiments, the formation of 36 as an intermediate
appears unlikely. On the other hand, transition states
such as 40 or 41 might also be formed directly from the
acylphosphonium salt 35. There would be little formal
charge separation in either of these phosphorane-like
structures (40, 41), a feature that is in better accord with
the striking similarity of activation parameters in the
acetonitrile and toluene experiments (Table 3). However,
caution is appropriate in the interpretation of our data
in the context of any of these structures. The geometric
differences among 39, 40, and 41 need not be large,
depending on the orientation of the carboxylate subunit
and the relative distances between phosphorus and
oxygen. Furthermore, we have no basis to assign the
configuration of the transient tetrahedral (formally asym-
metric) carbon in 39, 40, or 41, nor do we know the
placement or the exact role of the carboxylate anion.
Another problem is that activation parameters for the
analogous DMAP-catalyzed acyl transfer have not been
reported. Pending comparisons based on that informa-
tion, it will be difficult to evaluate a mechanism via a
tight ion pair variant of transition state 39.
The activation parameters determined for the acyla-
tions catalyzed by 24a are qualitatively consistent with
a highly associative nucleophilic catalysis mechanism via
39, 40, or 41. Prior studies have shown that activation
entropy for a variety of nucleophile-assisted acyl-transfer
reactions is usually in the range of -10 to -40 eu for
reactions in hydroxylic solvents.21 Acyl transfer in aceto-
nitrile has also been studied (catalyzed aminolysis of
ethyl aryl carbonates), and was found to have very large,
negative values for the entropy of activation (-65 to -67
eu) and minimal activation enthalpy (1.8-1.9 kcal/mol).22
As already mentioned, the activation parameters for
DMAP-catalyzed esterification of alcohols have not been
reported. However, DMAP-catalyzed phosphorylations
have been investigated, and the kinetic parameters
feature a similar combination of large, negative activation
entropy, and minimal activation enthalpy.23
F IGURE 5. Transition-state bonding options.
the anhydride and alcohol reactants. Several other criti-
cal features remain undefined, and a detailed discussion
of the acyl-transfer process would not be justified.
However, a qualitative discussion of some key features
in the more plausible scenarios is possible.
From the large, negative activation entropy, a highly
associative mechanism can be proposed, having all three
components (anhydride, alcohol, catalyst) present in the
transition state. The lack of a significant solvent dielectric
effect on the activation parameters from toluene to
acetonitrile (Table 3; compare entries 1 and 2) suggests
that there is little change in charge separation in the
transition state compared to the reactants. If the reaction
proceeds via 35 and not the phosphorane 36, then the
transition state would have to involve tightly ion paired
phosphonium and carboxylate subunits interacting with
the alcohol, as shown in 39 (Figure 5). This option
maintains the mechanistic analogy with DMAP-catalyzed
acyl transfer, to the extent that the latter process is
understood.26
In terms of the potential for applications of chiral
phosphines as nucleophilic catalysts, the results pre-
sented above reveal that bicyclic P-arylphospholanes
have excellent reactivity as well as promising enantiose-
lectivity. Optimized phosphabicyclo[3.3.0]octane (PBO)
analogues having exceptional enantioselectivy for the
kinetic resolution of alcohols have been prepared in our
laboratory using similar methods, as described else-
where.18 These catalysts are useful for the enantioselec-
tive acylations of allylic as well as benzylic alcohols.18,28
The alternative transition-state descriptions 40 and 41
represent pathways that might be accessible from the ion
pair 35, or from a transient phosphorane, 36. There is
no direct precedent for a P-acylphosphorane such as 36
to help evaluate its role as a potential intermediate.
However, if 36 were formed reversibly as a higher energy
intermediate from 35, then the analogy with simpler
alkoxyphosphoranes would suggest that pseudorotation
would result in facile epimerization at phosphorus27 and
interconversion of 24a and 24b. Because interconversion
does not occur (<3%) on the time scale of the catalytic
Ack n ow led gm en t. This work was supported by the
National Science Foundation.
(26) (a) Reviews: Ho¨fle, G.; Steglich, W.; Vorbru¨ggen, H. Angew.
Chem., Int. Ed. Engl. 1978, 17, 569. Scriven, E. F. V. Chem. Soc. Rev.
1984, 12, 129. (b) Kirichenko, A. I.; Litvinenko, L. M.; Dotsenko, I. N.;
Kotenko, N. G.; Nikkel’sen, E.; Berestetskaya, V. D. Dokl. Akad. Nauk.
SSSR 1979, 244, 65 (Russian 1125). Bondarenko, A. I.; Kirichenko, L.
M.; Litvinenko, L. N.; Dmitrenko, L. N.; Kobets, V. D. J . Org. Chem.
USSR (Engl. Transl.) 1981, 2310. Connors, K. A.; Albert, K. S. J .
Pharm. Sci. 1973, 62, 845. Lin, S.-F.; Connors, K. A. J . Pharm. Sci.
1981, 70, 235. Eboka, C. J .; Connors, K. A. J . Pharm. Sci. 1983, 72,
366. Connors, K. A.; Eboka, C. J . J . Pharm. Sci. 1983, 72, 369.
(27) Vedejs, E.; Marth, C. F. J . Am. Chem. Soc. 1989, 111, 1519.
Su p p or tin g In for m a tion Ava ila ble: Experimental pro-
cedures, characterization data, and X-ray data tables. This
material is available free of charge via the Internet at
http://pubs.acs.org.
J O030007+
(28) Vedejs, E.; Daugulis, O. J . Am. Chem. Soc. 1999, 121, 5813.
Vedejs, E.; Rozners, E. J . Am. Chem. Soc. 2001, 123, 2428. Vedejs, E.;
MacKay, J . A. Org. Lett. 2001, 3, 535.
J . Org. Chem, Vol. 68, No. 13, 2003 5027