Supercritical and liquid solvent effects on the enantioselectivity of asymmetric cyclopropanation with tetrakis[1-[(4-tert-butylphenyl)-sulfonyl]-(2S)-pyrrolidinecarboxylate] dirhodium(II)

Article abstract of DOI:10.1021/ja000894n

The enantioselectivity of the cyclopropanation of styrene and methyl phenyldiazoacetate in the presence of a dirhodium carboxylate catalyst was found to be dependent on both the dielectric constant and the coordinating ability of the solvent. The enantiom

Full text of DOI:10.1021/ja000894n

7638
J. Am. Chem. Soc. 2000, 122, 7638-7647
Supercritical and Liquid Solvent Effects on the Enantioselectivity of
Asymmetric Cyclopropanation with Tetrakis[1-[(4-tert-butylphenyl)-
sulfonyl]-(2S)-pyrrolidinecarboxylate]dirhodium(II)
Dolores C. Wynne, Marilyn M. Olmstead, and Philip G. Jessop*
Contribution from the Department of Chemistry, UniVersity of California, DaVis, California 95616-5295
ReceiVed March 13, 2000
Abstract: The enantioselectivity of the cyclopropanation of styrene and methyl phenyldiazoacetate in the
presence of a dirhodium carboxylate catalyst was found to be dependent on both the dielectric constant and
the coordinating ability of the solvent. The enantiomeric excess is pressure dependent in supercritical fluoroform,
changing from as low as 40% ee at pressures above 100 bar to nearly 80% ee at pressures near the critical
point. The pressure dependence is caused by the changing dielectric constant of the fluoroform. Reactions in
some coordinating liquid solvents had greater enantioselectivities than expected on the basis of dielectric constant
alone. The crystal structure of the catalyst was obtained along with spectroscopic data in order to elucidate the
causes of the solvent-dependent enantioselectivity.
Introduction
The choice of solvent can have a dramatic effect on the
enantioselectivity of asymmetric homogeneous catalysis.1-5 To
understand and optimize solvent-dependent enantioselectivity,
one normally performs a solvent study in which the reaction is
tested in a variety of different solvents and the results are
rationalized by noting relationships between the properties of
the solvent and the selectivity. A complication with this
approach is that solvents differ in more than one property;
individual properties of the solvent can be difficult to isolate.
For example, a change from a solvent with a low dielectric
constant (ꢀr) to a solvent with a high ꢀr will usually be
accompanied by a change in the coordinating ability, Lewis
basicity, acidity, and other properties of the solvent.6-8 This
problem can be overcome by utilizing supercritical fluids
(SCFs). Because the solvent properties of supercritical fluids
are pressure dependent, selected solvent properties can be varied
while keeping the chemical nature of the solvent unchanged.
Figure 1. The dielectric constant of fluoroform as a function of
pressure and temperature.63 Curves for temperatures below 70 °C were
drawn from published data,9,64 while the curves for T g 70 °C are
based upon the equation of Rhodes et al.65 and published densities.66,67
(1) The following references are given as examples: (a) Trost, B. M.;
Bunt, R. C. J. Am. Chem. Soc. 1994, 116, 4089-4090. (b) Imanishi, H.;
Katsuki, T. Tetrahedron Lett. 1997, 38, 251-254. (c) Namyslo, J. C.;
Kaufmann, D. E. Chem. Ber.-Rec. 1997, 130, 1327-1331. (d) Gross, Z.;
Ini, S. J. Org. Chem. 1997, 62, 5514-5521. (e) Keck, G. E.; Krishnamurthy,
D. J. Am. Chem. Soc. 1995, 117, 2363-2364. (f) Kureshy, R. I.; Khan, N.
H.; Abdi, S. H. R.; Patel, S. T.; Iyer, P. J. Mol. Catal. A-Chem. 1999, 150,
163-173. (g) Yamamoto, K.; Ikeda, K.; Yin, L. K. J. Organomet. Chem.
1989, 370, 319-332. (h) Yang, T. K.; Lee, D. S. Tetrahedron: Asymmetry
1999, 10, 405-409. (i) Jendralla, H. Tetrahedron: Asymmetry 1994, 5,
1183-1186.
For example, the dielectric constant of supercritical fluoroform
(scCHF3) is a strong function of pressure at 30 °C; it is a
medium-polar solvent (ꢀr ) 7) at 120 bar but a relatively
nonpolar solvent (ꢀr ) 3) at 52 bar (Figure 1).9 Supercritical
solvent studies are complementary to liquid studies in that one
can use the results in SCFs to isolate the effect of a property
such as the dielectric constant and consequently elucidate the
effects of the other properties. Liquid solvents, on the other hand,
have the advantages of ease of use and a wider range of
dielectric constants. The information derived from combined
supercritical and liquid solvent studies can be used to elucidate
the role of solvent and to optimize the selectivity.
The tunable properties of supercritical solvents offer an
exciting new method of performing solvent studies. Chemical
synthesis in supercritical fluids has become an important new
field in chemistry and chemical engineering,10,11 and the effect
of pressure on solvent properties and therefore reaction perfor-
(2) Davies, H. M. L.; Hutcheson, D. K. Tetrahedron Lett. 1993, 34,
7243-7246.
(3) Davies, H. M. L.; Bruzinski, P. R.; Lake, D. H.; Kong, N.; Fall, M.
J. J. Am. Chem. Soc. 1996, 118, 6897-6907 and references therein.
(4) Doyle, M. P.; Zhou, Q.-L.; Charnsangavej, C.; Longoria, M. A.
Tetrahedron Lett. 1996, 37, 4129-4132.
(5) Kitagaki, S.; Matsuda, H.; Watanabe, N.; Hashimoto, S. Synlett 1997,
10, 1171-1174.
(6) Reichardt, C. SolVents and SolVent Effects in Organic Chemistry,
2nd ed.; VCH: Weinheim, 1988.
(7) Mathematical fitting of empirical data has been used to separate the
solvent effects on physicochemical observables such as spectral shifts.8
(8) Mu, L.; Drago, R. S.; Richardson, D. E. J. Chem. Soc., Perkin Trans.
2 1998, 159-167.
(9) Downing, R. C. Fluorocarbon Refrigerants Handbook; Prentice
Hall: Englewood Cliffs, NJ, 1988.
10.1021/ja000894n CCC: $19.00 © 2000 American Chemical Society
Published on Web 07/28/2000

SolVent Effect on Cyclopropanation EnantioselectiVity
J. Am. Chem. Soc., Vol. 122, No. 32, 2000 7639
mance in supercritical fluids is one of the factors which is most
often explored. Pressure dependence of catalytic selectivity in
SCFs has been observed previously and attributed to dilution12
or changes in tuning functions or activation volumes of
competing transition states.13,14 The effect of the dielectric
constant change of SCFs on the selectivity of homogeneous
catalysis has not been explored nor has the effect of pressure
on enantioselective homogeneous catalysis in SCFs.15 We set
out to demonstrate a case of pressure-dependent enantioselective
homogeneous catalysis in a SCF and then to demonstrate the
utility of such a discovery as part of a solvent study.
Asymmetric cyclopropanation, a synthetically important reac-
tion,3,16 has been shown to exhibit interesting solvent effects in
liquid reactions.2-5,17 One of the most successful cyclopropa-
nation catalysts is the complex tetrakis[1-[(4-tert-butylphenyl)-
sulfonyl]-(2S)-pyrrolidinecarboxylate]dirhodium(II) ([Rh2(TBSP)4],
1) developed by Davies2,3 as a hexane-soluble analogue of
McKervey’s tetrakis[1-[phenylsulfonyl]-(2S)-pyrrolidinecarbox-
ylate]dirhodium(II) catalyst.18
effect was due to the difference between the dielectric constants
of the two solvents (CH2Cl2, ꢀr ) 9.08; C5H12, ꢀr ) 1.84),19
although this is not the only property by which CH2Cl2 and
pentane differ (they also differ in their ability to coordinate to
metal complexes).20-22 We suspected therefore that the enan-
tiomeric excess might be pressure dependent in supercritical
fluoroform (scCHF3; Tc ) 25.9 °C, Pc ) 48.2 bar). A
communication describing our preliminary findings was pub-
lished recently.23 Here, we present, along with a complete
analysis of the pressure dependent enantioselectivity, new data
which describe coordination effects on selectivity. This study
provides additional insight into the solvent -dependent enanti-
oselectivity seen with the [Rh2(TBSP)4] catalyst.
Experimental Section
Materials. The [Rh2(S-TBSP)4] catalyst was prepared according to
the literature method24 for the pressure dependence studies. However,
for the later coordinating effect studies (Tables 1 and 2), commercially
available catalyst was used (Aldrich). The styrene (Aldrich) contained
10-15 ppm of 4-tert-butyl catechol, which was not removed by
distillation because this was found to have no effect on the enantiose-
lectivity of the reaction. The methyl phenyldiazoacetate was synthesized
from phenylglycine methyl ester and isoamyl nitrite.25 The CHF3 (AGA
Specialty Gas, 99.995% pure) and CO2 (Air Products and Chemicals,
Inc., 99.9999% pure) were passed through an oxygen trap (Alltech)
before use. Nitrous oxide (Nellcor Puritan Bennett, 99.998%, O2 < 2
ppm) was used as received. Liquid solvents (98-99% pure) were
purchased from a variety of manufacturers and were dried by distillation
from sodium benzophenone ketyl (hexane, THF) or molecular sieves
(DMF, CH2Cl2, NEt3, MeCN). Chloroform (Spectrophotometric grade
with amylene inhibitor) was dried with oven-dried K2CO3. The trialkyl
phosphines and triethylphosphine oxide were stored and used under
nitrogen in a drybox.
Cyclopropanation involving diazo decomposition in the
presence of this catalyst was studied by Davies2,3 (eq 1) and
Doyle4 (eq 2). Both studies found a dramatic increase in the
Equipment/Spectroscopy. The supercritical experimental apparatus
is presented schematically in Figure 2. The gases were pressurized via
an ISCO syringe pump (model 500D) and were delivered through 1/16
in. stainless steel HPLC tubing and a Rheodyne model 7725 HPLC
injector to the reaction vessel. The reactions were carried out in a Parr
160 mL stainless steel vessel fitted with a pressure transducer,
thermocouple, burst disk, and two reagent addition ports. The apparatus
also includes an acetone reservoir and a wash pump to allow cleaning
of the gas lines. The injector was cleaned after every reaction. The
vessel was heated by a water bath fitted with a Fisher Isotemp
recirculator. A magnetic stirrer custom-made by Glas-Col was placed
underneath the water bath and directly below the vessel; tests showed
that this stirrer was able to reproducibly couple with a magnetic stir
bar inside the vessel. All enantiomeric excesses were determined by
enantiomeric excess (ee) of the major diastereomer upon
changing the reaction solvent from methylene chloride to
pentane, from 74% ee to 90% ee and from 61% ee to 85% ee,
for eqs 1 and 2, respectively. Davies proposed that the solvent
(10) (a) Chemical Synthesis using Supercritical Fluids; Jessop, P. G.,
Leitner, W., Eds.; VCH/Wiley: Weinheim, 1999. (b) Special issue of Chem.
ReV. 1999, 99, issue 2.
(19) CRC Handbook of Chemistry and Physics, 63rd ed.; Weast, R. C.,
Ed.; CRC Press: Boca Raton, FL, 1982.
(20) CH2Cl2 has been found to bind to several transition metal
complexes,20a-e but complexes containing alkanes as ligands usually have
extremely short lifetimes near room temperature.20f,g (a) Arndtsen, B. A.;
Bergman, R. G. Science 1995, 270, 1970-1973. (b) Butts, M. D.; Scott,
B. L.; Kubas, G. J. J. Am. Chem. Soc. 1996, 118, 11831-11843. (c)
Huhmann-Vincent, J.; Scott, B. L.; Kubas, G. J. Inorg. Chem. 1999, 38,
115-124. (d) Leoni, P. Organometallics 1993, 12, 2432. (e) Peng, T.-S.;
Winter, C. H.; Gladysz, J. A. Inorg. Chem. 1994, 33, 2534-2542. (f) Lee,
D. W.; Jensen, C. M. J. Am. Chem. Soc. 1996, 118, 8749-8750. (g) Sun,
X.-Z.; Grills, D. C.; Nikiforov, S. M.; Poliakoff, M.; George, M. W. J. Am.
Chem. Soc. 1997, 119, 7521-7525.
(21) Coordinating additives or solvents have been shown to affect the
rate of O-H insertion reactions using diazo compounds. Nelson, T. D.;
Song, Z. J.; Thompson, A. S.; Zhao, M.; DeMarco, A.; Reamer, R. A.;
Huntington, M. F.; Grabowski, E. J. J.; Reider, P. J. Tetrahedron Lett. 2000,
41, 1877-1881.
(22) We are assuming that the coordinating ability of CHF3 is negligible
at all pressures.
(23) Wynne, D.; Jessop, P. G. Angew Chem., Int. Ed. 1999, 38, 1143-
1144.
(24) Callot, H. J.; Metz, F. Tetrahedron 1985, 41, 4495-4501.
(25) Takamura, N.; Mizoguchi, T.; Koga, K.; Yamada, S. Tetrahedron
Lett. 1975, 31, 227-230.
(11) The history of reactions in supercritical fluids is the subject of
reviews covering the years up to 1945,11a 1945-1985,11b and 1986-1994.11c
(a) Jessop, P. G.; Leitner, W. In Chemical Synthesis using Supercritical
Fluids; Jessop, P. G., Leitner, W., Eds.; VCH/Wiley: Weinheim, 1999; pp
1-36. (b) Subramaniam, B.; McHugh, M. A. Ind. Eng. Chem., Proc. Des.
DeV. 1986, 25, 1-12. (c) Savage, P. E.; Gopalan, S.; Mizan, T. I.; Martino,
C. J.; Brock, E. E. AIChE J. 1995, 41, 1723-1778.
(12) Fu¨rstner, A.; Koch, D.; Langemann, K.; Leitner, W.; Six, C. Angew.
Chem., Int. Ed. Engl. 1997, 36, 2466-2469.
(13) Guo, Y.; Akgerman, A. J. Supercrit. Fluids 1999, 15, 63-71.
(14) Oakes, R. S.; Heppenstall, T. J.; Shezad, N.; Clifford, A. A.; Rayner,
C. M. Chem. Commun. 1999, 1459-1460.
(15) There is a report of pressure-dependent enantioselectivity of
enzymatic catalysis: Kamat, S. V.; Beckman, E. J.; Russell, A. J. J. Am.
Chem. Soc. 1993, 115, 8845-8846.
(16) Doyle, M. P.; McKervey, M. A.; Ye, T. Modern Catalytic Methods
for Organic Synthesis with Diazo Compounds; John Wiley & Sons: New
York, 1998.
(17) Davies, H. M. L.; Panaro, S. A. Tetrahedron Lett. 1999, 40, 5287-
5290.
(18) McKervey, M. A.; Ye, T. J. Chem. Soc., Chem. Commun. 1992,
823-824.

7640 J. Am. Chem. Soc., Vol. 122, No. 32, 2000
Wynne et al.
Styrene (440 µmol), [Rh2(TBSP)4] (0.69 µmol), and a stirbar were
placed, under air, into a flat-bottomed glass liner (5.5 cm diameter) in
a 160 mL steel vessel equipped with diametrically opposed sapphire
windows. Methyl phenyldiazoacetate (56 µmol) was placed in a 0.7
mL microbeaker inside the glass liner. The vessel was purged with
N2O and then placed in a water bath at 28 °C. Liquid nitrous oxide
was pumped into the vessel until the liquid level rose above the top of
the microbeaker. Stirring was then started. To stop the reaction after 2
h, the vessel was very slowly vented and then opened; the colorless
liquid inside was dissolved in acetone and analyzed by GC.
Solubility Studies. Solubility of the substrates in scCO2 (35 °C) or
scCHF3 (30 °C) was verified by the following methods. The dissolution
of the styrene and diazoacetate (twice the concentration used for
cyclopropanation reactions) was confirmed visually in a 160 mL vessel
fitted with sapphire windows. The reagents were found to be dissolved
completely by 51 bar. For this reason the lowest pressure used for the
catalytic experiments was 52 bar. The solubility was verified in this
manner for both of the substrates separately and for a mixture of the
styrene and methyl phenyldiazoacetate in the proportions used in
experiments but again at double the normal concentration. Because of
the small quantity of catalyst used, the catalyst solubility could not be
detected visually. The solubility of the catalyst in scCHF3 was confirmed
by UV/vis spectroscopy by dissolving the catalyst in the SCF in the
sapphire-windowed vessel. The absorbance (at 290 nm) at 51 bar was
identical to the absorbance at 122 bar, indicating that all of the 1.3 mg
(0.90 µmol) had dissolved at both pressures. As a further demonstration
that the catalyst was homogeneous,30 the catalyst was put into a small
beaker in the vessel and the vessel was pressurized to 60 bar for 1 h
without stirring. The beaker was then removed from the vessel and a
reaction was performed in the vessel without the addition of more
catalyst. That the reaction proceeded asymmetrically is considered
evidence that the active catalyst was soluble.
Crystallography. [Rh2(TBSP)4(DMF)2] crystals were grown in a 7
in. NMR tube in the following manner. About 1 mL of a concentrated
solution of [Rh2(TBSP)4] in toluene was placed into the NMR tube,
and 20 µL of dimethylformamide was added to this portion and allowed
to mix. About 1 mL of pentane was slowly layered on top of the toluene
solution. The NMR tube was capped and allowed to sit for 2 weeks.
Long, thin, blue needles formed. A needle of dimensions 0.05 × 0.05
× 0.35 mm was mounted in the CRYO Industries cold stream of a
Bruker SMART 1000 diffractometer equipped with a sealed Mo tube
and graphite monochromator. Approximately a 1/2 sphere of data (98%
completeness) were collected to a maximum 2Θ of 63°. No decay in
the intensities of 50 standard frames was observed. Of 76 290 reflections
measured, 24 884 were unique, R(int) ) 0.114. The structure was solved
using direct methods. Refinement was by full-matrix least-squares
methods, based on F2, using all data. Hydrogen atoms were located on
a difference map and constrained during refinement. The solvent
molecules of toluene and n-pentane occupy the same site. They were
initially refined as rigid groups but were fixed in the final cycles of
refinement. No chemically significant peaks were present in the final
difference map. No extinction or absorption corrections were applied.
The handedness of the structure was determined by calculation of the
Flack parameter, -0.02(2). Crystallographic programs were those of
SHELXTL 5.1.31 Tables of neutral atom scattering factors, f ′ and f′′,
and absorption coefficients are from International Tables for Crystal-
lography.32
Figure 2. Experimental apparatus.
GC on a Chirasil-Dex CB column (25 m × 0.25 mm × 0.25 µm, T )
170 °C, split ratio 7.7).
All liquid UV/vis measurements were obtained with a Hewlett-
Packard 8452A diode array spectrophotometer for solutions in 1.0 cm
path length quartz cuvettes. Airtight cuvettes filled under nitrogen were
used for solutions containing trialkyphosphines or triethylphosphine
oxide. The UV/vis measurements in SCFs were obtained in a 160 mL
vessel with diametrically opposed sapphire windows (path length )
14.3 cm) on an Olis (Rapid Scanning Monochromator) spectrometer.
All NMR data was collected on a General Electric 300 MHz
spectrophotometer.
Cyclopropanation in SCFs. To the 160 mL steel vessel were added
440 µmol of styrene and 0.69 µmol of [Rh2(TBSP)4] along with a
stirbar. The vessel was placed in a water bath at the desired temperature
and allowed to equilibrate (30 °C for reactions in scCHF3 and 35 °C
for reactions in scCO2). The vessel was purged with the gas to be used
and stirring was started. The vessel was then pressurized to above the
critical pressure but below the desired pressure. After allowing for the
temperature of the vessel to equilibrate for 10 min, 56 µmol of methyl
phenyldiazoacetate was added via the HPLC injector and carried into
the vessel with the remaining amount of SCF. To stop the reaction
(usually after 1 h), the vessel was cooled in a dry ice/acetone bath
until the pressure dropped to <5 bar, vented, and allowed to warm to
room temperature. The vessel was opened; the colorless liquid inside
was dissolved in acetone and analyzed by GC.
Cyclopropanation in Liquid Solvents. The liquid-phase reactions
were performed by dissolving styrene (440 µmol), [Rh2(TBSP)4] (0.69
µmol), and any added ligand (14 µmol, 20 mol per mol of catalyst)
into 4 mL of solvent in a 10 mL vial at room temperature under
nitrogen. The mixture was stirred until the catalyst appeared to be
dissolved, and then the methyl phenyldiazoacetate (56 µmol) was added.
The reaction time required depended on the solvent, but all reactions
were given at least 1 h. The reaction was considered complete when
the bright orange of the diazoacetate completely disappeared.
Cyclopropanation in Liquid Nitrous Oxide. SAFETY WARN-
ING: Nitrous oxide is a thermodynamically powerful oxidant. Never
mix high concentrations of organic compounds with liquid or super-
critical N2O. Explosions have occurred with a 9 vol % solution of
ethanol in scN2O, with a mixture of 25 mL of cyclohexene in 75 mL
of scN2O, and with a mixture of 1 g of ground coffee in 2.5 mL of
scN2O.26-29 To minimize the risk, we choose to keep the combustible
substrate to microscale quantities and Very low concentrations. In
addition, we employ a burst disk, blast shield, and eye protection in
all experiments. Diluting N2O with CO2 may further enhance the safety.
Combustible cosolvents should not be used with compressed N2O under
any circumstances. Never use an oil-based compressor to pressurize
nitrous oxide; if the N2O were to leak into the oil, an explosion could
result.27
Results
The asymmetric cyclopropanation reaction in eq 2 proceeds
readily in scCHF3, producing compound 2 and no traces of
diastereomer 3. The reaction was performed in scCHF3 at several
pressures, and the enantiomeric excess (ee) of the product was
(30) Jessop, P. G.; Leitner, W. In Chemical Synthesis using Supercritical
Fluids; Jessop, P. G., Leitner, W., Eds.; Wiley-VCH: Weinheim, 1999; pp
351-387.
(31) Sheldrick, G. SHELXTL 5.1; Bruker-AXS: Madison, WI, 1997.
(32) International Tables for Crystallography; Wilson, A. J. C., Ed.;
Kluwer Academic Publishers: Dordrecht, 1992; Vol. C, Tables 6.1.1.3 (pp
500-502), 4.2.6.8 (pp 219-222), and 4.2.4.2 (pp 193-199).
(26) Raynie, D. E. Anal. Chem. 1993, 65, 3127-3128.
(27) Bridson-Jones, F. S.; Buckley, G. D.; Cross, L. H.; Driver, A. P. J.
Chem. Soc. 1951, 2999-3008.
(28) Hansen, B. N.; Hybertson, B. M.; Barkley, R. M.; Sievers, R. E.
Chem. Mater. 1992, 4, 749-752.
(29) Sievers, R. E.; Hansen, B. Chem. Eng. News 1991, 69 (29), 2.

SolVent Effect on Cyclopropanation EnantioselectiVity
J. Am. Chem. Soc., Vol. 122, No. 32, 2000 7641
Figure 3. The dependence of the dielectric constant9,64 of fluoroform
and the enantiomeric excess of cyclopropanation performed in scCHF3
on the pressure of fluoroform at 30 °C.
Figure 5. The dependence of the dielectric constant of carbon dioxide33
and the enantiomeric excess of cyclopropanation performed in scCO2
on the pressure of carbon dioxide at 35 °C.
SCFs was slightly lower than in liquids. We therefore tested
the effects of the following: temperature, injection method,
ability of the steel vessel walls to catalyze the reaction, and the
role of trace water. These studies are described in the Supporting
Information. The conclusions were that the temperature and
injection method have only a negligible effect on the ee but
that the use of a steel vessel lowers the ee, possibly because of
a competing steel-catalyzed cyclopropanation.
Coordination Effects. The effect of solvent coordinating
ability can be isolated from the effect of the dielectric constant
by comparing two solvents with differing coordinating ability
but identical dielectric constants. The reaction was performed
in THF (ꢀr ) 7.6 at 25 °C)34 to see if its coordinating ability
causes the selectivity to differ from CHF3 at the same dielectric
constant. With the help of published dielectric constant data,9
we were able to tune the dielectric constant of CHF3 to match
that of THF. The dielectric constant matches when the CHF3 is
at 25 °C and 156 bar (Figure 1), which happens to be just inside
the liquid portion of the CHF3 phase diagram. Because of the
extremely dilute conditions, the reagent contribution to the
dielectric constant of the medium could be neglected. The
enantiomeric excess of the product of the reaction in the liquid
CHF3 was 35%, which falls along the line for the results in
scCHF3 (Figure 4). Surprisingly, the enantioselectivity in THF
was 81%, far higher than the result in CHF3 and higher than
would be expected (∼70%) in a noncoordinating liquid solvent
with a dielectric constant of 7.6. These results demonstrate that
the dielectric constant is not the only parameter which controls
the ee; another parameter such as coordinating ability is
involved.
Figure 4. The dependence of the enantiomeric excess of 2 on the
dielectric constant of scCHF3 (b), scCO2 ([), liquid CHF3 (9), or
noncoordinating liquid solvents (literature4 data O, our data 0). The
two data points for CHCl3 correspond to the amylene- (lower point)
and ethanol-stabilized (upper point) solvents.
found to be a function of the pressure at which the reaction
was performed (Figure 3). The ee is as low as 40% at pressures
above 100 bar and as high as 78% at pressures close to the
critical point. This is the first example of pressure-dependent
enantioselectivity in homogeneous catalysis in SCFs or indeed
in any medium at pressures below thousands of bar. As seen in
the plot of ee vs dielectric constant (Figure 4), the results in
CHF3 show the same trend as the results in noncoordinating
liquid solvents. This suggests that the pressure-dependent
enantioselectivity is due to dielectric constant effects rather than
other pressure-related effects such as dilution or solute-solvent
clustering. As further evidence, when the reaction is performed
in scCO2 (Tc ) 31 °C, Pc ) 73.8 bar) at 35 °C, the enantio-
selectivity is essentially pressure independent (Figure 5), as
expected because the dielectric constant of scCO2 changes only
very slightly with pressure.33 Although the dielectric constant
dependence was evident, it was unclear why the selectivity in
Other potentially coordinating solvents, including acetonitrile,
triethylamine, N,N-dimethylformamide (DMF), and compressed
liquid nitrous oxide (N2O), also had an effect on the enanti-
oselectivity (Table 1). The enantioselectivity was found to vary
over a wide range and have no correlation to the dielectric
constant of the solvent, probably because the other solvent
properties overwhelm the dielectric constant effect. For example,
(33) Kita, T.; Uosaki, Y.; Moriyoshi, T. In High-Pressure Liquids and
Solutions; Taniguchi, Y., Senoo, M., Hara, K., Eds.; Elsevier: Amsterdam,
1994; pp 181-198.
(34) Coetze, J. F.; Chang, T.-H. Pure Appl. Chem. 1985, 57, 633.

7642 J. Am. Chem. Soc., Vol. 122, No. 32, 2000
Wynne et al.
Table 1. Enantiomeric Excess of Product 2 after Cyclopropanation
Table 2. The Wavelengths (λ, nm) and Extinction Coefficients (ꢀ,
cm-1 M-1) of the UV/Vis Maxima for [Rh2(TBSP)4] in Various
Solvents, with or without Additivesa
with [Rh2(TBSP)4] in Various Solventsa
slvent
additive
none
OP(Oct)3
OPPh3
OP(OEt)3
OPEt3
color
ee, %
solvent
additive
none
OP(n-Oct)3
OPEt3
OPPh3
color
λ
max1 (ꢀ)
λmax2 (ꢀ)
hexane
hexane
hexane
hexane
hexane
hexane
hexane
hexane
hexane
hexane
hexane
hexane
hexane
hexane
N2Ob
green
green
green
green
green
green
orange
orange
orange
orange
yellow
green
green
green
green
green
purple
blue
90
90
90
89
82
90
90
89
88
46
21
67
89
65
84
81
73
69
67
55
CHCl3
CHCl3
CHCl3
CHCl3
CHCl3
CHCl3
CHCl3
CHCl3
CHCl3
CHCl3
CHCl3
DMF
green
green
green
green
green
orange
orange
green
orange
yellow
purplec
blue
612 (250)
622 (240)
618 (220)
614 (270)
612 (270)
544 (300)
578 (500)
610 (250)
472 (1,600)
354 (4,700)
552 (230)
590 (320)
596 (260)
556 (280)
656 (390)
548 (460)
440 (140)
442 (130)
442 (130)
438 (170)
440 (170)
P(mes)3
(R)-BINAP
(S)-BINAP
PCy3
OP(OEt)3
(R)-BINAP
(S)-BINAP
P(mes)3
440 (150)
325 (8,700)
b
PPh3
PPh3
PEt3
PEt3
PCy3
none
none
none
none
none
H2O
c
H2O + OP(Oct)3
H2O + OP(Oct)3
none
454 (110)
448 (130)
444 (170)
424 (290)
404 (250)
d
THF
green
purple
green
purple
MeCN
CH2Cl2
NEt3
THF
MeCN
DMF
CH2Cl2
NEt3
none
none
none
a Additive:1 molar ratio ) 20:1. Spectra with additives were
measured in CHCl3 rather than hexane because of the greater solubility
of 1 in CHCl3. b Additive:1 molar ratio ) 10:1. c The solution in CHCl3
is orange for a few seconds and then turns purple. In pentane and hexane
it is orange.
none
green
purple
none
a Performed in glassware at 20 °C. Additive:1 molar ratio ) 20:1.
b Performed in a glass liner inside a steel vessel at 28 °C. c H2O:
OP(Oct)3:1 molar ratio ) 16:20:1. d H2O:OP(Oct)3:1 molar ratio ) 100:
20:1.
contained only 5% of 2 and 20% starting material, while with
NEt3 the mixture contained less than 1% of 2 and 68% starting
material, despite the fact that these reactions were run for 48 h.
Although solutions of 1 in CH2Cl2, pentane, hexane, and THF
are bright green, solutions of complex 1 are a wide variety of
colors in the presence of coordinating solvents or reagents
(Tables 1 and 2), the color changes presumably being indicative
of reactions between the solvent or additive and the rhodium
complex. The UV/vis maxima for the catalyst in different
solvents or in CHCl3 with added coordinating ligands were
measured (Table 2). The λmax for the two peaks in the visible
region were 612 and 440 nm in CHCl3. The peak at 440 nm
did not change greatly with different coordinating species (other
than phosphines). Only with NEt3 did the λmax change signifi-
cantly, where the peak was slightly blue-shifted. The 612 nm
peak varied widely depending on the solvent. It ranged from as
low as 556 nm in MeCN to as high as 656 nm in CH2Cl2. In
the presence of most phosphines, the spectrum completely
changed. Interestingly, a few solvents or additives had little or
no effect on the spectrum; these included CH2Cl2, CHCl3, THF,
the phosphine oxides, and trimesitylphosphine. One can con-
clude that these solvents or reagents failed to bind to the Rh
complex or bound so weakly that they did not significantly affect
the electronic properties of the complex.
The Effect of Water. The enantioselectivity we obtained in
Na/benzophenone-dried alkane (hexane or pentane) was sig-
nificantly greater than that reported in the literature4 and also
greater than we found in commercially available “anhydrous”
alkane. We therefore tested the effect of controlled amounts of
water on the enantioselectivity in hexane and found that the ee
dropped precipitously (Table 1). However, the addition of
phosphine oxides or triethyl phosphate counteracted the effect
of water. It is therefore not necessary to use rigorously dried
hexane if instead one adds 20 equiv of tri-n-octylphosphine
oxide.
the ee of the product from a run in acetonitrile was 73%, much
higher than would be expected for a solvent of a dielectric
constant of 36. The trend in ee is hexane > N2O > THF .
MeCN > DMF ≈ CH2Cl2 > NEt3. The trend in dielectric
constants is N2O ≈ hexane < NEt3 , THF < CH2Cl2 , MeCN
≈ DMF. The trend in donor number is CH2Cl2 < MeCN <
THF < DMF < NEt3 (value not known for hexane or N2O)
while the trend in acceptor number is hexane ≈ NEt3 < THF
< DMF < MeCN ≈ CH2Cl2 (value not known for N2O).6 It is
evident that the enantioselectivity does not correlate with any
of these properties (nor does it correlate to other related8 solvent
scales). If one were to compare the observed ee to the predicted
ee (based on the dielectric constant), one could group the
solvents into those that gave a much greater ee than predicted
(DMF, MeCN), those that gave a moderate increase (THF), and
those that gave a moderate decrease (NEt3). The difference
between observed and predicted ee correlates to neither the
donor number nor the acceptor number of the solvent.
Solutions of coordinating reagents in hexane were also tested
as reaction media. The reagents were triethylphosphine, tricy-
clohexylphosphine, trimesitylphosphine, triphenylphosphine,
(R)-(+)-2,2′-bis(diphenylphosphino)-1,1′-binaphthyl (R-BINAP),
(S)-(-)-2,2′-bis(diphenylphosphino)-1,1′-binaphthyl (S-BINAP),
triethylphosphine oxide, triphenylphosphine oxide, and tri-n-
octylphosphine oxide. The results are listed in Table 1. The
bulkiest phosphine ligands (trimesitylphosphine, tricyclohexy-
lphosphine, and the BINAP ligands) did not seem to affect the
ee significantly. However, the smaller phosphines (PPh3 and
PEt3) caused a significant drop in the ee and caused the
formation of several unidentified byproducts. Phosphine oxides
and triethyl phosphate had no detrimental effect and in fact
greatly enhanced the ee of reactions performed in solvents that
had not been rigorously dried.
Product yield was also affected by the presence of coordinat-
ing reagents. In hexane, the conversion was 100% (after 1 h)
and the spectroscopically determined yield was 84%, with the
remainder being unidentified byproducts. In somewhat more
coordinating solutions, the yield was slightly lower (61% in
MeCN after 5 h). However, in the presence of strongly basic
coordinating reagents, the yields and conversions dropped
dramatically; with PPh3 for example, the product mixture
Rates of Reaction. The orange color of the diazoacetate took
longer to disappear in some reactions performed in coordinating
solvents or in the presence of coordinating additives such as
DMF, NEt3, THF, and PEt3 (but not trimesitylphosphine or the
phosphine oxides). NMR spectroscopy was used to track the
progress of the reaction in CDCl3, CD3CN, and d8-THF
([diazoacetate] ) 11 mM, [1] ) 0.14 mM, [styrene] ) 44 mM,

SolVent Effect on Cyclopropanation EnantioselectiVity
J. Am. Chem. Soc., Vol. 122, No. 32, 2000 7643
Figure 6. Log plot of the area of the methyl peak for methyl
1
phenyldiazoacetate in the H NMR spectrum in CD3CN as a function
of time during the cyclopropanation reaction at 20 °C.
Figure 8. Crystallographically determined molecular structure of [Rh2-
(TBSP)4(DMF)2]‚0.5toluene‚0.5n-pentane. A view emphasizing the
placement of the tert-butylphenylsulfonylprolinate ligands relative to
the Rh(1)-O(1)-O(5)-O(9)-O(13) surface. The hydrogen atoms and
the solvate molecules have been omitted.
Table 3. Crystal Data and Structure Refinement for
[Rh2(TBSP)4(DMF)2]‚0.5Toluene‚0.5n-Pentane
identification code
empirical formula
formula weight
temperature
mn1028
C72H104N6O18Rh2S4
1675.67
91(2) K
wavelength
0.71073 Å
crystal system
space group
unit cell dimensions
orthorhombic
P212121
a ) 10.6911(5) Å, R ) 90°
b ) 24.4434(11) Å, â ) 90°
c ) 29.7425(13) Å, γ ) 90°
7772.5(6) Å3
Figure 7. Crystallographically determined molecular structure of [Rh2-
(TBSP)4(DMF)2]‚0.5toluene‚0.5n-pentane. The hydrogen atoms and the
solvate molecules have been omitted.
volume
Z
4
density (calculated)
absorption coefficient
F(000)
1.432 Mg/m3
0.602 mm-1
T ) 20 °C). The peaks monitored were the methyl peaks of the
diazoacetate and the cyclopropane methyl ester (at 3.90 and 3.66
ppm, respectively, in CDCl3). At the addition of the diazoacetate,
the vial was shaken and then transferred to an NMR tube where
the first scan was taken. At approximately 15 min intervals,
the NMR tube was shaken and then another scan was taken. In
CDCl3, the reaction went so quickly that by the time the first
scan was taken (approximately 5 min after the start of the
reaction), the reaction was essentially complete. Therefore,
assuming first-order kinetics, kobs g 1 × 10-2 s-1. In CD3CN,
the reaction took much longer, more than 5 h. The log plot
(Figure 6) was found to be linear for at least 2 half-lives, show-
ing that the reaction was first order with respect to the diazo-
acetate. The observed rate constant, 9.6 × 10-5 s-1, was 2 orders
of magnitude lower than in CDCl3. In THF, the reaction was
3504
crystal size
0.35 × 0.05 × 0.05 mm3
crystal color and habit
diffractometer
Θ range for data collection
index ranges
blue needle
Bruker SMART 1000
1.60-31.52°
-15 e h e 15, -35 e k e 35,
-43 e l e 43
24950
reflections collected
independent reflections
24884 [R(int) ) 0.1144]
completeness to theta ) 31.52° 98.0%
absorption correction
max. and min. transmission
solution method
none
0.9705 and 0.8169
SHELXS-97 (Sheldrick, 1990)
SHELXL-97 (Sheldrick, 1997) full
matrix least-squares on F2
24884/0/868
refinement method
data/restraints/parameters
goodness-of-fit on F2
final R indices [I > 2σ(I)]
R indices (all data)
absolute structure parameter
largest diff. peak and hole
0.902
R1 ) 0.0590, wR2 ) 0.1096
R1 ) 0.1310, wR2 ) 0.1306
-0.02(2)
also slow with an observed rate constant of 2.2 × 10-4 s-1
.
1.850 and -1.174 e.Å-3
Crystal Structure of [Rh2(TBSP)4]. Crystals of the bis(N,N-
dimethylformamide) adduct of [Rh2(TBSP)4] were prepared and
crystallographically characterized in order to determine the Rh-
Rh bond length and to note the orientation of the ligands
(Figures 7 and 8, Tables 3 and 4). Although this catalyst is
commercially available, its X-ray structure has not previously
been published, but an illustration of the structure of the bis-
(H2O) adduct of the closely related complex tetrakis(N-benze-
nesulfonyl-L-prolinate)dirhodium(II) has been published.35,36
(35) Ferguson, G.; Lydon, K.; McKervey, M. A. Unpublished results,
as reported in Doyle, M. P.; McKervey, M. A.; Ye, T. Modern Catalytic
Methods for Organic Synthesis with Diazo Compounds; John Wiley &
Sons: New York, 1998.

7644 J. Am. Chem. Soc., Vol. 122, No. 32, 2000
Wynne et al.
Table 4. Selected Bond Lengths [Å] and Angles [deg] for
[Rh2(TBSP)4(DMF)2]‚0.5Toluene‚0.5n-Pentane
Scheme 1
Rh(1)-Rh(2)
Rh(1)-O(13)
Rh(1)-O(5)
Rh(1)-O(1)
Rh(1)-O(9)
Rh(1)-O(17)
2.3910(5)
2.020(3)
2.040(3)
2.043(3)
2.055(3)
2.264(3)
Rh(2)-O(10)
Rh(2)-O(2)
Rh(2)-O(14)
Rh(2)-O(6)
Rh(2)-O(18)
2.020(3)
2.040(3)
2.056(3)
2.060(3)
2.262(3)
O(13)-Rh(1)-O(5)
O(13)-Rh(1)-O(1)
O(5)-Rh(1)-O(1)
O(13)-Rh(1)-O(9)
O(5)-Rh(1)-O(9)
O(1)-Rh(1)-O(9)
O(13)-Rh(1)-O(17)
O(5)-Rh(1)-O(17)
O(1)-Rh(1)-O(17)
O(9)-Rh(1)-O(17)
O(13)-Rh(1)-Rh(2)
O(5)-Rh(1)-Rh(2)
O(1)-Rh(1)-Rh(2)
O(9)-Rh(1)-Rh(2)
177.28(14) O(10)-Rh(2)-O(2)
89.16(14) O(10)-Rh(2)-O(14)
89.70(14) O(2)-Rh(2)-O(14)
90.90(14) O(10)-Rh(2)-O(6)
90.07(13) O(2)-Rh(2)-O(6)
175.79(13) O(14)-Rh(2)-O(6)
89.34(14) O(10)-Rh(2)-O(18)
93.12(14) O(2)-Rh(2)-O(18)
89.61(12) O(14)-Rh(2)-O(18)
94.60(12) O(6)-Rh(2)-O(18)
88.99(10) O(10)-Rh(2)-Rh(1)
88.51(10) O(2)-Rh(2)-Rh(1)
177.05(13)
90.53(14)
89.53(15)
90.23(15)
89.49(15)
175.56(13)
86.53(12)
96.41(12)
92.23(14)
92.19(14)
88.87(9)
88.19(9)
87.65(10)
87.98(10)
88.11(9)
87.67(9)
O(14)-Rh(2)-Rh(1)
O(6)-Rh(2)-Rh(1)
O(17)-Rh(1)-Rh(2) 177.20(9)
O(18)-Rh(2)-Rh(1) 175.40(8)
Extensive data has been published on the effect of various
ligands L on the Rh-Rh bond length in Rh2(O2CR)4L2 (R )
Me or Pr).37-42 The Rh-Rh distances found by crystallography
vary from 2.4505 (L ) PPh3, R ) Me)43 down to 2.366 (L )
none, R ) Pr).44 The Rh-Rh bond distance found in complex
1 is 2.3910 Å, which is near the short end of the range but
comparable to that found45 in the complex Rh2(O2CCH3)4-
(DMF)2 and longer than the estimated Rh-Rh bond length of
Rh2(O2CR)4 in solution (2.33 Å).39
essentially no steric effect on the active sites. The sulfonyl
groups, on the other hand, are oriented to be able to influence
reactions of ligands bound to Rh(1). The closest distance
between a sulfonyl oxygen and the DMF oxygen atom is 4.5 Å
between O(15) and O(17). This represents the distance between
the sulfonyl O atom and the electrophilic carbon of a carbene
intermediate; too far for any kind of interaction but potentially
close enough to supply steric hindrance to an incoming olefin.
Discussion
Although in solution the floppy tails of the carboxylate ligands
are likely to be continuously and rapidly changing conformation,
the solid-state structure gives a snapshot of one possible
conformation. The proline portion of the ligands was found to
be fairly ordered; all of the proline rings were oriented such
that the four sulfonyl sulfur atoms were coplanar with Rh(1)
and the four oxygen atoms attached thereto. If viewed down
the Rh(2)-Rh(1) bond, these sulfonylproline fragments appear
to form a swastika-type arrangement. Rotation about the N-S
bond should allow the tert-butylphenyl groups to swing freely
in solution. In the solid state, two of these tert-butylphenyl
groups extend past Rh(1) so that they appear to be roughly
parallel to the DMF ligand on that Rh. The other two
tert-butylphenyl groups are oriented out and away from the
dirhodium core. The ability of these groups to swing so far away
from the active sites on either Rh suggests that they have
Cyclopropanation with dirhodium tetracarboxylates46,47 has
been studied quite extensively in terms of mechanism and
structure/selectivity relationships.3,16,48-50 Rhodium(II) carboxy-
lates have two open sites for catalysis, one on each electronically
unsaturated rhodium. In the currently accepted mechanism for
rhodium-catalyzed cyclopropanation (Scheme 1),16,48 the diaz-
oacetate binds to the metal and releases the dinitrogen to form
a carbene. This step is rate determining.51 The electrophilic
carbene can react with a nucleophilic olefin to form a cyclo-
propane ring. This is generally thought to occur in a nonsyn-
chronous manner where the carbene first interacts with the more
electron rich olefinic carbon.3 In asymmetric cyclopropanation,
the chirality transfer is from the chiral carboxylate ligands, which
block some orientations of the incoming olefin. There has been
much research into the relationship between the catalyst/alkene
structure and diastereoselectivity16 and some research into the
effects of structure on the enantioselectivity.3,50 For reaction 2,
the E diastereomer 2 is favored as is common with diazocar-
bonyl substrates and is formed with excellent diastereoselec-
tivity. The enantioselectivity is found to depend greatly on the
nature of the chiral ligands. For the [Rh2(TBSP)4] catalyst and
its analogues, it was determined that the stuctural features that
most enhanced the enantioselectivity were the aliphatic ring
attached to the carboxylate group and the existence of the
arylsulfonyl group.3
(36) X-ray structures of three other chiral dirhodium carboxylate catalysts
have appeared: [Rh2((R)-R-methoxy-R-phenylacetate)4(THF)2],36a [Rh2((S)-
mandalate)4(EtOH)2],36a and [Rh2(N-phthaloyl-(S)-phenylalaninate)4(4-
tbutylpyridine)2].36b (a) Agaskar, P. A.; Cotton, F. A.; Falvello, L. R.; Han,
S. J. Am. Chem. Soc. 1986, 108, 1214-1223. (b) Hashimoto, S.; Watanabe,
N.; Sato, T.; Shiro, M.; Ikegami, S. Tetrahedron Lett. 1993, 34, 5109.
(37) Christoph, G. G.; Halpern, J.; Khare, G. P.; Koh, Y. B.; Ro-
manowski, C. Inorg. Chem. 1981, 20, 3029-3037.
(38) Bursten, B. E.; Cotton, F. A. Inorg. Chem. 1981, 20, 3042-3048.
(39) Cotton, F. A.; Dikarev, E. V.; Feng, X. Inorg. Chim. Acta 1995,
237, 19-26.
(40) Alarco´n, C. J.; Lahuerta, P.; Peris, E.; Ubeda, M. A.; Aguirre, A.;
Garc´ıa-Granda, S.; Go´mez-Beltra´n, F. Inorg. Chim. Acta 1997, 254, 177-
181.
(46) Paulissen, R.; Reimlinger, H.; Hayez, E.; Hubert, A. J.; Teyssie, P.
Tetrahedron Lett. 1973, 2233.
(47) Hubert, A. J.; Noels, A. F.; Anciaux, A. J.; Teyssie, P. Synthesis
1976, 600.
(41) Cotton, F. A.; Felthouse, T. R.; Klein, S. Inorg. Chem. 1981, 20,
3037-3042.
(42) Cotton, F. A.; Kim, Y. Eur. J. Solid State Inorg. Chem. 1994, 31,
525-534.
(48) Yates, P. J. Am. Chem. Soc. 1952, 74, 5376.
(49) Pirrung, M. C.; Morehead, A. T. J. Am. Chem. Soc. 1994, 116,
8991-9000.
(43) Christoph, G. G.; Halpern, J.; Khare, G. P.; Koh, Y. B.; Ro-
manowski, C. Inorg. Chem. 1981, 20, 3029-3037.
(44) Cotton, F. A.; Shiu, K.-B. ReV. Chim. Miner. 1986, 23, 14.
(45) Moszner, M.; Glowiak, T.; Ziolkowski, J. J. Polyhedron 1985, 4,
1413-1417.
(50) Yoshikawa, K.; Achiwa, K. Chem. Pharm. Bull. 1995, 43, 2048-
2053.
(51) Doyle, M. P. Recl. TraV. Chim. Pays-Bas. 1991, 110, 305-316.

SolVent Effect on Cyclopropanation EnantioselectiVity
J. Am. Chem. Soc., Vol. 122, No. 32, 2000 7645
Scheme 2
Scheme 3
as the one shown as 1b in Scheme 2, leave one rhodium site
sterically unencumbered and therefore achiral (the rotomer found
for [Rh2(TBSP)4(DMF)2] in the solid state is shown as 1c). The
rationale is that in pentane a highly selective rotomer such as
1a is stabilized, while in CH2Cl2 a less selective rotomer closer
to 1b is the predominant form. In support of this theory is
Doyle’s finding that with rigid dirhodium carboxamide ligands
no solvent-dependent enantioselectivity was seen.4 To test this
theory, Davies17 synthesized catalysts 4 and 5, similar to the
most highly selective dirhodium catalysts but with carboxylate
ligands tethered to each other to hinder rotation (Scheme 3).
Although catalyst 4 did not have solvent-dependent enantiose-
lectivity for reaction 1, strong solvent dependence was observed
for catalyst 5 (from 74% ee in pentane to 90% in CH2Cl2) and
the trend was opposite to that observed with 1 or related
catalysts. The reason for this is not at all clear. It appears,
nevertheless, that the solvent dependence is caused by more
factors than just rotation within the carboxylate ligands.
The other explanation that Davies has suggested involves
electronic effects. It has been amply demonstrated that changing
the electronic nature of both the substrates and catalyst can affect
selectivity.2-5,16,49,50 The electronic nature of the substrate and
catalyst can affect the charge separation at the transition state
in which the carbene attracts the more electron rich carbon atom
of the olefin while the other alkene carbon becomes more
electropositive (Scheme 1). Davies3 suggested that the solvent
polarity could play a role in the timing of the transition state.
According to this explanation, solvents of high dielectric
constant favor an early transition state in which the styrene
molecule is farther away from the rhodium and therefore in a
less chiral environment. In pentane, the charge separation is
not stabilized at greater distances, the transition state must be
later, and the olefin must be closer to the carbene and therefore
in a more chiral environment.
Pressure-Dependent Enantioselectivity. The results confirm
that the dependence of the dielectric constant of a SCF upon
pressure can cause the enantioselectivity of homogeneous
catalysis to be pressure dependent.22 Because all SCFs have
pressure-dependent dielectric constants and many reactions have
dielectric-constant-dependent enantioselectivity, one can predict
the following. At any temperature close to but above Tc, the
enantioselectivity of an asymmetric synthesis near the critical
pressure of the reaction mixture may deviate from that at higher
pressure. If the principal cause of the deviation for a particular
system is known, then the direction of the deviation can be
predicted. In particular, if the cause of the deviation is the
pressure dependence of the dielectric constant of the SCF, then
the enantioselectiVity in supercritical fluids will be greater near
the critical pressure than at higher pressures if the enantiose-
lectiVity in liquids is greater in nonpolar rather than in polar
solVents. This prediction will be true for all supercritical fluids,
because they all have lower dielectric constants near Pc than at
higher pressures. However, the magnitude of the deviation in
enantioselectivity will be proportional to the magnitude of the
change in dielectric constant (a property of the SCF) and
proportional to the magnitude of the dielectric-constant depen-
dence observed in liquid solvents (a property of the reaction).
While the change in dielectric constant is large in SCFs which
have significant dipole moments, it is small in scCO2. Thus,
one would not expect a significant pressure dependence in scCO2
based on dielectric constant change. It may be possible, however,
to obtain a large dielectric constant change by pressure changes
on a mixture of scCO2 with a polar cosolvent. This is the subject
of future research.
The Effect of Noncoordinating Solvents. There has been a
small amount of work on solvent effects in enantioselective
cyclopropanation, most notably the elegant work by Davies and
co-workers.2,3 Davies has proposed two possible explanations
for the solvent effects on enantioselectivity seen in this system.
The first explanation invokes steric arguments, in which the
chiral ligands are stabilized as different rotomers in different
solvents. The carboxylate groups have free rotation about the
proline carboxylate C-C bond and the N-S bond between the
proline nitrogen and the sulfonyl group. Therefore, the floppy
ligands could potentially have many orientations. Davies sug-
gests that the greatest chiral induction would be obtained from
the rotomer shown as 1a in Scheme 2 due to its high symmetry
and more hindered catalytic sites. Other possible rotomers, such
Our results in both supercritical fluids and noncoordinating
solvents confirm a dependence of the enantioselectivity on the
dielectric constant. Even with the new data, it is difficult to
determine whether the dependence is due to steric or electronic
effects.
The Effect of Coordinating Solvents or Reagents. The new
results demonstrate convincingly that in coordinating solvents
or in solutions of coordinating reagents the dielectric constant
is not the sole factor affecting the selectivity. It is obvious that
in some manner the ability of the solvent or additive to
coordinate affects the enantioselectivity. However, there is no

7646 J. Am. Chem. Soc., Vol. 122, No. 32, 2000
Wynne et al.
obvious trend. Some solvents, such as THF and acetonitrile,
cause the enantioselectivity to be greater than one would expect
on the basis of the dielectric constants of the solvents. Other
solvents or mixtures such as NEt3 or PPh3/hexane cause the
enantioselectivity to be poorer than expected. The ee does not
correlate exactly to any single characteristic property of the
solvent (such as donor number or ET(30)). How does the
coordinating ligand affect selectivity? The two possibilities are
still steric and electronic effects.
For the steric argument, one might suggest that the coordinat-
ing solvent or ligands bind to one site and influence the
arrangement of the chiral auxiliaries so that catalysis occurring
at the other Rh is in a sterically altered environment. Such an
argument would suggest that a very large phosphine bound to
one Rh would force the prolinate ligands to the other Rh and
therefore increase the enantioselectivity. The results are not
definitive. PPh3 (cone angle52 ) 145°) and trimesitylphosphine
(212°) were tested, but the trimesitylphosphine was too large;
the failure of the UV/vis spectrum to change upon addition of
that phosphine indicates that it did not bind to the complex
(Table 2). PCy3 and PEt3 were also tested; they are electronically
similar to each other, but the former is sterically much larger
(a cone angle of 170° compared to 132°). The UV spectral
changes show that both reacted with the catalyst. The ee of the
cyclopropanation in the presence of PCy3, however, is much
greater than that in the presence of PEt3. In general, the larger
phosphines gave better enantioselectivity than the smaller
phosphines, but no phosphine raised the ee above that observed
in the absence of phosphine.
solvents allowed higher ee’s than would have been expected
on the basis of their dielectric constants.
Pirrung and co-workers found a relationship between the
electronic nature of the carboxylate ligands on dirhodium
catalysts and the selectivity between ylide formation and C-H
insertion.49 They found that the ∆∆G‡ between the transition
states for the two pathways varied with the nature of the
carboxylate ligands. It was found that the selectivity depended
directly on the polarizability of the carboxylate ligands, where
the best selectivity was seen with carboxylate ligands of low
field effects and high polarizability. The polarizability was
hypothesized to help stabilize the carbene intermediate and
accept back-bonding electron donation. While electronic effects
of this type may be involved with a few of the coordinating
ligands, it is difficult to adapt this theory to explain the effect
of most of the coordinating solvents or reagents on the
enantioselectivity of cyclopropanation. For example, trieth-
ylphosphine is more polarizable than triethylamine, but both
greatly decrease the enantioselectivity.
(2) In the presence of particularly donating ligands, one or
more of the carboxylates in [Rh2(TBSP)4L2] could become
monodentate (dangling), reducing the number of chiral ligands
proximate to the active site. Pyridine, for example, reacts with
[Rh2(O2CCF3)4] at room temperature to form [Rh2(O2CCF3)4-
(py)4] with two dangling carboxylates (structure 6).54 [Rh2(O2-
The possibility of the coordinating species influencing the
selectivity due to electronic effects is likely. Three possible ways
that the coordinating solvent can affect the electronic properties
of the catalyst are (1) coordination to one rhodium affects the
electronic density at the other rhodium, therefore possibly
changing the nature of rhodium-carbene bond and thus the
electronic nature of the transition state, (2) coordinating species
may disrupt the dirhodium tetracarboxylate core, and (3)
coordinating species might interact directly with the carbene.
These three scenarios will now be discussed.
(1) Presumably, in the presence of excess coordinating ligand
(L), the diadduct [Rh2(TBSP)4L2] is formed. This should be a
catalytically inactive species (hence the slower reaction in
strongly coordinating solvents) and would become active only
upon loss of 1 equiv of L. Thus, the catalysis (carbene formation
and subsequent cyclopropanation) would take place at the
unsaturated Rh, while 1 equiv of L would remain bound to the
other Rh. The presence of this bound ligand L could affect the
catalyst by donating electron density into the Rh-Rh σ*
antibonding orbital and therefore weaken the Rh-Rh bond and
increase the electron density at both metals. For example, PPh3
donates electron density far better than weaker ligands such as
H2O and hence the Rh-Rh distance is much greater in the bis-
(PPh3) adduct41,43 than the bis(H2O) adduct42,53 of dirhodium
tetracarboxylates. The existence of the axial ligand on one Rh
atom in LRh(µ-O2CR)4RhdCPhCO2Me could therefore change
the electrophilicity and ability of the other Rh to stabilize charge
as well as change the nucleophilicity of the carbene carbon.
While this is probably true in the present system, it is not
sufficient to explain the results. Although weakly coordinating
ligands in general allowed high ee’s while strongly coordinating
ligands did not, a few strongly coordinating ligands (PCy3 and
BINAP) allowed high ee’s. In addition, almost all coordinating
CR)4(PPh3)2] reacts even further (eq 3, L ) RCO2H, R ) CF3),
resulting eventually in a complex containing two orthometalated
-
and bridging Ph2PC6H4 ligands.54-56 This isomerization se-
quence is known to take place even at room temperature.54
While we have no direct evidence for the formation of dangling
carboxylates or orthometalated phosphines in the cyclopropa-
nation, the time period required for the cyclopropanation in the
presence of PPh3 is so long that there could be time for partial
isomerization of the bis(PPh3) adduct.
(3)
(54) Lahuerta, P.; Ubeda, M. A.; Paya´, J.; Garc´ıa-Granda, S.; Gomez-
Beltra´n, F.; Anillo, A. Inorg. Chim. Acta 1993, 205, 91-97.
(55) Chakravarty, A. R.; Cotton, F. A.; Tocher, D. A. J. Chem. Soc.,
Chem. Commun. 1984, 501-502.
(56) Chakravarty, A. R.; Cotton, F. A.; Tocher, D. A.; Tocher, J. H.
Organometallics 1985, 4, 8-12.
(52) Tolman, C. A. Chem. ReV. 1977, 77, 313-348.
(53) Cotton, F. A.; DeBoer, B. G.; LaPrade, M. D.; Pipal, J. R.; Ucko,
D. A. Acta Crystallogr., Sect. B 1981, B27, 1664.

SolVent Effect on Cyclopropanation EnantioselectiVity
J. Am. Chem. Soc., Vol. 122, No. 32, 2000 7647
Dangling carboxylate ligands are even more likely to form
if a bidentate additive is added. For example, (R)-(+)-2,2′-bis-
(diphenylphosphino)-1,1′-binaphthyl (R-BINAP) and its S-
enantiomer were able to bind to [Rh2(S-TBSP)4], as indicated
by a change in the color and spectrum. The R-BINAP and
S-BINAP adducts of [Rh2(S-TBSP)4] had similar UV/vis.
spectra, but the spectra were significantly different from that
of the PPh3 adduct. The BINAP adducts also catalyzed the
cyclopropanation with far greater enantioselectivity than the
PPh3 adduct. These observations strongly suggest that the
manner of binding of BINAP differed significantly from the
manner of binding of PPh3. We suggest that the BINAP ligand
was likely to have chelated and thereby forced at least one of
the carboxylate ligands into a dangling position. The reason this
does not cause a decrease in the enantioselectivity is not known.
(3) The third possibility is that the coordinating solvent or
additive might directly interact with the carbene before or even
during the transition state of the cyclopropanation. For example,
it is likely that the electrophilic carbene would attract the Lewis
basic phosphine oxide. This could stabilize the carbene in the
form of an ylide (eq 4). While there is no precedent for an ylide
Further study will be required to determine the reasons for
the widely differing enantioselectivities in the presence of these
various coordinating solvents and additives. It is likely that more
than one reason is involved; the manner in which PPh3 lowers
the enantioselectivity should be different from the manner in
which weaker ligands increase the enantioselectivity. Neverthe-
less, the results are promising because they indicate new ways
in which enantioselectivity can be enhanced.
Conclusions
1. The dependence of the dielectric constant of a SCF upon
pressure can cause the enantioselectivity of homogeneous
catalysis to be pressure dependent. Cyclopropanation in fluo-
roform has been described as the first example of this effect on
homogeneous catalysis. Predictions have been made that future
examples will be found, in which, at a temperature close to but
above Tc, the enantioselectivity near the critical pressure of the
reaction mixture will deviate from that at higher pressure. If
the cause of the deviation is the pressure dependence of the
dielectric constant of the SCF, then the enantioselectivity in
supercritical fluids will be greater near the critical pressure than
at higher pressures if the enantioselectivity in liquids is greater
in nonpolar rather than in polar solvents.
(4)
2. This pressure dependence was used as part of a combined
liquid/supercritical solvent study for the purposes of mechanistic
understanding and selectivity optimization for the cyclopropa-
nation reaction. Our results in both supercritical fluids and
noncoordinating solvents confirm a strong dependence of the
enantioselectivity on the dielectric constant of the medium.
However, in coordinating solvents or solvent/additive mixtures,
the dielectric constant is not the sole factor affecting the
selectivity. In fact, the enantioselectivity in such solutions is,
in some manner, strongly dependent on the coordinating ability
or nucleophilicity of the solvent or additive L. Although
coordinating solvents enhance the enantioselectivity relative to
what one would expect on the basis of the dielectric constant,
the best enantioselectivity was obtained in solvents or in the
presence of reagents which showed no UV/vis evidence of
binding to the catalyst. Three different explanations are offered
for the effect of coordinating solvents or reagents; coordination
of the solvent to form a LRh(µ-O2CR)4RhdCPhCO2Me com-
plex, disruption of the Rh2(O2CR)4 core to create dangling
carboxylates, or coordination of L to the carbene to form an
ylide. There is as yet insufficient evidence to distinguish between
these; it is possible that all three occur with different coordinat-
ing species.
of exactly this type, trimethylamine oxide is known to accelerate
carbonyl substitution by nucleophilic attack on the carbonyl
carbon57 (trialkyphosphine oxides are believed to accelerate
reactions of metal carbonyl by binding to the metal and not by
forming an ylide with the carbonyl).58,59 Rhodium carboxylates
are known to act as catalysts for the formation of oxonium ylides
from diazo compounds and ethers.16 In fact, dirhodium tetra-
carboxylate-catalyzed ylide formation between a carbene and
an ether group can compete with cyclopropanation.60 Ylide
formation in this manner could protect the carbene from
adventitious water, which would explain the beneficial effects
of phosphine oxides in the presence of trace water.61
Free ylide formation may also occur in the presence of
phosphines62 or amines16 plus a catalyst; this may be the reason
for the very low yields and large number of byproducts observed
when the cyclopropanation was performed in the presence of
PPh3, PEt3, and NEt3.
(57) Albers, M. O.; Coville, N. J. Coord. Chem. ReV. 1984, 53, 227-
259 and references therein.
(58) Webb, S. L.; Giandomenico, C. M.; Halpern, J. J. Am. Chem. Soc.
1986, 108, 345-347.
Acknowledgment. This material is based upon work sup-
ported by the EPA/NSF Partnership for Environmental Research
under NSF Grant 9815320. We also gratefully acknowledge Drs.
Huw Davies and Susan Tucker for helpful discussions, Mrs.
Heidi Sickafoose for assistance in equipment design, and
especially Dr. Michael Doyle for advice and a preliminary
supply of catalyst.
(59) Darensbourg, D. J.; Walker, N.; Darensbourg, M. Y. J. Am. Chem.
Soc. 1980, 102, 1213-1214.
(60) Doyle, M. P.; Peterson, C. S. Tetrahedron Lett. 1997, 38, 5265-
5268.
(61) The phosphine oxides and phosphates could be acting as drying
agents, although we are not aware of studies in which they are used as
such or shown to be effective.
(62) Seyferth, D.; Grim, S. O.; Read, T. O. J. Am. Chem. Soc. 1960, 82,
1510-1511.
(63) This plot is likely to be more accurate than an earlier version based
entirely on the equation. Jessop, P. G.; Ikariya, T.; Noyori, R. Chem. ReV.
1999, 99, 475-493.
Supporting Information Available: Description of tests of
the effects of temperature, injection method, trace water, and
steel vessel walls on enantioselectivity of cyclopropanations in
SCFs. Tables of atomic coordinates, equivalent isotropic
displacement parameters, anisotropic displacement parameters,
bond lengths, and bond angles for [Rh2(TBSP)4(DMF)2]‚
0.5toluene‚0.5n-pentane. Log plot for reaction 2 in THF-d8. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(64) Makita, T.; Kubota, H.; Tanaka, Y.; Kashiwagi, H. Gakujutsu
Koenkai Koen Rombunshu, Refrigeration Tokyo Nippon Reito Kyokui 1976,
52, 543.
(65) Rhodes, T. A.; O’Shea, K.; Bennett, G.; Johnston, K. P.; Fox, M.
A. J. Phys. Chem. 1995, 99, 9903-9908.
(66) Aizpiri, A. G.; Rey, A.; Da˜vila, J.; Rubio, R. G.; Zollweg, J. A.;
Streett, W. B. J. Phys. Chem. 1991, 95, 3351-3357.
(67) Thermophysical Properties of Freons. Methane Series, Part 1;
Altunin, V. V., Geller, V. Z., Petrov, E. K., Rasskazov, D. C., Spiridinov,
G. A., Selover, T. B., Jr., Eds.; Hemisphere Publishing Corp.: Washington,
1987.
JA000894N