Mechanistic Study of the Jacobsen Asymmetric Epoxidation of Indene

Article abstract of DOI:10.1021/jo961735i

The asymmetric epoxidation of indene using aqueous NaOCl, catalyzed by Jacobsen's chiral manganese salen complex, provides indene oxide in 90% yield and 85-88% enantioselectivity. The axial ligand, 4-(3-phenylpropyl)pyridine N-oxide (P₃NO), increases the rate of epoxidation without affecting enantioselectivity and also stabilizes the catalyst. These two effects afford a reduction in catalyst loading to <1%. The turnover-limiting step in the catalytic cycle has been determined to be the oxidation of the manganese catalyst, based on reaction orders of 0 in indene and 1 in catalyst and also based on the dependence of the rate on the hypochlorite concentration. In the presence of the ligand P₃NO, this rate-limiting oxidation occurs in the organic phase with HOCl as oxidant, as shown by the dependence of the rate on the NaOH concentration. P₃NO assists the transport of HOCl to the organic layer as demonstrated by titration studies and by measuring the rates of oxidation of a redox indicator, diphenylbenzidine. On the other hand, stirring speed studies indicate that, in the absence of the ligand, oxidation occurs at the interface. Thus, the axial ligand plays at least two roles in the epoxidation of indene: it stabilizes the catalyst, presumably by ligation, and it increases the epoxidation reaction rate by drawing the active oxidant, HOCl, into the organic layer.

Full text of DOI:10.1021/jo961735i

2222
J . Org. Chem. 1997, 62, 2222-2229
Mech a n istic Stu d y of th e J a cobsen Asym m etr ic Ep oxid a tion of
In d en e
David L. Hughes,* George B. Smith,* J i Liu, George C. Dezeny, Chris H. Senanayake,
Robert D. Larsen, Thomas R. Verhoeven, and Paul J . Reider
Process Research, Merck Research Laboratories, P.O. Box 2000, Rahway, New J ersey 07065
Received September 10, 1996X
The asymmetric epoxidation of indene using aqueous NaOCl, catalyzed by J acobsen’s chiral
manganese salen complex, provides indene oxide in 90% yield and 85-88% enantioselectivity. The
axial ligand, 4-(3-phenylpropyl)pyridine N-oxide (P3NO), increases the rate of epoxidation without
affecting enantioselectivity and also stabilizes the catalyst. These two effects afford a reduction in
catalyst loading to <1%. The turnover-limiting step in the catalytic cycle has been determined to
be the oxidation of the manganese catalyst, based on reaction orders of 0 in indene and 1 in catalyst
and also based on the dependence of the rate on the hypochlorite concentration. In the presence
of the ligand P3NO, this rate-limiting oxidation occurs in the organic phase with HOCl as oxidant,
as shown by the dependence of the rate on the NaOH concentration. P3NO assists the transport
of HOCl to the organic layer as demonstrated by titration studies and by measuring the rates of
oxidation of a redox indicator, diphenylbenzidine. On the other hand, stirring speed studies indicate
that, in the absence of the ligand, oxidation occurs at the interface. Thus, the axial ligand plays
at least two roles in the epoxidation of indene: it stabilizes the catalyst, presumably by ligation,
and it increases the epoxidation reaction rate by drawing the active oxidant, HOCl, into the organic
layer.
In tr od u ction
The mechanism of the transition metal-catalyzed ep-
oxidation has been the subject of speculation for the last
decade. Recent controversy has centered on whether the
reaction takes place as a direct transfer of the oxygen
from the metal to the substrate or whether an intermedi-
ate oxametallocycle is formed. The hypothesis for the
oxametallocycle was first advanced by Collman and co-
workers and was based on kinetic measurements.9a,b The
reactions were independent of olefin concentration, yet
different olefins exhibited variable epoxidation rates. This
suggested that the olefin and catalyst reversibly formed
an intermediate that broke down in the rate-determining
step, and as expected from this reasoning, the reactions
followed Michaelis-Menten kinetics.9a,b However, the
conclusions from this paper were later withdrawn as
results from further experimental work demonstrated
poor mass balance and instability of the oxidant under
the conditions used for the kinetic experiments.9b,c On
the other hand, Norrby and colleagues recently argued
that the oxetane was a viable intermediate based on
theoretical considerations,10 and Katsuki observed non-
linearity in Eyring plots for the epoxidation of alkenes
using modified chiral manganese salen catalysts,11 which
is consistent with an intermediate on the reaction
pathway. However, these results were questioned by
J acobsen,12 who noted that the variation in ee is less than
10%, making interpretation difficult. On the other hand,
J acobsen has demonstrated a linear temperature/ee
The versatility of asymmetric epoxides as intermedi-
ates in organic synthesis has sparked considerable effort
in their synthesis over the past 2 decades. In the 1970s
Sharpless1 and others2 discovered that achiral epoxides
could be generated from alkenes using metal catalysts
and alkyl peroxides. As an extension of this work, the
first significant breakthrough in realizing asymmetric
epoxidations came in 1980 when Sharpless and Katsuki
reported that the reaction of allylic alcohols with t-
BuOOH, Ti(O-i-Pr)4, and (+)- or (-)-diethyl tartrate
provided high yields of epoxides with enantioselectivities
up to 90%.3 Soon after, Groves discovered that chiral
transition metal porphyrin complexes were able to ep-
oxidize styrenes with enantioselectivities up to 51%.4
Building on the work of Kochi5 and Burrows6 in the
1980s, who used achiral manganese salen catalysts for
epoxidations, J acobsen reported in 1990 that unfunction-
alized alkenes could be epoxidized with high enantio-
selectivities using chiral manganese salen catalysts.7
High ee’s, straightforward catalyst preparation, use of
simple oxidants such as aqueous hypochlorite, and suc-
cess of the method with a wide variety of olefins have
made the preparation of chiral epoxides practical and
efficient using the J acobsen protocol.8
X Abstract published in Advance ACS Abstracts, March 15, 1997.
(1) Sharpless, K. B.; Michaelson, R. C. J . Am. Chem. Soc. 1973, 95,
6136.
(2) (a) Hart, H.; Lavrik, P. B. J . Org. Chem. 1974, 39, 1793. (b)
Sheldon, R. A. Recl. Trav. Chim. Pays-Bas 1973, 92, 253.
(3) Katsuki, T.; Sharpless, K. B. J . Am. Chem. Soc. 1980, 102, 5974.
(4) Groves, J . T.; Meyers, R. S. J . Am. Chem. Soc. 1983, 105, 5791.
(5) Srinivasan, K.; Michaud, P.; Kochi, J . K. J . Am. Chem. Soc. 1986,
108, 2309-2320. Srinivasan, K.; Perrier, S.; Kochi, J . K. J . Mol. Catal.
1986, 36, 297-317.
(9) (a) Collman, J . P.; Brauman, J . I.; Meunier, B.; Raybuck, S. A.;
Kodadek, T. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 3245-3248. (b)
Collman, J . P.; Brauman, J . I.; Meunier, B.; Hayashi, T.; Kodadek, T.;
Raybuck, S. A. J . Am. Chem. Soc. 1985, 107, 2000-2005. (c) Collman,
J . P.; Brauman, J . I.; Hampton, P. D.; Tanaka, H.; Bohle, D. S.;
Hembre, R. T. J . Am. Chem. Soc. 1990, 112, 7980-7984.
(10) Norrby, P.-O.; Linde, C.; Akermark, B. J . Am. Chem. Soc. 1995,
117, 11035-11036.
(6) Yoon, H.; Burrows, C. J . J . Am. Chem. Soc. 1988, 110, 4087.
(7) Zhang, W.; Loebach, J . L.; Wilson, S. R.; J acobsen, E. N. J . Am.
Chem. Soc. 1990, 112, 2801.
(11) Hamada, T.; Fukuda, T.; Imanishi, H.; Katsuki, T. Tetrahedron
1996, 52, 515-530.
(8) J acobsen, E. N. In Catalytic Asymmetric Synthesis; Ojima, I.,
Ed.; VCH Publishers: New York, 1993; pp 159-202.
(12) Finney, N. S.; Pospisil, P. J .; Chang, S.; Palucki, M.; Hansen,
K. B.; J acobsen, E. N. Private communication.
S0022-3263(96)01735-5 CCC: $14.00 © 1997 American Chemical Society

J acobsen Asymmetric Epoxidation of Indene
J . Org. Chem., Vol. 62, No. 7, 1997 2223
relationship over a 100 °C range, which argues against
an intermediate in the reaction.12
Another curious and ill-defined aspect of these epoxi-
dations is the role of the cocatalysts, which include
variously substituted pyridines, imidazoles, and N-oxides,
that accelerate the reactions and sometimes increase the
enantioselectivities. These additives are generally thought
to act as axial ligands on the transition metal catalyst,
which help to activate the catalyst either toward oxida-
tion or toward reactivity with the olefin. J acobsen has
recently presented convincing evidence that the N-oxides
behave as axial ligands, since the catalyst with a strapped
N-oxide ligand is more active that the standard catalyst
and is no longer affected by added N-oxide ligands.12
The asymmetric epoxidation of indene is a key step in
the formation of 1,2-aminoindanol, a component of the
HIV-protease inhibitor CRIXIVAN. J acobsen reported
that indene could be epoxidized in the presence of
4-phenylpyridine N-oxide with 1.5 mol % of the chiral
manganese salen catalyst, providing an 80% yield and
84% ee of chiral indene oxide.13 Recently, Senanayake
and co-workers found that the use of 4-(3-phenylpropyl)-
pyridine N-oxide as ligand allowed for a more reproduc-
ible reaction and the catalyst loading could be reduced
to 0.4 mol %.14
F igu r e 1. Rate of epoxidation of indene with variable initial
levels of indene. Reactions run with 25 mM catalyst and 100
mM P3NO at 860 rpm, by method 2 as described in the
Experimental Section.
The present contribution describes mechanistic studies
of the asymmetric epoxidation of indene and the role of
the N-oxide. The N-oxide is involved both in accelerating
the epoxidation and in stabilizing the catalyst and,
unexpectedly, in transporting the oxidant HOCl into the
organic layer.
Sch em e 1. Ca ta lytic Cycle for Ep oxid a tion of
In d en e by Hyp och lor ite
Resu lts a n d Discu ssion
Rea ction Over view . The standard epoxidation of
indene (eq 1) is carried out in a two-phase mixture that
includes (S,S)-(salen)MnIIICl catalyst (1), 4-(3-phenyl-
propyl)pyridine N-oxide (P3NO) (2), indene, and chlo-
robenzene as the organic components and 1.5 M aqueous
sodium hypochlorite as the terminal oxidant. The cata-
throughout the reaction were all measurably the same
at 85-88%. All of the components of the reaction
(indene, indene oxide, P3NO, and catalyst) were moni-
tored over the course of the reactions using both normal
and reverse phase HPLC.
Since the catalyst degraded as the epoxidation reaction
proceeded, it was vital to have in place an assay for the
catalyst in order to interpret kinetic data. A normal
phase HPLC assay was developed using a Whatman
Partisil 10 PAC column, as described in the Experimental
Section. Using this assay the catalyst levels were
routinely determined in epoxidation experiments. The
results are discussed in conjunction with the epoxidation
kinetics in the following sections.
Rea ction Or d er s. 1. Ca ta lyst. Reactions run at
variable catalyst levels indicated that the epoxidation is
roughly first order in catalyst concentration.
2. In d en e. A kinetic profile of indene concentrations
for reactions run using 1.6, 3.2, and 4.8 M indene in
chlorobenzene is shown in Figure 1. In each case, the
catalyst (25 mM) and P3NO (100 mM) levels were kept
constant and the reaction was carried out at -5 °C. For
each curve, the disappearance of indene was zeroth order
for most of the reaction, with slowing only at the end.
The abrupt disappearance of indene and the similarity
in initial rates for the three concentrations examined
were also consistent with an order of 0 for indene. The
kinetic profile was also similar to those observed by
Collman and co-workers for epoxidations of olefins using
a manganese(III) porphyrin catalyst in a two-phase
system with LiOCl as oxidant and a phase transfer agent.
lytic cycle is shown in Scheme 1. The optimum conditions
for ee, yield, and productivity include the use of 0.75 mol
% catalyst vs indene, 3 mol % P3NO vs indene, 3 M
indene in the chlorobenzene, and 1.5 M NaOCl in the
aqeuous phase; reaction is complete within 2.5 h at -5
°C with a yield of >90% and an ee of 85-88%. The ee
values of the indene oxide taken at 0.5 h intervals
(13) J acobsen, E. N.; Zhang, W.; Muci, A. R.; Ecker, J . R.; Deng, L.
J . Am. Chem. Soc. 1991, 113, 7063.
(14) Senanayake, C. H.; Smith, G. B.; Ryan, K. M.; Fredenburgh,
L. E.; Liu, J .; Roberts, F. E.; Hughes, D. L.; Larsen, R. D.; Verhoeven,
T. R.; Reider, P. J . Tetrahedron Lett. 1996, 37, 3271-3274.

2224 J . Org. Chem., Vol. 62, No. 7, 1997
Hughes et al.
They were able to fit their kinetic data to the Michaelis-
Menton equation and determined Km and Vmax constants
for several olefins.9a,b However, these results were
subsequently retracted as further experimentation re-
vealed that the hypochlorite was degrading under the
conditions used for the kinetic experiments.9c In the
event, we have been able to use the Michaelis-Menten
kinetics (eq 2) to fit our data, with values of Km ) (k-1
+
k2)/k1 ) 0.8 M and k2 ) 100 h-1
.
k2
98
catalyst + indene yk1z complex
k-1
catalyst + indene oxide (2)
Using a kinetic model which takes into account the rate
of catalyst degradation and avoids the steady state
assumption for the complex, the values changed slightly
to Km ) 0.4 M and k2 ) 90 h-1. However, as discussed
below, this validity of this model becomes a mute point,
as we show that the rate-determining step is catalyst
oxidation and therefore the kinetics can provide no
information on steps occurring after the rate-determining
step.
Effect of th e N-Oxid e P 3NO on In d en e Ep oxid a -
tion . After surveying a wide range of additives for the
epoxidation, Senanayake and co-workers found that the
best results in terms of yield and conservation of catalyst
were obtained with the lipophilic N-oxide 2, termed P3-
NO.14 Thus, this N-oxide was chosen for further inves-
tigations. A series of epoxidations was run using P3NO
at levels varying from 0 to 10 equiv vs the catalyst. The
reaction profiles for indene epoxidation are shown in
Figure 2a. For these reactions, the P3NO, Mn salen
catalyst, chlorobenzene, and indene were mixed at ambi-
ent temperature such that all solids dissolved. This
solution was then cooled to -10 °C and added to the
hypochlorite solution at -5 °C to initiate the reaction.
Although the P3NO was supersaturated in the mixture
when P3NO/catalyst > 1, it remained in solution for the
duration of the experiments except for the case where
P3NO/catalyst ) 10, where precipitation occurred after
30 min of reaction.15 The curves in Figure 2a indicate
that the epoxidation rate is increased with an increasing
level of P3NO. With 1 equiv or less of P3NO, an initial
burst is followed by a slower conversion to indene oxide,
while with greater than 1 equiv the reactions stay near
zeroth order in indene for most of the reaction. The
enantioselectivity was unchanged within experimental
error at 85-88%, indicating that the ee is not dependent
on the P3NO level. In another experiment, the reaction
was started with no P3NO present, then 4 equiv of P3-
NO vs catalyst was added at 3 h, and the epoxidation
accelerated immediately (Figure 3). Thus, the effect of
P3NO on the indene epoxidation was felt instantaneously.
P3NO also has a significant effect on the stability of
the catalyst, as shown for the same set of reactions in
Figure 2b. These data are consistent with a model in
which P3NO serves as an axial ligand for the catalyst,
as depicted in 3, where the ligand activates the catalyst
for the epoxidation and also stabilizes the catalyst.
F igu r e 2. (a) Effect of variable P3NO levels on the rate of
indene epoxidation. Reactions run with 0.75 mol % catalyst
vs indene at 860 rpm, by method 1 as described in the
Experimental Section. (b) Effect of variable P3NO levels on
the rate of catalyst degradation.
However, experiments discussed below suggest that this
is not the only role of P3NO.
(15) If P3NO is not dissolved at room temperature but added first
to the cold aqueous NaOCl, the initial concentration of P3NO is 6 mg/
mL, which is equivalent to 1 mol/mol of catalyst at a catalyst loading
of 0.75% vs indene. As the reaction proceeds, the concentration
increases, since the indene oxide product increases its solubility. The
kinetics in this case are somewhat different than those in which all
P3NO is dissolved initially, and interpretation is more complex since
the level changes throughout the reaction.
Effect of Hyd r oxid e a n d Hyp och lor ite Con cen -
tr a tion s. Prepared by reaction of molecular chlorine

J acobsen Asymmetric Epoxidation of Indene
J . Org. Chem., Vol. 62, No. 7, 1997 2225
F igu r e 5. Effect of variation in catalyst, NaOH, and NaOCl
levels on the rate of indene epoxidation in the absence of P3-
NO.
F igu r e 3. Effect of addition of P3NO to an indene epoxidation
at 3 h time point. Conditions: catalyst, 0.75 mol % vs indene;
stirring rate, 420 rpm; 3 mol % P3NO added as a solid at 3 h
point.
with water and then adding back NaOH), the reaction
rate slowed by roughly 2-fold. In another experiment,
both the NaOCl and NaOH concentrations were halved
(by simply diluting with water), and this had no effect
on the epoxidation rate. These data can be readily
rationalized based on the equilibrium equations for HOCl
and water dissociation (eqs 3, 4). They demonstrate that
the rate of the epoxidation is directly proportional to the
amount of HOCl present, i.e., that HOCl is involved in
the rate-determining step. From eq 5, the HOCl con-
centration is inversely proportional to the hydroxide
concentration, which is consistent with the reaction
profiles shown in Figure 4 that show the reaction slows
with increasing NaOH. The HOCl concentation is also
Keq
HOCl y z H+ + OCl-
(3)
(4)
Kw
H2O y z H+ + OH-
Kw[H2O][OCl-]
[HOCl] )
(5)
Keq[OH-]
proportional to the OCl- concentration, and we observe
a rate decrease when the OCl- concentration is decreased
if the NaOH concentration is held constant. In addition,
when OCl- and HO- are decreased proportionally, as
occurs with just a simple water dilution, then eq 5
predicts that there will be no effect on the HOCl
concentration, and this is consistent with the observation
that diluting the NaOCl by one-half with water causes
no rate effect.
In contrast to the above results, reactions in the
absence of P3NO behaved completely differently. As
shown in Figure 5, variation in the concentrations of
hydroxide and hypochlorite had essentially no effect on
the rate of epoxidation. However, increasing the con-
centration of catalyst increased the rate. These results
suggest that, in the absence of the N-oxide, the catalyst
is still involved in the rate-determining step, but [HOCl]
no longer controls the reaction rate. As discussed in the
F igu r e 4. Effect of NaOH concentration on the rate of indene
epoxidation. Reactions run with 0.75 mol % catalyst and 3 mol
% P3NO, both vs indene, at 860 rpm, by method 2 as described
in the Experimental Section.
with aqueous NaOH, commercial aqueous sodium hy-
pochlorite contains residual levels of hydroxide that vary
among sources. The commercial material from Spectrum
Chemicals consistently contained 0.13 M NaOH in the
1.5-1.6 M NaOCl. The effects of the hypochlorite and
hydroxide concentrations on the epoxidation were deter-
mined both with and without P3NO present. Figure 4
illustrates the effect of increasing the hydroxide concen-
tration from 0.13 to 0.3 M in the presence of 4 equiv of
P3NO vs catalyst at a catalyst level of 0.75 mol % vs
indene. At -5 °C the initial reaction rate was slowed by
1.7-fold when the NaOH concentration was increased.
When the concentration of NaOCl was cut in half, but
the NaOH kept constant at 0.13 M (by diluting NaOCl

2226 J . Org. Chem., Vol. 62, No. 7, 1997
Hughes et al.
next section, these data along with the stirring speed
studies suggested that the reaction in the absence of P3-
NO occurs at the liquid-liquid interface.
In the presence of P3NO the above results indicated
that the reaction rate was dependent on the level of
HOCl. Thus, the oxidation of catalyst must be the
turnover-limiting step, as it is the only step in the cycle
in which hypochlorite is involved. These data are
consistent with the reaction orders in indene and catalyst
discussed above. Since the oxidation of catalyst is
turnover limiting, the amount of catalyst present should
affect the rate, and we find an order in catalyst close to
unity. On the other hand, the epoxidation of indene is
not the rate-limiting step, so we would expect a zeroth-
order dependence on indene, as observed. This hypoth-
esis is also consistent with results reported by J acobsen,
who found that the epoxidation of styrene using manga-
nese salen catalysts could be carried out at -78 °C when
an organic oxidant such as m-CPBA is used.16 In this
case, oxidation in the homogeneous organic solution is
fast and the epoxidation of the alkene must also be rapid
since the reactions readily occur at -78 °C. Thus, if the
oxidation of the catalyst is rapid, then the overall
epoxidation can occur at low temperature. If the oxida-
tion of catalyst is slow, as in the two-phase system with
NaOCl as oxidant, then the overall catalytic cycle is slow
and requires temperatures near 0 °C for a suitable
reaction rate. Our kinetic results, however, provide no
information regarding the oxametallocycle pathway.
Since the limiting step in the catalytic cycle is catalyst
oxidation, the kinetics provide no information on the
other steps in the cycle.
Effect of Stir r in g Sp eed . Since the epoxidation is a
two-phase reaction, possible interfacial reactions were
probed by variation of the agitation speed. In the first
set of experiments, shown in Figure 6a, the epoxidation
in the presence of 0.75 mol % catalyst and 3 mol % P3-
NO (both vs indene) was conducted at speeds of 420, 860,
and 2180 rpm. In the runs shown in previous figures,
the agitation speed had always been 860 rpm, so we chose
a speed above and below our standard conditions.
The results shown in Figure 6a reveal that the stirring
speed has only a small effect on the rate of epoxidation
in the presence of P3NO. Moderately faster epoxidation
at higher stirring speed indicated some reaction at the
liquid-liquid interface. At 860 rpm the rates of both the
epoxidation and the catalyst decomposition were es-
sentially unchanged when Aliquat 336 (tricaprylmethyl-
ammonium chloride) (4 mol/mol of Mn salen catalyst) was
included in the reaction mixture (not shown). This result
ruled out rate-determining phase transfer of OCl- and
confirmed that HOCl reacts in the slow step of the
epoxidation rather than OCl-.
In runs with P3NO, the rate at which catalyst disap-
peared was measurably independent of the stirring
speed, indicating that the catalyst was decomposing only
in the organic phase (Figure 6b).
The effect of stirring speed in runs without P3NO was
quite different. Strong dependence of the epoxidation
rate and the catalyst decomposition rate on the stirring
speed indicated the importance of both reactions at the
liquid-liquid interface, Figure 7. At the medium stirring
speed of 860 rpm, the rates of both reactions were
essentially unchanged when Aliquat 336 (4 mol/mol of
F igu r e 6. (a) Effect of stirring speed on the rate of indene
epoxidation in the presence of P3NO. Reactions run with 0.75
mol % catalyst and 3 mol % P3NO, both vs indene, at 860 rpm,
by method 2 as described in the Experimental Section. (b)
Effect of stirring speed on the rate of catalyst degradation in
the presence of P3NO.
Mn salen catalyst) was included in the reaction mixture.
Thus, rate-determining phase transfer of OCl- in either
reaction was ruled out.
These data suggest that the turnover-limiting step in
the epoxidation without P3NO is a liquid-liquid inter-
facial oxidation of the catalyst, while the reaction in the
presence of P3NO occurs mainly in the organic phase with
a small interfacial component.
Su m m a r y on th e Role of P 3NO for th e Ep oxid a -
tion . Several lines of evidence point toward the role of
P3NO in accelerating the indene epoxidation as being to
transport HOCl into the organic layer, where it oxidizes
Mn(III) to the catalytically active MnVdO in the rate-
determining step. The zeroth-order indene kinetics
indicate that indene is not involved in the turnover-
limiting step, and the dependence of the rate on hydrox-
ide and hypochlorite concentrations indicates that HOCl
(16) Palucki, M.; Pospisil, P. J .; Zhang, W.; J acobsen, E. N. J . Am.
Chem. Soc. 1994, 116, 9333-9334.

J acobsen Asymmetric Epoxidation of Indene
J . Org. Chem., Vol. 62, No. 7, 1997 2227
the active oxidant. These data suggest that the oxidation
of the catalyst in the absence of P3NO is occurring at the
interface with NaOCl as oxidant.
Titr a tion of HOCl in th e Or ga n ic La yer . Since the
kinetic experiments above suggested that P3NO was
transporting HOCl into the organic layer, we sought to
quantitate the oxidant level in the organic layer by redox
titration using thiosulfate/iodine. These experiments
were run in the same manner as the kinetic experiments,
except that the catalyst was omitted so that no epoxida-
tion would take place. In the absence of P3NO, no
oxidant (<10-5 M) could be measured in the organic layer
after separation of the aqueous layer and centrifugation
to remove all second-phase water. This is not surprising
since the pKa of HOCl is 7.5, and only very small amounts
of HOCl would be expected to be found from an aqueous
solution that contains 0.13 M NaOH even if the distribu-
tion coefficient for the organic layer was large. On the
other hand, in the presence of the standard level of P3-
NO (3% vs indene), the quantity of oxidant in the organic
phase was quantifiable at (7 ( 1) × 10-5 M. Using one-
quarter of the level of P3NO (0.75% vs indene), the
quantity of oxidant was too low to accurately measure.
Control experiments showed that P3NO itself was not
titrated by the thiosulfate/iodine system. These experi-
ments quantified the total amount of oxidant that P3NO
transports into the organic phase but provided no infor-
mation on whether this oxidant is HOCl or OCl-.
However, the above kinetic experiments suggested that
HOCl was the active oxidant in the indene epoxidation.
Kin etic Mea su r em en ts Usin g a Red ox In d ica tor .
Since the level of oxidant measured by titration was small
and there was no information as to the ionization of the
oxidant (HOCl or ClO-), a redox indicator was used to
monitor the level of oxidant drawn into the organic phase
by P3NO. The indicator used was diphenylbenzidine,
which is colorless in the reduced state but is oxidized to
the violet-colored quinoid structure (eq 6). The oxidation
was followed by monitoring the increase in absorbance
at 437 nm.
F igu r e 7. Effect of stirring speed on the rate of (a) indene
epoxidation in the absence of P3NO and (b) catalyst degrada-
tion in the absence of P3NO.
is the active oxidant and that HOCl is involved in the
rate-limiting step. Additionally, since the rate is nearly
independent of stirring speed in the presence of P3NO,
the turnover-limiting step is occurring primarily in the
organic phase.17
In the absence of P3NO, the epoxidation exhibits quite
different behavior. The reactions are now dependent on
the rate of agitation, indicating that the turnover de-
pendent step is occurring at the interface. Moreover, the
epoxidation is no longer dependent on hydroxide or
hypochlorite concentrations, so HOCl appears not to be
Experiments with this indicator were not definitive,
but the results supported the argument that P3NO
transports HOCl to the organic phase. A number of
probe experiments were conducted to establish conditions
in which the oxidation was reproducible. Initial experi-
ments carried out in aqueous NaOCl/PhCl or aqueous
NaOCl/PhCl/indene were difficult to interpret because
the oxidized form of the indicator was not stable in these
media, as the color deepened initially and then faded.
The best system found was aqueous NaOCl/dichlo-
romethane, wherein the oxidized form of the indicator
was stable long enough for valid kinetic measurements
to be obtained. Previous work with the epoxidation of
indene has shown that the epoxidation characteristics are
very similar in PhCl and dichloromethane. The experi-
ments described below were carried out at -5 °C with
the indicator level at 1 mM in dichloromethane, which
(17) A reviewer has suggested that a synergistic effect between P3-
NO and HOCl may accelerate the catalyst oxidation; that is, complex-
ation of P3NO with catalyst, as in 3, may result in the catalyst being
more readily oxidized. On the other hand, as the structure 3 suggests,
the oxygen from P3NO may be transferred to the catalyst aided by
HOCl as acid catalyst. However, experiments using 4-(3-phenylpropyl)-
pyridine14 indicated that the pyridine cannot be oxidized to the N-oxide
under the reaction conditions, so it is unlikely that a catalytic cycle
exists involving the pyridine and its N-oxide.

2228 J . Org. Chem., Vol. 62, No. 7, 1997
Hughes et al.
F igu r e 10. Effect of NaOH concentration on oxidation of
diphenylbenzidine, with and without P3NO.
F igu r e 8. Effect of P3NO levels on the rate of diphenylben-
zidine oxidation in dichloromethane/aqueous NaOCl.
result again opposite to what was observed in the
epoxidation and consistent with the oxidation occurring
by an interfacial mechanism.
Experiments conducted at a slower agitation speed of
420 rpm were more consistent with the results observed
with the epoxidation. Displayed in Figure 10 are results
for experiments run at two concentrations of hydroxide,
0.125 and 0.25 M. Without P3NO present, there was no
difference in oxidation rate between the two levels of
hydroxide. However, with P3NO, the rate was slowed as
the concentration of hydroxide increased, a result in
concert with the findings for the epoxidation. Moreover,
reducing the concentration of NaOCl by one-half, but
keeping the hydroxide constant, also reduced the rate by
one-half, which is consistent with eq 5, where the
concentration of HOCl is proportional to the hypochlorite
concentration. Thus, the experiments at 420 rpm are
consistent with P3NO transporting HOCl to the organic
layer, with oxidation of diphenylbenzidine occurring in
the organic layer. However, these experiments also
indicate that P3NO serves as a surfactant, and at higher
stirring speeds, it is this property that overwhelms the
HOCl transportation in controlling the oxidation.
F igu r e 9. Effect of stirring speed on the rate of diphenylben-
zidine oxidation in the presence of 0.1 M P3NO in dichloro-
methane/aqueous NaOCl.
Con clu sion s
is close to the solubility limit. The first set of experiments
at the intermediate stirring speed of 860 rpm probed the
effect of P3NO on the rate of oxidation. As illustrated in
Figure 8, P3NO accelerated the oxidation, although not
in a proportional fashion, since 5 and 25 mM solutions
provided nearly the same level of rate increase. However,
P3NO was clearly capable of accelerating the oxidation
reaction. How it accelerates the oxidation is more
complicated. As shown in Figure 9, for the reactions at
0.1 M P3NO, the rates were faster at higher stirring
speed. This is opposite to what was observed in the
epoxidation and suggests that the oxidation of diphenyl-
benzidine is occurring at least partially by an interfacial
mechanism. Also supporting this conclusion were the
results obtained with the phase transfer agent and
surfactant Aliquat 336. Addition of Aliquat 336 to the
reaction mixture also increased the rate of oxidation, a
We have identified an unexpected role of an N-oxide
in the epoxidation of indene in the two-phase system of
aqueous NaOCl and chlorobenzene, which is to transport
HOCl into the organic phase, where HOCl oxidizes MnIII
to MnVdO in the turnover-limiting step. This conclusion
is supported by kinetic experiments on the epoxidation,
which show that the epoxidation rate is not dependent
on indene but is dependent on HOCl concentration, by
titrations of the oxidant level in the organic layer and
by the kinetics of oxidation of the redox indicator,
diphenylbenzidine, that closely resemble the kinetics of
the epoxidation. The transfer of HOCl to the organic
phase may involve a hydrogen-bonded complex 4. P3NO
itself crystallizes as a hydrate 5, which provides some
support for P3NO as a hydrogen bond acceptor. The

J acobsen Asymmetric Epoxidation of Indene
J . Org. Chem., Vol. 62, No. 7, 1997 2229
°C; then the precooled organic mixture was added to initiate
the reaction. The slight exotherm was compensated for by
having the organic mixture at -10 °C before adding, such that
the reaction temperature after addition was very close to -5
°C. Using this procedure, P3NO was supersaturated in the
organic layer at the reaction temperature of -5 °C but
remained in solution for the duration of the run. Aliquots,
about 0.5 mL, of the mixtures of the two liquid phases were
removed periodically with the stirrer running and were
centrifuged. Control experiments indicated that the reaction
stopped once the aliquots were pulled, since the layers were
no longer mixing. The upper organic layer was diluted 2000-
fold with acetonitrile for reverse phase HPLC analysis and
100-fold with dichloromethane for the normal phase HPLC
analysis.
N-oxide also plays other roles in the epoxidation, includ-
ing stabilization of the catalyst and as a surfactant.
Exp er im en ta l Section
Gen er a l. NaOCl (1.5 M) was purchased from Spectrum
Chemicals, chlorobenzene and diphenylbenzidine were sourced
from Aldrich, and the Mn(salen)Cl catalyst was provided by
Sepracor. 4-(3-Phenylpropyl)pyridine N-oxide was prepared
by Oxone oxidation of the corresponding pyridine,14 which was
available from Lancaster.
Meth od 2 Usin g 4 equ iv of P 3NO vs Ca ta lyst. Aqueous
1.64 M NaOCl (60 mL, 98 mmol, 0.13 M in NaOH) and P3NO
(274 mg, 1.28 mmol) were charged to the vessel at ambient
temperature and cooled to -5 °C. The Mn salen catalyst (204
mg, 0.32 mmol) was dissolved in chlorobenzene (7.5 mL), cooled
to -5 °C, and then added to the reaction vessel. Then, indene
(5.0 mL, 40.7 mmol), cooled to -5 °C, was added to the mixture
to initiate the reaction. Sampling and assaying were done as
described above. In this procedure, only 25% of the P3NO is
dissolved at the beginning of the run, but the percentage slowly
increases as the reaction proceeds since it is more soluble in
the product mixture than in the initial reaction mixture, as
determined by control experiments.
Hyp och lor ite Titr a tion s. Hypochlorite was titrated io-
dometrically. Chlorobenzene or chlorobenzene/indene (3:2,
v:v), 15 mL, was stirred in the epoxidation vessel described
above with 40 mL of 1.5 M aqueous NaOCl at room temper-
ature. In one run, 274 mg of P3NO was included, while none
was added in another. In a control run, just chlorobenzene
and P3NO were combined, leaving out the NaOCl. After 15
min, the stirrer was stopped and the layers were allowed to
separate. The organic layer was then centrifuged to further
remove any second-phase water and then titrated. For the
titrations, 5 mL of HOAc, 5 mL of water, and 5 mL of 10%
aqueous potassium iodide were mixed and titrated using 0.001
M sodium thiosulfate. Then three 5-mL aliquots of the organic
layer were added and titrated in succession, all in the same
vessel. All operations were carried out under a nitrogen
atmosphere to avoid autooxidation of KI to iodine.
A normal phase HPLC assay for the Mn salen catalyst was
developed using a Whatman Partisil 10 PAC column with a
gradient elution from 10% ethanol in isooctane to 30% ethanol
over 20 min, with a flow rate of 2 mL/min. MnIIIsalen eluted
at 8.5 min, while P3NO eluted at 12 min. A diode array
detector provided multiwavelength detection. A guard column
of the same packing was found to be essential, as reaction
samples fouled the column after 15-20 injections causing split
peaks for the catalyst. With fresh packing in the guard
column, the catalyst eluted as a fairly sharp peak, and
quantitation of catalyst in reaction mixtures was readily
accomplished.
Assays for indene, indene oxide, and P3NO were accom-
plished by reverse phase HPLC using a Zorbax RX-C8 column,
isocratic elution of 60% acetonitrile/40% 0.01 M aqueous KH2-
PO4, a flow rate of 1 mL/min, and detection at 220 nm. The
asymmetric induction in indene oxide was determined by
HPLC with a Chiracel-OB column with a mobile phase of 97%
hexane/3% isopropyl alcohol, a flow rate of 1 mL/min, and
detection at 254 nm.
Gen er a l P r oced u r e for Ep oxid a tion Kin etic Exp er i-
m en ts. Epoxidations were performed under nitrogen atmo-
sphere at -5 °C on a 75-mL scale using a jacketed vessel
equipped with a mechanical stirrer, whose stirring speed was
calibrated using a tachometer. Ethylene glycol-water at -7
°C circulating from a constant temperature bath controlled the
reaction temperature at -5 °C. For standard epoxidation
runs, reaction mixtures were composed in two ways. In one
all the organic components were mixed and dissolved at room
temperature; in the other P3NO was added as a solid to the
cold aqueous NaOCl.
Meth od 1 Usin g 4 m ol of P 3NO vs Ca ta lyst. P3NO (274
mg, 1.28 mmol), Mn salen catalyst (204 mg, 0.32 mmol), indene
(5.0 mL, 40.7 mmol), and chlorobenzene (7.5 mL) were
combined at room temperature, mixed to dissolve P3NO and
catalyst, and then cooled to -10 °C in the constant tempera-
ture bath. Aqueous 1.64 M NaOCl (60 mL, 98 mmol, 0.13 M
in NaOH) was added to the reaction vessel and cooled to -5
Kin etic Exp er im en ts w ith Dip h en ylben zid in e. Runs
were performed using the epoxidation vessel described above
starting with 60 mL of aqueous 1.5 M NaOCl and adding 15
mL of dichloromethane containing 1 mM of diphenylbenzidine,
at -5 °C. P3NO was added in various amounts. Appearance
of the oxidized form of the indicator was followed by visible
spectroscopy (λmax 445 nm) of samples diluted from 1- to 10-
fold, depending on the extent of oxidation.
J O961735I