Mechanistic Studies on the Mukaiyama Epoxidation

Article abstract of DOI:10.1021/jo030345a

A detailed mechanistic study on the Mukaiyama epoxidation of limonene with oxygen as oxidant, bis(acetylacetonato)nickel(II) as catalyst, and an aldehyde as co-reagent is reported. All major products of the reaction have been quantitatively identified, both with isobutyraldehyde and 2-methylundecanal as co-reacting aldehydes. Limonene epoxide is formed in good yield. The main products evolving from the aldehyde are carboxylic acid, CO₂, CO, and lower molecular weight ketone and alcohol (K + A). A mechanism is proposed in which an acylperoxy radical formed by the autoxidation of the aldehyde is the epoxidizing species. The observation of carbon dioxide and (K + A) in a 1: 1 molar ratio supports this mechanism. CO₂ and (K + A) are formed in molar amounts of 50-60% with respect to the amount of epoxide produced, indicating that epoxidation takes place not only via acylperoxy radicals but also via a peracid route. Cyclohexene epoxidation was also investigated with a number of different metal complexes as catalysts. Cyclohexene is very sensitive for allylic oxidation, which provides information about the action of the catalyst, e.g., metals that form strongly oxidizing stable high-valence complexes are more likely to induce allylic oxidation. Color changes in the reaction mixture indicate the presence of such high-valence species. In the case of nickel, it was found that low-valence compounds predominate during the reaction, which is in line with the fact that this metal displays the highest selectivity for epoxide. A mechanism that accounts for the observations is presented.

Full text of DOI:10.1021/jo030345a

Mech a n istic Stu d ies on th e Mu k a iya m a Ep oxid a tion
†
‡
,†
†
Bastienne B. Wentzel, Paul L. Alsters, Martinus C. Feiters,* and Roeland J . M. Nolte
Department of Organic Chemistry, University of Nijmegen, Toernooiveld 1, 6525 ED Nijmegen,
The Netherlands, and DSM Pharma Chemicals, Advanced Synthesis, Catalysis, and Development,
P.O. Box 18, 6160 MD Geleen, The Netherlands
mcf@sci.kun.nl
Received November 10, 2003
A detailed mechanistic study on the Mukaiyama epoxidation of limonene with dioxygen as oxidant,
bis(acetylacetonato)nickel(II) as catalyst, and an aldehyde as co-reagent is reported. All major
products of the reaction have been quantitatively identified, both with isobutyraldehyde and
2
-methylundecanal as co-reacting aldehydes. Limonene epoxide is formed in good yield. The main
products evolving from the aldehyde are carboxylic acid, CO , CO, and lower molecular weight
2
ketone and alcohol (K + A). A mechanism is proposed in which an acylperoxy radical formed by
the autoxidation of the aldehyde is the epoxidizing species. The observation of carbon dioxide and
(
K + A) in a 1:1 molar ratio supports this mechanism. CO and (K + A) are formed in molar amounts
2
of 50-60% with respect to the amount of epoxide produced, indicating that epoxidation takes place
not only via acylperoxy radicals but also via a peracid route. Cyclohexene epoxidation was also
investigated with a number of different metal complexes as catalysts. Cyclohexene is very sensitive
for allylic oxidation, which provides information about the action of the catalyst, e.g., metals that
form strongly oxidizing stable high-valence complexes are more likely to induce allylic oxidation.
Color changes in the reaction mixture indicate the presence of such high-valence species. In the
case of nickel, it was found that low-valence compounds predominate during the reaction, which is
in line with the fact that this metal displays the highest selectivity for epoxide. A mechanism that
accounts for the observations is presented.
In tr od u ction
referred to as the Mukaiyama epoxidation or Mukaiyama-
Yamada epoxidation. The method is very mild; usually
the reactions are performed at room temperature and
often display epoxide selectivities up to 100% at full
conversion of the alkene. Furthermore, the product
formation is complete in a few hours, whereas the
uncatalyzed reaction takes a day or more to reach the
maximum conversion. These characteristics make the
Mukaiyama method attractive for industrial applications,
despite the fact that more than stoichiometric amounts
of the co-reacting aldehyde are needed.
The aerobic epoxidation of alkenes with an aldehyde
as a co-reagent is an efficient and useful method for the
production of fine chemicals. In general, there is a
distinction between methods that do not use transition
metal catalysts (see, for example, Kaneda and Lassila )
and the more widely explored systems that make use of
1
2
such a catalyst. The latter method was investigated in
detail by Mukaiyama et al.3
-18
and is therefore often
†
‡
University of Nijmegen.
DSM Pharma Chemicals.
The uncatalyzed epoxidation of alkenes using molec-
ular oxygen with co-oxidation of an aldehyde appears to
proceed via a mechanism related to aldehyde autoxida-
tion (Scheme 1, eqs 1 and 2). The acylperoxy radical
formed in eq 2 has been shown2 to be the epoxidizing
species (eq 3, 4a). It was suggested that the carbon
(1) Kaneda, K.; Haruna, S.; Imanaka, T.; Hamamoto, M.; Nishiyama,
Y.; Ishii, Y. Tetrahedron Lett. 1992, 33, 6827-6830.
(2) Lassila, K. R.; Waller, F. J .; Werkheiser, S. E.; Wressell, A. L.
Tetrahedron Lett. 1994, 35, 8077-8080.
(
3) Mukaiyama, T. Aldrichimica Acta 1996, 29, 59-76.
,19
(4) Yamada, T.; Takai, T.; Rhode, O.; Mukaiyama, T. Bull. Chem.
Soc. J pn. 1991, 64, 2109-2117.
5) Yamada, T.; Takahashi, K.; Kato, K.; Takai, T.; Inoki, S.;
Mukaiyama, T. Chem. Lett. 1991, 641-644.
6) Hata, E.; Takai, T.; Yamada, T.; Mukaiyama, T. Chem. Lett.
994, 535-538.
7) Yamada, T.; Takai, T.; Rhode, O.; Mukaiyama, T. Chem. Lett.
991, 1-4.
8) Takai, T.; Yamada, T.; Rhode, O.; Mukaiyama, T. Chem. Lett.
991, 281-284.
9) Takai, T.; Hata, E.; Yamada, T.; Mukaiyama, T. Bull. Chem. Soc.
J pn. 1991, 64, 2513-2518.
10) Yamada, T.; Imagawa, K.; Nagata, T.; Mukaiyama, T. Bull.
Chem. Soc. J pn. 1994, 67, 2248-2256.
11) Yamada, T.; Imagawa, K.; Mukaiyama, T. Chem. Lett. 1992,
109-2112.
12) Yamada, T.; Imagawa, K.; Nagata, T.; Mukaiyama, T. Chem.
Lett. 1992, 2231-2234.
(
(
(13) Nagata, T.; Imagawa, K.; Yamada, T.; Mukaiyama, T. Bull.
Chem. Soc. J pn. 1995, 68, 1455-1465.
1
1
1
(
(14) Nagata, T.; Imagawa, K.; Yamada, T.; Mukaiyama, T. Bull.
Chem. Soc. J pn. 1995, 68, 3241-3246.
(
(15) Mukaiyama, T.; Yamada, T.; Nagata, T.; Imagawa, K. Chem.
Lett. 1993, 327-330.
(
(16) Imagawa, K.; Nagata, T.; Yamada, T.; Mukaiyama, T. Chem.
Lett. 1994, 527-530.
(
(17) Inoki, S.; Takai, T.; Yamada, T.; Mukaiyama, T. Chem. Lett.
1991, 941-944.
(
(18) Yanai, K.; Irie, R.; Ito, Y.; Katsuki, T. Mem. Fac. Sci., Kyushu
Univ. 1992, 18, 213-222.
2
(
(19) Valov, P. I.; Blyumberg, E. A.; Emanuel, N. M. Izv. Akad. Nauk
SSSR, Ser. Khim. 1966, 1334-1339.
1
0.1021/jo030345a CCC: $27.50 © 2004 American Chemical Society
Published on Web 04/20/2004
J . Org. Chem. 2004, 69, 3453-3464
3453

Wentzel et al.
SCHEME 1
SCHEME 2
SCHEME 3
ic acid in a nonradical epoxidation pathway. Peracid may
also react with aldehyde, yielding two molecules of
carboxylic acid (eq 12); this reaction consumes aldehyde,
involves peracid as an intermediate, and produces car-
boxylic acid, but does not yield the desired epoxide.
Furthermore, the acyl radical generated during the
autoxidation steps (eqs 1 and 10) may decarbonylate in
an endothermic reaction to form an alkyl radical and
carbon monoxide (eq 13). All these reactions consume
aldehyde without generating epoxide via the radical
epoxidation pathway, and therefore do not generate
dioxide that is found as a side product originates from
the unstable carboxyl radical generated during the ep-
oxidation. This radical is so unstable that it decarboxyl-
ates before it can abstract a hydrogen atom, forming CO
and an alkyl radical (path 4b). The latter radical is
subsequently oxidized by O to give an alkyl peroxyradi-
2
2
cal, which abstracts a hydrogen atom and forms an
alkylhydroperoxide (Scheme 2, eqs 6 and 7). Alterna-
tively, two alkylperoxy radicals may combine to form a
tetroxide. This compound rearranges via a so-called
Russell termination20 to generate a ketone and an alcohol
2
ketone and alcohol (K + A) and CO .
The transition-metal-catalyzed epoxidation of alkenes
using molecular oxygen and isobutyraldehyde as a co-
reagent (the Mukaiyama epoxidation) was studied in
detail by Nam et al.21 As substrates, limonene, stilbene,
styrene, and cyclohexene were used, and as catalysts
several cyclam and porphyrin complexes of, e.g., nickel-
(II), cobalt(II), manganese(III), and iron(III) were tested.
It was proposed that autoxidation of aldehyde plays an
important role in this metal-catalyzed reaction, just as
it does in the uncatalyzed oxidation of alkene and
aldehyde with molecular oxygen. On the basis of results
of cis-stilbene epoxidation22 it was concluded that the
predominant oxidizing species is an acylperoxyl radical
and not a peroxy acid; a mixture of cis- and trans-stilbene
epoxides, with the latter as the major product, was
(
2
K + A), provided that R-hydrogens are present (Scheme
, eqs 8 and 9). If no R-hydrogens are available, the
alkylperoxy radical abstracts a hydrogen and forms an
alkylhydroperoxide. Lassila found t-butylhydroperoxide
and CO in a 1:1 ratio as the products of the epoxidation
2
of diisobutylene with pivaldehyde (R ) tert-butyl) as co-
reagent, supporting eqs 4-7. The epoxidation-decarboxyl-
ation reaction is likely to proceed concertedly (eq 5) as
Lassila has shown.2
In addition to epoxidation, the acylperoxy radical that
is formed in eq 2 may abstract a hydrogen atom from
another aldehyde molecule to give a peroxy acid and an
acyl radical, thus propagating the autoxidation chain
(Scheme 3, eq 10). Peroxy acid (or peracid for short) is a
competing epoxidizing agent (eq 11), generating carboxyl-
(21) Nam, W.; Kim, H. J .; Kim, S. H.; Ho, R. Y. N.; Valentine, J . S.
Inorg. Chem. 1996, 35, 1045-1049.
(22) Wentzel, B. B.; Gosling, P. A.; Feiters, M. C.; Nolte, R. J . M. J .
Chem. Soc., Dalton Trans. 1998, 2241-2246.
(
20) Sheldon, R. A.; Kochi, J . K. Metal-Catalyzed Oxidations of
Organic Compounds; Academic Press: New York, 1981.
3
454 J . Org. Chem., Vol. 69, No. 10, 2004

Mukaiyama Epoxidation
SCHEME 4
CHART 1
obtained, indicating that a radical in which there is free
rotation around the bond between the reacting carbons
(as in eq 3) is the main intermediate. Acylperoxy radicals
are known to preferentially react with the double bonds
of alkenes yielding epoxides (Scheme 4, eq 14), whereas
hydroxy and alkylperoxy radicals tend to abstract allylic
hydrogens giving allylic oxidation products (Scheme 4,
eq 15). A good substrate to investigate whether the
oxidizing species has a preference for allylic oxidation or
epoxidation is cyclohexene; this molecule is known to be
very sensitive to allylic oxidation. Using cyclohexene as
a substrate, Nam et al. found epoxide as the predominant
product. The product distributions appeared not to
depend on the type of metal complex that was used as
the catalyst.21 It was concluded that the only role of the
metal complex was the stabilization the acylperoxy
radical. Unfortunately, the products evolving from the
aldehyde were not isolated as was done for the uncata-
SCHEME 5
catalysts. Limonene epoxidation is of interest as the first
step in a new industrial route for the manufacture of
carvone [2-methyl-5-(1-methylethenyl)-2-cyclohexen-1-
one], an important spearmint flavor compound, and was
therefore investigated as the main substrate in this
study. As a co-reacting aldehyde in the Mukaiyama
epoxidation, isobutyraldehyde 4 is widely applied. How-
ever, since it was expected that low molecular weight
ketone and alcohol (isopropyl alcohol and acetone in this
case, Scheme 2, eq 9), which are volatile and not easily
analyzed quantitatively, will evolve from the reaction, we
chose to use a higher molecular weight aldehyde as well
for our studies. The results obtained with this aldehyde
will be compared to those obtained with isobutyralde-
hyde.
lyzed co-oxidation of the alkene and aldehyde by Lassila
et al.2 It was assumed that carboxylic acid, formed
through hydrogen abstraction by the carboxyl radical
from either the substrate or the aldehyde, was the
exclusive product.
In an earlier study,22 we investigated the scope and
mechanism of the Mukaiyama epoxidation. We provided
new evidence for the radical nature of the reaction and
proposed a tentative mechanism in which the nickel
catalyst may serve to stabilize the epoxidizing species,
i.e., the acylperoxy radical.
Mizuno et al.23 also investigated the Mukaiyama
epoxidation using three different polyoxometalates as
catalysts, isobutyraldehyde as co-reagent, and cyclohex-
ene as substrate. In contrast to Nam et al.,21 they noted
the formation of allylic oxidation products, i.e., 2-cyclo-
hexen-1-ol and 2-cyclohexen-1-one, and observed that the
uncatalyzed co-oxidation of cyclohexene and isobutyral-
dehyde gave a very high selectivity ratio (SR) of epoxide
to allylic oxidation product, albeit in a very slow reaction.
On the other hand, the catalyzed reaction gave lower
selectivity ratios but was much faster. The SR was found
not to vary among the three different catalysts.
Furthermore, we felt that in addition to Mizuno’s
23
studies much can be learned from a detailed study of
the use of various transition metal catalysts in the
Mukaiyama epoxidation. Preliminary results from our
group had indicated that there is indeed a difference in
selectivity ratio (SR) between different metal catalysts
in the Mukaiyama epoxidation of cyclohexene, in contrast
to Mizuno’s conclusions.23 This would suggest that the
metal catalyst has a more complex role than just stabiliz-
ing the oxidizing species in the reaction as Nam et al.
concluded.21 For these studies cyclohexene instead of
limonene was used as the substrate, since the former
compound in contrast to the latter is more sensitive to
allylic oxidation (vide infra).
Despite the interesting results of the study of Lassila
2
et al. on the uncatalyzed reaction, no comparable studies
have been published for the metal-catalyzed epoxidation
of alkenes using molecular oxygen as oxidant and an
aldehyde as co-reagent. With this paper, we intend to fill
this gap by presenting a quantitative study of all the
products evolving from the Mukaiyama epoxidation. We
were particularly interested in the aerobic epoxidation
of S-limonene (Scheme 5, S-1), under Mukaiyama’s
conditions using nickel(II) â-diketonate complexes (3) as
Exp er im en ta l Section
Ma ter ia ls. Dipentene (RS-limonene, [1-methyl-4-(1-meth-
ylethenyl)cyclohexene]), toluene, and cyclohexene were used
as received; dichloromethane, isobutyraldehyde, and 2-methyl-
undecanal were distilled before use. Bis(pentane-2,4-dionato)-
2
cobalt(II) (Co(acac) ), bis(pentane-2,4-dionato)nickel(II) (Ni-
(23) Mizuno, N.; Weiner, H.; Finke, R. G. J . Mol. Catal. A: Chem.
1
996, 114, 15-28.
(acac)
2
, 3a ), and bis(3-(p-tert-butylbenzyl)pentane-2,4-dionato)-
J . Org. Chem, Vol. 69, No. 10, 2004 3455

Wentzel et al.
nickel(II) (3b) were prepared as described earlier24 and dried
by azeotropic distillation with toluene. All other metal salts
and complexes were commercial samples and were used as
received. Oxygen and nitrogen gases were used as received.
In str u m en ta tion . The GC analyses were performed with
a fused-silica capillary column (15 m length, 35 µm i.d., d
f
)
1
.0) containing a FFAP stationary phase. GC analyses of the
cyclohexene oxidation products were performed with a Chrom-
pack fused silica CP-Sil 5CB column (25 m length, 32 µm i.d.,
d ) 1.2 µm). GC-MS analyses were performed with a WCOT
f
fused silica capillary column (25 m length, 25 µm i.d., d ) 0.2
f
µm) containing a FFAP stationary phase (CP-Wax 58) using
an ion-trap MS detector.
The epoxidation reactions were carried out in a Premex
autoclave reactor with Hastelloy C276 wet parts, equipped
with a HC276 Dean-Stark water separator, a 4-blade stirrer
(max 1500 rpm), a sintered HC276 gas inlet (5 µm frit), and a
sampling tube. The temperature was regulated with a Premex
C-M2 control unit to (0.1°C. The nitrogen and oxygen gas inlet
was regulated by mass-flow controllers (MFCs) and could be
controlled to (0.1% v/v O
-80 °C at atmospheric pressure) to condense any vapor that
was present. The gas was then analyzed for carbon dioxide
and carbon monoxide with a Maihak Multor 610 CO /CO
analyzer (IR detection) and for oxygen content with a Ser-
vomex 570A O analyzer.
2 2
in N . The exhaust gas was cooled
(
2
2
Ep oxid a tion Ru n s. A standard epoxidation run was
performed as follows. Approximately 60 mmol of dipentene (a
mixture of R- and S-limonene, 1) was accurately weighed and
dissolved in 75 or 150 mL of toluene. This solution was loaded
into the autoclave, and the Dean-Stark cooler was filled with
toluene. This mixture was equilibrated for at least 1 h at 25.0
F IGURE 1. Limonene epoxidation with oxygen and 2-methyl-
°
C under an atmosphere of approximately 8% v/v O
total pressure 7 bar. No reaction was observed during this
period. The autoclave was opened, Ni(acac) and aldehyde 4
2 2
in N ,
undecanal catalyzed by Ni(acac)
limonene 1, 93.5 mmol aldehyde 5, 0.1 mol % Ni (acac)
in 75 mL of toluene, 7 bar of 8% O in N , 25 °C. Analyses by
GC. Top: (a) 2-methylundecanal (5), (b) O uptake, (c) li-
2
. Conditions: 32.8 mmol
II
2
(3b)
2
or 5 (corresponding to 0.1 mol % and 3 molar equiv with
respect to alkene respectively) were added quickly, and the
autoclave was closed and pressurized again. This point was
taken as t ) 0. During the reaction the pressure was kept at
2
2
2
monene epoxide (2), (d) limonene (1). Bottom: (e) 2-methyl-
undecanoic acid (8), (f) (ketone 10 + alcohol 9), (g) 2-unde-
canone (10), (h) 2-undecanol (9).
7
.0 ( 0.1 bar, the temperature was 25.0 ( 0.1 °C, and the
stirring rate was 1500 rpm. Samples were taken regularly and
analyzed with GC using 1,2,4-trichlorobenzene as the external
standard and 2-(tert-butyl)-4-methylphenol as a stabilizer for
the samples. The oxygen content of the exhaust gas was
registered, and the CO and CO contents were read on-line
2
with a personal computer. The experimental error (deviations
of the measuring apparatus) was much less than 1%.
The epoxidations of cyclohexene catalyzed by different metal
complexes were performed in a glass vessel at room temper-
ature. A 50 mL two-neck flask was loaded with 0.03 mmol of
the appropriate metal complex and subsequently equipped
with a balloon, a septum, and a 1 cm magnetic stirring bar.
The flask was thermostated with a water bath at 25°C. The
2-methylundecanal (5) to study the products that evolve
from the reaction. The choice for 5 was based on the high
boiling point of this compound and of the potential
oxidation products, which ensures that these products
do not evaporate during their analysis, allowing a
quantitative identification by GC and GC-MS. In the
epoxidation run shown in Figure 1, limonene was reacted
with 2.9 equiv of 2-methylundecanal 5 (R ) C
Scheme 5) in toluene in an autoclave; 0.1 mol % of Ni-
acac) (3a ) was used as the catalyst. In the upper panel
9
H19 in
(
2
of Figure 1, the consumptions of limonene 1 and aldehyde
5 are plotted, together with the formation of epoxide 2.
Furthermore, the consumption of O2 is plotted (an
increasing curve, not a decreasing one). In Figure 1 (lower
panel) the formation of 2-methylundecanoic acid (8) as
an oxidation product of aldehyde 5 is plotted. Moreover,
vessel was flushed with 100% O
solution containing cyclohexene (0.30 M), 1,3-dichlorobenzene
0.25 M, internal standard), and isobutyraldehyde (0.84 M,
redistilled quality) in dichloromethane was freshly prepared.
Of this solution, 10 mL was injected in the vessel at t ) 0.
The reaction was stirred at 1200 rpm. Samples were taken
regularly and analyzed with GC.
2
at least three times. A
(
2
-undecanol (9, alcohol, A) and 2-undecanone (10, ketone,
K) were detected in the reaction mixture by GC-MS
analysis (m/z ) 171 and 170). These products were
identified by comparison with authentic samples. Their
formation is also plotted in Figure 1 (lines g and h). When
the molar amounts of ketone and alcohol (K + A) are
added, line f in Figure 1 is obtained. Small amounts of
Resu lts
Ep oxid a tion of Lim on en e u sin g 2-Meth ylu n d e-
ca n a l a s Co-Rea gen t. As a co-reagent in the Mu-
kaiyama epoxidation an R-branched aliphatic aldehyde
is most suitable. Although isobutyraldehyde (4) is com-
monly employed in this reaction (vide infra), we used
epoxides 6 and 7 were also found (not shown in this
1
figure). In this particular experiment, CO
2
formation was
not recorded although it is expected to be occur on the
basis of Scheme 1 and Scheme 2.
(
24) Gosling, P. A. Catalyst Systems for the Epoxidation of Alkenes
by Molecular Oxygen; University of Nijmegen: Nijmegen, 1996.
3
456 J . Org. Chem., Vol. 69, No. 10, 2004

Mukaiyama Epoxidation
CHART 2
After the reaction had been allowed to run for 130 min,
the conversion of limonene was 94% (30.9 mmol of the
initial 32.8 mmol had reacted), the consumption of
aldehyde being 56.9 mmol (note that this is approxi-
mately 2 equiv with respect to limonene). The selectivity
for epoxide 2 reached a maximum (74%) after 2 h of
reaction time. Thereafter, limonene epoxide was further
oxidized into diepoxide 7 and other unidentified oxidation
products. Formation of the diepoxide 7 started after
nearly all limonene had been converted into the mono-
epoxide 2, i.e., the maximum selectivity for monoepoxide
was obtained at >95% conversion of limonene. 2-Methyl-
undecanoic acid 8 was formed at approximately the same
rate as epoxide during at least the first 2 h of the reaction
F IGURE 2. Formation of epoxide (9) correlated with CO
formation (s) in the epoxidation of limonene with O and
2
2
2-methylundecanal catalyzed by Ni(acac) . Conditions: 18
2
mmol limonene, 22 mmol aldehyde, 75 mL of toluene, 0.3 mol
II
% Ni (acac)
2
, 7 bar of 8% O
2
in N
2
, 25 °C.
(
2
O
Figure 1, bottom); after 130 min, 19.5 mmol of 8 and
2.8 mmol of 2 had formed. At this point, 54.6 mmol of
had been consumed, i.e., approximately two times the
2
amount of consumed alkene. An important observation
was that in the first 2 h of the reaction (up to 80%
conversion), ketone and alcohol were formed in equimolar
amounts and that the cumulative amount of alcohol 9
and ketone 10 (K + A) was approximately one-half of the
amount of epoxide in these first 2 h.
F IGURE 3. Epoxidation of limonene with isobutyraldehyde
and oxygen catalyzed by Ni(acac) . Conditions: limonene (1)
(67 mmol), isobutyraldehyde (4) (184 mmol), 150 mL toluene,
0.1 mol % Ni(acac) (0.4 mM), 25.0 °C, 7 bar of 8% v/v O in
. Analysis by GC. (a) Oxygen uptake, (b) isobutyric acid, (c)
CO , (d) limonene epoxide, (e) limonene.
2
CO
present in slight excess with respect to limonene. A
solution of limonene in toluene was treated with 1.2 equiv
of 2-methylundecanal 5 and 0.3 mol % of Ni(acac) 3a in
2
formation was measured with methylundecanal
5
2
2
N
2
2
2
the autoclave under otherwise standard conditions (see
Experimental Section and Figure 2). After an induction
period that is usually observed at low aldehyde concen-
trations, limonene epoxide was formed at the same rate
mixture of R- and S-limonene 1) in toluene with 3 equiv
of isobutyraldehyde and 0.1 mol % of Ni(acac) at 7 bar
of 8% O in N (standard conditions, see Experimental
2
2
2
as CO
2
. At 76% conversion, 68% of limonene epoxide had
(8.6 mmol). At this
Section and Figure 3). As can be seen in Figure 3, the
conversion of limonene reaches 88% after 24 h with an
epoxide selectivity of 54%. The latter is at maximum after
been formed, as well as 62% of CO
2
point, the cumulative amount of 9 and 10 was 59% (8.1
mmol, not shown), i.e., under these conditions employing
a slight excess of aldehyde, approximately equimolar
amounts of epoxide, CO , and K + A (combined amounts
of alcohol plus ketone) are generated.
3
h (90% at 78% conversion of alkene) after which the
exocyclic double bond is oxidized to form 6 and further
oxidation of limonene monoepoxide is observed (i.e., 7 is
formed).
2
Ep oxid a tion u sin g i-Bu tyr a ld eh yd e a s Co-Re-
a gen t. Although 2-methylundecanal 5 is a good co-
reagent for studying the progress of the reaction and the
products of the Mukaiyama epoxidation of limonene, it
is not a very practical additive for use in industrial
processes, for which the inexpensive isobutyraldehyde 4
is preferred. Isobutyraldehyde has been widely studied
as a co-reagent in this epoxidation system, but only
limited quantitative information is available concerning
the fate of this aldehyde. We decided, therefore, to
perform some quantitative studies using dipentene (a
Because it was difficult to quantitatively determinate
the low molecular weight, volatile products formed from
isobutyraldehyde during the oxidation, the amount of
carbon dioxide and carbon monoxide present in the
exhaust gas were measured instead to follow the course
of the isobutyraldehyde conversion. From Scheme 1 and
Scheme 2 and the data presented above it can be
concluded that CO and K + A are formed in equimolar
2
amounts since they evolve from the same reaction step
(eq 4). CO is expected as a byproduct (Scheme 3), but at
J . Org. Chem, Vol. 69, No. 10, 2004 3457

Wentzel et al.
F IGURE 5. Reaction rate as a function of aldehyde concen-
F IGURE 4. Formation of CO
2
(symbols, left axis) and CO
tration in the epoxidation of limonene with O and isobutyral-
2
II
(drawn lines, right axis) as a function of time, at four different
dehyde catalyzed by Ni complexes. For conditions see Figure
temperatures: (a) 45.0, (b) 35.0, (c) 25.0, (d) 18.8 °C. For other
conditions see Figure 3 and Experimental Section.
3. (a) Catalyst 3a , rate of epoxide formation. (b) catalyst 3a ,
rate of CO formation, (c) catalyst 3b, rate of CO formation.
2
2
Solid lines: linear fits of the data.
the standard reaction temperature of 25 °C, no carbon
monoxide was detected in the exhaust gas. As can be
ligands and is not able to rearrange into a trimer.25,30 This
complex is square planar and diamagnetic and has a
purple color.
The rate of epoxidation of limonene as a function of
the isobutyraldehyde concentration catalyzed by the two
Ni-complexes was monitored by measuring the rate of
judged from Figure 3, the amount of CO
the first 3 h (up to 70% conversion of the reaction at 25
C) is approximately 60% of the amount of epoxide
formed. CO and CO formation was also determined at
three other temperatures, viz., at 18.8, 35, and 45.0 °C
Figure 4). The experiments were performed in duplicate,
2
formed during
°
2
2
CO formation. In the case of catalyst 3a , this rate was
(
also measured directly by determining the rate of epoxide
formation. In the analysis, the reaction rate was taken
as the slope of the linear part of the concentration-time
and the deviation was found to be less than (5%. CO
formation was negligible at 18.8 and 25.0 °C, but at
higher temperatures appreciable amounts of this gas
were formed (solid lines in Figure 4) as expected for an
endothermic decay of an intermediate acyl radical (see
Introduction). At 45 °C, decarbonylation had increased
to 4 mmol (about 10% of the amount of epoxide formed).
2
plot (the part where CO formation follows a zero-order
rate law), usually from ca. 5% to 80% conversion. With
this method, the dependence of the reaction rate on
aldehyde concentration was determined. In Figure 5, the
rate of the reaction is plotted against the concentration
of isobutyraldehyde. The concentrations of alkene and all
other parameters were kept constant, so that the amount
of aldehyde varied with respect to limonene from 1 molar
equiv (an aldehyde concentration of 0.4 M) to 10 molar
equiv (2.8 M). The lines (b) and (c) represent experiments
with respectively catalyst 3a and 3b, respectively. Line
The temperature dependence of the CO and CO
tion was found to be different. Whereas CO formation
increased rapidly with temperature, CO formation ceased
to increase on raising the temperature from 35.0 to 45.0
2
forma-
2
°
°
C, after an initial increase on going from 18.8 to 35.0
C. We presume that the rate of the epoxidation reaction
(a) shows the rate of epoxide formation in the experi-
(eqs 3-5) does increase with increasing temperature but
ments with 3a as a catalyst and should thus be compared
to line (b).
From Figure 5 it appears that for both catalysts the
reaction is first order in aldehyde concentration between
that the rate of formation of the oxidizing species, viz.,
the acylperoxy radical (eq 2), is reduced at these tem-
peratures due to premature decarbonylation of the acyl
radical (eq 13). We may conclude that 25 °C is the
optimum temperature for this reaction, because no side
reactions leading to CO are taking place.
1
and 4 molar equiv of this co-reagent with respect to
alkene. In this range, the linear fits afford a first-order
-
5
-1
rate constant of k1,3a ) (2.46 ( 0.12) × 10
s
for 3a
(Figure 5, b) and k1,3b ) (3.00 ( 0.15) × 10 s-1 for 3b
(Figure 5, c). Thus, the reaction with catalyst 3b is only
slightly faster than that with 3a . From the rate of epoxide
-5
E p oxid a t ion u sin g Differ en t Nick el(II) Com -
p lexes. Two different bis(acetylacetonato)nickel(II) cata-
lysts, 3a and 3b, were studied in the present Mukaiyama
epoxidation system to investigate the role of the structure
of the nickel complex on the reaction. These two nickel
complexes were chosen because they have different
(25) Addison, A. W.; Graddon, D. P. Aust. J . Chem. 1968, 21, 2003-
012.
2
(26) Fackler J r., J . P.; Cotton, F. A. J . Am. Chem. Soc. 1961, 83,
structures in toluene solution. Ni(acac)
2
(3a ) is known
3775-3778.
to be a trimer in noncoordinating solvents,2
5-28
(27) Cotton, F. A.; Fackler, J . P. J . Am. Chem. Soc. 1961, 83, 2818-
825.
with the
2
nickel center being octahedrally coordinated. The complex
has a green color and is paramagnetic with a magnetic
(
28) Bullen, G. J .; Mason, R.; Pauling, P. Inorg. Chem. 1965, 4, 456-
462.
(
29) Cotton, F. A.; Fackler, J . P. J . Am. Chem. Soc. 1960, 83, 2818-
moment of 3.27 µ
bulky and more electron-withdrawing acetylacetonato
at 27 °C.29 Nickel catalyst 3b has more
B
2
825.
(30) Cotton, F. A.; Wise, J . J . Inorg. Chem. 1966, 5, 1200-1207.
3
458 J . Org. Chem., Vol. 69, No. 10, 2004

Mukaiyama Epoxidation
limonene (60 mmol) and aldehyde (180 mmol) in 150 mL
of toluene, 0.1 mol % 3a , 8% O in N , 7 bar, 25 °C.) This
2
2
was found to have no influence on the formation of any
of the products or the rates of their formation. In an
earlier study22 we had found that, using a magnetically
stirred (1000 rpm) glass vessel, the reaction rate did
depend on the oxygen pressure between 0.2 and 1.0 bar.
This implies a diffusion-limited reaction. In the present
setup we did not observe such a dependence. This
discrepancy illustrates the importance of a good mixing
of the phases when the kinetics of a gas-liquid reaction
are studied.
The total pressure in the autoclave was varied between
2
and 15 bar, which again had no effect on the reaction
at all. The stirring rate was changed from 500 to 1500
rpm and the rate of epoxide formation was measured.
Going from 500 to 1000 rpm, the reaction rate increased,
but from 1000 to 1500 rpm, the rate remained constant.
We may conclude, therefore, that the epoxidation reaction
was not physically limited under the conditions we
applied to study the mechanism (1500 rpm; 7 bar of 8%
O in N ; 25 °C).
2
F IGURE 6. The rate of CO formation as a function of the
concentration of the nickel catalyst in the epoxidation of
limonene with O and isobutyraldehyde as co-reagent. For
2
conditions see Figure 3. (a) Catalyst 3a , (b) catalyst 3b.
formation the first-order rate constant of the 3a -catalyzed
-
5
reaction was calculated to be k1,3a ) (2.8 ( 0.2) × 10
(Figure 5, a), which is similar to that calculated from
the CO formation curve. We may conclude that the
catalysts differ in reaction rate at a given aldehyde
concentration, although the values of k are similar.
2
2
-
1
s
To establish the need for nickel(II) as an initiator in
the Mukaiyama epoxidation, a blank reaction was run
under standard conditions but in the absence of nickel
catalyst (limonene (60 mmol) and aldehyde (180 mmol)
in 150 mL of toluene, 0.1 mol % 3a , 8% O in N , 7 bar,
2
1
The rate of the reaction was also measured at different
catalyst concentrations varying from 0.01 to 0.5 mol %
with respect to alkene at otherwise standard conditions
2
2
25 °C). At ambient temperature this reaction did not
start. Only after addition of a radical initiator (m-CPBA,
0.25 mol % with respect to alkene), could a reaction be
detected. This reaction was much slower than the nickel-
catalyzed reaction, reaching a maximum of 86% alkene
conversion after 53 h with a selectivity for epoxide of 81%.
An uncatalyzed and uninitiated reaction could only be
established at elevated temperature (45 °C). (The reaction
under standard conditions afforded 25 mmol (42%) of
epoxide and 40 mmol of isobutyric acid after 4 h. The
induction period was ca. 50 min. When only isobutyral-
dehyde (180 mmol) and toluene (150 mL) were loaded
(Experimental Section). The results are shown in Figure
6
3
. The rates of the reactions catalyzed by both 3a and
b level off when the amount of nickel(II) becomes higher.
The reaction catalyzed by 3b is faster than that of 3a at
low catalyst concentrations, but the former catalyst
seems to become more easily deactivated at high concen-
trations than the latter (see Discussion).
Con tr ol Exp er im en ts. Mizuno et al.23 have pointed
out that it is extremely important in radical oxidations
to ensure that the reactions are not physically limited
(
(
i.e., by diffusion limitation) and that all compounds
substrates and products) are quantitatively identified.
into the autoclave and O was added, no reaction took
place; no oxygen was taken up and neither carboxylic acid
2
Furthermore, a blank reaction in the absence of a catalyst
should be carried out in order to establish the presence
or absence of an uncatalyzed oxidation process. These
control experiments were performed in the present study
and are described below.
nor CO were formed. When Ni(acac) was added to this
2
2
mixture, CO was formed in only very small amounts (ca.
2
4.5 mmol after 22 h), and isobutyric acid was formed (117
mmol) as the exclusive product. It is concluded, therefore,
that the uncatalyzed autoxidation of isobutyraldehyde is
too slow to be observed at room temperature.)
By comparing the masses of all incoming reagents
(
including O
after reaction (including CO
as expressed in the formula (M, mass in g) M(limonene)
M(toluene) + M(aldehyde) + M(Ni(acac) ) + M(O ) )
M(reaction mixture) + M(CO ) + M(CO), was deter-
2
) with the weight of the reaction mixture
Ep oxid a tion ver su s Allylic Oxid a tion . Mizuno et
al. noted recently23 that the uncatalyzed epoxidation of
the frequently studied substrate cyclohexene with isobu-
tyraldehyde and oxygen is almost as efficient as the
metal-catalyzed reaction (76-88% versus 88-94% yield
of epoxide under their reaction conditions of 1 bar, 100%
2
and CO), the mass balance,
+
2
2
2
mined. In all experiments reported here, more than 94%
of the initial mass was recovered after the reaction had
ended. This means that no large amounts of compounds
had escaped from the reaction vessel by evaporation or
otherwise. Typically, 85-90% of the products in the
recovered reaction mixture could be identified. Among
the unspecified products were several terpenes in amounts
less than 0.5%.
In separate experiments the oxygen content of the
incoming gas mixture was varied from 6% to 15% v/v,
corresponding to a partial oxygen pressure of 0.4 to 1.0
bar, under otherwise standard conditions. (Conditions:
O , and 38 °C). The selectivity ratio (SR) of the reaction,
2
which is defined as the ratio of the desired product
cyclohexene epoxide to the side products 2-cyclohexen-
1-ol and -1-one, was determined for both the catalyzed
and the uncatalyzed reaction. Interestingly, this ratio is
10-15 for reactions catalyzed by an iridium polyoxoanion
and 28-37 for the uncatalyzed epoxidation initiated by
alkylhydroperoxide. Mizuno concludes that the nature of
the metal catalyst has only a relatively small effect on
the selectivity, in agreement with the results of Nam et
2
1
al. This conclusion was, however, derived from experi-
J . Org. Chem, Vol. 69, No. 10, 2004 3459

Wentzel et al.
TABLE 1. Meta l-Ca ta lyzed Aer obic Oxid a tion of Cycloh exen ea
entry
catalyst
color before reaction
color after reaction
conv(%)b
yield of epoxide (%)b
SRc
1
2
3
4
5
6
7
8
9
1
1
1
1
none
colorless
colorless
99d
98
100
99
82
84
95
49
97
93
85
85
79
83
67
58
71
33
71
72
59
30
6
18
Ni(acac)2 3a
Ni(ptbbacac)2 3b
Ni(OAc)2
Ni(OAc)2‚4H2O
Co(acac)2
Co(OAc)2‚4H2O
Mn(acac)2
Mn(OAc)2‚4H2O
Cu(acac)2
pale green
pale purple
light green
light green
pale pink
pink
pale green
pale green
light green
light green
dark green
dark green
dark brown
n.d.f
bright blue
bright orange
bright yellow
purple
12.5
11.3
12.7
10.6
6.5
9.8
6.0
6.9
11.4
9.8
e
light brown
n.d.f
e
0
1
2
3
pale blue
Fe(acac)3
VO(acac)2
Cr(acac)3
dark red.
dark green
purple
78
70
12
2.9
9.9
d
a
Reaction conditions: 3.0 mmol of cyclohexene, 8.4 mmol of isobutyraldehyde, 0.03 mmol of catalyst (1 mol %), 1,3-dichlorobenzene as
b
c
an internal standard, 10 mL of CH2Cl2, stirred glass vessel equipped with an O2 reservoir, 25°C GC analysis after 5 h. SR ) selectivity
ratio, mmol of epoxide/mmol of allylic oxidation products (2-cyclohexen-1-one + 2-cyclohexen-1-ol) After 22 h. e Catalyst is only slightly
d
f
soluble in the reaction mixture Not determined.
ments on only three different, structurally related poly-
oxometalates. In our previous work, we tested a number
of transition metal â-diketonate complexes as catalysts
in the Mukaiyama epoxidation of limonene,2 but no
allylic oxidation products were found. Limonene is not a
good substrate, however, to establish the occurrence of
allylic oxidation (vide infra). Therefore, we decided to
study the oxidation of the more sensitive substrate
cyclohexene and to compare the selectivity ratios of a
number of different transition metal salts and metal
â-diketonate complexes (Table 1). These experiments
were aimed at finding a catalyst that combines a reduced
tendency for (unwanted) allylic oxidation with a high rate
of epoxidation.
changes in the reaction conditions. For example, the use
of nondehydrated or dehydrated nickel acetate as a
catalyst resulted in a considerable SR difference (entries
4 and 5). Not all differences in SR, however, can be
ascribed to the presence of water in the reaction mixture.
The SR was already shown by Mizuno23 to be solvent-
dependent, a noncoordinating solvent such as dichlo-
romethane giving a SR higher than that of a coordinating
one such as acetonitrile.
2
Discu ssion
Mech a n ism of Aer obic Lim on en e Ep oxid a tion
w ith Co-Oxid a tion of a n Ald eh yd e. The mechanism
of alkene epoxidation by aerobic co-oxidation with an
aldehyde was thoroughly investigated in the early seven-
We found that the reaction in the absence of a catalyst
gave a very high epoxide selectivity (entry 1) in agree-
ment with Mizuno’s results23 but proceeded only very
slowly at room temperature. (It should be noted that we
did not deliberately initiate the oxidation reaction, which
results in variable induction times but not in different
oxidation rates and reported conversions.) Highly inter-
esting was the observation that at the catalyst concentra-
tion used in our experiments (1 mol %), the selectivity of
epoxidation versus allylic oxidation was influenced by the
type of metal catalyst (see Table 1, last column). Nickel-
3
1
ties (see, for example, Vreugdenhil and Reit
and
3
2
Tsuchiya and Ikawa ) and has been reviewed by Sheldon
and Kochi.20 For the uncatalyzed reaction, Lassila et al.
have envisaged both a radical and a nonradical pathway.
On the basis of the results presented in the previous
2
II
sections we propose that the mechanism of the Ni -
(acetylacetonate)-catalyzed Mukaiyama epoxidation pro-
ceeds for the larger part via a radical pathway, i.e.,
autoxidation of the aldehyde (Scheme 1, eqs 1 and 2)
followed by epoxidation of the alkene by an acylperoxy
radical. Support for this mechanism is the formation of
(II) appeared to be the most efficient catalyst in these
experiments, combining the highest selectivity for ep-
oxide with the fastest reaction (full conversion was
obtained after 5 h, entries 2-4, except in the case of
hydrated nickel(II) acetate, entry 5). The observed se-
lectivity ratios of 10.6-12.5 are comparable to those
CO
2
and the mixture of ketone and alcohol (K + A)
(Scheme 2). A smaller part of the epoxidation follows the
nonradical peracid epoxidation route (Scheme 3), result-
ing in the formation of carboxylic acid.
In the radical pathway, the active epoxidizing species
are acylperoxy radicals formed by radical chain autoxi-
dation of the aldehyde (Scheme 1).20 The ratio of the
radical and nonradical pathways has an influence on the
type of products derived from the aldehyde. Thus, in the
nonradical peracid epoxidation pathway, the aldehyde is
expected to be converted into the corresponding carboxyl-
ic acid, whereas in the radical epoxidation route lower
molecular weight products are formed from the aldehyde
2
3
pusblished by Mizuno. Nickel was followed closely by
2
Cu(acac) , which displayed an SR of 11.4 at 93% conver-
sion (entry 10). The three metals that have been most
frequently studied in oxidations, cobalt(II), manganese-
(II), and iron(III), were somewhat less selective and less
efficient (entries 6-9 and 11). VO(acac)2 (entry 12) gave
epoxide and allylic oxidation products in an SR of 3, and
Cr(acac)3 gave almost no conversion (entry 13). Neither
of these complexes are suitable as a catalyst for the
Mukaiyama oxidation. Remarkably, in all experiments
where a reaction took place, with the exception of the
nickel- and copper-catalyzed reactions, clear color changes
were observed (see Table 1). We found that the selectivity
of the oxidation reaction was very sensitive to small
(Scheme 2). Formation of these lower molecular weight
compounds is accompanied by CO formation as was
2
(
31) Vreugdenhil, A. D.; Reit, H. Rec. Trav. Chim. Pays-Bas 1972,
1, 237-245.
(32) Tsuchiya, F.; Ikawa, T. Can. J . Chem. 1969, 47, 3191-3197.
9
3
460 J . Org. Chem., Vol. 69, No. 10, 2004

Mukaiyama Epoxidation
shown by Lassila et al.2 for pivaldehyde, which is
degraded into tert-butylhydroperoxide and tert-butanol.
These authors rationalized their findings by assuming a
concerted decomposition of an acylperoxy-alkene adduct
the type of aldehyde. This conclusion is supported by our
results for the catalyzed reaction shown in Figure 2,
which shows that the CO /epoxide ratio increases to a
2
1:1 level if the aldehyde/alkene ratio is lowered to 1.2:1,
in contrast to the 3:1 ratio used in the other experiments.
In flu en ce of Ald eh yd e/Alk en e Ra tio on th e Mu -
k a iya m a Ep oxid a tion . The Mukaiyama epoxidation
proceeds smoothly when an excess of aldehyde with
into epoxide, CO
2
, and an alkyl radical (eq 5), which is
rapidly trapped by dioxygen (eq 6). The fact that aromatic
6
-1 33
carboxyl radicals decarboxylate much slower (10 s
)
accounts for the observation22 that aromatic aldehydes
such as benzaldehyde are not active as co-reagents in the
Mukaiyama system. The alkyl radical that results from
the decarboxylation may be stabilized by the nickel center
and form a Ni-alkyl complex. This is known to occur in
nature in the chemistry of the cofactor F430, which
converts thioethers to methane via a methyl-Ni com-
plex.34
respect to alkene is applied, as many researchers have
noted.1
-18
This may be clarified by Scheme 3. In eq 10,
an acylperoxy radical abstracts a hydrogen atom from
another aldehyde molecule, forming peroxy acid and an
acyl radical, thus propagating the radical chain but not
generating epoxide. Subsequently, two molecules of car-
boxylic acid may be formed from the reaction of a peroxy
acid with an aldehyde (eq 12). Aldehyde is consumed
again without the formation of epoxide. The amount of
acid formed and thus the amount of aldehyde consumed
could depend on the catalyst, the substrate, and the
reaction conditions. In the system that we studied, only
approximately half of the aldehyde is converted into
carboxylic acid.
A consequence of the concerted decomposition shown
in Scheme 1, eq 5, is that equimolar amounts of epoxide,
CO , and lower molecular weight alkyl radical oxidation
2
products are expected if epoxidation proceeds exclusively
through the radical mechanism. In the case of 2-methyl-
undecanal, we do not find alkylhydroperoxide besides
2
CO but instead 2-undecanol and 2-undecanone, which
There are, however, more reasons why the ratio of
aldehyde to alkene in the Mukaiyama epoxidation is
larger than 1. For example, as can be seen in Figure 5,
it is evident that epoxidation under our conditions only
proceeds if the aldehyde/alkene ratio exceeds a certain
are formed in almost equal amounts during the first 2 h
of the reaction. Their equimolar formation is strong
evidence for the intermediacy of unstable alkylperoxy
radicals, which decompose into a 1:1 mixture of ketone
and alcohol via a Russell termination, which is outlined
in Scheme 2, eqs 8 and 9. The fact that somewhat more
ketone is formed is explained by the easy oxidation of
2
0
value (∼0.5 molar equiv). In an interesting series of
investigations by Wittig and co-workers,3
6-42
it was
shown that in the uncatalyzed co-oxidation of alkene and
aldehyde some alkenes are capable of retarding the
autoxidation of benzaldehyde. With an increasing ratio
of alkene to benzaldehyde, the rate of autoxidation
decreased. These results may bear upon the requirement
of a minimum aldehyde/alkene ratio in our experi-
ments: when this ratio is too low, i.e., when a large
amount of alkene is present with respect to aldehyde, the
alkene might inhibit the aldehyde autoxidation (eqs 1 and
the alcohol to the ketone under the autoxidation condi-
tions. Howard35 has provided a similar explanation of the
formation of a slight excess of ketone in the Russell
termination on the basis of a bicyclic tetroxide that
decomposes to form ketone and hydrogen peroxide.
Because the combined amount of undecanone and
2
-undecanol is only 50-60% of the total amount of
epoxide observed during the first hours of the epoxidation
(
∼90% conversion), we presume that ca. 40-50% of the
2
) and hence no oxidizing acylperoxy radical is generated.
epoxide is formed through the nonradical peracid path-
way. This pathway generates 2-methylundecanoic acid,
which was also detected as a major product derived from
the aldehyde (Scheme 3). The combined occurrence of
both peracid and radical epoxidation is also deduced from
Furthermore, at an aldehyde/alkene ratio where epoxi-
dation does not proceed, the epoxidation cannot be
II
induced by raising the concentration of Ni catalyst.
Thus, the aldehyde/alkene ratio seems to be more im-
II
portant than the ratio of aldehyde to Ni . A further
the amount of CO
of limonene in the presence of isobutyraldehyde. Figure
shows that the amount of CO is approximately 60% of
2
formed during the aerobic epoxidation
rationalization for the requirement of a minimum alde-
hyde/alkene ratio can be found in the work of Filippova
and Blyumberg.43 They observed that the rate of the
uncatalyzed alkene epoxidation with aldehyde co-oxida-
tion ceases to depend on the alkene concentration above
a certain threshold value of this concentration. They
assume that the acylperoxy radical and the alkene form
an adduct (Scheme 1, eq 3) so that at a certain alkene
3
2
the amount of epoxide formed during the first 2.5 h of
the reaction (at 70% conversion), whereas an equimolar
amount is expected if radical epoxidation is the exclusive
pathway. Thus, as with 2-methylundecanal, aerobic
epoxidations with isobutyraldehyde are also likely to
proceed via concomitant radical and peracid pathways.
In a study of the uncatalyzed co-oxidation of alkene
and aldehyde Vreugdenhil and Reit31 found that the
percentage of radical epoxidation is around 45%. This
percentage was influenced by the ratio of olefin to
aldehyde and by the reactivity of the alkene but not by
(
36) Wittig, G.; Pieper, G. J ustus Liebigs Ann. Chem. 1947, 558,
218-230.
(37) Wittig, G.; Lange, W. J ustus Liebigs Ann. Chem. 1938, 536,
2
66-284.
(
130-144.
(
38) Wittig, G.; Henkel, K. J ustus Liebigs Ann. Chem. 1939, 542,
39) Wittig, G.; Pieper, G. J ustus Liebigs Ann. Chem. 1940, 546,
1
42-171.
(
33) Bravo, A.; Bjorsvik, H.-R.; Fontana, F.; Minisci, F.; Serri, A. J .
Org. Chem. 1996, 61, 9409-9416.
34) Wackett, L. P.; Honek, J . F.; Begley, T. P.; Shames, S. L.;
(40) Wittig, G.; Pieper, G. J ustus Liebigs Ann. Chem. 1940, 546,
172-179.
(41) Wittig, G. J ustus Liebigs Ann. Chem. 1947, 558, 201-206.
(42) Wittig, G.; Pieper, G. J ustus Liebigs Ann. Chem. 1947, 558,
207-218.
(43) Filippova, T. V.; Blyumberg, E. A. Russ. Chem. Rev. 1982, 51,
582-591.
(
Niederhoffer, E. C.; Hausinger, R. P.; Orme-J ohnson, W. H.; Walsh,
C. T. In The Bioinorganic Chemistry of Nickel; Lancaster, J . R., J r.,
Ed.; VCH: Weinheim, 1988; pp 249-274.
(35) Howard, J . A. ACS Symp. Ser. 1978, 69, 413-432.
J . Org. Chem, Vol. 69, No. 10, 2004 3461

Wentzel et al.
concentration all acylperoxy radicals are trapped in this
adduct. As a result, the concentration of free, unbound
acylperoxy radicals, which can act as chain carriers
during aldehyde autoxidation and as oxidizing agents,
is too low to allow for efficient epoxidation.
SCHEME 6
At high aldehyde concentrations (more than 4 equiv
with respect to alkene, i.e., at an aldehyde concentration
>
1.5 M), the epoxidation reactions show a sudden drop
in rate (Figure 5). An explanation may be given based
on the work by Mizuno,23 who noted that the aerobic
oxidation of aliphatic aldehydes becomes limited by mass
transfer at aldehyde concentrations higher than 2-5 mol
%. The points where the curves in Figure 5 bend
correspond to 15 mol % of aldehyde (1.6 mol) in 150 mL
of toluene, and mass transfer might become a serious
problem in that case. A solvent effect on the reaction rate
may also play a role, since a 2.7 M solution of aldehyde
in toluene (10 equiv of aldehyde with respect to alkene)
consists of 30% v/v aldehyde.
However, we favor a different explanation for the
decreased reaction rate at high aldehyde concentration.
It is important to note that the rate of reaction in Figure
that result from the formation of higher oxidation state
species derived from vanadium, manganese, and cobalt
catalysts.
Scheme 6 provides a reaction sequence for nickel-
catalyzed aldehyde oxidation that explains why there are
no measurable concentrations of high oxidation state
III
Ni , despite the oxidizing conditions of the epoxidation
III
reaction and the probable role (see below) of Ni in the
5
was measured by the formation of epoxide or CO
2
; in
initiation of the radical chain reaction. This scheme is
based on earlier work of M a´ rta et al., who studied the
O2 oxidation of benzaldehyde catalyzed by cobalt and
other words, when no epoxide or CO is formed in the
2
oxidation process in the reaction vessel, no reaction is
observed. One such a process is the reaction of the
aldehyde with the acylperoxy radical yielding a peroxy
acid and an acyl radical (Scheme 3, eq 10), thus propa-
gating the autoxidation chain (i.e., conversion of alde-
4
6
nickel salts. These two catalysts differ in that Co is
III
mainly present in the form of Co , whereas for Ni the
lower, bivalent oxidation state prevails; the reason for
II
III
this is that oxidation of Ni to Ni by intermediate
peracid (eq 16) is slow compared to the rapid reduction
2
hyde) without generating epoxide or CO . Competing
with this reaction is the oxidation of alkene by the
III
of Ni by the aldehyde, a situation that is reversed for
acylperoxy radical (Scheme 1, eqs 3-5), which does yield
Co and most other metal ions (except Cu). This reduction
III
CO
2
and epoxide that are measured and plotted in Figure
of Ni by the aldehyde with formation of acyl radicals is
III
5
. The relative rates of these two reactions, which depend
proposed to proceed via a Ni -acyl intermediate, whose
on the alkene and aldehyde concentrations, determine
the overall rate that is measured. When the aldehyde
concentration is relatively high with respect to the alkene
concentration, aldehyde oxidation is much faster than
epoxidation (with CO formation) and an epoxidation rate
2
of zero is eventually measured.
formation is preceded by aldehyde coordination to the
III
Ni (OH) species, which is in turn formed by oxidation
II
of Ni by the peracid (eqs 17 and 18).
In earlier work,22 we obtained evidence for aldehyde
coordination to nickel by UV/vis measurements, thus
making eq 17 plausible. The formation of a metal acyl
species from a group 10 metal compound and an aldehyde
is supported by the work of Pregosin et al., who have
isolated stable metal acyl complexes from the reaction
In flu en ce of Meta l Ca ta lyst on th e Mu k a iya m a
Ep oxid a tion . Mizuno et al. have observed that the metal
catalyst influences the ratio of epoxidation versus allylic
oxidation but that the nature of the metal has no effect
on this ratio.23 Our cyclohexene oxidation experiments
with various metal catalysts confirm that there is an
influence of the metal catalyst, but contrary to Mizuno
we find a dependence of the selectivity ratio on the nature
of the metal (see Table 1), i.e., metals with a tendency
for high oxidation states (vanadium, manganese, cobalt)
induce more allylic oxidation than metals that do not
have such a tendency (nickel and copper). For nickel, the
of Pd and Pt compounds with quinoline-8-carboxalde-
47,48
hyde.
The initiation of the radical chain by peracid
II
oxidation of Ni is similar to the initiation mechanism
proposed by Kholdeeva et al. for Co -catalyzed alkene
oxidation by dioxygen in the presence of an aldehyde.
With EPR spectroscopy we were able to show that only
oxygen-centered radicals are trapped in the reaction
mixture, which means that all carbon-centered radicals
formed in Scheme 6 are quickly trapped by dioxygen
II
4
9
2
2
III
absence of a significant concentration of the Ni high
oxidation state during aerobic epoxidation is evident from
the absence of an EPR signal in the reaction mixture
(
45) Gmelins Handbuch der Anorganischen Chemie; Meyer, R. J .,
2
2
Pietsch, E. H. E., Kotowski, A., Eds.; Verlag Chemie: Weinheim, 1969;
Vol. 57-C2.
II
III
(
Ni is silent in conventional EPR, whereas Ni com-
(
(
46) M a´ rta, F.; Boga, E.; Mat o´ k, M. Faraday Discuss. 1968, 46, 173.
47) Anklin, C. G.; Pregosin, P. S. J . Organometallic Chem. 1983,
4
4
plexes do exhibit EPR signals ). In addition, both for
nickel as well as for copper catalysts, no color changes
are observed during the reaction, suggesting that no
II
II
243, 101-109.
48) Anklin, C. G.; Pregosin, P. S.; Wombacher, F. J .; R u¨ egg, H. J .
Organometallics 1990, 9, 1953-1958.
49) Kholdeeva, O. A.; Khavrutskii, I. V.; Romannikov, V. N.;
(
III
III
appreciable amounts of a highly colored Ni or Cu are
(
present.45 This contrasts with the marked color changes
Tkachev, A. V.; Zamaraev, K. I. Selective Alkene Epoxidation by
Molecular Oxygen in the Presence of Aldehyde and Different Type
Catalysts Containing Cobalt; Grasselli, R. K., Oyama, S. T., Gaffney,
A. M., Lyons, J . E., Eds.; Elsevier Science: New York, 1997; pp 947-
955.
(
44) Coyle, C. L.; Stiefel, E. I. In The Bioinorganic Chemistry of
Nickel; Lancaster, J . R., J r., Ed.; VCH: Weinheim, 1988; pp 1-28.
3
462 J . Org. Chem., Vol. 69, No. 10, 2004

Mukaiyama Epoxidation
Scheme 1, eq 2). The formation of acyl radicals (eq 18)
is supported by the observation that at higher reaction
temperatures CO is generated (see Scheme 3, eq 13),
since at these high temperatures the decarbonylation of
(
down. This feature has been observed earlier by our
group24 and also by Kholdeeva et al.
49,55
and is consistent
with the radical autoxidation mechanism. Kholdeeva
et al. concluded that the observed dependence may
be interpreted as being the result of the participation of
2
the acyl radical is faster than its trapping by O .
n+
The reaction sequence of Scheme 6, which explains the
the catalyst (M ) in a chain termination reaction, for
III
very low concentration of nickel in its high Ni oxidation
example, when acylperoxy radicals react to form
M(
n+1)+
-RCO
-
state, rationalizes the low amount of allylic oxidation
products (high SR value) observed for Ni-catalyzed ep-
oxidation with isobutryaldehyde and dioxygen, since it
is known that transition metal complexes in a high
oxidation state are capable of abstracting allylic hydro-
gens (in contrast to their more reduced counterparts),50
thus initiating allylic oxidation. The observed SRs cor-
related with the Irving-Williams series of formation
3
, which is inactive. Nickel(II) cyclam com-
5
6
plexes have also been shown to exhibit this behavior.
In our system, the monomeric complex 3b apparently is
more efficient in trapping radicals than the (partially)
trimeric complex 3a , which becomes evident at higher
catalyst concentrations. Interestingly, the concentration
at which a negative effect of the nickel catalyst is
observed is much higher than that of cobalt catalysts:
5
1-53
II
-1
49,55
-3
constants of complexes of divalent ions.
The Ni and
10-4 mol L for cobalt
versus more than 2 × 10
II
-1
Cu complexes have the highest formation constants and
mol L for nickel (Figure 6). Thus, nickel is fortunately
under our conditions not an efficient chain terminating
agent, which allows us to use much higher concentrations
of this catalyst to increase the reaction rate. Free
alkylperoxy radicals can combine in a Russell termina-
II
II
II
gave the highest SRs, whereas the Co , Fe , and Mn
complexes, in order of decreasing formation constants,
gave lower SRs. The vanadyl ion cannot be placed in this
series. This correlation indicates once more that the high
valent complexes of NiII and Cu are much less stable
than the high valent complexes of the other studied
metals and will not be present in large amounts in the
reaction mixture. The uncatalyzed reaction produces only
acylperoxy and carboxyl radicals (Schemes 1 and 2),
which are much more likely to add to a double bond than
to abstract an allylic hydrogen atom, thus giving a very
high selectivity ratio in favor of epoxidation.
II
35
tion step (eq 9) to form alcohol, ketone, and molecular
oxygen.
Con clu sion
We have shown that the aerobic epoxidation of li-
monene with nickel(II) â-diketonate as a catalyst and an
aldehyde as co-reagent shows some unexpected and
interesting features when studied in detail. First of all,
low molecular weight alcohol and ketone together (K +
A) are formed in a less than 1:1 ratio (ca. 50-60%) with
respect to epoxide. Concomitant formation of carbon
dioxide resulting from the decarboxylation of a carboxyl
radical was detected quantitatively and also in a less
than 1:1 ratio with respect to epoxide, as expected from
Schemes 1 and 2. It is important to note that this ratio
changes to a 1:1 ratio when the ratio of aldehyde to
In an earlier study we proposed that the epoxidizing
I
acylperoxy radical is stabilized by a Ni species (a low
oxidation state species).22 It was assumed that in the
initiation step, proton abstraction is accompanied by the
II
uptake of an electron from the aldehyde by Ni , thus
I
generating an acyl radical and Ni . In our discussion we
III
suggested that an intermediate Ni species would also
be conceivable.22 From our present results, we may
III
tentatively conclude that the latter possibility, i.e., Ni ,
is more likely, although only as an elusive transient
intermediate species. The role of small amounts of Ni
2
alkene is reduced from 3 to 1. The products (CO , ketone
III
and alcohol) are formed after radical epoxidation of
alkene by an acylperoxy radical, which is formed through
autoxidation of aldehyde (Scheme 1). A nonradical ep-
oxidation pathway via a peracid route (Scheme 3) plays
an appreciable role when the ratio of aldehyde to alkene
is 3:1. This produces carboxylic acid as the oxidation
product from the aldehyde. Radical autoxidation becomes
the exclusive pathway when the reaction conditions are
altered to low aldehyde-to-alkene ratios, e.g., 1:1.
is analogous to that proposed for the mechanism of
aerobic oxidation of DNA by Ni complexes with sulfite
as a coreactant.5
4
It may be expected that coordination of the aldehyde
to the nickel center is hindered when the Ni is 6-coordi-
nate (i.e., as in 3a ), which thus would slow the reaction.
Since the reaction with the square planar complex 3b is
faster at low catalyst concentrations (Figure 6), it may
be concluded that aldehyde coordination is influenced by
the catalyst structure in solution. (Note that in Figure 5
it appears as if there is only a slight difference in reaction
rate between the two nickel catalysts. However, the
nickel concentration at which these experiments were
done corresponds to the point in Figure 6 where the lines
cross, i.e., the observation of equal rates in Figure 5 is a
coincidence.) The same figure shows that at higher
catalyst concentration, the rate of the reaction slows
From the study of the oxidation of cyclohexene as
substrate we conclude that the metal catalyst in the
Mukaiyama epoxidation not only influences the initiation
and the rate of the reaction but also the selectivity for
epoxide versus allylic oxidation products. Nickel(II) ap-
pears to be the best epoxidation catalyst in a series of
metal salts and metal complexes tested as catalysts. This
metal enhances the rate of oxidation with respect to the
blank reaction and also gives the highest epoxidation/
allylic oxidation ratio. In other words, for a substrate that
II
(50) Shilov, A. E. Activation of Saturated Hydrocarbons by Transi-
is sensitive to allylic epoxidation, a Ni catalyst is the
tion Metal Complexes; D. Reidel Publishing Company: Dordrecht, 1984.
(
(
(
51) Irving, H.; Williams, R. J . P. J . Chem. Soc. 1953, 3192-3210.
52) Irving, H.; Williams, R. J . P. Nature 1948, 162, 746-747.
53) Shriver, D. F.; Atkins, P. W.; Langford, C. H. Inorganic
(55) Kholdeeva, O. A.; Grigoriev, V. A.; Maksimov, G. M.; Fedotov,
M. A.; Golovin, A. V.; Zamaraev, K. I. J . Mol. Catal. A: Chem. 1996,
114, 123-130.
(56) Nam, W.; Baek, S. J .; Lee, K. A.; Ahn, B. T.; Muller, J . G.;
Burrows, C. J .; Valentine, J . S. Inorg. Chem. 1996, 35, 6632-6633.
Chemistry, 1st ed.; Oxford University Press: Oxford, 1990.
54) Lepentsiotis, V.; Domagala, J .; Grgic, I.; van Eldik, R.; Muller,
J . G.; Burrows, C. J . Inorg. Chem. 1999, 38, 3500-3505.
(
J . Org. Chem, Vol. 69, No. 10, 2004 3463

Wentzel et al.
best choice in the Mukaiyama epoxidation. The low
degree of allylic oxidation with Ni catalysts is rationalized
by the very low concentration of nickel in its NiIII high
oxidation state during the reaction. The very strong
quently, a significant degree of allylic oxidation is ob-
served.
Ack n ow led gm en t. The authors thank Dr. P. J . M.
van Kan for EPR measurements and Dr. P. H. M.
Budzelaar for valuable discussions. This research was
financed by The Netherlands IOP Foundation for In-
novative Research and The Netherlands Ministry of
Economic Affairs (project IKA 94025).
II
III
predominance of Ni species over Ni species is thought
III
to result from the fast reduction of Ni by the aldehyde,
II
III
compared to the slower oxidation of Ni to Ni by
peracid. For metals that are less readily reduced by the
aldehyde, such as vanadium, manganese, and cobalt, the
high oxidation state of the catalyst prevails. Conse-
J O030345A
3
464 J . Org. Chem., Vol. 69, No. 10, 2004