Mechanistic studies of copper-catalyzed alkene aziridination

Article abstract of DOI:10.1021/ja993246g

The mechanism of the copper-catalyzed aziridination of alkenes using [N-(p-toluenesulfonyl)imino]-phenyliodinane (PhINTs) as the nitrene source has been elucidated by a combination of hybrid density functional theory calculations (B3LYP) and kinetic exper

Full text of DOI:10.1021/ja993246g

J. Am. Chem. Soc. 2000, 122, 8013-8020
8013
Mechanistic Studies of Copper-Catalyzed Alkene Aziridination
Peter Brandt,†,§ Mikael J. S o1 dergren, Pher G. Andersson,* and Per-Ola Norrby*
‡
,‡
,†
Contribution from the Department of Medicinal Chemistry, Royal Danish School of Pharmacy,
UniVersitetsparken 2, DK-2100 Copenhagen, Denmark, and Department of Organic Chemistry,
Uppsala UniVersity, Box 531, S-751 21 Uppsala, Sweden
ReceiVed September 7, 1999
Abstract: The mechanism of the copper-catalyzed aziridination of alkenes using [N-(p-toluenesulfonyl)imino]-
phenyliodinane (PhINTs) as the nitrene source has been elucidated by a combination of hybrid density functional
theory calculations (B3LYP) and kinetic experiments. The calculations could assign a Cu(I)/Cu(III)-cycle to
the reaction and demonstrate why a higher oxidation state of copper cannot catalyze the reaction. A mechanism
whereby Cu(II)-catalyst precursors can enter the Cu(I)/Cu(III)-cycle is suggested. Three low-energy pathways
were found for the formation of aziridines, where the two new N-C bonds are formed either in a nonradical
concerted or consecutive fashion, by involvement of singlet or triplet biradicals. A close correspondence was
found between the title reaction and the Jacobsen epoxidation reaction in terms of spin-crossings and the
mechanism for formation of cis/trans isomerized products. The kinetic part of the study showed that the reaction
is zero order in alkene and that the rate-determining step is the formation of a metallanitrene species.
Introduction
Scheme 1. Copper-Catalyzed Aziridination of Alkenes
Enantioselective functionalization of simple alkenes by oxida-
tive methods has experienced an explosive growth in the past
1
decade. Reactions such as the Sharpless dihydroxylation and
2
3
aminohydroxylation, and Jacobsen epoxidation of unfunction-
alized alkenes have had a tremendous impact in the field of
4
asymmetric synthesis. Transition-metal catalyzed aziridination
alkenes, whereas the Jacobsen system using dibensylidene
diimines 2 is better suited for aziridination of cis-alkenes.
has been less exploited despite the significant utility of aziridines
in organic synthesis. A recent development of this reaction
5
makes use of [N-(p-toluenesulfonyl)imino]phenyliodinane
(PhINTs) as the nitrene source and a metal catalyst which
mediates the cycloaddition to the alkene, yielding N-tosylaziri-
dines (Scheme 1). The synthetic utility of the sulfonylated
products has recently been enhanced by development of methods
6
for selective cleavage of the protecting group and introduction
of alternative nitrogen sources.7
8
In the title reaction, both Cu(I) and Cu(II) metal sources form
competent catalysts and several observations strongly indicate
that a common oxidation state is reached in both cases. Evans
and co-workers showed that degree of cis/trans isomerization
in the aziridination of (Z)-alkenes as well as the relative
reactivity of various alkenes are independent of the initial
The literature on the metal-catalyzed aziridination of alkenes
is dominated by the work of Evans9 and Jacobsen,
,10
11,12
who
also designed chiral ligands for the asymmetric reaction. The
bisoxazoline ligands 1 of Evans are most successful for trans-
†
Royal Danish School of Pharmacy.
Uppsala University.
Present address: Karo Bio AB, Novum, S-141 57, Huddinge, Sweden.
‡
10
catalyst oxidation state. In the asymmetric version of the
§
reaction, bisoxazolines 1 result in the same levels of asymmetric
(
1) Kolb, H. C.; Sharpless, K. B. Asymmetric Dihydroxylation. In
9
induction with CuOTf and Cu(OTf)2.
Transition Metals for Organic Synthesis; Beller, M., Bolm, C., Eds.; Wiley-
VCH: Weinheim, Germany, 1998; pp 219-242.
(2) Kolb, H. C.; Sharpless, K. B. Asymmetric Aminohydroxylation. In
(8) Manganese: (a) Noda, K.; Hosoya, N.; Irie, R.; Ito, Y.; Katsuki, T.
Synlett 1993, 469. (b) Nishikori, H.; Katsuki, T. Tetrahedron Lett. 1996,
37, 9245. (c) Lai, T.-S.; Kwong, H.-L.; Che, C.-M.; Peng, S.-M. Chem.
Commun. 1997, 2373. Ruthenium: (d) Au, S.-M.; Fung, W.-H.; Cheng,
M.-C.; Che, C.-M.; Peng, S.-M. Chem. Commun. 1997, 1655. Rhodium:
(e) M u¨ ller, P.; Baud, C.; Jacquier, Y. Can. J. Chem. 1998, 76, 738.
(9) (a) Evans, D. A.; Faul, M. M.; Bilodeau, M. T. J. Org. Chem. 1991,
56, 6744. (b) Evans, D. A.; Faul, M. M.; Bilodeau, M. T., Anderson, B.
A.; Barnes, D. M. J. Am. Chem. Soc. 1993, 115, 5328.
(10) Evans, D. A.; Faul, M. M.; Bilodeau, M. T. J. Am. Chem. Soc.
1994, 116, 2742.
Transition Metals for Organic Synthesis; Beller, M., Bolm, C., Eds.; Wiley-
VCH: Weinheim, Germany, 1998; pp 243-260.
(3) Jacobsen, E. N. Asymmetric Catalytic Epoxidation of Unfunction-
alized Olefins. In Catalytic Asymmetric Synthesis; Ojima, I., Ed.; VCH
Publishers: New York, 1993; pp 159-202.
(
4) M u¨ ller, P. Transition Metal-Catalyzed Nitrene Transfer. In AdVances
in Catalytic Processes; JAI Press Inc.: Greenwich, CN, 1997; Vol. 2, pp
1
13-151.
(
(
(
5) Tanner, D. Angew. Chem., Int. Ed. Engl. 1994, 33, 599.
6) Alonso, D. A.; Andersson, P. G. J. Org. Chem. 1998, 63, 9455.
7) (a) S o¨ dergren, M. J.; Alonso, D. A.; Bedekar, A. V.; Andersson, P.
(11) Li, Z.; Conser, K. R.; Jacobsen, E. J. J. Am. Chem. Soc. 1993, 115,
5326.
(12) Li, Z.; Quan, R. W.; Jacobsen, E. J. J. Am. Chem. Soc. 1995, 117,
5889.
G. Tetrahedron Lett. 1997, 38, 6897. (b) S o¨ dergren, M. J.; Alonso, D. A.;
Andersson, P. G. Tetrahedron: Asymmetry 1997, 8, 3563. (c) Dauban, P.;
Dodd, R. H. J. Org. Chem. 1999, 64, 5304.
1
0.1021/ja993246g CCC: $19.00 © 2000 American Chemical Society
Published on Web 07/29/2000

8014 J. Am. Chem. Soc., Vol. 122, No. 33, 2000
Brandt et al.
The authors suggest that the active catalyst is in the +II
oxidation state and that PhINTs may act as an oxidant for Cu(I).
This was supported by experiments where treatment of the
CuOTf precatalyst with PhINTs in the absence of alkene gave
a species indistinguishable (UV-vis) from that produced with
Cu(OTf)2 under identical conditions.8
b,9
On the basis of kinetic studies using Cu(I)-bis(imine)
catalysts, Jacobsen has concluded that the reaction is strictly
first order in alkene and suggests a Cu(III)-nitrene species to
1
2
be the reactive intermediate in a Cu(I)/Cu(III) catalytic cycle.
On the other hand, P e´ rez and co-workers suggest the catalyti-
cally active species to be Cu(II) in systems with monoanionic
hydridotrispyrazolylborate ligands, thus involving a Cu(II)/Cu-
(
gII) cycle.13
The carbon-nitrogen bond formation step(s) may be either
Figure 1. Diimine 3, used in the kinetic study and systems 4a-c used
in the calculations (4b:X ) H, 4c:X ) Cl).
concerted or consecutive. P e´ rez investigated the Cu(II)-
hydridotrispyrazolylborate-catalyzed aziridination by means of
Hammet studies, which suggested a reaction between the alkene
and an electrophilic radical species. The experimental data could
functions were used for iodine (BSII). An optimized f-polarization
function was obtained by minimizing the B3LYP energy of the free
atom in the ground state, resulting in an exponent of 0.2.19 Final energies
were determined using B3LYP together with the 6-311+G(2d) for Cu,
N, O, S and C(alkene); 6-311G(d,p) was used for other atoms and BSII
for iodine (BSIII). Normal modes and zero point corrections (ZPC)
were calculated for some selected points using B3LYP/BSI.
•
be nicely fitted to a linear combination of Jackson’s σ
substituent constants and Hammet σ constants.
+
13
Counterions markedly influence the stereochemical outcome
of aziridination of (Z)-alkenes. Strongly coordinating counterions
such as acetylacetonate and chloride favor isomerization of cis-
stilbene and cis-â-methylstyrene to the trans-aziridine products,
Model System. A neutral ligand was chosen due to the high
enantioselectivities obtained using such ligands. The diimine ligand 3
possibly indicating an involvement of radical intermediates in
1
2
the system.10 In an attempt to find evidence for such species,
of Jacobsen was selected due to the published kinetic study and the
simple synthesis of this ligand.1
1,20
The calculations were performed
aziridinations were carried out using a so-called hypersensitive
radical probe as substrate. These experiments suggest a con-
certed reaction pathway for alkyl-substituted alkenes using
CuClO4 as catalyst in acetonitrile. However, cis-4-octene did
not isomerize in the same system.10 Thus, in parallel with results
for the Jacobsen epoxidation reaction,14 radicals might only be
intermediates for substrates where significant stabilization
occurs.
Further development of this reaction requires an increased
understanding of the reaction mechanism. We have therefore
performed a theoretical study using quantum chemical calcula-
tions in combination with kinetic experiments.
both on a minimal model system 4a with methyl or hydrogen as models
for the aryl moieties, and on larger systems where all aromatic moieties
were retained, 4b and c (Figure 1).
The calculations were mainly performed using Cu(I) as the initial
12
oxidation state as postulated by Jacobsen. A catalytic cycle starting
with Cu(II) reacting with PhINTs would formally yield a Cu(IV)-
nitrene species, which in our opinion seems unlikely to be a catalytically
active species. Some calculations were performed on the dicationic
Cu(II) system in order to compare its relevance with that of the Cu(I)/
Cu(III) catalytic cycle.
Kinetic Experiments. Kinetic data were obtained from the aziri-
dination of 1,2-dihydronaphthalene by (N-(p-toluenesulfonyl)imino)-
phenyliodinane (PhINTs) with a catalyst generated in situ from
Methods
Cu(CH
3 4 6
CN) PF and diimine 3, and the reaction was performed at 0-2
°C. Alkene consumption was monitored by gas chromatography, using
Computational Details. All calculations reported in this work were
conducted using the Gaussian 98 program. Geometry optimizations
were performed using the B3LYP hybrid functional, together with
n-dodecane as internal standard.
1
5
1
6
Experimental Section
1
7
the LANL2DZ ECP and basis set (BSI). Intermediate energies were
determined using B3LYP together with the 6-311+G(d) for Cu, N, O,
S and alkene carbons; 6-31G was used for other atoms, whereas the
General Methods. All reactions were run under an argon atmo-
sphere, using oven-dried glassware and magnetic stirring. Molecular
sieves were activated at 250 °C and 0.5 µbar for 24 h and then stored
in a drybox. Methanol was heated at reflux over magnesium turnings
for several hours and then distilled and stored over activated 3 Å
molecular sieves under argon. Dichloromethane was distilled from
1
8
SDD ECP and basis set augmented with one set of f-polarization
(13) D ´ı az-Requejo, M. M.; P e´ rez, P. J.; Brookhart, M.; Tempelton, J. L.
Organometallics 1997, 16, 4399.
14) Linde, C.; Arnold, M.; Norrby, P.-O.; Åkermark, B. Angew Chem.
Int. Ed. Engl. 1997, 36, 1802-1803.
15) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
(
2
powdered CaH under nitrogen just prior to use. 1,2-Dihydronaphthalene
(
was filtered through neutral alumina and freshly distilled. Diimine 3
was prepared according to a literature procedure.11 Analytical GC was
carried out using a Varian OV-5 column (30 m, 0.32 mm i.d., 0.25 µm
film) and N as carrier gas. Neutral alumina (Merck, activity I) was
2
used for the filtration of GC samples.
Typical Procedure for the Kinetic Analyses. In a 5 mL pear-shaped
flask Cu(CH
mL) and stirred for 10 min under argon. The solution was filtered
through a plug of glass wool to a three-necked flask containing a
suspension of 1,2-dihydronaphthalene (130-260 mg, 1.0-2.0 mmol),
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A.; Stratmann,
R. E., Jr.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J.
V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.;
Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng,
C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.;
Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.; Head-
Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.3;
Gaussian, Inc.: Pittsburgh, PA, 1998.
3
CN)
4
PF
6
2 2
(16 mg, 43 µmol) was dissolved in CH Cl (3.0
(
16) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652. (b) Lee,
(19) For a similar determination, see Ehlers, E. W.; B o¨ hme, M.; Dapprich,
S.; Gobbi, A.; H o¨ llwarth, A.; Jonas, V.; K o¨ hler, K. F.; Stegmann, R.;
Veldkamp, A.; Frenking, G. Chem. Phys. Lett. 1993, 208, 111.
(20) Quan, R. W.; Li, Z.; Jacobsen, E. J. J. Am. Chem. Soc. 1996, 118,
8156.
C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789.
(
17) Hay; P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299.
(18) Andrae, D.; H a¨ ussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor.
Chim. Acta 1990, 77, 123.

Studies of Copper-Catalyzed Alkene Aziridination
J. Am. Chem. Soc., Vol. 122, No. 33, 2000 8015
diimine ligand 3 (26 mg, 50 µmol), internal standard (n-dodecane, ∼110
mg) and activated MS 4 Å powdered molecular sieves ( ∼1 g) in
Scheme 2. Species Evaluated in the Quantum Chemical
Study
a
2 2 6
CH Cl (10.0 mL). Complete transfer of CuPF was ensured by washing
2 2
the glass wool with one portion of CH Cl , (1.0 mL). The mixture was
stirred at room temperature for 45 min. The three-necked flask was
immersed in an ice-water bath and its internal temperature allowed to
stabilize at +0.4 °C An aliquot was taken at this point to establish the
2
1
initial alkene/dodecane ratio, and then PhINTs (500 mg, 1.3 mmol)
2
2
was added in 4-6 portions during ∼90 s. Aliquots of 25 µL were
taken at intervals of ca. 40 s, diluted with EtOAc ( ∼1 mL), and filtered
through a plug of neutral alumina.
[
N-(p-Toluenesulfonyl)imino]phenyliodinane (PhINTs), Modified
2
3
Procedure. A suspension of p-toluenesulfonamide (6.8 g, 40 mmol)
and 85% KOH (6.5 g, 99 mmol) in methanol (60 mL), was cooled in
an ice-water bath, and iodosobenzene diacetate (15 g, 47 mmol) was
added portionwise over a period of 10-20 min, so that the internal
temperature was maintained below +10 °C. The reaction mixture was
refrigerated at +5 °C overnight and then filtrated. The filter-cake was
washed with ice-cold MeOH (20 mL), and the solids were then dried
at 0.5 µbar for 10 h in a Kugelrohr oven at 25 °C, affording PhINTs
as an off-white to yellowish solid (12 g, 80%), with properties as
2
3
previously reported in the literature.
a
Results and Discussion
Small system: R ) R′ ) H, R′′ ) Me, large systems: R ) Ph or
2
-Cl-Ph, R′ ) R′′ ) Ph.
The Dicationic System. Calculations on dicationic systems
are inherently less reliable than neutral or monocationic systems,
as solvent and counterions (neglected in the calculations) can
be expected to have a larger influence for the more charged
systems. Nevertheless, some important clues to the mechanistic
puzzle can be obtained from the calculations. The Cu-PhINTs
complex 7 (Scheme 2) is easily formed, and will dissociate PhI
in an exothermic reaction (∆E ) -71 kcal/mol). However, the
dicationic form of 8 is unstable, with a very high electron
affinity, more than sufficient to oxidize PhI or the alkene. Thus,
Table 1. Potential Energy Surface of System 4a (kcal/mol)
Including ZPC
species
ma
BSI
BSII
BSIII
5
6
7
8
8
9
TS1
10
10
1
1
1
1
3
3
3
3
-30.5
0.0
-32.0
0.0
-31.8
0.0
-53.4
-43.3
-50.4
-54.4
-52.0
-68.6
-71.2
-70.0
-87.9
-63.4
-44.4
-38.6
-38.7
-41.1
-38.7
-59.5
-59.1
e
-44.0
-39.0
-38.9
-40.0
-37.3
-56.7
-57.2
e
+
•
a more facile reaction of 7 is to dissociate PhI , yielding
monocationic 8 (∆E ) -148 kcal/mol). Further calculations
on the dicationic PES also indicate that complex 11 would act
as a thermodynamic sink for the reaction, requiring more than
b
c
1
1
b
c,d
TS2
1
6
1
1
1
-85.1
-60.6
-79.5
-55.0
+product
3
0 kcal/mol to liberate the product and re-form 8. To conclude,
a
m ) multiplicity. b Open-shell singlet with biradical character
the calculations do not support Cu(II) as the active oxidation
state but instead show a favorable path for conversion of Cu(II)
to Cu(I) under the reaction conditions. Thus, the results
rationalize why Cu(I) and Cu(II) salts yield the same active
species and indicate that the catalytic cycle involves Cu(I)/
Cu(III), as suggested by Jacobsen.12 For details on the dicationic
system, see Supporting Information.
c
converged in Gaussian 94. ZPC transferred from the triplet of 10.
d
q
e
∆
E ) 1.2 kcal/mol calculated from the open-shell singlet 10. Did
not converge.
Table 2. Potential Energy Surface of System 4b and ca
system
4
b
4c
The PES of the Monocationic Systems 4a-c. The calcula-
tions were performed for the species depicted in Scheme 2 and
the respective energies are shown in Tables 1 and 2. Energies
cited in the text have been calculated at B3LYP/BSIII with ZPE
at B3LYP/BSI unless otherwise stated (see Figure 2).
species
mb
BSI
BSI
BSII
BSIII
5
6
8
8
0
1
6
1
1
1
3
3
1
1
-21.5
0.0
-12.5
0.0
-14.6
0.0
-13.7
0.0
-34.1
-45.5
-70.4
-89.6
-51.9
-23.4
-33.3
-55.0
-73.0
-51.9
c
-22.3
-49.3
-73.0
-55.8
-23.5
-47.8
-69.4
-54.3
1
1
(21) The quality of PhINTs proved to have a significant influence on
the reaction. Reproducible rates and full alkene conversions were more
securely obtained using PhINTs prepared by the modified version of the
Yamada protocol (see Experimental Section and ref 23).
a
Zero point corrections from system 4a. b m ) multiplicity Not
c
(22) Due to the exothermic nature of the aziridination reaction, the
calculated.
addition PhINTs was added in portions. A simple setup that we found useful
for the portionwise, yet quick addition of solids under inert atmosphere is
described as follows: Onto the reaction vessel is mounted a Schlenck-type
assembly having a female joint with the inlet in parallel with the surface of
the reaction medium. In this joint is fixed a rubber septum with a hole of
appropriate size to provide a gastight fit around the body of an open-front,
Copper-Alkene Complex 5. Before the catalytic cycle is
initiated, the catalyst is mixed with the alkene, resulting in a
reversible formation of a Cu(I)-alkene complex. Such com-
plexes are well-known and the structure of a complex between
styrene and a Cu(I)-complex of an analogue of 3 has recently
1
mL plastic syringe (i.e., cut off near the position of the 0 mL mark)
containing PhINTs. The apparatus is flushed with argon and the PhINTs
addition may in this way be efficiently controlled. The internal temperature
is monitored via a thermocouple probe, and by vigorous stirring, the
temperature could easily be kept between +0.5 and 1.5 °C, variations, which
were found not to interfere with the results at this level of precision.
2
0
been determined (see also Figure 3). The calculations on the
small system 4a show a strong binding of ethylene to copper
by 30.5 kcal/mol (BSI). This complexation energy is dramati-
cally reduced (to 21.5 kcal/mol) by introducing steric bulk at
(23) Yamada, Y.; Yamamoto, T.; Okawara, M. Chem. Lett. 1975, 361.

8016 J. Am. Chem. Soc., Vol. 122, No. 33, 2000
Brandt et al.
Figure 2. Potential free energy surface of the small model system 4a (BSIII/298 K).
Figure 3. Styrene complex 5c highlighting the Cu(I)-chlorine
attraction.
Figure 4. Cu-[N-(methylsulfonyl)imino]phenyliodinane complex 7a.
the imine and increasing the size of the alkene as in system 4b.
Further, in system 4c, the complexation energy drops to only
the triplet state results in immediate dissociation of iodobenzene
forming triplet Cu-nitrene 8.
1
2.5 kcal/mol because of a favorable copper-chlorine interac-
Cu-Nitrene 8. It is generally accepted that metal-bound
nitrenes are involved in the Cu/PhINTs aziridination reaction
of alkenes as evidenced by the equal selectivity in reactions
using PhINTs or photogenerated TsN-nitrene from TsN3.12
Jacobsen and co-workers have also convincingly demonstrated
that the PhI moiety of PhINTs does not affect the outcome of
the reaction. Hence, a Lewis-acid mechanism involving PhI in
tion. The entropic and solvation gain upon dissociation is
expected to negate this small energy difference.24
Copper-PhINTs Complex 7. Addition of PhINTs to the
catalyst/alkene mixture results in a fast aziridination reaction,
starting with decomplexation of the alkene and coordination of
PhINTs to the catalyst. Binding of the iminoiodinane is
considerably stronger than ethylene coordination, 44.0 vs 31.8
kcal/mol for the small system 4a. According to the B3LYP
calculations, the NSO2-moiety shows a strong preference for
an N,O-bidentate coordination to the Cu-diimine complex and
this N,O-binding mode was used throughout the investigation
12
the selectivity-determining TS could be ruled out. It may thus
be concluded that even though the Cu-[N-(methylsulfonyl)-
imino]phenyliodinane complex 7 is very nitrene-like, this species
is not reactive enough to attack the alkene before dissociation
of PhI.
Figure 4).25 It should be noted at this point that this coordination
(
Aziridination of cis-substituted styrenes often gives significant
amounts of trans-aziridines, which indicates that a radical
pathway could be involved. This is supported by the Hammet
creates a chiral center at sulfur, which might be of importance
in asymmetric versions of the reaction. In this complex, the N-I
bond of the iodinane is elongated from 2.11 to 2.50 Å, indicating
a very weak bond. The barrier for breaking this bond equals
the bond dissociation energy (5.0 kcal/mol at B3LYP/BSIII),
thus a real transition state is missing. Considering the favorable
free energy of the N-I bond-breaking process (see Figure 2),
1
3
study of P e´ rez. In contrast to these observations, experiments
performed by Evans using an aliphatic radical trap as substrate
did not show any indications of radical intermediates.10 These
facts indicate that a concerted pathway exists but do not rule
out a radical pathway for certain stabilized substrates.
-
3.6 kcal/mol, the Cu-PhINTs complex could not be the
The singlet and triplet states of the nitrene differ very slightly
in energy, for system 4a only by 0.1 kcal/mol; hence, a
branching radical pathway could be envisioned. According to
the B3LYP/BSI energies, the triplet states seem to be more
stable for the larger systems 4b and c (Tables 1 and 2). The
singlet state of nitrene 8a shows a short Cu-N distance and a
square planar coordination of the ligands as would be expected
resting state for the catalytic cycle. Excitation of this species to
(
24) The entropic contribution to the free energy of dissociation is around
6
-8 kcal/mol. The solvation cannot be reliably estimated for a metal-
containing species by any methods available to us, but the increase in surface
area and coordination of a solvent molecule to Cu will definitely provide
several kcal/mol of stabilization.
(25) No monodentate nitrenes were ever located in this study.

Studies of Copper-Catalyzed Alkene Aziridination
J. Am. Chem. Soc., Vol. 122, No. 33, 2000 8017
Figure 5. Cu-nitrene 8a; singlet (top) and triplet (bottom). Selected
Mulliken spin densities (from B3LYP/BSIII) are shown in italics for
the triplet.
Figure 7. Two spin states of intermediate 10a. Mulliken spin densities
from B3LYP/BSIII in italics.
force for these systems (Table 2), it is reasonable to assume
even lower activation energies.
Intermediate Carbon-Centered Radical 10. The interme-
diacy of an ethyl radical is probably a prerequisite for cis/trans
isomerization in the aziridination of cis-alkenes. The triplet state
of this species has one unpaired electron at Cu and one at the
terminal carbon of the former alkene. This biradical was found
to be stable with respect to ring-closure as expected from the
very high singlet-to-triplet excitation energy for product 11 (vide
infra). Thus, an efficient isomerization of cis-alkenes could be
expected for this state. The structure together with Mulliken
spin densities of system 10a is shown in Figure 7.
An alternative electronic state for this intermediate would be
a biradical singlet. Most interestingly, this wave function could
be converged, and the energy was found equal to that of the
triplet state. The reaction may thus take place in a biradical
fashion without the involvement of spin crossing. The TS for
the formation of this singlet biradical could not be located due
to convergence problems, but considering the equal energy of
the singlet and triplet nitrenes as well as of the two ethyl radicals
it is reasonable to assume a similar activation energy for the
open-shell singlet pathway.
Figure 6. The triplet TS for the reaction between 8a and ethene.
Mulliken spin densities from B3LYP/BSIII in italics.
8
of a formal d Cu(III) species. The restricted closed-shell singlet
wave function showed no instability toward unrestricted solu-
tions. The triplet state of nitrene 8a has an elongated Cu-N
bond and a slightly distorted square planar coordination
9
geometry, thus resembling a formal d Cu(II) complex. Struc-
tures are shown in Figure 5 together with Mulliken spin densities
for the triplet. In systems 4b and c, these species are destabilized
in a way similar to that for the alkene complex 5, but still, no
minimum with monodentate coordination (i.e., lacking the
Cu-O bond) could be located.
Transition State for the Formation of Nitrogen-Carbon
Bond(s) TS1. For the addition of ethylene to the Cu-nitrenes
8
a, barriers can only be found on the unrestricted PES’s. This
is not surprising, considering the very large driving force for
the reaction at the closed-shell PES (47 kcal/mol using BSI).
On the triplet surface, the presence of an intermediate ethyl
radical 10 reduces the driving force, and an early TS could be
located at an N-C distance of 2.26 Å (Figure 6). This TS is
slightly lower in energy than the free reactants, indicating the
presence of a molecular complex (Table 1). Calculating the
activation energy from this van der Waals complex, the barrier
was found to be 2.5 kcal/mol using BSI, and insignificantly
higher (2.7 kcal/mol) using BSIII. The free energy of activation
calculated from the triplet nitrene is 12.7 kcal/mol. TS optimiza-
tions of systems 2b and c could not be performed due to the
large size of these systems. Considering the increased driving
The Transition State for Ring Closure, TS2. Due to the
presence of the singlet biradical intermediate 10, we could locate
a transition state for the subsequent ring closure to form the
final aziridine product. The TS was found by means of a scan
where we started from structure 10 and the corresponding
UB3LYP singlet wave function and decreased the N-C-C
angle. This resulted in an activation energy of only 1.2 kcal/
mol at BSI, a barrier that most likely will be shifted by radical
stabilizing substituents (e.g., phenyl) on the alkene. Ring closure
at the triplet surface is efficiently hindered by the high
endothermicity of this process (Figure 8).
The Catalyst Bound Aziridine Product 11. The formation
of the aziridine is accomplished without loss of coordination to

8018 J. Am. Chem. Soc., Vol. 122, No. 33, 2000
Brandt et al.
Scheme 3. Catalytic Cycle Used in the Derivation of
Kinetic Rate Laws
Figure 8. The transition state for ring-closure, TS2.
workers,28 exploring the Jacobsen epoxidation reaction by means
of hybrid density functional theory (B3LYP), the isomerization
process has been rationalized in terms of the location of an
obligatory spin crossing. Thus, the formation of cis-products is
accompanied by an early triplet-quintet spin crossing taking
place before the isomerization of an ethyl radical intermediate.
These arguments are not readily applicable to the copper-
catalyzed aziridination, as the reactant Cu-nitrene 8 exhibits
isoenergetic singlet and triplet states. This implies that the
aziridination reaction does not actually require the involvement
of a spin crossing; the singlet surface shows overall low
activation energies and favorable thermodynamics. The fact that
cis/trans isomerization is still observed for some substrates might
reflect the fact that, for larger systems, the triplet Cu-nitrene
is instead the spin state lowest in energy. This is indicated by
the fact that the triplet-singlet splitting at the B3LYP/BSI level
is increased for the larger systems 4b and c (vide supra). In
this case, the same arguments will apply to both the copper-
catalyzed aziridination and the Jacobsen epoxidation of cis-
Figure 9. Cu-aziridine complex 11a.
copper, leading to complex 11. This species has a singlet ground
state, with a vertical excitation energy of 67 kcal/mol to the
triplet state (B3LYP/BSI, 11a (Figure 9)). The bond energy for
the Cu-aziridine complex is 25 kcal/mol, reduced to 15 kcal/
mol for the larger system 11c due to attractive interactions
between copper and the chlorine substituents. As for the alkene
complex 5, the free energy of binding cannot be reliably
predicted, mainly due to the difficulties in determining the
solvation contribution. From the computational results, either
26
alkenes. However, in contrast to the Jacobsen epoxidation,
cis/trans isomerization might also occur in the singlet biradical
of 10 for the aziridination reaction.
5
, 6 (with coordinated solvent molecules), or 11 could be
Kinetics. A postulated catalytic cycle used in the discussion
of the kinetics is depicted in Scheme 3. The computational
results indicate that the rate-determining step (rds) in the reaction
is formation of the copper-nitrene 8. This result is seemingly
in conflict with the reported first-order dependence on the alkene
plausible resting states of the catalytic system. The actual resting
state has important implications for the kinetics. A strong
coordination of aziridine to copper would result in product
inhibition, providing an alternative rationalization for the
1
2
apparent first-order alkene dependence reported previously
vide infra). In the case of the analogous copper-catalyzed
12
concentration. A true first-order reaction would result from a
(
rate-limiting addition of alkene to 8, forming 11 (equation 1; x
is used for the alkene concentration). Postulating steady-state
concentration of the metal nitrene 8 and integrating gives the
standard expression for a first-order linear plot (equation 2; x0
is the initial alkene concentration).
cyclopropanation reaction, it has been shown that the reaction
is inhibited by addition of bipyridine.26 This was interpreted in
terms of a ligand dissociation preequilibrium before the copper-
ethyldiazoacetate interaction.
Cis/Trans Isomerization. Cis/trans isomerization in the
copper-catalyzed aziridination of cis-â-methylstyrene is well-
∂[aziridine]
∂x
∂t
4
r )
) - ) k [8]x; x ) [alkene] (1)
known experimentally, and the close resemblance between this
2
∂t
reaction and Jacobsen epoxidation of cis-alkenes2 calls for a
comparison of the two reactions. In the epoxidation reaction,
isomerization has been suggested to be a consequence of an
ethyl radical intermediate which rotates before collapsing to the
epoxide, and the cis/trans ratio of the products simply reflecting
the partition between the possible pathways, direct collapse (to
give cis-epoxide) and rotation/collapse (leading to trans-
7
x0
ln ) k [8]t
(2)
2
x
Our calculations do not support the suggestion that the
reaction of 8 with alkene is rate-limiting (Table 1). Furthermore,
a reaction truly first-order in alkene would require either a
reversible formation of 8 or a buildup of 8 in the reaction
mixture. In the case of a catalytic cycle starting with a Cu(I)-
diimine complex, the reaction between Cu-aziridine complex
11 and PhINTs is calculated to be exothermic by ∼8 kcal/mol
for systems 4b and c and is further favored by entropic
contributions due to the increased molecularity. Thus, in
epoxide).9 However, both epoxidation and aziridination
1d
10
experiments using phenyl-substituted radical traps give reason
to believe that non-isomerized products are formed by a
concerted mechanism. In a recent study by Svensson and co-
(
26) D ´ı as-Requejo, M. M.; Belderrain, T. R.; Nicasio, M. C.; Prieto, F.;
P e´ rez, P. J. Organometallics 1999, 18, 2601.
27) Zhang, W.; Lee, N. H.; Jacobsen, E. N. J. Am. Chem. Soc. 1994,
16, 425.
(
(28) Linde, C.; Åkermark, B.; Norrby, P.-O.; Svensson, M. J. Am. Chem.
Soc. 1999, 121, 5083-5084.
1

Studies of Copper-Catalyzed Alkene Aziridination
J. Am. Chem. Soc., Vol. 122, No. 33, 2000 8019
Table 3. Kinetic Data for the Aziridination of
1
,2-dihydronaphthalene Using Ligand 3 and Cu(CH
3
CN)
4 6
PF to
Generate the Catalyst
[
alkene]t)0
[aziridine]t)0
[Cu]Tot
(mM)
initial rate/[Cu]Tot
-
1
(mM)
(mM)
s
6
5
0
0
0
10
79
3.1
3.1
6.2
3.1
3.1
0.22 ( 0.02a
0.20
0.21
0.17
0.26
1
1
28
25
6
6
4
7
a
Standard deviation, experiment performed four times.
combination with the low calculated barrier for the further
reaction between nitrene and alkene, formation of 8 is expected
to be effectively irreversible.
Figure 10. Plots of ln(x
concentrations x and metal concentrations m and also with added
aziridine product.
0
/x) vs mt for two different initial alkene
0
An irreversible formation of 8 in combination with a rate-
limiting reaction between 8 and alkene would imply that 8 is
the resting state of the catalyst. It is our experience that complex
experimental uncertainty) is observed. To illustrate the observed
behavior, all data points are plotted according to eq 2 in Figure
6
will consume PhINTs rapidly even in the absence of alkene,
1
0.
It is clear from Figure 10 that in the absence of added
presumably by a reaction of 8 with other species present in the
reaction mixture (solvent, PhINTs, or more 8). Furthermore,
the barrier for the reaction between the alkene and the copper-
nitrene is extremely low for both the singlet and the triplet state.
The mechanism suggested by the calculations is instead that
the rds is formation of the copper-nitrene 8, with either free
copper-ligand complex 6, product complex 11, or alkene
complex 5 as the resting state (the latter can be excluded based
on the observed kinetics). It should be noted, however, that the
selectiVity-determining step, affecting the ratio of aziridine
isomers, must by necessity involve the alkene.
aziridine, the fit to apparent first order is very good within each
run but varies proportionally to the initial concentration of alkene
(the slope is halved by doubling the alkene concentration).
However, the reactions with added aziridine fit less well. It can
be seen that the catalyst is deactivated more rapidly at a higher
aziridine concentration. It has been verified that the aziridine
is stable to the reaction end conditions. Together with the
observed catalyst deactivation upon aziridine addition (Figure
1
0) this indicates that the active catalyst is somehow decom-
posed by the aziridine into a form that will neither give
aziridination nor aziridine breakdown. Such a decomposition
upon increasing aziridine concentration might also rationalize
the apparent first order in alkene observed for each separate
run.
In a simplified view, a rds that does not involve the alkene
should result in a zero-order rate law, in conflict with observa-
tions. However, the observations only show that some essential
reactant is depleted, not necessarily the alkene. Strong com-
plexation of copper by a product or byproduct in the reaction
or catalyst decomposition could lead to the same apparent
behavior. To test whether the rate-determining step of the
reaction does involve the alkene, we studied the kinetics of the
reaction while varying the initial conditions. Due to insolubility,
the effective concentration of PhINTs (and of course the
molecular sieves) could not be controlled. Every effort was made
to keep the reaction conditions as constant as possible, to
minimize the variation in effective concentration of these
reactants. Alkene and metal-ligand complex were tested at two
different concentrations each. To test the hypothesis of inhibi-
tion, aziridine product was also added in two different concen-
trations. In all cases, the reaction displayed apparent first order
in alkene, at least in the initial stages of the reaction. Apparent
first-order rate constants were determined by fitting to eq 2.
Points at the end of the run (where the catalyst had presumably
decomposed) were discarded, on the basis of maximizing the
information content according to an F-test. From the apparent
rate constants initial rates were calculated by multiplication with
the initial alkene concentration. The initial rates divided by the
total metal concentration are listed for all reactions in Table 3.
Conclusions
The mechanism of the copper-catalyzed aziridination of
alkenes has been elucidated by high-level quantum chemical
calculations. The calculations strongly indicate that the active
catalyst is a Cu(I) species, as suggested by Jacobsen.12 It has
also been shown how Cu(II) can enter the Cu(I)/Cu(III) cycle
through reaction with PhINTs.
In the calculations, three different spin states have been
characterized along the reaction path; a closed-shell singlet
surface that leads directly to aziridine in a concerted manner
from the singlet nitrene 8, an open-shell singlet pathway that
could lead to an extremely short-lived N-ethyl singlet biradical
intermediate, and finally, a triplet surface leading to an N-ethyl
radical intermediate that has to change spin to singlet before
ring-closure. The resting state of the catalyst is a copper(I)
complex, possibly with coordinating solvent molecules, but not
involving either the product aziridine or the alkene. The
calculated reaction profile suggests that the rate-determining step
in the reaction is the formation of metallanitrene 8, in conflict
1
2
with a previous report.
The data in Table 3 conclusively shows that the initial reaction
rate is independent of alkene concentration and directly pro-
portional to the metal concentration, in accordance with the
calculated reaction path. The effect of aziridine product addition
is not entirely clear, but from the last result, it is obvious that
the apparent first order is not a result of product inhibition. If
product inhibition was effective, the last entry should correspond
approximately to entry 2 at 50% conversion, with about half
the initial rate, but instead a slight rate increase (probably within
Kinetic measurements confirmed that the reaction is indeed
zero order in alkene, despite the fact that single experiments
appear to obey a first-order kinetics. This behavior could be
rationalized in terms of decomposition of the catalytically active
metal complex.
Acknowledgment. We are indebted to professor David
Tanner for many helpful discussions. Financial support from
the Danish Technical Research Council (STVF), the Swedish

8020 J. Am. Chem. Soc., Vol. 122, No. 33, 2000
Brandt et al.
Natural Science Research Council (NFR), the Foundation for
Strategic Research (SSF), the Swedish Research Council for
Engineering Sciences (TFR), the Swedish Foundation for
International Cooperation in Research and Higher Education
and Parallelldatorcentrum (PDC), Royal Institute of Tech-
nology.
Supporting Information Available: Results and discussion
for dicationic complexes (PDF). This material is available free
of charge via the Internet at http://pubs.acs.org.
(STINT) (postdoctoral fellowship for P.B.) is gratefully ac-
knowledged. Access to supercomputer time was provided by
the Swedish Council for High Performance Computing (HPDR)
JA993246G