The asymmetric aza-claisen rearrangement: development of widely applicable pentaphenylferrocenyl palladacycle catalysts

Article abstract of DOI:10.1002/chem.200900712

Systematic studies have been performed to develop highly efficient catalysts for the asymmetric aza-Claisen rearrangement of trihaloacetimidates. Herein, we describe the stepwise development of these catalyst systems involving four different catalyst generations finally resulting in the development of a planar chiral pentaphenylferrocenyl oxazoline palladacycle. This complex is more reactive and has a broader substrate tolerance than all previously known catalyst systems for asymmetric aza-Claisen rearrange-ments. Our investigations also reveal that subtle changes can have a big impact on the activity. With the enhanced catalyst activity, the asymmetric aza-Claisen rearrangement has a very broad scope: the methodology not only allows the formation of highly enantioenriched primary allylic amines, but also secondary and tertiary amines; al-lylic amines with N-substituted quaternary stereocenters are conveniently accessible as well. The reaction conditions tolerate many important functional groups, thus providing stereoselective access to valuable functionalized building blocks, for example, for the synthesis of unnatural amino acids. Our results suggest that face-selective olefin coordination is the enantioselectivitydetermining step, which is almost exclusively controlled by the element of planar chirality.

Full text of DOI:10.1002/chem.200900712

DOI: 10.1002/chem.200900712
The Asymmetric Aza-Claisen Rearrangement: Development of Widely
Applicable Pentaphenylferrocenyl Palladacycle Catalysts
Daniel F. Fischer,[b] Assem Barakat,[a, b] Zhuo-qun Xin,[a, b] Matthias E. Weiss,[b] and
Renꢀ Peters*[a]
8722
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 8722 – 8741

FULL PAPER
Abstract: Systematic studies have been
performed to develop highly efficient
catalysts for the asymmetric aza-Clais-
en rearrangement of trihaloacetimi-
dates. Herein, we describe the stepwise
development of these catalyst systems
involving four different catalyst genera-
tions finally resulting in the develop-
ment of a planar chiral pentaphenylfer-
rocenyl oxazoline palladacycle. This
complex is more reactive and has a
broader substrate tolerance than all
previously known catalyst systems for
asymmetric aza-Claisen rearrange-
ments. Our investigations also reveal
that subtle changes can have a big
impact on the activity. With the en-
hanced catalyst activity, the asymmetric
aza-Claisen rearrangement has a very
broad scope: the methodology not only
allows the formation of highly enan-
tioenriched primary allylic amines, but
also secondary and tertiary amines; al-
lylic amines with N-substituted quater-
nary stereocenters are conveniently ac-
cessible as well. The reaction condi-
tions tolerate many important function-
al groups, thus providing stereoselec-
tive access to valuable functionalized
building blocks, for example, for the
synthesis of unnatural amino acids. Our
results suggest that face-selective olefin
coordination is the enantioselectivity-
determining step, which is almost ex-
clusively controlled by the element of
planar chirality.
Keywords: allylic compounds
amines aza-Claisen rearrange-
ment · cyclopalladation · palladium
·
·
Introduction
Soft Lewis acids such as PdII, PtII, or AuI have become more
and more important for organic synthesis, because coordina-
tion of these carbophilic late-transition-metal cations to ole-
fins or alkynes results in a net transfer of electron density to
the metal center, thus activating the unsaturated system for
the attack by nucleophiles.[1] The soft Lewis acid can there-
fore be regarded as a chemoselective (owing to its low oxo-
philicity) and possibly chiral proton substitute.
For example, the PdII-catalyzed aza-Claisen rearrange-
ment[2] of allylic trichloro-[3] and trifluoroacetimidates[4] ena-
bles the transformation of achiral allylic imidates 2, readily
prepared in a single step from allylic alcohols 1, to chiral
enantioenriched allylic amides 5 (Scheme 1). Since the triha-
loacetamide-protecting groups can be readily removed, the
overall transformation leads to valuable allylic amine build-
ing blocks 6. There is ample evidence that these reactions
proceed by means of a cyclization-induced rearrangement
mechanism, in which the olefin moiety of 2 coordinates to
the PdII complex and is thereby activated in 3 for a nucleo-
philic attack by the imidate N atom.[5] In 1974, 37 years
after the first description of the thermal aza-Claisen rear-
rangement of allylic benzimidates by Mumm and Mçller,[6]
Overman reported the first application of trichloroacetimi-
date substrates and demonstrated that the rearrangement
can be catalyzed by soft Lewis acids such as HgII.[7]
Scheme 1. The aza-Claisen (Overman) rearrangement of trihaloacetimi-
dates 2.
catalyst for the rearrangement of trihaloacetimidates, the
planar chiral oxazoline-based palladacycle COP-X (7, X=
Cl).[3a,4a] This catalyst was found to be superior to its ferro-
cene analogue FOP-X (8)[4c] in a number of aspects such as
preparation (i.e., FOP-X could not be prepared by direct cy-
clopalladation due to oxidative decomposition by PdII, but
required two low-temperature lithiations), catalyst stability,
catalytic activity, and substrate scope.[8]
In 2003, Overman, Richards, and co-workers disclosed an
additional breakthrough: the first highly enantioselective
[a] A. Barakat, Z.-q. Xin, Prof. Dr. R. Peters
Institut fꢁr Organische Chemie, Universitꢂt Stuttgart
Pfaffenwaldring 55, 70569 Stuttgart (Germany)
Fax : (+49)711-685-64321
E-mail: rene.peters@oc.uni-stuttgart.de
[b] Dr. D. F. Fischer, A. Barakat, Z.-q. Xin, M. E. Weiss
ETH Zꢁrich, Hçnggerberg, Laboratory of Organic Chemistry
Wolfgang-Pauli-Strasse 10, 8093 Zꢁrich (Switzerland)
Unfortunately, in the case of trifluoroacetimidate sub-
strates, which appear to be synthetically slightly more attrac-
tive than the corresponding trichloro derivatives due to
milder deprotection conditions for the resulting amide,
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200900712.
Chem. Eur. J. 2009, 15, 8722 – 8741
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8723

R. Peters et al.
COP-Cl still required loadings that are not attractive for
large-scale applications (ꢀ10 mol% PdII) and long reaction
times were necessary for high conversion. Moreover, the
scope was limited to substrates bearing a-unbranched alkyl
substituents (R’) at the 3-position of the allylic imidate.
To fully capitalize on the potential of the synthetically at-
tractive allylic imidate rearrangement concept, we have sys-
tematically addressed the issues of catalytic activity and
scope, ultimately resulting in the development of highly
active chiral catalysts for the rearrangement of trifluoro-
and trichloroacetimidates, thereby providing allylic trihalo-
acetamides with excellent enantioselectivities. In this full
paper, we describe the stepwise development of these cata-
lyst systems involving four different catalyst generations.
Results and Discussion
The previous investigations pioneered by the Overman
group had shown that planar chirality is in principle an ef-
fective means of achieving high enantioselectivities. Our in-
vestigations were based upon the assumption that a highly
modular design of a carbophilic Lewis acid system would ar-
guably allow for an adjustment of electronic and steric prop-
erties over a wide range but also for a final fine-tuning for
ultimate optimization. Although all ferrocene-based cata-
lysts such as FOP-X were previously not as efficient as
COP-X, in the present study ferrocene moieties were
chosen as the backbone for planar chiral catalyst systems for
two main reasons: 1) the cobalt-based sandwich complexes
proved to be very sensitive with regard to the cyclopallada-
tion outcome depending on their substitution pattern in
terms of reactivity and stereochemistry, thus limiting the
number of accessible catalyst structures;[4b] 2) for the cobalt-
based sandwich backbone the electron density could not be
extensively variegated, in contrast to ferrocenes, for which,
for example, the oxidation potentials are strongly dependent
upon the substitution pattern. Along these lines, imidazoline
rather than oxazoline coordination sites fit better into a
modular concept to understand how the electronic proper-
ties ideally should be, since electron density can be readily
adjusted by the choice of the amino-N substituent.
Scheme 2. Synthesis of N-methyl 2-ferrocenyl imidazoline palladacycles
14 and 15, 19 and 20.
gel column under the exclusion of light. Oxidative addition
First- and second-generation catalysts: The investigations
started with catalyst systems closely related to FOP-X so as
to study the effect of an electron-rich donor system in N-
methyl 2-ferrocenyl imidazolines. Since, analogously to FOP,
all attempts failed to obtain the corresponding palladacycles
of [Pd2ACTHUNGTRENNU(G dba)3] (dba=dibenzylideneacetone) provided
iodide-bridged palladacycles 14 and 15 in 54 and 42% over-
all yield, respectively.
Unfortunately, activating 14 and 15 with silver salts such
as silver trifluoroacetate (AgTFA) for the rearrangement of
N-PMP (p-methoxyphenyl)-trifluoroacetimidates resulted in
low enantioselectivities and poor reproducibility, which was
ascribed to insufficient catalyst stability (formation of Pd
black), whereas the iodide bridged dimers were catalytically
almost inactive without activation.
The sterically more demanding pentamethylferrocenyl
(Fc*) imidazoline palladacycle 19 was not only expected to
allow for a more pronounced steric discrimination of the
space above and below the Pd square plane, but also to pro-
by direct cyclometallation with PdII salts as both Na
and Pd(OAc)2 gave practically no diastereoselectivity, palla-
2ACHTUNGTNER[NUNG PdCl4]
ACHTUNGTRENNUNG
dacycles 14 and 15 were prepared by a two-step procedure
(Scheme 2). tBuLi and a combination of tBuLi and lithium
diisopropylamide (LDA) allowed for diastereoselective
ortho-lithiations,[9] and subsequent reaction with 1,2-diiodo-
ethane as iodinating agent provided selective access to both
planar chiral diastereomers 12 and 13 in the form of air- and
light-sensitive solids that had to be purified by using a silica
8724
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 8722 – 8741

Asymmetric Aza-Claisen Rearrangement
FULL PAPER
vide a relatively stable ferrocenium ion generated by the
silver-activating agent (kinetic stability due to shielding of
FeIII by C5Me5 (Cp*) and thermodynamic stability due to
the electron-donating effect of the five methyl groups). As
PdII salts oxidatively decomposed Fc* imidazoline ligand 17
or resulted in low diastereoselectivity for cyclopalladations,
19 was also prepared by means of a lithiation procedure
(Scheme 2).[4d] The diastereomeric ratio (dr) of 19:1 for iodo
derivative 18 was improved to >99:1 in the iodide-bridged
palladacycle complex. Treatment of 19 with PPh3 led to the
diastereomerically pure monomeric species 20, the structure
of which was determined by X-ray crystallography
(Scheme 2), thereby revealing the Sp configuration with
regard to the planar chirality.[10,11] The observed configura-
tion is in harmony with a mechanistic model presented by
Richards et al. for ortho-lithiations of ferrocenyl oxazo-
lines.[12] The Pd center adopts a distorted square-planar co-
ordination geometry, and the PPh3 ligand is exclusively posi-
tioned trans to the imino-type nitrogen atom.
The absolute configuration was assigned by NOESY ex-
periments of 24 (see the Supporting Information) because
suitable crystals could not be obtained. Like lithiations of
the Fc* imidazolines 17,[4d] the cyclopalladation reaction
occurs in a way that the phenyl group at the imidazoline 4-
position points toward the spectator Cp ligand (endo).
Complexes 19 and 23 were examined in the aza-Claisen
rearrangement of allylic trifluoroacetimidates 25 (Scheme 4)
after activation with AgTFA or AgNO3 in the presence of
In addition, we were interested in sterically even more de-
manding, yet less electron-rich pentaphenylferrocenyl (FcF)
systems like 23 (Scheme 3), which should be more resistant
Scheme 4. Aza-Claisen rearrangement of trifluoroacetimidates 25 using
23 as precatalyst, which presumably generates 27 as the catalytically
active species.
substoichiometric amounts of proton sponge (PS, N,N,N’,N’-
tetramethyl-1,8-diaminonaphthalene) to prevent decomposi-
tion of the substrate by traces of acid.[4c] AgNO3 generally
provided a more active species than AgTFA, and 6 equiv
AgI per palladacycle dimer gave the best results. Whereas
Fc* derivative 19 (10 mol%, reaction time 20 h at 408C in
CH2Cl2) gave at best only 50% yield and a moderate ee
value of 70% for model substrate 25a (R=nPr), 5 mol% of
FcF derivative 23 provided good enantioselectivities and ex-
cellent yields under the same conditions.[15]
Activation of the precatalyst dimer by AgNO3 is assumed
to take place as follows: 2 equiv per dimer are necessary to
exchange both of the strongly binding Cl anions by weakly
binding nitrate. Another 2 equiv oxidize the two Fe centers
from FeII to FeIII, thus leading to a far less electron-rich
(paramagnetic) ligand and resulting in a stronger Lewis
acid. The remaining Ag ions might partly oxidize the imida-
zoline moieties to form imidazoles as donor (as in 27 in
Scheme 4), which renders the Pd centers even less electron-
rich.
Scheme 3. Synthesis of N-methyl 2-(pentaphenyl)ferrocenyl imidazoline
palladacycles 23 and 24 by direct diastereoselective cyclopalladation.
towards oxidative decomposition. All attempts to ortho-lith-
iate 22 failed and the starting material was always recovered
unchanged. However, unlike Fc* derivative 17, 22 could be
directly and diastereoselectively ortho-palladated with Na2-
ACHTUNGTRENNUNG[PdCl4]/NaOAc furnishing the air-stable palladacycle 23 as a
dimeric chloride-bridged complex. To determine the diaste-
Despite the fact that the diastereomeric ratio of the pre-
catalyst was only 16:1 (i.e., diastereomeric excess (de) =
88%), enantioselectivity is only slightly lower than that for
COP-Cl.[4a] However, the catalyst has two main technical
disadvantages: 1) it could not be generated in diastereomeri-
cally pure form by a more selective metalation or by crystal-
lization or chromatography to obtain very high enantioselec-
tivities; 2) the diamines 10a/b used for imidazoline forma-
tion are not commercially available.
reoselectivity of the cyclopalladation, 23 was treated with an
excess of Na
ACHTUNGTRENNUNG(acac) (acac=acetylacetonate), thereby leading
to the monomeric complex 24 (dr=16:1).[13,14]
Chem. Eur. J. 2009, 15, 8722 – 8741
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8725

R. Peters et al.
Third-generation catalysts: Because these initial two catalyst
generations had shown that reduced electron density of the
ligand results in higher stability, activity, and enantioselectiv-
ity, a third-generation catalyst was targeted in which a
strongly electron-withdrawing N-sulfonyl residue is the
structural key element. The Pd center consequently displays
a higher Lewis acidity while the ferrocene body is more
stable toward oxidative decomposition. In addition, the sul-
fonyl group permits a diastereoselective direct cyclopallada-
tion of ligand systems 28: steric repulsion between the resi-
due R1 at the imidazoline 5-position and the sulfonyl group
effects a transfer of chirality by transmitting stereoinforma-
tion to the sulfonylated nitrogen atom (Scheme 5), which
Scheme 6. Modular concept of N-sulfonyl 2-ferrocenyl imidazoline palla-
dacycles 29.
zoline part synthesized from an enantiomerically pure C2-
symmetric diamine 30, 3) the sulfonyl residue, 4) the coun-
teranion XÀ, and 5) the oxidation state of the Fe ion (ferro-
cene versus ferrocenium).
The air-stable and crystalline imidazolines 28 were pre-
pared in good yield from carbamides 33[4d] by activation by
means of the corresponding iminium ether intermediate and
a subsequent sulfonylation with various commercially avail-
able sulfonyl chlorides (Scheme 7, Table 1).
Scheme 5. Top: Conformational equilibrium for 28 caused by chirality
transfer to the sulfonylated N atom and steric repulsion between the sul-
fonyl residue and the Cp spectator ligand. Bottom: Confirmation by X-
ray crystal structure analysis. Due to crystal packing effects, two N–S ro-
À
tamers are formed, in which the C S bond is either syn- (left) or anti-per-
iplanar (right) to the N-atom lone pair. Both species adopt a similar con-
À
formation around the C(imidazoline) Cp bond.
Scheme 7. Synthesis of N-sulfonyl ferrocenyl imidazoline palladacycles 29
by direct diastereoselective cyclopalladation (p-Tol=p-toluene; 1-Naph=
1-naphthyl).
adopts a significantly pyramidalized geometry and results in
a preferred conformation in which the sulfonyl group points
away from the ferrocenyl floor and, as a result, allows for a
diastereoselective cyclopalladation. In contrast to the direct
cyclopalladation of FcF imidazoline 22, the stereochemical
outcome can thus be explained by the reaction of the ther-
modynamically most stable conformer.
Yields for the formation of ligands 28 were comparable
for all imidazolines based on 1,2-diphenylethylenediamine
regardless of the size of the Cp’ spectator ligand (Table 1,
entries 1–3 and 5–8), whereas the more bulky 1,2-di-tert-bu-
The hypothesis of chirality transfer to the N-sulfonyl
group was confirmed by X-ray crystal structure analysis
(Scheme 5) of imidazoline 28-Cp-tBu-Ts (numbering
system: 28-(C5R5)-R1-(SO2R2); Ts=tosyl).[16] The sulfonylat-
ed N atom is significantly pyramidalized, thus minimizing
unfavorable steric interactions with the neighboring sub-
stituents.
The third-generation palladacycles 29 are created based
on a modular design that allows adjustment of both the
steric and the electronic properties of the Lewis acid by five
different modules (Scheme 6):[4e] 1) the Cp’ spectator ligand
C5R5 (C5H5 =Cp, C5Me5 =Cp*, C5Ph5 =CpF), 2) the imida-
tylethylenediamine resulted in
entry 4).
Cyclopalladation at RT with Na2AHCTUNGTERG[NNUN PdCl4]/NaOAc in MeOH
a
poor yield (Table 1,
(plus benzene if the solubility of the imidazoline is insuffi-
cient in MeOH) provided the air-stable dimeric complexes
29 in good yield and with high diastereoselectivity with
regard to the planar chirality (dr=7:1–20:1 and 9:1–83:1
before and after chromatography, respectively). The cyclo-
palladations usually proceeded best for Cp’-unsubstituted
imidazolines (Table 1, entries 9–11), thereby providing high
yields with the exception of 28-Cp-tBu-Ts carrying two tBu
substituents on the imidazoline backbone (Table 1,
8726
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 8722 – 8741

Asymmetric Aza-Claisen Rearrangement
FULL PAPER
Table 1. Preparation of imidazolines 28 and palladacycles 29.
Interestingly, upon addition of PPh3 in CDCl3, Fc*- and
FcF-complexes 29 rapidly provided a mixture of monomeric
isomers with the phosphine in a cis and trans position to the
N-donor atom (Scheme 8), but completely isomerized after
several hours at RT to yield, as expected from comparable
complexes, the trans isomers as a result of the so-called
trans effect.[17] Using a less bulky phosphine (MePPh2), or
working with Cp’-unsubstituted complexes 29, only one
isomer was detected directly after addition. These findings
can be rationalized by the better (thus kinetically favored)
steric accessibility of the position cis to N, which is, however,
thermodynamically less favored for p-acidic ligands.
Entry
Product
C5R5
R1
R2
Yield[a] [%]
dr[b]
dr[c]
1
2
3
4
5
6
7
8
28
28
28
28
28
28
28
28
29
29
29
29
29
29
29
29
Cp
Cp
Cp
Cp
Cp*
Cp*
Cp*
CpF
Cp
Cp
Cp
Cp
Cp*
Cp*
Cp*
CpF
Ph
Ph
Ph
tBu
Ph
Ph
Ph
Ph
Ph
Ph
Ph
tBu
Ph
Ph
Ph
Ph
CF3
72[d]
73[d]
73[d]
18[d]
71[d]
81[d]
65[d]
77[d]
92
93
91
55
69
–
–
–
–
–
–
–
–
9:1
18:1
12:1
20:1
15:1
20:1
7:1
20:1
–
–
–
–
–
–
–
–
p-Tol
1-Naph
p-Tol
CF3
p-Tol
C6F5
p-Tol
CF3
p-Tol
1-Naph
p-Tol
CF3
9
9:1
18:1
12:1
72:1
83:1
20:1
14:1
38:1
10
11
12
13
14
15
16
The structures of 35-CpF-Ph-Ts and 34-Cp*-Ph-Ts were
determined by X-ray crystal structure analysis (Figure 1),[18]
thus confirming the predicted Sp configuration for the major
diastereomer.[19,20] The same configuration was confirmed
p-Tol
C6F5
p-Tol
85
60
50
[a] Isolated yield after chromatography. [b] The dr values of the crude
product as determined by 1H NMR spectroscopy of complex 34 or 35.
1
[c] The dr values of the isolated product as determined by H NMR spec-
troscopy of complex 34 or 35. [d] Yield over 2 steps from 33.
entry 12). For reactions with incomplete conversion, most of
the unreacted ligands 28 can be readily recovered after the
reaction. To determine the ratio of diastereomers with
regard to the introduced planar chirality, the dimers 29 were
transformed into monomeric complexes 34 and 35 by treat-
ment with PPh3 or NaACTHNUTRGNEUGN(acac) (Scheme 8).
Figure 1. X-ray crystal structures of the monomeric complexes 35-CpF-
Ph-Ts and 34-Cp*-Ph-Ts.
for 35-Cp-Ph-Ts and 35-CpF-Ph-Ts, as determined by
NOESY experiments (see the Supporting Information).
Complexes 29 were then investigated in the aza-Claisen
rearrangement of N-PMP-trifluoroacetimidate model-sub-
strate 25a (Table 2). All five modules of the catalyst, the sol-
vent, and different additives were examined to find the best
conditions. The Cl-bridged dimeric palladacycle 29-Cp-Ph-
Ts was found to be nearly unreactive as catalyst (Table 2,
entry 1). Therefore, different silver salts AgX (X=TFA,
OTf, OTs, BF4, NO3; entries 2–9 in Table 2; OTf=trifluoro-
methanesulfonate) were evaluated for their catalyst-activat-
ing properties by exchanging chloride with a more labile co-
ordinating ligand.
Although AgTFA is the most general activating reagent
of those investigated (Table 2, entries 3, 6), AgOTf proved
to be superior in the case of 29-Cp catalyst precursors for Z-
configured imidates 25 (Table 2, entry 8). Treatment of 29-
Cp-Ph-Ts with 2 equiv AgTFA led to the formation of the
corresponding TFA complex, of which an 1H NMR spec-
trum could be recorded displaying a single set of signals
(one geometric isomer around the Pd square plane), where-
as addition of another 2 equiv of AgTFA led to a color
change from dark red to brown and the complete disappear-
Scheme 8. Synthesis of the monomeric complexes 34 and 35 with PPh3 or
Na(acac) to determine the diastereomeric ratio of 29 with regard to the
planar chirality.
ACHTUNGTRENNUNG
Determination of the diastereomeric ratio is complicated
at the dimer stage by the existence of geometrical isomers
around the Pd–Cl square planes resulting in up to 6 signal
sets in NMR spectroscopy ((Sp/Sp)-cis, (Sp/Sp)-trans, (Sp/Rp)-
cis, (Sp/Rp)-trans, (Rp/Rp)-cis, (Rp/Rp)-trans).
Chem. Eur. J. 2009, 15, 8722 – 8741
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8727

R. Peters et al.
Table 2. Investigation of the silver salt for precatalyst activation.
centration of catalyst. Gratify-
ingly, the activated catalyst
stays in solution even in neat
substrate.
As mentioned for the first-
and second-generation precata-
lysts, the rearrangements were
performed in the presence of
the noncoordinating, non-nucle-
ophilic base proton sponge (1–
2 equiv/dimer 5) to prevent
elimination of N-PMP-trifluor-
oacetamide triggered by traces
of acid that are possibly formed
by the hydrolysis of AgTFA
with moisture to give small
amounts of trifluoroacetic acid.
The use of PS also results in
higher ee values (higher catalyst
Entry
25a
29
AgX (x [mol%])
t [h]
T [8C]
Yield[a] [%]
ee[b] [%]
1
2
3
4
5
6
7
8
9
E
E
E
E
E
Z
Z
Z
E
29-Cp-Ph-Ts
29-Cp-Ph-Ts
29-Cp-Ph-Ts
29-Cp-Ph-Ts
29-Cp-Ph-Ts
29-Cp-Ph-Tf
29-Cp-Ph-Tf
29-Cp-Ph-Ts
29-Cp-Ph-Ts
1
–
120
48
48
24
24
24
48
24
48
20
20
20
20
40
40
20
20
20
0
5
–
AgTFA (10)
AgTFA (20)
AgOTf (20)
AgBF4 (20)
AgTFA (20)
AgOTs (20)
AgOTf (20)
AgNO3 (20)
n.d.[c]
67
56
56
73
81
91
69
88
95
66
100
76
90
92
[a] Yield determined by H NMR spectroscopy. [b] The ee values determined by HPLC after hydrolysis of 26a
to the secondary amine (see the Supporting Information). [c] n.d.=not determined.
ance of the 1H NMR spectroscopic signals, indicating the
complete oxidation of FeII to FeIII. A paramagnetic ferroce-
nium species is thus formed, which is also the catalytically
active species as the rearrangement proceeds only very
slowly employing 2 equiv of the silver salt per dimer 29.[21]
Whereas 5 mol% of 29-Cp-Ph-Ts, activated by 2 equiv of
AgTFA, catalyzed the reaction of 25a to 26a with only 5%
yield after 48 h (Table 2, entry 2), activation with 4 equiv
AgTFA led to a complex that performed the same reaction
with 88% yield (Table 2, entry 3). Similar results were ob-
tained with AgNO3 (Table 2, entry 9). Alternatively, thalli-
um(I) triflate, which is a less potent oxidizing agent as com-
pared to AgOTf, was investigated as halide scavenger. As
expected, the precatalyst treated with TlOTf was not active.
Generally speaking, the counterion should not transfer
too much electron density to the Pd center, it should also
prevent formation of kinetically inert dimers like in the case
of halide counterions. Furthermore, it should selectively
block one defined position—most likely the cis position to
the N donor—in the square-planar complex so as to achieve
selective coordination of the substrate molecules at the
same defined coordination site, most likely the trans position
to the N donor.[22] On the other side, a coordinatively unsa-
turated PdII center, as presumably formed with noncoordi-
stability) at the expense of somewhat decreased conversion
rates, which might be ascribed to a weak coordination to the
catalytically active center or a partial reduction of the ferro-
cenium core, but allowing for significantly cleaner reaction
outcomes. In addition, prior to use the activated catalyst so-
lution is usually filtered over CaH2 to trap traces of acid.
During the investigation of the sulfonyl module, the p-tol-
uenesulfonyl group emerged as the best residue of those in-
vestigated (Table 3, entries 2, 5, 9, 11, 12). Larger groups
like 1-naphthylsulfonyl lowered the reaction rate significant-
ly (Table 3, entries 3 and 6), whereas groups with a stronger
electron-withdrawing character did not result in considera-
bly higher yields or ee values (Table 3, entries 1, 4, and 10),
but in some cases (29-Cp*-Ph-Tf) led either to decomposi-
tion of the catalyst and thus a racemic rearrangement prod-
uct (Table 3, entry 8), or even impeded the cyclopalladation.
For instance, 28-CpF-Ph-Tf (R2 =CF3) did not undergo a cy-
clopalladation at all with either Na
anol (508C, 16 h), Pd(OAc)2 in benzene at 608C, or Pd-
(OAc)2 (24 h) in acetic acid at 708C (8 h).
No significant difference in enantioselectivity and yield of
2ACHTUNGTREN[NNUG PdCl4]/NaOAc in meth-
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
26a was found for the residue R1 being either Ph or tBu
(Table 3, entries 2 and 7). Because the imidazolines with
R1 =Ph are prepared in much higher yields and the required
diamine is commercially available at a reasonable price for
larger amounts,[23] they were selected for the investigation of
the influence of the cyclopentadienyl spectator ligand
module, which has the highest impact on the rearrangement
outcome.
À
nating counterions like BF4 (Table 2, entry 5), may not only
lead to an unstable Pd complex, but also to a less regioselec-
tive coordination mode.
After a solvent screening, the noncoordinating CH2Cl2
was selected for further catalyst optimization since it al-
lowed for the highest enantioselectivities and a promising
catalytic activity (general trend: 1) catalyst activity: 1,2-di-
chloroethane (DCE)>CH2Cl2 >CHCl3 >PhH>CH2Cl2 and
DMF (1%)>MeCN>Et2O (catalyst not soluble); 2) enan-
tioselectivity: CH2Cl2, CH2Cl2 and DMF (1%)>PhH>
DCE, MeCN>CHCl3). The catalyst activation is also best
performed in CH2Cl2. The active catalyst solution is subse-
quently added to neat substrate and most of the solvent is
then removed to enhance the reaction rate by a higher con-
Whereas catalysts possessing an unsubstituted Cp’ ring
did not provide useful ee levels for the E-configured model
substrate 25a (Table 3, entries 1–3, 7), the rearrangement of
(Z)-25a enabled formation of amide 26a, reaching the
90% ee benchmark (Table 3, entries 4, 5). By contrast, with
a more bulky, yet more electron-rich Cp* spectator ligand,
both (E)- and (Z)-25a gave about 90% ee (Table 3, en-
tries 9, 11) although the reaction mixtures had to be heated
to 408C to obtain useful conversion rates with a precatalyst
8728
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 8722 – 8741

Asymmetric Aza-Claisen Rearrangement
FULL PAPER
Table 3. Catalyst screening with model substrate (E)-/(Z)-25a.
not only in significantly higher enantioselectivity for (E)-
25a (ee=97%, entry 12 in Table 3) but also in a highly
active catalyst, thus allowing for a reduction of the precata-
lyst amount to an unprecedented 0.5 to 0.05 mol% without
largely effecting the enantioenrichment yet still providing
useful reaction rates (Table 3, entries 13–15). In contrast, 29-
CpF-Ph-Ts is not a useful catalyst for (Z)-25a, which was
converted in only 36% yield and 85% ee using high catalyst
loadings (Table 3, entry 16).
Having identified the most active and selective catalyst of
this series, attempts were made to synthesize the complex in
a diastereomerically pure form. Crystallization was not suc-
cessful since two main isomers (the geometrical isomers of
the Sp,Sp diastereomer around the Pd–Cl squares, ca. 2:1)
plus four minor isomers are present (see above). However,
it was found that the acac monomer 35-CpF-Ph-Ts can be
conveniently crystallized on a preparative scale, thereby
leading to diastereomerically pure material in the form of
purple-red needles (Scheme 10). Treatment with a mixture
of benzene (to dissolve the complex), MeOH (to get misci-
bility with the aqueous phase), LiCl (as Cl source), and con-
centrated aqueous HCl (to protonate [acac]À) then leads
within minutes to the regeneration of 29-CpF-Ph-Ts in a dia-
stereomerically pure form with respect to planar chirality.
Only minor decomplexation of PdII was observed (<5%),
which means that the Pd–C s bond is remarkably stable,
even to strong acids.
Entry 25a C5R5 R1 R2
x
T [8C] Yield[a] [%] ee[b] [%]
1
2
3
E
E
E
Z
Z
Z
E
E
E
E
Z
E
E
E
E
Z
Cp
Cp
Cp
Cp
Cp
Cp
Cp
Ph CF3
5
5
5
5
5
5
5
5
5
5
5
1.0
0.5
0.1
20
20
20
20
20
20
20
40
40
40
40
20
20
20
96[c]
88[c]
54[c]
90[e]
75[c]
19[c]
76[c]
80[g]
91[e]
88[e]
72[e]
96[e]
95[e]
94[e]
95[h]
36[e]
64 (R)
67 (R)
65 (R)
91 (S)
89 (S)
72 (S)
66 (R)
0
89 (R)
74 (R)
93 (S)
97 (R)
98 (R)
97 (R)
95 (R)
85 (S)
Ph p-Tol
Ph 1-Naph
Ph CF3
Ph p-Tol
Ph 1-Naph
tBu p-Tol
4[d]
5[d]
6[d]
7
8[f]
9[f]
10[f]
11
12
13
14
15
16
Cp* Ph CF3
Cp* Ph p-Tol
Cp* Ph C6F5
Cp* Ph p-Tol
CpF Ph p-Tol
CpF Ph p-Tol
CpF Ph p-Tol
CpF Ph p-Tol
CpF Ph p-Tol
0.05 40
40
5
[a] Yield determined by 1H NMR spectroscopy. [b] The ee values deter-
mined by HPLC after hydrolysis of 26a to the secondary amine (see the
Supporting Information). [c] Reaction time 2 d. [d] AgOTf was used for
activation. [e] Reaction time 1 d. [f] Catalyst precursor doped with 10%
28-Cp*-Ph. [g] Reaction time 5 h. [h] Reaction time 3 d.
loading of 5 mol%. Interestingly, to obtain these high levels
of asymmetric induction for (E)-25a, the catalyst precursor
29-Cp*-Ph-Ts (dr=20:1) had to be doped with 10% 28-
Cp*-Ph-Ts (i.e., 0.5 mol% with regard to (E)-25a). In the
absence of 28-Cp*-Ph-Ts, the ee decreased significantly to
73%. We assume that the Rp-configured catalyst (the minor
palladacycle isomer) is preferentially deactivated by the for-
mation of a monomeric complex with the imidazoline imino
N atom of 28-Cp*-Ph-Ts (Scheme 9).[24]
Scheme 10. Formation of geometrically pure 29-CpF-Ph-Ts.
The diastereomerically pure catalyst precursor leads only
to a negligible further improvement of the ee of the rear-
rangement product, since the minor diastereomer is argua-
bly less catalytically active. The scope of the rearrangement
of imidates 25 was then studied in detail using 29-Cp-Ph-Ts
and 29-Cp*-Ph-Ts for Z- and 29-CpF-Ph-Ts for E-substrates
25 (Table 4).
Whereas the former two catalyst precursors provide good
conversions and yields only for a-unbranched allylic sub-
strates (Z)-25 requiring a precatalyst loading of 5 mol%
(Table 4, entries 1–6), the latter has a broad applicability
even at low catalyst loadings. The rate of the rearrangement
depends primarily on the steric bulk of the residue R’. With
unbranched alkyl substituents (R1 =Me, nPr, (CH2)2Ph,
Scheme 9. Proposed deactivation of the Rp-configured isomer of 29-Cp*-
Ph-Ts by imidazolines 28-Cp*-Ph-Ts.
A breakthrough was achieved with the less electron-rich
and even more bulky CpF spectator ligand, which resulted
Chem. Eur. J. 2009, 15, 8722 – 8741
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8729

R. Peters et al.
Table 4. Screening of substrates 25 bearing different groups R1 with 29-Cp-Ph-Ts/29-Cp*-Ph-Ts (Z-substrates)
and 29-CpF-Ph-Ts (E substrates).
Using the deprotection proto-
col from the literature, that is,
treatment with 1m NaOEt in
EtOH at 45 to 508C overnight
(Scheme 11),[4a] allylic trifluor-
oacetamides 26 could be gener-
ally deacylated with good
yields, except for 26h. In that
case complete decomposition to
phenylethylketone (39) was
found, probably through base-
catalyzed tautomerization to
form an enamine. However,
amide 26h could be deacylated
in 88% yield by reaction with
NaBH4 in EtOH at RT.[25] This
method could also be applied
as an alternative method to all
other allylic amides 26.
Opposite absolute configura-
tions were obtained for the
major enantiomers of rear-
rangement products 26 starting
from either (E)- or (Z)-25. This
is accounted for by the working
model depicted in Scheme 12.
Assuming that the olefin as
p-acidic ligand will preferential-
ly coordinate trans to the imida-
zoline N atom due to the trans
effect,[17] and that the cis posi-
tion is blocked by the negative-
ly charged counterion,[22] the
imidate N atom, which is not
Entry
25
R1
C5R5
x
T [8C]
Yield[a] [%]
ee[b] [%]
1[c]
(Z)-25a
(Z)-25c
(Z)-25d
(Z)-25a
(Z)-25c
(Z)-25d
(E)-25a
(E)-25a
(E)-25a
(E)-25b
(E)-25b
(E)-25c
(E)-25c
(E)-25d
(E)-25d
(E)-25e
(E)-25e
(E)-25 f
(E)-25 f
(E)-25g
(E)-25h
(E)-25h
nPr
Cp
Cp
Cp
5
5
5
5
5
5
20
20
20
40
40
40
20
40
40
20
40
20
40
40
40
40
40
40
40
40
40
40
75
70
69
72
95
82
95
89
95
91
98
94
99
96
95
75
81
96
82
96
99
97
89 (S)
90 (S)
96 (S)
93 (S)
95 (S)
96 (S)
98 (R)
97 (R)
95 (R)
95 (R)
92 (R)
99.7 (R)
98 (R)
98 (R)
98 (R)
96 (R)
93 (R)
99 (S)
97 (S)
97 (R)
88 (S)
84 (S)
2[c]
A
3[c]
iBu
nPr
4[c]
Cp*
Cp*
Cp*
CpF
CpF
CpF
CpF
CpF
CpF
CpF
CpF
CpF
CpF
CpF
CpF
CpF
CpF
CpF
CpF
5[c]
A
6[c]
iBu
nPr
nPr
nPr
Me
Me
7[c]
0.5
0.1
0.05
0.1
0.05
1.0
0.05
0.2
0.1
0.5
0.1
1.0
0.5
0.35
1.0
0.5
8[c]
9[d]
10[d]
11[d]
12[c]
13[d]
14[c]
15[c]
16[c]
17[d]
18[d]
19[d]
20[d]
21[c]
22[c]
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
iBu
iBu
iPr
iPr
c-Hex
c-Hex
AHCTNURTGEG(NNNU CH2)2CO2Et
Ph
Ph
[a] Isolated yield. [b] The ee values determined by HPLC after hydrolysis of 26 to the secondary amines (see
the Supporting Information). [c] Reaction time 1 d. [d] Reaction time 3 d.
iBu), (R)-26 is formed in excellent yield with 0.05 to
0.1 mol% catalyst loading, providing very high enantioselec-
tivities (Table 4, entries 7–15). The most difficult aliphatic
substrate in terms of the enantioselectivity prepared from
crotonylic alcohol (i.e., R1 =Me) furnished (R)-26b with
95% ee (Table 4, entry 10). Even in the case of the branched
iPr or cyclohexyl (c-Hex) substituents, a case that has not
been described so far, the rearrangement proceeds with ac-
ceptable rates with 0.1 to 1.0 mol% catalyst precursor
(Table 4, entries 16–19). An ester functionality somewhat
lowered the reaction rate but still permitted good ee values
and yields with 0.35 mol% precatalyst (Table 4, entry 20).
Also, the aromatic Ph substituent (best reported results:[4b]
yield: 72%; ee=81% with 5 mol% COP-TFA) is well toler-
ated (Table 4, entries 21 and 22). The comparatively lower
ee values of 84–88% obtained with 25h using precatalyst
loadings of 0.5–1.0 mol% are the consequence of a relative-
ly rapid noncatalyzed, thermal rearrangement (vide infra),
which was not observed for 3-alkyl-substituted substrates
and that might also be the reason for the low enantioselec-
tivity obtained with previous less active catalysts.
Scheme 11. a) Procedures for the cleavage of the trifluoroacetamide
group; b) phenylethylketone formation from 26h by cleavage attempts
with NaOEt.
coordinated to the Pd center, will approach the olefin by
means of an outer-sphere attack (remote to the Pd center).
Increased steric interactions of the coordinated olefin with
bulky substituted Cp’ spectator ligands presumably results in
a better face selectivity of the olefin coordination, thus ex-
8730
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 8722 – 8741

Asymmetric Aza-Claisen Rearrangement
FULL PAPER
The enhanced catalyst activity of 29-CpF in combination
with the stereospecific reaction outcome explained by the
working model shown in Scheme 12 led to the investigation
of the enantioselective formation of N-substituted quaterna-
ry stereocenters by means of the aza-Claisen rearrangement
using 3,3-disubstituted allylic trifluoroacetimidates 42.
2.0 mol% of the catalyst precursor 29-CpF-Ph-Ts activat-
ed in situ with 3.75 equiv of AgTFA[34] (relative to 29) were
generally sufficient to give high conversion after 2.5 days at
508C with substrates in which one of the substituents R or
R’ is a methyl group (Table 5).[4e] The allylic trifluoroaceta-
mides 43 were formed with high to excellent ee values and
in good yield (Table 5, entries 1–10). As expected, the rear-
Scheme 12. An explanation of the stereospecific outcome of the rear-
rangement of 25.
Table 5. Highly enantioselective rearrangement of 3,3-disubstituted allyl-
ic imidates 42.[a]
plaining the higher enantioselectivity as compared to deriva-
tives bearing the unsubstituted Cp. The exceptionally high
catalytic activity is mainly attributed to five factors: 1) the
possibility to apply almost solvent-free reaction conditions,
2) the robustness of the active catalyst under the reaction
conditions even at enhanced temperatures, 3) the bulk of
the CpF ligand, which might lead to a facilitated monomer
formation of the active species, but also in less product in-
hibition due to destabilization of the corresponding olefin
complex, 4) the formation of a ferrocenium system as cata-
lytically active species by oxidation of the ferrocene core by
the silver salt, and 5) the electron-withdrawing nature of the
five phenyl substituents on the CpF spectator ligand.[26,27]
The last two factors enhance the Lewis acidity of the Pd
center in 29-CpF.
Prior to this work, a longstanding problem has been the
catalytic enantioselective formation of N-substituted quater-
nary stereocenters by aza-Claisen rearrangement or allylic
substitution reactions.[28] In the former case, this can be at-
tributed to insufficient catalyst activity, as the formation of a
quaternary center is slowed for steric reasons. Previously,
the aza-Claisen rearrangement forming quaternary carbon
centers has only been reported in a nonenantioselective
way.[29] Since quaternary carbon atoms bearing a nitrogen
substituent are a widespread structural motif for bioactive
natural and unnatural compounds,[30] the asymmetric con-
struction of those stereocenters by means of asymmetric cat-
alysis is an important challenge. A key application of enan-
tiopure allylic amines might become the formation of qua-
ternary amino acids and derivatives, of which many can be
found in biologically active natural products and pharma-
ceuticals.[31] These targets are attractive due to their ability
to, for example, induce helical peptide structures,[32] and also
owing to the fact that peptides incorporating quaternary
amino acid residues possess enhanced stability to proteas-
es.[33] To this end, various methods have been developed in
the past for the formation of quaternary amino acids, some
of them also utilizing asymmetric catalysis, while the large
majority of reactions is still relying on stoichiometric con-
cepts.[31a]
#
42 R’
R
Cat.
[mol%] [%]
Yield[b] ee[c] [%]
1
2
3
4
5
6
7
8
9
a
a
b
c
N
Me
Me
Me
2
0.5
2
94[d]
79[e,f]
63
99.6 (R)
97 (R)
93 (R)
98 (R)
96 (R)
98 (R)
93 (R)
96 (R)
98 (S)
99 (R)
98 (R)
>99.5 (R)
98 (R)
97 (R)
2
74
d
e
f
G
E
Me
2
73
2
84
Me
Me
(CH2)3OSi
CH2OBn
CH2OBn
CH2OBn
CH2OBn
CH2OBn
2
64
g
h
A
2
50
E
2
74
10 i Me
11 j Et
12 k nPr
13 l nBu
2
84
4
4
4
4
81
63[g]
61[g]
51[g]
14 m
(CH2)3OSi
(iPr)3
[a] The reactions were performed on a 0.03–0.08 mmol scale (reaction
time 2.5 d) unless otherwise noted. [b] Yield of isolated 43. [c] The ee
values determined by HPLC. [d] 0.3 mmol scale. [e] Reaction time 10 d.
[f] 1.0 mmol scale. [g] Reaction time 3.5 d.
rangement of 3,3-disubstituted substrates 42 is significantly
slower relative to substrates in which R=H.[4e] The lower
turnover frequency is attributed to the additional double-
bond substituent hampering the attack of the imidate nitro-
gen atom on the sterically more hindered olefin. It is possi-
ble, though, to lower the catalyst loading to 0.5 mol%
(Table 5, entry 2), albeit at the price of a prolonged reaction
time of 10 d, but still attaining an excellent ee (97% for sub-
strate 42a with R=Me, R’=(CH2)2Ph). This implies that
the thermal rearrangement at 508C is at most 3% within
10 d.
Whereas in previous experiments with 29-CpF-Ph-Ts, Z-
configured 3-monosubstituted imidates had given signifi-
cantly lower reaction rates as compared to E-configured
substrates (see above), this difference vanishes if both R
and R’ are larger than H.[4e] For example, both the E- and
Z-configured substrates 42d and 42h bearing a bulky
Chem. Eur. J. 2009, 15, 8722 – 8741
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8731

R. Peters et al.
(CH2)3OTIPS (TIPS=triisopropylsilyl) moiety gave practi-
cally the same yield and conversion. In both cases the prod-
uct was formed under identical conditions with very high ee
values (96 and 98% respectively; Table 5, entries 5 and 9),
but with the opposite configuration.
already sufficiently electron-withdrawing to prevent the
facile formation of an allylic carbocation 45).
Raising the reaction temperature above 508C was investi-
Table 6. Temperature dependence for the enantioselective rearrange-
Substrates 42 were in general found to be much more sen-
sitive towards (Brçnsted or Lewis) acid-catalyzed elimina-
tion of trifluoroacetamide than 3-monosubstituted imidates
25, since the additional 3-substituent can stabilize an inter-
mediary allylic cation 45 (Scheme 13). Acid-catalyzed elimi-
nation of PMP-trifluoroacetamide 48 is thus a significant
side reaction.[35] However, by the use of proton sponge as
Brçnsted acid scavenger, this competitive reaction pathway
can be suppressed to a large degree (Table 5).
ment of 3,3-disubstituted allylic imidate 42a.
Entry
T [8C]
x
t [h]
Yield[a] [%]
ee[b] [%]
1
2
3
50
70
80
2.0
1.0
1.0
48
30
30
94
82
76
>99
96
93
[a] Yield of isolated 43a. [b] The ee values determined by HPLC.
gated in the case of substrate 42a to enhance the rearrange-
ment rate (Table 6). Whereas at 508C a reaction time of
48 h with 2 mol% catalyst precursor was necessary (Table 6,
entry 1), 1 mol% at 708C is sufficient for full conversion
with a reduced reaction time of 30 h (Table 6, entry 2). In-
creasing the reaction temperature further is not useful since
the enantioselectivity is dropping and side reactions like
elimination and [1,3] shift, possibly due to catalyst decompo-
sition, are observed (Table 6, entry 3).
Since the enantioselectivity of the rearrangement is appa-
rently largely independent from the steric discrimination of
the two residues R and R’ at the imidate 3-position and
since the rearrangement proceeds in a stereospecific fashion
(opposite enantiomers are formed from geometric isomers,
see Table 5, entries 5 and 9), the enantioselectivity-deter-
mining step should be the face-selective coordination of the
olefin moiety to the PdII center (Scheme 14).[36] Again, the
olefin is assumed to coordinate trans to the imidazoline N
atom in 49, thereby triggering an attack of the imidate N
atom at the olefin from the face remote to the Pd center. In
the stereoelectronically preferred orientation of the olefin
part parallel to the ferrocene axis, that is, perpendicular to
the Pd square plane, the sterically undemanding C-1 methyl-
ene moiety should point towards the bulky CpF spectator
ligand to minimize unfavorable steric interactions, while the
bulky imidate moiety is avoiding the ferrocene core. Coordi-
nation of the enantiotopic olefin face is less favorable again
owing to steric arguments. As a consequence of these as-
Scheme 13. Acid-triggered elimination of trifluoroacetamide 48.
If both substituents R and R’ are larger than a methyl
group, 4.0 mol% of 29-CpF-Ph-Ts was employed to obtain
synthetically useful yields and the reaction time was pro-
longed to 3.5 d. In that case, one of the substituents should
display an electron-withdrawing character, since otherwise
decomposition is becoming predominant. The attack by the
imidate N is retarded with increasing size of the substituents
and cannot compete with the formation of an allylic cation,
whereas destabilization of the allylic cation by electron-
withdrawing groups can overcome this problem.
With R’=CH2OBn, R was varied from Me to Et, nPr,
nBu, and (CH2)3OTIPS, showing practically no difference in
enantioselectivity (Table 5, entries 10–14). Excellent enan-
tioselectivities were attained even in those cases in which R
and R’ have a similar size (Table 5, entries 12–14). Gratify-
ingly, the rearrangement is compatible with important func-
tional groups such as olefin, ester, carbonate, silyl ether,
benzyl ether, or Boc-protected secondary amino moieties.
An azide-functionalized substrate (R=Me, R’=(CH2)3N3)
gave only very little conversion, probably owing to catalyst
poisoning by the azide functionality.
If one of the two substituents at C-3 is Me, branches in
the other residue are tolerated only starting from the 6-posi-
tion. Substrates bearing isobutyl or benzyl residues thus did
not react to the desired product. An iBu group mainly re-
sulted in decomposition, whereas the substrate with a Bn
moiety simply did not react (the phenyl substituent might be
Scheme 14. Assumed coordination mode of the olefinic moiety of allylic
imidates 42.
8732
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 8722 – 8741

Asymmetric Aza-Claisen Rearrangement
FULL PAPER
sumptions, it was expected that even substrates in which R
and R’ have an identical size should provide almost perfect
enantioselectivities.
To verify this hypothesis, the use of the geometrically
pure allylic imidate 42n in which R and R’ have practically
the same size, namely, CH3 and CD3 (Scheme 15), was inves-
tigated. Also in this case the product 43n is formed with an
ee value of 96%, thus confirming the mechanistic hypothe-
sis.[37]
Scheme 17. Use of allylic amide 43a for the preparation of a,a-disubsti-
tuted a-amino acid 54 and b,b-disubstituted b-amino acid 55 (Fmoc-
OSu=N-(9-fluorenylmethyloxycarbonyloxy)succinimide; DMS=dimeth-
yl sulfide).
Scheme 15. Highly enantioselective formation of CH3/CD3-substituted al-
lylic amide 43n (200 mg scale), a substrate with practically no steric dif-
ference between both of its 3-substituents.
As mentioned above, 29-CpF-Ph-Ts provides a highly effi-
cient catalyst for E-configured 3-monosubstituted substrates
25, but it is comparatively slow for the Z-configured coun-
terparts, whereas if both R and R’ in 42 are larger than H,
there is no clear difference in rate between E- and Z-config-
ured substrates. This finding can be rationalized by the as-
sumed six-membered cyclic transition state (Scheme 16):[2a]
moved by NaBH4 in iPrOH/H2O (10:1), followed by oxida-
tive cleavage of the PMP group with CAN in MeCN/H2O
(1:1). Due to limited stability of the free amine, the crude
reaction mixture was directly used for Fmoc protection. Al-
ternatively, the amine could be purified as hydrochloride.
The vinyl system was cleaved by ozonolysis, followed by
oxidation with NaClO2 to form a-amino acid 54 in 85%
yield. Alternatively, hydroboration with 9-borabicyclo-
AHCTUNGTREN[GNUN 3.3.1]nonane (9-BBN) and oxidative workup with H2O2 and
NaOAc gave the corresponding amino alcohol in 73%
yield; it was subsequently converted to b-amino acid 55 in
80% yield by CrO3 in 6n H2SO4.
The absolute configuration of a-amino acid 54 was deter-
mined after removal of the Fmoc group with tetrabutylam-
monium fluoride (TBAF) by comparison with the reported
specific optical rotation[38] and found to correspond to the
expected configuration. Since a uniform mechanism for the
stereospecific rearrangement step can be expected, the abso-
lute configuration of all compounds in Table 5 can be as-
signed.
The approaches described above and in previous publica-
tions were limited to trifluoroacetimidate substrates in
which the N atom is protected by a PMP moiety that can be
oxidatively removed to furnish free primary allylic amines.
A variation of the imidate N substituent would allow for the
direct formation of chiral secondary allylic amines without
the need for a protecting group and a subsequent alkylation
step. To our knowledge, the N substituent was previously
always considered to be a protecting group. Moreover, even
in the case of the more reactive benzimidate substrates (aryl
instead of CY3 in 2 in Scheme 1) providing the synthetically
less useful benzoylamides, only a small number of examples
has been described. In those cases the N substituent was re-
stricted to either a carbobenzyloxy (Cbz) group (no enantio-
selectivity)[4c] or to a few phenyl residues.[39] Since hydrolysis
Scheme 16. Transition states for E- and Z-configured substrates. If both
3-substituents are larger than H, the energetic difference for positioning
the more bulky one in an axial position is much lower than if one sub-
stituent is H.
in the case of an E-configured 3-monosubstituted substrate,
the substituent R is in an equatorial position whereas H
adopts an axial position in 50. In the case of a Z-configura-
tion, R is axially and H equatorially oriented in 51 (energet-
ically less favored). If, however, both R and R’ are larger
than H, the difference in energy of the two transition states
is considerably lower, resulting in similar reaction rates for
E as well as Z configuration (see 52).
To showcase the utility of the rearrangement products, al-
lylic amide 43a was employed to synthesize 9-fluorenyl-
methoxycarbonyl (Fmoc)-protected a,a-disubstituted a-
amino acid 54 and b,b-disubstituted b-amino acid 55
(Scheme 17). The PMP group in amides 43 could not be di-
rectly removed, since the oxidation by ceric ammonium ni-
trate (CAN) requires sufficient electron density of the aro-
matic system. Hence, the trifluoroacetyl group was first re-
Chem. Eur. J. 2009, 15, 8722 – 8741
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8733

R. Peters et al.
of the resulting benzamides is very difficult, no examples for
secondary amines were reported. For the application of N-
alkyl substituents, there was previously no example in the
literature, which can be attributed to a reduced reactivity of
these substrates as our results below reveal. 29-CpF-Ph-Ts
allowed for the first catalytic asymmetric preparation of un-
protected secondary allylic amines by aza-Claisen rearrange-
ment of N-aryl- or N-alkyl-substituted trifluoroacetimida-
tes.[4i]
Scheme 18. Explanation of line broadening in NMR spectra of imidates
bearing aromatic residues on nitrogen.
The imidate substrates 60 were prepared in good yield by
condensation of iminochlorides 58 with geometrically pure
allylic alcohols 59 after deprotonation of the hydroxy group
with NaH or lithium hexamethyl disilazide (LHMDS) in
THF (Table 7). The iminochlorides 58 were readily available
by a modified literature procedure.[40]
the C=N bond. In contrast, with aliphatic residues, such a
resonance structure is not possible and as a consequence the
C=N bond is configurationally more stable.
To find the lowest catalyst loadings that still provide good
yields for most of the rearrangement products 61 within a
reaction time of 3 d (Table 7), various substrates were inves-
tigated at 50 and 708C using 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, or
5.0 mol% of precatalyst 29-CpF-Ph-Ts.
Allylic trifluoroacetimidates bearing an aromatic residue
on nitrogen generally exhibit a very broad 19F NMR spectro-
scopic signal at RT for the CF3 group and broad signals in
1H NMR spectroscopy for all protons in close proximity to
the imidate functionality. N-Alkyl-substituted imidates, in
contrast, often have two separated, rather sharp signals for
the CF3 group and protons in proximity, pointing to the co-
existence of two geometric isomers. Attempts to determine
the geometry of the C=N double bond in N-aryl imidates by
NOE experiments failed, whereas for imidate 60l (N-
CH2CH2CO2Et), an NOE between the NCH2 protons and
the CF3 group was observed, thus suggesting that the major
isomer is E-configured (see the Supporting Information).
The line broadening in the case of N-aryl-substituted imi-
dates can be explained by a fast isomerization that is slow
on the NMR spectroscopic timescale for N-alkyl imidates.
This finding can be explained by the resonance structure 56’
(Scheme 18), which results in a facilitated rotation around
The E(C C)-configured allylic imidates 60a–e carrying N-
=
aryl residues with different electronic or steric properties
such as 4-F-C6H4, 4-I-C6H4, 1-naphthyl, 2,4-dimethoxyphen-
yl, or 3,4-dimethoxyphenyl require similar precatalyst load-
ings (0.05–0.2 mol%; Table 7, entries 1–5) as substrates bear-
ing the N-PMP moiety. Interestingly, the nonenantioselec-
tive [PdCl2ACHTNUGRTENUNG(NCMe)2]-catalyzed rearrangement was not feasi-
ble for dimethoxyphenyl-substituted imidates 60d/e: In all
cases, a complete decomposition was observed, probably by
PdII-promoted oxidation of the electron-rich aromatic sys-
tems. Such a decomposition was not observed using chiral
palladacycle 29-CpF-Ph-Ts. Racemic samples for HPLC
analysis were thus prepared through a thermal rearrange-
ment (2 h at 1608C, neat or in mesitylene 1:1 w/w). Gratify-
ingly, N-alkyl-substituted imidates also undergo the aza-
Table 7. Synthesis of imidates 60, enantioenriched trifluoroacetamides 61, and the free secondary allylic amines 62.
Entry
R1
R2
R3
60
Yield [%][a]
61
T [8C]
Y[b]
Yield [%][c]
ee [%]
62
Yield [%][c]
1
2
3
4
5
6
7
8
4-I-C6H4
4-F-C6H4
1-naphthyl
2,4-OMe-C6H3
3,4-OMe-C6H3
nPr
nPr
nPr
nPr
nPr
Me
iPr
H
H
H
H
H
H
H
H
H
H
H
Me
60a
60b
60c
60d
60e
60 f
60g
60h
60i
87 (A)
68 (A)
85 (A)
87 (B)
89 (B)
80 (A)
85 (A)
74 (A)
77 (A)
67 (A)
58 (B)
70 (A)
61a
61b
61c
61d
61e
61 f
61g
61h
61i
70
50
70
50
50
50
70
50
50
70
70
70
0.2
0.1
0.05
0.05
0.1
2.0
2.0
0.2
1.0
1.0
1.0
2.0
85
91
90
96[d]
96[d]
97[d]
98
62a
62b
62c
62d
62e
62 f
62g
62h
62i
90[e]
82
86
96[f]
97[f]
98
91[g]
89[g]
77
98
T
94[h]
97[h]
99[d]
98[d]
98[h]
99[h]
98[h]
N
51
91
99
92
84
42
75
89
84
78
80
76
A
9
N
N
10
11
12
N
N
60k
60l
60m
61k
61l
61m
62k
62l
62m
E
T
G
[a] Isolated yield (in brackets the method which was used is given, see graphic). [b] Precatalyst loading (see Scheme 3). [c] Isolated yield. [d] The ee
values determined by HPLC after hydrolysis of 61 to 62. [e] Amide cleavage with NaOEt. [f] Reaction time 48 h. [g] Amide cleavage with MeLi, THF,
À788C to RT. [h] The ee values determined by HPLC.
8734
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 8722 – 8741

Asymmetric Aza-Claisen Rearrangement
FULL PAPER
Claisen rearrangement with excellent enantioselectivity and
in most cases with high yield. The required catalyst loadings
depend largely upon the steric bulk of the alkyl substituent
R1; whereas only 0.2 mol% precatalyst were required for
the case of the unbranched alkyl residue n-hexyl (Table 7,
entry 8), with branched or functionalized N-alkyl groups
precatalyst loadings of 1.0 to 2.0 mol% were needed. Never-
theless, also in those cases the products were formed in high
yield and with almost perfect stereocontrol (Table 7, en-
tries 6, 9–11). A moderate yield was obtained for the combi-
nation of an alkyl residue R1 with an a-branched residue R2
(Table 7, entry 7). No product was formed employing a-
branched alkyl substituents R1. In contrast to catalytic allylic
substitutions, secondary allylic amines with a quaternary N-
substituted stereocenter can also be generated highly enan-
tioselectively (Table 7, entry 12), albeit in only moderate
yield.
Fourth-generation catalysts: To understand if the higher cat-
alytic activity and selectivity of FcF imidazoline palladacycle
29-CpF-Ph-Ts as compared with the COP-Cl system 7 is
mainly caused by the imidazoline or by the pentaphenyl fer-
rocenium moiety, FcF oxazolines 65a (R1 =iPr, R2 =H) and
65b (R1 =R2 =Ph) were prepared according to Scheme 19 in
83 and 66% yield (over two steps), respectively, starting
from carboxylic acid 21 and (S)-valinol (63a) or (1R,2S)-1,2-
diphenylethanolamine (63b).
The clearly lower reaction rate of N-aliphatic imidates as
compared to N-aryl-substituted substrates is most likely
caused by electronic effects. One possible explanation might
be that an imidate nitrogen bearing an aliphatic residue is
more electron-rich and thus a better competing ligand for
PdII, thereby lowering the concentration of active, accessible
catalyst. An alternative explanation is the resonance stabili-
zation of the positive charge on the N atom in the benzylic
position in the cyclic intermediate 4 (Scheme 1), although
using classical resonance structures, a transfer of electron
density from the N-aryl substituent to N cannot be achieved
without charge separation.
Reductive removal of the trifluoroacetamide protecting
group releasing the free secondary allylic amines was usually
achieved in high yield with NaBH4 (Table 7). The reductive
cleavage has the general advantage of shorter reaction times
as compared to hydrolytic procedures utilizing, for example,
NaOEt in EtOH (4 h versus 1–2 d). For 61a though
(Table 7, entry 1), the iodo atom was partially removed
under reductive conditions and therefore basic hydrolysis
(K2CO3, MeOH/H2O (10:1)) was preferred for this exam-
ple.[41] Alternatively, MeLi in THF at À788C leads to a
smooth deprotection if there are no other functional groups
sensitive to this reagent (Table 7, entries 4, 5).
The absolute configuration was determined by chemical
correlation. The R-configured compounds 62h and 62m
were independently prepared by aza-Claisen rearrangement
of the corresponding PMP-substituted allylic trifluoroaceti-
midates 25c and 42a (R1 =PMP, R2 =(CH2)2Ph, R3 =H or
Me), with full deprotection yielding the free primary amine
and subsequent reductive monoalkylation with hexanal/
NaBH4. Comparison of HPLC retention times revealed that
this material had the same configuration as the material re-
ported in Table 7, entries 8 and 12, respectively. Since we
have shown that the aza-Claisen rearrangement of allylic tri-
fluoroacetimidates catalyzed by 29-CpF-Ph-Ts is a stereo-
specific reaction and a uniform reaction mechanism can be
assumed, the configuration has been assigned to all allylic
amines 62a–m.
Scheme 19. Synthesis of FcF palladacycles 66/67 by diastereoselective cy-
clopalladation.
Upon heating oxazolines 65 with PdACTHNUTRGENUN(G OAc)2 in HOAc, pal-
ladacycles 66a and 66b precipitated in diastereomerically
pure form and could be isolated by filtration, whereas by-
products and possibly also the minor diastereomer stayed in
solution. Further purification was carried out by recrystalli-
zation, thus delivering pure 66a and 66b in 93 and 72%
yield, respectively. Conversion of the acetate-bridged palla-
dacycles to the Cl-bridged complexes 67a/b took place in 95
and 93% yield, respectively, by stirring a suspension of 66
and LiCl in MeOH/benzene. Whereas the acetate-bridged
dimers exist as a single diastereomer relative to the Pd–ace-
tate square planes, the chloride-bridged complexes were
formed as approximately 2:1 mixtures of geometrical iso-
mers. Alternative conditions for the cyclopalladation, nota-
bly Na
[PdCl4] in combination with NaOAc in MeOH/ben-
(OAc)2 in CH2Cl2 or benzene,
zene or CH2Cl2, as well as PdAHCTNUGTRENNNUG
failed to give any cyclopalladated product. Instead of cyclo-
palladation, a complete decomposition was found to take
place within less than 30 min.
Chem. Eur. J. 2009, 15, 8722 – 8741
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8735

R. Peters et al.
Table 8. Optimization of the rearrangement of model substrates (E)-/
(Z)-25a catalyzed by 67.
The absolute configuration of palladacycle 66a was deter-
mined by X-ray crystal structure analysis (Figure 2).[42,43] In
contrast to COP-OAc (OAc=X in 7), the isopropyl group is
pointing towards the sandwich core.
Entry 67
x
AgX (y)
E/Z t [h] T [8C] Yield[a] [%] ee[b]
[%]
1
2
3
4
a
a
a
a
a
a
a
b
b
a
a
a
a
0.5
0.5
0.5
0.5
0.2
1.0
0.2
0.2
0.1
0.1
AgTFA (3.8)
AgNO3 (3.8)
AgOTs (3.8)
AgNO3 (2.0)
AgNO3 (2.0)
–
AgNO3 (3.8)
AgNO3 (3.8)
AgNO3 (3.8)
AgNO3 (3.8)
E
E
E
E
E
E
E
E
E
E
E
Z
Z
24
24
24
24
24
24
24
24
24
24
24
24
24
40
40
40
40
40
40
40
40
40
RT
40
40
40
44
>99
98
99
45
90
97
77
96
n.d.
n.d.
97
97
n.d.
97
5
6[c]
7
6
>99
>99
70
99
99
8
9
10
11
12
13
0.05 AgNO3 (3.8)
2
1
97
88
88
AgNO3 (3.8)
AgNO3 (3.8)
97
87
Figure 2. X-ray crystal structure of the acetate-bridged dimeric complex
66a.
[a] Yield determined by NMR spectroscopy. [b] The ee values determined
by HPLC after hydrolysis to the corresponding amine (see the Support-
ing Information). [c] 66a was used.
The absolute configuration of palladacycle 66b was tenta-
tively assigned by the known absolute configuration of the
aza-Claisen rearrangement product of imidate 25a. The
same major enantiomer as with complex 66a and 29-CpF-
Ph-Ts was obtained, suggesting an Sp-configuration, since
mainly the element of planar chirality is decisive for the
configuration of the catalysis product. This assumption was
In addition to iPr-substituted complex 67a, the 4,5-di-
phenyl-substituted oxazoline palladacycle 67b was exam-
ined. While there is practically no difference in enantioselec-
tivity, the catalytic activity slightly decreases (Table 8, en-
tries 7–10). Whereas 0.1 mol% 67a completely convert the
test substrate within 24 h at RT (Table 8, entry 10), the same
amount of 67b at 408C resulted in only 70% yield due to in-
complete conversion (Table 8, entry 9). Since complex 67b
was prepared in lower yields and from a more expensive
amino alcohol than 67a, all further investigations were car-
ried out with 67a.
A screening of the catalyst amount was performed for
both the E- and the Z-configured test substrates (E)-/(Z)-
25a. Like for the imidazoline palladacycle 29-CpF-Ph-Ts,
the E-configured substrate not only reacts considerably
faster than its Z-configured counterpart, it also provides sig-
nificantly higher ee values. Although (E)-25a reacts com-
pletely within 1 d at 408C using only 0.05 mol% precatalyst
(Table 8, entry 11), full conversion for (Z)-25a requires
2 mol% (Table 8, entry 12). Ferrocenyl bisimidazoline bis-
palladacycle complexes thus remain the catalysts of choice
for Z-configured substrates.[4f,h]
The enantioselectivity obtained for (E)-25a did not signif-
icantly change within the investigated temperature range.
Catalyst loadings below 0.05 mol% were studied as well,
but the results were not reliable, which is most probably ex-
plained by a partial catalyst deactivation by trace impurities
of the substrate. A certain threshold of catalyst is thus nec-
essary on laboratory scale. The same was found with 29-
CpF-Ph-Ts, in which catalyst loadings below 0.05 mol% had
resulted in low conversions even at prolonged reaction
times.
1
confirmed by H NOESY experiments (see the Supporting
Information).
Utilizing the optimized conditions for activating imidazo-
line catalyst 29-CpF-Ph-Ts, that is, 3.75 equiv of AgTFA per
precatalyst dimer, oxazoline 66a (0.5 mol%) rearranged the
3-monosubstituted model substrate 25a with only 44% yield
and 90% ee (Table 8, entry 1) at 408C for 24 h (83% conver-
sion after 72 h), whereas 29-CpF-Ph-Ts had given practically
1
full conversion and 95% ee with = of the catalyst amount
10
(Table 4, entry 9). As the nature of the Ag salt has a large
impact on the reaction rate and selectivity, further Ag salts
were examined. It was found that AgNO3 generates a highly
active and selective catalyst (Table 8, entry 2), whereas
AgOTs allows for high activity (Table 8, entry 3) but consid-
erably lower enantioselectivity. Further studies were thus
carried out with AgNO3.
By use of 2 equiv of AgNO3 per Cl-bridged dimer 67a,
full conversion and 96% ee were obtained with 0.5 mol%
precatalyst, whereas 0.2 mol% led to only 45% conversion
(Table 8, entries 4, 5). Part of this catalytic activity might be
explained by oxidation of 67a to the corresponding active
ferrocenium system on the surface of precipitated AgCl.
However, nonactivated 66a also has some, yet low, catalytic
activity (6% conversion with 1 mol% catalyst at 408C for
24 h; Table 8, entry 6). Complete oxidation to the corre-
sponding ferrocenium species is accomplished with 4 equiv
of AgNO3 per Cl-bridged dimer resulting in the best reactiv-
ity.
8736
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 8722 – 8741

Asymmetric Aza-Claisen Rearrangement
FULL PAPER
The rearrangement could be scaled up without any prob-
lem. Using 10.0 mmol of substrate (E)-25a (3.01 g) in com-
bination with 0.1 mol% 67a (0.4 mol% AgNO3, 408C) re-
sulted in complete conversion and an isolated yield of 98%
with 97.5% ee.[44]
Oxazoline 67a activated by AgNO3 is in general more re-
active than imidazoline 29-CpF-Ph-Ts (Table 9, entries 2–
15). For instance, with the (E)-3-Ph-substituted substrate
the N-cyclohexyl-substituted imidate 60n rearranged with
reasonable catalyst loadings (1.0 or 4.0 mol%) and excellent
enantioselectivities, thus yielding the protected secondary al-
lylic amine 61n.
The formation of a tertiary allylic amine bearing three dif-
ferent N substituents would require even more steps with
the traditional protecting group/alkylation strategy (vide
supra). Rearrangement of an allylic imidate already contain-
ing the complete carbon skele-
ton would only entail a reduc-
Table 9. Investigation of the substrate scope of the rearrangement of trifluoroacetimidates with 67a.
tion of the amide functionality,
for example, with LiAlH4, to
provide the tertiary amine. To
establish proof of principle, al-
lylic imidate 70 (Scheme 20)
was prepared in a one-pot reac-
tion from acetanilide 68 to
study if enolizable imidates
Entry R1
R2
R3
CX3
Product
x
t [h] T [8C] Yield[a] [%] ee [%]
1
nPr
H
H
H
H
H
H
Me
CH3
CH3
PMP
PMP
PMP
PMP
PMP
PMP
PMP
PMP
PMP
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
CF3
26a
26h
26h
26e
26 f
26i
43o
43a
43a
43i
0.05 24
40
20
40
20
40
50
50
50
50
50
70
50
50
50
70
50
60
50
99
98
96
76
99
2
12
71
90
98
74
95
95
90
91
96
99
85
97[b]
98[b]
90[b]
99[b]
99[b]
n.d.
n.d.
98[c]
99[c]
97[c]
96[c]
97[d]
97[d]
99[b]
98[b]
94[c]
95[c]
97[c]
bearing
a-acidic
hydrogen
2
Ph
0.5
0.2
0.5
0.5
2.0
2.0
2.0
2.0
1.0
0.5
4.0
2.0
4.0
1.0
0.2
24
24
24
24
24
24
24
72
24
24
48
72
48
72
48
atoms can be employed. Imi-
date 70 was found to be in equi-
librium with its tautomer, the
ketene N,O-acetal 71 (ratio 70/
71=ca. 3:1). With 0.2 mol%
67a activated by AgNO3 the re-
3
Ph
4
iPr
5
6
c-Hex
tBu
7
CH2Ph
8
A
9
A
arrangement
formed in high yield and with
excellent enantioselectivity
product
was
10
11
12
13
14
15
16
17
18
CH3
CH3
A
CH2OBn PMP
CH2OBn PMP
43i
43m
43m
61n
61n
72
N
(Table 9, entry 16). A Claisen
rearrangement product was not
detected.
E
H
H
H
H
H
c-Hex CF3
c-Hex CF3
Ph
H
ACHTUNGTRENNUNG
CH3
CCl3 74
CCl3 74
29-CpF-Ph-Ts is not able to
rearrange allylic trichloroaceti-
midates with a free NH, the
substrates of the original Over-
man rearrangement, presuma-
bly since it is deactivated by co-
ordination with the free NH
T
0.25 24
0.5 24
A
H
[a] Isolated yield. [b] ee determined by HPLC after hydrolysis to the corresponding amine (see the Supporting
Information). [c] ee determined by HPLC. [d] ee determined by HPLC after cleavage of TIPS with excess
TBAF (see the Supporting Information).
26h, a temperature of 408C was necessary using 29-CpF-Ph-
Ts, whereas 67a is effective already at RT (Table 9, entry 2).
group. In contrast, 67a was found to be a highly active cata-
lyst for these compounds (Table 9, entries 17, 18), producing
Also,
the
3,3-disubstituted
substrate
42m
(R’=
(CH2)3OTIPS, R=CH2OBn), which required 4 mol% 29-
CpF-Ph-Ts and a reaction time of 3.5 d to give a yield of
51%, reacted within 2 d in a nearly quantitative yield (or
within 3 d with 2 mol% catalyst; Table 9, entries 12, 13).
Like with 29-CpF-Ph-Ts, N-substituted quaternary stereo-
centers can be generated with almost perfect stereocontrol
(Table 9, entries 8–13).
Due to the enhanced catalytic activity, substrates that
could not be processed with 29-CpF-Ph-Ts or any other re-
ported chiral catalyst were investigated: whereas allylic tri-
fluoroacetimdates bearing a tBu or a Bn moiety as R1 still
did not react at useful rates (Table 9, entries 6, 7) although
mono-a-branched aliphatic residues R1 such as iPr or c-Hex
are well tolerated (Table 9, entries 4, 5), this catalyst is for
the first time able to rearrange substrates bearing an a-
branched aliphatic N-substituent R3 (Table 9, entries 14, 15):
Scheme 20. Synthesis of the enolizable acetimidate 70 and its use for the
catalytic asymmetric formation of tertiary amine 73 (LAH=lithium alu-
minum hydride).
Chem. Eur. J. 2009, 15, 8722 – 8741
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8737

R. Peters et al.
trichloroacetamide 74 in 24 h at 608C with a quantitative
yield and 95% ee using 0.25 mol% of precatalyst.
p-methoxyphenyl group, reacts simply too quickly, even at
48C (Table 10, entry 8).
Furthermore, problematic trifluoroacetimidate substrates
were investigated, which react already to a considerable
degree in the absence of a soft transition-metal ion by ther-
mal rearrangement. Allylic trifluoroacetimidates tend to un-
dergo a thermal rearrangement at ambient temperature if
the O–allyl bond is reactive due to formation of a particular-
ly stabilized allylic carbocation as result of a terminal aro-
matic substituent (Scheme 21).
A kinetic study (Figure 3) of the thermal rearrangement
at 408C in CDCl3 for the three 3-aryl-substituted imidates
25h/k/l allowed the determination of rate constants for the
Figure 3. Thermal rearrangement of three aryl-substituted imidates 25h/
k/l at 408C in CDCl3. Conversions were determined by 19F NMR spec-
troscopy.
Scheme 21. Acceleration of the thermal rearrangement of 3-aryl-substi-
tuted allylic imidates by stabilization of the positive charge in transition
state 75.
first-order kinetics, which fit well with the Hammet s+
values (1=À0.47; see the Supporting Information).
Phenyl-substituted imidate 25h thus rearranged in 72%
yield and only 81% ee employing 5 mol% of COP-Cl 7
(Cl=X),[4b] whereas 29-CpF-Ph-Ts (1 mol%) was able to im-
prove these values to 99% yield and 88% ee. Better enan-
tioselectivities were not possible without increased catalyst
loadings because a temperature of 408C was necessary to
obtain reasonable conversion rates. On the other hand, 67a
is already able to rearrange 25h at RT, thereby producing
26h with an ee value of 98% (Table 10, entry 1). To investi-
Similar to the 3-aryl-substituted substrates, pentadienylic
imidates such as 25n, which possess a conjugated diene
system, are able to form particularly stabilized cations. Nev-
ertheless, a synthetically useful 86% ee value was attained
under the conditions depicted in Scheme 22.
Table 10. Rearrangement of 3-aryl-substituted imidates with 67a.
Scheme 22. Regioselective rearrangement of diene substrate 25n.
The steric environment of COP-X 7 and 29-CpF-Ph-Ts
around the catalytic palladium site mainly differs in a
phenyl (29-CpF-Ph-Ts) and an iPr group (7) next to the co-
ordinating N site and the type and distance of the spectator
ligand (CpF for 29-CpF-Ph-Ts, tetraphenylcyclobutadiene
for COP). Although the distance between the two sandwich
ligands differs only marginally between COP and 29-CpF-
Ph-Ts (3.4 vs. 3.3 ꢃ), oxidation of the ferrocene to the ferro-
cenium species is expected to shorten this distance further.
Overall, the steric hindrance to access the Pd center is more
distinct for 29-CpF-Ph-Ts. These steric effects can be used to
explain the higher ee value obtained with 29-CpF-Ph-Ts, but
not necessarily the higher rearrangement rate. One would
have to assume that the rate-determining step, if accelerated
by steric bulk, is either the collapse of the anticipated cyclic
intermediate 4 (Scheme 1), which might also be a transition
state, or the decomplexation of the resulting vinyl group
from the catalyst. Since turnover frequencies are in the
range of approximately 1–40 per h, one of these intermedi-
Entry 25 X (x [mol%])
R
t [h] T [8C] Yield[a] [%] ee[b] [%]
1
2
3
4
5
6
7
8
h
h
j
k
k
l
0.5
0.2
1.0
1.0
0.2
1.0
1.0
1.0
H
H
CF3
Cl
Cl
Me
OMe 24
OMe 48
24
24
48
24
24
48
20
40
20
20
40
20
20
4
98
96
99
95
80
99
93
52
98
90
92
98
87
98
28
34
m
m
[a] Isolated yield. [b] The ee values determined by HPLC.
gate the electronic influence of the aryl substituents, a series
of 3-aryl-substituted imidates 25j–m was applied with vary-
ing substituents (Table 10, entries 3–8). Substrates bearing
electron-donating (p-Me) or -withdrawing substituents (p-
Cl, p-CF3) display an intrinsic thermal rearrangement rate
constant low enough to allow for a successful asymmetrical-
ly catalyzed reaction at RT, whereas 25m, bearing a
8738
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 8722 – 8741

Asymmetric Aza-Claisen Rearrangement
FULL PAPER
ates should be detectable, which was not the case in these
studies. However, the sterically overcrowded dimeric FcF
complexes might more readily form the catalytically active
monomeric species.
aza-Claisen rearrangement fulfilled this model. However,
the excellent ee values and activities obtained with 67a
show that at least the position of block (1) is not necessarily
relevant. Apparently, much more important is the choice of
(2) and (3) to completely shield the bottom, as in 77, to ach-
ieve a face-selective olefin coordination.
From an electronic point of view, the differences are more
pronounced since the type of N donor (oxazoline versus N-
Ts-imidazoline) as well as the overall charge of the ligand
(À1 for COP, 0 for 29-CpF-Ph-Ts after oxidation to the fer-
rocenium species) change. As PdII most likely acts as a car-
bophilic Lewis acid coordinating to an olefin, the lower elec-
tron density in 29-CpF-Ph-Ts is suitable to explain the
higher reactivity found for this complex. This is supported
by the solubility: 29-CpF-Ph-Ts is soluble in practically neat
substrate, thus allowing the reaction to be run at the highest
possible concentration.
Conclusion
The systematic studies presented herein have enabled us to
attain a more detailed understanding of the mode of action
of the catalytic asymmetric aza-Claisen rearrangement,
thereby resulting in the development of highly efficient cata-
lysts. Compound 67a, a hybrid system between Overmanꢄs/
Richardsꢄ COP-Cl and our FcF imidazoline system 29-CpF-
Ph-Ts, is more reactive than both systems and has a broader
substrate tolerance. However, one should be aware that FcF
oxazoline palladacycle 67a and FcF imidazoline palladacycle
29-CpF-Ph-Ts are not directly comparable for two main rea-
sons: the stereocenter next to the N-donor atom has the op-
posite configuration and with the current technology directly
comparable compounds with identical configurations cannot
be accessed. Moreover, subtle changes can have a big
impact: using the same mode of activation for 29-CpF-Ph-Ts
and 67a by virtue of the properties of AgTFA, imidazoline
palladacycle 29-CpF-Ph-Ts seemed to be superior and only
an additional screening of activation agents disclosed the ex-
cellent activity of 67a.
Comparing COP-X 7 to FcF oxazoline 67a, there are two
major differences: 1) the isopropyl residue on the oxazoline
moiety in 67a is pointing towards to the Cp’-spectator
ligand, thereby leaving the space unhindered above the Cp–
oxazoline–Pd plane, whereas in COP, the substituent is
pointing away from the sandwich core; 2) 67a can be oxi-
dized to a ferrocenium species, again providing a less elec-
tron-rich ligand.
Because a FcF oxazoline complex with an identical config-
uration as in COP could not be prepared so far, a direct
comparison is not possible. However, Overman and Ri-
chards reported a tert-butyl-substituted COP derivative in
which the substituent points towards the spectator ligand.[4b]
By using AgTFA for activation, significantly lower yields
and enantioselectivities were obtained relative to the analo-
gous complex with the iPr pointing away from the sandwich
core. However, a different catalyst-activation method might
lead to different results.
The origin of the higher rate of 67a as compared to 29-
CpF-Ph-Ts might be explained by the better accessibility of
the PdII center in the former, as the exo face above the Pd
square plane is open, thus resulting in a facilitated olefin co-
ordination by means of an associative mechanism. Since the
ee values obtained with 67a are practically identical to those
obtained with 29-CpF-Ph-Ts, enantioselectivity originates
nearly exclusively from the planar chiral FcF backbone.
According to the previously accepted hypothesis about
the ideal catalyst design represented by 76 (Figure 4, left
side), the residue (1) next to the N-binding site should point
away from the sandwich complex core (2), and all previously
highly enantioselective sandwich complexes used for the
With the enhanced reactivity of 29-CpF-Ph-Ts and 67a,
the catalytic asymmetric aza-Claisen rearrangement has a
very broad scope. Now, the methodology not only allows for
the formation of highly enantioenriched primary allylic
amines, but also secondary and tertiary amines. Allylic
amines with quaternary N-substituted stereocenters are also
conveniently accessible. The reaction conditions tolerate
many important functional groups, thus providing stereose-
lective access to valuable functionalized building blocks, for
example, for the synthesis of unnatural amino acids. Our
studies suggest that face-selective olefin coordination is the
enantioselectivity-determining step, which is almost exclu-
sively controlled by the element of planar chirality.
Experimental Section
General procedure
Precatalyst activation: A solution of the dimeric precatalyst (1 equiv) in
dry CH2Cl2 (1 mL/3 mmol) was added to the corresponding silver salt
(3.75 equiv if nothing else is specified) in a dry, pear-shaped flask under
N2, sealed, shielded from light, and stirred for 3 h (Cp- and Cp*-based
complexes) or overnight (CpF-based complexes) at RT. The resulting sus-
pension was filtered under an N2 atmosphere through Celite/CaH2
(ꢁ1:1, ca. 5 mm thickness) and the filter cake was washed with dry
CH2Cl2 (1 mL/3 mmol). Proton sponge (as 0.1m solution in CH2Cl2, 2 to
4 equiv) was added.
Figure 4. Schematic representation of Overmanꢄs original model 76 (left
side) for the ideal catalyst geometry and the revised version 77 (right
side) after the present studies.
Catalysis: The calculated amount of activated catalyst as stem solution in
CH2Cl2 was transferred to a flask containing the imidate (1 equiv, for
Chem. Eur. J. 2009, 15, 8722 – 8741
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8739

R. Peters et al.
most experiments 60 mmol). A stream of N2 was passed through the flask
until the solvent volume reached 0.2 mL or less (for a 60 mmol experi-
ment). The flask was sealed with a plastic cap and stirred for the indicat-
ed time at the indicated temperature. After the reaction, the residue was
purified by filtration over silica gel (CyH/EtOAc 9:1 or pentane/EtOAc
9:1).
Weiss (Universitꢂt Stuttgart) for skillful experimental contributions
during their internships in our laboratory.
[1] Recent reviews: PdII: a) P. M. Henry, Handbook of Organopalladi-
um Chemistry for Organic Synthesis, Vol. 2, Wiley-Interscience, New
York, 2002, p. 2119; b) K. N. Fanning, A. G. Jamieson, A. Suther-
land, Curr. Org. Chem. 2006, 10, 1007; c) AuI/III and PtII: A. Fꢁrstner,
P. W. Davies, Angew. Chem. 2007, 119, 3478; Angew. Chem. Int. Ed.
2007, 46, 3410; d) Au: A. S. K. Hashmi, Chem. Rev. 2007, 107, 3180;
e) PtII: A. R. Chianese, S. J. Lee, M. R. Gagnꢅ, Angew. Chem. 2007,
119, 4118; Angew. Chem. Int. Ed. 2007, 46, 4042; f) C. Liu, C. F.
Bender, X. Han, R. A. Widenhoefer, Chem. Commun. 2007, 3607.
[2] Reviews: a) T. K. Hollis, L. E. Overman, J. Organomet. Chem. 1999,
576, 290; b) L. E. Overman, N. E. Carpenter, Org. React. 2005, 66,
1–107.
[3] a) C. E. Anderson, L. E. Overman, J. Am. Chem. Soc. 2003, 125,
12412; b) S. F. Kirsch, L. E. Overman, M. P. Watson, J. Org. Chem.
2004, 69, 8101; c) H. Nomura, C. J. Richards, Chem. Eur. J. 2007, 13,
10216; d) M. D. Swift, A. Sutherland, Tetrahedron 2008, 64, 9521.
[4] a) L. E. Overman, C. E. Owen, M. M. Pavan, C. J. Richards, Org.
Lett. 2003, 5, 1809; b) R. S. Prasad, C. E. Anderson, C. J. Richards,
L. E. Overman, Organometallics 2005, 24, 77; c) C. E. Anderson, Y.
Donde, C. J. Douglas, L. E. Overman, J. Org. Chem. 2005, 70, 648;
d) R. Peters, Z.-q. Xin, D. F. Fischer, W. B. Schweizer, Organometal-
lics 2006, 25, 2917; e) M. E. Weiss, D. F. Fischer, Z.-q. Xin, S. Jautze,
W. B. Schweizer, R. Peters, Angew. Chem. 2006, 118, 5823; Angew.
Chem. Int. Ed. 2006, 45, 5694; f) S. Jautze, P. Seiler, R. Peters,
Angew. Chem. 2007, 119, 1282; Angew. Chem. Int. Ed. 2007, 46,
1260; g) D. F. Fischer, Z.-q. Xin, R. Peters, Angew. Chem. 2007, 119,
7848; Angew. Chem. Int. Ed. 2007, 46, 7704; h) S. Jautze, P. Seiler,
R. Peters, Chem. Eur. J. 2008, 14, 1430; i) Z.-q. Xin, D. F. Fischer, R.
Peters, Synlett 2008, 1495.
Di-m-acetatobis
dienyl,1-C,3’-N}-(h5-pentaphenylcyclopentadiene)ferrocene]dipalladium
(66a): A solution of oxazoline 65a (400 mg, 0.59 mmol) and Pd(OAc)2
ACHTUNGTRENNUNG
[{h5-(S)-(Sp)-2-[2’-(4’-methylethyl)oxazolinyl]cyclopenta-
AHCTUNGTRENNUNG
(132 mg, 0.59 mmol, 1 equiv) in glacial acetic acid (2 mL) was heated in a
preheated oil bath to 958C for 30 min, furnishing a red precipitate. The
mixture was cooled to room temperature and the solid product was sepa-
rated by filtration and washed with further glacial acetic acid (2 mL),
1
showing only one diastereomer in H NMR spectroscopy. For further pu-
rification, 66a was dissolved in a minimum of DCE (ca. 2 mL for 0.5 g of
crude complex) and transferred into a crystallization beaker (ca. 3 mm
height of the solution, 5 cm diameter), which was placed into a desiccator
containing n-pentane (height ca. 2 cm, ca. 25 cm diameter). After gener-
ally 1 d, dark red crystals of chemically and diastereomerically pure 66a
had formed. The supernatant was decanted and the crystals were dried in
vacuo
(460 mg,
0.27 mmol,
93%).
C96H82Fe2N2O6Pd2;
Mr =
1684.14 gmolÀ1
;
m.p. 230.0–231.38C (decomp); [a]2D3:4 =À1018.5 (c=
0.274 gdLÀ1 in CHCl3); 1H NMR (300 MHz, CDCl3, 218C): d=7.25–7.01
(m, 25H; arom. H), 4.25 (d, J=2.1 Hz, 1H; m-C5H3R3), 4.24 (d, J=
2.1 Hz, 1H; o-C5H3R3), 3.95 (d, J=1.5 Hz, 1H; o-C5H3R3), 3.84–3.69 (m,
2H; OCH2), 2.90–2.83 (m, 1H; NCH), 2.06 (s, 3H; CH3COO), 2.00–1.95
(m, 1H; CHACHTUNGTRENNUNG(CH3)2), 0.61 (d, J=7.2 Hz; CH3CHCH3), 0.03 ppm (d, J=
6.9 Hz; CH3CHCH3); 13C NMR (75 MHz, CDCl3, 218C): d=181.1, 177.1,
135.4, 132.7, 132.4, 126.9, 126.2, 90.8, 88.5, 77.8, 77.7, 75.8, 71.2, 70.8, 66.8,
28.2, 24.2, 20.7, 14.5 ppm; IR (film): n˜ =3057, 2961, 1576, 1503, 1413, 909,
738, 699 cmÀ1; MS (MALDI): m/z (%): 782 (100) [M+H]+; HRMS
(MALDI): m/z: calcd for C46H39FeNOPd (loss of bridging OAc ligand):
782.1350; found: 782.1365; elemental analysis calcd (%) for
C96H82Fe2N2O6Pd2: C 67.45, H 5.17, N 1.71; found: C 67.73, H 5.14, N
1.67.
[5] L. E. Overman, Angew. Chem. 1984, 96, 565; Angew. Chem. Int. Ed.
Engl. 1984, 23, 579.
Di-m-chlorobis
dienyl,1-C,3’-N}(h5-pentaphenylcyclopentadiene)ferrocene]dipalladium
(67a): To suspension of acetate bridged complex 66a (145 mg,
ACHTUNGTRENNUNG
[{h5-(S)-(Sp)-2-[2’-(4’-methylethyl)oxazolinyl]cyclopenta-
[6] O. Mumm, F. Mçller, Ber. Dtsch. Chem. Ges. 1937, 70, 2214.
[7] a) L. E. Overman, J. Am. Chem. Soc. 1974, 96, 597; b) L. E. Over-
man, J. Am. Chem. Soc. 1976, 98, 290; c) L. E. Overman, Angew.
Chem. 1984, 96, 565; Angew. Chem. Int. Ed. Engl. 1984, 23, 579.
[8] The use of COP-Cl was independently reported for the rearrange-
ment of allylic benzimidates: a) J. Kang, T. H. Kim, K. H. Yew,
W. K. Lee, Tetrahedron: Asymmetry 2003, 14, 415. For alternative
catalysts for this substrate class, see reference [2a] and: b) A.
Moyano, M. Rosol, R. M. Moreno, C. Lꢆpez, M. A. Maestro,
Angew. Chem. 2005, 117, 1899; Angew. Chem. Int. Ed. 2005, 44,
1865; c) J. Kang, K. H. Yew, T. H. Kim, D. H. Choi, Tetrahedron
Lett. 2002, 43, 9509.
a
0.086 mmol) in MeOH (20 mL), benzene (5 mL) and LiCl (250 mg) were
added. The reaction mixture was stirred at RT for 1 h, then diluted with
water and the phases were separated. The organic phase was washed
with brine and dried over MgSO4. All volatiles were removed under re-
duced pressure to yield the title product 67a as dark red solid (138 mg,
0.082 mmol, 95%), which did not require further purification.
C92H76Fe2N2O2Cl2Pd2;
Mr =1636.36 gmolÀ1
;
m.p.
204.5–205.58C
(decomp); [a]D22:7 =À1129.4 (c=0.275 gdLÀ1 in CHCl3); 1H NMR
(300 MHz, CDCl3, 218C): d=7.36–7.24 and 7.09–6.99 (m, 25H; arom. H),
4.46–4.10 (m, 5H; o-C5H3R3, m-C5H3R3, OCH2), 3.87–3.84 (m, 1H;
NCH), 2.41 (m, 1H; CHACTHNUTRGNE(UNG CH3)2), 0.81–0.75 (m, 3H; CH3CHCH3), 0.08–
[9] R. Peters, D. F. Fischer, Org. Lett. 2005, 7, 4137.
0.05 ppm (m, 3H; CH3CHCH3); 13C NMR (75 MHz, CDCl3, 218C): d=
178.0, 135.0, 134.8, 132.6, 126.7, 126.2, 125.9, 94.8, 88.8, 88.7, 76.5, 75.8,
71.4, 71.1, 67.9, 28.8, 20.7, 14.1 ppm; IR (film): n˜ =3057, 2960, 2365, 1602,
1505, 1444, 1372, 1185, 1028, 909, 737, 700 cmÀ1; MS (MALDI): m/z (%):
1636.3 (100) [M+H]+; HRMS (MALDI): m/z calcd for
C92H76Fe2N2O2Cl2Pd2: 1636.3646; found: 1636.2240; elemental analysis
calcd (%) for C92H76Fe2N2O2Cl2Pd2: C 67.42, H 4.80, N 1.71; found: C
67.23, H 5.04, N 1.79.
[10] K. Schlçgl, Top. Stereochem. 1967, 1, 39.
[11] CCDC 299669 (20) contains the supplementary crystallographic data
for this paper. This data can be obtained free of charge from The
Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/
data_request/cif.
[12] C. J. Richards, T. Damalidis, D. E. Hibbs, M. B. Hursthouse, Synlett
1995, 74.
[13] Only few direct diastereoselective cyclopalladations of chiral ferro-
cenes are known. For a recent overview, see: J.-P. Djukic, A. Hijazi,
H. D. Flack, G. Bernardinelli, Chem. Soc. Rev. 2008, 37, 406–425.
[14] Recent reviews about palladacycles: a) J. Dupont, M. Pfeffer, J.
Spencer, Eur. J. Inorg. Chem. 2001, 1917; b) R. B. Bedford, Chem.
Commun. 2003, 1787; c) M. E. van der Boom, D. Milstein, Chem.
Rev. 2003, 103, 1759; d) J. T. Singleton, Tetrahedron 2003, 59, 1837;
e) F. Bellina, A. Carpita, R. Rossi, Synthesis 2004, 2419; f) R. B.
Bedford, C. S. J. Cazin, D. Holder, Coord. Chem. Rev. 2004, 248,
2283; g) I. Omae, Coord. Chem. Rev. 2004, 248, 995; h) I. P. Belet-
skaya, A. V. Cheprakov, J. Organomet. Chem. 2004, 689, 4055; i) J.
Dupont, C. S. Consorti, J. Spencer, Chem. Rev. 2005, 105, 2527; j) J.
Dupont, M. Pfeffer, Palladacycles, Wiley-VCH, Weinheim, 2008.
Acknowledgements
This work was financially supported by Eidgençssische Technische Hoch-
schule (ETH) Zurich research grants TH-30/04-2 and TH-01 07-1, F.
Hoffmann-La Roche, Universitꢂt Stuttgart, ETH Zurich, Novartis (Mas-
ter’s fellowship to M.E.W. and Ph.D. fellowship to Z.-q.X.), and the Poli-
tecnico di Milano (Ph.D. exchange fellowship for A.B.). We thank Lukas
Dialer, Katrin Niedermann, and Nico Santschi (all ETHZ); and Marcel
8740
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 8722 – 8741

Asymmetric Aza-Claisen Rearrangement
FULL PAPER
[15] With AgTFA only 27% yield was obtained for 26a under identical
conditions.
[16] CCDC 606393 (28-Cp-tBu-Ts) contains the supplementary crystallo-
graphic data for this paper. This data can be obtained free of charge
from The Cambridge Crystallographic Data Centre via www.ccdc.
cam.ac.uk/data_request/cif.
addition on ketones and ketimines: O. Riant, J. Hannedouche, Org.
Biomol. Chem. 2007, 5, 873; e) for the catalytic asymmetric forma-
tion of N-substituted quaternary stereocenters, see, for example: X.
Liu, H. Li, L. Deng, Org. Lett. 2005, 7, 167, and references therein.
[32] a) A. Aubry, D. Bayeul, G. Precigoux, M. Pantano, F. Formaggio, M.
Crisma, C. Toniolo, W. H. J. Boesten, H. E. Schoemaker, J. Kam-
phuis, J. Chem. Soc. Perkin Trans. 2 1994, 525; b) T. S. Yokum, T. J.
Gauthier, R. P. Hammer, M. L. McLaughlin, J. Am. Chem. Soc.
1997, 119, 1167; c) B. Jaun, M. Tanaka, P. Seiler, F. N. M. Kuhnle, C.
Braun, D. Seebach, Liebigs Ann./Recl. 1997, 1697.
[17] F. R. Hartley, Chem. Soc. Rev. 1973, 2, 163.
[18] CCDC 604468 (35-CpF-Ph-Ts) and 606392 (34-Cp*-Ph-Ts) contain
the supplementary crystallographic data for this paper. This data
can be obtained free of charge from The Cambridge Crystallograph-
ic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
[19] The stereodescriptors with regard to the planar chirality are used ac-
cording to ref. [10].
[33] M. C. Khosla, K. Stachowiak, R. R. Smeby, F. M. Bumpus, F. Piriou,
K. Lintner, S. Fermandjian, Proc. Natl. Acad. Sci. USA 1981, 78,
757.
[20] Despite the different bulkiness of the substituents on the Cp rings,
the distance between the ring centers is almost constant in all mole-
cules.
[34] 3.75 equiv AgX were used to avoid an accidental excess (i.e.,
>4 equiv) of unreacted Ag+ ions at the expense of somewhat re-
duced catalytic activity.
[21] For similar results, see: T. P. Remarchuk, Ph.D. Thesis, University of
California (USA), 2003, pp. 175–235.
[22] See ref. [4h] and S. Jautze, R. Peters, Angew. Chem. 2008, 120, 9424;
Angew. Chem. Int. Ed. 2008, 47, 9284.
[35] Whereas 3-monosubstituted imidates 25 decompose only slowly on a
slightly acidic HPLC column (<0.1% [Bu4N]HSO4), 3,3-disubstitut-
ed imidates 42 decompose completely to 4-methoxytrifluoroacetani-
lide and the corresponding diene.
[23] Also readily prepared on 10 g scale from 1,2-di
G
[36] Recently, Overman and Bergman determined binding constants and
kinetic data for the rearrangement of allylic trichloroacetimidates
catalyzed by COP-X 7, see: M. P. Watson, L. E. Overman, R. G.
Bergman, J. Am. Chem. Soc. 2007, 129, 5031. In combination with a
computational analysis and according to the mechanistic model pre-
viously presented by us (see ref. [4e]), it was suggested that a coordi-
nation of the olefin occurs trans to N. The stereo-determining (as
well as the rate-determining) step was identified as the nucleophilic
attack of the N atom onto the activated olefin. To our understand-
ing, the different energies for olefin complexes resulting from coor-
dination to the enantiotopic olefin faces were not included in these
calculations.
dione (benzil) using a resolution step with mandelic or tartaric acid:
S. Pikul, E. J. Corey, Org. Synth. 1993, 71, 30.
[24] N-Methylimidazole is, for example, known to cleave palladacycle
dimers: C. Lꢆpez, R. Bosque, X. Solans, M. Font-Bardꢇa, D. Tra-
muns, G. Fern, J. Silver, J. Chem. Soc. Dalton Trans. 1994, 3039.
[25] F. Weygand, E. Frauendorfer, Chem. Ber. 1970, 103, 2437.
[26] For other applications of imidazolines in asymmetric catalysis, see
ref. [4d–i] and, for example: a) F. Menges, M. Neuburger, A. Pfaltz,
Org. Lett. 2002, 4, 4713; b) C. A. Busacca, D. Grossbach, R. C. So,
E. M. OꢄBrien, E. M. Spinelli, Org. Lett. 2003, 5, 595; c) S. Bhor, G.
Anilkumar, M. K. Tse, M. Klawonn, C. Dobler, B. Bitterlich, A.
Grotevendt, M. Beller, Org. Lett. 2005, 7, 3393; d) K. Ma, J. You,
Chem. Eur. J. 2006, 12, 1863; e) T. Arai, T. Mizukami, A. Yanagisa-
wa, Org. Lett. 2007, 9, 1145; f) S. Enthaler, B. Hagemann, S. Bhor,
G. Anilkumar, M. K. Tse, B. Bitterlich, K. Junge, G. Erre, M. Beller,
Adv. Synth. Catal. 2007, 349, 853; g) S. Nakamura, K. Hyodo, Y. Na-
kamura, N. Shibata, T. Toru, Adv. Synth. Catal. 2008, 350, 1443–
1448.
[27] For the first preparation of chiral ferrocenyl imidazolines, see
ref. [9]. For selected further preparations and/or applications, see
refs. [4d–i] and a) J. Ma, X. Cui, L. Gao, Y. Wu, Inorg. Chem.
Commun. 2007, 10, 762; b) J. Ma, X.-L. Cui, B. Zhang, M.-P. Song,
Y.-J. Wu, Tetrahedron 2007, 63, 5529.
[28] Review on regioselective allylic substitutions: G. Helmchen, A.
Dahnz, P. Dꢁbon, M. Schelwies, R. Weihofen, Chem. Commun.
2007, 675.
[29] a) D. B. Berkowitz, B. Wu, H. Li, Org. Lett. 2006, 8, 971. See also
b) P. Metz, C. Mues, A. Schoop, Tetrahedron 1992, 48, 1071; c) K.
Estieu, J. Ollivier, J. Salaun, Tetrahedron Lett. 1995, 36, 2975.
[30] Selected books and reviews: a) S. Yamamura, in The Alkaloids,
Vol. 29. (Ed.: A. Brossi), Academic Press, New York, 1986; b) M.
Hesse, in The Alkaloids: Natureꢀs Curse or Blessing?, Wiley-VCH,
Weinheim, 2002; c) G. A. Cordell, in The Alkaloids: Chemistry and
Biology, Vol. 60, Elsevier, San Diego, 2003; d) J. Kobayashi, H.
Morita, Alkaloids 2003, 60, 165; e) D. J. Ramon, M. Yus, Curr. Org.
Chem. 2004, 8, 149; f) Y. Ohfune, T. Shinada, Eur. J. Org. Chem.
2005, 5127; g) T. Reynolds, Phytochemistry 2005, 66, 1399.
[31] a) Recent reviews about the stereoselective formation of a,a-disub-
stituted a amino acids: H. Vogt, S. Brꢂse, Org. Biomol. Chem. 2007,
5, 406; b) C. Cativiela, M. D. Dꢇaz-de-Villegas, Tetrahedron: Asym-
metry 2007, 18, 569; c) review about the catalytic asymmetric synthe-
sis of a-branched chiral amines: S. Brꢂse, T. Baumann, S. Dahmen,
H. Vogt, Chem. Commun. 2007, 1881; d) review about the catalytic
asymmetric synthesis of quaternary carbon centers by nucleophilic
1
[37] The ee value of 43n was determined by H NMR spectroscopy after
removal of the trifluoroacetamide and PMP protecting groups and
condensation of the free amine with Mosherꢄs acid chloride (see the
Supporting Information).
[38] L. M. Harwood, K. J. Vines, M. G. B. Drew, Synlett 1996, 1051.
[39] See references [4c,29b] (nonenantioselective method) and a) M.
Calter, T. K. Hollis, L. E. Overman, J. Ziller, G. G. Zipp, J. Org.
Chem. 1997, 62, 1449; b) T. K. Hollis, L. E. Overman, Tetrahedron
Lett. 1997, 38, 8837; c) Y. Uozumi, K. Kato, T. Hayashi, Tetrahedron:
Asymmetry 1998, 9, 1065; d) Y. Jiang, J. M. Longmire, X. Zhang,
Tetrahedron Lett. 1999, 40, 1449; e) Y. Donde, L. E. Overman, J.
Am. Chem. Soc. 1999, 121, 2933; f) use of formimidates: P.-H.
Leung, K.-H. Ng, Y. Li, A. J. P. White, D. J. Williams, Chem.
Commun. 1999, 2435.
[40] K. Tamura, H. Mizukami, K. Maeda, H. Watanabe, K. Uneyama, J.
Org. Chem. 1993, 58, 32.
[41] This side reaction is believed to be catalyzed by traces of palladium
that remained in the batches despite chromatographic purification.
[42] CCDC 703503 (67a) contains the supplementary crystallographic
data for this paper. This data can be obtained free of charge from
The Cambridge Crystallographic Data Centre via www.ccdc.cam.
ac.uk/data_request/cif.
[43] For the cyclopalladation of a chiral 2-ferrocenyl oxazoline including
an X-ray crystal structure analysis, see the Supporting Information
of J.-B. Xia, S.-L. You, Organometallics 2007, 26, 4869.
[44] Catalyst recycling by precipitation was not efficient, because the
precipitated material is almost inactive. We assume slow catalyst de-
À
composition by olefin insertion into the Cp Pd bond followed by b-
hydride elimination to be the reason, as we observe the formation
of Pd black. Purification of the ferrocenium species by chromatogra-
phy was also not successful.
Received: March 19, 2009
Published online: August 18, 2009
Chem. Eur. J. 2009, 15, 8722 – 8741
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
8741