Azaphilic versus Carbophilic Coupling at C=N Bonds: Key Steps in Titanium-Assisted Multicomponent Reactions

Article abstract of DOI:10.1002/chem.201503732

Consecutive C- and N-arylation of N-heterocyclic nitriles is mediated by titanium(IV) alkoxides. The carbo- and azaphilic arylation step may be separated by choosing the order in which the two equivalents of aryl transfer reagent are added. In the course of this transformation, the ancillary N-heterocycle acts as both a directing anchor group and electron reservoir. In the selectivity-determining step, the selectivity is governed by a choice between (direct) C- and Ti-arylation; the latter opens up a reaction pathway that allows further migration to the nitrogen atom. The isolation of metal-containing aggregates from the reaction mixture and computational studies gave insights into the reaction mechanism. Subsequently, a multicomponent one-pot protocol was devised to rapidly access complex quaternary carbon centers.

Full text of DOI:10.1002/chem.201503732

DOI: 10.1002/chem.201503732
Full Paper
&
Synthesis Design
Azaphilic versus Carbophilic Coupling at C=N Bonds: Key Steps in
Titanium-Assisted Multicomponent Reactions
[a]
[a]
[b]
[a]
Torsten Roth, Hubert Wadepohl, Eric Clot,* and Lutz H. Gade*
In memoriam Peter Hofmann
Abstract: Consecutive C- and N-arylation of N-heterocyclic
nitriles is mediated by titanium(IV) alkoxides. The carbo- and
azaphilic arylation step may be separated by choosing the
order in which the two equivalents of aryl transfer reagent
are added. In the course of this transformation, the ancillary
N-heterocycle acts as both a directing anchor group and
electron reservoir. In the selectivity-determining step, the se-
lectivity is governed by a choice between (direct) C- and Ti-
arylation; the latter opens up a reaction pathway that allows
further migration to the nitrogen atom. The isolation of
metal-containing aggregates from the reaction mixture and
computational studies gave insights into the reaction mech-
anism. Subsequently, a multicomponent one-pot protocol
was devised to rapidly access complex quaternary carbon
centers.
Introduction
the substrate; this results in a conformation that favors N-addi-
tion as well as polarization of the C=N bond due to the elec-
tron-withdrawing properties of the adjacent functional group
(in a-iminoesters). Another qualitative rationale refers to the
similarity of this structural motif in a-iminoesters to the well-
known nucleophilic addition to vinylogous systems. Unfortu-
nately, there is limited insight into this pattern of reactivity
available from studies focusing on the underlying organome-
tallic chemistry, experimental mechanistic work, or computa-
tional modeling.
In general, the reaction of nucleophilic alkyl or aryl metal com-
pounds with C=N groups yields the corresponding C-alkylated
[
1]
or -arylated products. Nevertheless, a number of cases are
known in which this “normal” regioselectivity is suppressed in
[
2,3]
favor of an azaphilic addition (“umpolung”).
An early exam-
ple of this unexpected addition mode was reported by Kagan
and Fiaud in 1971 for the reaction of Grignard reagents with
[
4a]
a-iminoesters,
a viable route to a-amino esters.
in special electronic environments, such as cyclopentadiene-
and since then has been expanded into
[
4,5]
Additionally, imino groups
Herein, we demonstrate how the polarization and orienta-
tion of metal-coordinated ketimides may influence the regio-
chemistry of Grignard additions to CN multiple bonds. The
focal point in the assembly is tetravalent titanium, which ef-
fects the fixation and orientation of the substrates, and thus
[
6]
imines, have been found to follow similar reactive patterns.
The observed azaphilic regioselectivity has been attributed,
among other things, to the potentially chelating coordination
2
+
[7,8]
of a Lewis acidic metal cation (Mg for Grignard reagents) by
controls the relative approach of key reagents. The presence
of additional functional groups that act as ligating units may
significantly alter the preferred coordination mode of the keti-
mide and, as will be demonstrated, the observed pattern of re-
activity (Figure 1).
[
a] T. Roth, Prof. H. Wadepohl, Prof. L. H. Gade
Anorganisch-Chemisches Institut, Universität Heidelberg
Im Neuenheimer Feld 270, 69120 Heidelberg (Germany)
E-mail: lutz.gade@uni-heidelberg.de
[
b] Dr. E. Clot
Institut Charles Gerhardt, CNRS 5253, cc 1501
UniversitØ de Montpellier, Place Eug›ne Bataillon
4000 Montpellier (France)
3
E-mail: eric.clot@umontpellier.fr
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201503732. It contains preparative methods,
spectroscopic data, and full computational details.
Figure 1. Schematic representation of opposing patterns of reactivity.
ꢀ
2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA.
This is an open access article under the terms of Creative Commons Attri-
bution NonCommercial-NoDerivs License, which permits use and distribu-
tion in any medium, provided the original work is properly cited, the use is
non-commercial and no modifications or adaptations are made.
Chem. Eur. J. 2015, 21, 18730 – 18738
18730 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Results and Discussion
Table 1. Variation of the N-heterocyclic unit.
Consecutive carbo- and azaphilic addition of Grignard re-
agents to N-heterocyclic nitriles
Ketimide compounds have been generated in situ by the
group of de Meijere and others through the reaction of
Grignard reagents with various nitriles, including heterocyclic
derivatives (3- and 4-pyridine-, 2-thiophene-, 2-furan-carboni-
[
9]
trile, etc.) to yield the corresponding trityl amine products.
The starting point for our study was the observation that the
reaction of 2-pyridine carbonitrile with aryl Grignard reagents
under the same conditions as those described by de Meijere
et al., and subsequent hydrolytic workup gave N-arylated
amines, and thus, the products of azaphilic addition
(
Scheme 1).
Scheme 1. Regioselection due to chelation control.
Notably, and crucially, for the selectivity of this transformation,
the magnesium imide generated in this reaction step is not
susceptible to further reaction with excess Grignard reagent.
A single-crystal X-ray structure analysis of the ketimido–mag-
nesium compound 3 established the structural details of an oc-
tameric aggregate (Figure 2), in which the 2-pyridylimido unit
We hypothesized that this unexpected regioselectivity was
due to chelation control by the pyridyl ring, which generated
reactive intermediates coordinated to the metal center that, in
turn, could act as key species in metal-complex-induced, multi-
component one-pot reactions. The critical role played by the
N-heterocyclic ring within the substrate for this reactive pat-
tern was supported by control reactions (see the Supporting
Information) as well as the isolation of a series of other N-aryl
benzhydryl amines by the same titanium-mediated route
[
10]
acts as a chelating ligand for magnesium.
This leads to a bent structure of the metal–ketimido frag-
ment, which exposes the imido-N atom that acts as a bridging
ligating unit to a neighboring magnesium center. The ob-
served reactivity in subsequent transformations indicates at
least partial fragmentation of the octamer under the reaction
conditions (Figure 2).
(
Table 1).
Significantly, we were able to separate the carbo- and aza-
philic arylation steps, and thus, selectively introduce different
aryl substituents, by choosing the order in which the two
equivalents of Grignard reagent were added (Table 2): the first
equivalent of Grignard reagent cleanly added to the nitrile
carbon, whereas the second addition only took place in the
The subsequent reaction step, transmetalation with the tita-
nium reagent, was studied by reacting 3 with [Ti(OiPr) ] and
4
1
13
[TiBr(OiPr) ] (Scheme 2). In both cases, in situ H and C NMR
3
spectroscopy as well as mass spectrometry results indicated
the formation of a mixture of molecular species that defied
presence of [Ti(OiPr) ] and occurred at the nitrogen atom. The
characterization. However, with [TiBr(OiPr) ], we isolated a crys-
4
3
reaction is not limited to Grignard reagents nor to the pres-
ence of magnesium salts, since the corresponding lithium
compounds are equally applicable in this transformation (com-
pounds 2j and 2k).
talline reaction product that could be characterized, inter alia,
by XRD (Figure 3). The product was a Ti/Mg heterodinuclear
complex 4, in which the Ti atom is coordinated by a bromide
ligand, two remaining isopropoxido units and two chelating 2-
pyridylimido units, one of which bridges the titanium and
magnesium centers. The coordination sphere of magnesium is
completed by a halogenido ligand and a coordinated THF mol-
ecule.
Isolation and structural characterization of a magnesium
imide and a magnesium–titanium heterobimetallic imido
complex
The 2-pyridylimides act as bidentate chelates, and thus, fea-
ture bent ketimido–metal units. This form of ligation, which is
significant for the subsequent transformations discussed
below, should be seen in comparison with the large number of
structurally characterized ketimides of the early transition
metals, most of which adopt near-linear C=NÀmetal arrange-
To obtain mechanistic insights into this intriguing transforma-
tion, we aimed to isolate crystalline products of the individual
reaction steps that might represent or be related to reaction
intermediates. First, Grignard addition onto the nitrile moiety
was investigated for the addition of 4-fluorophenylmagnesium
bromide to 2-pyridine carbonitrile in THF at room temperature.
[
11]
ments.
Chem. Eur. J. 2015, 21, 18730 – 18738
www.chemeurj.org
18731 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 2. Variation of the alkyl/aryl organometallics. The molecular struc-
ture of product 2i is shown in the bottom-right corner. Thermal ellipsoids
are set at the 50% probability level.
Figure 2. Top: In situ formation of oligomeric Mg–ketimido clusters. Bottom:
Octameric Mg–ketimide cluster 3 arranged around a central Mg Br core.
4 4
Thermal ellipsoids are set at the 50% probability level. Hydrogen atoms and
cocrystallized solvent molecules have been omitted for clarity. The central
cube has been highlighted. For selected bond lengths and angles, see the
Supporting Information.
The following key reaction step of the transformation, lead-
ing to the reaction products represented in Table 1 and 2, in-
volves attack of the second Grignard reagent and determines
whether carbo- or azaphilic coupling is favored. Given the
complex equilibria of aggregates encountered in this type of
chemistry, which precludes in situ characterization by spectro-
scopic means as well as systematic kinetic studies, additional
insight had to be drawn from a computational study (PBE0 cal-
culations) by employing an appropriately chosen model
system (see the Supporting Information for computational de-
tails).
Scheme 2. In situ formation of labile titanium- and magnesium-containing
heterometallic species.
plex features three alkoxy ligands (herein modeled by OMe)
À
and one ketimide N =C (Py)(Ar) (Py=2-pyridine; Ar=para-
a
b
fluorophenyl) with either coordination of the pyridine substitu-
ent (A-cyc) or lack of pyridine coordination (A). The two opti-
mized geometries are shown in Figure 4 and A-cyc is comput-
DFT modeling of aza- versus carbophilic coupling to C=N
bonds
À1
ed to be more stable than A by DG=À3 kcalmol . The geom-
In the first stage, the simplest product of transmetalation be-
etry of A has a pseudo-C3 geometry with three almost identi-
v
tween 3 and Ti(OiPr) (or TiBr(OiPr) ) was considered. This com-
cal TiÀO bonds (1.784, 1.788, and 1.802 Š) and a significantly
4
3
Chem. Eur. J. 2015, 21, 18730 – 18738
www.chemeurj.org
18732 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
tween N and C is delocalized on the Ti center with overall
a
b
weights of 2.7, 57.1, and 37.1% on Ti, N , and C , respectively.
a
b
For the same NLMO, the respective weights are 1.9, 52.2, and
2.1% in A-cyc. This indicates that the N =C bond is more po-
4
a
b
larized toward C in A-cyc than in A. Therefore, the carbon
b
atom is more electron-rich in A-cyc than that in A, and thus,
less prone to attack by a nucleophile. This is also reflected in
the natural population analysis (NPA) charge for C , which is
b
more positive in A than that in A-cyc (0.286 vs. 0.215). In other
words, the binding of pyridine lowers the intrinsic reactivity of
Cb towards a nucleophilic reagent.
Coordination of Mg(Br)(Ar) (Ar=parafluorophenyl) to either
À1
A or A-cyc is exergonic by DG=À7.7 and À17.5 kcalmol , re-
spectively. The geometries of adducts B and B-cyc are shown
in Figure 4. In adduct B, the Mg cation is bridging between Na
Figure 3. Isolated heterometallic titanium- and magnesium-containing
adduct 4. Thermal ellipsoids are set at the 50% probability level. Hydrogen
atoms and cocrystallized solvent molecules have been omitted for clarity.
For selected bond lengths and angles, see the Supporting Information.
and one methoxy ligand (N ···Mg=2.286 Š and Mg···O=
a
2
.122 Š). The TiÀN ÀC linkage deviates from linearity in
a
b
adduct B with Ti-N -C =1488. The N =C bond is slightly elon-
a
b
a
b
gated (1.284 Š) and the NLMO associated with the p bond is
more strongly polarized toward
N in adduct B than that in the
a
mononuclear reference system A
(
60.1% N , 34.2% C , 2.9% Ti;
a b
charge on C in B: 0.334). Coor-
b
dination of Mg(Br)(Ar) has in-
creased the reactivity of Cb
toward a nucleophile. In B-cyc,
the Mg cation is bridging be-
tween Na and the apical me-
thoxy ligand (Mg···N =2.089 Š,
a
Mg···O=2.027 Š).
The initial trigonal bipyramidal
geometry of A-cyc is altered
upon coordination of MgBr(Ar)
to a pseudo-octahedral geome-
try with creation of a Ti···Br inter-
action (2.842 Š); thus explaining
the larger binding energy. As in
the case of B, the formation of
the adduct increases the reactivi-
ty of Cb toward a nucleophile.
The geometry of B-cyc is qualita-
tively similar to that observed
for the isolated hetereobimetallic
adduct (Figure 3).
Figure 4. Computed structures of the Ti complex fragments A and A-cyc and the TiÀMg heterodinuclear model
complexes B and B-cyc. The optimized geometries are available in a single xyz file (Geom.xyz) in the Supporting
Information. Color code: green=Ti, purple=Br, orange=Mg, yellow=F, red=O, blue=N, dark gray=C, light
gray=H.
The transition state, TS-B-C
shortened TiÀN distance of 1.865 Š, which is indicative of
(Figure 5), corresponding to the arylation of C in B has been
a
b
[12]
°
some double bond (i.e., azavinylidene) character. Upon pyri-
dine coordination in A-cyc, the geometry becomes a trigonal
bipyramidal with the apical pyridine unit and a methoxy
located on the potential energy surface with DG =25.9 kcal
À1
mol . In TS-B-C, the forming C···C bond length is 2.386 Š,
whereas the N =C bond is elongated to 1.333 Š and the MgÀ
a
b
ligand. The TiÀN bond is significantly elongated at 1.988 Š,
Ar bond length increases from 2.110 Š in B to a value of
a
whereas the N =C distance remains unchanged (1.269 Š, A;
2.185 Š in TS-B-C. The C-arylation is exergonic with DG=
a
b
À1
1
.268 Š, A-cyc).
A natural bonding orbital (NBO) analysis of the electronic
À12.9 kcalmol
and, in the product of C-arylation, C
(Figure 5), the Mg cation is bridging N (2.105 Š) and one me-
a
structure of A and A-cyc highlighted significantly different
electronic properties of the N =C bond in A and A-cyc. The
natural localized molecular orbital (NLMO) for the p bond be-
thoxy unit (1.976 Š).
In the transition state TS-B-C-cyc (Figure 5), in which pyri-
dine is coordinated to Ti, the forming C···C bond has an intera-
a
b
Chem. Eur. J. 2015, 21, 18730 – 18738
www.chemeurj.org
18733 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
With pyridine coordinated, the
activation barrier associated with
TS-B-D-cyc is slightly lower
°
À1
(
DG =4.8 kcalmol ) and the
transformation is still endergon-
ic, with D-cyc lying at DG=
À1
2
.8 kcalmol above B-cyc. Here
again, the transferring aromatic
ring is bridging Mg and Ti in the
product D-cyc (Mg···C=2.366 Š
and Ti···C=2.317 Š, see Figure 6).
From intermediate D, N-aryla-
tion is effective through TS-D-E
with an activation barrier of
Figure 5. Optimized geometries of the various extrema located along the pathway for direct C-arylation without
left) or with (right) pyridine coordination. The optimized geometries are available in a single xyz file (Geom.xyz)
in the Supporting Information. Color code: green=Ti, purple=Br, orange=Mg, yellow=F, red=O, blue=N, dark
gray=C, light gray=H.
(
°
À1
tomic distance similar to that observed in TS-B-C (2.380 Š).
However, the N =C bond is less elongated (1.317 Š) and the
DG =20.8 kcalmol and leads to the N-arylated product E
À1
with DG=À12.0 kcalmol . In TS-D-E, the migrating aryl
a
b
breaking Mg···Ar bond is longer (2.290 Š) in TS-B-C-cyc. This is
group is bridging the TiÀN bond (Ti···C=2.190 Š and N ···C=
a
a
in agreement with a greater reactivity of C toward nucleophil-
1.784 Š) and the N =C bond is elongated to 1.342 Š.
b
a
b
ic attack in B than that in B-cyc.
The product of the reaction, E, features a ketimine moiety,
N (Ar)=C (Py)(Ar), in which the Ti center interacts both with
As a matter of fact, the activation barrier associated with TS-
a
b
°
À1
B-C-cyc is greater (DG =30.4 kcalmol ) than that for TS-B-C
the N ÀC (Ti···N =2.081 Š and Ti···C =2.112 Š) and C ÀAr
a
b
a
b
b
°
À1
(
DG =25.9 kcalmol ). There is thus a significant decrease in
bonds (Ti···Ar=2.482 Š), whereas the Mg cation interacts with
the C-arylation reactivity upon coordination of pyridine.
N (2149 Š) and two methoxy groups.
a
From B or B-cyc, no transition-state structures associated
Interestingly, the energetics associated with the correspond-
with direct transfer of Ar from Mg to N could be located on
ing N-arylation from D-cyc is similar, with TS-D-E-cyc lying at
a
°
À1
the potential energy surface. However, in both cases, a transi-
tion-state structure associated with Ar transfer from Mg to Ti
could be located. In the case of TS-B-D (Figure 6), when pyri-
DG =21.0 kcalmol above D-cyc and E-cyc lying at DG=
À1
À13.1 kcalmol below D-cyc. However, this energetic similari-
ty does not translate into similar geometric parameters. The
migrating aromatic ring is further away from both Ti (Ti···C=
2.463 Š) and N (N ···C=1.930 Š) in TS-D-E-cyc compared with
°
dine is not coordinated, the activation barrier is low, DG =
À1
8
6
.9 kcalmol , and the transformation is endergonic, DG=
a
a
À1
.9 kcalmol for D relative to B. The product of the reaction,
TS-D-E. Only the N =C bond exhibits a similar lengthening
a b
D, features an aromatic ring bridging Mg and Ti (Mg···C=
.247 Š and Ti···C=2.353 Š, see Figure 6).
(1.344 Š). Because of pyridine coordination, the ketimine
2
moiety, N (Ar)=C (Py)(Ar), in product E-cyc only interacts with
a
b
Ti through the N =C bond
a
b
(
Ti···N =2.057 Š and Ti···C =
a b
2
.272 Š). The Mg cation is bridg-
ing Na (Mg···N =2.088 Š) and
a
one methoxy group.
In Figure 7a, a comparison be-
tween the pathways for C- and
N-arylation when pyridine is not
coordinated is shown (decoord
pathway). The C-arylation path-
way is preferred both kinetically
and
thermodynamically.
Al-
though aromatic ring transfer is
intrinsically easier from Ti to Na
than it is from Mg to C , the
b
energy needed to transfer the
aromatic ring from Mg to Ti
overall destabilizes the N-aryla-
tion pathway. A comparison be-
tween the pathways for C- and
N-arylation when pyridine is co-
ordinated (metallacycle pathway)
is shown in Figure 7b . Although
the product of C-arylation is pre-
Figure 6. Optimized geometries of the various extrema located along the pathway for N-arylation without (top) or
with (bottom) pyridine coordination. The optimized geometries are available in a single xyz file (Geom.xyz) in the
Supporting Information. Color code: green=Ti, purple=Br, orange=Mg, yellow=F, red=O, blue=N, dark
gray=C, light gray=H.
Chem. Eur. J. 2015, 21, 18730 – 18738
www.chemeurj.org
18734 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 7. Comparison between the pathways for C- and N-arylation without (a) and with pyridine coordination (b). Gibbs free energy values are expressed in
À1
À1
kcalmol at 333 K relative to B (path a) or to B-cyc (path b); B-cyc is 12.7 kcalmol more stable than B.
ferred thermodynamically, the formation of the N-arylated
product is preferred kinetically.
The kinetic preference for the N-arylation pathway is the
result of two cooperative effects. As already explained, pyridine
coordination reduces the intrinsic reactivity of C toward direct
b
reaction with a nucleophile. This leads to an increase in the
energy of TS-B-C-cyc with respect to B-cyc, compared with the
energy of TS-B-C with respect to B. In addition, aromatic ring
Scheme 3. Internal rearrangement of the titanium–enediamido species E-cyc
transfer from Mg to Ti is slightly favored when pyridine is coor-
to the corresponding energetically favorable chelated magnesium/titanaazir-
dine complex F-cyc. Gibbs free energy values (kcalmol , 333 K) are ex-
pressed relative to B-cyc.
À1
À1
dinated (by ca. 4 kcalmol ; compare TS-B-D-cyc in Figure 7b
with TS-B-D in Figure 7a ).
Consequently, the transition state of the rate-determining
step for the C-arylation pathway is destabilized, whereas that
of the rate-determining step for N-arylation is stabilized when
pyridine is coordinated. This eventually leads to inversion of
the kinetically preferred pathway in favor of N-arylation when
a substituent on the arylated ketimide can coordinate to titani-
um, as observed experimentally (Table 1). When such coordina-
tion is not possible, the expected C-arylation product will be
obtained.
cies represents a key intermediate for the subsequent conver-
sions discussed below.
In recent work by Müller, Beckhaus and co-workers, aromatic
aldimine and ketimine complexes of titanocene were studied
to highlight the accessibility and close structural relationship
[
13a]
of three- and five-membered titanacycles (Figure 8).
A struc-
[
13b]
turally related example was reported by Rosenthal et al.
It
In summary, the proposed mechanism of the azaphilic CÀN
coupling has revealed that the initial idea of an umpolung, in
which a nucleophile is added onto the more electronegative
nitrogen, has to be revised. In fact, in the selectivity-determin-
ing step, the choice is not between N- and C-arylation, but be-
tween C- and Ti-arylation; the latter opens up a reaction path-
way that allows further migration to the nitrogen atom.
The reaction product for the pathway involving azaphilic
coupling, E-cyc, represents a local minimum on the free
energy hypersurface, and may rearrange to a pyridyl-ligated ti-
tanaaziridine complex, F-cyc, as depicted in Scheme 3. Com-
plex F-cyc is computed to be DG=À7.7 kcalmolÀ1 more stable
than E-cyc. It is therefore reasonable to assume that such a spe-
was rationalized that steric factors determined the preferred
coordination mode (titanaaziridine vs. 1-aza-2-titanacyclopent-
[
13]
4-ene species).
Figure 8. Structurally characterized titanaaziridine and azatitanacyclopentene
complexes.
Chem. Eur. J. 2015, 21, 18730 – 18738
www.chemeurj.org
18735 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Development of multicomponent reactions based on se-
quential C- and N-arylation steps
A key insight of the computational study described above has
been the likely formation of the titanaaziridine species F-cyc,
which may thus be viewed as a key intermediate for subse-
quent reaction steps in the multicomponent transformations
described herein. The reactivity of metallacyclic aziridines of
the Group 4 metals has been studied previously by several re-
search groups and has given rise to exciting examples of
[14]
metal-induced transformations. The most common pattern
of reactivity involves the coupling of the metal-bonded carbon
atom with various electrophilic reagents, including aldehydes
or alkyl halides. This opened up the possibility of developing
the one-pot formation of such species into multiple-compo-
nent reactions for synthetic targets, in which quaternary
carbon centers with four different substituents would be ac-
[15]
cessible in a chemoselective way.
Initially, we examined the addition of a range of different
electrophiles to the reaction mixture prior to hydrolytic
workup.
Upon the addition of an excess of paraformaldehyde, the ox-
azolidines 5a,b were isolated as the main products, depending
on the electronic nature of the N-aryl substituent (Scheme 4).
The oxazolidine is thought to be formed by a Mannich-type
mechanism from an initial b-aminoalcohol intermediate.
Subsequently, the reactive potential of the titanaaziridine in-
termediate towards various alkyl halides was investigated.
Both allyl and benzyl bromides afforded the corresponding tri-
substituted amines (Scheme 5).
Scheme 5. In situ trapping of the intermediary titanium species with alkyl
halides. The molecular structure of 6c is shown in the bottom-right corner.
Hydrogen atoms, except for N(1)ÀH have been omitted for clarity; thermal
ellipsoids are shown at the 50% probability level.
pot procedure, the halides were found to be inert under the
reaction conditions chosen for the formation of the active tita-
nium species, and therefore, could be added during the initial
setup of the reaction, rendering this a true one-pot, multicom-
ponent coupling.
Remarkably, and in contrast to the reaction involving formal-
dehyde, which had to be added in the final step of the one-
Recently, Micalizio and co-workers reported the reaction of
in situ generated azatitanacyclopropanes (generated from
[
16]
imines and the Sato reagent ) with allylic or allenylic alkox-
[
17a–d]
ides to yield homoallylic amines or dienes.
This approach
was used, inter alia, for the synthesis of complex natural prod-
ucts. Reacting allylic alkoxides with the titanium intermediate,
we observed the formation of the corresponding homoallylic
amines (Scheme 6).
Again, the allylic alkoxide could be added directly to the re-
action mixture and did not interfere with the TiÀN aryl transfer.
Using chiral allylic alkoxides as coupling partners, we obtained
the corresponding enantioenriched reaction products, albeit
[
18]
with a slight erosion of chirality. Chirality is thus transferred
from an easily accessible chiral allylic alcohol to a metallacycle,
thereby creating an enantioenriched quaternary carbon atom.
The selectivity observed in this transformation is consistent
with an empirical model based on a formal metallo-[3,3]-rear-
[
17]
rangement.
Finally, we exchanged the allylic alkoxides for allenylic alkox-
[
19]
ides. Instead of the expected diene-containing product, the
,5-aminoalkohols 8a and 8b were isolated (Scheme 7). The re-
sults of a workup in D O (8a-d ) suggest a reaction mechanism
1
2
1
through 1,2-carbometalation of the terminal allene p bond, in
contrast to the previously reported formal metallo-[3,3]-rear-
rangement.
Scheme 4. In situ trapping of the intermediary titanium species with formal-
dehyde. Hydrogen atoms in the molecular structure of 5a have been omit-
ted for clarity; thermal ellipsoids are shown at the 50% probability level.
Chem. Eur. J. 2015, 21, 18730 – 18738
www.chemeurj.org
18736 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Conclusion
We have developed a powerful new multicomponent strategy
for the rapid convergent assembly of structural complexity
starting from simple N-heterocyclic nitriles, Grignard reagents,
and electrophilic coupling agents. Notably, this process relies
on inexpensive, readily available starting materials, is highly
modular, and yields environmentally benign and easily separa-
ble byproducts.
The range of products thus accessible comprises valuable ni-
trogen-containing heterocycles, enantioenriched quaternary
carbon centers, and densely functionalized structural motifs of
pharmaceutical and biomedical relevance, as exemplified by
the class of cholesterylester trans-
fer protein (CETP) inhibitors shown
[
20,21]
in Figure 9.
Whereas the reaction products
described herein are accessible
through conventional approaches,
namely, nucleophilic addition to di-
arylimines, these methods require
Figure 9. CETP inhibitors com-
multiple steps and tedious workup prising an N-aryl tritylamine
[
22]
motif (Ms=mesyl).
procedures. The reaction mecha-
nism of the unexpected azaphilic
addition of nucleophiles to keti-
Scheme 6. Multicomponent reactions employing allylic alkoxides. The re-
spective starting materials are given in brackets. The molecular structure of
7
b is shown in the bottom-right corner. Hydrogen atoms, except for N(1)ÀH,
have been omitted for clarity; thermal ellipsoids are shown at the 50%
probability level. A rationale for the observed E selectivity of these transfor-
mations based on an empirical model is depicted in the Supporting Informa-
tion.
mide titanium complexes was explored by DFT methods,
which revealed the formal umpolung of a CÀN group as the
key step toward reactive intermediary metallacycles (titanaziri-
dines and azatitanacyclopentenes). A central role is attributed
to the tethered nitrogen donor moiety, which serves as a pre-
coordinating directing group and, in conjunction with an ancil-
lary p system, as an electron reservoir.
Finally, the reactivity profile of the titanacyclic intermediates
was assessed by four mechanistically distinct follow-up reac-
tions, namely, by insertion reactions with a C=O function, by
nucleophilic substitution of halogenated alkanes, by allylic alk-
oxides featuring a formal metallo-[3,3]-rearrangement, and by
allenic alkoxides resulting in a 1,2-carbometalation reaction.
Notably, these transformations show a high degree of regiose-
lectivity, although the control of absolute stereochemistry in
this process remains a challenge. Moreover, the step-economic
concept presented in this work shows the great utility of
Group 4 metals for mediating challenging CÀC and CÀN bond
formations.
Acknowledgements
We gratefully acknowledge the award of a Ph.D. grant to T.R.
from the Landesgraduiertenfçrderung (LGF Funding Program
of the state of Baden-Württemberg) and the University of Hei-
delberg for generous funding.
Scheme 7. A multicomponent reaction employing an allenylic alkoxide. The
molecular structure of 8a is shown in the bottom-right corner. Hydrogen
atoms, except for N(1)ÀH and O(1)ÀH, have been omitted for clarity; thermal
ellipsoids are shown at the 50% probability level.
Keywords: heterocycles · metallacycles
reactions · synthetic methods · titanium
· multicomponent
Chem. Eur. J. 2015, 21, 18730 – 18738
www.chemeurj.org
18737 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[
[
[
1] For review articles, see: a) S. Kobayashi, Y. Mori, J. S. Fossey, M. M. Salter,
Chem. Rev. 2011, 111, 2626–2704; b) G. K. Friestad, A. K. Mathies, Tetra-
hedron 2007, 63, 2541–2569; c) S. Kobayashi, H. Ishitani, Chem. Rev.
[11] For a review of Group 4 ketimide complexes, see: a) M. J. Ferreira, A. M.
Martins, Coord. Chem. Rev. 2006, 250, 118–132; for seminal work on
early-transition-metal ketimido complexes, see: b) K. Farmery, M. Kilner,
C. Midcalf, J. Chem. Soc. A 1970, 2279–2285; c) M. Kilner, C. Midcalf, J.
Chem. Soc. D 1970, 552–553; d) M. R. Collier, M. F. Lappert, J. McMeek-
ing, Inorg. Nucl. Chem. Lett. 1971, 7, 689–694.
[12] a) J. R. Hagadorn, J. Arnold, Organometallics 1998, 17, 1355–1368;
b) D. M. Giolando, K. Kirschbaum, L. J. Graves, U. Bolle, Inorg. Chem.
1992, 31, 3887–3890.
[13] a) F. Loose, I. Plettenberg, D. Haase, W. Saak, M. Schmidtmann, A. Schä-
fer, T. Müller, R. Beckhaus, Organometallics 2014, 33, 6785–6795; b) D.
Thomas, W. Baumann, A. Spannenberg, R. Kempe, U. Rosenthal, Organo-
metallics 1998, 17, 2096–2102.
[14] a) L. D. Durfee, I. P. Rothwell, Chem. Rev. 1988, 88, 1059–1079; Group 4:
b) S. L. Buchwald, M. W. Wannamaker, B. T. Watson, J. Am. Chem. Soc.
1989, 111, 776–777; c) S. L. Buchwald, B. T. Watson, M. W. Wannamaker,
J. C. Dewan, J. Am. Chem. Soc. 1989, 111, 4486–4494; d) R. B. Grossman,
W. M. Davis, S. L. Buchwald, J. Am. Chem. Soc. 1991, 113, 2321–2322;
e) M. C. J. Harris, R. J. Whitby, J. Blagg, Tetrahedron Lett. 1995, 36, 4287–
4290; see also refs. [7, 16, 17, 19]; for Group 5, see: f) J. M. P. Lauzon,
L. L. Schafer, Dalton Trans. 2012, 41, 11539–11550; g) E. Chong, P.
Garcia, L. L. Schafer, Synthesis 2014, 46, 2884–2896; h) A. L. Reznichen-
ko, K. C. Hultzsch, J. Am. Chem. Soc. 2012, 134, 3330–3311; i) P. Eisen-
berger, R. O. Ayinla, J. M. P. Lauzon, L. L. Schafer, Angew. Chem. Int. Ed.
2009, 48, 8361–8365; Angew. Chem. 2009, 121, 8511–8515; j) E. Chong,
J. W. Brandt, L. L. Schafer, J. Am. Chem. Soc. 2014, 136, 10898–10901.
[15] a) J. Christoffers, A. Baro, Quaternary Stereocenters: Challenges and Solu-
tions for Organic Synthesis, Wiley-VCH, Weinheim, 2005; b) J. P. Das, I.
Marek, Chem. Commun. 2011, 47, 4593–4623; c) J. Christoffers, A. Baro,
Adv. Synth. Catal. 2005, 347, 1473–1482; d) I. Denissova, L. Barriault, Tet-
rahedron 2003, 59, 10105–10146; e) K. Fuji, Chem. Rev. 1993, 93, 2037–
2066.
1999, 99, 1069–1094; d) R. Bloch, Chem. Rev. 1998, 98, 1407–1438;
e) D. Enders, U. Reinhold, Tetrahedron: Asymmetry 1997, 8, 1895–1946.
2] The term umpolung has been defined by Seebach as “any process by
which donor and acceptor reactivity of an atom are interchanged”: D.
Seebach, Angew. Chem. Int. Ed. Engl. 1979, 18, 239–258; Angew. Chem.
1979, 91, 259–278. See also: a) J. J. Eisch, J. Organomet. Chem. 1995,
5
00, 101–115; b) H. R. Guan, Curr. Org. Chem. 2008, 12, 1406–1430.
3] For a review of approaches relying on electrophilic amination, see: a) A.
Ricci, Modern Amination Methods, Wiley-VCH, Weinheim, 2000; b) T. J.
Barker, E. R. Jarvo, Synthesis 2011, 3954–3964; c) C. Greck, B. Drouillat,
C. Thomassigny, Eur. J. Org. Chem. 2004, 1377–1385; d) C. Greck, J. P.
GenÞt, Synlett 1997, 741–748; e) E. Erdik, M. Ay, Chem. Rev. 1989, 89,
1
5
947–1980; f) P. Starkov, T. F. Jamison, I. Marek, Chem. Eur. J. 2015, 21,
278–5300; for a selection of specific methods, see g) T. J. Barker, E. R.
Jarvo, Angew. Chem. Int. Ed. 2011, 50, 8325–8328; Angew. Chem. 2011,
123, 8475–8478; h) P. Dembech, G. Seconi, A. Ricci, Chem. Eur. J. 2000,
6, 1281–1286; i) I. Sapountzis, P. Knochel, Angew. Chem. Int. Ed. 2004,
43, 897–900; Angew. Chem. 2004, 116, 915–918.
[
4] Seminal work: a) J.-C. Fiaud, H. B. Kagan, Tetrahedron Lett. 1971, 12,
019–1022; recent work: b) M. Hatano, K. Yamashita, K. Ishihara, Org.
1
Lett. 2015, 17, 2412–2415; c) J. M. Curto, J. S. Dickstein, S. Berritt, M. C.
Kozlowski, Org. Lett. 2014, 16, 1948–1951; d) Y. Mizutani, H. Tanimoto, T.
Morimoto, Y. Nishiyama, K. Kakiuchi, Tetrahedron Lett. 2012, 53, 5903–
5
906; e) M. Hatano, K. Yamashita, M. Mizuno, O. Ito, K. Ishihara, Angew.
Chem. Int. Ed. 2015, 54, 2707–2711; Angew. Chem. 2015, 127, 2745–
749; f) J. S. Dickstein, M. W. Fennie, A. L. Norman, B. J. Paulose, M. C.
2
Kozlowski, J. Am. Chem. Soc. 2008, 130, 15794–15795.
5] For review articles, see: a) J. S. Dickstein, M. C. Kozlowski, Chem. Soc.
Rev. 2008, 37, 1166–1173; b) Y. Yamamoto, W. Ito, Tetrahedron 1988, 44,
[
[
5
415–5423; c) M. Shimizu, Pure Appl. Chem. 2006, 78, 1867; d) M. Shimi-
[16] a) F. Sato, S. Okamoto, Adv. Synth. Catal. 2001, 343, 759–784; b) F. Sato,
H. Urabe, S. Okamoto, Chem. Rev. 2000, 100, 2835–2886; c) F. Sato, H.
Urabe, S. Okamoto, Synlett 2000, 0753–0775.
zu, I. Hachiya, I. Mizota, Chem. Commun. 2009, 874–889; e) S. Hata, M.
Shimizu, J. Synth. Org. Chem. Jpn. 2011, 69, 1134–1144.
6] For cyclopentadieneimines, see: a) R. A. Hagopian, M. J. Therien, J. R.
Murdoch, J. Am. Chem. Soc. 1984, 106, 5753–5754; for fluorenylimines,
see: b) W. Dai, R. Srinivasan, J. A. Katzenellenbogen, J. Org. Chem. 1989,
[17] a) M. Takahashi, M. McLaughlin, G. C. Micalizio, Angew. Chem. Int. Ed.
2009, 48, 3648–3652; Angew. Chem. 2009, 121, 3702–3706; b) M. A.
Tarselli, G. C. Micalizio, Org. Lett. 2009, 11, 4596–4599; c) M. Z. Chen, M.
McLaughlin, M. Takahashi, M. A. Tarselli, D. Yang, S. Umemura, G. C. Mi-
calizio, J. Org. Chem. 2010, 75, 8048–8059; d) D. Yang, J. Belardi, G. C.
Micalizio, Tetrahedron Lett. 2011, 52, 2144–2147; e) I. L. Lysenko, H. G.
Lee, J. K. Cha, Org. Lett. 2009, 11, 3132–3134; for a brief rationalization
of the observed selectivities, see the Supporting Information; for exam-
ples with allenic alcohols, see ref. [19].
5
4, 2204–2208; for quinone imines, see: c) J. Honzl, M. Metalovµ, Tetra-
hedron 1969, 25, 3641–3652; d) J. Brachi, A. Rieker, Synthesis 1977,
08–711; for N-heterocycles, see: e) J. J. A. Campbell, S. J. Noyce, R. C.
7
Storr, J. Chem. Soc. Chem. Commun. 1983, 1344–1346; f) D. Hunter,
D. G. Neilson, J. Chem. Soc. Perkin Trans. 1 1984, 2779–2783; g) J.
Faragó, Z. Novµk, G. Schlosser, A. Csµmpai, A. Kotschy, Tetrahedron
2
004, 60, 1991–1996; h) S. Tolshchina, G. Rusinov, V. Charushin, Chem.
Heterocycl. Compd. 2013, 49, 66–91; i) M. C. Wilkes, J. Heterocycl. Chem.
991, 28, 1163–1164; j) Q. Zhou, P. Audebert, G. Clavier, F. Miomandre,
[18] Similarly, Micalizio et al. observed an erosion of optical purity in the
enantioselective allylic alcohol–imine coupling when examining allylic
alcohols without terminal substitution of the alkene, see ref. [17].
[19] For examples demonstrating the use of allenic alcohols in carbometala-
tion and formal metallo-[3,3]-rearrangments, see: a) H. L. Shimp, G. C.
Micalizio, Chem. Commun. 2007, 4531–4533; b) M. McLaughlin, H. L.
Shimp, R. Navarro, G. C. Micalizio, Synlett 2008, 735–738; c) H. L. Shimp,
A. Hare, M. McLaughlin, G. C. Micalizio, Tetrahedron 2008, 64, 3437–
3445. See also ref. [16b].
1
J. Tang, RSC Adv. 2014, 4, 7193–7195; k) D. Hunter, D. G. Neilson, T. J. R.
Weakley, J. Chem. Soc. Perkin Trans. 1 1985, 2709–2712; l) A. Ohsawa, T.
Kaihoh, H. Igeta, J. Chem. Soc. Chem. Commun. 1985, 1370–1371; for
bis(imino)pyridines, see: m) I. Khorobkov, S. Gambarotta, G. P. A. Yap, Or-
ganometallics 2002, 21, 3088–3090; n) G. K. B. Clentsmith, V. C. Gibson,
P. B. Hitchcock, B. S. Kimberley, C. W. Rees, Chem. Commun. 2002, 1498–
1
499.
[20] The construction of nitrogen–aryl bonds is of particular interest, see:
S. D. Roughley, A. M. Jordan, J. Med. Chem. 2011, 54, 3451–3479.
[21] Y. Wang, W. Yang, H. J. Finlay, L. S. Harikrishnan, J. Jiang, M. G. Kamau, K.
Van Kirk, D. S. Nirschl, D. S. Taylor, A. Y. A. Chen, X. Yin, P. G. Sleph, R. Z.
Yang, C. S. Huang, L. P. Adam, R. M. Lawrence, R. R. Wexler, M. E. Salvati,
Bioorg. Med. Chem. Lett. 2014, 24, 860–864.
[22] In particular, the 1,2-addition of organometallic reagents to imines is
a challenging transformation due to the innate low reactivity of the
substrate and generally requires activating groups; for recent reports,
see: a) A. Desmarchelier, P. Ortiz, S. R. Harutyunyan, Chem. Commun.
2015, 51, 703–706; b) R. Reingruber, S. Bräse, Chem. Commun. 2008,
105–107; see also ref. [1].
[
[
7] I. Marek, Titanium and Zirconium in Organic Synthesis, Wiley-VCH, Wein-
heim, 2002.
8] For reviews on metallacycle-mediated synthesis, see: a) U. Rosenthal,
V. V. Burlakov, M. A. Bach, T. Beweries, Chem. Soc. Rev. 2007, 36, 719–
7
2
5
28; b) S. Roy, U. Rosenthal, E. D. Jemmis, Acc. Chem. Res. 2014, 47,
917–2930; c) H. A. Reichard, G. C. Micalizio, Chem. Sci. 2011, 2, 573–
89; d) K. M. Doxsee, J. K. M. Mouser, J. B. Farahi, Synlett 1992, 13–21.
[
9] a) O. Tomashenko, V. Sokolov, A. Tomashevskiy, A. de Meijere, Synlett
007, 0652–0654; b) O. Tomashenko, V. Sokolov, A. Tomashevskii, A. Po-
2
tekhin, A. de Meijere, Russ. J. Org. Chem. 2007, 43, 1421–1426; c) R.
Wang, B. T. Gregg, W. Zhang, K. C. Golden, J. F. Quinn, P. Cui, D. O. Ty-
moshenko, Tetrahedron Lett. 2009, 50, 7070–7073; d) O. A. Tomashenko,
A. E. Rudenko, V. V. Sokolov, A. A. Tomashevskiy, A. de Meijere, Eur. J.
Org. Chem. 2010, 1574–1578.
[
10] For a (dimeric) solid-state structure of a magnesium ketimide complex,
see: K. Manning, E. A. Petch, H. M. M. Shearer, K. Wade, J. Chem. Soc.
Chem. Commun. 1976, 107–108.
Received: September 17, 2015
Published online on November 6, 2015
Chem. Eur. J. 2015, 21, 18730 – 18738
www.chemeurj.org
18738 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim