Mechanistic investigation of the iridium-catalysed alkylation of amines with alcohols

Article abstract of DOI:10.1039/c2ob06603c

The [Cp*IrCl₂]₂-catalysed alkylation of amines with alcohols was investigated using a combination of experimental and theoretical methods. A Hammett study involving a series of para-substituted benzyl alcohols resulted in a line with a negative slope. This clearly documents that a positive charge is built up in the transition state, which in combination with the measurement of a significant kinetic isotope effect determines hydride abstraction as being the selectivity-determining step under these conditions. A complementary Hammett study using para-substituted anilines was also carried out. Again, a line with a negative slope was obtained suggesting that nucleophilic attack on the aldehyde is selectivity-determining. A computational investigation of the entire catalytic cycle with full-sized ligands and substrates was performed using density functional theory. The results suggest a catalytic cycle where the intermediate aldehyde stays coordinated to the iridium catalyst and reacts with the amine to give a hemiaminal which is also bound to the catalyst. Dehydration to the imine and reduction to the product amine also takes place without breaking the coordination to the catalyst. The fact that the entire catalytic cycle takes place with all the intermediates bound to the catalyst is important for the further development of this synthetic transformation. The Royal Society of Chemistry 2012.

Full text of DOI:10.1039/c2ob06603c

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
Organic &
Biomolecular
Chemistry
Cite this: Org. Biomol. Chem., 2012, 10, 2569
www.rsc.org/obc
PAPER
Mechanistic investigation of the iridium-catalysed alkylation of amines with
alcohols†
Peter Fristrup,* Matyas Tursky and Robert Madsen*
Received 20th September 2011, Accepted 10th January 2012
DOI: 10.1039/c2ob06603c
The [Cp*IrCl ] -catalysed alkylation of amines with alcohols was investigated using a combination of
2
2
experimental and theoretical methods. A Hammett study involving a series of para-substituted benzyl
alcohols resulted in a line with a negative slope. This clearly documents that a positive charge is built up
in the transition state, which in combination with the measurement of a significant kinetic isotope effect
determines hydride abstraction as being the selectivity-determining step under these conditions. A
complementary Hammett study using para-substituted anilines was also carried out. Again, a line with a
negative slope was obtained suggesting that nucleophilic attack on the aldehyde is selectivity-
determining. A computational investigation of the entire catalytic cycle with full-sized ligands and
substrates was performed using density functional theory. The results suggest a catalytic cycle where the
intermediate aldehyde stays coordinated to the iridium catalyst and reacts with the amine to give a
hemiaminal which is also bound to the catalyst. Dehydration to the imine and reduction to the product
amine also takes place without breaking the coordination to the catalyst. The fact that the entire catalytic
cycle takes place with all the intermediates bound to the catalyst is important for the further development
of this synthetic transformation.
3
iridium catalysts where the iridium complex [Cp*IrCl ] has
Introduction
2 2
9
been one of the most effective catalysts. The proposed mechan-
ism involves dehydrogenation of the alcohol to the correspond-
ing carbonyl compound followed by imine formation and
hydrogenation to the amine with the hydride-complex formed in
the first step (Scheme 1). The reaction only produces a molecule
of water as by-product and does not require any stoichiometric
additives but only a catalytic amount of a base, e.g. carbonate.
The dehydrogenation of an alcohol and the hydrogenation of an
imine are also the key transformations in the transfer hydrogen-
ation of ketones and imines with isopropanol or formic acid–
The carbon–nitrogen bond is one of the most abundant covalent
bonds in organic chemistry and biochemistry and methods for
formation of C–N bonds belong to the very heart of organic syn-
thesis. In most cases, C–N bonds are formed by substitution of a
suitable leaving group or by reductive amination from a carbonyl
compound. These reactions, however, employ stoichiometric
reagents or produce stoichiometric amounts of waste. In recent
years, there has been an increasing awareness of the environ-
mental impact of many chemical processes and research has
been directed towards more atom economical and sustainable
10
triethyl amine. In this connection, the mechanism has been
1
11
12
reactions. This has led to the development of new methods for
studied both experimentally and theoretically, but mainly
with ruthenium complexes. Although, a number of iridium
the formation of C–N bonds and one of the most promising is
2,3
the direct metal-catalysed alkylation of amines with alcohols.
Besides alkylating simple amines this transformation can also be
4
used for synthesis of various heterocyclic ring systems such as
5
5
6
7
pyrrolidines, piperidines, piperazines, quinolines
and
8
2
indoles. The reaction can be achieved with ruthenium and
Department of Chemistry, Technical University of Denmark, Building
2
01, 2800 Kgs, Lyngby, Denmark. E-mail: pf@kemi.dtu.dk, rm@kemi.
dtu.dk; Fax: (+45) 4593 3968
Electronic supplementary information (ESI) available: Experimental
†
data from kinetic runs, KIE determination, XYZ coordinates, SCF ener-
gies, Gibbs free energies and IRC scan calculations. See DOI: 10.1039/
c2ob06603c
Scheme 1 Simplified catalytic cycle for iridium-catalysed alkylation of
amines with alcohols.
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 2569–2577 | 2569

View Article Online
complexes catalyse the transfer hydrogenation of ketones and
imines, the mechanism of these reactions has not been as exten-
1
3
sively studied as in the case of ruthenium. Notably, all the
investigated iridium complexes were shown to transfer hydrogen
from secondary alcohols to ketones through an iridium–monohy-
1
3
dride pathway.
Scheme 2 Alkylation of aniline with para-substituted benzyl alcohols
as part of the Hammett study.
Recently, the alkylation of amines with alcohols was studied
by Crabtree, Eisenstein and co-workers using density functional
theory. In this investigation [CpIrCl ] with the unsubstituted
Cp ligand was employed and methanol and methylamine were
used as the models for the alcohol and the amine, respectively.
Interestingly, the carbonate counterion was found to participate
actively in the deprotonation of the alcohol and also in the proto-
nation of the formed amine. Furthermore, this computational
study predicts that dissociation of the amine is the rate-determin-
ing step, which was shown to correlate with experimental data as
14
Table 1 Overview of Hammett study with para-substituted benzyl
alcohols
2
2
2
Entry
X
σ
k
X
/k
H
log(k
X
/k
H
)
R
1
2
3
4
5
6
7
8
NMe
OMe
Me
2
−0.83
−0.27
−0.17
0.23
0.45
0.54
5.45
1.94
1.61
0.71
0.44
0.20
0.27
0.22
0.74
0.29
0.21
−0.17
−0.36
−0.70
−0.57
−0.66
0.996
0.999
0.999
0.998
0.985
0.998
0.990
0.989
Cl
COOMe
CF
the weakly basic TsNH was found to react faster than aniline.
3
2
CN
NO
0.66
0.78
Herein, we describe a combined experimental and theoretical
mechanistic investigation of the alkylation of amines with alco-
hols in the presence of the [Cp*IrCl ] catalyst. The experimen-
2
2
2
tal study is conducted under conditions relevant for application
in organic synthesis with equimolar amounts of amine and
alcohol. The electronic influence on the rate of the reaction was
investigated by performing two Hammett studies using compe-
tition experiments. Furthermore, the primary kinetic isotope
effect was determined and the nature of the iridium–hydride
complex was investigated. The computational part of the present
study includes the full Cp* ligand in combination with benzyl
alcohol and aniline substrates, which allows a direct comparison
between experiment and theory.
Results and discussion
Alcohol oxidation
Fig. 1 Hammett plot obtained from competition experiments with a
series of para-substituted benzyl alcohols.
The electronic influence on the rate of the initial alcohol oxi-
dation was studied using several para-substituted benzyl alco-
2
plots, which all resulted in straight lines having R > 0.98
(
1
5
hols, i.e. a Hammett study. These Hammett studies were
Table 1), and thus giving rise to accurate determinations of the
1
6
performed as a series of competition experiments, where each
para-substituted benzyl alcohol was allowed to compete with the
parent benzyl alcohol in the alkylation of aniline. If no signifi-
cant side reactions are occurring, disappearance of the starting
material (either alcohol or aniline) can be used to monitor reac-
tion progress. In this study we carried out GC analysis of the
reaction mixtures, which allowed the consumption of each
alcohol to be followed. If the reactions were of first order with
respect to alcohol, the concentrations of the alcohols should
follow the equation
relative reactivity (k /k ). For the ester substrate in entry 5 the
X
H
lower correlation is probably due to partial degradation of the
substrate through competing pathways such as transesterification
and amidation. However, these side reactions are much slower
than the amination and will therefore only play a role at very
high conversions.
With the relative reactivities in hand and σ values from the lit-
15
erature the Hammett plot could be constructed (Fig. 1). The
correlation is very good using the standard σ values, indicating
that neither cations nor radicals are involved in the selectivity-
determining step. Furthermore, the negative slope of the line
indicates that a small positive charge is built up in the transition
state.
ꢀ
ꢁ
ꢀ
ꢁ
X0
kX
H0
In
¼
In
ð1Þ
X
kH
H
thus resulting in a series of linear plots; one for each para sub-
stituent. The subscript “0” stands for initial concentration, X is
the para-substituted alcohol and H is benzyl alcohol itself. From
these plots the relative reactivity of each para-substituted benzyl
alcohol (k /k ) can easily be obtained as the slope of the line.
In addition to the information gathered from the Hammett
study, we were also interested in probing whether hydride
abstraction takes place in the selectivity-determining step. This
was investigated by measuring the primary kinetic isotope effect
(KIE), using benzyl alcohol α,α-d in competition with para-
2
X
H
The reaction conditions for the Hammett study employing
para-substituted benzyl alcohols are shown in Scheme 2. The
initial assumptions regarding reaction orders are justified by the
methoxybenzyl alcohol under the standard reaction conditions.
After correcting for the reactivity of the para-methoxybenzyl
alcohol the kinetic isotope effect (k /k ) was found to be 2.48.
H D
2570 | Org. Biomol. Chem., 2012, 10, 2569–2577
This journal is © The Royal Society of Chemistry 2012

View Article Online
Scheme 3 Alkylation of para-substituted anilines with benzyl alcohol
as part of the second Hammett study.
Table 2 Overview of Hammett study with para-substituted anilines
a
2
Entry
X
σ
k
X
/k
H
log(k
X
/k
H
)
R
b
1
NMe
OMe
Me
2
−0.83
−0.27
−0.17
0.23
0.45
0.54
29.12
6.59
2.64
0.44
0.22
0.07
0.05
1.46
0.82
0.42
−0.35
−0.66
−1.18
−1.33
0.995
0.998
0.999
0.993
0.984
0.960
0.979
2
3
4
5
6
Fig. 2 Hammett plot obtained from competition experiments using a
series of para-substituted anilines.
Cl
COOMe
CF
CN
3
c
7
0.66
a
reduction of the intermediate imine by a metal–hydride complex.
p-Nitroaniline was completely insoluble and did not react under the
b
standard reaction conditions. Competition experiment was performed
with p-methoxyaniline as the competing substrate. Competition
experiment was performed with p-chloroaniline as the competing
substrate.
Instead, it may suggest that formation of the imine is indeed the
c
18
selectivity-determining event. This could both be in the initial
nucleophilic addition to form a hemiaminal or in the subsequent
elimination of water since both steps should be accelerated by an
electron-donating group. From the results obtained it was not
possible to fully eliminate an alternative explanation, which
involved a rapid exchange of the amine part in the imine. This
Imine reduction
19
According to the proposed mechanism (Scheme 1), the aldehyde
reacts with the amine to form an imine, which is then reduced by
the iridium–hydride complex to form an amine. In the simplest
possible scenario this elementary step is just the reverse of the
alcohol oxidation. However, we decided to carry out a separate
Hammett study to elucidate this in further detail. Assuming that
the formation of the imine is a fast reaction, the reduction of
imines can be probed by competition experiments under similar
conditions as in the previous study. Thus, different para-substi-
tuted anilines were allowed to compete with aniline in the reac-
tion with benzyl alcohol as shown in Scheme 3. Again the
competition experiments resulted in linear plots with a good cor-
has been reported to take place for imines, but under the
current reaction conditions, the formed imine was readily
reduced to the amine, thus minimising the possibility for
exchange to occur. In a control experiment N-benzyl aniline and
p-toluidine were subjected to the standard reaction conditions
and GC did not reveal any exchange, strongly suggesting that
such an exchange is slow compared to the forward reaction (i.e.
reduction).
In both of the Hammett studies benzyl alcohol(s) were
employed as substrates. However, if the catalytic cycle described
so far is indeed operating under the actual experimental con-
ditions, we envisioned that it should be possible to study the
imine reduction step independently of the alcohol oxidation. In
this way, we intended to use a sacrificial alcohol (e.g. 2-propa-
nol) to form the iridium–hydride intermediate, which would then
deliver the hydride to an imine already present in the reaction
mixture. We were hoping to conduct competition experiments
and perhaps even determining the kinetic isotope effect (using
deuterated 2-propanol), but it was soon realised that such “trans-
fer reduction” of imines does not occur under the standard exper-
imental conditions. While this work was in progress, Yamaguchi
and co-workers reported that a preformed imine is not transfer
hydrogenated with 2-propanol, and suggested that the coupling
of the amine and the aldehyde therefore must occur in the
2
relation coefficient in most cases (R > 0.98), thus verifying that
the reaction is also first order in aniline. In two cases a lower cor-
relation coefficient was obtained (Table 2, entries 6 and 7) and
this was due to large differences in rates between the two com-
peting substrates. When the difference in rate is larger than ∼10
the inaccuracies in the determination of the remaining substrate
1
7
concentration hampers the correlation. However, since the
Hammett equation employs the logarithm of the relative reactiv-
ities the impact of these uncertainties is small.
For the anilines there was a significantly larger reactivity span,
which necessitated the use of substrates other than the parent
aniline for the competition experiments with the most and the
least reactive substrates. The data derived from the competition
experiments involving different para-substituted anilines have
been collected in Table 2.
3g
coordination sphere of iridium. Consequently, it may even be
possible to run the reaction in the presence of a preformed imine
without cross-reaction between the iridium–hydride and the pre-
formed imine. We carried out this test experiment by reacting
2 mmol of benzyl alcohol and aniline under the standard con-
ditions, and after 2 hours (conversions: benzyl alcohol: 0.24,
aniline: 0.22) N-benzylidene p-methoxyaniline was added.
Although the rate of the reaction decreased it did not stop and,
more importantly, there was no detection of the reduced N-
benzyl-p-methoxyaniline which clearly shows that the imine is
The kinetic data from Table 2 allows the construction of the
Hammett plot using the standard σ values (Fig. 2).
Again the best correlation is obtained with the standard σ
values, indicating that neither cations nor radicals are involved in
this reaction. To our surprise the Hammett plot indicates that a
positive charge is built up in the selectivity-determining tran-
sition state, which is not what would be expected from a simple
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 2569–2577 | 2571

View Article Online
formed and consumed within the coordination sphere of the
catalyst.
(PCM) with parameters suitable for benzene (dielectric constant
2.284 and probe radius 2.6 Å).
26
For the most accurate determination of Gibbs free energies in
solution we have chosen to use the single-point solvation energy
for the fully optimised structure in vacuo. To this energy is then
added the entropy contributions from analytical frequency calcu-
Metal monohydride or dihydride
Previously, Pàmies and Bäckvall have developed a method that
makes it possible to investigate whether a metal-catalysed hydro-
gen transfer goes via a monohydride or a dihydride intermedia-
27
lations in vacuo. This was originally suggested by Wertz, and
recently successfully applied by Lau and Deubel for palladium
28
complexes. In the current study we denote it Gcomb(298 K)
since it is a combination of solution phase energies and gas
phase entropy. This approach has earlier been found to give
results in accordance with experimental selectivities in cases
where neither Gibbs Free energies in vacuo nor solvation ener-
1
3g
te.
The experiment involves the racemisation of a chiral
alcohol, for which the amount of deuterium scrambling can be
measured after the reaction. Absence of deuterium scrambling is
indicative of a monohydride mechanism, in which the deuterium
is always abstracted from and delivered to the same carbon atom,
whereas complete scrambling indicates a dihydride intermediate
where exchange of H and D can take place on the metal atom.
For this purpose (R)-1-deutero-1-phenylethanol was synthesised
16b
gies were sufficiently accurate.
Transition states displayed a
single imaginary frequency, as expected, and was further charac-
terised by IRC scan calculations.
In recent years it has become clear, however, that the B3LYP
functional does not adequately describe the non-bonded inter-
actions. This is a well-known deficiency of the B3LYP functional
and the problem have been addressed by either appending a
(81% ee; 1% H by NMR) and was subjected to the [Cp*IrCl2]2
catalyst in the presence of an equimolar amount of acetophe-
none. Heating this mixture to 110 °C resulted in racemisation, as
expected, and after five hours of reaction time (8% ee) 1-phenyl-
29
classic dispersion term or by using a functional which incor-
1
ethanol was reisolated. H NMR spectroscopy revealed only
30
porates kinetic energy density terms. Among the most success-
ful of the latter approaches are the M0x family of functionals
reported by Zhao and Truhlar, and here we have chosen the M06
functional which have been optimised with particular focus on
little proton incorporation (6%), which is a strong indication that
a monohydride intermediate is involved.
To further investigate the rate of the initial β-hydride abstrac-
tion a new KIE determination was performed and followed by
GC and GC–MS. When competing p-chlorobenzyl alcohol with
2
0
31
organometallic systems. Recently the M06 functional was
necessary for the correct prediction of enantioselectivity in the
benzyl alcohol α,α-d we found that deuterium exchange does
2
32
palladium-catalysed allylic alkylation with a chiral ligand.
occur, however, the rate of this exchange is slower than the
overall consumption of the benzyl alcohol. Thus, when analysing
the initial part of the experiment we were able to obtain a KIE
value close to the one determined earlier (k /kD = 1.94). In the
H
Dimer–monomer equilibrium for [Cp*IrCl
2 2
]
ESI† we have included GC–MS data for both benzyl alcohol
substrates and the obtained amine products, which confirms that
the initially formed unsubstituted amine indeed contains a high
amount of deuterium, as expected.
Initially, the X-ray structure of [Cp*IrCl ] was acquired from
2
2
33
the CCDB database, hydrogen atoms were added and the struc-
ture optimised in the gas phase. The overall agreement between
the X-ray structure and computed structure is good (RMSD =
0
.0951 Å, excluding hydrogens), with the DFT structure having
Computational study
slightly longer Ir–Cl and Ir–Cp* bonds (ca. 0.005 Å in both
cases). The dimeric pre-catalyst [Cp*IrCl ] , which is an 18-
2
2
As an extension of the experimental mechanistic study we have
conducted a complementary computational investigation of the
title reaction using density functional theory (DFT). In the
current investigation we have used DFT in combination with the
electron complex, must dissociate to a monomeric 16-electron
complex to allow coordination of the substrate alcohol. The
simple dimer–monomer equilibrium (Scheme 4) was calculated
using DFT/B3LYP, and the dimer is favoured by 10 kJ mol− in
the gas phase (Gibbs free energy, 298 K). In solution phase the
dimer is also more stable by 30 kJ mol− , but when the resulting
1
2
1
B3LYP hybrid functional, as implemented in the Jaguar from
22
Schrodinger Inc., in line with earlier work on organometallic
1
23
systems. This computational method has been used extensively
during the last decade, and is characterised by a favourable ratio
between accuracy and computational intensity. Recently, there
have been concerns about the validity of DFT when dealing
with, for example, enthalpies of formation, electrocyclic reac-
G
(298 K) is calculated the monomer is favoured by 22 kJ
comb
−
1
mol . When applying the M06 functional that takes into
account additional stabilising dispersion energy the dimer
increases in stability by 65 kJ mol− . When the reaction is
carried out in refluxing toluene, formation of a small amount of
1
2
4
tions and C–C bond breaking reactions. However, when com-
paring isomeric structures many of the inherent short-comings of
the B3LYP method will cancel out, and good agreement with
experimental results can be obtained.
34
the catalytically active monomer is thus favoured by entropy.
In conjunction with the B3LYP method, we have used the
Hay–Wadt effective core potential (ECP) for iridium (denoted
25
LACVP* in Jaguar), which have been parameterised to also
include relativistic effects. Solvation is described using a self-
consistent reaction field (SCRF) polarised continuum model
Scheme 4 Dimer–monomer equilibrium.
2572 | Org. Biomol. Chem., 2012, 10, 2569–2577
This journal is © The Royal Society of Chemistry 2012

View Article Online
Both the alcohol and the aniline substrates are competent
ligands, which must be taken into consideration, in particular
due to their high concentration under typical experimental con-
ditions (catalyst loading 0.5–5 mol%). In contrast, the chloride
counter-ion is only present in catalytic amounts and will be dis-
placed by the better coordinating amines. Calculating the relative
energies of Cp*Ir complexes with ammonia, water and chloride
ligands showed that ammonia is a far better ligand than the other
two. Under the experimental conditions employed here a colour
change from colourless or slightly yellow to green is observed
upon addition of the iridium catalyst. After a few minutes the
solution turns brown, which may indicate a change of ligation
state of the metal, suggesting that chloride is indeed being dis-
positive charge on the benzylic carbon in going from the pre-
complex (1a: +0.10) to the TS (1b: +0.31). This build up of
positive charge is in excellent agreement with results from the
experimental Hammett study using para-substituted benzyl
alcohols.
When the β-hydride elimination has been completed, an
iridium–aldehyde complex is formed (1c, Fig. 5) which is 6 kJ
−1
mol lower in energy than the initial iridium–alkoxide complex
1a. While most earlier suggestions have indicated that dis-
sociation of the aldehyde takes place at this point, we and others
have shown experimentally that this exchange is slow compared
to the forward reaction of the aldehyde (vide supra).
The aldehyde complex 1c can now undergo internal or exter-
nal attack by an amine, which results in the formation of a hemi-
aminal after a proton shuttle from nitrogen to oxygen. If the
attack is external the formation of the hemiaminal can be
favoured by a bidentate coordination to the iridium centre and
result in simultaneous or subsequent dissociation of the coordi-
nated amine. Due to the many possibilities and the inherent low
barriers for addition to the aldehyde we have focused on locating
a reasonable structure for this hemiaminal intermediate (1d,
Fig. 6).
5
placed. We assume the anionic, η -coordinated Cp* ligand
remains coordinated to the metal during the entire catalytic
cycle.
Catalytic cycle
In the organometallic chemistry of Ir(III) the “piano-stool” com-
plexes are ubiquitous, which for the current reaction with a
Cp*Ir moiety suggests that three additional ligands can be
present. As already mentioned, one of them is probably an
amine, which leaves two available coordination sites for the
reacting alcohol. With a cationic metal centre the deprotonation
of benzyl alcohol is favourable, although even for state-of-the-art
computational methods the prediction of such an acid–base equi-
librium is a formidable challenge and the inaccuracy is likely to
be high. In any case, if the deprotonation has a lower barrier
than the following β-hydride elimination it will not affect the
overall rate of the reaction (Curtin–Hammett conditions). This
rationale provides us with complex 1a as a plausible starting
point for the catalytic reaction (Fig. 3).
With the computational methods applied here the relative
energy for this intermediate is 42 kJ mol− , and an additional
proton shift to the oxygen results in spontaneous imine for-
mation with extrusion of water (1e, 31 kJ mol− , Fig. 7). The
short H–H distance (1.89 Å) between the hydride and the hydro-
gen on the phenyl group of the imine indicates a rather strong
electrostatic interaction.
1
1
Interestingly, we found the cis-imine to be energetically
slightly more favourable than the trans-imine, which is probably
due to better coordination to the metal center. However, the inter-
conversion between the two isomers were found to have a high
barrier (see ESI† for details), but perhaps a pathway involving
dissociation of the loosely bound water ligand is more facile.
From 1a β-hydride elimination takes place through transition
−1
state 1b with a barrier of 39 kJ mol (ΔGcomb(298 K)). The dis-
tances to the migrating hydride are typical for such transform-
ations (C–H: 1.38 Å and Ir–H: 1.77 Å) as shown in Fig. 4. The
3
5
calculated ESP charges showed a significant increase of
Fig. 3 Complex 1a with deprotonated benzyl alcohol coordinated to
iridium.
Fig. 4 Transition state 1b for β-hydride elimination.
Org. Biomol. Chem., 2012, 10, 2569–2577 | 2573
This journal is © The Royal Society of Chemistry 2012

View Article Online
Fig. 5 Iridium–aldehyde complex 1c formed after β-hydride
elimination.
Fig. 7 Structure 1e with expelled water molecule coordinated to
iridium.
Fig. 8 Iridium–imine complex 1f with imine in position to receive the
hydride.
Fig. 6 Hemiaminal intermediate 1d coordinating to iridium in a biden-
tate fashion.
In order for the reduction of the imine by the Ir–H complex to
take place, the complex must rearrange to a conformation where
the hydride has the correct Bürgi–Dönitz angle for attack on the
2
sp -hybridised imine carbon. At the same time a ligand
exchange from water to aniline could take place resulting in the
structure 1f, Fig. 8).
From this complex the reduction can take place through a TS
resembling the one found for oxidation of the alcohol (1g,
Fig. 9).
The relative energy of the transition state 1g is 94 kJ mol ,
which is the highest energy throughout the entire transformation
Fig. 9 Transition state 1g for reduction of imine by iridium-hydride.
−1
formed with iridium coordinated to the deprotonated nitrogen of
the amine (1h, Fig. 11), and a simple protonation will liberate
the product and re-generate the catalyst.
(Fig. 10). After the reduction a relatively stable complex is
2574 | Org. Biomol. Chem., 2012, 10, 2569–2577
This journal is © The Royal Society of Chemistry 2012

View Article Online
Fig. 10 Overview of the entire reaction pathway from benzyl alcohol complex 1a to benzyl amine complex 1h.
Fig. 11 Iridium–amine complex 1h formed after imine reduction.
Scheme 5 Revised catalytic cycle for iridium-catalysed alkylation of
amines with alcohols.
With amines present in large excess under experimental con-
ditions we do not favour the direct coordination of carbonate to
vibrational analysis with T = 110 °C results in a calculated KIE
of 2.70 which is in good agreement with the experimental value.
14
iridium proposed by Crabtree, Eisenstein and co-workers. Fur-
thermore, the reaction can be performed with a variety of differ-
ent bases without large variations in reactivity as shown by
3
g
Yamaguchi and co-workers.
Conclusion
Both experimental and theoretical studies suggest a catalytic
cycle where the intermediate aldehyde does not leave the cata-
lyst, but instead is attacked by the amine resulting in an iridium-
hemiaminal intermediate (Scheme 5). This hemiaminal can then
dehydrate to the imine, which is subsequently reduced to the
amine, both steps taking place without breaking the coordination
to the metal. The absence of conventional solution-phase imine
formation may not have large implications for the model systems
investigated here, but can be important when extending the reac-
tion to more complicated substrates. The increased understanding
of the mechanism will be valuable in order to further optimise
and fine-tune this catalyst system.
Determination of theoretical KIE
If we assume the C–H(D) bond is fully broken in the transition
state, one should expect a KIE of 4.5 at 110 °C based solely on
−1
the differences in stretching frequencies (C–H: 2900 cm , C–D:
1
36
2
100 cm− ). The value obtained by the competition exper-
iment was significantly lower (2.48), suggesting that the C–H–D
bond indeed is broken in the selectivity-determining step, but
that there is significant stabilisation of the formed hydride. With
the calculated intermediates and transition states for the reaction
sequence it is also possible to determine the KIE using transition
state theory. For the initial β-hydride elimination a full
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 2569–2577 | 2575

View Article Online
(
c) K. Alfonsi, J. Colberg, P. J. Dunn, T. Fevig, S. Jennings, T.
Experimental
A. Johnson, H. P. Kleine, C. Knight, M. A. Nagy, D. A. Perry and
M. Stefaniak, Green Chem., 2008, 10, 31–36.
(a) A. J. A. Watson, A. C. Maxwell and J. M. J. Williams, J. Org. Chem.,
Toluene was distilled from sodium–benzophenone ketyl under a
nitrogen atmosphere. Reactions were carried out in oven-dried
Schlenk tubes equipped with a cooling finger under an argon
atmosphere. Aniline and p-(trifluoromethyl)aniline were distilled
under reduced pressure; p-anisidine, p-toluidine were recrystal-
lised from ethanol (with activated charcoal); p-(dimethylamino)
benzyl alcohol was prepared according to a literature pro-
2
2
011, 76, 2328–2331; (b) S. Bähn, S. Imm, K. Mevius, L. Neubert,
A. Tillack, J. M. J. Williams and M. Beller, Chem.–Eur. J., 2010, 16,
590–3593; (c) M. H. S. A. Hamid, C. L. Allen, G. W. Lamb, A.
3
C. Maxwell, H. C. Maytum, A. J. A. Watson and J. M. J. Williams, J.
Am. Chem. Soc., 2009, 131, 1766–1774; (d) D. Hollmann, A. Tillack,
D. Michalik, R. Jackstell and M. Beller, Chem.–Asian J., 2007, 2, 403–
410; (e) Y. Tsuji, K.-T. Huh and Y. Watanabe, J. Org. Chem., 1987, 52,
37
1
673–1680.
cedure. [Cp*IrCl ] was obtained from ABCR, Germany. N-
2
2
3
(a) I. Cumpstey, S. Agrawal, K. E. Borbas and B. Martín-Matute, Chem.
Commun., 2011, 47, 7827–7829; (b) R. Kawahara, K.-i. Fujita and
R. Yamaguchi, Adv. Synth. Catal., 2011, 353, 1161–1168; (c) O. Saidi,
A. J. Blacker, M. M. Farah, S. P. Marsden and J. M. J. Williams, Chem.
Commun., 2010, 46, 1541–1543; (d) N. Andrushko, V. Andrushko,
P. Roose, K. Moonen and A. Börner, ChemCatChem, 2010, 2, 640–643;
Benzylidene anilines and N-benzyl anilines were prepared by lit-
erature methods.
commercial suppliers and used without further purification. GC
analyses were performed on a Shimadzu GC-2010 instrument
equipped with a Supelco Equity-1 capillary column (30 m ×
3
8,39
All other chemicals were obtained from
(e) B. Blank, S. Michlik and R. Kempe, Chem.–Eur. J., 2009, 15, 3790–
0
.25 mm × 0.25 μm) and a FID detector. NMR spectra were
3799; (f) A. Prades, R. Corberán, M. Poyatos and E. Peris, Chem.–Eur.
J., 2008, 14, 11474–11479; (g) K.-i. Fujita, Y. Enoki and R. Yamaguchi,
Tetrahedron, 2008, 64, 1943–1954; (h) G. Cami-Kobeci, P. A. Slatford,
M. K. Whittlesey and J. M. J. Williams, Bioorg. Med. Chem. Lett., 2005,
recorded on a Varian Mercury 300 spectrometer in CDCl3
solution.
1
5, 535–537.
R. Yamaguchi, K.-i. Fujita and M. Zhu, Heterocycles, 2010, 81, 1093–
140.
5 K.-i. Fujita, T. Fujii, A. Komatsubara, Y. Enoki and R. Yamaguchi, Het-
erocycles, 2008, 74, 673–682.
4
General procedure for competition experiment with benzyl
alcohols
1
In an oven-dried Schlenk tube were placed K CO (14 mg,
6 (a) K.-i. Fujita, Y. Kida and R. Yamaguchi, Heterocycles, 2009, 77,
2
3
1
5
371–1377; (b) L. U. Nordstrøm and R. Madsen, Chem. Commun., 2007,
054–5056.
0
1
.1 mmol), aniline (180 μL, 2 mmol), benzyl alcohol (105 μL,
mmol), p-substituted benzyl alcohol (1 mmol), naphthalene
7
8
(a) R. N. Monrad and R. Madsen, Org. Biomol. Chem., 2011, 9, 610–
615; (b) H. Aramoto, Y. Obora and Y. Ishii, J. Org. Chem., 2009, 74,
(128 mg, 1 mmol) and dry toluene (1 mL). The mixture was
628–633.
placed in an oil bath preheated to 110 °C and a GC sample was
taken out. Then, [Cp*IrCl ] (40 mg, 0.05 mmol) was added.
(a) M. Tursky, L. L. R. Lorentz-Petersen, L. B. Olsen and R. Madsen,
Org. Biomol. Chem., 2010, 8, 5576–5582; (b) S. Whitney, R. Grigg,
A. Derrick and A. Keep, Org. Lett., 2007, 9, 3299–3302; (c) K.-i. Fujita,
K. Yamamoto and R. Yamaguchi, Org. Lett., 2002, 4, 2691–2694.
K.-i. Fujita and R. Yamaguchi, Synlett, 2005, 560–571.
2
2
This resulted in an immediate colour change from original col-
ourless/yellow to dark green, which in a few minutes changed to
brown, and then remained brown for the rest of the experiment.
The mixture was stirred at 110 °C and samples were taken out
for GC analysis as shown in the ESI† for each experiment.
9
1
0 (a) B. L. Conley, M. K. Pennington-Boggio, E. Boz and T. J. Williams,
Chem. Rev., 2010, 110, 2294–2312; (b) A. Comas-Vives, G. Ujaque and
A. Lledós, THEOCHEM, 2009, 903, 123–132; (c) J. S. M. Samec, J.-
E. Bäckvall, P. G. Andersson and P. Brandt, Chem. Soc. Rev., 2006, 35,
237–248.
General procedure for competition experiment with anilines
11 For recent examples, see: (a) M. C. Warner, O. Verho and J.-E. Bäckvall,
J. Am. Chem. Soc., 2011, 133, 2820–2823; (b) C. A. Sandoval, F. Bie,
A. Matsuoka, Y. Yamaguchi, H. Naka, Y. Li, K. Kato, N. Utsumi,
K. Tsutsumi, T. Ohkuma, K. Murata and R. Noyori, Chem.–Asian J.,
In an oven-dried Schlenk tube were placed K CO (14 mg,
2
3
0
.1 mmol), aniline (90 μL, 1 mmol), p-substituted aniline
2010, 5, 806–816; (c) J. Wettergren, E. Buitrago, P. Ryberg and
(
(
1 mmol), benzyl alcohol (210 μL, 2 mmol), naphthalene
128 mg, 1 mmol) and dry toluene (1 mL). The mixture was
H. Adolfsson, Chem.–Eur. J., 2009, 15, 5709–5718; (d) N. Pannetier, J.-
P. Sortais, P. S. Dieng, L. Barloy, C. Sirlin and M. Pfeffer, Organometal-
lics, 2008, 27, 5852–5859; (e) X. Wu, J. Liu, D. D. Tommaso, J. A. Iggo,
C. R. A. Catlow, J. Bacsa and J. Xiao, Chem.–Eur. J., 2008, 14, 7699–
placed in an oil bath preheated to 110 °C and a sample for GC
was taken out. Then, [Cp*IrCl2]2 (40 mg, 0.05 mmol) was
added. This resulted in an immediate colour change from orig-
inal colourless/yellow to dark green, which in few minutes
changed to brown and then remained brown for the rest of the
experiment. The mixture was stirred at 110 °C and samples were
taken out for GC analysis as shown in the ESI† for each
experiment.
7715; (f) W. Baratta, K. Siega and P. Rigo, Chem.–Eur. J., 2007, 13,
7479–7486; (g) B. Martín-Matute, J. B. Åberg, M. Edin and J.-
E. Bäckvall, Chem.–Eur. J., 2007, 13, 6063–6072; (h) J. S. M. Samec, A.
H. Éll, J. B. Åberg, T. Privalov, L. Eriksson and J.-E. Bäckvall, J. Am.
Chem. Soc., 2006, 128, 14293–14305; (i) C. P. Casey, G. A. Bikzhanova
and I. A. Guzei, J. Am. Chem. Soc., 2006, 128, 2286–2293.
12 For recent examples, see: (a) J. Bosson, A. Poater, L. Cavallo and S.
P. Nolan, J. Am. Chem. Soc., 2010, 132, 13146–13149; (b) A. Nova,
D. Balcells, N. D. Schley, G. E. Dobereiner, R. H. Crabtree and
O. Eisenstein, Organometallics, 2010, 29, 6548–6558; (c) N. Sieffert and
M. Bühl, J. Am. Chem. Soc., 2010, 132, 8056–8070; (d) A. Comas-
Vives, G. Ujaque and A. Lledós, Organometallics, 2007, 26, 4135–4144;
Acknowledgements
(
3
e) J.-W. Handgraaf and E. J. Meijer, J. Am. Chem. Soc., 2007, 129,
099–3103; (f) T. Privalov, J. S. M. Samec and J.-E. Bäckvall, Organo-
We thank the Danish National Research Foundation and the
Danish Council for Independent Research, Technology and Pro-
duction Sciences for financial support.
metallics, 2007, 26, 2840–2848.
3 (a) K.-i. Fujita, T. Yoshida, Y. Imori and R. Yamaguchi, Org. Lett., 2011,
1
1
3, 2278–2281; (b) H. Li, G. Lu, J. Jiang, F. Huang and Z.-X. Wang,
Organometallics, 2011, 30, 2349–2363; (c) C. Azerraf and D. Gelman,
Chem.–Eur. J., 2008, 14, 10364–10368; (d) F. Hanasaka, K.-i. Fujita and
R. Yamaguchi, Organometallics, 2005, 24, 3422–3433; (e) J.-
W. Handgraaf, J. N. H. Reek and E. J. Meijer, Organometallics, 2003,
References
1
(a) C.-J. Li and B. M. Trost, Proc. Natl. Acad. Sci. U. S. A., 2008, 105,
3197–13202; (b) R. A. Sheldon, Chem. Commun., 2008, 3352–3365;
22, 3150–3157; (f) M. Albrecht, J. R. Miecznikowski, A. Samuel, J.
1
2576 | Org. Biomol. Chem., 2012, 10, 2569–2577
This journal is © The Royal Society of Chemistry 2012

View Article Online
W. Faller and R. H. Crabtree, Organometallics, 2002, 21, 3596–3604;
25 LACVP* uses the 6-31G* basis set for all light elements and the Hay–
Wadt ECP and basis set for Ir, see: P. J. Hay and W. R. J. Wadt, J. Chem.
Phys., 1985, 82, 299–310.
(g) O. Pàmies and J.-E. Bäckvall, Chem.–Eur. J., 2001, 7, 5052–5058;
h) K. Mashima, T. Abe and K. Tani, Chem. Lett., 1998, 1201–1202. See
(
also ref. 3b.
26 (a) B. Marten, K. Kim, C. Cortis, R. A. Friesner, R. B. Murphy, M.
N. Ringnalda, D. Sitkoff and B. Honig, J. Phys. Chem., 1996, 100,
11775–11788; (b) D. J. Tannor, B. Marten, R. Murphy, R. A. Friesner,
D. Sitkoff, A. Nicholls, M. Ringnalda, W. A. Goddard, III and B. Honig,
J. Am. Chem. Soc., 1994, 116, 11875–11882.
1
4 D. Balcells, A. Nova, E. Clot, D. Gnanamgari, R. H. Crabtree and
O. Eisenstein, Organometallics, 2008, 27, 2529–2535.
5 C. Hansch, A. Leo and R. W. Taft, Chem. Rev., 1991, 91, 165–195.
6 (a) J. H. Dam, P. Fristrup and R. Madsen, J. Org. Chem., 2008, 73,
1
1
3
228–3235; (b) P. Fristrup, M. Kreis, A. Palmelund, P.-O. Norrby and
27 D. H. Wertz, J. Am. Chem. Soc., 1980, 102, 5316–5322.
28 J. K.-C. Lau and D. V. Deubel, J. Chem. Theory Comput., 2006, 2, 103–
106.
R. Madsen, J. Am. Chem. Soc., 2008, 130, 5206–5215; (c) L. Keinicke,
P. Fristrup, P.-O. Norrby and R. Madsen, J. Am. Chem. Soc., 2005, 127,
1
5756–15761.
29 (a) M. Elstner, P. Hobza, T. Frauenheim, S. Suhai and E. Kaxiras, J.
Chem. Phys., 2001, 114, 5149–5155; (b) S. Grimme, J. Comput. Chem.,
2004, 25, 1463–1473.
1
7 P. Fristrup, G. H. Jensen, M. L. N. Andersen, D. Tanner and P.-
O. Norrby, J. Organomet. Chem., 2006, 691, 2182–2198.
1
1
8 E. F. Pratt and M. J. Kamlet, J. Org. Chem., 1961, 10, 4029–4031.
9 (a) G. Reddelien, Ber., 1920, 53B, 355–358; (b) N. Giuseppone, J.-
L. Schmitt, E. Schwartz and J.-M. Lehn, J. Am. Chem. Soc., 2005, 127,
30 Y. Zhao, N. E. Schultz and D. G. Truhlar, J. Chem. Theory Comput.,
2006, 2, 364–382.
31 (a) Y. Zhao and D. G. Truhlar, Theor. Chem. Acc., 2008, 120, 215–241;
(b) Y. Zhao and D. G. Truhlar, Acc. Chem. Res., 2008, 41, 157–167.
32 X. Bantreil, G. Prestat, A. Moreno, D. Madec, P. Fristrup, P.-O. Norrby, P.
S. Pregosin and G. Poli, Chem. Eur. J., 2011, 17, 2885–2896.
33 M. R. Churchill and S. A. Julis, Inorg. Chem., 1977, 16, 1488–
1494Refcode: DCPMIR; See ESI† for further details.
34 D. S. Gill, C. White and P. M. Maitlis, J. Chem. Soc., Dalton Trans.,
1978, 617–626.
35 L. E. Chirlian and M. M. Francl, J. Comput. Chem., 1987, 8, 894–905.
36 F. H. Westheimer, Chem. Rev., 1961, 61, 265–273.
37 M. R. Pitts, J. R. Harrison and C. J. Moody, J. Chem. Soc., Perkin Trans.
1, 2001, 955–977.
38 (a) H. Neuvonen, K. Neuvonen and F. Fülöp, J. Org. Chem., 2006, 71,
3141–3148; (b) J. H. Billman and A. C. Diesing, J. Org. Chem., 1957,
22, 1068–1070.
39 For characterisation data of N-benzyl anilines, see: C.-Y. Tsai, R. Sung,
B.-R. Zhuang and K. Sung, Tetrahedron, 2010, 66, 6869–6872;
S. C. A. Sousa and A. C. Fernandes, Adv. Synth. Catal., 2010, 352,
2218–2226; C.-T. Yang, Y. Fu, Y.-B. Huang, J. Yi, Q.-X. Guo and L. Liu,
Angew. Chem., Int. Ed., 2009, 48, 7398–7401; ref. 3b and 3g.
5
528–5539.
2
0 Very recently, Yamaguchi and co-workers reported the alkylation of
amines with alcohols in water in the presence of the related complex
3 3 2
[Cp*Ir(NH ) ]I (ref. 3b). Deuterium scrambling experiments did also in
this case indicate a reaction pathway with an iridium monohydride
intermediate.
2
2
2
1 (a) A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652; (b) A. D. Becke,
J. Chem. Phys., 1993, 98, 1372–1377; (c) C. Lee, W. Yang and R.
G. Parr, Phys. Rev. B, 1988, 37, 785–789.
2 Jaguar, version 7.0, Schrödinger, LLC, New York, NY, 2007,
release 107. For the most recent version, see: http://www.schrodinger.
com..
3 (a) P. Fristrup, M. Ahlquist, D. Tanner and P.-O. Norrby, J. Phys. Chem.
A, 2008, 112, 12862–12867; (b) N. Svensen, P. Fristrup, D. Tanner and
P.-O. Norrby, Adv. Synth. Catal., 2007, 349, 2631–2640; (c) M. Ahlquist,
P. Fristrup, D. Tanner and P.-O. Norrby, Organometallics, 2006, 25,
2
066–2073.
4 For an overview, see: P. R. Schreiner, Angew. Chem., Int. Ed., 2007, 46,
217–4219.
2
4
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 2569–2577 | 2577