Experimental and theoretical mechanistic investigation of the iridium-catalyzed dehydrogenative decarbonylation of primary alcohols

Article abstract of DOI:10.1021/ja5106943

The mechanism for the iridium-BINAP catalyzed dehydrogenative decarbonylation of primary alcohols with the liberation of molecular hydrogen and carbon monoxide was studied experimentally and computationally. The reaction takes place by tandem catalysis through two catalytic cycles involving dehydrogenation of the alcohol and decarbonylation of the resulting aldehyde. The square planar complex IrCl(CO)(rac-BINAP) was isolated from the reaction between [Ir(cod)Cl]₂, rac-BINAP, and benzyl alcohol. The complex was catalytically active and applied in the study of the individual steps in the catalytic cycles. One carbon monoxide ligand was shown to remain coordinated to iridium throughout the reaction, and release of carbon monoxide was suggested to occur from a dicarbonyl complex. IrH₂Cl(CO)(rac-BINAP) was also synthesized and detected in the dehydrogenation of benzyl alcohol. In the same experiment, IrHCl₂(CO)(rac-BINAP) was detected from the release of HCl in the dehydrogenation and subsequent reaction with IrCl(CO)(rac-BINAP). This indicated a substitution of chloride with the alcohol to form a square planar iridium alkoxo complex that could undergo a β-hydride elimination. A KIE of 1.0 was determined for the decarbonylation and 1.42 for the overall reaction. Electron rich benzyl alcohols were converted faster than electron poor alcohols, but no electronic effect was found when comparing aldehydes of different electronic character. The lack of electronic and kinetic isotope effects implies a rate-determining phosphine dissociation for the decarbonylation of aldehydes.

Full text of DOI:10.1021/ja5106943

Article
pubs.acs.org/JACS
Experimental and Theoretical Mechanistic Investigation of the
Iridium-Catalyzed Dehydrogenative Decarbonylation of Primary
Alcohols
Esben P. K. Olsen,† Thishana Singh,‡ Pernille Harris,† Pher G. Andersson,*,‡ and Robert Madsen*,†
†Department of Chemistry, Technical University of Denmark, 2800 Kgs. Lyngby, Denmark
‡Department of Organic Chemistry, Stockholm University, 106 91 Stockholm, Sweden
S
* Supporting Information
ABSTRACT: The mechanism for the iridium−BINAP catalyzed
dehydrogenative decarbonylation of primary alcohols with the
liberation of molecular hydrogen and carbon monoxide was studied
experimentally and computationally. The reaction takes place by
tandem catalysis through two catalytic cycles involving dehydrogen-
ation of the alcohol and decarbonylation of the resulting aldehyde. The
square planar complex IrCl(CO)(rac-BINAP) was isolated from the
reaction between [Ir(cod)Cl]2, rac-BINAP, and benzyl alcohol. The
complex was catalytically active and applied in the study of the
individual steps in the catalytic cycles. One carbon monoxide ligand
was shown to remain coordinated to iridium throughout the reaction, and release of carbon monoxide was suggested to occur
from a dicarbonyl complex. IrH2Cl(CO)(rac-BINAP) was also synthesized and detected in the dehydrogenation of benzyl
alcohol. In the same experiment, IrHCl2(CO)(rac-BINAP) was detected from the release of HCl in the dehydrogenation and
subsequent reaction with IrCl(CO)(rac-BINAP). This indicated a substitution of chloride with the alcohol to form a square
planar iridium alkoxo complex that could undergo a β-hydride elimination. A KIE of 1.0 was determined for the decarbonylation
and 1.42 for the overall reaction. Electron rich benzyl alcohols were converted faster than electron poor alcohols, but no
electronic effect was found when comparing aldehydes of different electronic character. The lack of electronic and kinetic isotope
effects implies a rate-determining phosphine dissociation for the decarbonylation of aldehydes.
Recently, two of us presented an iridium catalyst that
combines the dehydrogenation and the decarbonylation in one
transformation.9 In this way, a primary alcohol is converted into
the corresponding one-carbon shorter adduct with the release
of both molecular hydrogen and carbon monoxide. The
optimized conditions employed [Ir(coe)2Cl]2, rac-BINAP,
and LiCl in refluxing mesitylene saturated with water (Scheme
1).9 The same conditions also facilitated the decarbonylation of
aldehydes, which suggests that independent dehydrogenation
and decarbonylation cycles are operating, and the reaction
therefore is an example of a tandem catalytic transformation.10
Considerable attention has been devoted in recent years to the
development of tandem catalytic processes where several
INTRODUCTION
■
The acceptorless dehydrogenation of an alcohol into the
corresponding carbonyl compound and molecular hydrogen is
an important transformation that can be used both for
hydrogen production1 and for in situ generation of carbonyl
compounds in the presence of nucleophiles.2 The dehydrogen-
ation is usually catalyzed by various ruthenium and iridium
complexes and can be achieved in the absence of a
stoichiometric additive such as a base.1,2 The mechanism has
gained significant attention and has been the subject of both
experimental3 and theoretical4 investigations. Depending on the
nature of the ligands, two pathways for hydrogen abstraction
have been identified where the alcohol either coordinates to the
metal (inner-sphere) or to a basic ligand (outer-sphere).3,4
On the contrary, the decarbonylation of an aldehyde into the
one-carbon shorter adduct and carbon monoxide has been less
thoroughly studied. This transformation can be performed with
rhodium,5 iridium6 and palladium7 catalysts in the absence of a
stoichiometric additive, but the mechanism has only been
studied in detail with a rhodium catalyst.8 In this case, the
transformation was shown to proceed by oxidative addition into
the aldehyde C−H bond followed by extrusion of carbon
monoxide and reductive elimination.8
Scheme 1. Acceptorless Dehydrogenative Decarbonylation
Catalyzed by Iridium and BINAP
Received: October 17, 2014
© XXXX American Chemical Society
A
DOI: 10.1021/ja5106943
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
reactions are performed consecutively in the same operation
with one or more metal catalysts.10 However, thorough
mechanistic studies of these transformations are still relatively
rare in order to identify the catalytically active species and the
interplay between the catalytic cycles.11
Recently, we have applied the dehydrogenative decarbon-
ylation as an ex situ generator of syngas in the hydroformylation
of styrenes.12 Furthermore, the transformation of primary
alcohols into their dehydroxymethylated counterparts has also
been accomplished by a rhodium complex in a photocatalytic
pathway13 and by a ligand free palladium catalyst.14 It is a
transformation that has been suggested to play a future role in
the defunctionalization of renewables in the replacement of
fossil resources.15
Herein, we present a combined experimental and computa-
tional mechanistic investigation of the iridium-catalyzed tandem
reaction. The studies employ synthesis and detection of the
major catalytic species, isotope-labeling experiments, kinetic
studies and DFT calculations. The complex IrCl(CO)(BINAP)
(1) has been identified as the key catalytically active
intermediate and has been found to serve an interesting triple
role in the transformation. It reacts with the alcohol in the
dehydrogenation, with the aldehyde in the decarbonylation and
with HCl as a mild and reversible acid scavenger.
Figure 1. Crystal structure of IrCl(CO)(S)-BINAP.
scrambling of deuterium-labeled substrates,9 could now be
excluded from the reaction mixture.
The KIE of the overall reaction was measured by comparing
2-hydroxymethylnaphthalene and d3-2-hydroxymethylnaphtha-
lene in a noncompetitive experiment. The formation of
naphthalene was followed by GC-MS and an average of two
RESULTS AND DISCUSSION
■
experiments resulted in a KIE of 1.42
0.07. Similarly, the
To study the catalytic intermediates formed during the
dehydrogenative decarbonylation, benzyl alcohol was chosen
as the substrate since the product benzene could be easily
removed by evaporation. Thus, a 1:2:3 ratio of [Ir(cod)Cl]2,
rac-BINAP, and benzyl alcohol was stirred in mesitylene at 110
°C for 60 min followed by removal of the solvent by distillation
under reduced pressure along with dissociated cyclooctadiene
and benzene. The remaining orange residue was analyzed by IR
and 31P NMR. IR spectroscopy revealed a strong peak at 1986
decarbonylation of 2-naphthaldehyde and 2-α-d-naphthalde-
hyde were compared in a noncompetitive experiment and
resulted in a KIE of 1.0 0.05. The KIE of 1.42 was therefore
assigned to the dehydrogenation cycle, and the value is
considered too high for a normal secondary KIE.18
A Hammett study provides precious knowledge about the
mechanism, but unfortunately due to the high reaction
temperature and the low boiling point of the generated aryl
compounds, a regular Hammett plot was not obtainable.
Instead, three aryl alcohols affording high boiling products and
with very different electronic properties were compared in
competitive rate experiments. By measuring initial rates, it was
found that 3,4,5-trimethoxybenzene was formed 1.5 times faster
than naphthalene, which was formed 2.5-fold faster than methyl
benzoate as outlined in Figure 2.
cm−1 corresponding to a carbonyl ligand, and the very clean 31
P
NMR spectrum showed only two doublets with a coupling
constant of 29 Hz. Applying α-[13C]-benzyl alcohol with
[Ir(cod)Cl]2 and rac-BINAP resulted in a splitting of the two
different doublets into two double doublets. The two new
couplings constants were 123 Hz corresponding to a trans-
coupling and 11.2 Hz corresponding to a cis-coupling. The
same coupling constants were found in a double duplet at 180.8
ppm, when recording a 13C NMR spectrum of the [13C]-
enriched residue. These data comply with what is reported for
complexes of the type IrXCO(P−P), where X is a halide and
(P−P) is a cis chelating phosphine.16
Figure 2. Rate comparison of electronically different benzyl alcohols.
The structure was also confirmed by X-ray of a crystal
obtained by slow addition of Et2O into a solution of
IrCl(CO)(S)-BINAP in CH2Cl2 (Figure 1). The strong
coordination between carbon monoxide and iridium could
not be disrupted by long-term refluxing in mesitylene or long-
term storage on a vacuum line. This observation is consistent
with the high bond dissociation energies reported for square
planar iridium carbonyl complexes.17 The complex was
catalytically active and kinetically very similar to the in situ
formed catalyst. It could be stored under argon, but lost
catalytic activity when exposed to air over a couple of days. The
similarities in rate between the in situ formed and the
preformed catalyst excluded any participating role of cyclo-
octadiene. This also made it possible to perform KIE
measurements since cyclooctadiene, that had earlier caused
Interestingly, the decarbonylation of 3,4,5-trimethoxybenzal-
dehyde and methyl 4-formylbenzoate was equally fast
(Supporting Information), which also suggests that the
electronic effect stems from the dehydrogenation cycle.
The reaction order in catalyst was determined to be 0.75
0.15 for the dehydrogenative decarbonylation of 2-(2-
naphthyl)ethanol. The aliphatic substrate was applied because
aliphatic aldehydes do not accumulate in the tandem reaction
and the reaction is thereby zeroth order in the alcohol. The
reaction order in catalyst for the decarbonylation of 2-
naphthaldehyde was determined to be 0.85
0.08. The
noninteger reaction orders lower than 1.0 in both cases may
indicate a higher degree of inactivation with increased catalyst
B
DOI: 10.1021/ja5106943
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
concentration. However, the result can be used to preclude any
theory of second order kinetics.
transition state in the H2 oxidative addition.16 Interestingly, the
ratio between the two diastereomers remained 3:1 over time,
whereas Eisenberg who investigated the same reaction but with
iodide as the counterion, observed a rapid scrambling of the
isomers in deuterated benzene.16
Experimental Study of Dehydrogenation. The liberated
molecular hydrogen is expected to be reductively eliminated
from an iridium dihydride species. The group of Eisenberg has
investigated the oxidative addition of hydrogen to IrX(CO)(P−
P) complexes in order to form IrH2X(CO)(P−P).16,19 When
IrCl(CO)(rac-BINAP) (1) was stirred in CDCl3 under a
hydrogen atmosphere, IrH2Cl(CO)(rac-BINAP) isomers were
rapidly formed. As discovered by Eisenberg, and illustrated in
Figure 3, molecular hydrogen can add from above or below, or
Reductive elimination of hydrogen from IrH2Cl(CO)(rac-
BINAP) is essential to regenerate the catalytic starting material.
To investigate the conditions necessary to accomplish this, the
following procedure was employed: the mixture of 2A and 2B,
formed from oxidative addition of hydrogen in CDCl3, was
divided into three portions followed by removal of CDCl3 by
evaporation. The residues were taken up in benzene, toluene,
and mesitylene, respectively. The aromatic solvents were
removed by distillation at atmospheric pressure, and the
residues were dissolved in CDCl3 and analyzed by 1H NMR. As
seen in the spectrum (Figure 4B), recorded after the removal of
benzene, new peaks assigned to isomers 3A and 3B appeared.
Increasing the temperature, as was done when toluene was
removed by distillation, resulted in complete isomerization
(Figure 4C). Interestingly, the conditions were not harsh
enough for molecular hydrogen to be released, but when
mesitylene was removed by distillation, the hydride signals
vanished completely and a 31P NMR spectrum confirmed the
regeneration of complex 1.
Since refluxing in toluene is not sufficient for the reductive
elimination of hydrogen, we envisioned that isomers of the
dihydride complexes could be detected along the catalytic
pathway by reacting 1 with a primary alcohol. Thus, after
refluxing 1 with benzyl alcohol for 4 h, toluene was removed by
distillation and the residue taken up in CDCl3 and analyzed by
1H NMR. Only the hydride region is shown in Figure 5 (full
spectrum included in the Supporting Information).
Figure 3. Oxidative addition of dihydrogen leading to four different
isomers.
along the P−Ir−CO axis or the P−Ir−Cl axis, giving rise to
four different isomers when a chiral ligand like BINAP is used.
Labeling the starting complex with [13C]CO makes it possible
to distinguish between the diastereomers since complexes 2A
and 2B have a trans coupling between H and [13C]CO, while
1
complexes 3A and 3B only have cis couplings. A H NMR
spectrum recorded after 30 min of reaction time with hydrogen
revealed that only isomers 2A and 2B were formed (Figure
4A). The effect is electronic and is caused by the ability of CO
to stabilize the Ir(H2)Cl(CO)(rac-BINAP) trigonal bipyramidal
Figure 5. Hydride region recorded after removal of toluene from a
reaction of 1 with benzyl alcohol.
The spectrum clearly shows formation of the dihydride
complexes 3A and 3B. Only a small amount of benzyl alcohol
was converted at this temperature, which may indicate the
formation of dihydride to be of high activation energy or the
alcohol activation to be reversible. To determine a potential
equilibrium, the dihydride complexes 2A and 2B were stirred
with a stoichiometric amount of 2-naphthaldehyde. An aliquot
was taken after 3 h and analyzed by GC−MS, which revealed
that 32% of the aldehyde had been converted to 2-
hydroxymethylnaphthalene. The detection of 3A and 3B
isomers in the reaction of 1 with benzyl alcohol, and the
evidence for reversibility, strongly suggests the CO ligand to
remain coordinated to iridium throughout the dehydrogenation
cycle. With this in mind, several possible routes were
envisioned as illustrated in Scheme 2.
Figure 4. Hydride region in 1H NMR spectra. (A) Spectrum recorded
after 30 min where only hydrides from the kinetic isomers 2A and 2B
are observed. (B) Spectrum recorded after removal of benzene, where
an isomerization had started to occur. (C) Spectrum recorded after
removal of toluene, where complete isomerization into the
thermodynamic isomers 3A and 3B had occurred.
C
DOI: 10.1021/ja5106943
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
1
analyzed by H, 13C, and 31P NMR. Complete conversion was
readily obtained and the coupling pattern in the spectra
indicated formation of the oxidative addition product
IrHCl2(CO)(rac-BINAP) (4). The structure was also con-
firmed by X-ray crystallography of a crystal obtained by slow
addition of Et2O into a solution of 4 in CH2Cl2 (Figure 6).
Scheme 2. Possible Routes to Formation of 3A and 3B
a
(A) Pathway without a vacant site. (B) Chloride dissociation followed
by β-hydride elimination and ligand substitution. (C) Hydride extracts
the α-proton forming a dihydrogen complex. (D) Direct β-hydride
elimination from the square planar iridium alkoxo complex. (E)
Phosphine dissociation prior to β-hydride elimination. (F) Ionic
pathway.
Figure 6. Crystal structure of IrHCl2(CO)(R)-BINAP.
A re-examination of the spectrum in Figure 5, which was
recorded from the reaction between 1 and benzyl alcohol at
110 °C, shows the peaks from the two isomers of 4 at −16.8
and −16.0 ppm, indicating HCl formation under these reaction
conditions.
The pathways are separated into two major subgroups:
oxidative addition of alcohol to iridium20 affording fully
saturated complexes and substitution of the chloride ligand
with the alcohol yielding square planar iridium alkoxo
complexes by the release of HCl. The oxidative addition
products lack a vacant site for a β-hydride elimination. In
pathway A, a vacant site is not necessary as a similar case was
reported by Milstein.21 In pathway B, chloride dissociates to
create the vacant site in an unsaturated cationic complex. In
path C, molecular hydrogen is formed directly. A common
feature in pathways D, E, and F is the formation of the square
planar iridium alkoxo complex which is closely related to the
Vaska analog Ir(CO)(PPh3)2(OR) investigated by Hartwig.22
In pathway D, the β-hydride elimination takes place directly
from the tetra-substituted square planar complex. In pathway E,
the phosphine dissociates prior to β-hydride elimination, and
pathway F represents an ionic route where the alkoxide could
be stabilized by hydrogen bonding or a Lewis acid. Hartwig
found that electron donating groups enhance the rate of β-
hydride elimination from iridium alkoxo complexes and
rationalized the observation with either greater stability of the
resulting more electron rich aldehyde complex or lowering the
barrier for migration of the hydrogen with partial hydridic
character.22 Hence, the electronic effect, illustrated in Figure 2,
does not only dictate positive charge buildup as a result of
chloride dissociation, but can also be used to exclude pathway F
from the speculations. In the substitution pathways, HCl is
formed and liberated, but has to remain in solution to reform
the dihydride complex after β-hydride elimination. In 1968
The iridium monohydride complexes depicted in pathway D
and E were investigated by reacting 1 with sodium benzyl
alkoxide at 110 °C for 3 h. Benzaldehyde formation was
observed without the concomitant formation of 3A and 3B.
Instead an unknown iridium hydride complex was observed at
1
−11.2 ppm in the H NMR spectrum. The pattern showed a
triplet with coupling constants of 39.2 Hz, which lies in-
between the coupling constants that had been observed for P−
H cis- and trans-couplings in this study. Applying [13C]-labeled
1 splits the triplet into a triple quintet indicating a complicated
cluster formation (Scheme 3). The pattern from this cluster
Scheme 3. Unstable Monohydride Species Leading to
Cluster Formation
1
and a similar one at −11.6 ppm could also be found in the H
NMR spectrum in Figure 5. This pattern is also observed when
forming 2A and 2B from 1 under a hydrogen atmosphere
(Figure 4A). The clusters are not expected to be on the
catalytic pathway, since their patterns do not disappear upon
heating. Catalytically inactive iridium clusters have been
reported in many cases.24 Tetra-coordinated iridium mono-
hydride carbonyl complexes appear to be rather unstable as the
Vaska noticed that the oxidative addition of HCl to Vaskas
́
complex IrCl(CO)(PPh3)2 is reversible,23 which persuaded us
to investigate whether complex 1 would have similar chemical
properties. In a CDCl3 solution of 1, HCl was formed in situ
from ethanol and acetyl chloride. The crude mixture was
D
DOI: 10.1021/ja5106943
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
related complex IrH(CO)(PPh3)2 also seemed to aggregate in a
study of β-hydride eliminations from Ir(n-octyl)(CO)-
(PPh3)2.25
These observations favor route D since both 4 and the
iridium cluster byproduct from the decomposition of the square
planar iridium alkoxo complex are detected in the reaction of 1
with benzyl alcohol (Figure 5). Complex 4 does catalyze the
dehydrogenative decarbonylation, but it is not faster than 1,
although it is easier to handle since it can be stored on the shelf,
outside a glovebox, and without any sign of decomposition after
several months.
In addition, a 3:1 mixture of 1 and 4 was stirred with benzyl
alcohol at 110 °C to investigate changes in the hydride region.
The hydride region resulting from this reaction is depicted in
Figure 7 and by comparison with Figure 5, it is clear that
formation of the cluster at −11.2 ppm has been inhibited.
Figure 8. Energy profile for the dehydrogenation pathway C.
Pathway D is initiated by the substitution of chloride with
benzyl alcohol (Scheme 4). The transition state that results in
Scheme 4. Substitution of Chloride with Benzyl Alcohol
the formation of HCl has a free energy of 21.1 kcal/mol, and
upon dissociation of HCl into solution, the energy of the alkoxo
complex d2 is 28.9 kcal/mol. With a base present, the energy
should be substantially lower.
Figure 7. Hydride region recorded after removal of toluene from a
reaction of 1 and 4 with benzyl alcohol.
If the parent complex 1 acts as the base and absorbs HCl, the
energy is lowered by 10.7 kcal/mol (Figure 9). The β-hydride
Computational Study of Dehydrogenation. The
dehydrogenation pathways in Scheme 2 were also evaluated
by DFT calculations using the crystal structure of 1 as the
starting point and benzyl alcohol as the substrate. From the
calculations, a total energy ΔGtot was calculated using the
formula ΔGtot = ΔG − Escf + Esolv, where ΔG represents the
Gibbs free energy, Escf is the gas phase energy, and Esolv is the
solution phase energy in mesitylene. This method, originally
proposed by Wertz,26 has been applied in many theoretical
investigations of transition metal catalytic cycles.3d,8
The two routes with the lowest activation energies were
pathways C and D. We did not succeed in calculating the
energy profile for path E, because as soon as we forced one of
the phosphines in BINAP to dissociate, it reassociated in the
following optimization. Pathway C is initiated by an oxidative
addition, which has a very high activation barrier of 50.3 kcal/
mol (Figure 8, cTS1). The depicted isomer of the oxidative
addition product (c1) is not the most stable isomer, but it has
the lowest barrier of formation among the different isomers.
The following hydride elimination forming molecular hydrogen
directly has a barrier of 32.3 kcal/mol (cTS2). Notably, the
formation of molecular hydrogen occurs simultaneously with an
increase in the distance between the iridium center and oxygen
giving rise to complex c2 where the aldehyde is no longer
coordinated to iridium. The experimental observation of
dihydride complexes 3A and 3B can be explained by oxidative
addition of molecular hydrogen to form 2A and 2B followed by
an isomerization to 3A and 3B which was proven possible
experimentally. The elimination of molecular hydrogen has a
low activation energy and since the overall dehydrogenation is
endothermic it must be driven by the removal of hydrogen.
Figure 9. Energy profile for the dehydrogenation pathway D.
elimination is 29.8 kcal/mol higher in energy than d2 and the
overall energetic span is slightly lower than in pathway C. In
addition, the substitution with the alcohol is considerably more
favorable than the oxidative addition in pathway C. As a result,
the calculations also suggest pathway D to be the most likely
route with complex 1 serving the additional role as acid
scavenger (Figure 9).27
Experimental Study of Decarbonylation. Convinced
that carbon monoxide is too tightly bound to 1 under the
reaction conditions, it was thus expected that CO will be
liberated from an iridium dicarbonyl species. Isolation of the
dicarbonyl complex IrCl(CO)2(rac-BINAP) (5) under the
reaction conditions has not been accomplished, probably
E
DOI: 10.1021/ja5106943
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
because CO addition is reversible. Instead, the Skrydstrup CO
generating system28 was used to add 1 equiv of CO to a CDCl3
solution of [13C]-1 in a two-chamber setup. Mesitylene was
added and then removed by distillation to completely release
one of the carbon monoxide ligands. A 31P NMR spectrum of
the resulting residue was recorded and revealed a 1:1.3 ratio
between 1 and [13C]-1 (Scheme 5). This observation is in
Scheme 6. Possible Pathways for the Decarbonylation of
Aldehydes
a
compliance with the ability of Vaskas
́
complex to coordinate an
additional carbonyl ligand.29
Scheme 5. CO Scrambling with the Skrydstrup CO
Generator in a Two-Chamber Setup
a
(G) Substitution pathway. (H) Oxidative addition followed by
dissociation of chloride. (I) A concerted migration pathway. (J) R
dissociates from the acyl complex. (K) Phosphine dissociation
pathway.
Even though IrCl(CO)2(rac-BINAP) is not sufficiently stable
to be isolated, Eisenberg had synthesized [Ir(CO)2(R)-
BINAP][SbF6] and found that this complex was stable enough
to be isolated by crystallization.16 With this is mind, complex 1
was mixed with benzaldehyde and stirred in benzene at 80 °C
for 3 h. Benzene was removed by evaporation, and a solution of
AgSbF6 in CH2Cl2 was added to the remaining residue. After it
stirred for 1 h at room temperature, the reaction mixture was
filtered, and dry Et2O was added to the filtrate. The resulting
precipitate matched the spectroscopic data for [Ir(CO)2(R)-
BINAP][SbF6].16
trimethoxybenzaldehyde was decarbonylated in the presence of
two equivalent α,α-dimethylbenzyl alcohol, a tertiary alcohol
not able to undergo dehydrogenation. Furthermore, the same
deuterated aldehyde was decarbonylated by 33 mol % of 4,
which would dissociate HCl prior to the decarbonylation. The
isolated products observed in both experiments had no proton
incorporation which eliminates pathway J. Pathway I was
investigated by DFT calculations since it was the only neutral
pathway, besides the phosphine dissociation pathway, that
could be directly compared to the energies which were
obtained in the dehydrogenation cycle. It is difficult to compare
ionic pathways to neutral pathways using computational
methods, but the apolar reaction environment created by
mesitylene should disfavor ionic pathways.
Computational Study of Decarbonylation. A similar
methodology that was used for the dehydrogenation, was also
applied for the decarbonylation: the crystal structure of 1 was
used as the starting point and benzaldehyde was used as the
substrate.
The oxidative addition of benzaldehyde to 1 leads to
saturated iridium(III) species. All the possible isomers were
optimized, and the relative Gibbs free energies were compared
(Figure 10). Even though the isomers with the acyl group in
the apical position with respect to BINAP have the highest
energies, the barriers for their formation were significantly
lower than for the two other isomers. The lowest energy of 34.5
It is generally accepted that in transition metal mediated
decarbonylations of aldehydes, the aldehyde oxidatively adds to
the transition metal by activation of the C−H bond.8,30
However, oxidative addition of an aldehyde to 1 will form a
saturated iridium(III) complex with no vacant sites for
migration. Since no kinetic isotope effect and no charge
buildup were found, the steps in-between oxidative addition
and CO dissociation are not immediately apparent. Five
different pathways, illustrated in Scheme 6, have been
evaluated. Radical pathways have been omitted, since a broad
range of substrates including aryl halides are fully converted
under the catalytic conditions.9 Pathway G is inspired by the
mechanism suggested for the rhodium-catalyzed decarbon-
ylation8 where a substitution initiates a cationic pathway. In
pathway H, chloride dissociates to create a vacant site for the
migratory extrusion. Pathway I and J are similar since the R
group does not migrate to iridium, and both cases are triggered
by an analogous example reported by Bergman.31 In pathway K,
one of the phosphines in BINAP dissociates to create a vacant
site for migration. In principle, a similar pathway with CO
dissociation could also be envisioned, but since the
experimental studies show tight bonding of one CO ligand to
iridium, it was decided not to include this possibility in the
evaluation.
In the decarbonylation reported by Bergman,31 deuterium
incorporation into the product was observed when using
CD3OD as the solvent which strongly supports their proposed
mechanism with dissociation of an anionic R group into
solution. With this in mind, the deuterated aldehyde, α-d-3,4,5-
Figure 10. Four different geometrical isomers formed from oxidative
addition of benzaldehyde.
F
DOI: 10.1021/ja5106943
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
kcal/mol (Figure 11, iTS1) was found for the formation of the
isomer with the acyl group coordinated trans to chloride.
presence of an excess of H2O and LiCl, the clusters at −11.2
and −11.6 ppm were still observed, but the formation of 4 was
inhibited, which suggests the effect may be due to the basic
properties of H2O and LiCl. Stronger bases, though, such as
K2CO3, retard the overall transformation.9 To distinguish
between H2O and LiCl, two additional experiments were
conducted: one with an excess of H2O and one with an excess
of extensively dried LiCl. The H2O experiment showed
inhibition of 4, while the experiment with LiCl gave a higher
conversion of benzyl alcohol, but only traces of 3A and 3B
1
could be observed in the H NMR spectrum. A 31P NMR
spectrum of the LiCl experiment revealed complete conversion
of 1 and the formation of a new complex with two singlet peaks
of equal integrals at 44.2 and 25.2 ppm. The IR spectrum
showed a strong absorbance at 1955 cm−1, proving that the CO
ligand was still coordinated. To obtain more information, the
newly observed species was labeled with [13C]CO from α-
[13C]-benzyl alcohol, which afforded a double doublet in the
13C NMR spectrum at 167.4 ppm and changed the two singlets
in the 31P NMR spectrum into two doublets with couplings of
3.5 and 14.3 Hz, respectively. The disappearance of the C−P
trans-coupling and the P−P cis-coupling from 1 indicates that
one of the phosphines has dissociated. Unfortunately, we were
not successful in obtaining a crystal structure. The complex
does possess catalytic activity, although with a lower conversion
rate than 1. Starting from [Ir(cod)Cl]2 and rac-BINAP, a crude
31P NMR spectrum was recorded after 15 turnovers in the
decarbonylation of 3,4,5-trimethoxybenzaldehyde at 164 °C in
mesitylene/H2O (150 ppm) which verified that 1 is still the
major iridium species present. These observations show that
phosphine dissociation is indeed possible under the reaction
conditions as required in pathway K and if the phosphine
dissociation is rate determining it also explains the lack of
electronic and kinetic isotope effects observed for the
decarbonylation.
Figure 11. Lowest energy profile of the decarbonylation pathway I.
The barrier for the reductive elimination through a four-
membered transition state was very high for all the isomers.
The lowest was found to be 52.2 kcal/mol (Figure 11, iTS2),
and again this was observed for the isomer with the acyl group
coordinated trans to chloride. We would, however, expect an
energy profile with relative free energies closer to those in the
dehydrogenation cycle and hence this makes pathway I
unlikely.
The lowest energy profile for the decarbonylation through
pathway K was also calculated and is illustrated in Figure 12.
In 2008 Tsuji et al. reported the decarbonylation of
aldehydes with 2.5% of [Ir(cod)Cl]2 and 5% of PPh3 in
refluxing dioxane, i.e., with an iridium/phosphine ratio of 1:1.6c
Lower yields were observed when the same reaction was
performed with an iridium/phosphine ratio of 1:2 by using
either 10% of PPh3 or 5% of rac-BINAP. This indicates that the
catalytically active species has less than two phosphines
coordinated to iridium in the optimum system. Furthermore,
a KIE of 1.70 was measured with 2.5% of [Ir(cod)Cl]2 and 5%
of PPh36c which further illustrates the kinetic differences to the
present system with [Ir(cod)Cl]2 and rac-BINAP. Most likely,
the higher reaction temperature needed in the present system is
explained by the higher energy required for phosphine
dissociation in pathway K.
The synthesis of iridium carbonyl complexes from [Ir(cod)-
Cl]2 and benzyl alcohol could be extended to a number of other
phosphine ligands. In the presence of BIPHEP, the complex
IrCl(CO)BIPHEP was isolated as an orange solid in 66% yield.
This is in line with our original ligand screening where BIPHEP
gave results which were comparable with BINAP.9 1,2-
(Diphenylphosphino)benzene produced an unstable complex
that rapidly went from orange to green, which is probably the
reason that this ligand only gave a few turnovers in the
dehydrogenative decarbonylation.9 1,3-(Diphenylphosphino)-
propane (dppp), which should resemble BINAP with regard to
the bite angle,32 afforded a yellow dimeric carbonyl complex.33
The complex [IrCl(CO)dppp]2 coordinated molecular oxygen
very easily which can be seen in the crystal structure
Figure 12. Lowest energy profile of the decarbonylation pathway K.
Unfortunately, the transition state between the saturated
oxidative addition product k1 and the phosphine dissociated
species k2 could not be located. The migratory extrusion of
carbon monoxide takes place through transition state kTS1
with a relative Gibbs free energy of 41.6 kcal/mol. The
following reductive elimination of benzene from k3 occurs
rather readily through transition state kTS2.
The Effect of LiCl and H2O. We had proposed earlier that
the rate enhancement effect of H2O and LiCl could be
explained by the inhibition of inactive iridium clusters.9
However, when 1 was reacted with benzyl alcohol in the
G
DOI: 10.1021/ja5106943
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 13. Proposed catalytic cycles.
́
(Supporting Information). Vaskas complex could not be
ASSOCIATED CONTENT
* Supporting Information
Experimental procedures as well as spectroscopic, crystallo-
graphic and computational data. This material is available free
of charge via the Internet at http://pubs.acs.org.
■
formed with benzyl alcohol as the CO source, but with
benzaldehyde, IrCl(CO)(PPh3)2 was formed cleanly in the
presence of [Ir(cod)Cl]2 and triphenylphosphine. The
decarbonylation was carried out in refluxing mesitylene since
the same reaction in toluene only afforded the complex
S
IrCl(cod)PPh3. It has previously been observed that Vaskas
́
AUTHOR INFORMATION
■
complex does not dehydrogenate an alcohol without the
presence of a strong base.34 These results again illustrate that
monodentate phosphine ligands are less effective in the
dehydrogenation cycle.
Corresponding Authors
phera@organ.su.se
rm@kemi.dtu.dk
Notes
The authors declare no competing financial interest.
CONCLUSION
■
ACKNOWLEDGMENTS
In summary, the overall proposed catalytic cycles are composed
of pathway D for the dehydrogenation and pathway K for the
decarbonylation as illustrated in Figure 13. The formation of 1
occurs within seconds after initiation of the reaction.
Experimental observations and DFT calculations suggest
substitution of chloride with the alcohol to form HCl, which
can be temporarily stored by reversible oxidative addition to 1.
Complex 6 is able to undergo β-hydride elimination according
to DFT calculations and the resulting aldehyde dissociates from
complex 7 followed by oxidative addition of HCl. DFT
calculations indicate that the involvement of the IrHCl2(CO)-
BINAP lowers the energetic span of the dehydrogenation cycle.
The reductive elimination of molecular hydrogen from complex
3 occurs readily under the reaction conditions, as was
concluded after heating the dihydride complex in mesitylene.
The decarbonylation is initiated by oxidative addition of the
aldehyde to complex 1. The coordination of carbon monoxide
was shown to be reversible which indicates that CO is liberated
from the dicarbonyl complex 5. The phosphine dissociation
pathway explains the lack of a KIE and electronic effects in the
decarbonylation. The rate enhancement effect of H2O was
explained by its basic properties and the effect of LiCl is
speculated to aid in the phosphine dissociation. This completes
the two catalytic cycles and highlights the multiple roles of a
single complex in a tandem catalytic transformation.
■
We thank The Danish Council for Strategic Research, The
Swedish Energy Agency, and Nordic Energy Research (N-
INNER II) for financial support. Thishana Singh acknowledges
the NRF South Africa and the College of Agriculture,
Engineering and Science at the University of KwaZulu-Natal,
South Africa for a postdoctoral research fellowship. All
computations were carried out using the computational cluster
resources at the National Supercomputer Centre based at
Linkoping University, Sweden.
REFERENCES
■
(1) (a) Nielsen, M.; Alberico, E.; Baumann, W.; Drexler, H.-J.; Junge,
H.; Gladiali, S.; Beller, M. Nature 2013, 495, 85−89. (b) Rodríguez-
Lugo, R. E.; Trincado, T.; Vogt, M.; Tewes, F.; Santiso-Quinones, G.;
Grutzmacher, H. Nat. Chem. 2013, 5, 342−347. (c) Nielsen, M.;
̈
Kammer, A.; Cozzula, D.; Junge, H.; Gladiali, S.; Beller, M. Angew.
Chem., Int. Ed. 2011, 50, 9593−9597.
(2) (a) Gunanathan, C.; Milstein, D. Science 2013, 341, 1229712.
(b) Bahn, S.; Imm, S.; Neubert, L.; Zhang, M.; Neumann, H.; Beller,
̈
M. ChemCatChem 2011, 3, 1853−1864. (c) Obora, Y.; Ishii, Y. Synlett
2011, 30−51. (d) Dobereiner, G. E.; Crabtree, R. H. Chem. Rev. 2010,
110, 681−703.
(3) (a) Ye, X.; Plessow, P. N.; Brinks, M. K.; Schelwies, M.; Schaub,
T.; Rominger, F.; Paciello, R.; Limbach, M.; Hofmann, P. J. Am. Chem.
Soc. 2014, 136, 5923−5929. (b) Warner, M. C.; Backvall, J.-E. Acc.
̈
H
DOI: 10.1021/ja5106943
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Chem. Res. 2013, 46, 2545−2555. (c) Polukeev, A. V.; Petrovskii, P. V.;
Peregudov, A. S.; Ezernitskaya, M. G.; Koridze, A. A. Organometallics
2013, 32, 1000−1015. (d) Makarov, I. S.; Fristrup, P.; Madsen, R.
Chem.Eur. J. 2012, 18, 15683−15692. (e) Fristrup, P.; Tursky, M.;
Madsen, R. Org. Biomol. Chem. 2012, 10, 2569−2577.
Chem., Int. Ed. 2009, 48, 582−585. (b) Dervisi, A.; Carcedo, C.; Ooi,
L. Adv. Synth. Catal. 2006, 348, 175−183. (c) Smidt, S. P.;
Zimmermann, N.; Studer, M.; Pfaltz, A. Chem.Eur. J. 2004, 10,
4685−4693. (d) Smidt, S. P.; Pfaltz, A.; Martínez-Viviente, E.;
Pregosin, P. S.; Albinati, A. Organometallics 2003, 22, 1000−1009.
(e) Crabtree, R. Acc. Chem. Res. 1979, 12, 331−337.
(4) (a) Qu, S.; Dang, Y.; Song, C.; Wen, M.; Huang, K.-W.; Wang,
Z.-X. J. Am. Chem. Soc. 2014, 136, 4974−4991. (b) Li, H.; Hall, M. B.
J. Am. Chem. Soc. 2014, 136, 383−395. (c) Yang, X. ACS Catal. 2014,
4, 1129−1133. (d) Zeng, G.; Sakaki, S.; Fujita, K.; Sano, H.;
Yamaguchi, R. ACS Catal. 2014, 4, 1010−1020. (e) Cho, D.; Ko, K.
C.; Lee, J. Y. Organometallics 2013, 32, 4571−4576. (f) Sandhya, K. S.;
Suresh, C. H. Organometallics 2013, 32, 2926−2933. (g) Sieffert, N.;
(25) Schwartz, J.; Cannon, J. B. J. Am. Chem. Soc. 1974, 96, 2276−
2278.
(26) Wertz, D. H. J. Am. Chem. Soc. 1980, 102, 5316−5322.
(27) An additional “outer-sphere” mechanism for the dehydrogen-
ation was also considered (see ref 4g). In this pathway, alkoxo complex
d2 reacts with another molecule of benzyl alcohol through a six-
membered ring with the simultaneous transfer of the hydride and the
proton to iridium and oxygen, respectively. Unfortunately, we were
never able to locate a suitable transition state for this pathway.
(28) Hermange, P.; Lindhardt, A. T.; Taaning, R. H.; Bjerglund, K.;
Lupp, D.; Skrydstrup, T. J. Am. Chem. Soc. 2011, 133, 6061−6071.
(29) Burk, M. J.; Crabtree, R. H. Inorg. Chem. 1986, 25, 931−932.
Buhl, M. J. Am. Chem. Soc. 2010, 132, 8056−8070.
̈
(5) (a) Kreis, M.; Palmelund, A.; Bunch, L.; Madsen, R. Adv. Synth.
Catal. 2006, 348, 2148−2151. (b) Beck, C. M.; Rathmill, S. E.; Park, Y.
J.; Chen, J.; Crabtree, R. H.; Liable-Sands, L. M.; Rheingold, A. L.
Organometallics 1999, 18, 5311−5317. (c) Doughty, D. H.; Pignolet,
L. H. J. Am. Chem. Soc. 1978, 100, 7083−7085.
(30) (a) Sieffert, N.; Reo
́
creux, R.; Lorusso, P.; Cole-Hamilton, D. J.;
(6) (a) Adams, J. J.; Arulsamy, N.; Roddick, D. M. Organometallics
2012, 31, 1439−1447. (b) Geilen, F. M. A.; vom Stein, T.; Engendahl,
B.; Winterle, S.; Liauw, M. A.; Klankermayer, J.; Leitner, W. Angew.
Chem., Int. Ed. 2011, 50, 6831−6834. (c) Iwai, T.; Fujihara, T.; Tsuji,
Y. Chem. Commun. 2008, 6215−6217.
Buhl, M. Chem.Eur. J. 2014, 20, 4141−4155. (b) Shuai, Q.; Yang, L.;
̈
́
Guo, X.; Basle, O.; Li, C.-J. J. Am. Chem. Soc. 2010, 132, 12212−
12213. (c) Garralda, M. A. Dalton Trans. 2009, 3635−3645. (d) Guo,
X.; Wang, J.; Li, C.-J. J. Am. Chem. Soc. 2009, 131, 15092−15093.
(31) Cordaro, J. G.; Bergman, R. G. J. Am. Chem. Soc. 2004, 126,
16912−16929.
(7) Modak, A.; Deb, A.; Patra, T.; Rana, S.; Maity, S.; Maiti, D. Chem.
Commun. 2012, 48, 4253−4255.
(32) Birkholz, M.-N.; Freixab, Z.; van Leeuwen, P. W. N. M. Chem.
Soc. Rev. 2009, 38, 1099−1118.
(8) Fristrup, P.; Kreis, M.; Palmelund, A.; Norrby, P.-O.; Madsen, R.
J. Am. Chem. Chem. 2008, 130, 5206−5215.
(33) Zotto, A. D.; Costella, L.; Mezzetti, A.; Rigo, P. J. Organomet.
Chem. 1991, 414, 109−118.
(9) Olsen, E. P. K.; Madsen, R. Chem.Eur. J. 2012, 18, 16023−
16029.
(34) Huang, J.-L.; Dai, X.-J.; Li, C.-J. Eur. J. Org. Chem. 2013, 6496−
6500.
(10) (a) Shindoh, N.; Takemoto, Y.; Takasu, K. Chem.Eur. J. 2009,
15, 12168−12179. (b) Alcaide, B.; Almendros, P.; Luna, A. Chem. Rev.
2009, 109, 3817−3858. (c) Wasilke, J.-C.; Obrey, S. J.; Baker, R. T.;
Bazan, G. C. Chem. Rev. 2005, 105, 1001−1020. (d) Fogg, D. E.; dos
Santos, E. N. Coord. Chem. Rev. 2004, 248, 2365−2379.
(11) (a) Leitch, D. C.; Labinger, J. A.; Bercaw, J. E. Organometallics
2014, 33, 3353−3365. (b) Young, A.; Vincent, M. A.; Hillier, I. H.;
Percy, J. M.; Tuttle, T. Dalton Trans. 2014, 43, 8493−8498.
(c) Higman, C. S.; Plais, L.; Fogg, D. E. ChemCatChem. 2013, 5,
3548−3551. (d) Pijnenburg, N. J. M.; Cabon, Y. H. M.; van Koten, G.;
Klein Gebbink, R. J. M. Chem.Eur. J. 2013, 19, 4858−4868.
(12) (a) Christensen, S. H.; Olsen, E. P. K.; Rosenbaum, J.; Madsen,
R. Org. Biomol. Chem. 2015, 13, 938−945. (b) Verendel, J. J.;
Nordlund, M.; Andersson, P. G. ChemSusChem 2013, 6, 426−429.
(13) Ho, H.-A.; Manna, K.; Sadow, A. D. Angew. Chem., Int. Ed. 2012,
51, 8607−8610.
(14) Modak, A.; Naveen, T.; Maiti, D. Chem. Commun. 2013, 49,
252−254.
(15) Behr, A.; Vorholt, A. J.; Ostrowski, K. A.; Seidensticker, T. Green
Chem. 2014, 16, 982−1006.
(16) Atesin, A. Ç.; Duckett, S. B.; Flaschenriem, C.; Brennessel, W.
W.; Eisenberg, R. Inorg. Chem. 2007, 46, 1196−1204.
(17) Rosini, G. P.; Liu, F.; Jespersen, K. K.; Goldman, A. S.; Li, C.;
Nolan, S. P. J. Am. Chem. Soc. 1998, 120, 9256−9266.
(18) Gom
4963.
́
ez-Gallego, M.; Sierra, M. A. Chem. Rev. 2011, 111, 4857−
(19) (a) Kunin, A. J.; Johnson, C. E.; Maguire, J. A.; Jones, W. D.;
Eisenberg, R. J. Am. Chem. Soc. 1987, 109, 2963−2968. (b) Kunin, A.
J.; Farid, R.; Johnson, C. E.; Eisenberg, R. J. Am. Chem. Soc. 1985, 107,
5315−5317. (c) Johnson, C. E.; Eisenberg, R. J. Am. Chem. Soc. 1985,
107, 3148−3160.
(20) Blum, O.; Milstein, D. J. Am. Chem. Soc. 2002, 124, 11456−
11467.
(21) Blum, O.; Milstein, D. J. Organomet. Chem. 2000, 593−594,
479−484.
(22) Zhao, J.; Hesslink, H.; Hartwig, J. F. J. Am. Chem. Soc. 2001, 123,
7220−7227.
(23) Vaska, L. Acc. Chem. Res. 1968, 1, 335−344.
(24) (a) Xu, Y.; Celik, M. A.; Thompson, A. L.; Cai, H.; Yurtsever,
M.; Odell, B.; Green, J. C.; Mingos, D. M. P.; Brown, J. M. Angew.
I
DOI: 10.1021/ja5106943
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX