Precise supramolecular control of selectivity in the Rh-catalyzed hydroformylation of terminal and internal alkenes

Article abstract of DOI:10.1021/ja4046235

In this study, we report a series of DIMPhos ligands L1-L3, bidentate phosphorus ligands equipped with an integral anion binding site (the DIM pocket). Coordination studies show that these ligands bind to a rhodium center in a bidentate fashion. Experiments under hydroformylation conditions confirm the formation of the mononuclear hydridobiscarbonyl rhodium complexes that are generally assumed to be active in hydroformylation. The metal complexes formed still strongly bind the anionic species in the binding site of the ligand, without affecting the metal coordination sphere. These bifunctional properties of DIMPhos are further demonstrated by the crystal structure of the rhodium complex with acetate anion bound in the binding site of the ligand. The catalytic studies demonstrate that substrate preorganization by binding in the DIM pocket of the ligand results in unprecedented selectivities in hydroformylation of terminal and internal alkenes functionalized with an anionic group. Remarkably, the selectivity controlling anionic group can be even 10 bonds away from the reactive double bond, demonstrating the potential of this supramolecular approach. Control experiments confirm the crucial role of the anion binding for the selectivity. DFT studies on the decisive intermediates reveal that the anion binding in the DIM pocket restricts the rotational freedom of the reactive double bound. As a consequence, the pathway to the undesired product is strongly hindered, whereas that for the desired product is lowered in energy. Detailed kinetic studies, together with the in situ spectroscopic measurements and isotope-labeling studies, support this mode of operation and reveal that these supramolecular systems follow enzymatic-type Michaelis-Menten kinetics, with competitive product inhibition.

Full text of DOI:10.1021/ja4046235

Article
pubs.acs.org/JACS
Precise Supramolecular Control of Selectivity in the Rh-Catalyzed
Hydroformylation of Terminal and Internal Alkenes
Paweł Dydio,† Remko J. Detz,†,‡ and Joost N. H. Reek*,†
†van’t Hoff Institute for Molecular Sciences, University of Amsterdam, Science Park 904, 1098 XH, Amsterdam, The Netherlands
‡InCatT B.V., Science Park 904, 1098 XH, Amsterdam, The Netherlands
S
* Supporting Information
ABSTRACT: In this study, we report a series of DIMPhos
ligands L1−L3, bidentate phosphorus ligands equipped with
an integral anion binding site (the DIM pocket). Coordination
studies show that these ligands bind to a rhodium center in a
bidentate fashion. Experiments under hydroformylation
conditions confirm the formation of the mononuclear
hydridobiscarbonyl rhodium complexes that are generally
assumed to be active in hydroformylation. The metal
complexes formed still strongly bind the anionic species in
the binding site of the ligand, without affecting the metal
coordination sphere. These bifunctional properties of DIMPhos are further demonstrated by the crystal structure of the rhodium
complex with acetate anion bound in the binding site of the ligand. The catalytic studies demonstrate that substrate
preorganization by binding in the DIM pocket of the ligand results in unprecedented selectivities in hydroformylation of terminal
and internal alkenes functionalized with an anionic group. Remarkably, the selectivity controlling anionic group can be even 10
bonds away from the reactive double bond, demonstrating the potential of this supramolecular approach. Control experiments
confirm the crucial role of the anion binding for the selectivity. DFT studies on the decisive intermediates reveal that the anion
binding in the DIM pocket restricts the rotational freedom of the reactive double bound. As a consequence, the pathway to the
undesired product is strongly hindered, whereas that for the desired product is lowered in energy. Detailed kinetic studies,
together with the in situ spectroscopic measurements and isotope-labeling studies, support this mode of operation and reveal that
these supramolecular systems follow enzymatic-type Michaelis−Menten kinetics, with competitive product inhibition.
This can be achieved by using bifunctional ligands that can
coordinate to the metal center and bind noncovalently to a
substrate molecule.6 This supramolecular substrate binding can
in principle preorganize the reactive functionality at the
catalytic center (Figure 1) such that one of the competitive
reaction pathways is favored over the competing ones,
INTRODUCTION
■
Transition metal catalysis is a powerful enabling technology for
the sustainable preparation of chemical compounds.1 However,
the value of the individual catalytic transformations depends
largely on the access to catalysts displaying the required
selectivity, activity, and stability.2 Various tools to control these
decisive features by modification of ligands coordinated to the
catalytically active metal centers have been introduced.3 The
traditional approach to catalyst development involves knowl-
edge-supported trial-and-error protocols, for which combina-
torial and high-throughput screening methods of putative
catalysts have demonstrated their added value.4 If the required
selectivity is not obtained using these strategies, “directing
groups” (DGs) can be used. Such groups should be introduced
to the substrate molecules, and during the reaction they steer
the selectivity by coordination to the metal center, directing the
reaction toward the desired product.5 Although effective, this
method is limited to substrates with DGs spatially close to the
reactive functionality, imposing limitations. Moreover, the
reaction occurring at the metal center should be compatible
with the DGs, further limiting its potential. For these reasons, it
is highly interesting to develop alternative methodologies with
directing groups that operate via interactions between the
functional groups of the substrate and the ligand of the catalyst.
Figure 1. General concept of substrate preorganization by a catalyst
with a bifunctional ligand, consisting of a donating function for
catalytic center coordination and a specific recognition site for binding
to a functional group of a substrate.
Received: May 15, 2013
Published: June 25, 2013
© 2013 American Chemical Society
10817
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
controlling the overall selectivity. This principle was demon-
strated in selective oxidation reactions catalyzed by metal-
loporphyrins by Breslow and co-workers,7 using hydrophobic
or coordination interactions as the driving force. Hydrogen
bonding is highly directional and can lead to relatively strong
interactions; hence, it provides a powerful tool for the rational
design of selective catalysts that operate via substrate
orientation.8 As such, it was elegantly used for controlling the
regioselectivity in the Ru-catalyzed hydration of alkynes,9 the
Mn-catalyzed C−H oxidation at sp3-carbon atoms,10 and the
Rh-catalyzed hydroformylation of β,γ-unsaturated carboxylic
acids.11 A two-point hydrogen-bonding interaction allowed one
also to control regio- and enantioselectivity in the Ru-catalyzed
epoxidation of olefinic double bonds,12 and a simple single
hydrogen bond was shown to improve the selectivity in the Co-
catalyzed cyclopropanation13 and the enantioselectivity in the
Rh-catalyzed hydrogenation of olefins.14
precedents of precise selectivity control.11,18,19 DFT calcu-
lations of the decisive intermediates reveal the mode of
operation of this new catalyst. The anion binding in the DIM
pocket restricts the rotational freedom of the reactive double
bound required during the hydride migration step. As a result,
the pathway to the undesired product is strongly hindered,
whereas that for the desired product is lowered in energy. The
kinetic studies and the in situ spectroscopic measurements
support this mechanism, and reveal that the system follows
Michaelis−Menten kinetics. Full details of these studies are
presented in the following sections of this Article.20
RESULTS AND DISCUSSION
■
Coordination and Anion Binding Properties. We first
investigated the anion binding and coordination properties of
the DIMPhos ligands. Strong anion binding of L1 was apparent
as the presence of acetate anions triggered a significant
downfield shift of the NH signals (Δδ = 2.4−3.9 ppm) in
The examples reported so far9−14 addressed successfully
quite small or rigid substrate molecules of rather limited scope.
We anticipated that the “remote control” for more flexible and
bigger substrates requires catalytic systems based on well-
defined, rigid complex structures, in which the positions of the
controlling unit and the catalytic centers are precisely defined.
In this respect, the use of functionalized bidentate ligands has
an advantage over functionalized monodentate ligands11 as the
number of possible complexes that can be formed is
significantly lower. We considered that neutral anion
receptors15 that operate with hydrogen bonding are excellent
candidates for the substrate-directing motifs for supramolecular
catalysts, as they can interact with one of the most common
functional groups in organic compounds, that is, carboxyl
moiety. Therefore, we chose the 7,7′-diamido-2,2′-diindolyl-
methane (the DIM pocket),16 a tailor-made receptor for
carboxylate and phosphate anions, as a scaffold to prepare new
bidentate DIMPhos ligands L1−L3 (Figure 2). In this Article,
we report in depth studies that demonstrate that these new
ligands can control the regioselectivity in hydroformylation of
alkenes by substrate orientation in the binding site. Especially,
control over selectivity in the hydroformylation of internal
alkenes is challenging,17 and there are only a few literature
1
the H NMR spectra in CD2Cl2. Further, the titration studies
revealed the high association constant, Ka ≫ 105 M−1, for the
formation of a 1:1 ligand−acetate anion complex. Upon the
addition of the rhodium precursor, [Rh(acac)(C2H4)2] (acac =
acetylacetonate), the rhodium−ligand complex, the precursor
of the active hydroformylation catalyst17a with the acetate
remained bound in the DIM pocket, [Rh(L1·AcO)(acac)], was
formed. Also, the addition of [RhCl(CO)2]2 as metal precursor
led to the formation of the complex with both phosphorus
atoms coordinated to the rhodium center in a trans mutual
orientation, with the acetate anion bound in the DIM pocket.
The structure of this complex was also elucidated by the X-ray
crystallography of TBA[Rh(L1·AcO)(CO)(Cl)] crystals (Fig-
ure 2 (right), TBA+ = tetrabutylammonium cation).20a As
anticipated, the acetate is bound in the binding site with four
strong hydrogen bonds (the N−O distances are 2.737(3) and
3.006(3) Å, for the amide and indole N atoms, respectively),
and, importantly, its aliphatic group points toward the metal
center.
High-pressure (HP) NMR studies reveal that a 1:1 mixture
of ligand L1 and [Rh(acac)(CO)2] under hydroformylation
conditions, 5 bar CO/H2 (1:1), results in exclusive formation of
a trigonal bipyramidal hydrido complex [Rh(L1)(CO)2H], the
catalytically active complex for hydroformylation.17 A doublet
1
of triplets for the hydride signal at δ = −9.5 ppm in the H
NMR spectrum indicates that the hydride couples both with
rhodium and the two phosphorus donor atoms. This signal
1
simplifies in a phosphorus-decoupled H{31P} NMR experi-
ment, and the observed doublet is consistent with the hydride
coupled to rhodium. The 31P{1H} NMR spectrum displays only
one doublet at δ = 36.7 ppm, indicative of the phosphine
coupling with rhodium and showing the equivalency of both
phosphorus atoms. Upon lowering the temperature from 25 to
−95 °C, the signals in both H and 31P NMR spectra broaden
1
and split into two sets.21 These experiments establish that the
bidentate ligand L1 coordinates in both equatorial−equatorial
(eq−eq) and equatorial−axial (eq−ax) fashion and that at
room temperature these isomeric complexes are in fast
equilibrium on the NMR time scale. The low value of the
phosphorus−hydride coupling (4.0 Hz) indicates that the eq−
eq isomer dominates the equilibrium.22 In line with this, high-
pressure infrared (IR) studies using either H2/CO or D2/CO
(both 1:1, 20 bar) show absorption bands in the carbonyl
region corresponding to both eq−eq and eq−ax isomeric
complexes.23 Furthermore, DFT calculations (BP86, SV(P))
Figure 2. Structure of DIMPhos ligands L1−L3 and DIM anion
receptor R1 (left) and X-ray structure of the supramolecular complex
[Rh(L1·AcO)(CO)Cl]− (right); TBA+ counterion, disordered solvent
molecules, and most hydrogen atoms are omitted for clarity.
10818
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
indicate that the complex can adopt both conformations, and
that the eq−eq isomer is more favored by 11.4 kJ/mol.24
Importantly, HP NMR studies show that the coordination
geometry around the rhodium center does not change in the
Table 1. Hydroformylation of Anionic Substrates 1a−8a and
a
Control Experiments
conversion
(%)
regioselectivity
(l/b ratio)
no.
ligand
L1
substrate
n
−
presence of anions (acetate or H2PO4 ) that are bound in the
1
2
1a
2a
3a
4a
5a
6a
7a
8a
1b
1c
2b
2c
2c
2c
2c
2a
2a
2a
1
2
3
4
5
6
7
8
1
1
2
2
2
2
2
2
2
2
23
2.6
binding site of the ligand. The signals of the NH groups of the
b
b
L1
95 (80)
40 (66)
1
ligand are shifted toward lower fields in the H NMR spectra
3
L1
85
77
58
56
45
42
46
46
49
42
43
34
27
100
100
13
22
(Δδ = 2.5−3.1 ppm), confirming the formation of strong
hydrogen bonds between the anions and the binding site of the
ligand.15 The carbonyl absorption bands of the rhodium
complex show only a small shift in the HP IR spectra to lower
wavenumbers (Δν up to 5 cm−1) upon anion binding,
indicating a slightly increased electron density at the metal
complex. The binding constants for carboxylate and phosphate
anions to the DIM binding site of [Rh(L1)(CO)2H] were
determined from titration experiments performed at 5 bar CO/
H2 (1:1) in CD2Cl2 by using HP NMR spectroscopy. These
studies reveal that only one anion is bound in the DIM pocket
of [Rh(L1)(CO)2H], and that the association constants for
4
L1
27
5
L1
19
6
L1
24
7
L1
20
8
L1
15
9
L1
1.4
1.6
3.7
3.6
3.6
3.9
3.5
2.9
3.1
1.8
10
11
12
L1
L1
L1
c
13
L1
d
14
L1
e
15
L1
CH3COO− and H2PO4 are higher than 105 M−1 and around
−
16
17
18
R1/PPh3
PPh3
103.7 M−1, respectively. In contrast to these anionic species, the
acidic (CH3COOH and H3PO4) and the (alkyl) esters
analogues are not bound in the DIM pocket of the ligand,24
and therefore they are well suited for control experiments.
Regioselective Hydroformylation of Terminally Un-
saturated Aliphatic Acids. We next studied the performance
of ligand L1 in the rhodium-catalyzed hydroformylation of a
series of deprotonated ω-unsaturated carboxylic acids 1a−8a
(Scheme 1), varying the aliphatic chain length between the
a
Reagents and conditions: [Rh(acac)(CO)2]/L1(R1 and/or PPh3)/
substrate = 1:3(3 and/or 6):100; c(Rh) = 2 mM, 20 bar CO/H2 (1:1),
CH2Cl2, 40 °C, 24 h, N,N-diisopropylethylamine (DIPEA, 1.5 equiv)
was used as a base for anionic substrate generation; triethylamine
(TEA) can be used alternatively. Regioselectivity and conversion (%)
were determined by 1H NMR analysis of the reaction mixture.
Isomerization and hydrogenation side reactions were not observed.
b
Values between parentheses are for the reaction at room temperature
c
d
for 72 h. Reaction in the presence of DIPEA (0.3 M). Reaction in
the presence of acetic acid (0.2 M). Reaction in the presence of a
Scheme 1
e
mixture of acetic acid and DIPEA (0.2 and 0.3 M, respectively). For
full experimental details, see the Supporting Information.
(4-pentenoate anion) that precisely spans the distance between
the metal and the DIM pocket is hydroformylated with
unprecedented selectivity for the linear aldehyde (l/b ratio of
40). If the reaction is carried out at room temperature (instead
of 40 °C), the l/b ratio reaches 66. In contrast, the neutral acid
(2b) and the methyl ester (2c) analogues of 2a, substrates of
the same size that do not bind to the DIM binding site of the
ligand, form the aldehyde with the typical low selectivity (l/b
ratios of ca. 3), verifying the importance of the anion binding.
These results demonstrate that for anionic 2a the reaction
barrier for the formation of the linear aldehyde is effectively
lowered, with respect to that for the branched product, by the
substrate binding event. Interestingly, along with the higher
selectivity, the conversion is also much higher when the
substrate binds in the DIM pocket of the ligand. The observed
rate acceleration can result from the overall lowered reaction
barrier due to the substrate preorganization, as well as from the
higher concentration of the olefin near the metal center due to
substrate prebinding to the DIM pocket of the ligand; that is,
the so-called effective concentration25 of the olefin is
substantially higher than the actual concentration of the olefin
in solution. Obviously, both effects can contribute simulta-
neously to the overall increase in rate.
carboxyl moiety and the double bond, that is, from 3-butenoic
to 10-undecenoic acid (Table 1). Molecular modeling reveals
that the shortest substrate 1a (3-butenoate anion) cannot
simultaneously bind to the anion binding site and coordinate to
the Rh center with its double bond. The homologue substrate
2a that is one carbon longer (4-pentenoate anion) precisely
spans the distance between the metal and the binding site of
the catalyst, whereas the other substrates 3a−8a fit easily. To
verify the influence of the anion binding on the reaction
selectivity, we performed control experiments with the neutral
acids 1b−8b and their methyl esters 1c−8c (Table 1 and Table
S2), substrates that do not bind in the DIM pocket of the
catalyst (vide supra). As anticipated, the shortest substrate 1a is
hydroformylated with poor selectivity, and hardly any difference
in reactivity is observed between this anionic substrate and its
acid (1b) and ester (1c) analogues. The linear/branched
selectivity (l/b product ratio) is in the expected range for these
substrates, between 1.6 and 2.6. In sharp contrast, substrate 2a
Control experiments confirmed that the presence of acetate
ions has negligible effect on the regioselectivity and activity of
the Rh(L1) catalyst for nonanionic substrates, for example,
methyl 4-pentenoate (2c) (Table 1, entries 12−15). To further
verify that the anion binding site and the catalytic center must
10819
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
a
be present as an integrated system, we performed a control
experiment using a mixture of the anion receptor R1 (Figure 2)
and triphenylphosphine (PPh3). In this case, substrate 2a (4-
pentenoate anion) is hydroformylated with low selectivity,
nearly the same as displayed by the catalyst based on only
triphenylphospine (Table 1, entries 16 and 17). In the absence
of any phosphorus ligand, substrate 2a is also hydroformylated
with poor selectivity (l/b = 1.8) and low conversion (Table 1,
entry 18).
Table 2. Hydroformylation of Substrates 9−13
no.
substrate
n
conversion (%)
regioselectivity (l/b ratio)
1
2
3
4
5
6
7
8
9
9
79
84
22
>100
28
10
11
13
b
b
12a
12b
12c
13a
13b
13c
1
1
1
2
2
2
10(69)
− (1.6)
c
0
3
0.6
100
>40
Substrates 3a−8a (5-hexenoate through 10-undecenoate
anions) that are the longer homologues of the optimal
substrate 2a also experience an effect of binding in the DIM
pocket of the ligand (Table 1). These substrates are
hydroformylated with higher selectivity for linear products
than their acid 3b−8b or ester analogues 3c−8c. In these
reactions, the l/b ratios are larger than 15 and therefore all
considerably higher than that for the substrates that do not bind
to the DIM pocket of the ligand (Table S2). Notably, both the
regioselectivity and the rate enhancement (based on con-
version) gradually drop with the increasing distance between
the anion and the alkene of the substrate (i.e., from 2a to 8a).
This trend can be explained by the aforementioned effective
concentration once the alkene is bound in the DIM pocket of
the catalyst, which depends on the inverse cube of the linker
length.25 Consequently, with larger substrates the effective
concentration of alkene is lower, hence the reaction is slower,
and the alternative pathway via the nonbound species (that is
the nonselective “background reaction”) contributes effectively,
lowering to some extent the overall selectivity of the reaction.
Also, the longer substrates lead to complexes with less
perturbed alkene coordination, which could affect their
reactivity (vide infra). Interestingly, for the longest substrate
8a, the conversion is similar to that for ester 8c; however, the
selectivity is still enhanced. Notably, among the series of
substrates 1a−8a, the highest regioselectivity and the highest
rate enhancement were achieved for substrate 2a, 4-pentenoate
anion that fits precisely in the catalytic cavity, which is between
the DIM pocket and the rhodium center.
c
0
12
4.0
a
Reagents and conditions: [Rh(acac)(CO)2]/L1/substrate = 1:3:100;
c(Rh) = 2 mM, 20 bar CO/H2 (1:1), CH2Cl2, 40 °C for entries 1−3
and room temperature for entries 4−9, 24 h, N,N-diisopropylethyl-
amine (DIPEA, 1.5 equiv for 9−11 and 3 equiv for 12a and 13a) was
used as a base for anionic substrate generation. Regioselectivity and
conversion (%) were determined by H and/or 13C NMR analysis of
1
the reaction mixture. Isomerization and hydrogenation side reactions
were not observed. At room temperature, the conversion is too low
to determine the selectivity; values between parentheses are for the
reaction at 40 °C. Additional experiments with 1-octene showed that
under these strongly acidic conditions the catalyst is inactive. For full
experimental details, see the Supporting Information.
b
c
Scheme 3
To further investigate the substrate scope, we also evaluated
a small series of (deprotonated) substituted 4-pentenoate acids
9−11 (Scheme 2 and Table 2). These experiments show that
the anionic substrate is used (Table 2, entries 7−9). As
expected, the homologue 12a, allyl-phosponate anion that is
too short to span the distance between the catalytic and the
binding site, is converted with low regioselectivity (Table 2,
entries 4 and 6).
Scheme 2
Kinetic Studies and Mode of Action. To gain a deeper
insight into the reaction mechanism, we studied the hydro-
formylation of substrate 2a by Rh(L1) in more detail. In situ
HP IR spectroscopy identifies the hydrido complex [Rh(L1)-
(CO)2H] as the resting state of the catalyst throughout the
whole catalytic experiment.24 Monitoring reaction progress by
the gas-uptake for experiments with different partial pressures
of CO and H2 reveals the (nearly) zero-order dependence of
the reaction rate (turnover frequency, TOF, consumed
substrate/catalyst/time in mol mol−1 h−1) on the hydrogen
pressure and the negative dependence of the TOF on the
pressure of CO (Table 3). Furthermore, experiments with
different substrate concentrations reveal the positive depend-
ency of the TOF on the alkene concentration (Table 3). Thus,
both in situ HP IR spectroscopy and gas-uptake experiments
are in agreement with a rate-determining step early in the
catalytic cycle (Scheme 4), being either alkene coordination or
hydride migration.3
the Rh(L1) catalyst stays highly selective as long as the
substituents introduced do not hamper the bifunctional
substrate binding (Table 2, entries 1,2 vs entry 3).
In view of the high affinity of the DIM binding site for
phosphate anions (vide supra), we next extended the scope of
substrates to alkenes functionalized with the phosphate group
(Scheme 3 and Table 2). One might expect that substrate 13a,
3-butenylphosphonate anion that is a phosphate analogue of 2a,
would react with similar high selectivity. Indeed, substrate 13a
is hydroformylated by the Rh(L1) catalyst to form the linear
aldehyde with excellent regioselectivity (l/b > 40). Again, the
high selectivity and higher conversion are observed only when
10820
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
a
Table 3. Hydroformylation of Substrate 2a by Rh(L1)
no.
c(2a) (M)
P(H2) (bar)
P(CO) (bar)
TOF
1
2
3
4
5
6
7
1.0
1.0
1.0
1.0
0.5
0.2
0.1
10
10
20
20
10
10
10
10
20
10
20
10
10
10
24
11
25
13
19
15
13
a
Reagents and conditions: [Rh(acac)(CO)2]/L1 = 1:1.5; c(Rh) = 2
mM, CH2Cl2, 40 °C, triethylamine (TEA, 1.5 equiv) was used as a
base for anionic substrate generation. Turnover frequency, TOF (mol
mol−1 h−1), was determined from gas-uptake profiles at 10%
conversion. For full experimental details, see the Supporting
Information.
Scheme 4. Catalytic Cycle of the Rhodium-Catalyzed
Hydroformylation
Reaction progress kinetic analysis26 for experiments under
the standard conditions (CO/H2 1:1, 20 bar (constant
pressure), 40 °C) provides further insight into the reaction
mechanism. The initial studies were performed for ester 2c
(methyl 4-pentenoate), that is, the substrate, which does not
bind in the DIM pocket of the Rh(L1) catalyst. These
experiments reveal, common for hydroformylation,3b the first-
order dependence of the reaction rate on the substrate
concentration (the first-order constant, k1 = 0.0458 h−1; eq
1). Furthermore, no catalyst inhibition by the product was
observed, as indicated by the overlying curves, in Figure 3a.26
Figure 3. Graphical representation of the kinetic profiles: reaction rate
versus substrate concentration plots from reaction at different initial
substrate concentrations for hydroformylation of ester 2c (a) and
anionic 2a (b) using Rh(L1) as the catalyst, determined by gas uptake
methods. Reagents and conditions: 20 bar CO/H2 (1:1), CH2Cl2, 40
°C, [Rh(acac)(CO)2]/L1 = 1:1.5; c(Rh) = 2 mM, triethylamine (TEA,
1.5 equiv) was used as a base for anionic substrate generation, and
pentanoic acid was used to mimic the product (the aldehyde group is
proven not to affect the kinetics) (* = repeated experiment with a
longer reaction time). For full experimental details, see the Supporting
Information.
V = k1·[S]
(1)
for an individual experiment; yet the reaction kinetics is also
strongly dependent on the initial substrate concentration
(nonoverlying blue, red, orange, and green curves, Figure
3b). This observation could indicate either slow catalyst
deactivation during the course of the reaction or catalyst
inhibition by the product formed.26 To discriminate between
these scenarios, we performed an additional experiment using a
mixture of 0.5 M of the substrate and 0.5 M of the product,
simulating the reaction performed with the initial substrate
concentration of 1 M being halfway executed. As is clear from
Figure 3b, the reaction rate versus substrate concentration plots
from these experiments overlay perfectly (green and black
curves, Figure 3b), in contrast to that for the experiment with
the initial substrate concentration of 0.5 M (orange curve,
Figure 3b). These experiments indicate that the catalyst is
Additionally, this analysis shows that the catalyst activation
period is negligible and that there is no catalyst deactivation
occurring during the reaction triggered by the substrate or the
product. Both of these findings stay in agreement with the in
situ HP IR and HP NMR studies, which showed the fast
catalyst activation (<15 min at 40 °C) and no catalyst
decomposition within at least 48 h.
In contrast, the reaction progress kinetic analysis for the
hydroformylation of anionic alkene 2a (4-pentenoate), which is
the substrate that does bind in the binding site of the Rh(L1)
catalyst, demonstrates a different kinetic behavior (under
otherwise identical conditions, Figure 3b). Experiments at
different initial substrate concentrations (0.1−1.0 M of 2a)
reveal a linear dependence of rate on substrate concentration
10821
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
stable and that the reaction is inhibited to some extent by the
product that is formed. Such behavior is in line with a
mechanism in which the substrate is preorganized in the
binding site of the catalyst prior to reacting on the metal center.
Such system should follow Michaelis−Menten kinetics with a
competitive product inhibition (eq 2; V = reaction rate (in M
To investigate the relative substrate selectivity, we next
performed competition experiments with mixtures of sub-
strates. Competitive hydroformylation of substrate 2a that
precisely fits to the Rh(L1) system, and of its methyl ester
analogue (2c), shows that at first mostly anionic substrate is
consumed, while during that reaction period ester 2c reacts
slowly (Figure 4a). During the course of the reaction, when a
h−1), Vmax = maximum reaction rate (in M h−1), Kmm
=
Michaelis constant (in M), Ki = product inhibition constant (in
M), [S] = substrate concentration (in M), and [P] = product
concentration (in M)).27
Vmax ·[S]
Kmm + [S] +
V =
Kmm
Ki
·[P]
(2)
Indeed, the reaction progress data gave a good fit to eq 2,
revealing the following kinetic parameters: Vmax = 0.063 M h−1,
Kmm = 0.168(5) M, and Ki = 0.158(4) M. Nearly equal values of
the Michaelis constant and the product inhibition constant
show that both the substrate and the product interact with the
catalyst in a similar fashion. This indicates that they compete
for the binding in the DIM pocket (rather than for the metal
center). Thus, the product inhibits the reaction via partial
expelling of the substrate from the DIM pocket, hence lowering
its “local concentration”. In view of the nearly equal values of
Kmm and Ki, and the stoichiometry of the reaction (one
molecule of the product formed for one molecule of the
substrate reacted, hence the sum of [S] and [P] being equal to
the initial substrate concentration (c0)), eq 2 can be simplified:
Kmm ≅ Ki
and
[S] + [P] = c0
Thus, in this case, eq 2 simplifies to:
Vmax ·[S]
V =
Kmm + c0
(3)
Equation 3 shows and rationalizes the pseudo-first-order
dependence of the reaction rate on the substrate concentration
observed experimentally for individual experiments (Figure 3b)
and shows the influence of the initial substrate concentration
on the reaction kinetics.
The evaluation of the influence of the concentration of
substrate 2a on its hydroformylation reveals that the selectivity
gradually drops with higher initial concentrations; however, it
does not change during the single experiment (the l/b ratio =
52, 44, 22, and 15 at 0.1, 0.2, 0.5, and 1.0 M solution of 2a,
respectively, and the l/b = 15 at 0.5 M solution of 2a in the
presence of 0.5 M of the product). This shows that at higher
concentrations a greater part of the reaction occurs along a
nonselective pathway with more substrate and product
molecules involved. Higher concentrations favor a scenario in
which the anionic functionality of the reactive substrate is not
bound in the DIM binding site, because the latter is already
occupied by another molecule (substrate or product formed).
Thus, this transformation does not experience any substrate
preorganization and leads to a mixture of both products (in
analogy to reaction with nonanionic substrates). This stays in
agreement with DFT calculations, which suggest that the
branched aldehyde cannot be formed when its anionic moiety is
bound in the DIM binding site of the catalyst (vide infra).
Figure 4. Substrate competition experiments: hydroformylation of a
1:1 mixture of substrates 2a and 2c (a) and 1a and 1c (b) by Rh(L1).
Reagents and conditions: 20 bar CO/H2 (1:1), CH2Cl2, 40 °C,
[Rh(acac)(CO)2]/L1/substrate I/substrate II = 1:3:100:100; c(Rh) =
2 mM, N,N-diisopropylethylamine (DIPEA, 1.5 equiv) was used as a
base for anionic substrate generation. For full experimental details, see
the Supporting Information.
greater part of 2a is consumed (and the anionic product formed
competes for the binding to the DIM binding site with
remaining substrate 2a (vide supra)), ester substrate 2c is
converted more quickly (Figure 4a). Kinetic analysis shows that
anion 2a reacts with the overall first-order kinetics (Figure
S20), despite the competition with ester 2c, as in the single
substrate experiments (vide supra). Interestingly, ester 2c reacts
at first with the seeming negative order kinetics that during the
10822
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
course of the reaction (conversion of 2a > 60%) switches to the
expected first-order kinetics (Figure S20). This indicates that
initially the preassociation of 2a with the DIM pocket of
Rh(L1) increases its “effective concentration” around the
catalytic metal center, outcompeting substrate 2c, but in the
final phase of the reaction this effect is attenuated due to
product inhibition (vide supra).
In contrast, the competition experiment with the shortest
substrate 1a, which is too short for bifunctional binding to
Rh(L1), and with its methyl ester 1c, shows that both
substrates react rather independently of each other (Figure 4b),
and both follow first-order kinetics (Figure S21). Interestingly,
in this case, ester 1c reacts a bit faster than anionic 1a.
Presumably, the latter being bound in the anion binding site
needs to dissociate prior to reacting on the metal center,
resulting in its lower reactivity. Furthermore, the competition
experiment between anionic substrates 1a and 2a shows similar
reactivity trends (Figure 5) with the first-order kinetics (Figure
in the competition experiment, as well as in the rate
enhancement for shorter substrate 2a as compared to 8a.
Taking all of these results together, these experiments reveal
the order of the events taking place on the catalyst. First, the
substrate molecule is bound via the anionic group to the
binding site of the catalyst (rather than the double bond
coordinates first). Next, if the CO dissociates from the rhodium
center, the double bond can coordinate to it and follow the
catalytic cycle (Scheme 4), which is finished by the product
release. The equilibrium between the free substrate and that
bound in the DIM pocket is fast compared to the alkene
conversion.
DFT Calculations and Origin of Selectivity. To gain
deeper understanding of the origin of the selectivity, we studied
the Rh(L1) catalytic system with DFT (BP86, SV(P)). Both in
situ IR and kinetic studies (vide supra) show that the rate-
determining step is early in the catalytic cycle, which in
combination with the absence of double bond isomerization
indicates that the regioselectivity for this catalytic system is
defined during insertion of the olefin into the Rh−H bond. We
additionally confirmed this by performing experiments using
D2/CO. Under these conditions, deuterium scrambling was not
observed (Figure S24), which indicates that the hydride
migration step is indeed irreversible.24,29 Consequently, this
selectivity determining hydride migration step was further
studied in detail.
We first calculated several possible structures of the
substrate−catalyst complex [RhH(CO)(L1)]−(2a) with differ-
ent geometries around the metal center.30 We found that the
eq−eq coordination geometry is preferred over the eq−ax
isomeric complex by 17.7 kJ/mol (I and V in Figure 6). The
optimal eq−eq complex structure shows that the carboxylate
group of the substrate is strongly bound in the DIM pocket of
the ligand with four hydrogen bonds (dN−O = 2.7−2.9 Å), and
the coordinated alkene moiety is tilted out of the P−Rh−P
plane of the trigonal bipyramidal rhodium complex (I in Figure
6). This perturbation results from the carboxylate moiety being
anchored in the binding site of the ligand. Importantly, the
anionic group binding severely restricts the movement of the
coordinated double bond. However, the double bond can easily
rotate toward the transition state, leading to the linear alkyl Rh
complex, and hence to the linear aldehyde product. In fact, the
geometry of the complex in the calculated early transition state
(ΔG°⧧ = 11.2 kJ mol−1) is almost unperturbed (the Rh−H
bond elongates by only 0.036 Å), with the alkene rotated only a
little further out of the equatorial plane (II in Figure 6).
Restrictions on the movement of the double bond imposed by
the bifunctional substrate binding block its rotation in the
direction necessary for the reaction pathway toward the
branched alkyl Rh;31 hence the branched aldehyde product
cannot be formed from this complex conformer. The
alternative conformer of the substrate−catalyst complex in
which the carbonyl and hydride positions are inverted (IV in
Figure 6), hence for which the favored rotation of the alkene
would direct the reaction toward the branched product, was
also evaluated. This conformation has a much higher energy
(15.8 kJ mol−1)32 that is even higher than the transition state
leading to the linear product from the former substrate−catalyst
conformer (ΔΔG° = 4.6 kJ mol−1). These calculations suggest
that the branched aldehyde product that is formed during the
reaction follows a pathway in which the anion moiety of the
substrate is not bound in the DIM pocket of the ligand (e.g.,
Figure 5. Competitive hydroformylation of a 1:1 mixture of substrates
1a and 2a by Rh(L1). Reagents and conditions: 20 bar CO/H2 (1:1),
CH2Cl2, 40 °C, [Rh(acac)(CO)2]/L1/1a/2a = 1:3:100:100; c(Rh) =
2 mM, triethylamine (TEA, 1.5 equiv) was used as a base for anionic
substrate generation. For full experimental details, see the Supporting
Information.
S22), demonstrating that both compete for the binding site
equally. However, substrate 2a reacts faster, because it can react
more easily on the metal center when its anionic group is
bound in the DIM pocket, while substrate 1a needs to
dissociate from the DIM binding site prior to reacting on the
metal center, lowering its reactivity. Additionally, the
competition experiment between the longest anionic substrate
8a and ester 1c shows that both substrates seem to react
independently of one another, both following first-order
kinetics (Figure S23). This reveals that, at the concentrations
used, the effective concentration of the long substrate 8a is
comparable to the actual concentration in solution.28 There-
fore, the substrate prebinding does not lead to the effective
competition with the nonbinding substrate 1c, and hence does
not inhibit the intermolecular hydroformylation. This confirms
that the effective concentration that depends inversely on the
cube of the linker length25 is of crucial importance in
determining the (relative) reaction rate, which is found both
10823
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
Figure 6. Calculated reaction pathway (DFT, BP86, SV(P)) of the
regioselectivity-determining hydride migration step in the hydro-
formylation of substrate 2a by the Rh(L1) catalyst. Notation: catalyst−
substrate complex I, transition state toward linear product II, and
linear alkyl Rh complex III, alternative structures of the catalyst−
substrate complex IV and V. G298: Gibbs free energy at 298 K (relative
to the catalyst-substrate complex I) in kJ mol−1. For full computational
details, see the Supporting Information.
Figure 7. Calculated reaction pathway (DFT, BP86, SV(P)) of the
regioselectivity-determining hydride migration step in the hydro-
formylation of substrate 3a (a) and 4a (b) by the Rh(L1) catalyst.
Notation: catalyst−substrate complex I, transition state toward linear
product II, and linear alkyl Rh complex III. G298: Gibbs free energy at
298 K (relative to the catalyst−substrate complex I) in kJ mol−1. For
full computational details, see the Supporting Information.
the anion binding site is occupied by another substrate
molecule or by the product formed, vide supra).
Calculations for reactions with longer substrates 3a and 4a
reveal similar trends for the hydride migration step (Figure 7).
The main difference is that the longer aliphatic linkers between
the double bond and the carboxylic moiety allow one to more
easily span the distance between the metal center and the anion
binding site, resulting in less perturbed coordination of the
alkene moiety in the equatorial plane of the Rh(L1) complex.
Thus, the double bond needs to rotate slightly further to reach
the transition state leading to the linear product. However, this
is still the privileged direction of the alkene rotation, due to the
restrictions imposed by the substrate anchoring in the DIM
pocket. Moreover, the lower perturbation of the alkene
coordination for longer substrates results in the greater
difference between the substrate−catalyst complex and the
decisive transition state. This in turn results in a higher energy
barrier. This, together with the change in the effective
concentration, explains why the longer anionic substrates
have a lower rate enhancement with respect to their ester
analogues in comparison to the substrate 2a that fits best (vide
supra).
reaction pathway that leads to the linear aldehyde and hinders
the competing pathway that would lead to the isomeric
product.
Regioselective Hydroformylation of Internally Unsa-
turated Aliphatic Acids. Selective hydroformylation of
internal alkenes is highly challenging, as it involves an internal
double bond with inherent lower reactivity. More forcing
conditions in turn lead to possible isomerization side reactions,2
which are deteriorating the selectivity.33 Moreover, to be
selective, the catalyst needs to precisely differentiate between
carbon atoms of the double bond whose electronic properties
are nearly identical. Importantly, analysis of the mechanism
controlling the regioselectivity for hydroformylation of terminal
olefins 1a−8a with the Rh(L1) catalyst (Figures 6 and 7)
allows one to postulate that the approach should be also
operative for substrates with internal double bonds. In
principle, one may expect that the restricted movement of
the reactive functionality should allow for selective introduction
of the aldehyde moiety on the carbon atom of the double bond,
which is more distant from the carboxylic group. Unfortunately,
our initial hydroformylation experiments proved that the
Rh(L1) catalyst is not active enough for substrates with
internal double bonds. Therefore, we next investigated the
optimization of the activity for this catalytic system by
modifying the DIMPhos ligand.
These DFT studies corroborate our assertion that the high
regioselectivity obtained for size-matching substrates with the
Rh(L1) catalyst originates from substrate preorientation
imposed by the hydrogen bonds between the anionic
functionality and the DIM pocket. This interaction highly
restricts the movement of the reactive double bond during the
decisive selectivity-determining step. Preorganization favors the
10824
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
Less basic phosphine and more π-accepting phosphite
ligands are known to afford more active rhodium complexes
for hydroformylation.34,35 We therefore designed and prepared
ligand L2, a close analogue of L1 functionalized with two
strongly electron-withdrawing trifluoromethyl groups on each
of the four phenyl rings (Figure 2). We also obtained ligand L3,
equipped with phosphite donor atoms (Figure 2), which are
known to lead to highly reactive hydroformylation catalysts.35a
To evaluate the influence of the modifications introduced,
ligands L2 and L3 were first studied in the hydroformylation of
anionic terminal olefins 1a−8a (under otherwise identical
conditions). In general, ligands L1−L3 allow for hydro-
formylation of all substrates 1a−8a with remarkable regiose-
lectivities, with the l/b ratios all above 38. The trends of relative
selectivity are somewhat different (Figure 8), and they cannot
substrate 2a, revealing TOFs of 24, 86, and 100 mol mol−1 h−1,
for L1, L2, and L3, respectively, proving the successful design
of new ligands.
With more active catalysts in hand, we evaluated the
hydroformylation of aliphatic carboxylates with internal double
bonds. Although more reactive than Rh(L1), the catalyst with
phosphine ligand L2 did not afford sufficient activity to convert
internal alkenes. Fortunately, the catalyst based on phosphite
L3 proved to be active toward these substrates, providing close
to full conversion for most of the reactions (Scheme 5 and
Scheme 5
Table 4). Interestingly, the catalyst is highly precise, presenting
unprecedented regioselectivities for the whole range of
substrates that differ in the positions of the reactive double
bond with respect to the carboxylic functionality and in size of
the substituents (Table 4). In addition, both E and Z isomers of
the substrates are converted with high selectivity. The analysis
of the isolated products shows that, as anticipated, in all cases
Figure 8. Hydroformylation of anionic substrates 1a−8a using Rh−
ligands L1−L3 catalysts. Full conversion in all cases for catalysts with
ligands L2,L3; for conversion with ligand L1, see Table 1. Conditions
are as described in the footnote to Table 1, with L2/Rh = L3/Rh =
1.1. For full experimental details, see the Supporting Information.
a
Table 4. Hydroformylation of Internal Alkenes 14−19
be fully rationalized at this point. Interestingly, when phosphite
ligand L3 was applied, also the shortest substrate 1a reacts with
very high selectivity (l/b = 38), in contrast to results for
catalysts with phosphine ligands L1, L2. To rationalize this, we
performed DFT studies for the Rh(L3)−(1a) complex. The
molecular modeling shows that the higher flexibility of this
phosphite-based ligand allows one to reduce the distance
between the metal center and the anion binding site such that
the Rh(L3) can adjust for bifunctional binding to substrate 1a
(Figure 9). Again, the double bond can rotate more easily in
one direction, leading to regioselective hydroformylation of the
bound substrate (Figures S22). The activities displayed by
catalysts with all DIMPhos ligands L1−L3 were further
compared quantitatively in hydroformylation of model
a
Reagents and conditions: [Rh(acac)(CO)2]/L3/substrate =
Figure 9. Calculated structure (DFT, BP86, SV(P)) of the catalyst−
substrate complex [RhH(CO)(L3)]−(1a). For reaction pathways of
the regioselectivity-determining hydride migration step in the
hydroformylation of substrate 1a by the Rh(L3) catalyst, see Figure
S26.
1:1.1:100; [Rh] = 2 mM, 20 bar CO/H2 (1:1), CH2Cl2, 40 °C, 72
h, triethylamine (TEA, 1.5 equiv) was used as a base for anionic
substrate generation. Regioselectivity and conversion (%) were
determined by H NMR analysis of the reaction mixture. For full
experimental details, see the Supporting Information.
1
10825
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
(2) (a) Cornils, B., Herrmann, W. A., Eds. Applied Homogeneous
Catalysis with Organometallic Compounds: A Comprehensive Handbook
in Three Volumes, 2nd ed.; Wiley-VCH: Weinheim, 2002. (b) Beller,
M., Bolm, C., Eds. Transition Metals for Organic Synthesis; Wiley-VCH:
Weinheim, 2004.
(3) (a) Hartwig, J. F. Organotransition Metal Chemistry: From Bonding
to Catalysis; University Science Books: Sausalito, 2009. (b) van
Leeuwen, P.W. N. M. Homogeneous Catalysis: Understanding the Art;
Kluwer Academic Publishers: Dordrecht, 2004.
the major product has the aldehyde functionality on the carbon
atom of the double bond more distant from the carboxylate
group (Table 4). The relative selectivity of addition of the
aldehyde group across the double bond is above 23 in all cases;
hence the major product is formed with regioselectivity above
95%!36 For comparison, typical catalysts provide close to
equimolar mixtures of alternative products. For example, for
Rh−PPh3 catalyst, the ratio between products is in the range
0.8 and 1.7 (under otherwise identical conditions).24
Importantly, for reactions with Rh(L3) at higher temperatures,
the regioselectivity is retained at a similar level, allowing for
greater catalyst activity, however, at the expense of some side/
consecutive reactions.24 These results further confirm the
general operational model of the Rh−DIMPhos catalysts.
(4) For reviews, see: (a) Goudriaan, E. P.; van Leeuwen, P.W. N. M.;
Birkholz, M.-N.; Reek, J. N. H. Eur. J. Inorg. Chem. 2008, 2939−2958.
(b) Breit, B. Pure Appl. Chem. 2008, 80, 855−860. (c) Reetz, M. T.
Angew. Chem., Int. Ed. 2008, 47, 2556−2588. (d) de Vries, J. G.; de
Vries, A. H. M. Eur. J. Org. Chem. 2003, 799−811. (e) Gennari, C.;
Piarulli, U. Chem. Rev. 2003, 103, 3071−3100. (f) Tang, W.; Zhang, X.
Chem. Rev. 2003, 103, 3029−3070. (g) Jakel, C.; Paciello, R. Chem.
̈
Rev. 2006, 106, 2912−2942. For supramolecular ligand strategies, see:
(h) Wilkinson, M. J.; van Leeuwen, P. W. N. M.; Reek, J. N. H. Org.
Biomol. Chem. 2005, 3, 2371−2383. (i) Breit, B. Angew. Chem., Int. Ed.
2005, 44, 6816−6825. (j) Sandee, A. J.; Reek, J. N. H. Dalton Trans.
2006, 3385−3391. (k) Meeuwissen, J.; Reek, J. N. H. Nat. Chem.
2010, 2, 615−621.
(5) (a) Whisler, M. C.; MacNeil, S.; Snieckus, V.; Beak, P. Angew.
Chem., Int. Ed. 2004, 43, 2206−2225. (b) Snieckus, V. Chem. Rev.
1990, 90, 879−933. (c) Reetz, M. T. Acc. Chem. Res. 1993, 26, 462−
468. (d) Engle, K. M.; Mei, T.-S.; Wasa, M.; Yu, J.-Q. Acc. Chem. Res.
2012, 45, 788−802. (e) Colby, D. A.; Tsai, A. S.; Bergman, R. G.;
Ellman, J. A. Acc. Chem. Res. 2012, 45, 814−825. (f) Daugulis, O.; Do,
H.-Q.; Shabashov, D. Acc. Chem. Res. 2009, 42, 1074−1086. (g) Chen,
X.; Engle, K. M.; Wang, D.-H.; Yu, J.-Q. Angew. Chem., Int. Ed. 2009,
48, 5094−5115.
(6) (a) Carboni, S.; Gennari, C.; Pignataro, L.; Piarulli, U. Dalton
Trans. 2011, 40, 4355−4373. (b) Dong, Z.; Luo, Q.; Liu, J. Chem. Soc.
Rev. 2012, 41, 7890−7908. (c) van Leeuwen, P. W. N. M.
Supramolecular Catalysis; Wiley-VCH: Weinheim, 2008. (d) Crabtree,
R. H. New J. Chem. 2011, 35, 18−23.
(7) (a) Yang, J.; Gabriele, B.; Belvedere, S.; Huang, Y.; Breslow, R. J.
Org. Chem. 2002, 67, 5057−5067. (b) Yang, J.; Breslow, R. Angew.
Chem., Int. Ed. 2000, 39, 2692−2694. (c) Breslow, R.; Zhang, X.;
Huang, Y. J. Am. Chem. Soc. 1997, 119, 4535−4536. (d) Breslow, R.;
Zhang, X.; Xu, R.; Maletic, M.; Merger, R. J. Am. Chem. Soc. 1996, 118,
11678−11679. (e) Fang, Z.; Breslow, R. Org. Lett. 2006, 8, 251−254.
(f) Breslow, R.; Brown, A. B.; McCullough, R. D.; White, P. W. J. Am.
Chem. Soc. 1989, 111, 4517−4518.
CONCLUSION
■
In summary, we have reported a series of DIMPhos ligands
L1−L3, which are bidentate phosphorus ligands equipped with
an integral anion binding site (the DIM pocket). We have
shown that these bifunctional ligands form well-defined
rhodium complexes that can bind anionic species in the
binding site of the ligand. This interaction can be used to
preorganize a substrate molecule, that is, an alkene with an
anionic group that can be remote from the reactive double
bond (even 10 bonds!), which leads to its highly selective
hydroformylation. For the hydroformylation of substrates with
internal double bonds, the current system gives the highest
selectivity reported in the literature,37 clearly demonstrating the
power of supramolecular control of the selectivity for catalysis.
Importantly, the mode of operation is well understood by the
detailed studies provided in this Article. This enables rational
design of selective catalysts for desired reactions, clearly
complementing trial-and-error approaches in the field of
transition metal catalysis. In principle, it should be possible to
transmit the current system to other transition metal-catalyzed
processes involving a migration in the selectivity-determining
step, giving rise to other selective transformations in chemical
catalysis. Research along these lines is continued in our
laboratories.
ASSOCIATED CONTENT
* Supporting Information
Details concerning materials and methods, catalytic and kinetic
studies, coordination and binding studies, DFT studies,
experimental procedures, and spectral data for new compounds
(including images of NMR spectra). This material is available
free of charge via the Internet at http://pubs.acs.org.
(8) For reviews, see: (a) Taylor, M. S.; Jacobsen, E. N. Angew. Chem.,
Int. Ed. 2006, 45, 1520−1543. (b) Sohtome, Y.; Nagasawa, K. Chem.
Commun. 2012, 48, 7777−7789. For selected examples, see:
(c) Zuend, S. J.; Coughlin, M. P.; Lalonde, M. P.; Jacobsen, E. N.
Nature 2009, 461, 968−970. (d) Xu, H.; Zuend, S. J.; Woll, M. G.;
Tao, Y.; Jacobsen, E. N. Science 2010, 327, 986−990. (e) Singh, R. P.;
Foxman, B. M.; Deng, L. J. Am. Chem. Soc. 2010, 132, 9558−9560.
(f) Klauber, E. G.; De, C. K.; Shah, T. K.; Seidel, D. J. Am. Chem. Soc.
2010, 132, 13624−13626. (g) Wang, Y.; Yu, T.-Y.; Zhang, H.-B.; Luo,
Y.-C.; Xu, P.-F. Angew. Chem., Int. Ed. 2012, 51, 12339−12342. (h) Lu,
Y.; Johnstone, T. C.; Arndtsen, B. A. J. Am. Chem. Soc. 2009, 131,
■
S
AUTHOR INFORMATION
Corresponding Author
j.n.h.reek@uva.nl
Notes
■
11284−11285. (i) Bauer, A.; Westkamper, F.; Grimme, S.; Bach, T.
̈
Nature 2005, 436, 1139−1140. (j) Muller, C.; Bauer, A.; Bach, T.
̈
Angew. Chem., Int. Ed. 2009, 48, 6640−6642. (k) Muller, C.; Maturi,
̈
The authors declare no competing financial interest.
M. M.; Bauer, A.; Cuquerella, M. C.; Miranda, M. A.; Bach, T. J. Am.
Chem. Soc. 2011, 133, 16689−16697. (l) Wiegand, C.; Herdtweck, E.;
Bach, T. Chem. Commun. 2012, 48, 10195−10197.
ACKNOWLEDGMENTS
■
(9) (a) Grotjahn, D. B.; Miranda-Soto, V.; Kragulj, E. J.; Lev, D. A.;
Erdogan, G.; Zeng, X.; Cooksy, A. L. J. Am. Chem. Soc. 2008, 130, 20−
21. (b) Grotjahn, D. B.; Kragulj, E. J.; Zeinalipour-Yazdi, C. D.;
Miranda-Soto, V.; Lev, D. A.; Cooksy, A. L. J. Am. Chem. Soc. 2008,
130, 10860−10861.
We kindly acknowledge the NRSC-C for financial support,
Prof. Dr. Bas de Bruin and Dr. J. I. van der Vlugt for helpful
discussions, and Dr. W. I. Dzik and Dr. T. Gadzikwa for critical
comments and valuable suggestions.
(10) (a) Das, S.; Incarvito, C. D.; Crabtree, R. H.; Brudvig, G. W.
Science 2006, 312, 1941−1943. (b) Das, S.; Brudvig, G. W.; Crabtree,
R. H. J. Am. Chem. Soc. 2008, 130, 1628−1637. (c) Hull, J. F.; Sauer, E.
REFERENCES
(1) Beller, M. Chem. Soc. Rev. 2011, 40, 4891−4892.
■
10826
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
L. O.; Incarvito, C. D.; Faller, J. W.; Brudvig, G. W.; Crabtree, R. H.
Inorg. Chem. 2009, 48, 488−495.
Goubitz, K.; Fraanje, J.; Schenk, H.; Bo, C. J. Am. Chem. Soc. 1998,
120, 11616−11626. (b) Kamer, P. C. J.; van Rooy, A.; Schoemaker, G.
C.; van Leeuwen, P. W. N. M. Coord. Chem. Rev. 2004, 248, 2409−
2424. (c) Kubis, C.; Ludwig, R.; Sawall, M.; Neymeyr, K.; Borner, A.;
Wiese, K. D.; Hess, D.; Franke, R.; Selent, D. ChemCatChem 2010, 2,
287−295. (d) Diebolt, O.; van Leeuwen, P. W. N. M.; Kamer, P. C. J.
ACS Catal. 2012, 2, 2357−2370.
(11) (a) Smejkal, T.; Gribkov, D.; Geier, J.; Keller, M.; Breit, B.
Chem.-Eur. J. 2010, 16, 2470−2478. (b) Smejkal, T.; Breit, B. Angew.
Chem., Int. Ed. 2008, 47, 311−315.
(12) (a) Fackler, P.; Berthold, C.; Voss, F.; Bach, T. J. Am. Chem. Soc.
2010, 132, 15911−15913. (b) Fackler, P.; Huber, S. M.; Bach, T. J.
Am. Chem. Soc. 2012, 134, 12869−12878.
(24) For details, see the Supporting Information.
(13) (a) Dzik, W. I.; Xu, X.; Zhang, X. P.; Reek, J. N. H.; de Bruin, B.
J. Am. Chem. Soc. 2010, 132, 10891−10902. (b) Zhu, S.; Ruppel, J. V.;
Lu, H.; Wojtas, L.; Zhang, X. P. J. Am. Chem. Soc. 2008, 130, 5042−
5043.
(14) (a) Breuil, P.-A. R.; Patureau, F. W.; Reek, J. N. H. Angew.
Chem., Int. Ed. 2009, 48, 2162−2165. (b) Breuil, P.-A. R.; Reek, J. N.
H. Eur. J. Org. Chem. 2009, 6225−6230. (c) Dydio, P.; Rubay, C.;
Gadzikwa, T.; Lutz, M.; Reek, J. N. H. J. Am. Chem. Soc. 2011, 133,
17176−17179.
(25) For the description of effective concentrations and of
multivalency in supramolecular chemistry, see: Mulder, A.; Huskens,
J.; Reinhoudt, D. N. Org. Biomol. Chem. 2004, 2, 3409−3424.
(26) Blackmond, D. G. Angew. Chem., Int. Ed. 2005, 44, 4302−4320.
(27) Copeland, R. A. Enzymes: A Practical Introduction to Structure,
Mechanism, and Data Analysis; Wiley-VCH: New York, 2000.
(28) The effective concentration for substrate 8a bound to the DIM
pocket of the catalyst can be roughly estimated: (i) for the initial
concentrations used and the association constant of Ka > 105 M−1, the
DIM pocket is nearly fully occupied by the substrate (>99.995%); (ii)
the maximal distance between the Rh center and the alkene in the
supramolecular complexes is about 15 Å; thus the “probing volume” is
about 1.7 × 10−24 dm3, which translates to the effective concentration
Ceff ≈ 0.9 M. With this rough estimate, one can understand that the
nonbound substrate present in 0.2 M will considerably compete with
the bound substrate. For comparison, for substrate 2a, with the
maximal Rh−alkene distance of 8Å, the effective concentration is
estimated to Ceff ≈ 6 M, which is substantially higher than the actual
alkene concentration in solution (0.2 M), hence allowing for the
effective competition. For a detailed discussion on the influence of the
effective concentrations, see ref 25.
(15) (a) Sessler, J. L.; Gale, P. A.; Cho, W.-S. Anion Receptor
Chemistry; Royal Society of Chemistry: Cambridge, 2006. (b) Dydio,
P.; Lichosyt, D.; Jurczak, J. Chem. Soc. Rev. 2011, 40, 2971−2985.
(c) Wenzel, M.; Hiscock, J. R.; Gale, P. A. Chem. Soc. Rev. 2012, 41,
480−520.
(16) (a) Dydio, P.; Zielin
4560−4562. (b) Dydio, P.; Zielin
1076−1078. (c) Jurczak, J.; Chmielewski, M. J.; Dydio, P.; Lichosyt,
D.; Ulatowski, F.; Zielin
́
́
́
ski, T. Pure Appl. Chem. 2011, 83, 1543−1554.
́
(d) Dydio, P.; Lichosyt, D.; Zielinski, T.; Jurczak, J. Chem.-Eur. J. 2012,
18, 13686−13701.
(17) (a) van Leeuwen, P. W. N. M.; Claver, C. Rhodium Catalyzed
Hydroformylation; Kluwer Academic Publishers: Dordrecht, 2000.
(29) Horiuchi, T.; Shirakawa, E.; Nozaki, K.; Takaya, H. Organo-
metallics 1997, 16, 2981−2986.
(b) Franke, R.; Selent, D.; Borner, A. Chem. Rev. 2012, 112, 5675−
̈
5732. (c) Breit, B.; Seiche, W. Synthesis 2001, 2001, 1−36.
(30) We started a search for possible conformations of the substrate−
catalyst complex for both eq−eq and eq−ax coordination geometries,
using a simplified model with a ligand in which the 4 phenyl rings were
removed from the phosphine. Subsequently, the structures lowest in
energy were supplemented with the phenyl rings and optimized again.
(The geometries with much higher energies, >16 kJ mol−1, were
omitted.) For full computational details, see the Supporting
Information.
(18) For regioselective hydroformylation of unfunctionalized internal
alkenes, see: (a) Kuil, M.; Soltner, T.; van Leeuwen, P. W. N. M.;
Reek, J. N. H. J. Am. Chem. Soc. 2006, 128, 11344−11345. For regio-
and enantioselective hydroformylation of unfunctionalized internal
alkenes, see: (b) Bellini, R.; Chikkali, S. H.; Berthon-Gelloz, G.; Reek,
J. N. H. Angew. Chem., Int. Ed. 2011, 50, 7342−7345. (c) Gadzikwa,
T.; Bellini, R.; Dekker, H. L.; Reek, J. N. H. J. Am. Chem. Soc. 2012,
134, 2860−2863. (d) Noonan, G. M.; Fuentes, J. A.; Cobley, C. J.;
Clarke, M. L. Angew. Chem., Int. Ed. 2012, 51, 2477−2480.
(31) Despite many attempts, we were not able to find a transition
state for the formation of the branched alkyl complex from this
substrate−catalyst complex conformer.
(19) For regioselective hydroformylation of alkenes, using reversible
covalent bonding of catalytic ligand-like directing groups, see:
(32) The complex structure is distorted from the ideal coordination
geometry (a trigonal bipyramid or a square pyramid) due to the
geometrical constrains imposed by the ligand and by the substrate
bound to the phosphorus ligand (the virtually tridentate L1−2a
ligand). Consequently, the two axial positions occupied by the hydride
and CO are electronically different. Therefore, changing their positions
leads to a complex with different energy, and also modifies the
coordination geometry around the rhodium center. As indicated by the
change of the τ parameter value from 0.39 to 0.27, for geometries I
and IV, respectively, the inversion of the CO and H ligands pushes the
geometry of the metal center to a structure closer to a square pyramid
and results in its destabilization. (The τ value indicates the idealized
square pyramid with τ = 0, and the trigonal bipyramid with τ = 1). For
the description of the τ parameter, see: Addison, A. W.; Rao, T. N.;
Reedijk, J.; van Rijn, J.; Verschoor, G. C. J. Chem. Soc., Dalton Trans.
1984, 1349−1356.
(33) The isomerization of the internal double bond to the chain
terminus, followed by its hydroformylation, gives access to linear
products, see: (a) van der Veen, L. A.; Kamer, P. C. J.; van Leeuwen, P.
W. N. M. Angew. Chem., Int. Ed. 1999, 38, 336−338. (b) Klein, H.;
Jackstell, R.; Wiese, K.-D.; Borgmann, C.; Beller, M. Angew. Chem., Int.
Ed. 2001, 40, 3408−3411. (c) Seayad, A.; Ahmed, M.; Klein, H.;
Jackstell, R.; Gross, T.; Beller, M. Science 2002, 297, 1676−1678.
(d) Bronger, R. P. J; Kamer, P. C. J.; van Leeuwen, P. W. N. M.
Organometallics 2003, 22, 5358−5369. (e) Yu, S.; Chie, Y.-M.; Guan,
Z.-h.; Zhang, X. Org. Lett. 2008, 10, 3469−3472. (f) Selent, D.; Franke,
(a) Grunanger, C. U.; Breit, B. Angew. Chem., Int. Ed. 2008, 47,
̈
7346−7349. (b) Grunanger, C. U.; Breit, B. Angew. Chem., Int. Ed.
̈
2010, 49, 967−970. (c) Lightburn, T. E.; Dombrowski, M. T.; Tan, K.
L. J. Am. Chem. Soc. 2008, 130, 9210−9211. (d) Worthy, A. D.;
Gagnon, M. M.; Dombrowski, M. T.; Tan, K. L. Org. Lett. 2009, 11,
2764−2767. (e) Sun, X.; Frimpong, K.; Tan, K. L. J. Am. Chem. Soc.
2010, 132, 11841−11843. (f) Lightburn, T. E.; De Paolis, O. A.;
Cheng, K. H.; Tan, K. L. Org. Lett. 2011, 13, 2686−2689. For a review
on removable directing groups in organic synthesis and catalysis, see:
(g) Rousseau, G.; Breit, B. Angew. Chem., Int. Ed. 2011, 50, 2450−2494
and references therein.
(20) Preliminary results of this study have been communicated:
(a) Dydio, P.; Dzik, W. I.; Lutz, M.; de Bruin, B.; Reek, J. H. N. Angew.
Chem., Int. Ed. 2011, 50, 396−400. For recent application of ligand L3
in β-selectivity hydroformylation of vinyl arenes, see: (b) Dydio, P.;
Reek, J. H. N. Angew. Chem., Int. Ed. 2013, 52, 3878−3882.
(21) For low-temperature NMR studies of a mixture of eq−eq and
eq−ax complexes, see: (a) Molina, D. A. C.; Casey, C. P.; Muller, I.;
̈
Nozaki, K.; Jakel, C. Organometallics 2010, 29, 3362−3367. (b) Gual,
̈
A.; Godard, C.; Castillon, S.; Claver, C. Adv. Synth. Catal. 2010, 352,
463−477.
(22) van der Veen, L. A.; Keeven, P. H.; Schoemaker, G. C.; Reek, J.
N. H.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Lutz, M.; Spek, A.
L. Organometallics 2000, 19, 872−883.
(23) For high-pressure IR studies, see: (a) van der Veen, L. A.; Boele,
M. D. K.; Bregman, F. R.; Kamer, P. C. J.; van Leeuwen, P.W. N. M.;
10827
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828

Journal of the American Chemical Society
Article
R.; Kubis, C.; Spannenberg, A.; Baumann, W.; Kreidler, B.; Borner, A.
̈
Organometallics 2011, 30, 4509−4514.
(34) (a) van der Veen, L. A.; Boele, M. D. K.; Bregman, F. R.; Kamer,
P. C. J.; van Leeuwen, P. W. N. M.; Goubitz, K.; Fraanje, J.; Schenk,
H.; Bo, C. J. Am. Chem. Soc. 1998, 120, 11616−11626. (b) Casey, C.
P.; Paulsen, E. L.; Buettenmueller, E. W.; Proft, B. R.; Petrovich, L. M.;
Matter, B. A.; Powell, D. R. J. Am. Chem. Soc. 1997, 119, 11817−
11825.
(35) (a) van Leeuwen, P. W. N. M.; Kamer, P. C. J.; Claver, C.;
Pam
̀
ies, O.; Dieg
́
uez, M. Chem. Rev. 2011, 111, 2077−2118. (b) Gual,
A.; Godard, C.; Castillon
21, 1135−1146 and references therein.
́
, S.; Claver, C. Tetrahedron: Asymmetry 2010,
(36) Although both ligand L3 and the aldehyde products are chiral,
there was no enantioselectivity observed in these reactions.
(37) Breit and co-workers reported selective hydroformylation of
(Z)-pent-3-enoic acid, with relative regioselectivity of 18.1:1; see ref
11.
10828
dx.doi.org/10.1021/ja4046235 | J. Am. Chem. Soc. 2013, 135, 10817−10828