Polycationic ligands in gold catalysis: Synthesis and applications of extremely π-acidic catalysts

Article abstract of DOI:10.1021/ja411146x

Very often ligands are anionic or neutral species. Cationic ones are rare, and, when used, the positively charged groups are normally appended to the periphery of the ligand. Here, we describe a dicationic phosphine with no spacer between the phosphorus atom and the two positively charged groups. This structural feature makes its donor ability poorer than that of phosphites and only comparable to extremely toxic or pyrophoric compounds such as PF ₃ or P(CF₃)₃. By exploiting these properties, a new Au catalyst has been developed displaying a dramatically enhanced capacity to activate π-systems. This has been used to synthesize very sterically hindered and naturally occurring 4,5-disubstituted phenanthrenes. The present approach is expected to be applicable to the development and improvement of many other transition metal catalyzed transformations that benefit from extremely strong π-acceptor ligands. The mechanism of selected catalytic transformations has been explored by density functional calculations.

Full text of DOI:10.1021/ja411146x

Article
pubs.acs.org/JACS
Polycationic Ligands in Gold Catalysis: Synthesis and Applications of
Extremely π‑Acidic Catalysts
Javier Carreras, Gopinadhanpillai Gopakumar, Liangu Gu, Ana Gimeno, Pawel Linowski,
Jekaterina Petuskova, Walter Thiel, and Manuel Alcarazo*
̌
Max-Planck-Institut fur Kohlenforschung, Kaiser Wilhelm Platz 1, Mulheim an der Ruhr, 45470, Germany
̈
̈
S
* Supporting Information
ABSTRACT: Very often ligands are anionic or neutral species. Cationic
ones are rare, and, when used, the positively charged groups are normally
appended to the periphery of the ligand. Here, we describe a dicationic
phosphine with no spacer between the phosphorus atom and the two
positively charged groups. This structural feature makes its donor ability
poorer than that of phosphites and only comparable to extremely toxic or
pyrophoric compounds such as PF3 or P(CF3)3. By exploiting these
properties, a new Au catalyst has been developed displaying a dramatically
enhanced capacity to activate π-systems. This has been used to synthesize
very sterically hindered and naturally occurring 4,5-disubstituted
phenanthrenes. The present approach is expected to be applicable to
the development and improvement of many other transition metal
catalyzed transformations that benefit from extremely strong π-acceptor
ligands. The mechanism of selected catalytic transformations has been explored by density functional calculations.
As part of our program devoted to the development of more
amenable extreme electron-poor phosphines with potential
applications in catalysis, we envisaged a new strategy based on
the introduction of positively charged homo- or heteroaromatic
moieties directly attached to the central phosphorus atom.6
This article discloses the successful implementation of this
concept through the synthesis of (dialkylamino)-
cyclopropenium substituted phosphines. Specifically, when
two of these three-member ring substituents are introduced,
the donor ability of the resulting ligand 1a bench stable
crystalline solidperfectly matches that of P(CF3)3. Taking
advantage of these properties, we also develop herein a new
gold catalyst 6 able to transform in excellent yields ortho-biaryl
substituted alkynes into bent phenantrenes containing sub-
stituents in both internal positions, 4 and 5. We also
demonstrate the impact of this catalyst on the synthesis of
phenanthrene-derived natural occurring products such as
Bulbophylantrin, Marylaurencinol A, Ochrolide, and Coelogi-
nin. Finally, we report density functional theory (DFT)
calculations to analyze the electronic structure of the catalyst
and to elucidate the mechanism of a prototypical cyclo-
isomerization reaction catalyzed by 6.
1. INTRODUCTION
Phosphines are arguably the most important ancillary ligands in
organometallic chemistry probably due to the easy modification
of their electronic and steric properties in a predictable manner.
Specifically, in homogeneous catalysis this tuning is a key
attribute that allows a remarkable degree of control over the
outcome of transition metal promoted transformations. Just to
mention some examples, bulky phosphines are known to
stabilize low-coordinated intermediates,1 electron-rich ones
facilitate oxidative additions,2 and the employment of strong
acceptor ligands favors the coordination of substrates or other
coligands. Unfortunately, while steric hindrance can be
modified quite independently, the tuning of electron properties
is more subtle and any change in the σ-donation ability of a
ligand is coupled with a variation of its π-acceptor properties,
normally, in an opposite direction.3
Many electron-rich phosphines with different steric require-
ments are available. If moderate π-acceptor ligands are
necessary, phosphites or polyhalogenated phosphines can be
still employed, but there are only a few phosphorus-based
ligands that are feeble σ-donor and strong π-acceptor ligands. In
fact, only PF3, P(CF3)3, and PCl3 can be considered members
of this category. However, despite this potential, their
coordination chemistry is unexploited4 for obvious reasons:
PF3 is a very poisonous gas, difficult to prepare and to handle,
P(CF3)3 is a liquid with a quite low boiling point (17.3 °C, 760
mm) that burns in contact with air, and PCl3 is a very irritant
liquid readily prone to hydrolysis.5 Moreover, the metal
complexes resulting from their coordination are often as
reactive or even more so than the corresponding free ligands.
2. RESULTS AND DISCUSSION
2.1. Bis[(diisopropylamino)cyclopropenium]-Substi-
tuted Phosphines. In phosphines, the nonshared electron
pair at phosphorus accounts for the σ-donor ability while the
Received: July 9, 2013
Published: December 5, 2013
© 2013 American Chemical Society
18815
dx.doi.org/10.1021/ja411146x | J. Am. Chem. Soc. 2013, 135, 18815−18823

Journal of the American Chemical Society
Article
σ*(P−C) orbital accepts electron density from the metal
through π-backdonation. In a first approach, introduction of a
positively charged substituent directly on phosphorus lowers
the energy of both orbitals by the inductive effect.
Consequently, the resulting phosphine must be a worse σ-
donor but a better π-acceptor ligand. This rationale is correct in
general lines but still incomplete. If the positively charged
group is an aromatic system, it will contain low-lying π* orbitals
that can interact with the lone pair at phosphorus and thereby
change its shape. This can be viewed as a back-donation from
phosphorus to its substituents, which will reduce the overlap
with the orbitals of the metal (M) and weaken or, in an extreme
case, even cancel the σ component of the P-M bond. This
scenario explains why all monocationic phosphines that are
known to form complexes with metals depict phosphite-type
behavior.7 Obviously, when two or more -onium substituents
are attached to the central P-atom, the resulting polycationic
phosphines should be even poorer donating ligands. Probably
for this reason these compounds rarely coordinate to metals. In
fact, the only two complexes reported to date containing
polycationic ligands share the same [L→PtCl3]n+ (n = 1,2)
structure and their stability presumably relies on the exquisite
π-back-donation capacity of the anionic Pt-moiety which is able
to overcompensate the feeble σ-bond component.6b,8 However,
for a general impact on catalysis, polycationic ligands able to
coordinate a broader metal panoply are required.
Thus, it seems that at least two -onium substituents will be
necessary to engender very strong π-acceptor properties in a
phosphine, but importantly, not all positively charged
substituents will be equally appropriate. Those endowed with
the highest lowest unoccupied molecular orbital (LUMO)
should minimize the undesired (onium-π*)-(P-lone pair)
overlap and should therefore be most prone to coordinate an
array of transition metals. For these reasons, the bis-
(diisopropylaminocyclopropenium)-substituted phosphine 1
was selected as the most promising target ligand.9 This
compound could be prepared in a two-step sequence: First,
condensation of the readily available chlorocyclopropenium
tetrafluoroborate 2 with phenylphosphine 3 in refluxing
tetrahydrofuran afforded salt 4 in 76% yield. Deprotonation
of 4 with a strong base such as potassium hexamethyldisilazide
at −30 °C and in situ trapping of the phosphalkene
intermediate with a second equivalent of 2 finally yielded 1
as an air and water stable white solid (Figure 1a). Its dicationic
character was reflected in the shielding of the P center (31P
NMR resonance at δ = −48.3 ppm) comparable to that in a
dicationic bis(imidazolium) substituted sister compound (δ =
−50.8 ppm).10 Subsequently, single crystal X-ray diffraction
analysis unambiguously confirmed the proposed connectivity
for 1 (Figure 1b).
Figure 1. (a) Synthesis, structure, and complexation of 1. (b) X-ray
crystal structures of 1 and 5 in the solid state; hydrogen atoms and
tetrafluoroborate counteranions are omitted for clarity.
synthesis of complex organic molecules through the promotion
of structural rearrangements.13 The low-lying LUMO and the
poor back-donation capacity of cationic Au species account for
their high efficiency in promoting these processes. It seems
then reasonable to assume that this natural π-acidity of Au
catalysts can be significantly enhanced by the introduction of
strong π-acceptor ancillary ligands such as 1.
Recently, we developed an efficient protocol for the
previously known Pt(II)-catalyzed 6-endo-dig cyclization of 2-
ethynyl-1,1′-biphenyls into polysubstituted phenanthrenes that
dramatically expanded its scope.6a,14 Unfortunately, when using
biphenyls precursors of 4 and/or 5 substituted phenanthrenes,
the yields of the desired products dropped significantly, or no
cycloisomerization could be detected. The torsion imposed
onto the phenanthrene system by the unfavorable steric
interactions between the two internal substituents is probably
responsible for the difficulties attending the synthesis of these
seemly simple molecules. Hence, an even stronger activation of
the alkyne moiety by a metal catalyst appears necessary to
overcome this limitation and thus provide a useful entry for the
synthesis of 4,5-disubstituted phenanthrenes. This route will
intrinsically lead to the formation of only the desired
compound, in contrast with the established methodology
based on the irradiation of stilbenes in the presence of an
external oxidant that affords mixtures of positional isomers.15
In an attempt to address this issue, we focused on the
performance of precatalysts 6 containing the more π-acid Au(I)
center and the extremely strong π-acceptor ligand 1. Initial
trials were carried out using 2-ethynyl-2′,6-dimethylbiphenyl 7
as model substrate (Figure 3a). After screening the reaction
using a series of five solvents (THF, MTBE, toluene,
acetonitrile, and dichloromethane), different catalyst loadings,
and silver salts for the abstraction of the chloride moiety of 6,
We were pleased to observe that 1 was not only able to react
with K2PtCl4 to give complex 5, but also with neutral metal
fragments such as AuCl to afford compound 6. Encouraged by
this distinctive reactivity, the Tolman cone angle (θ) and the
electronic parameter (TEP) of 1 were determined. From the X-
ray structure of 5, θ11 was calculated to be 183° while the
TEP12 was estimated to be 2101 cm−1. This value reveals the
tremendous influence of the two cyclopropenium substituents
on the donor capacity of 1. In fact, 1 can be ranked as weaker
donor than any phosphite ligand; only P(CF3)3 depicts similar
donor properties (Figure 2).
2.2. Catalysis. The electrophilic activation of alkynes by
gold species has emerged as an extremely powerful tool for the
18816
dx.doi.org/10.1021/ja411146x | J. Am. Chem. Soc. 2013, 135, 18815−18823

Journal of the American Chemical Society
Article
Figure 2. Tolman stereoelectronic map for P-based ligands. Comparison with other phosphines and phosphites highlights the exceptional π-acceptor
properties of 1 (shown in blue).
Figure 3. Ligand effect on the Au-catalyzed cyclization of 2-ethynylbiphenyls into phenanthrenes. (a) Plot of conversion versus time in the reaction
of 2-ethynyl-6,2′-dimethylbiphenyl into 4,5-dimethylphenanthrene. Conditions: 7 (0.05 M), Au precatalysts 2 mol %, AgSbF6 2 mol %, CH2Cl2,
room temperature. Conversions were determined by GC. (b) Effect of phosphine ligands with different donor−acceptor properties on the Au-
catalyzed cyclization of biphenyl 9. Product ratios and yields measured after complete consumption of the starting material. Isolated yields of the 10/
11 mixtures: 92%, (6); 89%, (PhO)3P; 93%, Ph3P; 90%, Me3P.
we found optimum conditions (2 mol % of 6 and AgSbF6,
CH2Cl2, room temperature), under which the desired 4,5-
dimethylphenantrene 8 was obtained in excellent yield (97%)
after only 10 min. For comparison, Figure 3a also depicts the
conversion-versus-time plot for the archetypal Au precatalysts
Ph3PAuCl (red) and (PhO)3PAuCl (green) under otherwise
identical conditions. The vastly superior behavior of 6
demonstrates the remarkable ability of 1 to increase the π-
acidity of Au (see the Computational Section for a detailed
theoretical analysis). In addition, the advantages derived from
the employment of 6 as precatalyst are not only of kinetic
nature. Figure 3b shows the product distribution when 9 is
employed as substrate in the presence of different Au
precatalysts. Whereas Me3PAuCl, Ph3PAuCl or (PhO)3PAuCl
lead to the formation of cycloheptatriene 11 as the major
product through a parasite 7-exo-dig process, 6 furnished the
desired 4-phenylphenanthrene 10 with excellent selectivity.
This switch from an originally preferred pathway to another
18817
dx.doi.org/10.1021/ja411146x | J. Am. Chem. Soc. 2013, 135, 18815−18823

Journal of the American Chemical Society
Article
Table 1. Scope of the Au-Catalyzed Cycloisomerization of Ortho-Alkynylated Biaryls Employing Catalyst 6
*
†
Conditions: biphenyl (0.05M), 6 (2 mol %), AgSbF6 (2 mol %) CH2Cl2, room temperature. Isolated yield. Four products were detected by GC−
MS analysis of the reaction crude (4-methyl-8-isopropylphenanthrene 36, 66%; 4-methyl-5-isopropylphenanthrene 37, 22%; 4-methyl-6-
isopropylphenanthrene 38, 3%; 4-methylphenanthrene 26, 7%). Preparative HPLC allowed the isolation of the four compounds and subsequent
NMR analysis. The structure of the major isomer, 36, was also confirmed by X-ray difraction analysis of a monocrystal.
one depending on the π-acidity of the catalysts is consistent
with the following mechanistic interpretation.16 The reaction is
thought to commence with a nucleophilic attack of the arene
ring on the activated alkyne. A strong activation of the alkyne
by the catalysts will give rise to a rather early transition state
where steric issues are not crucial; hence, the 6-endo-dig
cyclization takes place almost exclusively. In the case of weaker
π-acid catalysts, a closer approach of the reacting fragments is
needed to reach the transition state. Sterics may then play an
important role and increase the barrier to the 6-endo-dig
cyclization so that the competitive 7-exo-dig process gets
favored.
Having identified the optimized reaction conditions, further
experiments were conducted to explore the generality of this
cycloisomerization protocol regarding functional group toler-
ance and the size of the groups that could be introduced. We
18818
dx.doi.org/10.1021/ja411146x | J. Am. Chem. Soc. 2013, 135, 18815−18823

Journal of the American Chemical Society
Article
first prepared a collection of 2-ethynylbiphenyls 9−14 only
substituted at position 2′ (see the Supporting Information) and
performed a reactivity screen. The results were highly
encouraging as all the substrates were completely converted
to the desired phenanthrenes 23−28 in good to excellent yields
(Table 1, entries 1−6). Interestingly, methyl-, phenyl-,
methoxy-, iso-propyl-, and even tert-butyl-substituents in
position 4 were found to be compatible with the cyclization
process. Moreover, the presence of free −OH groups was
tolerated, although competitive coordination of the oxygen
atom to Au could have been an issue.
With these results in hand, we then focused our attention on
the reactivity of biphenyl substrates that simultaneously bear
substituents at positions 2′ and 6. To our delight, full
conversions and high isolated yields of the desired phenan-
threnes were also obtained after short reaction times (Table 1,
entries 7−14). X-ray structure analysis of compounds 8, 30, and
35 further demonstrates the exquisite ability of 6 to accomplish
the desired cyclizations encompassing sterically demanding
substrates. The clash between substituents in positions 4 and 5
generates destabilizing twists of up to 31.7° within the
phenanthrene moiety (measured through the C4−C4a−C5a−
C5 torsion) but nonetheless, the desired ring closure takes
place cleanly (Figure 4).
products, possibly through migration of the isopropyl moiety.
Efforts directed toward the understanding of this phenomenon
are currently underway.
2.3. Synthetic Applications. A large number of naturally
occurring polyoxygenated phenanthrenes with different sub-
stitution patterns have been isolated from plants belonging to
the Orchidaceae family.17 They exhibit interesting biological
activities as antiinflammatory, antiallergic, antimicrobial, anti-
fungal, and cytotoxic reagents. In view of their relevance, we
decided to explore the utility of our methodology for the
synthesis of some representative examples. Specifically, we
focused on the preparation of the synthetically more
challenging 4,5-disubstituted derivatives.
As first targets, we chose Bulbophylantrin 39,18 a diphenol
isolated from the orchid Bulbophyllum leopardium and its 9,10-
dihydro- derivative Marylaurencinol A 40,19 recently obtained
from Cymbidium Great Flower Marie Laurencin. Our total
synthesis of these products started from the already described
2-bromo-3-hydroxy-benzaldehyde 4120 which was cross-
coupled with the benzyl-protected boronic acid 4221 to form
the corresponding biphenyl derivative 43. Treatment with
Ohira−Bestmann reagent22 generated the desired alkyne 44
that was subsequently submitted to the key Au-catalyzed
cycloisomerization. In line with our expectations, the desired
phenanthrene 45 was obtained in excellent yield after a few
minutes. Removal of the benzyl group by hydrogenolytic
cleavage in ethyl acetate cleanly afforded Bulbophyllantrin 39.
Hydrogenation in a more polar solvent such as methanol not
only caused cleavage of the benzyl ether but also reduction of
the phenanthrene skeleton at the more activated 9- and 10-
positions giving Marylaurencinol A. This material crystallized
quite readily and X-ray analysis could be performed confirming
the expected connectivity (Scheme 1).
The naturally occurring phenantropyrone Ochrolide 53, first
isolated from the orchid Coelogine orchracea, was also selected
as an interesting synthetic target. In this case, the final molecule
is planar however, it can be envisaged that if the twisted
phenanthrene 52 could be obtained, removal of the benzyl-
protecting groups should also spontaneously induce the desired
transesterification (Scheme 2). After preparation of the two
coupling partners 46 and 47 (see the Supporting Information),
a Negishi reaction afforded the corresponding biphenyl
derivative 48 in which the aldehyde moiety is already
deprotected. Subsequently, the Ohira−Bestmann reaction
produced the expected alkyne 49.
The stage was then set for the key Au-catalyzed cyclization.
Although the formation of 51 could be observed, the benzyl-
deprotected product 52 was also formed in similar quantities. It
is well-known that Lewis acids promote deprotection of
benzylated phenols having a carbonyl group in the ortho-
position. Hence, the formation of 52 could be attributed to the
exceptional Lewis acidity of Au when coordinated to phosphine
1. Analytically pure 52 could be obtained after selective
monodebenzylation of 49 with AlCl3 followed by treatment
with 6/AgBF4. In any case, Pd/C catalyzed cleavage of the
benzyl ethers in 52 or mixtures containing 51 and 52 cleanly
generated Ochrolide 53. As expected, hydrogenation of 52
under more forcing conditions (H2, 20 bar) also allowed the
isolation of the 9,10-dihydrophenanthrene derivative Coelogi-
nin 54, another phenanthrenoid that was first isolated from the
high-altitude Himalayan orchid Coelogyne cristata.23
Figure 4. X-ray crystal structures of 8 and 35 in the solid state.
Dihedral angles between the two terminal rings are defined in terms of
the torsion angle C4−C4a−C5a−C5. Hydrogen atoms are omitted for
clarity, and ellipsoids are shown at 50% probability.
Interestingly, the biphenyl substrate 23 behaves differently
when submitted to the standard reaction conditions. Up to four
different products with phenanthrene architecture could be
isolated from the reaction mixture in which the isopropyl group
ended up in positions 5 (37), 6 (38), 8 (36), or was not even
present as shown by the isolation of 26. Apparently, with this
sterically very hindered substrate, new competitive reaction
pathways emerge leading to the formation of more stable
2.4. Computational Results. Up to this point, our
explanations and mechanistic arguments, although based on
18819
dx.doi.org/10.1021/ja411146x | J. Am. Chem. Soc. 2013, 135, 18815−18823

Journal of the American Chemical Society
Article
used to characterize synthetically inaccessible compounds and
to elucidate reactions pathways.24
Scheme 1. Synthesis of Bulbophyllantrin and
Marylaurencinol A
In an attempt to gain insight into the origin of the
remarkable reactivity of the actual catalyst 55, we studied its
electronic structure by performing a fragment molecular orbital
(MO) analysis employing the Amsterdam Density Functional
(ADF) program package (see the Supporting Information).25
In this analysis, the MOs are expanded in terms of Slater-type
orbitals, employing a triple-ζ basis set with one polarization
function (TZP); relativistic effects were included using the
zero-order regular approximation (ZORA) approach,26 as
implemented in the ADF program package.
In catalyst 55, the interaction between the fragment MOs
from 1 and Au+ generates a low-lying antibonding orbital (σ*
P−Au, LUMO) that has a strong contribution from the Au 6s
orbital (46%) (see Figure 5 (left) and Figure S54 of the
Conditions: (a) Pd2(dba)3 5 mol %, Cy3P 11 mol %, Cs2CO3 (2.5
equiv), dioxane:toluene (2:3), 85 °C, 16 h, 85%; (b) Ohira−Bestmann
reagent (1.5 equiv), K2CO3 (2 equiv), MeOH, 16h, room temperature,
91%; (c) 6 2 mol %, AgSbF6 2 mol %, CH2Cl2 (0.05 M), 10 min, room
temperature, 90%; (d) Pd/C (10%), 20 mol %, AcOEt, 24 h, H2 (1
atm.), 81%; (e) Pd/C (10%), 50 mol %, MeOH, 16 h, H2 (1 atm.),
97%.
Figure 5. Left, LUMO of 55; right, LUMO of [(MeO)3PAu]+ (56).
Scheme 2. Synthesis of Ochrolide and Coeloginin
Supporting Information). For comparison, Figure 5 also depicts
the LUMO of [(MeO)3PAu]+ (56) (right). In both cases, the
LUMO has a similar shape (σ* P−Au), but it lies much lower
in 55 (−11.89 eV) than in 56 (−8.03 eV). As in both systems,
the LUMO is the main orbital involved in the coordination
with the alkyne substrate, its energetic position thus governs
the π-acidity of both catalysts. Therefore, the lower LUMO
energy confers a stronger π-acidity to 55 and accounts for its
higher reactivity when compared with other phosphorus-
containing Au complexes such as 56.
The mechanistic pathways for the cycloisomerization of 7 to
8 catalyzed by 55 (green) and [Ph3PAu]+ (57) (red) were
explored by DFT calculations (see Figure 6). Geometry
optimizations were carried out using BP8627,28 functional in
combination with def2-SVP basis sets.29 In the case of gold, the
60 inner-shell core electrons were replaced by an effective core
potential (ECP) generated for the neutral atom using quasi-
relativistic methods, and the explicitly treated electrons were
described by the standard def2-ECP basis set.30 The resolution-
of-identity (RI) approximation31 was applied in conjunction
with appropriate auxiliary basis sets to speed up the
calculations. All low-energy conformations of the substrate
and the catalyst were considered during initial screening, and all
relevant stationary points were characterized as minima or first-
order transition states by evaluating the harmonic vibrational
frequencies at the same level (RI-BP86/def2-SVP) that had
been applied for geometry optimization.
Conditions: (a) 46, n-BuLi, THF, 1 h; then ZnCl2 and Pd2(dba)3 2.5
mol %, 2-dicyclohexylphosphino-2′,6′-di(isopropyl)oxybiphenyl 6 mol
% and 47, 64% ; (b) Ohira−Bestmann reagent (1.5 equiv), K2CO3 (2
equiv), MeOH, 16 h, room temperature, 93%; (c) AlCl3, benzene,
reflux, 73%; (d) 6/AgSbF6 5 mol %, Cl(CH2)2Cl, 75%; (e) Pd/C
(10%), AcOEt, H2 (1 atm), 93%; (f) Pd/C (10%), AcOEt, H2 (20
atm), 92%.
kinetic data, have been qualitative in nature. Computational
methods nowadays provide quantitative information that can be
The influence of the solvent environment (dichloromethane,
dielectric constant ε = 8.93) on the relative energies was
18820
dx.doi.org/10.1021/ja411146x | J. Am. Chem. Soc. 2013, 135, 18815−18823

Journal of the American Chemical Society
Article
Figure 6. Free energy profiles (kcal/mol) for the cyclization of 2-ethynyl-2′,6-dimethylbiphenyl 7 to 4,5-dimethylphenanthrene 8. The green and red
lines represent the calculated profiles for catalysts 55 and 57, respectively. Both diasteromeric protons were considered in the calculation of TS4a and
TS4b.
investigated through single-point calculations with the con-
ductor-like screening model (COSMO)32 at RI-BP86/def2-
SVP level. To evaluate the best estimate of total energies, all
located stationary points were reoptimized at RI-BP86 level
employing the def2-TZVPP basis set. The empirical Grimme-
type dispersion corrections were also incorporated during this
step using the latest parametrization (DFT-D3).33 Relative free
energies (ΔG) at standard pressure (1 bar) and 273.15 K were
determined at the RI-BP86/def2-SVP level. The required
thermal and entropic contributions were evaluated within the
rigid-rotor harmonic-oscillator approximation. All geometry
optimizations were performed using the TURBOMOLE
(version 6.4) suite of program.34 To check the sensitivity of
the computed energy profile, single-point energies at optimized
RI-BP86/def2-TZVPP geometries were evaluated by using
more advanced functionals (B3LYP, M06), and larger basis sets
(def2-SVP, def2-TZVPP, and 6-31+G*).The single-point
results (see the Supporting Information) confirm the qualitative
conclusions drawn from the standard approach outlined above
and will therefore not be discussed in the following.
Figure 6 depicts the calculated pathways for the cyclo-
isomerization of 7 to 8. For both catalysts, the reaction begins
with the coordination of the alkyne to Au forming a catalyst-
substrate complex Int1, which is much more exergonic when
55 is employed as catalyst, Int1a. The higher thermodynamic
stability of Int1a has its origin in the very low-lying LUMO of
the catalyst 55 that favors the coordination of substrate (Figure
5). The next step along the reaction coordinate involves in both
cases the nucleophilic attack of the arene ring on the activated
employed: The free energy of activation (relative to the
preceding intermediate) is only 7.6 kcal/mol compared with
20.9 kcal/mol when using the archetypical catalyst 57. This is
the main reason why the use of ligand 1 significantly fosters the
whole process. The following steps, cyclopropane ring-opening
and 1,2-H shift to Int4a/b, are quite facile in both cases (free
energy barriers of less than 10 kcal/mol) and do not depend
much on the nature of the ancillary ligand.
The two intermediates Int4a/b are more stable than their
precursors Int3a/b by 14.8 and 21.3 kcal/mol, respectively, and
the energies for the following transition states TS4a (−59.1
kcal/mol) and TS4b (−29.7 kcal/mol) are much lower than
those of TS3a (−48.7 kcal/mol) and TS3b (−15.8 kcal/mol),
respectively. Thus, the formation of the Au-carbenes Int4a/b is
likely to be irreversible. In the final step, a second 1,2-hydrogen
shift (involving either one of the diasteromeric H atoms) yields
the phenanthrene−Au π-complexes, the dissociation of which
will provide the desired product and regenerate the catalysts.
The interaction between catalyst 55 and the product is
calculated to be very strong (92.4 kcal/mol), which may be
responsible for the catalyst deactivation of the catalysts
observed at high substrate conversion, once the phenanthrene
product accumulates (Figure 3a).
We have examined in an analogous manner the cyclo-
isomerization of the substrate 9 to the 6-endo product 10 and
the 7-exo product 11 using 55 and Me3PAuCl as catalysts. The
computed mechanistic pathway for the conversion 9 →10 (see
Figure S25 of the Supporting Information) is similar to that
established for 7→8 (see Figure 6), with one minor
modification: The cyclization does not proceed through the
formation of a cyclopropyl intermediate, instead the six-
membered ring is formed in one step. For both conversions,
alkyne to form the expected cyclopropyl intermediates Int2a/b
.
Judging from the computed relative free energies, this step is
much easier when the strongly activating catalyst 55 is
18821
dx.doi.org/10.1021/ja411146x | J. Am. Chem. Soc. 2013, 135, 18815−18823

Journal of the American Chemical Society
Article
55 outperforms the other considered catalysts (Ph3PAuCl and
Me3PAuCl) by a large margin, with significantly lower rate-
limiting barriers. When using catalyst 55, the initial cyclization
is computed to be more facile for 7 than for 9 (barriers of 7.6
and 13.8 kcal/mol, respectively).
fellowship for J.C.) and the Chinese Scholarship Council
(doctoral fellowship to L.G.).
REFERENCES
■
(1) Christmann, U.; Vilar, R. Angew. Chem., Int. Ed. 2005, 44, 366.
(2) (a) Hartwig, J. F. Inorg. Chem. 2007, 46, 1936. (b) Barder, T. E.;
Buchwald, S. L. J. Am. Chem. Soc. 2007, 129, 12003.
We have also located pathways for the conversion 9 →11
(see Figure S44 of the Supporting Information). When using
55 as catalyst, the relevant barriers are higher than those for the
cyclization 9 →10, consistent with the observed preferential
formation of the 6-endo product 10 (see Figure 3). When
Me3PAuCl is used as catalyst, the computed barriers are
generally higher than in the case of 55 (as expected), but the
qualitative preferences remain the same (6-endo product
favored). Hence, the observed reversal of selectivity when
changing the catalyst (see Figure 3) is not reproduced by the
present calculations. Despite extensive scans of the potential
energy surface, we apparently still miss a low-energy pathway
for the Me3PAuCl-catalyzed conversion 9 →11. Further work is
needed to find such a pathway (possibly with the use of more
extended model systems).
(3) Tolman, C. A. Chem. Rev. 1977, 77, 313.
(4) Carbon monoxide, a pivotal ligand in transition metal chemistry,
is also a very strong π-acceptor ligand. However, it presents a series of
disadvantages such as gaseous nature and toxicity. More importantly, it
is often a noninnocent ligand and its structure does not allow any
tuning of the stereoelectronic properties.
́
(5) (a) Bennet, F. W.; Brandt, G. R. A.; Emeleus, H. J.; Haszeldine, R.
N. Nature 1950, 166, 225. (b) Drews, T.; Rusch, D.; Siedel, S.;
Willemsen, S.; Seppelt, K. Chem.Eur. J. 2008, 14, 4280.
(6) (a) Carreras, J.; Patil, M.; Thiel, W.; Alcarazo, M. J. Am. Chem.
̌
Soc. 2012, 134, 16753. (b) Petuskova, J.; Patil, M.; Holle, S.; Lehmann,
C. W.; Thiel, W.; Alcarazo, M. J. Am. Chem. Soc. 2011, 133, 20758. (c)
c) Weigand, J. J.; Feldmann, K.; Henne, F. D. J. Am. Chem. Soc. 2010,
132, 16321.
(7) (a) Andrieu, J.; Azuori, M.; Richard, P. Inorg. Chem. Commun.
2008, 11, 1401. (b) Digard, E.; Andrieu, J.; Cattey, H. Inorg. Chem.
Commun. 2012, 25, 39. (c) Petuskova, J.; Bruns, H.; Alcarazo, M.
̌
3. CONCLUSIONS
Angew. Chem., Int. Ed. 2011, 50, 3799. (d) Canac, Y.; Debono, N.;
Vendier, L.; Chauvin, R. Inorg. Chem. 2009, 48, 5562. (e) Abdellah, I.;
Lepetit, C.; Canac, Y.; Duhayon, C.; Chauvin, R. Chem.Eur. J. 2010,
16, 13095. (f) Canac, Y.; Maaliki, C.; Abdellah, I.; Chauvin, R. New J.
Chem. 2012, 36, 17.
(8) Coles, M. P.; Hitchcock, P. B. Chem. Commun. 2007, 5229.
(9) (a) Schoeller, W. W.; Frey, G. D.; Bertrand, G. Chem.Eur. J.
2008, 14, 4711. For related bis(imidazolium)-substituted phosphines
see (b) Weigand, J. J.; Feldmann, K. O.; Henne, F. D. J. Am. Chem. Soc.
2010, 132, 16321.
(10) Azuori, M.; Andrieu, J.; Picquet, M.; Cattey, H. Inorg. Chem.
2009, 48, 1236.
(11) Smith, J. M.; Coville, N. J. Organometallics 2001, 20, 1210.
(12) Gusev, D. G. Organometallics 2009, 28, 763.
This study provides a conceptually new approach to the design
of extreme π-acceptor ligands for transition metal catalysis
based on the use of dicationic phosphines with two
cyclopropenium groups directly attached to the phosphorus
atom. The remarkable ability of this ligand to enhance the
natural π-acidity of Au(I) was demonstrated in a series of
examples, which achieve levels of reactivity and selectivity that
clearly surpass the known catalytic systems. Moreover, and as
an additional advantage, complex 6 can be weighed and
handled in air and it is stable for months when stored under
inert gas. Considering the multitude of transformations that
might benefit from the use of extreme π-acceptor ligands, we
believe that this approach shows a vast potential in the
development and improvement of transition metal catalyzed
reactions.
(13) For recent reviews see: (a) Jimen
M. Chem. Rev. 2008, 108, 3326. (b) Gorin, D. J.; Sherry, B. D.; Toste,
́
ez-Nunez, E.; Echavarren, A.
́
̃
F. D. Chem. Rev. 2008, 108, 3351. (c) Furstner, A.; Davies, P. W.
̈
Angew. Chem., Int. Ed. 2007, 46, 3410. (d) Rudolph, M.; Hashmi, A. S.
ASSOCIATED CONTENT
K. Chem. Soc. Rev. 2012, 41, 2448.
■
(14) Mamane, V.; Hannen, P.; Furstner, A. Chem.Eur. J. 2004, 10,
̈
S
* Supporting Information
4556.
Experimental procedures including characterization data for all
new compounds, NMR spectra, CIF files for 1, 5, 8, 30, 35, and
40, as well as computational procedures and detailed
computational results. This material is available free of charge
via the Internet at http://pubs.acs.org.
(15) Singh, S. B.; Pettit, G. R. J. Org. Chem. 1989, 54, 4105.
(16) Soriano, E.; Marco-Contelles, J. Organometallics 2006, 25, 4542.
(17) Kovac
1084.
́
s, A.; Vasas, A.; Hohmann, J. Phytochemistry 2008, 69,
(18) Majumder, P. L.; Kar, A.; Shoolery, J. N. Phytochemistry 1985, 9,
2083.
(19) Yoshikawa, K.; Ito, T.; Iseki, K.; Baba, C.; Imagawa, H.; Yagi, Y.;
Morita, H.; Asakawa, Y.; Kawano, S.; Hashimoto, T. J. Nat. Prod. 2012,
75, 605.
(20) Stavrakov, G.; Keller, M.; Breit, B. Eur. J. Org. Chem. 2007, 5726.
(21) Potenziano, J.; Spitale, R.; Janik, M. E. Synth. Commun. 2005, 35,
2005.
(22) (a) Ohira, S. Synth. Commun. 1989, 19, 561. (b) Muller, S.;
Liepold, B.; Roth, G. J; Bestmann, H. J. Synlett. 1996, 521.
(23) Majumder, P.; Brandyopadhyay, D.; Joardar, S. J. Chem. Soc.
Perkin I 1982, 1131.
(24) For theoretical calculations on the activation of alkynes by Au
and Pt complexes see: (a) Pernpointner, M.; Hashmi, A. S. K. J. Chem.
Theory Comput. 2009, 5, 2717. (b) Lein, M.; Rudolph, M.; Hashmi, S.
K.; Schwerdtfeger, P. Organometallics 2010, 29, 2206.
(25) (a) Baerends, E. J.; Ellis, D. E. M; Ros, P. Chem. Phys. 1973, 2,
41. (b) Velde, G. T.; Bickelhaupt, F. M.; Baerends, E. J.; Guerra, C. F.;
Van Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T. J. Comput. Chem.
2001, 22, 931. (c) SCM. ADF 2010.02; Theoretical Chemistry,Vrije
AUTHOR INFORMATION
■
Corresponding Author
*alcarazo@mpi-muelheim.mpg.de
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Dedicated to Prof. Manfred Reetz on the occasion of his 70th
birthday. The authors thank C. Fares
assistance, C. W. Lehmann, R. Goddard, and J. Rust for X-ray
crystallographic analysis, and A. Degge for HPLC assistance. A.
Furstner is thanked for continuous support and encourage-
ment. Financial support for this work was provided by the Max
Planck Gesellschaft, the Deutsche Forschung Gemeinschaft, the
̀
for NMR spectroscopic
̈
Spanish Ministerio de Educacion y Ciencia (postdoctoral
́
18822
dx.doi.org/10.1021/ja411146x | J. Am. Chem. Soc. 2013, 135, 18815−18823

Journal of the American Chemical Society
Article
Universiteit: Amsterdam, The Netherlands, 2008; http://www.scm.
com/.
(26) (a) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. J. Chem. Phys.
1994, 101, 9783. (b) van Lenthe, E.; Ehlers, A.; Baerends, E. J. J. Chem.
Phys. 1999, 110, 8943.
(27) Becke, A. D. Phys. Rev. A 1988, 38, 3098.
(28) Perdew, J. P. Phys. Rev. B 1986, 33, 8822.
(29) (a) Schafer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97,
̈
2571. (b) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7,
3297. (c) Weigend, F. Phys. Chem. Chem. Phys. 2006, 8, 1057.
(30) Andrae, D.; Haussermann, U.; Dolg, M.; Stoll, H.; Preuss, H.
̈
Theor. Chim. Acta 1990, 77, 123.
(31) (a) Eichkorn, K.; Treutler, O.; Ohm, H.; Haser, M.; Ahlrichs, R.
Chem. Phys. Lett. 1995, 242, 652. (b) Eichkorn, K.; Weigend, F.;
Treutler, O.; Ahlrichs, R. Theor. Chem. Acc. 1997, 97, 119.
(c) Weigend, F. Phys. Chem. Chem. Phys. 2002, 4, 4285.
̈
̈
(32) Klamt, A.; Schuurmann, G. J. Chem. Soc. Perkin Trans. 2 1993, 5,
̈
̈
799.
(33) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys.
2010, 132, 154104.
(34) (a) Ahlrichs, R.; Bar, M.; Haser, M.; Horn, H.; Kolmel, C. Chem.
̈
̈
̈
Phys. Lett. 1989, 162, 165. (b) TURBOMOLE version 6.4; University
of Karlsruhe and Forschungszentrum Karlsruhe GmbH: Germany,
2012; http://www.turbomole.com.
18823
dx.doi.org/10.1021/ja411146x | J. Am. Chem. Soc. 2013, 135, 18815−18823