Ruthenium-catalyzed allylic alkylation reactions: Carbonate-based catalysts and intermediates

Article abstract of DOI:10.1002/anie.200500967

(Chemical Equation Presented) Branching out: A previously unrecognized ruthenium carbonate intermediate is formed during the reaction of an allylic tert-butoxycarbonate with the ruthenium complex [Ru(Cp*) (CH ₃CN)₃]PF₆ (1;

Full text of DOI:10.1002/anie.200500967

Angewandte
Chemie
Homogeneous Catalysis
Ruthenium-Catalyzed Allylic Alkylation
Reactions: Carbonate-Based Catalysts and
Intermediates**
RenØ Hermatschweiler, Ignacio Fernµndez,
Frank Breher, Paul S. Pregosin,* Luis F. Veiros, and
Maria JosØ Calhorda
We have recently reported[8] that the source of the
observed high branched-to-linear regioselectivity (when X =
Cl), has an electronic origin. These findings are based on X-
ray crystallography and NMR spectroscopy data, together
with DFT calculations. In an extension of this study we show
herein that a) the carbonate anion can remain in the
coordination sphere of the ruthenium and b) an isolated,
well characterized RuIV allyl carbonate complex is a better
catalyst than 1.
Reaction of the branched tert-butyl carbonates 4a and 4b
(Scheme 1) with 1 in DMFat ambient temperature affords the
RuIV cationic complexes 3a and 3b, respectively, in excellent
yield.[9]
Transition-metal-catalyzed allylic alkylation reactions con-
tinue to attract attention. This chemistry can be accomplished
using a variety of (primarily) late-transition-metal com-
plexes[1–3] and affords high yields of product. In contrast to
palladium(ii)-catalyzed allylic alkylation reactions,[1] which
often afford the linear (less substituted) product, the use of
ruthenium-based catalysts affords primarily branched organic
products; that is, the nucleophile prefers to attack at the more
substituted carbon,[4] immediately adjacent to the phenyl
group [Eq. (1) Cp* = C5Me5; X = halide or carbonate].
½½RuðCp*ÞðCH3CNÞ3ŠPF6
PhCH¼CHCH2X þ NuÀ
ƒƒƒƒƒƒƒƒƒƒƒƒ!
DMF or acetone
ð1Þ
PhCHðNuÞCH¼CH2 þ XÀ
Trost et al.[4] have shown that the cationic tris(nitrile)
complex [Ru(Cp*)(CH3CN)3]PF6 (1), is an excellent catalyst
for the chemistry shown in Equation (1) and that carbonate is
the preferred leaving group, X, in terms of the observed
regioselectivity.[4] Bruneau et al.[5] have recently reported that
the bipyridine derivatives, 2, are also efficient catalysts.
Further, we note that conventional wisdom suggests that an
oxidative addition mechanism is operating in this allylation
chemistry,[4,6,7] although no explanation for the role of the
anion (chloride or carbonate) has been forthcoming.
Scheme 1. Synthesis of the new carbonate complexes, 3a (R=H) and
3b (R=OMe).
The solid-state structure[10] of 3b (Figure 1) reveals the
expected distorted piano-stool arrangement of the various
ligands in this coordinatively saturated complex. The tert-
butyl carbonate ligand clearly coordinates in a bidentate
fashion through two oxygen atoms, and the allyl ligand shows
an endo configuration such that the phenyl substituent is
remote to the bulky Cp* ligand. Thus the oxidative addition
affords the monocationic carbonate and not the bis(acetoni-
trile) dicationic complex.
[*] R. Hermatschweiler, Dr. I. Fernµndez, Dr. F. Breher,
Prof. Dr. P. S. Pregosin
Laboratory of Inorganic Chemistry
ETHZ, Hönggerberg
8093 Zürich (Switzerland)
Fax: (+41)44-633-1071
E-mail: pregosin@inorg.chem.ethz.ch
The characteristic chemical shifts in the 1H and 13C NMR
spectra[9] of pure 3a have been recorded and compared to the
analogous data from the in situ measurement involving
stoichiometric amounts of 1 and the branched tert-butyl
carbonate. These results show that 3a is formed immediately
on mixing, which indicates that a) the oxidative addition is not
likely to be the rate-determining step in the catalysis and
b) the isolated carbonate complex is not formed during the
work-up procedure.
The reaction between a DMF solution of 3a and one
equivalent of the dimethyl malonate anion (DMM) leads to
the formation of a mixture of branched and linear organic
products [Eq. (2)]. This reaction shows 3a to be reactive
towards nucleophiles.
Prof. Dr. L. F. Veiros
Centro de Química Estrutural
Complexo I, Instituto Superior TØcnico
Av. Rovisco Pais 1, 1049-001 Lisbon (Portugal)
Prof. M. J. Calhorda
Departamento de Química e Bioquímica
Faculdade de CiÞncias
Universidade de Lisboa, 1749-016 Lisbon (Portugal)
[**] P.S.P. thanks the Swiss National Science Foundation, the
“Bundesamt für Bildung und Wissenschaft”, and the ETHZ for
financial support. I.F. thanks the Junta de Andalucía for a research
contract. We also thank Johnson Matthey for the loan of precious
metals. We acknowledge the COST working group D24/0014/02.
Supporting information for this article (computational details, the
corresponding list of references, and the atomic coordinates for the
optimized structures) is available on the WWW under http://
www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2005, 44, 4397 –4400
DOI: 10.1002/anie.200500967
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4397

Communications
We believe the difference between the reactivity of 4a and
5 is related to the differing stabilities of the h6-arene
complexes formed by these substrates under the reaction
conditions.[11] These results suggest that, not only is 3a a
relevant structure in the catalytic cycle, but that the presence
of the nitrile ligands in solution, when using catalyst 1, slows
the reaction in Equation (1), presumably as these ligands
compete with the olefin for a position on the ruthenium
center. We note that for the complexes 2, Bruneau reports
reaction times of around 16 h at ambient temperature
whereas with 3a the reactions are complete within a few
minutes.
DFT calculations[12] were carried out on model complexes,
where the Cp* ligand was replaced by Cp (Cp = C5H5), and
the tert-butyl substituent by a methyl group. Three isomers
were considered, namely, an analogue of allyl/carbonates 3a
and 3b, [Ru(Cp){OC(OMe)O}(h3-PhCHCHCH2)]+ (3’), and
the two ruthenium h6-arene complexes, 8 (derived from
coordination of the linear carbonate, 5) and 9 (derived from
coordination of the branched carbonate 4). The three
optimized structures are represented in Figure 2. The calcu-
lated geometry of the carbonate complex 3’ compares well
with the experimentally determined structure of 3b (Figure 1)
Figure 1. Molecular structure for the cation of 3b. Thermal ellipsoids
are set at 30% probability. The PF6 ion has been omitted for clarity.
À
Selected bond lengths [pm] and angles [8]: Ru1-O1 214.8(3), Ru1-O2
211.6(3), Ru1-C20 216.2(5), Ru1-C21 213.7(5), Ru1-C22 230.3(5), Ru1-
Ct 183.8(5), C1-O1 126.0(6), C1-O2 127.1(6), C1-O3 131.1(6); O1-C1-
O2 117.9(4), O1-C1-O3 125.9(5), O2-C1-O3 116.2(4), C20-C21-C22
116.5(5); f=32.48; Ct=centroid of the Cp* ligand; f is the intersec-
tion of the planes described by Ru1, O1, C1, O2 and the atoms of the
five-membered Cp* ring.
We have used complex 3a as a catalyst for the allylic
alkylation reaction with dimethyl malonate (Scheme 2,
Table 1). With the branched carbonate, 4a, the reaction was
Scheme 2. Alkylation of branched and linear carbonates by dimethyl malonate/
NaH with 1 or 3a as the catalyst (see Table 1).
25-times faster with complex 3a than with 1. Using the linear
substrate 5, the reaction with complex 3a is only a factor of
two faster than 1.
Table 1: Yields and regioselectivities of the alkylation of 4a and 5 by
dimethyl malonate (see Scheme 2).
Catalyst
Substrate
t [min]
Yield [%]
6/7[a]
1
3a
1
5
5
4a
4a
40
20
25
1
98
96
98
97
90:10
90:10
90:10
90:10
Figure 2. Optimized geometries (B3LYP) of the allyl/carbonate, 3’ and
the h6-arene complexes, 8 (linear carbonate) and 9 (branched
carbonate. The Ru and the O atoms are gray and relevant
distances [Š], Wiberg indices (italics), and the relative energies of the
three isomers are indicated.
3a
[a] The ratio 6/7 (branched/linear) was determined by 1H NMR
spectroscopy.
4398
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 4397 –4400

Angewandte
Chemie
with maximum and mean absolute deviations of 0.15 and
0.07 Š, respectively, for the interatomic distances of the metal
coordination sphere.
The coordination of the allyl ligand in 3’ is distorted, and is
similar to what was observed for [Ru(Cp*)Cl(CH3CN)(h3-
PhCHCHCH2)]+.[8] The bond length from the metal center to
the allyl carbon atom C22 adjacent to the phenyl group is
0.26 Š longer than the corresponding distance to the meth-
use does not lead to the presence of competing ligands in the
reaction mixture.
Received: March 16, 2005
Published online: June 16, 2005
Keywords: allylic alkylation · carbonate complex · density
.
functional calculations · ruthenium · structure elucidation
À
ylene allyl carbon atom C20. This Ru C22 distance reflects a
weaker bond, as shown by the Wiberg indices (Figure 2), and
supported by natural population analysis (NPA) charges[12]
(À0.32 for C20 and À0.08 for C22) indicative of a less
efficient back-donation from the metal to C22.
[1] B. M. Trost, M. L. Crawley, Chem. Rev. 2003, 103, 2921; B. M.
Trost, D. L. van Vranken, Chem. Rev. 1996, 96, 395; H. Tye, J.
Chem. Soc. Perkin Trans. 1 2000, 275; L. F. Tietze, H. Ila, P. Bell.
Chem. Rev. 2004, 104, 3453; O. Reiser, Angew. Chem. 1993, 105,
576; Angew. Chem. Int. Ed. Engl. 1993, 32, 547.
[2] B. M. Trost, I. Hachiya, J. Am. Chem. Soc. 1998, 120, 1104; B. M.
Trost, N. G. Andersen, J. Am. Chem. Soc. 2002, 124, 14320.
[3] D. A. Evans, J. D. Nelson, Tetrahedron Lett. 1998, 39, 1725; D. A.
Evans, J. D. Nelson, J. Am. Chem. Soc. 1998, 120, 5581.
[4] B. M. Trost, P. L. Fraisse, Z. T. Ball, Angew. Chem. 2002, 114,
1101; Angew. Chem. Int. Ed. 2002, 41, 1059; B. M. Trost, C. M.
Older, Organometallics 2002, 21, 2544; S. W. Zhang, T. Mitsudo,
T. Kondo, Y. Watanabe, J. Organomet. Chem. 1993, 450, 197 –
207.
Interestingly, both the charges calculated for the allylic
carbon atoms and the nature of the LUMO of 3’, with a larger
contribution from C22 than from C20 (Figure 3), are in
agreement with the observed preference of nucleophilic
attack on C22 and the experimentally observed regioselec-
tivity of the reaction.
[5] M. D. Mbaye, B. Demerseman, J. L. Renaud, L. Toupet, C.
Bruneau, Angew. Chem. 2003, 115, 5220; Angew. Chem. Int. Ed.
2003, 42, 5066.
[6] T. Kondo, H. Ono, N. Satake, T. Mitsudo, Y. Watanabe,
Organometallics 1995, 14, 1945; H. Kondo, Y. Yamaguchi, H.
Nagashima, Chem. Commun. 2000, 1075.
[7] K. Onitsuka, Y. Ajioka, Y. Matsushima, S. Takahashi, Organo-
metallics 2001, 20, 3274; Y. Matsushima, K. Onitsuka, S.
Takahashi, Organometallics 2004, 23, 3763.
[8] R. Hermatschweiler, I. Fernµndez, P. S. Pregosin, E. J. Watson,
A. Albinati, S. Rizzato, L. F. Veiros, M. J. Calhorda, Organo-
metallics 2005, 24, 1809.
[9] Synthesis of 3a. [RuCp*(CH3CN)3]PF6 (64 mg, 0.128 mmol) and
4a (30 mg, 0.128 mmol) were stirred in DMF (3 mL) for 30 min
at room temperature. The solution volume was reduced in
vacuum and diethyl ether was added precipitating an orange-
brown powder. The solid was washed with diethyl ether and
dried under vacuum to yield 74 mg (94%) of the crude product.
1H NMR (500 mHz, [D7]DMF, 298 K): d = 1.53(9H), 1.72
(15H), 3.52 (1H, J = 10 Hz), 4.66 (1H, J = 6.5 Hz), 5.11 (1H,
J = 11.0 Hz), 6.36 (1H, J = 11.0, 10.0, 6.5 Hz), 7.42 (2H, J = 7.7,
7.3Hz), 7.62 (1H, J = 7.7 Hz), 7.74 ppm (2H, J = 7.3Hz);
13C NMR (126 mHz, [D7]DMF, 298 K): d = 8.8 (CH3), 28.2
(CH3), 65.8 (H2Callyl), 85.9 (C), 90.2 (HCallyl), 99.9 (HCallyl),
107.2 (C), 129.2 (HCAr), 130.5 (HCAr), 130.8 (HCAr), 135.2
(Cipso), 164.4 ppm (CO3). Elemental analysis (%) calcd for
C24H33O3F6PRu: C 46.83, H 5.40; found: C 46.15, H 5.25. ESI
MS: 471.2 [M+], 427.2 [M+ÀCO2], 371.2, 353.2 [M+ÀC5H9O3,
M+ÀC9H9]. Synthesis of 3b. [RuCp*(CH3CN)3]PF6 (55 mg,
0.110 mmol) and 4b (29 mg, 0.110 mmol) were stirred in DMF
(3mL) for 30 min at room temperature. The solution volume
was reduced in vacuum and diethyl ether was added precipitat-
ing an orange-red powder. The solid was washed with diethyl
ether and dried under vacuum to yield 67 mg (95%) of the crude
Figure 3. LUMO of 3’, showing a large contribution from C22 of the
allyl ligand.
The calculated relative energies of the isomeric h6-arene
complexes 8 and 9 (Figure 2) indicate that the linear
carbonate, 8, forms a complex which is 3.0 kcalmolÀ1 more
stable than the branched analogue, 9, corroborating the
observed differences in reactivity for the corresponding
substrates 4a and 5. In addition, the geometrical features of
the two species might also play a role in the course of the
reaction. For the isomer with the branched substrate, 9, the
carbonate group is considerably closer to the metal than in the
linear carbonate analogue, 8. The formation of the allyl/
carbonate complex 3’ should hence be considerably easier for
9 than 8, in agreement with the differences in observed
reaction rates. An h6-arene complex with the branched
carbonate could not be detected experimentally.[11]
1
product. H NMR (500 mHz, [D7]DMF, 298 K): d = 1.53(9H),
1.70 (15H), 3.50 (1H, J = 9.6 Hz), 4.61 (1H, J = 6.4 Hz), 5.21
(1H, J = 11.1 Hz), 6.20 (1H, J = 11.1, 9.6, 6.4 Hz), 7.00 (2H, J =
8.6 Hz), 7.71 ppm (2H, J = 8.6 Hz); 13C NMR(126 mHz,
In conclusion, we have shown that the carbonate anion
coordinates to the ruthenium atom, yielding a previously
unknown reaction intermediate in the allylic alkylation
catalytic sequence. Moreover, when isolated, this ruthenium
carbonate species serves as a new, even more efficient catalyst
than the reported nitrile complexes, presumably because its
[D7]DMF, 298 K): d = 8.8 (CH3), 28.3(CH ), 55.7 (OCH3), 65.3
3
(H2Callyl), 85.7 (C), 93.2 (HCallyl), 98.6 (HCallyl), 106.3(C), 114.8
(HCAr), 127.1 (Cipso), 132.8 (HCAr), 162.0 (Cipso), 164.3ppm
(CO3). Elemental analysis (%) calcd for C25H35O4F6PRu: C
Angew. Chem. Int. Ed. 2005, 44, 4397 –4400
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4399

Communications
46.51, H 5.46; found: C 46.31, H 5.40. ESI MS: 501.1 [M+], 457.2
[M+ÀCO2], 401.1, 383.1 [M+ÀC5H9O3), 353.3 [M+ÀC10H11O].
Catalysis: Dimethyl malonate (76 mL, 0.65 mmol) and NaH
(26 mg, 0.64 mmol, 60% dispersion in mineral oil) were stirred
in DMF (1.5 mL) for 30 min at room temperature, after which
time the carbonate 4a or 5 (0.21 mmol) and the catalyst 1 or 3a
(0.02 mmol, 3mol%) were added. The resulting solution was
stirred at room temperature (see Table 1 for reaction times) and
then diluted with diethyl ether and water (ca. 10 mL). After
three extractions with diethyl ether, the combined organic
extracts were washed with water and brine, dried over
MgSO4,and concentrated. The crude product was purified by
chromatography on silica gel (hexane/ethyl acetate 6/1). The
branched/linear ratio was determined by 1H NMR spectroscopy.
[10] Crystal structure of 3b: Orange single crystals were obtained
from a CH2Cl2 solution of 3b which was layered with n-pentane;
C25H35F6O4PRu, orthorhombic, space group Pbca; a = 11.030(1),
b = 21.378(2), c = 23.626(2) Š; V= 5570.9(7) Š3; Z = 8; 1calcd
1.539 MgmÀ3; crystal dimensions 0.58 ” 0.23” 0.13mm; diffrac-
=
tometer Bruker SMART Apex; MoKa radiation, 200 K, 2Vmax
49.428; 41497 reflections, 4749 independent (Rint = 0.0451),
direct methods; empirical absorption correction SADABS (v.
2.03); refinement against full matrix (versus F2) with SHELXTL
(v. 6.12) and SHELXL-97, 343 parameters, R1 = 0.0596 and wR2
(all data) = 0.1231, max./min. residual electron density 0.877/
À1.067 eŠÀ3. The hydrogen atoms were placed in idealized
positions and included as riding atoms. CCDC-265478 (3b)
contains the supplementary crystallographic data for this paper.
These data can be obtained free of charge from the Cambridge
Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_
request/cif.
=
[11] Starting from substrate 5, several 18-electron h6-arene RuII
6
=
complexes of the type, [Ru(Cp*)(h -RC6H4CH CHCH2OC(O-
tBu)O)]PF6 where R is a para substituent can be isolated. This
type of arene complex is well known, see Organometallics (Eds.:
C. Elschenbroich, A. Salzer), 2nd ed., VCH, Weinheim, 1992,
pp. 356 – 357. However, for the analogous reactions using
branched carbonate 4, the oxidative addition reaction is too
rapid for us to isolate the h6-arene complex.
[12] DFT calculations were performed with the Gaussian 98 software
package using the B3LYP hybrid functional. The basis set used
for the geometry optimizations (VDZP) included a standard
LanL2DZ augmented with a polarization function for Ru and 4-
31G(d) for the other atoms. Single point energies were
calculated for the optimized complexes using the same func-
tional and a VTZP basis set: SDD with a polarization function
for Ru and 6-311G(d,p) for the remaining elements. The atomic
charges and Wiberg indices resulted from a Natural Population
Analysis (NPA). Computational details and the corresponding
list of references are given in the Supporting Information.
4400 ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 4397 –4400