Kinetic and density functional studies on alkyl-carbene elimination from PdII heterocylic carbene complexes: A new type of reductive elimination with clear implications for catalysis

Article abstract of DOI:10.1021/ja003861g

A number of new methyl-Pd^II complexes of heterocyclic carbenes of the form [PdMe(tmiy)L₂]BF₄ have been prepared, and their reaction behavior has been studied (tmiy = 1,3,4,5-tetramethylimidazolin-2-ylidene, L = cyclooctadiene (8), methyldiphenylphosphine (9), triphenyl phosphite (10), triphenylphosphine (11)). In common with other hydrocarbyl-M carbene complexes (M = Pd, Ni) the complexes are predisposed to a facile decomposition process. A detailed mechanism for the process and of the decomposition pathway followed is presented herein. All complexes decompose with first-order kinetics to yield 1,2,3,4,5-pentamethylimidazolium tetrafluoroborate and Pd⁰ species. The kinetic investigations combined with density functional studies show that the complexes decompose via a mechanism of concerted reductive elimination of the methyl group and carbene. The reaction represents a new type of reductive elimination from transition metals and also represents a low-energy pathway to catalyst deactivation for catalysts based on heterocyclic carbenes. The theoretical studies indicate extensive involvement of the p(π) orbital on the carbene carbon in the transition structure. Methods of stabilizing catalysts based on heterocyclic carbene complexes are suggested, and the possibility of involvement of carbene species during catalysis in ionic liquids is discussed.

Full text of DOI:10.1021/ja003861g

J. Am. Chem. Soc. 2001, 123, 4029-4040
4029
Kinetic and Density Functional Studies on Alkyl-Carbene Elimination
from PdII Heterocylic Carbene Complexes: A New Type of
Reductive Elimination with Clear Implications for Catalysis
David S. McGuinness,† Nadja Saendig,‡ Brian F. Yates,*,† and Kingsley J. Cavell*,†,§
Contribution from the School of Chemistry, UniVersity of Tasmania, GPO Box 252-75 Hobart,
Tasmania 7001, Australia, and Institut fu¨r Organische Chemie, Technische UniVersita¨t Berlin,
Strasse des 17. Juni 135, D-10623 Berlin, Germany
ReceiVed NoVember 2, 2000
Abstract: A number of new methyl-PdII complexes of heterocyclic carbenes of the form [PdMe(tmiy)L2]BF4
have been prepared, and their reaction behavior has been studied (tmiy ) 1,3,4,5-tetramethylimidazolin-2-
ylidene, L ) cyclooctadiene (8), methyldiphenylphosphine (9), triphenyl phosphite (10), triphenylphosphine
(11)). In common with other hydrocarbyl-M carbene complexes (M ) Pd, Ni) the complexes are predisposed
to a facile decomposition process. A detailed mechanism for the process and of the decomposition pathway
followed is presented herein. All complexes decompose with first-order kinetics to yield 1,2,3,4,5-
pentamethylimidazolium tetrafluoroborate and Pd0 species. The kinetic investigations combined with density
functional studies show that the complexes decompose via a mechanism of concerted reductive elimination of
the methyl group and carbene. The reaction represents a new type of reductive elimination from transition
metals and also represents a low-energy pathway to catalyst deactivation for catalysts based on heterocyclic
carbenes. The theoretical studies indicate extensive involvement of the p(π) orbital on the carbene carbon in
the transition structure. Methods of stabilizing catalysts based on heterocyclic carbene complexes are suggested,
and the possibility of involvement of carbene species during catalysis in ionic liquids is discussed.
1. Introduction
similar to that of the ubiquitous tertiary phosphine ligands,
although it is thought that the carbenes are stronger donors than
even the highly basic phosphines such as PMe3 and PCy3.17,20,24-28
Hence, the reversible dissociation that is a characteristic of the
phosphines probably does not occur for heterocyclic carbenes.
A common problem encountered with phosphine-based
catalysts is decomposition resulting from P-C bond cleavage.29
One of the most active areas of research in chemistry is the
search for new homogeneous catalysts for organic transforma-
tions. Frequently, this involves the preparation and application
of new ancillary ligands to influence the properties of metal-
based catalysts.1 In recent years there has been much interest
in the application of heterocyclic carbenes as ancillary ligands
in various transition metal-catalyzed reactions; these include
furan synthesis (Ru),2 olefin metathesis (Ru),3-8 hydrosilylation
(Rh),9-13 copolymerization (Pd),14 and Heck-type coupling
reactions (Pd, Ni).15-23 The role of the carbene is thought to be
(12) Enders, D.; Gielen, H.; Breuer, K. Tetrahedron: Asymmetry 1997,
8, 3571.
(13) Enders, D.; Gielen, H.; Runsink, J.; Breuer, K.; Brode, S.; Boehn,
K. Eur. J. Inorg. Chem. 1998, 913.
(14) Gardiner, M. G.; Herrmann, W. A.; Reisinger, C.-P.; Schwarz, J.;
Spiegler, M. J. Organomet. Chem. 1999, 572, 239.
(15) Herrmann, W. A.; Elison, M.; Fischer, J.; Ko¨cher, C.; Artus, G. R.
J. Angew. Chem., Int. Ed. Engl. 1995, 34, 2371.
(16) Herrmann, W. A.; Reisinger, C.-P.; Spiegler, M. J. Organomet.
Chem. 1998, 557, 93.
(17) Schwarz, J.; Bo¨hm, V. P. W.; Gardiner, M. G.; Grosche, M.;
Hermann, W. A.; Hieringer, W.; Raudaschl-Sieber, G. Chem. Eur. J. 2000,
6, 1773.
(18) Enders, D.; Gielen, H.; Raabe, G.; Runsink, J.; Teles, J. H. Chem.
Ber. 1996, 129, 1483.
(19) McGuinness, D. S.; Green, M. J.; Cavell, K. J.; Skelton, B. W.;
White, A. H. J. Organomet. Chem. 1998, 565, 165.
(20) McGuinness, D. S.; Cavell, K. J.; Skelton, B. W.; White, A. H.
Organometallics 1999, 18, 1596.
(21) McGuinness, D. S.; Cavell, K. J. Organometallics 2000, 19, 741.
(22) Zhang, C.; Huang, J.; Trudell, M. L.; Nolan, S. P. J. Org. Chem.
1999, 64, 3804.
† University of Tasmania.
‡ Technische Universita¨t Berlin.
§ Present address: Department of Chemistry, Cardiff University, PO Box
912, Cardiff, CF1 3TB, UK. E-mail: cavellkj@cf.ac.uk.
(1) Herrmann, W. A.; Cornils, B. Angew. Chem., Int. Ed. Engl. 1997,
36, 1048.
(2) Ku¨cu¨kbay, H.; Cetinkaya, B.; Salaheddine, G.; Dixneuf, P. H.
Organometallics 1996, 15, 2434.
(3) Weskamp, T.; Schattenmann, W. C.; Spiegler, M.; Herrmann, W.
A. Angew. Chem., Int. Ed. 1998, 37, 2490.
(4) Weskamp, T.; Kohl, F. J.; Hieringer, W.; Gleich, D.; Herrmann, W.
A. Angew. Chem., Int. Ed. 1999, 38, 2146.
(5) Ackermann, L.; Fu¨rstner, A.; Weskamp, T.; Kohl, F. J.; Herrmann,
W. A. Tetrahedron Lett. 1999, 40, 4787.
(6) Huang, J.; Stevens, E. D.; Nolan, S. P.; Petersen, J. L. J. Am. Chem.
Soc. 1999, 121, 2674.
(7) Huang, J.; Schanz, H.-J.; Stevens, E. D.; Nolan, S. P. Organometallics
1999, 18, 5375.
(8) Jafarpour, L.; Schanz, H.-J.; Stevens, E. D.; Nolan, S. P. Organo-
metallics 1999, 18, 5416.
(9) Hill, J. E.; Nile, T. A. J. Organomet. Chem. 1977, 137, 293.
(10) Lappert, M. F.; Maskell, R. K. J. Organomet. Chem. 1984, 264,
217.
(11) Herrmann, W. A.; Goossen, L. J.; Ko¨cher, C.; Artus, G. R. J. Angew.
Chem., Int. Ed. Engl. 1996, 35, 2805.
(23) Calo´, V.; Del Sole, R.; Nacci, A.; Schingaro, E.; Scordari, F. Eur.
J. Org. Chem. 2000, 869.
(24) Lappert, M. F. J. Organomet. Chem. 1975, 100, 139.
(25) Lappert, M. F.; Pye, P. L. J. C. S. Dalton 1977, 2172.
(26) Lappert, M. F. J. Organomet. Chem. 1988, 358, 185.
(27) O¨ fele, K.; Herrmann, W. A.; Mihalios, D.; Elison, M.; Herdtweck,
E.; Scherer, W.; Mink, J. J. Organomet. Chem. 1993, 459, 177.
(28) Huang, J.; Schanz, H.-J.; Stevens, E. D.; Nolan, S. P. Organome-
tallics 1999, 18, 2370.
10.1021/ja003861g CCC: $20.00 © 2001 American Chemical Society
Published on Web 04/06/2001

4030 J. Am. Chem. Soc., Vol. 123, No. 17, 2001
McGuinness et al.
Accordingly, it is often necessary to use a large excess of the
ligand relative to the metal in catalyst formulations. In contrast,
one of the reasons for the success of heterocyclic carbenes in
catalysis is thought to be the high stability of the subsequent
complexes, which results from the strength of the M-carbene
bond. However, we have previously discovered a facile mode
by which the M-carbene bond in Pd complexes may be broken,
leading to decomposition.19,20,30 In particular, decomposition is
observed when the Pd complexes also contain a hydrocarbyl
group, and the process is characterized by the elimination of a
hydrocarbylimidazolium salt and M0 species (Reaction 1). This
presents a potentially serious drawback in the application of
complexes of heterocyclic carbenes in catalysis reactions
involving hydrocarbon substrates.
Scheme 1
Scheme 2. Possible Mechanisms for Elimination of
Alkyl-imidazolium Salts from Pd-Carbene Complexes
The decomposition reaction was first observed when the
thermal behavior of [PdMe(dmiy)(cod)]BF4 (1) was monitored
(dmiy ) 1,3-dimethylimidazolin-2-ylidene).19 Unexpectedly, it
was found that the cod ligand simply dissociates during
decomposition, and the sole products of the reaction are 1,2,3-
trimethylimidazolium tetrafluoroborate and Pd0 (Reaction 2).
acyl-imidazolium salt from Pd during carbonylation studies of
a Me-Pd carbene complex.30
In subsequent mechanistic studies on the Heck reaction we found
that neutral aryl PdII complexes 2 and 3 are also prone to
decompose via elimination of imidazolium salts (Reaction 3).20
Several mechanisms by which this mode of decomposition
occurs may be envisaged (Scheme 2). For example, migratory
insertion of the carbene into the Pd-R bond (path a) would
give a complex in which the carbene ligand has (formally)
become an anionic alkyl ligand. Examples of migration of
Fischer and Schrock carbenes into both alkyl- and hydride-
metal bonds are common.31-41 Furthermore, the carbene moiety
is isoelectronic with isocyanide, for which insertion into the
Pd-Me bond is known.42 In the case of R being an alkyl group
the next step in decomposition could involve â-hydride elimina-
tion to give a complex containing a coordinated enediamine
fragment and a hydride ligand (path b). From this complex
reductive elimination of H+, aided by the basic enediamine
ligand,43 would give the imidazolium salt and Pd0 (path c). When
R is an aryl or acyl group, then a pathway involving â-hydride
elimination is not available. In this case (and also possibly in
Moreover, it was found that halide abstraction from 3, to give
a cationic complex, resulted in a much higher rate of decom-
position when compared to the neutral complex. Ensuing studies
into the Heck reaction showed that at lower temperatures the
reaction is competitive with â-hydride elimination; halide
abstraction from 3 in the presence of n-butylacrylate gave both
the imidazolium salt 4 and cinnamate 5, presumably through
insertion intermediate 6 (Scheme 1). Significantly, the elimina-
tion reaction has also been observed for the NiII complex
Ni(Me)(tmiy)2I (7, tmiy ) 1,3,4,5-tetramethylimidazolin-2-
ylidene), which could not be isolated pure because of the ease
at which it eliminated the pentamethylimidazolium cation even
at -40 °C (Reaction 4).20
(29) Garrou, P. E. Chem. ReV. 1985, 85, 171.
(30) McGuinness, D. S.; Cavell, K. J. Organometallics 2000, 19, 4918.
(31) Threlkel, R. S.; Bercaw, J. E. J. Am. Chem. Soc. 1981, 103, 2650.
(32) Jernakoff, P.; Cooper, N. J. J. Am. Chem. Soc. 1984, 106, 3026.
(33) Green, J. C.; Green, M. L. H.; Morley, C. P. Organometallics 1985,
4, 1302.
(34) Osborn, V. A.; Parker, C. A.; Winter, M. J. J. Chem. Soc., Chem.
Commun. 1986, 1185.
(35) Ziegler, T.; Versluis, L.; Vincenzo, T. J. Am. Chem. Soc. 1986,
108, 612.
(36) Carter, E. A.; Goddard, W. A., III. J. Am. Chem. Soc. 1987, 109,
579.
(37) Carter, E. A.; Goddard, W. A., III. Organometallics 1988, 7, 675.
(38) Winter, M. J. Polyhedron 1989, 8, 1583.
(39) Winter, M. J.; Woodward, S. J. Organomet. Chem. 1989, 361, C18.
(40) Davey, C. E.; Devonshire, R.; Winter, M. J. Polyhedron 1989, 8,
1863.
(41) Fischer, H.; Jungklaus, H. J. Organomet. Chem. 1999, 572, 105.
(42) Yamamoto, Y.; Yamazaki, H. Bull. Chem. Soc. Jpn. 1970, 43, 3.
(43) Kuhn, N.; Bohnen, H.; Henkel, G.; Kreutzberg, J. Z. Naturforsch.
1996, 51 b, 1267.
Furthermore,we have recently reported the elimination of an

Alkyl-Carbene Elimination from Pd-Carbene Complexes
J. Am. Chem. Soc., Vol. 123, No. 17, 2001 4031
Scheme 3. Synthesis of Me-Pd Carbene Complexesa
the case where R ) alkyl group) reductive elimination of the
alkyl ligand that results from the initial insertion would yield
the observed decomposition products (path d).
N-heterocylic carbenes bind with metals in a manner different
to that observed for Fischer and Schrock carbenes. As a
consequence of electron density donated from the two N atoms,
heterocyclic carbenes act as pure σ-donors, with little if any of
the π-back-bonding that occurs in other classes of carbene
complexes.44,45 Consistent with this view, M-Ccarbene bond
lengths for heterocyclic carbenes lie in the range for normal
M-C(sp3/sp2) single bonds.15,20 In light of this observation it
would seem not unreasonable that the M-Ccarbene bond could
show aspects of chemical reactivity similar to other M-C single
bonds. Thus, the decomposition of alkyl, aryl, or acyl Pd carbene
complexes could arguably occur via a concerted reductive
elimination to yield in one step the imidazolium ion and Pd0
(path e) in an analogous manner to the well-known reductive
elimination of R-R′ and R-X from transition metals.46,47
One of the most important goals of a catalyst design study is
to understand the mechanism by which the catalytic reaction
occurs, allowing a logical approach to the design process.
Likewise, understanding how an active catalyst is deactivated
may allow the deactivation process to be controlled. The
decomposition reaction we have described above appears to be
a common mode of carbene loss for hydrocarbyl-carbene
complexes and indicates a significant pathway for catalyst
deactivation. Hydrocarbyl species are expected to be intermedi-
ates in many catalytic reactions including those carried out with
transition metal-carbene complexes (for example Heck-type
coupling,15-23 ethylene-CO copolymerization14 and ethylene
polymerization48,49). It was thus of considerable interest to us
to elucidate the detailed mechanism for the reaction such that
steps can be taken to limit its occurrence. We report here further
details of the reaction together with combined kinetic and
theoretical studies aimed at deducing the mechanism of the
reaction.
a i) AgBF4, cod; ii) AgBF4, 2 PMePh2; iii) AgBF4, 2 P(OPh)3; iv)
AgBF4, 2 P(Ph)3.
and 0.8 ppm (J ) 33 Hz). Complexes 10 and 11 could also be
prepared via addition of 2 equiv of the appropriate phosphorus
ligand to 8. 1H NMR spectroscopy showed the clean formation
of the phosphite/phosphine complex along with free cyclo-
octadiene.
2.2. Kinetic Studies. On warming, complexes 8-11 were
found to decompose cleanly to the 1,2,3,4,5-pentamethyl
imidazolium ion with rates measurable by 1H NMR. The
imidazolium cation exhibits signals at 3.64, 2.62, and 2.21 ppm
for the N-CH3, C2-CH3, and backbone CH3 protons, respec-
tively. During the decomposition of complexes 9-11, bis-
(phosphite/phosphine) Pd0 complexes, Pd(PMePh2)2, Pd-
[P(OPh)3], and Pd(PPh3)2,50 form initially and then slowly
decompose to give Pd deposits (Reaction 5).
2. Results and Discussion
2.1. Synthesis. To study the kinetics of the decompositon
reaction a number of complexes (8-11) were prepared via halide
19
The 1H NMR spectrum of decomposing 10 shows the formation
of pentamethylimdazolium accompanied by extra signals at 6.7-
7.2 ppm, arising from the formation of Pd[P(OPh)3]2. In the
31P NMR spectrum of 10 the two doublets are replaced by a
singlet at 141.0 ppm, due to Pd[P(OPh)3]2. For complex 11 these
extra peaks appear at 7.60-7.85 ppm in the 1H NMR spectrum
and as a singlet a 35.4 ppm in the 31P spectrum, resulting from
Pd(PPh3)2. Complex 8 decomposes to give metallic Pd and free
cyclooctadiene.
abstraction from the dimer [PdMeCl(tmiy)]2 in the presence
of coordinating ligands (Scheme 3). Complexes have been
1
characterized by H-, 13C-, and 31P NMR spectroscopy, mass
spectroscopy and elemental analysis with the exception of 11,
the instability of which prevented elemental analysis. Conse-
quently, a high-resolution MS has been obtained for 11.
A cis carbene/methyl group arrangement for complexes 9-11
is expected from the relative trans influences of the carbene
versus the phosphite/phosphine, the carbene being a stronger
σ-donor than the phosphorus ligands. The 31P spectra of 9-11
confirm this viewpoint. The 31P NMR spectrum of complex 10
contains two doublets at 120.3 and 113.4 ppm with a P-P
coupling constant of 72 Hz, showing the inequivalence of each
phosphite ligand. Complex 11 shows two doublets at 34.2 and
21.4 ppm (J ) 30 Hz), and complex 9, two doublets at 12.3
The rates of decomposition of 8-11 were obtained by
monitoring the disappearance of the N-CH3 or C-CH3 signals
1
in the H NMR spectrum at 20 °C. In all cases linear first-
order rate plots were obtained (Table 1). Rate constants were
also measured for the decomposition of 10 and 11 in the
presence of 1 equiv of COD. This was achieved by adding 2
equiv of the respective phosphorus ligand to an NMR solution
of 8. Table 1 shows that the rates of decomposition followed
the order: complex 9 < 8 < 10 < 11. The presence of 1 equiv
of COD slows the decomposition of complex 11 somewhat.
Attempts to make the tricyclohexylphosphine (PCy3) analogue
12 using the same technique failed. NMR studies of the product
solution showed that it contained pentamethylimidazolium
(44) Fro¨hlich, N.; Pidun, U.; Stahl, M.; Frenking, G. Organometallics
1997, 16, 442.
(45) Boehme, C.; Frenking, G. Organometallics 1998, 17, 5801.
(46) Stille, J. K. In The Chemistry of the Metal Carbon Bond; Hartley,
F. R., Patai, S., Eds.; Wiley-Interscience, 1985; Vol. 2, p 742.
(47) Collman, J. P.; Hegedus, L. S.; Norton, J. R.; Finke, R. G. Principles
and Applications of Organotransition Metal Chemistry; University Science
Books: Mill Valley, CA, 1987.
(48) Do¨hring, A.; Go¨hre, J.; Jolly, P. W.; Kryger, B.; Rust, J.; Verhovnik,
G. P. J. Organometallics 2000, 19, 388.
(49) McGuinness, D. S.; Cavell, K. J. Unpublished results.
(50) Urata, H.; Suzuki, H.; Moro-oka, Y.; Ikawa, T. J. Organomet. Chem.
1989, 364, 235.

4032 J. Am. Chem. Soc., Vol. 123, No. 17, 2001
Table 1. Decomposition Kinetics for Complexes 8-12
McGuinness et al.
complex
k/10-4 s-1 a
t1/2/103 s θ (deg)b
pKab
8
9
10
0.71 ( 0.03 (0.995)
0.518 ( 0.009 (0.998)
1.08 ( 0.02 (0.999)
9.8
13.4
6.42
6.86
1.1
1.54
136
128
4.57
-2.00
10 + COD 1.01 ( 0.04 (0.993)
11 6.6 ( 0.2 (0.995)
11 + COD 4.49 ( 0.09 (0.999)
12c
too fast to measure
145
2.73
170
9.70
a Correlation coefficient in parentheses. b From ref 52. c Complex
12 not isolated due to rapid decomposition, see text.
tetrafluoroborate only, along with signals between 1 and 2 ppm
attributed to PCy3. We conclude that 12 forms initially but
decomposes too rapidly to allow isolation of the complex. Thus,
it seems that PCy3 promotes very rapid decomposition. Further
qualitative information on the rate effect of different phosphines
is obtained from [PdMe(dmiy)(dppp)]BF4 13 (dppp ) 1,3-bis-
(diphenylphosphino)propane).51 A solution of this complex
failed to show any decomposition even after 24 h when
Figure 1. Arrhenius plot for the decomposition of complex 10.
The kinetics of the reaction could be accounted for by
assuming an alkyl migration (path a, Scheme 2), which is
followed by rapid further reactions (path d or b,c) to give the
products. If this is the case, then migratory insertion must be
the rate-determining step, giving a low steady-state concentration
of the insertion intermediatesno intermediate in which the
methyl group has migrated to the carbene has been observed.
Consistent with this idea, â-hydride eliminations are known to
be facile, and it is likely that reductive elimination following
either path c or d will also be fast.
1
monitored by H NMR spectroscopy.
Insertion reactions (of for example, CO and ethylene) are
thought to proceed via opening of the angle between the cis
nonparticipatory ligands with concomitant closing of the angle
between the reacting groups. Chelating ligands can slow the
insertion process by locking-in a rigid bite angle.53-55 The
remarkable rate suppression observed when the dppp ligand is
bound to Pd is therefore consistent with an insertion process,
as is the effect of bulky phosphines, which can cause the angle
between the cis phosphine ligands to open with a consequent
rate increase. However, previous studies into CO and ethylene
insertion have shown that migration of the alkyl group is
energetically favored by ligands with a high trans influence
located in the trans position to the migrating alkyl group53 (i.e.,
the rate of decomposition should increase with strong donor
ligands for a mechanism involving alkyl migration). Therefore,
in contradiction to the previous arguments, which conform to
an insertion process, we find a strong donor ligand stabilizes
the carbene complexes to an extent that does not fit a migratory
insertion mechanism.
Moreover, aspects of the kinetics are consistent with a
concerted reductive elimination (path e, Scheme 2), as first-
order rate constants are observed for elimination from dialkyl
PdII, NiII, and AuIII complexes.47,56-60 For this type of reaction
it is known that leaving groups with a strong σ-donor capability
favor reductive elimination;59 hence, the σ-donating ability of
heterocyclic carbenes could likewise favor reductive elimination.
It has also been noted that positive charge on the metal increases
the rate of reductive elimination,47 which is in agreement with
our observations that cationic carbene complexes eliminate
Table 1 also lists the steric bulk of the different phosphorus
ligands, expressed as the cone angle θ, and the basicity of the
ligands, expressed as the pKa of the conjugate acid,52 which
gives a good measure of the ability of the ligand to donate σ
electrons to a metal. It is apparent that the rate of reaction does
not correlate well with ligand basicity. The most basic ligand
used, PCy3, results in the fastest decomposition, while the second
most basic ligand, PMePh2, gives the slowest rate. If the cone
angle of the phosphines is considered, it appears that greater
steric bulk favors elimination; however complex 9, containing
the phosphine PMePh2 with a cone angle of 136°, decomposes
more slowly than complex 10, which contains P(OPh3) with a
cone angle of 128°. It is manifest that ligand basicity and steric
bulk have opposing influences on the rate of decomposition.
Bulky phosphines clearly destabilize the complexes, but the
observed slow rate of decomposition of 9 indicates that a
strongly donating phosphine can stabilize the complex, providing
the steric bulk of the ligand is not too great. The stability of
complex 13, which contains dppp, suggests that chelation
strongly retards the reaction. The relatively slow rate of
decomposition of the COD complex 8 is consistent with this
premise.
The effect of excess ligand on the kinetics was also studied.
The rate of decomposition of 10 on the presence of 0.3, 0.5,
and 1.7 equiv of excess P(OPh)3 was found to be the same
within experimental error. Thus, the rate is unaffected by excess
phosphite, and it seems a mechanism involving prior dissociation
of a phosphorus ligand in unlikely, vide infra.
(53) Cavell, K. J. Coord. Chem. ReV. 1996, 155, 209.
(54) Thorn, D. L.; Hoffmann, R. J. Am. Chem. Soc. 1978, 100, 2079.
(55) Dekker: G. P. C. M.; Elsevier: C. J.; Vrieze, K.; van Leeuwen, P.
W. N. M. Organometallics 1992, 11, 1598.
(56) Komiya, S.; Albright, T. A.; Hoffmann, R.; Kochi, J. K. J. Am.
Chem. Soc. 1976, 98, 7255.
Rate constants for the decomposition of 10 at 20, 25, 30, 35,
and 40 °C were obtained which gave a linear Arrhenius plot as
shown in Figure 1. From this an activation barrier (∆Hq) of
23.5 ( 1.3 kcal mol-1 was calculated along with an entropy of
activation (∆Sq) of 4.3 ( 0.3 kcal mol-1 K-1
.
(57) Tatsumi, K.; Hoffmann, R.; Yamamoto, A.; Stille, J. K. Bull. Chem.
Soc. Jpn. 1981, 54, 1857.
(51) Green, M. J.; Cavell, K. J.; Skelton, B. W.; White, A. H. J.
Organomet. Chem. 1998, 554, 175.
(52) Matiur Rahman, M.; Liu, H.-Y.; Ericks, K.; Prock, A.; Giering, W.
P. Organometallics 1989, 8, 1.
(58) Gillie, A.; Stille, J. K. J. Am. Chem. Soc. 1980, 102, 4933.
(59) Moravskiy, A.; Stille, J. K. J. Am. Chem. Soc. 1981, 103, 4182.
(60) Ozawa, F.; Ito, T.; Nakamura, Y.; Yamamoto, A. Bull. Chem. Soc.
Jpn. 1981, 54, 1868.

Alkyl-Carbene Elimination from Pd-Carbene Complexes
J. Am. Chem. Soc., Vol. 123, No. 17, 2001 4033
imidazolium more easily than their neutral counterparts. Coupled
with this, and consistent with our observations, is the report
that strong σ-donor phosphorus ligands inhibit reductive elimi-
nation from PdR2L2.60
calculations. For the optimized geometries, harmonic vibrational
frequencies were calculated at the B3LYP level and zero-point
vibrational energy corrections obtained using unscaled frequen-
cies. The vibrational frequencies were also used to obtain
thermodynamic corrections and entropies. All transition struc-
tures possessed one and only one imaginary frequency, and they
were further characterized by following the corresponding
normal mode toward each product and reactant. Single-point
energies on B3LYP/LANL2DZ optimized geometries were
calculated at the B3LYP level with the LANL2augmented:
6-311+G(2d,p) basis set,79 which incorporates the LANL2
effective core potential and the large f-polarized valence basis
set of Bauschlicher and co-workers80 on palladium together with
the 6-311+G(2d,p) basis set76-78 on all other atoms. Calculations
on the large triphenyl phosphite system were carried out with
the ONIOM81-83 method in which the high layer (at B3LYP/
LANL2DZ) employed PH3 and the low layer (at RHF/
LANL2MB) employed P(OPh)3. All calculations were carried
out with the Gaussian 9884 program.
However, for reductive elimination from dialkyl Pd com-
plexes, PdR2L2, it has been shown that a dissociative pathway
is preferred such that elimination occurs from an intermediate
with either a T or Y geometry generated after dissociation of a
ligand, L.57-60 Thus, addition of excess ligand (usually phos-
phine) severely inhibits reductive elimination. This is in contrast
to the decomposition reaction of the present study, which
remains unaffected by excess phosphite. Nonetheless, at least
one study has shown that reductive elimination from four
coordinate PdR2L2 is possible when one of the R groups is
styryl.61 It is worth noting in this context that the carbene carbon
is sp2 hybridized with a formally empty but in reality appreciably
occupied p-orbital perpendicular to the plane of the carbene (vide
infra)44,45,62 and in this respect it has similarities to a phenyl or
alkenyl ligand.63 Theoretical studies have shown that reductive
elimination from four coordinate Pd is possible.57 The same
study showed that reductive elimination from a four-coordinate
species is accompanied by an increase in the angle between
the two nonparticipatory ligands as the reaction proceeds. This
would help account for the rate enhancement afforded by more
bulky phosphines, as well as accounting for the strong stabilizing
effect that the dppp ligand has on the carbene complex 13.
Furthermore, several theoretical studies64-66 have shown what
is known from experiment,67-69 that cis chelating ligands result
in a more exothermic oxidative addition, or stated differently,
a chelating ligand will result in a more endothermic reductiVe
elimination.
2.3.2. Calculations on a Model System. As the kinetics of
the reaction were in part consistent with both a concerted
reductive elimination and a multistep insertion process, the
reaction was modeled theoretically using [PdMe(dmiy)(PH3)2]+
14 as the model complex, giving 1,2,3-trimethylimidazolium
and Pd(PH3)2 as decomposition products (Reaction 6).
2.3. Theoretical Studies. 2.3.1. Computational Methods.
Full geometry optimizations were carried out with the use of
the B3LYP70-72 density functional level of theory combined
with two different basis sets: the LANL2DZ basis set (which
incorporates the Hay and Wadt73 small-core relativistic effective
core potential and double-ú valence basis set on palladium and
phosphorus together with the Dunning/Huzinaga74 double-ú
basis set on other atoms), and the DSP basis set (which
incorporates the small-core relativistic effective core potential
and triple-ú valence basis set from the Stuttgart group of Dolg,
Stoll, Preuss, and co-workers75 on palladium together with the
6-311G(d,p)76-78 basis set on other atoms). Sets of five
d-functions were used in the basis sets throughout these
The PES was initially probed for intermediates in both a
multistep migratory process and a concerted reductive elimina-
tion. We were unable to find a pathway that corresponded to
alkyl migration on the PES, however stationary points corre-
sponding to reductive elimination were found. Figure 2 shows
the optimized geometries of the reactant 14, the transition
structure 14-15, the post reductive elimination encounter
complex 15, and the products. In Table 2 the calculated gas-
phase bonding parameters of 14 have been compared with the
experimental values of some Me-Pd carbene complexes;
PdMeCl(dmiy)2,19 PdMeCl(tmiy)2 and PdMe(dmiy)(NC5H4-
85
CO2-2).51 It may be seen that the agreement between experiment
(78) Frisch, M. J.; Pople, J. A.; Binkley, J. S. J. Chem. Phys. 1984, 80,
3265.
(79) Frankcombe, K. E.; Cavell, K. J.; Yates, B. F.; Knott, R. B. J. Phys.
Chem. 1996, 100, 18363.
(61) Loar, M. K.; Stille, J. K. J. Am. Chem. Soc. 1981, 103, 4174.
(62) Heinemann, C.; Mu¨ller, T.; Apeloig, Y.; Schwarz, H. J. Am. Chem.
Soc. 1996, 118, 2023.
(63) Low, J. J.; Goddard W. A., I. J. Am. Chem. Soc. 1986, 108, 6115.
(64) Sakaki, S.; Biswas, B.; Sugimoto, M. J. Chem. Soc., Dalton Trans.
1997, 803.
(65) Sakaki, S.; Biswas, B.; Sugimoto, M. Organometallics 1998, 17, 7,
1278.
(80) Langhoff, S. R.; Petterson, L. G.; Bauschlicher, C. W.; Partridge,
H. J. J. Chem. Phys. 1987, 86, 268.
(81) Maseras, F.; Morokuma, K. J. Comput. Chem. 1995, 16, 1170.
(82) Svensson, M.; Humbel, S.; Froese, R. D. J.; Matsubura, T.; Sieber,
S.; Morokuma, K. J. Chem. Phys. 1996, 100, 19357.
(66) Su, M.-D.; Chu, S.-Y. Inorg. Chem. 1998, 37, 3400.
(67) Hackett, M.; Ibers, J. A.; Jernakoff, P.; Whitesides, G. M. J. Am.
Chem. Soc. 1986, 108, 8094.
(68) Hackett, M.; Ibers, J. A.; Whitesides, G. M. J. Am. Chem. Soc. 1988,
110, 1436.
(69) Hackett, M.; Whitesides, G. M. J. Am. Chem. Soc. 1988, 110, 1449.
(70) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(71) Stephens, P. J.; Devlin, J. F.; Chabalowski, C. F.; Frisch, M. J. J.
Phys. Chem. 1994, 98, 11623.
(72) Hertwig, R. H.; Koch, W. Chem. Phys. Lett. 1997, 268, 345.
(73) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299.
(74) Dunning, T. H.; Hay, P. J. In Modern Theoretical Chemistry;
Schaefer, H. F., Ed.; Plenum: New York, 1976; Vol. 3, p 1.
(75) Andrae, D.; Haeussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor.
Chim. Acta 1990, 77, 123.
(83) Dapprich, S.; Komaromi, I.; Byun, K. S.; Morokuma, K.; Frisch,
M. J. J. Mol. Struct. (THEOCHEM) 1999, 462, 1.
(84) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A.
D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Otcherski, J.; Petersson, G. A.; Ayala, P. J.; Cui, Q.; Morokuma, K.; Malick,
D. K.; Rabuk, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.;
Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi,
I.; Gomberts, R.; Martin, R. L.; Fox, D. J.; Kieth, T.; Al-Laham, M. A.;
Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M.
W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon,
M.; Replogle, E. S.; Pople, J. A. Gaussian 98, rev. A.1.; Gaussian, Inc:
Pittsburgh, PA, 1998.
(76) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys.
1980, 72, 650.
(77) McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72, 5639.
(85) Magill, A. M.; McGuinness, D. S.; Cavell, K. J.; Britovsek, G. J.
P.; Gibson, V. C.; White, A. J. P.; Williams, D. J.; White, A. H.; Skelton,
B. W. J. Organomet. Chem 2001, 617-618, 546.

4034 J. Am. Chem. Soc., Vol. 123, No. 17, 2001
McGuinness et al.
Figure 2. Optimized geometries (B3LYP/LANL2DZ) corresponding to reductive elimination from complex 14 (left) and complexes 16 and 17
(right). Bond lengths are given in Å and angles in degrees, values in parentheses represent B3LYP/DSP optimized geometries.
and theory is in most cases excellent. Furthermore, Table 2
shows that the geometries obtained with both the LANL2DZ
and DSP basis sets are almost the same. Table 3 compares the
calculated geometry of 1,2,3-trimethylimidazolium with the
experimental geometries of 1,2-dimethyl-3-propylimidazolium
chloride (dmpim)Cl86 and pentamethylimidazolium iodide (pmi-
m)I,87 and again the level of agreement between experiment
and theory, and the two basis sets is excellent.
Reductive elimination with prior dissociation of one phos-
phorus ligand was modeled by removing one phosphine from
the complex. The optimized geometries of the stationary points
corresponding to this pathway are also shown in Figure 2. Two
(86) Scordilis-Kelly, C.; Robinson, K. D.; Belmore, K. A.; Atwood, J.
L.; Carlin, R. T. J. Crystallogr. Spectrosc. Res. 1993, 23, 601.
(87) Kuhn, N.; Henkel, G.; Kreutzberg, J. Z. Naturforsch. 1991, 46b,
1706.

Alkyl-Carbene Elimination from Pd-Carbene Complexes
J. Am. Chem. Soc., Vol. 123, No. 17, 2001 4035
Table 2. Comparison of Calculated Bond Lengths (Å) and Bond Angles (deg) in PdMe(dmiy)(PH3)2 14 with Experimental Values of Several
Me-Pd Carbene Complexes19,85,51
parameter
13 (DSP)a
13 (LANL2DZ)b
PdMeCl(dmiy)2
PdMeCl(tmiy)2
PdMe(dmiy) (NC5H4CO2-2)
Pd-Ccarbene
N-Ccarbene
N-CH3
2.06
1.35
1.46
2.04
1.38
1.47
1.999(7), 2.009(8)
1.36(1), 1.369(9), 1.345(9)
1.48(1), 1.46(1), 1.45(1)
1.33(1), 1.32(1)
2.033(4)
1.345(5), 1.347(5)
1.456(7), 1.472(6)
1.971(2)
1.347(3), 1.348(3)
1.456(4), 1.461(4)
1.339(4)
CdC
1.35
1.37
N-C-N
Cc-N-CH3
Cc-N-C
105.2
125.5
110.6
105.0
125.2
110.7
102.1(6), 103.2(6)
122.9(7)-123.8(6)
111.5(6)-112.5(7)
103.1(3)
123.7(3), 124.5(3)
104.6(2)
124.7(2), 124.0(2)
110.9(2)
a B3LYP/DSP optimization. b B3LYP/LANL2DZ optimization.
Figure 3. Potential energy profile for reductive elimination from four-coordinate Pd (solid line) and with prior phosphine dissociation (dashed
line). Energies are from B3LYP/augmented LANL2DZ:6-311+G(2d,p)//B3LYP/LANL2DZ calculations. Values in parentheses represent B3LYP/
DSP energies, and those in brackets, B3LYP/LANL2DZ energies, see text.
Table 3. Comparison of Calculated Bond Lengths (Å) and Bond
Angles (deg) in 1,2,3-trimethylimidazolium with Experimental
Values of (dmpim)Cl86 and (tmim)I87
which was also predicted by Hoffmann et al. using extended
Hu¨ckel calculations on PdR2L.57
Figure 3 shows the potential energy profile for the elimination
from both complex 14 and the three coordinate complexes 16
and 17. All energies are relative to the starting complex 14,
and in the case of elimination with phosphine dissociation, the
energies given are with free PH3 added. The energies in
parentheses are from B3LYP/DSP calculations and those in
brackets represent B3LYP/LANL2DZ energies. Figure 3 shows
that the level of agreement is in general quite good, and in
further discussion the B3LYP/augmented LANL2DZ:6-311+G-
(2d,p) energies will be used. Reductive elimination from 14 is
predicted to occur with an enthalpy of activation of 22.3 kcal
mol-1, which seems very reasonable given the experimental
value of 23.5 kcal mol-1 for the triphenyl phosphite complex
in solution.88 The reaction is predicted to be slightly exothermic,
the products lying 3.7 kcal mol-1 below the reactant.
parameter DSPa LANL2DZb
(dmpim)Cl
(tmim)I
C-CH3
N-C
N-CH3
CdC
1.48
1.35
1.47
1.36
1.49
1.36
1.48
1.37
1.486(6)
1.339(5), 1.334(6)
1.488(6)
1.48(2)
1.32(2), 1.35(2)
1.47(2), 1.50(2)
1.38(2)
1.327(6)
N-C-N
107.4
107.2
125.6
109.3
107.1(4)
125.8(4)
109.4(4)
106.2(11)
124.0(11), 125.8(11)
110.5(10), 112.2(10)
C-N-CH3 125.5
C-N-C
109.2
a B3LYP/DSP optimization. b B3LYP/LANL2DZ optimization.
isomeric starting complexes were found, corresponding to
phosphine dissociation trans to the methyl group (16) and
dissociation of the phosphine trans to the carbene (17). It is
expected that there would be a small barrier to interconversion
of 16 and 17, although we did not attempt to find this. Only
one transition structure, TS(16,17-18), was found, which leads
to the post reductive elimination encounter complex 18.
Presumably complex 16 proceeds to the transition structure
through 17, which is higher in energy (vide infra). Reductive
elimination is predicted to occur from a T shaped geometry,
Dissociation of a phosphine not surprisingly, is an endother-
mic process. It takes 26.8 kcal mol-1 for the phosphine trans
(88) As noted by a reviewer, the theoretical ∆Hq and ∆Sq values are
calculated for the gas-phase and do not take into account the solvation
effects.

4036 J. Am. Chem. Soc., Vol. 123, No. 17, 2001
McGuinness et al.
to the carbene to dissociate compared to 20.8 kcal mol-1 for
that trans to the Me group. The methyl group therefore has a
higher trans effect than the carbene, and this is reflected in the
differing Pd-P bond lengths in 14. The energy required to
remove a phosphine makes this pathway higher in energy along
the complete potential energy profile, relative to the reaction
without prior dissociation. The activation barrier, ∆Hq, for TS-
(16,17-18) is predicted to be 35.1 kcal mol-1, which is 12.8
kcal mol-1 higher than that with two coordinated phosphines,
TS(14-15). The encounter complex 18 is similarly higher in
energy, this time 24.8 kcal mol-1 higher than 15. It is also
interesting to note the much closer approach of the imidazolium
ion in 18 compared to 15 which is stabilized by an additional
phosphine ligand (Figure 2). The important conclusion from
the potential energy profile is that reductive elimination occurs
from the four coordinate complex 14, and that prior dissociation
of a phosphine is neither required nor favorable. This outcome
is in accord with the kinetic results.
Geometry changes during the reaction include a decrease in
the Me-Pd-Ccarbene angle from 85.7° in 14 to 52.0° in TS-
(14-15) and a concomitant but smaller increase in the P-Pd-P
angle from 95.3° to 98.3°. The Pd-Me distance increases from
2.09 to 2.29 Å while the Pd-Ccarbene distance barely changes
in the transition structure. These changes indicate an early,
reactant like transition structure. The Me-Ccarbene distance in
the transition structure is 1.90 Å which decreases to 1.49 Å in
the encounter complex, and in the free imidazolium cation.
During reductive elimination the P-Pd-P angle increases from
that in the transition structure (98.3°) to 166.0° in the encounter
complex, and to 179.9° in the product, Pd(PH3)2. These changes
are very similar to those predicted for conventional reductive
elimination,64-66 and indicate that a chelating ligand may well
hinder the reaction to an extent, as found experimentally.
Figure 4. Optimized geometries corresponding to reductive elimination
from complex 19.
2.3.3. Calculations on the Real System. Using PH3 as
spectator ligand in the calculations was useful for expediting
the elucidation of possible reaction pathways. However this
ligand differs from tertiary phosphines/phosphites both in terms
of steric and electronic effects. To verify that the same
mechanism is likely when ‘real’ ligands are coordinated, the
energetics of reductive elimination from PdMe(dmiy){P(OPh)3}2
19 were calculated. The only difference between 19 and the
isolated complex 10, for which activation parameters have been
experimentally obtained, is the presence of methyl groups in
the C4,5 positions of the carbene, which is not expected to
greatly influence the energetics of the reaction.
The optimized geometries of 19, the transition structure TS-
(19-20), the encounter complex 20 and the product, Pd-
{P(OPh)3}2 are shown in Figure 4. The increased steric bulk of
P(OPh)3 results in an increased P-Pd-P angle (103.5°) in 19
relative to that in complex 14 (95.3°). This is accompanied by
a decrease in the Me-Pd-carbene angle (81.9° compared to
85.7°). In the transition structure TS(19-20) the P-Pd-P angle
increases to 114.1°, an increase of 10.6° from that of the initial
complex. This increase is greater than that which occurs in going
from 14 to TS(14-15). Despite this, the positions of the Me
and carbene groups in the transition structure are very similar
in both TS(14-15) and TS(19-20). The Pd-Me, Pd-Ccarbene and
Me-Ccarbene distances in TS(19-20) are 2.31, 2.04 and 1.97 Å
respectively, compared to distances of 2.29, 2.03 and 1.90 Å in
TS(14-15).
Figure 5. Potential energy profile for reductive elimination from
complex 19.
cal mol-1 K-1 compares well with the experimental value of
4.3 ( 0.3 cal mol-1 K-1. Compared to the model system, the
reaction is predicted to be more exothermic with an enthalpy
of reaction of -9.2 kcal mol-1. The differences in geometry
between the reactant and transition structure for complexes 14
and 19 demonstrate that the more sterically demanding phos-
phine ligand, P(OPh)3, results in a greater P-Pd-P angle in
both the reactant and transition structure, thus favoring the
approach of Me and carbene in the transition structure, and
lowering the activation energy as predicted theoretically and
found experimentally.
A potential energy profile for the reaction is shown in Figure
5. The activation barrier is predicted to be 14.1 kcal mol-1
,
somewhat lower than the experimental value found for complex
2.3.4. Bonding in the Reactant, Transition State, and
Encounter Complex. The bonding situation in the reactant 14,
10 (23.5 ( 1.3 kcal mol-1). The theoretical ∆Sq value of 3.8

Alkyl-Carbene Elimination from Pd-Carbene Complexes
J. Am. Chem. Soc., Vol. 123, No. 17, 2001 4037
Table 4. Population of the Valence 4d and 5s Orbitals in the
Reactant (14), the Transition Structure (TS(14-15)) and the Product
(15)
2
2
2
complex
14
dxy
dxz
dyz
dz
dx -y
∑d
s
1.92 1.97 1.98 1.95
1.40
1.62
1.90
9.22 0.52
9.40 0.43
9.68 0.61
Figure 7.
TS(14-15) 1.88 1.97 1.98 1.95
15
1.90 1.98 1.94 1.96
decreased and is more localized on the nitrogens (Table 6).
Furthermore, there is now involvement of a d-function (19.2%)
on the carbon side of the N p(π) f C p(π) donor bond,
indicating a substantial increase in the polarization of the
electron density away from the carbon nucleus. When consider-
ing the Pd-Ccarbene bond, significant d character (41.1%) at the
Ccarbene atom is also found (Table 5). The occupancy of this
bond has decreased and is localized more on Pd, which now
interacts with Ccarbene almost purely through d orbitals (97.6%).
The Pd s orbital is hardly involved (2.3%) which explains the
decrease in the s orbital population. The Pd-CMe NBO has gone
in the transition structure and instead a CMe-Ccarbene bond is
found (Table 5). On the CMe side the expected s and p orbitals
are involved, while on the Ccarbene there is also involvement of
a d-orbital (23.8%).
Figure 6.
TS(14-15), and the product 15 was analyzed by the NBO
partitioning scheme.89 Table 4 lists the occupation of the Pd
valence 4d and 5s orbitals. The 5p populations are not listed as
they were found to be vacant (population < 0.002 in all cases)
and are not involved in Pd-ligand bonding. Table 4 shows that
2
2
the dx -y orbital in 14 is much less occupied than the other d
orbitals, and that the s orbital has a small population. These are
the Pd orbitals which are involved in bonding with the ligands,
as borne out in the NBO analysis. The methyl group bonds to
Pd through interaction of a roughly sp3 orbital on C with a Pd
sd hybrid. Likewise the carbene bonds with Pd through
interaction of the expected sp2 lone pair of carbon with an sd
hybrid of Pd. This is shown in Table 5 which contains the
occupancy, polarization and hybridization of the Pd-Ccarbene
bond in 14 and TS(14-15).
The above results can be interpreted as follows: in the
transition structure there is mixing of the Ccarbene p(π) orbital,
the CMe orbital and the Pd d orbitals. This mixing with the p(π)
and d orbitals allows the Ccarbene-CMe bond to form (Figure 7).
Due to the angle of the carbene plane relative to the coordination
plane, the p(π) orbital, which is directed perpendicular to the
carbene plane, is in the correct orientation to interact with the
methyl ligand. The increase in the Ccarbene p(π) population is
thus explained by its involvement in bonding with CMe in the
transition state. The d orbital with the correct orientation for
this 3-center interaction is the dxy orbital, and we note a slight
decrease in its population in the transition state (Table 4).
Recently Sakaki and co-workers94 have studied oxidative
addition of (HO)2B-CH3 to Pd(PH3)2 and have found a similar
involvement of the B(OH)2 p(π) orbital in the transition state.
It should also be noted that there are some similarities between
our reaction and reductive coupling of sp2 phenyl or alkenyl
ligands with alkyl ligands. It has been noted that sp2 hybrids
make a less directional bond with the metal than sp3 hybrids,
and therefore more multicentered bonding is possible in the
transition structure, which results in lower activation energies
for alkyl-alkenyl or alkyl-phenyl coupling.63 Similar involvement
of the p(π) orbitals in the transition structures can be envisaged.
Looking at the product, encounter complex 15, the population
Previous theoretical and experimental studies by a number
of groups have shown that the formally empty p orbital
perpendicular to the plane of the carbene (p(π)) is in fact
appreciably occupied due to donation from the p(π) lone pairs
on the nitrogens.62,90-92 This N-C π-donation is apparently
responsible for the high stability of N-heterocyclic carbenes. It
has also been realized that this π-donation is increased on
formation of carbene-metal complexes; electron donation from
carbene to metal upon complexation is compensated by in-
creased π-donation from the nitrogens. The result is an increase
in the population of the C p(π) orbital (Figure 6).44,45 The same
situation occurs in complex 14, and is shown in Table 6, which
contains the population, polarization and hybridization of the
N-Ccarbene π bond. The occupation of the p(π) NAO on Ccarbene
is 0.83, significantly higher than in the free carbene (0.66).93
Along with the expected C-N σ bond, there is a bonding orbital
between N and Ccarbene with a population of 1.89. It is purely
p(π) in character and 25% localized on Ccarbene. This is the N
p(π) f C p(π) donation, and results in a low occupation of the
N p(π) orbitals (1.54).
2
2
of the dx -y orbital increases further as does the population of
the 5s orbital. The Ccarbene-CMe bond is fully occupied and there
is no longer any d orbital involvement. The N p(π) f C p(π)
is purely p in character as in the reactant complex. The p(π)
occupancy of the carbon (0.92) has increased due to greater
donation for the N p(π) orbitals, which is expected from the
greater delocalization of the imidazolium cation.
In the transition structure there is a large increase in the
2
2
population of the dx -y orbital, and a similar increase in the
total d orbital population, along with a smaller decrease in the
s orbital population (Table 4). This observation is consistent
with the notion of a reductive elimination. The occupancy of
the Ccarbene p(π) orbital has increased somewhat (0.89), but this
increase does not come from increased donation from the N
p(π) orbitals, as the occupancy of the N-C π bond has
The changes in the partial charges on the ligands and metal
are shown in Table 7. The degree of positive charge on the
carbene and the methyl ligands increases from start to finish,
while the charge on Pd decreases. The changes are thus
consistent with the notion of reductive elimination. The total
charge on the imidazolium ion (+0.96) and that on Pd(PH3)2
(+0.04) are very close to those on the separated species, and
this indicates that the bonding situation in the encounter complex
is very similar to that in the separated species.
(89) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. ReV. 1988, 88,
899.
(90) Heinemann, C.; Thiel, W. Chem. Phys. Lett. 1994, 217, 11.
(91) Boehme, C.; Frenking, G. J. Am. Chem. Soc. 1996, 118, 2039.
(92) Lehmann, J. F.; Urquhart, S. G.; Ennis, L. E.; Hitchcock, A. P.;
Hatano, K.; Gupta, S.; Denk, M. K. Organometallics 1999, 18, 1862.
(93) The NBO analysis of 1,3-dimethylimidazolin-2-ylidene was per-
formed at the B3LYP/6-311+G(2d,p)//B3LYP/LANL2DZ level.
2.4. Homogeneous Catalysis with Carbene-Metal Com-
plexes. We have shown above that alkyl-carbene coupling from
(94) Sakaki, S.; Kai, S.; Sugimoto, M. Organometallics 1999, 18, 4825.

4038 J. Am. Chem. Soc., Vol. 123, No. 17, 2001
McGuinness et al.
Table 5. Occupancy, Polarization, and Hybridization of the Pd-Ccarbene and Ccarbene-Me Bonds in 14, TS(14-15) and 15
complex
occup
%Ccarbene
%Pd
%s(C)
%p(C)
%d(C)
%s(Pd)
%d(Pd)
Pd-Ccarbene
38.6
14
1.88
1.70
71.9
13.8
28.1
86.2
61.4
40.5
0
41.1
52.4
2.3
47.4
97.6
TS(14-15)
18.3
%Ccarbene
%CMe
%s(Cc)
%p(Cc)
%d(Cc)
%s(CMe)
%p(CMe)
Ccarbene-CMe
23.5
TS(14-15)
15
1.61
1.98
42.2
52.8
57.8
47.2
52.6
59.2
23.8
0
18.5
27.3
81.3
72.5
40.8
Table 6. Occupancy of the Ccarbene and N p(π) Orbitals, and
Occupancy, Polarization, and Hybridization of the N-Ccarbene p(π)
Donor Bond
An understanding of the mechanism by which Pd- and Ni-
carbene complexes eliminate the imidazolium cation allows us
to suggest possible ways in which catalysts may be made more
stable, with the anticipation that improved catalytic performance
will result. Theoretical results show that for reductive elimination
to occur, the carbene and alkyl group must be in a cis
arrangement, and the angle between them must decrease in the
transition structure as they react together. Hence, coordinating
functional groups attached to the carbene nitrogens, as in
complex 21, will hinder the necessary cis arrangement.
p(π) occupancy
N-Ccarbene p(π) donor bond
complex
14
TS(14-15) 0.89
15 0.92
C
N
occup %C %N %p(C) %d(C) %p(N)
0.83
1.54
1.56
1.48
1.89 25.0 75.0 99.8
1.79 17.0 83.0 80.5
1.89 27.8 72.2 99.8
0.2 100.0
19.2
99.9
0.2 100.0
Table 7. Partial Charges (q) on the Ligands and Metal in 14,
TS(14-15), and 15
complex
14
TS(14-15)
q(carbene)
q(Me)
q(Pd(PH3)2
q(Pd)
+0.42
+0.46
+0.86
-0.18
0.00
+0.10
+0.76
+0.54
+0.04
+0.24
+0.15
-0.31
15
heterocyclic carbene complexes of Pd can be facile and occurs
via a mechanism of concerted reductive elimination. Although
not studied in this work, it seems likely that the same mechanism
may operate for carbene-hydride, -acyl, and -phenyl couplings
on Pd,20,30,95 and that Ni will also eliminate the imidazolium
cation via the same mechanism. Both Pd and Ni-carbene
complexes have been applied as catalysts in C-C coupling
reactions, so this reaction provides a route by which such
catalysts may deactivate. The high activity of Pd-carbene
complexes in C-C coupling reactions has been attributed to
the high bond dissociation energy of heterocyclic carbenes.15,17,96
A strongly coordinating ligand will stabilize a catalyst, providing
ligand dissociation is the main cause of catalyst decomposition.
However, while the bond dissociation may be high, this does
not necessarily equate to high catalyst stability, as there are often
other low-energy pathways to decomposition not dependent on
dissociation. In this paper we have identified such a pathway
for catalysts based on hydrocarbyl carbene complexes.
A number of Pd-carbene complexes are among the most
active catalysts reported for Heck couplings.21,85 However, if
Pd complexes so readily reductively eliminate, how can this
high activity be rationalized? While studying olefin insertion
into phenyl-Pd carbene complexes we found that at low
temperatures carbene-phenyl reductive coupling predominates,
while at higher temperatures insertion and â-hydride elimination
effectively competes with reductive coupling.20 It seems likely
that under catalytic conditions (elevated temperature, excess
substrate) the insertion/â-hydride elimination steps of the Heck
reaction may be faster than reductive elimination, hence the high
activity of the catalysts. Our observation that decomposition is
observed shortly after the substrates are exhausted is in
agreement with the concept of competing pathways.19 Additional
stability would come from anionic coordinating ligands present
in solution, and it is also possible that the catalyst resting state
is zerovalent, from which reductive coupling cannot occur.
We have recently prepared a number of complexes of this type
and have found that when the donor group is suitably coordinat-
ing the complex does appear to have higher stability. When
weaker donor groups are attached, the complex is much less
stable, and the imidazolium cation is readily eliminated.85
Sterically demanding substituents on the carbene will hinder
close approach of the alkyl group and carbene. This is likely to
be most effective if the carbene contains out of plane steric bulk,
as the carbene interacts with the alkyl group in a side-on manner.
t
Nitrogen substituents such as Bu, 2,6-diisopropylphenyl, and
adamantyl would achieve this. It is interesting to note that
Herrmann et al.96 found that bulky carbene ligands were
important in achieving high activity with mixed carbene-
phosphine Pd complexes for Suzuki coupling. Moreover, the
approach of the carbene and methyl groups is accompanied by
an attendant increase in the angle between the two nonpartici-
patory ligands in the transition structure. We have shown that
fixing this “bite angle” with a chelating diphosphine (complex
13) slows the decomposition reaction to such an extent that it
is not observed at ambient temperature.
2.4.1. Metal-Catalyzed Reactions in Ionic Liquids. A
potentially important method to limit decomposition of Pd-
carbene complexes, via reductive elimination, is to force the
reaction in the reverse direction by operating in a large excess
of imidazolium ionsany Pd0 that forms as a consequence of
reductive elimination may re-form the Pd-carbene complex
through oxidative addition of the imidazolium ion. To operate
in molten imidazolium salt97 as the solvent would ensure an
appropriate excess of ion.
A variety of catalytic reactions have been carried out in ionic
liquids, including Ni-catalyzed olefin dimerization98,99 and Pd-
catalyzed Heck reactions.100-102 One of the most common types
of ionic liquids used are those based on substituted imidazolium
(97) Welton, T. Chem. ReV. 1999, 99, 2071.
(98) Chauvin, Y.; Einloft, S.; Olivier, H. Ind. Eng. Chem. Res. 1995,
34, 1149.
(99) Ellis, B.; Keim, W.; Wasserscheid, P. J. Chem. Soc., Chem.
Commun. 1999, 337.
(95) In the case of (sp2) phenyl and acyl coupling with carbenes
involvement of the occupied p(π) orbitals is possible.
(96) Weskamp, T.; Bo¨hm, V. P. W.; Herrmann, W. A. J. Organomet.
Chem. 1999, 585, 348.

Alkyl-Carbene Elimination from Pd-Carbene Complexes
J. Am. Chem. Soc., Vol. 123, No. 17, 2001 4039
4. Experimental Section
salts,97 providing an ideal opportunity for reaction between the
metal and imidazolium ion. Indeed, in one application of
imidazolium salts as solvents for the Heck reaction the possible
involvement of Pd-carbene species has been suggested,100 and
in a study on Suzuki coupling reactions with Pd2(dba)3 as the
precatalyst, the apparently successful in situ conversion of
imidazolium salt into free carbene has been reported.22 Very
recently, Xiao et al. isolated Pd-carbene complexes from an
ionic liquid solvent in which Pd(OAc)2 was used as a pre-
catalyst for the Heck reaction.102 The formation of carbene
complexes in the Heck reaction most likely resulted from
deprotonation of the imidazolium C2 proton under the basic
conditions of the reaction. However, the apparent formation of
Pd-carbene complexes from Pd2(dba)3 raises the question, can
carbene-metal complexes form via an alternative mechanism
in which prior deprotonation of the imidazolium cation is not
necessary? The present study suggests that they cansthrough
a process of oxidative addition to a low-valent metal. It is known
that chelating ligands on zerovalent Pd and Pt favor oxidative
addition.65-69,94 This is a consequence of a bent geometry for
the M0(L∧L) fragment, which has a high d-orbital energy
relative to linear ML2. Thus, the reverse reaction (in this case,
reductive elimination from a carbene complex containing a
chelating ligand) is expected to be disfavored. Consistent with
this expectation, our preliminary calculations on the model
system PdMe(dmiy)(P∧P) (P∧P ) 1,2-diphosphinoethane) show
that the reactant is lower in energy than the Pd0(P∧P) and
imidazolium products, that is, oxidative addition of imidazolium
ion to Pd(P∧P) should be spontaneous. Thus, in principle it is
possible to change the energetics of a system to favor oxidative
addition to give a carbene complex, especially in a huge excess
of imidazolium ion. It is important to note that not only would
oxidative addition of an imidazolium ion produce a carbene
complex, but it would also generate a reactive metal-hydride
that could initiate a catalytic cycle such as dimerization or C-C
coupling. Further computational and experimental studies on
the oxidative addition reaction103 and on nickel-carbene
catalyzed dimerization reactions will be reported shortly.
4.1. General Comments. All manipulations were carried out using
standard Schlenk techniques or in a nitrogen glovebox (Innovative
Technology Inc.). All solvents were purified by standard procedures
and distilled under nitrogen immediately prior to use. Nuclear magnetic
resonance (NMR) spectra were recorded at ambient temperature unless
otherwise stated, and peaks are labeled as singlet (s), doublet (d), triplet
(t), multiplet (m), and broad (br). Elemental analysis, MS and GC-
MS were carried out by the Central Science Laboratory, University of
Tasmania. [PdMeCl(tmiy)]2 was prepared as previously described.19,51
4.2. Synthesis. [PdMe(tmiy)(COD)]BF4, 8. A solution of [PdMeCl-
(tmiy)]2 (71 mg, 0.25 mmol) in 10 mL of DCM was treated with COD
(35 µL, 0.29 mmol) and syringed into a suspension of AgBF4 (73 mg,
0.37 mmol) in 3 mL of DCM at -20 °C. The solution was stirred at
this temperature for 1 h before being filtered through Celite and the
solvent removed in vacuo at -20 °C until ∼2 mL remained. Diethyl
ether (4 mL) was added, and the solvent was decanted off the precipitate
that formed. After being washing twice with 4 mL of ether the product
was dried in vacuo to yield a white powder. Yield: 0.100 g (90%).
Anal. Calcd for C16H27N2BF4Pd: C, 43.61; H, 6.18; N, 6.36. Found:
1
C, 43.62; H, 6.16; N, 6.08. MS (LSIMS) m/z: 353 [M]+ (100%). H
NMR (200 MHz, CDCl3, -20 °C): δ 5.79 (s, br, 2H, HCdC), 5.66
(s, br, 2H, CdCH), 3.75 (s, 6H, NCH3), 2.68 (m, br, 8H, CH2), 2.16
(s, 6H, CCH3), 0.60 (s, 3H, PdCH3). 13C NMR (50 MHz, CDCl3 CDCl3,
-20 °C): δ 165.3 (NCN), 126.8 (CdCcarbene), 118.4, 116.2 (CdCCOD),
35.5 (NCH3), 29.4, 28.9 (CH2), 9.2 (H3CCdCCH3), 1.7 (PdCH3).
[PdMe(tmiy)(PMePh2)2]BF4, 9. This complex was prepared in a
manner similar to that for 8 using 51.0 mg (0.0.181 mmol) of [PdMeCl-
(tmiy)]2, 37 mg (0.19 mmol) of AgBF4 and 70 µL (0.37 mmol) of
PMePh2. Yield: 0.113 g (85%). Anal. Calcd for C34H41N2P2PdBF4:
C, 55.72; H, 5.64; N, 3.82. Found: C, 55.07; H, 5.87; N, 3.60. MS
(LSIMS) m/z: 645 [M]+ (30%), 506 [Pd(PMePh2)2]+ (10%), 307 [Pd-
(PMePh2)]+ (10%), 201 [PMePh2]+ (100%). 1H NMR (400 MHz, CD2-
Cl2): δ 7.6-7.1 (m, 20H, phenylH), 3.49 (s, 6H, NCH3), 1.88 (s, 6H,
CCH3), 1.86 (d, J ) 8 Hz, 3H, PCH3), 1.45 (d, J ) 7 Hz, 3H, PCH3),
0.17 (dd, J ) 7 Hz, 8 Hz, 3H, PdCH3). 13C NMR (100 MHz, CD2Cl2):
δ 132.6 (d, J ) 11 Hz, phenylC), 131.4 (d, J ) 12 Hz, phenylC),
130.8 (d, J ) 2 Hz, phenylC), 130.5 (d, J ) 2 Hz, phenylC), 129.0 (d,
J ) 10 Hz, phenylC), 128.8 (d, J ) 9 Hz, phenylC), 126.2 (d, J ) 3
Hz, CdC), 35.1 (NCH3), 13.2 (d, J ) 27 Hz, PCH3), 11.9 (d, J ) 21
Hz, PCH3), 8.9 (s, H3CCdCCH3), -0.27 (d, J ) 91 Hz, PdCH3). 31P
NMR (162 MHz, CD2Cl2): δ 12.3 (d, J ) 33 Hz), 0.8 (d, J ) 33 Hz).
3. Conclusions
[PdMe(tmiy){P(OPh)3}2]BF4, 10. This complex was prepared in a
manner similar to that for 8 using 65.2 mg (0.232 mmol) of [PdMeCl-
(tmiy)]2, 66 mg (0.34 mmol) of AgBF4, and 125 µL (0.477 mmol) of
P(OPh)3. Yield: 0.192 g (87%). Anal. Calcd for C44H45N2O6P2PdBF4:
C, 55.45; H, 4.76; N, 2.94. Found: C, 55.51; H, 4.88; N, 3.11. MS
(LSIMS) m/z: 865 [M]+ (3%), 726 [Pd{P(OPh)3}2]+ (1%), 139
[pentamethylimidazolium]+ (100%). 1H NMR (200 MHz, CDCl3, -20
°C): δ -0.23 (t, J ) 8.6 Hz, 3H, PdCH3), 1.88 (s, CCH3), 2.92 (s,
NCH3), 7.5-7.0 (m, 30H, phenylH). 13C NMR (50 MHz, CDCl3, -20
°C): δ 150.1 (d, J ) 5 Hz, ipso-phenylC), 149.8 (d, J ) 8 Hz, ipsoC),
130.1 (d, J ) 5 Hz, phenylC), 126.4 (d, J ) 5 Hz, phenylC), 125.8 (d,
J ) 15 Hz, phenylC), 120.3 (dd, J ) 6 Hz, 24 Hz, CdC), 34.4 (NCH3),
8.7 (H3CCdCCH3), 2.7 (d, J ) 130 Hz, PdCH3). 31P (162 MHz, CD2-
Cl2, -30 °C): δ 120.3 (d, J ) 72 Hz), 113.4 (d, J ) 72 Hz). After
some decomposition had occurred signals at 6.7-7.2 ppm appear in
A number of novel cationic methyl-PdII-carbene complexes
incorporating different phosphine ligands have been prepared
and fully characterized. The complexes decompose cleanly to
yield methylimidazolium salts as has been observed previously
for a variety of Pd and Ni complexes, and this shows that
carbene-based catalysts incorporating a hydrocarbyl group in
the active species have available a low-energy pathway to
decomposition. Combined kinetic and DFT studies show that
the mechanism is one of concerted reductive elimination, and
not one of carbene insertion into M-R bonds. This new type
of reductive elimination suggests there are similarities in the
reactivity of M-R bonds and M-heterocyclic carbene bonds,
and again highlights the differences between conventional
Fischer and Schrock carbenes and N-heterocyclic carbenes.
Finally, an understanding of the mechanism by which this
reaction occurs has allowed us to postulate methods by which
it can be impeded. These methods and the catalysts that result
from their application are currently under investigation in our
laboratories.
1
the H NMR spectrum corresponding to Pd{P(OPh)3}2, along with a
singlet at 141.0 ppm in the 31P NMR spectrum.
[PdMe(tmiy)(PPh3)2]BF4, 11. This complex was prepared in a
manner similar to that for 8 using 52.2 mg (0.186 mmol) of [PdMeCl-
(tmiy)]2, 44 mg (0.23 mmol) of AgBF4, and 100.4 mg (0.383 mmol)
of PPh3. Yield: 106 mg (67%). HRMS Calcd for C44H45N2P2104Pd:
767.209826. Found: 767.21313. MS (LSIMS) m/z: 769 [M]+ (7%),
631 [M and MH cluster for Pd(PPh3)2]+ (25%), 139 [pentamethylimi-
(100) Carmichael, A. J.; Earle, M. J.; Holbrey, J. D.; McCormac, P. B.;
Seddon, K. R. Org. Lett. 1999, 1, 997.
(101) Herrmann, W. A.; Bo¨hm, V. P. W. J. Organomet. Chem. 1999,
572, 141.
(102) Xu, L.; Chen, W.; Xiao, J. Organometallics 2000, 19, 1123.
(103) McGuinness, D. S.; Cavell, K. J.; Yates, B. F. J. Chem. Soc., Chem.
Commun. In press.
1
dazolium]+ (100%). H NMR (200 MHz, CD2Cl2, -50 °C): δ 7.0-
7.5 (m, 30H, phenylH), 3.52 (s, 6H, NCH3), 1.89 (s, 6H, CCH3), 0.23
(“t”, J ) 6 Hz, 3H, PdCH3). 31P NMR (162 MHz, CD2Cl2, -30 °C):
δ 34.2 (d, J ) 30 Hz), 21.4 (d, J ) 30 Hz). After some decomposition
has occurred signals at 7.6-7.85 ppm appear in the 1H NMR spectrum

4040 J. Am. Chem. Soc., Vol. 123, No. 17, 2001
McGuinness et al.
corresponding to Pd(PPh3)2, along with a singlet at 35.4 ppm in the
31P NMR spectrum.
of the Computational Chemistry group (University of Tasmania)
are thanked for their assistance.
Acknowledgment. We are indebted to the Australian
Research Council for financial support and for providing an
Australian Postgraduate Award for D.S.M. We also acknowl-
edge the generosity of Johnson-Matthey for providing a loan
of Pd salts. The staff of the Central Science Laboratory
(University of Tasmania) are gratefully acknowledged for their
assistance in a number of instrumental techniques, and members
Supporting Information Available: Complete kinetic plots
for the decomposition of complexes 8-11 and optimized
geometries and absolute energies of complexes 14-20 and 1,3-
dimethylimidazolin-2-ylidene (PDF). This material is available
free of charge via the Internet at http://pubs/acs.org.
JA003861G