Mechanism of Pd(NHC)-catalyzed transfer hydrogenation of alkynes

Article abstract of DOI:10.1021/ja1062407

The transfer semihydrogenation of alkynes to (Z)-alkenes shows excellent chemo- and stereoselectivity when using a zerovalent palladium(NHC)(maleic anhydride)-complex as precatalyst and triethylammonium formate as hydrogen donor. Studies on the kinetics under reaction conditions showed a broken positive order in substrate and first order in catalyst and hydrogen donor. Deuterium-labeling studies on the hydrogen donor showed that both hydrogens of formic acid display a primary kinetic isotope effect, indicating that proton and hydride transfers are separate rate-determining steps. By monitoring the reaction with NMR, we observed the presence of a coordinated formate anion and found that part of the maleic anhydride remains coordinated during the reaction. From these observations, we propose a mechanism in which hydrogen transfer from coordinated formate anion to zerovalent palladium(NHC)(MA)(alkyne)-complex is followed by migratory insertion of hydride, after which the product alkene is liberated by proton transfer from the triethylammonium cation. The explanation for the high selectivity observed lies in the competition between strongly coordinating solvent and alkyne for a Pd(alkene)-intermediate.

Full text of DOI:10.1021/ja1062407

Published on Web 11/08/2010
Mechanism of Pd(NHC)-Catalyzed Transfer Hydrogenation of
Alkynes
Peter Hauwert,† Romilda Boerleider,† Stefan Warsink,† Jan J. Weigand,‡ and
Cornelis J. Elsevier*,†
Molecular Inorganic Chemistry, Van’t Hoff Institute for Molecular Sciences, UniVersity of
Amsterdam, Science Park 904, Postbus 94157, 1090 GD Amsterdam, The Netherlands, and
Institut fu¨r Anorganische und Analytische Chemie, Westfa¨lische Wilhelms-UniVersita¨t Mu¨nster,
Corrensstrasse 30, 48149, Mu¨nster, Germany
Received July 27, 2010; E-mail: c.j.elsevier@uva.nl
Abstract: The transfer semihydrogenation of alkynes to (Z)-alkenes shows excellent chemo- and
stereoselectivity when using a zerovalent palladium(NHC)(maleic anhydride)-complex as precatalyst and
triethylammonium formate as hydrogen donor. Studies on the kinetics under reaction conditions showed a
broken positive order in substrate and first order in catalyst and hydrogen donor. Deuterium-labeling studies
on the hydrogen donor showed that both hydrogens of formic acid display a primary kinetic isotope effect,
indicating that proton and hydride transfers are separate rate-determining steps. By monitoring the reaction
with NMR, we observed the presence of a coordinated formate anion and found that part of the maleic
anhydride remains coordinated during the reaction. From these observations, we propose a mechanism in
which hydrogen transfer from coordinated formate anion to zerovalent palladium(NHC)(MA)(alkyne)-complex
is followed by migratory insertion of hydride, after which the product alkene is liberated by proton transfer
from the triethylammonium cation. The explanation for the high selectivity observed lies in the competition
between strongly coordinating solvent and alkyne for a Pd(alkene)-intermediate.
this reaction are known based on Ru-,7,8 Rh-,9 or Fe-
complexes.10 However, this process is often plagued by over-
Introduction
The partial hydrogenation of alkynes to cis-alkenes is a very
important transformation in synthetic organic chemistry.1 It is
particularly relevant for the synthesis of biologically important
molecules such as natural products, pharmaceuticals, and
fragrance chemicals, since many of these molecules incorporate
carbon-carbon double bonds with defined Z or E configurations.
Despite the usefulness of this reaction, it has been studied far
less extensively than the similar hydrogenation of carbon-carbon
double bonds.1,2 Several homogeneous Pd-complexes that
catalyze this reaction have been reported by us3-5 and others,6
and examples of other metal-complexes that are able to catalyze
reduction after full conversion, leading to saturated compounds.
Also cis/trans isomerization usually occurs to a considerable
extent, and issues with reproducibility have been encountered.1,11
Some examples of chemo- and stereoselective alkyne semihy-
drogenation have been reported, but very few studies have
addressed the mechanism of the reaction or explained the
observed selectivity.4,12,13 For the development of better catalytic
systems, it is of vital importance to understand the mechanism,
enabling a directed search for new catalysts.
Recently, we described a system for semihydrogenation of
alkynes that cleanly and reproducibly leads to Z-alkenes, using
Pd0(IMes)(MA) complex 1 as precatalyst, which is prepared in
situ from our standard Pd0-precursor Pd(tBuDAB)(MA) 2, 1,3-
bis(mesityl)imidazolium chloride, and t-potassium butoxide
(Scheme 1). The over-reduction to alkanes is fully inhibited
† University of Amsterdam.
‡ Westfa¨lische Wilhelms-Universita¨t Mu¨nster.
(1) Kluwer, A. M.; Elsevier, C. J. In Handbook for Homogeneous
Hydrogenation, 1st ed.; de Vries, J. G., Elsevier, C. J., Eds.; Wiley-
VCH: Weinheim, 2007; Vol. 1, pp 375-411.
(2) Ager, D. J. In Handbook for Homogeneous Hydrogenation, 1st ed.;
de Vries, J. G., Elsevier, C. J., Eds.; Wiley-VCH: Weinheim, 2007;
Vol. 2, pp 745-772.
(7) (a) Horva´th, H. H.; Joo´, F. React. Kinet. Catal. Lett. 2005, 85, 355–
360. (b) Shvo, Y.; Goldberg, I.; Czerkie, D.; Reshef, D.; Stein, Z.
Organometallics 1997, 16, 133–138.
(3) Sprengers, J. W.; Wassenaar, J.; Clement, N. D.; Cavell, K. J.; Elsevier,
C. J. Angew. Chem., Int. Ed. 2005, 44, 2026–2029.
(8) Gao, Y.; Jennings, M. C.; Puddephatt, R. J. Can. J. Chem. 2001, 79,
915–921.
(4) Kluwer, A. M.; Koblenz, T. S.; Jonischkeit, T.; Woelk, K.; Elsevier,
C. J. J. Am. Chem. Soc. 2005, 127, 15470–15480.
(9) (a) Kameda, N.; Yoneda, T. J. Chem. Soc. Jpn. 1999, 1, 33–36. (b)
Schrock, R.; Osborn, J. A. J. Am. Chem. Soc. 1976, 98, 2143–2147.
(10) Bianchini, C.; Meli, A.; Peruzzini, M.; Frediani, P.; Bohanna, C.;
Esteruelas, M. A.; Oro, L. A. Organometallics 1992, 11, 138–145.
(11) (a) Molna´r, A.; Sa´rka´ny, A.; Varga, M. J. Mol. Catal. A: Chem. 2001,
173, 185–221. (b) Lindlar, H. HelV. Chim. Acta 1952, 35, 446–450.
(12) Andriollo, A.; Esteruelas, M. A.; Meyer, U.; Oro, L. A.; Sa´nchez-
Delgado, R. A.; Sola, E.; Valero, C.; Werner, H. J. Am. Chem. Soc.
1989, 111, 7431–7437.
(5) (a) van Laren, M. W.; Duin, M. A.; Klerk, C.; Naglia, M.; Rogolino,
D.; Pelagatti, P.; Bacchi, A.; Pelizzi, C.; Elsevier, C. J. Organometallics
2002, 21, 1546–1553. (b) van Laren, M. W.; Elsevier, C. J. Angew.
Chem., Int. Ed. 1999, 38, 3715–137.
(6) (a) Jurcˇ´ık, V.; Nolan, S. P.; Cazin, C. S. L. Chem.sEur. J. 2009, 15,
2509–2511. (b) Costa, M.; Pelagatti, P.; Pelizzi, C.; Rogolino, D. J.
Mol. Catal. A: Chem. 2002, 178, 21–26. (c) Pelagatti, P.; Venturini,
A.; Leporati, A.; Carcelli, M.; Costa, M.; Bacchi, A.; Pelizzi, G.;
Pelizzi, C. J. Chem. Soc., Dalton Trans. 1998, 2715–2721. (d) Trost,
B. M.; Braslau, R. Tet. Lett. 1989, 30, 4657–4660.
(13) Kovacs, G.; Ujaque, G.; Lledos, A.; Joo, F. Eur. J. Inorg. Chem. 2007,
2879–2889.
9
16900 J. AM. CHEM. SOC. 2010, 132, 16900–16910
10.1021/ja1062407  2010 American Chemical Society

Mechanism of Alkyne Transfer Hydrogenation
A R T I C L E S
Scheme 1. Transfer Semihydrogenation of 1-Phenyl-1-propyne Using Triethylammonium Formate and in Situ Generated Complex 1
Table 1. Effect of Various Hydrogen Donors on the Hydrogenation of 1-Phenyl-1-propyne to Z-ꢀ-Methylstyrene with Precatalyst 1a
entry
hydrogen donor
[donor] mol L-1
base
[base] mol L-1
[alkyne] mol L-1
[catalyst] (× 10-3) mol L-1
rate (× 10-3) mol h-1
selectivity (Z/E/alkane)b
1
2
3
4
5
6
HCO2H
HCO2H
HCO2H
HCO2H
HCO2H
iPrOH
0.87
0.96
0.94
0.61
0.82
13.06
NEt3
0.87
0.41
0.11
0.61
0.59
-
0.16
0.18
0.18
0.10
0.13
0.25
0.90
1.6
1.6
1.2
1.3
2.5
1.50
0.50
0.045
0.041
0.011
0.014
95.2/3.9/0.9
96.4/3.7/0.0
96.7/3.3/0.0
95.3/4.0/0.7
93.2/2.9/3.9
1/-/-
NEt3
NEt3
NH3
Na2CO3
-
a Conditions: Reactions are performed in refluxing MeCN, using in situ generated precatalyst 1. b Selectivity reported after 24 h.
when formic acid is used as hydrogen donor in strongly
coordinating solvents,14,15 whereas the use of hydrogen gas still
gives over-reduction. This increased chemoselectivity strongly
hints at different modes of operation for the formic acid-
mediated transfer hydrogenation versus hydrogenation with
molecular hydrogen.
inhibited in our system? We decided to study the kinetics of
the transfer hydrogenation of 1-phenyl-1-propyne and the role
of hydrogen donor and base on rate and selectivity. Furthermore,
several catalytically relevant metal complexes were isolated and
catalytic intermediates were investigated using in situ NMR-
spectroscopy and deuterium-labeling of the precatalyst and the
hydrogen donor.
Mechanistic studies on transfer hydrogenation have focused
on ketones and imines as the abundant substrates, and several
mechanisms are known to operate depending on the catalyst,
substrate, and the hydrogen donor.16,17 However, carbonyls are
more reactive than alkenes or alkynes because of the high degree
of polarization of the CdO bond, and the only known alkenes
subject to transfer hydrogenation are also highly polarized.18,19
According to calculations by Comes-Vives and Lledo´s, the
activation barriers for nonpolar substrates are approximately 10
kcal/mol higher than for polar substrates.17 This difference is
reflected in the observation that transfer hydrogenation of
acetophenone has become a standard reaction for testing new
catalytic systems, whereas relatively few examples are known
of hydrogen transfer to alkenes18,19 or alkynes.8,14,15,20 Hence,
the mechanism of transfer hydrogenation of carbon-carbon
multiple bonds has been studied only marginally.17,19
Results and Discussion
1. Kinetic Studies. Knowledge about the kinetics of a reaction
is essential for discussion of the mechanism. Hence, the reaction
was monitored by GC and the rate of reaction and its
dependence on the concentration of the reactants were obtained
from the reaction profiles, using the initial rates method. This
method is straightforward, reliable, and especially suitable for
reactions that are not very fast. A 5-fold excess of HCO2H/
NEt3 was used to ascertain that product selectivity toward alkene
is not due to a shortage of hydrogen donor.
We initially investigated whether the acid/base-ratio affects
the rate and what the influence of the base strength is. As can
be seen from Table 1, the presence of base is essential for an
efficient reaction, as the rate is drastically lower if HCO2H is
in excess to NEt3, even if NEt3 is still in excess to alkyne (Table
1, entries 1-3).
We also performed the reaction with Na2CO3 as base or
ammonium formate as hydrogen donor (Table 1, entries 4-5),
but both give inferior results compared to triethylammonium
formate (TEAF). The reaction with ammonium formate is much
slower than with in situ generated TEAF; this difference is
attributed to the relative pKb values of ammonia and triethy-
lamine. Na2CO3 is too basic; the reaction was even slower than
without added base. The active hydrogen donor is probably an
ammonium formate species, as the reaction becomes very
sluggish once the amount of NEt3 is decreased relative to
HCO2H (Table 1, entries 1-3); therefore, we used TEAF in
the kinetic studies. Isopropyl alcohol, the standard hydrogen
donor for transfer hydrogenation of ketones, is not an efficient
hydrogen donor for the transfer hydrogenation of alkynes (Table
1, entry 6).
This scarcity in profound mechanistic understanding for both
alkene transfer hydrogenation and partial alkyne hydrogenation
prompted us to thoroughly investigate the mechanism of the
formic acid mediated partial alkyne transfer hydrogenation. The
main issues to be addressed are (1) what is the mechanism of
the Pd-catalyzed hydrogen transfer and (2) why is over-reduction
(14) Hauwert, P.; Maestri, G.; Sprengers, J. W.; Catellani, M.; Elsevier,
C. J. Angew. Chem., Int. Ed. 2008, 47, 3223–3226.
(15) Warsink, S.; Hauwert, P.; Siegler, M. A.; Spek, A. L.; Elsevier, C. J.
Appl. Organometal. Chem. 2009, 23, 225–228.
(16) (a) Blacker, A. J. In Handbook for Homogeneous Hydrogenation, 1st
ed.; de Vries, J. G., Elsevier, C. J., Eds.; Wiley-VCH: Weinheim, 2007;
Vol. 3, pp 1215-1244; (b) Gladiali, S.; Alberico, E. Chem. Soc. ReV.
2006, 35, 226–236. (c) Samec, J. S.; Ba¨ckvall, J.-E.; Andersson, P. G.;
Brandt, P. Chem. Soc. ReV. 2006, 35, 237–248.
(17) Comas-Vives, A.; Ujaque, G.; Lledo´s, A. J. Mol. Struct: THEOCHEM
2009, 903, 123–132.
(18) (a) Gnanamgari, D.; Moores, A.; Rajaseelan, E.; Crabtree, R. H.
Organometallics 2007, 26, 1226–1230. (b) Black, P. J.; Cami-Kobeci,
G.; Edwards, M. G.; Slatford, P. A.; Whittlesey, M. K.; Williams,
J. M. J. Org. Biomol. Chem. 2006, 4, 116–125. (c) Basu, B.; Bhuiyan,
M. M. H.; Das, P.; Hossain, I. Tet. Lett. 2003, 44, 8931–8934.
(19) (a) Xue, D.; Chen, Y. C.; Cui, X.; Wang, Q. W.; Zhu, J.; Deng, J. G.
J. Org. Chem. 2005, 70, 3584–3591. (b) Lange, S.; Leitner, W.
J. Chem. Soc., Chem. Commun. 2002, 752–758.
1.1. Dependence of the Reaction Rate on the Catalyst
Concentration. The concentration of precatalyst 1 was varied
between 0.1 and 8.3 mM, keeping the initial concentration of
hydrogen donor constant at 0.8 M and substrate concentration
at 0.18 M. The reaction rate shows a linear dependency on the
9
J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010 16901

A R T I C L E S
Hauwert et al.
Figure 1. (a) Plot of rate of hydrogenation versus catalyst concentration. (b) Plot of ln(rate) versus ln[catalyst].
Table 2. Kinetic Data for Hydrogenation of 1-Phenyl-1-propyne to
Z-ꢀ-Methylstyrene by in Situ Generated 1, Varying [Precatalyst 1]a
Table 3. Kinetic Data for Hydrogenation of 1-Phenyl-1-propyne to
Z-ꢀ-Methylstyrene by in Situ Generated 1, Varying [TEAF]a
[catalyst] (x 10-3
mol L-1
)
rate (× 10-3
)
kobs L2.3 h-1
mol-2.3
selectivity
(Z/E/alkane)b
selectivity
(Z/E/alkane)b
entry
mol h-1
entry [TEAF] mol L-1 rate (× 10-3) mol h-1 kobs L2.3 h-1 mol-2.3
1
2
3
4
5
0.13
0.48
0.90
3.67
8.26
0.15
0.64
1.5
4.9
15.0
2.1
2.6
3.1
2.4
3.4
98.8/1.2/0
98.3/1.7/0
95.2/3.9/0.9
77.2/13.2/9.6c
68.6/16.3/15.1d
1
2
3
4
5
0.14
0.24
0.45
0.83
1.76
0.33
0.79
1.19
1.24
1.91
4.2
5.9
4.8
2.8
2.1
98.3/1.7/0
97.2/2.8/0
96.6/2.9/0.5
97.2/2.8/0
97.4/2.4/0.2
a Conditions: Reactions are performed in refluxing MeCN, using in
situ generated precatalyst 1, [alkyne] ) 0.18 M, [TEAF] ) 0.90 M.
b Selectivity is reported after 24 h. c Product selectivity at 80%
conversion (20 min): 97.2/2.2/0.6. d Product selectivity at 80%
conversion (10 min): 95.3/3.3/1.4.
a Conditions: Reactions are performed in refluxing MeCN, using in
situ generated precatalyst 1, [1-phenyl-1-propyne] 0.18 M;
[precatalyst 1] ) 0.92 mM. b Selectivity is reported after 24 h.
)
1.3. Dependence of Reaction Rate on Substrate Concentration.
The alkyne concentration was varied between 0.03 and 0.43
M, keeping the concentrations of catalyst and TEAF constant
at 0.9 mM and 0.8 M, respectively. At alkyne concentrations
below 0.15 M, the reaction rate shows a positive linear
dependency with respect to the alkyne concentration, while
above 0.15 M, an increase in alkyne concentration results in a
decrease of the rate (Figure 3a; rates are listed in Table 4).
Plotting ln(rate) versus ln[alkyne], as presented in Figure 3b,
yields two straight lines, with slope +0.28 ([alkyne] < 0.15 M)
and -0.75 ([alkyne] > 0.15 M). The broken orders in alkyne
concentration imply that the substrate coordinates to palladium
before the rate-determining step and that an equilibrium is
involved between the Pd(solvent)-complex + alkyne and the
Pd(alkyne) complex. This is in accordance with the strong
solvent effect we had already observed in the optimization of
the catalytic system.14
palladium concentration (see Figure 1a; rates are listed in Table
2). Plotting ln(rate) versus ln[Pd], as presented in Figure 1b,
yields a straight line of slope 1.09, showing that the transfer
semihydrogenation of 1-phenyl-1-propyne is first order in
palladium. This implies that the active catalyst is a homogeneous
species, and not a precatalyst that is converted into catalytically
active Pd nanoparticles. It seems that a low catalyst loading
improves the product selectivity for alkenes (Vide infra).
1.2. Dependence of the Reaction Rate on Hydrogen Donor
Concentration. The concentration of TEAF was varied between
0.1 and 1.6 M, keeping the concentrations of catalyst and
substrate constant at 0.90 mM and 0.16 M, respectively. The
reaction rate shows a linear dependency on the TEAF concen-
tration, which levels off above 0.5 M (Figure 2a; rates are listed
in Table 3). Plotting ln(rate) versus ln[TEAF], as presented in
Figure 2b, yields a straight line of slope 1.06, showing that the
hydrogenation of 1-phenyl-1-propyne is first order in triethy-
lammonium formate. It has been established that the concentra-
tion of hydrogen donor has no effect on the selectivity.
We have attempted to study the association constant of
1
1-phenyl-1-propyne for the catalyst using H NMR, but no
change in chemical shift was observed upon addition of
precatalyst to a sample of 1-phenyl-1-propyne in DMSO-d6.
Figure 2. (a) Plot of rate of hydrogenation versus hydrogen donor concentration. (b) Plot of ln(rate) versus ln[TEAF].
9
16902 J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010

Mechanism of Alkyne Transfer Hydrogenation
A R T I C L E S
Figure 3. (a) Plot of rate of hydrogenation versus substrate concentration. (b) Plot of ln(rate) versus ln[1-phenyl-1-propyne]; circles represent values at
[alkyne] < 0.15 M, triangles represent values at [alkyne] > 0.15 M.
Table 4. Kinetic Data for Hydrogenation of 1-Phenyl-1-propyne to
From the leftmost part of Figure 3a, one sees that the slope
Z-ꢀ-Methylstyrene by in Situ Generated 1, Varying
slightly decreases with increasing alkyne concentration. At
sufficiently high concentrations ([alkyne] > 0.15 M), the order
in alkyne changes from a positive to a negative order. We
believe this change in order occurs because the alkyne is
involved in two concurrent reactions: a productive reaction with
a Pd(solvento)-species to form a Pd(alkyne)-species that leads
toward hydrogenation (with [alkyne]0.28), and a side-reaction
with that Pd(alkyne)-species to form a catalytically inactive
complex (with [alkyne]-1). At [alkyne] > 0.15 M, the second
reaction effectively competes with the catalytically productive
reaction for the alkyne present; this decreases the concentration
of actiVe catalyst, so the net reaction becomes slower at higher
substrate concentration. The observed order in alkyne is the sum
of the separate orders, rendering the order in alkyne -0.75. At
[alkyne] , 0.15 M, the concentration of the Pd(alkyne)-species
is too low to significantly affect the reaction rate. The change
in sign of the substrate order at higher alkyne concentration
may be explained by the formation of Pd(alkyne)2-complexes
(substrate inhibition) or palladacylopentadiene species that act
as a sink for catalyst species (Scheme 2).
[1-phenyl-1-propyne]a
selectivity (Z/E/alkane)b
rate
(× 10-3
[alkyne]
mol L-1
)
kobs L2.3 h-1
mol-2.3
entry
mol h-1
>90% conversion
24 h
1
2
3
4
5
6
7
8
9
0.03
0.06
0.09
0.11
0.14
0.16
0.23
0.32
0.44
1.6
2.0
2.2
2.3
2.4
2.0
1.7
1.3
1.0
6.2
6.5
6.2
5.9
5.8
4.7
3.6
2.4
1.9
74.9/9.4/15.6
90.9/3.7/5.3
93.4/3.9/2.7
94.6/3.2/2.2
92.7/3.4/3.9
n.a.
n.a.
n.a.
n.a.
47.1/17.1/35.8
47.2/21.1/31.7
47.2/21.1/31.7
61.2/17.7/21.4
46.2/11.7/42.0
97.0/2.1/1.0
97.1/2.4/0.6
98.7/1.3/0.0
97.9/1.6/0.5
a Conditions: Reactions are performed in refluxing MeCN, using in
situ generated precatalyst 1; [TEAF] ) 0.77 M; [precatalyst 1] ) 0.95
mM. b Selectivity is reported after 24 h. For higher concentrations this
was only determined after 24 h, even if full conversion was not attained
(n.a.). Selectivity at >90% conversion is at last point before full
conversion (between 90 and 99% conversion).
We have noted this behavior before in the Pd(Ar-BIAN)(alk-
ene)-catalyzed semihydrogenation of 4-octyne with molecular
hydrogen.4 In that system, the optimum concentration of alkyne
was 0.32 M; more electron-poor alkynes give more stable
metallacycles, so the concentration above which palladacycle
formation effectively competes with hydrogenation will be lower
for aromatic alkynes than for aliphatic alkynes.23 Attempts to
isolate these palladacyclic species from the reaction mixture
were not successful. (Triphenyl)mesitylenes formed by reaction
(20) (a) Li, J.; Hua, R. M.; Liu, T. J. Org. Chem. 2010, 75, 2966–2970.
(b) Barrios-Francisco, R.; Garc´ıa, J. J. Appl. Catal., A 2010, 385, 108–
113. (c) Hori, J.; Murata, K.; Sugai, T.; Shinohara, H.; Noyori, R.;
Arai, N.; Kurono, N.; Ohkuma, T. AdV. Synth. Catal. 2009, 351, 3143–
3149. (d) Barrios-Francisco, R.; Garc´ıa, J. J. Inorg. Chem. 2009, 48,
386–393. (e) Campan˜a, A. G.; Este´vez, R. E.; Fuentes, N.; Robles,
R.; Cuerva, J. M.; Bun˜uel, E.; Ca´rdenas, D.; Oltra, J. E. Org. Lett.
2007, 9, 2195–2198. (f) Shirakawa, E.; Otsuka, H.; Hayashi, T. Chem.
Commun. 2005, 5885–5886. (g) Tani, K.; Iski, A.; Yamagata, T. Chem.
Commun. 1999, 1821–1822. (h) Tani, K.; Ono, N.; Okamoto, S.; Sato,
F. J. Chem. Soc., Chem. Commun. 1993, 386–387. (i) Cortese, N. A.;
Heck, R. F. J. Org. Chem. 1978, 43, 3985–3987.
Figure 4. Lineweaver-Burke plot: 1/[alkyne] vs 1/rate. Circles represent
values at [alkyne] < 0.15 M, triangles represent values at [alkyne] >
0.15 M.
Amatore and Jutand have already found that the affinity of
internal alkynes for coordinatively unsaturated Pd(PPh3)2 is very
low, significantly lower than that of terminal alkynes, so at low
concentrations the equilibrium will be on the side of noncoor-
dinated alkyne.21 Because of the observed substrate inhibition,
we analyzed the data for Michaelis-Menten kinetics. This gave
a good fit and Figure 4 shows the Lineweaver-Burke plot of
1/[rate] as a function of 1/[alkyne], which is linear at [alkyne]
< 0.15 M. Herewith, we found that Vmax ) 2.85 mmol/h (if
there were no catalyst deactivation at high alkyne concentration)
and KM ) 25.5 mM (substrate concentration at which rate )
1/2Vmax).22
(21) Amatore, C.; Bensalem, S.; Ghalem, S.; Jutand, A.; Medjour, Y. Eur.
J. Org. Chem. 2004, 366–371.
(22) Because linear extrapolation of the trendline from an inverse relation-
ship does not give equal weight to points at different concentrations
or reaction rates, Vmax and KM were obtained from the so-called Eadie-
Hofstee plot of rate/[substrate] vs rate. See: Hofstee, B. H. J. Nature
1959, 184, 1296–1298.
(23) (a) Ito, T.; Hasegawa, S.; Ishii, Y. J. Organomet. Chem. 1974, 73,
401–409. (b) Moseley, K.; Maitlis, P. M. Chem. Commun. 1971, 1604–
1605.
9
J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010 16903

A R T I C L E S
Hauwert et al.
Scheme 2. The Equilibrium between Precatalyst 1 and Substrate Gives Several Potentially Active Species
of palladacyclopentadiene-species with an additional equivalent
of alkyne have not been observed by NMR, nor by GC.
At high catalyst loading (Pd/alkyne-ratio), the selectivity for
Z-alkene is seriously affected (compare entries 1, 4, and 8 in
Table 4). The effect of concentration of alkyne on catalytic
behavior had already been observed in the solvent screening,
where more strongly coordinating solvents gave rise to a slower
but more selective reaction.14 When performing the reaction in
THF, even the addition of 1 mol % of MeCN induces a higher
selectivity and lower rate, and with 1 equiv of MeCN, near-
full selectivity is observed (Figure 5). It is well-known that
palladium(0)-complexes show highly dynamic behavior in
solution and they can easily exchange weakly coordinating
ligands, the equilibrium conditions depending on the nature and
relative concentrations of all ligating species.24 A lower ratio
of alkyne-to-palladium will on the whole lead to less-protected
palladium, which enhances the reactivity toward alkynes but
also toward the product alkene, thus, increasing over-reduction
to alkane.
react with another molecule of TEAF to give the saturated product,
which is immediately displaced by solvent and thus regenerates
complex I. This reasoning implies that the chemoselectivity of the
reaction is dictated by the lifetime of complex III, which is
determined not only by the coordinative properties of the solvent,
but also by the strength of the Pd-alkene bond. The strength of
the Pd-alkene bond depends on the electron density on the metal
and the pi-accepting properties of the alkene. Substituents on the
alkene (and alkyne) that increase the pi-accepting capacity will then
lead to decreased chemoselectivity. This reasoning is in agreement
with our previous finding that alkynes with electron-withdrawing
substituents (e.g., dimethyl butynedioate) give significant over-
reduction, while most simple aliphatic and aromatic alkynes give
the Z-alkene in 90-99% yield.14 Also the observed solvent-
competition fits very well with this explanation. As these are all
equilibrium reactions (except for the hydrogenation), part of the
alkene will also be displaced by alkyne, directly transforming III
to II. So, at higher catalyst/substrate ratios, the lifetime of species
III will increase and thus give a higher tendency for over-reduction,
which is what was indeed observed in the kinetics of substrate
and catalyst.
1.4. Rate Equation. Combination of the observed rate de-
pendencies allows us to set up an experimental rate law for
palladium-catalyzed semihydrogenation of 1-phenyl-1-propyne
with TEAF in acetonitrile:
Concerning the role of maleic anhydride, several possibilities
can be envisaged a priori; (i) it may remain coordinated to the
metal, (ii) it might dissociate, or (iii) it could be hydrogenated
to succinic anhydride. The latter is seen in alkyne hydrogena-
d(substrate)
) -kcat · [alkyne]0.28 · [TEAF] · [catalyst]
dt
(1)
Equation 1 is only valid at alkyne concentrations below 0.15
M, and shows that the reaction rate depends on the concentra-
tions of hydrogen donor, substrate, and catalyst in a positive
fashion. This means that alkyne and formate will both coordi-
nate to a Pd species before the rate-determining step. When the
alkyne concentration is increased to above 0.15 M, part of the
catalyst reacts with alkyne to form a catalytically inactive
species. This makes the concentrations of catalyst and alkyne
interdependent, so setting up a rate law in this concentration
range would be meaningless.
1.5. MechanismsCompetition. A first interpretation of the
kinetic data leads to a simplified image of the catalytic cycle, shown
in Scheme 3, which focuses on the role of competition between
substrate and solvent for palladium. Starting from in situ generated
complex I, solvent is replaced by a substrate molecule to give a
complex II, after which the alkyne is hydrogenated by TEAF (Vide
infra), giving a complex of type III. The alkene product can be
displaced by the more strongly coordinating solvent, closing the
catalytic cycle to obtain the desired alkene product. For a more
weakly coordinating solvent, complex III is longer lived and could
Figure 5. Transfer hydrogenation of 1-phenyl-1-propyne in THF, with
increasing amounts of MeCN; MeCN/alkyne ratio vs rate and alkene
selectivity. Rate is depicted in diamonds and relates to values on the primary
y-axis; alkene selectivity is depicted in squares and relates to values on the
secondary y-axis. Conditions: [alkyne] ) 0.15 M, [catalyst] ) 15 mM,
[TEAF] ) 0.75 M in refluxing THF.
9
16904 J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010

Mechanism of Alkyne Transfer Hydrogenation
A R T I C L E S
Scheme 3. Simplified Catalytic Cycle, Focusing on the Effect of
Competition between Solvent, Alkynes, and Alkenes for the
Catalyst
isotope effect does not necessarily imply that the transfers of
the acidic and formic hydrogens occur in a single elementary
step. They showed that the transfer of hydrogens may take place
in a concerted fashion or it may involve two separate hydrogen
transfer steps with approximately equal barriers, and that these
processes can be distinguished.25 For a concerted reaction in
which both hydrogens are transferred in a single step, the kinetic
isotope effect for the doubly labeled material should be equal
to the product of the two individual isotope effects. In our case,
KIECHOH/CHOD × KIECHOH/CDOH ) 5.95, which is appreciably
larger than the experimentally observed KIECHOH/CDOD of 3.60.
Hence, we conclude that the acidic hydrogen and the hydridic
hydrogen from formic acid are transferred in two distinct steps,
and that these transfers have approximately equal activation
energies.
Apart from the effect of isotopic labeling on the reaction rate,
this experiment also allows us to observe whether the hydrogens
retain their identity during the reaction, and if so, whether a
distinction is made in the mechanism between the two carbons
of the alkyne. For this, we looked at the 1H NMR (integral and
multiplicity) and 2H NMR (integral) of the products. In the 1H
NMR of the hydrogenation products, we see that the distribution
of protons between the two olefinic positions is roughly equal:
for both monodeuterated formic acids, the integrals of both
olefinic protons add up to one hydrogen compared to the five
aromatic hydrogen nuclei. For the dideuterated formic acid, a
minor amount of not-fully deuterated product is observed in
1H NMR; the integral of the olefinic protons accounts for
approximately 0.12 hydrogens relative to five aromatic hydro-
gens. The presence of unlabeled product is probably caused by
the faster reaction with nonlabeled formic acid. This partly
protonated product shows a broad singlet and a broadened
Table 5. Effect of Deuterated Formic Acids on the Rate of
Transfer Hydrogenation of 1-Phenyl-1-propyne to
Z-ꢀ-Methylstyrene by in Situ Generated 1a
entry formic acid [formic acid] mol L-1 rate (× 10-3) mol h-1 kcat L2.3 mol-2.3 h-1
1
2
3
4
HCO2H
HCO2D
DCO2H
DCO2D
0.38
0.37
0.36
0.36
1.32 ( 0.12
0.54 ( 0.09
0.52 ( 0.10
0.34 ( 0.04
5.81 ( 1.38
2.39 ( 0.82
2.38 ( 0.88
1.61 ( 0.39
a Conditions: Reactions are performed in refluxing MeCN, using in
situ generated catalyst 1; [1-phenyl-1-propyne] ) 0.11 M, [NEt3] ) 0.50
M, [catalyst] ) 1.05 mM, in refluxing MeCN, using in situ generated
precatalyst 1. All reactions were done in triplicate, values are averaged.
Table 6. Kinetic Isotope Effects of Transfer Hydrogenation of
1-Phenyl-1-propyne to Z-ꢀ-Methylstyrene by in Situ Generated 1
1
quartet in H NMR, as would be expected when the vicinal
position is occupied by deuterium instead of hydrogen. In the
product of unlabeled formic acid, a doublet and a doublet of
quartets are seen for the olefinic protons, whereas for mono-
deuterated formic acids, these peaks are broadened due to the
presence of mono-, as well as nondeuterated products.
KH/KD
HCO2H/DCO2H
HCO2D/DCO2D
HCO2H/HCO2D
DCO2H/DCO2D
HCO2H/DCO2D
2.43 ( 0.33
1.48 ( 0.33
2.45 ( 0.35
1.47 ( 0.35
3.60 ( 0.27
The 2H NMR of the products shows that the deuterium
distribution depends on the position of the 2H in the formic acid:
for HCO2D, the integral of the deuteron at the R-position was
15% larger than that of the deuteron at the ꢀ-position, whereas
with DCO2H and with DCO2D, the integral of the R-deuteron
was 15% smaller than that of the ꢀ-deuteron.26 First of all, this
means that the hydrogens retain their identity during the reaction.
Furthermore, the first hydrogen is delivered to the more
accessible ꢀ-position of the alkyne and originates from the
formic position, assuming the sterically most accessible position
is attacked first. In case of the 1-phenyl-1-propyne, the difference
in steric demand between the Ph-C and C-CH3 sites is not
very large, which may explain why a preference of ‘only’ 15%
for the ꢀ-position has been observed. For all labeled formic
tions with Pd(Ar-Bian)(dmfu).4 In the current case, we have
neither observed maleic anhydride nor succinic anhydride in
the GC of aliquots of the reaction, nor in the NMR-spectrum
of the product, so we postulate that maleic anhydride remains
coordinated to the complex during reaction (Vide infra).
2. Isotopic Labeling of the Hydrogen Donor. We investigated
the role of the two different hydrogens in formic acid by
performing the reaction with the various deuterium-labeled
formic acids (HCO2D, DCO2H, and DCO2D). It appears that
the reaction rate decreases with increasing deuterium-content;
note that reactions of both monodeuterated formic acids (HCO2D
and DCO2H) proceed at the same rate (Table 5). Comparing
the rates of the reactions with labeled formic acids, a primary
kinetic isotope effect (KIE) of 2.4 is observed on both the formic
(CH/CD) and the acidic (OH/OD) position, while the kinetic
isotope effect upon double isotopic substitution is 3.6 (Table
6).
2
acids, approximately 10% of the signal in H NMR is seen at
the γ position. Indeed, monitoring the deuteration experiments
(24) Clement, N. D.; Cavell, K. J.; Ooi, L.-L. Organometallics 2006, 25,
4155–4165.
The observation of a kinetic isotope effect due to single
isotopic substitution, on either the formic or the acidic position,
means that both the bond to the formic hydrogen and that to
the acidic hydrogen are broken or formed in the rate-determining
step. Casey et al., in their studies on transfer hydrogenation of
imines, have shown that the observation of a primary kinetic
(25) Casey, C. P.; Singer, S. W.; Powell, D. R.; Hayashi, R. K.; Kavana,
M. J. Am. Chem. Soc. 2001, 123, 1090–1100.
(26) We assume that the product is mostly monodeuterated, as one of the
referees noted. We believe this is a reasonable assumption, based on
the integrals in the 1H-NMR of the product. Also, if a substantial
amount would be non- or dideuterated, this would be seen in the
1
multiplicity of the peaks in the H-NMR.
9
J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010 16905

A R T I C L E S
Hauwert et al.
Scheme 4. Synthesis of Various Pd(IMes)(MA)-Complexes
with GC reveals a trace of allylbenzene as a transient species,
2
which accounts for the H-scrambling.
2.1. Catalyst Isolation. In previous work, we have found that
in situ generated Pd0(NHC)(MA)-complexes are more selective
catalysts for the hydrogenation of alkynes than isolated Pd-
(NHC)(MA)2-complexes.3 Therefore, the transfer semihydro-
genation of alkynes was developed with in situ generated
Pd0(IMes)(MA)-complexes,14 where the open coordination site
is occupied by solvent. As the reaction is slower but more
selective in MeCN than in THF, we postulated that in aceto-
nitrile a [Pd0(IMes)(MA)(MeCN)] species is the source of active
catalyst. To verify this, we attempted the isolation of such a
species. Indeed, by concentration of a catalyst solution in MeCN,
an off-white solid could be precipitated, which appeared from
NMR to be complex 3, a Pd(IMes)(MA)-complex with two
coordinated acetonitrile molecules (Scheme 4).
Figure 6. Simulated structure for compound 3.
Scheme 5. Synthesis of Palladacyclopentadiene(NHC)-Complexes
In 1H NMR, we observe that the maleic anhydride hydrogens
in 3 have become inequivalent (2.91 and 3.45 ppm), while at
elevated temperatures, these peaks coalesce to a broad singlet
at 3.0 ppm (60 °C, 300 MHz). This indicates that at room
temperature there is hindered rotation around the Pd-alkene
bond and the alkene hydrogens experience inequivalent magnet-
ic environments. This effect has been seen in similar Pd0-
(NHC∧N)(MA)-complexes, and there the X-ray structure shows
that one of the MA-proton signals is near the shielding region
of the aromatic ring, explaining the lower chemical shift.15 Also,
the methyl peaks on the ortho-position of the IMes ligand
experience different environments (2.08 and 2.10 ppm), indicat-
ing a spatial orientation of the IMes that differentiates between
the two aromatic groups.
maleic anhydride was needed, giving the previously reported
complex Pd(IMes)(MA)2 4.3
2.2. Palladoles as Substrate-Inhibited Catalyst. In Scheme
2, we postulated the formation of palladacyclopentadiene species
as a sink for active catalyst, to explain the inverse order in
substrate at high concentrations. Although we were not able to
isolate such palladoles from the reaction mixture, the synthesis
of palladacycle 5 with an IMes-moiety was possible under
certain conditions (Scheme 5).
Using NOESY NMR-experiments, we found many NOE
cross-peaks, most notably that of the NHC-backbone protons
with the maleic anhydride protons. For a complex with square
planar geometry, these C-H bonds are perpendicular to each
other and further away than in a tetrahedral geometry, as is
common for 18-electron Pd complexes. Despite our best efforts,
attempts to grow X-ray quality crystals of 3 failed. We have
optimized the geometry of this complex with density functional
theory, with which we observed dissociation of one of the
acetonitrile ligands (in the gas phase). In solution, the second
molecule of acetonitrile may still be weakly coordinated, in
accordance with the relatively small shift upon coordination and
the integral of 6H of the MeCN protons. Additionally, the
inequivalence of both maleic anhydride protons and both
aromatic rings is clearly seen in the simulated structure of 3
(Figure 6). We have also tried to obtain the solvent-stabilized
complex from a catalyst solution in THF, but this only gave
decomposition to palladium black. An additional equivalent of
For simple aromatic alkynes, the reaction of several Pd-
precursors with a carbene and/or excess alkyne failed to produce
the desired palladacyclopentadiene(NHC)-complexes. Substitu-
tion of alkene from solvent complexes 1 or 3, or formation of
a Pd(IMes)-complex from Pd0-precursors 2 or 6 followed by
cycloaddition of alkyne proved unsuccessful. Cycloaddition for
Pd(tBuDAB)(alkyne)-complex 7 does not occur, as already
shown by tom Dieck,27 but coordination of IMes to alleviate
the steric demand of the diimine ligand and subsequent
cyclization with excess alkyne also did not yield the desired
product. We were able to isolate complex 5 when using the
more electron-withdrawing dimethyl butynedioate (dmbd), but
only if the palladacyclopentadiene moiety had already been
formed prior to coordination of the NHC. This shows that the
(27) tom Dieck, H.; Munz, C.; Mu¨ller, C. J. Organomet. Chem. 1990, 384,
243–255.
9
16906 J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010

Mechanism of Alkyne Transfer Hydrogenation
A R T I C L E S
Table 7. Catalysts Tested in Transfer Hydrogenation of
1-Phenyl-1-propyne (Structures in Chart 1)
Chart 1. Precatalysts Tested in Transfer Hydrogenation of
1-Phenyl-1-propyne
entry
catalyst
TOF h-1
selectivity (Z/E/alkane)a
1
2
3
4
5
6
7
8
1
100.6
74.6
18.0
25.6
108.6
21.2
14.7
1.0
95.2/3.9/0.9
87.3/9.2/3.5
92.3/6.1/1.6
80.1/10.0/9.9c
91.6/5.4/3.0
96.0/3.0/1.0
96.4/2.7/0.9
95.8/4.2/0.0
aged 1b
3
4
5
5 + 1% MA
8
8, no base
a Selectivity determined by GC, after 24 h, conditions in Supporting
Information. b Two months aged stock solution. c Selectivity after 90%
conversion (3.5 h) is 93.5/5.0/1.6, TOF ) 15.9 mol/(mol/h).
formation of these proposed cyclopalladated species is possible,
but not straightforward under catalytic conditions, and may only
occur in the presence of >150 equiv of alkyne. Reaction of in
situ generated complex 1 with 160 equiv of 1-phenyl-1-propyne
gives a complex that spectroscopically resembles 5. Complex
5 is the first example of a palladacyclopentadiene species with
a coordinated N-heterocyclic carbene, while complex 7 is one
of the few examples of Pd-alkyne complexes that have been
crystallized.28 Because of the high electron-density on the metal,
it shows very strong back-bonding, with an alkyne carbon-carbon
bond distance of 1.287(2) Å) (see Supporting Information).
2.3. Catalytic Studies with Isolated Catalysts. To confirm
whether the isolated solvent-complex 3 indeed resembles the
catalytically active species, we compared its activity and
selectivity with the in situ generated precatalyst 1. Complex 3
exhibits the same excellent chemoselectivity as the in situ
prepared catalyst (Table 7, entry 1 vs 3). The Z-selectivity for
complex 3 is somewhat lower, but the absence of over-reduction
is seen for both precatalysts. Employing the isolated complex
3 as precatalyst leads to a lower rate of transfer hydrogenation
of 1-phenyl-1-propyne compared to the same reaction with the
in situ generated precatalyst 1; still no induction period is
observed. Although one would predict the same solution-phase
structures when starting from 1 and 3, the reaction with in situ
generated species still contains the diimine ligand from 2. So
in 1 additional ligand is present that probably facilitates a
dissociative step in the mechanism to increase the rate of
reaction, while at the same time aiding MeCN in the competition
for Pd(alkene) intermediate III to give higher alkyne selectivity.
We have also repeated the reaction with a catalyst stock
solution in MeCN that had been standing for 2 months and it
shows only little decomposition: a decrease of rate is observed
and Z-selectivity is slightly lower, but over-reduction is still
marginal (Table 7, entry 2). This shows that the solvated catalyst
species is very stable and that after partial decomposition it is
able to isomerize alkenes but does not reduce them. In
comparison, the palladium bis(maleic anhydride) complex 4 as
precatalyst is neither as fast nor as selective as solvent-stabilized
complexes 1 or 3.
reasonable to assume that the catalytic cycle starts from a Pd0-
complex. For complex 5, this implies that the formation of the
palladacyclopentadiene from a Pd(IMes)(alkyne)2-complex is
reversible under these conditions, even for the electron-poor
dimethyl butynedioate. After retrocycloaddition from 5, the
formed Pd(IMes)(dmbd)2 complex loses its alkynes either by
dissociation or hydrogenation and generates Pd0(IMes)-species
with ligands that are only weakly coordinating compared to
complex 3. When the reaction is performed with 5 and 1 equiv
of maleic anhydride to protect the ‘naked’ Pd0(IMes)-species
formed, the same excellent selectivity is observed as for 3, along
with a decrease in rate (Table 7, entry 6). Surprisingly, the
presence of maleic anhydride on the complexes moderates the
reactivity toward alkynes, and thus decreases the tendency
toward over-reduction of product alkene to alkane.
From the TEAF-kinetics, we found that an equimolar amount
of base to acid is required to obtain high reaction rates, but
when using Pd(NHC-amine) complex 8, the reaction also
proceeds without additional base (Table 7, entries 7-8).15 With
this ligand, the hemilabile amine functions as an internal base
during hydrogenation. After hydrogenation of alkyne, the amine
can re-coordinate to stabilize the complex and promote the
dissociation of alkene, completely suppressing the over-reduc-
tion to obtain full chemoselectivity for alkynes.
3. NMR Studies. To gain more insight into the nature of the
species present during catalysis, we have monitored the reaction
by 1H NMR, which had to be done in DMSO-d6 as the solubility
of complex 3 in acetonitrile is too low. Prior to the reaction, a
control experiment was performed without alkyne but with 1
equiv of TEAF at 60 °C. Signals that could belong to a metal
hydride were not identified, but an upfield shift was observed
in the N-ethyl signals, while the formic proton shifted downfield,
indicating coordination of the formate anion to Pd and the
presence of a triethylammonium cation. In the absence of alkyne,
the complex decomposed to the corresponding imidazolium salt
and Pd black. Also, a signal is seen at 5.97 ppm, of which the
carbon resonates at 136.57 ppm (1JCH ) 158 Hz), indicative
for maleic anhydride alkene carbons (Vide infra). We then
performed the transfer hydrogenation reaction, adding 2 equiv
of 1-phenyl-1-propyne and then 2 equiv of triethylammonium
formate to complex 3 in DMSO-d6 (Figure 7). Within 10 min
after addition, the formation of Z-ꢀ-methylstyrene was observed
Although palladacyclopentadiene-complexes could not be
isolated from the catalysis reaction mixture, the reaction does
proceed with a relatively high reaction rate when employing
palladole 5 as a catalyst precursor at [alkyne] < 0.15 M (Table
7, entry 5). The chemoselectivity is lower for 5, compared to
in situ generated 1 or isolated 3, but complex 5 is equally fast
as precatalyst 1. We have previously seen that PdII-complexes
are not selective transfer hydrogenation catalysts,15 so it is
(28) McGinnety, J. A. J. Chem. Soc., Dalton Trans. 1974, 59, 1038–1043.
9
J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010 16907

A R T I C L E S
Hauwert et al.
Figure 8. (a) 46 MHz 2H NMR spectra of 3-d2 during reaction with
1-phenyl-1-propyne and triethylammonium formate in DMSO at 25 °C;
top spectrum is of pure, noncoordinated 2,3-dideuterio maleic anhydride.
The different species are marked above the spectra: diamonds (starting
catalyst 3-d2), triangles (active catalyst), circles (catalyst after reaction) and
squares (uncoordinated maleic anhydride-d2).
Figure 7. The 300 MHz 1H NMR spectra of 3 and 1-phenyl-1-propyne in
DMSO-d6 at 60 °C, after addition of TEAF in DMSO-d6, and during
reaction. The different species are marked above the spectra: squares (formic
hydrogen), circles (aromatic hydrogens and imC4,5-hydrogens of the catalyst),
pentagons (alkene and methyl hydrogens of the product), triangles (am-
monium proton), and diamonds (methyl and methylene peaks of triethyl-
ammonium cation).
2
the reaction in an NMR-tube while monitoring by H NMR.
The reaction was carried out at 25 °C to decrease the reaction
rate, because the low solubility of the catalyst requires longer
acquisition time, and spectra were recorded every hour over a
72 h period. Figure 8 shows that during the reaction only part
of the catalyst is in the native state (diamonds, 2.4 ppm), while
part of the catalyst bears maleic anhydride with a more
downfield shift (triangles) around 5.5 ppm. The pi-backdonation
should be significantly diminished to shift the maleic anhydride
signal downfield to such an extent; either an additional very
strong pi-acceptor is coordinated, or the orbital overlap is
decreased because of steric congestion by incoming substrate,
or a Pd(II) hydride is formed. To the best of our knowledge
there is no precedent of coordinated maleic anhydride signals
at such low field, and we are hesitant to claim this with such
limited evidence. Another possibility is that hydrolysis of
(uncoordinated) maleic anhydride to maleic acid occurs, which
(1H-signals at 6.39 ppm, 5.76 ppm and 1.85 ppm); no formation
of either the corresponding E-isomer or propylbenzene was
observed. The formic hydrogen shifted the same as in the control
experiment (8.35 f 8.55 ppm), but now another very broad
signal was observed at 5.66 ppm which is attributed to the proton
on the triethylammonium cation. As the reaction proceeds, this
signal shifts to 3.36 ppm, indicating that the positive charge is
distributed over more than one amine as the reaction mixture
becomes more basic. This shift is also seen in the ethyl signals
of NEt3, though less pronounced.
The remainder of the complex proton signals behaved in a
similar manner to the control experiment, with an upfield shift
of the triethylammonium peaks and downfield shift of the
formate anion (Figure 7, diamonds resp. square). Notably, some
decomposition of the catalyst is seen but remarkably less than
in the control experiment. As the ratio of alkyne/TEAF ) 1,
the over-reduction at high catalyst loading is not observed.
From this experiment, we conclude that a palladium formate
species is formed, which in the presence of alkyne transfers a
hydride to the coordinated alkyne, probably via Pd. In the
absence of alkyne, the palladium formate decarboxylates to form
a Pd(II) hydride which decomposes upon reductive elimination
of imidazolium salt. In neither case signals were observed in
the hydride region. This could indicate that the hydride is rapidly
transferred to the alkyne, making the decarboxylation the first
rate-determining step. However, another reason for not observing
a hydride signal is fast relaxation of the hydride because of the
quadrupole Pd; hence, we cannot rule out that the subsequent
migratory insertion of alkyne into Pd-H is the first rate-
determining step.
1
causes an upfield shift of the alkene H signals to around 6.0
ppm. However, after the reaction, this signal shifts to 6.8 ppm,
and upon addition of an additional equivalent of MA-d2 after
the reaction, the signal at 6.8 ppm increases and only after
addition of more than 5 equiv of MA-d2 some uncoordinated
MA-d2 is observed at 7.4 ppm. This observation shows that
exchange of coordinated and uncoordinated maleic anhydride
is fast on the NMR time scale, implying that free maleic
anhydride is already present during the reaction. By comparison
2
with H NMR spectra of authentic samples of succinic anhy-
dride, we have ascertained that this is not present in the reaction
medium, so maleic anhydride is not hydrogenated off to activate
the complex. Although we cannot conclusively identify the
nature of the formed species, it is clear that part of the maleic
anhydride remains coordinated during the reaction.
3.1. Labeled Catalyst. In previous research, we have shown
that during alkyne hydrogenation with Pd0(Ar-BIAN)(alkene)-
complexes as catalysts, the alkene is hydrogenated off to form
the active catalyst.4 We wondered whether this would also be
the case for our current NHC-based precatalysts 1 or 3.
3.2. Proposed Catalytic Cycle. With the catalytic cycle of
alkyne hydrogenation proposed for Pd(diimine)-complexes in
mind,29 we initially thought that the mechanism would consist
of oxidative addition of the formic acid O-H-bond, migratory
insertion of the hydride into the Pd-alkyne bond, decarboxy-
lation of the formyl anion to CO2, and reductive elimination
from a Pd(alkenyl)(hydride). On the basis of our current
1
However, by H NMR this cannot be verified, because the
hydrogens of noncoordinated maleic anhydride would be
obscured by the aromatic hydrogens of the mesityl-substituent
on the carbene ligand. Therefore, we prepared complex 3-d2 in
which the maleic anhydride is deuterium-labeled, and performed
(29) Dedieu, A.; Humbel, S.; Elsevier, C.; Grauffel, C. Theor. Chem. Acc.
2004, 112, 305–312.
9
16908 J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010

Mechanism of Alkyne Transfer Hydrogenation
A R T I C L E S
Scheme 6. Proposed Catalytic Cycle for Transfer Hydrogen of Alkynes to Z-Alkenes, Catalyzed by Complex 1 with Triethylammonium
Formate As Hydrogen Donor
experimental results, this hypothesis must clearly be revised.
For the hydrogen donor, we found that (1) a Pd(formate)
intermediate is involved, (2) the nature and concentration of
base is important, and (3) the kinetic isotope effects dictate two
rate-determining steps with similar activation energies, involving
both hydrogens of formic acid. The studies of the catalyst and
kinetics showed that (4) maleic anhydride is not hydrogenated
off as expected previously, but is partly coordinated during
reaction and moderates the catalytic activity; also (5) substrate
and formate are coordinated to Pd before the rate-determining
steps. Furthermore, (6) the coordination of solvent versus the
alkyne is competitive and strongly influences the chemoselec-
tivity. We therefore propose a mechanism that starts from
species I, a Pd0-complex containing an N-heterocyclic carbene
ligand, maleic anhydride, and acetonitrile (Scheme 6).
by the order in which the proton and hydride are added to the
alkyne, and this mechanism resembles our proposed catalytic
cycle.13 For a mechanism in which the hydride is added first
and the Pd-alkenyl bond is broken by protonolysis, the product
is the Z-alkene; the E-alkene is formed if the proton is added to
alkyne prior to hydride transfer. The reason they give for the
high Z-selectivity is that the approach of the proton to the
Pd(alkenyl)-species is hindered by the aryl-group, disfavoring
formation of E-alkene. This rationalizes two of our observations,
namely, that (1) bulky diphenylacetylene is hydrogenated in
higher (>99% Z) selectivity than 1-phenyl-1-propyne, and (2)
Pd(NHC∧amine)(MA) complex 8 is a more selective catalyst
toward Z-alkene in the absence of an external base.
Conclusion
Displacement of solvent from I by alkyne, and subsequent
coordination of formate anion forms complex III, probably with
dissociation of maleic anhydride to retain a 16-electron con-
figuration. Jutand and Amatore have shown that similar anionic
Pd(0)-species containing, for example, chloride or acetate as
anions are the active species in Heck and cross-coupling
reactions, much more reactive than their neutral counterparts.30
From species III, the coordinated formate decarboxylates to give
palladium hydride species IV, which after migratory insertion
into the Pd-alkyne bond gives species V; one of these is the
first rate-determining step. The alkene is obtained by protolytic
cleavage, which is the second rate-determining step to give the
Pd-alkene species VI, and the alkene is displaced by solvent
or alkyne to close the catalystic cycle. The lifetime of species
VI determines the amount of over-reduction, and thus the
chemoselectivity. The groups of Lledo´s and Joo´ have shown
by calculations that for alkyne hydrogenation with
[{RuCl2(mtppms)2}2] in water the Z/E-selectivity is determined
The mechanism of Pd0(IMes)-catalyzed transfer hydrogena-
tion of alkynes has been elucidated for the case of 1-phenyl-
1-propyne. Kinetics and several observations allow us to propose
a viable catalytic cycle for this transformation. A number of
similarities but also differences are seen when comparing to
transfer hydrogenation of carbonyls or to alkyne hydrogenation
using molecular hydrogen. We have found that the transfer of
hydrogens from the hydrogen donor to the substrate takes place
in two separate rate-determining steps. These hydrogens keep
their identity during the reaction, as seen from the distribution
of the 2H-label across the alkene bond when formic acid
isotopomers were used. When these observations are compared
with transfer hydrogenation of ketones, a resemblance with an
inner-sphere monohydride mechanism comes to mind, but a
metal-hydride species for such a mechanism could not be
observed. Concerning other known alkyne semihydrogenations,
a comparison of Pd0(NHC)(alkene) with Pd0(Ar-BIAN)(alkene)
is obvious. However, in the present study, we have a mono-
dentate NHC ligand instead of a bidentate BIAN-ligand; hence,
in the current case, the alkene need not be hydrogenated off to
(30) Amatore, C.; Jutand, A. Acc. Chem. Res. 2000, 33, 314–321.
9
J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010 16909

A R T I C L E S
Hauwert et al.
obtain a catalytically active complex. This allows the maleic
anhydride ligand to moderate the electron-density of the metal,
a property we found of importance for the chemoselectivity of
the reaction. The main difference between the known hydro-
genation of alkynes with molecular hydrogen and the current
catalytic system involving transfer hydrogenation is a strong
competition between substrate and solvent for a Pd(alkene)
intermediate occurring in the transfer hydrogenation, which is
not observed in the hydrogenation with molecular H2. This
competition ensures negligible concentration of unsaturated
Pd(alkene) species, decreasing the chance of reaction with a
second equivalent of hydrogen. As such this competition
between substrate and solvent is important for the selectivity
of the transfer hydrogenation and in fact determines the complete
absence of over-reduction, a selectivity that has proven to be
difficult to obtain for other systems for catalytic hydrogenation
of alkynes.
the reaction itinerary, but it also provides new insights, upon
which new catalysts for transfer hydrogenation reactions may
be designed.
Acknowledgment. This research was funded by the National
Research School Combination Catalysis (project no. 2004-2008-
UVA-Elsevier-02/03). The authors thank Jan Meine Ernsting and
Jan Geenevasen for help with the heteronuclear NMR-measure-
ments, Han Peeters for mass spectroscopic measurements and Anna
Pavlova and Evert-Jan Meijer for the geometry optimization of
complex 3. We thank the International Research Training Group
1444 for collaboration and additional funding.
Supporting Information Available: Concentrations and reac-
tion rates for kinetic measurements depicted in Tables 2-5 and
7, kinetic study methodology, general experimental methods,
experimental details and characterization for 3, 3-d2, 5, 7 and
the palladacyclopentadiene bis(acetonitrile)-complex (PDF);
compound I (CIF). This material is available free of charge via
the Internet at http://pubs.acs.org.
Prospectively, the stability of the Pd(product alkene) inter-
mediate VI might be tuned by adjusting the electronic properties
of the catalyst precursor, which may lead to higher selectivity
for alkynes with electron-withdrawing groups. In this way, our
mechanistic investigation does not only explain the details of
JA1062407
9
16910 J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010