Gas-Phase energetics of reductive elimination from a Palladium(II) N-heterocyclic carbene complex

Article abstract of DOI:10.1002/chem.200902929

Energy-resolved collision-induced dissociation experiments using tandem mass spectrometry are reported for an phenylpalladium N-heterocyclic carbene (NHC) complex. Reductive elimination of an NHC ligand as a phenylimidazolium ion involves a barrier of 30.9(14) kcal mol^-1, whereas competitive ligand dissociation requires 47.1(17) kcal mol^-1. The resulting threecoordinate palladium complex readily undergoes reductive C-C coupling to give the phenylimidazolium π complex, for which the binding energy was determined to be 38.9(10) kcalmol^-1. Density functional calculations at the M06L//BP86/TZP level of theory are in very good agreement with experiment. In combination with RRKM modeling, these results suggest that the rate-determining step for the direct reductive elimination process switches from the C-C coupling step to the fragmentation of the resulting a complex at low activation energy.

Full text of DOI:10.1002/chem.200902929

DOI: 10.1002/chem.200902929
Gas-Phase Energetics of Reductive Elimination from a Palladium(II)
N-Heterocyclic Carbene Complex
Erik P. A. Couzijn, Eva Zocher, Andreas Bach, and Peter Chen*[a]
ꢀ
Abstract: Energy-resolved collision-in-
duced dissociation experiments using
tandem mass spectrometry are report-
ed for an phenylpalladium N-heterocy-
clic carbene (NHC) complex. Reduc-
tive elimination of an NHC ligand as a
phenylimidazolium ion involves a bar-
rier of 30.9(14) kcalmolꢀ1, whereas
competitive ligand dissociation requires
47.1(17) kcalmolꢀ1. The resulting three-
coordinate palladium complex readily
undergoes reductive C C coupling to
give the phenylimidazolium p complex,
for which the binding energy was deter-
mined to be 38.9(10) kcalmolꢀ1. Densi-
ty functional calculations at the M06-
L//BP86/TZP level of theory are in
very good agreement with experiment.
In combination with RRKM modeling,
these results suggest that the rate-de-
termining step for the direct reductive
elimination process switches from the
ꢀ
C C coupling step to the fragmenta-
ꢀ
Keywords: carbene ligands · C C
tion of the resulting s complex at low
coupling · density functional calcu-
lations · mass spectrometry · palla-
dium
activation energy.
Introduction
ꢀ
The formation of C C bonds under mild conditions by pal-
ladium-catalyzed cross-coupling reactions has become
widely used in organic synthesis.[1] The archetypal catalytic
cycle involves a Pd0 species that undergoes oxidative addi-
tion, transmetalation, and reductive elimination (Scheme 1).
However, the exact mechanism is still poorly understood,
mainly because the transmetalation and successive reductive
elimination are generally too fast to be studied individual-
ly.[2] Only recently were the solution-state kinetics for reduc-
tive elimination of two isolable cis-diaryl palladium(II) com-
plexes reported.[3] Cavell et al.[4] showed that also N-hetero-
cyclic carbene (NHC) hydrocarbyl palladium(II) complexes
can readily decompose by concerted reductive elimination
of hydrocarbyl imidazolium ions, in contrast with the
common view of NHCs as mere spectator ligands.[5] This re-
activity parallels biaryl formation if one envisages the in-
ꢀ
Scheme 1. Simplified palladium-catalyzed C C cross-coupling cycle.
volvement of an imidazolium-2-yl palladium(II) resonance
structure (Scheme 2). The superior catalytic activity found
for NHC-based precursors, which is usually attributed to the
strong s-donor properties of NHCs, may alternatively stem
from the thus-generated low-coordinate Pd0 species.
Here we report on the use of electrospray tandem mass
spectrometry (ESI-MS)[6] to study the gas-phase reactivity of
a palladium NHC complex. Energy-resolved collision-in-
duced dissociation (CID) experiments are reported for the
reductive elimination of an imidazolium species and for
competitive ligand dissociation (cf. Scheme 2). The three-co-
ordinate complex generated on loss of an NHC can still un-
dergo reductive elimination, and we present a model to de-
scribe the energy dependence for such sequential-reaction
[a] Dr. E. P. A. Couzijn, Dr. E. Zocher, Dr. A. Bach, Prof. Dr. P. Chen
Laboratorium fꢀr Organische Chemie, ETH Zꢀrich
Wolfgang-Pauli-Strasse 10, 8093 Zꢀrich, Switzerland
Fax : (+41)44-632-1281
E-mail: peter.chen@org.chem.ethz.ch
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200902929. It includes the 24-
pole ion guide design, mass spectra, and cross sections, details of the
threshold CID measurements and fitting procedures and of the
RRKM rate calculations, and optimized geometries and energies.
5408
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 5408 – 5415

FULL PAPER
activation with argon, 2a,b underwent reductive elimination
of 1,3-dimethyl-2-phenyl-1H-imidazolium (3, m/z 173) and
competitive decomplexation of one N-heterocyclic carbene
(2a) or the triphenylphosphine ligand (2b) to give tricoordi-
nate palladium complex 4 at m/z 375. Density functional cal-
culations (see below) indicated that 4 can undergo rather
ꢀ
facile reductive C C coupling to give the more stable phe-
nylimidazolium p complex 4-p, so that both species are ac-
cessible at high energy. By either increasing the tube lens
voltage to 175 V or using 30 V source collision-induced dis-
sociation (CID), 4/4-p was generated in the spray and was
confirmed to eliminate 3 selectively on CID.
Scheme 2. Resonance structures for cationic palladium NHC complexes,
and associated fragmentation reactivity.
CID threshold measurements: Energy-resolved reaction
cross sections were acquired on a customized Finnigan MAT
TSQ-700 (see Experimental Section). The zero-pressure ex-
trapolated cross sections are shown in Figure 1; Table 1 sum-
marizes the fitting results for deconvolution with L-CID[8]
(see the Supporting Information for details and results with
other transition-state models). Collision of 2a with xenon
gas was carried out to monitor the reductive elimination of
3 (Figure 1A, green diamonds) and competitive NHC ligand
dissociation giving 4/4-p (red circles). Above a collision
energy ECM of about 5 eV, the experimental cross section for
formation of 3 rises more steeply at the expense of that of 4/
4-p, which indicates the occurrence of sequential reductive
elimination of 3 from 4.[9] Therefore, we performed two-
channel fits of the primary reaction cross sections up to
cross sections. The experimental results are complemented
by a detailed computational study, which indicates that the
direct reductive elimination channel proceeds via a s com-
plex, whose fragmentation becomes rate-limiting at low acti-
vation energy. Additionally, we evaluated several density
functionals for accurate reproduction of the gas-phase reac-
tion barriers, a validation of importance for our ongoing
ꢀ
mechanistic research on the palladium-catalyzed C C cou-
pling cycle.
Results and Discussion
Gas-phase reactivity: Treatment of iodo-bridged dimeric
palladium complex 1 with 1,3-dimethyl-1H-imidazol-2-yli-
dene silver iodide [(NHC)AgI] in acetonitrile generated cat-
ionic complex 2a, as observed at m/z 471 by ESI-MS
(Scheme 3). The only other intense signal corresponded to
E
CM =4.5 eV, and then simulated the reaction cross sections
over the full collision-energy range (solid lines in Fig-
ure 1A). Thanks to the physically realistic treatment of the
collisional activation in L-CID,[8] the simulated curves relia-
bly reflect the primary reaction cross sections in the absence
of the sequential reductive elimination. Thus, the difference
between the simulated and experimental cross section for
formation of 4/4-p (red solid line vs. red circles) affords an
estimate of the cross section for sequential conversion of 4
to 3 (Figure 1B).
To fit the sequential CID threshold curve[10] with L-CID,
we implemented the following model (see the Supporting
Information): the collisional activation was treated for 2a
with the previously fitted parameters for the primary reac-
tion giving 4. The primary reaction barrier was then sub-
tracted and the remaining internal energy scaled to account
for the energy partitioning on NHC ligand dissociation. Fi-
nally, the fitting procedure was performed with the degrees
of freedom and number of free rotors for 4. As the sequen-
tial reaction sets in at energies where the primary reaction
cross section has not yet leveled off, the kinetic shift was
treated with the appropriate overall rate equation. In this
way, all aspects of the sequential reaction are treated except
for the additional internal energy broadening due to the pri-
mary NHC dissociation. Whereas the extra assumptions
made are realistic, the model is less rigorous than that for
fitting of the primary reaction cross sections, and therefore
we find it more appropriate to list the barrier with only two
significant digits and without confidence interval.
Scheme 3. Formation and gas-phase reactivity of complexes 2a,b (NHC=
1,3-dimethyl-1H-imidazol-2-ylidene). Inset: calculated (red bars) and ex-
perimental isotope pattern (blue) of 2a.
the analogous palladium complex 2b (m/z 637), in which the
original triphenylphosphine ligand is not replaced by a third
NHC group.[7] The identity of these ions is corroborated by
their isotope patterns (see inset Scheme 3 and the Support-
ing Information) and gas-phase reactivities. On collisional
Chem. Eur. J. 2010, 16, 5408 – 5415
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5409

P. Chen et al.
Table 1. L-CID fitting results [kcalmolꢀ1] for reaction of 2a to give 3 and
4/4-p.[a]
z
Process
TS model
d[b]
DoF[c]
Rotors
6
E0
tight
loose
2
1
165
30.9(14)
47.1(17)
A
E
[1]
2
G
[6]
4
ACHTUNGTRENNUNG
!3+(NHC)Pd0
loose[a]
4-p!3+(NHC)Pd0
loose
1
120
4
38.9(10)
[a] The primary NHC dissociation (in brackets) was treated with fixed
parameters from the two-channel fit up to 4.5 eV, and the remaining
excess internal energy was corrected for this ligand loss; see main text.
[b] Reaction-path degeneracy. [c] Degrees of freedom.
state model to use (i.e., tight when an intramolecular rear-
rangement is rate-limiting, or loose when product dissocia-
tion is), and to assess the performance of various density
functionals for the system studied. We will discuss the re-
sults for the following density functionals: B3PW91; the
one-parameter mPW1K[11] designed for improved kinetics[12]
and used with some success in transition-metal chemistry;[13]
and M06-L,[14] which was designed to accurately describe
main-group and transition-metal thermochemistry, kinetics,
and nonbonding interactions at a reasonable computational
cost. The first two density functionals were employed for op-
timizations and energy calculations in Gaussian,[15] whereas
energies for the last-named functional were evaluated in
ADF.[16]
The M06-L//BP86/TZP potential-energy diagram for the
reactivity of 2a is shown in Figure 2 and the calculated rela-
tive energies are listed in Table 2. We will first discuss the
formation of trans-4 and the sequential reductive elimina-
tion of 3 (upper profile in Figure 2), because three experi-
mental values are available, namely, the binding energies of
an NHC ligand in 2a and of 3 in 4-p, as well as the sequen-
tial reductive elimination barrier. Thereafter, we will focus
on direct reductive elimination from 2a.
Figure 1. Cross sections and L-CID fits for A) reductive elimination of 3
from 2a (green diamonds) and competitive ligand dissociation to give 4/
4-p (red circles), fitted up to 4.5 eV center-of-mass frame collision energy
(dotted vertical line). B) Sequential reductive elimination of 3 from 4,
derived as the difference between fitted and experimental cross sections
for formation of 4/4-p. C) Dissociation of 3 from 4-p.
Ligand dissociation from 2a and sequential reductive elimi-
nation: A loose transition-state model should be appropriate
for the fragmentation of 2a into trans-4 and an NHC
ligand.[17] The experimental dissociation energy of
47.1(17) kcalmolꢀ1 is well reproduced by both M06-L and
mPW1K (45.8 and 44.4 kcalmolꢀ1, respectively, Table 2),
whereas B3PW91 underestimates it at 39.7 kcalmolꢀ1. Ac-
cording to our calculations, the resulting trans-4 isomerizes
The sequential formation of 3 was also studied separately
by generating 4/4-p in the spray at 60.0 V tube-lens voltage,
presumably affording the more stable 4-p on thermalization
(see below), which was then subjected to CID threshold
measurements (Figure 1C). Argon was used as the collision
gas to allow working at higher laboratory-frame energies, so
that the CID threshold region would not overlap with the
region in which the ion beam is truncated.
ꢀ
to cis-4 along the way to reductive C C coupling to give
complex 4-p, in which the product 3 is still ipso,ortho p-co-
ordinated to the (NHC)Pd0 fragment (see Figure 2).[18] All
three methods give very similar results for the transition
states and intermediates involved in the reductive elimina-
ꢀ
tion reaction. Specifically, the barrier for reductive C C
coupling (4-TS, 14.6 kcalmolꢀ1 at M06-L) is rather low, so
that when NHC ligand dissociation is induced by a harsh
tube lens setting, most likely the initially generated trans-4
is sufficiently activated to rearrange to the more stable 4-p.
DFT calculations: Quantum-chemical calculations were per-
formed to confirm the nature of the observed species and
their reactivities, in particular with respect to the transition-
5410
www.chemeurj.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 5408 – 5415

Reductive Elimination from a PdII N-Heterocyclic Carbene Complex
FULL PAPER
agree within 5 kcalmolꢀ1 for
the two direct dissociation reac-
tions, but these underestimate
the sequential reaction barrier
by 10–12 kcalmolꢀ1.
Direct reductive elimination
ꢀ
from 2a: Reductive C C cou-
pling in four-coordinate palladi-
um complex 2a (Figure 2,
bottom) initially affords the ex-
pected ipso,ortho p complex 2-
p.[18] At this point, it is instruc-
tive to compare the Pd···phenyl
bonding in 4-p and 2-p, as ana-
lyzed with ADF,[16] in terms of
their
fragments
3
and
(NHC)nPd0 (n=1, 2; Table 3).
The Hirshfeld charges[19] on the
formally neutral (NHC)nPd0
fragments of 4-p (0.243) and 2-
p (0.434) indicate strong back-
donation to the phenylimidazo-
lium fragment, especially in 2-
p, reflecting increasing elec-
tron-richness of these d10 Pd
species with the number of
strong s-donor NHC ligands.
This trend is paralleled by the
Figure 2. M06-L//BP86/TZP potential-energy diagram, including selected optimized geometries with C···C and
Pd···H distances in angstrom; other hydrogen atoms are omitted for clarity. Dashed arrows denote dissociation
reactions without reverse activation barriers.
Table 2. Calculated relative energies [kcalmolꢀ1] for reaction of trans-4 to give 3 (top) and for reaction of 2a
to trans-4 and 3 (bottom).
Method
trans-4!
4-TSx
cis-4
4-TS
4-p[a]
3+(NHC)Pd
4-p ! 3 + (NHC)Pd[a]
M06-L[b]
0.0
0.0
0.0
11.2
12.6
11.0
8.7
8.3
8.2
14.6
16.8
15.9
ꢀ6.1 (ꢀ5.7)
ꢀ7.7 (ꢀ8.5)
ꢀ6.3 (ꢀ7.5)
32.8
22.3
23.2
38.9 (38.5)
30.0 (30.8)
29.6 (30.6)
mPW1K[c]
B3PW91[c]
ortho
extent of Cipso
bond elon-
ꢀ
C
2a!
0.0
0.0
0.0
trans-4+NHC
2-TS
2-p[a]
2-TS2
2-s
3+(NHC)2Pd
gation (1.440 ꢂ in 4-p and
1.468 ꢂ in 2-p versus 1.408 ꢂ in
3) and the interaction energies
M06-L[b]
45.8
44.4
39.7
23.6
30.2
27.3
12.1 (14.2)
12.1 (12.8)
10.9 (10.7)
13.3
21.4
17.6
7.3
9.6
6.2
28.7
22.8
18.3
mPW1K[c]
B3PW91[c]
(4-p:
DEint =ꢀ38.5,
2-p:
[a] Values for ortho,meta p-coordinated complexes 4-pb and 2-pb are given in parentheses. [b] ADF M06-L
energy evaluations of BP86/TZP geometries, including Gaussian mPW1K/SDD(d) zero-point energy correc-
tions. [c] Gaussian SDB-cc-pVTZ single-point energies, including zero-point energy corrections of SDD(d) ge-
ometries.
ꢀ55.6 kcalmolꢀ1 at BP86/TZP).
However, the net p-coordina-
tion strength in 4-p is much
larger than in 2-p (DENet
=
On collisional activation, 4-p can again interchange with
trans/cis-4, but this does not affect the measured barrier ac-
cording to the Curtin–Hammett principle. The dissociation
of 3 from 4-p, which was thus experimentally determined to
require 38.9(10) kcalmolꢀ1 by using a loose TS model, is ex-
actly matched by M06-L, whereas both mPW1K and
B3PW91 underestimate it by about 9 kcalmolꢀ1. Finally, the
measured sequential reaction barrier of 37 kcalmolꢀ1 corre-
sponds reasonably well with the M06-L energy of [3+
(NHC)Pd] relative to trans-4 (32.8 kcalmolꢀ1), but is again
underestimated by mPW1K and B3PW91 by about
14 kcalmolꢀ1.
Table 3. BP86/TZP (NHC)nPd–3 bond analyses [kcalmolꢀ1] and fragment
Hirshfeld charges for 4-p, 2-p, and 2-s.
4-p
2-p
2-s
DEPauli
DVelstat
DEoi
DEint
DEprep(3)
118.8
ꢀ92.5
ꢀ64.9
ꢀ38.5
4.9
171.7
21.7
ꢀ127.6
ꢀ99.7
ꢀ55.6
13.7
ꢀ21.7
ꢀ16.6
ꢀ16.6
0.9
[a]
DE
DENet
(LnPd)
2.1
32.5
1.4
[b]
ꢀ31.5
ꢀ9.3
ꢀ14.6
q
N
0.243
0.434
0.118
[a] DEint =DEPauli +DVelstat +DEoi.
SEfragments
[b] DENet =DEint +SDEprep =Ecomplex
ꢀ
.
Thus, the M06-L density functional gives best agreement
with the experimental barriers, mainly because of its superi-
or performance for dissociation energies,[6e,f] which can be
attributed to the improved treatment of medium-range non-
covalent interactions.[14] We note that of all 69 density func-
tionals screened with ADF,[16] only PW91 and M06 also
ꢀ31.5 vs. ꢀ9.3 kcalmolꢀ1), mainly because of the large prep-
aration energy that is required to deform the preferably
linear (NHC)2Pd fragment to its bent geometry in 2-p (C-
Pd-C=103.948). This strain is released when the two frag-
Chem. Eur. J. 2010, 16, 5408 – 5415
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5411

P. Chen et al.
ments partially separate to give intermediate 2-s (C-P-C
171.308), which features end-on s coordination of the Pd
atom with both an acidic imidazolium methyl H and a
phenyl ortho-H atom of 3. The near-linearity of the Pd-
ACHTUNGTRENNUNG(NHC)2 fragment in 2-s and the recovered phenyl aromatic-
ity result in a very small preparation energy, compensating
for the much weakened interaction energy and making 2-s
more stable than 2-p.
For the experimental barrier of 30.9 kcalmolꢀ1 for reduc-
tive elimination of 3 from 2a a tight TS model was initially
ꢀ
assumed, that is, that the C C coupling barrier (2-TS)
would be rate-determining. This is supported by both
mPW1K and B3PW91, which also reproduce the experimen-
tal barrier well (Table 2; 30.2 and 27.3 kcalmolꢀ1, respective-
ly). Then again, as was the case for 4-p, these density func-
tionals presumably underestimate the energy for dissocia-
tion into 3+(NHC)2Pd, so that in reality it may be very sim-
ꢀ
ꢀ
ilar to the C C coupling barrier. M06-L gives a C C cou-
pling barrier that is even 5.1 kcalmolꢀ1 below the
dissociation energy of 28.7 kcalmolꢀ1, but the latter is much
lower than that obtained from fitting the experimental data
with a loose TS model (44.3(16) kcalmolꢀ1, see the Support-
ing Information). At first sight, M06-L thus surprisingly
seems to fail, but the situation may be more complicated.
The threshold curves were fitted by assuming unimolecular
first-order kinetics, which is a valid approximation for a mul-
tistep reaction only if one step is much slower than all the
others. However, for the reductive elimination from 2a the
Figure 3. Experimentally derived (thick solid lines) and M06-L based rate
constants for reductive elimination (green) and NHC ligand dissociation
(red) from 2a. The individual RRKM rate constants for rearrangement
of 2a to 2-s (blue long dashes) and vice versa (cyan dots), as well as for
subsequent elimination of 3 from 2-s (black dashes) are also included.
and can thus be ignored to simplify the overall reaction ki-
netics. In Figure 3, the individual RRKM rate constants are
ꢀ
ꢀ
“tight” C C coupling to 2-p becomes rate-limiting at high
plotted against the collision energy for the tight C C cou-
pling of 2a (kCꢀC, blue long dashes), the net reverse process
from 2-s (kC···C, cyan dots), and the loose elimination of 3
collision energies, as the rate of the only slightly less favora-
ble “loose” elimination of 3 increases more rapidly with in-
creasing excess energy.[20,21] Moreover, in the intermediate
regime where both steps are slow on the experimental time-
scale, the overall reaction does not obey pseudo-first-order
from 2-s (kelim, black dashes). Indeed, kCꢀ rises less steeply
than kelim and is already the smaller of the two from a colli-
sion energy of 1.3 eV.
C
ꢀ
kinetics. Only few reactant ions will undergo both C C cou-
Next, to evaluate the kinetic shift we derived the overall
rate equation for the multistep reductive elimination from
2a. For each set of individual RRKM rate constants, the for-
mation of 3 within the detection time window (ca. 60 ms)
was calculated and converted back to an apparent first-
order rate constant kapp for comparison with the experimen-
tally derived rate constant (Figure 3, thin and thick green
curves, respectively). The apparent rate constant has its
onset just after that of the loose dissociation and is consider-
pling and subsequent elimination before reaching the detec-
tor, leading to a much larger kinetic shift than for a single-
step first-order reaction. Thus, a loose TS model is inappro-
priate in cases where rearrangement precedes fragmentation
with an activation barrier that is comparable to the dissocia-
tion energy.
Transition-state switching: To probe the role of such switch-
ing of the rate-limiting transition state[20] in the reductive
elimination from 2a, we extracted the rate constants for this
reaction and the competitive ligand dissociation from the L-
CID fits (Figure 3, thick green and red curve, respectively).
The RRKM[21] rate constants for the individual steps were
calculated in CRUNCH[22] by using the M06-L relative ener-
gies with mPW1K zero-point energy corrections, frequen-
cies, and rotational constants (see Supporting Information
for details). As a control, we also calculated the RRKM rate
constant kdiss for NHC ligand dissociation from 2a and
found fair agreement with the corresponding first-order rate
constant derived from experiment (thin and thick red curve,
respectively). For the reductive elimination from 2a, the less
stable intermediate 2-p will rapidly interconvert with 2-s
ably smaller than both individual rate constants kCꢀ and
C
kelim up to 1.85 eV; at higher excess energy kapp follows the
ꢀ
profile of the tight C C coupling step. As a result, the pro-
file of kapp resembles that for a tight TS model but with a
barrier height slightly above that of the loose step, in rea-
sonable agreement with the rate constant extracted from the
L-CID fits (thick green curve).[23] In other words, in this
case it is appropriate to fit the CID data with a tight TS
model, but this gives an upper bound to the dissociation
energy rather than the barrier for the preceding rearrange-
ment. Hence, the M06-L results for the direct reductive
elimination of 3 from 2a are not at variance with the experi-
mentally determined threshold. More generally, this RRKM
modeling illustrates how CID thresholds for multistep reac-
5412
www.chemeurj.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 5408 – 5415

Reductive Elimination from a PdII N-Heterocyclic Carbene Complex
FULL PAPER
the first mass-selection quadrupole.[29] The ion guide is operated without
an external longitudinal field; ions move through it because of a weak
longitudinal potential induced by the space charge from the continuous
beam of incoming ions. Thus, a well-defined near-Gaussian distribution
of the ionsꢄ kinetic and, presumably, internal energies is achieved
(Figure 4), which is required for subsequent deconvolution of the cross-
section curves.
tions are affected when the barrier for rearrangement ap-
proaches the final product dissociation energy.
Conclusion
A phenylpalladium(II) N-heterocyclic carbene complex was
studied by CID threshold measurements to provide quanti-
tative gas-phase thermochemical data for reductive elimina-
tion and ligand dissociation. The analogous three-coordinate
ꢀ
complex readily undergoes reductive C C coupling to give a
p complex, for which the product binding energy was also
determined. Our results complement recent solution-state
evidence[4] that NHCs are not always mere spectator ligands,
but can be reductively eliminated as imidazolium salts.
Quantum chemical calculations show that the barrier for re-
ꢀ
ductive C C coupling is similar to the energy required for
dissociation of the produced 1,3-dimethyl-2-phenyl-1H-imi-
dazolium, which supposedly leads to transition-state switch-
ing. An RRKM treatment showed that in this case one
should fit the energy-resolved CID data with a “tight” TS
model, but then actually obtains an upper bound to the dis-
sociation energy. The M06-L density functional reproduces
the experimental energetics well, which is an important vali-
dation for our ongoing investigations of the cross-coupling
mechanism.
Figure 4. Distribution of ion kinetic energies in the laboratory frame, de-
termined for ion 2a by retarding potential measurements after the gas-
filled 24-pole ion guide; the Gaussian fit (red line) has a full width at
half-maximum of 1.71 eV.
For the CID threshold measurements, the parent ion was mass-selected
in the first quadrupole and allowed to react with xenon or argon (30–
110 mTorr) in the octapole collision cell. Intensities of the reactant and
product ions were recorded as a function of the collision offset voltage.
An ICL[30] script was developed to sample all species in an interleaved
fashion rather than individually, which suppresses discrepancies between
the ion intensity curves due to variations in experimental conditions. The
script also monitors the collision gas pressure as measured by a hot-cath-
ode gauge, allowing for improved zero-pressure extrapolations. The ion
intensities were converted to cross sections according to Ervin et al.,[31]
extrapolated to zero collision-gas pressure, and fitted with L-CID.[8] For
cases where the nature of the rate-limiting reaction step was uncertain,
both tight and loose transition states (i.e., with or without reverse activa-
tion barrier) were considered. Methyl groups were taken as free rotors
Experimental Section
General procedures: Solvents were distilled from the appropriate drying
agents.[24] Starting materials were synthesized and handled under inert at-
mosphere by standard Schlenk techniques or in a glove box. Their struc-
ture and purity was confirmed by 1H and, where applicable, 31P NMR
spectroscopy on Varian Mercury 300 MHz instruments. Iodide-bridged
palladium dimer 1 was prepared in two steps in 84% overall yield from
bis(triphenylphosphine)palladium(II) dichloride following the procedures
of Grushin and Alper.[25] 1,3-Dimethyl-1H-imidazol-2-ylidene silver
iodide [(NHC)AgI] was prepared from 1-methyl-1H-imidazole according
to literature procedures.[26, 27]
ꢀ
and a reaction-path degeneracy of two was used for the C C coupling
steps.[32] Further details of the experimental setup, data acquisition and
processing, and L-CID deconvolution are contained in the Supporting In-
formation.
Tris(1,3-dimethyl-1H-imidazol-2-ylidene)phenylpalladium(II) iodide (2a):
Fresh stock solutions of 2a in dry acetonitrile were prepared in a glove
box by dissolving 1 (1.7 mg, 1.5 mmol) and [(NHC)AgI] (3 mg, 4.5 mmol)
in acetonitrile (0.50 mL). The produced dark brown precipitate was re-
moved by filtration through cotton wool and washed with acetonitrile
(2ꢃ0.50 mL). For the ESI-MS studies, the clear colorless filtrate was di-
luted to approximately 50 mm. Stock and spray solutions were stored at
ꢀ358C in the glove box. For the qualitative MS measurements on a
Thermo Finnigan TSQ Quantum tandem mass spectrometer, optimal ESI
settings were a spray rate of 5 mLminꢀ1 at 5 kV spray voltage, heating the
capillary at 1708C at 35 V capillary voltage, and 25 V tube lens voltage;
CID experiments were conducted with 0.5 mTorr argon.
Computational methods: Density functional theory (DFT) calculations
were performed with the Gaussian suite[15] employing the B3PW91 and
mPW1K[11] density functionals. Geometry optimizations were performed
with the Stuttgart/Dresden effective-core potential and associated basis
set[33] for palladium and the Cartesian Dunning 95 full double-zeta basis
set[34] with additional d polarization functions for main-group elements,
which we abbreviate as SDD(d). The Geometry DIIS algorithm[35] was
applied in combination with an ultrafine integration grid and tight SCF
and geometry convergence criteria. The nature of each stationary point
was confirmed by a frequency analysis; for the transition states, the vibra-
tional mode of the imaginary frequency corresponded to the expected re-
action coordinate. Subsequent single-point energies were calculated with
the SDB-cc-pVTZ basis set,[36] to which the SDD(d) zero-point energy
corrections were added.[37]
CID threshold measurements: Energy-resolved CID threshold measure-
ments were performed on a customized Finnigan MAT TSQ-700 tandem
mass spectrometer, whose original transfer octapole was replaced by a
long radio-frequency 24-pole ion guide to thermalize the ions with
5 mTorr argon to the manifold temperature of 708C. The previously re-
ported design[28] was improved to extend the ion transmission range
below m/z 200. Specifically, rods instead of plates were used for the poles
to lower the capacitance and hence allow for higher radio frequencies, in
combination with a movable thermalization chamber that was positioned
somewhat closer to the skimmer to improve the background vacuum in
Additionally, density functionals were screened with the Amsterdam
Density Functional (ADF) suite.[16] Geometries were reoptimized with
the BP86 density functional using an all-electron Slater-type basis set of
triple-zeta quality with added polarization functions (TZP), where rela-
tivistic effects were treated with the scalar zeroth-order regular approxi-
mation (ZORA). The integration accuracy was set to 6.0, and geometry
convergence criteria were used of 2ꢃ10ꢀ5 on the energy and 3ꢃ10ꢀ4 on
Chem. Eur. J. 2010, 16, 5408 – 5415
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5413

P. Chen et al.
the gradients. Subsequent single-point calculations with the keywords
MetaGGA and HartreeFock[16] afforded energy evaluations for 69 densi-
ty functionals, including Truhlarꢄs M06 suite,[14] to which the Gaussian
mPW1K/SDD(d) zero-point energy corrections were added.[37,38]
Int. Ed. 2010, 49, 2873–2877; h) A. Fedorov, L. Baptiste, E. A. P.
Couzijn, P. Chen, ChemPhysChem, 2010, 11, 1002–1005.
[7] M06-L//BP86/TZP calculations favor placement of the phosphine
ligand trans to phenyl group by 4.5 kcalmolꢀ1 over the cis arrange-
ment.
Bond analyses in terms of the separate fragments were performed with
ADF[16] by using the BP86/TZP results. According to the extended transi-
tion-state model,[39] the net bond energy DENet can be decomposed into
four contributions: the preparation energy DEprep needed to deform each
fragment from its equilibrium geometry to that in the complex; the steric
interactions between the fragments due to Pauli repulsion (DEPauli) and
electrostatic attraction (DVelstat); and the orbital interaction energy DEoi;
the last three are usually summed to give the total interaction energy
DEint. Furthermore, the net transfer of electron density between the frag-
ments is indicated by their Hirshfeld charges.[19]
[8] S. Narancic, A. Bach, P. Chen, J. Phys. Chem. A 2007, 111, 7006–
7013.
[9] We note that the decline of the cross section for ligand dissociation
at high collision energies cannot be an effect of increased competi-
tion by the primary reductive elimination reaction, because the rate
of the former, loose reaction should increase more rapidly with
excess energy than that of the latter, causing the opposite effect.
[10] Armentrout et al. recently reported examples of statistical modeling
of sequential CID thresholds with CRUNCH: a) P. B. Armentrout,
J. Chem. Phys. 2007, 126, 234302; b) A. L. Heaton, R. M. Moision,
P. B. Armentrout, J. Phys. Chem. A 2008, 112, 3319–3327; c) A. L.
Heaton, P. B. Armentrout, J. Am. Chem. Soc. 2008, 130, 10227–
10232.
[11] B. J. Lynch, P. L. Fast, M. Harris, D. G. Truhlar, J. Phys. Chem. A
2000, 104, 4811–4815.
Acknowledgements
[12] a) S. Parthiban, G. de Oliveira, J. M. L. Martin, J. Phys. Chem. A
2001, 105, 895–904; b) B. J. Lynch, D. G. Truhlar, J. Phys. Chem. A
2001, 105, 2936–2941.
This work has been supported by the ETH Zꢀrich. We thank Renꢅ
Dreier for constructing the 24-pole ion guide and Esther Quintanilla for
assistance with its integration into the TSQ-700 mass spectrometer.
[13] See, for example: a) M. A. Iron, H. C. Lo, J. M. L. Martin, E.
Keinan, J. Am. Chem. Soc. 2002, 124, 7041–7054; b) M. A. Iron,
J. M. L. Martin, M. E. van der Boom, J. Am. Chem. Soc. 2003, 125,
11702–11709, but also: c) M. M. Quintal, A. Karton, M. A. Iron,
A. D. Boese, J. M. L. Martin, J. Phys. Chem. A 2006, 110, 709–716.
[14] a) Y. Zhao, D. G. Truhlar, J. Chem. Phys. 2006, 125, 194101; b) Y.
Zhao, D. G. Truhlar, Theor. Chem. Acc. 2008, 120, 215–241; c) Y.
Zhao, D. G. Truhlar, Acc. Chem. Res. 2008, 41, 157–167.
[15] Gaussian 03, Revision E.01, M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgom-
ery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S.
Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani,
N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K.
Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda,
O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian,
J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W.
Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J.
Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari,
J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cio-
slowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaro-
mi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng,
A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W.
Chen, M. W. Wong, C. Gonzalez, J. A. Pople, Gaussian, Inc., Wall-
ingford, CT, 2004.
[16] a) ADF2008.01, SCM, Theoretical Chemistry, Vrije Universiteit,
Amsterdam, http://www.scm.com; b) G. te Velde, F. M. Bickelhaupt,
S. J. A. van Gisbergen, C. Fonseca Guerra, E. J. Baerends, J. G.
Snijders, T. Ziegler, J. Comput. Chem. 2001, 22, 931–967; c) C. Fon-
seca Guerra, J. G. Snijders, G. te Velde, E. J. Baerends, Theor. Chem.
Acc. 1998, 99, 391–403.
[17] P. B. Armentrout, J. Simons, J. Am. Chem. Soc. 1992, 114, 8627–
8633.
[18] As compared to 2-p and 4-p, the ortho,meta p-coordinated species
2-pb and 4-pb are not significantly more stable (values in parenthe-
ses in Table 2); such isomerizations along the reaction pathways can
[1] Selected books and reviews: a) Metal-catalyzed Cross-coupling Re-
actions, 2nd ed. (Eds.: A. de Meijere, F. Diederich), Wiley-VCH,
Weinheim, 2004; b) S. R. Chemler, D. Trauner, S. J. Danishefsky,
Angew. Chem. 2001, 113, 4676–4701; Angew. Chem. Int. Ed. 2001,
40, 4544–4568; c) N. Miyaura, A. Suzuki, Chem. Rev. 1995, 95,
2457–2483; d) I. P. Beletskaya, A. V. Cheprakov, Chem. Rev. 2000,
100, 3009–3066; e) J.-P. Corbet, G. Mignani, Chem. Rev. 2006, 106,
2651–2710; f) K. C. Nicolaou, P. G. Bulger, D. Sarlah, Angew.
Chem. 2005, 117, 4516–4563; Angew. Chem. Int. Ed. 2005, 44, 4442–
4489; g) H.-U. Blaser, A. Indolese, F. Naud, U. Nettekoven, A.
Schnyder, Adv. Synth. Catal. 2004, 346, 1583–1598; h) A. O. King,
N. Yasuda, Organometallics in Process Chemistry, Springer, Berlin,
2004, pp. 205–245.
[2] Most knowledge of reductive elimination is based on kinetic studies
of the less reactive platinum complexes, see, for example: a) J. F.
Hartwig, Inorg. Chem. 2007, 46, 1936–1947, and references therein;
b) P. S. Braterman, R. J. Cross, G. B. Young, J. Chem. Soc. Dalton
1977, 1892–1897; c) R. K. Merwin, R. C. Schnabel, J. D. Koola,
D. M. Roddick, Organometallics 1992, 11, 2972–2978.
[3] K. Osakada, H. Onodera, Y. Nishihara, Organometallics 2005, 24,
190–192.
[4] a) D. S. McGuinness, N. Saendig, B. F. Yates, K. J. Cavell, J. Am.
Chem. Soc. 2001, 123, 4029–4040; b) N. D. Clement, K. J. Cavell,
Angew. Chem. 2004, 116, 3933–3935; Angew. Chem. Int. Ed. 2004,
43, 3845–3847; c) D. C. Graham, K. J. Cavell, B. F. Yates, Dalton
Trans. 2007, 4650–4658; d) A. T. Normand, A. Stasch, L.-l. Ooi,
K. J. Cavell, Organometallics 2008, 27, 6507–6520; e) K. J. Cavell,
Dalton Trans. 2008, 6676–6685; f) A. T. Normand, M. S. Nechaev,
K. J. Cavell, Chem. Eur. J. 2009, 15, 7063–7073.
[5] a) W. A. Herrmann, Angew. Chem. 2002, 114, 1342–1363; Angew.
Chem. Int. Ed. 2002, 41, 1290–1309; b) W. A. Herrmann, K. ꢆfele,
D. von Preysing, S. K. Schneider, J. Organomet. Chem. 2003, 687,
229–248; c) E. A. B. Kantchev, C. J. OꢄBrien, M. G. Organ, Angew.
Chem. 2007, 119, 2824–2870; Angew. Chem. Int. Ed. 2007, 46, 2768–
2813.
ꢀ
thus be discounted. C C coupling in 4 could alternatively afford ini-
[6] For related work from our group, see: a) P. Chen, Angew. Chem.
2003, 115, 2938–2954; Angew. Chem. Int. Ed. 2003, 42, 2832–2847;
b) E. Zocher, R. Dietiker, P. Chen, J. Am. Chem. Soc. 2007, 129,
2476–2481; c) M.-E. Moret, P. Chen, Organometallics 2007, 26,
1523–1530; d) E. Zocher, R. Sigrist, P. Chen, Inorg. Chem. 2007, 46,
11366–11370; e) S. Torker, D. Merki, P. Chen, J. Am. Chem. Soc.
2008, 130, 4808–4814; f) A. Fedorov, M.-E. Moret, P. Chen, J. Am.
Chem. Soc. 2008, 130, 8880–8881; g) M.-E. Moret, D. Serra, A.
Bach, P. Chen, Angew. Chem. 2010, 122, 2935–2939; Angew. Chem.
tial product 4-pN, in which 3 is p-coordinated by the (NHC)Pd frag-
ment at C2 and one nitrogen atom of the imidazolium moiety, but
this species is calculated to be 6.4 kcalmolꢀ1 less stable than 4-p at
the M06-L//BP86/TZP level. Computational searches for an analo-
gous isomer 2-pN afforded 2-s instead.
[19] F. L. Hirshfeld, Theor. Chim. Acta 1977, 44, 129–138.
[20] a) W. J. Chesnavich, L. Bass, T. Su, M. T. Bowers, J. Chem. Phys.
1981, 74, 2228–2246; b) M. T. Bowers, M. F. Jarrold, W. Wagner-Re-
5414
www.chemeurj.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 5408 – 5415

Reductive Elimination from a PdII N-Heterocyclic Carbene Complex
FULL PAPER
deker, P. R. Kemper, L. M. Bass, Faraday Discuss. Chem. Soc. 1983,
75, 57–76; c) M. F. Jarrold, W. Wagner-Redeker, A. J. Illies, N. J.
Kirchner, M. T. Bowers, Int. J. Mass Spectrom. Ion Processes 1984,
58, 63–95.
[31] K. M. Ervin, P. B. Armentrout, J. Chem. Phys. 1985, 83, 166–189.
[32] This reflects the fact that the phenyl moiety can be coupled to
either of the two neighboring NHC ligands (see Scheme 3).
[33] M. Dolg in Modern Methods and Algorithms of Quantum Chemistry,
Vol. 1 (Ed.: J. Grotendorst), John von Neumann Institute for Com-
puting, Jꢀlich, 2000, pp. 479–508.
[21] W. Forst, Unimolecular Reactions - A Concise Introduction, Cam-
bridge University Press, Cambridge, 2003.
[22] a) CRUNCH, Version 04.1600D, P. B. Armentrout, K. M. Ervin,
2002; b) M. T. Rodgers, K. M. Ervin, P. B. Armentrout, J. Chem.
Phys. 1997, 106, 4499; c) V. F. DeTuri, K. M. Ervin, J. Phys. Chem. A
1999, 103, 6911.
[23] We note that the exact profile of kapp is quite sensitive to both the
relative and absolute values of the individual rate constants, and
hence to the exactness of the calculated energy barriers as well as of
the RRKM treatment.
[34] T. H. Dunning, Jr., P. J. Hay in Modern Theoretical Chemistry, Vol. 3
(Ed.: H. F. Schaefer III), Plenum, New York, 1976, pp. 1–28.
[35] ꢆ. Farkas, H. B. Schlegel, J. Chem. Phys. 1999, 111, 10806–10814.
[36] J. M. L. Martin, A. Sundermann, J. Chem. Phys. 2001, 114, 3408–
3420.
[37] The counterpoise method afforded negligible basis-set superposition
errors of ꢁ1.2 and ꢁ1.8 kcalmolꢀ1 for the SDB-cc-pVTZ and TZP
basis sets, respectively.
[24] W. L. F. Armarego, D. D. Perrin, Purification of Laboratory Chemi-
cals, 4th ed., Butterworth-Heinemann, Oxford, 1997.
[25] V. V. Grushin, H. Alper, Organometallics 1993, 12, 1890–1901.
[26] B. L. Benac, E. M. Burgess, A. J. Arduengo III, Org. Synth. 1990, 7,
195.
[27] D. M. Khramov, V. M. Lynch, C. W. Bielawski, Organometallics
2007, 26, 6042–6049.
[28] a) L. A. Hammad, G. Gerdes, P. Chen, Organometallics 2005, 24,
1907–1913; b) C. R. Hinderling, Ph.D. thesis, ETH Zꢀrich (Switzer-
land), 1999, Chapter 4.2.
[38] ADF analytical frequency calculations afforded spurious imaginary
frequencies for several species, although their geometries were fully
optimized. Therefore, we used the zero-point energy corrections
from Gaussian 03.
[39] a) K. Morokuma, Acc. Chem. Res. 1977, 10, 294–300; b) T. Ziegler,
A. Rauk, Inorg. Chem. 1979, 18, 1755–1759; c) T. Ziegler, A. Rauk,
Theor. Chim. Acta 1977, 46, 1–10; d) F. M. Bickelhaupt, E. J. Baer-
ends in Reviews in Computational Chemistry, Vol. 15 (Eds.: K. B.
Lipkowitz, D. B. Boyd), Wiley, New York, 2000, pp. 1–86.
[29] E. M. Zocher, Ph.D. Thesis, ETH Zꢀrich (Switzerland), 2008, Chap-
ter 4.2.
[30] F. P. Mason, Instrument Control Language Manual, Revision A, Fin-
nigan MAT, San Jose, 1992.
Received: October 22, 2009
Revised: February 4, 2010
Published online: April 1, 2010
Chem. Eur. J. 2010, 16, 5408 – 5415
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5415