Gas-phase thermochemistry of ruthenium carbene metathesis catalysts

Article abstract of DOI:10.1021/ja078149z

Quantitative energy-resolved collision-induced dissociation cross-sections by tandem ESI-MS provide absolute thermochemical data for phosphine binding energies in first- and second-generation ruthenium metathesis catalysts of 33.4 and 36.9 kcal/mol, respectively. Furthermore a study of the ring-closing metathesis in the second-generation system to liberate norbornene by forming the 14-electron reactive intermediate from the intramolecular π-complex gives an estimate of the olefin binding energy to the 14-electron complex of around 18 kcal/mol, assuming a loose transition state. The results reported here are in remarkably good agreement with the latest DFT calculations using the M06-L functional.

Full text of DOI:10.1021/ja078149z

Published on Web 03/15/2008
Gas-Phase Thermochemistry of Ruthenium Carbene
Metathesis Catalysts
Sebastian Torker, Daniel Merki, and Peter Chen*
Laboratorium fu¨r Organische Chemie, ETH Zu¨rich, Wolfgang-Pauli-Strasse 10,
CH-8093 Zu¨rich, Switzerland
Received October 24, 2007; E-mail: peter.chen@org.chem.ethz.ch
Abstract: Quantitative energy-resolved collision-induced dissociation cross-sections by tandem ESI-MS
provide absolute thermochemical data for phosphine binding energies in first- and second-generation
ruthenium metathesis catalysts of 33.4 and 36.9 kcal/mol, respectively. Furthermore a study of the ring-
closing metathesis in the second-generation system to liberate norbornene by forming the 14-electron
reactive intermediate from the intramolecular π-complex gives an estimate of the olefin binding energy to
the 14-electron complex of around 18 kcal/mol, assuming a loose transition state. The results reported
here are in remarkably good agreement with the latest DFT calculations using the M06-L functional.
Introduction
energy-resolved collision-induced dissociation cross-sections of
electrosprayed ruthenium carbene complexes. The gas-phase
Quantitative thermochemistry of organometallic complexes
involved in homogeneous catalysts is central to clear mechanistic
thinking and catalyst design and, at the same time, uncommon
because many of the experimental approaches, for example,
reaction calorimetry, fail for solutions where the catalysts
themselves may be part-per-thousand to part-per-million com-
ponents in a complex reaction system with many other com
plexes and substrates simultaneously present in much higher
concentrations. We have in recent years prepared the active
species in catalytic cycles by a combination of chemical synthe
sis, electrospray ionization, and gas-phase ion-molecule reac-
tions in a tandem mass spectrometer.1 In many cases, the gas-
phase chemistry resembled the reactions in solution, providing
experimental data for mechanistic models and quantum chemical
studies. Among the reactions we have investigated with these
techniques is the olefin metathesis reaction,2 catalyzed by
ruthenium carbene complexes.3 While the mass spectrometric
experiments proved informative, specifically with respect to the
reactivity of the 14-electron reactive intermediate, the methods
and the analysis suffered from the inability of the experiment
to produce absolute activation energies and reaction thermo-
chemistry for the elementary steps observed in the mass
spectrometer.
energetics are consistent with the most recent quantum chemical
calculations as well as solution-phase results if one factors in
observed trends in solvation.
Experimental Section
General Remarks. Unless otherwise stated, all manipulations were
carried out under an argon atmosphere on a vacuum line using standard
Schlenk techniques. The solvents were dried by distillation from the
following drying agents prior to use and were transferred under N2:
diethyl ether (Na/K), CH2Cl2 (CaH2), acetonitrile (CaH2). 1,2-Dichlo-
robenzene was deoxygenated by N2 bubbles and dried with molecular
sieves (3 Å). Low-resolution ESI-MS measurements were done on a
Finnigan MAT TSQ Quantum mass spectrometer. NMR measurements
are reported for a Varian Mercury XL 300 (1H: 300 MHz, 31P: 121
MHz) spectrometer. Chemical shifts (δ-values) are reported in ppm
with respect to Me4Si (δ ) 0 ppm), used as an internal standard for 1H
NMR, and an 85% aqueous H3PO4 solution, used as an external standard
for 31P NMR. Coupling constants (J) are given in hertz. 31P NMR
spectra were proton broad-band-decoupled. The multiplicities of peaks
are denoted by the following abbreviations: s: singlet, d: doublet,
dd: double doublet, m: multiplet.
Syntheses. Tricyclohexyl(4-vinylbenzyl)phosphonium Chloride.4
Tricyclohexylphosphine (0.99 g, 3.53 mmol) was dissolved in aceto-
nitrile (4 mL) in a Teflon-valve sealed bomb, and 4-vinylbenzyl chloride
(0.54 g, 3.55 mmol) was added. The reaction mixture was stirred at 80
°C for 3 days. The solvent was removed in Vacuo, the pale yellow
sticky residue was dissolved in chloroform (2 mL), and diethyl ether
(15 mL) was added dropwise to force precipitation. In the beginning,
a sticky precipitate was formed at the bottom, and after a few minutes
of stirring, a fine white powder also precipitated. This powder was
separated by decantation followed by filtration and was dried in Vacuo
for 4 h: 0.92 g (67%). (The sticky residue was analyzed separately
and it turned out to be mostly the same as the white powder, but it was
not used any further.)
We report here the experimental measurement of the acti-
vation energies for phosphine dissociation and ring-closing
metathesis for cationized Grubbs first and second-generation
catalysts, 1 and 3, in the gas phase, by deconvolution of the
(1) Chen, P. Angew. Chem. 2003, 115, 2938.
(2) Handbook of Metathesis; Grubbs, R. H., Ed.; Wiley-VCH: Weinheim,
Germany, 2003.
(3) (a) Adlhart, C.; Chen, P. HelV. Chim. Acta 2000, 83, 2192. (b) Adlhart,
C.; Volland, M. A. O.; Hofmann, P.; Chen, P. HelV. Chim. Acta 2000, 83,
3306. (c) Adlhart, C.; Hinderling, C.; Baumann, H.; Chen, P. J. Am. Chem.
Soc. 2000, 122, 8204. (d) Volland, M. A. O.; Adlhart, C.; Kiener, C. A.;
Chen, P.; Hofmann, P. Chem. Eur. J. 2001, 7, 4621. (e) Adlhart, C.; Chen,
P. HelV. Chim. Acta 2003, 86, 941.
(4) Synthesized according to an analogous procedure: Colabufo, N. A.; Berardi,
F.; Perrone, R.; Rapposelli, S.; Digiacomo, M.; Balsamo, A. J. Med. Chem.
2006, 49, 6607.
9
4808
J. AM. CHEM. SOC. 2008, 130, 4808-4814
10.1021/ja078149z CCC: $40.75 © 2008 American Chemical Society

Ruthenium Carbene Metathesis Catalysts
A R T I C L E S
Scheme 1
Scheme 2
1H NMR (300 MHz, CDCl3): δ ) 7.38 (s, 4H, HAr), 6.67 (dd, J )
17.6 Hz, 11.0 Hz, 1H, olefinic), 5.75 (d, J ) 17.7 Hz, 1H, olefinic),
5.28 (d, J ) 10.8 Hz, 1H, olefinic), 4.36 (d, J ) 14.4 Hz, 2H), 2.74 (q,
T-CID Experiments. Dilute solutions (10-5 mol/L) of the charged
complexes in CH2Cl2 were freshly prepared in a glovebox. All
measurements were performed on a Finnigan MAT TSQ-700 tandem
mass spectrometer modified by replacement of the transfer octopole
in the original triple quad instrument with a long radiofrequency 24-
pole ion guide, which was run at gas pressures of 5 mTorr argon or
norbornene for thermalization to the 70 °C manifold temperature or
reaction. No external field is applied, and the ions move through the
ion guide because of a weak longitudinal potential induced by the space
charge from the continuous beam of incoming ions. The principle
significance of this modification is the achievement of a well-defined,
narrow distribution of the ions’ kinetic and internal energies. The ions
were selected in the first quadrupole and then injected into the octopole
collision cell where they undergo CID with Argon at low pressures
(30, 50, 70, and 90 µTorr) before mass analysis in the second
quadrupole. The instrument has been described previously.5c
J ) 6.0 Hz, 3H, HCy), 1.88 (m, 15H, HCy), 1.37 (m, 15H, HCy). 31
P
NMR (121 MHz, CDCl3): δ ) 29.8 (s, 1P). MS (ESI, CH2Cl2): 397
-
(100, M+), 830 (9, [2M+ + Cl-]+), 117 (6, [M+ PCy3]+).
[(PCy3)2Cl2RudCH(p-CH2PCy3)Ph]+Cl-, 1. Grubbs first genera-
tion catalyst (100 mg, 0.12 mmol) and tricyclohexyl(4-vinylbenzyl)-
phosphonium chloride (40 mg, 0.09 mmol) were dissolved in dichlo-
romethane (6 mL). This mixture was allowed to stir at room temperature
for 20 min. The solvent was removed in Vacuo, and the formed styrene
was coevaporated with 1,2-dichlorobenzene (7 × 2 mL) to shift the
equilibrium. Furthermore dichloromethane (2 × 2 mL) was added to
remove residual 1,2-dichlorobenzene, and the residue was dried in Vacuo
overnight. After dissolving in dichloromethane (1 mL), diethyl ether
(15 mL) was added dropwise to force precipitation. The mixture was
filtered, and the purple solid was dried in Vacuo: 68 mg (64%).
Results
1H NMR (300 MHz, CD2Cl2): δ ) 20.02 (s, 1H, carbene), 8.53 (s,
2H, HAr), 7.47 (s, 2H, HAr), 4.19 (d, J ) 13.8 Hz, 2 H), 2.69 (m, 9H,
HCy), 1.84 (m, 45H, HCy), 1.28 (m, 45H, HCy). 31P NMR (121 MHz,
CD2Cl2): δ ) 37.1 (s, 2P, phosphine), 27.9 (s, 1P, phosphonium). MS
Collision-induced dissociation (CID) of electrosprayed com-
plexes5 provides access to two of the important elementary
reactions in olefin metathesis: the initial dissociation of the
phosphine ligand to produce the key 14-electron intermediate
(Schemes 1 and 2), and ring-closing metathesis of a carbene
complex with a pendant olefinic moiety (Scheme 3). Daughter-
ion spectra produced by CID of mass-selected parent ions are
shown in Figures 1-4. The suitability of a given CID process
for the subsequent energy-resolved CID cross-section measure-
ments can be read off of the spectra in Figures 1-4 in a
straightforward manner. The present deconvolution procedure,
implemented in the L-CID program,6 can handle one or two
-
(ESI, CH2Cl2): 836 (100, [M+ PCy3]+), 1118 (55, M+), 383 (41,
[stilbene derivative]2+).
[(H2IMes)(PCy3)Cl2RudCH(p-CH2PCy3)Ph]+Cl-, 3 was prepared
from Grubbs second generation catalyst (51 mg, 0.06 mmol) and
tricyclohexyl(4-vinylbenzyl)phosphonium chloride (21 mg, 0.05 mmol)
according to the above procedure which yielded a dark purple, fine
powder after drying in Vacuo overnight: 39 mg (67%). The product
was not as clean as 1 due to the higher metathesis activity of second
generation systems and thus giving a considerable amount of stilbene
side products. The quality however was sufficient for MS experiments
where the ions were totally separated.
1H NMR (300 MHz, CD2Cl2, only downfield region): δ ) 19.22
(s, 1H, carbene). 31P NMR (121 MHz, CD2Cl2): δ ) 30.5 (s, 1P,
phosphine), 27.8 (s, 1P, phosphonium). MS (ESI, CH2Cl2): 862 (100,
(5) (a) Zhang, X.; Narancic, S.; Chen, P. Organometallics 2005, 24, 3040. (b)
Zocher, E. M.; Dietiker, R.; Chen, P. J. Am. Chem. Soc. 2007, 129, 2476.
(c) Hammad, L.; Gerdes, G.; Chen, P. Organometallics 2005, 24, 1907.
(d) Moret, M.-E.; Chen, P. Organometallics 2007, 26, 1523.
-
[M+ PCy3]+), 1144 (35, M+), 384 (34, [carbene]+).
(6) Narancic, S.; Bach, A.; Chen, P. J. Phys. Chem. A 2007, 111, 7006.
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4809

A R T I C L E S
Scheme 3
Torker et al.
parallel dissociation channels. Moreover, there should be no
major loss processes other than the CID process under
investigation. In an important control experiment with the
bis (dicyclohexylethyl) analog of 1, only loss of PCy2Et was
seen upon CID, indicating that the PCy3 lost from 1 was one
of the ligands on ruthenium and not a loss of the charge label
from the carbene moiety. The CID curves at progressively lower
collision gas pressure are shown in the Supporting Information.
Extrapolation to zero pressure and deconvolution with L-CID
produces the energy-dependent cross-sections depicted in
Figures 5 and 6. As had been discussed in our previous work,6
the deconvolution itself cannot distinguish between a tight and
loose transition state. In the cases examined thus far, either
assumption can yield a statistically acceptable fit. Nevertheless,
as will be seen below, chemical arguments usually do lead to
an unambiguous transition state model. It should also be noted
that L-CID requires as input the number of free rotors in the
molecule. While the number of free rotors is relatively easy to
ascertain for the N-heterocyclic carbene (NHC) ligand, the same
cannot be said for tricyclohexylphosphine in a metathesis
catalyst because of strongly hindered rotations.7 Accordingly,
we performed a sensitivity analysis in which the number of
rotors was run from zero up to the number that would include
all groups that could formally rotate around a single bond. We
observed that there is a modest difference in the deconvoluted
E0 as the number of rotors increases from zero to moderate
values, such as 6 for ions as large as the metathesis catalysts,
but even that small effect of additional rotors diminishes rapidly
as the number of rotors is further increased. Almost no
dependence on the number of rotors is observed with a tight
transition state assumption. The end effect is a minor increase
Figure 1. Daughter spectra of 1 (m/z ) 1116) at a collision offset of -25
eV at different pressures of xenon (30, 50, 70, and 90 µTorr) in the octopole
collision chamber. Comparable spectra are obtained with argon at collision
energies of about -100 eV.
Figure 2. Daughter spectra of 3 (m/z ) 1142) at a collision offset of -25
eV at different pressures of xenon (30, 50, 70, and 90 µTorr) in the octopole
collision chamber. Comparable spectra are obtained with argon at collision
energies of about -100 eV.
9
4810 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

Ruthenium Carbene Metathesis Catalysts
A R T I C L E S
Discussion
Olefin metathesis presented a broad range of mechanistic
problems, of which many have been addressed using a variety
of experimental and computational methods.2 It is generally
agreed now that the initial ligand exchange of the tricyclohexyl-
phosphine for the olefin substrate, the initiation step in catalysis,
occurs by way of a dissociative mechanism via a 14-electron
intermediate ruthenium carbene complex. The best current
quantum chemical calculations support this mechanism.10 The
strongest experimental evidence comes from the series of kinetic
and NMR experiments by Grubbs and co-workers in which the
enthalpies, and more importantly, the entropies of activation
were measured.11 A key recognition from that work was, that
despite being strong σ-donors, the NHC-ligands in the second-
generation catalysts do not labilize a tricyclohexylphosphine in
the trans position when the corresponding first-generation system
is taken as a reference. Effectively, the conclusion is that the
first- and second-generation catalysts operate by similar mech-
anisms but with different rate-determining steps.12 For a simple,
near-thermoneutral cross-metathesis, the ligand exchange, olefin
for phosphine, is rate-limiting in the second-generation catalysts,
whereas the rate-determining step in the first-generation catalysts
occurs later within the sequence of reactions defining the
metathesis itself. The recognition matters a great deal in attempts
to design stereo-, regio-, or chemoselectivity into a catalytic
metathesis reaction. Our chemoselective catalyst for alternating
copolymerization of two different cyclic olefins was built on a
first-generation chassis for precisely this reason.13 Under these
circumstances, the amount of quantitative data on the elementary
steps in olefin metathesis is surprisingly small, the paucity being
attributable to technical difficulties in doing detailed kinetic and
thermochemical experiments on low-abundance intermediates
in a catalytic cycle. The activation parameters from Grubbs are
the only extant experimental data.
Figure 3. Mass spectrum after spraying compound 3 from DCM (m/z )
1142) and reaction at 5.7 mTorr of norbornene in the 24-pole ion guide.
in the uncertainty bounds for E0. The final results for the three
reactions in Schemes 1-3 are listed in Table 1, along with fitting
parameters and assumptions. For the dissociative activation
reaction of the first-generation catalyst, 1 f 2 + PCy3, we take
E0 for the loss of tricyclohexylphosphine to be 1.45 ( 0.1 eV,
or 33.4 ( 2.3 kcal/mol for a loose transition state. The
comparable reaction for the second-generation system, 3 f 4
+ PCy3, shows E0 ) 1.60 ( 0.1 eV, or 36.9 ( 2.3 kcal/mol,
also for the loose transition state. The comparable dissociation
energies, assuming tight transition states, are 0.84 and 0.96 eV,
respectively. The error bounds contain the uncertainty of the
fit due to the statistical scatter of the data points and day-to-
day fluctuations in calibration and operation of the mass
spectrometers. One accordingly can claim with confidence that,
although the listed ranges of the two dissociation energies just
overlap, the difference between them, ∼0.15 eV or ∼3.5 kcal/
mol, is real and can be reproduced. Comparable measurements
published for copper complexes clearly show the ability to
distinguish reliably between dissociation thresholds that differ
by 0.1 eV.8 For the ring-closing metathesis, or backbiting
reaction, 7 f 4 + norbornene, the CID was only clean enough
for a reliable fit in the case of the second-generation catalyst,
shown in Scheme 3. Side reactions in the first-generation system
increased the number of significant channels beyond what
L-CID could reliably handle. The measured threshold energy
is E0 ) 1.45 or 0.79 ( 0.1 eV (33.4 or 18.2 ( 2.3 kcal/mol)
depending on whether the transition state is modeled as loose
or tight.
Being able to directly manipulate in the gas phase some of
the reactive intermediates in homogeneous catalytic cycles, we
believed that it should be possible to obtain quantitative kinetic
and thermochemical data for key elementary steps in the olefin
metathesis reaction by means of energy-resolved CID cross-
section measurements. Unfortunately, the deconvolution of the
experimental curves had not been possible for species of the
size and complexity of the first- and second-generation ruthe-
nium metathesis catalysts, but we have recently produced a
deconvolution program, L-CID,6 which is generally applicable
to the treatment of CID thresholds for organometallic ions of
the size and structural complexity commonly found in homo-
geneous catalysis.
An approximation to the reaction coordinate by means of a
linear synchronous transit at the BP86/ZORA-TZP level of
theory using ADF 2006 showed neither credible evidence for a
reverse barrier in the dissociation reaction nor any indication
of a rate-determining development of an intramolecular agostic
interaction in the reaction, 1 f 2 + PCy3, which suggests
strongly that the dissociation of a phosphine should be treated
with a loose transition state as has been argued in other reports.9
The energy profiles, as well as the coordinates and energies
along the dissociation path, are given in the Supporting
Information.
Application of L-CID to the energy-resolved CID cross-
section curves produced the activation energies for phosphine
loss from both the first- and second-generation ruthenium
metathesis catalysts, which, in the case of no reverse barrier,
correspond to the ligand binding energies. While the data
analysis itself can be done for either a loose or a tight transition
state, one would expect a loose transition state for a simple
(10) Zhao, Y.; Truhlar, D. G. Org. Lett. 2007, 9, 1967.
(11) (a) Sanford, M. S.; Ulman, M.; Grubbs, R. H. J. Am. Chem. Soc. 2001,
123, 749. (b) Love, J. A.; Sanford, M. S.; Day, M. W.; Grubbs, R. H. J.
Am. Chem. Soc. 2003, 125, 10103.
(7) Adlhart, C.; Chen, P. Angew. Chem., Int. Ed. 2002, 41, 4484.
(8) Zocher, E.; Sigrist, R.; Chen, P. Inorg. Chem., in press.
(9) Tsipis, A. C.; Orpen, A. G.; Harvey, J. N. Dalton Trans. 2005, 2849.
(12) Adlhart, C.; Chen, P. J. Am. Chem. Soc. 2004, 126, 3496.
(13) (a) Bornand, M.; Chen, P. Angew. Chem., Int. Ed. 2005, 44, 7909. (b)
Bornand, M.; Torker, S.; Chen, P. Organometallics 2007, 26, 3585.
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4811

A R T I C L E S
Torker et al.
Figure 4. Daughter spectrum of 7 (m/z ) 956) at a collision offset of -60 (bottom) and -120 eV (top) at a pressure of 90 µTorr of argon in the octopole
collision chamber. The latter energy was the highest energy taken for the fit. It is evident that new dissociation channels open at the highest energies. At
lower collision energies used in the deconvolution of the threshold, only negligible amounts of second and third channel products (cleavage of the C-Ru
bond; m/z ) 383 and 477) are formed.
Figure 6. Energy-dependent cross-section curve after extrapolation to zero
Figure 5. Energy-dependent cross-section curves after extrapolation to zero
pressure and deconvolution with L-CID for process 7 f 4.
pressure and deconvolution with L-CID for processes 1 f 2 (red) and 3 f
4 (blue).
for phosphine dissociation from the first- and second-generation
ligand dissociation. Grubbs entropies of activation also indicate
complexes were found to be 33.4 and 36.9 kcal/mol, respec-
a loose transition state. We nevertheless considered the pos-
tively. Both values are systematically higher than the values
sibility that there could be an intramolecular formation of an
reported by Grubbs for the same process in the solution-phase
agostic bond, either preceding or synchronous with phosphine
reaction.11 In that previous work, it was reported that the
departure, which could lead to a rate-determining transition state
initiation rate, in which the dissociation of tricyclohexylphos-
that would be tight. We had been alerted to the possibility of
phine plays the dominant role, is dependent on the solvent,
an intramolecular agostic bond by structural distortions in the
showing a marked dependence on the dielectric constant of the
solvent. The increasing dissociation rate as the medium is made
more polar was rationalized by the greater polarity of the 14-
electron intermediate relative to its 16-electron precursor. It is
therefore no surprise that the threshold energies for dissociation
in the gas phase are higher than in the solution-phase reaction.
Interestingly, the difference in activation energies for phosphine
X-ray structures of many of the ruthenium carbene complexes.13
The reaction coordinate, approximated by a linear synchronous
transit, showed no evidence for the incipient formation of an
agostic bond, and there was furthermore no evidence for a
reverse barrier along the energy profile, all of which indicates
that the treatment of the phosphine departure via a loose
transition state is very likely to be correct. The threshold energies
9
4812 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

Ruthenium Carbene Metathesis Catalysts
A R T I C L E S
Table 1. Threshold Energies (E0) after Deconvolution with L-CID.
Values Arise from Three Different Experimental Data Sets and 15
Fits per Number of Rotors. ∆-Values Represent a 95% Confidence
Interval with Respect to a Gaussian Distribution. R′-Values Cover
a Huge Range within the Limit for the TS-model (5000-5600 for
loose TS and 0-495 for tight TS)
1
1
TS-model
rotors
E
0 (eV)
∆E
0 (eV)
ν
eff (cm-
)
∆νeff (cm-
)
1 f 2
loose
0
2
13
0
2
13
0
6
8
18
0
6
8
18
0
6
8
0
6
8
1.43
1.45
1.53
0.84
0.84
0.85
1.53
1.58
1.60
1.63
0.95
0.96
0.96
0.96
1.38
1.45
1.43
0.78
0.79
0.80
0.03
0.04
0.05
0.01
0.02
0.01
0.04
0.05
0.05
0.06
0.02
0.02
0.02
0.02
0.04
0.04
0.06
0.02
0.02
0.02
956
972
988
985
980
980
923
951
963
978
977
984
973
973
523
529
533
539
547
549
57
26
17
21
23
18
51
46
37
24
24
21
28
24
24
20
25
29
29
33
tight
3 f 4
loose
tight
Figure 7. Energetic surface of the reaction of the second generation Grubbs
catalyst 3 with norbornene relative to the 14-electron complex 4. Green
points correspond to experimental values based on loose transition state
assumptions for both CIDs (3 f 4 and 7 f 4). The dashed line gives the
region that is not accessible by this experiment at this time. Truhlar’s energy
of 3 is based on M06-L/TZQ-CP, and the complex 5 shows the binding
energy of ethylene with M06-L/DZQ. Geometries of the BP86 surface are
given in the Supporting Information.
7 f 4
loose
tight
3 calculated with BP86/ZORA-TZP using ADF 2006 (geom-
etries are given in the Supporting Information) in comparison
with the experimental value based on a loose transition state
assumption, together with Truhlar’s value for the phosphine
binding energy of 38.2 kcal/mol. The subsequent steps along
the reaction for the overall metathesis reaction with norbornene
from 4 f 7 require a more detailed examination. Again, full
DFT energies at the BP86/ZORA-TZP level of theory have been
calculated for comparison with the experiment. This common
level of theory does show the qualitatively correct orders of
stability, but the errors in absolute energies are very large.
The energy-resolved CID cross-section curve in Figure 6
depicts the threshold for a composite reaction which amounts
to a ring-closing metathesis culminating in the loss of the cyclic
olefin product, in this case, norbornene. All of the reliable
computational evidence indicates that the resting state is the
π-bound intramolecular olefin complex, and the experimentally
observable product is the 14-electron complex resulting from
dissociation of norbornene. There are several steps between these
two points, and the question may be rightly asked as to which
step is rate-determining. The answer to the question determines
whether one takes the deconvoluted E0 for a loose versus a tight
transition state. The candidates for the rate-determining step are
(i) formation of the metallacyclobutane from the intramolecular
π-complex, 7 f 6, (ii) cleavage of the metallacyclobutane to a
carbene complex with coordinated norbornene, 6 f 5, or (iii)
dissociation of the coordinated norbornene, 5 f 4. Both i and
ii would have tight transition states, whereas iii would have a
loose one. The bulk of the ring strain incurred in step i. One
can also reason that, starting from the metallacyclobutane, 6,
going backward to the ring-opened intramolecular π-complex,
7, should have a lower activation energy than the step going
forward to coordinated norbornene, 5. This application of the
Hammond Postulate suggests that step i cannot be rate-limiting,
which is likely to be legitimate given that both the forward and
reverse reaction of 6 are of the same type. Given that the strain
release upon ring-opening of norbornene, as defined by the
homodesmotic equation in Scheme 4, comes out to 15 kcal/
mol,15 one can accordingly consider the consequences of the
loss between the first- and second-generation catalysts is
independent of whether the reaction occurs in the gas phase or
in toluene.
Several quantum chemical calculations have examined the
loss of tricyclohexylphosphine from the first- and second-
generation catalysts, with the reported dissociation energies
being uniformly lower than the values from our gas-phase
experiment.14 The very low values from truncated model
compounds are likely unrepresentative of the situation for the
fully elaborated, complicated ligands in the experimental
systems, but, more fundamentally, it has been claimed that the
usual DFT methods systematically underestimate the phosphine
binding,10 at least for the ruthenium carbene complexes. Given
the previously mentioned solvent dependence of the phosphine
binding, one would expect the gas-phase values to be higher
than those reported by Grubbs in solution. The typical DFT
calculation, however, produces values for the gas-phase which
are lower than those measured in solution, which strongly
suggests that these calculations are an unreliable guide to the
absolute thermochemistry of this class of organometallic
complexes. Interestingly, Zhao and Truhlar have reported a new
density functional, M06-L, which has been designed to better
handle the medium-range correlation energy.10 It is claimed that
this treatment is particularly well-suited to dissociation energies
in organometallic complexes. Application of the M06-L density
functional, with a triple-ú quality basis set and counterpoise
correction, to the first- and second-generation ruthenium carbene
complexes produces tricyclohexylphosphine dissociation ener-
gies of 34.2 and 38.2 kcal/mol, respectively, which are in
remarkable agreement with the measurements in the present
report. Figure 7 shows the bond dissociation energy of unlabeled
(14) (a) Cavallo, L.; J. Am. Chem. Soc. 2002, 124, 8965. (b) see reference 6.
(c) See reference 7. (d) See reference 8. (e) van Rensburg, W. J.; Steynberg,
P. J.; Kirk, M. M.; Meyer, W. H.; Forman, G. S. J. Organomet. Chem.
2006, 691, 5312. (f) Jordaan, M.; van Helden, P.; Sittert, C. G. C. E.;
Vosloo, H. C. M. J. Mol. Catal. A: Chemical 2006, 254, 145. (g) Forman,
G. S.; McConnell, A. E.; Tooze, R. P.; van Rensburg, W. J.; Meyer, W.
H.; Kirk, M. M.; Dwyer, C. L.; Serfontein, D. W. Organometallics 2005,
24, 4528.
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4813

A R T I C L E S
Scheme 4
Torker et al.
have the same consequence in that it is clear from the reaction
in which a ring-closing metathesis of the intramolecular
π-complex, 7, goes to the 14-electron complex, 4, and nor-
bornene, that the ligand binding energy of a π-bound olefin is
very significantly less than that of tricyclohexylphosphine. This
is consistent with the great difficulty in observing the olefin
π-complexes in ruthenium-catalyzed metathesis reactions. The
relative binding energies are furthermore important in judging
the extent to which phosphine binding and (re)dissociation is
kinetically important during turnover of metathesis catalysts.16
Considering the considerable dependence of the phosphine
binding energy with solvent polarity, and, presumably, a
comparable but not necessarily parallel dependence in the olefin
binding energies, it would not be unexpected that the extent of
intervention of phosphine ligands in multiple turnovers is a
delicate balance determined by concentrations, solvents, tem-
perature, and other conditions. The solution-phase experiment,
as well as computational studies, indicate a rate-limiting phos-
phine dissociation, i.e., step iii is rate-determining with a loose
transition state, for the second-generation system. Accordingly,
the experiment indicates that the olefin binding energy to the
14-electron, second-generation ruthenium carbene complex
should be slightly more than 18 kcal/mol, which can be
compared to Zhao and Truhlar’s computed binding energy of
18.8 kcal/mol for ethylene. One would expect norbornene to
bind somewhat more strongly than ethylene because of the
pyramidalization in the strained olefin, so the degree of agree-
ment here is again more than satisfactory.
two remaining choices. If one were to assume that step ii is
rate-limiting, i.e., the step in which the metallacyclobutane, 6,
is cleaved to a π-bound norbornene complex, 5, then a tight
transition state would be appropriate. The 18 kcal/mol threshold
energy extracted by L-CID would be just enough to account
for the strain in either the metallacyclobutane, 6, or the bound
norbornene relative to intramolecular π-complex, 7. The assump-
tion of rate-limiting step ii would also place a bound on the
ligand binding energy of norbornene to the 14-electron complex,
4. Consider that a rate-limiting step ii would necessarily mean
that the transition state for dissociation of norbornene from the
16-electron complex, 5, cannot lie more than 18 kcal/mol above
the intramolecular π-complex. At the same time, the assumption
means that the π-complex, 5, cannot lie below the intramolecular
π-complex, 7, because, were that to be the case, then the former
and not the latter would be the resting state, and the overall
reaction by which norbornene is lost from this resting state by
CID would proceed necessarily by way of a loose transition
state. The two conditions mean that an assumption of a rate-
limiting step ii would constrain the ligand binding energy of
norbornene to the 14-electron carbene complex to be less than
18 kcal/mol. The other choice, rate-limiting step iii via a loose
transition state, would place the transition state for dissociation
of norbornene 33 kcal/mol above the intramolecular π-complex,
7. The transition states for steps i and ii are constrained to lie
below this limit, but this is not a difficult condition to meet. If
we were to neglect the small energy difference between a
ruthenium benzylidene and a ruthenium alkylidene, then the
difference in energy between the intramolecular π-complex, 7,
and the norbornene complex, 5, is principally the 15 kcal/mol
strain of norbornene relative to a monocyclic reference. This
would place the ligand binding energy of norbornene to the 14-
electron carbene complex around 18 kcal/mol, or perhaps
slightly higher.
Conclusions
Mass spectrometric methods lead to the experimental deter-
mination of the phosphine binding energy in first- and second-
generation ruthenium metathesis catalysts. An experimental
study of a gas-phase ring-closing metathesis sets bounds on the
binding energy of the olefinic substrate to the 14-electron active
species. The absolute binding energies agree remarkably well
with the latest DFT calculations using a density functional
specifically suited to organometallic thermochemistry.
Acknowledgment. The authors acknowledge support from
the ETH Zu¨rich and the Swiss Nationalfonds.
Supporting Information Available: Data acquisition and
processing (kinetic energy distribution, scaled ion intensities as
a function of the collision energy and pressure, cross-section
curves at different pressures and extrapolation to zero pressure).
Linear transit calculations for 1 f 2 and 3 f 4 and minimum
geometries of 1 and 3 as well as of the end points at Ru-P
distances of 5.00 Å. Furthermore, energy surfaces of the reaction
of unlabeled 1 and 3 with norbornene at the BP86/ZORA-TZP
level of theory using ADF 2006. This material is available free
of charge via the Internet at http://pubs.acs.org.
There is no a priori reason in the experiment itself that favors
one or the other transition states, i.e., step ii or step iii, but both
(15) (a) Lebedev, B.; Smirnova, N.; Kiparisova, Y.; Makovetsky, K. Makromol.
Chem. 1992, 193, 1399. The heat of polymerization of norbornene has been
measured to be about 62.2 kJ/mol (∼15 kcal/mol), which is essentially
determined by strain release. (b) Khoury, P. R.; Goddard, J. D.; Tam, W.
Tetrahedron 2004, 60, 8103. The total ring strain in norbornene has been
determined to be 19.2 kcal/mol and 6.2 kcal/mol in cyclopentane (slightly
less in substituted cyclopentanes), which gives a strain release after ROMP
of norbornene of slightly more than 13 kcal/mol (19.2-6.2).
JA078149Z
(16) Sanford, M. S.; Love, J. A.; Grubbs, R. H. J. Am. Chem. Soc. 2001, 123,
6543.
9
4814 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008