A combined gas-phase, solution-phase, and computational study of C-H activation by cationic Iridium(III) complexes

Article abstract of DOI:10.1021/ja970995u

A combination of electrospray ionization MS/MS techniques, isotopic labeling experiments in the gas-phase and solution, and ab initio calculations is used to study the C-H activation reactions of [Cp*Ir(PMe₃)(CH₃)]⁺ and [CpIr(PMe₃)(CH₃)]⁺. The reaction in the gas phase was found to proceed through a Cp or [Cp*Ir(η²-CH₂-PMe23)]⁺ intermediate. Quantitative collision-induced dissociation (CID) threshold measurements were used along with general models for ion-molecule reactions to construct potential energy diagrams which rationalize the gas-phase results. The comparison between the two complexes, and between the reactions in the gas phase and in solution, suggests that the reaction through the intermediacy of a metallaphosphacyclopropane could be favored over the conventional (and simpler) oxidative addition/reductive elimination mechanism when the Ir(III) complex is rendered more electron deficient.

Full text of DOI:10.1021/ja970995u

J. Am. Chem. Soc. 1997, 119, 10793-10804
10793
A Combined Gas-Phase, Solution-Phase, and Computational
Study of C-H Activation by Cationic Iridium(III) Complexes
Christian Hinderling, Derek Feichtinger, Dietmar A. Plattner,* and Peter Chen*
Contribution from the Laboratorium fu¨r Organische Chemie der Eidgeno¨ssischen Technischen
Hochschule (ETH) Zu¨rich, CH-8092 Zu¨rich, Switzerland
ReceiVed March 28, 1997X
Abstract: A combination of electrospray ionization MS/MS techniques, isotopic labeling experiments in the gas-
phase and solution, and ab initio calculations is used to study the C-H activation reactions of [Cp*Ir(PMe3)(CH3)]+
and [CpIr(PMe3)(CH3)]+. The reaction in the gas phase was found to proceed through a Cp or [Cp*Ir(η2-CH2-
PMe2)]+ intermediate. Quantitative collision-induced dissociation (CID) threshold measurements were used along
with general models for ion-molecule reactions to construct potential energy diagrams which rationalize the gas-
phase results. The comparison between the two complexes, and between the reactions in the gas phase and in
solution, suggests that the reaction through the intermediacy of a metallaphosphacyclopropane could be favored
over the conventional (and simpler) oxidative addition/reductive elimination mechanism when the Ir(III) complex is
rendered more electron deficient.
Introduction
Scheme 1
The chemical activation of alkane C-H bonds under mild
conditions has been a long-standing goal in organic and
organometallic chemistry.1 Two mechanisms have been sug-
gested by Bergman and co-workers2,3 for the recently reported
σ-bond metathesis reaction by cationic Ir(III) complexes,4 cited
as the most facile C-H activation reaction yet found for an
organometallic complex (Scheme 1). Although one would
ordinarily expect late transition metals to engage in an oxidative
addition/reductive elimination sequence of reactions, the con-
certed mechanism comes into consideration for the Ir(III)
complex because it might be unfavorable for the strongly
electron-deficient 16-electron Ir(III) complex to go to an 18-
electron Ir(V) species in an oxidative addition step. These
complexes, extensively investigated recently by Bergman and
co-workers,2,3,5 perform C-H activation without prior photo-
chemical excitation, and are therefore particularly interesting
as model systems leading to catalytic cycles. On the basis of
the increase in reactivity as the ligand binding is weakened,
going from triflate to dichloromethane, the actual reactive
species was presumed to be 2, which would be formed from 1
in a pre-equilibrium in the solution-phase studies. For the
dichloromethane-ligated complex, activation of C-H bonds was
observed well below room temperature. According to ab initio
calculations by Hall and co-workers,6 an oxidative addition/
reductive elimination mechanism is more likely than the
concerted metathesis, although solid experimental evidence for
either alternative has yet to be presented. A similar pattern of
reactivity, with the same set of mechanistic possibilities, has
also been reported for Pt(II) complexes.7
We report here experimental and computational results which
clearly demonstrate that the C-H activation reaction by cationic
Ir(III) complexes like 2 and 6, in the gas phase, proceeds by a
third mechanism, namely, the elimination of CH4 to produce
an intermediate metallaphosphacyclopropane, and then, ulti-
mately, reaction with the C-H bond of alkanes and arenes
(Scheme 2). A preliminary report of the gas-phase chemistry
of organometallic ions 5-8 (Scheme 3) by electrospray tandem
mass spectrometry has recently appeared.8 We expand and
complement that work in the present report with an extension
of the study to the complexes 1-4, solution-phase isotopic
labeling experiments, ab initio quantum chemical calculations,
and quantitative collision-induced dissociation (CID) threshold
measurements. The comparison of the Cp and Cp* series of
complexes, as well as the comparison of gas-phase to solution-
phase reactivity, illuminates the features determining the choice
between competing mechanisms for C-H activation.
X Abstract published in AdVance ACS Abstracts, October 15, 1997.
(1) Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Petersen, T. H.
Acc. Chem. Res. 1995, 28, 154.
(2) Arndtsen, B. A.; Bergman, R. G. Science 1995, 270, 1970.
(3) Burger, P.; Bergman, R. G. J. Am. Chem. Soc. 1993, 115, 10462.
(4) Recent progress in C-H activation by Ir(III) complexes has been
reviewed: Lohrenz, J. C. W.; Jacobsen, H. Angew. Chem. 1996, 108, 1403;
Angew. Chem., Int. Ed. Engl. 1996, 35, 1305.
(5) Veltheer, J. E.; Burger, P.; Bergman, R. G. J. Am. Chem. Soc. 1995,
117, 12478.
(6) Strout, D. L.; Zaric, S.; Niu, S.; Hall, M. B. J. Am. Chem. Soc. 1996,
118, 6068.
Experimental Section
Preparation of the organometallic complexes in this study was done
by standard Schlenk techniques or in an argon-purged glovebox.
Characterization of the compounds by 1H and 13C NMR (Varian
GEMINI 2000, 300 MHz), X-ray crystallography (Enraf Nonius Mach
3), and electrospray mass spectrometry (Finnigan MAT TSQ-7000) was
done. Detailed synthetic procedures for the various complexes, their
(7) Holtcamp, M. W.; Labinger, J. A.; Bercaw, J. E. J. Am. Chem. Soc.
1997, 119, 848.
(8) Hinderling, C.; Plattner, D. A.; Chen, P. Angew. Chem. 1997, 109,
272; Angew. Chem., Int. Ed. Engl. 1997, 36, 243.
S0002-7863(97)00995-5 CCC: $14.00 © 1997 American Chemical Society

10794 J. Am. Chem. Soc., Vol. 119, No. 44, 1997
Hinderling et al.
Scheme 2
Scheme 3
physical characterization, and the procedures by which the solution-
phase reactions were done are given in the Supplemental Information.
Complexes 1 (L ) -OTf, CH3CN) were prepared according to the
procedures by Burger and Bergman,3 starting from the dimeric
complex,9 [Cp*Ir(Cl)2]2, and proceeding through the equilibration of
the bis-methyl complex,10 Cp*Ir(PMe3)(CH3)2, with the bis-triflate,11
Cp*Ir(PMe3)(OTf)2. Complexes 5 (L ) -OTf, CH3CN) were prepared
analogously,8 starting from CpIr(C2H4)2, described by Heinekey,12 which
was converted to the polymeric complex [CpIr(Cl)2]n (not isolated),
and then taken on to 5 as detailed in the Supplemental Information.
Cations 2 and 6, the [Cp*Ir(PMe3)(CH3)]+ and [CpIr(PMe3)(CH3)]+
complexes, were prepared in the gas phase by electrospray of CH2Cl2
or CHCl3 solutions containing the [Cp*Ir(PMe3)(CH3)]+ or the [CpIr-
(PMe3)(CH3)]+ cations. The slightly modified Finnigan MAT TSQ7000
electrospray tandem mass spectrometer was set up with an octopole,
quadrupole, octopole, quadrupole (O1/Q1/O2/Q2) configuration behind
a conventional electrospray ion source.13 The extent of desolvation
and collisional activation prior to thermalization, reaction, and/or
analysis was controlled by the tube lens potential in the electrospray
source, which typically ranged from 35 (mild conditions) to 75 V (hard
conditions).14 The first octopole was fitted with an open cylindrical
sheath around the rods into which a collision gas could be bled for
thermalization or reaction at pressures up to 20 mTorr. Control
of a single mass from among all of the ions produced in O1, which are
then collided or reacted with a target gas in O2, and mass-analyzed by
Q2. This mode is used to determine the specific reactivity of a species
of a given mass. The third mode of operation, the RFD mode, is used
to record CID thresholds for quantitative thermochemical measure-
ments.16,17 Ions produced in O1 are again mass-selected, but with Q1
acting as a high-pass filter. The selected ions are then collided with a
target gas in O2, and mass-analyzed in Q2. In the RFD mode, the
reactions of ions in a particular mass range are isolated by setting the
mass cutoff above and below the desired range and subtracting the
latter from the former spectra. The more complicated RFD mode
operation is used instead of the daughter-ion mode for the threshold
measurements because the ion kinetic energy distribution is narrower
(∼0.4-0.6 versus ∼1.3-2.0 eV fwhm in the laboratory frame).
Furthermore, the kinetic energy distribution in the RFD mode is nearly
Gaussian, in comparison to a distribution in daughter-ion mode that,
while also not too far from Gaussian, has a non-negligible high-energy
tail. Both factors combine to make deconvolution of the threshold
measurement more reliable (even though we can use the measured
kinetic energy distribution in the fit). In both the daughter-ion and
RFD modes (unless otherwise specified), the pressure in O2 was kept
low enough so as to minimize the chance of multiple collisions in the
collision cell.18 A kinetic energy distribution of the ions in the
laboratory frame of <0.6 eV fwhm corresponds to a distribution of
<0.1 eV in the center-of-mass frame when the collision partner is argon
or xenon, and is a non-negligible contributor to the cited (2 kcal/mol
uncertainty in the CID threshold determinations. As a control to check
that we could correctly measure CID thresholds, we reproduced
experiments by Kebarle19 and Armentrout20,21 for the reaction (H2O)n-
(H3O+) f (H2O)n-1(H3O+) + H2O. The threshold for n ) 1 was
+
experiments with Mn(CO)6 have found that ∼10 mTorr of Ar in O1
are sufficient15 for thermalization of ions produced by the electrospray
source to the 70 °C manifold temperature. It should be noted here
that heating of the external inlet lines for the thermalization gas is
important to ensure that the gas is fully equilibrated to the manifold
temperature. The second octopole was operated as a conventional
collision-induced dissociation (CID) cell, and run as specified in the
commercial instrument for MS/MS experiments. Spectra were recorded
in three different modes. In the normal ES-MS mode, only one
quadrupole is operated, and an ordinary mass spectrum of the
electrosprayed ions is recorded. This mode, exemplified by entries
1-3 and 13-18 in Table 1, serves primarily to characterize the ions
produced by a given set of conditions. In the daughter-ion mode, used
in entries 4-12 and 19-22 in Table 1, Q1 is used to mass-select ions
(16) The present state of organometallic thermochemistry for Ir com-
plexes is reviewed by: Halpern, J. Inorg. Chim. Acta 1985, 100, 41. In
particular, Stoutland et al. (Stoutland, P. O.; Bergman, R. G.; Nolan, S. P.;
Hoff, C. D.; Polyhedron 1988, 7, 1429) find reasonably certain Ir-H and
Ir-C bond strengths for (Cp*)(PMe3)Ir(cyclohexyl)(H) of 75 and 52
kcal/mol, respectively.
(17) Thermochemical work by mass spectrometric techniques has been
done for small organometallic ions. For several examples, see: Halle, L.
F.; Armentrout, P. B.; Beauchamp, J. L. Organometallics 1982, 1, 963.
Armentrout, P. B.; Halle, L. F.; Beauchamp, J. L. J. Am. Chem. Soc. 1981,
103, 6501. Armentrout, P. B.; Beauchamp, J. L. J. Am. Chem. Soc. 1981,
103, 784. For some more recent work on small complexes, see: Sunderlin,
L. S.; Wang, D.; Squires, R. R. J. Am. Chem. Soc. 1993, 115, 12060.
Sunderlin, L. S.; Squires, R. R. J. Am. Chem. Soc. 1993, 115, 337.
(18) Pressure in the collision cell was kept below 4 × 10-5 torr (usually
argon), as measured by a cold cathode gauge. The interaction region was
18.5 cm long. Typically, thresholds were measured at four different pressures
and extrapolated to zero pressure. Even at the highest pressures used, the
amount of fragmentation of the parent ions remained below 15%.
(19) Anderson, S. G.; Blades, A. T.; Klassen, J.; Kebarle, P. Int. J. Mass
Spectrom. Ion Proc. 1995, 141, 217. We observed the threshold for loss of
H2O from protonated water dimer to be E0 ) 1.28 eV (298 K value). The
absolute value of the energy scale was determined by stepping the potential
of the collision cell, and then fitting the derivative of the resulting retardation
curve to a Gaussian function. The maximum of the Gaussian was taken as
the absolute zero of the energy scale. It coincided with the potential of the
first octopole.
(9) White, C.; Yates, A.; Maitlis, P. M. Inorg. Syn. 1992, 29, 228.
(10) Buchanan, J. M.; Stryker, J. M.; Bergman, R. G. J. Am. Chem. Soc.
1986, 108, 1537.
(11) Stang, P. J.; Huang, Y.; Arif, A. M. Organometallics 1992, 11, 231.
(12) Heinekey, D. M.; Miller, J. M.; Koetzle, T. F.; Payne, N. G.; Zilm,
K. W. J. Am. Chem. Soc. 1990, 112, 909.
(13) The basic electrospray source for mass spectrometry is described
by: Whitehouse, C. M.; Dreyer, R. N.; Yamashita, M.; Fenn, J. B. Anal.
Chem. 1985, 57, 675.
(14) Approximate ranges for tube lens potentials are the following: low
≈30-44 V, medium ≈45-60 V, high ≈60-80 V.
(15) Thermalization was judged to be complete when an observed (but
relatively weak) dependence of the CID threshold curves for the loss of
one CO ligand on the temperature of the heated capillary (from 100 to 240
°C) in the electrospray ion source was eliminated. N2 in the sheath gas was
found to give a small amount of a dinitrogen complex with the 16-electron
complexes. For mass spectra in the “normal” mode, CO2 was therefore used
as the sheath gas. For daughter-ion and RFD modes, in which there is a
mass-selection step in the experiment, N2 could be used.

C-H ActiVation by Cationic Iridium(III) Complexes
J. Am. Chem. Soc., Vol. 119, No. 44, 1997 10795
Table 1. Schematic Description of the Gas-Phase Tandem Mass Spectrometric Experiments on Complexes 1 and 5a

10796 J. Am. Chem. Soc., Vol. 119, No. 44, 1997
Table 1. (Continued)
Hinderling et al.

C-H ActiVation by Cationic Iridium(III) Complexes
Table 1. (Continued)
J. Am. Chem. Soc., Vol. 119, No. 44, 1997 10797
a Entries without items in O2 and Q2 are ordinary electrospray mass spectra. Items with the designation “mass selection...” indicate daughter-ion
spectra.
extracted from our data with Armentrout’s CRUNCH program, yielding
a value in agreement with the published data.19,21
Ab initio calculations of geometries and frequencies for the orga-
nometallic complexes in this report were performed with Gaussian 9422
on IBM RS/6000-590 or DEC Alpha AXP 8400 5/300 workstations.
The initial calculations at the Hartree-Fock level employed a basis
set comprising the Hay and Wadt effective core potential23 (ECP) for
iridium and a modest 3-21G basis set for the light atoms, comparable
to that found by Frenking and co-workers,24 to give good geometries
and relative energies for other organometallic systems. The larger
LANL2DZ basis set22 (D95 on first row, Los Alamos ECP plus double-ú
on Na-Bi) was used for subsequent Hartree-Fock and DFT (with the
B3LYP density functional) calculations. Full geometry optimization
and frequency calculation was done at the Hartree-Fock level and with
DFT, yielding very similar results throughout. The DFT calculations
closely parallel those by Hall,6 and by Siegbahn,25 for other metal
systems. The consistency of the geometries at differing levels of theory
and their intuitive reasonableness lend credibility to the computational
results. Accordingly, the DFT calculations with the large basis were
deemed sufficient for the computation of frequencies to be used in the
RRKM kinetic shift calculations needed to extract thermochemistry
from CID threshold measurements. Even at this relatively modest level
of theory, the computation of sufficiently realistic geometries and
frequencies represents one of the real-life limiting factors in the entire
experiment.
Figure 1. Electrospray mass spectrum of [Cp*Ir(PMe3)(CH3)-
(NtCCH3)]+ with the m/e 458, 460 doublet due to the 1:2 natural
abundance of 191Ir and 193Ir, taken with a very low tube lens potential,
showing the intact organometallic molecular ion transferred from
solution to the gas phase. The spectrum corresponds to entry 14 in
Table 1.
Results
The gas-phase results are summarized in Table 1. Some
representative mass spectra, keyed to the table entries, are shown
(20) Dalleska, N. F.; Honma, K.; Armentrout, P. B. J. Chem. Phys. 1989,
90, 5466.
(21) Dalleska, N. F.; Honma, K.; Armentrout, P. B. J. Am. Chem. Soc.
1993, 115, 12125.
(22) Gaussian 94 (Revision A.1); M. J. Frisch, G. W. Trucks, H. B.
Schlegel, P. M. W. Gill, B. G. Johnson, M. A. Robb, J. R. Cheeseman, T.
A. Keith, G. A. Petersson, J. A. Montgomery, K. Raghavachari, M. A. Al-
Laham, V. G. Zakrzewski, J. V. Ortiz, J. B. Foresman, J. Cioslowski, B. J.
Stefanov, A. Nanyakkara, M. Challacombe, C. Y. Peng, P. Y. Ayala, W.
Chen, M. W. Wong, J. L. Andres, E. S. Replogle, R. Gomperts, R. L. Martin,
D. J. Fox, J. S. Binkley, D. J. DeFrees, J. Baker, J. P. Stewart, M. Head-
Gordon, C. Gonzalez, and J. A. Pople; Gaussian, Inc.: Pittsburgh, PA, 1995.
(23) Hay, R. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270, 284, 299.
(24) Jonas, V.; Frenking, G.; Reetz, M. T. J. Comp. Chem. 1992, 13,
919. Jonas, V.; Frenking, G.; Reetz, M. T. J. Comp. Chem. 1992, 13, 935.
Veldkamp, A.; Frenking, G. J. Comp. Chem. 1992, 13, 1184.
(25) Siegbahn, P. E. M. J. Am. Chem. Soc. 1996, 118, 1487.
Figure 2. Electrospray mass spectrum of [Cp*Ir(PMe3)(CH3)]+ with
the m/e 417, 419 doublet due to the 1:2 natural abundance of 191Ir and
193Ir, taken with the application of a larger tube lens potential to remove
the weakly bound solvent ligand. The spectrum corresponds to entry
15 in Table 1.
in Figures 1-6 to illustrate the quality of the original data. With
some small differences (indicated below), the reactions of the
[Cp*Ir(PMe3)(CH3)]+ and [CpIr(PMe3)(CH3)]+ complexes 2 and

10798 J. Am. Chem. Soc., Vol. 119, No. 44, 1997
Hinderling et al.
Figure 3. Electrospray mass spectrum of cyclic cation, [Cp*Ir(η2-
(CH2PMe2)]+, with the m/e 401, 403 doublet due to the 1:2 natural
abundance of 191Ir and 193Ir, taken with the application of a high enough
tube lens potential to induce the elimination of CH4 by CID. The
spectrum corresponds to entry 16 in Table 1.
Figure 6. Electrospray mass spectrum, in daughter-ion mode, of the
products from the gas-phase reaction of mass-selected [Cp*Ir(η2-
CD2PMe2-d6)]+ (only the 193Ir isotopic species) and benzene. The peaks
at m/e 411 and 489 correspond to the ion without and with addition of
benzene. The spectrum corresponds to entry 22 in Table 1.
(2) The ion assigned as 6, prepared by collisional removal
of CH3CN from 5 and thermalized, picks up neutral 2-e- ligands
such as acetonitrile or CO with high efficiency at very low
collision energies. Also evident (at much lower intensity) are
the ions due to 7 and (7 + acetonitrile) [Table 1, entry 12].
(3) After collisional removal of the acetonitrile ligand from
1 or 5, the resulting ions 2 and 6 readily lose CH4 with
cyclometalation upon further CID [Table 1, entries 2, 5, 16,
and 21].
(4) In the gas phase, all detectable cyclometalation proceeds
exclusively on the methyl groups of the phosphine, and not on
the Cp or Cp* [Table 1, entries 11 and 21].
(5) The metallaphosphacyclopropanes 3 and 7 react readily
with pentane and benzene to yield the products (with respect
to the original 2 and 6) of an overall σ-bond metathesis [Table
1, entries 5 and 22].
(6) The addition of 3 or 7 to benzene is reversible upon further
CID [Table 1, entries 6, 7, and 20].
Figure 4. Electrospray mass spectrum of [Cp*Ir(PMe3)(CH3)]+ at m/e
417, 419 and [Cp*Ir(PMe3)(Ph)]+-d6 at m/e 485, 487, prepared by
reaction of a mixture (such as shown in Figure 3) of [Cp*Ir(PMe3)(CH3)]+
and [Cp*Ir(η2-CH2PMe2)]+ with benzene-d6. The spectrum corresponds
to entry 18 in Table 1.
(7) When â-hydride elimination is possible from 4 or 8, it
occurs upon CID [Table 1, entries 8, 9, and 19].
(8) The cyclometalated ions 3 and 7 are much more reactive,
with regard to C-H activation in the gas phase, than the
precursor ions 2 and 6 [Table 1, entries 3, 4, and 18].
The differences in the gas-phase chemistry of the Cp*
complexes 1-4 and the Cp analogs 5-8 are confined to two
areas. Comparing entry 3 or 5 to entry 17, one sees that, while
the Cp-substituted complexes add to pentane, cyclohexane, or
benzene, the Cp*-substituted analogs react with pentane or
benzene only. No cyclohexane adducts were observed in the
gas-phase reactions of 2 or 3. Secondly, comparison of entries
7 and 19 shows that, in the CID of the two phenyl-substituted
complexes 4 and 8 (R ) Ph), there is complete deuterium
scrambling in 8, but only partial scrambling in 4. An explana-
tion for both observations, based on general features of ion-
molecule reactions, will be given below in the Discussion. It
should also be noted that reactions of 7 with methane produced
only the barest hint of addition products.
The quantitative CID threshold (argon as collision partner)
for loss of CH4 from 6 to form 7 is shown, with the fit to an
analytical threshold function, in Figure 7. The data were
recorded as described in the RFD mode, and deconvoluted with
the CRUNCH program from Armentrout and co-workers,20,21
with the internal energy of the ions set by the 70 °C manifold
temperature to which they were thermalized. The RRKM
calculation comprising part of the deconvolution requires the
frequencies for the product ions, as well as estimates for the
five normal modes in the transition state that become, at the
Figure 5. Electrospray mass spectrum, in daughter-ion mode, of the
products from the collision-induced dissociation of mass-selected
[Cp*Ir(PMe3)(Ph)]+-d6 (only the 193Ir isotopic species) with xenon as
the collision partner. The peaks at m/e 404 and 487 correspond to the
ion with and without loss of benzene-d5. The expanded inset showing
the m/e 404 peak also shows a small peak at m/e 403, corresponding
to loss of benzene-d6 upon CID. The spectrum corresponds to entry 20
in Table 1.
6 are entirely parallel, with the Cp series of complexes showing
qualitatively higher reactivity than their Cp* analogs. One can
draw a series of broad conclusions from the observed gas-phase
reactions.
(1) Electrospray ionization cleanly produces the molecular
ion of organometallic complexes, with even weakly-bound
ligands still intact [Table 1, entries 1, 13, and 14].

C-H ActiVation by Cationic Iridium(III) Complexes
J. Am. Chem. Soc., Vol. 119, No. 44, 1997 10799
lost. The reported threshold is the average of five independent
determinations. For each determination, four separate threshold
measurements were done. A similar measurement for the 2 f
3 + CH4 reaction was not attempted because the five additional
methyl rotors on the Cp* ligand would result in a much larger
kinetic shift, already ∼0.4 eV in the reaction with 6, making
accurate determination of the threshold more difficult.
The solution-phase experiments showed somewhat different
chemistry. Starting with the undeuterated complex [Cp*Ir-
(PMe3)(CH3)(OTf)] in CD2Cl2, reaction with benzene-d6 through
the putative intermediacy of [Cp*Ir(η2-CH2PMe2)]+, complex
3, should lead to deuterium incorporation in the phosphine ligand
and loss of CH4 instead of CH3D (Scheme 4). The mirror-
image of the deuterium “wash-in” experiment would be a
deuterium “wash-out” from [Cp*Ir(PMe3-d9)(CH3)(OTf)] by
reaction with C6H6 in CD2Cl2 solution. In both experiments,
1
Figure 7. Representative collision-induced dissociation (CID) threshold
curve for the reaction 6 f 7 + CH4, showing the dependence of the
dissociation cross-section (σ, 10-16 cm2) again collision energy (eV,
center-of-mass frame), and the fit to Armentrout’s threshold function
(described in ref 17). The fit covered data between 0 and 3 eV, and
yielded the following parameters (0 K): E0 ) 0.58(2) eV, σ ) 52.8(3),
and n ) 1.42(4).
replacement of the H NMR signals of 1 by those of 4 (R )
Ph) was complete and quantitative after several hours upon
reaction at either room temperature or 45 °C. However, in
neither the former experiment, probed by H NMR, nor the
2
latter, probed by 1H NMR, was there any evidence for deuterium
“wash-in” or “wash-out”, respectively, in the resonances as-
signed to the methyl groups on the trimethylphosphine ligand.
An attempted preparation of the [CpIr(PMe3)(CH3)(OTf)]
complex 5 in CD2Cl2 resulted only in the rapid darkening of
the solution and the production of uncharacterizable products.
One presumes that the qualitatively greater reactivity of the Cp
versus Cp*-substituted complex leads to reaction with the
solvent and/or decomposition.
asymptotic limit of dissociation, two translations and three
rotations of the two departing fragments.26 Numerical estimates
for these five frequencies were taken from computations of
analogous oxidative addition,6 reductive elimination,27 and
σ-bond metathesis28 transition states. Fortunately, the cyclic
transition states for all of these chemical processes are relatively
tight, showing for the desired modes similar frequencies. With
use of 80, 135, 540, 1285, and 2050 cm-1 for the five transition-
state frequencies,29 (anything over ∼200 cm-1 makes very little
difference anyways), the deconvoluted threshold for the 6 f 7
+ CH4 reaction, corrected to 0 K, is E0 ) 0.72 ( 0.10 eV, if
all the vibrations are treated as harmonic. If a physically more
realistic model is used in which the methyl groups, the
phosphine as a whole, and the Cp ligand are treated as internal
rotors,30 then the deconvoluted threshold drops to E0 ) 0.59 (
0.10 eV, or 13.6 ( 2 kcal/mol (2σ error bounds). The drop in
the deconvoluted threshold upon inclusion of internal rotations
in the fit is expected because, in going from 6 to any of the
conceivable transition states for methane loss, free rotation about
two of the four methyl rotors, and the phosphine as well, is
Because a small amount of reaction of 6 with CH2Cl2 could
be seen in the gas phase under certain conditions, it was thought
that use of a more inert solvent might be advisable. Accord-
ingly, treatment of the [CpIr(PMe3)(CH3)(I)] complex with
AgOTf in purified C6F6 in the presence of C6D6 led to the
disappearance of the 1H NMR signals for starting material, and
the appearance of a red precipitate that unfortunately exhibited
such low solubility in C6F6 that no 1H NMR spectrum could be
seen. When the precipitate was collected and redissolved in
acetonitrile, the electrospray mass spectrum showed [CpIr-
(PMe3)(CH3)(CH3CN)]+. (We find that acetonitrile traps any
of the 16-e- complexes and stops any further reactions.)
Evidently, ionization of the CpIr(PMe3)(CH3)(I) to the triflate
salt had occurred, but poor solubility of the complex led to
precipitation before any reaction could occur. On the other
hand, reaction of ∼7 mg of CpIr(PMe3)(CH3)(Cl) in 1.5 mL of
C6F6 with 1 equiv of Ag(O3SC8F17), instead of AgOTf, led to
the appearance of the AgCl precipitate above which stood a
supernatant solution, which was filtered and added rapidly to a
3-fold excess of CD2Cl2/C6H6 (30:1 v/v) at room temperature.
The 1H NMR spectrum showed CH4 and the appearance of three
distinct species (by both the Cp and the -PMe3 resonances)
whose relative amounts changed over a period of a day.
Periodic quenches of an aliquot with acetonitrile, followed by
electrospray mass spectrometry, also showed the formation of
three ionic complexes: the expected [CpIr(PMe3)(CH3)(CH3-
CN)]+ and [CpIr(PMe3)(C6H5)(CH3CN)]+ complexes, and a
doubly-charged dimeric species,31 ([CpIr(PMe2)(CH2)(CH3-
CN)]+)2, of as yet undetermined structure. After 12 h at room
temperature, the reaction was complete, with only the latter two
complexes remaining. When the experiment was done in the
(26) This methodology is described in: Loh, S. K.; Hales, D. A.; Lian,
L.; Armentrout, P. B. J. Chem. Phys. 1989, 90, 5467.
(27) Obara, S.; Kitaura, K.; Morokuma, K. J. Am. Chem. Soc. 1984,
106, 7482.
(28) Rappe´, A. K.; Upton, T. H. J. Am. Chem. Soc. 1992, 114, 7507.
The transition state geometries for the Cl2M(CH3)(H)(CH3) (M ) Al, Sc)
were reoptimized at the CAS(4,4)/LANL1DZ level. Frequencies at this level
were then computed and used to guide the estimates for the current system.
(29) These frequencies came from the calculation described in ref 28.
An alternative set of five frequencies taken from the calculation in ref 6
(149, 184, 488, 499, and 2243 cm-1) produced deconvoluted thresholds
that differed by much less than the stated (0.1 eV error bounds.
(30) The internal rotor energy levels were computed from the reduced
moments of inertia for rotation of the Cp ligand, the trimethylphosphine
unit as a whole, and the individual methyl groups, using the computed
equilibrium geometry of 6. See: Berry, R. S.; Rice, S. A.; Ross, J. Physical
Chemistry, 1st ed.; John Wiley: New York, 1980; p 771. Two methyl rotors,
and rotation of the phosphine unit, are lost at the transition state, which
accounts for the marked effect on the computed threshold energy.
(31) The dimeric nature of the complex is unambiguous from the isotope
peaks in the mass spectrum. Because the 1H NMR spectrum for the dimeric
complex shows a single Cp resonance with no new splittings, and two
different aliphatic peaks split by coupling to the phosphorus, both of which
disappear in the dimer derived from the complex with a deuterated
trimethylphosphine, one can place some constraints on the structure of the
complex. Moreover, it is clear from the mass alone that methane loss has
occurred. While a final judgment awaits further work, we believe the
structure has a six-membered ring analogous to that found by X-ray
crystallography for the dimer of (dmpe)2Ru, characterized by: Cotton, F.
A.; Frenz, B. A.; Hunter, D. L. J. Chem. Soc., Chem. Commun. 1974, 755.
(32) Care must be taken that the C6F6 and C5F5N solvents are freshly
distilled from CaH to remove all traces of protic acid, presumably HF. We
find the CpIr(PMe3)(CH3)(OTf) complex to be very susceptible to proto-
nation and subsequent loss of methane.
(33) An intramolecular addition would simply be the 6 f 7 reaction,
and would produce CH3D. An intermolecular addition to a C-D bond in
the deuterated trimethylphosphine ligand, done twice, would be one way
to the dimeric complex of ref 31, and would also produce CH3D.

10800 J. Am. Chem. Soc., Vol. 119, No. 44, 1997
Hinderling et al.
Scheme 4
absence of benzene, only the first and third complexes were
produced. The same chemistry is also observed when CpIr-
(PMe3)(CH3)(Cl) is treated with AgOTf in perfluoropyridine.
In that case, the AgOTf was dissolved in C5F5N prior to addition
to a solution of the iridium complex. Treatment of CpIr(PMe3-
d9)(CH3)(Cl) in an identical fashion gave the corresponding
CpIr(PMe3-d9)(CH3)(OTf) complex in C5F5N solution, which
again slowly dimerized over a period of 12 h in the absence of
hydrocarbon traps. Methane32 produced in the reaction without
added hydrocarbon was clearly shown by 1H NMR to be CH3D,
although it is not clear whether the deuterium incorporation came
from an intra- or intermolecular addition to the deuterated
trimethylphosphine ligand.33 Lastly, we find that, with C6D6
present in the C5F5N solution as the complex is prepared, CpIr-
(PMe3)(CH3)(OTf) reacts with the evolution of both CH4 and
CH3D, although again, the presence of significant amounts of
the dimeric complex makes it difficult to identify the reaction
producing CH4. The electrospray mass spectrum of the
[CpIr(PMe3)(Ph)(CH3CN)]+ product in the reaction with C6D6,
after acetonitrile quench, showed five deuteriums rather than
six, making it clear that if 6 and 7 were both present, then all
of the 7 would have had to have ended up in the dimer rather
than in the adduct with benzene.
connecting the various energy minima is beyond the scope of
the present report and will be the subject of future investigations.
Looking at the DFT results, one sees that the relative energies
of 6 and 7 are not too well described, with the cyclometalated
complex 7 (plus CH4) predicted to lie 26 kcal/mol above the
open structure 6. As a comparison, the CID threshold experi-
ment places an upper-bound on the energy difference of 13.6
kcal/mol. Three stationary points were found with an Ir(V)
structure 9 containing the same three-membered ring, as well
as Ir-CH3 and Ir-H bonds, designated 9 with cis-anti, cis-
syn, and trans configuration of the Ir-CH3 and Ir-H bonds
(Scheme 5). Although the two cis structures are almost identical
in energy (within 1 kcal/mol), with the trans structure 24 kcal/
mol higher in energy, only the cis-anti (on geometrical grounds)
structure can be the first-formed intermediate (or transition-state)
The geometries and relative energies of several of the
important organometallic structures, computed by the ab initio
methods described above, are shown in Figures 8-10. For all
three structures, the structures were verified to be minima by
computation of the frequencies, which are tabulated in the
Supporting Information along with the Cartesian coordinates
of each atom. If one draws a “bond” from the iridium atom in
6 or 7 to the center of the Cp ring, then the sum of C-Ir-P,
Cp-Ir-P, and Cp-Ir-C angles is a measure of the “planarity”
at the iridium center. In Figure 8, the view of 6 is chosen to
emphasize the nearly “planar” arrangement of ligands around
the metal (sum of angles ∼358°). Similarly, the view of 7 in
Figure 9 shows that the metal is now “pyramidalized” (sum of
angles ∼339°). While a definitive judgment will require a more
thorough computational study, one can suspect that the differ-
ences in reactivity between 6 and 7 could stem from a
combination of better steric access to the metal in 7 and
electronic differences imposed by the different geometry about
the iridium center. A full exploration of the hypersurface
Figure 9. Computed (DFT with B3LYP functional and LANL2DZ
basis set) structure of 7 viewed along the plane formed by the C-Ir-P
atoms (Ir-C bond in front) to emphasize the near “pyramidalization”
at the central iridium atom. The hydrogen atoms are not shown.
Important bond lengths: Ir-Cp (center of ring) 1.93 Å, Ir-C 2.10 Å,
Ir-P 2.41 Å, P-C 1.88 and 1.87 Å (shorter bond on “concave” side).
Important bond angles: C-Ir-P 47.56°, Cp-Ir-P 147.06°, Cp-Ir-C
144.25°. Note the sum of the bond angles about the central iridium
atom is 338.87°. Total energy: -423.56517 hartrees for 7 and
-40.51448 hartrees for methane.
Figure 10. Computed (DFT with B3LYP functional and LANL2DZ
basis set) structure of cis-anti-9, the putative product of an intramo-
lecular oxidative addition by 6, viewed along the plane formed by the
C(ring)-Ir-P atoms (Ir-C bond in front) to show the relationship of
the hydrido and methyl ligands to the ring. Only the Ir-H hydrogen
atom is shown. Important bond lengths: Ir-Cp (center of ring) 2.02
Å, Ir-C(ring) 2.18 Å, Ir-C(methyl) 2.15 Å, Ir-P 2.41 Å, P-C 1.87
Å (for both methyls), Ir-H 1.57 Å. Important bond angles: C(ring)-
Ir-P 46.16°, Cp-Ir-P 136.20°, Cp-Ir-C(ring) 123.69°, Cp-Ir-
C(methyl) 116.41°, Cp-Ir-H 123.85°, C(ring)-Ir-C(methyl) 119.70°,
C(ring)-Ir-H 77.32°, P-Ir-H 97.29°, P-Ir-C(methyl) 86.77°,
C(methyl)-Ir-H 73.95°. Note the two important distances C(ring)-H
and C(methyl)-H are 2.39 and 2.29 Å, respectively. Total energy:
-464.09308 hartrees for cis-anti-9. cis-syn- and trans-9 lie at -464.09426
and -464.05455 hartrees, respectively.
Figure 8. Computed (DFT with B3LYP functional and LANL2DZ
basis set) structure of 6 viewed along the plane formed by the C-Ir-P
atoms (Ir-C bond in front) to emphasize the near “planarity” at the
central iridium atom. The hydrogen atoms are not shown. Important
bond lengths: Ir-Cp (center of ring) 1.95 Å, Ir-C 2.05 Å, Ir-P 2.41
Å, P-C 1.88 Å (for all three methyls). Important bond angles: C-Ir-P
89.96°, Cp-Ir-P 134.15°, Cp-Ir-C 133.54°. Note the sum of the
bond angles about the central iridium atom is 357.65°. Total energy:
-464.12127 hartrees.

C-H ActiVation by Cationic Iridium(III) Complexes
J. Am. Chem. Soc., Vol. 119, No. 44, 1997 10801
Scheme 5
Scheme 6
for a hypothetical reaction from 6 to 7 via a discrete Ir(V)
species. The DFT calculations place cis-anti-9 approximately
18 kcal/mol above 6. While the numbers, taken literally, place
cis-anti-9 below (7 + CH4), the disagreement with experiment
(see below) on the 6 versus (7 + CH4) energy difference,
combined with the well-known problems in computing bond
energies (BSSE, size-consistency, etc.), means that one can only
say with certainty that both 7 and cis-anti-9 are energetically
accessible structures that could plausibly be important in the
C-H activation chemistry of 6.
literature, an alternative structure, an isomeric L3Ir(PR2)(CH3)
phosphido complex, has been shown to rearrange irreversibly
to L3Ir(H)(η2-CH2PMe3),43 which indicates at least some relative
stability. Furthermore, a reversible intramolecular C-H inser-
tion reaction on the trialkylphosphine ligand has also been
reported for several complexes of iron, nickel, manganese,
rhenium, ruthenium, and osmium,45 with the equilibrium lying
toward the cyclometalated product. The cyclometalation reac-
tions of 2 and 6 differ from these primarily in the concomitant
loss of methane, removing any possibility of reversibility. The
observed threshold for methane loss from 6 of 13.6 kcal/mol is
also consistent with the structural assignment. The facile
addition of 7 to alkanes and arenes indicates that the reactions
of 7 occur over a rather low barrier. Accordingly, one would
expect the heat of formation of (7 + CH4) to be something like
10 kcal/mol above that of 6. The principal difference between
the two complexes is just ring strain, with a small contribution
from the difference between the C-H bond strength in methane
Discussion
At the nonquantitative level, the mass spectrometric observa-
tion of organometallic complexes, including reactive intermedi-
ates, is straightforward,34-37 and qualitative observation of
collision-induced reductive elimination from electrosprayed Pd-
(IV) complexes, for example, has even been reported.38 The
range of organometallic systems for which electrospray mass
spectrometry can been done suggests broad applicability as both
an analytical and mechanistic tool. While there have been CID
threshold measurements on other organometallic species,39,40 the
present work represents the first application of the method to
organometallic complexes that appear as reagents or catalysts
in solution-phase chemistry.
and that in the methyl group on the phosphine ligand.
A
The gas-phase reactions of the cationic Ir(III) complexes, as
observed in the mass spectra, clearly point to a previously
unreported mechanism for their observed σ-bond metathesis
reactions. Previous discussions had considered a two-step
mechanism involving intermolecular oxidative addition of either
2 or 6 to the C-H bond of an alkane or arene producing an
Ir(V) intermediate, followed by reductive elimination of meth-
ane, or a concerted σ-bond metathesis reaction similar to that
seen in early transition metals. The electrospray mass spectra
show another mechanism involving initial elimination of
methane, followed by addition to an alkane or arene, with all
detected reactive intermediates retaining a formal +3 oxidation
state.
Structural Assignment. The structures of the intermediate
species, 3 and 7, are assigned on the basis of literature precedent,
ab initio calculations, and the present experimental observations.
The masses of 3 and 7, as well as isotopic labeling experiments,
strongly suggest a cyclic structure with cyclometalation occur-
ring on the phosphine and not the Cp or Cp* ligand. Numerous
examples of transition metal complexes with the η2-CH2PMe3
ligand have been reported and characterized by X-ray crystallog-
raphy.41-44 For at least one of the iridium complexes from the
difference of ∼10 kcal/mol is just about what one would expect.
An alternative structural possibility for 7 was considered, and
then excluded by the experimental evidence. A methylidene
complex bearing a phosphido ligand, isomeric with 3 or 7,
differs from the cyclometalated structures primarily in the P-C
distance. A number of methylidene complexes of iridium have
been reported.46 Furthermore, precedent for conversion of C-H
activation products to the Fischer carbene complexes has been
found for cationic Ir(III)47 and Pt(II)7 complexes. One can
therefore envision a case where the carbene complex is formed
from 3 or 7. One can even consider the possibility that the
carbene complex lies on the reaction coordinate from 2 or 6 as
the first-formed reactive intermediate. The latter possibility can
be excluded in the gas-phase reaction, 2 f 3 + CH4. Entries
21 and 22 in Table 1 clearly indicate that [Cp*Ir(PMe3-d9)-
(42) Al-Jibori, S.; Crocker, C.; McDonald, W. S.; Shaw, B. L. J. Chem.
Soc., Dalton Trans. 1981, 1572.
(43) Fryzuk, M. D.; Joshi, K. J. Am. Chem. Soc. 1989, 111, 4498. Fryzuk,
M. D.; Joshi, K.; Chadha, R. K.; Rettig, S. J. J. Am. Chem. Soc. 1991, 113,
8724.
(44) Hester, D. M.; Yang, G. K. Organometallics 1991, 10, 369.
(45) Karsch, H. H.; Klein, H.-F.; Schmidbauer, H. Angew. Chem. 1975,
87, 630. Rathke, J. W.; Muertterties, E. L. J. Am. Chem. Soc. 1975, 97,
3272. Klein, H.-F. Angew. Chem., Int. Ed. Engl. 1980, 19, 362. Morvillo,
A.; Turco, A. J. Organomet. Chem. 1981, 208, 103. Werner, H.; Werner,
R. J. Organomet. Chem. 1981, 209, C60. Lindner, E.; Starz, K. A.; Eberle,
H. J.; Hiller, W. Chem. Ber. 1983, 116, 1209. Lindner, E.; Ku¨ster, E. U.;
Hiller, W.; Fawzi, R. Chem. Ber. 1984, 117, 127. Karsch, H. H. Chem.
Ber. 1984, 117, 3123. Bergman, R. G.; Seidler, P. F.; Wenzel, T. T. J. Am.
Chem. Soc. 1985, 107, 4358. Wenzel, T. T.; Bergman, R. G. J. Am. Chem.
Soc. 1986, 108, 4856. Hartwig, J. F.; Andersen, R. A.; Bergman, R. G. J.
Am. Chem. Soc. 1991, 113, 6492.
(34) Colton, R.; Traeger, J. C. Inorg. Chim. Acta 1992, 201, 153. Colton,
R.; D’Agostino, A.; Traeger, J. C. Mass Spectrom. ReV. 1996, 14, 79.
(35) Kane-Maguire, L. A. P.; Kanitz, R.; Sheil, M. M. J. Organomet.
Chem. 1995, 486, 243.
(36) Lipshutz, B. H.; Stevens, K. L.; James, B.; Pavlovich, J. G.; Snyder,
J. P. J. Am. Chem. Soc. 1996, 118, 6796.
(37) Spence, T. G.; Burns, T. D.; Posey, L. A. J. Phys. Chem. A 1997,
101, 139.
(38) Canty, A. J.; Traill, P. R.; Colton, R.; Thomas, I. M. Inorg. Chim.
Acta 1993, 210, 91.
(39) Freiser, B. S. Organometallic Ion Chemistry; Kluwer: Dordrecht,
1996.
(40) Armentrout, P. B. Acc. Chem. Res. 1995, 28, 430.
(41) Lindner, E.; Neese, P.; Hiller, W.; Fawzi, R. Organometallics 1986,
5, 2030.
(46) Clark, G. R.; Roper, W. R.; Wright, A. H. J. Organomet. Chem.
1984, 273, C17. Fryzuk, M. D.; McNeil, P. A.; Rettig, S. J. J. Am. Chem.
Soc. 1985, 107, 6708. Klein, D. P.; Bergman, R. G. J. Am. Chem. Soc.
1989, 111, 3079. Fryzuk, M. D.; Gao, X.; Joshi, K.; McNeil, P. A.; Massey,
R. L. J. Am. Chem. Soc. 1993, 115, 10581.
(47) Lu¨cke, H. F.; Arndtsen, B. A.; Burger, P.; Bergman, R. G. J. Am.
Chem. Soc. 1996, 118, 2517.

10802 J. Am. Chem. Soc., Vol. 119, No. 44, 1997
Hinderling et al.
Scheme 7
(CH3)]+ loses CH3D upon collisional activation, indicating that
the carbon leaving as methane came from the methyl group
directly attached to the metal. If the carbene complex were to
be the first-formed intermediate, one would expect the departing
carbon to come from one of the methyls in the trimethylphos-
phine ligand, giving loss of CHD3 instead. The characteristic
C-H activation chemistry we report is moreover inconsistent
with the reactions reported for the known methylidene com-
plexes,46 meaning, that if there were to be interconversion
between 3 or 7 and the corresponding carbene complexes, it
would be a nonproductive equilibrium lying off of the main
reaction pathway.
distance for a charge-dipole interaction and 1/r4 for a charge-
induced dipole interaction, and amounts to a potential well in
which the reacting partners of a bimolecular ion-molecule
reaction are trapped prior to reaction (Scheme 7).48 In the
absence of subsequent thermalizing collisions, the charge-dipole
or charge-induced dipole interaction can provide the energy
needed to surmount an activation barrier for the trapped reactants
as long as the transition state is not substantially higher in energy
than the separated reactants. Neglecting for the moment any
other shallow minima that may lie on the reaction coordinate,
the potential energy diagram above rationalizes the very high
reactivity in the reaction of 7 with benzene, as well as the
extensive isotopic exchange between the trimethylphosphine and
the hydrocarbon ligand. By microscopic reversibility, CID of
8 (R ) Ph) to produce 7 and benzene should go through the
same charge-induced dipole complex. As long as the transition
state for the conversion of the complex to the adduct lies below
the asymptotic limit for separate molecules, the first step of the
dissociation should be reversible, leading to the extensive
isotopic scrambling. Going to the Cp* series, the potential
energy diagram is drawn with the same energy difference
between the two minima, and the same activation barrier
between them (for argument’s sake), but a shallower well for
the complex relative to the separated molecules. One would
justify this change by invoking both the increased steric bulk
of the Cp* ligand and the greater degree of charge-delocalization
in 3 (relative to 7) as factors making the charge-induced dipole
interaction less stabilizing. Having lowered the asymptote
relative to the two minima and the transition state for reaction,
a small activation energy now appears for the bimolecular gas-
phase reaction. It may or may not be possible to measure the
resulting barrier to the 3 + benzene addition reaction if it is
only few kilocalories per mole, but importantly, the reverse
reaction, CID of 4 (R ) Ph), would now proceed without
reversiblity in the first step of the dissociation, limiting the
isotopic exchange to zero or one label.
The experiment listed as entry 12 in Table 1 was done to
check the structure of 6. One could have imagined that 9, with
the same mass as 6, could be the species actually prepared by
electrospray ionization of 5. While not yet definitive, the
experiment is strongly suggestive that the ion is the 16-electron
complex 6 rather than the 18-electron complex 9. Entry 12
corresponds to a daughter-ion experiment in which thermalized
ions with the mass corresponding to 6 are selected, and collided
with acetonitrile or CO in O2 at very low collision energies. In
this instance, the pressure of acetonitrile vapor or CO in O2
was set to correspond to approximately 100 collisions by the
average ion before it exits into the mass analyzer. Nearly all
molecules of 6 pick up an additional acetonitrile or CO ligand
under these conditions. The observation of a very minor mass
peak due to 7 indicates that there is sufficient energy in the
collisions to induce a small amount of cleavage of CH4. A
larger (but still smaller than the peaks due to 6 and (6 +
acetonitrile)) peak due to (7 + acetonitrile) indicates that 7 also
picks up a 2-electron ligand. Not all molecules of the
unambiguously 16-electron complex 7 pick up an acetonitrile
or CO despite multiple collisions because, presumably, the
reactive cross section in the collision is smaller than the hard-
sphere cross section. Taken altogether, these observations
strongly suggest that the ion with the mass of 6 is indeed the
16-electron structure 6 and not the 18-electron structure 9. A
control experiment whereby the unambiguously 18-electron
complex 5 (L ) CH3CN) is tested under the same conditions
shows no pickup of an additional ligand. This rules out the
formation of electrostatically bound ion-dipole complexes as
the explanation for the acetonitrile or CO “trapping” results
above.
One should note that the reaction of 7 with CH4 (daughter-
ion mode, mass-selected 7, collision in O2, not reported in Table
1 because the extent of reaction was very small or zero) must
undoubtedly follow a potential energy diagram similar to that
shown in Scheme 8 for 3 and benzene, i.e. the transition state
for the interconversion between the two minima must lie above
the separated molecules. Presumably, the small size and
consequent low polarizability of CH4 leads to an even less
stabilizing charge-induced dipole interaction. Accordingly, even
complex 7 reacts poorly, if at all, with methane. The CID
Comparison of the Cp and Cp* Series of Complexes. The
difference in the extent of isotopic exchange between the
trimethylphosphine and the phenyl ligand upon CID of 4 versus
8 (R ) Ph) very likely comes from a general feature of ion-
molecule reactions. In the gas phase, ion-molecule reactions
proceed through a charge-dipole or charge-induced dipole
complex bound by anywhere from a few to several kilocalories
per mole relative to the separated molecules. The non-negligible
attractive potential is relatively long ranged, i.e. 1/r2 falloff with
(48) Farneth, W. E.; Brauman, J. I. J. Am. Chem. Soc. 1976, 98, 7891.
Olmstead, W. N.; Brauman, J. I. J. Am. Chem. Soc. 1977, 99, 4219.
Asubiojo, O. I.; Brauman, J. I. J. Am. Chem. Soc. 1979, 101, 3715. Pellerite,
M. J.; Brauman, J. I. J. Am. Chem. Soc. 1980, 102, 5993. Sharma, S.;
Kebarle, P. J. Am. Chem. Soc. 1982, 104, 19. Caldwell, G.; Magnera, T.
F.; Kebarle, P. J. Am. Chem. Soc. 1984, 106, 959. McMahon, T. B.; Heinis,
T.; Nicol, G.; Hovey, J. K.; Kebarle, P. J. Am. Chem. Soc. 1988, 110, 7951.

C-H ActiVation by Cationic Iridium(III) Complexes
J. Am. Chem. Soc., Vol. 119, No. 44, 1997 10803
Scheme 8
threshold for the reverse reaction, loss of methane from 6, is
therefore interpreted as a ∆Gq rather than a ∆G for the reaction.
Similarly, one can postulate that the difference between 3 and
7 in their reactions with cyclohexane may stem from this same
model. The extra methyl groups in 3 may force the intermediate
complex in the addition reaction of 3 and cyclohexane to adopt
a geometry where the two molecules are farther apart on average
than they are in the complex of 7 and cyclohexane. The
potential surfaces for addition of 7 and 3 to cyclohexane would
show the same differences as between Scheme 7 and Scheme
8 except that, in the case of reaction with cyclohexane, the
activation barrier in Scheme 8 would be high enough to stop
reaction at all accessible collision energies. In short, benzene
is highly reactive with the cationic metallaphosphacyclopropanes
because, with its large π-system, it is highly polarizable.
Pentane reacts because the 1° C-H bonds at its termini can
approach even a sterically hindered ion. Cyclohexane, with its
2° C-H bonds, cannot approach as closely, so the reaction is
worse. Lastly, methane, although it is small, is the least
polarizable of the aliphatic hydrocarbons, and therefore forms
only weakly-bound charge-induced dipole complexes, giving
the lowest reactivity.
which also suggest that the mechanism we observe in the gas
phase does not occur to a significant extent in solution for 1.
We can postulate an explanation for the difference between
the gas-phase and solution-phase results. A concerted σ-bond
metathesis was originally offered as an alternative to the more
conventional oxidative addition/reductive elimination mecha-
nism for the observed C-H activation chemistry of the Ir(III)
complexes because it was thought that an electron-deficient Ir-
(III) might find it unfavorable to proceed via an even more
electron-deficient Ir(V) intermediate.2 A subsequent search for
an accessible transition state for the concerted mechanism by
ab initio computational techniques was unsuccessful, leading
to the conclusion that the reaction proceeded by addition-
elimination after all.6 The mechanism found in the present gas-
phase work was not considered in that study, but nevertheless
could have provided yet another alternative to the conventional
oxidative addition/reductive elimination mechanism that could
again avoid the Ir(V) intermediate if the methane loss and C-H
activation steps each were to proceed concertedly with the
formation or cleavage of the three-membered ring. If one
presumes that both an addition-elimination and an elimination-
addition mechanism were to be plausible, and, furthermore, that
they proceed through transition states that are not too different
in energy, then electron donation by solvent or simple electro-
static stabilization of charged species in a dielectric continuum
could tip the balance in favor of a route through an Ir(V)
intermediate. External stabilization of an electron-deficient
species is absent, of course, in the gas-phase experiments. Our
preliminary solution-phase results for CpIr(PMe3)(CH3)(OTf)
in C5F5N solution indicate that the more electron-deficient
complexes undergo some different chemistry. While we could
exclude isotope exchange between the phosphine ligand and
the incoming hydrocarbon in reactions of CpIr(PMe3)(CH3)-
(OTf) with C6D6, the observation of CH4 evolution associated
with dimer formation indicates a mechanism involving an
addition to the C-H bond on the trimethylphosphine. What is
as yet unclear is whether the reaction is intra- or intermolecular.
In the former case, intramolecular insertion would produce 7,
which presumably then dimerizes or reacts with another
molecule of uncyclized complex. In the latter case, one could
have an ordinary oxidative addition/reductive elimination
sequence, done twice, which would give the same CH4 as well
as the same dimeric complex. The experiments to date do not
distinguish between the two possibilities, although it would be
odd for an intermolecular addition to occur to the exclusion of
the intramolecular version. While a definitive conclusion will
require much further work, the preliminary results suggest that
the more electron deficient Cp-substituted complexes may be
reacting in solution, at least in part, by the same elimination-
addition mechanism seen in the gas phase. The proposition can
be tested in two ways. By rendering the complexes even more
One last comment derived from the qualitative potential
surfaces needs to be made. In solution, the dielectric screening
of electrostatic interactions reduces the charge-dipole or charge-
induced dipole interaction between the two reactant molecules
to near zero until the reacting species are practically in contact.
The effect on the qualitative potential surface would be to pull
the asymptote representing separated molecules down relative
to the electrostatic complex until there is only a very shallow
minimum (which presumably is the physical basis for a cage
effect in solution-phase reactions). Accordingly, isotopic
exchange between the trimethylphosphine and a phenyl ligand
in both the Cp and Cp* series of compounds should involve no
more than one labeled atom per event, when the reaction is
performed in solution. Of course, multiple exchange would still
be possible if a reversible solution-phase reaction were to occur
repeatedly, but that is a trivial case.
Comparison of the Gas-Phase to Solution-Phase Results.
The results are unambiguous that, in the gas-phase, loss of
methane precedes reaction with C-H bonds for both the Cp
and Cp*-substituted complexes. The solution-phase results for
[Cp*Ir(PMe3)(CH3)(OTf)] are equally compelling in that the
absence of isotopic exchange between the trimethylphoshine
ligand and the incoming hydrocarbon rules out the participation
of an intermediate of structure 3. We have also been recently
apprised of unpublished results from the Bergman group,49
(49) The reaction of 1 and benzene or deuteriobenzene in solution shows
a primary deuterium kinetic of kH/kD ∼ 4, which can be interpreted to mean
that C-H cleavage on the arene occurs in the rate-determining step. R. G.
Bergman, private communication.

10804 J. Am. Chem. Soc., Vol. 119, No. 44, 1997
Hinderling et al.
Scheme 9
in an experiment designed to produce it (if it were an
intermediate), which means that, in accordance with the
prediction from theory, one would have to place cis-anti-9
energetically above 6 by more than a few kilocalories per mole.
Meanwhile, we also have the experimentally determined activa-
tion energy for the loss of CH4 from 6 at only 13 kcal which
would be a rigorous upper-bound on the amount above 6 that
cis-anti-9 could lie if it were on the reaction coordinate. One
can be reasonably certain in concluding, therefore, that if cis-
anti-9 were to be an intermediate in the reaction 6 f 7 + CH4,
it would have to reside in, at best, a shallow minimum no more
than a few kilocalories per mole deep.
Conclusions
Electrospray ionization tandem mass spectrometry, in con-
junction with solution-phase experiments and ab initio calcula-
tions, is a new approach to the study of organometallic reactions
and reactive intermediates. Both the similarities and the
differences between the observed gas-phase and solution-phase
chemistry hold information useful to the dissection of complex
reaction pathways. In the particular case of C-H activation
by cationic Ir(III) complexes, a novel mechanism involving
metallaphosphacyclopropane intermediates has been found. The
balance favoring this mechanism over competing mechanisms
not involving cyclometalation depends on the presence or
absence of solvent, and may be adjustable by perturbations in
structure or surroundings. The multiplicity of competing
mechanisms and energetically similar structural types in orga-
nometallic chemistry has often been blamed for the poor state
of mechanistic understanding in the field, but with the introduc-
tion of new, powerful physical techniques, this problem can
potentially become an asset. We see that changes in mechanism
can be induced by small changes in the molecule or its
surroundings, which if properly understood can then be used
to fine-tune reactivity rationally. This possibility will be tested
in subsequent investigations.
electron deficient through use of functionalized phosphines, the
oxidative addition through an Ir(V) intermediate could be
increasingly disfavored, allowing competing mechanisms to
appear. A more direct approach is the independent syntheses
of 3 and 7, and an investigation of their reactivity in the gas
phase and solution. Both experiments are underway.
Mechanism of Methane Loss. Presumably, the preference
for 2 and 6 to react in the gas phase via the elimination-addition
mechanism comes from the aforementioned reluctance of the
formal Ir(III) center to go to the requisite Ir(V) of an oxidative
addition product. The claim really only pushes the question
back one step, though. In going from 2 to 3, or 6 to 7, one can
again imagine a two-step mechanism involving intramolecular
oxidative addition through an intermediate cis-anti-9, followed
by reductive elimination of CH4, or a concerted mechanism
through a four-center transition state. The absence of any C-H
activation chemistry by the CpFe(CO)(η2-CH2PEt2) complex44
(an 18-e- complex) could be interpreted to mean that the metal
center is involved in an addition reaction. Nevertheless, the
present experimental results do not distinguish decisively
between the two possibilities shown in Scheme 9. The ab initio
calculations do find cis-anti-9 to be a minimum. Of special
interest are the H-Ir, H-CH2, and H-CH3 distances in that
structure, which are computed to be 1.57, 2.39, and 2.29 Å.
Comparing these to the H-Ir and H-CH3 distances in the
computed structure of the [CpIrV(PH3)(CH3)2(H)]+ intermediate
by Hall and co-workers,6 one sees that the present numbers more
closely resemble the 1.567 and 2.194 Å for a product of an
oxidative addition than the 1.611 and 1.535 Å for a transition
state. Computed to lie 18 kcal/mol above 6, cis-anti-9 is
predicted to be lower in energy than the sum of the computed
energies for 7 and methane. However, given the basis set
superposition error, size consistency, etc. problems at this level
of theory, and given that the calculations are only single points
and not a complete surface, one still cannot judge with certainty
whether or not cis-anti-9 is on the reaction coordinate from 6
to 7. On thermochemical grounds, however, one can place real
limits on cis-anti-9, if it were to be an intermediate in the
reaction. The structural arguments cited earlier can be inter-
preted to mean that we see no evidence for an Ir(V) structure
Acknowledgment. We would like to acknowledge support
for this work by the Eidgeno¨ssische Technische Hochschule
Zu¨rich, the Schweizerischer Nationalfonds zur Fo¨rderung der
wissenschaftlichen Forschung, and a graduate fellowship for
C.H. from the Stipendienfonds der Basler Chemischen Industrie.
We thank Prof. P. B. Armentrout for generously providing the
CRUNCH program, and for helpful discussions on CID
threshold determinations. We also thank Dr. D. L. Strout and
Prof. M. B. Hall for kindly supplying additional details from
their computation on oxidative additions by Ir(III) complexes.
Lastly, we acknowledge discussions with Prof. R. G. Bergman
on reactivity issues in the Ir(III) system and the interpretation
of results, including unpublished data from his laboratory.
Supporting Information Available: Geometries, absolute
energies, and the frequencies for 6, 7, and 9 from the ab initio
calculations with the HF and DFT methods, as well as the
synthetic details and physical characterization of the complexes
in this study (10 pages). See any current masthead page for
ordering and Internet access instructions.
JA970995U