Unifying the mechanisms for alkane dehydrogenation and alkene H/D exchange with [IrH2(O2CCF3)(PAr3)2]: The key role of CF3CO2 in the "sticky" alkane route

Article abstract of DOI:10.1039/b101715m

To understand photochemical and thermal alkane activation with IrH₂(O₂CCF₃)(PAr₃)₂ (Ar = p-FC₆H₄), H/D isotope scrambling between alkenes and IrD₂(O₂CCF

Full text of DOI:10.1039/b101715m

View Article Online / Journal Homepage / Table of Contents for this issue
Unifying the mechanisms for alkane dehydrogenation and alkene
H/D exchange with [IrH (O CCF )(PAr ) ] : the key role of
2 2 3 3 2
CF CO in the ““stickyÏÏ alkane route
3
2
Helene Gerard,a Odile Eisenstein,*a Dong-Heon Lee,¤b Junyi Chenb and Robert H. Crabtree*b
a L aboratoire de Structure et Dynamique des Systemes Moleculaires et Solides (CNRS UMR
5636), Universite Montpellier 2, 34095 Montpellier cedex 05, France.
E-mail: odile.eisenstein=lsd.univ-montp2.fr
b Department of Chemistry, Y ale University, 225 Prospect Street, New Haven, CT 06511-8107,
USA. E-mail: robert.crabtree=yale.edu
Received (in Montpellier, France) 21st February 2001, Accepted 30th May 2001
First published as an Advance Article on the web 21st August 2001
To understand photochemical and thermal alkane activation with IrH (O CCF )(PAr ) (Ar \ p-FC H ), H/D
2
2
3
3 2
6 4
isotope scrambling between alkenes and IrD (O CCF )(PAr ) was studied. No unique interpretation of the
2
2
3
3 2
experimental data was possible, so DFT(B3PW91) calculations on the exchange process in
Ir(H) (O CCF )(PH ) (C H ) were carried out to distinguish between the possibilities allowed by experiment. Of
2
2
3
3 2 2 4
several possible mechanisms for H/D scrambling, one was strongly preferred and is therefore proposed here. It
involves the insertion of the oleÐn to give an alkyl hydride that reductively eliminates to lead to a transition state
that contains an g3-bound alkane. This transition state, which achieves a 1,1@ geminal H/D exchange, is
signiÐcantly lower in energy than a dihydrido carbene, located as a secondary minimum, eliminating the alternative
carbene mechanism. The unexpectedly large binding energy (BDE) of the alkane (““sticky alkaneÏÏ) to the
Ir(O CCF )(PH ) fragment (BDE \ 11.9 kcal mol~1) in this transition state is ascribed in part to the presence of a
2
3
3 2
weakly r- and p-donating (CF CO ) group trans to the alkane binding site. The H/D exchange selectivity observed
3
2
requires that 1,1@-shifts (i.e., M moving to a geminal CÈH bond), but not 1,3-shifts, be allowed in the alkane
complex. In a key Ðnding, a 1,3-shift in which the metal moves down the alkane chain is indeed found to have a
much higher activation energy than the 1,1@-process and is therefore slow in our system. A 1,2-shift has not been
considered since it would involve a strong steric hindrance at a tertiary carbon in this system. The mechanism via
an alkane path provides an insight into the closely related photochemical and catalytic thermal alkane
dehydrogenation processes mediated by IrH (O CCF )(PAr ) ; the thermal route requires tBuCH2CH as the
hydrogen acceptor. These two alkane reactions are intimately related mechanistically to the isotope exchange
because they are proposed to have the same intermediates, in particular the sticky alkane complex. Remarkably,
the rate determining step of the thermal (150 ¡C) alkane dehydrogenation process is predicted to be substitution of
the hydrogen acceptor-derived alkane by the alkane substrate.
2
2
3
3 2
2
Alkane complexes have remained elusive and hard to charac-
terize since alkanes are weak ligands for a transition metal
fragment. Such complexes were Ðrst inferred from detailed
studies of CÈH activation processes and related hydrogen
isotope scrambling experiments.1,2 In very early work, alkane
by IR methods,11 an X-ray determination12 and NMR mea-
surements.13,14
These difficulties in characterization mean that theoretical
studies can provide complementary information for identiÐca-
tion of reaction paths involving alkane complexes. In several
such studies an alkane adduct was proposed as a starting
reactant for further chemistry. Several oxidative additions are
proposed to start from a weakly bound alkane complex.15h17
Such an alkane complex (a secondary minimum 3 kcal
mol~1 above the separated reactants) is found for alkane
complexes were observed at low temperatures for M(CO)
5
(M \ Cr, Mo, W) fragments in matrices, solution and in the
gas phase.3h5 In spite of the difficulties, some understanding of
the behavior of alkane complexes has been acquired.5h7
Recent Ñash kinetic studies were carried out on an alkane
complex of rhodium.8 Several proposed organometallic
mechanisms require the intermediacy of an alkane
complex.4,9,10 For example, alkane complexes were suggested
in several CÈH activation reactions,9 as well as in the substi-
dehydrogenation catalyzed by (PCP@)Ir(H)
[PCP@ \
2
3-C H (CH PH ) -1,3].18 Weak alkane binding was also cal-
6
3
2
2 2
culated for W(CO) .19 Stronger alkane binding (7È10 kcal
5
mol~1 binding energies) was calculated for [CpM(PH )(R)]`
(Cp \ g5-C H ; M \ Rh, Ir; R \ H, CH ).18,20 Because of
3
tution of H in an Ir(III) complex10a and in the reductive elimi-
2
5
5
3
nation
of
Tp@RhL(R)(H)
(Tp@ \ tris-3,5-
the weak binding, proposals involving alkane intermediates
have been avoided if an alternative was available. Computa-
tional chemistry now permits a reliable comparison of di†er-
ent reaction paths via combined theoreticalÈexperimental
studies like the one reported here.
dimethylpyrazolylborate, L \ CNCH CMe ).9c The culmi-
nation of this work has been the recent discovery that some
metal fragments can form stable alkane complexes as shown
2
3
Although many examples of alkane oxidative addition are
known,21 alkane functionalization reactions are still rare.22h25
¤ Present address: Department of Chemistry, School of Natural Sci-
ences, Chonbuk National University, Chonju, 561-756, Korea.
DOI: 10.1039/b101715m
New J. Chem., 2001, 25, 1121È1131
1121
This journal is ( The Royal Society of Chemistry and the Centre National de la Recherche ScientiÐque 2001

View Article Online
Alkane dehydrogenation catalysis is the longest-known case
nisms.33a We also Ðnd faster 1,1@- than 1,3-migration in an
alkane complex. A key transition state is found to have an
alkane bonded via two CÈH bonds (g3) to the metal center.
This leads us to propose a uniÐed mechanism that accounts
both for the isotope exchange in alkenes and for the photo-
chemical or thermal dehydrogenation of alkanes by
[IrH (O CCF )(PAr ) ].
and we have shown that [IrH (O CCF )(PAr ) ] (Ar \ p-
2
2
3
3 2
FC H ) can dehydrogenate alkanes photochemically [eqn.
6
4
(1)], or thermally [eqn. (2)] in the presence of a hydrogen
acceptor such as tBuCH2CH .26,27 Isotope exchange has
2
given important information about transient intermediates in
organometallic reactions. For example, in [L M(D)(CH )]
n
3
2
2
3
3 2
(L M \ Cp*(PMe )Ir, Cp W, Cp Re`), isotope exchange
n
3
2
2
occurs to give L M(H)(CH D).9a,28,29 The intermediacy of
alkane complexes of the type L M(CH D) was proposed,
n
2
Results and discussion
n
3
although
the
intermediacy
of
carbene
complexes
L M(2CH )(H)(D) or L M(2CH )(HÈD) could have given the
same outcome. Carbene formation would have required Cp
slippage to avoid exceeding 18 valence electrons and this
Choice of experimental and theoretical systems
n
2
n
2
Choice of system and synthesis. The Ir(III) complex,
[IrH (j2-O CCF )(PAr ) ] (1, Ar \ p-FC H ) has previously
possibility was therefore rejected without comment.
H
exchange in Cp*Os(dmpm)(CH )(H)` could also be explained
2
2
3
3 2
6 4
3
been shown to react with alkanes and alkenes.26 The d2-1 iso-
topomer is best prepared by reaction of [(cod)Ir-
(PAr ) ](O CCF ) (cod \ 1,5-cyclooctadiene) with D for
by the intermediacy of a carbene or alkane complex30 but
DFT calculations suggested that the latter was more likely.31
Carbene intermediates explain preferential 1,1-polydeuteration
of alkanes by Pt(II)/DOAc in Shilov32 chemistry.
3 2
2
3
2
20 min, but it is also formed from [IrD (PAr ) -
2
3 2
(O2CMe ) ]BF (2) and NaO CCF .
2 2
4
2
3
Complex d2-1 showed no proton NMR signals in the high-
Ðeld hydride region but a strong D signal appears at [30.1
ppm in the 2H NMR spectrum. At incomplete levels of deu-
teriation, the 31P NMR resonances of all three species, d0-1,
d1-1 and d2-1, could be distinguished at 22.12, 22.24 and 22.37
ppm as a result of the isotope shifts. The bidentate acetate
ligand in 1 is hemilabile, allowing addition of external ligands
to give [IrH (j1-O CCF )(L)(PAr ) ] (3, L \ CO or g2-
(1)
2
2
3
3 2
alkene).
Choice of
a computational model. The reaction of
[IrH (O CCF )(PH ) ] (4) and ethylene was studied at the
(2)
2
2
3
3 2
DFT level (B3PW91, see Computational details). When neces-
sary, propene was used as a model oleÐn. PH was used as a
3
The goal of this study is to obtain mechanistic information
for the alkane dehydrogenation via eqn. (1) and (2) from the
experimental evidence for the isotope exchange of eqn. (3). The
connection between eqn. (1) and (2) and eqn. (3) comes from
the more stable dihydrido oleÐn system of eqn. (3) exploring
part of the reaction path of eqn. (1) and (2) in reverse due to
the microreversibility principle. We studied and brieÑy
reported H/D isotope exchange data for eqn. (3), the
model phosphine, as is usually the case.17 Unless otherwise
mentioned, the reference energy (in kcal mol~1) is that of the
separated reactants in their most stable geometry (i.e., 4 with
j2-coordinated triÑuoroacetate, 4j2, plus free oleÐn).
Experimental isotope exchange reactions with 1
Monitoring the reaction mixtures by 2H NMR spectroscopy
over 2 h at 25 ¡C showed that the deuteriated complex d2-1
reacts with a variety of alkenes, leading to isotope exchange
with the alkene. Both organic (2H NMR) and inorganic (31P
NMR) products could be monitored. Alkane is a very minor
(\5%) product and when this is formed, the resulting reactive
Ir(O CCF )(PAr ) fragment is irreversibly trapped by the
reaction between the hydrogen acceptor tBuCH2CH and
2
[IrD (j2-O CCF )(PAr ) ], but at least two alternate
2
2
3
3 2
pathways remained possible.27
2
3
3 2
solvent to give Ir(Ph)(H)(O CCF )(PAr ) .33b
3 2
Because disubstituted alkenes R C2CH behave di†erently,
(3)
2
3
2
2
we Ðrst consider monosubstituted alkenes such as PhCH -
CH2CH . This shows simultaneous isomerization to the more
stable isomer PhCH2CHCH and D incorporation at all the
side chain sites (1, 2 and 3 positions). The isomerization [eqn.
2
We originally proposed the intermediacy of a carbene for
isotope exchange via eqn. (3) since no free alkane was
observed and Ðrmly bound (““stickyÏÏ) alkane complexes were
not considered possible at the time; in addition, carbene for-
mation was feasible without ligand slip.27 The current situ-
ation is therefore unsatisfactory in that the same general type
of reaction (alkane dehydrogenation chemistry and H/D
isotope exchange) currently has two di†erent published expla-
nations.
2
3
(4)] is attributed to a Markovnikov MÈD insertion into the
C2C bond (M bonded to the more substituted carbon) to give
a secondary alkyl, followed by b-elimination; this process
labels the 1 position. Concomitant insertion/elimination in an
anti-Markovnikov fashion [not shown in eqn. (4)] in the reac-
tant alkene labels the 2 position. Labeling at C-3 can arise
from further insertion/elimination. Everything can be inter-
preted via insertion/b-elimination and nothing else is required,
so monosubstituted alkenes are not mechanistically useful.
This paper reports experimental data for the reactions in
eqn. (3) and also DFT (B3PW91) calculations that allow us to
exclude both carbene27 and vinylic oxidative addition mecha-
1122
New J. Chem., 2001, 25, 1121È1131

View Article Online
(4)
Disubstituted alkenes, R C2CH , in contrast, only show
2
2
isotope exchange with d2-1 at the unsubstituted carbon
without signiÐcant (\2%) alkene isomerization. For methyl-
enecyclohexane, the endocyclic isomerization product, 1-
methylcyclohexene, known to be more stable than the starting
alkene, should have been seen if a pathway were available.
For isobutene, the isomerization is degenerate, but it would
still label the methyl groups; this did not happen to any sig-
niÐcant (\2%) extent.
Fig. 1 Optimized structures for type 4 complexes: 4j1 right and 4j2
left. For clarity, the hydrogens of PH are not shown. Energies are in
3
kcal mol~1.
The reactions described here only occur in hydrocarbon
solvents. In halogenated solvents such as CH Cl , the metal
2
2
complex degrades by oxidation. In polar solvents like acetone,
the CF COO~ ligand is displaced and the resulting
3
[IrH (solvent) L ]CF COO gives quite di†erent chemistry.
kcal mol~1 above 4j2. This is consistent with an energetically
facile opening of 4j2 to allow the coordination of an incoming
ligand.
2
2 2
3
Experimental probe of triÑuoroacetate coordination
When ethylene binds, two adducts, each with j1-tfa and
trans phosphines, are located as minima (Fig. 2). In line with
experiment, the more stable ethylene isomer, 5c ([16.2 kcal
mol~1 relative to 4j2 plus free ethylene), has cis hydrides and
therefore the oleÐn is cis to tfa (Fig. 2). Isomer 5t with trans
hydrides is not observed experimentally, in agreement with its
high energy relative to 5c (]11.4 kcal mol~1 above 5c). In
both complexes, ethylene is predicted to be perpendicular to
the IrÈP direction. Paradoxically, ethylene is more weakly
bound to Ir in the more stable 5c than in 5t as indicated by
the geometric parameters: longer C2C and shorter IrÈC dis-
tances in 5t. The greater stability of 5c is thus not associated
with stronger oleÐn binding but with a better mutual arrange-
ment of the two H ligands. With their large trans inÑuence,
mutually trans hydrides as in 5t are especially unfavorable.
To see if the triÑuoroacetate ligand (tfa) in 1 remained bound,
we studied the reaction of methylenecyclohexane with
[IrD (PAr ) (O2CMe ) ]BF , d2-2, under identical condi-
2
3 2
2 2
4
tions.33b 2 is expected to be a good model for the case in
which the tfa completely dissociates since O2CMe is a good
2
leaving group. In sharp contrast with 1, 2 gave a very rapid
(seconds) isomerization to 1-methylcyclohexene [eqn. (5)]. The
1-methylcyclohexene is deuterium-labeled, principally at the
CH group. The tfa in 1 must therefore remain bound, at least
3
under the conditions used, or isomerization of the oleÐn
would have been seen.
(5)
Experimental observation of intermediate alkene complexes
As reported earlier,33b we examined the reactions of the tfa
complex, 1, with alkenes at low temperature by 1H NMR in
an attempt to directly observe reaction intermediates, to help
infer the reaction path. In all cases we saw either no change or
only the formation of an alkene complex of the type [IrH (j1-
2
O CCF )(L)(PAr ) ] (3, L \ g2-alkene) in equilibrium with
2
3
3 2
the starting complex 1. These were characterized by 1H NMR
spectroscopy because attempts to isolate them led to reversion
to the parent complex 1. CO did bind more strongly to give 3
(L \ CO), isolated as a moderately stable complex. In both
cases, 1H NMR spectroscopic data were consistent with the
presence of cis hydrides.
Equilibrium constants, also reported earlier,33b show steric
and electronic e†ects; small alkenes (e.g., C H , K [ 5
2
4
eq
] 103 L mol~1 in CD Cl at [80 ¡C) and those with elec-
2
2
tron withdrawing groups (e.g., styrene, K \ 32) bind best. As
an example of steric e†ects, ethylene binds strongly but the
eq
bulky tBuCH2CH binds weakly (K \ 0.58). Electronic
2
eq
e†ects are indicated by the fact that Me SiCH2CH (K
\
3
2
eq
1.3 ] 103) binds much better than tBuCH2CH (K \ 0.58)
and styrene (K \ 32) binds better than 1-hexene (K \ 17).
2
eq
eq
eq
Theoretical study of intermediate alkene complexes
Calculations on the model starting complex [IrH (O C-
2
2
Fig. 2 Reaction paths and optimized structures and transition states
(distances in A, angles in ¡) for insertion of C H into the IrÈH bond
CF )(PH ) ], 4, show that the tfa is hemilabile. The bidentate
3
3 2
j2 and the monodentate j1 forms are local minima close in
energy on the potential energy surface (PES). 4j2 is pseudooc-
tahedral while 4j1 is a Y-type distorted trigonal bipyramid
(TBP)34h36 with apical phosphines (Fig. 1). 4j1 lies only 5.3
2
4
for cis-IrH (j1-tfa)(C H )(PH ) , 5c (left) and for trans IrH (j1-
3 2
2
2
4
2
tfa)(C H )(PH ) , 5t (right). Hydrogens of PH are not shown for
2
4
3 2
3
clarity. Energies are given with respect to 4j2 plus free ethylene (in
kcal mol~1).
New J. Chem., 2001, 25, 1121È1131
1123

View Article Online
Experimental study of oxidative addition mechanism
vation by 2H NMR, k /k was found to be 2.5( ^ 0.1). This
CH CD
value is sufficiently di†erent from that at long reaction times
The failure of alkenes of the type R C2CH to isomerize
to suggest a di†erent mechanism.
2
2
during H/D exchange could be explained by an oxidative
These are independent, converging arguments that show
that the mechanism of H/D exchange by IrH (tfa)L at room
addition to the vinyl CÈH bond, as found experimentally for
IrH L ,33a so we looked for experimental methods for dis-
5 2
2
2
temperature does not allow oxidative addition at the Ir(III)
tinguishing this mechanism (Fig. 3). In the authentic vinyl
level and does not form (at short times, \2 h) Ir(I)(tfa)L ,
activation pathway, strong selectivity is seen for exchange of
the vinyl CÈH trans to the bulky tBu group and fast exchange
also occurs readily with arenes. In contrast, in our system,
IrH (tfa)L at room temperature with no added tBuCH2CH ,
2
which is capable of oxidative addition. We therefore moved to
a theoretical study to compare the possible routes.
Theoretical studies of vinylic oxidative addition to Ir
2
2
2
there is no selectivity at all in alkene exchange and arenes are
unreactive (\2 h). One could perhaps dismiss the failure to
distinguish between the cis and trans positions of the
The experimental results clearly suggest that Ir(III)H (tfa)L ,
2
2
1, is unlikely to oxidatively add, even to a reactive CÈH bond.
We decided to see if the calculations could conÐrm this point.
The CÈH oxidative addition to 4j1, to give vinyl
[IrH (CH2CH )(PH ) (j1-tfa)], 6, was calculated. Two
unhindered terminal CH group, but the highly hindered
2
a-vinyl position of tBuCH2CH is also attacked equally fast.
2
In methylenecyclohexane, there can be no selectivity, but it
3
2
3 2
does not seem reasonable to postulate a di†erent mechanism
minima, close in energy, di†ering only in the orientation of tfa,
for tBuCH2CH and methylenecyclohexane.
were located for 6 (Fig. 4). They both have trans phosphines
2
The absence of reaction with arenes in the absence of an H
and an IrH(H ) ligand set, showing a preference for Ir(III) over
2
acceptor is especially informative. Arene CÈH bonds are very
reactive in oxidative addition, so IrIIIH (tfa)L cannot
undergo oxidative addition; the required Ir(V) intermediate is
Ir(V). The energies are ]10.6 (6a, tfa eclipsing IrÈP) and
]11.5 kcal mol~1 (6b, tfa perpendicular to IrÈP), respectively,
above 4j2 plus free ethylene. A third Ir(III) vinyl complex, 6c,
with cis phosphine ligands and an IrH(H)(tfaH) ligand set was
located 10.4 kcal mol~1 above the energy reference. Any
mechanism going through these vinylic intermediates is thus
unlikely because their energy is signiÐcantly higher than that
of the separated reagents, 4j2 and free ethylene.
2
2
expected to be unfavorable. In the reaction of [IrD -
5
(PCy ) ]/tBuCH2CH where oxidative addition does occur,
3 2
2
this can take place at the Ir(I) level after loss of dihydrogen
(Fig. 3).33a If the same mechanism is considered in the case of
IrH (tfa)L , the loss of H to a sacriÐcial oleÐn would lead to
2
2
2
IrI(tfa)L , which has no hydride left and thus cannot be
To test the role of the anionic ligand in the vinylic CÈH
activation, we replaced tfa by H. As expected, the vinylic
species [IrH (CH2CH )L ] is a pentagonal bipyramidal Ir(V)
2
involved in H/D exchange. This is in full agreement with
experimental observations. Under the conditions we use
(room temperature, \2 h) and in the absence of a hydrogen
acceptor, no signiÐcant arene deuteration was observed with
benzene or naphthalene.
4
2 2
tetrahydro vinyl with apical phosphines. Its energy (30.1 kcal
mol~1 above the more stable ethylene complex) shows that
this reaction is disfavored. Thus, whatever the anionic ligand,
the 5-coordinate Ir(III) fragment does not give vinyl CÈH acti-
vation.
The IrH (PCy ) system shows an induction period during
5
3 2
which formation of tBuEt is observed; neither an induction
period nor tBuEt is seen in our system. The induction period
is proposed33a to involve hydrogen transfer from Ir to give a
coordinatively saturated intermediate, plausibly Ir(I), which
gives vinyl oxidative addition. The vinyl activation route
might be expected to show a signiÐcant primary isotope e†ect,
because although a range of primary isotope e†ects have been
seen for CÈH(D) oxidative additions to late transition
metals,37 our particular system [1 after a long (10 h) induction
Initial step: alkene insertion into the Ir–H bond
Insertion of methylenecyclohexane into the IrÈH bond is left
as the most likely path. This is expected to be anti-
Markovnikov to give the primary alkyl, because the opposite
orientation would lead to an unfavorable bulky tertiary alkyl
and 1-methylcyclohexene, not seen experimentally, as Ðnal
products. The labile and less hindered cationic acetone
complex 2 does give a Markovnikov insertion, however,
because the products of such an alkene isomerization are seen.
Since methylenecyclohexene is never seen for 1, we propose
that the very hindered tertiary alkyl intermediate required for
isomerization never forms. With this in mind, we did not
introduce any substituents into the model compounds because
the bulky groups are always remote from the metal.
period] allows the formation of IrI(tfa)L , which can oxidati-
2
vely add to arene. The Ir(I) intermediate retains a hydride in
the coordination sphere of the metal and can therefore give
H/D exchange. This system shows33b a k /k
CH CD
C H /C D , attributed to the oxidative addition step. When
of 4.5 for
6
6
6 6
we ran the reaction in eqn. (3) in reverse mode at room tem-
perature over 2 h, using deuteriated d2-alkene (formed from
Ph PCD
and cyclohexanone by
a
standard Wittig
Ethyl complexes of type 7 result from ethylene insertion
into one IrÈH bond in 4. Several minima close in energy were
predicted either with j2-tfa (7j2) or with j1-tfa (type 7j1) (Fig.
5). The two 7j2 complexes that were located as minima are
pseudooctahedral and di†er only in the orientation of the
3
2
reaction)37b,c and protonated metal complex, d0-1, with obser-
Fig. 4 Optimized structures for type 6 vinyl complexes: 6a, 6b and
Fig. 3 Reaction mechanism for H/D exchange between CH 2CHtBu
6c. For clarity, the hydrogens of PH are not shown. Energies are
2
3
and IrL D .33a
given with respect to 4j2 plus free ethylene (in kcal mol~1).
2
5
1124
New J. Chem., 2001, 25, 1121È1131

View Article Online
hydride ethylene complex, 5t, to the agostic alkyl, 7j1T via
H
TS(5t-7j1T ), only 7.6 kcal mol~1 above 5t (Fig. 2, right). In
H
going from 5t to TS(5t-7j1T ), one of the hydrides
approaches the ethylene (C É É ÉH 1.65 A) and the other
hydride moves away (C É É ÉH 2.86 A). One reason why inser-
tion of ethylene is so easy from 5t is the higher nucleophilicity
H
b
a
of the hydride trans to another hydride. A hydride trans to a
weak trans e†ect ligand like tfa is expected to be much less
nucleophilic. Another factor also helps to lower the transition
state from the less stable oleÐn isomer. Starting from 5c the
new r IrÈC bond necessarily develops pseudo trans to the
remaining hydride (Fig. 2); having two high trans e†ect
ligands mutually trans is unfavorable. When inserting from
the less stable oleÐn complex, 5t, the new IrÈC bond develops
cis to IrÈH, a more favorable situation. The ethylene complex
that is both experimentally observed and predicted to be most
stable is thus a dead-end for the reaction since the insertion is
energetically inaccessible from this isomer and the whole reac-
tion is proposed to go through the less stable ethylene adduct
5t. Related results were previously calculated for the reaction
of ethylene with (PCP@)Ir(H) 18 and for RhH (Cl)(C H ).15,16
2
2
2 4
This shows a sigiÐcant inÑuence of the ancillary ligands on the
insertion of an oleÐn into a metalÈhydride bond.
Pathway for H/D exchange
In the methylenecyclohexane and the isobutene cases, after the
insertion takes place to give the primary alkyl, the elimination
has to occur via the removal of the same D label originally
added to the alkene in the insertion step because no alterna-
tive is available. This would lead to the formation of
unlabeled alkene and so does not explain our observation of
label incorporation. Two leading mechanisms need to be con-
sidered for the isotope exchange, one involving an alkane
complex (modeled later by 8) and the other a carbene interme-
diate (modeled later by 9) (Fig. 6). In the carbene pathway,
Fig. 5 Optimized structures (distances in A, angles in ¡) for type 7
alkyl complexes Ir(H)(tfa)(C H )(PH ) . For clarity, the hydrogens of
2
5
3 2
PH are not shown. Energies are given with respect to 4j2 plus free
3
ethylene (in kcal mol~1).
ethyl ligand; they are 4.1 and 2.5 kcal mol~1 below the most
stable ethylene complex 5c. All 16-electron 7j1-type com-
plexes are calculated to be less stable than the 18-electron 7j2
species although the energy di†erence is not large for hemi-
labile tfa. In the complexes with j1-tfa and trans phosphine
ligands, the three other ligands can form either a square
pyramid (T) or distorted TBP (Y) arrangement (Fig. 5).34h36
Two complexes were found with a Y arrangement, tfa occupy-
ing the foot of the Y, at 2.0 (7j1Y) and 4.0 kcal mol~1 (7j1Yº)
a-elimination causes exchange between IrÈD and the a-CH
2
group of the alkyl via a carbene intermediate. In the alkane
pathway, the H/D exchange takes place via reductive elimi-
nation to give an alkane complex in which the metal then
migrates from one CH bond to a geminal one. These are not
readily distinguishable experimentally, hence the need for
theory. As shown in Fig. 6, we expect the alkene product to be
labeled with a single D in either mechanism, in line with
experiment since we Ðnd that the majority ([90%) of the
labeled isobutene formed contains just one and not two deute-
riums per molecule.
above 5c; they di†er only by rotation about the IrÈC bond
a
and the ethyl group is not agostic. Two T-shaped alkyls were
also found with hydride, 7j1T , or alkyl, 7j1T , trans to the
H
C
vacancy at 5.9 and 3.9 kcal mol~1 above 5c, respectively. The
former, 7j1T , forms a b-agostic CÈH structure (IrÈC ÈC \
H
a
b
90.7¡; C ÈH \ 1.13 A). This interaction appears to be weak
since removing it by rotating around the C ÈC bond while
keeping the angles between the ligand bond at the metal as in
the agostic complex does not signiÐcantly alter (0.2 kcal
Carbene complex as intermediate
b
a
b
In the carbene path, fast a-elimination to give a carbene dihy-
dride or dihydrogen complex allows H/D exchange by a
reverse a-migration to regenerate the alkyl, now deuterium
labeled. A b-elimination step completes the process and intro-
duces one D per molecule, as observed. This pathway seemed
the most reasonable to us in the early work and we initially
suggested it for the isotope exchange.27
mol~1) the energy of 7j1T . The similarity of the energies
H
found for the Y- and T-shaped alkyl complexes conÐrms prior
studies showing the high Ñuxionality of d6 ML complexes36
5
and suggests interconversion between the di†erent alkyl
minima is easy with monodentate j1-tfa.
The search for the transition state (TS) for the insertion of
the ethylene into the IrÈH bond gave remarkable results. No
TS could be found that links any oleÐn complex of type 5 to
the 18-electron 7j2 alkyl complex. All transition states located
link an oleÐn complex (type 5) to an ethyl complex with
monodendate tfa (type 7j1). A TS, denoted TS(5c-7j1Yº) was
found between the ethylene complex with cis hydrides 5c and
The b-elimination of a transition metal alkyl and its reverse
reaction, 1,2-insertion of an alkene into an MÈH bond, are
very common. Much rarer are a-elimination of an alkyl and
its reverse, 1,1-insertion of a carbene into an MÈH bond,
about which relatively little is known.38h40 Attention has been
drawn to this problem following several signiÐcant results.
Schrock and co-workers38a reported a series of paramagnetic
Mo(IV) alkyls where a-elimination is [106 times faster than
b-elimination. Other early transition metal cases involve rapid
a-elimination. Among the late transition metals, a tetra-
coordinate 14e Ru hydride reacts with a vinyl hydride to give
a Fischer carbene hydride [eqn. (6)]. DFT calculations of the
reaction path for an ethylene model ligand suggests a multi-
the Y-shaped 7j1Yº (Fig. 2, left). Likewise, TS(5c-7j1T ) was
C
also located (Fig. 2, middle). These two transition states,
which are 38.5 and 34.9 kcal mol~1 above 5c, respectively, are
too high to be reachable, being more than 18 kcal mol~1
above the separated reactants. The lowest energy insertion
path goes from the less stable and experimentally unseen trans
New J. Chem., 2001, 25, 1121È1131
1125

View Article Online
Fig. 6 The mechanism for H/D exchange in methylenecyclohexane via the carbene and the alkane routes. The structures shown are schematic
and are not intended to represent preferred geometries. The alkane pathway is proposed in this work. The species in curly brackets may not be a
stable minimum.
step path involving insertion to form an alkyl that rearranges
to the Ðnal carbene hydride by a-elimination.39,40
complex is predicted to be more stable with trans (9t) than
with cis hydrides (9c). A more stable carbene isomer, 9o, with
cis phosphines was located 5.3 kcal mol~1 above separated
HRuClL ] CH 2CH(OEt) ] RuClL [CHMe(OEt)] ]
4j2 and ethylene. Even though PAr is not very bulky and cis
2
2
2
3
arrangements are known in closely related cases,41 9t prob-
HRuClL [2CMe(OEt)] (6)
2
ably lies very close in energy to 9o and could be preferred
when steric e†ects are considered. For this reason only 9t will
be considered below as representing the carbene complex. The
transformation of the more stable ethylene adduct 5c into the
carbene complex 9t is thus endothermic by 23.5 kcal mol~1
(12.1 kcal mol~1 above the less stable 5t isomer). High endo-
thermicity was also calculated for other metal fragments:
RuHCl(PH ) (16.0 kcal mol~1)39,40 and RuH(CO)(PH )
In pursuit of this alternate mechanism for the present
system, three dihydride carbene complexes (type 9) with
monodentate tfa were located as minima on the PES (Fig. 7).
No minimum with bidentate tfa or dihydrogen ligands could
be located. The two minima with trans phosphines have either
cis hydrides (carbene 9c, 12.9 kcal mol~1 above 4j2 plus
ethylene) or trans hydrides (carbene 9t, 7.3 kcal mol~1 above
4j2 plus ethylene). Thus, unlike the ethylene case, the carbene
3 2
3 2
(23.4 kcal mol~1).42 The carbene being unlikely, we turn to
the alternate alkane pathway for H/D exchange.
The alkane route
Free alkane is not a signiÐcant product (\5%), so any Ir(I)È
alkane species must be stable enough to prevent alkane disso-
ciation before CÈH oxidative addition takes place. Once the
alkane species forms, the metal has to undergo facile 1,1@-
migration of the alkane ligand so the initial IrÈ(DÈCH R)
2
alkane species becomes IrÈ(HÈCHDR). The subsequent steps
reverse the previous ones: a CÈH oxidative addition to give
the alkyl hydride Ir(III) complex is followed by a b-H-elimi-
nation to yield the Ðnal labeled alkene (Fig. 6).
Calculations support the alkane mechanism. The alkane
complex 8 was found to be square planar with a side-on g2-
alkane (Fig. 8) located 0.9 kcal mol~1 below separated 4j2
plus ethylene. The binding energy of the ethane is large (14.4
kcal mol~1), consistent with the signiÐcant elongation of the
Fig. 7 The optimized structures (distances in A) of the carbene dihy-
dride complex Ir(tfa)(CH2CH )(PH ) . For clarity hydrogens of PH
2
3 2
3
are not shown. Energies are given with respect to 4j2 plus free ethyl-
ene (in kcal mol~1).
1126
New J. Chem., 2001, 25, 1121È1131

View Article Online
complex questionable as a distinct intermediate (Fig. 8).
Adding the zero point energy (ZPE) to the total energies
reverses their relative stabilities with TS(7j1Yº-8) now being
0.7 kcal mol~1 below 8. The shape of the PES involving
TS(7j1Yº-8) and the alkane complex 8 was thus computed
with another functional. With B3LYP, the alkane bonding
dissociation energy (BDE) is 11.3 kcal mol~1, consistent with
its shorter (1.18 A) CÈH bond (B3PW91: 1.22 A). For com-
parison, free C H has a CÈH distance of 1.095 A with both
2
6
methods. With B3LYP, TS(7j1Yº-8) is 1.4 kcal mol~1 above 8
without ZPE. Including ZPE with B3LYP (this functional will
no longer be used in what follows) results in the two species
having essentially the same energy (di†erence of 0.1 kcal
mol~1). Consequently, the pathway for the reductive coupling
of the alkyl group and H followed by a 1,1@-exchange of
hydrogens is best described as a concerted CÈH bond metath-
esis in the alkyl complex. The transition state for 1,1@-
exchange, TS(8-8º), has the character of a chelated alkane
complex with two short C ÈH bonds (B3PW91: 1.13 A)
a
Fig. 8 The reaction path for hydride exchange in the alkyl complex
Ir(H)(C H )(tfa)(PH ) via the alkane route (distances in A). For
bound to the metal and with an sp3 C center (HÈCÈ
a
H \ 112.0¡). The H/D exchange should still be considered as
2
5
3 2
clarity, the hydrogens of PH are not shown. Energies (including ZPE
3
going via an alkane-type transition state, TS(8-8º). Precisely
this kind of 1,1@-shift has been proposed to explain the NMR
in parentheses) are given with respect to 4j2 plus free ethylene (in kcal
mol~1).
spectrum of the alkane complex, CpRe(CO) (cyclo-
2
CÈH bond from 1.10 A in free C H to 1.22 A in the alkane
pentane).13,14
2
6
complex. The large binding energy contrasts with the usual
weak binding to transition metals4,9,10,18,25,43 but is sup-
ported by recent observations of moderately stable alkane
complexes.12h14 Entropy e†ects cannot be computed quanti-
tatively in the harmonic approximation, but these should
lower the *G of binding. The calculated *G of dissociation
(]2.3 kcal mol~1) should therefore only be viewed as a
further indication that alkane should not be lost easily.
Further insight into the alkane path
Special role of the tfa ligand. Even though the g2-bonded
alkane is at best a very shallow secondary minimum, it is con-
venient to consider for discussion purposes. The tfa ligand has
a central role in enhancing the binding energy of ethane to
Ir(j1-tfa)(PH ) . Replacing tfa by H lowers the predicted
2 2
Formation of the alkane complex starts with reductive
coupling (reductive elimination of alkyl and hydride in which
the alkane remains bound to the metal). The starting alkyl
7j1Yº is proposed to go to the alkane complex, 8, via a tran-
sition state TS(7j1Yº-8) only 0.2 kcal mol~1 above 8 (Fig. 8).
The TS geometry naturally greatly resembles that of 8, with
the exception of small changes in the IrÈCÈH region. As
expected, the CÈH bond is longer and the IrÈC and IrÈH
bonds are shorter in TS(7j1Yº-8) than in 8. SigniÐcantly, the
reductive coupling starts from a Y-shaped alkyl complex
having a short distance between the alkyl and H to be
coupled. This Y complex is easily obtained from any T struc-
ture owing to the well known Ñatness36 of the PES for d6
binding energy to 5.2 kcal mol~1 and shortens the CÈH bond
to 1.13 A. The binding energy is thus larger for tfa than for H
since tfa is an electronegative ligand with weak trans inÑuence
and some p-donating ability. The CÈHÈmetal interaction can
also be quantiÐed by examining the agostic CÈH vibration
frequencies. Binding to the metal weakens the CÈH bond. The
average calculated l
of the free CÈH in the ethane complex
ChH
is 3114 cm~1 (^40 cm~1) whereas the l
frequency of the
ChH
metal-bound CÈH is lowered to 1722 cm~1. Replacing tfa by
H raises l
of the g2-bonded CÈH to 2624 cm~1, in agree-
ChH
ment with the shorter CÈH distance.
Comparison with previous calculations of alkane complexes
highlights some important features. Prior discussions have
emphasized the hapticity,19,44 the role substituents,19 and the
solvent.17,45 Binding energies and geometries not only depend
on the metal and ligand but also on the level of calculations.
Alkane complexes have been calculated for metals with
various electron counts (d2, d6, d8, d10). Bound alkanes in d10
complexes have very low bond dissociation energies (\5 kcal
mol~1) and very small CÈH elongations (CÈH \ 1.11 A).46,47
Coordinated alkane in d6 and d8 complexes have larger
MÈalkane BDEs, which vary signiÐcantly with the level of
ML species. It is probably for this reason that we were
5
unable to locate a transition between 7j1T and 7j1Yº. Faster
H
reductive elimination from low coordinate organometallic
species is a common Ðnding; the greater Ñexibility of these
intermediates may allow distortions that lead to the appropri-
ate transition states.
The isotope exchange cannot be completed without 1,1@-
exchange within the alkane complex, converting CÈDÈM to
CÈHÈM so that D is retained in the alkene. The TS for 1,1@-
exchange, TS(8-8º), with C symmetry and two CÈH bonds
calculations. For instance, the BDE of CH in M(Cp)(CO)q
s
4
proximate to the metal, is located only 2.5 kcal mol~1 above 8
(M \ Os, q \ [1) is only 10.1 kcal mol~1 at the B3LYP level
(prime labels describe structures equivalent to non-prime
labels after H exchange). In this g3-ethane species, the two
CÈH bonds are moderately stretched to 1.13 A (Fig. 8). In
TS(8-8º) the alkane is still strongly bound (11.9 kcal mol~1),
allowing the H/D exchange to occur without formation of free
ethane. The activation energy for this exchange is very close to
but 9.3 kcal mol~1 larger at the MP2 level.48 Changing the
correlation functional in hybrid DFT methods from B3LYP
to B3P86 increases the BDE by nearly 5 kcal mol~1 in the
same type of compound (M \ Ir, q \ 0).49 Similar variations
have been calculated in this work for Ir(tfa)(PH ) (C H ), with
3 2
2 6
B3LYP giving a lower BDE than B3PW91.
that calculated in CpOs(dmpm)(CH )H` (2.0 kcal mol~1).31
Ligands greatly inÑuence the alkane BDEs. Methane
3
The highest transition state of the alkane path is TS(8-8º),
binding energies to Ir(PH ) X (X \ H, Cl), calculated at the
3 2
which is at lower energy than the intermediates in the carbene
path. If so, this leads to the conclusion the that the alkane
path is most likely for the isotope exchange.
The small predicted energy di†erence between TS(7j1Yº-8)
and alkane complex 8 makes the existence of the alkane
RHF level are 6.8 (X \ H) and 15.6 kcal mol~1 (X \ Cl).50
Replacing H by Cl thus has similar e†ects to those found on
replacing H by tfa in this work. Both tfa and Cl have a lower
r-donor ability than H and both lead to a larger alkane BDE.
The RHF level also reproduces the lowering of the l
,
ChH
New J. Chem., 2001, 25, 1121È1131
1127

View Article Online
larger for Cl than for H, although the extent of this lowering is
much smaller at the RHF level than at the B3PW91.
1,n-Shifts (n = 1 to 3) in agostic alkane complexes. Although
1,1@-migration of Ir in the alkane complex has to be fast,
neither 1,2- nor 1,3-migration of the metal in the proposed
alkane complex can be fast because these processes would
lead to net alkene isomerization, which is not observed. Since
the H atoms of an alkane are essentially close packed around
the carbon core, this restriction initially seemed unreasonable
and originally led us to disfavor the alkane complex pathway.
Fast 1,1@- vs. 1,2- and 1,3-migration must apply to our case
and has been seen in NMR data on alkane complexes,13,14
but, as Periana and Bergman have shown,9a some alkane
complexes can undergo 1,2- and 1,3-shifts. Because all the
alkane complexes in this paper with slow 1,2- and 1,3-migra-
tion are derived from disubstituted alkenes, all have tertiary
CÈH bonds adjacent to the CÈH bond formed by reductive
coupling. Coordination through the tertiary CÈH bond could
be strongly disfavored because the bulky R groups might
hinder the required side-on binding, which could well prevent
a 1,2-shift from occurring. Calculations on propane vs. ethane
Alkane as protecting group. The calculations suggest that
when X is tfa, the alkane acts as a protecting group for the
highly reactive MIrXL N fragment, which would otherwise
2
undergo oxidative addition17,20 with an external substrate or
via orthometallation. This protection is e†ective at 25 ¡C but
on heating, the alkane departs because the thermal (150 ¡C)
alkane dehydrogenation system does give both alkane and
arene CH activation. When X is H, the weakly bound alkane
is no longer a protecting group. This protection is especially
important against vinylic oxidative addition with the 14-
electron tricoordinate IrIXL . The vinylic oxidative addition,
2
calculated to be thermodynamically favorable with the two
anionic ligands, will be presented in a future paper.
show that one additional methyl group on C decreases the
a
alkane binding energy by 2.4 kcal mol~1 whereas it has no
An overview of the reaction path for H/D exchange
inÑuence when substituted at C . This trend is thus opposite
b
from that calculated for W(CO) (alkane)19 but energy varia-
The entire path for H/D exchange, presented in Fig. 9, has a
mirror element that goes through the g2-alkane transition
state. The Ðrst part (A) includes the insertion of the oleÐn into
the IrÈH bond and the coupling of the ethyl and hydride to
form the alkane. In the second part (B), a CÈH bond of the
alkane is cleaved to form an ethyl hydride complex, which
itself b-eliminates an oleÐn. The key characteristic of this
entire path is the presence of two steps that are equally impor-
tant for determining the kinetics of the reaction: the Ðrst is
reached from the ethyl hydride complex, 7j1Yº, going to the
alkane transition state [TS(8-8º)] in part A and the second is
reached from the same 7j1Yº going to the b-elimination tran-
sition state TS(5t-7j1Tº) in part B. These two steps are ener-
getically comparable since they necessarily start from the same
alkyl complex (7j1Yº) and go to transition states that are of
5
tions in the latter unhindered system were too small to be
conclusive. No further calculations were thus carried out on
the 1,2-shift.
The TS for 1,3-migration of the metal on propane was
found to be 11.7 kcal mol~1 above the primary propane
complex and only 2.9 kcal mol~1 below the dissociation of
propane. No direct 1,3-shift is thus expected to occur. In this
poorly bonded transition state, the hydrogens of the two CÈH
bonds are pointing directly toward the metal in a chelated
g1 ] g1 mode, quite unlike the usual side-on g2 arrangement
of the r-bonded alkane. We interpret the high barrier of the
1,3-shift vs. the 1,1@-shift to mean that decoordination of the
metal from the carbon of the CÈH bond, required for the 1,3-
but unnecessary for 1,1@-shift, is unfavorable in this case (10).
Fig. 9 Full energy proÐle (kcal mol~1) for 1,1@-H exchange in ethylene in the presence of IrH (tfa)(PH ) . The energy reference corresponds to
3 2
separated reactants IrH (tfa)(PH ) and C H in their optimized geometries. The dotted line indicates that structure 8 may not be a true
3 2
minimum. Prime should be assumed on the right hand side of the diagram, since B is the mirror image of A.
2
2
2 2
1128
New J. Chem., 2001, 25, 1121È1131

View Article Online
Relation between isotope exchange and alkane dehydrogenation
Mechanism of photochemical alkane dehydrogenation. We
originally investigated the H/D exchange reactions in the
hope of using the results to understand the alkane functional-
iztions of eqn. (1) and (2). This is possible because the isotope
exchange is very closely related mechanistically to alkane
dehydrogenation with [IrH (PR ) (O CCF )] [eqn. (1) and
2
3 2
2
3
(2)] since both reactions involve the same intermediates. For
example, Fig. 10 shows the path proposed for the photo-
chemical alkane dehydrogenation via eqn. (1). The three pro-
posed intermediates involved are precisely those that take part
in the H/D exchange process.
In the photochemical version [eqn. (1)], carried out in
alkane solution at room temperature, irradiation of starting
material 1 at 254 nm leads to conversion of cyclohexane to
cyclohexene and methylcyclohexane to methylenecyclohexane
and 1-methylcyclohexene (55 : 45 ratio). We interpret the
photochemical step as the photoextrusion of H , well known
2
in similar systems.9a The resulting Ir(I) fragment can then bind
alkane to give the same alkane complex as we saw in Fig. 6.
The reaction to release alkene can then proceed as discussed
above and shown in Fig. 9. The crucial di†erence between
this reaction and the dehydrogenation catalyzed by
[(PCP)Ir(H) ]18 is that initial extrusion of H permits the
reaction to go downhill from the alkane adduct to the oleÐn
complex. The experimental product ratio indicates that the
14e Ir(I) intermediate in this reaction slightly prefers to bind
methylcyclohexane by the less hindered methyl group. The
preferential formation of methylenecyclohexane indicates that
Ir(I) attacks the alkane at the methyl group to give the
primary alkyl after oxidative addition. Any other attack
would tend to give the much stabler 1-methylcyclohexene.
Fig. 10 The mechanism for photochemical alkane dehydrogenation
with Ir(H) (tfa)(PAr ) . The structures shown are schematic and are
2
3 2
not intended to represent preferred geometries.
2
2
similar energies. The multistep nature of this reaction con-
trasts with the case of IrH (PCy ) where the rate determining
5
3 2
step is believed to be the oxidative addition of the CÈH bond
to the IrIHL fragment. The multistep nature of the present
2
reaction prevents any simple interpretation of the experimen-
tal isotope e†ect, k /k . The observed di†erence in the values
H
D
(2.5 in this work and 4.5 under di†erent experimental
conditions33b), however, is consistent with di†erent mecha-
nisms in these two sets of experiments.
Mechanism of thermal alkane dehydrogenation. In the
thermal version of the reaction (Fig. 11), tBuCH2CH dehy-
2
Fig. 11 The mechanism for thermal dehydrogenation of alkanes with Ir(H) (tfa)(PAr ) . The structures shown are schematic and are not
3 2
2
intended to represent preferred geometries.
New J. Chem., 2001, 25, 1121È1131
1129

View Article Online
drogenates the starting metal dihydride complex, 1. The Ðrst
part of the reaction is predicted to go uphill from the
tBuCH2CH adduct to the tBuCH CH complex. Substitution
problems even where intermediates defy experimental detec-
tion.
2
2
3
of the alkane is achieved as a result of the high temperature in
the reaction (150 ¡C) and is probably helped by the presence of
the bulky tBu group. Once substitution of the alkane occurs,
the second part of the reaction is downhill from the methyl-
cyclohexane adduct to the methylenecyclohexane adduct. The
knowledge gained from the isotope exchange studies above
thus allows us to suggest that the tBuCH CH complex of
Technical section
Computational details
The calculations were carried out using the GAUSSIAN 98
set of programs52 within the framework of DFT at the
B3PW91 level of the theory.53,54 Extra calculations were also
carried out using the LYP correlation functional.55 Ir was
represented using the HayÈWadt relativistic electron core
potential (ECP) for the 46 innermost electrons and its associ-
ated double f basis set.56,57 Cl and P were also represented
with the Los Alamos ECPs and their associated double f aug-
mented by a d polarization function.58 A 6-31G(d,p) basis set
was used for all other atoms.59 This basis set has been used
for a large number of similar studies and gives a proper
description of bonding in organometallic species.60 The BSSE
correction is not calculated since it has been shown to be
small with DFT calculations and this basis set.60 Full opti-
mizations without symmetry constraints have been carried
out. The nature of all transition states was assigned by ana-
lytical frequency calculations. ZPE corrections were calculated
for each extrema. The corresponding values are given on the
Ðgures but discussion has been limited to cases where it had
an inÑuence. The transition state structures were slightly per-
turbed then further geometrically optimized to ensure they
connect the reactant and product of interest.
2
3
Ir(tfa)L is a likely intermediate instead of 4j1 and that a
2
much higher temperature is evidently required, not for the
CÈH activation step (which the calculations suggest has no
activation energy), as might have been assumed, but to cause
substitution of the reagent-derived alkane, tBuCH CH to
2
3
give the methylcyclohexane adduct. Since the BDE of the
alkane in [(alkane)IrXL ] is calculated to be lower when
2
X \ H, rather than X \ tfa, it should be easier to decoordin-
ate the alkane in the case of X \ H. In agreement with this
step being rate determining, a much lower reaction tem-
perature (100 vs. 150 ¡C) was reported for alkane dehydroge-
nation with [IrH (PiPr ) ]/tBuCH2CH than with the present
5
3 2
2
tfa system.51
Conclusions
A re-examination of the mechanism of H/D oleÐn exchange in
the presence of Ir(H) (O CCF )(PAr ) via a combined experi-
2
2
3
3 2
mental and theoretical approach leads to the proposal that
the preferred route goes via a transition state having an alkane
coordinated through two CÈH bonds (Fig. 8). After opening of
the tfa chelate ring, an easy process, the alkene coordinates.
Indeed, the whole pathway seems to go through j1-tfa species.
Insertion cannot occur from the most stable oleÐn complex,
but only from a trans dihydride isomer in which the MÈH
bond is more nucleophilic. Rather than a-elimination to give
the carbene, experimental and theoretical data suggest that
the alkyl hydride prefers to reductively eliminate to give a
bound alkane TS. The unusually strong binding in the TS
accounts for the lack of free alkane as product and may con-
tribute to the much higher barrier calculated for the 1,3- vs.
the 1,1@-migration in the alkane complex. Rather than the
carbene mechanism previously invoked, a sticky alkane route
could explain the Ðnding32 of preferential 1,1@-polydeuteration
of alkanes by Pt(II)/DOAc.
Experimental details
Syntheses were carried out under puriÐed N or Ar as
2
described in before.33b NMR spectra were obtained on Bruker
WM500, HX490, GE-Omega 300 and WM250 instruments
and IR spectra on a Nicolet 5-SX FT-IR. Dideuteriomethyl-
enecyclohexane (C H CD ) was prepared from cyclo-
6
10
2
hexanone (Aldrich) and Ph PCD
by the literature
3
2
method37b,c and puriÐed using a GOW-MAC series 350 gas
chromatograph. Other materials were obtained from Aldrich
Chemical Co. and Strem Chemical Co. and puriÐed by stan-
dard techniques.
NMR observation of deuteriation. In a typical case, d2-1
(Ar \ p-FC H ; 35 mg, 0.037 mmol) was dissolved in C H
6
5
6 6
(0.5 mL) and alkene (3 mol equiv.) added by syringe in an
NMR tube under Ar. The reaction was monitored by 2H
NMR over 3 h. Monitoring the 31P resonances allowed quan-
tiÐcation of the metal complexes present [orthometallated
complex: ]17.1, [44.3 ppm; phenyl hydride: ]17.2 or
]17.3 ppm (C D form)] relative to a known concentration
These results suggest that it is the presence of the tfa ligand
that allows the Ir fragment to bind the alkane so strongly.
This allows the H/D exchange to occur without alkane
decoordination. The hemilabile character of tfa allows coordi-
nation of tBuCH2CH . The tfa also indirectly inhibits the
2
6 5
vinyl CÈH oxidative addition, a process that is not feasible on
of an external PPh standard. The authentic compounds were
3
the Ir(III) starting material even if the tfa is monodentate and
not feasible at the Ir(I) stage because no exchangeable D
atoms are left bound to the metal.
used to identify the resonances. In all cases, D incorporation
happened at essentially the same rate in all the vinylic hydro-
gen positions and for all the alkenes. Alkane was not a signiÐ-
cant product (\5%) in any case. Isomerization of the C2C
double bond was signiÐcant only for PhCH CH2CH ; in
other cases it was \5%. GC-MS showed that the labeled
alkene was ca. 85% d2 for the monosubstituted alkenes and d1
for the 1,1-disubstituted alkenes. The results were independent
of solvent (benzene, toluene, dichloromethane). Kinetic
isotope e†ects were determined by 2H-NMR spectroscopy
using Ðve determinations over the Ðrst 10% of reaction, where
product formation is essentially linear with time.
These results permit us to unify the mechanism of alkane
dehydrogenation with that for H/D atoms exchange in oleÐns
by IrH (O CCF )(PAr ) . In the alkane dehydrogenation the
2
2
2
2
3
3 2
two H ligands of the catalyst are removed either by photoex-
trusion or by transfer to a hydrogen acceptor oleÐn. The cata-
lytic reaction only occurs at 150 ¡C, because only then does
thermal substitution of the alkane derived from the hydrogen
acceptor becomes possible. After dissociation of this alkane,
the resulting highly reactive species, trans-Ir(O CCF )(PAr ) ,
2
3
3 2
is proposed to coordinate the substrate alkane, which is then
dehydrogenated. The 14e species is not formed in the H/D
exchange reaction, however, where the acceptor alkane acts as
protecting group. This study not only uniÐes reactions pre-
viously regarded as distinct but it also shows that combining
calculations with experiment can suggest answers to chemical
Study of k /k for Ir–D/C H CH vs. Ir–H/C H CD
2
CH CD
6
10
2
6
10
exchange. To determine k , d2-1 (Ar \ p-FC H ; 35 mg,
CH
6 5
0.037 mmol) was dissolved in C H (0.5 mL) and C H CH
6
6
6
10
2
(20 mol equiv) added by syringe in an NMR tube under Ar.
The reaction was monitored by 2H NMR over 2 h at 20 ¡C.
1130
New J. Chem., 2001, 25, 1121È1131

View Article Online
29 G. L. Gould and D. M. Heinekey, J. Am. Chem. Soc., 1989, 111,
5502.
The value of k was determined similarly by following the
CD
reaction of 1 (20 mg, 0.02 mmol) with C H CD (20 mol
6
10
2
30 C. L. Gross and G. S. Girolami, J. Am. Chem. Soc., 1998, 120,
equiv.) in C D (0.5 mL) at 20 ¡C via 1H NMR over 2 h.
6605.
6
6
Thirty data points were taken and analyzed by the GE-
Omega kinetics program.
31 R. L. Martin, J. Am. Chem. Soc., 1999, 121, 9459.
32 A. E. Shilov and G. B. ShulÏpin, Activation and Catalytic Reac-
tions of Saturated Hydrocarbons in the Presence of Metal Com-
plexes, Kluwer, Dordrecht, 2000, p. 267 †.
33 (a) J. W. Faller and H. Felkin, Organometallics, 1985, 4, 1488; (b)
M. J. Burk and R. H. Crabtree, J. Am. Chem. Soc., 1987, 109,
8025.
34 Y. Jean and O. Eisenstein, Polyhedron, 1988, 7, 405.
35 I. E.-I. Rachidi, O. Eisenstein and Y. Jean, New J. Chem., 1990,
14, 671.
36 J.-F. Riehl, Y. Jean, O. Eisenstein and M. Pelissier,
Organometallics, 1992, 11, 729.
Acknowledgements
We thank Dr Mark Burk for initiating work in this area,
which was supported by the NSF (R. H. C.), the DOE
(D. H. L.), University Montpellier 2 and the CNRS (H. G.,
O. E.).
37 (a) R. H. Schultz, A. A. Bengali, M. J. Tauber, B. H. Weiller, E. P.
Wasserman, K. R. Kyle, C. B. Moore and R. G. Bergman, J. Am.
Chem. Soc., 1994, 116, 7369 and references therein; (b) G. Wittig
and A. Hesse, Org. Synth., 1970, 50, 66; (c) G. W. Guchanan and
A. E. Gustafson, J. Org. Chem., 1973, 38, 2910.
38 (a) R. R. Schrock, S. W. Seidel, N. C. Mosch-Zanetti, K. Y. Shih,
M. B. OÏDonoghue and W. M. Davis, J. Am. Chem. Soc., 1997,
119, 11871; (b) D.-H. Lee, J. Chen, J. W. Faller and R. H.
Crabtree, Chem. Commun., 2001, 213.
39 J. N. Coalter, G. J. Spivak, H. Gerard, E. Clot, E. R. Davidson,
O. Eisenstein and K. G. Caulton, J. Am. Chem. Soc., 1998, 120,
9388.
40 J. N. Coalter, J. C. Bollinger, J. C. Hu†man, U. Werner-
Zwanziger, K. G. Caulton, E. R. Davidson, H. Gerard, E. Clot
and O. Eisenstein, New J. Chem., 2000, 24, 9.
41 M. Loza, J. W. Faller and R. H. Crabtree, Inorg. Chem., 1995, 34,
2937.
References
1
S. E. Bromberg, H. Yang, M. C. Asplund, T. Lian, B. K. McNa-
mara, K. T. Kotz, J. S. Yeston, M. Wilkens, H. Frei, R. G.
Bergman and C. B. Harris, Science, 1997, 278, 260.
2
T. A. Mobley, C. Schade and R. G. Bergman, J. Am. Chem. Soc.,
1995, 117, 7822.
3
4
R. N. Perutz and J. J. Turner, Inorg. Chem., 1975, 14, 262.
C. E. Brown, Y. Ishikawa, P. A. Hackett and J. Rayner, J. Am.
Chem. Soc., 1990, 112, 2530.
(a) C. Hall and R. N. Perutz, Chem. Rev., 1996, 96, 3125; (b) A. E.
Shilov and G. B. ShulÏpin, Chem. Rev., 1997, 97, 2879.
R. H. Crabtree, Chem. Rev., 1995, 95, 987.
R. H. Crabtree, Angew. Chem., Int. Ed. Engl., 1993, 32, 789.
B. K. McNamara, J. S. Yeston, R. G. Bergman and C. B. Moore,
J. Am. Chem. Soc., 1999, 121, 6437.
5
6
7
8
9
(a) R. A. Periana and R. G. Bergman, J. Am. Chem. Soc., 1986,
108, 7332; (b) P. J. Alaimo and R. G. Bergman, Organometallics,
1999, 18, 2707; (c) D. D. Wick, K. A. Reynolds and W. D. Jones,
J. Am. Chem. Soc., 1999, 121, 3974; (d) S. S. Stahl, J. A. Labinger
and J. A. Bercaw, J. Am. Chem. Soc., 1996, 118, 5961; (e) D. F.
Schafer and P. T. Wolczanski, J. Am. Chem. Soc., 1998, 120, 4881;
( f ) J. C. Green and C. N. Jardine, J. Chem. Soc., Dalton T rans.,
1998, 1057; (g) L. Johansson, M. Tilset, J. A. Labinger and J. E.
Bercaw, J. Am. Chem. Soc., 2000, 122, 10846.
42 D. Huang, H. Gerard, E. Clot, V. Young, Jr., W. E. Streib, O.
Eisenstein and K. G. Caulton, Organometallics, 1999, 18, 5441.
43 P. Burger and R. G. Bergman, J. Am. Chem. Soc., 1993, 115,
10462.
44 R. Jimenez-Catano and M. B. Hall, Organometallics, 1996, 15,
1889.
45 P. E. M. Siegbahn and R. H. Crabtree, J. Am. Chem. Soc., 1996,
118, 4442.
10 (a) D. W. Lee and C. M. Jensen, J. Am. Chem. Soc., 1996, 118,
8749; (b) R. N. Perutz and J. J. Turner, J. Am. Chem. Soc., 1975,
97, 4791; (c) D. C. Grills, X. Z. Sun, G. I. Childs and M. W.
George, J. Phys. Chem., 2000, 104, 4300; (d) A. Chernega, J.
Cook, M. L. H. Green, L. Labella, S. J. Simpson, J. Souter and
A. H. H. Stephens, J. Chem. Soc., Dalton T rans., 1997, 3225; (e)
E. F. Walsh, V. K. Popov, M. W. George and M. Poliako†, J.
Phys. Chem., 1995, 99, 12016.
46 M. D. Su and S. Y. Chu, Inorg. Chem., 1998, 37, 3400.
47 S. Sakaki, B. Biswas and M. Sugimoto, Organometallics, 1998, 17,
1278.
48 M. D. Su and S. Y. Chu, Organometallics, 1997, 16, 1621.
49 P. E. M. Siegbahn, J. Am. Chem. Soc., 1996, 118, 1487.
50 T. R. Cundari, J. Am. Chem. Soc., 1994, 116, 340.
51 H. Felkin, T. Fillebeen-Khan, R. Holmes-Smith and L. Yingrui,
T etrahedron L ett., 1985, 26, 1999.
52 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A.
Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A. Montgomery, Jr.,
R. E. Stratmann, J. C. Burant, S. Daprich, J. M. Millam, A. D.
Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi, V.
Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C.
Adamo, S. Cli†ord, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q.
Cui, K. Morokuma, D. K. Malick, A. D. Rabuck, K. Raghava-
chari, J. B. Foresman, J. Cioslowski, J. V. Ortiz, A. G. Baboul,
B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.
Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe,
P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, J. L. Andres,
C. Gonzalez, M. Head-Gordon, E. S. Replogle and J. A. Pople,
GAUSSIAN 98, Rev. A.7, Gaussian, Inc., Pittsburgh, PA, 1998.
53 A. D. Becke, J. Chem. Phys., 1993, 98, 5648.
54 C. Lee, W. Yang and R. G. Parr, Phys. Rev. B, 1988, 37, 785.
55 J. P. Perdew and Y. Wang, Phys. Rev. B, 1992, 45, 13244.
56 P. G. Hay and W. R. Wadt, J. Chem. Phys., 1985, 82, 299.
57 W. R. Wadt and P. G. Hay, J. Chem. Phys., 1985, 82, 184.
58 A. H. Hollwarth, M. B. Bohme, S. Dapprich, A. W. Ehlers, A.
Gobbi, V. Jonas, K. F. Kohler, R. Stegman, A. Veldkamp and G.
Frenking, Chem. Phys. L ett., 1993, 208, 237.
59 P. C. Hariharan and J. A. Pople, T heor. Chim. Acta, 1973, 28,
213.
60 A. Rosa, A. W. Ehlers, E. J. Baerends, J. G. Snijders and G. te
Velde, J. Phys. Chem., 1996, 100, 5690.
11 X. Z. Sun, D. C. Grills, S. Nikiforov, M. Poliako† and M. W.
George, J. Am. Chem. Soc., 1997, 119, 7521.
12 D. R. Evans, T. Drovetskaya, R. Bau, C. A. Reed and P. D. W.
Boyd, J. Am. Chem. Soc., 1997, 119, 3633.
13 S. Geftakis and G. E. Ball, J. Am. Chem. Soc., 1998, 120, 9953.
14 S. Geftakis and G. E. Ball, J. Am. Chem. Soc., 1999, 121, 6336.
15 K. Koga and K. Morokuma, Chem. Rev., 1991, 91, 823.
16 D. G. Musaev and K. Morokuma, in Advances in Chemical
Physics, ed. S. A. Rice and I. Prigogine, John Wiley & Sons, New
York, NY, 1996, vol. XCV, p. 61.
17 S. Niu and M. B. Hall, Chem. Rev., 2000, 100, 353.
18 S. Niu and M. B. Hall, J. Am. Chem. Soc., 1999, 121, 3992.
19 S. Zaric and M. B. Hall, J. Phys. Chem., 1997, 101, 4646.
20 M. D. Su and S. Y. Chu, J. Am. Chem. Soc., 1997, 119, 5373.
21 A. J. Lees and A. A. Purwoko, Coord. Chem. Rev., 1994, 132, 155.
22 K. M. Waltz and J. F. Hartwig, Science, 1997, 277, 211.
23 R. A. Periana and R. G. Bergman, Organometallics, 1984, 3, 508.
24 R. A. Periana, D. J. Taube, S. Gamble, H. Taube, T. Satoh and
H. Fujii, Science, 1998, 280, 560.
25 B. A. Arndtsen and R. G. Bergman, Science, 1995, 270, 1970.
26 M. J. Burk, R. H. Crabtree and D. V. McGrath, J. Chem. Soc.,
Chem. Commun., 1985, 1829.
27 M. J. Burk, D. V. McGrath and R. H. Crabtree, J. Am. Chem.
Soc., 1988, 110, 620.
28 R. M. Bullock, C. E. L. Headford, S. E. Kegley and J. R. Norton,
J. Am. Chem. Soc., 1985, 107, 727.
New J. Chem., 2001, 25, 1121È1131
1131