On the mechanism of (PCP)Ir-catalyzed acceptorless dehydrogenation of alkanes: A combined computational and experimental study

Article abstract of DOI:10.1021/ja012460d

Pincer complexes of the type (^RPCP)IrH₂, where (^RPCP)Ir is [η³-2,6-(R₂PCH₂)₂ C₆H₃] Ir, are the most effective catalysts reported to date for the "acceptorle

Full text of DOI:10.1021/ja012460d

Published on Web 08/31/2002
On the Mechanism of (PCP)Ir-Catalyzed Acceptorless
Dehydrogenation of Alkanes: A Combined Computational and
Experimental Study
Karsten Krogh-Jespersen,* Margaret Czerw, Nadine Summa, Kenton B. Renkema,
Patrick D. Achord, and Alan S. Goldman*
Contribution from the Department of Chemistry and Chemical Biology, Rutgers, The State
UniVersity of New Jersey, New Brunswick, New Jersey 08903
Received October 31, 2001
R
R
3
Abstract: Pincer complexes of the type ( PCP)IrH
2
2 2 2 6 3
, where ( PCP)Ir is [η -2,6-(R PCH ) C H ]Ir, are the
most effective catalysts reported to date for the “acceptorless” dehydrogenation of alkanes to yield alkenes
Me
2 2
and free H . We calculate (DFT/B3LYP) that associative (A) reactions of ( PCP)IrH with model linear
(
propane, n-PrH) and cyclic (cyclohexane, CyH) alkanes may proceed via classical Ir(V) and nonclassical
2
2 2
Ir(III)(η -H ) intermediates. A dissociative (D) pathway proceeds via initial loss of H , followed by C-H
addition to ( PCP)Ir. Although a slightly higher energy barrier (∆E ) is computed for the D pathway, the
calculated free-energy barrier (∆G ) for the D pathway is significantly lower than that of the A pathway.
Under standard thermodynamic conditions (STP), C-H addition via the D pathway has ∆G° ) 36.3 kcal/
Me
q
q
q
mol for CyH (35.1 kcal/mol for n-PrH). However, acceptorless dehydrogenation of alkanes is thermodynami-
cally impossible at STP. At conditions under which acceptorless dehydrogenation is thermodynamically
-
7
q
Me
possible (for example, T ) 150 °C and PH2 ) 1.0 × 10 atm), ∆G for C-H addition to ( PCP)Ir (plus a
molecule of free H ) is very low (17.5 kcal/mol for CyH, 16.7 kcal/mol for n-PrH). Under these conditions,
the rate-determining step for the D pathway is the loss of H ) 27.2
2
Me
q
q
2
Me
from ( PCP)IrH
2
with ∆G
on the A pathway is 35.2 kcal/mol
32.7 kcal/mol for n-PrH). At catalytic conditions, the calculated free energies of C-H addition are 31.3
and 33.7 kcal/mol for CyH and n-PrH addition, respectively. Elimination of H from the resulting “seven-
D
≈ ∆H
D
q
kcal/mol. For CyH, the calculated ∆G° for C-H addition to ( PCP)IrH
2
(
2
coordinate” Ir-species must proceed with an activation enthalpy at least as large as the enthalpy change
of the elimination step itself (∆H ≈ 11-13 kcal/mol), and with a small entropy of activation. The free energy
q
of activation for H
2
elimination (∆G
A
) is hence found to be greater than ca. 36 kcal/mol for both CyH and
n-PrH under catalytic conditions. The overall free-energy barrier of the A pathway is calculated to be higher
R
than that of the D pathway by ca. 9 kcal/mol. Reversible C-H(D) addition to ( PCP)IrH
2
is predicted to
lead to H/D exchange, because the barriers for hydride scrambling are extremely low in the “seven-
coordinate” polyhydrides. In agreement with calculation, H/D exchange is observed experimentally for several
deuteriohydrocarbons with the following order of rates:
cyclohexane-d12. Because H/D exchange in cyclohexane-d12 solution is not observed even after 1 week at
80 °C, we estimate that the experimental barrier to cyclohexane C-D addition is greater than 36.4 kcal/
6 6
C D > mesitylene-d12 > n-decane-d22 .
1
mol. This value is considerably greater than the experimental barrier for the full catalytic dehydrogenation
cycle for cycloalkanes (ca. 31 kcal/mol). Thus, the experimental evidence, in agreement with calculation,
strongly indicates that the A pathway is not kinetically viable as a segment of the “acceptorless”
dehydrogenation cycle.
Introduction
progress has been made over the past two decades toward the
development of solution-phase transition-metal-based systems
The selective catalytic functionalization of alkanes and alkyl
groups remains one of the foremost challenges in the field of
2
-7
capable of thermally effecting this reaction.
Most such
1
systems involve transfer of hydrogen to a sacrificial olefin (eq
catalysis. Dehydrogenation of alkanes to the corresponding
olefins is a reaction of particular potential value. Significant
(
2) (a) Baudry, D.; Ephritikine, M.; Felkin, H.; Holmes-Smith, R. J. Chem.
Soc., Chem. Commun. 1983, 788-789. (b) Burk, M. J.; Crabtree, R. H.;
Parnell, C. P.; Uriarte, R. J. Organometallics 1984, 3, 816-817. (c) Felkin,
H.; Fillebeen-Khan, T.; Gault, Y.; Holmes-Smith, R.; Zakrzewski, J.
Tetrahedron Lett. 1984, 25, 1279-1282. (d) Burk, M. J.; Crabtree, R. H.;
McGrath, D. V. J. Chem. Soc., Chem. Commun. 1985, 1829-1830.
*
To whom correspondence should be addressed. E-mail: goldman@
rutchem.rutgers.edu or krogh@rutchem.rutgers.edu.
(
1) For some recent reviews of homogeneously catalyzed hydrocarbon func-
tionalization, see: (a) Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.;
Peterson, T. H. Acc. Chem. Res. 1995, 28, 154-162. (b) Shilov, A. E.;
Shul’pin, G. B. Chem. ReV. 1997, 97, 2879-2932. (c) Sen, A. Acc. Chem.
Res. 1998, 31, 550-557. (d) Guari, Y.; Sabo-Etiennne, S.; Chaudret, B.
Eur. J. Inorg. Chem. 1999, 1047-1055.
(3) (a) Maguire, J. A.; Goldman, A. S. J. Am. Chem. Soc. 1991, 113, 6706-
6708. (b) Maguire, J. A.; Petrillo, A.; Goldman, A. S. J. Am. Chem. Soc.
1992, 114, 9492-9498. (c) Wang, K.; Goldman, M. E.; Emge, T. J.;
Goldman, A. S. J. Organomet. Chem. 1996, 518, 55-68.
11404
9
J. AM. CHEM. SOC. 2002, 124, 11404-11416
10.1021/ja012460d CCC: $22.00 © 2002 American Chemical Society

Acceptorless Dehydrogenation of Alkanes
A R T I C L E S
1
), but a smaller number of catalysts have been shown effective
or
5,7,8
for “acceptorless” dehydrogenation (eq 2).
(
PCP)IrH + RH f [(PCP)IrH R]
(4a)
(4b)
2
3
catalyst
alkane + acceptor
8
[
(PCP)IrH R] f (PCP)Ir(R)(H) + H2
3
alkene + H • acceptor
(transfer) (1)
2
Associative (A) pathway
catalyst
alkane
8 alkene + H2v
(“acceptorless”) (2)
In either case, the overall reaction is that shown in eq 5:
PCP)IrH + RH f (PCP)Ir(R)(H) + H2 (5)
Aside from the obvious and important advantage that a mole
of olefin is not sacrificed, acceptorless systems are, in principle,
simpler than transfer systems with respect to potential recycling
and separation problems.
The family of solution-phase catalysts that have proven most
effective for acceptorless dehydrogenation are those that have
also been found most effective for transfer-dehydrogenation,
(
2
Overall reaction of eq 4 or 5; pathway unspecified
The question whether the acceptorless dehydrogenation
proceeds dissociatively (D pathway, Ir(I)/Ir(III) couple) or
associatively (A pathway; possibly via Ir(V)-dihydride or
nonclassical Ir(III)-dihydrogen intermediates, or possibly via
a concerted pathway) has relevance extending beyond this
particular system. C-H bond activation by late metals is
generally assumed to occur via low oxidation states and to be
favored by increased electron density at the metal center. Yet,
Bergman has found that cationic Cp*Ir(III) species can add
R
R
viz., pincer catalysts of the type ( PCP)IrH2, where ( PCP)Ir is
η -2,6-(R2PCH2)2C6H3]Ir. Transfer-dehydrogenation is more
3
4-7
[
1
4
alkane C-H bonds, and Hall has calculated that this system
15
proceeds via an Ir(V) intermediate. Unlike the (PCP)Ir systems,
an Ir(I) pathway is not accessible in the Cp*Ir case, and the
alternative of σ-bond-metathesis by Cp*Ir(III) was calculated
to be less favorable than the Ir(V) pathway. Further, M(I)/M(III)
pathways (M ) Co, Rh, Ir) are assumed to be operative for a
wide range of important reactions catalyzed by group 9 metals.16
Establishing the operation of a M(III)/M(V) pathway for
dehydrogenation would have obvious implications for hydro-
genation with further extensions to hydroformylation, hydro-
silation, and other related catalyses.
amenable to experimental mechanistic study, mainly because
the acceptorless reaction must be conducted in an open system
while vigorously refluxing to purge the solution of H2. Thus,
we have, to date, focused our experimental mechanistic studies
on the transfer system, and we have elucidated a fairly detailed
free-energy profile. The reaction cycles of the two systems
undoubtedly share several steps, including â-hydrogen elimina-
9
tion by the presumed alkyl hydride intermediates ((PCP)Ir-
(alkyl)(H)) and subsequent loss of olefin to give (PCP)IrH2. The
key difference between the acceptorless and transfer systems
lies in the loss of H2 from the iridium dihydride. In the case of
the transfer systems, we have shown that the dihydride loses
hydrogen via insertion of the acceptor into the Ir-H bond
followed by C-H elimination. Hence, an Ir(I)/Ir(III) couple is
In this paper, we confirm, experimentally and computation-
ally, that C-H addition to the Ir(III) species can indeed occur.
However, the experimental results and a detailed computational
model lead firmly to the conclusion that the acceptorless
dehydrogenation system does not operate via an associative
pathway. Instead, loss of H2 occurs dissociatiVely via an Ir(III)/
Ir(I) pathway. Furthermore, this dissociative loss of H2 may be
the rate-determining step under the actual catalytic conditions,
a result which has significant implications for any attempts at
the development of related catalysts with higher turnover
operative in accord with independent computations by Hall and
by us.1
0,11
The acceptorless system, however, must obviously
lose free dihydrogen. This can, in principle, proceed either
dissociatively (with H2 loss prior to alkane addition; eqs 3a,b)
or associatively (H2 loss after, or concertedly with, alkane
addition; eqs 4a,b). Previous computational studies have pre-
dicted that the reaction proceeds associatively, specifically via
1
7
frequencies.
Results and Discussion
12,13
C-H oxidative addition to the dihydride:
Calculated Energies (∆E) of C-H Addition. Our compu-
Me
(
PCP)IrH + RH f (PCP)Ir + H + RH
(3a)
(3b)
tational modeling uses ( PCP) as the basic “pincer” ligand.
2
2
(
9) Goldman, A. S.; Renkema, K. B.; Kissin, Y.; Czerw, M.; Krogh-Jespersen,
K. 222nd ACS National Meeting; American Chemical Society, Washington,
DC: Chicago, IL, 2001; AN 2001:637418.
(
PCP)Ir + H + RH f (PCP)Ir(R)(H) + H
2
2
Dissociative (D) pathway
(10) Li, S.; Hall, M. B. Organometallics 2001, 20, 2153-2160.
(
11) Krogh-Jespersen, K.; Czerw, M.; Kanzelberger, M.; Goldman, A. S. J.
Chem. Inf. Comput. Sci. 2001, 41, 56-63.
(
4) (a) Gupta, M.; Hagen, C.; Flesher, R. J.; Kaska, W. C.; Jensen, C. M. Chem.
Commun. 1996, 2083-2084. (b) Gupta, M.; Hagen, C.; Kaska, W. C.;
Cramer, R. E.; Jensen, C. M. J. Am. Chem. Soc. 1997, 119, 840-841. (c)
Gupta, M.; Kaska, W. C.; Jensen, C. M. Chem. Commun. 1997, 461-462.
5) Xu, W.; Rosini, G. P.; Gupta, M.; Jensen, C. M.; Kaska, W. C.; Krogh-
Jespersen, K.; Goldman, A. S. Chem. Commun. 1997, 2273-2274.
6) Liu, F.; Pak, E. B.; Singh, B.; Jensen, C. M.; Goldman, A. S. J. Am. Chem.
Soc. 1999, 121, 4086-4087.
(12) (a) Niu, S. Q.; Hall, M. B. J. Am. Chem. Soc. 1999, 121, 3992-3999. (b)
Niu, S. Q.; Hall, M. B. Chem. ReV. 2000, 100, 353-405.
(13) Haenel, M. W.; Oevers, S.; Angermund, K.; Kaska, W. C.; Fan, H.-J.; Hall,
M. B. Angew. Chem., Int. Ed. 2001, 40, 3596-3600.
(
(
(14) Klei, S. R.; Tilley, T. D.; Bergman, R. G. J. Am. Chem. Soc. 2000, 122,
1816-1817.
(15) Niu, S. Q.; Hall, M. B. J. Am. Chem. Soc. 1998, 120, 6169-6170.
(16) Collman, J. P.; Hegedus, L. S.; Norton, J. R.; Finke, R. G. Principles and
Applications of Organotransition Metal Chemistry; University Science
Books: Mill Valley, CA, 1987; pp 523-576.
(
(
7) Liu, F.; Goldman, A. S. Chem. Commun. 1999, 655-656.
8) (a) Fujii, T.; Saito, Y. J. Chem. Soc., Chem. Commun. 1990, 757-758. (b)
Fujii, T.; Higashino, Y.; Saito, Y. J. Chem. Soc., Dalton. Trans. (1972-
(17) Presented in part at the Pacifichem 2000 Congress, December, 2000,
Honolulu; Abstract 90004907. A preliminary account of some of this work
has also been communicated as ref 11.
1
999) 1993, 517-520. (c) Aoki, T.; Crabtree, R. H. Organometallics 1993,
2, 294-298.
1
J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002 11405

A R T I C L E S
Krogh-Jespersen et al.
Figure 3. Illustration (full PCP ligand shown) of representative transition
state for C-H addition via the A pathway (eq 4a; TSE-endo) and the
corresponding addition product (intermediate E). See Figures 4 and 5 for
a more detailed illustration of the equatorial planes of these species and of
the other, isomeric, transition states and intermediates.
Figure 1. Energy profile for the dissociative pathway D of eq 3 (R ) Cy
)
cyclohexyl).
Figure 4. Schematic illustrations of the plane perpendicular to the P-Ir-P
Me
axis for the five minima located for C-H addition to ( PCP)IrH2
(
internuclear distances (Å) and energies for R ) Cy). CPCP is the carbon
Figure 2. Energy profile for the associative pathway A of eq 4 (R ) Cy).
The structures of the five calculated energy minima (in the plane
perpendicular to the P-Ir-P axis) are shown schematically. Pathways for
C-H addition (eq 4a) are indicated by blue lines, H2 elimination (eq 4b) is
indicated by red lines, and interconversions of isomers are indicated by
black lines.
forming the Ir-PCP linkage; CCy is the carbon forming the Ir-Cy linkage.
Ir-P bond lengths (not shown) lie in the narrow range 2.32-2.33 Å.
geometrical features associated with the coordination around
the Ir metal are shown for the five minima in Figure 4.
The geometrical parameters (Figure 4) show that potential
intermediates C and E are clearly classical Ir(V) C-H addition
Methyl groups on phosphorus presumably capture the electronic
effects of the bulky groups used experimentally (t-butyl,
i-propyl) quite well,18 but in terms of exerting steric effects these
may not be much better than the hydrogen atoms most often
used in theoretical studies involving (model) phosphines. We
will comment briefly on the Me versus t-Bu issue later in this
report. Because most experimental studies on the (PCP)Ir-
catalyzed dehydrogenation have employed cycloalkanes, we
use cyclohexane(CyH) as our exemplary model substrate in the
discussion; however, results for propane (n-PrH, 1° C-H bond),
benzene, and toluene (benzylic) C-H addition are reported as
well when appropriate.
products, while intermediates A, B, and D are nonclassical Ir-
2
(III)(η -H2) species (for convenience, we will refer to both
groups of intermediates as “seven-coordinate”). Minima A, C,
D, and E are almost isoenergetic, whereas minimum B lies ca.
4
kcal/mol higher. Importantly, interconversions among the
species A and C and between the species D and E are calculated
to have energy barriers that are lower (e2 kcal/mol) than the
activation energy for the reverse reaction (C-H elimination,
11
∼
3-5 kcal/mol). Furthermore, and most important later (see
below), the barrier to rotation of the dihydrogen ligands around
the Ir-(H2) axis is calculated to be even lower (e1 kcal/mol),
11
in accord with experimental and previously calculated values
2
2
The calculated value of ∆E for loss of dihydrogen from
for η -H rotation for closely related Ir(III)-dihydrogen com-
(MePCP)IrH2 (eq 3a) is 28.7 kcal/mol. Repeated attempts to
19
plexes.
locate a transition state for the dissociation were unsuccessful.
The reverse reaction apparently experiences no energetic barrier
in an idealized gas phase, and thus the calculated activation
Three transition states have been located for cyclohexane
C-H addition (equatorial C-H bond cleavage) to ( PCP)-
Me
Ir(H) . The respective structures are shown schematically in
2
q
energy for H2 loss, ∆E , is also 28.7 kcal/mol. Subsequent
Figure 5 (see Figure 3 for full structure of TSE-endo).
The transition state of lowest energy (TSE-endo, ∆E ) 20.1
kcal/mol) has the cyclohexane adduct approaching the ( PCP)-
Me
addition of cyclohexane to three-coordinate ( PCP)Ir has a
q
barrier of 7.5 kcal/mol. Hence, the overall reaction under study
Me
q
(
eq 5) has a calculated ∆ED ) 36.2 kcal/mol for RH ) CyH,
Ir(H)2 species between the two hydrides, trans to the carbon
when it proceeds via the dissociative pathway of eq 3 (Figure
forming the Ir-PCP linkage (CPCP), and it connects directly to
Me
1
). The barrier to addition of n-PrH to ( PCP)Ir is only 6.6
minimum E. The transition state labeled TS
A/B/C
lies only 2
q
q
kcal/mol, and eq 5 has a calculated ∆E ) 35.3 kcal/mol.
kcal/mol higher in energy (∆E ) 22.4 kcal/mol) than TSE-endo.
Me
Several configurational isomers can be envisioned as inter-
mediates on the associative pathway (eq 4). We have located
In TSA/B/C, cyclohexane is approaching ( PCP)Ir(H)2 from the
Me
(18) van W u¨ llen, C. J. Comput. Chem. 1997, 18, 1985-1992.
five minima with the composition ( PCP)IrH3(Cy) as indicated
(
19) Li, S. H.; Hall, M. B.; Eckert, J.; Jensen, C. M.; Albinati, A. J. Am. Chem.
in Figure 2 (see Figure 3 for full structure). Some essential
Soc. 2000, 122, 2903-2910 and references therein.
11406 J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002

Acceptorless Dehydrogenation of Alkanes
A R T I C L E S
Figure 5. Schematic illustrations of the plane perpendicular to the P-Ir-P
Me
axis for the three transition states for C-H addition to ( PCP)IrH2
(
internuclear distances (Å) and energies for R ) Cy). CPCP is the carbon
forming the Ir-PCP linkage; CCy is the carbon forming the Ir-Cy linkage.
Ir-P bond lengths (not shown) lie in the narrow range 2.32-2.33 Å.
Figure 6. Computed free-energy profiles for D and A pathways at STP (T
side (cis to CPCP), and this approach may, in principle, lead to
minimum A, B, or C. Judging from the structural parameters
of TSA/B/C (Figure 5), we may anticipate that minimum C is
the most likely product from this TS. Intrinsic reaction
coordinate (IRC) calculations (see Computational and Experi-
mental Section) were successfully employed to confirm this idea.
)
298 K; all reactants and products at P ) 1 atm) for the reaction of
cyclohexane with (MePCP)IrH . For the A pathway, only the lowest-energy
2
TS for C-H addition is shown (TSE-endo; see Figures 2-4). Likewise,
only the lowest energy of the two intermediates (D and E) preceded by
Me
TSE-endo is shown (see Figure 2). The TS for H2 loss from either ( PCP)-
IrH2 or D was not located; see ref 22 for a more detailed discussion of the
estimated values.
q
The transition state of highest energy, TSD/E-exo (∆E ) 26.8
Me
kcal/mol), also has the cyclohexane approaching ( PCP)Ir(H)2
Calculated Gibbs Free Energies (∆G) of C-H Addition:
Standard Thermodynamic Conditions. As expected, the
from the side (cis to CPCP) and could lead to minimum D or E.
On the basis of its geometry and confirmed by IRC calculations,
TSD/E-exo actually connects to minimum E (as does TSE-endo).
Thus, the located transition states all lead to formation of the
classical “seven-coordinate” isomers (C and E). As stated above,
formation of the nonclassical isomers A and D may proceed
from C and E, respectively, with very low barriers. However,
regardless of which C-H addition pathway is traversed, the
highest energy point along the overall A pathway corresponds
to elimination of H2 from this manifold of intermediates.
Because there is no barrier calculated for addition of H2 to
q
calculated differences between ∆E and ∆H (or between ∆E
q
and ∆H ) are generally found to be small, though they can be
as much as ca. 3 kcal/mol (mostly due to the inclusion of
vibrational zero-point energies in the enthalpies). However,
entropic factors result in very large differences between ∆H
q
q
and ∆G (or between ∆H and ∆G ).
We consider first the dissociative pathway (D). As stated
above, there is apparently no purely electronic energy barrier
Me
for H addition to ( PCP)Ir in the gas phase (eq 3a reversed).
2
Thus, the enthalpic barrier to H loss may be estimated to be
2
Me
q
equal to the thermodynamic enthalpy difference for eq 3a, ∆Hq
(
PCP)IrH(Cy), the activation energy (∆E ) for elimination
of H2 from any of the minima A-E is equal to the thermody-
namic energy difference (∆E) for the elimination step (12.5 kcal/
mol for RH ) CyH from minimum D). For the overall reaction
(eq 3a) ≈ ∆H (eq 3a). Being unable to locate a conventional
transition state structure from the electronic structure calcula-
q
tions, we cannot compute ∆S for H loss from the five-
2
q
q
(eq 4), ∆EA ) 28.4 kcal/mol when RH ) CyH (∆EA ) 28.0
coordinate species using standard statistical mechanical expres-
sions. However, we can estimate that the activation entropy for
H2 elimination must be very small on the basis of experimental
data obtained on closely related “seven-coordinate” systems,
kcal/mol when RH ) n-PrH).
It is not possible to predict which one of the two low-energy
associative transition states is really the most favorable for the
initial C-H addition step, because the computed energies are
too similar. However, the activation energy for the associative
q
21
that is, ∆S (eq 3a) ≈ 0. Overall, the free-energy difference
Me
q
between ( PCP)IrH2 and the TS for H2 elimination, ∆G (eq
3a), is thus estimated to be equal to the enthalpy difference,
∆H (eq 3a). Under standard thermodynamic conditions (T )
298 K, all reactants and products at P ) 1 atm), the computed
(
A) path, ∆EAq ) 28.4 kcal/mol, is determined by the H2
elimination step, and it is distinctly lower (by about 8 kcal/
q
mol) than that calculated for the dissociative (D) pathway, ∆ED
q
)
36.2 kcal/mol. But reaction rates are of course dependent
value is ∆H° (eq 3a) ) 27.0 kcal/mol ≈ ∆G° (eq 3a).
q
q
q
Me
upon differences in Gibbs free energy, ∆G ) ∆H - T(∆S ),
The TS for C-H addition to ( PCP)Ir (eq 3b) has a
q
20
and not on internal energy differences (∆E ). As will be shown
below, consideration of free energies leads to the conclusion
that the dissociative pathway is the more favorable, even under
standard thermodynamic conditions. Furthermore, the free-
energy difference favoring the D pathway becomes quite
substantial when the actual experimental conditions of the
catalytic reaction (elevated temperature, low pressure of dihy-
drogen, bulky groups in (PCP)Ir) are carefully considered and
accounted for.
significantly higher free energy at STP than does the TS for H
2
Me
q
loss from ( PCP)IrH (eq 3a). Thus, the ∆G° (eq 3a) value
2
estimated above does not enter into the overall activation free-
energy value for the complete D pathway, which is calculated
q
to be ∆G ° ) 36.3 kcal/mol for RH ) CyH at STP (Figure
D
(21) Activation entropies for H
2
loss have been determined for a series of “seven-
coordinate” iridium complexes, (PR ) IrX(H) (H ), which are particularly
3 2 2 2
closely related to the nonclassical intermediates calculated in this work.
q
Six values were determined for ∆S , ranging from 0.7 ( 1.0 to 3.7 ( 1.1
eu: (a) Hauger, B. E.; Gusev, D.; Caulton, K. G. J. Am. Chem. Soc. 1994,
1
16, 208-214. (b) Bakhmutov, V. I.; Vorontsov, E. V.; Vymenits, A. B.
(
20) We use E to represent the internal purely electronic energy. The thermal
electronic energy, Eth, may be obtained by correcting E for the effects of
zero-point vibrational energies and finite temperature via the translational,
Inorg. Chem. 1995, 34, 214-17. Another highly relevant study was
conducted on the neutral “seven-coordinate” complexes Ru(H) (H )(PR
from these complexes
2
2
3 3
)
q
4 3 3 2
and Os(H) (PR ) . ∆S values for elimination of H
rotational, and vibrational partition functions (Eth ) E + E
t
+ E
r
+ E
v
).
were determined to be 3 ( 1 and -2 ( 4 eu, respectively: (c) Halpern, J.;
Cai, L.; Desrosiers, P. J.; Lin, Z. J. Chem. Soc., Dalton Trans. (1972-
1999) 1991, 717-723.
The enthalpy H is then H ) Eth + PV ) Eth + RT, and the Gibbs free
energy is G ) H - TS.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002 11407

A R T I C L E S
). The computed value of ∆GD° (eq 5) includes an overall
value of ∆SD° ≈ -13 eu relative to the reactants (( PCP)-
IrH2 + CyH), reflecting the sum of the favorable entropy
attributable to the elimination of a free H2 molecule (26.8 eu at
Krogh-Jespersen et al.
q
6
kcal/mol higher than the barrier to the prior C-H addition step
(eq 4a) in the A pathway.
q
Me
Calculated Gibbs Free Energies (∆G) of C-H Addition:
Experimental Conditions. As just noted, under standard
thermodynamic conditions, ∆G° is calculated to be slightly
1
atm H2) and the quite unfavorable entropic contribution of
q
the C-H addition step (∼ -40 eu). The computed activation
lower for the D pathway than for the A pathway. The D pathway
would thus be predicted to be operative, but the computed
difference in free energies is so small (∼1 kcal/mol) that it is
certainly testing the confidence limits of the calculations.
However, the actual catalytic reaction certainly is not (and
cannot be) conducted under standard conditions (298 K, 1 atm
in all reactants and products). ∆G° of alkane dehydrogenation
is highly positive under 1 atm H2 (∼16 kcal/mol), and an
Me
entropy for C-H elimination from ( PCP)Ir(Cy)(H) is only
q
∆
S° (eq 3b) ) 0.3 eu, a very small value which provides some
q
additional support for the estimate ∆S° (eq 3a) ≈ 0 for H2
Me
loss from ( PCP)IrH2 made above. Apparently, the entropic
gain associated with the group (H2 or CyH) leaving the (PCP)-
Ir species is not realized until after the transition state has been
reached.
Turning now to the associative pathway (eq 4), we find, as
extremely low steady-state concentration of H (less than the
2
q
expected, that ∆S° for C-H addition is significantly negative,
equilibrium concentration value for eq 2) must be maintained
about -46 to -48 eu. The large difference between this value
and that for C-H addition via the dissociative pathway (∆S°
in solution or the reaction (eq 2) will proceed to the left, rather
than to the right. Furthermore, the high reaction endothermicity
inevitably requires elevated temperatures, which also help favor
the reaction thermodynamics; in no case has acceptorless
dehydrogenation been reported at temperatures below 151 °C
(and at that temperature only for the “easily” dehydrogenated
q
≈
-13 eu) reflects the absence of a free molecule of H2 in the
A pathway and a small additional contribution of ca. -7 eu
resulting from the greater steric crowding in the TS for C-H
Me
Me
addition to ( PCP)IrH2 as compared with addition to ( PCP)-
Ir. Factoring in the very negative entropy terms for C-H
addition via the A pathway produces large free energies of
activation. The TS of lowest free energy for C-H addition via
the A pathway (eq 4a) is calculated to be 32.2 kcal/mol for
cyclohexane (TSE-endo). This value is substantial, but the TS
for loss of H2 (eq 4b) from the resulting “seven-coordinate”
adducts (D/E) has an even higher enthalpy - eVen if it is
assumed to be no higher in enthalpy than the reaction product,
5
,7,8,24-26
substrate cyclooctane).
Clearly, to obtain free-energy parameters computationally that
are relevant to experiment, it will be necessary to model the
reactions as closely as possible to experimental conditions. From
the thermodynamic literature, the maximum steady-state con-
centration of H2 required for alkane dehydrogenation to produce
significant yield of olefin can be calculated to correspond to a
-14
H2 pressure of ca. 10
atm at 25 °C. At the elevated
Me
that is, the alkyl hydride. A barrier for addition of H2 to ( PCP)-
temperatures of the actual experiments, more relevant values
Ir(Cy)H cannot be found on the potential energy surface, and it
is thus not possible to directly calculate the entropy or free
-7
-5
are ca. 1.0 × 10 atm at 150 °C and ca. 1.0 × 10 atm at 200
25-27
°C.
We have used these two temperature/pressure combina-
energy of the transition state for H2 elimination (TSH elim) from
2
tions to calculate free-energy profiles for each of the hydro-
carbon substrates that we have studied (Table 1).
the “seven-coordinate” intermediate. In analogy to the consid-
erations made above, we assume that TSH elim from D/E has
2
It should be noted that any calculated differences in ∆G and
entropy similar to that of either intermediate D or TSE-endo (the
TS for C-H elimination from D/E) and obtain a free energy of
activation ∆G° ) 37.4 kcal/mol for H2 elimination.21,22
Therefore, this step (eq 4b) is apparently rate-determining for
the A pathway on the potential energy surface (Figure 2) as
well as on the Gibbs free-energy surface (Figure 6).
q
∆
G resulting from the application of a nonstandard pressure
of dihydrogen are completely rigorous in terms of fundamental
thermodynamic laws. They follow from the simple relationship
for the pressure dependence of the chemical potential for a
species X:
q
q
It is worth noting that the value of ∆GA° obtained above,
µ ) µ ° + RT[ln(P /P °)]
(6)
X
X
X
X
37.4 kcal/mol for eq 5 (RH ) CyH), is derived from calculations
which indicate that no point along the A pathway is of higher
Me
enthalpy than the products, ( PCP)IrRH + H2. Nevertheless,
∆
(
(23) Determination of the thermodynamics for reaction 5 involves the comparison
of two very similar species, that is, (PCP)Ir(R)(H) and (PCP)Ir(H)
GA°q is quite high due to the combination of the high
thermodynamic) reaction enthalpy of eq 5 (25.3 kcal/mol for
R ) Cy; 24.9 kcal/mol for R ) n-Pr) and the substantial
unfavorable entropic contribution. Because the entropic contri-
bution is essentially unavoidable in any bimolecular pathway,
the value thus obtained for ∆GA° may be regarded as a quite
This barrier (∆GA° ) 37.4 kcal/mol)
is found to be ca. 1 kcal/mol higher than ∆GD° , and it is ca. 5
2
.
Electronic structure calculations (DFT or otherwise) are well known to
give highly reliable results, when comparisons of similar species are made.
It is worth noting that our values for the thermodynamics of this reaction
are in good agreement with related calculations by Hall (refs 12 and 13),
and they have been calibrated and found to be quite accurate for the
t-Bu
t-Bu
2
observable species ( PCP)IrH and ( PCP)Ir(Ph)H (ref 9).
21
(
24) The dehydrogenation enthalpy of cyclooctane is approximately 23.8(5) kcal/
mol as determined by either direct measurement of hydrogenation or on
q
25,26
the basis of enthalpies of formation.
This can be compared, for example,
reliable lower limit.2
1,23
q
with the following enthalpies (kcal/mol) for the formation of alkenes from
2
5,26
q
n-butane: trans-2-butene, 27.5; cis-2-butene, 28.5; 1-butene, 30.1.
25) NIST Standard Reference Database Number 69, 1996, http://webbook.nist-
gov/chemistry/.
(
.
(
22) If we assume that the entropy of the transition state for H
2
elimination,
(26) Stull, D. R.; Westrum, E. F.; Sinke, G. C. The Chemical Thermodynamics
of Organic Compounds; Robert E. Kreiger Publishing: Malabar, FL, 1987.
(27) For example, for the dehydrogenation of n-hexane(liq) to give 1-hexene
and trans-2-hexene, the values of ∆G at 25 °C are 20.97 and 18.37 kcal/
TSH2elim, is the same as that of intermediate D, we can estimate the free-
energy difference on the basis of the enthalpy difference between the two
states, which gives a value of 36.0 kcal/mol for TSH2elim. The transition
state for C-H elimination, TSE-endo, presents an alternative model for the
entropy of TSH2elim. Extrapolating from the enthalpy difference in this case
gives a free energy of 38.8 kcal/mol. The average of these two values,
2
5
mol, respectively. At 1.0 M hexene and 10 M n-hexane, the equilibrium
-
15
-13
pressures of H
2
at 25 °C are thus 4.2 × 10
and 3.4 × 10
atm,
-
8
respectively. At 150 °C, the respective pressures are 1.5 × 10 and 3.4 ×
-7
-7
3
7.4 kcal/mol, is shown in Figure 7. The basis for this estimation method
is well established, particularly for related “seven-coordinate” iridium
complexes (see ref 21).
10 atm. At 200 °C, the respective pressures are 6.8 × 10 and 1.1 ×
-5
-7
10 atm. Throughout the course of this work, we use 1.0 × 10 atm at
-5
150 °C and 1.0 × 10 atm at 200 °C.
11408 J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002

Acceptorless Dehydrogenation of Alkanes
A R T I C L E S
Table 1. Calculated Free Energies of Activation for A and D
Pathways (H2 Elimination and C-H Addition), and Free Energies
of Lowest Intermediate for the A Pathway
D path
A path
T
(°C)
P
H
H
2
C−H
addc
C−H
addd
H
2
elimf
2
hydrocarbon
(atm)a
dissb
IrH
3
Re
benzene
benzene
benzene
toluene
toluene
toluene
propane
propane
25
150
200
25
150
200
25
150
200
25
150
200
STP
10
10
27.0
27.2
27.3
27.0
27.2
27.3
27.0
27.2
27.3
27.0
27.2
27.3
25.4
7.1
9.7
25.2
24.6
26.1
29.1
28.9
30.6
32.7
33.7
35.2
32.2
31.3
32.8
14.9
14.2
15.6
20.5
20.2
21.9
26.3
25.5
26.9
27.3
25.6
26.8
27.9
27.7
29.2
33.2
33.3
35.1
37.7
36.6
38.1
37.4
36.0
37.3
-
-
7
5
STP
31.5
13.7
16.5
35.1
16.7
19.2
36.3
17.5
20.0
-
-
7
10
10
5
STP
-
-
7
10
10
5
propane
Figure 7. Free-energy profiles for D and A pathways at T ) 150 °C, PH
cyclohexane
cyclohexane
cyclohexane
STP
2
-7
-
-
7
) 10 atm, and [cyclohexane] ) 10 mol/L for the reaction of cyclohexane
10
10
Me
5
with ( PCP)IrH2. See caption, Figure 6, for details. Note that, under these
conditions (in contrast with STP; Figure 6), the rate-determining step for
a
the D pathway is loss of H2, not C-H addition.
At STP, P ) 1 atm for all reactants and products. At other conditions,
PH is the value given, and a value of PRH is used to give the density of
2
C-H bonds in solution of the corresponding hydrocarbon (see text for
q
small changes due to the temperature-dependence of ∆H ), the
b
discussion). Estimated free energy of activation for loss of H2 from
Me
Me
TS for C-H addition to ( PCP)Ir (eq 3b) is calculated to be
at a much lower free energy under “experimental” conditions
17.5 kcal/mol, relative to ( PCP)IrH2 and cyclohexane) than
is the TS for H2 loss (eq 3a; 27.2 kcal/mol, see Figure 7). We
conclude that under conditions where acceptorless dehydroge-
nation (eq 2) is experimentally possible, the rate-determining
step for the D pathway is elimination of H2 (eq 3a, rather than
(
PCP)IrH2, equal to the enthalpy of H2 loss (see ref 21 and text for
c
Me
discussion). Free energy (relative to ( PCP)IrH2 plus RH) of the TS for
R-H addition to ( PCP)Ir plus a free molecule of H2. ∆Gq of C-H
Me
d
Me
(
Me
e
addition to ( PCP)IrH2 for the TS of lowest energy. Free energy (relative
Me
to ( PCP)IrH2 plus RH) of the lowest energy intermediate (with stoichi-
Me
ometry [( PCP)IrH3R]) preceded by the lowest energy TS for C-H
addition to ( PCP)IrH2. f Estimated free energy (relative to ( PCP)IrH2
plus RH) of the “TS” for H2 elimination via the associative path. The value
is assumed to have an enthalpy no higher than the enthalpy of the products.
The entropy is then estimated as an average of the entropy value calculated
Me
Me
q
eq 3b) with ∆GD ) 27.2 kcal/mol.
Me
for the preceding intermediate ([( PCP)IrH3R]) and that for the preceding
The A pathway is also favored by 6.7 kcal/mol from the
higher solution density of cyclohexane C-H bonds. However,
as compared with results obtained at STP, this factor is offset
(almost exactly, by coincidence) by the higher temperature (150
°C) which disfavors the entropically unfavorable addition
TS for C-H addition/elimination (see ref 21 and text for discussion).
Thus, for example, for a state that includes free H2, the use of
-
7
PH ) 1.0 × 10 atm instead of 1.0 atm results in a free-energy
2
-7
change equal to T(R ln 10 ) or, equivalently, an entropy change
of R ln 10 ≈ 32.0 eu.
-
7
q
reaction. At 150 °C, the computed value of ∆G for eq 4a (RH
We should also include statistical terms to account for the
increased density of C-H bonds in alkane solution vis- a` -vis
gas phase at 1 atm. Neat alkane solvent has a concentration of
ca. 10 mol/L (e.g., 9.3 mol/L for cyclohexane), whereas the
standard calculations assume 1 atm (∼0.041 mol/L at STP).
The gas-liquid concentration correction is equivalent to an
entropy correction of R ln(10/0.041). Finally, because the basic
statistical mechanical expressions only take into account a single
C-H bond of the hydrocarbon, we must also correct for the
statistically greater number of C-H bonds available per
molecule that may add to iridium. Thus, when the addendum
is cyclohexane (12 C-H bonds), we include an overall C-H
bond density factor of R ln[12 × (10/0.041)]. Although this
entropic correction is less rigorous than the one used for the
dihydrogen pressure, it is also less important in the context of
differentiating between the A and D pathways. For both
pathways, the respective C-H addition transition states are
favored by the same factor, which thus cancels out in evaluating
the difference between these two transition states.
) CyH) is 31.3 kcal/mol. The TS for H2 loss from the resulting
“seven-coordinate” species has an even higher free energy of
36.0 kcal/mol, relative to the iridium-dihydride and cyclohex-
ane (Figure 7). Thus, the computed ∆GA is 36.0 kcal/mol for
eq 5 (RH ) CyH), and the calculated difference between the
overall free-energy barriers of the two pathways, ∆∆GA/D , is
substantial, 8.8 kcal/mol for RH ) CyH, and is unequivocally
in faVor of the D pathway. Analogous calculations give a similar
value for propane (1° C-H), ∆∆GA/D ) 36.6 - 27.2 ) 9.4
q
q
q
kcal/mol (favoring D; Table 1).
Bulky Alkyl Groups. It is noteworthy that the very large
calculated preference for the D pathway results even though
the calculations are conducted using methyl groups, rather than
t-butyl or i-propyl groups, on the PCP phosphorus atoms.
Bulkier alkyl (or aryl) groups would undoubtedly serve to further
increase the free energy of the crowded transition states
encountered on the A pathway. Even if the TS for C-H addition
R
to ( PCP)Ir (i.e., the second step in the D pathway) could be
equally disfavored by steric factors, which seems highly unlikely
In the case of the dissociatiVe pathway (D), the use of
experimental temperature and hydrogen pressure conditions
results in a significantly lower calculated value of ∆G for C-H
because it is less crowded than the TS for C-H addition to
R
q
( PCP)IrH2, this would have no effect on the overall ∆GD until
steric factors become so severe that C-H addition (eq 3b) rather
than H2 loss becomes rate-determining for the D pathway. This
would require ca. 10 kcal/mol of additional steric repulsion in
the case of RH ) CyH (Figure 7). Adding this magnitude of
steric energy to the free energies of both A and D transition
q
addition, because the relevant transition state includes a molecule
of free H2. When compared with data obtained at STP, the use
-
7
of PH ) 1.0 × 10 atm and T ) 150 °C (423 K) results in a
2
q
decrease of 12.7 kcal/mol for ∆G . The entropic factor account-
q
ing for the density of C-H bonds yields a further decrease of
states for C-H addition would increase the value of ∆∆GA/D
6.7 kcal/mol (cyclohexane). As a result of these corrections (plus
to approximately 19 kcal/mol in favor of the D pathway for
J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002 11409

A R T I C L E S
Krogh-Jespersen et al.
Table 2. Thermodynamic and Activation Enthalpies and Entropies
of the Relevant Addition Reactions at STP
Scheme 1. General Reaction Scheme Used for Kinetic Modeling
To Obtain the Rate of H/D Exchange between
t-Bu
Deuterohydrocarbons and (
Section for Details)
PCP-dn)IrH2 (See Experimental
q
q
q
addendum
Ir species
∆H
∆H
∆S
∆S
∆G
∆G
Me
H2
benzene
toluene
propane(1°)
cyclohexane
benzene
toluene
propane(1°)
cyclohexane
(
(
(
(
(
(
(
(
(
PCP)Ir
PCP)Ir
PCP)Ir
PCP)Ir
PCP)Ir
PCP)IrH2 13.2
PCP)IrH2 14.3
PCP)IrH2 19.4
PCP)IrH2 18.4
0.0 -27.0 -26.8 -26.8 8.0 -19.0
-5.4 -12.3 -39.1 -30.6 6.3 -3.2
Me
Me
Me
Me
Me
Me
Me
Me
0.9 -7.5 -45.2 -43.3 12.5
4.7 -2.2 -38.0 -37.0 16.0
5.4 -1.7 -39.8 -40.0 17.2
4.3
8.8
10.2
15.0
20.5
26.3
27.3
1.7 -44.8 -44.7 25.2
5.7 -50.0 -49.8 29.1
13.5 -44.8 -43.0 32.7
15.1 -46.5 -40.8 32.2
than C-H elimination (eq 4a, reverse), then the rate of H/D
exchange will be faster than the rate of eq 5.
Me
The addition of benzene to ( PCP)Ir is calculated to be much
more thermodynamically favorable than addition of alkanes (see
Table 2); accordingly, the corresponding substitution of H2 by
R-H (eq 5) is thermodynamically much more favorable for
either cyclohexane (see Figure 7) or propane. Any additional
steric repulsion, if it affected the A and the D transition states
equally, would not change this very substantial value.
Experimental Determination of the Barrier to the A
Pathway: H/D Exchange. The calculations clearly predict that
the A pathway has a much greater barrier than the D pathway
Me
benzene. Addition of RH to ( PCP)IrH2 to produce the
Me
(
PCP)IrH3R intermediate is also thermodynamically more
favorable for benzene (by 13.4 kcal/mol) than for cyclohexane,
and the associative transition state for benzene addition (25 °C)
is calculated to be 5.2 kcal/mol lower than that for cyclohexane.
These results are in accord with all previous experimental
comparisons revealing that addition of benzene to late metal
centers is both kinetically and thermodynamically much more
(for eq 5). Furthermore, a considerably greater barrier is
predicted for the A pathway (36.0 kcal/mol for cyclohexane and
Me
(
PCP)IrH2 at 150 °C) than is experimentally observed for the
actual, overall, catalytic reactions (for example, 30.0 kcal/mol
2
9
q
favorable than addition of alkanes. The barrier (∆G° ) for
t-Bu
for cyclooctane and (
PCP)IrH2 at 150 °C). The problem of
Me
benzene addition to ( PCP)IrH2 is calculated to be 24.6 kcal/
experimentally testing the conclusions of these calculations is
quite intriguing. The reaction in question (eq 5) yields an
intermediate, which has never been observed when RH )
alkane; indeed, formation of the alkyl hydrides (with loss of
H2) is calculated to be endothermic, and, in addition, the alkyl
hydrides are presumably subject to relatively rapid â-hydrogen
elimination. Even for an observable model species, such as
mol, when corrections are applied for the C-H bond concentra-
tion of neat benzene. The reverse reaction (elimination of
benzene from the lowest energy intermediate, minimum A in
this case (see Figure 2)) has a barrier of ca. 11.8 kcal/mol, which
is substantially greater than the barrier calculated for rotation
1
9
Me
of H2. Thus, H/D exchange between C6D6 and ( PCP)IrH2
q
is calculated to occur with a barrier of ∆G° ) 25.2 kcal/mol
at 25 °C (∆G ) 20.9 kcal/mol when the concentration of C-H
t-Bu
28
(
PCP)Ir(Ph)(H), eq 5 is highly endothermic (calculated ∆H
-12.3 - (-27.0) ) 14.7 kcal/mol; Table 2), which
q
)
bonds in neat benzene is taken into account) before adjustments
are made for isotope effects and for statistical terms that would
account for nonproductive addition/elimination.
effectively precludes kinetic measurements. Thus, direct mea-
surement of the barrier of eq 5 seems a formidable task.
However, the nature of the associative pathway, irrespective of
q
Experimentally, H/D exchange is observed to occur between
the calculated values of ∆G , suggests a way to determine at
t-Bu
C6D6 and (
PCP)IrH2. The kinetic analysis of the data is
least an upper limit to its rate. Because the calculated associative
pathway involves intermediate nonclassical dihydrogen com-
plexes, or intermediates that are in rapid equilibrium with
dihydrogen complexes, (PCP)IrH2 should undergo H/D ex-
change with a deuteriohydrocarbon solvent at a rate at least
comparable to that of the associative reaction.
slightly complicated by incorporation of deuterium into the
t-Bu
t-butyl groups. The concentration of C6D5H, (
(
PCP-dn)IrH2,
t-Bu
PCP-dn)IrHD, as well as the total fraction of protium in
1
the PCP t-butyl groups, was monitored over time by H NMR.
t-Bu
Fortunately, there is no exchange with the
PCP methylene
groups, which provide a good integration standard for the
determination of percent-deuteration at the hydride and t-butyl
positions. Also fortunate is the fact that the dihydride and the
mixed hydride-deuteride signals are easily resolved in the
RD + (PCP)IrH f [(PCP)IrH D(R) h
2
2
2
(
PCP)IrH(η -HD)(R)] f RH + (PCP)IrHD (7)
1
Note that actual observation of H/D exchange (eq 7) does not
demonstrate that a full associative exchange has occurred (eq
hydride region of the H NMR spectrum. Analysis by kinetics
simulation programs gives a best fit to all of the species
-
5
5
), because the step after C-H addition (i.e., H2 or HD
monitored with a rate constant for eq 7 of kRD/IrH ) 6 × 10
-
1
-1
elimination from the adduct) may not necessarily have taken
place. In fact, as noted above, the calculations suggest that the
barrier to loss of H2 from the “seven-coordinate” alkane C-H
addition products (eq 4b) is higher than that for C-H addition/
elimination, eq 4a (see Figures 6 and 7), though the small
differences are perhaps within the uncertainty level of the
calculations. Thus, the rate of H/D exchange (eq 7) should
provide a good upper limit estimate to the rate of associative
M s . A simplified reaction scheme for the fit is shown in
Scheme 1; see Experimental Section for details. The fit is
insensitive to the value of kt-Bu/IrD as long as the value used is
much larger than kRD/IrH.
(
29) For reviews of alkane C-H bond activation by organometallic complexes,
see ref 1 and (a) Jones, W. D.; Feher, F. J. Acc. Chem. Res. 1989, 22,
91-100. (b) Crabtree, R. H. Chem. ReV. 1985, 85, 245. (c) Halpern, J.
Inorg. Chim. Acta 1985, 100, 41-48. Some more recent papers that have
addressed selectivity in particular and provide good lead references
include: (d) Harper, T. G. P.; Desrosiers, P. J.; Flood, T. C. Organometallics
q
q
exchange (eq 5); that is, ∆G7 provides a lower limit for ∆G5 .
If H2 elimination from the C-H adduct (eq 4b) is indeed slower
1
990, 9, 2523-2528. (e) Bennett, J. L.; Vaid, T. P.; Wolczanski, P. T.
Inorg. Chim. Acta 1998, 270(1-2), 414-423. (f) Wick, D. D.; Jones, W.
D. Organometallics 1999, 18, 495-505. (g) McNamara, B. K.; Yeston, J.
S.; Bergman, R. G.; Moore, C. B. J. Am. Chem. Soc. 1999, 121, 6437-
6443.
(
28) Kanzelberger, M.; Singh, B.; Czerw, M.; Krogh-Jespersen, K.; Goldman,
A. S. J. Am. Chem. Soc. 2000, 122, 11017-11018.
11410 J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002

Acceptorless Dehydrogenation of Alkanes
A R T I C L E S
Table 3. Experimental Rates and Free Energies of H/D Exchange (Eq 7, kRD/IrH Extrapolated from Scheme 1), and Corresponding
Calculated ∆G Values of C-H Addition to ( PCP)IrH2 (Eq 4a)a
q
Me
calculated (MePCP)IrH
experimental (t-BuPCP)IrH
rate7a (s-1
2
2
q
7(extrap) (M-1 s-1
q
R−H
∆G4a (kcal/mol)
R−H
T (°C)
k
)
)
∆G7a (extrap) (kcal/mol)
-4
-5
-5
-6
-3
-3
-4
-5
benzene
toluene
C3H7-H(1°)
cyclohexane
20.9 at 25 °C
23.5 at 25 °C
29.1 at 25 °C
27.6 at 25 °C
(21.1 at 30 °C)
(27.5 at 110 °C)
(32.3 at 130 °C)
(32.3 at 180 °C)
benzene
30
110
130
180
1.2 × 10
7.2 × 10
8.5 × 10
<3 × 10
2.8 × 10
1.0 × 10
21.3
27.9
29.5
mesitylene
n-C10H21-H
cyclohexane
8.7 × 10
<2.8 × 10
>36.4
a
k7 is extrapolated from the reaction scheme indicated (in simplified form) in Scheme 1 (see Table 5 for full details). The value thus obtained (a second-
order rate constant) is then multiplied by a statistical factor of 2.0 to allow for unproductive addition/elimination (i.e., elimination of RD, rather than of RH,
from the addition product) and multiplied by solvent concentration, to give a pseudo first-order rate constant, rate7a. These values are then substituted into
q
the Eyring equation to give free energies of activation for eq 7a, which are directly comparable with the calculated values for C-H addition (∆G4a ).
We estimate a very large error in the determination of these
rate constants due to experimental error and possible errors in
assumptions concerning isotope effects (see Computational and
Experimental Section). Nevertheless, thanks to the generously
forgiving nature of logarithms, the final error in the experimental
determination of ∆G is reasonably small and certainly suf-
ficiently small to provide a useful value for comparison with
the calculated data.
H/D exchange with other hydrocarbons, specifically mesity-
lene-d12 and decane-d22, requires significantly higher temper-
atures than C6D6 to achieve conveniently measured rates. The
values for the corresponding exchange-reaction rate constants
are more easily determined than for C6D6, because kRD/IrH is
certainly very slow in these cases relative to kt-Bu/IrD, and the
kinetics analysis becomes essentially a one-parameter fit. The
resulting rates are shown in Table 3.
fully confirm the major conclusion drawn from the electronic
structure calculations reported above (i.e., that the catalytic
reaction proceeds via a dissociative pathway), when they are
directly compared with experimentally measured catalytic rates
-
1
-3
(Table 4). The catalytic turnover rate is 11 h (3.1 × 10
q
-1
t-Bu
s ) for cyclooctane dehydrogenation catalyzed by ( PCP)IrH2
at 151 °C, corresponding to an overall activation barrier of 30.0
kcal/mol. For cyclodecane dehydrogenation, the turnover rate
-
1
-2 -1
is 90 h (2.5 × 10 s ) at 201 °C, corresponding to a barrier
of 31.7 kcal/mol. The failure of cyclohexane-d12 to undergo any
t-Bu
observable H/D exchange with ( PCP)IrH2, even after 1 week
at 180 °C, indicates that the barrier for C-H addition to
t-Bu
(
PCP)IrH2 is significantly higher than the experimentally
determined barrier to the overall catalytic reaction. Note that
cyclohexane has been reported by Bergman to undergo addition
more readily than cyclooctane, in separate reactions with two
3
2
Our most significant experimental result regarding H/D
exchange is a negative one. Even at 180 °C, cyclohexane-d12
undergoes no observable H/D exchange with (t-BuPCP)IrH2 after
different iridium species. Because a catalytic cycle obviously
cannot proceed more rapidly than any single step in the cycle,
these results are clearly inconsistent with an associative pathway
for the catalysis. Or, at the very least, they are inconsistent with
the associative pathways calculated in this work by us and
1
to (
week, indicating that addition of cyclohexane-d12 C-D bonds
t-Bu
PCP)IrH2 does not occur on this time scale. We can
1
2
therefore determine a lower limit for the free energy of activation
calculated by others for smaller substrates, because such
pathways are also calculated to lead to H/D exchange. Thus,
cycloalkane dehydrogenation must proceed via the D pathway
with an overall barrier of ca. 31 kcal/mol. The cycloalkane C-H
activation step of the D pathway can have a barrier no greater
than that (though possibly lower). The value for C-H activation
of n-alkanes is presumably lower still, on the basis of literature
t-Bu
q
for Cy-H addition to (
PCP)IrH2 (eq 4a) of ∆G > 36.4 kcal/
mol (see Table 3 and accompanying footnote, and Computa-
tional and Experimental Section).
The resulting exchange rates for the different hydrocarbons
decrease in the following order: phenyl-H > benzyl-H >
n-alkyl-H > cycloalkyl-H. This trend is familiar in the context
of C-H activation, especially by late metal centers. Indeed, to
our knowledge, every study involving C-H addition of these
substrates has yielded this exact order for both kinetics and
thermodynamics.2 Consequently, the relative barriers to H/D
exchange are all consistent with the assumption that the
exchange rate is a measure of the addition rate of the corre-
sponding C-H bond.
2
9
precedent for selectivity in C-H activation.
Considerations on the Remaining Steps of the Full
Catalytic Cycle; Relative Rates for Different Alkanes and
9,30
an Experimental Upper Limit for the Rate of H Loss from
2
(PCP)IrH . The data in Table 4 reveal that cyclooctane (COA)
2
and cyclodecane are dehydrogenated with significantly lower
barriers than n-undecane. At first glance, this ordering appears
to be inconsistent with either the A or the D pathway. It is
opposite that which would be expected from a rate-determining
associative reaction, because transition metal complexes add
cycloalkane C-H bonds more slowly than primary C-H bonds
The experimentally determined activation barriers are useful
in two distinct respects. First, they provide a comparison with
the computational values, and it can be seen from Table 3 that
the agreement is quite good although the calculated free energies
31
29
of activation appear to be slightly too high.
for all known C-H addition systems. The ordering, however,
Second, and independent of the good agreement with the
calculated values, the experimental values for H/D exchange
is also inconsistent with a simple picture involving the dis-
(
31) Most of the currently available functionals (including B3LYP) underestimate
intermolecular binding energies and charge-transfer effects at long range.
It is possible that the B3LYP functionals differentially overestimate the
energies in metal complexes with high coordination numbers and in the
transition states, which lead to the formation of these complexes. See, for
example: (a) Paizs, B.; Suhai, S. J. Comput. Chem. 1998, 19, 575. (b)
Ruiz, E.; Salahub, D. R.; Vela, A. J. Am. Chem. Soc. 1995, 117, 1141. (c)
Ref 13.
(
30) The relative differences in the addition rates among different alkane
substrates for kinetic measurements are usually much smaller than those
observed in the present case. This is probably because direct measurements
of addition rates necessarily involve exothermic additions, whereas in the
present case the rate-determining step is endothermic. Thus, the transition
state is late and, perhaps more importantly, our kinetic barriers include a
thermodynamic barrier. Thermodynamic variations among additions of
various hydrocarbons are often found to be of the same order of magnitude
as those in the present case.
(32) Peterson, T. H.; Golden, J. T.; Bergman, R. G. J. Am. Chem. Soc. 2001,
123, 455-462.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002 11411

A R T I C L E S
Krogh-Jespersen et al.
Table 4. Experimental Rates for Acceptorless Alkane
Dehydrogenation and Corresponding Values of ∆G
Scheme 2
q
rate (h-
1)
∆G
q
catalyst
substrate
T (°C)
t-Bu
t-Bu
(
(
(
(
(
(
PCP)Ir
PCP)Ir
COA
CDA
n-C11H24
COA
CDA
151
201
196
151
201
196
11
90
30
94
400
60
30.0
31.7
32.4
28.2
30.3
31.7
t-Bu/MeO
PCP)Ira
i-Pr
PCP)Ir
i-Pr
i-Pr
PCP)Ir
PCP)Ir
Expressed in terms of Scheme 2, k2[(PCP)Ir][alkane] appears
to be greater than k-1[(PCP)Ir][H2(sol)]. However, if the rate of
expulsion of H2 from solution, kexp[H2(sol)], is less than
k-1[(PCP)Ir][H2(sol)], the majority of H2 dissociation events
would be ultimately unproductive.
n-C11H24
a
(t-Bu/MeOPCP)Ir ) [η -4-CH3O-2,6-[(t-Bu2P)CH2)]2C6H2]Ir.
3
sociative pathway, if it is assumed that (a) (PCP)IrH2 is the
catalyst resting state (ground state) in all of the reactions and
that (b) H2 loss is irreversible for all of the reactions. On the
basis of those two assumptions, the D pathway is expected to
afford equal rates for all substrates. Thus, it is strongly implied
that H2 loss (regardless of the pathway) is not necessarily
irreversible, and subsequent steps in the reaction cycle may, at
least in some cases, contribute to the overall reaction barrier.
While a detailed characterization of all remaining steps in the
cycle is well beyond the scope of this paper, in this section we
will briefly consider other steps. In particular, we will consider
the one remaining step that the acceptorless system does not
have in common with the transfer-dehydrogenation system: the
expulsion of H2 from solution. We believe that consideration
of this step is critical in accounting for the apparent discrepancy
of the greater reactivity of the cycloalkanes versus n-alkanes.
Precise energy barriers for the chemical steps in the catalytic
cycle after C-H addition are presently being determined
experimentally in the context of transfer-dehydrogenation stud-
ies. However, we have previously shown that sources of
Thus, even assuming that loss of H2 is rate-determining for
the formation of (PCP)Ir(alkyl)(H) (eq 5), as is indicated by
the calculations in this paper, it is not necessarily the rate-
determining step of the catalytic cycle. Furthermore, even if
the loss of H2 is rate-determining for the overall reaction yielding
olefin and H2 in solution (and we believe this to be the case,9
although it falls outside the scope of the present work), the rate
of the catalytic cycle may still be lower than the rate of
dissociation of H2 from (PCP)IrH2. This condition would only
require that kexp[H2(sol)] is less than or not much greater than
k-1[(PCP)Ir][H2(sol)]. In the limiting case, which can be ex-
pressed as eq 10, expulsion of H2 from solution could actually
be the rate-limiting step.
kexp , k-1 [(PCP)Ir]
(for rate-limiting expulsion of H from solution)
2
(10)
Under this condition (eq 10), the preequilibria of eqs 11 and
12 would be in effect.
t-Bu
t-Bu
(
PCP)Ir (e.g., ( PCP)Ir(Ph)(H)) will readily dehydrogenate
28
alkanes even at 25 °C. We cannot yet experimentally determine
the rate of the overall reaction sequence of (
t-Bu
(PCP)IrH h (PCP)Ir + H
2(sol)
(11a)
PCP)Ir with
2
alkanes, but an upper limit of ca. 18 kcal/mol can be placed on
(
PCP)Ir + alkane h (PCP)Ir(alkyl)(H)
(11b)
(11c)
the overall barrier to this sequence (C-H addition, followed
9
by eqs 8 and 9).
(
PCP)Ir(alkyl)(H) h (PCP)IrH + alkene
2
â-H elimination
(
PCP)Ir(R)(H)
8 (PCP)IrH (alkene)
(8)
(9)
2
sum of eqs 11b,c: (PCP)Ir + alkane h
(
PCP)IrH (alkene) f (PCP)IrH + alkene
(PCP)IrH + alkene (11d)
2
2
2
This would imply that C-H addition to (t-BuPCP)Ir, as well as
â-H elimination and olefin removal, should all proceed more
rapidly than the back reaction with H2 under the conditions
sum of eqs 11a-c: alkane h alkene + H2(sol) (12)
The preequilibrium of eq 11d implies that the concentration of
(PCP)Ir would be dependent upon the dehydrogenation ther-
modynamics of the particular alkane (at a given concentration
of alkene). In other words, alkanes with higher dehydrogenation
-7
required for dehydrogenation (e.g., 150 °C, PH e ca. 10 atm).
2
This is in accord with (a) the calculations in this paper, which
show the free energy of the C-H addition transition state to be
lower than the free energy of the H2 elimination/addition TS
24,25
enthalpies (i.e., linear alkanes
) will result in higher concen-
(
see Figure 7), and (b) the simple assumption that even if H2
trations of (PCP)Ir in solution (eq 11d lies toward the left). The
higher concentration of (PCP)Ir results in greater rates of back
reaction of H2 with (PCP)Ir, relative to expulsion of H2 from
solution (i.e., the limit of eq 10 is approached). Thus, the limit
10
-1 -1
addition is diffusion controlled (k ≈ 10
M
s ), the low
-10
concentration of H2 (e ca. 10 M) gives a rate of back reaction
-
1
with H2 (ca. 1 s ) which is relatively slow (cf., the overall
rate of the sequence of C-H addition followed by eqs 8 and 9
of eq 10 should be more readily approached by the buildup of
-
1
q
24,25
is g130 s at 150 °C, on the basis of ∆G e 21 kcal/mol).
Hence the loss of H2 is probably followed by a relatively rapid
formation of olefin product for both cyclic and linear alkanes.
It must be remembered, however, that the acceptorless dehy-
drogenation (unlike transfer-dehydrogenation) is an inherently
heterogeneous reaction system. If H2 is not removed sufficiently
rapidly from solution, it is possible that the formation of olefin,
no matter how rapid, would be reversible.
linear alkene product as compared with cycloalkene.
It is
entirely possible, therefore, that the rate-determining step for
catalytic dehydrogenation of COA and cyclodecane is loss of
H2 from (PCP)IrH2, whereas under similar concentrations and
conditions, the rate-limiting step for n-alkane dehydrogenation
is much more likely to be expulsion of H2 from solution. This
same conclusion may be reached by considering that if reaction
11d proceeds at comparable rates with COA and n-alkanes
11412 J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002

Acceptorless Dehydrogenation of Alkanes
A R T I C L E S
9
(
which appears to be the case ), then the back reaction with
is, in itself, consistent with the rate of expulsion of H2 playing
an important role in determining the overall rate. Instrumentation
that will allow more quantitative control of the H2 expulsion
rate, which is required for detailed kinetic studies of the system,
is currently being designed.
linear alkenes is necessarily much faster, because the equilibrium
lies much further to the left.
An additional consideration is the possibility that both
cycloalkane and n-alkane systems might operate under preequi-
librium (eq 12) conditions. Because this would yield a much
Summary and Conclusion
24,25
greater equilibrium concentration of H2 for the cycloalkanes,
Our results may be summarized as follows: The dissociative
the rate of reaction under such conditions, kexp[H2(sol)], would
be much greater for cycloalkanes. Thus, it is possible that loss
of H2 from (PCP)IrH2 will be rate-determining for cycloalkanes
but not n-alkanes or, alternatively, that H2 expulsion will be
rate-determining for both classes of substrates; in either case,
cycloalkanes will giVe faster rates. Only in the limit where loss
of H2 from (PCP)Ir is rate-determining for the dehydrogenation
of both cycloalkanes and n-alkanes (kexp . k-1 [(PCP)Ir]) would
equal rates be obtained by the D pathway.
Although we have not yet developed methods to monitor the
acceptorless dehydrogenation in situ, experiments with alkene/
alkane mixtures in closed systems strongly indicate that in the
case of COA, and probably in the case of n-alkanes at low
alkene concentration, the resting state is (PCP)IrH2 at temper-
Me
pathway for the reaction of ( PCP)IrH2 with alkanes (cyclo-
hexane and propane) is calculated to have an energy barrier
q
(
∆E ) of ca. 36 kcal/mol. The highest energy TS for this path
Me
involves C-H addition to ( PCP)Ir, after H2 elimination. For
the same pathway, ∆G at standard thermodynamic conditions
q
is also ca. 36 kcal/mol, because the entropies of H2 loss and
C-H addition approximately cancel out. However, when the
free-energy barriers are calculated for conditions under which
dehydrogenation is thermodynamically possible (e.g., T ) 150
-
7
°
C; PH ) ca. 10 atm, with densities of hydrocarbon C-H
2
q
bonds equal to those in neat solution), ∆G for the C-H
activation step is much lower, ca. 18 kcal/mol. The large
difference is primarily due to the high entropy under these
conditions of the free H2 molecule produced prior to C-H
activation. Under these conditions, the rate-determining step is
9
atures of 150 °C or greater. The free-energy barrier to the loss
of H2 cannot be greater than the overall catalytic barrier
measured for any one of the alkanes. The loss of H2 (or any
other step in the cycle) can have a barrier no greater than the
Me
loss of H2 from ( PCP)IrH2, which has an enthalpy (and
estimated free energy) barrier of 27.2 kcal/mol (27.8 kcal/mol
t-Bu
for (
PCP)IrH2).
3
0.0 kcal/mol found for COA dehydrogenation (cf., the much
An associative pathway for the same reaction is calculated
greater barriers calculated for the A pathway). Preliminary
kinetic studies of acceptorless COA dehydrogenation indicate
that the reaction rate is independent of stirring rate. This would
suggest that expulsion of H2 is highly efficient under these
conditions (kexp . k-1 [(PCP)Ir]), specifically with COA as
substrate, and that 30.0 kcal is therefore the approximate
experimental value of ∆G for H2 loss at 151 °C. Calculations
using full bis(t-butyl)phosphino groups on the PCP ligand yield
a free energy of 27.8 kcal/mol for H2 addition to (t-BuPCP)Ir,
2
to proceed through Ir(V) and Ir(III)(η -H2) intermediates, in
H
12
accord with previous calculations on ( PCP)IrH2. The lowest
q
Me
energy barrier (∆E ) for direct C-H addition to ( PCP)IrH2
is calculated to be 20.1 kcal/mol (cyclohexane). The loss of H2
from the resulting manifold of adducts has a higher energy
barrier, 28.4 kcal/mol above the level of the reactants. Because
there is no barrier calculated for H2 addition to ( PCP)Ir(R)H,
E for the overall reaction is equal to ∆E ) 28.4 kcal/mol.
q
33
Me
q
∆
At standard thermodynamic conditions, the free-energy barriers
0
.6 kcal/mol greater than the calculated value for H2 addition
q
for this pathway are much higher: ∆G ) 32.2 kcal/mol for
Me
to ( PCP)Ir. The experimental value of 30.0 kcal/mol is in good
agreement with this calculated value. Alternatively, if H2 is in
fact not removed with a sufficiently high rate under these
conditions, that is, if kexp[H2(sol)] is not much greater than
k-1[(PCP)Ir][H2(sol)] (Scheme 2), then it is possible that dis-
sociation of H2 (and subsequent alkane dehydrogenation) is
occurring reversibly. In this case, the experimental value of 30.0
C-H addition and ca. 37.4 kcal/mol for subsequent loss of H2
from the C-H addition product. Under experimental conditions
-7
(
e.g., T ) 150 °C; PH ) ca. 10 atm, hydrocarbon C-H bond
2
densities equal to those in neat solution), the respective values
for C-H addition and H2 loss are 31.3 and 36.0 kcal/mol for R
)
Cy (33.7 and 36.6 kcal/mol for R ) n-Pr).
Consequently, under experimental conditions, the calculated
kcal/mol serves as an experimental upper limit for the barrier
overall free-energy barrier is ca. 9 kcal/mol greater for the A
pathway. It is further worth noting that even if a TS of lower
free energy for the C-H addition step could be located, this
would have no bearing on the highest free-energy barrier, the
one for the “transition state” for H2 elimination from the “seven-
coordinate” C-H adducts. Because some point on the reaction
coordinate must have an enthalpy at least equal to that of the
products (25.3 and 24.8 kcal/mol above reactants for R ) Cy
and n-Pr, respectively), and the entropy term (T∆S) due to the
loss of a free hydrocarbon molecule is substantial, especially
t-Bu
to loss of H2 from (
PCP)IrH2.
In the case of n-alkanes, preliminary kinetics data have been
less reproducible, but the slower rates indicate that loss of H2
from (PCP)IrH2 is not rate-determining. Thus, in accord with
the above considerations, the n-alkane system may operate under
conditions at or approaching the limit in which the preequilib-
rium of eq 12 is established, and the rate-determining step is
expulsion of H2. Note that the irreproducibility of the kinetics
(
33) In the case of n-alkanes, the slower rates must therefore be attributable to
a faster back reaction rehydrogenation of the olefin product as compared
with cycloalkenes. Because we have found that n-alkanes are dehydroge-
nated more rapidly than cycloalkanes, and because n-alkane dehydroge-
nation is thermodynamically less favorable, then hydrogenation of linear
olefins must indeed be much more rapid than that of cycloalkenes as would
be expected in view of both steric and thermodynamic considerations. In
at higher temperatures, the existence of this very high free-
21,34
energy barrier for the A pathway is unavoidable.
It should
also be noted that the presence of dialkylphosphino groups
bulkier than dimethylphosphino would be expected to further
increase the free-energy gap favoring the dissociative pathway.
The A pathway involves intermediates of the composition
2 2
terms of Scheme 2, k [(PCP)IrH ][olefin] is apparently greater than
3
k [H2(sol)] for linear olefins at the concentrations monitored (> ca. 10 mM),
but not for cyclooctene at concentrations in which the stirring rate was
R
varied (< ca. 100 mM).
( PCP)Ir(alkyl)H3 which are calculated to undergo rapid
J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002 11413

A R T I C L E S
Krogh-Jespersen et al.
scrambling among the hydrogen ligands. In the case of deuter-
be an important consideration in future efforts directed toward
rational development of related dehydrogenation catalysts.
ated hydrocarbons, this would lead to H/D exchange with
R
PCP)IrH2. Indeed, exchange with (t-BuPCP)IrH2 is observed
(
Computational and Experimental Section
for several hydrocarbons with rates decreasing in the following
order: C6D6 > mesitylene-d12 > n-decane-d10 . cyclohexane-
d12 (cyclohexane-d12 does not undergo observable exchange even
after 1 week at 180 °C). This order is in agreement with the
relative rates calculated for the different hydrocarbons in this
work. The absolute free-energy barriers extrapolated for the H/D
exchange are also in reasonably good agreement with the
calculated barriers, so there is strong evidence that C-H addition
Computational Methods. All calculations used DFT methodology35
with Becke’s three-parameter hybrid exchange functional (B3) and the
Lee-Yang-Parr correlation functional (LYP).3
6,37
The Hay-Wadt
relativistic, small core ECP and corresponding basis sets (split valence
3
8
double-ú) were used for the Ir atom (LANL2DZ model). We used
all-electron, full double-ú plus polarization function basis sets for the
3
9
second and third row elements C (Dunning-Huzinaga D95(d)) and
P (McLean-Chandler).40 Hydrogen atoms in the hydrocarbon, which
formally become hydrides in the product complexes, or in dihydrogen
R
to ( PCP)IrH2 can indeed occur along the pathways calculated.
However, the barrier to cyclohexane addition (both experimental
and computational) is much greater than the experimental barrier
to the full catalytic alkane dehydrogenation cycle. Thus, both
experimental methods (at least for cycloalkanes) and compu-
tational methods (for both propane and cyclohexane) demon-
strate that the A pathway is not kinetically viable.
41
were described by the triple-ú plus polarization 311G(p) basis set;
regular hydrogen atoms in alkyl or aryl groups (including PCP) carried
a double-ú quality 21G basis set.42
Reactant, transition state, and product geometries were fully
optimized with the ECP/basis set combination described above (B3LYP/
BasisA). The stationary points were characterized further by normal-
mode analysis, and the (unscaled) vibrational frequencies formed the
basis for the calculation of vibrational zero-point energy (ZPE)
corrections. Thermodynamic corrections were made to convert from
The D pathway for cycloalkane C-H addition must therefore
be operative experimentally with a free-energy barrier no greater
than ca. 30 kcal/mol, in good agreement with values calculated
for the D pathway. For n-alkanes, the experimental results alone
cannot exclude the A pathway. However, the experimental
results are consistent with the high barrier to the C-H addition
step and certainly not inconsistent with the substantially higher
barrier calculated for the subsequent H2 elimination. The D
pathway found for cycloalkanes (with a barrier of ca. 30 kcal/
mol) must also be accessible for dehydrogenation of n-alkanes;
thus all calculated and experimental results are consistent only
q
purely electronic reaction or activation energies (∆E, ∆E ; T ) 0 K,
q
q
no ∆ZPE) to enthalpies and free energies (∆H°, ∆H° ; ∆G°, ∆G° ;
∆
ZPE included, T ) 298 K, P ) 1 atm) according to standard statistical
43
mechanical expressions. Standard techniques were also applied to
obtain ∆G and ∆G at other combinations of temperature and pressure.
q
Additional single-point calculations at the B3LYP level used a more
extended basis set for Ir in which the default LANL2DZ functions for
the Ir(6p) orbital were replaced by the functions reoptimized by Couty
and Hall,44 and sets of diffuse d functions (exponent ) 0.07) and f
45
3
3
functions (exponent ) 0.938) were added as well (B3LYP/BasisB).
with the D pathway for either cyclic or linear alkanes.
The reoptimized/expanded Ir basis set preferentially favors structures
The development of an experimental energy profile for the
full catalytic cycle awaits (a) the completion of transfer-
dehydrogenation studies, which will yield the precise barriers
for C-H addition and â-H elimination steps, and (b) the
development of experimental methods suitable for monitoring
the reaction in situ and controlling the rate of H2 expulsion.
However, in the limit where H2 is rapidly expelled from solution,
the rate-determining step of the catalytic reaction is dissociation
of H2 from (PCP)IrH2, rather than addition of alkane. This must
with high Ir coordination numbers.4
4,46
All computed energy data
q
discussed in the text or presented in the tables are based on ∆E (∆E )
values from these B3LYP/BasisB calculations, followed by electronic
energy-enthalpy-free energy conversions (as appropriate) made in
an additive fashion with data derived at the B3LYP/BasisA level.
In a few cases, the exact nature of a particular transition state was
47
investigated further by intrinsic reaction coordinate calculations.
Approximately 10 steepest descent steps from the transition state were
executed in the forward and reverse direction given by the transition
vector. The resulting structures were then geometry optimized toward
the nearest minimum to assign proper reactants and products.
All calculations were executed using the Gaussian 98 series of
(
34) After completion of this work, Haenel, Kaska, and Hall reported an
experimental/computational study (ref 13) in which it was concluded that
the (anthraphos-PCP)IrH
proceeding through an associative transition state for C-H addition. The
computational model used in ref 13 has PH groups in the anthraphos-
2 2
analogue of (PCP)IrH undergoes reaction 5
48
computer programs.
2
H/D Exchange. (t-BuPCP)IrH
was prepared by subjecting
PCP ligand rather than dimethylphosphino groups (as in our calculations)
or bis(t-butyl)phosphino groups (as used in both the anthraphos and the
PCP-parent experimental systems), and a small model alkane (RH )
ethane). Haenel, Kaska, and Hall chose to classify the associative TS as
2
(t-BuPCP)IrH
4
4
to vacuum, according to reported methods. Deuterio-
an “interchange TS”, TS
PCP)IrEtH; that is, C-H cleavage and addition of ethane were concerted
with formation and expulsion of H . Some of the computed internuclear
distances (Å) in TS are Ir-Cethyl (2.176); Ir-H (1.641); Cethyl-H (2.58);
Ir-H (1.664); Ir-H (1.687); H -H (1.046). Thus, the ethyl carbon and
I
. TS
I
was proposed to lead directly to (anthraphos-
(35) Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and Molecules;
University Press: Oxford, 1989.
(36) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652.
(37) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(38) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299.
(39) Dunning, T. H.; Hay, P. J. In Modern Theoretical Chemistry; Schaefer, H.
F., III, Ed.; Plenum: New York, 1976; pp 1-28.
(40) McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72, 5639.
(41) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980,
72, 650.
(42) Binkley, J. S.; Pople, J. A.; Hehre, W. J. J. Am. Chem. Soc. 1980, 102,
939.
(43) McQuarrie, D. A. Statistical Thermodynamics; Harper and Row: New York,
1973.
2
I
a
a
b
c
b
c
all three hydrogens have normal distances to Ir and show significant bonding
with iridium. The C-H distance is 2.58 Å, indicating a fully cleaved C-H
bond. Using the exact model species and computational methods applied
in ref 13, we have verified the existence and structure of TS
graphical examination of the transition vector components for TS
no structural indications of H expulsion. Rather, the normal mode
displacements indicate that TS is a TS for interconversion of two “seven-
coordinate” isomers, the analogues of our structures B and C in Figures 2
and 4. Normal mode following from TS (intrinsic reaction coordinate
approach) led cleanly to the intermediate (anthraphos-PCP)IrEtH(H ) and
anthraphos-PCP)IrEt(H) species as the “reactant” and “product” connected
by TS . In our hands, the proposed rate-limiting TS in ref 13, TS , does not
lead to the final products of eq 5, (anthraphos-PCP)IrEtH + H . Also, the
energy obtained for TS is peculiar, if TS is associated with a TS leading
in a single step from reactants (anthraphos-PCP)IrH + EtH to the
is reported to be approximately
8 kcal/mol lower in enthalpy than these final products (Table 1 of ref
3).
I
. However,
I
shows
2
I
I
2
(44) Couty, M.; Hall, M. B. J. Comput. Chem. 1996, 17, 1359.
(45) Ehlers, A. W.; B o¨ hme, M.; Dapprich, S.; Gobbi, A.; H o¨ llwarth, A.; Jonas,
V.; K o¨ hler, K. F.; Stegmann, R.; Veldkamp, A.; Frenking, G. Chem. Phys.
Lett. 1993, 208, 111.
(
3
I
I
2
I
I
(46) If we let ∆∆E ) ∆E(B3LYP/BasisB) - ∆E(B3LYP/BasisA), we find ∆∆E
q
Me
2
) -3.5 kcal/mol and ∆∆E ) -4.8 kcal/mol for the reaction ( PCP)Ir +
Me Me Me
(
anthraphos-PCP)IrEtH + H
2
products. TS
I
2
CyH f ( PCP)Ir(Cy)(H). For the system ( PCP)IrH + CyH f ( PCP)-
q
1
1
Ir(Cy)(H)
3
, we find ∆∆E ) -3.3 kcal/mol and ∆∆E ) -4.8 kcal/mol.
(47) Gonzalez, C.; Schlegel, H. B. J. Phys. Chem. 1990, 94, 5523.
11414 J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002

Acceptorless Dehydrogenation of Alkanes
A R T I C L E S
Table 5. Complete Set of Reactions and Relative Rate Constants Used for Modeling the Kinetics of H/D Exchange between
t-Bu
a
Deuterohydrocarbons and ( PCP)IrH2
reactionb
rate constantc
fixed value (relative)d
numerical value (M-1
s )
-1 e
t-Bu
PCP)IrH2 + RD f (t-BuPCP)IrHD + RH
-4
(
(
(
(
(
(
(
(
(
(
(
(
k7
k-7
k2′
k-2′
k3′
k-3′
k4′
k-4′
k5′
k-5′
k6′
variable
1.2 × 10
t-Bu
t-Bu
t-Bu
t-Bu
t-Bu
t-Bu
t-Bu
t-Bu
t-Bu
t-Bu
t-Bu
t-Bu
-5
PCP)IrHD + RH f ( PCP)IrH2 + RD
2 × 0.5/n × k7
2.0 × 10
t-Bu
-5
PCP)IrHD + RD f ( PCP)IrD2 + RH
0.5 x k7
6.1 × 10
t-Bu
-5
PCP)IrD2 + RH f ( PCP)IrHD + RD
2/n × k7
4.0 × 10
t-Bu
PCP)IrH2 + PD f ( PCP)IrHD + PH
variable
1.0
1.0
t-Bu
PCP)IrHD + PH f ( PCP)IrH2 + PD
2 × 0.5 × k3′
t-Bu
-1
PCP)IrHD + PD f ( PCP)IrD2 + PH
0.5 × k3′
5.0 × 10
t-Bu
PCP)IrD2 + PH f ( PCP)IrHD + PD
2 × 0.5 × k3′
variable
2.0
t-Bu
-1
PCP)IrH2 + TD f ( PCP)IrHD + TH
1.2 × 10
t-Bu
-2
PCP)IrHD + TH f ( PCP)IrH2 + TD
0.5 × k-5′
0.5 × k-5′
1.0 × k-5′
6 × 10
t-Bu
-2
PCP)IrHD + TD f ( PCP)IrD2 + TH
6 × 10
t-Bu
-1
PCP)IrD2 + TH f ( PCP)IrHD + TD
k-6′
1.2 × 10
a
b
Numerical values determined for C6D6 are given; values for k7 given in Table 4 for other deuteriohydrocarbons. RD ) deuteriohydrocarbons; PH(D)
t-Bu
PCP)IrHxD4-x. c k7 and k-7 refer to eq 7. Other rate constants do not
)
PCP tert-butyl hydrogens (deuterons); TH(D) ) hydrides (deuterides) of (
d
correspond to equations in the text and are indicated as kn′. n ) statistical correction for number of H’s in RH (actually monoproteated deuteriohydrocarbon)
e
as compared with number of equivalent D’s in RD, for example, n ) 6 for reactions with C6D6. Numerical values for RD ) C6D6; for other
-
1
deuteriohydrocarbons, k3′ and k5′ were set to an arbitrary high value relative to k7 (1.0 s ; see Table 4 for other values obtained for k7).
hydrocarbon solvents were distilled from Na/K alloy. Solutions of ca.
fits based on one parameter, k
of certainty in k than if the fits were also sensitive to other parameters.
We believe that our results, for k , have an uncertainty of less than
(25%; this corresponds to an uncertainty in ∆G of (0.2 kcal/mol
(even a factor of 2.0 corresponds to an acceptable error level of only
(0.4 kcal/mol at 30 °C).
7
(see below). This allows a higher level
t-Bu
0
.030 M ( PCP)IrH
2
were prepared and transferred to an NMR tube
7
which was then flame-sealed.
7
q
Because our interest was specifically in determining the rate of C-H
t-Bu
2
addition to ( PCP)IrH , it was desirable to disfavor any other
pathways that might potentially lead to H/D exchange. In particular,
in the case of alkanes, there was concern that a degenerate transfer-
hydrogenation/dehydrogenation involving even trace amounts of cor-
responding alkene could lead to rapid H/D exchange without proceeding
The kinetics modeling scheme treated the PCP t-butyl hydrogens
(deuterons) as a pool of individual species with a total “concentration”
36 times that of total iridium. This was consistent with our experimental
method, which could only determine the fraction of total PCP t-butyl
deuteration, and this obviated the need to treat each of the 36 possible
isotopomers separately for kinetics modeling. A pseudo second-order
rate constant, k , for exchange between the hydrides and tert-butyl
via reaction 4a. To minimize this possibility, all runs were conducted
t-Bu
with small amounts of ( PCP)IrH
4
present (ca. 2-5 mol %). This
t-Bu
was obtained simply by not driving the conversion from ( PCP)IrH
to ( PCP)IrH to completion. Our hope was that the presence of
2
would keep the steady-state alkene concentration to an
4 2
absolute minimum. Because ( PCP)IrH loses free H , the thermo-
4
t-Bu
3
′
(
t-BuPCP)IrH
4
hydrogen is then obtained; this value is not physically meaningful,
assuming that the exchange is intramolecular, but it allows us to
t-Bu
dynamics of alkene hydrogenation must be very favorable and lie very
far to the right:
extrapolate k . In principle, a physically meaningful first-order rate
7
constant for exchange could be obtained by multiplying k by (36 ×
3′
t-Bu
[(
2 n
PCP)IrH -d ]). However, as this value was not directly relevant
to this work, such a conversion to first-order was unnecessary, and
because the reaction in all cases was fast relative to k , it could, in any
7
case, only serve as a very approximate lower limit. The hydrides/
We still cannot exclude the possibility that transfer-hydrogenation/
dehydrogenation or other undesired H/D exchange pathways were in
fact operative to at least a partial extent. However, because we use the
observed H/D exchange rates to obtain upper limits for the rate of eq
t-Bu
deuteride ligands of ( PCP)IrH
of individual species. Unlike the hydrides/deuterides of ( PCP-d )-
n
4
-d
n
were similarly treated as a pool
t-Bu
t-Bu
IrH
resolved by H NMR.
Rate constants k-7, k2′, and k-2′ are chemically equivalent to k
because it was assumed that statistically the
2 n 4 n
-d , the various isotopomers of ( PCP)IrH -d could not be
1
4
, any such pathways would not invalidate the conclusions obtained
7
. k2′
from this work.
was set equal to 0.5 × k
7
The total concentration of dihydride and tetrahydride was found to
t-Bu
rate of RD exchange with ( PCP)IrHD would be one-half the rate
1
remain unchanged during the kinetic runs, as determined by H NMR
t-Bu
with ( PCP)IrH
2
. Likewise, k-7 was set equal to 0.5 × k-2′. Both
t-Bu
n 2
observation of the PCP methylene protons (( PCP-d )IrH : δ
rate constants contained the statistical term, 1/n, where n is the number
t-Bu
-
)
3.47(vt), JPH ) 6.4 Hz, dihydride; ( PCP-d
n
)IrH
4
: 3.24(vt), JPH
of chemically equivalent H’s in the perproteo-hydrocarbon, to account
for the fact that the RH product was only monoproteated (e.g., C D H
6 5
where n ) 6).
Although kinetic isotope effects for C-H activation reactions vary
greatly, equilibrium isotope effects vary much less so because the
relevant vibrational frequencies of the respective ground states are far
1
3.4 Hz) and an external (capillary) standard. H NMR was also used
to monitor (a) the total concentration of iridium dihydride, ( PCP-
t-Bu
d
n
)IrH
2
(δ -19.194(t), JPH ) 8.7 Hz; n ) 0-36); (b) the total
t-Bu
concentration of hydridodeuteride, ( PCP-d )IrHD (δ -19.039(t), JPH
)
n
8.4 Hz; n ) 0-36); (c) the fraction of deuteration of the total t-BuPCP
tert-butyl hydrogens (δ 1.22(vt) JPH ) 6.4 Hz); (d) the fraction of
t-Bu
n m
deuteration of the total ( PCP-d )IrH D(4-m) (tetrahydride) hydrides
(
48) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.;
Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham,
M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.;
Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.;
Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian
98, revision A.9; Gaussian, Inc.: Pittsburgh, PA, 1998.
(d -9.48(t) JPH ) 9.8 Hz); and (e) the fraction of protiation of the
deuteriohydrocarbon. Thus, a total of five different experimental
parameters were measured.
The concentrations of all species were fit to the reaction scheme
indicated in Table 5.
At the outset of this discussion, it is important to note that we
required only a value for k , and that value was needed at only a very
7
approximate level. Ultimately, the data fits were shown to be highly
insensitive to the other variables and therefore essentially amounted to
J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002 11415

A R T I C L E S
Krogh-Jespersen et al.
more constant in value than those of the transition states. For the sake
of simplicity, we kept the same ratio, 2.0, for all reactions involving
transfer of H/D from C to Ir (e.g., k /k-7 or k3′/k-3′). This somewhat
7
for (t-BuPCP)IrH
:
2
nIrH ) IrH + (-k *IrH *RD - k *IrH *PD +
2
2
7
2
3
2
k-7*IrHD*RH + k-3*IrHD*PH - k *IrH *TD +
arbitrary value was obtained by taking a well-determined equilibrium
5
2
isotope effect for C-H addition, 2.4 for addition to Tp′Rh(CN-
neopentyl) at 26 °C, and extrapolating to the midpoint of the
temperature range most relevant to this work (30-180 °C). The final
k-5*IrHD*TH) dt
4
9
IrH
2
is [(t-BuPCP)IrH
] at the outset of each time interval, dt; nIrH is
2 2
the resulting concentration at the end of each time interval (other
abbreviations are defined in footnote a, Table 5). The use of time
intervals smaller than 0.5 s had no effect on the concentrations that
were calculated for the experimental times (ca. every 2000 s). Rate
constants were then calculated by iterating to minimize the sums of
value of k
isotope effect of 1.4 gave values within 10% of those obtained using
.0). It was assumed that there was no significant equilibrium isotope
7
was not very sensitive to the value used (e.g., an assumed
2
effect in the exchange between dihydride and tetrahydride.
The reactions were monitored at intervals of ca. 2000 s. Data for
benzene and n-decane (experimental and least-squares best-fit calcu-
lated) are given in the Supporting Information. Benzene gave the best
data probably because the low temperature of the reaction, 30 °C,
allowed continuous heating in the NMR spectrometer, whereas other
samples had to be removed, heated, and then returned to the
spectrometer. The resulting fit was found to be highly insensitive to
2
the squares (∑[(exp - calc) /expavg]), weighted for the time-averaged
concentrations of various species that were monitored. It was established
by iteration of k3′, k5′, and k
benzene (Table 5). For other deuteriohydrocarbons, where k
smaller and temperatures were much higher, it was assumed that k3′
and k5′ were large relative to k . Thus, k3′ and k5′ were set at an arbitrary
7
that k3′ and k5′ were large relative to k
7
for
7
was
7
-
1
high value (1.0 s ) for other deuteriohydrocarbons, and it was
subsequently confirmed that varying k3′ and k5′ in that regime had no
effect on the calculated fit. Raw data and corresponding calculated data
points are given in the Supporting Information for the exchange
the values for variables k3′ and k5′ because these rates were rapid relative
2
to k
7
. The sum of the squares of the differences (∑(exp - calc) )
decreased asymptotically with increasing values of k3′ (exchange
between t-butyls and hydrides).
6 6
reactions with C D and n-C10D22, respectively.
The factors used for k-7 and k-2′ were also unimportant in optimizing
the fit, because the corresponding rates were slow due to the very low
concentrations of RH (particularly after correction for the statistical
term, n, which is required because the “RH” product is only mono-
proteated RD). Thus, the kinetics simulation was essentially a fit to
Acknowledgment. We thank the Division of Chemical
Sciences, Office of Basic Energy Sciences, Office of Energy
Research, U.S. Department of Energy for support of this work,
and the National Science Foundation for a computer equipment
grant (DBI-9601851-ARI).
the one parameter of interest, k
applicable to the other deuteriohydrocarbons, because the corresponding
values of k were much smaller, relative to presumed values of k3′ and
5′ at the much higher temperatures used for these substrates.
Using a macro written for Microsoft Excel 98 for Macintosh, we
7
. This conclusion was even more
Supporting Information Available: Data (concentrations
7
1
determined by H NMR) used to obtain (least-squares best-fit)
k
rate constants for H/D exchange between deuteriohydrocarbons
t-Bu
calculated the concentrations of each species at small finite time
intervals (0.5 s) using the appropriate kinetic equation. For example,
(C6D6 and n-decane-d22) and (
PCP)IrH2 and corresponding
calculated values (PDF). This material is available free of charge
via the Internet at http://pubs.acs.org.
(
49) Northcutt, T. O.; Wick, D. D.; Vetter, A. J.; Jones, W. D. J. Am. Chem.
Soc. 2001, 123, 7257-7270.
JA012460D
11416 J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002