Olefin isomerization by iridium pincer catalysts. experimental evidence for an η3-allyl pathway and an unconventional mechanism predicted by DFT calculations

Article

pubs.acs.org/JACS

Oleﬁn Isomerization by Iridium Pincer Catalysts. Experimental

Evidence for an η³-Allyl Pathway and an Unconventional Mechanism

Predicted by DFT Calculations

Soumik Biswas,^†Zheng Huang,^‡Yuriy Choliy,^†David Y. Wang,^†Maurice Brookhart,*^,‡

Karsten Krogh-Jespersen,*^,†and Alan S. Goldman*^,†

^†Department of Chemistry and Chemical Biology, Rutgers, The State University of New Jersey, New Brunswick, New Jersey 08903,

United States

^‡Department of Chemistry, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599, United States

S

* Supporting Information

ABSTRACT: The isomerization of oleﬁns by complexes

of the pincer-ligated iridium species (^tBuPCP)Ir (^tBuPCP =

κ³-C₆H₃-2,6-(CH₂P^tBu₂)₂) and (^tBuPOCOP)Ir (^tBuPOCOP =

κ³-C₆H₃-2,6-(OP^tBu₂)₂) has been investigated by computa-

tional and experimental methods. The corresponding dihy-

drides, (pincer)IrH₂, are known to hydrogenate oleﬁns via

initial Ir−H addition across the double bond. Such an addition

is also the initial step in the mechanism most widely proposed

for oleﬁn isomerization (the “hydride addition pathway”);

however, the results of kinetics experiments and DFT

calculations (using both M06 and PBE functionals) indicate that this is not the operative pathway for isomerization in this

case. Instead, (pincer)Ir(η²-oleﬁn) species undergo isomerization via the formation of (pincer)Ir(η³-allyl)(H) intermediates; one

example of such a species, (^tBuPOCOP)Ir(η³-propenyl)(H), was independently generated, spectroscopically characterized, and

observed to convert to (^tBuPOCOP)Ir(η²-propene). Surprisingly, the DFT calculations indicate that the conversion of the η²-

oleﬁn complex to the η³-allyl hydride takes place via initial dissociation of the Ir−oleﬁn π-bond to give a σ-complex of the allylic

C−H bond; this intermediate then undergoes C−H bond oxidative cleavage to give an iridium η¹-allyl hydride which “closes” to

give the η³-allyl hydride. Subsequently, the η³-allyl group “opens” in the opposite sense to give a new η¹-allyl (thus completing

what is formally a 1,3 shift of Ir), which undergoes C−H elimination and π-coordination to give a coordinated oleﬁn that has

undergone double-bond migration.

Scheme 1. Schematic Illustration of the Two Mechanism

Classes Proposed for Oleﬁn Isomerization: (a) “Hydride”

Mechanism, (b) π-Allyl Mechanism

INTRODUCTION

■

Examples of transition-metal-catalyzed oleﬁn isomerization are

widespread and play a key role in numerous important chemical

processes.¹⁻⁹The mechanism most commonly accepted for

oleﬁn isomerization involves initial addition of a metal−H bond

across the oleﬁn double bond. If the resulting metal alkyl has a

β-carbon atom that is inequivalent to the carbon to which the

hydride was added, β-elimination of hydrogen from that carbon

will result in oleﬁn isomerization (Scheme 1a). An alternative

pathway, less commonly proposed and less well-studied,

involves the intermediacy of a π-allyl complex (Scheme 1b)

through which oleﬁn isomerization occurs via a formal 1,3-

hydride shift.^1,2,10−14If a coordinatively unsaturated metal

complex bears a metal-bound hydride, the hydride addition

pathway is believed to be the preferred mechanism for oleﬁn

isomerization.¹

of low molecular weight alkanes into higher molecular weight

alkanes; this will likely become increasingly important as conventional

We have recently reported a tandem catalytic system for the

metathesis of alkanes, comprising an iridium pincer catalyst

for alkane dehydrogenation/oleﬁn hydrogenation and a second

catalyst for oleﬁn metathesis (Scheme 2).^15,16Alkane meta-

thesis^17,18(AM) can potentially be employed for the transformation

Received: February 13, 2012

Published: July 5, 2012

13276

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Scheme 2. Tandem Catalytic Alkane Metathesis with Pincer-

Ligated Iridium Complexes and Oleﬁn Metathesis Catalysts

(Shown for Reaction of n-Hexane)

Scheme 3. Possible Oleﬁn Insertion Pathway (Hydride

Pathway) for Isomerization by Pincer−Iridium Complexes

petroleum reserves dwindle. For example, Fischer−Tropsch

product mixtures (primarily n-alkanes)¹⁹as well as natural gas

condensates²⁰include C₃−C₈n-alkane fractions that are not

directly suitable for most transportation fuels. Conversion to the

corresponding C_2n−2species, such as conversion of n-hexane to

n-decane plus ethane, could give access to heavier (C₉₋₁₉) n-alkanes,

which constitute a valuable clean-burning diesel fuel. Previously

reported AM systems, however, have not shown such desirable

selectivity for the formation of C_2n−2species from C_nsubstrates.^17,18

The AM systems developed in our laboratories are mainly

based on two types of pincer catalysts, (^RPCP)Ir (^RPCP = κ³-

C₆H₃-2,6-(CH₂PR₂)₂) and (^RPOCOP)Ir (^RPOCOP = κ³-C₆H₃-

2,6-(OPR₂)₂), employed in combination with a Schrock-type

oleﬁn metathesis catalyst.²¹Iridium pincer catalysts, including

mechanism is quite diﬀerent from that which is conventionally

proposed for π-allyl pathways. This conclusion has implications for

catalytic AM, for alkane dehydrogenation systems more generally,

and, potentially, for any system involving transition-metal-catalyzed

oleﬁn isomerization or the formation of an η³-allyl complex from

an oleﬁn.

(

^tBuPCP)Ir and (^iPrPCP)Ir, have been found to dehydrogenate

n-alkanes with kinetic selectivity to form α-oleﬁns.²²Thus,

these systems might be expected to selectively produce ethane

and C_2n−2species, as shown in the example in Scheme 2

(metathesis of n-hexane to yield ethane and n-decane); in fact,

AM of n-hexane generates a range of C₂to C₁₅n-alkanes.^15,16

Signiﬁcantly, the (^tBuPCP)Ir system gives moderate selectivity

for the formation of n-decane (ca. 50 mol % of the C_nproducts

where n > 6), whereas the (^tBuPOCOP)Ir system gives an

essentially stochastic distribution of n-alkanes.^15,16The most

obvious explanation for the initial formation of C₃₋₅and C₇₋₉

products from n-hexane is based upon isomerization/meta-

thesis of the oleﬁn intermediates prior to hydrogenation. A

signiﬁcant degree of isomerization of 1-alkenes has indeed been

previously reported when pincer−Ir complexes are used for

either acceptorless or transfer dehydrogenation of n-alkanes.²³

In consideration of these issues, we initiated a study of pincer−

iridium-catalyzed oleﬁn isomerization with an emphasis on

comparison between the (^tBuPCP)Ir and (^tBuPOCOP)Ir

systems.

RESULTS AND DISCUSSION

■

AM Catalyst Resting State. Typical AM reactions are

performed at 125 °C in n-alkane as solvent and reactant, with

10 mM of pincer−iridium dehydrogenation catalyst and ca.

16 mM Mo(N-2,6-(i-Pr)₂C₆H₃)(CHCMe₂Ph)[OC-

21

(CF₃)₂(CH₃)]₂(“Mo-F12”) oleﬁn metathesis catalyst in

n-alkane as solvent and reactant. When an iridium dihydride is

used as the catalyst, 2 equiv of t-butylethylene per mol iridium

dihydride is typically added; 1 equiv serves to dehydrogenate

the catalyst, while the second equivalent acts as an acceptor for

hydrogen from n-alkane and thereby serves to establish a

steady-state concentration of 1 equiv oleﬁn per mol catalyst.

No t-butylethylene is added when (^tBuPOCOP)Ir(C₂H₄) is

used as precursor since this precatalyst does not need to be

dehydrogenated and the mol of C₂H₄acts as an acceptor to

establish the steady-state concentration of oleﬁn.

Our investigation of the mechanistic aspects of AM,

including the oleﬁn isomerization component, began with in

situ observations of the catalyst resting state. ³¹P NMR

spectroscopy reveals that, under the typical conditions noted

above, the resting state for the (^tBuPCP)Ir system is

With either PCP or POCOP ligands, ﬁve-coordinate

(pincer)IrH₂species are intermediates in alkane dehydrogen-

ation.^22,24,25The insertion of C−C double bonds into the Ir−H

bonds of these complexes is a necessary step in the overall cycle

for alkane metathesis and for transfer dehydrogenation more

generally. Thus, catalytic oleﬁn isomerization by the pincer−

iridium complexes was initially assumed to proceed via

the insertion mechanism (i.e., Ir−H addition) illustrated in

Scheme 3. A 2,1 insertion of 1-alkene into the Ir−H bond of

the (pincer)IrH₂complex, followed by β-H elimination from

the n-alkyl group, would generate the isomeric 2-oleﬁn

product.²²Surprisingly, however, experimental and computa-

tional evidence supports a π-allyl pathway for both complexes

(Scheme 1b). Even more surprisingly, although π-allyl pathways

are well-established, our results strongly indicate that the stepwise

(

^tBuPCP)IrH₂(1); no other species are observed in the ³¹P

NMR spectrum.¹⁴In contrast, for the (^tBuPOCOP)Ir system,

only (^tBuPOCOP)Ir(oleﬁn) species are observed to be present

by ³¹P NMR spectroscopy.¹⁴Thus, if it were assumed that

oleﬁn isomerization proceeds via the hydride insertion

mechanism, it might be expected that the (^tBuPOCOP)Ir

system would be less active with respect to isomerization and

might therefore impart higher selectivity in AM, in sharp

contrast to the observed results. This apparent anomaly initially

led us to probe the mechanism of oleﬁn isomerization by each

of these pincer systems.

13277

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Kinetics of Catalytic Isomerization. Eﬀect of Alkane

versus Arene Solvent: Evidence against a Hydride

Mechanism. A kinetic study of the isomerization of 1-octene

catalyzed by (^tBuPCP)IrH₂(1) was conducted using n-octane or

p-xylene as solvents (eq 1). Addition of 1-octene (100 mM) to

1 (5 mM), in either solvent, results in apparently quantitative

conversion to (^tBuPCP)Ir(1-octene) within 10 min at 25 °C

(eq 2). However, as indicated by the equilibrium of eq 2, the

Scheme 4. Schematic π-Allyl Pathway for 1-Alkene

Isomerization with Dissociative Oleﬁn Exchange (Detailed

Mechanism Not Speciﬁed)

the concentration of added trans-2-hexene, with activation

parameters determined to be ΔH^⧧= 15.4 0.4 kcal/mol and

ΔS^⧧= −0.7

1.5 eu (from rates measured at temperatures

ranging from −17 to 6 °C). Extrapolation to 125 °C yields a

value (2 × 10⁴s⁻¹) that is orders of magnitude greater than the

overall rate of isomerization.

These kinetic experiments were repeated with (^tBuPOCOP)-

IrH₂(2) as the catalyst precursor. As was the case with

concentration of dihydride complex 1, although minor, would

presumably be much greater in n-octane than in p-xylene.

(The equilibrium expression for the concentration of 1 is not

rigorous because it neglects all sources of hydrogen other than

n-octane, but certainly any steady-state concentration of 1

would be much greater in alkane solvent than in p-xylene,

which cannot easily act as a hydrogen donor.) Therefore, if a

minor concentration of 1 were actually the major active oleﬁn

isomerization catalyst, oleﬁn isomerization would be much

faster in the alkane solvent. In fact, the rates of isomerization

are equal (within 1%) in n-octane and p-xylene solvent (Figure 1).

These results argue strongly against operation of the hydride

addition mechanism for (^tBuPCP)Ir-catalyzed oleﬁn isomerization.

The turnover frequency (k) for eq 1 in either solvent is

(

^tBuPCP)IrH₂, (^tBuPOCOP)Ir(1-octene) was formed rapidly

and apparently quantitatively and remained the major species in

solution during the catalysis, as determined by in situ ³¹P NMR

spectroscopy. No signiﬁcant diﬀerence in isomerization rates

was found using n-octane as solvent versus p-xylene (Figure 2);

4.2 × 10⁻³s⁻¹at 125 °C (rate = k[(^tBuPCP)Ir]; ΔG^⧧

∼

27.9 kcal/mol). A zero-order dependence on [1-octene] is

observed (Figure 1), which is consistent with an intramolecular

π-allyl mechanism in which (^tBuPCP)Ir(1-octene) is the major

resting state (as is observed by NMR spectroscopy), and the

rate-determining step lies within segments (a), (b), or (c) of the

cycle indicated in Scheme 4. (We deﬁne these as “segments” of

the cycle since each of them, and particularly (a) or (b), could

involve more than one elementary reaction step.)

Figure 2. Isomerization of 1-octene to internal octenes catalyzed by

(

solvent.

^tBuPOCOP)IrH₂at 125 °C: (a) n-octane solvent, (b) p-xylene

Consistent with the mechanism of Scheme 4, loss of alkene

from (^tBuPCP)Ir(alkene) is rapid and dissociative. Exchange

spectroscopy (EXSY) experiments reveal that at 25 °C the

rate of 1-hexene dissociation is 0.51 s⁻¹and independent of

the concentration of added 1-hexene. An Eyring plot based on

rates measured at temperatures between 20 and 46 °C aﬀords

activation parameters ΔH^⧧= 21.4 0.8 kcal/mol and ΔS^⧧=

12 3 eu for loss of 1-hexene. Dissociation of trans-2-hexene,

as would be expected of a dissociative process with a bulkier

ligand, is even faster; at 4 °C, the rate is 3.1 s⁻¹, independent of

again, this argues strongly against operation of the hydride

addition mechanism.

In contrast with the (^tBuPCP)Ir-catalyzed reaction, however,

in the case of (^tBuPOCOP)Ir, the reaction kinetics show a clear

positive dependence on 1-octene concentration. Given that the

1-octene complex is observed to be the major species in

solution, this is not easily reconciled with a hydride insertion

mechanism since the steady-state concentration of

(

^tBuPOCOP)IrH₂should be inverse-second-order in [1-octene]

Figure 1. Catalytic isomerization of 1-octene to internal octenes by 1 (5 mM) at 125 °C: (a) n-octane solvent, (b) p-xylene solvent.

13278

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

(and ﬁrst-order in n-octane) (cf. eq 2). This positive

dependence on [1-octene] is, however, also inconsistent with

a π-allyl mechanism operating with (^tBuPOCOP)Ir(1-octene) as

the resting state, unless the rate-determining step involves an

additional equivalent of 1-octene. While this would be the case

if loss of 2-alkene (step c in Scheme 4) were reversible and the

back-reaction with 2-octene were fast relative to addition of

1-alkene, these conditions seem highly unlikely given the lesser

bulkiness and initially higher concentration of 1-alkenes.

An alternative explanation for the positive dependence of rate

on [1-alkene] is that formation of (^tBuPOCOP)Ir(2-alkene) is

reversible, and that displacement of coordinated 2-alkene by

1-alkene proceeds through an associative rate-determining step;

such a pathway is indicated in Scheme 5. Activation enthalpies

(and the rate dependence on [1-alkene] which it implies), but

we note that it is fully reproducible.

In view of the above, we cannot assign a meaningful rate

constant or activation parameters to the (^tBuPOCOP)Ir-

catalyzed isomerization of α-oleﬁns. Importantly, however,

the rate is not signiﬁcantly faster than that of oleﬁn

isomerization catalyzed by (^tBuPCP)Ir; indeed, it is somewhat

slower at the low concentrations of oleﬁn present during typical

AM experiments. Thus, quite surprisingly, the diﬀerence

between these two catalysts in the product distribution

obtained from alkane metathesis apparently does not result

from a diﬀerence in oleﬁn isomerization rates. In view of these

results, we have begun to investigate in detail the

regioselectivity of dehydrogenation by these two catalysts.

Indeed, preliminary results indicate that (^tBuPOCOP)Ir shows

much less selectivity than either (^tBuPCP)Ir or (^iPrPCP)Ir for

dehydrogenation at the terminal position of the n-alkanes.²⁶

Intramolecular H/D Scrambling. A sample of

Scheme 5. Alternative π-Allyl Pathway for 1-Alkene

Isomerization, Involving Associative Displacement of

2-Alkene by 1-Alkene

(

^tBuPOCOP)Ir(propene-d₃) (3-d₃) was prepared containing a

deuterium-labeled methyl group in the propene ligand.

1

Monitoring a mesitylene-d₁₂solution of 3-d₃by H NMR

revealed no H/D exchange at room temperature over the

course of 24 h. However, heating the solution at 60 °C resulted

in H/D scrambling between the methyl group and the terminal

carbon of propene (Scheme 6 and Figure 3). The reaction

approaches equilibrium with a half-life of ca. 9 h, corresponding

to a free energy of activation, ΔG^⧧, of ca. 26.7 kcal/mol.

Increasing the temperature to 125 °C signiﬁcantly increases

the rate, and the reaction reaches equilibrium in ca. 10 min

(cf. comparable rates, on the order of 0.2 catalytic turnover/min,

observed for α-oleﬁn isomerization catalyzed by either

for self-exchange for (^tBuPOCOP)Ir(oleﬁn) were determined by

EXSY spectroscopy to be 29.4 1.2 and 26.0 1.2 kcal/mol

for 1-hexene and trans-2-hexene, respectively, while the

respective activation entropies were found to be 15 3 and

17 3 eu (from rates measured at temperatures between 107

and 126 °C and 57 and 79 °C, respectively). These oleﬁn

dissociation rates are in fact much slower than those of the

corresponding (^tBuPCP)Ir complexes. Nevertheless, extrapola-

tion to 125 °C for trans-2-hexene self-exchange (experimentally

determined at increments from 57 to 79 °C) aﬀords a rate of

2.1 × 10²s⁻¹, which is independent of excess oleﬁn con-

centration and still much faster than the isomerization of

(

^tBuPOCOP)Ir or (^tBuPCP)Ir).

The H/D exchange is proposed to occur via the Ir(III) allyl

deuteride/hydride intermediate, 4-d₃, in Scheme 6. As in the

case with α-oleﬁn noted above (in either alkane or arene

solvent), in the presence of excess propene, no iridium

dihydride is observable. Since mesitylene was used as solvent

in this experiment, there would be even less dihydride present

than if the solvent were alkane (cf. eq 2).

Insertion mechanisms for isomerization involve the addition

of a hydride derived from one oleﬁn molecule to the double

bond of a second oleﬁn molecule; that is, the mechanism is

(

^tBuPOCOP)Ir(trans-2-hexene) to (^tBuPOCOP)Ir(1-hexene)

(discussed below). Thus, at present, we have no good

explanation for the curvature seen in the plot of Figure 2

Scheme 6. Hydrogen/Deuterium Scrambling via an Ir(III) η³-Allyl Hydride Intermediate

13279

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

1

Figure 3. H NMR overlay showing the H/D scrambling between the methyl and methylene sites of the propene ligand in 3-d₃.

Scheme 7. Ligand Exchange Reaction of 3-d₆and Propene

Scheme 8. H/D Scrambling at the Central Carbon of Propene Involving a Metallacyclobutane Intermediate

intermolecular in contrast to the allyl mechanism in which a

hydride eﬀectively undergoes an intramolecular 1,3 shift. With

this in mind, (^tBuPOCOP)Ir(propene-d₆) (3-d₆) was treated

with 2 equiv of free perprotio-propene in mesitylene-d₁₂at

60 °C (Scheme 7). If an iridium hydride/deuteride species were

present in the system and “isomerization” (degenerate 1,2 shift

13280

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Scheme 9. Formation of (POCOP)Ir(H)(η³-Allyl) and (POCOP)Ir(η²-Propene)

of the double bond) proceeded through a hydride addition

mechanism, then intermolecular H/D scrambling would occur.

However, only ligand exchange was observed in this experi-

ment. Complex 3-d₆underwent ligand substitution with

propene (t_1/2ca. 1 h, 60 °C), giving a mixture of 1/3 equiv

of 3-d₆and 2/3 equiv of 3-d₀, as well as 2/3 equiv of propene-

d₆and 4/3 equiv of propene-d₀(the expected statistical ratio of

products; Scheme 7). No crossover of H/D between the

perdeuterio and perprotio species occurred under the reaction

conditions after 24 h. These results strongly argue against the

hydride addition mechanism for the H/D exchange reaction,

while a π-allyl mechanism, as shown in Scheme 6, fully accounts

for the intramolecular H/D scrambling.

Interestingly, in addition to the H/D exchange between the

propene terminal vinyl and methyl positions, we also observed

deuterium incorporation into the central methine position in the

(^tBuPOCOP)Ir(propene-d₃) (3-d₃) (Scheme 8 and Figure 3). This

scrambling process is revealed more clearly upon thermolysis of 3-

d₁(deuterium-substituted at the central methine site, Scheme 8).

Heating of this d₁species in mesitylene at 60 °C resulted in

incorporation of ¹H into the central site with a half-life of ca. 230 h

(Scheme 8 and Figure 19 in Experimental Details). This rate is

substantially slower than the H/D exchange between the terminal

carbon sites (t_1/2= ca. 9 h). The reaction presumably proceeds via

a metallacyclobutane intermediate (5), as depicted in Scheme 8.

Similar iridium(III) metallacyclobutane species have been reported

by Bergman,²⁷Ibers,²⁸and Stryker.²⁹

temperature, indicating that these two species are kinetic

products and that, at −88 °C, 3 and 4 form via independent

routes. We propose that at −88 °C 4 is formed via 1,2 insertion

of allene complex 6 and that 3 is formed by 2,1 insertion to

yield 2-propenyl hydride 7, followed by reductive elimination

(Scheme 9). Allene complex 6 is undetected during the

reaction, so it cannot be determined whether it is formed

reversibly or if its formation is rate-determining in producing 4

and 3.

When the mixture was warmed to −58 °C, 4 rearranged to

form 3 through the migration of the hydride to the terminal

carbon. The reaction is ﬁrst-order with a rate constant of 5 ×

10⁻⁴s⁻¹, corresponding to ΔG^⧧= 15.6 kcal/mol. A similar

hydrido(η³-allyl)iridium(III) complex, (η⁵-C₅Me₅)IrH(η³-allyl),

has been reported by McGhee and Bergman.²⁷In contrast to 4,

this hydrido(allyl) complex is thermally stable and could be

isolated; the 16-electron complex (η⁵-C₅Me₅)Ir(propene) was

postulated as a reversibly formed intermediate in the oxidative

addition of benzene to the iridium(I) center formed reversibly

from the η³-allyl hydride.²⁷The observation of the allyl hydride

species 4 and its rapid rearrangement to propene complex 3

provides additional evidence in support of a π-allyl oleﬁn

isomerization mechanism.

Stoichiometric Isomerization of Coordinated 2-Alkene

to 1-Alkene. Closely related to the 1,3-H(D) shift observed

for coordinated propene, stoichiometric and apparently intra-

molecular isomerizations of hexene were observed with

both (^tBuPCP)Ir and (^tBuPOCOP)Ir complexes. Interestingly,

the isomerizations proceed in the reverse of the direction

thermodynamically favored for the free oleﬁns.

Observation of an Ir(III) η³-Allyl Hydride Intermediate.

The intramolecular H/D exchange reactions are fully consistent

with a π-allyl oleﬁn isomerization mechanism. In addition,

the key intermediate in this isomerization, the Ir(III) η³-allyl

hydride species (4), has been generated separately and

characterized by low-temperature NMR techniques. Treatment

of dihydride complex 2 with allene (5 equiv) in methyl-

cyclohexane-d₁₄at −88 °C generated 18% of hydrido(allyl)

complex 4 and 82% of propene complex 3 (Scheme 9). The

At 60 °C, (^tBuPOCOP)Ir(η²-trans-2-hexene) underwent

isomerization in p-xylene solvent to give (^tBuPOCOP)Ir(1-

hexene). Free trans-2-hexene is more stable than 1-hexene by

ca. 2.5 to 3 kcal/mol.³⁰The diﬀerence in their binding

enthalpies must therefore exceed that value; presumably, this is

driven by decreased crowding in the 1-hexene complex. The

reaction kinetics were monitored by ³¹P NMR spectroscopy

and a ﬁrst-order rate constant of 5.6 × 10⁻⁶s⁻¹was obtained

using COPASI³¹kinetics ﬁtting software (Figure 4). The rate of

the (^tBuPOCOP)Ir(trans-2-hexene) isomerization is somewhat

slower than, but quite comparable to, that of 1,3-D migration

of (^tBuPOCOP)Ir(propene-d₃), which has a rate constant of

ca. 2.1 × 10⁻⁵s⁻¹at 60 °C. An Eyring plot of isomerization rates

of (^tBuPOCOP)Ir(trans-2-hexene) to (^tBuPOCOP)Ir(1-hexene),

based on rates measured at 10 degree increments from 60 to

1

characteristic H NMR data for complex 4 include a hydride

2

resonance at δ −13.1 ppm (triplet, J_P−H= 15.0 Hz) and ﬁve

separate H signals for the allyl unit H_anti, 2.0, 2.1; H_syn, 2.6, 2.8;

H₂, 4.9 ppm. The ³¹P NMR spectrum shows a pair of AB

pattern doublets at 158.5 and 152.5 ppm (²J_P−P= 341 Hz) for

two inequivalent phosphorus nuclei. The 1:5 ratio of complexes

4 and 3 is constant throughout the course of their formation

from 2 and persists unchanged for hours in solution at −88 °C.

Complex 4 rearranges quantitatively to 3 but only at higher

13281

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Figure 6. Isomerization of (^tBuPCP)Ir(trans-2-hexene) to (^tBuPCP)Ir-

(1-hexene) at 25 °C.

Figure 4. Isomerization of (^tBuPOCOP)Ir(trans-2-hexene) to

(

^tBuPOCOP)Ir(1-hexene) at 60 °C; curve shown represents a best

ﬁt to a ﬁrst-order rate constant of 5.6 × 10⁻⁶s⁻¹obtained using

COPASI³¹kinetics ﬁtting software.

greater than that of the stoichiometric reaction. Note that under

catalytic conditions the presumed uphill isomerization of bound 1-

hexene is followed by the very rapid loss of trans-2-hexene

(extrapolated to a rate of ca. 2 × 10⁴s⁻¹at 125 °C) and then

addition of free 1-hexene.

90 °C, gave activation parameters ΔH^⧧= 23.7( 0.7) kcal/mol

and ΔS^⧧= −9.3( 1.1) eu (Figure 5).

As will be discussed in detail below, the most signiﬁcant

diﬀerence between the (^tBuPCP)Ir and (^tBuPOCOP)Ir frag-

ments appears to be greater crowding in the former. The fact

that isomerization of bound trans-2-hexene proceeds more

rapidly in the case of (^tBuPCP)Ir therefore suggested that

the rate-determining transition states are less crowded than

the ground states (the oleﬁn π-complexes). In this context, we

investigated the eﬀect of increasing the steric demands of the

oleﬁn substrate by monitoring the isomerization kinetics of a

bound branched oleﬁn, trans-4-methylpent-2-ene (MP-2-ene).

(

^tBuPOCOP)Ir(MP-2-ene) was found to undergo isomer-

ization to the 4-methyl-1-pentene complex (^tBuPOCOP)Ir(MP-

1-ene) (eq 3, X = O) analogously to the isomerization of

Figure 5. Eyring plot for the isomerization of (^tBuPOCOP)Ir(trans-2-

hexene) to (^tBuPOCOP)Ir(1-hexene) in the temperature interval 60 to

90 °C.

The analogous isomerization reaction of (^tBuPCP)Ir(trans-2-

hexene) was also observed, and its kinetics were monitored. As

noted above, at 60 °C, dissociation of 1-hexene and trans-2-

hexene from (^tBuPCP)Ir is rapid on the NMR time scale,

resulting in broadening of the ³¹P NMR signals and impeding

kinetics measurements. The kinetics could be measured at

25 °C, however, and a ﬁrst-order rate constant of 3.2( 0.4) ×

10⁻⁵s⁻¹was obtained (Figure 6), corresponding to ΔG^⧧=

23.6 kcal/mol. This compares with the kinetics of isomerization

of the (^tBuPOCOP)Ir analogue, which can be extrapolated to

25 °C (based on the Eyring plot of Figure 5) to give a value of

2.5 × 10⁻⁷s⁻¹(2 orders of magnitude slower).

(

^tBuPOCOP)Ir(trans-2-hexene).Under the same conditions as

the reaction of the 2-hexene complex (60 °C, p-xylene

solution), the kinetics of the MP-2-ene complex were

monitored and the reaction was found to proceed with a

ﬁrst-order rate constant of 5 × 10⁻⁴s⁻¹(Figure 7), ca. 90-fold

faster than the reaction of the trans-2-hexene complex.

Assuming that the two complexes undergo isomerization

through analogous pathways, the greater rate of the branched

oleﬁn complex provides further evidence that the rate-

determining transition states are sterically less demanding

than the respective oleﬁn π-complex reactants.

If we assume the same value of ΔS^⧧for the (^tBuPCP)Ir(trans-

2-hexene) and (^tBuPOCOP)Ir(trans-2-hexene) isomerizations,

the rate of the (^tBuPCP)Ir(trans-2-hexene) isomerization

extrapolates to ca. 0.3 s⁻¹at 125 °C, with ΔG^⧧= 24.5 kcal/mol.

This compares with the rate of isomerization of 1-hexene to

trans-2-hexene, catalyzed by (^tBuPCP)Ir, 4.2 × 10⁻³s⁻¹at 125 °C,

corresponding to a barrier of ΔG^⧧= 27.9 kcal/mol. The

somewhat greater free energy of activation for the isomerization of

1-hexene to trans-2-hexene is consistent with our general

mechanistic picture which proposes (Scheme 4) that the catalytic

reaction proceeds via isomerization of (^tBuPCP)Ir(1-hexene) to

(^tBuPCP)Ir(trans-2-hexene). This step must be endergonic by

several kcal/mol since the reverse (stoichiometric) reaction

proceeds to apparent completion; therefore, the free energy

barrier of the forward (catalytic) reaction must be several kcal/mol

Analogously, (^tBuPCP)Ir(MP-2-ene) underwent isomeriza-

tion to give (^tBuPCP)Ir(MP-1-ene) (eq 3, X = CH₂) with a rate

constant of 7.9 × 10⁻⁵s⁻¹at 25 °C (Figure 8). This also is

more rapid than isomerization of the respective linear oleﬁn

complex, (^tBuPCP)Ir(trans-2-hexene), although only by a factor

of 2.5.

COMPUTATIONAL STUDIES AND MECHANISTIC

ANALYSIS

■

The isomerization or double-bond migrations discussed above

were modeled with DFT calculations³²using two of the most

widely applied functionals in transition metal chemistry, PBE³³

13282

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Table 1. Experimental and Calculated Parameters for Oleﬁn

Dissociation from Complexes of (^tBuPCP)Ir and

(

^tBuPOCOP)Ir

oleﬁn complex

^tBuPCP)Ir(1-hexene)

^tBuPCP)Ir(trans-2-hexene)

^tBuPOCOP)Ir(1-hexene)

^tBuPOCOP)Ir(trans-2-hexene)

ΔH^⧧(exp) ΔH° (PBE) ΔH° (M06)

(

21.4

15.4

29.4

26.0

20.6

12.7

29.6

24.8

30.1

24.5

36.9

35.0

than the experimental activation enthalpies, most likely

as the result of overestimating the strength of Ir−oleﬁn

π-interactions and/or underestimating the strength of steric

interactions.

Figure 7. Isomerization of (^tBuPOCOP)Ir(MP-2-ene) to (^tBuPOCOP)-

Ir(MP-1-ene) at 60 °C.

An accurate calculation of the energies of the π-complexes

is obviously important in obtaining reaction energies and

modeling kinetics, as the π-complexes are the resting state or

reagent in all of the isomerizations. Moreover, the superiority of

PBE in such calculations might well be expected to extrapolate

to the various intermediates and transition states in the

isomerization pathways since these species may have much in

common with the oleﬁn complexes. In this section, we will

therefore refer exclusively to values calculated using the PBE

functional. For all systems studied, however, calculations using

the M06 functional were also performed, and those results can

be found in Supporting Information. Importantly, the results

obtained with either functional support the same overarching

conclusions regarding the preferred isomerization pathway.

As noted above, the experimental kinetics of 1-octene

isomerization catalyzed by (^tBuPOCOP)Ir show a clear positive

dependence on 1-octene concentration, contrary to the proposed

mechanism for isomerization by (^tBuPCP)Ir (Scheme 4). As such,

calculations for a proposed mechanism for catalytic isomerization

by (^tBuPOCOP)Ir may not be applicable to the actual mechanism

of isomerization by (^tBuPOCOP)Ir. Accordingly, calculations for

the (^tBuPOCOP)Ir system are presented below primarily for

comparison with (^tBuPCP)Ir.

Figure 8. Isomerization of (^tBuPCP)Ir(MP-2-ene) to (^tBuPCP)Ir(MP-

1-ene) at 25 °C.

and M06³⁴(see Computational Methods for a full description).

The PBE functional contains pure DFT functionals based on

the generalized gradient approximation (GGA); it is parameter

independent, very robust, and eﬃcient. The more recently

developed M06 functional is a hybrid functional (28% exact

exchange), includes kinetic energy and dispersion corrections,

and is partially parametrized against experimental data.

Calculations were conducted on the reactions of (^tBuPCP)Ir

and (^tBuPOCOP)Ir fragments with propene, 1-hexene/trans-2-

hexene, and MP-1-ene/MP-2-ene.

Hydride Addition Mechanism. The pathways for

1-hexene isomerization via the hydride mechanism for both

(

^tBuPOCOP)Ir and (^tBuPCP)Ir (Scheme 3) were calculated

The two diﬀerent functionals employed here generally

yielded similar results with respect to the relative energies of

the diﬀerent pathways investigated, but the PBE functional

appeared to aﬀord considerably better agreement with those

absolute values that could be determined experimentally. For

example, using the PBE functional, the enthalpies of oleﬁn

binding, ΔH°, for (^tBuPCP)Ir(1-hexene), (^tBuPCP)Ir(trans-2-

hexene), (^tBuPOCOP)Ir(1-hexene), and (^tBuPOCOP)Ir(trans-2-

hexene) were calculated to be 20.6, 12.7, 29.6, and 24.8 kcal/

mol, respectively, while M06 gave values of 30.1, 24.5, 36.9,

and 35.0 kcal/mol (Table 1). As noted above, the activation

enthalpies for dissociation of the oleﬁns from the (pincer)Ir

fragments were determined experimentally, by EXSY NMR

spectroscopy. Assuming that the reverse reaction, oleﬁn

addition, has a positive enthalpy of activation, the (thermody-

namic) enthalpies of binding must be less than the activation

enthalpies of dissociation. Indeed, the correlation of exper-

imental values of ΔH^⧧with PBE-based values of ΔH° is

striking; in all four cases, the experimental ΔH^⧧for dissociation is

greater, by less than 3 kcal/mol, than the calculated value of ΔH°,

consistent with an expectedly small but positive value of ΔH^⧧

for oleﬁn addition to (pincer)Ir. In contrast, the M06-based

binding enthalpies are signiﬁcantly greater (by 7−9 kcal/mol)

using DFT. The calculations indicate that the barrier to the

hydride pathway for isomerization is prohibitively high, in

accord with the body of experimental results discussed above

arguing against a hydride mechanism for double-bond

migration. The barrier to insertion of oleﬁn into the Ir−H

bonds, to give (pincer)Ir(alkyl)(H), was found to be quite high

(ΔG^⧧> 30 kcal/mol for both (^tBuPOCOP)Ir and (^tBuPCP)Ir

and for propene and 1-hexene), but the highest (rate-

determining) barrier was found for subsequent rotation of the

alkyl group about the Ir−C bond with calculated free energy

barriers >40 kcal/mol above the corresponding η²-oleﬁn

complexes (44.7 and 46.3 kcal/mol for (^tBuPCP)Ir complexes

of propene and 1-hexene, respectively). By contrast, the

experimental value obtained for catalytic isomerization of

1-hexene by (^tBuPCP)Ir was found to be ΔG^⧧= 27.9 kcal/mol,

while the experimental value for the stoichiometric isomer-

ization of (^tBuPCP)Ir(trans-2-hexene) to (^tBuPCP)Ir(1-hexene)

at 25 °C was ΔG^⧧= 23.6 kcal/mol. A more detailed treatment

of the hydride pathway (including calculations for propene

isomerization) can be found in Supporting Information.

Comparison of PCP and POCOP Complexes: Steric

Eﬀects. Although both (^tBuPCP)Ir and (^tBuPOCOP)Ir

complexes show prohibitively high barriers for the hydride

13283

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

addition pathway, there are signiﬁcant diﬀerences between the

two in the factors determining the barrier magnitudes. In

particular, (^tBuPOCOP)IrH₂is ca. 5 kcal/mol higher in free

energy than (^tBuPCP)IrH₂, relative to their respective propene

or 1-hexene complexes. This accounts for the observation that

under alkane metathesis conditions (^tBuPOCOP)Ir is only

observable as the oleﬁn complex, while (^tBuPCP)Ir is only

observable as the dihydride. An investigation into the origin of

this diﬀerence, detailed below, leads to the conclusion that it is

largely due to steric factors.

Scheme 11. Space-Filling Models of Calculated Structures of

(

Indicated

^tBuPCP)Ir and (^tBuPOCOP)Ir with Closest H−H Distances

Although the two (pincer)Ir fragments are formally

isostructural, a comparison of the geometries of (^tBuPCP)Ir

and (^tBuPOCOP)Ir complexes reveals signiﬁcant diﬀerences.

The smaller size of the O-linker versus CH₂combined with the

tendency of O to form shorter bonds but greater bond angles

results in the phosphino groups being “pulled back” in

Scheme 12. Space-Filling Models of Calculated Structures of

(

^tBuPCP)IrH₂and (^tBuPOCOP)IrH₂(Ir-Bound Hydrides

Yellow) with H(^tBu)−H(Hydride) Distances Indicated

(

^tBuPOCOP)Ir. Thus, for the free (pincer)Ir fragments, the

calculated C_aryl−O and O−P bond lengths are 1.39 and 1.71 Å,

respectively, in (^tBuPOCOP)Ir, whereas the C_aryl−CH₂and

CH₂−P bond lengths in (^tBuPCP)Ir are longer at 1.51 and

1.86 Å, respectively (Scheme 10). The C_aryl−O−P bond angle

Scheme 10. Selected Distances (Å) and Angles in Calculated

Structures of (^tBuPCP)Ir and (^tBuPOCOP)Ir

distances (both 2.384 Å, just barely below 2.4 Å), indicating that if

there is any crowding in (^tBuPOCOP)IrH₂it is signiﬁcantly less

than in (^tBuPCP)IrH₂.

In order to further probe the energetic eﬀects of crowding,

we considered complexes in which the pincer t-Bu groups were

replaced with methyl groups (Table 2). The enthalpy of addition

of H₂to (^MePCP)Ir and (^MePOCOP)Ir diﬀers very little from that

of the t-Bu analogues. Moreover, the small diﬀerence between the

Me and t-Bu analogues is essentially equal for ^RPCP and

^RPOCOP complexes (1.2 and 1.1 kcal/mol favoring addition to

the complexes where R = Me). We conclude that there is no

signiﬁcant steric contribution to the energetics of H₂addition to

either complex. Given the short H−H nonbonding distances in

(^tBuPCP)IrH₂, it seems that H₂just barely “ﬁts” into the cleft

between the opposing (^tBuPCP)Ir phosphino-t-butyl groups; this

would suggest that addition of any substrate larger than H₂to

(^tBuPCP)Ir would result in signiﬁcant crowding and more so than

in the case of addition to (^tBuPOCOP)Ir.

is 113.7° in (^tBuPOCOP)Ir, whereas the C_aryl−CH₂−P angle is

108.3° in (^tBuPCP)Ir. These diﬀerences lead to smaller C_aryl

−

Ir−P angles in (^tBuPOCOP)Ir (81.8°) as compared with 84.9°

in (^tBuPCP)Ir. Thus, the P−Ir−P angle is calculated to be

approximately 6° smaller in (^tBuPOCOP)Ir than in (^tBuPCP)Ir

(163.6 vs 169.9°, respectively).

Experimental values for various (^tBuPOCOP)Ir and (^tBuPCP)-

Ir complexes indeed reveal slightly greater P−Ir−P angles for

the latter. For example, (^tBuPOCOP)IrHCl³⁵and (^tBuPCP)-

IrHCl³⁶complexes have P−Ir−P angles of 159.79(5) and

164.27(4)°, while the corresponding dinitrogen complexes have

P−Ir−P angles of 154.86(3)³⁷and 160.22(10)°,³⁸respectively.

While these structural diﬀerences may not seem large, the

resulting gap between the bulky phosphino-t-butyl groups is

substantially less in (^tBuPCP)Ir than in (^tBuPOCOP)Ir in the

region of the open coordination site (the site opposite to the

pincer aryl group). The closest distances between hydrogens of

the two phosphino groups ﬂanking the vacant coordination sites in

(^tBuPCP)Ir and (^tBuPOCOP)Ir are 3.01 and 4.43 Å, respectively

(Scheme 11). Taking the van der Waals radius of hydrogen as

1.2 Å^39,40would suggest a space of only 0.61 Å between the

corresponding van der Waals surfaces in (^tBuPCP)Ir.

Accordingly, whereas H₂adds to (^MePCP)Ir slightly more

favorably than to (^tBuPCP)Ir (by 1.2 kcal/mol), the addition of

1-hexene to (^tBuPCP)Ir is 13.8 kcal less favorable than to

(^MePCP)Ir, indicative of very signiﬁcant crowding. Addition of

1-hexene to (^tBuPOCOP)Ir is likewise less favorable than to the

^MePOCOP complex, but only by 9.2 kcal/mol. These results

suggest that crowding is important in both (^tBuPCP)Ir and

(

^tBuPOCOP)Ir 1-hexene complexes, but the penalty is about

4.6 kcal/mol greater for (^tBuPCP)Ir. Accordingly, addition of H₂

to (^tBuPCP)Ir is 8.4 kcal/mol more favorable than 1-hexene

addition, while the corresponding value is 3.6 kcal/mol for

As a result of the small gap between the trans-di-t-butyl-

phosphino groups of (^tBuPCP)Ir, addition of even the smallest

substrate, H₂, results in a complex with four contacts (2.288,

2.289, 2.331, 2.332 Å) between t-Bu hydrogen atoms and

the resulting hydrides that are less than twice the van der Waals

radius of hydrogen, suggesting that even H₂addition to

(^tBuPCP)Ir might be disfavored by steric crowding (Scheme 12).

In contrast, in (^tBuPOCOP)IrH₂, there are only two short H−H

(

^tBuPOCOP)Ir (eq 4). This diﬀerence of 4.8 kcal/mol (or

5.4 kcal/mol for the propene complexes; see Figure 9) can thus

13284

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

(

^tBuPCP)Ir(trans-2-hexene) than in (^tBuPOCOP)Ir(trans-2-

Table 2. Calculated Thermodynamic Parameters for

Addition of H₂and Hexenes to (^RPCP)Ir and (^RPOCOP)Ir,

hexene).

The signiﬁcantly greater steric crowding in (^tBuPCP)Ir versus

^tBuPOCOP)Ir complexes presumably has important implica-

a

t

R = Bu, Me, and H

(

addition product

ΔE

ΔH

ΔG

ΔS

tions in any comparison of the two fragments and indeed in

almost any reaction involving (^tBuPCP)Ir complexes in

particular.

(

^tBuPCP)IrH₂

−31.2

−22.9

−15.0

−32.0

−35.4

−33.2

−31.2

−36.8

−35.1

−35.3

−31.4

−27.2

−36.2

−40.0

−38.1

−36.3

−42.5

−41.4

−29.0

−20.6

−12.7

−30.2

−34.4

−32.2

−29.4

−35.7

−34.0

−33.2

−29.6

−24.8

−34.3

−38.8

−37.0

−34.4

−41.3

−40.4

−20.3

−5.3

2.6

−29.2

−51.9

−51.3

−27.8

−44.3

−46.1

−27.6

−42.5

−42.9

−27.9

−48.4

−53.3

−28.4

−42.7

−44.4

−28.9

−42.8

−44.1

^tBuPCP)Ir(1-hexene)

^tBuPCP)Ir(trans-2-hexene)

^MePCP)IrH₂

^MePCP)Ir(1-hexene)

^MePCP)Ir(trans-2-hexene)

(^HPCP)IrH₂

Allyl Mechanism for Oleﬁn Isomerization. Isomer-

ization mechanisms proceeding through η³-allyl intermediates,

while often proposed to be operative,^1,2,10−14have not been the

subject of thorough mechanistic studies or even detailed

mechanistic proposals. Formation of the η³-allyl hydride is

generally assumed to proceed via γ-hydride migration from a

π-coordinated oleﬁn; we will refer to this as an η²-η³step. If the

resulting η³-allyl ligand is symmetrical or approximately

symmetrical with respect to the hydride, migration of the

hydride to the terminal carbon C_b(Scheme 13a) is equivalent

−21.9

−21.2

−18.5

−21.2

−23.2

−21.3

−24.9

−15.2

−8.9

−25.8

−26.1

−23.8

−25.8

−28.6

−27.2

(^HPCP)Ir(1-hexene)

(^HPCP)Ir(trans-2-hexene)

(

^tBuPOCOP)IrH₂

^tBuPOCOP)Ir(1-hexene)

^tBuPOCOP)Ir(trans-2-hexene)

^MePOCOP)IrH₂

^MePOCOP)Ir(1-hexene)

^MePOCOP)Ir(trans-2-hexene)

(^HPOCOP)IrH₂

Scheme 13. Schematic Illustration of the π-Allyl Pathway for

Oleﬁn Isomerization Proceeding via Formation of an Allyl

Ligand That Is (a) Symmetrical or (b) Unsymmetrical with

Respect to the Hydride

(^HPOCOP)Ir(1-hexene)

(^HPOCOP)Ir(trans-2-hexene)

a

Units are kcal/mol for ΔE, ΔH, and ΔG; units are cal/(deg-mol)

for ΔS.

to the reverse of the η²-η³step and results in double-bond

migration.

Alternatively, however, the hydride may be located to one

“side” of the resulting allyl in such a way that it can migrate only

to the same terminal carbon to which it was originally bound,

that is, terminal carbon C_a(Scheme 13b), thus requiring one or

more additional rearrangement steps in order to eﬀect a net

double-bond migration. The calculations clearly indicate that

the allyl carbons of the (pincer)Ir(η³-allyl)(H) complexes are

inequivalent in this sense with respect to the hydride, in accord

with the spectroscopic characterization of (^tBuPOCOP)Ir(η³-

allyl)(H) (4).

Conceptually, the simplest rearrangement of the η³-allyl

ligand that would permit subsequent migration of the hydride

to the γ-carbon would be a 180° rotation during which the allyl

group remains bound in an η³fashion. Such a process has

in fact been well-established by ﬂuxional NMR studies.^42,43

Our calculations indicate, however, that the barriers for such

a rotation in the pincer−Ir systems are prohibitively high,

59.5 kcal/mol for (^tBuPOCOP)Ir(η³-propenyl)(H) (Figure 9)

and 41.0 kcal/mol for (^tBuPCP)Ir(η³-propenyl)H. (In the

latter case, the reaction proceeds through Ir−P bond cleavage,

which is facilitated by relief of severe steric crowding in

the (^tBuPCP)Ir((η³-allyl)H intermediate.) For isomerization of

1-hexene by (^tBuPOCOP)Ir, the barrier was calculated to be

even higher, 62.4 kcal/mol, while we were unable to locate a TS

for rotation of the corresponding (^tBuPCP)Ir intermediate.

An alternative pathway that would eﬀect a net “rotation” of

the allyl group could proceed via an initial η³-η¹“opening” of

the η³-allyl ligand, followed by rotation around the resulting

iridium−carbon σ-bond, and terminated by η¹-η³“closing”, as

Figure 9. An η²-η³rotation pathway (γ-H migration) calculated for

1,3-H migration in (^tBuPOCOP)Ir(η²-propene).

be attributed solely to steric eﬀects. It should be emphasized

that, as indicated above, in the absence of steric factors,

(^RPOCOP)Ir fragments add both oleﬁns and hydrogen more

favorably than (^RPCP)Ir fragments do,⁴¹by ca. 5 kcal/mol, but

these electronic factors do not contribute to the greater relative

preference for olefins versus H₂that is demonstrated by

(

^tBuPOCOP)Ir versus (^tBuPCP)Ir.

In accord with the above considerations, addition of trans-2-

hexene to (^tBuPCP)Ir is 19.5 kcal/mol less favorable than

addition to (^MePCP)Ir, while the diﬀerence between t-Bu and

Me analogues is only 12.2 kcal/mol for (^RPOCOP)Ir. This

suggests that crowding is ca. 7 kcal/mol more severe in

13285

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Figure 10. An η²-η³-η¹rotation pathway (γ-H migration) calculated for 1,3-H migration of bound propene, (^tBuPOCOP)Ir (X = O; blue, italicized

free energy values) and (^tBuPCP)Ir (X = CH₂; red).

shown in Figure 10.⁴⁴⁻⁴⁶(Processes involving η³-η¹-η³

opening/closing have also been proposed previously to account

for syn/anti exchange of allyl ligands; these processes are related

to but distinct from the present pathway.^47,48) The free energy

barriers (25 °C) calculated for such an η²-η³-η¹pathway for

1,3-H migration for propene are 28.7 and 25.0 kcal/mol for

Table 3. Path of η²-η³-η¹(γ-H Migration of η²-Coordinated

Alkene) for Double-Bond Migration

a

pincer ligand

POCOP

PCP

MP-

ene

MP-

ene

alkene

propene hexene

(

^tBuPOCOP)Ir and (^tBuPCP)Ir, respectively (Figure 10 and

Ir + 1-alkene

14.9

0.0

15.2

0.0

15.3

0.0

4.8

0.0

5.3

0.0

5.2

0.0

Ir(η²-1-alkene)

Table 3). At 60 °C, the value for (^tBuPOCOP)Ir is 29.0 kcal/

mol, which compares adequately with the barrier determined

experimentally for 1,3-deuterium migration in (^tBuPOCOP)Ir-

(CD₃CHCH₂) at 60 °C, ca. 26.7 kcal/mol.

η²-η³γ-C3−H

28.7

28.0

28.5

25.0

27.2

26.9

migration TS

Ir(H)(η³-allyl) (H

syn to C3)

13.3

23.2

16.0

18.5

15.7

18.0

13.3

28.7

0.0

13.7

20.2

13.3

16.0

12.9

17.2

16.0

29.9

1.9

17.0

22.7

13.1

18.9

12.8

17.7

17.4

33.0

6.1

8.8

18.3

14.3

17.0

13.0

16.7

8.8

11.0

16.2

11.7

14.5

10.8

17.5

9.6

9.9

17.3

11.5

15.5

10.8

18.2

9.6

η³-η¹-allyl TS (syn

For the isomerization of (pincer)Ir(trans-2-hexene) to

(pincer)Ir(1-hexene), the experimental value obtained for

to H)

Ir(H)(η¹-C1-allyl)

(H syn to vinyl)

(

^tBuPOCOP)Ir(trans-2-hexene) is 26.8 kcal/mol at 60 °C

(26.5 kcal/mol when extrapolated to 25 °C) and 23.6 kcal/mol

for (^tBuPCP)Ir(trans-2-hexene) at 25 °C. The calculated values

for this pathway, 28.0 and 24.6 kcal/mol for (^tBuPOCOP)Ir and

η¹-C1-allyl rotation

TS

Ir(H)(η¹-C1-allyl)

(H anti to vinyl)

(

^tBuPCP)Ir, respectively, are in very good agreement with

η¹-η³-allyl TS (anti

experimental values.

to H)

Alternative Pathway for Allyl Formation: A σ-Complex

Undergoing C−H Addition. The intermediacy of the η¹-allyl

species in the η²-η³-η¹pathway suggested an alternative process

that might avoid the relatively high-barrier η²-η³step, namely,

C−H addition of a non-π-bound oleﬁn leading directly to the η¹

species. Such a pathway is shown in Figure 11 for

Ir(H)(η³-allyl) (H

syn to C1)

η³-η²-γ-C1-H

25.0

0.0

29.0

4.4

30.7

7.4

migration TS

Ir(η²-propene or 2-

alkene)

a

Free energies (kcal/mol, 25 °C) for (^tBuPOCOP)Ir and (^tBuPCP)Ir

complexes of propene, 1-hexene/2-hexene, and MP-1-ene/MP-2-ene

(highest calculated TS free energies for formation of η³-allyl complex,

from each isomer, in bold font).

(

^tBuPOCOP)Ir(η²-propene) and (^tBuPCP)Ir(η²-propene). Ad-

dition of the allylic C−H bond is achieved via a slippage from

π-coordination to a σ-complex of the allylic C−H bond,

followed by C−H bond oxidative addition. The η¹-allyl hydride

can then close to an η³conﬁguration via coordination of the

double bond at the position anti to the hydride ligand. The

η³-η¹opening at the position syn to the hydride then aﬀords

an η¹-allyl hydride in which the carbon atom that initially

was C1 (vinylic) now becomes an sp³carbon bound to iridium.

C−H elimination gives a σ-complex, which slips back to an

η²-propene complex that is equivalent to the starting complex,

with the coordinated propene having undergone double-bond

migration. The overall barriers calculated for this pathway,

23.5 and 21.3 kcal/mol for (^tBuPOCOP)Ir and (^tBuPCP)Ir,

respectively (Figure 11), are signiﬁcantly lower (by 5.2 and 3.7

kcal/mol, Table 5) than those for the η²-η³-η¹rotation pathway.

Extrapolated to 60 °C, the barrier is 23.7 kcal/mol for the 1,3-H

13286

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Figure 11. An η²-η¹-η³pathway (C−H addition) calculated for 1,3-H migration of bound propene. (^tBuPOCOP)Ir (X = O; blue, italicized free

energy values) and (^tBuPCP)Ir (X = CH₂; red). See Supporting Information for pathway connecting the C−H addition transition states with the

propene complex.

shift in (^tBuPOCOP)Ir(η²-propene) via the η²-η¹-η³C−H

addition pathway, in reasonable agreement with the value of ca.

26.7 kcal/mol obtained experimentally for the 1,3-D shift.

Free energy values for the intermediates and TSs for the

η²-η¹-η³pathway shown in Figure 11, as well as the hexene and

methylpentene analogues, are given in Table 4. A summary of

experimentally obtained free energies of activation and the

calculated overall free energy barriers (calculated at 25 °C and

experimental temperatures for comparison purposes) is

presented in Table 5. Molecular structures of the (^tBuPOCOP)-

Ir intermediates and transition states are shown in Figures 12

and 13 ((^tBuPOCOP)Ir was chosen for visual simplicity because

the conﬁguration of the pincer framework is much closer to

planar than that of (^tBuPCP)Ir).

Table 4. Path of η²-η¹-η³(C(sp³)−H Addition of Non-π-

a

coordinated Alkene) for Double-Bond Migration

pincer ligand

POCOP

PCP

MP-

ene

MP-

ene

alkene

propene hexene

Ir + 1-alkene

Ir(η²-1-alkene)

14.9

0.0

15.2

0.0

15.3

0.0

4.8

0.0

5.3

0.0

5.2

0.0

Ir(2-alkene-C3−

H) σ-complex

17.3

23.1

24.9

13.8

17.7

21.7

C3−H add/elim

TS (H anti to

vinyl)

23.5

28.4

29.8

21.3

27.2

31.6

Ir(H)(η¹-allyl) (H

anti to vinyl)

15.7

18.0

21.3

22.7

24.9

27.5

13.0

16.7

20.2

21.9

22.0

30.1

η¹-η³allyl TS (anti

Since both η²-η³and η²-η¹-η³pathways involve cleavage of

the oleﬁn C(3)−H bond (formally as a β-H elimination in the

former case, as an oxidative addition in the latter), it may seem

surprising that the lower energy TS is the one in which the

double bond is not coordinated to iridium. However, there is

extensive precedent for addition to d⁸species in which the TS

for addition to a three-coordinate intermediate is lower in

energy than the TS for addition to the corresponding four-

coordinate species, despite the penalty of one less metal−ligand

bond.⁴⁹The archetypal example is Wilkinson’s catalyst,

Rh(PPh₃)₃Cl, which loses one PPh₃ligand before it adds H₂

and then recoordinates the PPh₃.⁵⁰Likewise, the microscopic

reverse reaction, reductive elimination from six-coordinate d⁶

complexes, typically proceeds via prior ligand loss.⁵¹More

directly related to the present system, we have shown that

addition of aryl C−H bonds ortho to coordinating groups (e.g.,

acetophenone) is more favorable when the ortho functional

group is not coordinated; coordination of the so-called

“directing” group occurs subsequent to C−H addition.⁵²If

the reaction with an alkene to give an allyl hydride is viewed as

to H)

Ir(H)(η³-allyl)

13.3

13.7

17.0

8.8

11.0

9.9

η³-η¹allyl TS (syn

23.2

20.2

22.7

18.3

16.2

17.3

to H)

Ir(H)(η¹-allyl) (H

syn to vinyl)

16.0

21.4

13.3

18.6

13.1

18.5

14.3

11.7

16.1

11.5

16.2

C1−H add/elim

TS (H syn to

vinyl)

19.2

Ir(2-alkene-C1-H)

17.3

0.0

14.7

1.9

14.3

6.1

13.8

0.0

10.5

5.3

10.0

7.4

σ-complex

Ir(η²-propene or 2-

alkene)

a

Free energies (kcal/mol, 25 °C) for (^tBuPOCOP)Ir and (^tBuPCP)Ir

complexes of propene, 1-hexene/2-hexene, and MP-1-ene/MP-2-ene

(highest calculated TS free energies for formation of η³-allyl complex,

from each isomer, in bold font).

the coordination of a double bond and the addition of an allylic

C−H bond, the precedents above would suggest that the η²-η¹

pathway would indeed be more favorable than η²-η³.

13287

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Table 5. Experimental and Calculated Free Energies of Activation (kcal/mol) for the Overall Isomerizations or Double-Bond

Migration (Propene) Reactions (Lowest Energy Calculated Barriers for Each Pathway in Bold Font)

Figure 12. Molecular structures of intermediates and transition states for double-bond migration in (^tBuPOCOP)Ir(propene) via the η²-η³-η¹

pathway (O = red, P = orange, Ir = purple, C = gray, selected H = blue; hydrogens on pincer ligand not shown).

The path from η²-complex to the η¹-allylic C−H bond

addition TS is not straightforward. We calculate that the oleﬁn

could fully (or nearly fully) dissociate from the metal center

and return to form the σ-complex, or the oleﬁn could “slip” to

form a vinylic C−H bond σ-complex, followed by isomerization

or “chain-walking” to give the allylic C−H bond σ-complex

leading to oxidative cleavage.⁵³For clarity, we have omitted

these steps from Figure 11; the details of these calculations can

be found in Supporting Information. In both cases, the

calculated barriers are less than the barrier to double-bond

migration, and therefore, such processes would not be rate-

limiting.

unsymmetrical as depicted in Figures 10 and 11. In particular,

in either of these two pathways, the formal opening or closing

of the allyl group (i.e., the η³/η¹transition) can ﬁrst occur

either syn or anti to the hydride, followed by the reverse

process (closing or opening) occurring anti or syn, respectively.

Thus, there are two propene 1,3-H migration pathways for

each of the two mechanisms; these pathways are simply those

represented by the forward and reverse direction in each

ﬁgure (Figures 10 and 11). In the case of 1,2-double-bond

migration of higher alkenes, however, where the η³-allyl

ligands are not symmetrical, the possibility of two distinct

pathways in each direction results, one initially opening syn

and the other initially opening anti. The diﬀerences in energy

between these pathways, however, are not large. For the case of

Hexene Isomerization. Note that both the η²-η¹-η³and

η²-η³-η¹pathways for propene double-bond migration are

13288

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Figure 13. Molecular structures, unique to the η²-η¹-η³pathway, of intermediates and transition states for double-bond migration in

^tBuPOCOP)Ir(propene) (for those species common to the η²-η³-η¹pathway, see Figure 12). (O = red, P = orange, Ir = purple, C = gray, selected

(

H atoms = blue; hydrogens on pincer ligand not shown.)

Figure 14. Pathway of η²-η³-η¹(γ-H migration) calculated for isomerization of bound 1-hexene/trans-2-hexene, (^tBuPOCOP)Ir (X = O; blue,

italicized free energy values) and (^tBuPCP)Ir (X = CH₂; red).

interconversion of bound 1-hexene and trans-2-hexene, the

pathway with the lower barrier is shown in Figures 14 and 15.

“Mixed” Pathways (Combined η²-η³-η¹and η²-η¹-η³).

Both η²-η³-η¹and C−H addition (η²-η¹-η³) pathways for

isomerization of 1-hexene proceed via the η³-allyl hydride in

which the n-Pr group bound to C3 is syn to the hydride

(Figures 14 and 15). Proceeding from the trans-2-hexene

adduct to that η³-allyl hydride (right to left in Figures 14 and

15), the transition states encountered via the η²-η³-η¹pathway

(29.9 and 29.0 kcal/mol; Figure 14) are much higher in energy

than those in the C−H addition pathway (18.6 and 16.1 kcal/

mol, Figure 15). The pronounced diﬀerences in TS energies are

probably well outside the error limits of the calculations. The

calculated energies of the TSs for the addition/elimination

of the secondary allylic C−H bond of 1-hexene, however, are

fairly high (28.4 and 27.2 kcal/mol, Figure 15) and essentially

equal to the values for the η²/η³transition for the 1-hexene

complex (28.0 and 27.2 kcal/mol, Figure 14). There is of

course no intrinsic reason that the operative pathway cannot be

unsymmetrical in the sense that one segment (i.e., trans-2-

hexene to η³-allyl) proceeds via C−H addition, while the other

segment (1-hexene to η³-allyl) proceeds via an η²/η³step. Thus,

the calculations very strongly support the C−H addition/

elimination step for π-oleﬁn/allyl-hydride interconversion for

one segment of the pathway (i.e., the reaction or formation of

the trans-2-hexene complex). The results are ambiguous for the

reaction or formation of the 1-hexene complex, which involves

addition/elimination of the secondary allylic (C3) C−H bond

of 1-hexene. The overall barriers for isomerization of the trans-

2-hexene complexes, calculated for either a symmetric pathway

(C−H addition/elimination in both directions) or a “mixed”

pathway (C−H addition of 2-hexene and an η³-η²transition to

give 1-hexene) are in good agreement with the experimental

activation free energies (Table 5).

The overall barriers to isomerization of (pincer)Ir(MP-2-

ene) to (pincer)Ir(MP-1-ene) proceeding via the η²-η³-η¹

pathway are calculated as 26.9 and 23.3 kcal/mol for

^tBuPOCOP and ^tBuPCP complexes, respectively (Figure 16).

The barriers for the C−H addition pathway (η²-η¹-η³) are

23.7 and 24.2 kcal/mol (Figure 17), not very diﬀerent from the

13289

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Figure 15. Pathway of η²-η¹-η³(C−H addition) calculated for isomerization of bound 1-hexene/trans-2-hexene, (^tBuPOCOP)Ir (X = O; blue,

italicized free energy values) and (^tBuPCP)Ir (X = CH₂; red).

η²-η³-η¹values, and consistent with experimental values (24.6

and 23.0 kcal/mol, respectively).

calculated diﬀerence is only 1.0 kcal/mol favoring isomerization

of MP-2-ene versus 2-hexene.

Note that if only “symmetrical” pathways were considered

(i.e., those proceeding via the same type of mechanism, either

η²-η³or η²-η¹-η³, for the reaction of both 1-alkene and 2-alkene

species), then the isomerization of (^tBuPCP)Ir(MP-2-ene)

would actually have a PBE-calculated barrier (η²-η³; 23.3

kcal/mol) greater than that calculated for (^tBuPCP)Ir(trans-2-

hexene) (η²-η¹-η³; 21.9 kcal/mol) (Table 5), in contradiction to

the experimentally observed slightly faster isomerization rate of

The computed values for MP-1-ene/MP-2-ene isomerization

are in good agreement with experimental, but as in the case of

1-hexene/2-hexene, they only indicate a clear preference for

initial C−H addition (η²-η¹-η³) step in one direction. For the

MP-2-ene isomer, the initial C−H addition is much more

favorable than the direct η²-η³step. This leads to formation of

the η³-allyl with the oleﬁn i-Pr group syn to hydride (Figures 16

and 17). Connecting the η³-allyl with the MP-1-ene complex,

however, the direct η³/η²path is more favorable owing to the

high energy of the very crowded secondary C−H addition/

elimination TSs. Thus, the overall most favorable isomerization

pathway (for both (^tBuPOCOP)Ir and (^tBuPCP)Ir) is mixed,

involving a direct C−H addition/elimination and a direct η²/η³

step; overall barriers via these mixed pathways are 22.4 and

20.9 kcal/mol, respectively. Note that if these values are

compared with those calculated for linear hexene, the mixed

pathways capture the favorable eﬀect of branching (i.e., the

observation that coordinated MP-2-ene isomerizes more rapidly

than coordinated trans-2-hexene). In particular, the mixed

pathway for (^tBuPOCOP)Ir(oleﬁn) isomerization is calculated

to be 3.7 kcal/mol lower for MP-2-ene than for trans-2-hexene,

which is consistent with the relative rate factor of 90 obtained

experimentally. In the case of (^tBuPCP)Ir, the experimentally

observed eﬀect of branching is much smaller and indeed, the

(

^tBuPCP)Ir(MP-2-ene).

Finally, we note that M06-based calculations also support the

η²-η¹-η³pathway and, in fact, show an even stronger preference

than do PBE-based calculations for η²-η¹-η³relative to η²-η³

paths. Of the six cases considered in this work (double-bond

migrations of propene, 1-hexene/2-hexene or MP-1-ene/MP2-

ene, bound to (^tBuPCP)Ir or (^tBuPOCOP)Ir), the M06-based

calculations favor the η²-η¹-η³sequence in ﬁve cases, by 6 to

8 kcal/mol versus the η²-η³path. (In the case of (^tBuPCP)Ir-

(MP-2-ene) isomerization, the M06-calculated η²-η¹-η³path is

still more favorable, but only by 2.8 kcal/mol. Data are found in

Table S23 in Supporting Information.)

SUMMARY AND CONCLUSION

■

(

We report that oleﬁn isomerization catalyzed by (^tBuPCP)Ir and

^tBuPOCOP)Ir, using the respective dihydrides as catalyst

precursors, occurs by reaction of a (pincer)Ir(η²-oleﬁn)

13290

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Figure 16. Calculated η²-η³-η¹pathway for isomerization of bound 4-methylpent-1-ene/4-methylpent-2-ene, (^tBuPOCOP)Ir (X = O; blue, italicized

free energy values) and (^tBuPCP)Ir (X = CH₂; red).

Figure 17. Calculated η²-η¹-η³pathway for isomerization of bound 4-methylpent-1-ene/4-methylpent-2-ene, (^tBuPOCOP)Ir (X = O; blue, italicized

free energy values) and (^tBuPCP)Ir (X = CH₂; red).

13291

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

complex to give a (pincer)Ir(η³-allyl)(H) intermediate, and not

via the reaction of oleﬁn with (pincer)IrH₂. This conclusion is

supported by in situ observations of the catalyst resting state,

deuterium exchange experiments, and the absence of a solvent

eﬀect (alkane vs arene) on the rate of oleﬁn isomerization. An

Ir(III) η³-allyl hydride intermediate was independently

generated, characterized by low-temperature NMR spectrosco-

py, and shown to isomerize to the η²-propene complex at low

temperature.

Extensive computational DFT studies were conducted using

two widely used combinations of functionals, PBE and M06.

EXSY NMR measurements provided the rates of oleﬁn

dissociation from (^tBuPCP)Ir(1-hexene), (^tBuPCP)Ir(trans-2-

hexene), (^tBuPOCOP)Ir(1-hexene), and (^tBuPOCOP)Ir(trans-

2-hexene). The resulting values of ΔH^⧧for dissociation are all

slightly higher than the PBE-derived enthalpies of binding,

consistent with a small positive activation enthalpy for oleﬁn

addition to the (pincer)Ir fragments; indeed, the correlation

between experimental and PBE-derived values is remarkably

good. In contrast, the M06-derived binding enthalpies are

signiﬁcantly greater than the activation enthalpies for

dissociation. Accordingly, we place greater weight on PBE-

based calculations in evaluating the potential mechanisms of

oleﬁn isomerization by (pincer)Ir fragments.

The DFT calculations support the conclusion that the

hydride mechanism for isomerization is not operative, as the

overall barriers calculated for such a pathway are prohibitively

high. An alternative mechanism in which the oleﬁn complex

reacts to give an η³-allyl hydride, followed by in-plane rotation

of the η³-allyl ligand, and then hydride migration back to give

an overall 1,3-H shift (i.e., double-bond migration) was also

considered, but the barrier to in-plane η³-allyl rotation was

found to be extremely high (ca. 60 kcal/mol or greater).

A mechanism has been investigated in which the coordinated

oleﬁn undergoes (i) migration of an allylic γ-hydrogen to give

an η³-allyl hydride complex (i.e., an η²-η³step), followed by (ii)

η³-η¹allyl “opening” (syn or anti to the hydride), (iii) rotation

around the Ir−C(η¹-allyl) bond, and then (iv) η¹-η³allyl

“closing” (anti or syn to the hydride, respectively), such that the

net eﬀect of steps (ii) to (iv) is the same as an in-plane η³-allyl

rotation, and (v) the reverse of the allylic γ-hydrogen migration

(i.e., an η³-η²step) to give the isomerized coordinated oleﬁn.

PBE-derived barriers for such a pathway are slightly higher than

experimental values, but not clearly outside the error limits of

the calculations. However, PBE-based calculations yielded

barriers that are lower and generally in better agreement with

experiment for a “direct C−H addition” (η²-η¹-η³) pathway. In

this latter case, formation of the η¹-allyl occurs via disruption of

the Ir−oleﬁn π-bond to give an allylic C−H σ-complex, which

then undergoes oxidative cleavage. The resulting η¹-allyl

undergoes η¹-η³allyl closing and then η³-η¹allyl opening in

the reverse sense to eﬀect a net 1,3-Ir shift, followed by

reductive elimination to aﬀord a C−H σ-complex, and then

formation of a π-complex in which the double bond has

migrated relative to the starting π-complex. Calculations using

the M06 functional predicted an even stronger preference than

PBE-based calculations for the “C−H addition” (η²-η¹-η³)

pathway versus the η²-η³mechanism.

Figure 18. Plot of concentration of ln[4] versus time showing ﬁrst-

order kinetics for decay of [4].

the pathways described in this work are relevant in this context

and, to our knowledge, have not been previously proposed: (i)

dissociation of an oleﬁn double bond from the metal center

followed by C−H addition and recoordination of the double

bond to give an η³-allyl group and (ii) a net “rotation” of an

η³-allyl group via an η³-η¹-η³opening−closing mechanism.

EXPERIMENTAL DETAILS

■

General Considerations. All manipulations were carried out

using standard Schlenk, high-vacuum, and glovebox techniques.

Pentane and toluene were passed through columns of activated

alumina. Mesitylene-d₁₂and methylcyclohexane-d₁₄were dried with

4 Å molecular sieves and degassed by freeze−pump−thaw cycles.

n-Hexane and tert-butylethylene were purchased from Aldrich, dried

with LiAlH₄or Na/K, and transferred under vacuum into sealed ﬂasks.

Deuteride propene gases, CD₃CHCH₂, CH₃CDCH₂, and

CD₃CDCD₂, were purchased from CDN and used as received.

Complexes 1⁵⁵and 2³⁷were synthesized as previously reported.

NMR spectra were recorded on BRUKER DRX-400, AVANCE-

400, and BRUKER DRX-500 MHz spectrometers. H and ¹³C NMR

1

spectra were referenced to residual protio solvent peaks. ³¹P NMR

chemical shifts were referenced to an external H₃PO₄standard.

Catalytic 1-Octene Isomerization. First, 3.0 mg of a pincer−

iridium dihydride complex (5.1 mM, (^tBuPCP)IrH₂or (^tBuPOCOP)-

IrH₂) was dissolved with 1-octene (17.0 μL, 108 mM) and mesitylene

(5.0 μL, 36 mM, internal standard) in a volume of n-octane or p-xylene

solvent such that the total volume of the reaction solution was 1 mL.

The reaction solution was transferred into an airtight, septa-sealed vial.

An initial aliquot was taken before heating the reaction to 125 °C.

Reaction data were obtained via GC analysis of aliquots taken at

various times during the reaction.

Isomerization of (^tBuPOCOP)Ir(2-Alkene) (trans-2-Hexene or

trans-4-Methylpent-2-ene). In a J. Young tube, 6.0 mg of

(

^tBuPOCOP)IrH₂(10 μmol) and 3.7 μL of the oleﬁn (29 μmol of

trans-2-hexene or 30 μmol of trans-4-methylpent-2-ene) were dissolved

in 0.5 mL of p-xylene. All volatiles were removed in vacuo, and the solid

residue was dissolved in p-xylene-d₁₀. No free oleﬁns were observed in the

¹H NMR spectrum. The reaction solution was heated to 60 °C in the

NMR spectrometer and was monitored by ³¹P NMR spectroscopy.

Isomerization of (^tBuPCP)Ir(trans-2-Hexene). In a J. Young tube,

3.0 mg of (^tBuPCP)IrH₂(5.1 μmol) and 32.0 μL of tert-butylethylene

(248 μmol) were dissolved in 0.5 mL of p-xylene. The volume of the

reaction solution was reduced in vacuo by 80%, and p-xylene-d₁₀was

then added such that the total volume of the reaction solution was

0.5 mL. Then, 2.5 μL of trans-2-hexene (20 μmol) was added to the

reaction solution. The reaction was monitored by ³¹P NMR

spectroscopy at 25 °C.

Oleﬁn isomerization has often been proposed to proceed via

η³-allyl complexes, but the mechanism of such reactions has

not been well-studied.^1,2,10−14More generally, the formation of

η³-allyl complexes from oleﬁns is a reaction of great importance,

but mechanistically, it is relatively unexplored.⁵⁴Two aspects of

Isomerization of (^tBuPCP)Ir(trans-4-Methylpent-2-ene). In a J.

Young tube, 3.0 mg of (^tBuPCP)IrH₂(5.1 μmol) and 12.3 μL of trans-

4-methylpent-2-ene (100 μmol) were dissolved in 0.5 mL of p-xylene-

d₁₀. The reaction was monitored by ³¹P NMR spectroscopy at 25 °C.

13292

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

1

Figure 19. Overlay of H NMR spectra showing the H/D scrambling between the central carbon and the terminal carbon of the propene ligand.

EXSY Experiments. First, 10.0 mg of a pincer−iridium dihydride

complex (17 μmol) and 10.0 μL of oleﬁn were dissolved in 0.7 mL of

p-xylene-d₁₀. Conversion to the pincer−iridium oleﬁn complex was

complete after 30 min at 25 °C for most pincer−ligand−oleﬁn

combinations. Formation of the (^tBuPCP)Ir(trans-2-hexene) complex

required heating at 50 °C for 30 min to eﬀect complete conversion to

the oleﬁn complex, and the rapid rate of oleﬁn exchange for this

complex necessitated a change in solvent to mesitylene-d₁₂. After

formation of the oleﬁn complex, a vinyl proton of the bound oleﬁn

complex was selectively irradiated in a 1D EXSY experiment, and

magnetization transfer to free oleﬁn was observed at diﬀerent mixing

times. The rate of oleﬁn dissociation was determined from a plot of

magnetization transfer versus mixing time.⁵⁶These 1D EXSY

experiments were performed at diﬀerent temperatures, and an Eyring

plot yielded activation energy parameters ΔH^⧧and ΔS^⧧. Oleﬁn

exchange was conﬁrmed to proceed via a dissociative mechanism by

repeating the EXSY experiment with a much larger excess of free oleﬁn

(4- to 10-fold excess) and observing negligible change in the rate.

Synthesis of 3-d₃, 3-d₁, and 3-d₆. The respective oleﬁns (3−5

equiv) were added to a solution of 2 (5 mg, 8.4 μmol) and toluene

(0.35 mL) in a medium-walled J. Young NMR tube. After 2 h at room

temperature, volatiles were evaporated under vacuum, and the

resulting red solid was dried under vacuum overnight. 3-d₃: ¹H

NMR (162 MHz, 23 °C, Mes-d₁₂) δ 1.15 (virtual triplet, apparent J =

6.4 Hz, 18H, 2 × ^tBu), 1.33 (virtual triplet, apparent J = 6.4 Hz, 18H,

2 × ^tBu), 2.39 (d, ³J_H−H= 8.0 Hz, 1H, CH₂), 3.74 (m, 1H, CH₂), 4.46

(0.35 mL) solution of dihydride complex 2 (9.5 mg, 16.0 μmol) in a

medium-walled J. Young NMR tube. The NMR tube was shaken

quickly, and when the solution was just beginning to melt, the tube

was inserted into the spectrometer at −88 °C. A mixture of 4, 2, and

1

propene complex 3 was observed by H and ³¹P NMR spectroscopy.

After the complete conversion of 2 into 4 and 3, the reaction was kept

at −88 °C and monitored by NMR for 3−5 h. The mixture was then

warmed to −58 °C, and the conversion of 4 to 3 was monitored by ¹H

1

and ³¹P{¹H} NMR spectroscopy (Figure 18): H NMR (500 MHz,

23 °C, methylcyclohexane-d₁₄) δ −13.13 (t, 1H, ²J_P−H= 15.0 Hz, IrH),

2.00 (br, 1H, H_syn), 2.13 (br, 1H, H_syn), 2.64 (br, 1H, H_anti), 2.78 (br,

3

1H, H_anti), 4.94 (br, 1H, H₂), 6.43 (d, J_H−H= 7.5 Hz, 1H, H at the

3

meta-position), 6.48 (d, J_H−H= 7.5 Hz, 1H, H at the meta-position).

The ^tBu groups and one H at the para-position of the arene backbone

were overlapping with those of complexes 2 and 3: ³¹P{¹H} NMR

(202 MHz, 23 °C, methylcyclohexane-d₁₄) δ 152.5 (d, J_P−P= 341 Hz),

158.5 (d, J_P−P= 341 Hz). 13.13 (t, 1H).

Intraligand Deuterium/Hydrogen Exchange. A mesitylene-d₁₂

(0.35 mL) solution of 3-d₃or 3-d₁(see Scheme 8) (8.4 μmol) in a

medium-walled J. Young NMR tube was heated at 60 or 125 °C.

Periodically, the tube was cooled to room temperature and monitored

by NMR spectroscopy (Figure 19).

Ligand Exchange Reaction of 3-d₆. Two equivalents of propene

was added to a mesitylene-d₁₂(0.35 mL) solution of 3-d₆(8.4 μmol)

in a medium-walled J. Young NMR tube. The tube was heated to

60 °C. Periodically, the tube was cooled to room temperature and

monitored by NMR spectroscopy.

3

(m, 1H, CH), 6.67 (d, J_H−H= 8.0 Hz, 2H), 6.84, (t, J_H−H= 8.0 Hz,

1H); ³¹P{¹H} NMR (400 MHz, 23 °C, Mes-d₁₂) δ 179.8 (s). 3-d₁: ¹H

NMR (400 MHz, 23 °C, Mes-d₁₂) δ 1.15 (virtual triplet, apparent J =

6.4 Hz, 18H, 2 × ^tBu), 1.33 (virtual triplet, apparent J = 6.4 Hz, 18H,

COMPUTATIONAL METHODS

■

All calculations used DFT methodologies implemented in the

Gaussian 09 series of electronic structure programs.³²Most of the

data presented in the text result from calculations which employed the

PBE set of functionals.³³We did, however, also examine most

pathways using the collection of functionals denoted M06.³⁴For Ir, we

applied the Hay−Wadt relativistic eﬀective (small) core potential⁵⁷

and the LANL2TZ basis set^57,58augmented by a diﬀuse d-type

function (exponent = 0.07645); all other atoms received 6-311G(d,p)

basis sets.^59,60The results obtained with the PBE functionals were

deemed overall superior to those obtained at the M06 level.

t

2 × Bu), 1.65 (s, 3H, CH₃in propene), 2.39, (s, 1H, CH₂), 3.74 (t,

³J_H‑D= 5.6 Hz, 1H, CH₂), 6.67 (d, ³J_H−H= 8.0 Hz, 2H), 6.84, (t, ³J_H−H

=

8.0 Hz, 1H); ³¹P{¹H} NMR (162 MHz, 23 °C, Mes-d₁₂) δ 179.8 (s).

1

3-d₆: H NMR (400 MHz, 23 °C, Mes-d₁₂) δ 1.15 (virtual triplet,

t

apparent J = 6.4 Hz, 18H, 2 × Bu), 1.33 (virtual triplet, apparent J =

t

3

6.4 Hz, 18H, 2 × Bu), 6.67 (d, J_H−H= 8.0 Hz, 2H), 6.84, (t, J_H−H

=

8.0 Hz, 1H); ³¹P{¹H} NMR (162 MHz, 23 °C, Mes-d₁₂) δ 179.8 (s).

Formation of the Ir(III) Hydride η³-Allyl Complex (4). Two

equivalents of allene was added to a frozen methylcyclohexane-d₁₄

13293

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

Standard geometry optimization procedures were followed to

obtain geometries and potential energies for stationary points along

the reaction paths. Normal mode analysis was performed to verify the

nature of a particular stationary point (minimum or transition state).

The resulting set of vibrational frequencies was employed (without

scaling) to determine zero-point energy corrections. Enthalpies (ΔH,

ΔH^⧧) and Gibbs’ free energies (ΔG, ΔG^⧧; T = 298.15 K, P = 1 atm)

were subsequently obtained from the potential energies (ΔE, ΔE^⧧)

using standard thermodynamic corrections.⁶¹Increased atomic grid

sizes were used for numerical integrations to enhance computational

stability and accuracy⁶²in geometry optimizations and normal mode

calculations (grid = ultraﬁne option).⁶³Geometries of stationary

points and tables containing energetic quantities are available as

Supporting Information. Solvation eﬀects were not considered

explicitly in the calculations since the experiments uniformly were

carried out in nonpolar hydrocarbons.

(15) Goldman, A. S.; Roy, A. H.; Huang, Z.; Ahuja, R.; Schinski, W.;

Brookhart, M. Science 2006, 312, 257−261.

(16) Bailey, B. C.; Schrock, R. R.; Kundu, S.; Goldman, A. S.; Huang,

Z.; Brookhart, M. Organometallics 2009, 28, 355−360.

(17) Burnett, R. L.; Hughes, T. R. J. Catal. 1973, 31, 55−64.

(18) (a) Vidal, V.; Theolier, A.; Thivolle-Cazat, J.; Basset, J.-M.

Science 1997, 276, 99. (b) Le Roux, E.; Taoufik, M.; Coperet, C.; de

́

Mallmann, A.; Thivolle-Cazat, J.; Basset, J.-M.; Maunders, B. M.;

Sunley, G. J. Angew. Chem. 2005, 44, 6755. (c) Basset, J. M.; Coperet,

C.; Lefort, L.; Maunders, B. M.; Maury, O.; Le Roux, E.; Saggio, G.;

Soignier, S.; Soulivong, D.; Sunley, G. J.; Taoufik, M.; Thivolle-Cazat,

J. J. Am. Chem. Soc. 2005, 127, 8604. (d) Le Roux, E.; Taoufik, M.;

Baudouin, A.; Coperet, C.; Thivolle-Cazat, J.; Basset, J.-M.; Maunders,

B. M.; Sunley, G. J. Adv. Synth. Catal. 2007, 349, 231.

(19) (a) Dry, M. E. Catal. Today 2002, 71, 227−241. (b) Schulz, H.

Appl. Catal., A 1999, 186, 3−12.

(20) (a) International Energy Agency (IAE) World Energy Outlook

2008, pp 250−261, http://www.worldenergyoutlook.org/2008.asp.

(b) Ayala, L. F.; Kouassi, J. P. Energy Fuels 2007, 21, 1592.

(21) (a) Schrock, R. R. Chem. Commun. 2005, 2773−2777. (b) Fox,

H. H.; Lee, J. K.; Park, L. Y.; Schrock, R. R. Organometallics 1993, 12,

759−768. (c) Schrock, R. R.; Murdzek, J. S.; Bazan, G. C.; Robbins, J.;

DiMare, M.; O’Regan, M. J. Am. Chem. Soc. 1990, 112, 3875−3886.

(22) Liu, F.; Pak, E. B.; Singh, B.; Jensen, C. M.; Goldman, A. S. J.

Am. Chem. Soc. 1999, 121, 4086−4087.

ASSOCIATED CONTENT

■

S

* Supporting Information

Tables of relative electronic energies, enthalpies, entropies and

free energies, and Cartesian coordinates and absolute energies

for all minima and transition states shown in the ﬁgures. Eyring

plots of EXSY kinetic data. Complete reference 32. This

material is available free of charge via the Internet at http://

pubs.acs.org.

(23) Choi, J.; MacArthur, A. H. R.; Brookhart, M.; Goldman, A. S.

Chem. Rev. 2011, 111, 1761−1779.

(24) Krogh-Jespersen, K.; Czerw, M.; Summa, N.; Renkema, K. B.;

Achord, P. D.; Goldman, A. S. J. Am. Chem. Soc. 2002, 124, 11404−

11416.

AUTHOR INFORMATION

■

Corresponding Author

mbrookhart@unc.edu; kroghjes@rutgers.edu; alan.goldman@

rutgers.edu

(25) Gottker-Schnetmann, I.; Brookhart, M. J. Am. Chem. Soc. 2004,

̈

126, 9330−9338.

(26) Biswas, S.; Brookhart, M.; Goldman, A. S.; Krogh-Jespersen, K.

Manuscript in preparation.

Notes

The authors declare no competing ﬁnancial interest.

(27) (a) McGhee, W. D.; Bergman, R. G. J. Am. Chem. Soc. 1985,

107, 3388. (b) McGhee, W. D.; Bergman, R. G. J. Am. Chem. Soc.

1988, 110, 4246.

(28) Tulip, T. H.; Ibers, J. A. J. Am. Chem. Soc. 1979, 101, 4201−

4211.

(29) Wakefield, J. B.; Stryker, J. M. J. Am. Chem. Soc. 1991, 113, 7057.

(30) NIST Chemistry WebBook; NIST Standard Reference Database

Number 69. http://webbook.nist.gov/chemistry/.

(31) Hoops, S.; Sahle, S.; Gauges, R.; Lee, C.; Pahle, J.; Simus, N.;

Singhal, M.; Xu, L.; Mendes, P.; Kummer, U. Bioinformatics 2006, 22,

3067−3074.

(32) Frisch, M. J.; et al., Gaussian 09, revision A.02; Gaussian, Inc.:

Wallingford, CT, 2009. See Supporting Information for the complete

reference to Gaussian 09.

ACKNOWLEDGMENTS

■

We acknowledge funding by the NSF (Grant CHE-0650456)

as part of the Center for Enabling New Technologies through

Catalysis (CENTC).

REFERENCES

■

(1) Herrmann, W. A.; Prinz, M. In Applied Homogeneous Catalysis

with Organometallic Compounds, 2nd ed.; Cornils, B., Herrmann, W. A.,

Eds.; Wiley-VCH Verlag GmbH: Weinheim, Germany, 2002; Vol. 3, p

1119−1130.

(2) Crabtree, R. H. In The Organometallic Chemistry of the Transition

Metals, 5th ed.; John Wiley & Sons: Hoboken, NJ, 2009; pp 229−231.

(3) Mol, J. C. J. Mol. Catal. A 2004, 213, 39−45.

(4) van Leeuwen, P. W. N. M. In Homogeneous Catalysis:

Understanding the Art; Kluwer Academic Publishers: Dordrecht, The

Netherlands, 2004; pp 101−107.

(5) Schmidt, B. Eur. J. Org. Chem. 2004, 1865−1880.

(6) Otsuka, S.; Tani, K. Transition Metals for Organic Synthesis, 2nd

ed.; Wiley: New York, 2004; Vol. 1, pp 199−209.

(7) Seayad, A.; Ahmed, M.; Klein, H.; Jackstell, R.; Gross, T.; Beller,

M. Science 2002, 297, 1676−1678.

(8) Scarso, A.; Colladon, M.; Sgarbossa, P.; Santo, C.; Michelin, R.

A.; Strukul, G. Organometallics 2010, 29, 1487−1497.

(9) Casey, C. P.; Cyr, C. R. J. Am. Chem. Soc. 1973, 95, 2240−2248.

(10) Manuel, T. A. J. Org. Chem. 1962, 27, 3941−3945.

(11) Alper, H.; LePort, P. C.; Wolfe, S. J. Am. Chem. Soc. 1969, 91,

7553−7554.

(12) Cowherd, F. G.; Von Rosenberg, J. L. J. Am. Chem. Soc. 1969,

91, 2157−2158.

(13) Misono, M.; Grabowski, W.; Yoneda, Y. J. Catal. 1977, 49, 363−

(33) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77,

3865.

(34) Zhao, C. Y.; Truhlar, D. G. Theo. Chem. Acc. 2008, 120, 215−

241.

(35) Gottker-Schnetmann, I.; White, P.; Brookhart, M. J. Am. Chem.

̈

Soc. 2004, 126, 1804−1811.

(36) Punji, B.; Emge, T. J.; Goldman, A. S. Organometallics 2010, 29,

2702−2709.

(37) Gottker-Schnetmann, I.; White, P. S.; Brookhart, M. Organo-

̈

metallics 2004, 23, 1766−1776.

(38) Lee, D. W.; Kaska, W. C.; Jensen, C. M. Organometallics 1998,

17, 1−3.

(39) Bondi, A. J. Phys. Chem. 1964, 68, 441−451.

(40) Pauling, L. The Nature of the Chemical Bond, 3rd ed.; Cornell

University Press: Ithaca, NY, 1960.

(41) Zhu, K.; Achord, P. D.; Zhang, X.; Krogh-Jespersen, K.;

Goldman, A. S. J. Am. Chem. Soc. 2004, 126, 13044−13053.

(42) Becconsall, J. K.; O’Brien, S. Chem. Commun. 1966, 720−722.

(43) Faller, J. W.; Incorvia, M. J. Inorg. Chem. 1968, 7, 840−842.

(44) Fish, R. W.; Giering, W. P.; Marten, D.; Rosenblum, M. J.

Organomet. Chem. 1976, 105, 101−118.

368.

(14) Casey, C. P.; Cyr, C. R. J. Am. Chem. Soc. 1973, 95, 2248−2253.

13294

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Journal of the American Chemical Society

Article

(45) Gibson, D. H.; Hsu, W.-L.; Steinmetz, A. L.; Johnson, B. V. J.

Organomet. Chem. 1981, 208, 89−102.

(46) Ariafard, A.; Bi, S. W.; Lin, Z. Y. Organometallics 2005, 24,

2241−2244.

(47) Sprinz, J.; Kiefer, M.; Helmchen, G.; Reggelin, M.; Huttner, G.;

Walter, O.; Zsolnai, L. Tetrahedron Lett. 1994, 35, 1523−1526.

(48) Trost, B. M.; Van, V. D. L. Chem. Rev. 1996, 96, 395−422.

(49) The classic explanation for the high barrier to addition of H−H

or C−H bonds to four-coordinate d⁸metal centers (or elimination

from six-coordinate d⁶centers) was proposed by Hoﬀmann: Saillard,

J.; Hoffmann, R. J. Am. Chem. Soc. 1984, 106, 2006−2026. For a more

recent lead reference concerning this issue, see: Crumpton-Bregel, D.

M.; Goldberg, K. I. J. Am. Chem. Soc. 2003, 125, 9442−9456.

(50) Halpern, J.; Wong, C. S. J. Chem. Soc., Chem. Commun. 1973,

629−630.

(51) Bartlett, K. L.; Goldberg, K. I.; Borden, W. T. J. Am. Chem. Soc.

2000, 122, 1456−1465.

(52) Zhang, X.; Kanzelberger, M.; Emge, T. J.; Goldman, A. S. J. Am.

Chem. Soc. 2004, 126, 13192−13193.

(53) Vetter, A. J.; Flaschenriem, C.; Jones, W. D. J. Am. Chem. Soc.

2005, 127, 12315−12322.

(54) Jensen, T.; Fristrup, P. Chem.Eur. J. 2009, 15, 9632−9636.

(55) Gupta, M.; Hagen, C.; Flesher, R. J.; Kaska, W. C.; Jensen, C. M.

Chem. Commun. 1996, 2083−2084.

(56) Perrin, C. L.; Dwyer, T. J. Chem. Rev. 1990, 90, 935−967.

(57) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299.

(58) Roy, L. E.; Hay, P. J.; Martin, R. L. J. Chem. Theory Comput.

2008, 4, 1029−1031.

(59) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys.

1980, 72, 650−654.

(60) McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72, 5639−

5648.

(61) McQuarrie, D. A. Statistical Thermodynamics; Harper and Row:

New York, 1973.

(62) Wheeler, S. E.; Houk, K. N. J. Chem. Theory Comput. 2010, 6,

395−404.

(63) Frisch, Æ.; Frisch, M. J.; Clemente, F. R.; Trucks, G. W.

Gaussian 09 User’s Reference, 147.

NOTE ADDED IN PROOF

■

We recently became aware of a report by Green and Hughes in

1976, in which it was proposed that the isomerization of an

iron-coordinated oleﬁn proceeded via an η¹-allyl complex that

was not formed via an η³-allyl intermediate. The evidence for

this proposal was based on elegant stereochemical studies and

is complementary to the computational support for the η¹-allyl

mechanism proposed in this work. See: Green, M.; Hughes, R. P.

J. Chem. Soc., Dalton Trans. 1976, 1907.

13295

dx.doi.org/10.1021/ja301464c | J. Am. Chem. Soc. 2012, 134, 13276−13295

Article Doi

Olefin isomerization by iridium pincer catalysts. experimental evidence for an η3-allyl pathway and an unconventional mechanism predicted by DFT calculations

DOI: 10.1021/ja301464c

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja301464c

Full text of DOI:10.1021/ja301464c

Products guided by the article

R&D Labs maybe for 111-67-1

Relevant to this article

Hot Product