Olefin hydroaryloxylation catalyzed by pincer-iridium complexes

Article abstract of DOI:10.1021/ja404566v

Aryl alkyl ethers, which are widely used throughout the chemical industry, are typically produced via the Williamson ether synthesis. Olefin hydroaryloxylation potentially offers a much more atom-economical alternative. Known acidic catalysts for hydroaryloxylation, however, afford very poor selectivity. We report the organometallic-catalyzed intermolecular hydroaryloxylation of unactivated olefins by iridium "pincer" complexes. These catalysts do not operate via the hidden Br?nsted acid pathway common to previously developed transition-metal-based catalysts. The reaction is proposed to proceed via olefin insertion into an iridium-alkoxide bond, followed by rate-determining C-H reductive elimination to yield the ether product. The reaction is highly chemo- and regioselective and offers a new approach to the atom-economical synthesis of industrially important ethers and, potentially, a wide range of other oxygenates.

Full text of DOI:10.1021/ja404566v

Article
pubs.acs.org/JACS
Olefin Hydroaryloxylation Catalyzed by Pincer−Iridium Complexes
Michael C. Haibach, Changjian Guan, David Y. Wang, Bo Li, Nicholas Lease, Andrew M. Steffens,
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
*
S Supporting Information
ABSTRACT: Aryl alkyl ethers, which are widely used throughout the
chemical industry, are typically produced via the Williamson ether synthesis.
Olefin hydroaryloxylation potentially offers a much more atom-economical
alternative. Known acidic catalysts for hydroaryloxylation, however, afford
very poor selectivity. We report the organometallic-catalyzed intermolecular
hydroaryloxylation of unactivated olefins by iridium “pincer” complexes.
These catalysts do not operate via the hidden Brønsted acid pathway
common to previously developed transition-metal-based catalysts. The reaction is proposed to proceed via olefin insertion into
an iridium−alkoxide bond, followed by rate-determining C−H reductive elimination to yield the ether product. The reaction is
highly chemo- and regioselective and offers a new approach to the atom-economical synthesis of industrially important ethers
and, potentially, a wide range of other oxygenates.
INTRODUCTION
In some cases, phase transfer catalysis can be used to avoid the
requirement of a polar aprotic solvent. The use of alkyl alcohols
in place of alkyl halides typically requires a gas-phase reaction or
dehydrating agent. For the industrially preferred route, 1 equiv of
alkali halide or alkali sulfonate waste is generated per equivalent
of product produced, in addition to the waste associated with
preparation of the alkali phenoxide and the alkyl halide, which is
typically prepared from the corresponding olefin.
■
The addition of H−X bonds across olefinic double bonds
catalyzed by transition-metal complexes represents a reaction
1
−3
class of great importance in organic chemical synthesis.
Recent years have seen significant developments in catalytic
4
−6
hydroamination; however, progress toward the development
of transition-metal complexes for catalytic addition of O−H
bonds to olefins has been much more limited.
additions of alcohol O−H bonds, especially intermolecular,
remain a particularly important and attractive challenge.
Alkyl aryl ethers are an important class of commodity
chemicals, with applications ranging from solvents to fragrances
to pharmaceutical building blocks. They are currently
synthesized primarily via the very classical Williamson ether
synthesis, whereby an alkali salt of the appropriate phenol
1
−3,6,7
Such
Despite these drawbacks, the Williamson ether synthesis is
widely used for both industrial and small-scale applications,
rather than the atom-economical olefin hydroaryloxylation route
shown in Scheme 1. This is due at least in part to the fact that,
until quite recently, the known catalysts for olefin hydro-
aryloxylation were all strong Brønsted or Lewis acids such as
H SO and BF ·OEt . While this class of catalysts is highly active,
8
9
2
4
3
2
its use suffers from competing Friedel−Crafts alkylations and
very poor chemoselectivity. For example, the reaction of propene
with phenol catalyzed by BF ·OEt affords comparable amounts
(
preformed or generated in situ) is coupled with an alkyl halide
or alkyl sulfonate ester, typically in a polar aprotic solvent
Scheme 1).
3
2
10
of both C- and O-isopropylphenol, even at 0 °C. Beginning
with He’s report in 2005, significant attention has been focused
11
(
on transition-metal precatalysts for hydroaryloxylation, such as
Scheme 1. Alternative Syntheses of Alkyl Aryl Ethers (Shown
for the Addition of Phenol to Propene)
(
PPh )Au(OTf). Despite early evidence that triflic acid was the
3
12,13
catalytically active species,
researchers continued to identify
numerous transition metal “precatalysts” that were later shown
by Hintermann to be Brønsted acid delivery systems (with
Ag(OTf) in chlorinated solvents serving as the most common
1
4,15
source).
Many recent and classic examples employing Lewis
acid catalysts, particularly lanthanide triflates, are also proposed
16
to operate via Yamamoto’s Lewis-assisted Brønsted acid mode
1
2,14,15
of activation.
Indeed, to our knowledge, at the outset of
this work there were no well-defined examples of intermolecular
Received: May 7, 2013
©
XXXX American Chemical Society
A
dx.doi.org/10.1021/ja404566v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
addition of alcohol O−H bonds across the double bond of simple
In a typical experiment, p-xylene solutions (100 μL) of 10 mM
( PCOP)IrH4 and 500 mM 3,5-dimethylphenol are heated at
7
,14,17
iPr
28
olefins directly catalyzed by a transition-metal complex.
In
this communication, we report the first such catalysts, specifically
150 °C in glass ampules sealed with 2 atm of propene. (Each
ampule has a total volume of 1.2 mL which holds 1.8 equiv of
propene in the gas phase per equivalent of alcohol, plus any
propene which may have dissolved in solution prior to sealing.)
After 24 h the appearance of 400 mM isopropyl aryl ether (40
catalytic turnovers (TO)) was observed (by GC), while in
another ampule 460 mM was observed after 48 h (Scheme 3). No
n-propyl aryl ether or other alkylphenols were detectable by GC,
indicating that the reaction is fully regio- and chemoselective.
The analogous reaction with ethylene yields exclusively ethyl aryl
ether (47 mM after 24 h, 128 mM after 72 h). 1-Butene also
yields only the iso product, albeit with lower conversion than
either ethylene or propene (Scheme 3). Running the same
reaction at 4 atm of 1-butene affords higher conversion.
Importantly, when (unsubstituted) phenol and propene are
allowed to react, only isopropyl phenyl ether (289 mM) is
observed after 24 h. Note that any catalyst operating by the
for the reaction of phenols, and support for a likely mechanism
18
on the basis of experimental and computational evidence.
These catalysts offer selectivity much greater than, and in some
cases orthogonal to, that of previously reported acid catalysts.
RESULTS AND DISCUSSION
■
We have previously reported that precursors of the fragment
tBu
3
R
3
(
PCP)Ir ( PCP = κ -C H -2,6-(CH PR ) ) could cleave aryl-
6 3 2 2 2
sp C−O bonds stoichiometrically via an initial C−H oxidative
1
9,20
addition step.
In the case of ethyl phenyl ether, for example,
this led to dehydroaryloxylation and formation of the iridium
adducts of ethylene and phenol (Scheme 2). The potential ability
Scheme 2. Stoichiometric Dehydroaryloxylation of Ethyl
Phenyl Ether by (tBuPCP)Ir
“
hidden Brønsted acid” mechanism proposed by Hintermann is
expected to afford at least some C-alkylphenol. Indeed, the
reaction of phenol and propene using 10 mM Hintermann’s
t
15
Brønsted acid catalyst (10 mM AgOTf and 40 mM BuCl)
yields no isopropyl phenyl ether after 24 h. Instead, 10 different
products are observable by GC, with both phenol and the p-
xylene solvent undergoing apparently unselective Friedel−Crafts
alkylations and isomerizations (eq 1).
of such species to undergo kinetically facile olefin loss and phenol
elimination suggested the possibility of a catalytic cycle; in the
thermodynamically favorable reverse direction such a cycle
would constitute olefin hydroaryloxylation.
tBu
Attempts to effect catalytic hydroaryloxylation by ( PCP)Ir
tBu
(
10 mM ( PCP)IrH catalyst precursor, 500 mM phenol, 1 atm
4
of ethylene or propene, p-xylene solvent, 100−150 °C) were
unsuccessful, yielding no new organic products. Investigation of
some previously reported derivatives of ( PCP)Ir, however,
tBu
Further evidence of the nonacidic nature of the iridium catalyst
system derives from a competition experiment between
isobutene and propene. Isobutene forms a much more stable
carbocation when protonated; hence, an acidic catalyst would be
expected to yield the tert-butyl phenyl ether as the major product.
successfully identified several catalysts active for the addition of
2
1
propene to 3,5-dimethylphenol at 150 °C (Figure 1). The three
tBu3Me
22
most active derivatives identified were (
PCP)Ir,
iPr
23
iPr
24
(MeO- PCP)Ir, and ( PCOP)Ir.
However, when a solution of 3,5-dimethylphenol and
iPr
(
PCOP)IrH4 is subjected to equal partial pressures of
isobutene and propene, the aryl isopropyl ether is formed with
40:1 selectivity (eq 2). The control experiment, using AgOTf/t-
>
BuCl, gave no aryl isopropyl ether, although only trace aryl tert-
butyl ether was observed (eq 3); the major products resulted
from Friedel−Crafts alkylation of the arene rings as in the case of
eq 1. Similarly arguing against any carbocation-derived
selectivity, although propene reacts significantly faster with 3,5-
dimethylphenol than does ethylene in independent experiments,
ethylene reacts preferentially vs propene with selectivity >3:1 in
an internal competition experiment (eq 4).
The apparently high regioselectivity for formation of i-PrOAr
vs n-PrOAr (Scheme 3) and the high chemoselectivity for
hydroaryloxylation of propene vs isobutene (eq 2) might be
attributed, a priori, to thermodynamic rather than kinetic factors.
In such a case the rate of the respective hydroaryloxylations
might be comparable to or even more rapid than the reaction to
give i-PrOAr, but the respective dehydroaryloxylation back-
reactions could be even faster. To test this hypothesis, we
investigated the possible back reactions.
Figure 1. Species found to be active for catalytic hydroaryloxylation.
This group of sterically less congested catalysts is about 1 order
of magnitude more active for alkane dehydrogenation than
tBu
25,26
(
PCP)Ir.
As in alkane dehydrogenation, the catalytically
active species can be generated from the iridium tetrahydride or
ethylene complexes under the reaction conditions or from the
27
corresponding (pincer)IrHCl complex and a base. Presumably
due to inhibition resulting from strong binding by ethylene, the
catalytic activity for each (pincer)Ir precatalyst followed the
trend (pincer)Ir(C H ) < (pincer)IrH ≈ (pincer)Ir(H)(Cl)/
2
4
4
t
NaO Bu.
B
dx.doi.org/10.1021/ja404566v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Scheme 3. Hydroaryloxylation of Olefins Catalyzed by ( PCOP)IrH
Article
iPr
4
Scheme 4. Schematic Free Energy Profile for
Hydroaryloxylation of Propene To Give i-PrOAr and n-
PrOAr and the Corresponding Dehydroaryloxylations
iPr
Catalyzed by ( PCOP)IrH at 150 °C
4
A p-xylene solution of i-PrOAr (500 mM), n-PrOAr (500
iPr
mM), and ( PCOP)Ir(C H ) was sealed under vacuum (100
2
4
μL in a 1 mL ampule) and heated to 150 °C. After 12 h, GC
analysis revealed the disappearance of 48% of the i-PrOAr and
8
% of the n-PrOAr, with commensurate appearance of ArOH
29
(270 mM) and the appearance of propene. Thus, the rates of
both propene hydroaryloxylation to give n-PrOAr and the
dehydroaryloxylation of n-PrOAr are much slower than the
corresponding reactions for i-PrOAr; this selectivity is therefore
clearly a kinetic phenomenon.
Note that the dehydroaryloxylation of i-PrOAr proceeds at
least 6-fold more rapidly than that of n-PrOAr (eq 5). Both
A competition experiment under the same conditions as for
the experiment of eq 5 was also conducted with i-PrOAr and t-
BuOAr. The dehydroaryloxylation of i-PrOAr was found to be
more rapid (by a factor larger than 2; eq 6). Thus, the much
greater rate of hydroaryloxylation of propene vs isobutene (>40-
fold; eq 2), like the regioselectivity to give i-PrOAr, is a kinetic
phenomenon.
30
experimental data and DFT calculations (discussed below)
indicate that the free energy of n-PrOAr is ca. 3 kcal/mol higher
⧧
than that of i-PrOAr. ΔΔGH for the hydroaryloxylation of
Finally, we note that, even in the presence of a large excess of
NaO Bu (125 mM), a solution of 3,5-dimethylphenol (0.5 M)
t
propene to give n-PrOAr must therefore be ca. 3 kcal/mol greater
⧧
than ΔΔGD for the dehydroaryloxylations (Scheme 4). In that
under propene (2 atm) undergoes clean catalytic hydro-
tBu3Me
case, a 6-fold rate difference for the latter, at 150 °C, would
aryloxylation by (
PCP)IrHCl (10 mM), yielding 250 mM
correspond to a rate difference of 6 exp(−3 kcal/mol/(RT)) ≈
i-PrOAr product after 24 h at 150 °C. This result argues strongly
against any mechanism involving Brønsted acid.
⧧
2
00 (i.e., ΔΔG ≈ 4.5 kcal/mol), in accord with our conclusion
H
of essentially complete kinetic selectivity for the formation of i-
Identification of the Catalyst Resting State. The reaction
iPr
PrOAr in the hydroaryloxylation.
of phenol with propene using ( PCOP)Ir as the catalyst was
C
dx.doi.org/10.1021/ja404566v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
1
31
investigated in detail by H and P NMR. Addition of 1 atm of
propene to a J. Young NMR tube with either a 0.1 or 0.01 M
Computational Investigation of the Mechanism.
21
Computational (DFT ) studies have been conducted which
shed light on the mechanism and selectivity of the hydro-
aryloxylation reactions. We employed the widely used M06 and
M06-L density functionals. Both functionals predicted regio- and
chemoselectivity in full agreement with our experimental results.
Since the M06-L functional provided slightly better quantitative
agreement, we will primarily discuss M06-L energies and present
those values in the figures shown here; energies obtained with the
M06 functional are given in tables in the Supporting Information.
We have focused on the reaction of phenol with propene by our
iPr
19
solution of ( PCOP)Ir(H)(OPh) in p-xylene-d (affording a
solution propene concentration of ca. 1 M) results in quantitative
conversion to the ( PCOP)Ir(η -propene) complex (see the
Supporting Information for characterization of ( PCOP)Ir(η -
propene) and ( PCOP)Ir(H)(OPh)) and free phenol within 15
min at 25 °C (Scheme 5). No change is observed upon heating to
1
0
iPr
2
iPr
2
iPr
iPr
2
Scheme 5. Identification of ( PCOP)Ir(η -propene) as the
Catalyst Resting State by NMR Studies
iPr
most effective catalyst, ( PCOP)Ir. Although the calculations
assume idealized gas-phase conditions, free energies have been
calculated at conditions (T, P) that are closer to those of the
actual catalytic experiments than are standard conditions (T =
2
98.15 K, P = 1.0 atm). Specifically, we use T = 150 °C = 423.15
K and, in order to approximate the concentrations of reagents in
solution, partial pressures of 34.7 atm were assumed, which
correspond to concentrations of 1 mol/L at 150 °C.
iPr
2
Experimentally, as noted above, ( PCOP)Ir(η -propene)
1a) was found to be the only major species in solution under the
(
standard reaction conditions. Using the M06 functional and the
thermodynamic conditions noted above (T = 150 °C, P = 34.7
iPr
2
atm), ( PCOP)Ir(η -propene) was indeed computed to be the
lowest energy species, 1.7 kcal/mol lower in free energy than
iPr
(
PCOP)Ir(H)(OPh) (3) (Table S4, Supporting Information)
1
20 °C in the NMR spectrometer. Importantly, no signals
iPr
2
1
and 3.9 kcal/mol below ( PCOP)Ir(H)(η -propene)(OPh) (4)
the lowest energy conformer, with propene coordinated trans to
the pincer aryl group; Table S5, Supporting Information). The
between δ 0 and −50 ppm in the H NMR spectrum,
characteristic of iridium hydrides, are observed, thus indicating
the lack of formation of any appreciable amounts of either
(
iPr
iPr
2
iPr
corresponding M06-L values for ( PCOP)Ir(H)(OPh) and
(
(
PCOP)Ir(η -propene)(H)(OPh) or ( PCOP)Ir[CH CH-
2
iPr
2
(
PCOP)Ir(H)(η -propene)(OPh), relative to 1a, are −2.3 and
OAr)CH ](H) under these conditions.
3
−
0.1 kcal/mol at 150 °C, respectively (Table S1, Supporting
A solution otherwise similar to that described above was
iPr
Information). In both cases, we judge the differences to be within
the error margins of the calculations when comparing species
that are significantly different (e.g., an Ir(I) complex and an
Ir(III) complex, π vs σ coordination, 4-coordination vs 5- or 6-
coordination). Accordingly, we will only consider energies
relative to the experimentally observed resting state, the olefin π-
complex 1a.
prepared, but with 0.5 M PhOH added (0.01 M ( PCOP)Ir-
(
H)(OPh), 0.5 M PhOH, ca. 1 M propene) to replicate the
concentrations typically used for the catalytic runs. Under these
conditions, the presence of PhOH resulted in a complex mixture
3
1
1
of iridium-containing complexes at 25 °C ( P and H NMR).
iPr
2
(
PCOP)Ir(η -propene) is present, while several signals
between δ −7 and −24 ppm (triplets with J values of ca. 15 Hz
which is typical for J ) are also observed in the H NMR
2
1
iPr
2
Under typical reaction conditions, ( PCOP)Ir(η -propene)
(1a) is calculated to be the major resting state (the kinetically
accessible species of lowest free energy) in the ( PCOP)Ir/
phenol/propene system using the M06 functional. However, at
25 °C the free energy of ( PCOP)Ir(H)(η -propene)(OPh) is
calculated to be −0.6 kcal/mol below the four-coordinate
propene adduct, whereas ( PCOP)Ir(H)(OPh) remains higher
in energy than 1a by 2.0 kcal/mol. The corresponding M06-L
values for ( PCOP)Ir(H)(OPh) and ( PCOP)Ir(H)(η -
propene)(OPh), relative to 1a, are −2.1 and −4.5 kcal/mol at
25 °C. These results are consistent (at least within the limits of
precision of the calculations) with the observation that a mixture
PH
spectrum. These signals are indicative of six-coordinate iridium
hydrides; it would seem likely that at least some of these are of
the composition ( PCOP)Ir(η -propene)(OPh)(H). The low
symmetry of these complexes, however, apparently generates
several isomers, making assignment of their structures by NMR
prohibitively difficult. (Note that even a single coordination
isomer of ( PCOP)Ir(η -propene)(OPh)(H) has at least four
possible conformers depending upon the orientation of the
propene ligand.) When this solution is heated to 120 °C,
however, a temperature at which there is catalytic activity, the
iPr
iPr
2
iPr
2
iPr
iPr
2
iPr
iPr
2
31
iPr
2
only species observable in solution is ( PCOP)Ir(η -propene).
Thus, the apparent equilibrium of ( PCOP)Ir(η -propene)-
iPr
2
iPr
2
iPr
of ( PCOP)Ir(η -propene), ( PCOP)Ir(H)(OPh), and
iPr
2
iPr
2
(
OPh)(H) isomers with ( PCOP)Ir(η -propene) plus free
( PCOP)Ir(H)(η -propene)(OPh) appear to be present in a
iPr
2
phenol (eq 7) is driven toward the side with free phenol at higher
typical reaction solution at 25 °C, whereas only ( PCOP)Ir(η -
propene) is observed at 120 °C.
iPr
2
temperature (as would be expected), and ( PCOP)Ir(η -
propene) is the resting state under catalytic conditions.
The results of the selectivity experiments discussed above
argue strongly against a Brønsted-acid catalyzed pathway, or any
pathway involving a carbocationic or carbocation-like inter-
mediate, and instead favor a genuinely “organometallic-
catalyzed” mechanism. Generally speaking, “organometallic”
mechanisms for hydrofunctionalization (addition of species H−
X across multiple bonds) may proceed via insertion of olefin into
a M−H bond followed by alkyl−X elimination (Figure 2a);
D
dx.doi.org/10.1021/ja404566v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 2. Typical “organometallic” pathways (proceeding via H−X addition, olefin insertion, and C−X or C−H elimination) for generic
hydrofunctionalization of an olefin (addition of H−X). Cycle a is shown giving anti-Markovnikov product, and cycle b is shown giving the Markovnikov
product. This represents the regioselectivity commonly expected of each pathway, but neither mechanism is necessarily limited to either type of
regioselectivity.
Figure 3. Free energy diagram (M06-L; values of ΔG in kcal/mol) for the proposed 1,2-Ir−O addition pathway for hydrophenoxylation of propene by
iPr
(
PCOP)Ir to give i-PrOAr (observed product; blue lines) and n-PrOAr (not observed; red lines).
1
known examples include X = SiR , BR , CN. Such mechanisms
our previous work, in which it was found that the barrier to direct
3
2
tBu
can favor formation of anti-Markovnikov products (e.g., CH X−
C−O bond oxidative addition to ( PCP)Ir is prohibitively
2
1
9,20
CH R from CH CHR plus HX). It is generally assumed that
2
2
high.
such selectivity is attributable to the preference of transition
Interestingly, although the pathway of Figure 2a is precluded
by the high barrier to C−O bond elimination, the initial steps
appear to be quite favorable. Addition of ArOD (0.5 M) and
3
2−35
metals for less substituted alkyl ligands
(e.g., primary vs
secondary) reflected in the TS preceding or perhaps following
the intermediate species L M(alkyl)X.
n
(
(
(
perprotio) propene (1 atm) to a p-xylene-d1 solution of
0
In the case of the present system, the free energy calculated for
the TS for the key step of C−X elimination as per Figure 2a (L M
( PCOP)Ir; X = OPh; alkyl = i-Pr) is 47.3 kcal/mol above that
of the calculated resting state, ( PCOP)Ir(propene) (M06-L;
Table S1, Supporting Information). This value is substantially
greater than the overall barrier indicated by experiment, ΔG ≈
2 kcal/mol (based on ca. 1.2 turnovers/h). The pathway of
Figure 2a is thus calculated not to be viable in this case, regardless
of the energies of any other intermediates and transition states in
that pathway. This result is consistent with and closely related to
iPr
PCOP)IrH , to give the mixture of species indicated in eq 6
4
n
and isotopologues thereof), results in rapid H/D exchange
iPr
=
iPr
between propene and ArOD (50% conversion to ArOH within
1
1
5 min at room temperature, as revealed in the H NMR
⧧
spectrum). This is most easily explained in terms of reversible
iPr
2
insertion of propene into the Ir−H/D bond of ( PCOP)Ir(η -
propene)(OPh)(H/D). The thermodynamics of this insertion
are calculated to be quite allowable for such an exchange
mechanism (ΔG = +7.3 and +8.6 kcal/mol for 1,2- and 2,1-Ir−H
3
E
dx.doi.org/10.1021/ja404566v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
iPr
Figure 4. Calculated distances (Å) for the H−Ir−C−C(propene)−O(phenoxide) unit for the 1,2-Ir−O insertion step: (a) trans-( PCOP)IrH-
iPr
(
OPh)(propene) (4a); (b) TS for propene insertion (TS-4a-5a); (c) insertion product ( PCOP)IrH[CH CH(CH )OPh] (5a).
2 3
addition, respectively), although we have been unable to locate
appropriate low-energy TS’s for these insertions.
calculated barrier height of 31.7 kcal/mol (TS-CH-elim-a), in
complete (and presumably fortuitously excellent) agreement
⧧
Rather than the mechanism indicated in Figure 2a, the
calculations are instead consistent with the hypothesis that led to
this work: namely, that the mechanism indicated in Scheme 2
could be implemented catalytically in the reverse direction (as
shown explicitly in Figure 2b). The mechanism of Figure 2b
proceeds via olefin insertion into the M−X (Ir−O) bond, rather
than insertion into the M−H bond as in Figure 2a, and is
followed by C−H rather than C−X (C−O) elimination. There
are relatively few well-characterized examples of insertion of
olefins into transition metal−oxygen bonds, but the reaction is
with the approximate experimental barrier, ΔG ≈ 32 kcal/mol.
Figure 3 also shows a pathway that proceeds via a 2,1-Ir−O
addition which would lead to n-PrOPh; this represents the
mechanism shown in Figure 2b but with the reverse
regioselectivity. The 2,1-Ir−O addition has a calculated barrier
substantially higher than the 1,2-Ir−O addition; TS-4b-5b is 32.0
kcal/mol above the resting state vs 21.5 kcal/mol for TS-4a-5a.
The energy of the resulting phenoxy-substituted secondary alkyl
hydride, 5b, is 7.1 kcal/mol above that of the primary alkyl
hydride, 5a, derived from the 1,2-addition (26.0 kcal/mol above
the resting state vs 18.9 kcal/mol). This is also in agreement with
Hartwig’s study, in which it was found that 1,2-addition of the
M−O bond was much more favorable than 2,1-addition (with
the difference being much greater than that found for M−C
36−40
certainly not without precedent.
Figure 3 shows the results of calculations of the catalytic cycle
iPr
(
as per Figure 2b) for the ( PCOP)Ir-catalyzed reaction of
propene and phenol to give i-PrOPh and n-PrOPh (all free
energies are expressed relative to the free three-coordinate pincer
iridium complex and free propene and phenol). 1,2-Addition of
39
addition). However, while these 1,2-Ir−O addition energies are
higher than the corresponding values for the 2,1-Ir−O addition,
they are not so high as to necessarily preclude formation of the n-
propyl ether at a rate comparable to that observed for formation
of i-PrOAr.
iPr
2
the Ir−OPh bond of ( PCOP)IrH(OPh)(η -propene) (4a)
across the double bond of coordinated propene is calculated to
have a barrier of only ca. 21 kcal/mol, with a transition state (TS;
TS-4a-5a) that is 21.5 kcal/mol above the propene complex
The calculations illustrated in Figure 3 predict that the
subsequent C−H elimination, not insertion into the Ir−OAr
bond, is both rate and product determining. The TS for the C−H
elimination, TS-CH-elim-b, is of higher energy for the secondary
alkyl hydride than for the primary, TS-CH-elim-a, by a
substantial margin of 3.8 kcal/mol. This difference would
39
resting state (1a). This is in agreement with a theoretical study
by Hartwig on olefin insertion into the Rh−X bond (X = CH ,
3
NH , OH) of (PMe ) RhX, in which it was calculated that the
2
3 2
barrier to 1,2-insertion of coordinated propene into a Rh−O
bond was 19.3 kcal/mol. Moreover, also in accord with Hartwig’s
3
9
results, in the present system the metal−oxygen bond remains
largely intact during and even after the insertion step. The Ir−O
correspond to a factor of greater than 90 in the rates for
⧧
formation of i-PrOAr (ΔG
= 31.7 kcal/mol) vs n-PrOAr
calc
iPr
⧧
bond distances in trans-( PCOP)IrH(OPh)(propene) (4a),
(ΔG
= 35.5 kcal/mol) at 150 °C. The calculations thus fully
calc
TS-4a-5a, and the insertion product 5a are 2.30, 2.33, and 2.42 Å,
respectively (Figure 4); thus, the Ir−O bond appears to
account for the observed rate of formation of i-PrOAr and for the
absence of n-PrOAr. Moreover, the same energy diagram
illustrates that the barrier to the back reaction (dehydroarylox-
ylation) is calculated to be slightly higher for the reaction of i-
PrOPh than for n-PrOPh (by 0.9 kcal/mol). This is also in
excellent agreement with experimental observations noted
above.
39
transition smoothly from formally covalent to dative.
A
conformer of 5a in which there is no significant Ir−O interaction
(d
= 4.4 Å) is a local minimum with a free energy 7.0 kcal/mol
Ir−O
above the lowest free energy conformer of 5a; this value
presumably represents the approximate strength of the dative
interaction.
The free energy difference between the two rate-determining
C−H elimination TS’s, which may determine the very high
regioselectivity for formation of i-PrOAr vs n-PrOAr, can perhaps
be most simply explained by considering the reaction proceeding
in the reverse direction. The difference of 3.8 kcal/mol can then
be viewed as resulting from a combination of two simple factors
(i) The energy of free i-PrOAr is lower than that of free n-PrOAr
(2.9 kcal/mol calculated difference, 3.35 ± 0.43 kcal/mol
The product of the 1,2-Ir-OPh addition to propene,
(iPrPCOP)Ir[CH CH(OPh)CH ](H) (5a), is 18.9 kcal/mol
2
3
above the propene resting state 1a (Figure 3). The TS for C−H
elimination from this species (TS-CH-elim-a), to give i-PrOPh,
is calculated to have a free energy 31.7 kcal/mol above the resting
state (i.e., 14.2 kcal/mol above the reference); this TS leads to a
C−H σ-bond complex (not shown in Figure 3) that is 22.9 kcal/
mol above the resting state (i.e., 5.4 kcal/mol above the
reference). We have been unable to locate a proper TS for
dissociation of this σ-bond complex. However, it seems likely
30
experimental ). (ii) The barrier for the oxidative cleavage of
primary C−H bonds is generally less than for secondary C−H
34
bonds; in this case addition of the primary C−H bond of i-
PrOPh is calculated to be 0.9 kcal/mol lower than that for the
secondary C−H (C2) bond of n-PrOPh.
(
although not certain) that loss of the σ-C−H-bound ether
product (which may proceed dissociatively or via displacement
by solvent, phenol, or propene) is fast relative to the back
reaction, C−H addition. In that case, the C−H elimination is the
rate-determining step for formation of i-PrOPh, with an overall
Overall, the calculated results presented above, obtained with
the use of the M06-L functional, strongly indicate that C−H
elimination is the rate-determining step in the cycle. The
F
dx.doi.org/10.1021/ja404566v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 5. Free energy diagram (M06-L; values of ΔG in kcal/mol) for the proposed 1,2-Ir−O addition pathway for hydrophenoxylation of propene by
iPr
(
PCOP)Ir to give i-PrOAr proceeding via “olefin-trans” (blue) and “olefin-cis” (red) pathways.
Figure 6. Free energy diagram (values in kcal/mol) for proposed pathway for hydrophenoxylation of ethylene (green lines) and propene (blue lines) by
iPr
(
PCOP)Ir. From a common resting state (as in a competition experiment) the barrier to the reaction of ethylene is lower, but in individual
experiments, the overall barrier is lower for the reaction of propene.
calculations are in excellent agreement with experimental results,
including the absolute rate and the selectivities for formation of i-
PrOAr vs n-PrOAr (very high) and for dehydroaryloxylation of i-
PrOAr vs n-PrOAr (ca. 6-fold). Calculations using the M06
functional lead to essentially the same predictions, including the
rate-determining nature of C−H elimination. However, whereas
G
dx.doi.org/10.1021/ja404566v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
the use of M06-L leads to barrier heights for C−H elimination
that are much higher than for insertion (by 10.2 and 3.5 kcal/mol
for 1,2-addition and 2,1-addition, respectively), the differences
are much less pronounced using M06 (3.5 and 0.7 kcal/mol for
addition to less substituted olefins make clear. DFT calculations
are strongly supportive of a mechanism proceeding via insertion
of olefin into the iridium−aryloxide Ir−O bond. A very high
degree of regioselectivity is observed. DFT calculations indicate
that this is determined by the energy of the respective TS’s for
C−H bond elimination; this derives in part from the same factors
that control selectivity for C−H bond addition.
The nature of the sterically congested and geometrically well-
defined pincer−metal unit, and the formation of secondary alkyl
ethers, suggest an entry into the development of olefin
hydroaryloxylation catalysis that may display unusual selectivity
or enantioselectivity. More generally, the discovery of these well-
defined nonacid catalysts suggests the possibility of catalytic
intermolecular O−H addition across multiple bonds with a scope
broader than that for phenols and simple olefins. Finally, we find
that the catalysts are also effective for the reverse, C−O bond
cleavage reaction, dehydroaryloxylation. Further research efforts
in these contexts are underway.
1,2-addition and 2,1 addition, respectively; see Tables S4 and S5
and Figure S4, Supporting Information). Thus, while DFT
calculations obtained using either functional indicate that C−H
elimination is rate determining for hydroaryloxylation (and C−H
addition rate determining for dehydroaryloxylation), future
studies to test this important conclusion seem warranted.
It should be noted that two geometrically distinct variants of
either the 1,2- or 2,1-Ir−O addition pathways have been
calculated. For each pathway there is the variant in which the
olefin is initially coordinated trans to the PCP aryl of
iPr
(
PCOP)Ir(OPh)H (shown in Figure 3), and another in
which olefin coordinates cis to the PCP aryl, while the phenoxy
group is coordinated trans (shown in Figure 5 for 1,2-addition
leading to i-PrOPh). The olefin-trans variant has a lower energy
TS for insertion of olefin into the Ir−O bond in for both 1,2- and
2
(
,1-additions. Each variant gives rise to a different isomer of
PCOP)Ir(phenoxypropyl)(H) upon Ir−O addition (5a vs 5c
ASSOCIATED CONTENT
Supporting Information
Text, figures, and tables giving experimental details and
■
iPr
*
S
in the case of the 1,2-addition shown in Figure 5). In both cases
the olefin-trans insertion TS is of higher energy than the olefin-
cis TS (TS-4c-5c vs TS-4a-5a in the case of 1,2-addition).
However, the intermediates resulting from Ir−O addition can
probably interchange readily (the barrier to decoordination of
the phenoxy group, as noted above, is only 7 kcal/mol). Thus
even if the olefin-cis insertion were more facile than the olefin-
trans pathway, since the insertion step is not rate determining the
distinction between these pathways would not necessarily be
significant.
iPr
procedures, including synthesis of ( PCOP)IrH , character-
4
iPr
ization of ( PCOP)Ir(H)(OPh), and characterization of
iPr
2
(
PCOP)Ir(η -propene), computational details, and molecular
energies and geometries. This material is available free of charge
via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Authors
kroghjes@rutgers.edu.
■
Finally, Figure 6 illustrates the results of calculations for the
addition of PhOH to ethylene proceeding via the mechanism of
Figure 2b, along with the calculated values for the addition to
propene in the presence of ethylene, thereby modeling the
competition experiment of eq 4. The overall barrier for PhOH
addition to ethylene (which is not affected by the presence of
propene) is calculated to be 35.6 kcal/mol (the difference
alan.goldman@rutgers.edu.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was carried out under the auspices of National Science
Foundation Grant CHE No. 1112456 and the Extreme Science
and Engineering Discovery Environment (XSEDE), which is
supported by National Science Foundation Grant No. OCI-
between the free energy of TS-CH-elim-d and the free energy of
iPr
(
PCOP)Ir(ethene) (1c)). The overall barrier for hydro-
aryloxylation of propene, in the presence of ethylene (which results
in an ethylene-bound resting state), is calculated to be 37.7 kcal/
mol (the free energy of TS-CH-elim-a minus the free energy of
the resting state ethene complex 1c) as compared with 31.7 kcal/
mol above the propene-bound resting state. Thus, these
calculations successfully capture both the greater reactivity of
ethylene in competition experiments and the greater reactivity of
propene in independent runs, providing additional support for
the proposed mechanism of Figure 2b. Interestingly, the TS for
ethene insertion TS-4d-5d is slightly higher than that for
propene insertion, TS-4a-5a. If that is in fact the case (although
the small difference of 0.9 kcal/mol is arguably not meaningful),
and if insertion into the Ir−O bond, not C−H elimination, were
rate-determining, then the competition experiment of eq 4 would
have yielded more i-PrOAr than EtOAr, in contrast with the
experimental result.
1
053575. M.C.H. thanks the NSF IGERT program (Renewable
and Sustainable Fuel Solutions for the 21st Century) for a
Graduate Training Fellowship. B.L. thanks the Rutgers−Jilin
exchange program for funding.
REFERENCES
■
(
1) Hartwig, J. F. In Organotransition Metal Chemistry; University
Science Books: Sausalito, CA, 2010; pp 667−699.
(
2) Yadav, J. S.; Antony, A.; Rao, T. S.; Subba Reddy, B. V. J.
Organomet. Chem. 2011, 696, 16−36.
3) Ananikov, V. P.; Beletskaya, I. P. In Hydrofunctionalization;
(
Ananikov, V. P., Tanaka, M., Eds.; Springer: Berlin, Heidelberg: 2013;
Vol. 43, pp 1−19.
(
4) Muller, T. E.; Hultzsch, K. C.; Yus, M.; Foubelo, F.; Tada, M. Chem.
Rev. 2008, 108, 3795−3892.
(5) Julian, L. D.; Hartwig, J. F. J. Am. Chem. Soc. 2010, 132, 13813−
1
3822.
(
6) Julian, L. D. Topics in Heterocyclic Chemistry, Springer: Berlin,
Heidelberg: 2013; pp 1−47.
7) For examples of lanthanide catalyzed intramolecular O−H addition
to multiple bonds see: (a) Seo, S.; Yu, X.; Marks, T. J. J. Am. Chem. Soc.
CONCLUSIONS
■
We report iridium pincer complexes that catalyze olefin
hydroaryloxylation with simple olefins and phenols. These
catalysts do not operate via a “hidden Brønsted acid” mechanism,
common to previously reported precatalysts for this reaction, as
their high selectivity for O- vs C-alkylation and the preference for
(
2
008, 131, 263−276. (b) Dzudza, A.; Marks, T. J. Org. Lett. 2009, 11,
1523−1526. (c) Dzudza, A.; Marks, T. J. Chem. Eur. J. 2010, 16, 3403−
3422.
H
dx.doi.org/10.1021/ja404566v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(
8) Fiege, H.; Voges, H.-W.; Hamamoto, T.; Umemura, S.; Iwata, T.;
the same propene complex. Unfortunately, however, the temperature
limits of our NMR spectrometers did not allow observation at this
temperature. (Note that when the temperature is taken back down to
ambient between intervals of heating at 150 °C, the original mixture is
again observed, which, on warming to 120 °C in the NMR spectrometer,
Miki, H.; Fujita, Y.; Buysch, H.-J.; Garbe, D.; Paulus, W. In Ullmann’s
Encyclopedia of Industrial Chemistry; Wiley-VCH: Weinheim, Germany,
2000.
(
9) Williamson, A. Philos. Mag. 1850, 37, 350−356.
iPr
2
(
10) Sowa, F. J.; Hinton, H. D.; Nieuwland, J. A. J. Am. Chem. Soc. 1932,
again yields exclusively ( PCOP)Ir(η -propene).
5
(
(
4, 3694−3698.
(32) Schwartz, J.; Labinger, J. A. Angew. Chem., Int. Ed. 1976, 15, 333−
11) Yang, C.-G.; He, C. J. Am. Chem. Soc. 2005, 127, 6966−6967.
340.
12) Rosenfeld, D. C.; Shekhar, S.; Takemiya, A.; Utsunomiya, M.;
(33) Reger, D. L.; Culbertson, E. C. Inorg. Chem. 1977, 16, 3104−3107.
(34) Some excellent lead references to organometallic C−H addition,
with particular emphasis on selectivity: (a) Bennett, J. L.; Vaid, T. P.;
Wolczanski, P. T. Inorg. Chim. Acta 1998, 270 (1−2), 414−423.
(b) Wick, D. D.; Jones, W. D. Organometallics 1999, 18, 495−505.
(c) Asbury, J. B.; Hang, K.; Yeston, J. S.; Cordaro, J. G.; Bergman, R. G.;
Lian, T. J. Am. Chem. Soc. 2000, 122, 12870−12871 and refs 4−11
therein. (d) Vetter, A. J.; Flaschenriem, C.; Jones, W. D. J. Am. Chem. Soc.
2005, 127, 12315−12322. (e) Balcells, D.; Clot, E.; Eisenstein, O. Chem.
Rev. 2010, 110, 749−823.
Hartwig, J. F. Org. Lett. 2006, 8, 4179−4182.
13) Li, Z.; Zhang, J.; Brouwer, C.; Yang, C.-G.; Reich, N. W.; He, C.
Org. Lett. 2006, 8, 4175−4178.
14) Hintermann, L. In C-X Bond Formation; Vigalok, A., Ed.; Springer:
Berlin, Heidelberg: 2010; Vol. 31, pp 123−155.
15) Dang, T. T.; Boeck, F.; Hintermann, L. J. Org. Chem. 2011, 76,
353−9361.
16) Yamamoto, H.; Futatsugi, K. Angew. Chem., Int. Ed. 2005, 44,
924−1942.
17) After this paper was submitted for publication, a report by Sevov
(
(
(
9
(
1
(
(35) Hartwig, J. F. In Organotransition Metal Chemistry; University
Science Books: Sausalito, CA, 2010; pp 85−146.
and Hartwig on a Segphos−iridium-catalyzed olefin hydroaryloxylation
appeared: Sevov, C. S.; Hartwig, J. F. J. Am. Chem. Soc. 2013, 135, 9303−
(36) Bryndza, H. E. Organometallics 1985, 4, 406−8.
(37) Woerpel, K. A.; Bergman, R. G. J. Am. Chem. Soc. 1993, 115,
9306.
7
(
1
(
1
(
888−7889.
38) Zhao, P.; Incarvito, C. D.; Hartwig, J. F. J. Am. Chem. Soc. 2006,
28, 9642−9643.
39) Tye, J. W.; Hartwig, J. F. J. Am. Chem. Soc. 2009, 131, 14703−
4712.
40) Hartwig, J. F. Nature 2008, 455, 314−322.
(
18) Some of this work has been presented preliminarily: Haibach, , M.
C.; Li, , B.; Wang, , D. Y.; Guan, , C.; Krogh-Jespersen, , K.; Goldman, , A.
S. Abstracts of Papers; 245thACS National Meeting & Exposition, New
Orleans, LA, United States,April 7−11, 2013,; American Chemical
Society: Washington, DC, 2013,; INOR-130.
(
19) Choi, J.; Choliy, Y.; Zhang, X.; Emge, T. J.; Krogh-Jespersen, K.;
Goldman, A. S. J. Am. Chem. Soc. 2009, 131, 15627−15629.
20) Kundu, S.; Choi, J.; Wang, D. Y.; Choliy, Y.; Emge, T. J.; Krogh-
(
Jespersen, K.; Goldman, A. S. J. Am. Chem. Soc. 2013, 135, 5127−5143.
(
(
21) See the Supporting Information for details.
22) Kundu, S.; Choliy, Y.; Zhuo, G.; Ahuja, R.; Emge, T. J.; Warmuth,
R.; Brookhart, M.; Krogh-Jespersen, K.; Goldman, A. S. Organometallics
009, 28, 5432−5444.
23) Zhu, K.; Achord, P. D.; Zhang, X.; Krogh-Jespersen, K.; Goldman,
A. S. J. Am. Chem. Soc. 2004, 126, 13044−13053.
24) Ahuja, R.; Punji, B.; Findlater, M.; Supplee, C.; Schinski, W.;
Brookhart, M.; Goldman, A. S. Nat. Chem. 2011, 3, 167−171.
25) Choi, J.; MacArthur, A. H. R.; Brookhart, M.; Goldman, A. S.
Chem. Rev. 2011, 111, 1761−1779.
26) Haibach, M. C.; Kundu, S.; Brookhart, M.; Goldman, A. S. Acc.
Chem. Res. 2012, 45, 947−958.
27) Pincer iridium dihydrides and tetrahydrides are well-known to
2
(
(
(
(
(
undergo dehydrogenation by olefin and are therefore presumed to act
simply as precursors of the corresponding (pincer)Ir fragment. Likewise,
the ethylene adducts and the hydrido chlorides in the presence of strong
base are known precursors of the same fragment. (a) Gupta, M.; Hagen,
C.; Kaska, W. C.; Cramer, R. E.; Jensen, C. M. J. Am. Chem. Soc. 1997,
1
19, 840−841. (b) Renkema, K. B.; Kissin, Y. V.; Goldman, A. S. J. Am.
Chem. Soc. 2003, 125, 7770−7771. (c) Gottker-Schnetmann, I.; White,
P.; Brookhart, M. J. Am. Chem. Soc. 2004, 126, 1804−1811.
̈
iPr
iPr
(
28) ( PCOP)IrH was generated from the reported ( PCOP)
4
Ir(C H ) complex and H at room temperature. See the Supporting
2
4
2
Information for experimental details and spectral data.
29) Dehydroalkoxylation, catalyzed by lanthanide triflates and
(
thermodynamically driven by hydrogenation of the olefin product
catalyzed by Pd nanoparticles (at 110 °C in the case of acyclic ethers),
was recently reported by Marks and co-workers: Atesin, A. C.; Ray, N.
A.; Stair, P. C.; Marks, T. J. J. Am. Chem. Soc. 2012, 134, 14682−14685.
(
30) Afeefy, H. Y.; Liebman, J. F.; Stein, S. E. Neutral Thermochemical
Data. In NIST Chemistry WebBook; Linstrom, P. J., Mallard, W. G., Eds.;
National Institute of Standards and Technology, Gaithersburg, MD,
NIST Standard Reference Database Number 69, 20899, http://
webbook.nist.gov (retrieved February 11, 2013).
iPr
(
31) Conversion from a mixture of products to exclusively ( PCOP)
2
Ir(η -propene) is observed upon raising the temperature from ambient
to 120 °C. It seems safe to assume that at 150 °C, the temperature of
most of our experimental runs, the major species is still predominantly
I
dx.doi.org/10.1021/ja404566v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX