C-H and C-O bond activation with a rhodium(i) β-diiminate complex

Article abstract of DOI:10.1039/c4dt00309h

Complex LRh(COE) [L = (2,6-Me₂C₆H₃NCMe) ₂CH; COE = cyclooctene] reacts with oxirane, methyloxirane and 2,2-dimethyloxirane to eventually produce LRh(COE)(CO) and alkane (methane, ethane, propane). This reaction and other aspects of the reactivity of the LRh fragment (hydrogenation of olefins and benzene) have been studied by density functional theory. The results indicate that for 2,2-dimethyloxirane (and probably also for methyloxirane) the reaction starts with C-H activation of a methyl group, followed by ring opening and decarbonylation. For the parent oxirane, where this path is not available, the reaction starts with C-O activation to form a 2-rhodaoxetane, which then undergoes β-elimination. More generally, easy C-H activation, which appears to be a recurring theme in this Rh chemistry, is due to a close energy matching between corresponding Rh(i) and Rh(iii) complexes. For the analogous Ir complexes, preference for higher oxidation states is larger, leading to significantly higher barriers for e.g. hydrogenation. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c4dt00309h

Dalton
Transactions
View Article Online
View Journal | View Issue
PAPER
C–H and C–O bond activation with a rhodium(I)
β-diiminate complex†
Cite this: Dalton Trans., 2014, 43,
11286
Nicolle N. P. Langer,a Gurmeet Singh Bindrab and Peter H. M. Budzelaar*b
Complex LRh(COE) [L = (2,6-Me2C6H3NCMe)2CH; COE = cyclooctene] reacts with oxirane, methyl-
oxirane and 2,2-dimethyloxirane to eventually produce LRh(COE)(CO) and alkane (methane, ethane,
propane). This reaction and other aspects of the reactivity of the “LRh” fragment (hydrogenation of olefins
and benzene) have been studied by density functional theory. The results indicate that for 2,2-dimethyl-
oxirane (and probably also for methyloxirane) the reaction starts with C–H activation of a methyl group,
followed by ring opening and decarbonylation. For the parent oxirane, where this path is not available, the
reaction starts with C–O activation to form a 2-rhodaoxetane, which then undergoes β-elimination. More
generally, easy C–H activation, which appears to be a recurring theme in this Rh chemistry, is due to a
close energy matching between corresponding Rh(I) and Rh(III) complexes. For the analogous Ir com-
plexes, preference for higher oxidation states is larger, leading to significantly higher barriers for e.g.
hydrogenation.
Received 29th January 2014,
Accepted 26th March 2014
DOI: 10.1039/c4dt00309h
www.rsc.org/dalton
the most difficult classes of substrates. Experimental evidence
pointed to a “chain walking” of the metal fragment (via an
Introduction
Many complexes of the platinum group metals display activity allyl-hydride intermediate) as part of the catalytic cycle. Inter-
in olefin hydrogenation and/or C–H activation, usually estingly, the corresponding Ir complex “LIr(COE)” (1b), which
through mechanisms involving oxidative addition/reductive actually prefers an allyl-hydride structure, was found to be
elimination steps.1 For such reactivity to happen, two con- inactive in catalysis, forming a rather unreactive LIr(COE)(H)2
ditions must be fulfilled:
“dead-end” species.
1. The reactive metal fragment must be low-valent and
(highly) coordinatively unsaturated, in order to break unreac-
tive H–H or C–H bonds.
2. The preference for a higher oxidation state must not be
too strong, to avoid the complex getting “stuck” after oxidative
addition.
The delicate balance between these two conditions explains
why tuning of catalysts is a decidedly non-trivial art.
Over 10 years ago, we reported on the reactivity of rhodium
β-diiminate complex 1a [LRh(COE), COE = cyclooctene] in stoi-
chiometric oxidative addition reactions,2 as well as on its cata-
lytic activity in olefin hydrogenation.3 One remarkable aspect
of the hydrogenation catalysis was the potential for hydro-
genating tetrasubstituted olefins, which are normally among
We recently revisited the chemistry of complex 1a and
found that it also reacts with Si–H4 and B–H bonds,5 and is
capable of activating even Si–Csp3 bonds.4 Indications were
obtained that complex 1a, or a species derived from it, is
capable of (reversibly) activating C–H bonds of THF. Intrigued
by this observation, we have now studied the reaction of
complex 1a with epoxides, which might be expected to
undergo either C–H or C–O bond activation. In fact, we find
that both reactions happen; interestingly, C–H activation
appears to precede C–O activation. Extensive DFT studies have
been carried out to elucidate the mechanism of this epoxide
activation as well as details of the above-mentioned hydrogen-
ation reactions. The results confirm earlier ideas about “chain
walking” and shed light on the distinctly different reactivity
patterns of analogous Rh and Ir complexes.
aDepartment of Chemistry, Radboud University, Heyendaalseweg 135,
6525 AJ Nijmegen, The Netherlands
bDepartment of Chemistry, University of Manitoba, Winnipeg, MB R3 T 2N2,
Canada. E-mail: Peter.Budzelaar@umanitoba.ca
†Electronic supplementary information (ESI) available: 1H and 13C NMR spectra
for reactions of 1a with oxiranes. Total energies and thermal corrections for all
species studied computationally. Energy profile for Ir analogue of Fig. 5. Atomic
coordinates for all species studied. See DOI: 10.1039/c4dt00309h
11286 | Dalton Trans., 2014, 43, 11286–11294
This journal is © The Royal Society of Chemistry 2014

View Article Online
Dalton Transactions
Paper
Results and discussion
Activation of epoxides
LRh(COE) reacts slowly (∼1 day at RT; ∼1 hour at 40 °C) with
methyloxirane in THF. The final product of this reaction is
LRh(COE)(CO) (2a, Scheme 1), which was characterized by
NMR. The structure of this product is presumably similar to
those of the structurally characterized complexes LRh(COE)-
(MeCN)3 and LiPrRh(COE)(N2).6
Ring opening of oxiranes mediated by Rh has been reported
before,7–9 but the emphasis has mostly been on metal-cata-
lyzed/metal-directed nucleophilic attack rather than the
present breakdown of the epoxide. Formation of 2a plausibly
involves “opening” of methyloxirane to propionaldehyde, fol-
lowed by decarbonylation (Scheme 2). Indeed, reaction of 1a
with propionaldehyde rapidly produces 2a in near-quantitative
yield (by 1H NMR). Decarbonylation of aldehydes at Rh has
been studied both experimentally10–13 and computationally10
and appears to be well-understood. However, the exact mode
of initial ring opening of methyloxirane is not obvious from
these results. If one assumes that the first step is oxidative
addition of a C–O bond to Rh, then the formation of 2a
implies a preference for attack at the substituted rather than
the unsubstituted C–O bond. Attack at the unsubstituted bond
would eventually produce acetone (see Scheme 2), which was
not observed; moreover, treatment of 1a with acetone did not
produce any 2a.
Scheme 3 Potential ring opening paths for dimethyloxirane.
contrast, the reaction of 1a with 2,2-dimethyloxirane was
found to be much faster (∼minutes at RT), while still yielding
mainly 2a. Assuming the reaction starts with C–O bond clea-
vage, these results would suggest a clear preference for break-
ing the more highly substituted C–O bond This is even more
remarkable considering that complex 1a has a rather bulky
β-diiminate ligand, which could be expected to strongly favour
attack at the less substituted bond. We also tested cyclohexene
epoxide, which interestingly did not react at all with 1a. This
led us to consider the possibility that maybe C–O activation is
not the first step of the reaction. In view of the reported C–H
activation of THF by 1a (or a complex derived from it),4 initial
C–H or C–C activation would also be conceivable. Based on
these considerations, we came up with the five paths shown in
Scheme 3 for opening of dimethyloxirane by 1a (dimethyl-
oxirane was chosen here because of its higher symmetry, and
also because it gives the cleanest and fastest reaction). At least
three of the paths could reasonably be expected to eventually
produce 2a. Distinguishing between these alternatives by
experiment would be hard, so we turned to DFT methods to
analyze all of them.
We then tested the reaction of 1a with unsubstituted
oxirane and with 2,2-dimethyloxirane. Reaction with oxirane
was found to be even slower than with methyloxirane, and was
also less clean. The main product was still 2a but NMR spectra
indicated that this was formed in only about 70% yield, and
was accompanied by several unidentified by-products. In
DFT results for oxirane opening
Transition states were located for all paths of Scheme 3 except
for a few β-eliminations, which were found to be virtually bar-
rierless. DFT results indicate that the preferred path for ring
opening of dimethyloxirane involves C–H activation of a methyl
group. Scheme 4 shows intermediates on this path. The reac-
tion starts with (associative) displacement of COE by the
oxirane. Methyl C–H activation comes next, following by C–O
cleavage. The resulting (enolate)(hydride) complex E under-
goes insertion of the CvC bond in the Rh–H bond, followed
by β-H elimination, forming (β-ketoalkyl)(hydride) species G.
Reductive elimination now produces side-on coordinated
isobutyraldehyde complex H, which undergoes oxidative
addition of the aldehyde C–H bond, CO deinsertion, reductive
elimination of alkane, and capture of COE to yield the final
product (L).
Scheme 1 Reaction of 1a with oxiranes.
Fig. 1 shows the geometries of important intermediates and
transition states along the preferred path for ring opening (a
complete set of figures for both the main path is included in
the ESI†). Coordination to Rh does not affect the geometry of
the oxirane much (C). The subsequent methyl C–H activation
Scheme 2 Possible mechanism for formation of 2a from 1a.
This journal is © The Royal Society of Chemistry 2014
Dalton Trans., 2014, 43, 11286–11294 | 11287

View Article Online
Paper
Dalton Transactions
accompanied by a further contraction of the Rh–O bond. In
the path depicted in Fig. 1, the metal-bound hydride and the
methyl group of the enolate in E end up on the same side of the
enolate ring plane, resulting in easy transfer of the hydride to a
ring carbon to form F. The initial C–H activation and ring
opening can also follow a diastereomeric path, with very similar
energies for each step, where the hydride and methyl groups of
E′ end up on opposite sides of the ring plane. Now, hydride
transfer to form F is much more difficult. Instead, the lowest-
energy path involves decoordination and face flipping of the
double bond via TSE′E to arrive at E and continuing from there
(see the ESI† for these structures). TSE′E is, at 10.1 kcal mol−1
,
only 8.9 kcal mol−1 above E′, so this face flip is perfectly feasible.
Fig. 2 shows the calculated free energies for relevant species
on all paths of Scheme 3 (a complete list of all energies can be
found in the ESI†).
Scheme 4 Preferred path (according to DFT) for ring opening of
dimethyloxirane at “LRh”.
Interestingly, the results show unambiguously that C–H
activation of an epoxide methyl C–H bond is the rate-limiting
step of the preferred ring-opening path. Once this bond has
been broken, C–O cleavage in the internally coordinated oxi-
ranylmethyl complex is facile, and from there on a cascade of
downhill reactions leads via the π-acyl complex to the final
product 2a (L). While there is not much precedent for this
mode of oxirane ring opening, it is rather similar to the well-
documented easy opening of oxiranylmethyl radicals.14
Initial activation of an oxirane ring C–H bond is nearly as
easy as methyl C–H activation. However, the subsequent C–O
cleavage step has a prohibitively high barrier, making this a
non-productive path. C–C cleavage of the oxirane is somewhat
more difficult; in addition, this step does not lead to signifi-
cantly more stable products and is therefore also non-pro-
ductive. Perhaps more surprisingly, direct C–O cleavage of the
substituted CMe2–O bond is not competitive either.15 Cleavage
of the unsubstituted CH2–O bond (which is indeed easier than
of the substituted bond, as expected on steric grounds) is non-
productive. Thus, we conclude that methyl C–H activation pre-
cedes ring opening; this explains why the larger “hindrance”
in dimethyloxirane does not result in a slower reaction.
For unsubstituted oxirane, the energy profile is subtly
different (Fig. 3). Methyl C–H activation is obviously not poss-
Fig. 1 Optimized structures for some intermediates and transition
states along the preferred ring opening path of Scheme 3. Much of the
ancillary ligand has been omitted for clarity.
occurs with retention of the O→Rh coordination, and is pre-
sumably assisted by it. The addition product (D) has a smaller
Rh–O distance and a clearly elongated ring C–O bond (by
about 0.1 Å), reflecting stronger intramolecular coordination
of oxygen to Rh(III). From there, ring opening via TSDE is
Fig. 2 Calculated free-energy profile for reaction of 1a with 2,2-
dimethyloxirane. The solid black line represents the preferred path,
involving methyl C–H activation.
11288 | Dalton Trans., 2014, 43, 11286–11294
This journal is © The Royal Society of Chemistry 2014

View Article Online
Dalton Transactions
Paper
Fig. 3 Calculated free-energy profile for reaction of 1a with oxirane.
The solid black line represents the preferred path, involving ring C–O
activation.
ible. Ring C–C activation has the lowest barrier but is non-pro-
ductive. Ring C–H activation is feasible but the subsequent
ring-opening reaction is not. However, ring C–O activation now
becomes a viable mechanism; the preferred path is summar-
ized in Scheme 5, and some relevant structures are shown in
Fig. 4. The initial product of C–O activation is 2-rhodaoxetane
M′. Only a handful of simple 2-rhodaoxetanes have been struc-
turally characterized;16–22 the majority derive from cyclo-
octadiene16,18,21 and are stabilized by intramolecular
coordination of the remaining CvC bond. The calculated struc-
ture of M′ agrees fairly well with the X-ray structures reported
for (TPA)Rh(CH2CH2O)+ and (MeTPA)Rh(CH2CH2O)+.17,19
Complex M′ easily undergoes β-elimination to give (enolate)-
(hydride) complex N′, which then undergoes reductive elimin-
ation to π-aldehyde complex H′, from which point the reaction
proceeds as for dimethyloxirane. This calculated path roughly
parallels that proposed by Milstein for reaction of oxirane with
Ir(PMe3)3(COE)Cl,23 the main difference being that in that Ir
case the reaction stops at the (hydrido)(enolate) stage unless
forcing conditions are employed. We tentatively ascribe the
difference in reactivity in part to the higher tendency of Ir to
remain at the Ir(III) oxidation state, and in part to the higher
degree of unsaturation of our LRh fragment.
Fig. 4 Optimized structures for some intermediates and transition
states along the preferred ring opening path of Scheme 5. Much of the
ligand has been omitted for clarity.
ence in reaction paths for the two substrates can be explained
as follows:
(a) Reaction via O–CH2 activation, preferred for oxirane, is
non-productive for dimethyloxirane because the resulting
rhodaoxetane does not have a ring β-hydrogen. Indeed,
Milstein has isolated a 4,4-dimethyl-2-rhodaoxetane and found
it to be quite stable.20,22
The calculated effective barrier for opening of oxirane
(28.9 kcal mol−1) is significantly higher than that for dimethyl-
oxirane (23.7 kcal mol−1), which qualitatively agrees with the
higher observed rate of dimethyloxirane opening. The differ-
(b) Reaction via O–CMe2 activation is blocked for dimethyl-
oxirane because of steric hindrance.
(c) Reaction via methyl C–H activation provides a low-
energy pathway for dimethyloxirane (and presumably mono-
methyloxirane) that is not available for the parent oxirane.
In this context, it should be noted that ring-lithiated oxiranes
are known to rearrange to enolates.24,25 This reaction has been
studied computationally26,27 and has a relatively high activation
energy (≈ 30 kcal mol−1). In view of the much lower oxophilicity
of Rh compared to Li, it is perhaps not surprising that oxiranyl-
rhodium species would not readily rearrange to enolates.
Hydrogenation and C–H activation
In the absence of additional donor molecules (like N2 or THF),
LRh(COE) is fluxional, showing only two broadened 1H reso-
nances (ratio 6 : 8) for the COE ligand.28 Based on calculations
Scheme 5 Preferred path (according to DFT) for ring opening of
oxirane at “LRh”.
This journal is © The Royal Society of Chemistry 2014
Dalton Trans., 2014, 43, 11286–11294 | 11289

View Article Online
Paper
Dalton Transactions
for a simple model system, we proposed that in the complex
Results for catalytic hydrogenation of DMB are significantly
the Rh “walks” around the ring via an allyl-hydride intermedi- different. We have studied both the direct hydrogenation path
ate; the Ir analog “LIr(COE)” actually prefers such an allyl- and an indirect path involving prior isomerization of the
hydride structure but is non-fluxional. In the present study we olefin within the coordination sphere. While 2-DMB is the
have modelled this fluxionality and subsequent hydrogenation thermodynamically preferred isomer of the free olefin, the pre-
activity, using the complete ligand in order to obtain results ferred form of the complexed olefin is 2,3-dimethyl-1-butene
comparable with experiment. The results support the allyl- (1-DMB). The results (summarized in Fig. 6) show that the
hydride ring walking mechanism proposed earlier; the calcu- direct hydrogenation path is more difficult than the one invol-
lated free energy of activation (14.6 kcal mol−1) is compatible ving prior isomerization (effective barrier 18.0 vs. 14.7 kcal
with the experimentally observed fluxionality,29 and much mol−1). Thus, it is likely that in hydrogenation of tetrasubsti-
smaller than the one calculated previously for a strongly sim- tuted olefins with 1a the Rh center would tend to migrate to a
plified model system (ΔH‡ ≈ 24 kcal mol−1).3
primary carbon, if possible, before being hydrogenated off. It
Catalytic hydrogenation with “LRh” was studied computa- should also be noted that for this more hindered olefin the
tionally for COE and 2,3-dimethyl-2-butene (2-DMB) as sub- resting state is not an olefin complex but LRh(H2).
strates. The mechanism of COE hydrogenation is
straightforward (see Fig. 5). LRh(COE) first binds a dihydrogen
molecule, giving LRh(COE)(H2), which is the resting state and
Arene hydrogenation
The reaction of LRh(COE) with hydrogen and benzene is rather
subtle.3,30 Reaction of LRh(COE) alone with H2 in an “inert”
solvent (THF or cyclohexane) initially leads to LRh(COE)(H2),
and then to a complex tentatively identified as [LRhH]2(μ-H)2.
Dissolution of LRh(COE) in benzene leads to slow (days) for-
mation of [LRh]2(μ–η4:η4-C6H6), which does not react with H2.
If a solution of LRh(COE) in benzene is treated with H2 for
10 min, and the H2 atmosphere is then removed, NMR shows
that the dominant species in solution is LRh(η4-C6H6), which
slowly (∼1 day) disproportionates to [LRh]2(μ–η4:η4-C6H6) and
C6H6. Finally, if a solution of LRh(COE) in benzene is kept
under H2 for one day, the main product is LRh(1,3-CHD)
(CHD = cyclohexadiene). Catalytic hydrogenation of benzene
was never observed, and would not be expected on thermo-
dynamic grounds.
1
has actually been observed by H NMR.3 This then undergoes
H–H cleavage and insertion of the olefin in one of the Rh–H
bonds; interestingly, our calculations indicate that there is no
local minimum corresponding to
a discrete dihydride
complex. In any case, after insertion the resulting (alkyl)-
(hydride) complex can either undergo immediate reductive
elimination, or first bind a dihydrogen molecule. These
alternatives appear to be close in energy; the path without
prior H2 coordination is preferred, and has an effective barrier
of 11.5 kcal mol−1. The resulting cyclooctane (COA) molecule
is only weakly bound to Rh and is easily lost; the remaining
LRh or LRh(H2) fragment picks up a new COE molecule to
close the catalytic cycle.
From the energies of the species involved, it is clear that
“ring walking” of LRh(COE) via the allyl-hydride mechanism is
not competitive with hydrogenation. On the other hand, rever-
sion of alkyl-hydride species back to LRh(COE)(H2) is easy
relative to the reductive elimination step, leading to the expec-
tation that the LRh fragment could walk along the substrate
backbone prior to elimination. These are important issues
when considering the design of stereoselective variations of
the catalysis.
We have modeled this arene reactivity of LRh(COE), again
using the complete ligand without simplifications. Calculated
energies are summarized in Fig. 7 (formation of “LRh”/“LRh-
(H2)” and COA from 1a and H2 has been described above and
Fig. 6 Calculated free-energy profile for catalytic 2-DMB hydrogen-
ation with “LRh”. The solid black line represents the preferred path,
involving isomerization and H2 coordination prior to C–H reductive
elimination. The red line corresponds to elimination without prior H2
coordination; the blue and violet paths are without prior DMB isomeri-
zation. The green step corresponds to non-productive isomerization to
a 2-iPr-allyl derivative.
Fig. 5 Calculated free-energy profile for catalytic COE hydrogenation
using 1a. The solid black line represents the preferred path; the red line
is the alternative path involving H2 coordination prior to C–H reductive
elimination. The green step corresponds to “ring walking” via the allyl-
hydride mechanism.
11290 | Dalton Trans., 2014, 43, 11286–11294
This journal is © The Royal Society of Chemistry 2014

View Article Online
Dalton Transactions
Paper
Benzene binds even more weakly than H2.32 Calculations
1
support the assignment (based on H and 13C NMR) of an η4-
bound structure for the adduct LRh(C6H6) (Fig. 8, A), but
judging from the ring bond lengths (1.36–1.44 Å) the loss of
aromaticity is only partial. Coordination of H2 to this complex,
with concomitant reorganization of the C6H6 moiety to an η2-
bound mode (B), is slightly exergonic. Migration of a hydride
to the benzene ring, forming LRh(η3-C6H7)(H) (C), is the most
difficult step, with a barrier of 22.0 kcal mol−1. The high
barrier of this step, compared to COE insertion in LRh(COE)-
(H2) (6.2 kcal mol−1, vide supra), is likely due to the complete
loss of aromaticity associated with benzene insertion. Complex
C can then rearrange to either a 1,3-CHD or a 1,4-CHD
complex; the former (D) is preferred on both kinetic and
thermodynamic grounds. The binding of the CHD ligand in
LRh(1,3-CHD) is much stronger than that of benzene in LRh-
(C6H6) (31.8 vs. 6.2 kcal mol−1), thus preventing ligand substi-
tution which might have rendered the system catalytic.33 In the
absence of excess hydrogen, i.e. if H2 is only used to generate
“LRh” from 1a, [LRh]2(μ–η4:η4-C6H6) is predicted to be the pre-
ferred product, in agreement with experimental observations.
Fig. 7 Calculated free-energy profile for reaction of “LRh” with
benzene and hydrogen.
is not included in the profile); geometries on the path to LRh-
(1,3-CHD) are shown in Fig. 8. In the absence of added
donors, formation of a dimer [LRhH]2(μ-H)2 is indeed found to
be feasible; the calculated structure shows some resemblance
to silane complexes reported by Tilley.31 LRh(H2) can also bind
a second H2 molecule, giving a bis(dihydrogen) complex (in
contrast to its Ir analogue, vide infra). This second H2 mole-
cule, like the first one, is only weakly bound; based on the
calculated energies one would expect easy dissociation.
A comparison of Rh and Ir
Experimentally, LRh(COE) is a good catalyst (precursor) for
olefin hydrogenation, while “LIr(COE)” on treatment with
hydrogen forms unreactive LIr(COE)H2.3 To further investigate
the differences in behaviour between the two metals, we calcu-
lated the Ir analogues of the Rh hydrogenation cycles dis-
cussed above. While the cycles are roughly similar, Ir shows a
much stronger preference for higher oxidation states, in par-
ticular in complexes with H2. Whereas for Rh we consistently
find dihydrogen complexes, with dihydrides not even corres-
ponding to local minima, the reverse is true for Ir, as pre-
viously reported by Chirik.34 As an illustration, Fig. 9 compares
the calculated structures of LRh(H2), LRh(COE)(H2), LRh(H2)2
with their Ir polyhydride analogues: calculated H–H bond
lengths for Rh(H2) complexes are all close to 0.88 Å, whereas
the H⋯H distances for the Ir polyhydrides vary from 1.42 to
1.64 Å. At the same time, metal–hydrogen distances are much
larger for Rh (1.67–1.70 Å) than for Ir (1.56–1.58 Å).
In order to further quantify this preference, we optimized
the Rh structures with H–H distances constrained to those of
their Ir analogs, and vice versa (see Table 1). The results indi-
cate a preference of Rh for dihydrogen complex structures of
about 3–4 kcal mol−1, and a similar preference of Ir for di-
hydride structures. Particularly noteworthy is the resistance of
Rh (by about 10 kcal mol−1) to form the formally pentavalent
LRhH4 structure.
In the context of olefin hydrogenation, the preference of Ir
for higher oxidation states causes LIr(COE)H2 to be a much
more stable resting state than LRh(COE)(H2), and results in a
rather high barrier for the path not involving extra H2.
However, the calculated effective barrier for the path with extra
H2 is low enough (11.6 kcal mol−1) that one still would expect
some hydrogenation activity via this route. It seems likely that
the method we use underestimates the barriers of hydrogen
Fig. 8 Calculated structures along the path from LRh(η4-C6H6) to LRh-
(η4-C6H8). Much of the diiminate ligand has been omitted for clarity.
This journal is © The Royal Society of Chemistry 2014
Dalton Trans., 2014, 43, 11286–11294 | 11291

View Article Online
Paper
Dalton Transactions
metallation variations, has so far precluded a definitive
conclusion.
Conclusions
LRh(COE) is a relatively simple molecule giving rise to a sur-
prising diversity of reactions, many of which involve C–H acti-
vation and several types of “chain walking”. Complex reaction
sequences are viable because the Rh(I) and Rh(III) oxidation
states are closely balanced in this ligand environment. An
example is the activation of methyloxirane and dimethyl-
oxirane, which passes through about 10 steps before ending
up in the “dead” species LRh(COE)(CO) in which finally the
π-acidic CO ligand “freezes” Rh in the univalent state. In con-
trast, β-diiminate Ir complexes have a much larger preference
for the Ir(III) [or even Ir(V)] oxidation state, leading to shorter
reaction sequences, less fluxionality and less efficient catalysis.
Fig. 9 Comparison of Rh dihydrogen complexes with Ir polyhydride
analogues: A LRh(H2); B LIrH2; C LRh(H2)2; D LIrH4; E LRh(COE)(H2); F
LIr(COE)H2. Much of the diiminate ligand has been omitted for clarity.
Experimental
General
All reactions were carried out under an atmosphere of Ar in
Schlenk glassware. Solvents were distilled from Na/benzo-
phenone prior to use. Acetone was distilled and stored under
N2 on 3 Å molecular sieve. Ethylene oxide (1.2 M in THF), ( )
propylene oxide and isobutylene oxide were purchased from
Sigma Aldrich/TCI chemicals; the latter two were distilled
under reduced pressure prior to use. LRh(COE) was prepared
according to a published procedure.3
Table 1 Relative energies (kcal mol−1) of constraineda hydride and
dihydrogen complexes
b
b
Constrained
Erel
Constrained
Erel
LRhH2
LRh(COE)H2
LRhH4
4.20
3.37
10.58
LIr(H2)
LIr(COE)(H2)
LIr(H2)2
4.62
4.06
3.90
a H–H distances for Rh complexes constrained at values for Ir
analogues, and vice versa, see Fig. 9. b Electronic energy relative to fully
optimized structure.
Reaction of LRh(COE) with methyloxirane
0.077 g of LRh(COE) was dissolved in 2.5 ml methyloxirane–
diethylether (1 : 8, v : v), corresponding to
a ratio Rh :
methyloxirane = 1 : 30. The solution was heated for 1 hour at
40 °C. The solvent was removed in vacuo and the residue was
dissolved in pentane. The solution was kept for one night at
−30 °C. The resulting suspension was filtered and the filtrate
was dried in vacuo, leaving LRh(COE)(CO) as a yellowish-white
solid.
transfer reactions,35 possibly more for Ir than for Rh. The DFT
results also produce virtually equal energies for the (cyclo-
octene) and (cyclooctenyl)(hydride) isomers of “LIr(COE)”,
with an interconversion barrier of only 6.6 kcal mol−1. Experi-
mentally, the equilibrium is clearly on the side of the (cyclo-
octenyl)(hydride) isomer, and the barrier must be significantly
larger than 6.6 kcal mol−1 to produce static NMR spectra. We
are not sure at present whether this problem is caused by the
choice of functional, basis set or pseudopotential for Ir.
1H NMR (300 MHz, THF-d8, 25 °C), δ (ppm): 7.1–6.8 (m,
6H, m + p), 5.13 (s, 1H, β), 3.22 (m, 2H, vCH), 2.39, 2.21 (s, 6H
each, o-CH3), 2.11, 1.49 and 1.21 (m, 2H each, CH2), 1.62, 1.51
(s, 3H each, NvCCH3).
13C NMR (75 MHz, THF-d8, 25 °C), δ (ppm): 189.7 (d, JRhC
73 Hz), 160.8, 159.1 (NvC), 159.0, 149.3 (i), 133.4, 131.4 (o),
129.2, 128.8 (m), 125.9, 125.5 (p), 98.5 (β), 79.9 (d, JRhC 11 Hz,
vCH), 32.0, 31.7, 31.2 (CH2), 27.0, 22.3 (NvCCH3), 19.5, 18.9
(o-CH3).
One reaction for which we do not yet have a satisfactory
explanation is the decomposition (disproportionation) of “LIr-
(COE)” to LIr(COE)H2 and LIr(1,4-COD). It is clear that some
form of C–H activation must be involved. Also, if the reaction
is carried out in THF-d8 significant D incorporation was seen
at the hydride position, indicating participation of a highly
reactive, C–H activating intermediate. However, the multitude
of conceivable paths, including radical reactions and ligand
IR (KBr pellet): νCO 1979 cm−1
.
C30H39N2ORh (546.53), calc (%) C 65.93, H 7.19, N 5.12;
found C 64.41, H 6.97, N 5.10.
11292 | Dalton Trans., 2014, 43, 11286–11294
This journal is © The Royal Society of Chemistry 2014

View Article Online
Dalton Transactions
Paper
Check for acetone formation
Acknowledgements
0.0505 g of LRh(COE) was dissolved in a solution of 7.60 ×
10−4 mole of methyloxirane (7.8 eq.) in THF-d8. The solution
was heated for 1 hour at 40 °C, after which an NMR spectrum
was recorded. No acetone was observed. 1 equivalent of
methyloxirane had been consumed to form LRh(COE)(CO),
and the remaining 7 eq. was still present in the solution.
We thank Mr Mark Cooper (University of Manitoba) for help
with X-ray structure determinations and Prof. Frank
Hawthorne for the use of their single-crystal X-ray diffracto-
meter. This work was supported in part by NSERC, CFI, and
MRIF grants (to P.H.M.B.).
Reaction of LRh(COE) with propanal
To LRh(COE) was added an excess of a solution of propanal in
diethyl ether (1 : 10). The solution was heated at 40 °C for Notes and references
1 hour. The solvent was removed in vacuo and the residue was
1 D. Steinborn, Fundamentals of Organometallic Catalysis,
Wiley-VCH, Weinheim, Germany, 2012.
identified as LRh(COE)(CO) by 1H and 13C NMR.
Reaction of LRh(COE) with oxirane
2 S. T. H. Willems, P. H. M. Budzelaar, N. N. P. Moonen,
R. de Gelder, J. M. M. Smits and A. W. Gal, Chem. – Eur. J.,
2002, 8, 1310–1320.
3 P. H. M. Budzelaar, N. N. P. Moonen, R. de Gelder,
J. M. M. Smits and A. W. Gall, Eur. J. Inorg. Chem., 2000,
753–769.
4 D. Zhu, D. J. Kozera, K. D. Enns and P. H. M. Budzelaar,
Angew. Chem., Int. Ed., 2012, 51, 12211–12214.
5 D. Zhu, G. S. Bindra and P. H. M. Budzelaar, to be
published.
To a purple-blue solution of LRh(COE) (0.100 g, 0.193 mmol)
in 2 mL diethyl ether under an argon atmosphere was added
4.8 mL of a 1.2 M oxirane solution in THF (30 eq., 5.76 mmol).
The reaction mixture turned brown immediately. After stirring
for 24 hours at room temperature, the solvents were removed
in vacuo, leaving a brownish solid which mostly consisted of
LRh(COE)(CO) (1H and 13C NMR).
Reaction of LRh(COE) with 2,2-dimethyloxirane
6 J. D. Masuda and D. W. Stephan, Can. J. Chem., 2005, 83,
324–327.
7 J. D. Ha, E. Y. Shin, S. K. Kang, J. H. Ahn and J. K. Choi,
Tetrahedron Lett., 2004, 45, 4193–4195.
8 K. Fagnou and M. Lautens, Org. Lett., 2000, 2, 2319–2321.
9 Y. Fukumoto, N. Chatani and S. Murai, J. Org. Chem., 1993,
58, 4187–4188.
10 P. Fristrup, M. Kreis, A. Palmelund, P.-O. Norrby and
R. Madsen, J. Am. Chem. Soc., 2008, 130, 5206–5215.
11 J. E. Baldwin, T. C. Barden, R. L. Pugh and W. C. Widdison,
J. Org. Chem., 1987, 52, 3303–3307.
12 M. C. Baird, C. J. Nyman and G. Wilkinson, J. Chem. Soc. A,
1968, 348–351.
13 M. C. Baird, J. T. Mague, J. A. Osborn and G. Wilkinson,
J. Chem. Soc. A, 1967, 1347–1360.
14 D. J. Pasto, J. Org. Chem., 1996, 61, 252–256.
15 We checked for solvent effects on the competition between
C–H activation and O–CMe2 cleavage; these were found to
be minimal (see ESI† for more details).
16 C. Tejel, M. A. Ciriano, E. Sola, M. P. del Rio, G. Rios-
Moreno, F. J. Lahoz and L. A. Oro, Angew. Chem., Int. Ed.,
2005, 44, 3267–3271.
Purple blue solid LRh(COE) (1.0 eq., 70 mg, 0.135 mmol) dis-
solved in 2 mL diethyl ether was added to anhydrous isobutyl-
ene oxide (20 eq., 0.36 mL, 4.05 mmol) under an argon
atmosphere. The reaction mixture initially turned deep green
and then brown. The reaction mixture was stirred at room
temperature under argon for 24 hours. The solvent was evapor-
ated in vacuo leaving a brownish solid which mostly consisted
of LRh(COE)(CO) (1H and 13C NMR).
Computational details
Geometry optimizations were carried out with Turbomole36
using the TZVP basis37 and the b3-lyp functional38–41 in com-
bination with an external optimizer (PQS OPTIMIZE42,43).
Vibrational analyses were carried out for all stationary points
to confirm their nature (1 imaginary frequency for transition
states, none for minima). Final energies were obtained using
the TZVPP basis.44 These were combined with thermal correc-
tions (enthalpy and entropy, 273 K, 1 bar) from the TZVP
vibrational analyses to arrive at the final free energies. To
account for the reduced freedom of movement in solution,
entropy contributions to the free energies were scaled to 2/3 of
their gas-phase values.45,46
17 B. de Bruin, M. J. Boerakker, J. A. W. Verhagen, R. de
Gelder, J. M. M. Smits and A. W. Gal, Chem. – Eur. J., 2000,
6, 298–312.
18 B. de Bruin, J. A. Brands, J. Donners, M. P. J. Donners,
R. de Gelder, J. M. M. Smits, A. W. Gal and A. L. Spek,
Chem. – Eur. J., 1999, 5, 2921–2936.
Abbreviations
COE
COA
Cyclooctene
Cyclooctane
19 B.
deBruin,
M.
J.
Boerakker,
J.
Donners,
1-DMB 2,3-Dimethyl-1-butene
2-DMB 2,3-Dimethyl-2-butene
B. E. C. Christiaans, P. P. J. Schlebos, R. deGelder,
J. M. M. Smits, A. L. Spek and A. W. Gal, Angew. Chem., Int.
Ed. Engl., 1997, 36, 2064–2067.
CHD
Cyclohexadiene
This journal is © The Royal Society of Chemistry 2014
Dalton Trans., 2014, 43, 11286–11294 | 11293

View Article Online
Paper
Dalton Transactions
20 M. J. Calhorda, A. M. Galvao, C. Unaleroglu, A. A. Zlota, 32 The level of theory employed probably underestimates the
F. Frolow and D. Milstein, Organometallics, 1993, 12, 3316–
3325.
benzene binding energy. According to 1H NMR, benzene
does not exchange rapidly at room temperature.
21 V. W. Day, W. G. Klemperer, S. P. Lockledge and D. J. Main, 33 Catalysis producing CHD would not have been expected
J. Am. Chem. Soc., 1990, 112, 2031–2033.
22 A. A. Zlota, F. Frolow and D. Milstein, J. Am. Chem. Soc.,
1990, 112, 6411–6413.
anyway since the reaction C6H6 + H2 → C6H8 is endergonic.
Complete hydrogenation to cyclohexane would be exergo-
nic but was never observed.
23 D. Milstein and J. C. Calabrese, J. Am. Chem. Soc., 1982, 34 W. H. Bernskoetter, E. Lobkovsky and P. J. Chirik, Chem.
104, 3773–3774. Commun., 2004, 764–765.
24 D. M. Hodgson, P. G. Humphreys and S. R. Hughes, Pure 35 G. Talarico, A. N. J. Blok, T. K. Woo and L. Cavallo, Organo-
Appl. Chem., 2007, 79, 269–279. metallics, 2002, 21, 4939–4949.
25 V. Capriati, S. Florio, F. M. Perna, A. Salomone, A. Abbotto, 36 TURBOMOLE V6.3, development of University of
a
M. Amedjkouh and S. O. N. Lill, Chem. – Eur. J., 2009, 15,
7958–7979.
26 B. R. Ramachandran, S. Waithe and L. M. Pratt, J. Org.
Chem., 2013, 78, 10776–10783.
27 A. Bouyacoub, R. Hadjadj-Aoul and F. Volatron, Eur. J. Org.
Chem., 2008, 4466–4474.
28 P. H. M. Budzelaar, R. de Gelder and A. W. Gal, Organome-
tallics, 1998, 17, 4121–4123.
Karlsruhe and Forschungszentrum Karlsruhe GmbH,
1989–2007, TURBOMOLE GmbH, since 2007, Karlsruhe,
2011.
37 A. Schäfer, C. Huber and R. Ahlrichs, J. Chem. Phys., 1994,
100, 5829–5835.
38 C. T. Lee, W. T. Yang and R. G. Parr, Phys. Rev. B: Condens.
Matter, 1988, 37, 785–789.
39 A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652.
29 Precise determination of activation parameters from the 40 A. D. Becke, J. Chem. Phys., 1993, 98, 1372–1377.
NMR spectra is not possible because the slow-exchange 41 The Turbomole functional “b3-lyp” is similar (but not iden-
spectrum is much too complicated to allow full simu-
lation of dynamic effects. However, an estimated 42 PQS and J. Baker, Parallel Quantum Solutions, Fayetteville,
“coalescence temperature” of ∼−10 °C and a “Δδ” of AR, 2001.
∼2 ppm produce an estimated free energy of activation 43 J. Baker, J. Comput. Chem., 1986, 7, 385–395.
tical) to the Gaussian “B3LYP” functional.
of ∼11.5 kcal mol−1
30 P. H. M. Budzelaar, N. N. P. Moonen, R. de Gelder,
.
44 F. Weigend, F. Furche and R. Ahlrichs, J. Chem. Phys., 2003,
119, 12753–12762.
J. M. M. Smits and A. W. Gal, Chem. – Eur. J., 2000, 6, 2740– 45 R. Raucoules, T. de Bruin, P. Raybaud and C. Adamo,
2747. Organometallics, 2009, 28, 5358–5367.
31 J. L. McBee and T. D. Tilley, Organometallics, 2009, 28, 46 S. Tobisch and T. Ziegler, J. Am. Chem. Soc., 2004, 126,
5072–5081.
9059–9071.
11294 | Dalton Trans., 2014, 43, 11286–11294
This journal is © The Royal Society of Chemistry 2014