How phenyl makes a difference: Mechanistic insights into the ruthenium(ii)-catalysed isomerisation of allylic alcohols

Article abstract of DOI:10.1039/c3sc52612g

[RuCl(η⁵-3-phenylindenyl)(PPh₃)₂] (1) has been shown to be a highly active catalyst for the isomerisation of allylic alcohols to the corresponding ketones. A variety of substrates undergo the transformation, typically with 0.25-0.5 mol% of catalyst at room temperature, outperforming commonly-used complexes such as [RuCl(Cp)(PPh₃) ₂] and [RuCl(η⁵-indenyl)(PPh₃) ₂]. Mechanistic experiments and density functional theory have been employed to investigate the mechanism and understand the effect of catalyst structure on reactivity. These investigations suggest a oxo-π-allyl mechanism is in operation, avoiding intermediate ruthenium hydride complexes and leading to a characteristic 1,3-deuterium shift. Important mechanistic insights from DFT and experiments also allowed for the design of a protocol that expands the scope of the transformation to include primary allylic alcohols.

Full text of DOI:10.1039/c3sc52612g

Chemical
Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
How phenyl makes a difference: mechanistic
insights into the ruthenium(II)-catalysed
Cite this: Chem. Sci., 2014, 5, 180
isomerisation of allylic alcohols†
Simone Manzini,a Albert Poater,b David J. Nelson,a Luigi Cavallocd
a
*
and Steven P. Nolan
[RuCl(h5-3-phenylindenyl)(PPh3)2] (1) has been shown to be a highly active catalyst for the isomerisation of
allylic alcohols to the corresponding ketones. A variety of substrates undergo the transformation, typically
with 0.25–0.5 mol% of catalyst at room temperature, outperforming commonly-used complexes such as
[RuCl(Cp)(PPh3)2] and [RuCl(h5-indenyl)(PPh3)2]. Mechanistic experiments and density functional theory
have been employed to investigate the mechanism and understand the effect of catalyst structure on
reactivity. These investigations suggest a oxo-p-allyl mechanism is in operation, avoiding intermediate
ruthenium hydride complexes and leading to a characteristic 1,3-deuterium shift. Important mechanistic
insights from DFT and experiments also allowed for the design of a protocol that expands the scope of
the transformation to include primary allylic alcohols.
Received 17th September 2013
Accepted 16th October 2013
DOI: 10.1039/c3sc52612g
www.rsc.org/chemicalscience
Introduction
The redox isomerisation of allylic alcohols catalysed by transi-
tion metal complexes represents a powerful, elegant and green
method to prepare carbonyl compounds where otherwise a two-
step sequence of oxidation and reduction reactions would be
required (Scheme 1), potentially using toxic and/or harsh
reagents.1–4 Numerous complexes of Fe, Os, Ru, Rh, Co, Ni, Mo,
Ir, and Pt have been shown to catalyse this rearrangement;
however, most of them show restricted scope with regard to
reaction conditions and substitution at the 1- and 3-positions
(R1 and R2).3–5 Primary allylic alcohols are typically the most
challenging substrates.
Various Ru complexes have been reported to date in the
literature for this transformation.1,6–8 The rst RuII–cyclo-
pentadienyl-like complex employed in this reaction was
[RuCl(Cp)(PPh3)2] by Kulawiec and Trost in the 1990s.9 This
complex, in combination with [Et3NH][PF6] in dioxane at
100 ꢀC, is able to catalyse the isomerisation of allylic alcohols,
with the exception of primary alcohols and unsubstituted
vinyl groups. Slugovc et al. improved the activity, reaching
Scheme 1
high conversions at 57 ꢀC using a well-dened cationic
complex of the form [Ru(Cp)(MeCN)2(PR3)][PF6] in CDCl3;
notably this system did not require the use of base.10
Another relevant step forward was shown by Cadierno
and Gimeno using a series of RuIV complexes such as
[Ru(h3:h2:h3-C12H18)Cl2] and derivatives, which show high
activity in aqueous media and in ionic liquids at 75 ꢀC.11,12
However, all of these complexes display activity with only a
limited substitution pattern and still require high tempera-
¨
aEaStCHEM, School of Chemistry, University of St. Andrews, Purdie Building, North
tures to perform the isomerisation. In 2005, Backvall reported
Haugh, St. Andrews, Fife, KY16 9ST, UK. E-mail: snolan@st-andrews.ac.uk
the use of [RuCl(1-methyl-2,3,4,5-tetraphenylcyclopentadie-
nyl)(CO)2], which is able to isomerise substituted allylic
alcohols at room temperature. However, this catalyst shows
activity at relatively high loadings (5 mol%), which increases
the cost of the chemistry and can lead to purication issues.13
In the same year, Ikariya introduced a series of Ru(Cp*)(P–N)
complexes that were very active at 30 ꢀC with 1 mol% of
catalyst.14
b
´
`
´
Institut de Quımica Computacional i Catalisi, Departament de Quımica, University of
Girona, Campus de Montilivi sn, 17071, Girona, Catalonia, Spain
cDepartment of Chemistry and Biology, University of Salerno, Via Ponte don Melillo,
Fisciano I-84084, Italy
dKAUST Catalyst Center, Physical Sciences and Engineering Division, King Abdullah
University of Science and Technology, Thuwal 23955-6900, Saudi Arabia
† Electronic supplementary information (ESI) available: Experimental details and
characterisation data. See DOI: 10.1039/c3sc52612g
180 | Chem. Sci., 2014, 5, 180–188
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
Scheme 3
isomerisation. Understanding the effects of catalyst structure is
of critical importance when designing the next generation of
complexes.
Ruthenium is an attractive metal for these reactions, due to
its excellent activity, oen at low loadings, and its relatively low
cost when compared to many other late transition metals, such
as iridium, that can also perform the reaction. We decided to
evaluate the activity of [RuCl(h5-3-phenylindenyl)(PPh3)] (1),
recently discovered in our laboratory, in allylic alcohol isomer-
isation.15 This complex has emerged as a highly active multi-
tasking complex (Scheme 2); such multi-purpose complexes are
highly valuable, as they minimise the number of precious metal
catalysts that must be held in stock in synthetic chemistry
laboratories. Complex 1 has already shown activity in the race-
misation of chiral alcohols,15 the hydrogenation of ketones,
aldehydes and imines,16 the chemoselective oxidation of
secondary alcohols,17 the formation of S–S, S–Si and S–B
bonds,18 and the isomerisation of alkenes19 (Scheme 2). More-
over, the straightforward synthesis of the complex starting from
RuCl2(PPh3)3 and propargylic alcohol (Scheme 3) makes it a
practical and cost effective complex.15 Importantly, the study of
this reaction allows the reactivity of 1 to be further explored,
yielding insights into the reactivity of complexes of this type.
Simple phenyl substitution of the indenyl ligand has previously
been shown to affect reactivity in this type of complex, so we
wished to examine whether this extended to the allylic alcohol
Optimisation, scope and limitations
Using the isomerisation of (E)-1,3-diphenylpropen-1-ol 2 to the
ketone 3 as a benchmark reaction, complex 1 was tested using the
same reaction conditions that were used for alcohol race-
misation. Substrate 2 is typically one of the more challenging
allylic alcohols.1–3 Pleasingly, it was discovered that 1 is very active
at room temperature, achieving total isomerisation in just 1 h
with only 0.25 mol% of catalyst and 4 mol% base (Scheme 4).
While the use of a base has an impact on the atom economy of
the process, only a very low loading is required here, presumably
to generate the active ruthenium species (vide infra).
To evaluate the effect of the phenylindenyl ligand on reac-
tivity, complex 1 was compared with analogues [RuCl(h5-
indenyl)(PPh3)2] (4) and [RuCl(Cp)(PPh3)2] (5), which have also
been employed for these types of reactions, under the same
reaction conditions (Table 1). Surprisingly, complexes 4 and 5
Scheme 4
Table 1 Catalyst comparisona
Entry
[Ru]
Catalyst loading
Time (h)
Conv. %b
1
2
3
1
4
5
0.25 mol%
0.25 mol%
0.25 mol%
1
1
1
>95
<5
<5
a
Conditions: substrate (0.5 mmol), base (0.02 mmol), catalyst (1.25
mmol), and toluene (1 mL) added. Reactions stirred at ambient
b
temperature for 1 h. Conversion determined from 1H NMR analysis
average of at least two experiments.
Scheme 2
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 180–188 | 181

View Article Online
Chemical Science
Edge Article
Table 3 Scope of the allylic alcohol isomerisation catalysed by 1a
show practically no activity under these conditions, meaning
that the phenyl ring in complex 1 makes a signicant difference
in activity. This result was intriguing, and stimulated us to
investigate the reaction in more detail (vide infra).
Entry Substrate
1 (mol%) t (h) Conv. (yield)b,c
The reaction conditions that were employed initially are
already highly effective, and therefore further optimisation
would be of little use. Instead, the use of milder conditions and
reagents was explored, in order to render this chemistry acces-
sible to as wide a range of synthetic chemists.
1
0.25
0.25
0.25
0.25
1
1
1
1
>99% (97%)
>99% (95%)
>99% (97%)
90% (89%)
A number of bases were screened for use in this reaction;
while these conditions allow for rapid and excellent conversions
at room temperature, the use of a strong base such as NaOtBu
could potentially be incompatible with some substrates, and
therefore milder bases are attractive. The results (Table 2) show
that the choice of counterion (K or Na) makes no difference to
the reaction, while milder bases such as KOH can still allow for
complete conversion. Much weaker bases, such as K2CO3, are
not suitable for this transformation. These results suggest that
there is a pKa threshold, below which the base is ineffective.
The system is compatible with some of the most common
industrially accepted solvents,20 reaching quantitative (or close
to quantitative) conversion in 1 h reaction time in each case.
While toluene is still acceptable on a large scale, more
sustainable solvents such as cyclopentyl methyl ether, 2-methy-
ltetrahydrofuran (2-MeTHF) and isobutylmethyl ketone are also
compatible with this chemistry.20 This renders this trans-
formation scalable, and further enhances its green credentials.
To understand the substrate compatibility of the catalytic
system, systematic variation of the substrates has been
considered (Table 3). It is important to note that optimisation of
the reaction conditions has taken place using one of the more
challenging substrates, however “easier” substrates, such as un-
substituted secondary allylic alcohols were evaluated.
2
3
4
5
6
0.5
0.5
1.5
>99% (81%)
>99% (94%)
2
7
8
0.5
6
>99% (96%)
96% (92%)
Complex
1 shows high compatibility with differently
substituted substrates (Table 3). 1 allows isomerisation of
internal and terminal benzyl allylic alcohols in only 1 h with
0.25 mol% of catalyst. Electronic effects were found to play an
important role in the isomerisation. Indeed, substrates bearing
electron-donating groups (e.g. 12) are more difficult to isomer-
ise than the corresponding substrates bearing electron-with-
drawing groups (e.g. 13). This suggests that the pKa of the
alcohol or the proton a- to the alcohol may be a key determinant
of reaction efficiency; electron-withdrawing groups would sta-
bilise a developing negative charge on the benzylic carbon and
0.25
2.5
9
0.25
1
>99% (92%)
a
Conditions: substrate (1 mmol), base (4 mol%) and toluene (1 mL)
b
Table 2 Screening of basesa
added. Reactions stirred at ambient temperature for 1 h. Conversion
determined from 1H NMR analysis. Yields are isolated yields.
c
Entry
Base
Conversion %b
1
2
3
4
5
NaOtBu
KOtBu
KHMDS
KOH
>99
>99
96
>99
31
therefore might be responsible for the observed difference in
reactivity.
With alkyl allylic alcohols it was necessary to increase the
catalyst loading to 0.5 mol% to obtain the same conversions.
With regard to the degree of substitution on the vinyl group, it is
easy to isomerise monosubstituted alkenes and 1,1- or
1,2-disubstituted alkenes; trisubstituted alkenes cannot be
K2CO3
a
Conditions: substrate (0.5 mmol), base (0.02 mmol), 1 (1.25 mmol) and
toluene (1 mL) added. Reactions stirred at ambient temperature for 1 h.
b
Conversion determined from 1H NMR analysis; average of at least two
experiments.
182 | Chem. Sci., 2014, 5, 180–188
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
isomerised using the system considered here, presumably due
to the degree of hindrance around the double bond making co-
ordination of the vinyl group to the metal difficult. However,
this methodology can potentially be used as a selective system
for internal and 1,1- and 1,2-disubstituted allylic alcohols,
which should render it of interest to, for example, natural
product chemists working on highly functionalised scaffolds.
To our disappointment, primary allylic alcohols cannot be iso-
merised using 1.
The scope and limitations have been explored at room
temperature, and represent the most accessible conditions for
synthesis. However, the performance of the system was evalu-
ated at lower catalyst loadings and higher temperatures in order
to probe the limits of the catalyst. 1 can catalyse the reaction of
2, selectively producing 3 (77% conversion) at 100 ꢀC with a
catalyst loading of 100 ppm in 15 minutes. This represents a
turnover number (TON) of 10 000, and a turnover frequency
(TOF) of 36 000 hꢁ1, which are highly competitive with the state-
of-the-art. Notably, this is in fact a conservative estimate of
TOF when compared to alternative measurements at low
conversion, as discussed by Kozuch and Martin.21 This result is
also made more remarkable by the use of 2 in this TON/TOF
determination, which is typically a more challenging substrate
for ruthenium complexes.1–3
Scheme 5
Table 4 Testing cationic complex 14 and hydride complex 15 versus
complex 1 in the isomerisation of 2 with and without basea
Catalyst loading
(mol%)
Entry
Complex
Base (mol%)
Conv. %b
1
2
3
4
5
6
1
0.25
0.25
0.5
0.5
0.5
4
4
4
0
0
0
>99
>99
0
0
>99
0
14
15
1
14
15
0.5
The role of additives
a
Conditions: substrate (0.5 mmol), base (0.02 mmol) and catalyst (125
mmol) and toluene (1 mL) added. Reactions stirred at ambient
In order to gain a better understanding of the action of complex
1 in this reaction, the effect of the base was investigated. In fact,
the role of the base was rather unclear, whether it acts as
chloride abstractor to generate the active species or promotes
the activation of the allylic alcohol through deprotonation of the
hydroxyl moiety. A series of exploratory reactions were con-
ducted, where the concentration of base and [nBu4N][Cl] (TBACl,
added as a chloride source) were varied.
As shown in Fig. 1, there is clearly an optimum loading of
base (4 mol%); the base is necessary for the reaction to proceed,
yet at high loadings the substrate completely decomposes in the
reaction medium. A more subtle effect was observed with
[nBu4N][Cl]: a slight inhibition of the reaction is observed, with
only 50% conversion when 1 equiv. TBACl was present. The
results suggest that either chloride abstraction is involved in the
temperature for 1 h. Conversion determined from 1H NMR; average
b
of at least two experiments.
initiation of the catalyst, or that one of the catalytic interme-
diates is sensitive to chloride. In order to have a clearer
understanding as to the need to remove the chloride in order to
activate the catalyst, two further complexes were prepared:
cationic 14 and hydride 15. The former complex has been
shown to be active in the isomerisation of terminal alkenes
(Scheme 5).19
As shown in Table 4, complexes 1 and 14 are both active in
the presence of base, while complex 15 is inactive. In the
absence of base, 14 remains active, albeit at a slightly higher
catalyst loading (0.5 mol%), conrming the necessity of the
chloride abstraction in order to generate the active species.
Mechanistic study
A detailed understanding of the reaction mechanism was
desired, in order to rationally design the next generation of
catalysts and fully understand the interplay of structure and
reactivity.
Three pathways have been suggested in the literature for this
class of transformations (Scheme 6).9,12,22 In each case the mech-
anism leads to an enol, which will tautomerise to the thermody-
namically-favoured ketone. Trost has shown that complex 4 does
not operate via a metal-hydride addition/elimination sequence,
via deuterium-labelling crossover experiments;9 attention was also
Fig. 1 Isomerisation of 2 in the presence of various equivalents of
NaOtBu (red points) and TBACl (blue points). Reaction conditions were
otherwise as per Scheme 4.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 180–188 | 183

View Article Online
Chemical Science
Edge Article
Scheme 6
drawn to the signicant improvement in activity when the cyclo- the same conditions as for the synthetic experiments
pentadienyl ligand was replaced with an indenyl moiety. A (Scheme 7(a)). The reaction showed only the 1,3-deuterium
particularly detailed study was presented by Cadierno et al., in shi, suggesting that the reversible metal hydride addition/
which DFT calculations (B3LYP/6-311+G(d,p)//B3LYP/6-31G(d,p)) elimination shown in Scheme 6(a) does not occur; if it did, both
were used to compare reaction mechanisms for the isomerisation 1,2- and 1,3-deuterium shis would be expected.
reaction using a Ru(IV) complex, favouring a p3-oxo-allyl mecha-
To further probe the potential presence of metal hydrides, a
nism.12 McGrath et al. used a simple Ru(II) system for allylic crossover experiment was performed (Scheme 7(b)) in which 2-d
alcohol isomerisation, and suggested a metal-hydride addition/ and 12 were exposed to the same charge of 1 (0.25 mol%, based
elimination mechanism on the basis of deuterium labelling and on the sum of substrates) for 24 h. There was no transfer of
crossover experiments.22
deuterium between compounds, ruling out the presence of an
In the alkyl mechanism (Scheme 6(a)), in which a metal unbound metal hydride (or deuteride) in the reaction. These
hydride complex is the active species, the reaction proceeds via experimental results permit the alkyl mechanism to be ruled
reversible addition of metal hydride across the alkene moiety, out (Scheme 6(a)).
followed by reversible b-hydride elimination to effectively move
the double bond down the chain. The net result is a 1,3-
hydrogen shi, but the reversible nature of the reaction means
Hydroxyl effect
In an attempt to discriminate between a mechanism involving a
simple h3-allyl intermediate (Scheme 6(b)) and one in which the
oxygen contributes to the reaction to form an h3-oxo-allyl
intermediate (Scheme 6(c)), the isomerisation was performed
using O-methylated allylic alcohol 16 (Scheme 8). If the reaction
involves only an h3-allyl intermediate, the isomerisation of 16
should yield enol ether 17, which is thermodynamically fav-
oured over the substrate by ca. 1.7 kcal molꢁ1 (see ESI† for
that a net 1,2-hydrogen shi may also result. This type of
mechanism has been shown to occur in metathesis reactions,
where ruthenium hydrides are generated via decomposition of
the active metathesis catalyst.23,24 Alternatively, a h3-allyl
mechanism might be in operation (Scheme 6(b)), which
proceeds via abstraction of an allylic proton by a metal which is
bound to the alkene via h2-coordination. The hydride is then
deposited at the terminus of the allyl complex, to yield a new
h2-complex. The h3-oxo-allyl mechanism (Scheme 6(c)) is
similar, but takes advantage of the presence of an alcohol in the
allylic position. The alcohol co-ordinates to the metal, which is
followed by h2-complexation of the alkene, abstraction of the
allylic proton, and generation of a new h3-allyl complex by
depositing the hydride on the terminus.
A number of experiments were therefore conducted to probe
the mechanism of this transformation.
Isotopic labelling studies
Selectively-labelled substrate 2-d was prepared by reduction of
the corresponding ketone using NaBD4 in THF, according to the
literature procedure.16 Two reactions were conducted with this
substrate. First, the allylic isomerisation was conducted using
Scheme 7
184 | Chem. Sci., 2014, 5, 180–188
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
Scheme 8
computational details). However, the reaction showed no
conversion, even aer 24 h, while the corresponding alcohol
shows complete conversion in only 1 h. Assuming that the limit
of detection of 1H NMR spectroscopy is at worst ca. 5%, this
corresponds to a greater than twenty-fold difference in turnover
number, and a ca. 500-fold difference in turnover frequency.
While this is a negative proof, rather than a positive proof, this
result suggests a role for the alcohol function, lending some
support to the h3-oxo-allyl mechanism (Scheme 6(c)).
Density functional theory calculations
In order to tie experimental observations together, thus
understanding fully the mechanism and setting the scene for
further catalyst development, we embarked on density func-
tional theory studies of the reaction mechanism (at the M06//
TZVP//BP86/SVP level of theory).
Potential energy surfaces (PES) were evaluated for the iso-
merisation of three model substrates using 1: allyl alcohol,
but-1-ene-3-ol, and a-vinyl benzyl alcohol. In each case, the h3-
oxo-allyl mechanism in Scheme 6(c) was found to be the most
favourable. In addition the isomerisation of but-1-ene-3-ol was
explored with [RuCl(indenyl)(PPh3)2] 4 and [RuCl(Cp)(PPh3)2] 5.
All PESs are normalised to pre-catalyst plus substrate plus base,
which is termed intermediate A for 1, A0 for 4, and A00 for 5.
Pre-catalyst 1 undergoes a series of steps to form the active
Scheme 9
alcohols is not a consequence of high kinetic barriers. The
energetic spans of the catalytic cycles can be calculated by
subtracting the energy of G (the turnover-determining inter-
mediate, or TDI) from that of LF‡ (the turnover-determining
transition state, or TDTS) according to the work of Kozuch and
Shaik (eqn (1)).25
dE ¼ TTDTS ꢁ ITDI
(1)
Values of 28.0, 31.5 and 28.9 kcal molꢁ1 can be obtained for
species, with the energetics somewhat dependent on the allyl alcohol, but-1-ene-3-ol and a-vinyl benzyl alcohol, respec-
substrate (Scheme 9). Phosphine dissociation is relatively facile tively. Therefore, it would be expected that primary alcohol
(16.8 kcal molꢁ1), and is followed by coordination of the isomerisation should be faster than that of the secondary
substrate and deprotonation by the base. Chloride abstraction alcohols (see the following section for further investigations
leads to intermediate F, and thus into the catalytic cycle. into this discrepancy).
Calculations on alternative mechanisms showed that the base is
With calculated intermediates and transition states in hand,
necessary at this point to remove the HCl produced by initiation we examined the effect of catalyst structure on reactivity
of the pre-catalyst. In addition, removal of the second phos- (Schemes 11 and 12). The energetics of the initiation sequence
phine (from C) leads to much higher energy intermediates are quite different. Small differences in phosphine dissociation
(Grel ¼ 25–35 kcal molꢁ1) on the PES of the catalytic cycle itself, and substrate binding energies were calculated. Surprisingly,
so this possibility was discounted.
even though the phosphine dissociation is favoured for 1, the
The reaction proceeds via abstraction of the second allylic deprotonation and chloride abstraction steps exhibited the
proton to yield the enone h2-coordinated to a ruthenium highest barriers for complex 1, despite the experimental
hydride species (G) in a potential energy well (Scheme 10). observations that 1 is a more efficient catalyst than 4 or 5.
Rotation of this h2-ligand has a considerable (20 kcal molꢁ1
)
The calculated energetics of the catalytic cycle (Scheme 12)
barrier via GH‡, yet presents the alkene terminus for delivery of show that, in general, the intermediates on the reaction pathway
the hydride via a rather facile process, with H to HI‡ being only catalysed by the cyclopentadienyl analogue 4 are much lower in
2–3 kcal molꢁ1. The complex then rearranges to bind the enol energy, although the overall energetic span (dE) is similar (31.5,
product via the oxygen, setting the scene for a series of steps to 27.3 and 28.7 kcal molꢁ1 for 1, 4 and 5, respectively). However, the
liberate the enol product and co-ordinate a subsequent mole- barriers to each of the individual steps for 1 are typically smaller,
‡
cule of substrate. Notably, the energetics of the isomerisation of compared to 4 and 5, e.g. J to JK‡ versus J0 to JK0 . Alternatively,
primary and secondary allylic alcohols are rather similar, sug- decomposition of 1 (or intermediates derived from 1) may be
gesting that the inability of 1 to isomerise primary allylic slower under the reaction conditions than for 4 and 5.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 180–188 | 185

View Article Online
Chemical Science
Edge Article
Scheme 10
Scheme 11
that this transformation should be both kinetically and ther-
modynamically feasible. DG values for the overall reactions were
calculated to be 6.4, 7.1 and 7.2 kcal molꢁ1 for allyl alcohol,
but-3-en-1-ol, and a-vinyl benzyl alcohol, respectively.
Isomerisation of primary allylic alcohols
One of the aims of the DFT study was to understand why
primary allylic alcohols were not successfully converted to the
corresponding aldehydes. Surprisingly, calculations indicated
186 | Chem. Sci., 2014, 5, 180–188
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
Table 5 Isomerisation of model primary allylic alcohols using 14a
Entry
1
Substrate
[14] (mol%)
0.25
Time (h)
1.5
Conv.b
>95%
2
3
4
5
6
1.5
17.5
43.5
16
<5%
24%
34%
>95%
33%
0.25
1
1.5
0.25
7
6
94%
8
9
10
0.25
0.25
0.25
1.5
6
12
<5%
<5%
<5%
a
Substrate solution (0.6 mL of a 0.5 mol Lꢁ1 solution in toluene-d8)
added to 14 in an J. Young type NMR tube and mixed at room
b
temperature. Conversion determined from 1H NMR; average of at
least two experiments.
Scheme 12
detected within 12 h at rt (Entries 8–10). The catalyst system is
therefore clearly very sensitive to substitution pattern,
particularly g- to the alcohol, which is in accordance with the
reactivity of 1.
In an attempt to reconcile experiments and theory, it was
hypothesised that either the base present is consumed in a side-
reaction, or that the presence of the base itself might promote
side reactions with one of the ruthenium intermediates. The
second alternative seems to be most likely, as even in the
presence of ca. 16 equivalents of base, no reaction occurs. This
suggested that a base-free protocol should be developed to
catalyse isomerisation of primary allylic alcohols. In addition,
the use of a base has implications for the overall atom economy
of a process; simple addition of a catalyst to a substrate is both
more convenient and atom-efficient than protocols requiring
the use of extra reagents such as bases.
Table 6 Isomerisation of primary allylic alcohols using 14a
Entry
Substrate
[14] (mol%)
Time (h)
Conv.b
>95%
1
1
4
This was explored using complex 14, which is already
capable of isomerising secondary allylic alcohols. In addition,
DFT calculations were performed to explain the inactivity of 15
in this transformation. These calculations suggested that
phosphine dissociation from 15 presents a considerable barrier
(30.4 kcal molꢁ1), yielding a series of intermediates that were
much higher in energy than those in the schemes presented
above (see the ESI† for details).
2
3
1
4
<5%
1
1
4
7
70%
>95%
Initial experiments were conducted at room temperature
with four model substrates (in NMR tubes) (Table 5), and
showed the ability of 14 to isomerise primary allylic alcohols.
Allyl alcohol was isomerised rapidly, yielding complete
conversion to propanal within 1.5 h with only 0.25 mol% of
catalyst (Entry 1). The addition of steric hindrance at the
g-position (crotyl alcohol) resulted in a signicant decrease in
reactivity, with only very modest conversion, even aer
extended periods of time (Entries 2–4). An increase of catalyst
loading to 1 mol% furnished complete conversion overnight
(Entry 5). Methylation at the alternative position had a less
dramatic effect on reactivity, but several hours were needed to
allow complete conversion (Entries 6 and 7). g,g-Dimethyla-
tion effectively shuts down the reaction, with no reaction
4
5
1
1
7
4
>95%
>95%
1
1
4
7
>28%
32%
6
a
Conditions: substrate (1 mmol), 14 and toluene (1 mL) added.
b
Reactions were stirred at ambient temperature for 1 h. Conversion
determined from 1H NMR; average of at least two experiments.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 180–188 | 187

View Article Online
Chemical Science
Edge Article
A number of cinnamyl alcohols were tested in the reaction,
to explore whether these conditions using 14 were synthetically
useful (Table 6). A catalyst loading of 1 mol% was employed, in
light of the challenging nature of primary allylic alcohol
substrates. While cinnamyl alcohol could be isomerised in a
straightforward manner, additional steric hindrance precluded
2 S.-I. Murahashi, Ruthenium in Organic Synthesis, Wiley-VCH
Verlag GmbH & Co. KGaA, 2005.
3 R. C. van der Dri, E. Bouwman and E. Drent, J. Organomet.
Chem., 2002, 650, 1–24.
4 S. Bellemin-Laponnaz and J.-P. Le Ny, C. R. Chim., 2002, 5,
217–224.
´
reaction (Entries
1
and 2). Electron-withdrawing groups
5 L. Mantilli, D. Gerard, C. Besnard and C. Mazet, Eur. J. Inorg.
rendered the reaction more challenging, although this could be
mitigated by increasing the reaction time (Entries 3–5). Inter-
estingly 14 shows reduced reactivity in the isomerisation of 27,
probably due to the combination of steric hindrance and
coordinating effect of the NMe2 on the aryl ring.
Catalysts based on the (3-phenylindenyl)bis(triphenyl phos-
phine)ruthenium motif are therefore capable of isomerising
primary alcohols, when a base is not present. Here, this has
been achieved via use of a cationic ruthenium species.
Chem., 2012, 2012, 3320–3330.
6 A. Azua, S. Sanz and E. Peris, Organometallics, 2010, 29, 3661–
3664.
7 A. P. da Costa, J. A. Mata, B. Royo and E. Peris,
Organometallics, 2010, 29, 1832–1838.
´
´ ´ ´
8 R. Garcıa-Alvarez, F. J. Suarez, J. Dıez, P. Crochet,
˜
´
´
V. Cadierno, A. Antinolo, R. Fernandez-Galan and
F. Carrillo-Hermosilla, Organometallics, 2012, 31, 8301–8311.
9 B. M. Trost and R. J. Kulawiec, J. Am. Chem. Soc., 1993, 115,
2027–2036.
¨
10 C. Slugovc, E. Ruba, R. Schmid and K. Kirchner,
Conclusions
Organometallics, 1999, 18, 4230–4233.
11 V. Cadierno, S. E. Garcia-Garrido and J. Gimeno, Chem.
Commun., 2004, 232–233.
Complex 1 has been evaluated in the isomerisation of allylic
alcohols, showing high activity and broad compatibility with
substituted allylic compounds. Indeed, 1 can be considered one
of the most active ruthenium complexes reported in the litera-
ture for this reaction, requiring only 0.25–0.5 mol% ruthenium
and short reaction times at room temperature. The potential
energy surface for this reaction has been explored, and
compared to that of the analogous reactions with similar cata-
lysts that show far less activity.
´
12 V. Cadierno, S. E. Garcıa-Garrido, J. Gimeno, A. Varela-
´
Alvarez and J. A. Sordo, J. Am. Chem. Soc., 2006, 128, 1360–
1370.
´
´
13 B. Martın-Matute, K. Bogar, M. Edin, F. B. Kaynak and
¨
J.-E. Backvall, Chem.–Eur. J., 2005, 11, 5832–5842.
14 M. Ito, S. Kitahara and T. Ikariya, J. Am. Chem. Soc., 2005,
127, 6172–6173.
15 S. Manzini, C. A. Urbina-Blanco, A. Poater, A. M. Z. Slawin,
L. Cavallo and S. P. Nolan, Angew. Chem., Int. Ed., 2012, 51,
1042–1045.
16 S. Manzini, C. A. Urbina-Blanco and S. P. Nolan, Adv. Synth.
Catal., 2012, 354, 3036–3044.
In addition, it was shown that the use of cationic species 14
can remove the need for added base, and widen the scope of
the reaction to include challenging primary allylic alcohol
substrates.
Further investigations into the applications of 1, the
synthesis and study of a systematic series of variants, and their
further application are currently underway in our laboratories.
17 S. Manzini, C. A. Urbina-Blanco and S. P. Nolan,
Organometallics, 2013, 32, 660–664.
´
18 J. A. Fernandez-Salas, S. Manzini and S. P. Nolan, Chem.
Commun., 2013, 49, 5829–5831.
19 S. Manzini, D. J. Nelson and S. P. Nolan, ChemCatChem,
2013, 5, 2848–2851.
20 R. K. Henderson, C. Jimenez-Gonzalez, D. J. C. Constable,
S. R. Alston, G. G. A. Inglis, G. Fisher, J. Sherwood,
S. P. Binks and A. D. Curzons, Green Chem., 2011, 13, 854–
862.
Acknowledgements
We thank the ERC (Advanced Investigator Award ‘FUNCAT’ to
SPN) and the EPSRC for funding. Umicore is thanked for gis of
materials. SPN is a Royal Society Wolfson Merit Award holder.
Melanja Smith and Dr Tomas Lebl are thanked for assistance
´
with NMR spectroscopy facilities. Dr Cesar A. Urbina-Blanco is
21 S. Kozuch and J. M. L. Martin, ACS Catal., 2012, 2, 2787–
2794.
22 D. V. McGrath and R. H. Grubbs, Organometallics, 1994, 13,
224–235.
acknowledged for useful discussions. LC thanks the HPC team
of Enea for using the ENEA-GRID and the HPC facilities
CRESCO in Portici (Italy) for access to remarkable computa-
´
tional resources. AP thanks the Spanish MICINN for a Ramon y
23 I. W. Ashworth, I. H. Hillier, D. J. Nelson, J. M. Percy and
M. A. Vincent, Eur. J. Org. Chem., 2012, 2012, 5673–
5677.
Cajal contract (RYC-2009-05226) and European Commission for
a Career Integration Grant (CIG09-GA-2011-293900).
24 F. C. Courchay, J. C. Sworen, I. Ghiviriga, K. A. Abboud and
K. B. Wagener, Organometallics, 2006, 25, 6074–6086.
25 S. Kozuch and S. Shaik, Acc. Chem. Res., 2010, 44, 101–110.
Notes and references
´
´
1 R. Uma, C. Crevisy and R. Gree, Chem. Rev., 2002, 103, 27–52.
188 | Chem. Sci., 2014, 5, 180–188
This journal is © The Royal Society of Chemistry 2014