Use of the tetrahydroborate ligand as gate-keeper and protected hydride ligand: Preparation and study of alkyl hydride and acyl hydride complexes of ruthenium(II)

Article abstract of DOI:10.1039/b600036c

Complex 3, [Ru(η²-BH₄)(CO)(Et)L₂] (L = PMe₂Ph) can be converted by nucleophiles L′ {a, PMe ₂Ph; b, P(OMe)₃; c, Me₃CNC; d, CO} to alkyl and acyl complexes [Ru(η¹-BH₄)(CO)(Et)L ₂L′] (4a), [Ru(η²-BH₄)(COEt)L ₂L′] (5a-d), and [Ru(η¹-BH₄)(COEt) L₂L′₂] (7d and isomers 7c and 10c). Deprotection can then be achieved under conditions mild enough to allow study of the resulting alkyl hydride complexes [Ru(CO)(Et)HL₂L′] (1a, 1b) and acyl hydride complexes [Ru(COEt)HL₂L′₂] (8c, 8d) prior to elimination of ethane and propanal respectively, with formation of ruthenium(0) complexes [Ru(CO)L₂L′₂] (6a, 6b, 6d). With Me ₃CNC, however, the final product is (depending on the solvent used) [Ru(CNCMe₃)₂{C(H)NCMe₃}(COEt)L₂] (9c) or [Ru(CNCMe₃)₃(COEt)L₂]⁺ (11c). Successive treatment of [Ru(η²-BH₄)(CO)HL ₂], 2, with ethene and then CO yields propanal, but turning this into a catalytic cycle is hindered by the greater readiness of 5d to yield propanal non-catalytically (reacting with CO) than catalytically (reacting with H ₂). The Royal Society of Chemistry 2006.

Full text of DOI:10.1039/b600036c

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/dalton | Dalton Transactions
Use of the tetrahydroborate ligand as “gate-keeper” and protected
hydride ligand: preparation and study of alkyl hydride and acyl
hydride complexes of ruthenium(II)†
Simon B. Duckett,* John P. Lowe and Roger J. Mawby
Received 4th January 2006, Accepted 13th February 2006
First published as an Advance Article on the web 6th March 2006
DOI: 10.1039/b600036c
2
Complex 3, [Ru(g -BH4)(CO)(Et)L2] (L = PMe2Ph) can be converted by nucleophiles Lꢀ {a,
1
PMe2Ph; b, P(OMe)3; c, Me3CNC; d, CO} to alkyl and acyl complexes [Ru(g -BH4)(CO)(Et)L2Lꢀ]
2
1
(4a), [Ru(g -BH4)(COEt)L2Lꢀ] (5a–d), and [Ru(g -BH4)(COEt)L2Lꢀ2] (7d and isomers 7c and 10c).
Deprotection can then be achieved under conditions mild enough to allow study of the resulting alkyl
hydride complexes [Ru(CO)(Et)HL2Lꢀ] (1a, 1b) and acyl hydride complexes [Ru(COEt)HL2Lꢀ2] (8c, 8d)
prior to elimination of ethane and propanal respectively, with formation of ruthenium(0) complexes
[Ru(CO)L2Lꢀ2] (6a, 6b, 6d). With Me3CNC, however, the final product is (depending on the solvent
used) [Ru(CNCMe3)2{C(H)NCMe3}(COEt)L2] (9c) or [Ru(CNCMe3)3(COEt)L2]+ (11c). Successive
2
treatment of [Ru(g -BH4)(CO)HL2], 2, with ethene and then CO yields propanal, but turning this into a
catalytic cycle is hindered by the greater readiness of 5d to yield propanal non-catalytically (reacting
with CO) than catalytically (reacting with H2).
complexes could play an important role in catalysis, since the
Introduction
vacant coordination site could allow attachment of a substrate to
the metal. Additionally, a site which became vacant at any stage in
a catalytic cycle could be protected by the reclosing of the M · · · HB
bridge. Marks and Kolb described the tetrahydroborate ligand
as “acting as a gate-keeper to the coordinative saturation of the
metal”.
Complexes containing a tetrahydroborate ligand which is attached
to the metal through two bridging hydrogen atoms are known for
most of the transition metals.1–12 With only a few exceptions,10,13
NMR studies of these complexes have provided evidence of
fluxional motion involving the tetrahydroborate ligand. Although
there are instances in which this motion involves (or includes) the
exchange of hydrogen atoms between the tetrahydroborate ligand
and other ligands,1,7,9,14 in the majority of cases the exchange is
simply between bridging and terminal hydrogen atoms within the
tetrahydroborate ligand. The mechanisms commonly proposed
for this exchange involve either the formation and subsequent
2
Since then there have been many reports of the use of g -
tetrahydroborate complexes in catalytic work, including instances
of the hydrogenation of octene16 and the asymmetric hydro-
genation of a-acetamidocinnamic acid,17 the polymerisation of
ethene,2 the oligomerisation of butadiene18 and the trimerisation
of methyl propynoate,19 hydrogen transfer from propan-2-ol to
ketones,9 the hydroboration of ethene20 and the hydrogenolysis
of benzothiophene to give 2-ethylthiophenol.21 In cases where
attempts have been made to establish a mechanism for the catalysis,
however, it is evident that the true active species very often does
not contain a tetrahydroborate ligand. Most commonly BH3 has
been abstracted from the ligand, often by a base which is added to
“activate” the catalyst or to increase its activity,16,17,21 sometimes
by one of the reactants,9 and sometimes by one of the other ligands
in the complex.18 Although such an abstraction from a complex
2
3
2
breaking of a third M · · · HB bridge (g → g → g ) or the
2
reversible breaking of one of the two existing bridges (g →
1
2
g → g ). Of the instances where the former mechanism has been
proposed, the most plausible cases deal with complexes which also
5
contain g -cyclopentadienyl ligands, and here it is suggested that
the formation of a third bridge is accompanied by a temporary
5
3
change from g - to g -bonding of a cyclopentadienyl ligand.3,4
The most generally accepted mechanism, however, is that which
involves the initial breaking of an M · · · HB bridge, and molecular
orbital calculations have indicated that this is the preferred
pathway for exchange of bridging and terminal hydrogens in the
2
containing an g -bonded tetrahydroborate ligand generates a
vacant site which can clearly be useful for catalytic purposes, the
“gate-keeper” effect of the ligand15 has been lost. One conclusion
which seems to emerge from these studies is that a tetrahydroborate
complex is unlikely to be the true active species in a catalytic
cycle if the reaction mixture contains any nucleophile L capable of
abstracting BH3 as the adduct H3B·L. This obviously includes any
cycle involving CO (an example being alkene hydroformylation),
since one must anticipate the abstraction of BH3 as H3B·CO.
In a recent paper22 we described a study of the kinetics
and mechanism of the reductive elimination of ethane from
[Ru(CO)(Et)H(PMe2Ph)3], 1a. Complex 1a can be conveniently
2
tetrahydroborate ligand in [Os(g -BH4)H3{P(CHMe2)3}2].14
In their early review of tetrahydroborate complexes, Marks and
Kolb15 noted that the apparent ease with which an M · · · HB
bridge could be broken meant that a vacant coordination site
could be created on the metal with very little expenditure of
energy, and suggested that this meant that tetrahydroborate
Department of Chemistry, University of York, Heslington, York, UK YO10
5DD. E-mail: sbd3@york.ac.uk; Fax: +01904 432516; Tel: +01904 432564
† Electronic supplementary information (ESI) available: Complete listing
of the NMR data for all new complexes. See DOI: 10.1039/b600036c
This journal is
The Royal Society of Chemistry 2006
Dalton Trans., 2006, 2661–2670 | 2661
©

View Article Online
obtained in two steps. First, the tetrahydroborate complex
2
2
[Ru(g -BH4)(CO)H(PMe2Ph)2], 2,23 is converted to [Ru(g -
BH4)(CO)Et(PMe2Ph)2], 3, by treatment with ethene at or below
273 K. This reaction is of interest in its own right, since NMR
studies showed23 that it proceeds by way of the g -tetrahydroborate
complex [Ru(g -BH4)(CO)(g -C2H4)H(PMe2Ph)2], making it a
nice example of Marks and Kolb’s gate-keeper effect.15
1
1
2
Then 3 is treated with PMe2Ph, yielding 1a and H3B·PMe2Ph.
Although 1a undergoes ethane elimination at a significant rate
even at temperatures below 273 K, this second step can be carried
out under mild enough conditions to avoid elimination. Whereas
this sequence was ideal for our needs, the second step appeared
to add further weight to the conclusion that the tetrahydroborate
ligand in a complex will not survive in solution in the presence
of a free nucleophile. Again, however, lower temperature studies
showed that- under the right conditions- the gate-keeper effect
of the tetrahydroborate ligand could be used to bring about a
reaction within the coordination sphere of the metal, with no
abstraction of the BH3. The whole sequence also demonstrated
how, in a synthetic procedure, a tetrahydroborate ligand can act
as a protected hydride ligand, with easy removal of the protection
by BH3 abstraction at the appropriate point.
Scheme 1 Reactions of 3 with Lꢀ = (a) PMe2Ph, (b) P(OMe)3,
(c) Me3CNC and (d) CO. L = PMe2Ph throughout. Complexes 4b–d were
not actually detected. *Only observed in CD3C6D5 solution: see Scheme 3
for reactions in CD3COCD3 solution.
These studies led us to explore the reactions of 3 with other
nucleophiles, and ultimately with CO (with a view to exploring
the possibility that 2 might be used as a catalyst for the hydro-
formylation of ethene). In the next section, we summarise briefly
the previously reported22 reaction of 3 with PMe2Ph, and then
describe those of 3 with other nucleophiles. These provided further
examples of (i) reactions which do not involve BH3 abstraction,
and (ii) reaction sequences in which the final stage deprotects a
hydride ligand, yielding new and reactive complexes containing
mutually cis hydride and alkyl or acyl ligands.
1
yielded [Ru(g -BH4)(CO)Et(PMe2Ph)3], 4a. Here the vacant site
on the metal created by the cleavage of one Ru · · · HB bridge was
occupied by the PMe2Ph. The effect of warming the solution to
253 K was to cause combination of the ethyl and carbonyl ligands
in 4a, allowing the Ru · · · HB bridge to reform and yielding [Ru(g -
2
BH4)(COEt)(PMe2Ph)3], 5a. Although, overall, the conversion
of 3 to 5a simply corresponds to the combination of ethyl and
carbonyl ligands and occupation of the vacated site by PMe2Ph,
the sequence of events demonstrates that it is the easy breaking of
an Ru · · · HB bridge which is the key to nucleophilic access to the
metal, and the reformation of the bridge which then helps to tip
the thermodynamic balance in favour of 5a rather than 4a.
When the solution of 4a was treated with PMe2Ph prior
to the increase in temperature, BH3 abstraction to give
[Ru(CO)(Et)H(PMe2Ph)3], 1a, was found to compete with (but
not entirely suppress) the formation of 5a. The reversibility of the
combination of ethyl and carbonyl ligands was demonstrated by
the fact that 5a was itself more slowly converted to 1a under these
conditions, presumably by way of 4a.
Results and discussion
Details of key resonances in the NMR spectra of the new com-
plexes are given in this section, and a complete listing of the NMR
1
data can be found in the ESI.† The H information was derived
from fully coupled spectra, but for each complex 1H spectra were
also recorded with both full and selective 31P decoupling. In every
case, the decoupling results were consistent with the assignments
given. Individual results will not be mentioned unless they
provided other key information. The patterns of resonances in the
31P, 1H and 13C NMR spectra of the new ruthenium(II) complexes
established that all of them contained a mutually trans pair of
PMe2Ph ligands.24,25 Except where stated otherwise, it was also
evident from the spectra that the Ru–P bonds to these ligands did
not lie in a molecular plane of symmetry.26 In the case of complexes
containing either an ethyl or a propanoyl ligand, assignments
Kinetic studies of ethane elimination from 1a in CD3C6D5
solution at temperatures between 273 and 293 K showed it to
be a simple first-order process. In the presence of free PMe2Ph, the
16-electron fragment [Ru(CO)(PMe2Ph)3] was trapped, yielding
[Ru(CO)(PMe2Ph)4], 6a (see Scheme 1).
(ii) Reaction of 3 with P(OMe)3
1
were confirmed with the help of 1H{ H}, 1H–1H COSY, 1H–
2
13C HMQC or DEPT-135 experiments.27–29 For all the g - and
When a CD3C6D5 solution of 3 was treated with an equimolar
1
g -tetrahydroborate complexes, the 1H NMR spectra provided
quantity of P(OMe)3 at 195 K, complete conversion to a single
2
product [Ru(g -BH4)(COEt)(PMe2Ph)2{P(OMe)3}], 5b, occurred.
evidence for fluxional motion of the tetrahydroborate ligand, and
the changes in spectra due to this motion were fully reversible.
No other new species was detected during the reaction. The
P(OMe)3 ligand in 5b was characterised by a typically low-field (d
150.3) resonance in the 31P spectrum30 (exhibiting a triplet splitting
due to coupling to the other two 31P nuclei in the complex),
and by doublet resonances in the 1H and 13C spectra for the
methyl substituents. The presence of the propanoyl ligand was
(i) Reaction of [Ru(g2-BH4)(CO)Et(PMe2Ph)2], 3, with PMe2Ph
As shown in Scheme 1, treatment of 3 with an equimolar amount
of PMe2Ph in either CD3COCD3 or CD3C6D5 solution at 193 K
2662 | Dalton Trans., 2006, 2661–2670
This journal is
The Royal Society of Chemistry 2006
©

View Article Online
indicated by resonances at d 1.19 (CH3) and 2.85 (CH2) in the
1H spectrum and at d 12.2 (CH3), 53.2 (CH2) and 258.5 (CO) in
[Ru(CO)(Et)H(PMe2Ph)2{P(OMe)3}], with a ligand arrangement
(see Scheme 1) equivalent to that in [Ru(CO)(Et)H(PMe2Ph)3], 1a.
For comparison, the methylene carbon atom resonance for 1a was
a doublet (|2JPC| = 48.6 Hz) of triplets (|2JPC| = 10.4 Hz) at
d −3.5.22
2
the 13C spectrum {the corresponding chemical shifts for [Ru(g -
BH4)(COEt)(PMe2Ph)3], 5a, were d 1.22, 2.92, 11.5, 53.6 and 261.8
respectively22}. The occupation of the other two coordination sites
in 5b by a bidentate tetrahydroborate ligand was confirmed by
As mentioned above, at 263 K complex 1b underwent a further
reaction not much slower than that by which it was formed from
5b. In this reaction ethane (characterised by singlet resonances in
the 1H and 13C NMR spectra at d 0.80 and 7.9, respectively) was
formed, indicating that 1b, like 1a, readily underwent reductive
elimination. The solution contained free P(OMe)3, and this
trapped the 16-electron fragment [Ru(CO)(PMe2Ph)2{P(OMe)3}]
as [Ru(CO)(PMe2Ph)2{P(OMe)3}2], 6b. Although we have been
unable to isolate 6b in a pure state, we obtained good quality NMR
spectra for the complex at 263 K. From the 31P spectrum, which
consisted of triplet resonances at d 18.4 and 173.9, with a mutual
|2JPP| value of 55.8 Hz, it appeared that the two PMe2Ph ligands
were equivalent, and that the same was true of the two P(OMe)3
ligands. This was confirmed by the fact that the resonance for the
carbonyl ligand, at d 221.7 in the 13C spectrum, was a triplet of
triplets (|2JPC| = 19.7 and 37.8 Hz), and by the observation of just
one resonance for all twelve PMe2Ph methyl protons and one for all
eighteen P(OMe)3 methyl protons in the 1H NMR spectrum, and
just one for each set of methyl carbon atoms in the 13C spectrum.
In a study of crystal structures and NMR spectra of trigonal
1
the presence in the H spectrum of broad resonances at d −5.4
and −5.3 (for the two inequivalent bridging hydrogens) and at
d 4.2 (for the two equivalent terminal hydrogens). The ligand
arrangement in 5b was clearly that shown in Scheme 1, and, as
expected, one bridging hydrogen resonance (that at d −5.4) showed
a doublet splitting (|2JPH| = ca. 102 Hz) by the 31P nucleus in the
P(OMe)3 ligand, placing this hydrogen atom trans to P(OMe)3.
The fluxionality of the tetrahydroborate ligand was demonstrated
by the broadening of all three resonances as the temperature of the
solution was raised. By 263 K this broadening had caused the two
bridging hydrogen resonances to overlap with one another, but 5b
was not sufficiently stable to enable us to reach the coalescence
temperature for bridging and terminal hydrogens. Here again, the
parallels between the spectra of 5b and 5a22 are marked, although
the change in the unique phosphorus ligand has quite significant
effects on the chemical shift for the bridging hydrogen trans to it
and the magnitude of the coupling constant |2JPH|.
When a solution of 5b in CD3C6D5 was warmed from 213 to
263 K in the presence of excess P(OMe)3, abstraction of BH3 as the
adduct H3B·P(OMe)3 occurred. Although not isolated, the adduct
was easily identifiable from its NMR spectra, with characteristic
1 : 1 : 1 : 1 quartet splittings by the main boron isotopomer 11B of
2
bipyramidal complexes [Ru(CO)2(g -alkene)LLꢀ], where L and Lꢀ
were both phosphorus ligands, Helliwell et al.30 showed that the
value of |2JPP| for complexes containing an axial PMe2Ph ligand
and an equatorial P(OMe)3 ligand was ca. 45 Hz, with a value of ca.
50 Hz where the positions of the two ligands were reversed. Where
both ligands were in equatorial positions the value was only about
8 Hz, while typical values of |2JPP| for two phosphorus ligands
in axial positions are around 200–300 Hz.30 Thus the NMR data
given above were compatible with a trigonal bipyramidal structure
for 6b, with one pair of phosphorus ligands {i.e. both PMe2Ph,
or both P(OMe)3 ligands} in axial positions and the other pair
in equatorial positions. The splitting patterns for the resonances
for the methyl protons and carbon atoms were, however, difficult
to reconcile with such a structure. Most noticeably, the methyl
1
the resonances for the atoms directly attached to it { H, d 1.00, qd,
1
1
| J BH| = 97.6 Hz, |2JPH| = 18.7 Hz; 31P, d 121.9, q, | J BP| =
94.5 Hz}. The corresponding atoms in the adduct H3B·P(OPh)3
have chemical shifts d 0.75 and 108.0, with 11B splittings of 100
11
11
and 90 Hz, respectively.31 There were also doublet resonances { H,
1
d 3.22, |3JPH| = 10.9 Hz; 13C, d 50.3, |2JPC| = 28.1 Hz} for the
methyl substituents.
Characterisation of the ruthenium product 1b of the reaction
was complicated by the fact that it underwent a further reaction
not much slower than its formation from 5b (this reaction is
discussed below). Nevertheless, reasonably good 1H and 31P NMR
data were obtained for 1b at 263 K, and a 13C spectrum was
obtained by allowing the reaction to reach the point at which
the concentration of 1b was at its highest, and then cooling the
solution to 243 K before recording the spectrum. Like 5b, 1b
was found to contain a P(OMe)3 ligand. As expected {given the
formation of H3B·P(OMe)3 during the reaction}, 1b contained a
proton resonances were both triplets {PMe2Ph, d 1.73, |2JPH
+
5
4JPH| = 7.2 Hz; P(OMe)3, d 3.18, |3JPH + JPH| = 11.8 Hz}.
This situation, where each methyl proton in a pair of equivalent
phosphorus ligands appears to be equally strongly coupled to
both phosphorus nuclei, requires that the value of |2JPP| between
4
the two PMe2Ph 31P nuclei be much larger than |2JPH − JPH|,
1
and that |2JPP| between the P(OMe)3 31P nuclei be much larger
hydride ligand, characterised by an apparent quartet in the H
5
than |3JPH − JPH|.24,25 Reference to the values given above for
NMR spectrum at d −7.00. Selective decoupling confirmed that
the simple appearance of the resonance resulted from very similar
values (|2JPH| = 25.1 Hz) for the coupling constants to all three 31P
nuclei, indicating that the hydride ligand in 1b was cis to all three
phosphorus ligands.32 The remaining two coordination sites were
occupied by a carbonyl ligand, represented by a poorly resolved
resonance at d 206.3 in the 13C spectrum, and an ethyl ligand.
Although the resonance for the methylene protons in the ethyl
ligand was partly obscured by a resonance for H3B·P(OMe)3, that
for the methylene carbon atom was a doublet of triplets at d −2.7,
and the fact that the doublet splitting (|2JPC| = 78.7 Hz) was
much larger than the triplet splitting (|2JPC| = 10.4 Hz) was a clear
indication that the ethyl ligand was trans to P(OMe)3. Thus 1b was
|2JPP| in trigonal bipyramidal ruthenium(0) complexes shows
that one can reasonably expect this condition to be met for two
axial ligands, but not for two equatorial ligands. As in the case of
[Ru(CO)(PMe2Ph)4], 6a,22 it is possible that 6b actually adopts a
square pyramidal structure, with mutually “trans” PMe2Ph ligands
and mutually “trans” P(OMe)3 ligands forming the base of the
pyramid. As shown in Scheme 2, however, it is also possible
that this square pyramidal structure is actually slightly higher in
energy than two isomeric trigonal bipyramidal structures, between
which there is rapid interconversion by Berry pseudorotation.33 In
1
support of this, we found that all the resonances in the H, 13C
and 31P NMR spectra of 6b broadened as the temperature of
Dalton Trans., 2006, 2661–2670 | 2663
This journal is
The Royal Society of Chemistry 2006
©

View Article Online
the resonances, but we did not reach a temperature high enough
to allow observation of a single averaged resonance for all four
2
hydrogens. Nevertheless the g -tetrahydroborate ligand was clearly
fluxional, and the behaviour of the bridging hydrogen resonances
suggested that one Ru · · · HB bridge (probably the one trans to the
acyl ligand- see later) might break more readily than the other.
The ligand arrangements in 5a, 5b and 5c are clearly all equivalent
(see Scheme 1).
Scheme 2 The complexes Ru(CO)(PMe2Ph)2Lꢀ2 {6a, Lꢀ = PMe2Ph;
6b, Lꢀ = P(OMe)3} may be square pyramidal but may alternatively
be undergoing rapid interconversion in solution between two trigonal
bipyramidal structures. L = PMe2Ph throughout.
Complex 7c contained two inequivalent Me3CNC ligands, each
the solution was lowered from 300 to 213 K, although we were
unable to reach a temperature at which any of the resonances
separated into the pairs of resonances which would be expected
for the individual isomers.
1
with a pattern of resonances in the H and 13C NMR spectra
similar to that for the Me3CNC ligand in 5c. The resonances for
the propanoyl ligand in 7c were also at chemical shifts similar
to those for the same ligand in 5c, except that the acyl carbon
resonance in 7c was at d 273.2, as opposed to d 256.9 for 5c, an
indication of the change in the ligand trans to the propanoyl ligand
(now, see Scheme 1, necessarily one of the Me3CNC ligands).35
The complete reaction sequence 3 → 5b → 1b → 6b is shown
in Scheme 1. It should be noted that we did not actually detect
1
the species [Ru(g -BH4)(CO)Et(PMe2Ph)2{P(OMe)3}] (shown as
4b) during either of the first two steps. Nevertheless, given the
1
In the H spectrum of 7c, the single bridging hydrogen in the
1
fact that the analogous complex [Ru(g -BH4)(CO)Et(PMe2Ph)3],
1
g -tetrahydroborate ligand was represented by a broad resonance,
4a, clearly does feature as an intermediate in the corresponding
reaction sequence 3 → 5a → 1a → 6a,22 it seems likely that 4b
plays a similar role here. This point is discussed further in the
Conclusions section.
visible at d −12.0 in spectra recorded at or below 233 K. This
resonance sharpened on cooling and disappeared on warming.
1
The corresponding resonance for [Ru(g -BH4)(CO)Et(PMe2Ph)3],
4a, has a chemical shift of d −12.3, and its appearance exhibits a
similar temperature dependence.22 Although we had been unable
to detect the resonance for the three terminal hydrogens in 4a,
for 7c it was visible at 213 K as a broad singlet (broader than
that for the bridging hydrogen at the same temperature) at d
0.8. The greater broadening may reflect a larger interaction with
the quadrupolar boron nucleus, since the terminal B–H bonds
would be expected to be shorter than bridging B–H. Like the
bridging hydrogen resonance, the terminal resonance sharpened
on cooling and disappeared on warming. We were unable to
reach a temperature at which an averaged resonance for the four
hydrogens could be detected (see later). The nature of the process
(iii) Reaction of 3 with Me3CNC: early stages
Treatment of solutions of 3 in either CD3COCD3 or CD3C6D5
with slightly less than an equimolar amount of Me3CNC at 213 K
resulted in the immediate formation of a mixture of two species,
5c and 7c. All the Me3CNC was consumed, and after the reaction
an appreciable quantity of 3 remained in the solution. Although
it was not possible to generate the major product 5c free from 7c,
controlled further addition of Me3CNC, still at 213 K, resulted in
the conversion of both 3 and 5c to 7c. 7c itself underwent a further
reaction even at relatively low temperatures, making its isolation
impossible, but 5c and 7c were satisfactorily identified by NMR
1
which exchanges the identities of the four hydrogens in the g -
tetrahydroborate ligand will also be discussed later.
2
spectroscopy as [Ru(g -BH4)(CNCMe3)(COEt)(PMe2Ph)2] and
1
[Ru(g -BH4)(CNCMe3)2(COEt)(PMe2Ph)2] respectively. Apart
In Scheme 1 we have implied that the conversion of 3 to
1
1
from minor differences in chemical shifts, the spectra were the
same in both solvents.
5c occurs by way of an g -tetrahydroborate species [Ru(g -
BH4)(CNCMe3)(CO)Et(PMe2Ph)2], 4c, similar to 4a, which we
observed in the reaction of 3 with PMe2Ph. Whether or not
this is the case (see Conclusions section), it is clear that here
again the nucleophile attaches itself to the metal rather than
abstracting BH3. Indeed, even the attack by a second molecule
of the nucleophile does not abstract BH3: instead, we see again the
breaking of one Ru · · · HB bridge (the one trans to the acyl ligand)
and attachment of the nucleophile to the metal, yielding 7c. Since
even the use of slightly less than one equivalent of Me3CNC in the
reaction with 3 yielded a significant amount of 7c as well as 5c,
either the rate of conversion of 5c to 7c must be comparable with
its rate of formation from 3, or 5c must rapidly undergo partial
disproportionation into 3 and 7c.
1
The H and 13C NMR spectra of 5c showed that it contained
one Me3CNC ligand, whose resonances included a singlet at d
1
1.39 in the H spectrum and singlets in the 13C spectrum at d
31.2 and 56.6 for the Me3C group. The metal-bonded carbon
atom in the ligand was represented by a broad triplet at d 162.6
(|2JPC| = ca. 17 Hz), a typical value for |2JPC| between mutually
cis isonitrile and phosphine ligands.34 Both the chemical shifts
and the broadening of the resonance for the metal-bonded carbon
atom (caused by the quadrupolar 14N nucleus) are characteristic
for ruthenium(II) complexes containing the ligand Me3CNC.35
1
The patterns of resonances in the H and 13C NMR spectra
for the propanoyl ligand in 5c were similar to those for 5a and
5b. Similarities in the spectra of the three complexes were also
2
evident for the g -tetrahydroborate ligand: for 5c at 213 K, the
(iv) Reaction of 3 with Me3CNC: later stages in CD3C6D5
solution
resonances for the two inequivalent bridging hydrogens were
at d −6.6 and −5.7, and that for the two equivalent terminal
hydrogens was at d 2.8. Raising the temperature of the solution
caused all three resonances to broaden, with the bridging hydrogen
resonance at d −5.7 broadening faster than that at d −6.6. At
263 K these two resonances had merged into one broad hump
at d −6.5. Further warming resulted in increased broadening of
As described above, the reaction between 3 and Me3CNC in a
1 : 2 molar ratio at 213 K ultimately yielded 7c, whether the
solvent used was CD3C6D5 or CD3COCD3. No further reaction
was observed in CD3C6D5 solution at this temperature, even if
more Me3CNC was added. When, however, a solution containing
2664 | Dalton Trans., 2006, 2661–2670
This journal is
The Royal Society of Chemistry 2006
©

View Article Online
equimolar quantities of 7c and Me3CNC was warmed to 293 K,
the resonances for 7c in the 31P and 1H spectra of the solution dis-
appeared, with the formation of one major ruthenium-containing
product, [Ru(CNCMe3)2(COEt)H(PMe2Ph)2], 8c. The lifetime of
this complex in solution was sufficient to allow 31P and 1H NMR
spectra to be recorded at this temperature, and a 13C spectrum
was obtained after cooling the solution to 213 K. Apart from
minor changes in chemical shift, the patterns of resonances in
the spectra of 8c were for the most part very similar to those in
the spectra of 7c, revealing the presence of two inequivalent (and
therefore mutually cis) Me3CNC ligands and a propanoyl ligand.
Only the resonances for the tetrahydroborate ligand in 7c were
The formation of 8c, which contains mutually cis acyl and
hydride ligands, is of interest since such species have been
postulated as intermediates in catalytic processes such as alkene
hydroformylation38,39 and aldehyde decarbonylation.40 Disregard-
ing cases in which the acyl ligand is stabilised either by chelation
or by bridging to another metal, there appear to be relatively few
well-characterised examples of complexes containing both an acyl
and a hydride ligand: these are mainly complexes of rhodium,41
iridium,42 platinum,43 tungsten37 and rhenium.44 As in the prepa-
ration of [Ru(CO)(Et)H(PMe2Ph)3], 1a, our route to 8c uses the
tetrahydroborate ligand as a protected hydride ligand which is then
deprotected when the other ligands required are in place. It is inter-
esting that the major reaction of 8c, in the presence of Me3CNC,
involves the combination of hydride and isonitrile ligands to yield
9c, rather than the elimination of propanal by the combination of
the hydride and propanoyl ligands. The reaction parallels that be-
tween [Ru(CNCMe3)(CO)(COMe)Me(PMe2Ph)2] and Me3CNC
to yield [Ru(CNCMe3)(CO){C(Me)NCMe3}(COMe)(PMe2Ph)2]:
here combination of methyl and isonitrile ligands occurs in
preference to the elimination of propanone.35
1
missing from the H spectrum of 8c: in their place was a sharp
triplet resonance (|2JPH| = 23.7 Hz) at d −6.81 for a hydride
ligand.
Clearly the ligand arrangement in 8c (see Scheme 1) matched
that in 7c, implying that the reaction simply involved the abstrac-
tion of BH3 from 7c by Me3CNC. We were not able to obtain the
adduct H3B·CNCMe3 in a pure state, but tentatively assigned a 1 :
1 : 1 triplet resonance (|3JNH| = 2.2 Hz) at d 0.64 in the 1H NMR
spectrum of the solution to the methyl protons, and singlets at d
29.3 and d 58.2 in the 13C spectrum to the carbon atoms of the
Me3C group.
(v) Reaction of 3 with Me3CNC: later stages in CD3COCD3
solution
When the reaction between 7c and Me3CNC in CD3C6D5 at
293 K was repeated using a 1 : 2 molar ratio of the reactants,
8c was again formed but then underwent a further reaction to
yield a new species 9c, together with a small amount of another
product which we have been unable to identify. Satisfactory
separation of 9c from its co-product and H3B·CNCMe3 was
not achieved, but 9c appeared to be reasonably long-lived in
solution at 293 K, and was characterised by 31P, 1H and 13C
NMR spectroscopy. Like its precursors, 7c and 8c, 9c contained a
propanoyl ligand. The presence of three singlet resonances, each
representing nine protons, at d 0.86, 1.16 and 1.39, seemed to imply
that 9c also contained three Me3CNC ligands, all inequivalent,
and it was evident that 9c did not contain a hydride ligand.
There was, however, a triplet resonance (|JPH| = 4.8 Hz) at d
Our discovery, that 7c represented the point at which the sequences
of reactions between 3 and Me3CNC in the two solvents diverged,
came about as a result of an attempt to detect an averaged
1
resonance for all four hydrogens in the g -tetrahydroborate ligand
in 7c. On warming a CD3COCD3 solution of the complex
(containing no free Me3CNC), we found that even at 240 K the
complex rearranged at a significant rate to a second isomer, 10c.
The isomerisation, which we did not observe in CD3C6D5 solution,
was complicated by an accompanying disproportionation reaction
(see later), and 10c was only of limited stability. Nevertheless we
1
were able to characterise it by 31P and H NMR spectroscopy.
The pattern of resonances for the mutually trans PMe2Ph ligands
showed that the Ru–P bonds now lay in a plane of symmetry. In
agreement with this, the two Me3CNC ligands were also equivalent
(and hence also mutually trans). As for 7c, triplet (d 0.40) and
quartet (d 2.11) resonances were observed for the methyl and
methylene protons of the propanoyl ligand in 10c (cf. d 0.54
and 2.51 for 7c). The significant difference between the chemical
shifts for the methylene protons in the two complexes presumably
reflected the change in the ligand trans to the propanoyl ligand.
1
10.83 in the H spectrum. The resonance integrated for a single
proton, and decoupling established that the splitting was due
to the 31P nuclei in the PMe2Ph ligands of 9c. The implication
was that the hydride ligand in 8c had been incorporated into
another ligand, most obviously Me3CNC, to give the imino-
formyl ligand C(H)NCMe3, and the chemical shift of d 10.83
is similar to that of d 10.32 for another ruthenium(II) complex
containing an iminoformyl ligand, [Ru(CO){C(H)NC6H4Me}-
1
For the g -tetrahydroborate ligand in 10c we observed a single
2
1
(g -O2CMe)(PPh3)2].36 A H–13C HMQC experiment established
a correlation between this resonance and a triplet (|2JPC| =
16.0 Hz) at d 210.2 in the 13C NMR spectrum for the metal-bound
carbon atom in the iminoformyl ligand. This chemical shift is
ca. 30 ppm below that of d 242.1 for the corresponding atom
in the iminoacetyl ligand in the related complex [Ru(CNCMe3)-
(CO){C(Me)NCMe3}(COMe)(PMe2Ph)2].35 A similar gap sepa-
rates the chemical shifts for the metal-bound carbon atom in a
pair of related formyl and acetyl tungsten complexes.37 Thus 9c
was [Ru(CNCMe3)2{C(H)NCMe3}(COEt)(PMe2Ph)2], with the
ligand arrangement shown in Scheme 1. Its formation from
8c involves the combination of the mutually cis hydride and
Me3CNC ligands in 8c, followed by occupation of the now vacant
coordination site by another molecule of Me3CNC.
averaged resonance for all four hydrogens. The resonance showed
signs of fine structure due to coupling to the major boron isotope
11B (31P decoupling had no effect on its appearance). The lifetime
of a quadrupolar nucleus in a particular spin state increases as
the temperature is raised, with the result that a resonance for a
hydrogen nucleus bonded to 11B can be a singlet at low temperature
and then on warming become first a broad doublet, then a broad
doublet with shoulders on the outer sides of the peaks, and
ultimately a 1 : 1 : 1 : 1 quartet. The effect is nicely illustrated
by the 1H NMR spectrum of Zr(BH4)4.15 As shown in Fig. 1, the
resonance for 10c was a doublet with shoulders at 293 K, a doublet
at 273 K and a singlet at 253 K. With further cooling, the slowing
of the fluxional process exchanging terminal and bridging hydro-
gens caused the resonance to broaden and finally to disappear.
This journal is
The Royal Society of Chemistry 2006
Dalton Trans., 2006, 2661–2670 | 2665
©

View Article Online
When a CD3COCD3 solution of 7c was treated with Me3CNC
and then warmed, 10c was not formed. Instead, a new complex
[Ru(CNCMe3)3(COEt)(PMe2Ph)2]+, 11c, was obtained, with the
liberation into the solution of BH4−, recognisable by the charac-
1
11
teristic pattern of superimposed 1 : 1 : 1 : 1 quartet (| J BH| =
1
10
81.3 Hz) and 1 : 1 : 1 : 1 : 1 : 1 : 1 septet (| J BH| = 27.3 Hz)
resonances at d 0.11 in the 1H NMR spectrum. Complex 11c could
also be produced, even at 213 K, by treating a CD3COCD3 solution
of 10c with Me3CNC. From the 1H and 31P NMR spectra of 11c
it was evident that the Ru–P bonds to the mutually trans PMe2Ph
ligands lay in a plane of symmetry, and that two of the three
Me3CNC ligands were equivalent (and therefore also mutually
trans). There were also the expected resonances for the propanoyl
ligand. Clearly the ligand arrangement in 11c was that shown in
Scheme 3. Although we could not obtain the tetrahydroborate salt
of 11c in crystalline form, treatment with NaBPh4 in propanone
solution yielded the tetraphenylborate salt as a white solid.
Analytical data for a recrystallised sample were consistent with
the formula [Ru(CNCMe3)3(COEt)(PMe2Ph)2](BPh4).
The simplest mechanism for the reaction sequence would
seem to be that shown in Scheme 3. The 16-electron species
[Ru(CNCMe3)2(COEt)(PMe2Ph)2]+, 12c, is the logical interme-
diate in the conversion of 10c to 11c, and the extreme lability
1
of the g -tetrahydroborate ligand when positioned trans to the
propanoyl ligand, demonstrated by the fact that 10c reacted
rapidly with Me3CNC in CD3COCD3 solution even at 213 K,
is unsurprising given the strong trans-labilising effect of an acyl
ligand in ruthenium(II) complexes.45 We have previously noted45
that the kinetically preferred direction of attack on a 16-electron
ruthenium(II) species containing an acyl ligand is trans to this
Fig. 1 1H NMR spectra, recorded in CD3COCD3, of the resonance for
the tetrahydroborate ligand in complex 10c.
1
ligand, so dissociation of the g -tetrahydroborate ligand from 7c
could provide both a pathway for the isomerisation of 7c to 10c
and a direct route from 7c to 11c. On this basis, the fact that these
reactions require a higher temperature than does the conversion of
10c to 11c can be attributed to the fact that the tetrahydroborate
ligand in 7c is not trans to the trans-labilising acyl ligand.
We cannot completely rule out an alternative mechanism for
the isomerisation of 7c to 10c which involves dissociation and
reattachment of an Me3CNC ligand, nor does the failure to
observe 10c during the reaction between 7c and Me3CNC to form
11c eliminate a route from 7c to 11c which passes through 10c,
since the conversion of 10c to 11c is clearly considerably faster
than the isomerisation of 7c to 10c. Nevertheless, the fact that we
did not observe either the isomerisation of 7c or the conversion
of 7c to 11c when propanone was replaced as the solvent by
the considerably less polar methylbenzene (see previous section)
provides support for the mechanism shown in Scheme 3. The
delicate thermodynamic balance between some of the species in
Scheme 3 was highlighted by the observation (mentioned earlier)
that the isomerisation of 7c to 10c which occurred on warming a
solution of 7c in CD3COCD3 solution was accompanied by the
appearance of significant quantities of 5c and 11c. This shows that
7c is easily reconverted to 5c, with the thermodynamic drive for
the reconversion being provided by use of the liberated Me3CNC
to generate 11c.
We were unable to reach a temperature low enough to allow
observation of separate resonances for terminal and bridging
protons. On the basis of the NMR data, we concluded that the
ligand arrangement in 10c was that shown in Scheme 3.
1
Reversible dissociation of the g -tetrahydroborate ligand in
7c and 10c could also account for the easy exchange of ter-
minal and bridging hydrogens in both of these complexes in
CD3COCD3 solution. On this basis, the observation of an averaged
Scheme 3 Later stages of the reaction between 3 and Me3CNC in
CD3COCD3 solution. L = PMe2Ph, Lꢀ = Me3CNC. The intermediate
12c (not detected) is assumed to be very short-lived.
2666 | Dalton Trans., 2006, 2661–2670
This journal is
The Royal Society of Chemistry 2006
©

View Article Online
Lꢀ = P(OMe)3, 5b; Lꢀ = Me3CNC, 5c}. In Scheme 1 we have
resonance for all four hydrogens in 10c at a temperature at
which separate resonances can still be observed for terminal and
bridging hydrogens in 7c would be attributable to the strong trans-
labilising effect of the propanoyl ligand. Two factors, however,
argue against complete dissociation. Firstly, the exchange process
in 7c was much faster than the isomerisation of 7c to 10c,
although this could simply imply that the 16-electron interme-
diate [Ru(CNCMe3)2(COEt)(PMe2Ph)2]+ formed by tetrahydrob-
orate dissociation is square-pyramidal, rather than the trigonal-
bipyramidal species shown as 12c in Scheme 3, and must rearrange
prior to reattachment of the tetrahydroborate ligand to give 10c.
On this basis the relative slowness of the isomerisation could
be attributed to competition between this rearrangement and a
simple reattachment of the tetrahydroborate ligand to reform 7c.
Secondly, we have also observed exchange of terminal and bridging
1
suggested that 5d is formed by way of the g -tetrahydroborate
1
species [Ru(g -BH4)(CO)2Et(PMe2Ph)2], 4d, but we were unable
to detect this species. Assuming that it is an intermediate in the
reaction, it must be very rapidly converted to 5d (see Conclusions
section).
The identification of 8d as [Ru(CO)2(COEt)H(PMe2Ph)2] was
straightforward, and the resonances for two inequivalent carbonyl
ligands, a propanoyl ligand and a hydride ligand indicated
that the ligand arrangement must be that shown in Scheme 1.
Thus the conversion of 5d to 8d must involve abstraction of
H3B·CO. As shown in Scheme 1, we assumed that this would
1
occur by way of an intermediate containing an g -tetrahydroborate
1
ligand, [Ru(g -BH4)(CO)2(COEt)(PMe2Ph)2], 7d, the equivalent
1
of [Ru(g -BH4)(CNCMe3)2(COEt)(PMe2Ph)2], 7c, formed in the
1
hydrogens in g -tetrahydroborate complexes of ruthenium(II) in
reaction between 3 and Me3CNC. We were unable to detect such
a species in CD3COCD3 solution, but repetition of the reaction
between 3 and a limited amount of CO in CD3C6D5 solution
the considerably less polar solvent CD3C6D5,22,23 and there did
not appear to be a very large difference in exchange rate at a
given temperature in the two solvents. We conclude that complete
dissociation is not necessary for the exchange process, although
it seems reasonable to assume that some charge separation
1
at 213 K led to the detection in the 31P and H NMR spectra of
resonances for a third species (in addition to 5d and 8d), tentatively
identified as 7d. These resonances included those expected for
the methyl and methylene protons in a propanoyl ligand {d 0.89
(t) and 2.30 (q) respectively, both with |3JHH| = 6.8 Hz}. Most
significantly, an extremely broad resonance was observed in the
1H NMR spectrum at d −10.5, and this sharpened markedly on
cooling. Both the chemical shift and the change in appearance with
temperature suggested that this was the single bridging hydrogen
1
occurs during the process. Studies of fluxionality in other g -
tetrahydroborate complexes have also led to the conclusion that
exchange of bridging and terminal hydrogens does not require
complete dissociation of the tetrahydroborate ligand.46,47
(vi) Reaction of 3 with CO
1
in an g -tetrahydroborate ligand (cf. 4a22 and 7c).
When CO was bubbled, very gently, through a CD3COCD3
When a CD3COCD3 solution of 8d was treated with CO and
solution of 3 for a brief period at 195 K, 31P and 1H spectra subs-
then warmed to 263 K, propanal was formed and characterised by
2
1
equently recorded at 213 K showed that two species, [Ru(g -BH4)-
comparison with a H spectrum of propanal in the same solvent
and at the same temperature {d 9.72 (t, 1H, |3JHH| = 1.2 Hz); d
2.46 (qd, 2H, |3JHH| = 7.3 Hz and 1.2 Hz); d 0.96 (t, 3H, |3JHH| =
7.3 Hz)}. The ruthenium product was [Ru(CO)3(PMe2Ph)2],23 6d,
presumably formed after propanal elimination by the capture of
the 16-electron species [Ru(CO)2(PMe2Ph)2] by CO. It should
be noted that propanal and 6d were also the products obtained
simply by treating a CD3COCD3 solution of 3 with CO at room
temperature. Treatment of a CD3COCD3 solution of 8d with H2,
followed by warming to 263 K, again yielded propanal, but the
ruthenium product was [Ru(CO)2H2(PMe2Ph)2],32 resulting from
capture of the [Ru(CO)2(PMe2Ph)2] by H2.
(CO)(COEt)(PMe2Ph)2], 5d, and [Ru(CO)2(COEt)H(PMe2Ph)2],
8d, had been formed, although some 3 remained in solution. It
proved impossible to achieve complete conversion of 3 to 5d with-
out some 8d being formed as well, and longer treatment with CO
(15 min at 213 K) led to both 3 and 5d being completely converted
to 8d. Complex 8d was also relatively unstable, decomposing at
263 K (see later). Nevertheless, we were able to obtain good quality
31P, 1H and 13C NMR spectra of 8d, and to detect and identify most
of the 31P and 1H resonances due to 5d.
The presence of an ethyl group in 5d was indicated by methyl and
methylene proton resonances {d 0.37 (t) and 1.58 (q), respectively}
with a value for |3JHH| of 7.1 Hz. Both the absence of splittings
by 31P and the relatively high chemical shift for the methylene
protons implied that the ethyl group was part of a propanoyl
ligand, rather than being directly attached to the metal. The
presence of a bidentate tetrahydroborate ligand was established
by the observation of two broad resonances, each of area 1, in the
1H spectrum at d −5.0 and −4.7, characteristic chemical shifts
for the bridging hydrogens in such a ligand. We were unable
to locate the resonance for the terminal hydrogens, probably
because of its breadth and the presence of other resonances in
the appropriate region. In view of the inevitable presence of 8d
in the solution, definite identification of resonances for 5d in the
13C NMR spectrum was difficult, so our assumption that the final
coordination site in the complex was occupied by a carbonyl ligand
was based on the fact that CO was the only other reactant and
that the reactions of 3 with all other nucleophiles tried had yielded
(vii) Feasibility of a catalytic cycle for ethene hydroformylation
We had now demonstrated that it was possible, under extremely
2
mild conditions, to obtain propanal by first treating [Ru(g -
BH4)(CO)H(PMe2Ph)2], 2, with ethene,23 and then treating the
product of this reaction, 3, with CO. This sequence is not,
however, catalytic, as the hydrogen required for the aldehyde
is derived from 2, not from H2, and BH3 is abstracted during
the sequence. Clearly it is necessary to introduce H2 into the
system before BH3 is abstracted. A possible catalytic cycle would
be of the type shown in Scheme 4. Some encouragement as
regards the feasibility of the step in which 5d adds H2 to form
1
[Ru(g -BH4)(CO)(COEt)H2(PMe2Ph)2], 13, was provided by the
2
discovery that [Ru(g -BD4)(CO)D(PMe2Ph)2], d5-2, underwent
exchange of the deuteride ligand with hydrogen on treatment with
H2 in C6D6 at room temperature.
2
complexes [Ru(g -BH4)(COEt)Lꢀ (PMe2Ph)2] {Lꢀ = PMe2Ph, 5a;
This journal is
The Royal Society of Chemistry 2006
Dalton Trans., 2006, 2661–2670 | 2667
©

View Article Online
as solvent, even attack by a third molecule of Me3CNC does
−
not cause BH3 abstraction: instead the complete BH4 ligand
is displaced to give [Ru(CNCMe3)3(COEt)(PMe2Ph)2]+, 11c (see
Scheme 3).
In 3, the vacant site on the metal necessary to allow attachment
of the nucleophile could be created either (i) by breaking an
Ru · · · HB bridge or (ii) by combining the ethyl and carbonyl
ligands. In the case of the reaction with PMe2Ph, the initial
Scheme 4 A possible cycle for ethene hydroformylation catalysed by 2.
We have not detected the presence of 13.
1
formation of [Ru(g -BH4)(CO)Et(PMe2Ph)3], 4a, and its subse-
quent conversion to 5a make it clear that the correct option is
(i). Our failure to observe the corresponding species 4b–d during
the conversion of 3 to 5b–d means that we cannot prove that
the same is true in these cases. It is known, however, that the
rate-determining step in the conversion of methyl complexes of
ruthenium(II) [Ru(CO)2(Me)X(PMe2Ph)2] (X = Me or Ph) to acyl
complexes [Ru(CO)(COMe)X(PMe2Ph)2Lꢀ] does not involve the
incoming ligand Lꢀ,35 and it seems likely that the same applies to the
opening of an Ru · · · HB bridge, particularly since the 1H spectrum
of 3 shows that bridge opening (and reclosing) occurs rapidly even
in the absence of a nucleophile and in a non-nucleophilic solvent
such as CD3C6D5.23 On this basis, the pathway from 3 to complexes
5a–d should be the same for all four complexes, which means that
the rate of conversion of the intermediates 4b–d to 5b–d must be
appreciably faster than that for the conversion of 4a to 5a. On
steric grounds, one would anticipate that combination of ethyl
and carbonyl ligands would be more rapid for 4a than for 4b and
4d, since PMe2Ph has a greater cone angle50 than either P(OMe)3
or CO. Thus it seems that the key factor here is electronic, namely
the greater p-acceptor ability of P(OMe)3, Me3CNC and CO, all
able to compensate better than PMe2Ph for the loss of much of
the p-acceptor character of the carbonyl ligand in 4 as it becomes
part of the propanoyl ligand.
A simple test would involve the treatment of a solution of 5d
with H2. Unfortunately, as mentioned in the previous section, in
the reaction between 3 and CO we were unable to achieve complete
2
conversion of 3 to [Ru(g -BH4)(CO)(COEt)(PMe2Ph)2], 5d, before
some abstraction of BH3 to yield [Ru(CO)2(COEt)H(PMe2Ph)2],
8d, occurred. This meant that H2 must be added to the solution of
3 prior to its treatment with CO. In order to maximise the chance
of capturing 5d with H2 before it could react further with CO,
we presaturated a CD3COCD3 solution of 3 with H2 at 195 K,
having previously checked to ensure that 3 itself would not react
with H2 at the temperature used for the reaction. A very slow
stream of CO was then bubbled through the solution for 30 s,
without stopping the flow of H2. The CO supply was then turned
1
off. After 2 min the H2 flow was also stopped, and a H NMR
spectrum of the solution was recorded at 213 K. There was no
evidence for the presence of a species such as 13 (see Scheme 4)
in the solution, but a small amount of 2, the expected ruthenium
end-product of the catalytic cycle, was now present. Several other
ruthenium species were also present, however, most notably 8d, the
immediate precursor of propanal elimination in the non-catalytic
sequence shown in Scheme 1.
Thus, even under conditions where H2 is present in a higher
concentration than CO, 5d appears to react at an appreciably faster
rate with CO than with H2. Nevertheless, the fact that formation
of 2 was observed suggests that, with further alterations to the
system, the reaction might be made catalytic. To counter the ease
with which the boron can apparently be abstracted by CO, one
might “tether” the boron to the rest of the complex by one or two
separate links in addition to the hydrogen bridges. In this respect,
some interesting complexes of ruthenium(II) and other metals have
recently been prepared using the ligand H3B·PPh2CH2PPh2, which
can chelate to the metal through both a phosphorus atom and one
or two M · · · HB bridges.48 Here again, though, there would be the
risk that cleavage of the B–P bond by CO could still lead to BH3
abstraction, and a boron atom attached to the metal through a
simple carbon chain or a B–C–P–Ru chain (as, for example, in the
work of Baker et al.49) might be more safely anchored.
The value of the ability of the tetrahydroborate ligand to
act as a protected hydride ligand is illustrated by the ease
2
with which the readily prepared [Ru(g -BH4)(CO)H(PMe2Ph)2],
2, can be converted first to the ethyl complex 323 and then to
complexes of the type 4 or 5, before the protection is removed
under conditions which are still mild enough to allow study of
the resulting alkyl hydride complexes [Ru(CO)(Et)H(PMe2Ph)2Lꢀ]
{1a, Lꢀ = PMe2Ph; 1b, Lꢀ = P(OMe)3} and acyl hydride complexes
[Ru(COEt)H(PMe2Ph)2Lꢀ2] {8c, Lꢀ = Me3CNC; 8d, Lꢀ = CO}.
Which type of complex is obtained presumably depends on the
relative stabilities of the precursors 4 and 5: as discussed above,
the greater the p-accepting ability of the ligand Lꢀ is, the less
serious should be the loss of much of the p-acceptor character
of the carbonyl ligand in 4 as it becomes part of the acyl ligand
in 5.
In the presence of the appropriate ligands Lꢀ, complexes
1a and 1b undergo ethane elimination, forming ruthenium(0)
complexes [Ru(CO)(PMe2Ph)2Lꢀ2]. Complex 8c reacts differ-
ently, with combination of mutually cis hydride and Me3CNC
ligands allowing uptake of Me3CNC to give [Ru(CNCMe3)2-
{C(H)NCMe3}(COEt)(PMe2Ph)2], 9c. Complex 8d, which could
react in a similar manner to 8c, with combination of mutually cis
hydride and carbonyl ligands to give a formyl complex, instead
undergoes propanal elimination. One factor here could be that the
presence of the two strongly p-accepting carbonyl ligands in 8d
favours the reduction to ruthenium(0) which propanal elimination
achieves.
Conclusions
2
As Scheme 1 indicates, the fact that reactions of [Ru(g -
BH4)(CO)Et(PMe2Ph)2], 3, with nucleophiles Lꢀ do not neces-
sarily result in the abstraction of BH3 has now been clearly
2
demonstrated by the identification of the products [Ru(g -
BH4)(COEt)(PMe2Ph)2Lꢀ] {5a, Lꢀ = PMe2Ph; 5b, Lꢀ = P(OMe)3;
5c, Lꢀ = Me3CNC; 5d, Lꢀ = CO}. With the nucleophiles Me3CNC
and CO, a second molecule of the nucleophile may be attached
1
to the metal without loss of BH3, giving complexes [Ru(g -
BH4)(COEt)(PMe2Ph)2Lꢀ2], 7c and 7d. Indeed, using CD3COCD3
2668 | Dalton Trans., 2006, 2661–2670
This journal is
The Royal Society of Chemistry 2006
©

View Article Online
2
The use of [Ru(g -BH4)(CO)H(PMe2Ph)2], 2, makes it possible
the gases, and a short needle served as a vent. Details of procedures
and reaction conditions have been given earlier in the text. Again
the intermediate species involved were too unstable to isolate. The
end-products, 6d, [Ru(CO)2H2(PMe2Ph)2] and, in the attempted
catalysis experiment, 2, were identified by comparison of their
NMR spectra with spectra of authentic samples recorded in the
same solvent.
to convert ethene to propanal at low temperature and under
1 atmosphere of CO. The difficulty in establishing a catalytic
cycle for ethene hydroformylation which uses 2 lies in persuading
2
the intermediate [Ru(g -BH4)(CO)(COEt)(PMe2Ph)2], 5d, to react
with H2 to generate 2 and propanal, rather than following the
non-catalytic route in which BH3 abstraction by CO precedes the
reductive elimination of propanal. Hopefully, as discussed earlier,
the use of a modified tetrahydroborate ligand, in which one or
both terminal hydrogens are replaced by “tethers”, may solve this
problem.
Exchange between d5-2 and H2
1
2
A H NMR spectrum of a C6D6 solution (0.7 cm3) of [Ru(g -
BD4)(CO)D(PMe2Ph)2] (d5-2, 90 mg), held in an NMR tube fitted
with a suba-seal, was recorded to check on the completeness
of deuterium incorporation. H2 was then bubbled through the
Experimental
1
solution for 2 min at room temperature. A H NMR spectrum
Except where CO and/or H2 were used as reactants, all experimen-
tal work was carried out under an atmosphere of N2. NMR spectra
were acquired using a Bruker AMX 500 FT spectrometer. The
preparations of complexes 1a, 2, d5-2, 3, 4a, 5a and 6a and details
of their characterisation have been given in earlier papers.22,23
recorded immediately after the H2 addition showed that the
intensities of both the resonance for the hydride ligand and
those for the tetrahydroborate ligand had already increased by
a small amount. Further treatment with H2 over a period of
90 min resulted in a steady increase in the intensities of all these
resonances.
The reactions of 3 with P(OMe)3 and Me3CNC
All the sequences of reactions were carried out in NMR tubes
fitted with suba-seals, using solutions of 3 (95 mg) in CD3C6D5
or CD3COCD3 (0.7 cm3). The appropriate amounts of P(OMe)3
or Me3CNC were added using a microlitre syringe. Details
of the amounts added and the reaction conditions have been
given earlier in the text. The reactions were monitored by 31P
Acknowledgements
We thank Johnson Matthey PLC (“JM”) for a generous loan of
ruthenium trichloride.
References
1
and H NMR spectroscopy. Given the ease of interconversion
1 T. J. Marks and J. R. Kolb, J. Am. Chem. Soc., 1975, 97, 3397.
2 J. A. Jensen and G. S. Girolami, Inorg. Chem., 1989, 28, 2114.
3 S. L. Conway, L. H. Doerrer and M. L. H. Green, Organometallics,
2000, 19, 630.
4 M. L. H. Green and L.-L. Wong, J. Chem. Soc., Dalton Trans., 1989,
2133.
5 S. W. Kirtley, M. A. Andrews, R. Bau, G. W. Grynkevich, T. J. Marks,
D. L. Tipton and B. R. Whittlesey, J. Am. Chem. Soc., 1977, 99, 7154.
6 D. Gusev, A. Llamazares, G. Artus, H. Jacobsen and H. Berke,
Organometallics, 1999, 18, 75.
7 C. A. Ghilardi, P. Innocenti, S. Midollini and A. Orlandini, J. Chem.
Soc., Dalton Trans., 1985, 605.
8 H. Werner, M. A. Esteruelas, U. Meyer and B. Wrackmeyer, Chem.
Ber., 1987, 120, 11.
9 M. A. Esteruelas, Y. Jean, A. Lledos, L. A. Oro, N. Ruiz and F. Volatron,
Inorg. Chem., 1994, 33, 3609.
and/or decomposition (even at low temperature) of the individual
intermediates in the reaction sequences, no attempt was made
to isolate them, but details of their characterisation by NMR
spectroscopy have been given in the text. Difficulty was encoun-
tered in isolating the final species in the P(OMe)3 sequence, 6b,
which appeared only to be stable in the presence of free P(OMe)3.
Although the major end-product from the reaction between 3
and Me3CNC in CD3C6D5 solution, 9c, seemed to be stable
in solution at ambient temperature, we were unable to achieve
complete removal of unidentified minor reaction products and the
adduct BH3·CNCMe3.
Isolation of 11c as its tetraphenylborate salt
10 H. D. Empsall, E. Mentzer and B. L. Shaw, J. Chem. Soc., Chem.
Commun., 1975, 861.
Complex 3 (50 mg) in CD3COCD3 (0.7 cm3) was treated with
Me3CNC (50 mm3) at 273 K. Conversion to the end-product of
the reaction, 11c, was complete within 1 h. The solution was then
stirred with NaBPh4 (50 mg) at 273 K for 1 h. The solution was
then filtered, and all volatile materials were removed from the
filtrate under reduced pressure. The residual oil was dissolved in
the minimum quantity of propanone, and a similar volume of
ethanol was added. The solution was kept at 253 K for 24 h.
The white crystals formed over this period were separated from
the solvents, washed with ethanol and dried (Found: C, 69.48; H,
7.44; N, 4.29. Calc. for C58H74BN3OP2Ru: C, 69.45; H, 7.44, N,
4.19%).
11 M. L. H. Green, H. Munakata and T. Saito, J. Chem. Soc. A, 1971,
471.
12 M. Grace, H. Beall and C. H. Bushweller, Chem. Commun., 1970, 701.
13 J. W. Bruno, J. C. Huffman and K. G. Caulton, Inorg. Chim. Acta, 1984,
89, 167.
14 I. Demachy, M. A. Esteruelas, Y. Jean, A. Lledos, F. Maseras, L. A.
Oro, C. Valero and F. Volatron, J. Am. Chem. Soc., 1996, 118, 8388.
15 T. J. Marks and J. R. Kolb, Chem. Rev., 1977, 77, 265.
16 J. B. Letts, T. J. Mazanec and D. W. Meek, J. Am. Chem. Soc., 1982,
104, 3898.
17 H. M. Lee, C. Yang and G. Jia, J. Organomet. Chem., 2000, 601, 330.
18 M. L. H. Green and H. Munakata, J. Chem. Soc., Dalton Trans., 1974,
269.
19 C. Bohanna, M. A. Esteruelas, J. Herrero, A. M. Lopez and L. A. Oro,
J. Organomet. Chem., 1995, 498, 199.
20 V. Montiel-Palma, M. Lumbierres, B. Donnadieu, S. Sabo-Etienne and
B. Chaudret, J. Am. Chem. Soc., 2002, 124, 5624.
21 C. Bianchini, A. Meli, S. Moneti and F. Vizza, Organometallics, 1998,
17, 2636.
The reaction of 3 with CO, and with CO and H2
These reactions were also carried out in NMR tubes fitted with
suba-seals. Needles dipping into the solution were used to supply
22 S. B. Duckett, J. C. Lowe, J. P. Lowe and R. J. Mawby, Dalton Trans.,
2004, 3788.
This journal is
The Royal Society of Chemistry 2006
Dalton Trans., 2006, 2661–2670 | 2669
©

View Article Online
23 B. Chamberlain, S. B. Duckett, J. P. Lowe, R. J. Mawby and J. C. Stott,
Dalton Trans., 2003, 2603.
24 R. K. Harris, Can. J. Chem., 1964, 42, 2275.
25 R. K. Harris, Inorg. Chem., 1966, 5, 701.
38 R. F. Heck and D. S. Breslow, J. Am. Chem. Soc., 1961, 83,
4023.
39 R. A. Sanchez-Delgado, J. S. Bradley and G. Wilkinson, J. Chem. Soc.,
Dalton Trans., 1976, 399.
26 A. J. Deeming and B. L. Shaw, J. Chem. Soc. A, 1970, 3356.
27 J. Keeler, Chem. Soc. Rev., 1990, 19, 381.
28 A. E. Derome, Modern NMR Techniques for Chemistry Research,
Pergamon, Oxford, 1987.
40 K. Ohno and J. Tsuji, J. Am. Chem. Soc., 1968, 90, 99.
41 C. Bianchini, A. Meli, M. Peruzzini, F. Vizza and F. Bachechi,
Organometallics, 1991, 10, 820.
42 D. Milstein, W. C. Fultz and J. C. Calabrese, J. Am. Chem. Soc., 1986,
108, 1336.
43 M. Gerisch, C. Bruhn, A. Vyater, J. A. Davies and D. Steinborn,
Organometallics, 1998, 17, 3101.
44 W. D. Jones, J. M. Huggins and R. G. Bergman, J. Am. Chem. Soc.,
1981, 103, 4415.
45 C. F. J. Barnard, J. A. Daniels and R. J. Mawby, J. Chem. Soc., Dalton
Trans., 1979, 1331.
46 C. Bianchini, P. J. Perez, M. Peruzzini, F. Zanobini and A. Vacca, Inorg.
Chem., 1991, 30, 279.
47 M. V. Baker and L. D. Field, J. Chem. Soc., Chem. Commun., 1984,
996.
29 S. Braun, H.-O. Kalinowski and S. Berger, 150 and More Basic NMR
Experiments, Wiley-VCH, New York, 2nd edn, 1998.
30 M. Helliwell, J. D. Vessey and R. J. Mawby, J. Chem. Soc., Dalton
Trans., 1994, 1193.
31 E. Rivard, A. J. Lough and I. Manners, J. Chem. Soc., Dalton Trans.,
2002, 2966.
32 J. M. Bray and R. J. Mawby, J. Chem. Soc., Dalton Trans., 1987, 2989.
33 R. S. Berry, J. Chem. Phys., 1960, 32, 933.
34 E. Lindner, M. Gepra¨gs, K. Gierling, R. Fawzi and M. Steimann, Inorg.
Chem., 1995, 34, 6106.
35 D. R. Saunders, M. Stephenson and R. J. Mawby, J. Chem. Soc., Dalton
Trans., 1984, 1153.
36 D. F. Christian, G. R. Clark, W. R. Roper, J. M. Waters and K. R.
Whittle, J. Chem. Soc., Chem. Commun., 1972, 458.
37 J. T. Gauntlett, B. E. Mann, M. J. Winter and S. Woodward, J. Chem.
Soc., Dalton Trans., 1991, 1427.
48 N. Merle, G. Koicok-Ko¨hn, M. F. Mahon, C. G. Frost, G. D. Ruggerio,
A. S. Weller and M. C. Willis, Dalton Trans., 2004, 3883.
49 R. T. Baker, J. C. Calabrese, S. A. Westcott and T. B. Marder, J. Am.
Chem. Soc., 1995, 117, 8777.
50 C. A. Tolman, Chem. Rev., 1977, 77, 313.
2670 | Dalton Trans., 2006, 2661–2670
This journal is
The Royal Society of Chemistry 2006
©