Phosphine - Amido complexes of ruthenium and mechanistic implications for ketone transfer hydrogenation catalysis

Article abstract of DOI:10.1021/om2004173

The air-sensitive phosphine-anilido complexes [RuR(η⁶-p- cymene)(P,N-Ph₂PAr^-)] (R = H, Et; Ar^- = o-C₆H₄NMe^-) have been prepared. While the precursor [RuCl(η⁶-p-cymene)(P,N-Ph₂PAr^-)] is a moderately active ketone transfer hydrogenation catalyst under basic conditions, the hydrido derivative is much less active, ruling out the possibility of an inner-sphere mechanism during catalysis. The possibility of an alternate mechanism, related to the Meerwein-Ponndorf-Verley-Oppenauer (MPVO) pathway, is discussed. While attempts to isolate intermediate alkoxo derivatives demonstrate their propensity toward β-hydride elimination to afford the hydrido complex, the ethyl derivative is remarkably stable, even in refluxing benzene, providing an interesting contrast between the labilities of the Ru-O and Ru-C bonds.

Full text of DOI:10.1021/om2004173

ARTICLE
pubs.acs.org/Organometallics
PhosphineꢀAmido Complexes of Ruthenium and Mechanistic
Implications for Ketone Transfer Hydrogenation Catalysis
Lindsay J. Hounjet,† Michael J. Ferguson,†,‡ and Martin Cowie*,†
†Department of Chemistry and ‡X-ray Crystallography Laboratory, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
S Supporting Information
b
ABSTRACT: The air-sensitive phosphineꢀanilido complexes
[RuR(η6-p-cymene)(P,N-Ph2PArꢀ)] (R = H, Et; Arꢀ
=
o-C6H4NMeꢀ) have been prepared. While the precursor
[RuCl(η6-p-cymene)(P,N-Ph2PArꢀ)] is a moderately active
ketone transfer hydrogenation catalyst under basic conditions,
the hydrido derivative is much less active, ruling out the
possibility of an inner-sphere mechanism during catalysis. The
possibility of an alternate mechanism, related to the Meer-
weinꢀPonndorfꢀVerleyꢀOppenauer (MPVO) pathway, is
discussed. While attempts to isolate intermediate alkoxo deri-
vatives demonstrate their propensity toward β-hydride elimina-
tion to afford the hydrido complex, the ethyl derivative is remarkably stable, even in refluxing benzene, providing an interesting
contrast between the labilities of the RuꢀO and RuꢀC bonds.
he transfer hydrogenation of ketones represents a useful
means of producing value-added alcohols under relatively
act as a scaffold, upon which the alkoxide (again, produced by adding
base to the reaction mixture) can transfer its hydride directly to the
electrophilic carbonyl functionality of a coordinated ketone, thereby
avoiding the intermediacy of a hydrido complex.
T
benign conditions, and many effective catalyst systems involving
phosphorus- and nitrogen-ligated ruthenium complexes have
been developed for this purpose.1ꢀ5 These metal-catalyzed
reactions often rely on the inclusion of a strong base in reaction
mixtures, either to activate the metal-containing precatalyst or to
play a more direct role as a cocatalyst.6
Although most transfer hydrogenation reactions catalyzed by
late-metal complexes can be rationalized on the basis of inner- or
outer-sphere mechanisms, one possible exception is Stradiotto’s
highly active zwitterionic catalyst [RuCl(η6-p-cymene)(P,N-1-
PiPr2-2-NMe2-C9H5ꢀ)],9 for which neither mechanism seems
applicable. Previously, we showed that an o-phosphinoanilido
complex of ruthenium, namely [RuCl(η6-p-cymene)(P,N-
Ph2PArꢀ)] (1; Arꢀ = o-C6H4NMeꢀ), not unlike Stradiotto’s
catalyst, functions as a moderately active ketone transfer hydro-
genation catalyst in the presence of base.10 The necessity for base
in conjunction with our amido catalyst appeared to rule out an
outer-sphere mechanism, but at the time, we did not investigate
this further. In the current work, we highlight new evidence that
also discounts an inner-sphere process as the catalytically domi-
nant pathway and discuss the possible operation of the less
explored (MPVO) alternative.
The two mechanisms most commonly involved in the transfer
hydrogenation of ketones are the inner-sphere and outer-sphere
cycles. During an inner-sphere mechanism (Scheme 1, left cycle),
the reagent ketone (acetophenone in the example shown) can be
inserted into an MꢀH bond (generated by β-hydride elimina-
tion from the isopropoxo group) with concomitant elimination
of acetone, to form a new alkoxide that is then protonated by the
incoming reagent alcohol, releasing the product alcohol from the
metal. In the outer-sphere mechanism, initially proposed by
Noyori,1 the inclusion of a strongly basic amido ligand within
ruthenium-containing catalysts can allow for deprotonation of the
alcohol (iPrOH in this example) by the nucleophilic nitrogen
donor with simultaneous hydride transfer to the adjacent metal
atom via a highly ordered transition state (Scheme 1, right cycle).
Transfer of the proton and hydride from the amine and metal,
respectively, to the polar substrate (acetophenone) then generates
the product alcohol (1-phenylethanol). In this case, the noninno-
cent amido ligand renders the catalyst exceptionally reactive with-
out requiring the use of an external base in reaction media.1
A less common mechanism that has been proposed for aluminum-
and tin-catalyzed processes but is rarely mentioned in the context of
late-metal systems is the MeerweinꢀPonndorfꢀVerleyꢀOppenauer
(MPVO) mechanism (Scheme 2).7,8 In this case the metal center can
’ EXPERIMENTAL SECTION
General Comments. All solvents were deoxygenated, dried (using
appropriate drying agents), distilled before use, and stored under
nitrogen. All reactions were performed under an Ar atmosphere using
standard Schlenk techniques. Isopropyl alcohol (iPrOH; >99%, distilled
over Mg turnings and stored under Ar) and acetophenone (99%,
deoxygenated and stored under Ar over 5 Å molecular sieves), used
Received: May 18, 2011
Published: July 12, 2011
r
2011 American Chemical Society
4108
dx.doi.org/10.1021/om2004173 Organometallics 2011, 30, 4108–4114
|

Organometallics
ARTICLE
Scheme 1. Inner- (Left Cycle) and Outer-Sphere (Right Cycle) Mechanisms of Transfer Hydrogenationa
a The reaction of acetophenone with iPrOH occurs upon the addition of base (tBuOK) to a metal halide precatalyst.
for transfer hydrogenation catalysis, as well as ethylmagnesium bromide
(EtMgBr; 3.0 M in diethyl ether) and lithium triethylborohydride
(Li[HBEt3]; 1.0 M in tetrahydrofuran) were purchased from Aldrich.
Sodium borohydride (NaBH4; 98%) was purchased from Strem Chemi-
Scheme 2. MPVO Mechanism of Transfer Hydrogenation
cals. The complex [RuCl(η6-p-cymene)(P,N-Ph2PArꢀ)] (1, Arꢀ
=
C6H4NMeꢀ), was prepared as previously reported.10 NMR spectra
were recorded on Varian Inova-400 and -500 and Varian Unity-500
spectrometers operating at 399.8, 498.1, and 499.8 MHz, respectively,
for 1H, at 161.8, 201.6, and 202.3 MHz, respectively, for 31P and at 100.6,
125.3, and 125.7 MHz, respectively, for 13C nuclei. J values are given in
hertz (Hz). Overlapping or unresolved aromatic signals, observed in the
typical 6ꢀ8 ppm range in the 1H NMR spectrum and found between 80
and 120 ppm in the 13C{1H} NMR spectrum, are not reported.
Spectroscopic data for complexes 2 and 3 are provided in Table 1.
The elemental analysis of 3 was performed by the Microanalytical
Laboratory within the Department.
Schlenk flask under anhydrous conditions and Ar atmosphere, 1 (84 mg,
149 μmol) was dissolved in 5 mL of benzene at ambient temperature and
the solution was stirred for 5 min. Ethylmagnesium bromide (3.0 M in
diethyl ether, 60 μL, 180 μmol) was added to the dark burgundy solution
via syringe to produce a cloudy, red mixture, which was stirred for 5 min
before removing the solvents in vacuo. While the mixture was stirred,
10 mL of benzene was added to produce a cloudy, orange-red mixture that
was then filtered through a Celite plug in a glass pipet (via Ar overpressure
through a cannula) into a prepared 25 mL Schlenk tube. The solvent
volume was reduced to approximately 5 mL in vacuo, and the red solution
was layered with 10 mL of n-pentane and left undisturbed for 18 h. The
supernatant was then removed via cannula, and the red crystalline product
was dried in vacuo (50 mg, 60% yield). Anal. Found: C, 66.73; H, 6.68; N,
2.66. Calcd for [C31H36NPRu]: C, 67.13; H, 6.54; N, 2.53%.
Preparation of Metal Complexes. (a). Reaction of [RuCl(η6-p-
cymene)(P,N-Ph2PArꢀ)] (1) with Li[HBEt3]. An NMR tube was charged
with 1 (22 mg, 39 μmol), sealed with a septum, and then evacuated and
back-filled with Ar three times. The compound was dissolved in 0.7 mL
of C6D6, producing a dark burgundy solution, and a 1.0 M solution of
Li[HBEt3] in tetrahydrofuran (42 μL, 42 μmol) was added via a
microliter syringe, resulting in a red mixture that was left for several
hours, allowing a white precipitate to settle from the bright red solution.
The supernatant was transferred via cannula under Ar to a prepared
NMR tube, and the sample was then analyzed by 1H and 31P{1H} NMR
spectroscopy. The resulting spectra indicated the formation of two
products, [RuH(η6-p-cymene)(P,N-Ph2PArꢀ)] (2) and [RuEt(η6-p-
cymene)(P,N-Ph2PArꢀ)] (3), in approximately equal proportions as
judged by the intensities of the only two 31P signals observed (at δ 70.1
and 68.7; see parts b and c for independent syntheses).
(d). Reactions of [RuCl(η6-p-cymene)(P,N-Ph2PArꢀ)] (1) with Alk-
oxides. In a 50 mL Schlenk flask under anhydrous conditions and Ar
atmosphere, 1(50mg, 89μmol) was dissolved in 5 mL of benzene at ambient
temperature with stirring. In a 25 mL Schlenk flask, KOH (5 mg, 90 μmol)
was dissolved in 5 mL of alcohol (either methanol or isopropyl alcohol) at
ambient temparature with stirring, and this solution was then transferred (via
Ar overpressure through a cannula) to the solution of 1 and the mixture was
stirred for 10 min at ambient temperature before removing the solvents in
vacuo. Benzene (10 mL) was then added to the resulting dark red residue, and
the mixture was filtered through a Celite plug in a glass pipet (via Ar
overpressure through a cannula) into a prepared 50 mL Schlenk flask. The
solvent was removed in vacuo, and the residue was dissolved in 1 mL of C6D6
and the solution transferred to a prepared NMR tube via cannula. NMR
spectra showed the exclusive formation of [RuH(η6-p-cymene)(P,N-
Ph2PArꢀ)] (2), when either alcohol (methanol or isopropyl alcohol)
was used. Similar reactions performed using tBuOK rather than KOH as
the base yielded identical spectroscopic observations.
(b). Hydrido(η6-p-cymene)(P,N-diphenyl(o-N-methylanilido)phos-
phine)ruthenium(II), [RuH(η6-p-cymene)(P,N-Ph2PArꢀ)] (2). In a
25 mL Schlenk tube under anhydrous conditions and Ar atmosphere,
[RuCl(η6-p-cymene)(P,N-Ph2PArꢀ)] (1; 177 mg, 315 μmol) and
sodium borohydride (26 mg, 687 μmol) were dissolved in 10 mL of
methanol with stirring. The mixture, which initially turned blue-green,
turned dark red after 10 min, and was stirred for a total of 30 min before
removing the solvent in vacuo. Benzene (10 mL) was added to the
remaining residue, and the resultant slurry was stirred for 10 min before
allowing the precipitate to settle. The supernatant was filtered through a
Celite plug and transferred to a prepared 50 mL Schlenk flask via cannula
transfer under Ar. The solvent was removed in vacuo, resulting in an
amorphous, red residue (156 mg). Although a satisfactory elemental
analysis could not be obtained for this compound, its NMR spectra (1H
NMR spectrum provided as Supporting Information) leave no doubt
about its formulation.
Ketone Transfer Hydrogenation Assay with [RuH(η6-p-
cymene)(P,N-Ph2PArꢀ)] (2). In a 50 mL Schlenk flask under
anhydrous conditions and Ar atmosphere, [RuCl(η6-p-cymene)-
(P,N-Ph2PArꢀ)] (1; 50.0 mg, 89.0 μmol) and NaBH4 (5.1 mg, 130 μmol)
(c). Ethyl(η6-p-cymene)(P,N-diphenyl(o-N-methylanilido)phosph-
ine)ruthenium(II), [RuEt(η6-p-cymene)(P,N-Ph2PArꢀ)] (3). In a 50 mL
4109
dx.doi.org/10.1021/om2004173 |Organometallics 2011, 30, 4108–4114

Organometallics
ARTICLE
Table 1. 31P{1H}, 1H, and 13C{1H} NMR Data for Ruthenium Compoundsa
δ(31P{1H})/ppmb
δ(1H)/ppmc
δ(13C{1H})/ppmc
[RuH(η6-p-cymene)(P,N-Ph2PArꢀ)] (2)
70.1 (s)
NCH3: 3.30 (s, 3H)
CH(CH3)2: 2.15 (sept, 3JHH = 6.8 Hz, 1H)
NCH3: 48.9 (s)
CH(CH3)2: 32.3 (s)
ArCH3: 1.56 (s, 3H)
CH(CH3)2: 24.1 (s), 23.6 (s)
ArCH3: 19.0 (s)
CH(CH3)2: 1.06 (d, 3JHH = 6.8 Hz, 3H), 1.04 (d, 3JHH = 6.8 Hz, 3H)
RuH: ꢀ7.96 (d, 2JPH = 46.5 Hz, 1H)
[RuEt(η6-p-cymene)(P,N-Ph2PArꢀ)] (3)
68.7 (s)
NCH3: 3.26 (s, 3H)
NCH3: 48.3 (s)
CH(CH3)2: 2.13 (sept, 3JHH = 6.8 Hz, 1H)
ArCH3: 1.68 (s, 3H)
CH2CH3: 1.38 (pt, 3JHH = 7.5 Hz, 3H)
CH2CH3: 1.10 (m, 1H), 0.80 (m, 1H)
CH(CH3)2: 0.96 (d, 3JHH = 6.8 Hz, 3H), 0.94 (d, 3JHH = 6.8 Hz, 3H)
CH(CH3)2: 31.6 (s)
CH(CH3)2: 24.2 (s), 22.6 (s)
CH2CH3: 23.4 (d, 3JPC = 4 Hz)
ArCH3: 17.5 (s)
CH2CH3: 14.0 (d, 2JPC = 14 Hz)
a NMR abbreviations: s = singlet, d = doublet, t = triplet, m = multiplet, sept = septet, p = pseudo. NMR data were recorded at 27 °C in C6D6. b 31
P
chemical shifts are referenced to external 85% H3PO4. c 1H and 13C chemical shifts are referenced to external tetramethylsilane. Chemical shifts for aryl
groups are not given.
were dissolved in 7.0 mL of MeOH at ambient temperature and stirred
for 15 min. The solvent was removed in vacuo, acetophenone (10.4 mL,
89.0 mmol) was added to the residue, and the cloudy, red mixture was
stirred for 15 min. The precipitate (NaCl) was allowed to settle from the
red solution, and a 2.08 mL aliquot (17.8 μmol of 2 dissolved in 17.8
mmol of acetophenone) was withdrawn via Gastight syringe and
transferred to a prepared 50 mL three-necked, round-bottom flask
equipped with a 0.5 in. stir bar and attached reflux condenser through
which an Ar overpressure was applied. While it was stirred, the solution
Scheme 3. Synthesis of Hydrido (2) and Ethyl (3)
Compounds
t
was heated to 90 °C for 10 min. A solution of BuOK (8.0 mg,
71.3 μmol) in iPrOH (13.6 mL, 178 mmol) was then added via cannula
transfer under Ar. A 1.0 mL aliquot of the reaction mixture was
immediately withdrawn and passed through a column containing 2 cm
of acidic alumina atop 2 cm of Florisil and collected in a vial so that less
than 30 s elapsed between removal of the sample from the mixture and
removal of catalyst from the sample. The vial was then capped and stored
at 0 °C until the mixture could be analyzed by NMR spectroscopy and
GC-EI-MS. Aliquots were withdrawn and treated as described above at
5, 60, and 120 min relative to the addition of iPrOH and tBuOK.
X-ray Structure Determination. Data were collected using Mo
KR radiation (λ = 0.710 73 Å) on a Bruker APEX-II CCD detector/D8
diffractometer11 with the crystal of 3 cooled to ꢀ100 °C. The data were
corrected for absorption through Gaussian integration from indexing of
the crystal faces. The structure was solved using direct methods
(SHELXS-97).11 Refinements were completed using the program
SHELXL-97.12 Hydrogen atoms were assigned positions based on the
sp2 or sp3 hybridization geometries of their attached carbon atoms and
were given thermal parameters 20% greater than those of their parent
atoms. A summary of the crystallographic experimental details for
[RuEt(η6-p-cymene)(P,N-Ph2PArꢀ)] (3) is provided as Supporting
Information.
operate in the absence of base by catalytic involvement of the
amido donor.1 We speculated that the active catalyst may form by
β-hydride elimination from a coordinated alkoxide to produce
the hydrido complex [RuH(η6-p-cymene)(P,N-Ph2PArꢀ)] (2,
Scheme 3), which could operate by an inner-sphere mechanism.
In the present study, we have attempted to determine the role of
base in the transfer hydrogenation reaction and to establish
whether or not 2 is a catalytically relevant intermediate.
Attempts to prepare the hydride species by reacting com-
pound 1 with 1 equiv of Li[HBEt3] under ambient conditions in
C6D6 led to the formation of two compounds in approximately
equimolar quantities; in addition to the targeted hydride (2), the
ethyl product [RuEt(η6-p-cymene)(P,N-Ph2PArꢀ)] (3) was also
obtained. Ethyl group transfer from superhydride to ruthenium
has previously been demonstrated in similar systems.13,14 Each
compound can be prepared independently as the sole product by
reaction of 1 with either NaBH4 in methanol to give the hydride
’ RESULTS AND DISCUSSION
In earlier work, we reported that the phosphineꢀamido
complex [RuCl(η6-p-cymene)(P,N-Ph2PArꢀ)] (1; Arꢀ = o-
C6H4NMeꢀ) functions as a ketone transfer hydrogenation
catalyst and found that the reaction occurred only in the presence
of tBuOK.10 This observation appeared to eliminate the possibi-
lity of an outer-sphere hydrogenation mechanism that should
4110
dx.doi.org/10.1021/om2004173 |Organometallics 2011, 30, 4108–4114

Organometallics
ARTICLE
Table 2. Selected Structural Parameters for Compound 3
Bond Lengths (Å)
RuꢀP
2.2837(6)
2.102(2)
2.319(2)
2.261(2)
2.238(2)
2.228(2)
2.186(2)
2.321(2)
2.154(2)
1.795(2)
PꢀC(21)
1.840(2)
1.833(2)
1.348(2)
1.430(3)
1.430(3)
1.378(3)
1.394(3)
1.387(3)
1.397(3)
1.517(3)
RuꢀN
PꢀC(31)
RuꢀC(1)
RuꢀC(2)
RuꢀC(3)
RuꢀC(4)
RuꢀC(5)
RuꢀC(6)
RuꢀC(18)
PꢀC(11)
NꢀC(12)
C(11)ꢀC(12)
C(12)ꢀC(13)
C(13)ꢀC(14)
C(14)ꢀC(15)
C(15)ꢀC(16)
C(11)ꢀC(16)
C(18)ꢀC(19)
Bond Angles (deg)
PꢀRuꢀN
PꢀRuꢀC(18)
81.30(5)
82.97(6)
NꢀRuꢀC(18)
RuꢀC(18)ꢀC(19)
82.52(8)
114.7(2)
solution with n-pentane. A crystallographic analysis of 3
(Figure 1) shows the chelating phosphineꢀamido ligand having
the characteristic planar geometry of the amido group,15 as
demonstrated by the sum of the angles at nitrogen (359.7°).
The ethyl group appears normal, and the lack of a β-agostic
interaction, made evident by the expanded RuꢀC(18)ꢀC(19)
Figure 1. ORTEP diagram of [RuEt(η6-p-cymene)(P,N-Ph2PArꢀ)]
(3). Gaussian ellipsoids for all non-hydrogen atoms are depicted at the
20% probability level. Hydrogen atoms are shown with arbitrarily small
thermal parameters for the methyl, methylene, and methine groups and
are not shown for the aryl rings.
angle (114.7(2)°, Table 2) and the Ru H separation of greater
3 3 3
than 3.2 Å, illustrates its k1 geometry in the solid state and is
consistent with coordinative saturation of the complex. The
length of the RuꢀC(18) bond (2.154(2) Å) appears normal,
being comparable to the RuꢀC distances within [RuEt(η6-
C6Me6)(S,S0-N(P(iPr)2S)2] (2.133(3) Å)13 and [RuEt2((5,50-
tBu)2-2,20-Bipy)2] (2.138(7) and 2.142(8) Å),16 and the
C(18)ꢀC(19) distance (1.517(3) Å) is also characteristic of a
CꢀC single bond. The RuꢀN distance (2.102(2) Å) is longer
than that within the coordinatively unsaturated anilido-contain-
ing complex [RuH(PPh3)(P,N,N0-Ph2PCH2P(Ph2)dN-(o)-
C6H4NHꢀ)] (2.031(2) Å,)17 suggesting that the presence of
the larger N-methyl group within our hexacoordinate complex
may hinder effective coordination of the amido nitrogen. This
suggestion is supported by the structural parameters for 3, which
show that repulsion between the p-cymene isopropyl and amido
methyl groups results not only in the somewhat elongated RuꢀN
bond but also in the unsymmetrical binding of the p-cymene
ligand, in which the RuꢀC(1) and RuꢀC(6) distances
(2.319(2) and 2.321(2) Å, respectively), adjacent to the iso-
propyl group, are elongated compared with the other Ruꢀ
Ccymene distances (2.186(2)ꢀ2.261(2) Å). Nevertheless, the
RuꢀN bond in 3 is still shorter than that within the related
phosphineꢀamine complex [RuCl(η6-p-cymene)(P,N-Ph2PAr)]Cl
(2.172(2) Å)10 and those within a series of related amine species,
[RuCl(η6-p-cymene)(P,N-Ph2PC6H4NH2)]+ (2.125ꢀ2.146 Å),18
consistent with this anionic amido group functioning as a better
donor than the amine and also possibly reflecting some degree of
π-donor character to ruthenium by the amido lone pair. It is
also noteworthy that the nitrogenꢀarene bond length within
amido complex 3 (NꢀC(12) = 1.348(2) Å) is much shorter than
those within the above amine compounds (1.458(4)10 and
1.422ꢀ1.460 Å18), suggesting additional delocalization of the
amido lone pair onto the aromatic ring. This proposed resonance
delocalization of the lone pair in 3 is further illustrated by
elongation of the aromatic CꢀC bonds involving the carbon
(2) or with EtMgBr in benzene to give the ethyl species (3), as
outlined in Scheme 3.
Compound 2 is represented by a signal at δ 70.1 in the
31P{1H} NMR spectrum, while the hydride signal is found at δ
ꢀ7.96 in the 1H spectrum as a doublet with 2JPH = 46.5 Hz (see
the Supporting Information). All other proton resonances
(Table 1) are similar to those of the precursor chloro species
(1)10 and are fully consistent with the proposed geometry.
Compound 3 shows a slightly more upfield chemical shift in
1
the 31P{1H} NMR spectrum at δ 68.7, while the H NMR
spectrum shows three resonances for the ethyl ligand, with a
pseudotriplet at δ 1.38 (3JHH = 7.5 Hz), representing the methyl
group, and two multiplets at δ 1.10 and 0.80, representing the
diastereotopic protons of the methylene unit. Compound 2
decomposes in CD2Cl2, reverting predominantly to 1 (among
other unidentified species) with concomitant production of
CHD2Cl (as made evident by a pentet signal at δ 3.06 in the
1H NMR spectrum) equal to the amount of 1 regenerated.
Numerous attempts to purify and isolate 2 as an analytically
pure solid were unsuccessful, since this product is quite unstable.
For example, attempts to crystallize the complex from dark red
(nearly black) benzene solutions using n-pentane, while employ-
ing our best efforts to provide completely inert conditions,
produced a pale green (almost white) precipitate that was much
less soluble in benzene, the NMR spectra for which reveal
decomposition to numerous unidentified species. However, a
benzene solution of 2 handled under strictly inert conditions
could be concentrated in vacuo to a dark red amorphous residue,
which was then quickly redissolved in anhydrous acetophenone
and iPrOH, in which the compound is stable, allowing us to study
its capabilities as a catalyst.
While attempts to crystallize complex 2 were unsuccessful,
compound 3, which is also highly air-sensitive, can be easily
isolated as large red crystals by layering a saturated benzene
4111
dx.doi.org/10.1021/om2004173 |Organometallics 2011, 30, 4108–4114

Organometallics
ARTICLE
atom ipso to nitrogen (C(11)ꢀC(12) and C(12)ꢀC(13) =
1.430(3) Å) in comparison to other CꢀC bonds within the same
group (ca. 1.39 Å) or adjacent phenyl groups (ca. 1.39 Å). The
delocalization of the amido lone pair onto the aryl ring has been
previously noted in a related series of phosphineꢀamido com-
plexes of Rh and was proposed to rationalize the low basicity of
the amido nitrogen in these species.15 A comparison with the
structure of 110 shows a similar trend in the amidoꢀaryl bond
lengths in this species, and the corresponding amido lone-pair
delocalization is presumably the reason this transfer hydrogena-
tion catalyst does not operate via an outer-sphere mechanism.
Also consistent with the electron-withdrawing character of the
anilido ring in 3 is the significantly shorter PꢀC(11) bond
(1.795(2) Å) compared to the two Pꢀphenyl distances
(1.840(2), 1.833(2) Å), indicating some resonance contribution
from a phosphonium ylide structure, as previously discussed.15
The electron-withdrawing arene also enhances the imine char-
acter of the adjacent nitrogen atom, making it a less effective
donor to the metal and possibly contributing to the slightly
longer RuꢀN bond discussed above.
Table 3. Transfer Hydrogenation of Acetophenone with
Catalysts 1 and 2
entry
complex
trxn (min)
conversion (%)a
TOF (hꢀ1 b
)
1c
2c
3c
4
1
1
1
2
2
2
5
60
13
34
46
1
1560
340
230
120
70
120
5
5
60
7
6
120
9
45
a Determined by GC and 1H NMR analyses. b Turnover frequency
determined at the corresponding reaction time (trxn) in column 3.
c Previously published results.10 Reaction conditions: temperature
90 °C; Ru/tBuOK/acetophenone/iPrOH = 1/4/1000/10 000.
aldehyde or a ketone. The involvement of such an elimination
process could explain the stability of the ethyl complex, which
contains a less polar and less labile RuꢀC bond. The dissociation
of alkoxide anion from ruthenium is consistent with our earlier
inference that chloro complex 1 undergoes enantiomerization in
dichloromethane solution via chloride ion dissociation to form a
coordinatively unsaturated intermediate.10
Under strictly inert conditions, compound 3 displays remark-
able stability, even in refluxing benzene. This species’ apparent
resistance to β-hydride elimination inspired our attempts to
prepare analogous alkoxo complexes, [Ru(OR)(η6-p-cymene)-
(P,N-Ph2PArꢀ)] (R = iPr, Me), which should be generated upon
the addition of a base to 1 in alcohols. However, numerous
attempts to prepare such derivatives by reactions of 1 with KOH
in MeOH or with 1 equiv or more of tBuOK or iPrONa in iPrOH
consistently reveal the presence of 2, due to its apparent
formation from the putative, but spectroscopically unobserved,
target alkoxide intermediates [Ru(OR)(η6-p-cymene)(P,N-
Ph2PArꢀ)]. Unfortunately, attempts to prepare an alkoxo com-
plex, not having a β-hydride, by reaction of 1 with 1 equiv of
tBuOK in C6D6 in the absence of alcohol resulted in a complex
mixture of products, as made evident by 31P{1H} NMR analysis.
Despite the stability of the ethyl derivative 3, it appears that
β-hydride elimination from alkoxo ligands of transient species,
[Ru(OR)(η6-p-cymene)(P,N-Ph2PArꢀ)], occurs (at least in the
absence of reagent ketone) over the time it takes to carry out
reactions and obtain NMR spectra. In this case, a classical
β-hydride elimination from the coordinated alkoxide would
require the generation of a vacant coordination site, either by
ring slippage of the η6-arene19 or by dissociation of either the
phosphorus or nitrogen donors. Although it may seem unlikely
that dissociation of the anionic amido donor would occur, the
presence of an electron-withdrawing arene seems to enhance the
imine character of the adjacent nitrogen atom, as noted above,
making it a less effective donor to the metal. Furthermore, amide
dissociation from Ru could be promoted by potassium ion
present.20 Ring slippage also seems viable on the basis of the
crystallographically observed distortions of the coordinated
arene in 3 (discussed above). However, it is difficult to rationalize
on these grounds why β-hydride elimination occurs so rapidly
from alkoxides, while the ethyl substituent of 3 is so robust.
Milstein et al. have offered another rationale for the apparent
β-hydride elimination from a coordinatively saturated alkoxo
complex.21 This proposal, adapted to our system, involves
dissociation of the alkoxide from the metal, allowing the substrate
alcohol to form a CꢀH σ complex while the alkoxide anion is
stabilized by exogenous alcohol. Subsequent deprotonation of
the alcohol by the alkoxide, accompanied by hydride transfer
from the alcohol to Ru, generates the RuꢀH bond and either an
A catalytic study of 2 showed its significantly reduced activity
relative to that of its precursor 1 for acetophenone transfer
hydrogenation in the presence of catalytic 4 equiv of tBuOK (see
Table 3), with turnover frequencies an order of magnitude less
for the hydrido species 2 compared to the chloro precursor 1 in
the presence of alkoxide ion.10 Such an observation suggests that
the hydrido complex 2 may represent a means of catalyst
deactivation, rather than the active species in the cycle involving
1. However, the fact that we do observe some (albeit significantly
reduced) substrate conversion in the presence of 2 could indicate
that partial decomposition of the unreactive hydrido species
occurs during its preparation or under the forcing reaction
conditions to generate small amounts of a more active species.
Alternatively, the small amount of conversion that we observe
using 2 as the catalyst could reflect the operation of a very slow
inner-sphere process. Consistent with our observations, Stra-
diotto et al. have recently reported a catalyst system comprised of
a zwitterionic indenyl species, [RuCl(η6-p-cymene)(P,N-1-
PiPr2-2-NMe2-C9H5ꢀ)], that is catalytically active in the pre-
sence of base and have also ruled out the involvement of the
related hydrido complex [RuH(η6-p-cymene)(P,N-1-PiPr2-2-
NMe2-C9H5ꢀ)], in transfer hydrogenation reactions carried
out under conditions similar to our own.9 The fact that these
hydrido complexes have low activity in both cases indicates that
an inner-sphere mechanism (Scheme 1, left cycle) is not the
catalytically dominant pathway for both 1 and Stradiotto’s
compound. The catalytic dependence on external base in our
reaction also rules out the involvement of an outer-sphere
mechanism for transfer hydrogenation,10 which is certainly not
possible for Stradiotto’s system. Pelagatti et al. have also exam-
ined the related amine species [RuCl(η6-p-cymene)(P,N-Ph2P-
(o-C6H4NH2)]Cl, as a ketone transfer hydrogenation catalyst,
which is also dependent on an excess of external base.18 How-
ever, in this case, base is required for conversion of the amine to
the active amido complex. Although it is possible that proton-
ation of the amido group in 1 occurs in alcohol and that base is
required to regenerate the active amido group, we have ruled out
this possibility, since compound 1 shows no evidence for
conversion to an amine in alcohols.
4112
dx.doi.org/10.1021/om2004173 |Organometallics 2011, 30, 4108–4114

Organometallics
ARTICLE
pure 2-butanol as the hydrogen donor and probing for enantio-
meric excess as additional support for an MPVO mechanism.
However, we did not conduct the experiment, because it became
apparent that this would fail to rule out an inner-sphere process
(some degree of enantioselectivity should be expected for either
pathway). Our chiral precatalyst 1 exists as a racemate; therefore,
coordination of an enantiomerically pure alkoxide may favor one
of two diastereoisomeric alkoxo complexes. In the case of an
MPVO mechanism, the resulting stereochemical influence on a
coordinated prochiral ketone is apparent. In the case of an inner-
sphere mechanism, two energetically distinct β-hydride elimina-
tion pathways also exist, one of which may be more kinetically
accessible, thereby favoring formation of one enantiomer of 2
over the other. The resulting stereochemical bias prevents us
from discounting an inner-sphere mechanism on the basis of the
observation of an enantiomeric excess. However, in the case of
Noyori’s complex, the fact that the enantioselectivity does not
differ when either racemic or enantiomerically pure alcohols are
used appears to rule out both inner-sphere and MPVO hydrogen
transfer processes. A comparison of the catalytic features of
complex 1 (containing a relatively inert amido functionality)15 to
those of Noyori’s catalyst shows how differences in the basicity of
amido groups can influence their involvement during catalysis,
resulting in entirely different reaction mechanisms.
Figure 2. Possible variations of an MPVO transition state.
A remaining mechanistic possibility that deserves consideration
is the MeerweinꢀPonndorfꢀVerleyꢀOppenauer (MPVO)2,7,8
pathway, in which an alkoxo ligand transfers its hydride directly
to a coordinated ketone without the intermediacy of a metal-bound
hydrogen atom (Scheme 2). In our case, an MPVO mechanism
again requires either dissociation of the phosphorus or nitrogen
donors from the (yet unobserved) coordinatively saturated iso-
propoxo complex [Ru(iPrO)(η6-p-cymene)(P,N-Ph2PArꢀ)] or
ring slippage19 of the η6-p-cymene group to generate a second
vacant coordination site for complexation of substrate ketone. In
the event of coordinative unsaturation at this complex, binding of
the O-donor substrate acetophenone to Ru would give rise to
mutually cis ketone and isopropoxo ligands, allowing direct hydride
transfer from the isopropoxo ligand to the ketone. While the MPVO
mechanism has been proposed to operate within transfer hydro-
genation reactions catalyzed by tin22 and aluminum23 species, there
is a lack of evidence supporting its involvement within processes
catalyzed by transition metals (although this possibility has been
suggested by Morris et al.).2
As noted above for the alkoxideꢀhydride transformation, the
requirement for coordinative unsaturation in the MPVO me-
chanism is somewhat troublesome in the context of the stability
of the ethyl complex 3. However, given the requirement for base
in reaction mixtures and the possibility for stabilization of outer-
sphere alkoxide by exogenous alcohol (or even potassium ions),
as noted above, a variant of the classical MPVO mechanism
might involve replacement of the dissociated isopropoxide anion
by acetophenone and subsequent direct hydride transfer from
alkoxide to ketone via a six-membered transition state without
requiring a vacant coordination site at ruthenium, as shown in
Figure 2. Adolfsson and co-workers have also proposed an
MPVO-like transition state for a hydrido ruthenium complex
in which a dianionic peptido/alkoxo ligand chelates at the metal
while the ligand’s oxygen atom also binds a lithium ion;20
however, this mechanism operates by hydride transfer from the
ruthenium center to the substrate ketone (rather than directly
from alkoxide to ketone) and is perhaps more reminiscent of an
outer-sphere transfer. When the process is viewed as an MPVO
reaction, the OꢀRu(Ln)ꢀH moiety is analogous to an alkoxide,
which binds to a lithium ion via oxygen, as does the substrate
ketone, allowing hydride transfer from the ruthenium center to the
ketone. Whenthe reactionisconsideredasanouter-sphereprocess,
hydride and lithium ion (rather than proton) are simultaneously
transferred from ruthenium and ligand, respectively, to the ketone.
Noyori’s diamido ruthenium catalyst [Ru(η6-p-cymene)(N,-
N-(S,S)-TsDPEN2ꢀ)], which operates in the absence of base,
cannot be rationalized by an MPVO pathway, on the basis of the
observation that its enantioselectivity is not affected when a chiral
hydrogen donor is used (during such a process, a chiral alkoxo
ligand should promote enantioselective hydrogen transfer to an
adjacent prochiral ketone).1,24 One might suggest conducting a
similar experiment with our catalyst 1 using enantiomerically
’ CONCLUSIONS
Hydrido and ethyl complexes of ruthenium, containing che-
lating phosphineꢀamido ligands, have been prepared and char-
acterized. Although reaction of the chloro precursor [RuCl(η6-p-
cymene)(P,N-Ph2PArꢀ)] (1) with Li[HBEt3] does yield the
targeted hydride species, [RuH(η6-p-cymene)(P,N-Ph2PArꢀ)]
(2), the ethyl complex [RuEt(η6-p-cymene)(P,N-Ph2PArꢀ)] (3)
is also obtained in approximately equimolar amounts, providing an
interesting example of carbonꢀboron bond activation (a funda-
mental process in SuzukiꢀMiyaura coupling reactions).25
The instant generation of the hydride species 2 upon reaction
of 1 with alkoxides is in contrast to the inertness of the ethyl
complex 3 and suggests that β-hydride elimination from a
putative, formally coordinatively saturated alkoxo intermediate
is not the route to the hydride 2. Instead, we suggest that alkoxide
dissociation precedes hydride abstraction by the metal, much as
proposed by Milstein et al.21
The hydrido complex [RuH(η6-p-cymene)(P,N-Ph2PArꢀ)]
(2) is found to be much less catalytically active than the
analogous chloro species 1 for the transfer hydrogenation of
i
acetophenone with PrOH in the presence of base, suggesting
that 2 is not a catalytically relevant intermediate and that the role
of base in the catalysis by 1 is not for generation of this hydride.
The catalytic dependence on an external base and the relatively
poor activity of the hydride species allow us to rule out both
traditional outer- and inner-sphere mechanisms, respectively, as
the catalytically dominant pathway within our system. We
instead suggest the possible role of an MPVO mechanism (or
variant thereof), the involvement of which in late-metal systems
could be more prominent than previously thought.
’ ASSOCIATED CONTENT
1
S
Supporting Information. A figure giving the H NMR
b
spectrum of 2 and a table and CIF file giving crystallographic data
4113
dx.doi.org/10.1021/om2004173 |Organometallics 2011, 30, 4108–4114

Organometallics
ARTICLE
for 3. This material is available free of charge via the Internet at
http://pubs.acs.org.
Shen, J. K.; Basolo, F.; Ju, T. D.; Lang, R. F.; Kiss, G.; Hoff, C. D.
Organometallics 1994, 13, 3692–3702.
(20) V€astil€a, P.; Zaitsev, A. B.; Wettergren, J.; Privalov, T.; Adolfsson,
H. Chem. Eur. J. 2006, 12, 3218–3225.
’ AUTHOR INFORMATION
(21) Blum, O.; Milstein, D. J. Organomet. Chem. 2000, 593ꢀ594,
479–484.
Corresponding Author
*E-mail: martin.cowie@ualberta.ca. Fax: 01 7804928231. Tel.:
01 7804925581.
(22) Boronat, M.; Corma, A.; Renz, M. J. Phys. Chem. B 2006, 110,
21168–21174.
(23) Campbell, E. J.; Zhou, H.; Nguyen, S. T. Org. Lett. 2001, 3,
2391–2393.
(24) Alonso, D. A.; Brandt, P.; Nordin, S. J. M.; Andersson, P. G.
J. Am. Chem. Soc. 1999, 121, 9580–9588.
’ ACKNOWLEDGMENT
(25) Yu, Y.; Brennessel, W. W.; Holland, P. L. Organometallics 2007,
26, 3217–3226.
We thank the Natural Sciences and Engineering Research
Council of Canada (NSERC) and the University of Alberta for
financial support for this research and for providing two Queen
Elizabeth II Graduate Scholarships to L.J.H. We thank the
NSERC for funding the Bruker D8/APEX II CCD diffrac-
tometer as well as the Department’s Analytical and Instrumenta-
tion and NMR Spectroscopy Laboratories for exceptional
assistance.
’ REFERENCES
(1) Haack, K.-J.; Hashiguchi, S.; Fujii, A.; Ikariya, T.; Noyori, R.
Angew. Chem., Int. Ed. Engl. 1997, 36, 285–288.
(2) Clapham, S. E.; Hadzovic, A.; Morris, R. Coord. Chem. Rev. 2004,
248, 2201–2237.
(3) Noyori, R. Angew. Chem., Int. Ed. 2002, 41, 2008–2022.
(4) Ohkuma, T.; Hattori, T.; Ooka, H.; Inoue, T.; Noyori, R. Org.
Lett. 2004, 6, 2681–2683.
(5) Gao, J.-X.; Ikariya, T.; Noyori, R. Organometallics 1996, 15,
1087–1089.
(6) (a) Noyori, R.; Ohkuma, T. Angew. Chem., Int. Ed. 2001,
40, 40–73. (b) Abdur- Rashid, K.; Faatz, M.; Lough, A. J.; Morris,
R. H. J. Am. Chem. Soc. 2001, 123, 7473–7474.
(7) Vinzi, F.; Zassinovich, G.; Mestroni, G. J. Mol. Catal. 1983,
18, 359–366.
(8) (a) Meerwein, H.; Schmidt, R. Justus Liebigs Ann. Chem. 1925,
444, 221. (b) Ponndorf, W. Z. Angew. Chem. 1926, 39, 138. (c) Verley, A.
Bull. Chem. Soc. Chim. Fr. 1925, 37, 537. (d) Oppenauer, R. V. Recl. Trav.
Chim. Pays-Bas 1937, 56, 137. (e) Wilds, A. L. Org. React. 1944, 2, 178.
(f) Djerassi, C. Org. React. 1951, 6, 207.
(9) Lundgren, R. J.; Rankin, M. A.; McDonald, R.; Schatte, G.;
Stradiotto, M. Angew. Chem., Int. Ed. 2007, 46, 4732–4735.
(10) Hounjet, L. J.; Bierenstiel, M.; Ferguson, M. J.; McDonald, R.;
Cowie, M. Inorg. Chem. 2010, 49, 4288–4300.
(11) Programs for diffractometer operation, unit cell indexing, data
collection, data reduction, and absorption correction were those sup-
plied by Bruker.
(12) Sheldrick, G. M. Acta Crystallogr. 2008, A64, 112–122.
(13) Cheung, W.-M.; Chiu, W.-H.; Williams, I. D.; Leung, W.-H.
Eur. J. Inorg. Chem. 2009, 792–798.
(14) Arikawa, Y.; Tashita, S.-Y.; Yamaguchi, S.; Miyahara, H.;
Asayama, T.; Umakoshi, K.; Onishi, M. Chem. Lett. 2010, 39, 940–941.
(15) Hounjet, L. J.; McDonald, R.; Ferguson, M. J.; Cowie, M. Inorg.
Chem. 2011, 50, 5361–5378.
(16) Black, S. I.; Skapski, A. C.; Young, G. B. J. Chem. Soc., Chem.
Commun. 1989, 911–913.
(17) Boubekeur, L.; Ulmer, S.; Ricard, L.; Mꢀezailles, N.; Le Floch, P.
Organometallics 2006, 25, 315–317.
(18) Bacchi, A.; Balordi, M.; Cammi, R.; Elviri, L.; Pelizzi, C.;
Picchioni, F.; Verdolino, V.; Goubitz, K.; Peschar, R.; Pelagatti, P. Eur.
J. Inorg. Chem. 2008, 4462–4473.
(19) (a) Muetterties, E. L.; Bleeke, J. R.; Wucherer, E. J. Chem. Rev.
1982, 82, 499–525. (b) Bianchini, C.; Caulton, K. G.; Chardon, C.;
Eisenstein, O.; Folting, K.; Johnson, T. J.; Meli, A.; Peruzzini, M.;
Rauscher, D. J. J. Am. Chem. Soc. 1991, 113, 5127–5129. (c) Zhang, S.;
4114
dx.doi.org/10.1021/om2004173 |Organometallics 2011, 30, 4108–4114