Unraveling the catalytic cycle of tertiary phosphine oxides reduction with hydrosiloxane and Ti(O i Pr)4 through EPR and 29Si NMR spectroscopy

Article abstract of DOI:10.1021/cs4002767

The reduction of tertiary phosphine oxides using tetramethyldisiloxane (TMDS) as a mild reducing agent and catalytic amount of Ti^IV isopropoxide has been studied in detail. An extensive EPR study has revealed the presence of at least five Ti^III species, and structures have been proposed. Thus, a single electron transfer (SET) mechanism consisting of a back and forth oxido-reduction of Ti from the IV to the III oxidation state could be proposed. Reduction of Ti^IV produces a Si^? that undergoes a O^2- abstraction from P^V compounds leading to P^III compounds and Si-O^? species. Reoxidation of Ti^III by the latter gives silanol species. This mechanism was further probed by ²⁹Si NMR analysis of the reaction mixture as a function of time and by the reduction of optically active P-stereogenic tertiary phosphine oxides. A practical reduction protocol of Ph₃PO with these environmentally benign reagents on a 100 g scale has been developed.

Full text of DOI:10.1021/cs4002767

Research Article
pubs.acs.org/acscatalysis
Unraveling the Catalytic Cycle of Tertiary Phosphine Oxides
29
Reduction with Hydrosiloxane and Ti(OiPr)4 through EPR and Si
NMR Spectroscopy
Christelle Petit,† Evelyne Poli,‡ Alain Favre-Reguillon,*,§,∥ Lhoussain Khrouz,‡ Sandrine Denis-Quanquin,‡
́
Laurent Bonneviot,‡ Gerard Mignani,⊥ and Marc Lemaire†
́
†Laboratoire de Catalyse et de Synthes
̀
e Organique, UMR 5246, ICBMS, Universite
1918, 69622 Villeurbanne Cedex, France
‡ENS Lyon, Laboratoire de Chimie, Groupe Mater
§Laboratoire de Gen
ie des Procedes Catalytiques, UMR 5285, CPE Lyon, 43 bld du 11 Novembre 1918, 69622 Villeurbanne Cedex,
France
∥Equipe CGP, Dep
⊥Rhodia Operations, Lyon Research Center, 85, avenue des Frer
́
Claude Bernard Lyon 1, 43 bld du 11 Novembre
́
́
iaux Hybrides, 46 allee d’Italie, 69364 Lyon cedex 7, France
́
́
́
́
artement CASER, Conservatoire National des Arts et Met
́ ́
iers, 2 rue Conte, 75003 Paris, France
̀
es Perret, BP 62, 69192 Saint-Fons Cedex, France
S
* Supporting Information
ABSTRACT: The reduction of tertiary phosphine oxides using
tetramethyldisiloxane (TMDS) as a mild reducing agent and catalytic
amount of TiIV isopropoxide has been studied in detail. An extensive EPR
study has revealed the presence of at least five TiIII species, and structures
have been proposed. Thus, a single electron transfer (SET) mechanism
consisting of a back and forth oxido-reduction of Ti from the IV to the III
oxidation state could be proposed. Reduction of TiIV produces a Si• that
undergoes a O2− abstraction from PV compounds leading to PIII
compounds and Si−O• species. Reoxidation of TiIII by the latter gives
silanol species. This mechanism was further probed by 29Si NMR analysis of the reaction mixture as a function of time and by the
reduction of optically active P-stereogenic tertiary phosphine oxides. A practical reduction protocol of Ph3PO with these
environmentally benign reagents on a 100 g scale has been developed.
KEYWORDS: phosphine oxide, reduction, mechanism, silane, EPR
Despite this effective procedure being used for decades by
numerous research teams, surprisingly little is known about the
mechanism of this reduction by hydrosiloxanes as detailed
mechanistic studies have not been undertaken.22 A catalytic
cycle has been proposed by Lawrence et al. for the reduction of
tertiary phosphine oxides by triethoxysilane involving a
titanium hydride-like complex.17 Recently, we have shown the
presence of TiIII in the mixture, which suggests that the
reduction of tertiary phosphine oxides may occur via a single
electron transfer (SET) mechanism.24 In this paper, we present
complementary studies on the intermediate titanium TiIII
species that appears under the reducing action of a TMDS
agent and on the mechanism of tertiary phosphine oxides
reduction by Ti(OiPr)4 and hydrosiloxanes.
INTRODUCTION
■
Tertiary phosphines are an important compound class for metal
ligation in transition-metal-catalyzed enantioselective trans-
formations1,2 or coupling reactions of aryl electrophiles.3−5
Often, however, these phosphines are prone to oxidation, and a
number of reduction strategies were developed.6 The existing
reducing agents, though efficient in producing high yields of
desired product, produce undesirable waste products or pose
safety problems.7−11 Finally, silanes12−16 and a mixture of
silanes with a Lewis acid16−18 remains the reagent of choice for
the reduction of phosphine oxides even if some silanes have
been demonstrated to be harmful since they generate a
dangerous, pyrophoric, and toxic SiH4 gas.19
In recent years, the rediscovery of the reductive abilities of
hydrosiloxanes when associated with a catalytic amount of
transition metal has renewed interest in using hydrosiloxanes as
stoichiometric reductive agents of tertiary and secondary
phosphine oxides.20−26 Indeed, hydrosiloxanes, in comparison
to most of the conventional reagents, have many advantages,
since they are commercially available, air- and moisture-stable,
nonpyrophoric, and soluble in most organic solvents.19
EXPERIMENTAL SECTION
■
Reduction of Triphenylphosphine Oxide. Triphenyl-
phosphine oxide (390 mg, 1.40 mmol) and methylcyclohexane
Received: October 29, 2012
Revised: May 17, 2013
© XXXX American Chemical Society
1431
dx.doi.org/10.1021/cs4002767 | ACS Catal. 2013, 3, 1431−1438

ACS Catalysis
Research Article
(1.9 mL) were placed in a dry 5 mL round-bottomed flask with
a magnetic stirrer. Then, TMDS (620 μL, 3.5 mmol) was added
to the reaction vessel followed by Ti(OiPr)4 (80 μL, 0.28
mmol), and the crude mixture was heated to 80 °C. After 14 h,
the 31P NMR analysis showed the complete conversion of the
starting reagent. The heterogeneous mixture was cooled to 0
°C and filtered over porous glass, and the solid was washed
with 4 × 5 mL of pentane. The resulting white solid was dried
under vacuum, yielding 365 mg of triphenylphosphine
(quantitative yield).
Safety. Unreacted TMDS in the filtrate must be destroyed
by the slow addition of a 3 M alcoholic solution of KOH at
room temperature. TMDS decomposes on contact with bases,
forming hydrogen.
EPR Studies. Methylcyclohexane (5 mL) and TMDS (49
μL, 0.28 mmol) were placed in a dry 10 mL round-bottomed
flask with a magnetic stirrer. The flask was capped with a
septum, and Ti(OiPr)4 (11.8 μL, 0.04 mmol) was added. The
crude mixture was stirred at room temperature until a blue
color appeared. Then, 1 mL of the mixture was transferred via
syringe into a standard 4 mm quartz dry EPR tube and quickly
placed in a small benchtop liquid nitrogen dewar.
dimer dynamic equilibrium at room temperature that converts
part of the tetrahedral Ti(PriO)4 species into dimers of
pentacoordinated TiIV sharing two μ-OiPr bridging groups,
[TiIV(OiPr)4μ-(OiPr)2TiIV(iPrO)4].27 Since the radius of the
TiIII ion is larger than that of the TiIV ion, the tendency for
dimerization is stronger after reduction favoring the formation
of TiIII−TiIV and TiIII−TiIII pairs at the expense of TiIV−TiIV
pairs. These pairs are EPR active and were observed in the
present system for a high level of reduction and high titanium
concentration obtained in optimum condition. Unfortunately,
TiIII−TiIII pairs exhibit a complicated and broad signal where
hyperfine interactions are not resolved (not reported here). For
the sake of simplification and to get better insights on the
constitutive TiIII species, it is preferred here to focus on the low
concentration conditions where no TiIII−TiIII species can be
found. In addition, to avoid a broadening effect, the
concentration was lower than in the previous study.24 As a
consequence, the room temperature EPR signal, though similar
to that obtained previously, presents a better resolved peak due
to less spin−spin interaction known to broaden the signal at
high concentrations.24
At 295 K, instead of two, there are now three well-resolved
narrow features pointing upward on the left-hand side of the
main peak in the range 3375−3410 G (Figure 1). In addition,
the second and third peaks are obviously split into two features
so that the triplet is better assigned as two intercalated
doublets. The main peak is quasi-symmetric pointing up- and
downward centered at 3430 G and a shoulder pointing
downward at 3445 G (Figure 1). When the temperature is
decreased down to 210 K, the methylcyclohexane (MCH)
becomes viscous, slowing down the free rotation of the solvated
species, and the signal broadens. A better resolved and much
more complex signal appears at 135 K, i.e., below the melting
point of MCH (147 K). Indeed, the signal is typical of a
powder pattern EPR signal obtained for randomly oriented
species in a frozen solvent matrix. Each species is characterized
by a complex signal where singularities correspond to the
projection of the g and A tensors on the main x, y, and z axes,
i.e., gx, gy, and gz and also Ax, Ay, and Az. Such an analysis
requires a simulation, particularly when a mixture of species is
at stake, as it appears here according to the number of features
on the spectra. There are two distinct narrow peaks at ca. 3360
G (upward) and 3470 G (downward) and also a broad one in
the middle around 3440 G. The latter results from the
superimposition of ca. eight secondary lines not regularly
spaced one with another. This complexity suggests the presence
of more than three species. In addition, the absence of any
signal at half field (g ∼ 4) indicates that there are no S = 1 pairs
as planned (vide supra, Figure S1).28−34 Therefore in the
following only S = 1/2 species were considered.
RESULTS AND DISCUSSION
■
Characterization of the Catalytic Species by EPR
Spectroscopy. The presence of a singlet state (S = 1/2) at
room temperature for the TMDS/Ti(OiPr)4 system had
already been noted using EPR spectroscopy.24 It was concluded
that these species correspond to a d1 TiIII species. On the basis
of the saturation test, it was claimed that it could not be an s
nor p type radical.24 This work is completed here by a low
temperature study (Figure 1) and EPR signal simulations in
order to identify and quantify the different TiIII species (Figure
2). The signal strongly depends on the concentration of species
with the appearance of TiIII−TiIII dimers with S = 1 mixed with
TiIII−TiIV pairs and also TiIV monomers characterized by S =
1/2. This is reminiscent of the natural tendency of TiIV alkoxide
to oligomerize. In particular, Ti(PriO)4 exhibits a monomer−
The simulation was performed assuming that the same
species could be found at both low and high temperatures. The
EPR contributions of the 47Ti (I = 5/2) and 49Ti (I = 7/2)
isotopes yielding 6 and 8 line hyperfine structures (hfs) were
included in the simulation taking into account their natural
abundance, 7.28 and 5.51%, respectively. Though both sets of
hfs structures were not fully resolved, they are centered on the g
values of the dominating signal of the species with the isotope
48Ti (I = 0), and their intensities are fully imposed by the
isotope abundance and the number of lines. Another constraint
was exploited to analyze the data knowing that at room
temperature the free rotating species exhibit isotropic signals
characterized by giso and Aiso split at low temperature into
Figure 1. X-band EPR spectrum of the TMDS/Ti(OiPr)4 system as a
function of temperature.
1432
dx.doi.org/10.1021/cs4002767 | ACS Catal. 2013, 3, 1431−1438

ACS Catalysis
Research Article
Figure 2. Experimental (black line) and simulated (red line) EPR spectrum of the TMDS/Ti(OiPr)4 system at 295 K (a) and 135 K (b). In b,
individual simulated signals A, B, C, D, and E provided the intensity used for the simulation of the experimental spectrum. The green spots stand for
the most specific feature of signal A. In c, full scale simulated signals A, B, and C with g1, g2, and g3 assignments and stars pointing the most obvious
features of the Ti hfs contributions.
Table 1. EPR Parameters and Relative Concentration of the EPR Active Species from the Simulation of the Experimental
a
Spectrum Measured at 295 K and 135 K (see Table S1 for the Full Set of Parameters)
295 K
135 K
Aav(Ti)/G
species
%
giso
Aiso(Ti)/G
Aiso(H)/G
%
gav
Aav(H)/G
A
B
C
D
E
83
5
1.961
1.958
1.962
1.986
1.977
8.3
10.0
6.7
25
10
46
6
1.946
1.953
1.964
1.965
1.968
8.3
9.4
6.4
5
4
13.3
13.3
13.3
13.3
3
3
a
The accuracy is 1 on the last digit provided for g and A values.
several components (g1, g2, g3, A1, A2, and A3) that are much
narrower due to longer relaxation times. Usually, apart from a
slight effect of the Jahn−Teller distortion, the average values gav
= (g1 + g2 + g3)/3 and Aav = (A1 + A2 + A3)/3 should match the
room temperature giso and Aiso. This was used as a guide to
assign a third component when only two of them were fully
resolved.
corresponded to a spin delocalization of ca. 13%.31 Such
doublets were consistent with the shfs of a single proton. It is
assigned to a monohydride TiIII species with respect to a series
of TiIII dihydride in [Cp*2Ti-(μ-H)2-MgR2] where Cp* = Cp,
MexCp (x = 0, 3, 4, 5), and R = μ-Cl, μ-Br, μ-OR, Cp, aryl, and
butyl, where the spin is delocalized onto two protons leading to
an Aiso(H) about 2 times smaller, ranging from 5 to 8 G.32,33
Accordingly, the simulation of the spectrum registered at 135
K was obtained similarly with three species exhibiting only Ti
hfs and two others with only H shfs (Table 1 and Figure 2). At
135 K, all the species exhibited a strong orthorhombicity
characteristic of a strong Jahn−Teller distortion typical of d1
TiIII species, which may explain a slight shift of gav in
comparison to giso. The detailed values of both g and A tensors
were provided in the Supporting Information (Table S1) as
well as the EPR signal of each simulated species (Figure 2).
Once again at 135 K, the hfs of Ti isotopes remained marginal
contributions on each side of the main g components (see, for
instance, simulated signal A in Figure 2). The relative position
of the g tensor components provided the ground state using a
Then, to reproduce most of the features of the spectra, five
different species were necessary and accounted for 94 and 91%
of the experimental spectrum at 295 and 135 K, respectively
(Table 1 and Figure 2). Reducing the simulation residue would
require supplementary signals. This was discarded because of a
lack of confidence and low accuracy on such solutions. Among
the species, the most intense were simulated from three species
characterized by giso = 1.958, 1.961, and 1.962 that reproduced
the slight asymmetrical aspect of the central peak at room
temperature. In addition, the little shoulders at the foot of the
peak on each side were obtained from different hyperfine
splittings (hsf) Aiso(Ti) = 10, 8.3, and 6.7 1 G for species
denoted as B, A and C, respectively (Table 1). Then, the latter
differed mostly from Aiso(Ti) hfs (vide infra). The remaining
two signals accounting for 4 and 3% of the overall fit were those
characterized by a doublet due to the presence of a nearby
single proton nucleus (I = 1/2) with a superhyperfine splitting
(shfs) constant, Aiso(H), of 13.3 and 13.3 G from species and
giso = 1.986 and 1.977 for species denoted as D and E,
respectively. The shift toward g = 2 was consistent with species
characterized by a higher crystal field stabilization than in A, B
and C species. Aiso(H) related to the Fermi contact term
2
first order analysis assuming a pseudoaxial like symmetry: dz
2
2
for species A and C (g2 ∼ g3 ∼ g⊥ < g1 ∼ g∥ ∼ 2) and dx −y for
species B (g1 ∼ g2 ∼ g⊥ > g3 ∼ g∥).31 Furthermore, gav values
indicated that the crystal field strength increases from species in
the order A, B and C. The parameters used to solve the
structure of each species were based on spectroscopic and
chemical criteria as well: (i) gav related to the average crystal
field strength and the number of ligands around TiIII, (ii) the
ground state provided by the symmetry at the first order (axial
1433
dx.doi.org/10.1021/cs4002767 | ACS Catal. 2013, 3, 1431−1438

ACS Catalysis
Research Article
a
Scheme 1. Proposed Structures for Species A, B, and C According to EPR Characteristics
a
TiIII (blue circle and blue bonds), TiIV (black circle and black bonds); the latter is smaller and systematically represented in a trigonal bipyramid
(tbp) double bridged to TiIII species. TiIV has no d electron and follows the VSEPR rule. A elongated along the Ti-HOR bond. B has a square
pyramidal TiIII with the weaker bonds in the xy plane and a Ti-OR bond in the z axis. TiIII in C is either pentacoordinated in an elongated tbp
symmetry or more likely hexacoordinated in an elongated octahedral environment. See text for species D an E; red arrows depict the first order axial
distortion.
approximation) and the strength of the distortion at the second
order, (iii) the number of available ligands for Ti in the present
system. The latter was limited to four isopropylates, one of
them being protonated during the reduction of TiIV into TiIII
species (vide infra). Bearing in mind that the alkoxide precursor
was a tetrahedral [TiIV(OiPr)4], species A, characterized by the
smallest gav and therefore the weaker crystal field stabilization,
was more likely a tetracoordinated [TiIII(OiPr)3(iPrOH)] or
possibly a tricoordinated [TiIII(OiPr)3] species. Both structures
formation of the monomeric species A was consistently favored
at the expense of pairs or higher adducts B and C at increasing
temperatures, suggesting a mixture of species in equilibrium,
one with another. Note that there is no resolved proton shfs
coupling in species A, B and C despite the presence of the
ligand ROH. Indeed such weak coupling was observed using
the ENDOR double resonance technique only. Moreover, the
pseudo-octahedral TiIII(H2O)6 species were shown recently to
be highly distorted with two energetically close conformations,
D3d and Ci, where the unequivalent protons led to the shfs
component, also resolved using the double resonance
technique.35
The relatively small proton shfs of species E and D precludes
the formation of a strong covalent Ti−H bond and would
rather fit with a weak Ti−H−Si type of interaction proposed
earlier when hydrosilane is in the presence of titanium
complexes.32 Then E and D will be likely the Lewis acid−
base adduct formed between TMDS and A and B, respectively.
These marginal species are not central here to explain the
reactivity and will be investigated in a forthcoming paper
focusing on the relation between A, B, C, D and E species.
Structure Analysis of the Reaction Products by 29Si
NMR Analysis. To further probe the proposed mechanism, the
reduction of Ph3PO was monitored by an in situ 29Si NMR
study. The 29Si NMR commonly used Cr(acac)3 as a spin
relaxant. 29Si NMR of the crude in CDCl3 using Cr(acac)3 was
shown in Figure S3. However, in order to minimize interactions
between spin relaxant and deuterated solvents with the
intermediate silicone species, 29Si NMR was operated using a
coaxial insert filled with D2O for locking. A DEPT sequence for
1H−29Si coupling was thus used to enhance the signal-to-noise
ratio. Shown in Figure 3 are 29Si NMR spectra for four time
points (t = 0, 10 min, 4 and 14 h) during the course of the
2
were compatible with a dz ground state assuming an elongated
tetrahedral symmetry (C3v) for the former and flat triangular
symmetry (D3h) for the latter. However, the strong rhombicity
of the signal was more compatible with an elongated tetrahedral
TiIII species (Scheme 1). For species A, the pseudo-axial
direction corresponds to the bond with a ligand ROH while the
two other directions are necessarily unequivalent due to the
presence of the other three ligands RO− located out of the xy
plane. The rationale for a stronger crystal field in species B and
C was a higher coordination number, i.e. 5 or 6. A
pentacoordinated d1 species with a dx −y ground state would
be obtained with a compressed pyramid trigonal geometry. A
2
2
1
2
hexacoordinated d species with a dz ground state would fit
with an elongated octahedral environment for species C.
Increasing further the coordination number without changing
the ratio of metal to ligands will be obtained by
oligomerization. A TiIV−TiIII−TiIV trimer as exemplified in
Scheme 1 is an example that possesses a TiIV−TiIV−TiIV trimer
analogue already reported in solution for the tetraethoxytita-
nium, [Ti(OEt)4].34 Alternatively, a pentacoordinated TiIII
elongated trigonal bipyramid conformer of species B would
2
also fit the ground dz ground state but would generate a weaker
crystal field. Other types of oligomers or conformers are likely
to be formed and may explain the residue of simulation. At this
stage of characterization, the assignment of species C and the
presence of other species remain open. Nonetheless, the
1434
dx.doi.org/10.1021/cs4002767 | ACS Catal. 2013, 3, 1431−1438

ACS Catalysis
Research Article
using CDCl3 as a solvent, and Cr(acac)3 was used as a spin
relaxant at t = 14 h (Figure S3). At t = 4 h, signals between −7
and −7.5 ppm could be attributed to silane species (Table S2,
species 4 and 5). In particular, the signal at −9.3 ppm could be
attributed to mono-oxidized TMDS species 5. Silanol groups
between −14 and −19 ppm could be also identified (Table S2,
species 6). It should be emphasized here that the presence of
silanol is not expected in the catalytic cycle proposed by
Laurence et al.17,18 Siloxane species (Table S2, species 8 and 9)
between −20 and −23 ppm originating from the condensation
of silanol groups were identified in the reaction mixture, and in
particular for high Ph3PO conversion, large signals between
−22 and −23 ppm were attributed to cyclic oligomers of
siloxane (i.e., cyclomethicone, Table S2, species 8). All the
species with different relative intensities are present in the 29Si
NMR spectra of the crude obtained in CDCl3 using Cr(acac)3
as a spin relaxant (Figure S3).
Proposed Mechanism. Following EPR and in situ 29Si
NMR analysis, an improved catalytic cycle, compared to the
catalytic cycle previously published,24 can be proposed
(Scheme 2). The first step consists of one electron reduction38
of TiIV(OiPr)4 into a tetracoordinated [TiIII(OiPr)3(iPrOH)] A
by TMDS. A could be stabilized by the formation of species of
higher coordination numbers, i.e. B, C, ... (Scheme 1). This one
electron reduction generates a silicon centered radical species 3
that readily reacts with PO yielding phosphorus radical
centered species 9, which evolved to give desired triphenyl-
phosphine and SiO• species 10. Therefore, the balance between
species 3 and 10 is a transfer O2− on a Si• yielding a Si−O bond
where oxygen is formally at the −1 oxidation state. Then, the
reoxidation of TiIII A by Si−O• furnished TiIV and a silanol 4
where oxygen comes back to the −2 oxidation state (Scheme
2). The silanol species 4 and 6 could be identified by 29Si NMR
(Figure 3). Incidentally, silicon centered radical species 3 react
together, leading to disilane species 2 identified by 29Si NMR
Figure 3. 29Si NMR of the crude reaction mixture of Ph3PO reduction
by TMDS/Ti(OiPr)4.
reaction. Silicon species (Table S2) were identified by
comparison with commercial products and literature data.36,37
At t = 0, the signal at −5.0 ppm is assigned to TMDS (1). At
t = 10 min, a small signal could be observed at −0.2 ppm that
could be assigned to a disilane species (2). It should be
emphasized here that no reduction of Ph3PO could be
determined at this point by 31P NMR. Such species could be
formed by the coupling of two Si radical centered species 3
(Scheme 2, vide supra). The relative signal intensity of 2
increased at t = 4 h, and 2 could be detected at t = 14 h. A
higher signal intensity at −0.2 pmm could be observed when
a
Scheme 2. Proposed Mechanism of Ti(OiPr)4-Catalyzed Reduction of Triphenylphosphine Oxide by Hydrosiloxanes
a
Underline numbers correspond to identifying silicone species by 29Si NMR, and letters correspond to identified titanium species by EPR.
1435
dx.doi.org/10.1021/cs4002767 | ACS Catal. 2013, 3, 1431−1438

ACS Catalysis
Research Article
(Figure 3). The condensation of silanol species is expected to
form siloxane species 5−8 and water. As already mentioned, an
equimolar amount of water is produced as a byproduct, and the
use of a drying agent improved the reactivity of the catalytic
system.24 Furthermore, the reduction of Ph3PO only requires
one Si−H function per PO group24 whereas two Si−H
functions per PO are needed in Lawrence’s mechanism.17,18
Reduction of P-Chiral Phosphines Oxides. The
proposed catalytic cycle (Scheme 2) involved the formation
of phosphorus radical centered species. While the reduction of
nonstereogenic phosphine oxides proceeds in high yield,23,24,26
the stereospecific reduction of pure P-chiral phosphine oxide
has never been examined with this system. Thus, the reduction
of (S,S)-DIPAMP oxide (11) was studied using standard
conditions (Scheme 3). After 24 h at 60 °C, 11 was
temperature of 80 °C, affording triphenylphosphine as the
only observable product by HPLC and NMR. The simplicity
and efficiency of the transformation led us to pursue the
development of a product reduction procedure that would
make this methodology adaptable to a 100 g scale.
The first and most critical step in the scale up procedure is to
undertake a risk assessment of the reaction. First of all, the
thermochemistry of the reduction of Ph3PO by TMDS/
Ti(OiPr)4 was studied using DSC (Figure S4 to S6). Low
exothermicity was determined for this reaction, down to 6 J/
mol (Figure S7). Then, the in situ monitoring of the reaction on
a 100 g scale was envisaged. However, Ph3PO and
triphenylphosphine were only slightly soluble in methylcyclo-
hexane. For a concentration up to 0.5 M L−1 and 10 wt % of
Na2SO4,24 the in situ monitoring of the reaction could only be
possible using Raman spectroscopy. An analytical method using
Raman spectroscopy was thus developed for the monitoring of
Ph3PO, TMDS, and triphenylphosphine concentration. Cyclo-
hexane was used instead of methylcyclohexane in order to
simplify Raman spectrum (Figure S8). Raman data were
correlated with a 31P NMR analysis of the aliquot as a function
of time (Figure 4).
Scheme 3. Reduction of Optically Active Tertiary Phosphine
Oxides
On the 100 g scale and for a concentration of 0.5 M L−1 at 60
°C and in the presence of 10 wt % Na2SO4, the reaction
proceeded smoothly, and a complete conversion was obtained
in 14 h. The heterogeneous reaction was cooled, and
triphenylphosphine was recovered by filtration using a Buchner
funnel. The obtained solid was washed with pentane to remove
silicone derivatives, and triphenylphosphine was obtained with
a quantitative yield after recrystallization from acetone. The
filtrate was cooled to 0 °C, and a 3 M alcoholic solution of
KOH was carefully added under vigorous stirring to destroy the
remaining Si−H functional group.
quantitatively reduced as estimated by 31P NMR, but as
expected for the proposed catalytic cycle (Scheme 3),
epimerization at phosphorus was observed. An e.e. of 86%
was determined by chiral HPLC for DIPAMP 12. The observed
racemization induced by TMDS/Ti(OiPr)4 could occur
through a pentacoordinate phosphorus intermediate (species
9 in Scheme 2) which can undergo pseudoracemisation.6
However, it should be noted that such partial racemization has
already been observed during the reduction of P-chirogenic
binaphthyl monophosphine oxide using classical methods, such
as hydride.39,40
CONCLUSION
■
In summary, the reduction of Ph3PO by the TMDS/Ti(OiPr)4
system was studied in detail for the first time, and a mechanism
involving a single electron transfer was proposed. The catalytic
species that involves a reduced state of titanium in the initiation
Scale Up. On a millimolar scale, the reduction of Ph3PO
proceeded smoothly to completion within 24 h at the
Figure 4. Monitoring of Ph3PO reduction on a 100 g scale using Raman spectroscopy. Experimental conditions: TMDS (Si−H/PO = 2.1),
Ti(OiPr)4 10% cat., Na2SO4 (10 wt %), cyclohexane, 60 °C.
1436
dx.doi.org/10.1021/cs4002767 | ACS Catal. 2013, 3, 1431−1438

ACS Catalysis
Research Article
step was investigated using EPR investigations at different
temperatures. The spectra exhibit a mixture of species
characteristic of TiIII, S = 1/2, species and TiIII−TiIII, S = 1
pairs. The spin S = 1 pairs produce a broad signal that could not
be isolated from the S = 1/2 systems. Nonetheless, at a low
reduction level, the S = 1 system disappears, yielding a mixture
of five different S = 1/2 species. Both room temperature and
low temperature (frozen matrix) spectra were simulated
providing some insights from the g and A tensors on the
symmetry and coordination number of the TiIII center of each
species. The rationale for such a variety of species is related to
the tendency of TiIV alkoxides to dimerize, a phenomenon
more favorable for TiIII ions that have a larger ionic radius than
TiIV ions. Then, monomeric tetrahedral TiIII species, TiIII−TiIV
pairs, and eventually TiIV−TiIII−TiIV trimers are likely to coexist
under partial reduction and were tentatively assigned. Indeed,
the dominating species at room temperature is the elongated
tetrahedral TiIII species A (∼ 80%) that is coordinatively
unsaturated and preferably associated with the catalytic activity.
The proposed mechanism is in agreement with the in situ 29Si
NMR analysis of the silicon based intermediate, the
stoichiometry of the reaction, EPR studies, and partial
racemization obtained with stereogenic phosphine oxides.
The mechanism is based on a back and forth oxido-reduction
of Ti from IV to III oxidation state.38 First, titanium(IV) acts as
a one-electron oxidation species on SiH producing Si•. Second,
the so-obtained titanium(III) species reduces the SiO•
intermediate into a silanol species. The reduction of tertiary
phosphine oxides thus consists of a transfer of O2− species on a
Si• yielding SiO• where oxygen is formally at the oxidation state
−1. Moreover, a reduction of Ph3PO on a 100 g scale was
described. These conditions may be attractive as a general way
to reduce tertiary phosphine oxides. This improved method
involved the following characteristic features: (1) An air- and
moisture-stable hydrosiloxane reagent was used as a hydride
source. (2) Water-insoluble “process friendly” solvents such as
toluene, methylcyclohexane, and cyclohexane were used as
solvents. (3) Crystals of triphenylphosphine were obtained
directly with a straightforward workup. (4) Treatment of the
waste stream by potassium hydroxide gave inert TiO2 and
siloxanes. (5) This reaction is general.23−26
REFERENCES
■
(1) Jacobsen, E. N.; Pfaltz, A.; Yamamoto, H. Comprehensive
asymmetric catalysis; Springer-Verlag GmbH: Berlin, 1999.
(2) Ojima, I. Catalytic asymmetric synthesis; Wiley-VCH: New York,
2000.
(3) Diederich, F.; Stang, P. J. Metal-catalyzed cross-coupling reactions;
Wiley-VCH: Weinheim, Germany, 1997.
(4) Tsuji, J. Palladium reagents and catalysts; Wiley & Sons:
Chichester, U. K., 2004.
(5) Beller, M.; Bolm, C. Transition metals for organic synthesis. Wiley-
VCH: Weinheim, Germany, 2004.
(6) Engel, R. Handbook of organophosphorus chemistry; Engel, R., Ed.;
Marcel Dekker: New York, 1992.
(7) Imamoto, T.; Kikuchi, S.-I.; Miura, T.; Wada, Y. Org. Lett. 2001,
3, 87−90.
(8) Imamoto, T.; Takeyama, T.; Kusumoto, T. Chem. Lett. 1985,
1491−1492.
(9) Bootle-Wilbraham, A.; Head, S.; Longstaff, J.; Wyatt, P.
Tetrahedron Lett. 1999, 40, 5267−5270.
(10) Busacca, C. A.; Raju, R.; Grinberg, N.; Haddad, N.; James-Jones,
P.; Lee, H.; Lorenz, J. C.; Saha, A.; Senanayake, C. H. J. Org. Chem.
2008, 73, 1524−1531.
(11) Rajendran, K. V.; Gilheany, D. G. Chem. Commun. 2012, 48,
817−819.
(12) Fritzsche, H.; Hasserodt, U.; Korte, F.; Friese, G.; Adrian, K.;
Arenz, H. J. Chem. Ber. 1964, 97, 1988−1993.
(13) Horner, L.; Balzer, W. D. Tetrahedron Lett. 1965, 1157−1162.
(14) Naumann, K.; Zon, G.; Mislow, K. J. Am. Chem. Soc. 1969, 91,
2788−2789.
(15) Krenske, E. H. J. Org. Chem. 2012, 77, 1−4.
(16) Organophosphorus Chemistry Series; Allen, D. W., Tebby, J. C.,
Eds.; Royal Society of Chemistry Cambridge: Cambridge, U. K.; pp
1970−2012.
(17) Coumbe, T.; Lawrence, N. J.; Muhammad, F. Tetrahedron Lett.
1994, 35, 625−628.
(18) Lawrence, N. J.; Drew, M. D.; Bushell, S. M. J. Chem. Soc., Perkin
Trans. 1 1999, 3381−3391.
(19) Wells, A. S. Org. Process Res. Dev. 2010, 14, 484−484.
(20) Li, Y.; Das, S.; Zhou, S.; Junge, K.; Beller, M. J. Am. Chem. Soc.
2012, 134, 9727−9732.
(21) Li, Y.; Lu, L.-Q.; Das, S.; Pisiewicz, S.; Junge, K.; Beller, M. J.
Am. Chem. Soc. 2012, 134, 18325−18329.
(22) Harris, J. R.; Haynes, M. T.; Thomas, A. M.; Woerpel, K. A. J.
Org. Chem. 2010, 75, 5083−5091.
(23) Berthod, M.; Favre-Reguillon, A.; Mohamad, J.; Mignani, G.;
Docherty, G.; Lemaire, M. Synlett 2007, 1545−1548.
(24) Petit, C.; Favre-Reguillon, A.; Albela, B. l.; Bonneviot, L.;
Mignani, G.; Lemaire, M. Organometallics 2009, 28, 6379−6382.
(25) Petit, C.; Favre-Reguillon, A.; Mignani, G.; Lemaire, M. Green
Chem. 2010, 12, 326−330.
(26) Dayoub, W.; Favre-Reguillon, A.; Berthod, M.; Jeanneau, E.;
Mignani, G.; Lemaire, M. Eur. J. Org. Chem. 2012, 3074−3078.
(27) Babonneau, F.; Doeuff, S.; Leaustic, A.; Sanchez, C.; Cartier, C.;
Verdaguer, M. Inorg. Chem. 1988, 27, 3166−3172.
(28) Samuel, E.; Harrod, J. F.; Gourier, D.; Dromzee, Y.; Robert, F.;
Jeannin, Y. Inorg. Chem. 1992, 31, 3252−3259.
(29) Lukens, W. W.; Andersen, R. A. Inorg. Chem. 1995, 34, 3440−
3443.
ASSOCIATED CONTENT
* Supporting Information
Synthesis and characterization of the products; simulation
parameters of the EPR signal spectra at 135 K; experimental
and simulated EPR spectrum at 295 K; 29Si NMR analysis of
the reactions products; DSC analysis; Raman spectra of
individual species. This material is available free of charge via
the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*E-mail: afr@lgpc.cpe.fr (A. F.-R.).
Notes
■
́ ́
(30) Horacek, M.; Gyepes, R.; Císarova, I.; Kubista, J.; Pinkas, J.;
Mach, K. J. Organomet. Chem. 2010, 695, 2338−2344.
(31) Weil, J. A.; Bolton, J. R. Electron paramagnetic resonance:
elementary theory and practical applications; Wiley: New York, 2007.
(32) Gyepes, R.; Hiller, J.; Thewalt, U.; Polasek, M.; Sindelar, P.;
Mach, K. J. Organomet. Chem. 1996, 516, 177−185.
(33) Horacek, M.; Cisarova, I.; Cejka, J.; Karban, J.; Petrusova, L.;
Mach, K. J. Organomet. Chem. 1999, 577, 103−112.
(34) Russo, W. R.; Nelson, W. H. J. Am. Chem. Soc. 1970, 92, 1521−
1526.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by Rhodia Operations through a
Ph.D. grant to C.P. We thank Dr. Nicolas Capelle, Jean Louis
Gustin, and Serge Henrot from Rhodia Operations for
assistance with 29Si NMR, DSC, and RAMAN spectroscopy.
1437
dx.doi.org/10.1021/cs4002767 | ACS Catal. 2013, 3, 1431−1438

ACS Catalysis
Research Article
(35) Maurelli, S.; Livraghi, S.; Chiesa, M.; Giamello, E.; Van
Doorslaer, S.; Di Valentin, C.; Pacchioni, G. Inorg. Chem. 2011, 50,
2385−2394.
(36) Kintzinger, J. P.; Marsmann, H. NMR Basic Principles and
Progress: Oxygen-17 and Silicon-29; Springer-Verlag: Berlin, 1981.
(37) Gupta, R. R.; Lechner, M. D.; Marsmann, H.; Mikhova, B.;
Uhlig, F. Chemical shifts and coupling constants for Silicon-29; Springer:
Berlin, 2008.
(38) Praneeth, V. K. K.; Ringenberg, M. R.; Ward, T. R. Angew.
Chem., Int. Ed. Engl. 2012, 51, 10228−10234.
(39) Higham, L. J.; Clarke, E. F.; Mueller-Bunz, H.; Gilheany, D. G. J.
Organomet. Chem. 2005, 690, 211−219.
(40) Kerrigan, N. J.; Dunne, E. C.; Cunningham, D.; McArdle, P.;
Gilligan, K.; Gilheany, D. G. Tetrahedron Lett. 2003, 44, 8461−8465.
1438
dx.doi.org/10.1021/cs4002767 | ACS Catal. 2013, 3, 1431−1438