2
M. Kuroboshi et al. / Tetrahedron Letters xxx (2015) xxx–xxx
Table 1
Table 3
Mg–Me
3
SiCl promoted reduction of 2a to 1a. Effect of solventsa
Reduction of several types of phosphine oxides
R1
R1
Entry
Solvent
Ratiob (%)
R1
Mg (8 mmol), Me
3
SiCl (6 mmol)
P
O
R1
P
1
a
2a
2
DMI (8 mL), Room temp., Time
R2
1
R
1
2
3
4
5
6
7
DMI
MeCN
DMF
DMAc
NMP
DMPU
THF
98 (96)c
1
91
93
100
100
100
100
2
(2 mmol)
9 (9)c
c
7 (7)
n.d.
n.d.
n.d.
R1
R2
Ratioa,b (%)
Entry
Time (h)
d
e
e
e
e
d
1
2
d
1
2
3
4
5
6
7
8
9
Ph
Ph
2a
2b
2c
2d
2e
2f
2g
2h
2i
2
4
6
4
2
2
2
2
2
2
97 (96)
95 (89)
91 (84)
95 (91)
94 (93)
Complex
3
4
7
5
d
n.d.
o-Tolyl
m-Tolyl
p-Tolyl
p-Anis
Ph
Ph
Ph
Et
Oct
o-Tolyl
m-Tolyl
p-Tolyl
p-Anis
a
b
c
3
Compound 2a (2 mmol), Mg (6 mmol), Me SiCl (6 mmol), solvent (8 mL), rt, 2 h.
Determined by P NMR.
Isolated yields are shown in the parentheses.
Not detected.
No peaks other than 2a were observed.
31
6
d
d
e
p-NCC
6
H
4
n.d.
22
4
57
98
i
c
Pr
37 (54 )
96 (75 )
c
Et
Ph
Oct
c
35 (70 )
d
10
2j
n.d.
1
a was obtained in 96% yield after purification by silica gel column
a
b
c
The ratios of 1 to 2 were determined by 31P NMR.
chromatography (hexane/AcOEt = 5:1) (Table 1, entry 1).
Remarkable solvent effect was found: among the solvents
examined, DMI was the only solvent to afford 1a efficiently
Yields referred to the isolated 1 are shown in the parentheses.
31
21
Yields were evaluated by P NMR.
d
Not detected.
(Table 1): 1a was obtained quantitatively in DMI (entry 1), whereas
yield of 1a was less than 10% in acetonitrile (entry 2) and DMF
Several phosphine oxides 2 were reduced under the optimized
conditions (Table 3). Triarylphosphine oxides 2a–2e gave the cor-
responding triarylphosphine 1a–1e in quantitative yields (entries
(
entry 3), and no 1a was obtained in DMAc (N,N-dimethylacet-
0
amide), NMP (N-methylpyrrolidone), DMPU (N,N -dimethylpropy-
leneurea), and THF (entries 4–7).
1
–5), whereas triarylphosphine oxide having an electron-with-
We investigated metal reductants. Mg powder (4 mol equiv)
gave 1a quantitatively (Table 2, entry 1). Mg turnings (Grignard
reaction grade) also gave 1a in 85% yield (entry 2). The reduction
did not occur with activated zinc powder and aluminum chips to
recover 2a unchanged (entries 3 and 4).
A decrease of the amount of Mg powder brought about a
decrease of the yield of 1a to 93% (3 mol equiv), 78% (2 mol equiv),
and 45% (1 mol equiv), respectively (entries 5–7).
drawing group such as (4-cyanophenyl)diphenylphosphine oxide
2
0
(
2
(
2f) gave only a complex mixture. Alkyldiarylphosphine oxides
g and 2h (entries 7 and 8) and dialkylarylphosphine oxide 2i
entry 9) were also reduced with the Mg–Me SiCl–DMI system to
afford the corresponding phosphine derivatives, 1g, 1h, and 1i, in
3
2
1
good yields. On the other hand, reduction of trialkylphosphine
oxide 2j did not proceed at all to recover 2j quantitatively.
1
8
A plausible mechanism is shown in Scheme 2. At the initial
stage of the reaction, one-electron reduction of 2 with Mg would
occur to afford the corresponding anion radical 4. The anion radical
Me
not proceed at all in the absence of Me
From these results, the optimized conditions were as follows:
Mg (4 mol equiv), Me SiCl (3 mol equiv), DMI, room temp. The
reduction proceeded in larger scale: to a mixture of Mg powder
45 mmol, 1.5 mol equiv), Me SiCl (80 mmol, 2.6 mol equiv), and
3
SiCl was indispensable for the reduction: the reduction did
SiCl.19
3
4
3
would be trapped immediately with Me SiCl to afford silylated
radical 5. Further one-electron reduction of 5 followed by elimina-
tion of hexamethyldisiloxane would occur to afford 1.
3
(
3
In conclusion, a direct reduction of phosphine oxides 2 to the
corresponding phosphines 1 was performed successfully by using
DMI (20 mL) was added 2a (8.40 g, 30 mmol). After short induction
period (ca. 5 min), a heavy exothermic reaction occurred. The
whole mixture was stirred for 2 h with cooling by a water bath.
After the workup, the reaction mixture was analyzed by P NMR
to find that 1a and 2a were obtained in 94:6 ratio. The desired
3
Mg–Me SiCl–DMI system. Though DMI was a special solvent to
31
promote the reduction, detail of the effect of DMI is still not clear
2
2–25
at present and will be discussed in near future.
The reduction
could be applicable to simple phosphine oxides as well as more
complicated phosphine oxides. Since Mg is stable and easy to han-
dle, and not so expensive, the present method gives a way to
recycle use of expensive phosphine reagents in organic syntheses.
1
a (7.08 g, 26 mmol, 87% yield) was obtained after purification
by the silica gel column chromatography.
Table 2
Effect of metal reductantsa
Entry
Metal (mmol)
Ratiob (%)
1
a
2a
R1
1
+
e-
.
R
.
1
2
3
4
5
6
7
Mg powder (8)
98 (96)c
85
n.d.
1
2
R2
P
O-
2
R
P
O SiMe3
Mg turnings (8)
15
100
100
7
22
55
R3
4
Zn powder (8)d
e
f
f
R3
Me Si Cl
5
e
3
Al chips (8)
n.d.
Mg powder (6)
Mg powder (4)
Mg powder (2)
93
78
45
1
+ e-
R
-
2
R
P
O
SiMe
P-O bond cleavage
3
R3
a
b
c
d
e
f
Me Si Cl
Compound 2a (2 mmol), Metal, Me
Determined by P NMR.
Isolated yield is shown in the parenthesis.
Zinc was activated by washing with aq dil HCl.
Not detected.
3
SiCl (6 mmol), DMI (8 mL), rt, 2 h.
6
3
31
1
+
Me Si O SiMe3
3
3
Scheme 2. A plausible mechanism of reduction of 2 to 1 by using Mg–Me SiCl–DMI
No peaks other than 2a were observed.
system.