Reduction of tertiary phosphine oxides with DIBAL-H

Article abstract of DOI:10.1021/jo7024064

(Chemical Equation Presented) The reduction of tertiary phosphine oxides (TPOs) and sulfides with diisobutylaluminum hydride (DIBAL-II) has been studied in detail. An extensive solvent screen has revealed that hindered aliphatic ethers, such as MTBE, are optimum for this reaction at ambient temperature. Many TPOs undergo considerable reduction at ambient temperature and then stall due to inhibition. ³¹P and ¹³C NMR studies using isotopically labeled substrates as well as competition studies have revealed that the source of this inhibition is tetraisobutyldialuminoxane (TIBAO), which builds up as the reaction proceeds. TIBAO selectively coordinates the TPO starting material, preventing further reduction. Several strategies have been found to circumvent this inhibition and obtain full conversion with this extremely inexpensive reducing agent for the first time. Practical reduction protocols for these critical targets have been developed.

Full text of DOI:10.1021/jo7024064

Reduction of Tertiary Phosphine Oxides with DIBAL-H
Carl A. Busacca,* Ravinder Raju, Nelu Grinberg, Nizar Haddad, Paul James-Jones,
Heewon Lee, Jon C. Lorenz, Anjan Saha, and Chris H. Senanayake
Department of Chemical DeVelopment, Boehringer-Ingelheim Pharmaceuticals, Inc.,
900 Ridgebury Road, Ridgefield, Connecticut 06877
cbusacca@rdg.boehringer-ingelheim.com
ReceiVed NoVember 7, 2007
The reduction of tertiary phosphine oxides (TPOs) and sulfides with diisobutylaluminum hydride (DIBAL-
H) has been studied in detail. An extensive solvent screen has revealed that hindered aliphatic ethers,
such as MTBE, are optimum for this reaction at ambient temperature. Many TPOs undergo considerable
reduction at ambient temperature and then stall due to inhibition. 31P and 13C NMR studies using
isotopically labeled substrates as well as competition studies have revealed that the source of this inhibition
is tetraisobutyldialuminoxane (TIBAO), which builds up as the reaction proceeds. TIBAO selectively
coordinates the TPO starting material, preventing further reduction. Several strategies have been found
to circumvent this inhibition and obtain full conversion with this extremely inexpensive reducing agent
for the first time. Practical reduction protocols for these critical targets have been developed.
Introduction
Results and Discussion
Tertiary phosphines are the single most important compound
class for metal ligation, and they are often prepared by reduction
of the corresponding tertiary phosphine oxides (TPOs).1 We
have recently developed the use of diisobutylaluminum hydride
(DIBAL-H) and triisobutylaluminum (TIBA) for the reduction
of secondary phosphine oxides (SPOs) and shown that these
reductants have superior substrate scope, mildness of conditions,
and cost2d when compared to other methods.2 We wished to
evaluate tertiary phosphine oxides with these reductants; in
addition, the scope and limitations of this reduction had never
been determined. Unlike SPOs, the reduction of TPOs with
neutral organoaluminums does have some limited precedent in
the patent literature, specifically in three patents.3 In the first,
from 1977, the single substrate triphenylphosphine oxide was
examined in a total of five solvents. In all cases, the reductions
cited did not go to completion. Heptane was chosen as the
optimum solvent in that study since the unreacted starting
material had the lowest solubility in heptane and could therefore
be removed by filtration after quenching.3a This is likely the
reason that DIBAL-H has not previously been utilized for TPO
reductions.
We started our research into TPO reductions with a solvent
study. Solvents can of course exert a profound effect on reactions
and are one of the simplest reaction parameters to systematically
(1) For recent, selected examples, see: Cl3SiH: (a) Xie, J.-H.; Duan,
H.-F.; Fan, B.-M.; Cheng, X.; Wang, L.-X.; Zhou, Q.-L. AdV. Synth. Catal.
2004, 346 (6), 625. (b) RajanBabu, T. V.; Nomura, N.; Jin, J.; Nandi, M.;
Park, H.; Sun, X. J. Org. Chem. 2003, 68 (22), 8431. (c) Baillie, C.; Xiao,
J. Tetrahedron 2004, 60 (19), 4159. Cl3SiH/PPh3: (d) Wu, H.-C.; Yu, J.-
Q.; Spencer, J. B. Org. Lett. 2004, 6 (25), 4675. BH3: (e) Keglevich, G.;
Chuluunbaatar, T.; Ludanyi, K.; Toke, L. Tetrahedron 1999, 56 (1), 1.
PhSiH3: (f) Botman, P. N. M.; Fraanje, J.; Goubitz, K.; Peschar, R.;
Verhoeven, J. W.; van Maarseveen, J. H.; Hiemstra, H. AdV. Synth. Catal.
2004, 346 (7), 743. (g) Pakulski, Z.; Koprowski, M.; Pietrusiewicz, K. M.
Tetrahedron 2003, 59 (41), 8219. MeOTf/LAH: (h) Imamoto, T.; Kikuchi,
S.-i.; Miura, T.; Wada, Y. Org. Lett. 2001, 3 (1), 87. (i) Guillen, F.; Rivard,
M.; Toffano, M.; Legros, J.-Y.; Daran, J.-C.; Fiaud, J.-C. Tetrahedron 2002,
58 (29), 5895. Ti(O-i-Pr)4/(EtO)3SiH: (j) Onodera, G.; Matsumoto, H.;
Milton, M. D.; Nishibayashi, Y.; Uemura, S. Org. Lett. 2005, 7 (18), 4029.
(k) Allen, A., Jr.; Ma, L.; Lin, W. Tetrahedron Lett. 2002, 43 (20), 3707.
(l) Coumbe, T.; Lawrence, N. J.; Muhammad, F. Tetrahedron Lett. 1994,
35 (4), 625. (m) Camps, P.; Colet, G.; Font-Bardia, M.; Munoz-Torrero,
V.; Solans, X.; Vazquez, S. Tetrahedron: Asymmetry 2002, 13 (7), 759.
Ti(O-i-Pr)4/PMHS: (n) Yin, J.; Buchwald, S. L. J. Am. Chem. Soc. 2000,
122 (48), 12051. AlH3: (o) Dandapani, S.; Curran, D. P. Tetrahedron 2002,
58 (20), 3855. (p) Bootle-Wilbraham, A.; Head, S.; Longstaff, J.; Wyatt,
P. Tetrahedron Lett. 1999, 40 (28), 5267. K/NH3: (q) Nycz, J.; Rachon, J.
Phos. Sulf. Sil. Rel. Elem. 2000, 161, 39.
* To whom correspondence should be addressed.
10.1021/jo7024064 CCC: $40.75 © 2008 American Chemical Society
Published on Web 01/16/2008
1524
J. Org. Chem. 2008, 73, 1524-1531

Reduction of Tertiary Phosphine Oxides with DIBAL-H
TABLE 1. Ambient Temperature Solvent Screening
entry
solvent
THF-d8
1-fluoroheptane
2,2-dimethyl THF
N-methylimidazole
DME
% conva
entry
solvent
PMPe
thiophene
tetrahydropyran
1,4-cineole
furan-d4
% conva
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
0
0
0
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
51
51
51
52
52
52
52
54
54
59
59
60
61
62
62
65
65
65
72
72
72
4
12
15
17
19
20
21
31
35
35
37
40
41
44
48
49
50
50
51
51
TEA
ETBEb
PhCl
anisole
t-Bu-SMe
thioanisole
t-amyl OMe
PhMe-d8
C6D6
Et2O-d10
none (neat)
p-xylene-d10
HFE-7100g
(+)-fenchyl OMe
i-Pr2O
n-Bu3P
2,2,5,5-TMTHFc
CDCl3
2-Me THF
2,5-dimethyl THF
tetrachloroethane-d2
PhNMe2
CD2Cl2
1-chloroheptane
PhF
PhCF3
cineole
C6D12
isoborneol OMe
c-Pent-OMef
MTBE
DMO OMed
p-dioxane-d8
n-Bu2O
a As determined by 31P NMR integration (standard delay time, d1 ) 2 s). b Ethyl tert-butyl ether. c Tetramethyl THF. d 3,7-Dimethyloctanol methyl ether.
e Pentamethylpiperidine. f Cyclopentyl methyl ether. g Perfluorobutyl methyl ethers.
vary.4a We chose not to limit ourselves to a few solvents as
was earlier done,3a but to cast a very wide net in an attempt to
gain as much understanding as possible. Our requirements for
inclusion into the solvent study were very simple: the material
should be a liquid near ambient temperature, and it should not
obviously react with DIBAL-H. We chose as our protypical
substrate diphenylethylphosphine oxide 1, and our initial screen-
ing conditions were 4 equiv of DIBAL-H at ambient temper-
ature. The results of this completely empirical approach, using
a total of 44 solvents, are shown in Table 1. The results are
listed in increasing order of percent conversion, as determined
by 31P NMR.
aliphatic ethers give higher conversion than unhindered aliphatic
ethers provided there are not two R-quaternary centers, and (8)
aliphatic ethers hindered on one side and unhindered on the
other side give the highest conversions (ca. 70%). MTBE and
cyclopentyl methyl ether4b fall in the last category. It is
interesting to note that performing the reaction neat in DIBAL-H
(entry 36) does not provide the highest conversion and that THF
(entry 1), a very common ethereal solvent, gives no conversion.
A typical 31P NMR reaction mixture spectrum (here C6D6) is
shown in Figure 1. The product phosphine is observed at δ -11
ppm. The resonanace for free phosphine oxide 1 is δ 32 ppm,
which is not observed in the reaction mixture. A cluster of 31
P
Several solvent trends were identified by analysis of the
ambient temperature data: (1) aliphatic hydrocarbons give
slightly higher conversions than aromatic hydrocarbons, (2)
halocarbons generally give poor conversion, (3) chlorocarbons
give better conversion than fluorocarbons, (4) hindered aliphatic
amines give better conversion than unhindered amines, (5) ethers
generally give better conversions than thioethers, (6) aliphatic
ethers give better conversion than aromatic ethers, (7) hindered
signals around δ 51 ppm are observed in this solvent and similar
behavior was seen in all of the solvents screened. When samples
such as these are quenched with aqueous NaOH, these multiple
downfield signals all convert to one signal for the starting
material (δ ∼32 ppm). These resonances are in fact “trapped”
starting material, as will be discussed shortly. Two critical
observations were made as a result of this solvent screen:
In most cases, no significant reduction occurred after 24 h
of reaction time, and in no case could additional conversion be
obtained by adding excess DIBAL-H at the 24 h time point.
In short, the reductions all stalled and could not be pushed
to completion at ambient temperature. The reductions were
clearly being inhibited, and the question was how this was
occurring.
(2) (a) Busacca, C. A.; Lorenz, J. C.; Grinberg, N.; Haddad, N.; Hrapchak,
M.; Latli, B.; Lee, H.; Sabila, P.; Saha, A.; Sarvestani, M.; Shen, S.;
Varsolona, R.; Wei, X.; Senanayake, C. H. Org. Lett. 2005, 7 (19), 4277.
(b) Busacca, C. A.; Lorenz, J. C. U.S. Pat. 7,256,314, 2007. Int. Appl. No.
PCT/US05/01526. (c) Busacca, C. A.; Lorenz, J. C.; Sabila, P.; Haddad,
N.; Senanayake, C. Org. Synth. 2007, 84, 242. (d) On a 100 kg scale, the
cost of DIBAL-H is ∼$1.50/mol from Albemarle.
(3) (a) GB 1,520,237, Appl. No. 0/18850/77, 1977. (b) Frey, F. W.; Lee,
J. Y. U.S. Patent 4,507,503, 1985. (c) Frey, F. W.; Lee, J. Y. U. S. Patent
4,507,504, 1985. (d) See also: Self, M. F.; Sangokoya, S. A.; Pennington,
W. T.; Robinson, G. H. J. Coord. Chem. 1990, 21 (4), 301.
Our initial working mechanistic model for the reduction of
TPO’s with DIBAL-H is shown in Scheme 1. It consists of
hydroalumination of the PdO bond to generate pentavalent
intermediate 3, followed by loss of H2 to furnish the product
phosphine 4. This mechanism would require 2 equiv of
DIBAL-H and would generate 1 equiv of tetraisobutyldialumi-
(4) (a) Reichardt, C. SolVents and SolVent Effects in Organic Chemistry;
Wiley-VCH: Weinheim, 2003. (b) Zeon Corporation, www.zeon.co.jp; This
solvent is undergoing commercialization. U.S representative: BetaChem,
contact: lu.john@beta-chem.com.
J. Org. Chem, Vol. 73, No. 4, 2008 1525

Busacca et al.
FIGURE 2. Reaction inhibition by aluminoxane.
SCHEME 2. Composition at 67% Conversion
FIGURE 1. 31P NMR spectra of TPO 1 reduction.
SCHEME 1. Initial Mechanistic Model of Reduction
The relative Lewis acidities of a number of organoaluminum
compounds toward the Lewis base PhOEt have been studied
1
by H NMR.5 DIBAL-H was found to be a better acid than
TIBAO toward this ether. This situation clearly does not extend
fully to tertiary phosphine oxides, however, or the buildup of
TIBAO in the reaction mixture would be of no consequence
and no inhibition would be observed.
Electron-rich aliphatic ether solvents might provide the
highest conversion under these conditions because they are
capable of displacing the TPO from TIBAO to a significant,
though not complete, extent. The preeminence of ethers hindered
on one side to positively affect these complexes is a subtlety
that we presently have no way of probing directly.
The observed effects with aliphatic ethers suggested a possible
strategy for increasing ambient temperature conversion: addition
of a “decoy” TPO that was more electron-rich than the TPO
substrate undergoing reduction. We originally chose to study
this by 31P and 13C NMR using 13C-labeled Ph2MePdO (5) as
substrate and commercially available tri-n-butylphosphine oxide
(TBPO, 6) as the decoy TPO. However, TBPO is in fact
extensively reduced at ambient temperature, as will be discussed
later. We therefore examined the reduction of 5 in the presence
noxane (TIBAO) per equivalent of tertiary phosphine produced.
Based on this model, the tertiary phosphine product and TIBAO
were both possible candidates for the inhibition observed. The
reduction of 1 to 2 was therefore carried out in toluene-d8 in
which 2 equiv of the product phosphine 2 were added to 1 prior
to addition of DIBAL-H. Quantitative 31P NMR after 24 h
showed the same amount of unreacted TPO was observed as
was seen in the control experiment without added phosphine.
Clearly, phosphine product inhibition was not taking place. We
next added 2 and 3 equiv of TIBAO as the commercial toluene
solution (0.29 M) to TPO 1 in toluene-d8, prior to the DIBAL-H
charge. The results are shown in Figure 2, in which percent
conversion (by 31P NMR) is plotted as a function of time.
Reaction inhibition is clearly seen, and 3 equiv of TIBAO
profoundly retards the reaction.
It is worth considering the stoichiometry of the reaction
mixture at ca. 2/3 conversion (Scheme 2): the ratio of TIBAO/
TPO at this conversion level would be 2:1, and significant
reaction inhibition would be expected.
(5) (a) Ushakova, T. M.; Meshkova, I. N.; Markevich, M. A. Organomet.
Chem. USSR 1988, 1 (4), 513.
1526 J. Org. Chem., Vol. 73, No. 4, 2008

Reduction of Tertiary Phosphine Oxides with DIBAL-H
and absence of the more sterically hindered decoy tricyclo-
hexylphosphine oxide (c-Hex3PdO, 7), which is also com-
mercially available. The proton-decoupled 31P spectrum of 5
(Supporting Information Figure 1A) is a doublet at δ 39.4 ppm,
with 1JPC ) 73 Hz for coupling to 13CH3. The 13C spectrum of
the methyl region of 5 shows the methyl signal to be a doublet
at δ 16.7 ppm with the same coupling constant. It is well-known
that 1JPC is a sensitive probe of phosphorus geometry.6 1JPC of
tetracoordinate phosphorus compounds like TPOs are generally
1
about 1 order of magnitude larger than JPC for tricoordinate
phosphorus compounds, such as tertiary phosphines.
A
FIGURE 3. Possible structure of inhibitory complex.
13C-labeled substrate was chosen so that 31P resonances
SCHEME 3. SPO Solvent Effect
associated with the substrate, which would be doublets with
J’s of ca. 75 Hz, could be readily differentiated from resonances
associated with the decoy TPO which was not isotopically
labeled and would therefore be present as singlets. In addition,
the geometric dependence of 1JPC mentioned above would also
provide insight into the structure of reaction intermediates. We
first reduced 5 with 4 equiv of DIBAL-H in C6D12 for 4 h. The
reaction stalled at ∼65% conversion, as expected. The product,
Ph213CH3P (8), was observed at δ -24 ppm as a doublet with
1
dramatically reduced JPC of 9 Hz, typical of tricoordinate
phosphorus. No uncomplexed starting material (δ 39 ppm) was
observed. We focused on the 31P signals at ∼49 ppm, since we
knew these were resonances of the “aluminoxane-trapped”
starting material, as determined by reaction quenching men-
tioned above. Two major signals were observed. Examination
of the resonances (Supporting Information Figure 2A) instantly
showed that both signals were doublets, with 1JPC values of 76
and 74 Hz: these are clearly starting material-derived reso-
nances, since only tetracoordinate phosphorus species will
possess these large coupling constants.
The reaction of labeled substrate 5 was then repeated, yet 1
equiv of the decoy c-Hex3PdO was added at the outset, prior
to addition of DIBAL-H. The 31P NMR spectrum (Supporting
Information Figure 2B) at the same 4 h time point showed the
product at δ -25 ppm, a new cluster of intense peaks at ∼67
ppm, and the absence of the resonances at ca. 49 ppm. Close
examination of the downfield resonances showed that all of them
were singlets. The signals near 67 ppm are therefore confirmed
to be derived from the unlabeled decoy, c-Hex3PdO. The NMR
conversion at 4 h was 95%, compared to 65% without the decoy.
As expected, NMR thus confirms displacement of the trapped
substrate TPO by the superior Lewis base c-Hex3PdO, validat-
ing the decoy concept. It should also be noted that no
tricyclohexylphosphine was produced under these reaction
conditions, and the decoy can be easily recovered and recycled
if desired. The 31P NMR spectrum of labeled substrate 5 in the
presence of 2 equiv of TIBAO only did not show the downfield
doublets of the reduction in the absence of decoy. If both TIBAO
and DIBAL-H were added to 5, however, a spectrum nearly
identical to the nondecoy reduction mixture was generated. It
seems possible therefore that the inhibitory complex contains
both TIBAO and DIBAL-H. One possible structure for this
complex is shown in Figure 3.
elimination of the TIBAO component could be induced, likely
at elevated temperature. Reduction of 1 in mesitylene-d12 was
studied first by 31P NMR at temperatures greater than ambient.
Following these reactions by NMR allowed us to determine with
precision the reaction times and temperatures required to achieve
full conversion. The temperature was slowly raised until
significant reduction was observed. 31P NMR spectrum of this
crude reaction mixture after 14 h at 120 °C was remarkably
clean (Supporting Information Figure 3), as only a single
resonance for the product (2) was observed. At 110 °C, the
boiling point of toluene, full conversion was not achieved. The
curious situation thus exists in that one can obtain 50-60%
yield by stirring the reaction overnight at ambient temperature,
yet only by overnight reaction at 120 °C can >90% yield be
obtained under purely thermal conditions.
It is known that hydroaluminations of unsaturated systems
are fastest in hydrocarbon solvents, where DIBAL-H is “naked”
and uncomplexed, while these reactions are far slower in donor
solvents.7 In hydrocarbons such as mesitylene, hydroalumination
is probably fast, yet the TIBAO-TPO complex is more stable,
and thus high temperatures are required. Conversely, it seems
likely that the use of ethers here probably slows down hydro-
alumination. This effect is apparently more than compensated
for, however, by efficient disruption of the TIBAO inhibitory
complex, leading to oVerall reduction under milder conditions.
Our previous work with secondary phosphine oxides did not
include such a broad solvent screen as employed here, since
SPOs are readily reduced with DIBAL-H in all organic solvents
examined. We did encounter one SPO substrate that reduced
slowly, however, due to chelation with DIBAL-H, namely bis-
(2-methoxyphenyl)phosphine oxide 9. In THF, 8 h at 35 °C
was needed to obtain full conversion to phosphine 10. Since
THF was one of the worst solvents for TPO reductions, we
repeated this reduction in MTBE-d3, the optimum ambient
temperature TPO reduction solvent. As shown in Scheme 3,
the results were dramatic: the reaction time was reduced from
8 h to 30 min simply by changing the solvent to MTBE.
We also wanted to examine an alternate method to obtain
full conversion in these reductions. Our basic premise was that
the TIBAO-TPO complex would be destroyed if â-hydride
(6) (a) Chou, W.-N.; Pomerantz, M. J. Org. Chem. 1991, 56 (8), 2762.
(b) Quin, L. D.; Verkade, J. G. Phosphorus-31 NMR Spectral Properties
in Compound Characterization and Structural Analysis; Wiley-VCH:
Hoboken 1994, p197; p 98. (c) Gorenstein, D. G. Phosphorus-31 NMR;
Academic Press: Orlando, 1984; pp 37-42.
(7) (a) Eisch, J. J.; Fichter, K. C. J. Am. Chem. Soc. 1974, 96 (21), 6815.
(b) Eisch, J. J.; Rhee, S.-G. J. Am. Chem. Soc. 1974, 96 (23), 7276. (c)
Eisch, J. J.; Foxton, M. W. J. Org. Chem. 1971, 36 (23), 3520.
J. Org. Chem, Vol. 73, No. 4, 2008 1527

Busacca et al.
TABLE 2. Reduction of Ph3PdO
TABLE 4. Reduction of Ph2-c-HexPdO
entry
solvent
C6D12
MTBE-d3
C6D12
C6D12
C6D12
T (°C)
time (h)
decoy
% conva
entry
solvent
C6D6
MTBE-d3
C6D12
C6D12
C6D12
C6D12
C6D12
c-Pent-OMe
C6D12
c-Pent-OMe
C6D12
T (°C)
time (h)
decoy
% conva
1
2
3
4
5
6
7
23
23
23
23
72
72
90
14
20
14
14
6/20
6
60
48
80
1
2
3
4
5
6
7
8
9
23
23
23
23
72
23
50
50
50
50
72
100
14
14
14
14
6/20
14
4
c-Hex3PdO
38
41
56
c-Hex3PdO
NMO
73
c-Hex3PdO
76
76/83
100
90
78/83
56
C6D12
c-Pent-OMe
c-Hex3PdO
NMO
c-Hex3PdO
30
85
66
94
67
100
87
4
a Percent conversion by 31P NMR integration.
20
20
7
c-Hex3PdO
c-Hex3PdO
10
11
12
TABLE 3. Reduction of Ph2CH3PdO
c-Pent-OMe
7
a Percent conversion by 31P NMR integration.
% conva
TABLE 5. Reduction of Ph2-t-BuPdO
entry
solvent
T (°C)
time (h)
decoy
1
2
3
4
5
6
7
8
MTBE-d3
MTBE-d3
C6D6
C6D12
C6D12
MTBE-d3
C6D12
C6D12
23
23
23
23
30
30
72
72
8
16
72
c-Hex3PdO
c-Hex3PdO
c-Hex3PdO
36
76
78
60
86
entry
solvent
T (°C) time (h)
decoy
% conva
4/20
4/20
6/20
4/20
c-Hex3PdO
c-Hex3PdO
90/100
89/99
93/100
95/99
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
MTBE-d3
C6D6
23
23
14
14
3
16
3
16
2
5
0
4
c-Hex3PdO
Et-c-Hex
Et-c-Hex
p-xylene-d10
p-xylene-d10
n-Bu2O
125
125
125
125
120
120
120
140
72
53
82
45
56
25
45
44
36
52
33
3
a Percent conversion by 31P NMR integration.
We felt that we had sufficient understanding of the parameters
of the reduction at this point to examine a series of TPO
substrates and to learn what conditions were required for their
successful reduction or where the limitations of this reduction
methodology would be found.
We first examined triaryl TPOs, which would be expected
to be the most reactive substrate class as they are not electron-
rich. The conversion results for the prototypical substrate
triphenylphosphine oxide (11) are shown in Table 2. Ambient
temperature conversion to Ph3P (12) in the presence of the decoy
was 80% (entry 3) and 100% conversion was achieved in 6 h
at 72 °C with the decoy (entry 7). NMO (entry 4) provided
slightly higher conversion than the reaction without decoy,
though it was less efficient than c-Hex3PdO.
Diphenylmethylphosphine oxide 13 was similarly examined
by 31P NMR, as shown in Table 3. Though more electron-rich
than the triaryl substrates, it was reduced more easily. Tweny
hours at 30 °C (entry 5) and 4 h at 72 °C with c-Hex3PdO
gave 100% and 95% conversion respectively for this substrate.
We next examined diphenylcyclohexylphosphine oxide 15
in some detail, and the conversion results are shown in Table
4. In the presence of the decoy TPO, hydrocarbon solvents were
found to be superior to ethers. This may be attributable to the
formation of a tight decoy TPO-TIBAO complex and thus
greater reduction of the freed substrate. High conversions were
obtained in 20 h at 50 °C (entry 9) and 7 h at 72 °C (entry 11)
with added c-Hex3PdO. The decoy lowers both the reaction
temperature and reaction time required to reach completion.
We then examined the more challenging diphenyl-tert-
butylphosphine oxide substrate 17, and the results are shown
in Table 5. The beneficial effect of added c-Hex3PdO observed
with all of the substrates described above was not observed with
this substrate. Attempted use of a decoy more sterically hindered
IB-OMeb
IB-OMeb
IB-OMeb
C6D12
C6D12
C6D12
14
4
7
c-Hex3PdO
c-Hex3PdO
Buchwald(O)c
72
72
21
5
c-Pent-OMe
c-Pent-OMe
c-Pent-OMe
100
100
100
2
8
18
55
66
55
a Percent conversion by 31P NMR integration. b Isoborneol methyl ether.
c Biphenyl-2-dicyclohexylphospine oxide.
than c-Hex3PdO (entry 13) was unsuccessful. In both hydro-
carbons and ethers, some product (18) decomposition was
observed after prolonged reaction times. The best reaction
conversion, 82%, was observed after 16 h at 125 °C (entry 4)
in ethylcyclohexane.
We next evaluated a series of dialkyl and trialkyl TPOs, and
these results are shown in Table 6. Phenyldimethylphosphine
oxide 19 was readily reduced (entries 1-6) under a variety of
conditions. The more sterically hindered dicyclohexyl substrate
20 was much more resistant. Sixteen hours at 125 °C were
required to obtain 95% conversion for this substrate (entry 8).
Reduction of the biphenyl-based Buchwald phosphine oxide8
21 gave only 17% conversion at 125 °C, as shown in entry 9.
Extremely sterically hindered substrates such as these are clearly
limitations of this reduction technology. The racemic TPO
PhMe-t-BuPdO (22) was completely reduced in 6 h at 72 °C
with decoy (entry 12), yet was untouched at ambient temperature
(entry 10). Tri-n-butylphosphine oxide 23 was significantly
reduced at ambient temperature in the absence of decoy (entry
(8) Prepared by oxidation of the commercial ligand; see the Experimental
Section.
1528 J. Org. Chem., Vol. 73, No. 4, 2008

Reduction of Tertiary Phosphine Oxides with DIBAL-H
TABLE 6. Reduction of Dialkyl and Trialkyl TPOs
entry
TPO
R1
R2
Me
Me
Me
Me
Me
Me
c-Hex
c-Hex
c-Hex
Me
Me
Me
n-Bu
n-Bu
Me
R3
Me
Me
Me
Me
Me
Me
c-Hex
c-Hex
c-Hex
t-Bu
t-Bu
t-Bu
n-Bu
n-Bu
Me
solvent
T (°C)
time (h)
decoy
% conva
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
19
19
19
19
19
19
20
20
21
22
22
22
23
23
24
phenyl
phenyl
phenyl
phenyl
phenyl
phenyl
phenyl
phenyl
2-biphenyl
phenyl
phenyl
phenyl
n-Bu
MTBE-d3
MTBE-d3
c-PentOMe
MTBE-d3
C6D12
50
50
100
23
72
72
125
125
125
23
50
72
16
20
16
16
6/20
6
4
16
4
16
20
6
14
14
15
54
100
87
c-Hex3PdO
c-Hex3PdO
c-Hex3PdO
26
93/100
100
79
C6D12
Et-c-Hex
Et-c-Hex
Et-c-Hex
MTBE-d3
MTBE-d3
C6D12
C6D12
C6D12
C6D12
95
17
0
25
100
68
94
c-Hex3PdO
c-Hex3PdO
c-Hex3PdO
23
23
23
n-Bu
c-Hex3PdO
Me
100
a Percent conversion by 31P NMR integration.
TABLE 7. One-Gram-Scale Reductions of TPOs
entry
TPO
R1
C6H5
R2
C6H5
R3
solvent
T (°C)
time (h)
decoy TPO
TP
yielda (%)
1
2
3
4
5
6
7
8
9
10
11
12
13
11
11
31
32
13
1
15
41
22
19
20
23
24
C6H5
C6H5
4-CF3C6H4
4-ClC6H4
CH3
C2H5
c-Hex
CH3
t-Bu
CH3
C6H12
C6H12
C6H12
C6H12
C6H12
C6H12
C6H12
C6H12
C6H12
C6H12
Et-c-Hex
C6H12
C6H12
72
72
72
72
72
72
72
72
72
72
125
72
23
7
7
7
7
8
14
16
7
7
4
c-Hex3PdO
c-Hex3PdO
c-Hex3PdO
c-Hex3PdO
12
12
36
37
14
2
16
42
28
25
26
29
30
91
C6H5
C6H5
99b
93
4-CF3C6H4
4-ClC6H4
C6H5
C6H5
C6H5
C6H5
C6H5
C6H5
C6H5
4-CF3C6H4
4-ClC6H4
C6H5
C6H5
C6H5
3,5-F2C6H3
CH3
CH3
c-Hex
n-Bu
CH3
98
98b
91b
90b
98
c-Hex3PdO
c-Hex3PdO
89
97b
90b
97b
99b
c-Hex
n-Bu
CH3
15
4
15
n-Bu
CH3
a Isolated yield of pure material on ca. 1 g scale. b Yield of phosphine borane adduct.
13), and 94% conversion was achieved for this substrate under
the same conditions with decoy present (entry 14). The most
sterically accessible TPO possible, Me3PdO (24), was trivially
reduced at ambient temperature without decoy, as shown in entry
15.
tography then furnished the pure phosphine boranes, generally
in very high yield.
The triaryl substrates (entries 1-4), as well as the dipheny-
lalkylphosphine oxides (entries 5-7), were all reduced in >90%
isolated yield. Two racemic targets (entries 8-9) were produced
in 98% and 89% isolated yield, respectively, under the standard
conditions.
Having established the requirements for reduction of the
various substrate classes, we then carried out gram-scale
experiments (Table 7) so that isolated yields for these processes
could be obtained. Triphenylphosphine oxide 11 was reduced
in 91% isolated yield (entry 1) after 7 h reaction time, and in
situ conversion of the product to its phosphine borane9 adduct
gave the desired product in 99% isolated yield. The reaction
mixtures were quenched at ∼0 °C with aqueous NaOH, giving
two clear phases, as previously described for the reduction of
secondary phosphine oxides.2 The aqueous/organic partition can
thus be carried out under non-inert conditions, which is a
significant operational advantage. Standard silica gel chroma-
Two pairs of dialkyl- and trialkylphosphine oxides were then
similarly reduced (entries 10-13), in high yield, and all were
isolated as the phosphine borane adducts.
Until this point, no optically active phosphine oxides had been
examined in the reduction. We were able to prepare optically
enriched (+)-41 by chiral preparative chromatography (Chiralcel
AD). We then subjected (+)-41 thus isolated to DIBAL-H
reduction, in the presence and absence of c-Hex3PdO, and then
oxidized the phosphine products back to the oxide. As shown
in Scheme 4, starting with (+)-41 of 79% ee led to regenerated
(+)-41 of only 10% ee in each case. Phosphine oxidation with
H2O2 is known to proceed with retention of configuration.10 In
at least some cases, the DIBAL-H reduction of tertiary phos-
phine oxides thus proceeds with retention of configuration, yet
very low stereoselectivity, probably due to pseudorotation of
(9) For reviews, see: (a) Imamoto, T.; Pure Appl. Chem. 1993, 65(4),
655-660. (b) Ohff, M.; Holz, J.; Quirmbach, M.; Boerner, A. Synthesis
1998, 10, 1391-1415. (c) Brunel, J. M.; Faure, B.; Maffei, M. Coord. Chem.
ReV. 1998, 178-180 (Part 1), 665-698. (d) Carboni, B.; Monnier, L.
Tetrahedron 1999, 55 (5), 1197-1248.
J. Org. Chem, Vol. 73, No. 4, 2008 1529

Busacca et al.
SCHEME 4. Reduction of Optically Active TPO
SCHEME 6. Reduction of Difunctional TPOs
SCHEME 5. Reduction of Phosphine Sulfides and Arsine
Oxides
SCHEME 7. Reactions of Bis-phosphineoxides
the pentacoordinate phosphorus intermediate (3 in Scheme 1).11
Pseudorotation is a low-energy process in which axial and
equatorial substituents are exchanged, pairwise, from a trigonal
bipyramidal structure. The existence of a pentacoordinate
intermediate does not guarantee racemization, although multiple
successive pseudorotations will cause racemization of an
organophosphorus compound.
We concluded our determination of the scope and limitations
of these reductions by examining phosphine sulfides, arsine
oxides, bis-phosphine oxides, and related substrates with a
potentially chelating second heteroatom. Triphenyl phosphine
sulfide (43) and BINAP-bis-sulfide 4412 were easily reduced at
50-100 °C, as shown in Scheme 5. It is known that phosphine
sulfides are easier to reduce than the analogous oxides with other
reducing agents,10,13 and the same holds true with DIBAL-H.
When triphenylphosphine sulfide was allowed to react with 2
equiv of TIBAO prior to the DIBAL-H charge, no inhibition
of the reduction was observed. It seems that the thiophilicity of
aluminum is far weaker than its oxophilicity. Triphenylarsine
oxide 46 was also trivially reduced, in 91% yield, at ambient
temperature to furnish arsine 47.
We then evaluated a series of potentially chelating substrates
including bis-phosphine monooxides (BPMOs)14 and amino
phosphineoxides. The monooxides of dppe and dppp were
readily reduced under conditions established for the analogous
TPOs as shown in Scheme 6. Amino-TPO 52 was also readily
converted to the desired aminophosphine 53 in excellent yield,
yet 15 h reaction time at 125 °C was required for full conversion.
The bis-oxide of BINAP15 (54) was treated with DIBAL-H
in a variety of solvents and studied by 31P NMR. 31P NMR
quickly established that this bisoxide forms an extremely stable
bidentate chelate with DIBAL-H (55, Scheme 7), as well as
other organoaluminum reagents such as triisobutylaluminum
(TIBA) and diisobutylaluminum chloride (DIBAL-Cl). The
complexes all resonate about 10 ppm downfield of the bisoxide
starting material. Clearly, both oxygens are required for this
inhibition, since the bisphosphine monooxides described above
are reduced without incident. Even at elevated temperature, no
(10) (a) Naumann, K.; Zon, G.; Mislow, K. J. Am. Chem. Soc. 1969, 91
(10), 2788. (b) Luckenbach, R. Tetrahedron Lett. 1976, 24, 2017. See also
ref 13a, p 300.
(11) a) Holmes, R. H. Pentacoordinate Phosphorus, V.I; American
Chemical Society: Washington 1980; pp 281-287. (b) Holmes, R. H.
Pentacoordinate Phosphorus, V.II; American Chemical Society: Wash-
ington 1980; pp 100-109, pp 154-155. For closely related results with
boron-based reductants, see: (c) Stankevic, M.; Pietrusiewicz, M. Synlett
2003, 7, 1012.
(12) Chapman, C. J.; Frost, C. G.; Gill-Carey, M. P.; Kociok-Kohn, G.;
Mahon, M. F.; Weller, A. S.; Willis, M. C. Tetrahedron: Asymmetry 2003,
14 (6), 705.
(13) (a) Quin, L. D.; Quin, G. S. A Guide to Organophosphorus
Chemistry; Wiley: New York, 2000; pp 116-117. (b) Yang, C.; Goldstein,
E.; Breffle, S.; Jin, S. THEOCHEM 1992, 91, 345-368.
(14) Grushin, V. V. Organometallics 2001, 20 (18), 3950.
(15) (a) Shimada, T.; Kurushima, H.; Cho, Y.-H.; Hayashi, T. J. Org.
Chem. 2001, 66 (26), 8854. For the reduction of dppe(O2) at elevated
temperature, see: (b) Self, M. F.; Sangokoya, S. A.; Pennington, W. T.;
Robinson, G. H. J. Coord. Chem. 1990, 21 (4), 301.
1530 J. Org. Chem., Vol. 73, No. 4, 2008

Reduction of Tertiary Phosphine Oxides with DIBAL-H
evolution of gas. The tube was then aged at ambient temperature
with an Ar line through the tube septum. The 31P spectrum were
then acquired with proton decoupling and a d1 delay of 5-10 s,
and the signals were integrated to determine percent conversion.
production of BINAP was observed. BINAP-bisoxide aluminum
complexes such as these have not been previously reported.
Although their potential use as chiral Lewis acids is intriguing,
they currently represent a limitation of this reductive methodol-
ogy. We were curious whether a bis-phosphine oxide with
sufficiently large bite angle might be successfully reduced. As
shown in Scheme 7, DIBAL-H in the presence of c-Hex3PdO
gave 100% conversion of DPPH(O2), 56, to DPPH in 2 h at
90 °C (Supporting Information Figure 4). Thus, if bidentate
chelation to aluminum becomes untenable, as in this six-carbon
spacer example, reduction of bis-phosphine oxides to bis-
phosphines can again be readily achieved.
In summary, the reduction of tertiary phosphine oxides by
DIBAL-H has been studied in detail for the first time. Significant
solvent effects have been found, and hindered aliphatic ethers
such as MTBE provide the best ambient temperature conver-
sions, while some common ethereal solvents such as THF are
ineffective. The origin of the inhibition of these reductions has
been uncovered, namely the aluminoxane that forms as a
byproduct of the reaction. More importantly, two different
strategies have been found to overcome this inhibition and obtain
full conversions in this transformation for the first time. A study
of the major structural classes of tertiary phosphine oxides has
been carried out, and in this way, the scope and limitations of
the reduction technology have been determined. Sterics, rather
than electronics, are the most important parameter for the
DIBAL-H reduction of tertiary phosphine oxides. Practical
experimental procedures, critically including the in situ conver-
sion of the tertiary phosphines to phosphine boranes, have been
developed which allow for the one-pot synthesis of these
important targets in high yield.
General Reduction Procedure. Diphenylcyclohexylphosphine
Borane 16. A 100 mL three-necked flask equipped with a
thermocouple and addition funnel was charged with 0.50 g of
phosphine oxide 15 (1.8 mmol, 1 equiv) and 0.52 g of c-Hex3Pd
O (1.8 mmol, 1 equiv). The flask was then evacuated/Ar filled (2×),
and 7.2 mL of 1 M DIBAL-H/C6H12 (7.2 mmol, 4 equiv) was added
via syringe. The reaction mixture was then placed in a pre-
equilibrated 72 °C oil bath under Ar. After 16 h, the mixture was
cooled to ambient temperature, and then 2.3 mL of 1 M BH3/THF
(2.3 mmol, 1.5 equiv) was added via syringe. The resultant mixture
was then stirred for 1 h at 23 °C, at which time 31P NMR showed
complete conversion to the phosphine borane. MTBE (14 mL) was
then added via syringe. After 5 min, the reaction mixture was cooled
to 0 °C, and then 11 mL of 6 N NaOH was slowly added
(CAUTION!:Vigorous reaction! Gas evolution!) via addition funnel.
The bath was then removed and the reaction mixture allowed to
warm to 23 °C. After 15 min, it was transferred to a separatory
funnel, and the phases were separated. The aqueous phase was
reextracted with MTBE (1 × 15 mL). The combined organics were
then dried (MgSO4) and filtered. The filtrate (unconcentrated) was
then passed through a pad of silica gel eluting with ∼100 mL of
MTBE. The solvents were then removed in vacuo, and the residue
was chromatographed on silica gel eluting with 4:1 Hex:CH2Cl2
to give 0.447 g of phosphine borane 16 (90%) as a colorless solid:
mp 93-94 °C; 1H NMR (400 MHz, CDCl3) δ 7.79-7.72 (m, 4H),
7.51-7.41 (m, 6H), 2.41 (q, J ) 13 Hz, 1H), 1.86-1.77 (m, 2H),
1.76-1.63 (m, 3H), 1.56-1.42 (m, 2H), 1.35-1.21 (m, 4H), 0.97
1
(q, 3H, JHB ) 89 Hz); 13C NMR (100 MHz, CDCl3) δ: 132.66
(d, JCP ) 8.4 Hz), 130.97 (d, JCP ) 2.3 Hz), 128.65 (d, JCP ) 9.6
1
Hz), 128.23 (s), 33.76 (d, JCP ) 36 Hz), 26.72 (t, JCP ) 12 Hz),
26.53 (t), 25.81 (t, JCP ) 1.2 Hz); 31P NMR (202 MHz, CDCl3) δ
20.93; HRMS (M - BH3 + H)+ C18H22P calcd 269.1453, obsd
269.1463, difference ) 3.4713 ppm. Anal. Calcd for C18H24BP:
C, 76.62; H, 8.57. Found: C, 76.50; H, 8.77.
Experimental Section
NMR Reactions. A 5 mm × 7 in. screw-cap NMR tube with
PTFE septum (Wilmad 535-TR-7) was charged with 46 mg of
diphenylethylphosphine oxide 1 (0.20 mmol, 1 equiv) and 0.80 mL
of solvent. The NMR tube was then agitated on a vortex mixer for
∼2 min, generally giving a suspension. To this was then added by
syringe 0.14 mL of neat DIBAL-H (CAUTION: Pyrophoric!, 0.80
mmol, 4 equiv). This mixture was then agitated for a full 5 min on
the vortex mixer, generally giving a clear, colorless solution, with
Supporting Information Available: Experimental procedures
and characterization data for all new and known compounds. This
material is available free of charge via the Internet at http://pubs.
acs.org.
JO7024064
J. Org. Chem, Vol. 73, No. 4, 2008 1531