Chromatography-free product separation in the Mitsunobu reaction

Article abstract of DOI:10.1016/j.tetlet.2006.05.070

Mitsunobu products can be isolated pure without column chromatography by first washing a solution of the crude reaction mixture in dichloromethane with 15% aqueous hydrogen peroxide followed by aqueous sodium sulfite. A final filtration through silica gel secures the pure Mitsunobu product.

Full text of DOI:10.1016/j.tetlet.2006.05.070

Tetrahedron Letters 47 (2006) 5151–5154
Chromatography-free product separation in the Mitsunobu reaction
a
b
b
a,
*
Anthony J. Proctor, Kevin Beautement, John M. Clough, David W. Knight
a
and Yingfa Li
a
School of Chemistry, Cardiff University, Main College, Park Place, Cardiff CF10 3AT, UK
Syngenta, Research Chemistry, Jealott’s Hill International Research Centre, Bracknell, Berkshire RG42 6EY, UK
b
Received 6 March 2006; revised 4 May 2006; accepted 11 May 2006
Abstract—Mitsunobu products can be isolated pure without column chromatography by first washing a solution of the crude reac-
tion mixture in dichloromethane with 15% aqueous hydrogen peroxide followed by aqueous sodium sulfite. A final filtration through
silica gel secures the pure Mitsunobu product.
ꢀ
2006 Elsevier Ltd. All rights reserved.
Despite enhanced levels of sophistication, chromatogra-
phy is increasingly a technique to be avoided if possible,
especially in these more environmentally conscious
times. Additionally, of course, it is often a tedious and
time-consuming procedure. This is most certainly true
and post-reaction conversion into insoluble material by
ring-opening metathetic polymerisation provides a solu-
tion which is useful in combinatorial chemistry at least.
3
However, the elaborate nature of many of these re-
agents, coupled with their lack of commercial availabil-
ity, probably precludes their use in ‘everyday’ synthesis.
1
of its use in the Mitsunobu reaction (Scheme 1), to such
2
4
an extent that a recent microreview has highlighted
Fundamentally a dehydration process, the Mitsunobu
chromatography-free procedures for this ubiquitous
and exceptionally useful reaction. The level of complex-
ity of some of these procedures attests both to the
importance of the Mitsunobu reaction and to the tedium
of most current separation procedures. Examples in-
clude the use of functionalised phosphines having pyr-
idyl, crown ether or fluorocarbon residues decorating
one of the aryl rings. Alternative azodicarboxylates also
feature the use of perfluorocarbon esterifying groups,
along with a number of otherwise highly modified deriv-
atives. Some of these are converted into relatively polar
by-products, which can assist with the final product
separation.
reaction is of central importance for ester formation
and as a means of introducing nitrogen into a molecule.
Usually proceeding with perfect S 2 inversion at a sec-
N
ondary alcohol site (1, Scheme 1), the reaction can also
be applied with great effect to various ring closures. It
has certainly been the saviour of many a synthetic
scheme which has produced the incorrect stereochemis-
try at a secondary stereogenic centre. The problem of
product isolation is generated by the typical reagent
combination of a tertiary phosphine, typically triphenyl-
phosphine, an azodicarboxylate, usually the diethyl or
diisopropyl ester (DEAD or DIAD, respectively), to-
gether with the reacting nucleophile. The latter is often
a carboxylic acid, hydrazoic acid or an amine derivative.
Unfortunately, it is often only after relatively Herculian
column chromatography that the desired product 2 can
be cleanly separated from the phosphine oxide and
hydrazine by-products and the (often slight) excess of
starting reagents which are necessary to obtain high
yields (Scheme 1). Further irritation is commonly
encountered by the presence of this slight excess of the
less polar phosphine, which is often of similar polarity
to the product 2.
Alternatively, the principle of ‘impurity annihilation’
using an azodicarboxylate having a norbornenyl residue
OH
R2
Nu
Ph3P, Nu-H
Ph3P=O
+
+
R1
R1
R2
RO2CN=NCO2R
RO2CNH-NHCO2R
1
2
Scheme 1.
*
Corresponding author. Fax: +44 02920 874 210; e-mail: knightdw@
cf.ac.uk
When we recently required a series of unsaturated, and
ultimately chiral, non-racemic, O-alkyl hydroxylamine
0040-4039/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2006.05.070

5152
A. J. Proctor et al. / Tetrahedron Letters 47 (2006) 5151–5154
O
O
O
N
O
N
HO
O
O
Mitsunobu
OH
N
O
O
+
HO
O
N
5
R1
R1
OH
R2
R2
O
Ph3P, DIAD
3
4
5
6
7
Scheme 2.
Scheme 3.
derivatives 3, a Mitsunobu reaction between (homochi-
ral) secondary propargylic alcohols 4 and N-hydroxy-
phthalimide 5 seemed the most viable option (Scheme
desired oxidations to completion. A typical procedure is
as follows for the reaction shown in Scheme 3.
2
). Although initial efforts delivered poor yields, an opti-
N-(3-Methylbut-2-en-1-yloxy)phthalimide 7: To a stirred
solution of triphenylphosphine (10.96 g, 41.785 mmol)
in dry tetrahydrofuran (300 ml), maintained at 0 ꢁC un-
der an atmosphere of dry nitrogen, was added DIAD
(6.92 ml, 35.17 mmol) dropwise. The resulting solution
was stirred at the same temperature for 15 min where-
upon it had become creamy white in colour. 3-Methyl-
but-2-en-1-ol 6 (3.00 g, 34.82 mmol) was next added
dropwise and stirring continued for 20 mins prior to
misation study revealed tetrahydrofuran to be a solvent
of choice and an optimum order of addition of triphen-
ylphosphine, DIAD, the alcohol 4 and, finally, the
hydroxy-phthalimide 5. Despite routinely isolating the
desired products 3 in yields in excess of 85%, there re-
mained the requirement for very careful column chro-
matography, made even more tedious by the need to
produce a diverse series of such hydroxylamines, in
order to properly test some novel chemistry.
the addition of N-hydroxyphthalimide
5 (5.76 g,
3
5.17 mmol) in one portion. The resulting mixture was
Even though our optimised recipe required only mar-
ginal reagent excesses (1.01–1.2 equiv) and, crucially tet-
rahydrofuran rather than the commonly used ethyl
acetate as solvent, in order to secure complete conver-
sion of the precursor propargylic alcohols 4 into the
phthalimides 3, contamination of the latter by small
quantities of triphenylphosphine was not uncommon,
despite due care being taken during the chromato-
graphic separation. This provoked the thought that per-
haps complete removal of this contaminant could be
facilitated by prior oxidation to the much more polar
phosphine oxide. Clearly, if such a tactic was to be via-
ble, the oxidising reagent would have to be readily avail-
able, cheap and, especially, easily destroyed afterwards,
as it would inevitably be present in some excess, because
of the uncertainty of precisely how much phosphine is
needed to be oxidised. We further speculated that such
an oxidant might interact with the by-product hydrazine
dicarboxylate and also any excess azodicarboxylate. If
these reactions were all to lead to more polar products,
such transformations could facilitate the final chromato-
graphic separation.
stirred for 16 h without further cooling and then the
bulk of the solvent was evaporated to leave a thick, or-
ange oil, which was dissolved in a minimum of dichloro-
methane. This was applied to a pad of silica gel (11 cm
diameter, 7 cm deep) packed using dichloromethane.
The pad was washed through with 750 ml of dichloro-
methane which was then evaporated to approximately
300 ml. This solution was washed sequentially with
15% aqueous hydrogen peroxide (300 ml), saturated
aqueous sodium sulfite (300 ml) and water (300 ml).
This final water wash was back-extracted with dichloro-
methane (200 ml) and the combined dichloromethane
solutions dried over dried magnesium sulfate for
5 min. The resulting mixture was filtered through a sec-
ond pad of silica gel (11 cm diameter, 5 cm deep), which
was finally washed with dichloromethane (300 ml).
Evaporation gave pure phthalimide 7 as a colourless
powdery solid (7.735 g, 96%), mp 94–95 ꢁC, [Found:
C, 67.38; H, 5.60; N, 6.01%. C H NO requires C,
1
3
13
3
67.52; H, 5.67; N, 6.06%], dH (CDCl , 400 MHz) 1.74
3
(3H, s, Me), 1.77 (3H, s, Me), 4.73 (2H, d, J 7.7 Hz, 1-
CH ), 5.54 (1H, br t, J ca. 7.7 Hz, 2-H), 7.75 (2H, dd,
2
J 5.5 and 3.1 Hz, 2 · ArH), 7.84 (2H, dd, J 5.5 and
In view of its ability to rapidly oxidise phosphines with-
out the need for a catalyst, we first examined the use of
hydrogen peroxide. We were delighted to observe that
a simple wash, using hydrogen peroxide, of a dichloro-
methane solution of a crude Mitsunobu reaction product
left only the desired product as a significantly mobile
spot on TLC analysis in dichloromethane. This sug-
gested the possibility that a simple filtration through sil-
ica gel could suffice to remove all the by-products. After
some experimentation, we developed the following sim-
ple and chromatography-free procedure for product iso-
lation. A subsequent wash with aqueous sodium sulfite
was included to ensure complete removal of any excess
peroxide. These trials also revealed that 15 wt % aqueous
hydrogen peroxide was quite strong enough to drive the
3.1 Hz, 2 · ArH), d (CDCl , 100 MHz) 18.1 (Me),
C
3
26.0 (Me), 74.1 (1-CH2), 117.1 (2-CH), 123.5
(2 · ArCH), 128.9 (C), 134.4 (2 · ArCH), 143.7
À1
(2 · C), 163.9 (2 · C@O), mmax/cm
(CHCl ) 1783,
3
1728, 1673, 1464, 1392, 1358, m/z (APCI) 232
(M +H, 100%) [Found: M +H, 232.0984. C H NO
+
+
13
14
3
requires M, 232.0973].
Occasionally, the initial solution from the first filtration
was slightly contaminated with DIAD, but this was eas-
ily removed during the second filtration. We found that
yields were reduced considerably if less dichloromethane
was used, especially in the first passage through silica
gel, hence occasional slight contamination by DIAD
was unavoidable. During both filtrations through silica

A. J. Proctor et al. / Tetrahedron Letters 47 (2006) 5151–5154
5153
gel, much larger quantities of dichloromethane could be
used, which indeed must be used with more polar prod-
ucts. However, this usually does not result in any further
contamination, in the case of the first column, given that
the highly coloured (orange) band remains on the col-
umn. It is also worth noting that larger scales can easily
be accommodated, either by employing a larger amount
of silica gel or, preferably, by repeating the first silica gel
filtration prior to washing, that is, performing the first
step of the above process twice. Using this modification,
we were able to isolate over 20 g of the phthalimide 7 in a
pure state in less than 30 min. Column chromatography
on this scale would certainly have taken much longer!
up procedure. Phthalimide itself also worked well with
2-octanol (entry 5). An alternative strategy for nitrogen
introduction was also successful using the N-tosyl t-
5
butylcarbamate reagent TsNHBoc introduced by Wein-
6
,7
reb (entry 6). By contrast, phenol delivered a poorer
8
yield from 2-octanol (entry 7). Optimisation of the
reaction between 2-octanol and acetic acid revealed that,
by using 2 equiv of the latter, an essentially quantitative
yield of the expected acetate could be obtained (entry
9
8). Finally, it has been established by the Ragnarsson
1
0
group that Mitsunobu nucleophiles should have pKa
values of less than 13.5, hence, the recommendation
made by Martin and Dodge to employ 4-nitrobenzoic
1
1
acid in Mitsunobu displacements. Using the new
work-up protocol, the latter with 2-octanol delivered
We have successfully applied this work-up procedure to
a range of exemplifying Mitsunobu reactions and the re-
sults from these are collected in Table 1. Some of the rel-
atively lower yields probably represent the minimum
obtainable, as these are unoptimised. N-Hydroxy-
phthalimide 5 worked well with both prenyl and geranyl
allylic alcohols (entries 1 and 2) and also with both sec-
ondary alcohols tested (entries 3 and 4 [Ref. 3]). It
should be noted that essentially correct microanalysis re-
sults for the products of entries 2 and 3 were obtained
without recrystallisation, following the present work-
1
2
an excellent yield of the desired product (entry 9).
As for what is happening during the peroxide wash,
there is no doubt from TLC evidence that any excess tri-
valent phosphine is being oxidised completely to the cor-
responding and much more polar oxide. The fate of the
hydrazine dicarboxylate and any excess azodicarboxy-
late is less certain. Hydrazine itself is well known as a
1
3
precursor to dimide [HN@NH] following treatment
with various oxidants, hydrogen peroxide among
Table 1. Examples of chromatography-free Mitsunobu reactions
Electrophile
Nucleophile
Product
Yield (%)
96
OH
1
O
N
ONPhth
7
HO
O
5
2
3
4
OH
5
ONPhth
63
85
81
OH
ONPhth
5
5
OH
ONPhth
NPhth
n-C6H13
n-C6H13
n-C6H13
OH
OH
OH
OH
OH
5
6
7
8
9
76
79
56
99
84
n-C6H13
n-C6H13
n-C6H13
n-C6H13
n-C6H13
O
N
H
O
TsNBoc
TsNHBoc
n-C6H13
n-C6H13
n-C6H13
OPh
OAc
PhOH
AcOH (2 equiv)
CO2H
O
O
O2N
n-C6H13
NO2

5154
A. J. Proctor et al. / Tetrahedron Letters 47 (2006) 5151–5154
Acknowledgements
O
H2O2
ArNHNHCO2Me
N=NCO2Me
Ar
We are very grateful to Mr. Robert Jenkins for experi-
mental assistance, to the EPSRC Mass Spectrometry
Service, University College, Swansea, for the provision
of high resolution mass spectrometric data and to Syn-
genta and the EPSRC for financial support.
8
9
Scheme 4.
them,14 but in the presence of a copper(I) salt. Standard
protocols for the oxidation of hydrazine dicarboxylates
to azodicarboxylates are to use a bromonium ion source
Br , NBS) in the presence of pyridine, or related
methods using chlorine. Alternatives include fuming
1
5
(
References and notes
2
1
6
1
7
nitric acid and, more recently hypervalent iodonium
1. Mitsunobu, O. Synthesis 1981, 1–28; Hughes, D. L. Org.
Prep. Proceed. Int. 1996, 28, 127–164; Hughes, D. L. Org.
React. 1992, 42, 335–365.
. Dembinski, R. Eur. J. Org. Chem. 2004, 2763–2772.
. Barrett, A. G. M.; Roberts, R. S.; Schr o¨ der, J. Org. Lett.
000, 2, 2999–3001.
. Mukaiyama, T. Challenges in Synthetic Organic Chemis-
try; Clarendon press: Oxford, 1994.
1
8
reagents. b-Hydroxy hydrazines have been oxidised
using air and copper(II) acetate as catalyst and the
resulting azo compounds further oxidised to a mono-
N-oxide using t-butyl hydroperoxide in the presence of
VO(acac)2. Although the reaction between azodicarb-
oxylates and peroxides appears not to have been re-
ported, hydrogen peroxide has been used to oxidise
the N-aryl carbazates 8 to the N-oxides 9 in the presence
2
3
2
1
9
4
5
. Mitsunobu, O.; Wada, M.; Sano, T. J. Am. Chem. Soc.
1972, 94, 679–680.
6. Henry, J. R.; Marcin, L. R.; McIntosh, M. C.; Scola, P.
2
0
of maleic anhydride (Scheme 4). In our experiments,
exposure of both diisopropyl azodicarboxylate and
hydrazide to the oxidative work-up induced no change,
M.; Harris, G. D.; Weinreb, S. M. Tetrahedron Lett. 1989,
3
0, 5709–5712.
7. Dandapani, S.; Curran, D. P. J. Org. Chem. 2004, 69,
751–8757.
. Downie, I.; Heaney, H.; Kemp, G. Angew. Chem. 1975,
7, 357–358.
. Brewster, J. H. J. Am. Chem. Soc. 1954, 76, 6364–6368.
0. Koppel, I.; Koppel, J.; Degerbeck, F.; Grehn, L.; Rag-
narsson, U. J. Org. Chem. 1991, 56, 7172–7174; Koppel,
I.; Koppel, J.; Leito, L.; Pihl, V.; Wallin, A.; Grehn, L.;
Ragnarsson, U. J. Chem. Soc., Perkin. Trans. 2 1993, 655–
658.
1
according to H NMR analysis. This is perhaps not sur-
8
prising given the rather brief exposure to peroxide in the
absence of a catalyst. Even so, it would seem advisable
to take the appropriate precautions when handling the
silica gel residues from this new work-up procedure, in
view of the possible presence of some level of N-oxides.
8
9
3
1
In summary, this simple procedure should obviate the
need for formal chromatography in Mitsunobu reactions
in general. No doubt, it will not be suitable for every
Mitsunobu reaction. For example, divalent sulfur atoms
might be vulnerable to at least partial oxidation during
the brief exposure to peroxide. However, the exposure
time required is less than 2 min. When a sample of 1,3-
dithiane was exposed to 15% hydrogen peroxide, in a
close mimic of the work-up procedure, there was no
1
1
1. Martin, S. F.; Dodge, J. A. Tetrahedron Lett. 1991, 32,
3
017–3020; Dodge, J. A.; Trujillo, J. I.; Presnell, M. J.
Org. Chem. 1994, 59, 234–236.
2. Barrett, A. G. M.; Braddock, D. C.; James, R. A.; Koike,
N.; Procopiou, P. A. J. Org. Chem. 1998, 63, 6273–6280.
3. Pasto, D. J.; Taylor, R. T. Org. React. 1991, 40, 91–155.
4. Buyle, R.; Van Overstraeten, A. Chem. Ind. (London)
1
1
1
964, 839.
1
oxidation at sulfur according to H NMR analysis.
15. Carpino, L. A.; Crowley, P. J. Org. Synth. 1973, Col. Vol.
5, 160–162.
16. Rabjohn, N. Org. Synth. 1955, Col. Vol. 3, 375–377.
Although we cannot be completely confident that this
will always be the case, the signs are clearly good. Amines
would normally be protected anyway, so should not un-
dergo oxidation, as indicated by entries 1–6 in Table 1.
The neutrality of the work-up also suggests that epoxida-
tions of olefinic bonds conjugated to carbonyl groups, by
a Michael addition mechanism, should not occur.
1
1
1
7. Cohen, S. G.; Zand, R.; Steel, C. J. Am. Chem. Soc. 1961,
8
3, 2895–2900.
8. Moriarty, R. M.; Prakash, I.; Penmasta, R. Synth.
Commun. 1987, 17, 409–413.
9. Nakata, M.; Kawazoe, S.; Tamai, T.; Tatsuta, K.;
Ishiwata, H.; Takahashi, Y.; Okuno, Y.; Deushi, T.
Tetrahedron Lett. 1993, 34, 6095–6098.
Hence, this procedure, while not completely free of the
need for silica gel, should represent a very significant
saving of time and effort.
2
0. Mortarini, V.; Calvino, R.; Gasco, A.; Ferrarotti, B.;
Sanfilippo, A.; Schioppacassi, G. Eur. J. Med. Chem.
1980, 15, 475–478.