The Wittig reaction with pyridylphosphoranes

Article abstract of DOI:10.1002/1099-0690(200007)2000:14<2601::AID-EJOC2601>3.0.CO;2-L

The Wittig reaction was studied by replacing one, two, or three phenyl rings in the triphenylethylidenephosphorane (8) with pyridyl rings, which were attached in the alpha position. It is shown that when using n- butyllithium as the base, the yield of the Wittig reaction is severely affected and it is proposed that the formation of a betaine salt adduct as intermediate suppresses the formation of oxaphosphetanes. In contrast, with sodium bis(trimethylsilyl)amide as the base, the yields are similar to the reaction with 8, although the Z-selectivity reaches values of over 18:1. The reactions have been monitored by ³¹p NMR spectroscopy and the position of the pyridyl ring in the oxaphosphetane 4a elucidated by ROESY NMR spectroscopy at low temperatures.

Full text of DOI:10.1002/1099-0690(200007)2000:14<2601::AID-EJOC2601>3.0.CO;2-L

FULL PAPER
The Wittig Reaction with Pyridylphosphoranes
[a]
[a]
Uwe Schröder and Stefan Berger
Keywords: Wittig reactions / NMR Spectroscopy / Oxaphosphetanes / Phosphorus ylides
The Wittig reaction was studied by replacing one, two, or
three phenyl rings in the triphenylethylidenephosphorane (8)
with pyridyl rings, which were attached in the alpha position.
It is shown that when using n-butyllithium as the base, the
yield of the Wittig reaction is severely affected and it is pro-
trast, with sodium bis(trimethylsilyl)amide as the base, the
yields are similar to the reaction with 8, although the Z-se-
lectivity reaches values of over 18:1. The reactions have been
monitored by 31P NMR spectroscopy and the position of the
pyridyl ring in the oxaphosphetane 4a elucidated by ROESY
posed that the formation of a betaine salt adduct as interme- NMR spectroscopy at low temperatures.
diate suppresses the formation of oxaphosphetanes. In con-
Introduction
The Wittig reaction[ is one of the fundamental trans-
formations in organic chemistry[ and is constantly used in
natural product synthesis to form an olefinic CϭC linkage.
The reaction is carried out either in the presence or absence
of lithium salts, resulting in different stereochemistries.[
In addition, one can distinguish between stabilized, half sta-
bilized and unstabilized (or ‘‘reactive’’) ylides.
1]
2]
3,4]
The oxaphosphetanes 4 are generally accepted as being
3
1
the only intermediates so far observed (by P NMR spec-
[
5]
troscopy) in the reaction, and the occurrence of the earlier
proposed betaines is difficult to ascertain. Computational
chemistry favors a cycloaddition-like reaction to yield ox-
[
6]
aphosphetanes. In our previous work on the mechanism
of the Wittig reaction we have shown NMR-detectable dy-
namic equilibria of oxaphosphetanes which are complexed
by the lithium ions in solution.[ In the special case of a Scheme 1
7]
[8]
Wittig reaction between triphenylmethylenephosphorane
(
the occurrence of a betaine salt adduct 3, since, in the pres-
ence of lithium salts the corresponding oxaphosphetane 4
is destabilized by complexation of the lithium ion to the
two pyridyl rings (Scheme 1).
1) and dipyridylketone 2, however, we have demonstrated[
9]
systematically. This is even more surprising, since phos-
phanes with ortho substituents were used to direct the ste-
reochemistry of the reaction,[ with the idea of complexing
the metal ion in order to obtain the extremely high Z/E
selectivities very recently reported by Schlosser.[ We have
therefore initiated a study to replace one, two, or all three
phenyl groups of the usual triphenylphosphane with alpha-
attached pyridyl rings.
11]
12]
Although this is of theoretical interest, in practice the use
of dipyridyl ketone as one of the reactants is of no general
importance. However, it occurred to us that the use of the
pyridyl rings could be of significant use in this reaction if
the pyridyl moiety were not placed in the ketone part, where
it ends up in the formed olefin, but rather in the ylide part,
where it ends up in the phosphane oxide after work up of
the reaction. Much to our surprise, and despite the huge
volume of variations investigated for the conditions of the
Wittig reaction (e.g. the recent replacement of one phenyl
group by an alkyl group)[ this idea has apparently not
been used or, to the best of our knowledge, not looked at
This investigation had the following objectives: (i) Since,
during the Wittig reaction of 2 a betaine salt adduct can be
isolated, it was of interest to see whether the same effect
can be observed by placing the pyridyl rings in the phos-
phorane part. (ii) Since ortho-substituted phosphoranes dir-
ect significantly the Z/E ratio, it was expected that an alpha-
attached pyridyl ring would also influence the stereochem-
istry.
10]
Results and Discussion
[
a]
Institut für Analytische Chemie, Universität Leipzig,
Linn e´ str.3, D-04103 Leipzig, Germany
Fax: (internat.) ϩ 49-(0)341/97-11833
E-mail: stberger@rz.uni-leipzig.de
Synthesis of Starting Products
For the synthesis of the pyridylphosphanes 5a؊c the
method of Keene et al.[ was chosen, although the trans-
13]
Eur. J. Org. Chem. 2000, 2601Ϫ2604
 WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
1434Ϫ193X/00/0714Ϫ2601 $ 17.50ϩ.50/0
2601

U. Schröder, S. Berger
FULL PAPER
formation from 2-bromopyridine to the three pyridylphos- and a Z/E ratio of about 2:1 in agreement with the literature
phanes is, with a yield of typically 35%, quite unsatisfactory. data.[ Replacing the phenyl rings with ortho-attached pyr-
Since the preparation of 5a؊c is currently the limiting step idyl rings and using n-butyllithium as base has a drastic
of this reaction, the subsequently formed phosphane oxides effect on the yield and, with more than one pyridyl ring, 1-
may have to be reduced and reused in order to avoid the phenylpropene is formed only in traces. The yields using
somewhat cumbersome first step. The quaternisation[ of NaHMDS are comparable with 8; however, the Z/E ratio is
15]
14]
5
6
a؊c with ethyl iodide to form the phosphonium salts increased enormously. From a practical point of view, one
a؊c (Scheme 2) is unproblematic and the corresponding pyridyl ring is sufficient to improve the stereochemical out-
yields are typically 90%.
come significantly compared to the result obtained by use
of 8, with insignificant loss in yield. Thus, this small modi-
fication of the Wittig protocol might be a quite useful al-
ternative in practice.
NMR Spectroscopy of the Intermediates
In order to clarify the dramatic loss of yield in case of
the use of n-butyllithium as a base in the presence of the
pyridyl rings, the reaction was monitored by temperature-
1
31
dependant H and P NMR spectroscopy between Ϫ100
and ϩ10 °C. Generally, in comparison with the spectra ob-
served with 8,[ it must be stated that the spectra obtained
with the pyridyl-substituted systems are much more com-
plicated due to a variety of complexation equilibria with
the metal ions. Even before their reaction with benzal-
dehyde the ylides formed show dynamic NMR spectra be-
tween Ϫ70 and Ϫ100 °C, depending on the number of pyri-
dyl rings attached and the metal ion in use. In Table 2 the
7]
Scheme 2
The deprotonation of compounds 6a؊c was always done
twice in parallel, using n-butyllithium or sodium bis(trime-
thylsilyl)amide (NaHMDS) as a base under otherwise
identical conditions at Ϫ75 °C in dry THF under nitrogen.
As the aldehyde component, freshly distilled benzaldehyde
was used throughout this work. Therefore, the alkylidene
part of the ylide and the aldehyde component remained
constant throughout the study, whereas the bases for the
generation of the ylide, and the number of pyridyl rings in
the phosphorane were varied (Scheme 3).
31
P NMR chemical shifts of the observed species are listed.
With n-butyllithium at Ϫ70 °C, compound 6a forms a
broad ylide signal at δ ϭ Ϫ2, which splits at Ϫ90 °C into
P
two sharpened signals at δP ϭ Ϫ4 and δP ϭ Ϫ0.6 in an
integral ratio of 4:1. Under these conditions compound 6b
yields only one sharp resonance at δP ϭ 6.7, whereas 6c
forms the ylide 7c (δP ϭ 7.2) only partly. In addition to
several other minor signals, one observes another main
component at δ ϭ Ϫ16. The relative intensity of this sig-
P
nals increases when an excess of butyllithium is used. The
ylide signals of 7a؊c disappear instantly after addition of
the aldehyde; however, an oxaphosphetane signal can only
be detected in the case of 7a, and the Wittig reaction does
Scheme 3
Yield and Stereochemistry of the Wittig Reaction
not occur in the case of 7b and 7c. As in our previous
The three pyridylphosphoranes 7a؊c and the standard study[ with 2, after addition of the aldehyde we observed
triphenylethylidenephosphorane 8 were subjected to the a very broad signal at δP ϭ Ϫ10, which we ascribe to a
Wittig reaction with benzaldehyde to form a mixture of E/Z betaine salt adduct. Addition of hexamethylphosphorous
9]
1
-phenylpropene 9, the composition of which was analyzed acid triamide (HMPT) or the crown ether 12-crown-4 to
1
by H NMR spectroscopy after chromatographic workup. these solutions yields the observation of an oxaphosphetane
As can be seen from the results given in Table 1, the ylide signal. Apparently, these reagents can complex the lithium
8
in our hands gives, with both bases, yields of about 80% ions, and therefore the Wittig pathway can be initiated.
Table 1. Yield and stereochemical outcome of the Wittig reaction between pyridylphosphoranes and benzaldehyde using n-butyllithium
and NaHDMS as deprotonating base
Compound
1-Phenylpropene
yield with nBuLi
Z/E ratio
with nBuLi
1-Phenylpropene
yield with NaHMDS
Z/E ratio
with NaHMDS
Ph
Ph
3
PEtI 8
76%
1.7:1
81%
79%
85%
65%
2:1
2
PyPEtI 6a
66%
4:1
11:1
PhPy
2
PEtI 6b
traces
traces
not determined
not determined
18.5:1
12.5:1
3
Py PEtI 6c
2602
Eur. J. Org. Chem. 2000, 2601Ϫ2604

The Wittig Reaction with Pyridylphosphoranes
Table 2. 31P NMR observations at Ϫ70 °C during the course of the Wittig reaction of the pyridylphosphoranes 7a؊c
FULL PAPER
System and base used
Ylides 7
Oxaphosphetanes or
betaine salt adducts
Phosphane oxides
at ϩ10 °C
Ph
2
PyPEtI 6a
PEtI 6b
n-butyllithium
NaHMDS
Ϫ2.0, exchanging
very broad signal at Ϫ12
Ϫ59.5, Ϫ60.7
18.3
17.7
11.0
6.7
PhPy
2
n-butyllithium
NaHMDS
very broad signal at Ϫ10
Ϫ60.7, Ϫ62.3. At Ϫ100 °C two
exchanging signals at Ϫ58/Ϫ63
very broad signal at Ϫ10
Ϫ61.6, Ϫ63.4. At Ϫ100 °C two
exchanging signals at Ϫ60/Ϫ64
8.7
15.2
Py
3
PEtI 6c
n-butyllithium
NaHMDS
7.2
1.0, exchanging
10.0
The situation is quite different with NaHDMS as a base.
Compound 7a shows only a single ylide signal at δ ϭ 11,
P
and, after addition of benzaldehyde, two sharp signals of
Z/E oxaphosphetanes at δP ϭ Ϫ59.5 and Ϫ60.7 are ob-
served instantaneously with an integration ratio of 15:1.
Compound 7b behaves very similarly to 7a with an initially
observed Z/E oxaphosphetane ratio of more than 30:1.
Compound 7c displays a dynamic ylide spectrum with a
very broad signal at Ϫ70 °C which splits at Ϫ90 °C into
two signals at δ ϭ Ϫ5 and 7.3 with an integral ratio of
P
3
:1. Thus the tripyridyl ylide has a similar behavior with
sodium ions as the monopyridyl ylide with lithium ions.
The addition of benzaldehyde generates two oxaphosphet-
ane signals in a Z/E ratio of 12:1. In the di- and tripyridyl
case, on lowering the temperature after the oxaphosphetane
formation yet another dynamic equilibrium is observed.
The oxaphosphetane signals broaden and a new broad ox-
aphosphetane signal at δ ϭ Ϫ57 appears, which increases
P
in intensity towards Ϫ100 °C, where the ratio of the two
oxaphosphetane signals is about 1:1. We ascribe these sig-
nals to the sodium-complexed and -uncomplexed oxaphos- Ϫ70 °C, mixing time (duration of spin-lock pulse) 300 ms
Figure 1. 2D-ROESY NMR spectrum of oxaphosphetane 4a at
phetanes as found earlier for the oxaphosphetanes 4 formed
from dipyridylketone. On warming the solutions above Ϫ10
trum is given in Figure 1 showing ROE contacts between
°
C, the formed oxaphosphetane signals disappear and the
the methyl group at the oxaphosphetane ring and the ortho
protons of one phenyl ring at δH ϭ 7.63. Both protons of
the oxaphosphetane ring show contacts to the ortho protons
at the other phenyl ring at δH ϭ 7.76. Very weak ROE con-
tacts can also be found between the methyl group and the
pyridyl protons H-1 (δH ϭ 8.56) and H-4 (δH ϭ 6.74), and
corresponding signals of the phosphane oxides appear.
We have attempted to shed some light on the alpha effect
and to explain why even one alpha pyridyl unit changes the
Z/E ratio from 2:1 to 11:1. For this purpose, we have pre-
pared an oxaphosphetane 4a starting from 7a (Scheme 4)
with one alpha pyridyl ring but using a perdeuterated
the oxaphosphetane proton H (at the same carbon atom
a
[
16]
benzaldehyde
in order to simplify the aromatic proton
as the methyl group) also shows a very weak contact to
the pyridine proton H-1. These data are consistent with the
pyridyl ring being placed in an equatorial position with
both phenyl rings placed in axial positions opposing each
other. These findings lead to the following interpretation of
the stereochemical outcome. In the ylide the metal ion che-
lates the electron pair of the pyridine nitrogen and that at
Cα in such a manner that a planar five-membered ring
region and therefore to enable safe proton assignments in
this region. The aim was to determine by ROESY[ spectra
at Ϫ70 °C how the pyridyl ring is situated with respect to
the oxaphosphetane protons. An expansion of this spec-
17]
(
MϪC ϪPϪC ϪN) is formed. The aldehyde can only ap-
α q
[18]
proach from the bottom, and, by an inspection of simple
molecular models, it can be easily demonstrated, that the Z
oxaphosphetane is favored by lower steric interaction with
the approaching aldehyde. The phenyl rings move in the bis-
axial position, whereas the pyridyl ring remains equatorial.
As found in the case of the Wittig reaction for dipyridylke-
Scheme 4
Eur. J. Org. Chem. 2000, 2601Ϫ2604
2603

U. Schröder, S. Berger
FULL PAPER
tone 2, the oxaphosphetane can be reopened by chelation NaHMDS, 1.0  in THF; Aldrich) was added at Ϫ75 °C and the
mixture was stirred for 30 min, after which time an equimolar
of the lithium ion between the pyridine nitrogen and the
amount of benzaldehyde was added. The solution was left stirring
to reach room temperature during 12 hours. The THF was evapor-
oxygen atom of the aldehyde to form a betaine lithium salt
complex, which gives rise to very broad signals at δ ϭ Ϫ10.
P
ated and the residue poured into 50 mL of water, followed by ex-
traction three times with diethyl ether (50 mL). The ether phase
was dried with Na SO . After filtration, the solvent was removed
2
4
Conclusion
and the mixture was worked up by column chromatography (silica
gel, petroleum ether as eluent). For the dynamic NMR measure-
We have shown in this work that replacement of the ments, the ylide solution was transferred at Ϫ70 °C to an NMR
phenyl rings in triphenylphosphane with pyridyl rings leads tube with exclusion of moisture and air before the addition of
to a significant increase in Z-selectivity during a Wittig re- benzaldehyde. One equivalent of benzaldehyde was added directly
to the NMR tube at Ϫ70 °C through a septum. In the case of using
action, without affecting the yields, as long as no lithium
6 6
the deuterated benzaldehyde C D
CHO[ dry perdeuterated THF
16]
base is used to generate the ylide. Even one pyridyl ring is
sufficient to create this effect. By ROESY NMR spectro-
scopy at low temperatures, we could determine the structure
of an oxaphosphetane intermediate and deduce a stereo-
was used as solvent for the Wittig reaction as well as for the base
NaHMDS.
chemical model to explain the Z-selectivity. Further struc- Acknowledgments
tural studies to confirm this stereochemical model are cur-
rently being undertaken.
This work was supported by the Fonds der Chemischen Industrie.
[
[
[
1]
2]
3]
G. Wittig, G. Geissler, Liebigs Ann. Chem. 1953, 580, 44Ϫ57.
E. Vedejs, M. J. Peterson, Top. Stereochem. 1994, 21, 1Ϫ157.
H. J. Bestmann, O. Vostrowsky, Top. Curr. Chem. 1983, 109,
Experimental Section
8
5Ϫ157;
[
4]
B. Maryanoff, A. Reitz, Chem. Rev. 1989, 89, 863Ϫ927.
E. Vedejs, K. A. J. Snoble, J. Am. Chem. Soc. 1973, 95,
5778Ϫ5780.
General Remarks. ؊ Dynamic NMR Measurements: Low temper-
ature 31P NMR measurements were performed in THF with a
Bruker DRX-600 NMR spectrometer using a 10 mm multinuclear
probe head under standard measurement conditions and proton
[5]
[
[
6]
7]
A. A. Restrepo-Cossio, C. A. Gonzalez, F. Mari, J. Phys.
Chem. A. 1998, 102, 6993Ϫ7000 and literature cited therein.
C. Geletneky, F.-H. Försterling, W. Bock, S. Berger, Chem. Ber.
decoupling. The shift values are referenced with respect to external
1
993, 126, 2397Ϫ2401.
17]
H
3
PO
4
at room temperature, The ROESY spectra[ were recorded
[8]
C. Subramanyam, Tetrahedron Letters 1995, 36, 9249Ϫ9252.
on a Bruker DRX-400 spectrometer at Ϫ70 °C using a spin-lock
mixing time of 300 ms.
[9] R. A. Neumann, S. Berger, Eur. J. Org. Chem. 1998,
1085Ϫ1087.
[
10]
11]
12]
F. Bangerter, M. Karpf, L. A. Meier, P. Rys, P. Skrabal, J. Am.
Chem. Soc. 1998, 120, 10653Ϫ10659.
X. Zhang, M. Schlosser, Tetrahedron Letters, 1993, 34,
1925Ϫ1928
Synthesis: The 2-pyridyl-/phenylphosphanes 5a؊c[19] were prepared
using the method described for tris(2-pyridyl)phosphane 5c by
[
[
[
13]
Keene et al.
6
The 2-pyridyl-/phenyl-ethylphosphonium iodides
a؊c were obtained according to Schmidbaur et al.[ by refluxing
14]
Q. Wang, M. El Khoury, M. Schlosser, Chem. Eur. J. 2000,
6
, 420Ϫ426.
the phosphanes 5a؊c with a 10-fold molar amount of ethyl iodide
and recrystallizing the product from methanol/diethyl ether.
[13]
R. F. Keene, M. R. Snow, P. J. Stephenson, E. R. Tiekink,
Inorg. Chem. 1988, 27, 2040Ϫ2046.
H. Schmidbaur, Y. Inoguchi, Z. Naturforsch. 1980, 35b,
1329Ϫ1334.
[
[
14]
15]
6
a: C19
H
19INP (M ϭ 419.22); yield 92%; m.p. 136Ϫ137.5 °C; 31P
): δ ϭ 23.9.
NMR (CDCl
3
E. Vedejs, K. A. J. Snoble, J. Am. Chem. Soc. 1973, 95,
5
778Ϫ5780; M. Schlosser, B. Schaub, Chimia 1982, 36,
6
b: C18
H
18IN
2
P (M ϭ 420.21); yield 93%; m.p. 161Ϫ163 °C; 31P
): δ ϭ 20.2.
396Ϫ397
[16]
M. Schlosser, J. M. Choi, S. Takagishi, Tetrahedron 1990, 46,
5633Ϫ5648.
NMR (CDCl
3
P (M ϭ 421.20); yield 88%; m.p. 164Ϫ166 °C; 31P
): δ ϭ 16.1.
[17]
6
c: C17
H
17IN
3
For details of the experiment, see: S. Braun, H.-O. Kalinowski,
S. Berger, ‘‘150 and more basic NMR Experiments’’, VCH
Wiley, Weinheim, 1998, Experiment 10.20.
18]
NMR (CDCl
3
[
M. Schlosser, B. Schaub, J. Am. Chem. Soc. 1982, 104,
5821Ϫ5823.
F. G. Mann, J. Watson, J. Org. Chem. 1948, 13, 502Ϫ531.
General Procedure for the Wittig Reactions: Under nitrogen,
mmol of the 2-pyridyl-/phenyl-ethylphosphonium iodide 6a؊c
6
[19]
was suspended in 50 mL dry THF. To this suspension, an equimo-
lar amount of the base solution (n-butyllithium, 1.6  in n-hexane;
Received March 6, 2000
[O00104]
2604
Eur. J. Org. Chem. 2000, 2601Ϫ2604