Comparative study of the phospha- and arsa-Wittig reaction using 1H, 75As and 17O NMR spectroscopy

Article abstract of DOI:10.1002/ejoc.200600432

The existence of oxaarsetanes during an arsa-Wittig reaction has been proved by ¹H and ¹⁷O NMR spectroscopy. ⁷⁵As NMR spectra were obtained from the corresponding arsonium salts and arsane oxides. The dynamic ¹H NMR spectra of phospha- and arsaylides were compared. Wiley-VCH Verlag GmbH & Co. KGaA, 2006.

Full text of DOI:10.1002/ejoc.200600432

FULL PAPER
DOI: 10.1002/ejoc.200600432
1
75
Comparative Study of the Phospha- and Arsa-Wittig Reaction Using H, As
17
and O NMR Spectroscopy
Christian Raeck[a] and Stefan Berger*[a]
Keywords: Oxaarsetanes / Stereoselectivity / Wittig reactions / NMR spectroscopy
The existence of oxaarsetanes during an arsa-Wittig reaction
has been proved by 1H and 17O NMR spectroscopy. 75As
NMR spectra were obtained from the corresponding ar-
(© Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
Germany, 2006)
1
sonium salts and arsane oxides. The dynamic H NMR spec-
tra of phospha- and arsaylides were compared.
From an NMR point of view, the phospha-Wittig reac-
tion is ideal to study since all the reaction steps can be con-
trolled. Starting with the 31P NMR spectrum of the phos-
phane 2a, its quaternisation to the phosphonium salt 4a can
be observed, followed by deprotonation to ylide 5a. Finally,
the 31P NMR spectrum of oxaphosphetane 1a and phos-
phane oxide 7a as end-products can be easily observed
(Scheme 2).
Introduction
The Wittig reaction for the olefination of aldehyde and
ketone functionalities using phosphorus ylides has a key
position in organic chemistry.[1] Since the recognition of the
potential of this reaction the mechanism has been widely
discussed.[2] Oxaphosphetanes 1a (Scheme 1) are the key in-
termediates of the reaction, as shown in 1973 by Vedejs and
Snoble using low-temperature 31P NMR spectroscopy.[3]
Scheme 1.
Besides phosphorus, there are other elements that can be
used in Wittig-type reactions and arsenic is one of the most
common alternatives. The arsa-Wittig reaction is one of the
most useful reactions in olefination chemistry because of its
high stereoselectivity and mild reaction conditions.[4] The
first publication to report on an arsa-Wittig reaction (1937)
was by Heffe and it preceded indeed the original paper by
Wittig by 16 years.[5] The first systematic investigation was
performed in 1960 by Henry and Wittig.[6] Two possible
reaction pathways of the arsa-Wittig reaction were ob-
served.[6,7] One pathway proceeds via an oxaarsetane, like
the phospha-Wittig reaction, with olefins being formed as
products. The other pathway is supposed to proceed via a
betaine as intermediate, like the reaction of sulfur-ylides
with carbonyl functions leading to epoxides.[8,9] Until now
no direct spectroscopic evidence for the intermediates of
either pathway has been available.
Scheme 2. Intermediates of the Wittig reaction.
The study of the arsa-Wittig reaction by 75As NMR
spectroscopy was anticipated to be more difficult as a result
of the large quadrupolar moment of the 75As nucleus and
its relatively low resonance frequency. The quadrupole mo-
ment creates broad lines in the spectra of arsenic deriva-
tives, especially for those that are not symmetrically substi-
tuted at the arsenic atom, and acoustic ringing prohibits the
acquisition of good spectra. Indeed, 75As NMR spectra
have only been reported for cases in which the substituents
are chemically very similar to each other and placed in a
tetrahedral sphere around the arsenic atom.[10]
Herein we report our attempts to obtain 75As NMR
spectra of several components during an arsa-Wittig reac-
tion and to provide spectroscopic proof of oxaarsetanes 1b
as the key intermediates of the reaction.
[a] Institut für Analytische Chemie, Fakultät für Chemie und
Mineralogie, Universität Leipzig,
Linné-Strasse 3, 04103 Leipzig, Germany
4934
© 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2006, 4934–4937

NMR Study of the Phospha- and Arsa-Wittig Reaction
FULL PAPER
intermediates of the arsa-Wittig reaction. All the antici-
pated intermediates of the arsa-Wittig reaction are shown
in Scheme 2.
Results and Discussion
In order to maintain the symmetry around the arsenic
nucleus we chose triphenylarsane as the starting compound
and quaternised it with benzyl bromide. In addition, ethyl-
idenetriphenylarsorane is known to be very reactive and
leads to epoxides via betaines as intermediates.[8,9] The par-
ticular arsa-Wittig reaction with benzylidenetriphenylarsor-
ane was to be compared with the phospha-Wittig reaction
involving the same substitution pattern. After preparing the
ylides from their arsonium and phosphonium salts, respec-
tively, benzaldehyde was added as the carbonyl compound
in all experiments at room temperature. In agreement with
the literature,[9,11] we observed the formation of stilbene 8
in both reactions, but with the expected differences in yields
and stereochemistry, as given in Table 1. The reason for the
high stereoselectivity of the arsa-Wittig reaction is the
higher reactivity of the arsenic ylide relative to the phos-
phorus ylide.[9] This high stereoselectivity was also notice-
able when the reaction temperature was lowered to –100 °C.
The high reactivity of the arsenic ylide is responsible for
the unchanged yield and stereoselectivity. In the case of the
phosphorus ylide, the stereoselectivity changes with tem-
perature.[11]
The arsenic species were investigated by 75As NMR spec-
troscopy using a modified spin-echo pulse sequence with a
reduced delay after the 180° pulse. Thus, the rising echo
signal was also recorded enhancing the signal-to-noise ratio
and in addition the acoustic ringing of the probe head was
suppressed. Typical 75As NMR spectra of benzyltri-
phenylarsonium bromide (4b) and triphenylarsane oxide
(7b) are shown in Figure 1. The NMR chemical shift
changes from 220 ppm for 4b to 324 ppm in the case of
7b. The linewidth (full width at half maximum) is nearly
unchanged at about 5 kHz indicating similar electronic en-
vironments around the quadrupolar nucleus of the arsenic
atom. Unfortunately, despite many attempts and modifica-
tions of the pulse sequences and spectrometer conditions,
no other intermediates such as the arsenic ylide 5b or the
oxaarsetane 1b gave reliable 75As NMR spectra. Therefore,
other NMR spectroscopic methods were used to obtain
more information about these intermediates.
We first focussed on the ylides. Ylides can be character-
ised by two mesomeric resonance structures, the ylide and
ylene structure (see Scheme 3). For benzylic ylides an ad-
ditional resonance structure is possible. This mesomeric
ylide structure stabilises the negative charge by delocalis-
ation in the aromatic ring. By recording the 1H NMR spec-
tra of the aromatic protons at low temperatures it should be
possible to observe slow rotation around the Cα–Cβ bond of
Table 1. Yields and stereochemical outcomes of Wittig reactions
with benzaldehyde starting from either phosphonium salt 4a or ar-
sonium salt 4b.
T [°C][a]
Ph3PBnBr (4a)
Ph3AsBnBr (4b)
(Z)/(E)[b] Yield [%][c] (Z)/(E)[b]
Yield [%][c]
1
the benzylic group. The H NMR spectra of the benzylic
Room temp.
–100
39:61
60:40
80
81
1:99
1:99
56
61
group of the arsorane 5b and phosphorane 5a are shown in
Figure 2. At low temperatures the rotation around the Cα–
Cβ bond of the benzylic group is indeed slow on the NMR
timescale. Therefore, NMR signals of all five protons can
be recorded separately. At higher temperatures the rotation
[a] Addition of benzaldehyde. [b] (Z)/(E) ratios were determined
on the isolated product by NMR spectroscopy. [c] All yields were
determined on isolated products.
Since the intermediates of the phospha-Wittig reaction becomes faster resulting in an exchange of the 1-H and 5-
were characterised by 31P NMR spectroscopy, then 75As H as well as the 2-H and 4-H atoms. This leads to broader
NMR spectra should provide related characteristics of the- signals and ultimately, after coalescence, only three sharp
Figure 1. 75As NMR spectra (0.05  in CD3OD) of a) benzyltriphenylarsonium bromide (4b) and b) triphenylarsane oxide (7b).
Eur. J. Org. Chem. 2006, 4934–4937
© 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
4935

C. Raeck, S. Berger
FULL PAPER
signals can be observed pertaining to the ortho, meta and
para protons. Line-shape analysis provides thermodynamic
data on the rotation barrier. For the arsorane ∆H# and ∆S#
are 37.3 1.5 kJ/mol and –10.9 7.2 J/molK yielding a
∆G#(298) value of 40.5 3.7 kJ/mol. The phosphorane pro-
vides ∆H# and ∆S# values of 38.6 2.1 kJ/mol and
13.3 11.5 J/molK yielding
34.6 5.5 kJ/mol.
a
∆G#(298) value of
Figure 3. 1H NMR spectra of the oxaarsetane 1b (0.2  in [D8]-
THF).
Scheme 3. Mesomeric resonance structures of benzylic ylides.
guished. Therefore, the benzaldehyde was labelled with 17O.
Whereas we assume that the small values of ∆S# essen- Since the benzylic oxaphosphetane is not stable we switched
tially mean a negligible activation entropy contribution, the for comparison to the oxaphosphetane generated from
∆G# values indicate that for arsenic the contribution of the ethylidenetriphenylphosphorane and benzaldehyde which
ylide versus the ylene structure is a bit more pronounced has been proven to be stable at low temperatures by 31P
compared with phosphorus, in accordance with the litera- NMR spectroscopy.[16,17] Replacing methyl by phenyl in the
ture.[12–14]
γ position with respect to the oxygen atom in the four-mem-
Having characterised the ylide 5b only the oxaarsetane bered ring was not expected to lead to a significant differ-
1b remained to be proven. A first indication of its existence ence in the 17O chemical shift. The 17O NMR spectrum
was found in the 1H NMR spectrum after addition of benz- from the reaction of ethylidenetriphenylphosphorane and
aldehyde to the ylide solution at –40 °C. Two signals arise [17O]benzaldehyde at –40 °C is displayed in part a of Fig-
as doublets coupling with each other (Figure 3). These pro- ure 4. The signal at δ = 22 ppm belongs to the oxaphosphet-
tons belong to the oxaarsetane ring. This is astonishing ane. The small signal next to the oxaphosphetane stems
since the analogous oxaphosphetane is not stable even at from the triphenylphosphane oxide generated during the
very low temperatures. When the temperature was raised addition of benzaldehyde to the ylide solution outside the
stepwise to 30 °C these signals slowly disappeared with a NMR spectrometer. After raising the temperature to 25 °C
concomitant increase in signals from stilbene. This corre- the oxaphosphetane signal was consumed and the tri-
sponds to a slow reaction at room temperature.
phenylphosphane oxide signal dominated (Figure 4, b).
These proton signals are not strict proof of an oxaarset- Repetition of the experiment with benzylidenetriphenylar-
ane because it is also conceivable that a betaine having two sorane (5b) provided the spectrum shown in Figure 4 (c).
protons in the neighbourhood will give a similar spectrum. At –40 °C again a signal at around 22 ppm indicates the
But the chemical shifts do indicate a four-membered ring. existence of the oxaarsetane 1b. This was verified by raising
See for comparison ref.[15]. The coupling constant of 8.6 Hz the temperature as the signal disappeared and the oxide sig-
do not allow the ring and betaine structures to be distin- nal at δ = 50 ppm increased.
1
Figure 2. Dynamic H NMR experiments (0.2  in [D8]THF) for a) benzylidenetriphenylarsorane (5b) and b) benzylidenetriphenylphos-
phorane (5a).
4936
www.eurjoc.org
© 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2006, 4934–4937

NMR Study of the Phospha- and Arsa-Wittig Reaction
FULL PAPER
of moisture and air. The THF used was dried with sodium/benzo-
phenone and freshly distilled prior to use. The benzaldehyde was
also distilled. All routine NMR spectra were recorded with a
Bruker DRX 400 spectrometer in CDCl3, with TMS as an internal
shift reference. 75As NMR spectra were also recorded with a Bruker
DRX 400 spectrometer and referenced to the Ξ scale. The 17O
NMR spectra were recorded with a Bruker DRX 600 spectrometer
with [D8]THF as solvent, shifts being referenced to the Ξ scale. The
IR spectrum was recorded with a Thermo Nicolet Avatar 360 FT-
IR E.S.P. spectrometer. Spectra simulations were performed by
using SpinWorks 2.4.[19]
[1]
a) G. Wittig, G. Geissler, Justus Liebigs Ann. Chem. 1953, 580,
44–57; b) G. Wittig, U. Schöllkopf, Chem. Ber. 1954, 87, 1318–
1330.
[2]
[3]
[4]
[5]
E. Vedejs, M. J. Peterson, Topics in Stereochemistry 1994, pp.
1–157.
E. Vedejs, K. A. J. Snoble, J. Am. Chem. Soc. 1973, 95, 5778–
5780.
H. S. He, C. W. Y. Chung, T. Y. S. But, P. H. Toy, Tetrahedron
2005, 61, 1385–1405.
W. H e ff e, Ph. D. Dissertation, University of Berlin, 1937;
quoted by G. Wittig, Pure Appl. Chem. 1964, 9, 249 (W. Heffe
was a student with F. Kröhnke).
[6]
[7]
[8]
M. C. Henry, G. Wittig, J. Am. Chem. Soc. 1960, 82, 563–564.
A. W. Johnson, J. Org. Chem. 1960, 25, 183.
A. Seyer, L. Alcaraz, C. Mioskowski, Tetrahedron Lett. 1997,
38, 7871–7874.
[9]
R. Broos, M. J. O. Anteunis, Bull. Soc. Chim. Belg. 1988, 97,
271–279.
[10]
G. Balimann, P. S. Pregosin, J. Magn. Reson. 1977, 26, 283–
289.
[11] H. Yamataka, K. Nagareda, K. Ando, T. Hanafusa, J. Org.
Chem. 1992, 57, 2865–2869.
[12] M. Shao, X. Jin, Y. Tang, Q. Huang, Y. Huang, Tetrahedron
Lett. 1982, 23, 5343–5346.
[13] I. Gosney, D. Lloyd, Tetrahedron 1973, 29, 1697–1710.
[14] E. E. Ernstbrunner, D. Lloyd, Chem. Ind. (London) 1971, 46,
1332.
Figure 4. 17O NMR spectra of a) the oxaphosphetane of ethyl-
idenetriphenylphosphorane at –40 °C, b) triphenylphosphane oxide
(7a) at 25 °C and c) oxaarsetane 1b at –40 °C.
[15] E. Vedejs, G. P. Meier, K. A. J. Snoble, J. Am. Chem. Soc. 1981,
103, 2823–2831.
Conclusions
[16] B. E. Maryanoff, A. B. Reitz, M. S. Mutter, R. R. Inners, H. R.
Almond Jr, R. R. Whittle, R. A. Olofson, J. Am. Chem. Soc.
1986, 108, 7664–7678.
In summary we have shown by various NMR methods
that oxaarsetanes exist and the reaction mechanism of the
arsa-Wittig reaction is identical to that of the phospha-Wit-
tig reaction.
[17] M. Appel, S. Blaurock, S. Berger, Eur. J. Org. Chem. 2002,
1143–1148.
[18] L. Horner, A. Mentrup, Justus Liebigs Ann. Chem. 1961, 646,
65–77.
[19] Kirk Marat, SpinWorks, version 2.4, University of Manitoba,
Canada, 1999–2004.
Experimental Section
All compounds were prepared according to literature pro-
cedures.[17,18] All reactions were carried out with careful exclusion
Received: May 17, 2006
Published Online: August 24, 2006
Eur. J. Org. Chem. 2006, 4934–4937
© 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
4937