Unequivocal experimental evidence for a unified lithium salt-free wittig reaction mechanism for all phosphonium ylide types: Reactions with β-heteroatom-substituted aldehydes are consistently selective for cis-oxaphosphetane-derived products

Article abstract of DOI:10.1021/ja300943z

The true course of the lithium salt-free Wittig reaction has long been a contentious issue in organic chemistry. Herein we report an experimental effect that is common to the Wittig reactions of all of the three major phosphonium ylide classes (non-stabilized, semi-stabilized, and stabilized): there is consistently increased selectivity for cis-oxaphosphetane and its derived products (Z-alkene and erythro-β-hydroxyphosphonium salt) in reactions involving aldehydes bearing heteroatom substituents in the β-position. The effect operates with both benzaldehydes and aliphatic aldehydes and is shown not to operate in the absence of the heteroatom substituent on the aldehyde. The discovery of an effect that is common to reactions of all ylide types strongly argues for the operation of a common mechanism in all Li salt-free Wittig reactions. In addition, the results are shown to be most easily explained by the [2+2] cycloaddition mechanism proposed by Vedejs and co-workers as supplemented by Aggarwal, Harvey, and co-workers, thus providing strong confirmatory evidence in support of that mechanism. Notably, a cooperative effect of ortho-substituents in the case of semi-stabilized ylides is confirmed and is accommodated by the cycloaddition mechanism. The effect is also shown to operate in reactions of triphenylphosphine-derived ylides and has previously been observed for reactions under aqueous conditions, thus for the first time providing evidence that kinetic control is in operation in both of these cases.

Full text of DOI:10.1021/ja300943z

Article
pubs.acs.org/JACS
Unequivocal Experimental Evidence for a Unified Lithium Salt-Free
Wittig Reaction Mechanism for All Phosphonium Ylide Types:
Reactions with β-Heteroatom-Substituted Aldehydes Are
Consistently Selective for cis-Oxaphosphetane-Derived Products
Peter A. Byrne and Declan G. Gilheany*
Centre for Synthesis and Chemical Biology, Conway Institute of Biomolecular and Biomedical Research,
School of Chemistry and Chemical Biology, University College Dublin, Belfield, Dublin 4, Ireland
S
* Supporting Information
ABSTRACT: The true course of the lithium salt-free Wittig reaction has
long been a contentious issue in organic chemistry. Herein we report an
experimental effect that is common to the Wittig reactions of all of the
three major phosphonium ylide classes (non-stabilized, semi-stabilized, and
stabilized): there is consistently increased selectivity for cis-oxaphosphetane
and its derived products (Z-alkene and erythro-β-hydroxyphosphonium
salt) in reactions involving aldehydes bearing heteroatom substituents in
the β-position. The effect operates with both benzaldehydes and aliphatic
aldehydes and is shown not to operate in the absence of the heteroatom
substituent on the aldehyde. The discovery of an effect that is common to reactions of all ylide types strongly argues for the
operation of a common mechanism in all Li salt-free Wittig reactions. In addition, the results are shown to be most easily explained
by the [2+2] cycloaddition mechanism proposed by Vedejs and co-workers as supplemented by Aggarwal, Harvey, and co-workers,
thus providing strong confirmatory evidence in support of that mechanism. Notably, a cooperative effect of ortho-substituents in the
case of semi-stabilized ylides is confirmed and is accommodated by the cycloaddition mechanism. The effect is also shown to
operate in reactions of triphenylphosphine-derived ylides and has previously been observed for reactions under aqueous conditions,
thus for the first time providing evidence that kinetic control is in operation in both of these cases.
of betaines as intermediates was alluring because non-Wittig
reactions deliberately designed to produce betaines gave the
same products as the analogous Wittig reactions. We know now
that both processes produce oxaphosphetane (OPA, Scheme 1).
Subsequent work has conclusively shown that solutions
containing only OPA (as confirmed by NMR monitoring of
the solution) give β-hydroxyphosphonium salt (β-HPS) upon
quenching with acid36 and react with LiBr to give a betaine−LiBr
complex.19 Thus, there is no need to invoke the involvement of
betaines in Wittig reactions. Furthermore, the non-involvement
of betaines has been unequivocally demonstrated in the reaction
of a particular dibenzophosphole (DBP)-derived non-stabilized
ylide.37 Irreversible formation of OPAs has been established for
representatives of all three types of ylide, which means also that
reversibility of Wittig reactionswhether they involve betaines
or notcannot be invoked to explain the high E-selectivity.26,35
It is significant (and ironic) that the very limited number of cases
in which genuine reversibility of OPA formation occurs involve
non-stabilized ylides.23−26
INTRODUCTION
■
The Wittig reaction1−3 of carbonyls with phosphorus ylides is
one of the most important and widely used methods for the
synthesis of alkenes4 and, even nowalmost 60 years after
its discoveryis still being intensively studied.5−13 Concur-
rently, its mechanism has been the focus of intense and vigorous
debate2,3,14,15 and could be described as one of the great long-
running investigations of organic chemistry. Johnson3 enum-
erated eight different mechanistic proposals that had been
advanced at various times, reflecting the tortuous path to the
currently emerging consensus on the mechanism. This was the
result of a discontinuous uncovering of the operation of several
different factors that can have a bearing on the mechanism and,
especially, on the stereoselectivity of the reaction. Only in
hindsight can we see that the results of valid and well-constructed
studies concerned with the involvement of betaines (salt-free or
otherwise),16−18 and the related issues of the effect of lithium
cation19−22 and reversible formation of intermediate(s),23−28 led
incorrectly to the twin hypotheses of the involvement of betaines
and the operation of thermodynamic control in Wittig
reactions.29 That these ideas persist in the modern literature30−34
perhaps may be attributed to their simplicity. For example, it is
easy to assume (but ill-founded on experimental evidence)35 that
reactions of stabilized ylides are under thermodynamic control
because they are E-selective. Similarly, the apparent involvement
The present emerging consensus on the mechanism centers
on [2+2] cycloaddition of the ylide and carbonyl to give OPA
directly (Scheme 1). As developed by Vedejs and co-workers,35,38
Received: January 29, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 1. Present Understanding of the Mechanism of the
a
Lithium Salt-Free Wittig Reaction
a
Oxaphosphetane formation is generally irreversible, with well-defined
exceptions (all from non-stabilized ylides, see text). Betaines are not
formed in the process of the salt-free Wittig reaction but, if
independently generated (e.g., from β-HPS), rapidly form OPAs. Each
OPA diastereomer undergoes stereospecific decomposition to alkene
and phosphine oxide, perhaps after undergoing pseudo-rotation.
and modified for reactions of stabilized ylides by Aggarwal,
Harvey, and co-workers,13 it is the mechanism that best accounts
for experimental observations.39,40 Consideration of this
mechanism grew from the initial observation by Vedejs and
Snoble41 that OPA was the only intermediate observable by low-
temperature NMR and that the high Z-selectivity from non-
stabilized alkylidenetriphenylphosphoranes could be explained by
a kinetically controlled [2+2] cycloaddition of ylide and aldehyde.
Subsequently the Vedejs group14,15,19,26,35,38,42−46 did extensive
work to establish the operation of the cycloaddition mechanism in
the reactions of non-stabilized,26,38 semi-stabilized, and stabilized
ylides.35 Maryanoff, Reitz, and co-workers2,20,23−25,47 and
Schlosser and co-workers48 also contributed significantly to its
development; the former group of workers also identified and did
extensive studies on the issue of stereochemical drift (vide infra).
More recently, extensive kinetic studies by Mayr and co-workers
confirmed that the reactivities of stabilized ylides were consistent
with the cycloaddition mechanism.49
Figure 1. Reaction coordinate diagrams for Wittig reactions. (a)
Reaction of a non-stabilized ylide (R3 = Ph, R2 = alkyl), indicating rate-
determining OPA decomposition. (b) Reaction of a stabilized ylide (e.g.,
R2 = CO2Me), indicating rate-determining OPA formation.
CONTEXT OF THE PRESENT EXPERIMENTS
cycloaddition mechanism. In the reactions of all three types of
ylide, the initially formed OPA (with apical carbon) undergoes
facile pseudo-rotation to place the ring oxygen in an apical
position in the OPA trigonal bypyramid.13,37,45,51 Finally, alkene
and phosphine oxide are formed by irreversible, stereospecific
syn-cycloreversion of the OPA.54,55 For reactions of non-
stabilized ylides, the straightforward formation of OPA and
substantial barrier to OPA decomposition mean that the latter is
the rate-determining step (see Figure 1a). For reactions of semi-
stabilized and stabilized ylides, the barrier to [2+2] cycloaddition
is increased and that to OPA cycloreversion is significantly
decreased (especially for OPAs derived from stabilized ylides),
and so OPA formation is rate-determining (Figure 1b).
Consequently, the OPA intermediates are reasonably long-
lived (at low temperature) and thus spectroscopically observable
in reactions of non-stabilized ylides, but not in those of semi-
stabilized or stabilized ylides, with the exception of OPAs formed
from DBP-derived semi-stabilized ylides.
■
The [2+2] Cycloaddition Mechanism. We may summarize
the emerging consensus as follows: It has been categorically
established that salt-free non-stabilized ylides react with
aldehydes by rapid and irreversible formation of cis- and trans-
OPA intermediates,23−26,46,50,51 with no involvement of betaine
intermediates (Scheme 1 and Figure 1a) prior to OPA
formation.37 In a limited number of cases (discussed below),
“Wittig reversal” of the cis-OPA to ylide and aldehyde has been
observed (Scheme 1), resulting in enhanced production of trans-
OPA and hence E-alkene.23−26,47 It is also established that the
OPAs (generated by non-Wittig processes) that are thought to
be necessary intermediates in the analogous Wittig reactions
of semi-stabilized35,42,52 and stabilized35,43 ylides do not
equilibrate under the typical experimental conditions for a
Wittig reaction. The presence of OPAs as intermediates has been
demonstrated in the reaction of a particular DBP-derived semi-
stabilized ylide35 but not, as yet, for any stabilized ylide. Reactions
of stabilized ylides have been observed to be slower in acetone
or dimethylformamide (DMF) than in benzene53 and to have
large negative activation entropy,53 which is consistent with a
Powerful indirect evidence supporting the [2+2] cycloaddition
as a unified Wittig mechanism is related to its use in a consistent
explanation of the variation of the Z/E ratios in the product
B
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
alkenes.3,14 It is postulated that alkylidenetriphenylphosphoranes
(non-stabilized ylides) react preferentially through an early,
puckered transition state (TS) in which the carbonyl substituent
occupies a pseudo-equatorial position. This minimizes steric
interactions between it and the P-phenyl groups, which are still
in a pseudo-tetrahedral arrangement about phosphorus (see
Figure 2a). In such a TS, the steric interactions of the ylide
TS in reactions of non-stabilized ylides (Figure 2a).13 This results
in a TS that has an electrostatically favorable anti-parallel orienta-
tion of the carbonyl C−O and ylide C−C(O) bond dipoles.
Minimization of both 1−2 and in particular 1−3 steric
interactions then dictates that the large aldehyde substituent
(R1) should be pseudo-equatorial, and so this TS is selective for
trans-OPA. The possible (late) cis-selective TSs (planar and
puckered) in these reactions of stabilized ylides (see Figure 2c,d)
were found to be significantly higher in energy than the trans-
selective TS.58,59
Confusion in the Secondary/Tertiary Literature about
the Mechanism. The evidence and arguments summarized
above are the basis for the consensus among mechanistic
organophosphorus chemists on the underlying simplicity and
unity of the mechanism of the salt-free Wittig reaction and how
it manifests itself in practice. It can be seen however that the
explanation of the source of stereoselectivity in the [2+2]
cycloaddition mechanism relies on fairly complex technical
arguments, with a number of caveats and exceptions to be
explained. This has not helped to dispel (previously accepted)
older descriptions of the mechanism involving reversible steps
and/or betaine intermediates.30−33,60−62 The persistence of both
of these issues may also be related to the fact that the mechanism
of the Li salt-containing Wittig reactions is still uncertain.33
Modern textbooks tend not to distinguish between Wittig
reactions conducted in the presence of Li+, for which the
mechanism is essentially unknown, and those that occur under
Li salt-free conditions, for which the mechanism is now well
established.
In the present study, for the first time, we demonstrate an
effect that is common to Li salt-free Wittig reactions of all three
classes of phosphonium ylide. This is powerful confirmatory
evidence that there is a unitary mechanism in operation in all
kinetically controlled Wittig reactions. In addition we shall
demonstrate that our results are entirely consistent with the
[2+2] cycloaddition mechanism. We hope that the unmasking of
an unexpected effect that is common to all ylide types and which
is easily explicable by the cycloaddition mechanism will enable
clarity on the mechanism of the Li salt-free Wittig reaction for
non-experts in organophosphorus mechanism.
Figure 2. Proposed [2+2] cycloaddition transition states for various
Wittig reactions. (a) Early puckered TS with pseudo-tetrahedral
arrangement of phosphorus substituents and pseudo-trigonal planar
geometry of the ylidic and carbonyl carbon substituents (for non-
stabilized ylides). (b) trans-selective TS for the reaction of a stabilized
ylide, with favorable anti-parallel orientation of carbonyl and C−CO2Me
dipoles. (c) Planar cis-selective TS for the reaction of a triphenylphos-
phine-derived stabilized ylide. (d) Disfavored cis-selective TS for the
reaction of a stabilized ylide, with approximately parallel orientation of
carbonyl and C−CO2Me dipoles.
α-substituent (with the P-phenyl groups in particular) are
minimized if it is in a pseudo-axial site, and hence this TS leads
to cis-OPA and Z-alkene. This TS simultaneously minimizes
1−2 and 1−3 interactions (see Figure 2a for numbering of ring
positions) for the particular arrangement of substituents about
phosphorus involved, as well as allowing the forming P−O bond
to avoid the P-phenyl group that is necessarily projecting in the
direction of the carbonyl approach to the ylide, and is thus highly
favored for alkylidenetriphenylphosphoranes.56
PRELIMINARY REMARKS ON THE EVALUATION OF
KINETIC DIASTEROSELECTIVITY IN WITTIG
REACTIONS
■
There are many variables that may affect the mechanism of the
Wittig reaction and thus the observed diastereomeric ratio of the
alkene product. An understanding of these variables has come
about through the substantial body of work carried out by Vedejs
and co-workers, Maryanoff, Reitz, and co-workers, Aggarwal,
Harvey, and co-workers, Schlosser and co-workers, and others.
The importance of the foundations laid by these workers is such
that it is only recently, in the aftermath of their work, that it has
become possible to conduct meaningful experiments on the
mechanism of the Wittig reaction with sufficient confidence that
the numerous variables at play are under control. We now briefly
discuss these variables, the effect that they exert in the reaction,
and how to prevent these effects giving rise to misleading results
in the context of the reaction mechanism.
In reactions of triphenylphosphine (TPP)-derived semi-
stabilized ylides, the somewhat more advanced nature of the
TS and the shape of the sp2-hybridized substituent on the
ylide result in decreased 1−3 and 2−3 steric interactions. As a
consequence, a planar trans-selective TS can become competitive
with a cis-selective puckered TS, and poor selectivity results. In
reactions of semi-stabilized ylides for which one or more of the
P-phenyl groups are replaced by alkyl group(s), the shape of the
phosphonium moiety is such that 1−3 and 2−3 interactions are
further reduced, resulting in a greater preference for the planar
trans-selective TS (which ensures minimal 1−2 interactions),
and hence the trans-OPA and E-alkene.
Stabilized ylides react through a relatively late TS.57 Recent
calculations by Aggarwal, Harvey, and co-workers indicate that
the trans-selective TS (shown in Figure 2b) in reactions of
stabilized ylides is puckered, but importantly that this puckering
is of the opposite sense to that proposed for the cis-OPA-selective
Operation of Kinetic Control. Conditions have been
established in which reactions of all three different ylide types
occur under kinetic controlmeaning that the OPA inter-
mediates are formed irreversibly and undergo stereospecific
C
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
decomposition to alkene and phosphine oxide. These conditions
are now described.
Stereochemical Drift. This is a more mechanistically
significant source of variation in stereochemistry. Under certain
circumstances, the proportion of trans-OPA present in the
initially formed mixture of OPA isomers (which reflects the
extent to which the trans-OPA is preferred kinetically) may be
augmented at the expense of the cis-OPA,14,25,26,47 leading to a
different ratio of cis- and trans-OPAs and therefore a Z/E alkene
ratio which is not reflective of the intrinsic kinetic selectivity of
the [2+2] cycloaddition step. This phenomenon has been
termed “stereochemical drift”.25 Reactions conducted in the
presence of additives such as salts that are soluble in the reaction
solventin particular Li cation,14,20 but also iodide anion,52
lithium halide with small amounts of alcohol,72 and benzoic
acid64have been shown to give Z/E ratios that are altered with
respect to reactions conducted in the absence of such additives.
The effect of Li ion is solvent dependent, with a profound effect
being observed for reactions in non-polar solvents, and
essentially no effect in solvents that effectively complex Li+.3,20
Hydroxylic solvents and high temperature have also been
implicated as possible initiators of OPA equilibration in reactions
of aromatic aldehydes.14
More challenging is that, in certain reactions of non-stabilized
ylides, stereochemical drift can occur in the absence of dissolved
salts (Li or otherwise). This has been observed to occur for OPAs
derived from trialkylphosphonium alkylides with tertiary or
aromatic aldehydes,23−26,47 and for OPAs derived from ethyl-
idenetriphenylphosphorane with benzaldehyde (although in the
latter of these, stereochemical drift only occurs at or above the
temperature at which OPA can decompose to alkene, while
below this temperature the OPA diastereomeric ratio remains
identical to the low-temperature kinetic ratio).19 By the use of
crossover experiments, the enhanced production of trans-OPA in
these examples was shown to arise from reversal of the cis-OPA
exclusively to ylide and aldehyde and subsequent recombination
of these reactants. In each of the above examples, the OPAs were
generated by deprotonation of β-HPS. In the course of the
present study the occurrence of this phenomenon has been
confirmed by comparison of the kinetic cis/trans ratio of the OPA
and the Z/E ratio of the alkene produced in the reactions of
non-stabilized ylides ethylidenetriphenylphosphorane,73
P-(ethylidene)ethyldiphenylphosphorane,73 and P-(ethylidene)-
P-phenyldibenzophospholane74 with each of benzaldehyde and
2-bromobenzaldehyde. In each case, a greater proportion of the
E-isomer was observed to be present in the alkene product than
would have arisen from the trans-OPA initially generated.
In reactions of non-stabilized ylides it is possible to observe the
OPA intermediates by low-temperature NMR, and also to
quench the OPA at low temperature with acid to give β-HPS
(whose erythro/threo ratio corresponds directly to the cis/trans
ratio of the OPA precursor). Both of these techniques have been
used to determine kinetic OPA cis/trans ratios, and in cases
where both techniques have been used, the ratios have been in
excellent agreement.2,36 Comparison of the diastereomeric ratio
of the intermediate determined by either method with the Z/E
ratio of the alkene product in all but a small number of
exceptional circumstances (see the Stereochemical Drift section
below) shows the ratios to be identical.23−26,46,50,51 Thus it can
be inferred that these reactions occur under kinetic control.
For Li salt-free reactions of semi-stabilized and stabilized
ylides, OPA intermediates are generally not sufficiently long-
lived to permit spectroscopic detection.63 Kinetic control has
been established in these challenging cases by generating the
OPA intermediates through processes independent of a Wittig
reaction and proving that these OPAs do not interconvert,
but instead undergo stereospecific decomposition to alkene and
phosphine oxide.35 TPP-derived OPAs are not amenable to
these routes (all of which require quaternization at phosphorus),
and so methyldiphenylphosphine-derived OPAs, which are
accessible by non-Wittig processes, have been employed in
these experiments.
Based on the above experiments, if one carries out a Wittig
reaction under conditions mimicking those for which the
operation of kinetic control has been verified, a direct
correspondence between the kinetic OPA cis/trans ratio and
the observed alkene Z/E ratio can be assumed (as long as steps
have been taken to ensure that no isomerization of the alkene
occurs after the reaction is complete; see the Dependability of Z/
E Ratios section below). The exceptional circumstances under
which the kinetic OPA cis/trans ratio changes from its original
value are well defined (see the Stereochemical Drift section
below), and steps can be taken to evaluate the relevant ratio
before it changes.
Dependability of Z/E Ratios. In the present work, the
kinetic selectivity of the OPA-forming step in Wittig reactions of
semi-stabilized and stabilized ylides is inferred from the observed
Z/E ratio of the alkene product. It is thus very important to
be sure that the alkene Z/E ratio is truly reflective of the kinetic
OPA cis/trans ratio, and to be aware of possible means by which
there may arise a non-correspondence between the two ratios.
Changes may occur to the Z/E ratio after completion of the
reaction and/or during workup or chromatographic purification
of the alkene product. It is not sufficiently recognized that Z-1,2-
disubstituted alkenes are quite easily converted, under a variety of
conditions, to a Z/E mixture and sometimes completely to the
E-isomer. Therefore the Z/E ratio resulting from the reaction is
fragile and can be affected by the presence of acids64 or strong
bases,65 the chromatographic stationary phase used, the solvent,
heat, and sunlight. Both ourselves66 and Vedejs and Peterson14
have identified multiple previous literature reports where
there was undoubtedly isomerization in favor of the E-alkene
subsequent to the actual Wittig reaction.67 It may even occur
simply if the reaction mixture is allowed to stand for a period.68
We have taken extensive precautions69,70 and performed a
substantial number of control experiments71 to ensure that our
observed Z/E ratio is truly reflective of that rendered by the
Wittig reaction in question.
Our Interest in This Area. This arose some years ago,66
when we reported on a curious phenomenon in the Wittig
reactions of triphenylphosphonium benzylides with ortho-
substituted benzaldehydes. Strong ortho-effects from substitu-
ents on phosphorus were already well known through the work
of McEwen and co-workers,75,76 and these had been extended to
the Wittig reaction, although with conflicting results.77,78 It was
also known that Z-selectivity in stilbene synthesis could be
induced by ortho-substituents with heteroatom lone pairs on the
aldehyde.79,80 However, remarkably, we found that this latter
Z-selectivity could be substantially augmented by an additional
ortho-substituent on the benzylide, despite the fact that such a
substituent would ordinarily lead to E-selectivity. This counter-
intuitive cooperative effect was strong enough to be preparatively
useful (Z/E up to 95:5), and the resulting Z-2,2′-disubstituted
stilbenes have been used to good effect in synthesis by
others.81,82 At the time, we rationalized the results within the
cycloaddition mechanism by proposing that the increased
D
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Table 1. Z/E Ratio for Stilbenes 6−27 Produced in the Reactions of Benzylidenemethyldiphenylphosphoranes 3a−g, Derived
b
from Phosphonium Salts 1a−g (with ortho-Substituent X) with Benzaldehydes 5a−i (with ortho-Substituent Y)
Z/E
ratio
Z/E
ratio
Z/E
ratio
Z/E
ratio
entry ylide X ald Y
entry ylide X ald Y
entry ylide X
ald Y
entry
ylide X ald Y
c
1
2
3
4
5
6
7
8
H
H
H
H
H
H
H
H
Cl
92:8
9
Cl
Br
I
H
H
H
H
H
H
H
34:66
33:67
28:72
41:59
7:93
16
17
18
19
20
21
22
23
Cl
Br
I
Cl
97:3
24
Me
Br
Cl
94:6
Br
94:6
10
11
12
13
14
15
Br
98:2
25
26
27
Me
Me
Me
Me
OMe
Br
70:30
73:27
52:48
66:34
95:5
I
96:4
I
99:1
Cl
d
c
F
84:16
35:65
86:14
77:23
42:58
F
F
Br
94:6
F
c
c
c
Me
OMe
OEt
SMe
Me
OMe
H
Me
OMe
Br
Br
Me
OMe
OEt
SMe
31:69
44:56
92:8
28
OMe
Br
e
12:88
15:85
29
30
OMe
72:28
61:39
a
Z/E ratio determined by 1H NMR analysis of the crude product obtained after minimal aqueous workup. See Supporting Information for full details
b
of the reaction, workup, Z/E analyses, and characterization of the starting materials and product alkenes. Counterion Z = Br− unless otherwise
c
d
noted. Counterion Z = Cl−. The corresponding result for X = Br, Y = F: Z/E = 84:16. The case X = Y = F is excluded because the alkene is
e
especially prone to isomerization. This reaction was also carried out using the phosphonium chloride salt, and the Z/E ratio was found to be 46:54.
Z-selectivity arose from the induction by the ortho-heteroatom of
a lower energy TS in the kinetically controlled cycloaddition step
leading to a lower energy cis-OPA. However, the experimental
conditions employed (aqueous, room temperature, presence of
sodium salts in solution, use of TPP-derived ylides), while very
convenient, were not such that kinetic control in the Wittig
reaction could be assumed and rendered the analysis provisional
at best. We have now re-investigated comprehensively the
original ortho-heteroatom phenomenon under conditions
ensuring kinetic control; much more significantly, we have
extended it to non-stabilized and stabilized ylides, and also to
aliphatic aldehydes.
Standard Wittig Reaction Conditions Used in This
work. For Wittig reactions of semi-stabilized and stabilized
ylides, we adopted a standard set of conditions designed to (i)
ensure kinetic control, (ii) avoid possible initiators of stereo-
chemical drift, and (iii) ensure that the Z/E ratio of the alkene
product rendered by the reaction remains unchanged after
completion of the reaction. For each reaction, the (Li salt-free)
ylide was pre-generated from the parent phosphonium bromide
or chloride salt using non-nucleophilic bases NaHMDS or
KHMDS in dry aprotic solvent (tetrahydrofuran (THF)) under
an atmosphere of dry nitrogen. The ylide solution was cooled
to −78 °C, and aldehyde (verified free of carboxylic acid by 1H
NMR) was then added dropwise. The operation of kinetic
control in reactions of methyldiphenylphosphine-derived semi-
stabilized and stabilized ylides has been unequivocally verified for
just these reaction conditions,35 and so the bulk of our reactions
involve such ylides. We have also carried out some reactions of
TPP-derived ylides for comparison. The inorganic salts produced
in these reactions (NaCl, NaBr, KCl, KBr) are insoluble in THF
and thus are out of solution and exert no effect on the reaction of
ylide with aldehyde.83 For reactions of semi-stabilized ylides, the
reaction mixture was allowed to warm to room temperature,
while reactions of stabilized ylides were quenched at −78 °C by
addition of NH4Cl solution to ensure that the reaction had
occurred at low temperature. The Z/E ratio was established on
the basis of integrations of characteristic signals in the 1H NMR
of the crude alkene obtained after minimal aqueous workup and
before chromatographic purification (unless otherwise indi-
cated). Thus, every effort has been made to ensure that the Z/E
ratios we have observed in these reactions correspond directly to
the kinetic OPA cis/trans ratio produced in the Wittig reaction.
For reactions of non-stabilized ylides with benzaldehydes, alkene
Z/E ratios are not used to infer the kinetic selectivity of OPA
formation, since in principle it may be possible for stereo-
chemical drift to occur at or above the temperature at which OPA
decomposition to alkene and phosphine oxide commences.
We instead rely on OPA cis/trans ratios (obtained by low-
1
temperature H and 31P NMR of the Wittig reaction mixture)
and β-HPS erythro/threo ratios (from low-temperature acid
quenching of Wittig reactions) to establish the kinetic cis/trans
ratio of OPA. In all cases, as expected,36 these two methods were
in agreement.
Finally we note that, in general, the OPA cis/trans ratio for a
Wittig reaction must be at least as high as the observed alkene
Z/E ratio since OPA decomposition is stereospecific and
irreversible, reflecting the fact that cis-OPAs are normally higher
in energy than trans-OPAs. Therefore, as long as the cis-OPA is
indeed higher in energy than the trans isomer, it can be assumed
that the reactions that are highly selective for the Z-alkene are
under dominant or total kinetic control.26 The consequence is
that it is not ordinarily possible to obtain Z-selectivity by accident
or by intervention of equilibration. Therefore, it is apposite that
our conclusions (vide infra) are dependent on high Z/E ratios,
which have a high likelihood of being the “true” values.
REACTIONS OF BENZALDEHYDES
■
Results for Semi-stabilized Ylides. Salts 1a−g (and
selected triphenyl analogues 2) were converted to the
corresponding ylides 3a−g (and analogues 4) and reacted with
benzaldehydes 5a−i. The Z/E ratios of the stilbene products
(6−27) obtained in these reactions are shown in Tables 1 and 2.
At the outset, we note that in reactions with benzaldehyde,
the unsubstituted benzylide of methyldiphenylphosphine shows
high E-selectivity (Table 1, entry 15), in good agreement with
E
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Table 2. Z/E Ratio for Stilbenes Produced in the Reactions of Selected Benzylidenetriphenylphosphoranes 4a,d−g, Derived from
b
Phosphonium Salts 2a,d−g (with ortho-Substituent X) with Selected Benzaldehydes 5a,c−f (with ortho-Substituent Y)
entry
ylide X
ald Y
Z/E ratio
entry
ylide X
ald Y
Z/E ratio
entry
ylide X
ald Y
Z/E ratio
c
c
c
1
H
H
H
H
Cl
90:10
87:13
90:10
88:12
5
Cl
H
H
H
H
51:49
42:58
42:58
59:41
9
Cl
Cl
94:6
94:6
90:10
94:6
95:5
2
3
4
Br
6
Br
10
11
12
13
Br
Br
d
OMe
I
7
OMe
H
OMe
I
OMe
I
e
8
Me
Cl
a
Z/E ratio determined by 1H NMR analysis of the crude product obtained after minimal aqueous workup. See Supporting Information for full details
b
of the reaction, workup, Z/E analyses, and characterization of the starting materials and product alkenes. Counterion Z = Br− unless otherwise
c
noted. Phosphonium salt used for this reaction was not dried in the standard manner and was not stored under argon; as a consequence, the yield of
d
e
alkene from the Wittig reaction was lower and the amount of phosphine oxide produced by ylide hydrolysis higher. Counterion Z = Cl−. From ref
14; KHMDS used to generate the ylide.
literature precedent,84 whereas that from TPP is known to show
slight Z-selectivity (Table 2, entry 8).84 Tables 1 and 2 show the
following trends:
(4) The reaction of 2-methylbenzaldehyde with the
unsubstituted methyldiphenylphosphine-derived benzylide
shows moderate E-selectivity (Table 1, entry 5), as does its
reaction with 2-methyl-substituted ylide (entry 20). Its reaction
with ortho-heteroatom-substituted benzylides shows poor to
moderate Z-selectivity (Table 1, entries 25−28). That high
Z-selectivity is not observed in these reactions shows that the
unusual effects observed in the reactions of ortho-heteroatom-
substituted benzaldehydes are dependent on the ortho-
substituent being lone-pair-bearing; i.e., the effect is not of steric
origin.
In summary, lone-pair-bearing ortho-substituents on benzal-
dehyde result in significantly enhanced Z-selectivity with
benzylides. There is also a counter-intuitive secondary effect
whereby this Z-selectivity is increased by ortho-substituents on
the benzylide.
The reactions in Table 1 have been carried out under condi-
tions for which irreversible OPA formation has been verified in
reactions of semi-stabilized ylides,35 and many of them show very
high Z-selectivity. Consequently, we are confident that kinetic
control is in operation in these reactions.26 Table 2 shows that
the same magnitudes and trends in Z/E ratios are obtained in the
analogous reactions of benzylidenetriphenylphosphoranes,
including the unusual ortho-heteroatom effects. This continuity
of unexpected effects is strong evidence that the triphenyl cases
too are under kinetic control. Furthermore, the trends in the
results shown in Table 2, again including the signature ortho-
effects, are entirely consistent with those obtained in our
previous report,66 except that the observed Z-selectivities are in
general higher here. This, again, is evidence that kinetic control
persists even under the aqueous bi-phasic conditions that we
previously used. The effect persists even when non-dry
phosphonium salt is used (Table 2, entries 1, 5, and 9), so the
THF solvent is wet to some degree due to water of crystallization
in the phosphonium salt. In these reactions of non-dry
phosphonium salts, a significant amount of phosphine oxide is
produced by ylide hydrolysis and, consequently, the yield of
alkene is lower. Thus the anhydrous conditions we adopted for
the purposes of ensuring kinetic control also improve the
preparative utility of the reactions.
(1) Reactions of unsubstituted benzylides with benzaldehydes
bearing an ortho-heteroatom (lone-pair-bearing) substituent
show markedly increased Z-selectivity (see Table 1, entries 1−
4, 6, and 7, and Table 2, entries 1−4). This effect is observed both
for benzaldehydes that are more electrophilic at the carbonyl
(“reactive”) than benzaldehyde itself and for ortho-alkoxy-
substituted benzaldehdyes, which are less electrophilic than
benzaldehyde. The effect is less pronounced with an ortho-
methylthio substituent (Table 1, entry 8). Considering their very
different starting points (Table 1, entry 15, vs Table 2, entry 8), the
magnitude of the shift toward Z-selectivity on ortho-substitution is
very much greater in the methyldiphenyl series than in the triphenyl
series (Table 1, entries 1−3 and 6, vs Table 2, entries 1−4).85 There
appears to be a trend in the observed Z-selectivity depending on the
identity of the aldehyde ortho-heteroatom substituent, increasing in
the order F < O < Cl < Br < I.
(2) Reactions of ortho-substituted benzylides with benzalde-
hyde are moderately E-selective. This is less pronounced in the
TPP series (Table 2, entries 5−7) than in the methyldiphenyl-
phosphine series (Table 1, entries 9−14), and in the latter series
ylides with electron-donating groups (“reactive”) give the highest
E-selectivity (Table 1, entries 13 and 14).
(3) Reactions of ortho-substituted benzylides with ortho-
heteroatom-substituted benzaldehydes show Z-selectivity equiv-
alent to or even greater than that of the corresponding reactions
of the same aldehydes mentioned in point 1. This is almost
always the case for either electron-withdrawing or -donating
substituents (Table 1, entries 16−19, 22, 24, 29, and footnote d;
Table 2, entries 9−13 and ref 69). The only notable exceptions
are the reactions of ortho-heteroatom-substituted benzaldehydes
with 2-methoxybenzylidenemethyldiphenylphosphorane 3f
(Table 1, entries 21 and 30). Once again the increase in
Z-selectivity in these reactions in comparison to the reaction of
the same ylide with benzaldehyde is generally much greater85 in
the methyldiphenylphosphine series (Table 1, entries 16−18 and
24, vs Table 2, entries 9, 10, 12, and 13), with the exception of
reactions of 3f. As with the reactions in point 1, the magnitude of
the Z-selectivity appears to depend on the identity of the
heteroatom, increasing in the order F < O < Cl ≤ Br ≤ I,
culminating in the particularly striking di-iodo cases (Table 1,
entry 18, and Table 2, entry 12). Although the reaction of
2-(methylthio)benzaldehyde (5i) with the ortho-bromobenzy-
lide 3d does show increased Z-selectivity (Table 1, entry 23 vs
entries 8 and 10), the magnitude of the increase is not as great as
with other aldehydes.
The high Z-selectivity in these reactions of semi-stabilized
ylides points to an energy lowering of the TS leading to the cis-
OPA (see later discussion).
Stabilized Ylides. We reasoned that, if a low-energy TS that
is strongly selective for the cis-OPA is available to the Wittig
reactions of benzylidenephosphoranes with ortho-heteroatom-
substituted benzaldehydes, there might be a similar low-energy
TS available to the reactions of stabilized ylides such as
F
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(alkoxycarbonylmethylidene)methyldiphenylphosphoranes 29.
Reactions of such ylides have previously been shown to be
under kinetic control by stereospecific formation of Z-alkene
from the erythro-β-HPSs derived from (ethoxycarbo-
nylmethylidene)methyldiphenylphosphorane and each of
several aldehydes (including benzaldehyde).35 Therefore we
examined several Wittig reactions of heteroatom-substituted
aldehydes with such ylides, generated in situ from the
corresponding phosphonium salt 28, under the same conditions
as were employed for the reactions of semi-stabilized ylides. The
Z/E ratios determined for these reactions are shown in Table 3.
These results are entirely consistent with those observed in the
reactions of the semi-stabilized ylides, with the aldehydes bearing
the larger bromo and iodo substituents showing the highest
Z-selectivity, again strongly implying that the reactions occur
under kinetic control.
Non-stabilized Ylides. The Li salt-free reactions of alkyli-
denetriphenylphosphoranes with aldehydes normally show
exceptionally high Z-selectivity.14 Therefore, it is experimentally
almost impossible to demonstrate unequivocally an enhance-
ment in Z-selectivity due to the presence of an ortho-heteroatom
on the benzaldehyde in such Wittig reactions. In our search for
suitable candidates we settled on the ylides shown in Chart 1,
Table 3. Z/E Ratios for Reactions of Selected
ab
,
(Alkoxycarbonylmethylidene)methyldiphenylphosphoranes
29f,h,j (Generated in Situ from the Corresponding
Phosphonium Salts 28) with Selected Benzaldehydes 5 To
Chart 1. Non-stabilized Ylides 47−50 Used in This Study
a
Give Enoates 30−42
a
Generated in situ by treatment of the parent phosphonium bromide
b
salts 43−46 with NaHMDS or KHMDS. The P-ethyl ylide 48 (rather
than the P-methyl analogue) had to be used to avoid transylidation.
entry
ylide OR
aldehyde Y
enoate Z/E ratio
b
1
OMe
H
36:64
36:64
40:60
66:34
77:23
77:23
70:30
79:21
77:23
83:17
83:17
85:15
84:16
which do not show intrinsically high selectivity for cis-OPA in
their reaction with benzaldehyde. This meant that there could be
a demonstrable change in the selectivity for reactions of these
ylides with an ortho-heteroatom-substituted benzaldehyde.
Although reactions of ethylidenetriphenylphosphorane (47, see
Chart 1) are typically highly Z-selective, there is still some scope
for increased Z-selectivity with this particular ylide that would
not be available with longer chain alkylidenetriphenylphosphor-
anes.
2
3
4
5
6
7
8
9
OEt
H
O(t-Bu)
OEt
H
OMe
OMe
OEt
SMe
Cl
O(t-Bu)
O(t-Bu)
O(t-Bu)
OMe
OEt
Cl
b
10
OMe
Br
However, a complication in the context of this project was that,
in several instances in the reaction of ethylidenetriphenylphos-
phorane with benzaldehyde, it has been shown that kinetic
control does not operate, even under Li salt-free conditions, and
so enhanced production of E-alkene is observed (see Stereo-
chemical Drift section above).19,25 Based on this, we felt that we
could not infer the kinetic OPA cis/trans ratio from alkene Z/E
ratio with certainty in Wittig reactions of non-stabilized ylides
with benzaldehydes. In order to circumvent these limitations,
we decided to deal with the stereochemical drift issue by
determining the kinetic selectivity of the reactions through
evaluation of low-temperature OPA cis/trans ratios or erythro/
threo ratios of β-HPS obtained by low-temperature acid
quenching of the reactions.
The erythro/threo ratio of the β-HPSs produced in the
reactions of some non-stabilized ylides with benzaldehydes are
shown in Table 4. The proportion of erythro-β-HPS was found to
be larger in each of the reactions of ethylidenetriphenylphos-
phorane (47) and of (ethylidene)ethyldiphenylphosphorane
(48) with 2-bromobenzaldehyde (see Table 4, entries 2 and 4)
than in the reactions of the same ylides with benzaldehyde (see
Table 4, entries 1 and 3). However, in the latter case the increase
is modest at best. The Z/E ratios for the alkenes produced in the
corresponding unquenched Wittig reactions are also shown in
Table 4 (and in Table 5 below). In certain cases, these provide
evidence for the occurrence of stereochemical drift in the process
of OPA decomposition (see below).
11
12
13
OEt
Br
O(t-Bu)
O(t-Bu)
Br
I
a
All reactions were carried out at −78 °C and subsequently quenched
with aqueous ammonium chloride at this temperature. Z/E ratio
determined by H NMR analysis of the crude product obtained after
1
aqueous workup of the reaction mixturesee Supporting Information
for full details. Phosphonium salt counterion Z = Br unless otherwise
b
indicated. Phosphonium salt counterion Z = Cl−.
The Z/E ratio for the reaction of ester-stabilized ylide
(ethoxycarbonylmethylidene)triphenylphosphorane86 with
benzaldehyde in THF at 20 °C has previously been found to
be 23:77.14,35 The reactions of benzaldehyde with the
(alkoxycarbonylmethylidene)methyldiphenylphosphoranes in-
vestigated here show slightly decreased E-selectivity compared
to this literature example (see Table 3, entries 1−3). The
reactions of the same ylides with ortho-heteroatom-substituted
benzaldehydes show significantly enhanced Z-selectivity, both in
reactions in which the aldehyde is more reactive than
benzaldehyde itself (Table 3, entries 4−6) and in those where
it is less reactive than benzaldehyde (Table 3, entries 8−12), as
well as in reactions of benzaldehydes of similar reactivity to
benzaldehyde itself (Table 3, entries 7 and 13). Such selectivity is
almost unprecedented outside of alcohol solvents.2,14 As in the
reactions of semi-stabilized ylides, 2-(methylthio)benzaldehyde
5i shows somewhat reduced Z-selectivity compared to other
ortho-heteroatom-substituted benzaldehydes.
The reactions of phenyldibenzophosphole (PhDBP)-derived
ylides (49, 50) with benzaldehydes were also investigated and
G
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 4. Results of Wittig Reactions of Non-stabilized Ylides
47 and 48 with Selected Benzaldehydes 5a and 5d To Give
OPA Initially and Subsequently β-HPS (51−54) after Low-
Table 5. Results of Wittig Reactions of Non-stabilized Ylides
48−50 with Benzaldehydes 5a and 5d To Give OPAs 55−60
a
Initially and Subsequently Alkenes 61−64
a
Temperature Acid Quenching of OPA
b
c
entry ylide R3 ald Y β-HPS erythro/threo ratio
alkene Z/E ratio
ylide
b
entry
Ra2RbP
R2
ald Y OPA cis/trans ratio alkene Z/E ratio
1
2
3
4
Ph
Ph
Et
H
92:8
97:3
85:15
79:21
32:68
56:44
d
c
d
Br
H
1
PhDBP
PhDBP
PhDBP
PhDBP
EtPh2P
EtPh2P
Me
Me
i-Pr
i-Pr
Me
Me
H
71:29
94:6
53:47
82:18
89:11
91:9
d
c
54:46
65:35
2
Br
H
e
Et
Br
3
89:11
94:6
e
a
4
Br
H
Ylides 47 and 48 (see Chart 1) were generated from the parent
f
5
6
−
32:68
56:44
phosphonium bromide salts 43 and 44, respectively, using NaHMDS
at 20 °C. All Wittig reactions were carried out at −78 °C and
subsequently quenched by cannulation of the reaction mixture into
HCl solution in THF/methanol. See Supporting Information for full
g
Br
64:36
a
Ylides 48−50 (see Chart 1) were generated from the parent
phosphonium bromide salts 44−46, respectively, using KHMDS unless
otherwsie indicated. All Wittig reactions were carried out at −78 °C. OPA
cis/trans ratios were determined by 31P NMR after cannula filtra-
tion of the reaction mixture into an NMR tube under an inert
atmosphere and addition of toluene-d8. The 31P NMR spectra all in-
dicated the presence of small amounts of phosphine oxide by-
product, which were shown to be derived from the ylide by control
reactions in which no aldehyde was added, and by the fact that no
b
details. The erythro/threo ratio was determined by H and 31P NMR
1
c
of the crude product after minimal aqueous workup. Alkene Z/E ratio
from the corresponding unquenched Wittig reaction allowed to warm
to room temperature after 15 min at −78 °C, as determined by
integration of characteristic signals in the 1H NMR of the crude
d
product. Stereochemical assignments verified from crystal structure of
the erythro isomer.87
b
alkene product could be observed by NMR prior to heating. Alkene
Z/E ratios were determined by integration of characteristic signals in
the 1H NMR of the crude product. DBP-derived OPAs were heated to
80 °C for 2 h to effect alkene formation, while the EtPh2P-derived
OPA began to decompose to alkene and phosphine oxide at ca. −10 °C.
provided more clear-cut evidence of the operation of the ortho-
heteroatom effect for non-stabilized ylides. These reactions have
the advantage of furnishing OPAs that are stable at room
temperature (although still air sensitive), and thus it is possible to
directly investigate kinetic OPA cis/trans ratios by NMR
at temperatures significantly higher than −78 °C. The OPA
cis/trans ratios for the reactions of benzaldehyde and
2-bromobenzaldehyde respectively with each of the ylides
P-(ethylidene)phenyldibenzophospholane (49), P-(iso-
butylidene)phenyldibenzophospholane (50), and (ethylidene)-
ethyldiphenylphosphorane (48) are shown in Table 5, as are
the alkene Z/E ratios from the same reactions after OPA
decomposition at higher temperature. The OPA cis/trans ratios
were assigned by integration of characteristic signals from −60
to −70 ppm in the 31P NMR spectrum of the reaction mixture.
All the reactions investigated gave Z-alkene as the major product,
and thus the largest OPA signal was assigned to the cis-OPA in
the 31P NMR spectra of the reaction mixtures, with the exception
of entry 5, for which OPA diastereomers could not be resolved
spectroscopically.
Comparison of Table 5, entries 2 and 4, with entries 1 and 3
again shows a distinct increase in the preference for the formation
of cis-OPA in the reactions of 2-bromobenzaldehyde relative
to the reactions of benzaldehyde. The diastereomeric ratio of
the OPA produced in the reaction of benzaldehyde with
(ethylidene)ethyldiphenylphosphorane (48) could not be
determined. However, the cis/trans ratio of the OPA from the
reaction of the same ylide with 2-bromobenzaldehyde closely
matches the erythro/threo ratio of the β-HPS produced in the cor-
responding low-temperature acid quenching experiment using
the same reactants, thus affirming conclusions drawn on the basis
of the results in Table 4.
c
The ylide was generated at −20 °C and then stirred for 0.5 h
at −45 °C before cooling to −78 °C for the reaction. The OPA
generated in this reaction was monitored by 31P NMR at −20 °C.
d
e
PhDBP = P-phenyldibenzophosphole moiety of ylide. NaHMDS
was used to generate the ylide from the parent phosphonium salt.
f
OPA signals have the same 31P chemical shifts at −40 and −20 °C, so
g
no diastereomeric ratio could be determined. OPA cis/trans ratio
determined at −40 °C.
A non-correspondence was observed between the kinetic OPA
cis/trans ratios (evaluated as the β-HPS erythro/threo ratio)
shown in Table 4 and the alkene Z/E ratio obtained in the
relevant Wittig reactions. Similar non-correspondence was
also observed in the decomposition of the OPA adducts of
P-(ethylidene)phenyldibenzophospholane (Table 5, entries 1
and 2) and of (ethylidene)ethyldiphenylphosphorane (Table 5,
entry 5). This demonstrates the operation of stereochemical drift
in these reactions, even under Li salt-free conditions, in keeping
with earlier observations of this phenomenon in reactions of
ethylidenetriphenylphosphorane with aromatic aldehydes.19
However, no variance of the OPA cis/trans ratio was observed
at temperatures well below that required to effect alkene
formation, so the diastereomeric ratios of the β-HPSs and OPAs
can reliably be equated to the kinetic ratios of the OPA-forming
steps of the reactions. Negligible stereochemical drift was
observed in the formation of alkene by heating the OPA adducts
of P-(isobutylidene)phenyldibenzophospholane (ylide 50, see
Table 5, entries 3 and 4). It appears that the irreversibility or
otherwise of OPA formation in reactions of non-stabilized
ylides with benzaldehydes is heavily dependent on the structure
of the non-stabilized ylide, and especially on the nature of the
alkylidene moiety. There may now be sufficient evidence to
suggest that the Li salt-free reactions of ethylides generally
Based on the low-temperature acid quenching experiments
detailed in Table 4 and the low-temperature NMR experiments
in Table 5, we conclude that the ortho-heteroatom-induced effect
is indeed in operation in the reactions of non-stabilized ylides.
H
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
undergo stereochemical drift. The reactions of longer chain
alkylidenetriphenylphosphoranes with benzaldehyde may gen-
erally be under kinetic control, although there are an insufficient
number of examples to allow definitive conclusions to be made.
Reactions of Non-aromatic Aldehydes. The results
described above are common to all three ylide classes. They
are also self-consistent and, as will be shown below, can be
explained by a common TS argument. Therefore they indicate a
common mechanism for the Wittig reaction of all ylide types.
However, it could be counter-argued that the effect is solely
confined to ortho-heteroatom-substituted benzaldehydes and
might not extend to other aldehydes. Therefore we were anxious
to find other examples.
Table 6. Z/E Ratios for Wittig Reactions of Aliphatic
a
Aldehydes 65 and 66 with Semi-stabilized Ylides 3a,d and 78,
b
Stabilized Ylides 29f,h,j, and Non-stabilized Ylide 50
As it turns out, we did not have far to look. Enhanced
Z-selectivity is a known effect in Wittig reactions of aldehydes
bearing a heteroatom substituent (typically an oxygen) on the
β-carbon relative to the carbonyl group (i.e., similarly disposed
relative to the carbonyl as the ortho-heteroatoms in benzalde-
hydes).88 The literature examples of this phenomenon for the
most part involve stabilized ylides2,14,89,90 (although in these, the
mechanistic origin of the high Z-selectivity has not been
identified), but there are also some examples involving semi-
stabilized ylides.2,88 In many of these examples, the carbonyl
and the β-substituent of the aldehyde are substituents on a ring,
and high Z-selectivity is observed only if the carbonyl and
β-heteroatom are oriented cis with respect to each other, and it is
highest in alcohol solvents. High E-selectivity is observed for
similar aldehydes in which the carbonyl and β-heteroatom have
trans relative orientation89−93 or if there is no β-heteroatom.89,90
We chose the aliphatic aldehyde, 1,2-O-isopropylidene-3-O-
methyl-α-D-xylopentodialdofuranose-(1,4)94 (65, see Chart 2),
ylide
entry
ald
Ra2RbP
R2
alkene Z/E ratio
1
2
65
65
65
65
65
65
66
65
66
MePh2P
MePh2P
MePh2P
MePh2P
MePh2P
PhDBP
PhDBP
PhDBP
PhDBP
Ph
95:5
2-BrC6H4
CO2Me
CO2(t-Bu)
COOEt
Ph
95:5
c
3
79:21
79:21
79:21
85:15
0:100
c
4
c
5
d
6
d
7
Ph
e
8
i-Pr
92:8
f
9
i-Pr
43:57
a
b
See Chart 2 for definition of R1 in structure of 65 and 66. Ylides 3,
29, and 50 were generated from the parent phosphonium bromide
salts 1, 28, and 46, respectively, using NaHMDS at 20 °C. All Wittig
reactions were carried out at −78 °C. Alkene Z/E ratios were
a
Chart 2. Structures of Aldehydes 65 and 66 and Ylide 78
1
determined by integration of characteristic signals in the H NMR of
c
the crude product. Reactions of stabilized ylides were quenched
at −78 °C by addition of aqueous NH4Cl in order to ensure the reaction
d
had occurred at this temperature. Ylide 78 was generated from the
correspoding phospholium bromide (77) at −20 °C by addition of
e
THF to a mixture of the salt and KHMDS. The kinetic OPA cis/trans
ratio was observed by 31P NMR at −20 °C and found to be 94:6. The
OPA solution was heated to 80 °C for 2 h to effect alkene formation.
a
f
The kinetic OPA cis/trans ratio was observed by 31P NMR at 30 °C
Ylide 78 was generated in situ from the parent phosphonium bromide
salt.
and found to be 45:55. The OPA solution was heated to 80 °C for 2 h
to effect alkene formation.
as our non-benzaldehyde test, and we reacted it under our
standard reaction conditions with some of the same ylides used in
the reactions described above. The carbonyl group of this
aldehyde is a substituent on a five-membered ring, and there is a
β-methoxy substituent oriented cis with respect to the carbonyl.88
High Z-selectivity is observed in the reactions of this aldehyde
with both semi-stabilized ylides 3a,d and 78 (Table 6, entries 1, 2,
and 6) and stabilized ylides 29f,h,j (Table 6, entries 3−5) to give
alkenes 67−72.95 The reaction of benzylide 78 with a control
aldehyde (66) lacking a β-heteroatom substituent showed
complete E-selectivity (Table 6, entry 7).
The reaction of aldehyde 65 with non-stabilized ylide
P-(isobutylidene)-P-phenyldibenzophospholane (50) was also
investigated,96 and 31P NMR observation of the kinetic cis/trans
ratio of the resulting OPA (73) at −20 °C showed it to be 94:6, as
corroborated by the Z/E ratio of 92:8 observed for the alkene
product 75 after heating of the OPA solution to effect OPA
decomposition (Table 6, entry 8). To show that this selectivity
is indeed out of the ordinary, the reaction of the same non-stabilized
ylide with cyclopentanecarbaldehyde (66) was carried out.
Observation of the cis/trans ratio of OPA 74 (by 31P NMR) for
this reaction showed it to be 45:55. Subsequent H NMR
observation of the Z/E ratio of the alkene (76) produced
by heating the OPA solution showed it to be 43:57 (Table 6,
entry 9), again indicating a close correspondence between the
OPA and alkene diastereomeric ratios.
Thus, a very striking shift in selectivity in favor of the Z-alkene
occurs in reactions of non-stabilized ylides with 1,2-O-
isopropylidene-3-O-methyl-α-D-xylopentodialdofuranose-(1,4)
(65) compared to a similar aliphatic aldehyde lacking a suitably
oriented β-heteroatom. This non-aromatic aldehyde was shown
not to undergo epimerization at the α-carbon under our reaction
conditions. Using 1D and 2D NOESY NMR, it was shown that
the hydrogen that had been at the α-carbon of the aldehyde
remains cis to the α-hydrogen in each of the Z-alkene products
(see Supporting Information for details). This is consistent with
earlier reports of non-epimerization of this and other related
1
I
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
aliphatic aldehydes in Wittig reactions carried out under similar
conditions.89−93,97
have been proposed to explain O−Se peri-interactions in 1,8-
substituted naphthalenes101,102 and anthroquinones103 and large
long-range PP coupling constants in 1,8-diphosphanaphtha-
lenes,104−106 a phosphorus-containing carborane,107 tetraphos-
phine ferrocenyl complexes,108 biaryl bisphosphines,109 and
calix[4]arene bisphosphites.110
Comparing phospholium ylides 50 and 78, it is noticeable that
there is a much greater shift from E to Z selectivity for the semi-
stabilized analogue 78 in its respective reactions with 66 and 65.
Thus it shows complete E-selectivity in its reaction with 66,98
but very high Z-selectivity in its reaction with 65 (albeit not quite
as high as the MePh2P-derived analogue), demonstrating a
very dramatic shift in the energy of the cis-selective TS as a
consequence of the presence of the heteroatom.
This high selectivity for cis-OPA and/or Z-alkene in reactions
of stabilized, semi-stabilized, and non-stabilized ylides with non-
aromatic aldehyde 65 is strikingly similar to the effect seen with
benzaldehydes. Therefore we ascribe it to be a consequence of
the same remote heteroatom effect.
Rationalization of the Effects within the Cycloaddition
Mechanism. The reactions detailed in this paper have been
carried out in conditions under which OPAs derived from reagents
similar to those used here have been shown not to equilibrate;26,35
therefore, they can be assumed to be under kinetic control.
We rationalize the signature aldehyde β-heteroatom effect within
the TS model for the [2+2] cycloaddition mechanism38 with the
single additional proposal of the existence of a stabilizing
phosphorus−heteroatom bonding interacction66,79 in the cis-selective
cycloaddition TS (see Figure 3). This results in a three-center,
The immediate consequence of the postulated phosphorus−
heteroatom bond is that the TS is forced to be puckered in order
to facilitate the existence of this bond (i.e., to get the heteroatom
within bonding range of phosphorus). The already sterically
encumbered environment around phosphorus becomes even
more crowded, and the six substituents around phosphorus
assume pseudo-octahedral geometry. There now exists the
potential for significant increases in the steric interactions
between the phosphorus substituents and the ylide α-carbon
substituent (2−3 interactions), in particular if the α-carbon
substituent is in a pseudo-axial site on the forming ring. These are
minimized when the α-carbon substituent R2 is in a pseudo-
equatorial site and thus points to the same side of the forming
ring as the aldehyde substituent (see Figure 3a,b). This then is
the source of the stabilization of the TS to cis-OPA. It is
noteworthy that this cis-selective TS is very similar to the trans-
selective TS proposed by Aggarwal, Harvey, and co-workers13 for
reactions of stabilized ylides (see Figure 2b above). The similarity
is that the ring-puckering angle is negativemeaning that
the dihedral angle between the C−O and C−R2 bonds is greater
than in a planar TS. This is the opposite of a cis-selective TS
in a reaction of a non-stabilized ylide, so the C−O bond is
approximately anti-parallel to the ylide C−R2 bond (Figure 3a).
The important difference in this case is that it is favorable for the
aldehyde substituent to be close to phosphorus (i.e., in a pseudo-
axial position in the forming ring) due to the phosphorus−
heteroatom bond. In the absence of this bond, 1−3 interactions
would dictate that the aldehyde substituent would preferentially
take up a pseudo-equatorial position. The geometry of the
proposed TS is consistent with the fact that high Z-selectivity is
observed even in reactions of stabilized ylides. Such a TS should
benefit from both the phosphorus−heteroatom bond and the
advantageous anti-parallel orientation of the carbonyl C−O and
ylide C−R2 bond dipoles that is normally only present in a trans-
selective TS.13
Apart from its precedence in organophosphorus chemistry,
several factors argue for the existence of phosphorus−
heteroatom bonding in the cycloaddition TS: (i) The effect is
observed for both aromatic and aliphatic aldehydes bearing a
suitably oriented remote heteroatom, whereas aldehydes with
no such heteroatom substituent do not show such high
Z-selectivity.89,90 (ii) The effect is seen for benzaldehydes with
both electron-donating and electron-withdrawing ortho-heter-
oatom substituents but hardly at all with a methyl substituent.
(iii) The effect increases as the heteroatom polarizability
increases and electronegativity decreases in the order F, OMe,
Cl, Br, I (e.g., see Table 1, entries 1−4 and 6, and Table 3, entries
4−13), which would correlate with the ability of the heteroatom
to bond to phosphorusthis has precedent in the reported
structures of 1,8-selenyl-substituted naphthalenes,101,102 where
the magnitude of the proposed 3c4e selenium−heteroatom
bonding interaction also appears to increase in the order F < Cl <
Br. (iv) Related to point (iii) is the fact that cis-selectivity also
increases in line with the bond length of the carbon−heteroatom
bond of the aldehydethe longer this bond and the larger and
more polarizable the heteroatom, the closer the heteroatom can
approach to phosphorus in the TS, thus facilitating stronger
Figure 3. (a,b) Different perspectives of the puckered cis-selective TS
with phosphorus−heteroatom bonding. The ylidic substituent R2 is
oriented as shown to minimize 2−3 steric interactions by avoiding the
phosphorus R3 substituents. (c) A trans-selective TS with phosphorus−
heteroatom bonding suffers from large 2−3 steric interactions.
four-electron (3c4e) bond, with the acceptor orbital being one of the
P−C bond σ* orbitals, analogous to the (orthogonal) inter-
action forming the P−O bond. The proposal of such a through-
space interaction has long-time precedent in organophosphorus
chemistry, having been used most notably by McEwen and co-
workers in simple rationalizations of the rates of quaternization of
ortho-substituted arylphosphines75 and ω-N,N-dimethylaminoal-
kylphosphines99 and hydrolysis of ortho-substituted arylphospho-
nium salts.76,100 More recently, similar through-space interactions
J
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
phosphorus−heteroatom bonding. (v) When the heteroatoms
are switched (Table 1, entry 19 and footnote d), the effect is
stroger for the bromoaldehyde than the fluoro. (vi) There must
be some significant effect lowering the energy of the cis-OPA-
selective cycloaddition TS particularly to make Z-alkene
formation predominant in the reactions of ester-stabilized ylides.
With reference to points (iii) and (iv) above, we emphasize
that the strength of the phosphorus−heteroatom bond is
expected to depend on the ability of the heteroatom to be in
close proximity to phosphorus, and therefore need not correlate
with the (much greater) strengths of the corresponding
phosphorus−heteroatom single bonds of stable compounds.111
It is also expected to be subject to the intervention of steric
effects, especially in reactions of semi-stabilized ylides where
2-ethoxybenzaldehyde and 2-(methylthio)benzaldehyde show
less dramatic shifts toward Z-selectivity than other ortho-
heteroatom-substituted benzaldehydes. In these cases there
also exists the possibility of π-symmetry repulsive steric
interactions between the heteroatom lone pairs or the S−Me
σ-bond and the P−C σ-bonds. Another factor that could militate
against phosphorus−heteroatom bonding in certain circum-
stances is that its existence in the cis-selective TS may require the
conjugation between the heteroatom and the benzaldehyde ring
to be broken, which would also have an associated energy cost.
The magnitude of the heteroatom-induced energy decrease
of the cis-selective TS also appears to be affected by the degree
of steric congestion about phosphorus. For reactions of semi-
stabilized ylides the shift is greater for benzylides derived from
PhDBP and methyldiphenylphosphine than for those derived
from TPP. For non-stabilized ylides, the shift toward Z-selectivity
decreases in the order PDBP > TPP > ethyldiphenylphosphine.
Stabilized ylides derived from methyldiphenylphosphine show
markedly increased Z-selectivity compared to those derived from
TPP.14,68
The strongest and most compelling evidence for a phosphorus−
heteroatom interaction, however, is its ability to easily
accommodate and explain the counter-intuitive co-operative effect
found in the semi-stabilized ylide cases. As alluded to above,
Z-selectivity in reactions of benzylides with ortho-heteroatom-
substituted benzaldehydes is consistently higher when the
benzylide also bears an ortho-substituent compared to the
corresponding reaction of the unsubstituted benzylide (see Tables 1
and 2 and the associated text). This can be explained as due to
the cis-selective TS being better able to accommodate the
increased steric demands (and especially 2−3 interactions) of
the bulkier ylidic substituent than is the trans-selective TS,
resulting in greater discrimination between the two. The high
Z-selectivity obtained in the reactions of the ortho-heteroatom-
substituted benzaldehydes with the ortho-methyl-substituted
benzylides (Table 1, entry 24, and Table 2, entry 13) shows that
this is a steric effect. This subtle augmentation of the remote
heteroatom effect in reactions of benzylides with benzaldehydes
appears to be specific to these reactions only. It must be very
dependent on the shape of the cycloaddition TSs in this particular
class of reactions because we found no noticeable change in cis- or
Z-selectivity in the reactions of bulky non-stabilized or ester-
stabilized ylides with heteroatom-substituted aldehydes. This
observation is consistent with others indicating that the energy
of the cis-selective TS in reactions of semi-stabilized ylides is
more sensitive to steric effects than analogous TSs in reactions of
non-stabilized and stabilized ylides leading to a greater influence
on selectivity from the nature of the phosphonium moiety,
the ylide C-substituent, and the aldehyde heteroatom substituent
(if present).
Some further insight into the nature of the proposed
cycloaddition TS can be gleaned from the 31P NMR shifts of
the OPA reaction intermediate in these reactions. We looked for
any evidence of phosphorus−heteroatom bonding in the OPAs
derived from the non-stabilized ylides P-(ethylidene)-
phenyldibenzophospholane (49) and P-(isobutylidenephenyldi-
benzophospholane (50) with 2-bromobenzaldehyde and also
with 1,2-O-isopropylidene-3-O-methyl-α-D-xylopentodialdofur-
anose-(1,4) for the latter ylide. These OPA intermediates have a
reasonable lifetime at low temperature. We found that the 31P
NMR chemical shifts of all of the OPAs generated in the course
of this project were in the range δ −60 to −70 ppm. An OPA with
a phosphorus−heteroatom bond (and thus hexacoordinate
phosphorus) would be expected to have a significantly more
negative chemical shift in the 31P NMR, analogous to
oxaphosphetanides, which have previously been reported to
have chemical shifts below −100 ppm.112,113 We conclude that
the phosphorus in the OPA intermediates in these reactions is
penta-coordinate. We reason that, although the phosphorus−
heteroatom bond could in principle be present in either a
puckered or a planar TS, it seems likely that if a planar TS was
capable of engaging in this stabilizing interaction, then the
resulting OPA should also benefit from such stabilization. The
lack of any noticeable phosphorus−heteroatom interaction in
the planar OPA is then consistent with the proposed puckered
cis-selective TS.
CONCLUSION
■
The high Z-selectivity observed here in the Wittig reactions of
heteroatom-substituted aldehydes with all ylide types is easily
explicable if all ylides react with aldehyde to form OPA by
irreversible direct cycloaddition through a puckered transition
state. The results obtained here then, in tandem with the
computational results of Aggarwal, Harvey, and co-workers,
corroborate the cycloaddition mechanism proposed by Vedejs
and strongly support the contention that there is a common
mechanism of the Wittig reaction for all ylide types and that,
ordinarily, it operates under kinetic control. In particular we
believe that the results for the reactions of ester-stabilized ylides
with ortho-heteroatom-substituted benzaldehydes provide strik-
ing evidence that the reactions of stabilized ylides occur by
irreversible cycloaddition of the reactants to give OPA. We hope
therefore that our results will ensure that the following points
become widely known:
(A) There is one mechanism of the salt-free Wittig reaction.
(B) It is an irreversible [2+2] cycloaddition to OPA followed
by a stereospecific syn-cycloreversion to give alkene and
phosphine oxide.
(C) The stereoselectivity of all Wittig reactions is explicable by
the single mechanism (especially stabilized ylide cases).
Corollaries of (A)−(C) are:
(D) OPAs are the first-formed and only intermediates in Li
salt-free Wittig reactions.
(E) With very limited exceptions, no salt-free Wittig reaction is
reversible.
Our view is that the now-established Li salt-free [2+2]
cycloaddition mechanism should be presented in textbooks, and
that a clear distinction should be made between this mechanism
and that of Wittig reactions conducted in the presence of Li salts,
for which the mechanism is as yet unknown.
K
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
49, 3492. (e) Appel, M.; Blaurock, S.; Berger, S. Eur. J. Org. Chem. 2002,
1143.
ASSOCIATED CONTENT
* Supporting Information
■
S
(7) New reaction conditions: (a) El-Batta, A.; Jiang, C. C.; Zhao, W.;
Anness, R.; Cooksy, A. L.; Bergdahl, M. J. Org. Chem. 2007, 72, 5244.
(b) McNulty, J.; Das, P. Eur. J. Org. Chem. 2009, 4031. (c) Bera, R.;
Dhananjaya, G.; Singh, S. N.; Kumar, R.; Mukkanti, K.; Pal, M.
Tetrahedron 2009, 65, 1300−1305 (this work describes the use of
microwave conditions and is a good leading reference to other
variations; see refs 7b, 7f, and 7h for further examples of microwave-
assisted Wittig reactions). Reactions “on” or “in” water: (d) McNulty, J.;
Das, P. Tetrahedron Lett. 2009, 50, 5737. (e) Das, P.; McNulty, J.
Tetrahedron Lett. 2010, 51, 3197. (f) McNulty, J.; Das, P.; McLeod, D.
Chem. Eur. J. 2010, 16, 6756. (g) Calzavara, J.; McNulty, J. Tetrahedron
Lett. 2011, 52, 5672. (h) Das, P.; McLeod, D.; McNulty, J. Tetrahedron
Lett. 2011, 52, 199. (i) Tiwari, S.; Kumar, A. Chem. Commun. 2008,
4445−4447. (j) Wu, J. L.; Li, D.; Zhang, D. Synth. Commun. 2005,
2543−2551. Phase-transfer conditions: (k) Pascariu, A.; Ilia, G.; Bora,
A.; Iliescu, S.; Popa, A.; Dehelean, G.; Pacureanu, L. Central Eur. J. Chem.
2003, 1, 491−534. (i) Moussaoui, Y.; Said, K.; Ben Salem, R. ARKIVOC
2006, Part (xii), 1−22. In a ball mill: (m) Balema, V. P.; Wiench, J. W.;
Pruski, M.; Pecharsky, V. K. J. Am. Chem. Soc. 2002, 124, 6244−6245.
Ultrasound-assisted: (n) Wu, L. Q.; Yang, C. G.; Yang, L. M.; Yang, L. J.
J. Chem. Res. (S) 2009, 183−185. Green Wittig reactions: (o) Nguyen,
K. C.; Weizman, H. J. Chem. Educ. 2007, 84, 119−121. (p) Martin, E.;
Kellen-Yuen, C. J. Chem. Educ. 2007, 84, 2004−2006.
(8) One pot reactions: (a) Orsini, F.; Sello, G.; Fumagalli, T. Synlett
2006, 1717. (b) Wu, J.; Yue, C. Synth. Commun. 2006, 36, 2939.
(c) Choudary, B. M.; Mahendar, K.; Kantam, M. L.; Ranganath, K. V. S.;
Athar, T. Adv. Synth. Catal. 2006, 348, 1977. (d) Li, C. Y.; Zhu, B. H.; Ye,
L. W.; Jing, Q.; Sun, X. L.; Tang, Y.; Shen, Q. Tetrahedron 2007, 63,
8046. (e) Lee, E. Y.; Kim, Y.; Lee, J. S.; Park, J. Eur. J. Org. Chem. 2009,
2943. See also refs 5fh and7ab.
(9) Synthetic studies: (a) Feist, H.; Langer, P. Synthesis 2008, 24, 3877.
(b) Okada, H.; Mori, T.; Saikawa, Y.; Nakata, M. Tetrahedron Lett. 2009,
50, 1276. (c) Wiktelius, D.; Luthman, K. Org. Biomol. Chem. 2007, 5,
603. (d) Molander, G. A.; Oliveira, R. A. Tetrahedron Lett. 2008, 49,
1266. (e) Rothman, J. H. J. Org. Chem. 2007, 72, 3945. (f) Hisler, K.;
Tripoli, R.; Murphy, J. A. Tetrahedron Lett. 2006, 47, 6293. (g) Ye, L.-W.;
Han, X.; Sun, X.-L.; Tang, Y. Tetrahedron 2008, 64, 8149. See also ref 8d.
(10) Catalytic Wittig reaction: O’Brien, C. J.; Tellez, J. L.; Nixon, Z. S.;
Kang, L. J.; Carter, A. L.; Kunkel, S. R.; Przeworski, K. C.; Chass, G. C.
Angew. Chem., Int. Ed. 2009, 48, 6836. For a review on the topic of the
phosphine-catalyzed Wittig reaction, see also: Fairlamb, I. J. S.
ChemSusChem 2009, 2, 1021.
(11) Wittig-type reactions of phosphonium ylides with N-sulfonyl
imines: (a) Dong, D.-J.; Li, H. H.; Tian, S. K. J. Am. Chem. Soc. 2010,
132, 5018. (b) Dong, D.-J.; Li, Y.; Wang, J.-Q.; Tian, S. K. Chem.
Commun. 2011, 47, 2158. (c) Fang, F.; Li., Y.; Tian, S. K. Eur. J. Org.
Chem. 2011, 1084. Amine and N-sulfonyl imine-promoted Wittig
reactions: (d) McNulty, J.; McLeod, D. Chem. Eur. J. 2011, 17, 8794.
(12) (a) Investigation of reaction kinetics and the relative
nucleophilicity of phosphonium ylides: Appel, R.; Loos, R.; Mayr, H.
J. Am. Chem. Soc. 2009, 131, 714. (b) Mechanistic investigation on
Wittig reactions of ketones: Ghosh, A.; Chakratborty, I.; Adarsh, N. N.;
Lahiri, S. Tetrahedron 2010, 66, 164. (c) Computational study on the
stability of heteroatom-stabilized ylides including phosphonium ylides:
Fu, Y.; Wang, H.-J.; Chong, S.-S.; Guo, Q.-X.; Liu, L. J. Org. Chem. 2009,
74, 810.
(13) (a) Robiette, R.; Richardson, J.; Aggarwal, V. K.; Harvey, J. N. J.
Am. Chem. Soc. 2005, 127, 13468. (b) Robiette, R.; Richardson, J.;
Aggarwal, V. K.; Harvey, J. N. J. Am. Chem. Soc. 2006, 128, 2394.
(c) Harvey, J. N. Faraday Discuss. 2010, 145, 487.
(14) Vedejs, E.; Peterson, M. J. In Topics in Stereochemistry; Eliel, E. L.,
Wilen, S. H., Eds.; Wiley: New York, 1994; Vol. 21, p 1.
Experimental procedures; details of the assignments of alkene Z/E
ratios, OPA cis/trans ratios, and β-HPS erythro/threo ratios; and
full characterization of all new phosphonium salts, alkenes, and β-
HPSs. This material is available free of charge via the Internet at
http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
declan.gilheany@ucd.ie
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank sincerely the Irish Research Council for Science,
Engineering and Technology (IRCSET) for funding this
chemistry through an EMBARK Scholarship to P.A.B. D.G.G.
thanks sincerely University College Dublin for a President’s
Research Fellowship during which a substantial part of the
conception of this work and initial analysis of the results took
place. The Fellowship was held partly in Stanford University in
the laboratory of Prof. James Collman, to whom D.G.G. is
warmly appreciative for both his hospitality and stimulating
intellectual discussions. We are also very grateful to UCD Centre
for Synthesis and Chemical Biology (CSCB) and the UCD
School of Chemistry and Chemical Biology for access to their
extensive analysis facilities and especially to Dr. Jimmy Muldoon
and Dr. Yannick Ortin for NMR. Finally we thank Prof. E. Vedejs,
University of Michigan, for helpful discussions and historical
background on the Wittig reaction.
REFERENCES
■
(1) Wittig, G.; Geissler, G. Justus Liebigs Ann. Chem. 1953, 580, 44.
(2) Maryanoff, B. E.; Reitz, A. B. Chem. Rev. 1989, 89, 863.
(3) Johnson, A. W. Ylides and Imines of Phosphorus; Wiley: New York,
1993; Chapters 8 and 9, pp 221−305.
(4) (a) Abell, A. D.; Edmonds, M. K. In Organophosphorus Reagents;
Murphy, P. J., Ed.; Oxford University Press: Oxford, UK, 2004; pp 99−
127. (b) Schobert, R. In Organophosphorus Reagents; Murphy, P. J., Ed.;
Oxford University Press: Oxford, UK, 2004; pp 12−148. (c) Edmonds,
M.; Abell, A. In Modern Carbonyl Olefination; Takeda, T., Ed.; Wiley-
VCH: Weinheim, Germany, 2004; Chapter 1. (d) Kolodiazhny, O. I.
Phosphorus Ylides: Chemistry and Application in Organic Synthesis; Wiley:
New York, 1999. (e) Gosney, I.; Rowley, A. G. In Organophosphorus
Reagents in Organic Synthesis; Cadogan, J. I. G., Ed.; Academic Press:
New York, 1979; Chapter 2, pp 17−153. (f) Schlosser, M. In Topics in
Stereochemistry; Eliel, E. L., Allinger, N. L., Eds.; Wiley: New York, 1970;
Vol. 5, p 1.
(5) New methods for ylide generation: (a) Aggarwal, V. K.; Fulton, J.
R.; Sheldon, C. G.; de Vicente, J. J. Am. Chem. Soc. 2003, 125, 6034.
(b) Fan, R.-H.; Hou, X.-L.; Dai, L.-X. J. Org. Chem. 2004, 69, 689.
(c) Lebel, H.; Pacquet, V.; Proulx, C. Angew. Chem., Int. Ed. 2001, 40,
2887. (d) Ramazani, A.; Kazemizadeh, A. R.; Ahmadi, E.;
Noshiranzadeh, N.; Souldozi, A. Curr. Org. Chem. 2008, 12, 59.
(e) Schwartz, B. D.; Williams, C. M.; Anders, E.; Bernhardt, P. V.
Tetrahedron 2008, 64, 6482. (f) Liu, D.-N.; Tian, S. K. Chem. Eur. J.
2009, 15, 4538. (g) Zhou, R.; Wang, C.; Song, H.; He, Z. Org. Lett. 2010,
12, 976. (h) Xu, S.; Zou, W.; Wu, G.; Song, H.; He, Z. Org. Lett. 2010, 12,
3556. (i) Das, P.; McNulty, J. Eur. J. Org. Chem. 2010, 3857.
(6) Methods for reactions showing atypical stereoselectivity:
(a) McNulty, J.; Keskar, K. Tetrahedron Lett. 2008, 49, 7054. (b) Oh,
J. S.; Kim, B. H.; Kim, Y. G. Tetrahedron Lett. 2004, 45, 3925. (c) Wang,
Z. G.; Zhang, G. T.; Guzei, I.; Verkade, J. G. J. Org. Chem. 2001, 61, 3521.
(d) Nishimura, Y.; Shiraishi, T.; Yamaguchi, M. Tetrahedron Lett. 2008,
(15) Vedejs, E.; Peterson, M. J. In Advances in Carbanion Chemistry;
Snieckus, V., Ed.; JAI: Greenwich, CT, 1996; Vol. 2.
(16) Isolation of betaine−lithium halide complex from Wittig reaction:
Schlosser, M.; Christmann, K. F. Liebigs Ann. Chem. 1967, 708, 1.
L
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(17) Isolation of β-HPS by acid quenching of Wittig reaction: Wittig,
low-temperature OPA cis/trans ratio and ultimate alkene Z/E ratio after
warming of the OPA (favouring production of the E-isomer,
“stereochemical drift”) has been observed in the reaction of Ph3P
CH(n-Pr) and benzaldehyde in the presence of LiBr.20 Positive
crossover experiments indicate that the conversion of cis-OPA into
trans-OPA in the examples mentioned above of reactions of
alkylidenetriphenylphosphoranes with benzaldehyde in the presence
of Li+ involves OPA reversal to ylide and aldehyde.19,25 LiBr has been
shown to exert an effect on the stereoselectivity of OPA formation in the
reaction of Ph3PCHCH3 and PhCH2CH2CHO, but not to affect the
stereochemical ratio of the OPA formed from these reactants if added to
a pre-formed solution of the OPA to give betaine−LiBr complex.19 Also
no stereochemical drift was observed in the reaction of Ph3PCH(n-Pr)
with hexanal in the presence of LiBr in THF.20 Thus it appears that OPA
formation is irreversible in reactions of aliphatic aldehydes, even in the
presence of Li+, and that the diminished Z-selectivity in these reactions
arises from the effect of Li+ on the initial formation of OPA. Although
the direct formation of betaine−LiBr complex from ylide and aldehyde
in the presence of LiBr is certainly plausible, the presence of such
complexes in Wittig reaction mixtures does not prove the involvement
of betaines (salt-free or as complexes with lithium halide) in the process
of OPA formation from ylide and aldehyde in the presence of Li salt,
since it has been shown that betaine−LiBr complex can be formed
directly from OPA.19 Similarly, the fact that reactions of benzaldehyde
with non-stabilized ylides in the presence of Li+ undergo reversal of OPA
to ylide and aldehyde19,25 (and that the betaine−LiBr complex produced
in the reaction of the semi-stabilized ylide mentioned above undergoes
reversal)22 does not necessarily mean that betaines are involved on the
pathway from OPA to ylide and aldehyderather it may occur by direct
cycloreversion. It should also be noted that the “Li salt-free” cyclo-
addition mechanism operates in reactions conducted in the presence of
Li salts in solvents that coordinate Li ion strongly (DMSO, DMF),33 in
reactions with low Li ion concentration in solvents that coordinate Li ion
poorly (THF, toluene),20 and in reactions of alkylidenetriphenylphos-
phoranes with tertiary aldehydes, on which Li salts appear to exert no
influence.14
G.; Schollkopf, U. Chem. Ber. 1954, 87, 1318.
̈
(18) Studies involving the reaction of phosphine with epoxide to give
alkene and phosphine oxide through an obligatory (Li salt-containing)
betaine intermediate: (a) Speziale, A. J.; Bissing, D. E. J. Am. Chem. Soc.
1963, 85, 3878. (b) Wittig, G.; Haag, W. Chem. Ber. 1955, 88, 1654.
(19) Vedejs, E.; Meier, G. P.; Snoble, K. A. J. J. Am. Chem. Soc. 1981,
103, 2823−2831.
(20) Reitz, A. B.; Nortey, S. O.; Jordan, A. D., Jr; Mutter, M. S.;
Maryanoff, B. E. J. Org. Chem. 1986, 51, 3302.
(21) Reference 14, pp 50−54.
(22) Vedejs, E.; Huang, W. F. J. Org. Chem. 1984, 49, 210.
(23) Maryanoff, B. E.; Reitz, A. B.; Mutter, M. S.; Inners, R. R.;
Almond, H. R., Jr. J. Am. Chem. Soc. 1985, 107, 1068.
(24) Reitz, A. B.; Mutter, M. S.; Maryanoff, B. E. J. Am. Chem. Soc. 1984,
106, 1873.
(25) Maryanoff, B. E.; Reitz, A. B.; Mutter, M. S.; Inners, R. R.;
Almond, H. R., Jr.; Whittle, R. R.; Olofson, R. A. J. Am. Chem. Soc. 1986,
108, 7664.
(26) Vedejs, E.; Marth, C. F.; Ruggeri, R. J. Am. Chem. Soc. 1988, 110,
3940.
(27) Jones, M. E.; Trippett, S. J. Chem. Soc. (C) 1966, 1090.
(28) (a) See reference 6b. (b) Anderson, R. J.; Henrick, C. A. J. Am.
Chem. Soc. 1975, 97, 4327.
(29) See, for example, ref 4e.
(30) See, for example, ref 4c, p 3.
(31) For example, see ref 4a, pp 99 and 104.
(32) The isolation from Wittig reactions of non-stabilized ylides of
betaine−lithium halide complexes16 or β-HPS17 after addition of acid,
along with the fact that alkene could be generated by the reaction of
epoxide with phosphine,18 led, not unreasonably, to the presumption
that betaines are necessarily involved as intermediates in the Wittig
reaction. Z-Selectivity in Wittig reactions of alkylidenetriphenylphos-
phoranes was proposed to arise from irreversible addition of ylide to
aldehyde to form erythro-betaine and hence cis-OPA and Z-alkene. Anti-
addition of ylide to aldehyde was proposed to be the kinetically preferred
route, giving anti-erythro-betaine. Bond rotation of the initially formed
betaine gives syn-erythro-betaine, which cyclizes to cis-OPA and hence
gives Z-alkene. E-Selectivity in Wittig reactions of stabilized ylides, and
mixed selectivity in reactions of semi-stabilized ylides, was explained by
reversible formation of the betaine intermediate or the subsequent OPA
intermediate, with kinetically favored cyclization of threo-betaine to
trans-OPA or decomposition of trans-OPA to E-alkene resulting in the
equilibrium between the betaines or OPAs being effectively biased
toward the (already thermodynamically favored) threo-betaine or trans-
OPA.29 However, the operation of kinetic control in Li salt-free Wittig
reactions of all three ylide types has been demonstrated (most
significantly for stabilized ylides),26,35 and there is no evidence for
kinetically preferred decomposition of threo-betaine or trans-OPA to
E-alkene, which renders these mechanistic arguments inoperative.
(33) Only a handful of Li salt-free Wittig reactions are known to
undergo reversible formation of OPA, and all such examples are
reactions of non-stabilized ylides. The proposal that Wittig reactions
(especially of semi-stabilized and stabilized ylides) are generally
reversible grew out of the betaine mechanism (see footnote 32) and
the effects exerted by Li cation on the stereoselectivities of the
reaction.19−21 However, even for reactions conducted in the presence of
Li salts, the involvement of betaine in the process of OPA formation
from ylide and aldehyde is not certain. Li+ causes diminished
Z-selectivity in reactions of alkylidenetriphenylphosphoranes,3,14 and
also results in altered selectivity in reactions of semi-stabilized ylides
compared to Li salt-free conditions, while little is known regarding the
effect of Li+ (or other cations) on reactions of stabilized ylides, which are
usually formed and purified (and are thus salt-free) prior to the
reaction.3 Betaine−LiBr complexes have been isolated from reactions of
non-stabilized ylides,16,19 and there exists one report in which betaine−
LiBr complexes have been formed in reactions of a semi-stabilized ylide
and undergo Wittig reversal, as judged by positive crossover
experiments.22 Furthermore, a non-correspondence between the initial
(34) The observation of crossover product in β-HPS deprotonation
experiments in alcohol solvent has also led to the suggestion of the
operation of reversibility in Wittig reactions in alcohol solvents.27
However, it has since been shown that, at least in certain circumstances,
the addition of methanol at low temperature (i.e., before OPA
decomposition occurs) to the Wittig reactions of non-stabilized ylides
causes very high E-selectivity in the reactions, which in the absence of
methanol (or if it is added after warming to room temperature) would
show high Z-selectivity.28
(35) Vedejs, E.; Fleck, T. J. Am. Chem. Soc. 1989, 111, 5861.
(36) Reitz, A. B.; Mutter, M. S.; Maryanoff, B. E. J. Am. Chem. Soc. 1984,
106, 1873.
(37) Vedejs, E.; Marth, C. F. J. Am. Chem. Soc. 1990, 112, 3905.
(38) Vedejs, E.; Marth, C. F. J. Am. Chem. Soc. 1988, 110, 3948.
(39) Although the necessity for the formation of oxaphosphetane was
recognized by Wittig at the outset (ref 1), he did not consider how
exactly it would be formed, nor did he specify whether it would be a true
intermediate or a transition state. Subsequently, in the absence of
definitive evidence, he settled on the betaine mechanism as being the
most plausible: Wittig, G.; Haag, A. Chem. Ber. 1963, 96, 1536.
(40) Among the earlier mechanistic workers, the fact that the [2+2]
mechanism superficially contravenes the Woodward−Hofmann rules
was a factor in slowing its acceptance/development: Trippett, S.,
personal communication to D.G.G., 1983.
(41) Vedejs, E.; Snoble, K. A. J. J. Am. Chem. Soc. 1973, 95, 5778.
(42) Vedejs, E.; Fuchs, P. L. J. Am. Chem. Soc. 1973, 95, 822.
(43) Vedejs, E; Fleck, T.; Hara, S. J. Org. Chem. 1987, 52, 4637.
(44) Vedejs, E.; Snoble, K. A. J.; Fuchs, P. L. J. Org. Chem. 1973, 38,
1178.
(45) Vedejs, E.; Marth, C. F. J. Am. Chem. Soc. 1989, 111, 1519.
(46) Vedejs, E.; Marth, C. F. Tetrahedron Lett. 1987, 28, 3445.
(47) Maryanoff, B. E.; Reitz, A. B.; Graden, D. W.; Almond, H. R.
Tetrahedron Lett. 1989, 30, 1361.
M
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(48) Schlosser, M.; Schaub, B. J. Am. Chem. Soc. 1982, 104, 5821.
(49) Reference 12a.
(50) See ref 6e.
(51) Bangerter, F.; Karpf, M.; Meier, L. A.; Rys, P.; Skrabal, P. J. Am.
Chem. Soc. 1998, 120, 10653.
(52) Ward, W. J.; McEwen, W. E. J. Org. Chem. 1990, 55, 493.
(53) Frøyen, P. Acta Chem. Scand. 1972, 26, 2163.
(54) Blade-Font, A.; Vanderwerf, C. A.; McEwen, W. E. J. Am. Chem.
Carey, F. A. Organic Chemistry, 7th ed.; McGraw-Hill: New York, 2008;
pp 729−730. Wade, L. G., Jr. Organic Chemistry, 7th ed.; Prentice-Hall:
Englewood Cliffs, NJ, 2010; p 836. See also refs 29 and 30.
(61) Some texts still retain the betaine mechanism as the only
explanation of the Wittig reaction mechanism, e.g.: Vollhardt, K. P. C.;
Schore, N. E. Organic Chemistry: Structure & Function, 6th ed.; Freeman:
2011; pp 806−807.
(62) Some texts do definitively show OPA as the first-formed
intermediate in the Wittig reaction: Bruice, P. Y. Organic Chemistry, 6th
ed.; Prentice-Hall: Englewood Cliffs, NJ, 2011; pp 805−806. Smith, J. G.
Organic Chemistry, 2nd ed.; McGraw-Hill: New York, 2008; pp 794−
795. However, even these do not tackle the source of stereoselectivity in
the reaction.
(63) OPAs derived from semi-stabilized ylides containing the
dibenzophosphole moiety have been observed by low-temperature
NMR. For these, stereospecific conversion to alkene has been
demonstrated by the same method as was used for reactions of non-
stabilized ylides.35
(64) Harcken, C.; Martin, S. F. Org. Lett. 2001, 3, 3591.
(65) Cotter, J.; Hogan, A-M. L.; O’Shea, D. F. Org. Lett. 2007, 9, 1493−
1496.
(66) Dunne, E. C.; Coyne, E. J.; Crowley, P. B.; Gilheany, D. G.
Tetrahedron Lett. 2002, 43, 2449.
(67) This can be synthetically convenient; for example, it may have
contributed to the results of recent microwave-accelerated reactions
(see ref 7c).
(68) Byrne, P. A.; Higham, L. J.; McGovern, P.; Gilheany, D. G.,
manuscript in preparation. In this study Z-enones, derived from keto-
stabilized ylides, were found to isomerize spontaneously under ambient
conditions. Similar isomerization has also been observed in the present
study to occur slowly for Z-2,2′-diiodostilbene.
Soc. 1960, 82, 2396.
(55) The rate of OPA decomposition equals the rate of alkene
formation in reactions of non-stabilized ylides,24,25,51 and the
diastereomeric ratios of the OPA and alkene are identical in these
reactions and in those of semi-stabilized ylides in which OPAs can be
observed spectroscopically.23−26,35,46,50,51 In experiments involving
generation of OPA by means independent of a Wittig reaction (via
transient betaine species) from precursors of defined stereochemistry
(e.g., β-HPS, epoxide), the stereochemistry of the precursor is retained
in the alkene product after OPA decomposition. This has been observed
in experiments involving OPAs derived from non-stabi-
lized,23−26,41,42,46,51 semi-stabilized,35,42,52 and stabilized ylides.35 In a
particular Wittig reaction of a P-stereogenic semi-stabilized phospho-
nium ylide, the stereochemistry at phosphorus was retained in the
phosphine oxide.54 All of this evidence is consistent with the operation
of a syn-elimination of phosphine oxide from a cyclic intermediate with
trigonal bipyramidal phosphorus.
(56) Non-stabilized ylides of the general structure RaRbRcPCHR1,
where R1 = alkyl and the phosphonium RaRbRcP moiety is not Ar3P or t-
BuAr2P, show much lower kinetic selectivity for cis-OPA.14,25,26 This can
be rationalized as being a consequence of decreased 1−3 interactions
and thus a lower propensity toward TS puckering. The lower Z-
selectivity generally observed in reactions of alkylidenetrialkylphosphor-
anes (see ref 14) is not only a consequence of the lower kinetic
selectivity for cis-OPA mentioned here. The cis-OPA reverts to ylide and
aldehyde while the trans-OPA does not, and thus the trans-OPA
accumulates. See refs13, 25, and 26.
(69) Byrne, P. A.; Rajendran, K. V.; Muldoon, J.; Gilheany, D. G. Org.
Biomol. Chem. 2012, 10, 3531.
(70) For example, although 2,2′-difluorostilbene can be obtained with
very high Z-selectivity (Z/E = 94:6) in the reaction of 2-
fluorobenzylidene-triphenylphosphorane and 2-fluorobenzaldehyde
(see ref 69), it is excluded from this study due to the propensity of
the Z-isomer to undergo isomerzationsee ref 66.
(57) The trans-selective TS was initially proposed to be planar, which
would be consistent with the observed high E-selectivity.35
(58) For example, a negatively puckered cis-selective TS with a
favorable relative orientation of reactant dipoles (Figure 1b with
aldehyde H and R1 swapped) suffers from strong 1−3 (as well as
significant 1−2) interactions, while one with the opposite sense of
puckering (Figure 1d) has an electrostatically disfavored disposition of
reactant dipoles. A planar cis-selective TS (Figure 1c) should not be
particularly disfavored electrostatically, but suffers from large 1−2
interactions and lacks the electrostatically favored anti-parallel
orientation of reactant dipoles that is present in the TS of Figure 1b.
Thus high E-selectivity is observed in general in Wittig reactions of
stabilized ylides in non-polar or polar-aprotic media, with selectivity
being extremely high for ylides in which phosphorus bears bulky
substituents (e.g., TPP-derived ylides), as the possibility of large 1−3
interactions dictates that these discriminate particularly well against
puckering of the cis-selective TS. If the phosphorus substituents are not
so bulky (e.g., if one phenyl is replaced by methyl), then a different cis-
selective TS with an anti-parallel orientation of reactant dipoles can be
envisaged (Figure 1b with the positions of R1 and H swapped) since
placement of the large substituent in the pseudo-axial position is not
discriminated against to the same extent by 1−3 interactions, consistent
with the lower E-selectivity observed (see ref 14).
(59) Diminished E-selectivity, and even predominant Z-selectivity, has
been observed in reactions of stabilized ylides in methanol (see refs 2,
14, and 35). It seems likely that this results from a solvent-induced
decrease in the importance of the interaction of reactant dipoles in the
cycloaddition TSs. In this scenario, the factors governing TS geometry
may be quite similar to those in reactions of semi-stabilized ylides.
(60) It is still common for modern undergraduate texts to state that one
or the other of OPA or betaine is the first-formed intermediate, but then
to qualify this by saying that some or all Wittig reactions may react by the
other intermediate under certain circumstances. For example see:
McMurry, J. Organic Chemistry, 8th ed.; Brooks/Cole: 2011; p 747.
(71) For example: (i) The non-isomerization of many of the other
stilbenes in contact with alumina was demonstrated by subjecting
samples of purified stilbene heavily enriched in the Z-isomer (or in some
cases containing only the Z-isomer) to the typical chromatographic
conditions used to purify the stilbenesi.e., elution through neutral
alumina using cyclohexane or pentane as solvent. In all cases the stilbene
Z/E ratio was unchanged after being chromatographed. (ii) For
reactions of ester-stabilized ylides, experiments were carried out to show
that the Z-isomer of the enoate product did not isomerise under the
reaction conditions. See the Supporting Information for further details
of these control experiments.
(72) See ref 28b.
(73) OPA cis/trans ratio determined from the erythro/threo ratio of the
β-HPS produced in the low-temperature acid quenching of the Wittig
reaction.
(74) OPA cis/trans ratio determined by low-temperature 31P NMR of
the Wittig reaction mixture.
(75) (a) McEwen, W. E.; James, A. B.; Knapczyk, J. W.; Killingstad, V.
L.; Shiau, W.-I.; Shore, S.; Smith, J. H. J. Am. Chem. Soc. 1978, 100, 7304.
(b) McEwen, W. E.; Fountaine, J. E.; Schulz, D. N.; Shiau, W. I. J. Org.
Chem. 1976, 41, 1684. (c) McEwen, W. E.; Shiau, W. I.; Yeh, Y. I.;
Schulz, D. N.; Pagilagan, R. U.; Levy, J. B.; Symmes, C.; Nelson, G. O.;
Granoth, I. J. Am. Chem. Soc. 1975, 97, 1787. (d) McEwen, W. E.;
Kyllingstad, V. L.; Schulz, D. N.; Yeh, Y. I. Phosphorus 1981, 1, 145.
(76) Keldsen, G. L.; McEwen, W. E. J. Am. Chem. Soc. 1978, 100, 7312.
(77) McEwen, W. E.; Cooney, J. V. J. Org. Chem. 1983, 48, 983.
(78) Zhang, X.; Schlosser, M. Tetrahedron Lett. 1993, 34, 1925.
(79) Yamataka, H.; Nagareda, K.; Ando, K.; Hanafusa, T. J. Org. Chem.
1992, 57, 2865.
(80) Cushman, M.; Nagarathnam, D.; Gopal, D.; Chakraborti, A. K.;
Lin, C. M.; Hamel, E. J. J. Med. Chem. 1991, 34, 2579−2588.
N
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(81) (a) Harrowven, D. C.; Guy, I. L.; Howell, M.; Packham, G. Synlett
2006, 2977. (b) Harrowven, D. C.; Guy, I. L.; Nanson, L. Angew. Chem.,
Int. Ed. 2006, 45, 2242.
(108) Hierso, J.-C.; Fihri, A.; Ivanov, V. V.; Hanquet, B.; Pirio, N.;
Donnadieu, B.; Rebier
2004, 126, 11077.
̀
e, B.; Amardeil, R.; Meunier, P. J. Am. Chem. Soc.
(109) Bonnafoux, L.; Ernst, L.; Leroux, F. R.; Colobert, F. E. J. Org.
(82) Takenaka, N.; Sarangthem, R. S.; Captain, B. Angew. Chem., Int.
Chem. 2011, 3387.
(110) Maji, P.; Krishnamurthy, S. S.; Nethaji, M. New J. Chem. 2010,
34, 1478.
Ed. 2008, 47, 9708−9710.
(83) Schlosser, M.; Schaub, B.; de Oliveira-Neto, J.; Jegenathan, S.
Chimia 1986, 40, 244.
(111) Mo,
́ ́ ́ ́
O.; Yanez, M.; Eckert-Maksic, M.; Maksic, Z. B.; Alkorta, I.;
̃
(84) Reference 14, pp 61−70.
Elguero, J. J. Phys. Chem. A 2005, 109, 4359.
(85) The reaction of benzylidenemethyldiphenylphosphorane 3a with
benzaldehyde favors E-selectivity to a much greater extent than the
corresponding reaction of benzyildenetriphenylphosphorane 4a, and
hence the greater shift toward Z-selectivity in reactions of the former
with ortho-heteroatom-substituted benzaldehydes represents a much
larger decrease in the activation energy of the cis-selective pathway in
these reactions.
(112) Kawashima, T.; Watanabe, K.; Okazaki, R. Tetrahedron Lett.
1997, 38, 551.
(113) Matsukawa, S.; Kojima, S.; Kajiyama, K.; Yamamoto, Y.; Akiba,
K.-y.; Re, S.; Nagase, S. J. Am. Chem. Soc. 2002, 124, 13154.
(86) KHMDS was used to generate the ylide from the precursor
phosphonium salt.
(87) Byrne, P. A.; Muller-Bunz, H.; Gilheany, D. G., manuscript in
̈
preparation.
(88) See ref 2, pp 882−887, in which can be found an extensive
collection of examples of atypical Wittig reaction stereoselectivity
apparently induced by the presence of a suitably placed heteroatom on
the aldehyde reactant. Amongst these are included numerous examples
of high Z-selectivity in reactions of β-alkoxy aldehydes. The vast majority
of these examples were carried out in polar protic solventsmost
commonly methanol. The same section of ref 2 also includes examples
of high Z-selectivity in reactions of α-heteroatom-substituted aldehydes.
Aldehyde 65, in addition to being a β-heteroatom-substituted aldehyde,
also has a heteroatom α to the carbonyl, which may have some impact on
the level of Z-selectivity observed with this aldehyde.
(89) Tronchet, J. M. J.; Gentile, B. Helv. Chim. Acta 1979, 62, 2091.
(90) Valverde, S.; Martin-Lomas, M.; Herradon, B.; Garcia Ochoa, S.
Tetrahedron 1987, 43, 1987.
(91) Suzuki, M.; Doi, H.; Kato, K.; Bjorkman, M.; Langstrom, B.;
̈
̊
̈
Watanabe, Y.; Noyori, R. Tetrahedron 2000, 56, 8263.
(92) Jarosz, S.; Skor
11, 1997.
́
a, S.; Szewczyk, K. Tetrahedron:Asymmetry 2000,
(93) Suzuki, M.; Kato, K.; Watanabe, Y.; Satoh, T.; Matsumura, K.;
Watanabe, Y.; Noyori, R. Chem. Commun. 1999, 307.
(94) Krueger, E. B.; Hopkins, T. P.; Keaney, M. T.; Walter, M. A.;
Boldi, A. M. J. Comb. Chem. 2002, 4, 229.
(95) The reaction of semi-stabilized ylide benzylidenetriphenylphos-
phorane (4a) with a variant of aldehyde 65 (with a benzyloxy group in
place of the methoxy group) in THF is reported on p 886 of ref 2 to give
a Z/E ratio of 95:5.
(96) This ylide was chosen as a test substrate for the reaction with 65
due to its low intrinsic kinetic Z-selectivity in reactions with aliphatic
aldehydes lacking a β-heteroatom substituent (see Table 6, entry 9).
(97) Hannesian, S.; Huang, G.; Chenel, C.; Machaalani, R.; Loiseleur,
O. J. Org. Chem. 2005, 70, 6721.
(98) Related P-methyldibenzophosphole-derived semi-stabilized
ylides have previously been reported to show exceptionally high E-
selectivity in reactions with cyclohexanecarboxaldehyde; see ref 46.
(99) McEwen, W. E.; Smith, J. H.; Woo, E. J. J. Am. Chem. Soc. 1980,
102, 2746.
(100) Cairns, S. M.; McEwen, W. E. J. Org. Chem. 1987, 52, 4829.
(101) Nakanishi, W.; Hayashi, S. J. Org. Chem. 2002, 67, 38.
(102) Yamane, K.; Hayashi, S.; Nakanishi, W.; Sasamori, T.; Tokitoh,
N. Polyhedron 2008, 27, 3557.
(103) Hayashi, S.; Yamane, K.; Nakanishi, W. J. Org. Chem. 2007, 72,
7587.
(104) Kilian, P.; Slawin, A. M. Z.; Woollins, J. D. Phosphorus, Sulfur,
Silicon Relat. Elem. 2004, 179, 999.
(105) Reiter, S. A.; Nogai, S. D.; Karaghiosoff, K.; Schmidbaur, H. J.
Am. Chem. Soc. 2004, 126, 15833.
́
(106) Owsianik, K.; Chauvin, R.; Balinska, A.; Wieczorek, M.;
Mikołajczyk, M. Organometallics 2009, 28, 4929.
(107) Zahn, S.; Frank, R.; Hey-Hawkins, E.; Kirchner, B. Chem Eur. J.
2011, 17, 6034.
O
dx.doi.org/10.1021/ja300943z | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX