Normal, Abnormal, and Cascade Wittig Olefinations of α-Oxoketenes

Article abstract of DOI:10.1002/chem.201801533

α-Oxoketenes generated in situ by a thermal Wolff rearrangement have been found to participate as 1,2- and 1,4-ambident C-electrophilic/O-nucleophilic reagents towards donor/acceptor carbonyl-stabilized Wittig ylides. This resulted in the very direct and practical syntheses of functionalized allenes by a normal Wittig olefination, 4H-pyran-4-ones by an abnormal Wittig olefination, or 4H-pyranylidenes following a Wittig/abnormal Wittig cascade sequence as a function of the substrates combination employed. Mechanistic experimental and computational studies provided a full rational for these reactivity switches. Some unusual mechanistic features for Lewis acid-free Wittig olefinations were identified in this series such as the involvement of betaine intermediates and some degree of reversibility in the normal Wittig olefination. The abnormal Wittig olefination was fully uncovered.

Full text of DOI:10.1002/chem.201801533

DOI: 10.1002/chem.201801533
Full Paper
&
Synthetic Methods |Hot Paper|
Normal, Abnormal, and Cascade Wittig Olefinations of
a-Oxoketenes
David Pierrot,[a] Marc Presset,[a, b] Jean Rodriguez,[a] Damien Bonne,*[a] and Yoann Coquerel*[a]
Abstract: a-Oxoketenes generated in situ by a thermal Wolff
rearrangement have been found to participate as 1,2- and
1,4-ambident C-electrophilic/O-nucleophilic reagents to-
wards donor/acceptor carbonyl-stabilized Wittig ylides. This
resulted in the very direct and practical syntheses of func-
tionalized allenes by a normal Wittig olefination, 4H-pyran-4-
ones by an abnormal Wittig olefination, or 4H-pyranylidenes
following a Wittig/abnormal Wittig cascade sequence as a
function of the substrates combination employed. Mechanis-
tic experimental and computational studies provided a full
rational for these reactivity switches. Some unusual mecha-
nistic features for Lewis acid-free Wittig olefinations were
identified in this series such as the involvement of betaine
intermediates and some degree of reversibility in the normal
Wittig olefination. The abnormal Wittig olefination was fully
uncovered.
Introduction
ketenes. Several examples of the anticipated a,a’-bisoxoallene
synthesis could indeed be realized. However, in a number of
cases the reaction between a-oxoketenes and carbonyl-stabi-
lized ylides followed alternative pathways involving either ab-
normal Wittig olefinations or cascade normal/abnormal Wittig
processes to afford 4H-pyran derivatives. In this article, we
report in full detail the results of our experimental and theoret-
ical investigations on the synthetically valuable chemical di-
chotomy in the reactions of a-oxoketenes with carbonyl-stabi-
lized Wittig ylides.
a-Oxoketenes are densely functionalized, highly reactive, elec-
trophilic reaction intermediates of past and current great inter-
est, not only because of mechanistic and theoretical considera-
tions, but also because of their rich chemistry as building
blocks in organic synthesis.[1] With very few exceptions, they
cannot be isolated under ordinary conditions and must be
generated in situ. The two most convenient and commonly
employed methods for the generation of a-oxoketenes are the
thermal decomposition of dioxinones,[2] and the thermal or
photochemical Wolff rearrangement of 2-diazo-1,3-dicarbonyl
compounds.[3] For the work described herein, and based on
some precedents with simple ketenes,[4] it was considered that
unsymmetrically functionalized a,a’-bisoxoallenes, for example,
4a–f, a class of chiral molecules difficult to obtain and only
available through relatively complex multistep synthetic se-
quences,[5] could be readily prepared by a Wolff rearrange-
ment/Wittig olefination cascade reaction from diazo com-
pounds and carbonyl-stabilized ylides via intermediate a-oxo-
Results and Discussion
Our investigations started with the acyclic diazo compounds
1a–c, which were allowed to react with the representative
Wittig ylides 2a–c in toluene at 140 or 1708C under microwave
irradiation conditions promoting rapid, quantitative, and regio-
selective (when applicable), Wolff rearrangements of the diazo
compounds
1 into the corresponding a-oxoketenes 3
(Scheme 1).[1e–h] As expected, the desired a,a’-bisoxoallenes
4a–f were obtained in fair to good yields, unlocking the practi-
cal synthesis of this valuable class of compounds. As an inter-
esting observation, we also found evidence for the formation
of a minor amount of g-pyrone 5l (15%) in the reaction be-
tween diazo 1a and ylide 2a. The formation of 5l can be ra-
tionalized by invoking a 1,4-C-electrophilic/O-nucleophilic be-
havior of the a-oxoketene 3a triggering an abnormal Wittig
olefination (see mechanistic discussion below).[6] g-Pyrones are
valuable chemical moieties with a relatively frequent occur-
rence in biologically active natural and non-natural products.[7]
Existing synthetic methods forging g-pyrones are often of lim-
ited breadth, hampering the exploration of their potential in
chemistry and chemical biology,[7,8] and alternative methods
are desirable.
[a] Dr. D. Pierrot, Dr. M. Presset, Prof. Dr. J. Rodriguez, Dr. D. Bonne,
Dr. Y. Coquerel
Aix Marseille Univ, CNRS, Centrale Marseille, iSm2
Marseille (France)
E-mail: damien.bonne@univ-amu.fr
yoann.coquerel@univ-amu.fr
[b] Dr. M. Presset
Present address: Universitꢀ Paris Est, ICMPE (UMR 7182), CNRS, UPEC
2-8 rue Henri Dunant, 94320 Thiais (France)
Supporting information and the ORCID identification number(s) for the au-
thor(s) of this article can be found under:
https://doi.org/10.1002/chem.201801533. It contains the compound num-
bering employed, general experimental considerations, some complementa-
ry experimental observations, the detailed experimental procedures, the
characterization data and copies of NMR spectra for all compounds, and
the full-details computational mechanistic study.
Chem. Eur. J. 2018, 24, 1 – 10
1
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
intermediate derived from 2-diazo-1,3-cyclohexanone (1d; R1,
R2 =-(CH2)3-) reacted with the methyl ketone-stabilized Wittig
ylide 2d (R3 =Me) to afford the fused bicyclic 4H-pyran-4-one
5a in 69% yield, the structure of which was confirmed by X-
ray diffraction analysis. Similarly, 4H-pyran-4-ones incorporating
either an alkyl (in 5b), hindered alkyl (in 5c, 5d and 5i), phenyl
(in 5e), carbethoxyl (in 5 f), or carbethoxymethyl (in 5g) sub-
stituent, or a simple hydrogen atom (in 5h) at the 6 position,
could be prepared. The reaction with the non-symmetric diazo
ketoamide 1c (R1 =NMe2, R2 =Me) afforded exclusively the 2-
dimethylamino-functionalized 4H-pyran-4-one 5j in 74% yield,
following a regioselective Wolff rearrangement of the diazo
compound. A variation of the reaction was used for the prepa-
ration of the 4H-pyran-4-one 5k having the 3 position unsub-
stituted (R2 =H), which involved the thermolytic decomposi-
tion of 2,2,6-trimethyl-4H-1,3-dioxin-4-one (1h) for the genera-
tion of the required monosubstituted a-oxoketene intermedi-
ate.[2] In a control experiment, the irradiation with microwaves
of a toluene solution of the diazo compound 1e (R1, R2 =
-CH2C(CH3)2CH2-) and the phosphorous ylide 2d for 15 minutes
at 80 or 1308C (which are temperatures that do not or only
slowly promote the Wolff rearrangement of 1e) left the start-
ing materials essentially unchanged, showing that the ylide is
inert toward the diazo compound at these temperatures. Over-
all, a simple and rapid synthetic method to prepare diversely
substituted and functionalized 4H-pyran-4-ones has been dis-
covered, revealing that the abnormal Wittig olefination is not
just a laboratory oddity[6] but a general reaction. It nicely com-
plements existing methods to synthesize g-pyrones.[7,8] Nota-
bly, ynol ethers have in the past been reported to undergo in-
verse-demand Diels–Alder cycloaddition with a-oxoketenes to
regioselectively afford g-pyrones of type 5 having an alkoxide
R3 substituent.[8c,d] As to the complementary method reported
herein, the Wittig ylides 2a,d–h may be regarded as synthetic
equivalents of terminal alkynes capable of undergoing regiose-
lective [4+2] cycloaddition reactions with a-oxoketenes
(Scheme 3).
Scheme 1. Synthesis of a,a’-bisoxoallenes 4 by normal Wittig olefination of
a-oxoketenes 3.
Stimulated and intrigued by these early results, a series of
representative in situ generated a-oxoketenes were allowed to
react with a selection of carbonyl-stabilized Wittig ylides. It
was actually found that a-oxoketenes generally react with
ketone-stabilized, and to some extent aldehyde-stabilized,
ylides 2 via abnormal Wittig olefinations rather than regular
Wittig olefinations, which resulted in the synthesis of polysub-
stituted and functionalized monocyclic and fused bicyclic g-py-
rones 5a–k in fair to good yields (Scheme 2). The a-oxoketene
Scheme 3. Carbonyl-stabilized Wittig ylides as synthetic equivalents of al-
kynes.
In sharp contrast to the above results, the combination of
the a-oxoketene derived from the diazo 1e and the ester-sta-
bilized Wittig ylide 2c did not produce the a,a’-bisoxoallene
4g or the g-pyrone 5m, but the 4H-pyranylidene 6a in 43%
yield together with a significant proportion of the a-oxoketene
dimer [Scheme 4, Eq. (a)]. Because two molecules of ylides
were clearly incorporated in the product 6a, the same reaction
was attempted with two equivalents of the ylide, which afford-
ed this time the 4H-pyranylidene 6a in 89% yield. This pseudo
three-component reaction could be generalized, allowing for
the expeditious synthesis of 4H-pyranylidenes 6a–f [Scheme 4,
Scheme 2. Synthesis of g-pyrones 5 by abnormal Wittig olefination of a-oxo-
ketenes 3. [a] The reaction also afforded the products 6 f (22%) and iso-6 f
(4%), see text and Scheme 4. [b] Obtained from 2,2,6-trimethyl-4H-1,3-
dioxin-4-one (1h) instead of 2-diazo-3-oxobutanal.
&
&
Chem. Eur. J. 2018, 24, 1 – 10
www.chemeurj.org
2
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
To identify a plausible reaction pathway accounting for the
formation of 4H-pyranylidenes 6a–f, two control experiments
were performed. First, a 1:1 mixture of 4H-pyran-4-one 5e and
Wittig ylide 2e in toluene was heated at 1708C for 15 min
[Scheme 5, Eq. (a)]. Both starting materials were nearly quanti-
Scheme 5. Control experiments.
tatively recovered, and no formation of the 4H-pyranylidenes
6 f or iso-6 f could be evidenced, ruling out the possibility for
a Wittig reaction between the ketone carbonyl group in 4H-
pyran-4-one 5e and phosphorous ylide 2e.[11] Second, a 1:1
mixture of allene 4e and Wittig ylide 2c in toluene was heated
at 1708C for 15 minutes [Scheme 5, Eq. (b)]. Pleasingly, the
analysis of the product reaction mixture evidenced the clean
and efficient formation of 4H-pyranylidene 6e, pointing to
allene 4e as a key intermediate capable of undergoing an ab-
normal Wittig olefination step in the cascade transformation
1a+2c!6e.
Based on the above experimental data, it was enticing to
postulate some empirical rules governing the fate of the reac-
tions of carbonyl-stabilized phosphorous ylides with a-oxoke-
tenes: 1) a,a’-bisoxoallenes of type 4 may be obtained by stan-
dard Wittig olefinations of acyclic-only a-oxoketenes;
2) ketone-stabilized Wittig ylides generally react with a-oxoke-
tenes to give g-pyrones of type 5 through abnormal Wittig ole-
finations; 3) ester-stabilized ylides react with a-oxoketenes by
a pseudo three-component Wittig/abnormal Wittig cascade
olefination to afford 4H-pyranylidenes of type 6 via the corre-
sponding a,a’-bisoxoallene intermediates 4; and 4) the phenyl
ketone-stabilized ylide 2e is a border case. However, the actual
reasons dictating the divergent reactivities observed in these
series warranted further research. With the intention to eluci-
date the mechanism of the abnormal Wittig olefination and to
identify the origin of the dichotomy in the reactions of carbon-
yl-stabilized Wittig ylides 2 with a-oxoketenes 3, we embarked
on a detailed mechanistic computational study using DFT
methods. This part of the study revealed several unusual fea-
tures for Wittig olefinations. As a short background on the
classical Wittig olefination, it is now generally accepted that
the lithium salt-free Wittig olefination of a carbonyl compound
with a carbonyl-stabilized phosphorous ylide is a three-step
process initiated by an irreversible regio- and stereoselective
[2+2] cycloaddition between the C=O bond of the carbonyl
compound and the C=P bond of the ylide, leading to an oxa-
phosphetane intermediate.[12] Given that bond-forming and
Scheme 4. Synthesis of 4H-pyranylidenes 6 by a Wittig/abnormal Wittig cas-
cade olefination of a-oxoketenes.
Eq. (b)]. In the case of product 6 f, a minor amount of the dia-
stereomer iso-6 f could also be isolated in 7% yield. The struc-
tures of 6e and 6 f were secured by X-ray diffraction analyses.
Notably, 4H-pyranylidenes of type 6 are donor-acceptors p-
conjugated molecules with potential applications as nonlinear
optical chromophores.[9] The 4H-pyranylidenes 6a–e, having a
2-methoxyl substituent, are relatively stable compounds show-
ing no significant decomposition when stored neat in closed
reaction vessels at ꢀ188C for several months. However, 4H-pyr-
anylidene 6c could be cleanly converted into the correspond-
ing 2H-pyran-2-one 7c upon prolonged exposure to air at
room temperature [Scheme 4, Eq. (c)]. Similarly, 4H-pyranyli-
dene 6e could be rapidly converted into 2H-pyran-2-one 7e in
the presence of aqueous HCl [Scheme 4, Eq. (c)]. These internal
redox transformations are supposedly promoted by water. By
extension, the synthetic approach to 4H-pyranylidenes 6a–e
depicted in Scheme 4 also constitutes a practical method to
prepare the corresponding substituted 2H-pyran-2-ones, which
is another important class of synthetically and biologically rele-
vant molecules.[10]
Chem. Eur. J. 2018, 24, 1 – 10
www.chemeurj.org
3
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
-breaking events at pentavalent bipyramidal phosphorus
atoms preferentially occur from apical positions, the oxaphos-
phetane intermediate then requires a low-energy-demanding
geometrical reorganization at the phosphorus atom, the so-
called Berry pseudorotations,[13] before its decomposition by a
retro-[2+2] cycloaddition affording the olefin product and tri-
phenylphosphine oxide. To get a realistic model for the Wittig
olefinations described herein, calculations were performed at
the B3LYP/6-311+ +G(d,p)//B3LYP/6-31+G(d) level of theory
including a continuum description of solvation effects, and, im-
portantly, with the full PPh3 ylides to account for all the steric
and electronic effects.[14] Both the normal and abnormal Wittig
olefinations of the a-oxoketene 3d having a blocked s-cis con-
formation were comparatively examined with the ketone-stabi-
lized ylide 2d and the ester-stabilized ylide 2c (Figures 1–5).
Importantly, all reactions were found to be initiated by the for-
mation of a covalent bond between the nucleophilic ylidic
carbon atom and the electrophilic ketene carbon atom to pro-
duce, respectively, the stabilized betaine intermediates I and V
with a reasonable barrier (Figure 1). All efforts directed at the
identification of alternative processes excluding the intermedi-
ary of the betaines I or V remained fruitless. Significantly, be-
taine intermediates were previously demonstrated not to form
in the course of standard Lewis acid-free Wittig olefinations,[12]
and on that matter the reactions described herein are peculiar.
The strain release resulting from the sp to sp2 rehybridization
at the addition carbon atom together with the formation of p-
extended zwitterionic ketoenolates leading to six-membered
chelates by hydrogen bonding certainly contribute to the sta-
bilization of betaines I and V.
Figure 1. Reactions of a-oxoketene 2a with acetyl-stabilized Wittig ylide 3a
and methyl ester-stabilized ylide 3h to produce betaines I and V, respective-
ly. All energies are Gibbs free energies expressed in kJmolꢀ1
.
For the synthesis of the g-pyrone 5a (Figure 2), betaine I
adopts the geometry Ibis suitable for the formation of the tricy-
clic oxaphosphetane intermediate II by a hemiacetalization/ox-
aphosphetane formation cascade. In oxaphosphetane II, the
oxygen atom of the former acetyl stabilizing group of the ylide
is in an apical position (with a 11.88 deviation from alignment)
at the phosphorus atom that adopts a trigonal bipyramid ge-
ometry. A geometrical reorganization at the pentavalent phos-
phorous atom (Berry pseudorotation) is then necessary to
obtain the oxaphosphetane intermediate IIbis, having this time
the former ylidic carbon atom in an apical position (with a
21.58 deviation from alignment). This conformational change
Figure 2. Energy profile of the reaction leading to g-pyrone 5a. All energies are Gibbs free energies expressed in kJmolꢀ1
.
&
&
Chem. Eur. J. 2018, 24, 1 – 10
www.chemeurj.org
4
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Figure 3. Energy profile of the reaction leading to allene 4h (not observed experimentally). All energies are Gibbs free energies expressed in kJmolꢀ1
.
was determined as the rate-limiting step of the overall trans-
formation, with an activation energy of +119.1 kJmolꢀ1 rela-
tive to the energy of the stabilized betaine intermediate I. It is
truly surprising that a simple Berry pseudorotation process is
the actual rate-limiting step of the reaction leading to 5a. Fi-
nally, decomposition of the oxaphosphetane IIbis occurs with a
low barrier (5.8 kJmolꢀ1) by a very asynchronous and exother-
mic retro-[2+2] cycloaddition with the rupture of the PꢀC
bond largely in advance to the one of the OꢀC bond, to give
the g-pyrone product 5a and triphenylphosphine oxide.
Overall, these calculations indicate that the g-pyrone abnor-
mal Wittig product 5a is the thermodynamic product by
91.1 kJmolꢀ1, and that the allene regular Wittig product 4h is
the kinetic product by only 14.6 kJmolꢀ1 [16]
To experimentally
.
demonstrate possible reversibility in the abnormal Wittig olefi-
nation leading to the g-pyrone 5l (Scheme 1), a deuterated
benzene solution of allene 4a was heated 30 minutes at 2008C
in the presence of triphenylphosphine oxide with an internal
1
standard (Scheme 6). The H NMR analysis of the resulting mix-
For the concurrent synthesis of the a,a’-bisoxoallene 4h
from the same substrates, which is not observed experimental-
ly, it was found that the betaine intermediate I can cyclize to
the oxaphosphetane intermediate III, having the former ylidic
carbon atom in an apical position (Figure 3). The expected oxa-
phosphetane intermediate, having the oxygen atom of the
former ketene carbonyl group in an apical position, is actually
formed as a shoulder on the raising energy profile from I, and
requires no significant activation energy (see the Supporting
Information for details). Very surprisingly, and in sharp contrast
to the well-established oxaphosphetane decomposition pro-
cess by asynchronous retro-[2+2] cycloaddition (e.g. IIbis!5a
in Figure 2), the only reaction path we could identify for the
decomposition of III is a two-step process. Indeed, oxaphos-
phetane III was found to evolve by rupture of the CꢀP bond
and some conformational change to afford the unusual be-
taine intermediate IV.[15] Intermediate IV, which shows both a
covalent OꢀP bond (dO-P =1.66 ꢁ) and an electrostatic Oꢀ/P+
stabilizing interaction (dO-P =1.91 ꢁ), may be regarded as a
Lewis pair adduct between triphenylphosphine oxide and the
a,a’-bisoxoallene 4h (with dC-O =1.40 ꢁ) stabilized by an elec-
tron retro-donation from the negatively charged oxygen atom
of the acetyl group in the apical position (with a 4.68 deviation
from alignment) to the electron-demanding phosphorus atom.
The separation of this Lewis pair can occur with a small barrier
(+37.8 kJmolꢀ1) to afford the a,a’-bisoxoallene 4h and triphe-
nylphosphine oxide.
Scheme 6. Experimental proof for the possible retro-Wittig process leading
to 5l.
ture revealed the formation of 8% of the expected g-pyrone
5l together with the presence of 67% unreacted 4a and some
unidentified products. A similar reaction performed without tri-
phenylphosphine oxide showed no significant reaction of 4a.
Although poorly efficient, the rearrangement of 4a into 5l in
the presence of triphenylphosphine oxide, presumably involv-
ing the betaine I’, points out the existence of a kind of retro-
Wittig process with a C=O bond being created from a C=C
bond and triphenylphosphine oxide.
We next examined computationally the reactions of the
model a-oxoketene 3d with the ester-stabilized Wittig ylide
2c. Paralleling the findings shown in Figure 2, betaine V was
found to evolve to the g-pyrone product 5n by a cascade
double cyclization, affording consecutively the two oxaphos-
phetane intermediates VI and VIbis, the decomposition of
which by
a retro-[2+2] cycloaddition gives the product
(Figure 4). Overall, the energy profiles depicted in Figure 2 and
Chem. Eur. J. 2018, 24, 1 – 10
www.chemeurj.org
5
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Figure 4. Energy profile of the reaction leading to g-pyrone 5n. All energies are Gibbs free energies expressed in kJmolꢀ1
.
Figure 4 are, as expected, very comparable; however, with a
significant difference: the rate limiting step lies
+119.1 kJmolꢀ1 above the energy of betaine I in Figure 2
whereas it is now calculated at +153.1 kJmolꢀ1 above the
energy of betaine V in Figure 4. In other words, in the ketone-
stabilized Wittig ylide series, the conversion of betaine I into g-
pyrone 5a can proceed at a rapid rate under the reaction con-
ditions, whereas under similar conditions in the ester-stabilized
Wittig ylide series the betaine V can only very slowly convert
into g-pyrone 5n. This difference in reactivity actually reflects
the lower reactivity of ester versus ketone carbonyl groups in
anionic cascade processes.[17]
shift of the triphenylphosphonium group to give intermediate
VII directly via a near planar oxaphosphetane transition state.
The latter actually corresponds to the rupture of the CꢀP bond
in an elusive oxaphosphetane intermediate having the former
ylidic carbon atom in an apical position at phosphorous (i.e.,
the analogue of III with a methyl ester group in place of the
acetyl group, see the Supporting Information for details). As
before, intermediate VII may be regarded as a Lewis pair
adduct between triphenylphosphine oxide and the a,a’-bis-
oxoallene 4i stabilized by an electron retro-donation from the
negatively charged oxygen atom of the carbonyl in the methyl
ester group in the apical position (with a 6.38 deviation from
alignment) to the electron demanding phosphorous atom.
However, this electrostatic Oꢀ/P+ interaction is much weaker
than above (dO-P =3.00 ꢁ), resulting in a less stabilized Lewis
pair (compare the relative stabilities of IV and VII).
Finally, we examined the concurrent pathway leading to
a,a’-bisoxoallene 4i from betaine V (Figure 5). In this case, it
was found that betaine V can readily undergo the C!O 1,3-
This is easily rationalized by the preferred delocaliza-
tion of the negative charge in VII on the distal
ketone carbonyl oxygen atom rather than the
methyl ester one. As above, the separation of the
Lewis pair VII occurs with
a small barrier (+
29.4 kJmolꢀ1) to afford a,a’-bisoxoallene product 4i
together with triphenylphosphine oxide. The rate-
limiting step of the overall transformation is the 1,3-
shift of the triphenyphosphonium group, with an ac-
tivation energy of +101.5 kJmolꢀ1 relative to the
energy of the betaine V. From this series of calcula-
tions, it was concluded that the preferential forma-
tion of a,a’-bisoxoallene 4i over g-pyrone 5n in the
reaction between a-oxoketene 2a and methyl ester-
stabilized Wittig ylide 3h is under sharp kinetic con-
trol, with the formation of the allene being ca.
106 times faster than the formation of the pyrone.
The above model study shows that ester-stabi-
lized Wittig ylides react with a-oxoketenes to afford
preferentially the corresponding a,a’-bisoxoallenes 4
Figure 5. Energy profile of the reaction leading to allene 4i. All energies are Gibbs free
energies expressed in kJmolꢀ1
.
&
&
Chem. Eur. J. 2018, 24, 1 – 10
www.chemeurj.org
6
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
by a normal Wittig olefination. In practice, however, and with
the exception of the reactions in the acyclic series presented in
Scheme 1, these reactions afforded the corresponding 4H-pyra-
nylidenes 6a–f by a pseudo three-component reaction involv-
ing two equivalents of ylide (Scheme 4). Given that the a,a’-bi-
soxoallene 4e has been shown to be an intermediate in the
synthesis of 4H-pyranylidene 6e [Scheme 5, Eq. (b)], we pro-
pose that 4H-pyranylidenes 6a–f are obtained by a normal
Wittig olefination/regioselective abnormal Wittig olefination
cascade. The a,a’-bisoxoallene 4 initially obtained by normal
Wittig olefination of a-oxoketene 3 with ylide 2 would react
with a second equivalent of ylide 2 to form betaine A
(Scheme 7). This betaine would then undergo the regioselec-
tive hemiacetalization/oxaphosphetane formation cascade dis-
cussed herein for the abnormal Wittig olefination to afford the
oxaphosphetane intermediate B, the regioselectivity of the cyc-
lization being dictated by the relative nucleophilicities of the
distal oxygen atoms (ketone enolate vs. ester enolate). Finally,
decomposition of B would give the 4H-pyranylidene product 6
and triphenylphosphine oxide.
diazo compounds 1a,c predominantly exist in their s-trans con-
formations, which here again slows down their abnormal
Wittig reactivity to the benefit of the normal Wittig reactivity,
allowing for the isolation of a,a’-bisoxoallenes 4a,d,f
(Scheme 1).
Conclusions
The reactions of a-oxoketenes, prepared in situ by a thermal
Wolff rearrangement of 2-diazo-1,3-dicarbonyl compounds,
with carbonyl-stabilized Wittig ylides have been studied in
detail. It was found that a-oxoketenes generally react with
ketone- and aldehyde-stabilized Wittig ylides as 1,4-ambident
C-electrophilic/O-nucleophilic reaction partners via an abnor-
mal Wittig olefination, unlocking the synthesis of 4H-pyran-4-
one (g-pyrone) derivatives with original substitution patterns.
This reaction presumably involves some degree of thermody-
namic control when compared to the normal Wittig olefination
of the a-oxoketenes, and some reversibility in the normal
Wittig olefination was demonstrated: a premiere! In sharp con-
trast, ester-stabilized Wittig ylides were found to react with a-
oxoketenes through a normal Wittig olefination, affording a,a’-
bisoxoallenes. This reaction was shown to be under kinetic
control when compared to the corresponding abnormal Wittig
olefination. The actual reasons at the origin of the divergent
reactivities observed for ketone-stabilized ylides on the one
hand and ester-stabilized ylides on the other hand were inves-
tigated, notably through
a comprehensive computational
mechanistic study, and were ultimately found to reside in the
higher electrophilicity of ketone carbonyl groups versus ester
ones. In most cases, the a,a’-bisoxoallene products were found
to be reactive under the reaction conditions and capable, in
turn, of undergoing an abnormal Wittig olefination with a
second equivalent of Wittig ylide, leading to 4H-pyranylidenes
by a pseudo three-component reaction involving a Wolff/
Wittig/abnormal Wittig cascade sequence. Until now, the ab-
normal Wittig olefination was a laboratory oddity and the pres-
ent work is the first report on its generalization, rationalization,
and concrete synthetic applications. Overall, the chemical
transformations described herein, involving the two venerable
and thoroughly explored Wolff rearrangement and Wittig olefi-
nation, carry a significant improvement to the current know-
how for the selective synthesis of functionalized 4H-pyran, and,
to some extent, 2H-pyran derivatives. In a broader meaning,
the abnormal Wittig olefination of a,b-unsaturated carbonyl
compounds with carbonyl-stabilized Wittig ylides can now be
added to the arsenal of synthetic methods to prepare 4H-
pyran derivatives. From a fundamental point of view, and con-
trary to the standard Lewis acid-free Wittig olefination of car-
bonyl compounds with carbonyl-stabilized phosphorous ylides,
the Wittig and abnormal Wittig reactions described herein are
likely to proceed via betaine intermediates.
Scheme 7. Mechanistic hypothesis for the formation of 4H-pyranylidenes 6
via a Wittig/abnormal Wittig cascade reaction.
As shown in the example in Equation (a), Scheme 4, the
a,a’-bisoxoallene 4g, having a blocked s-cis conformation of
the a-oxoallene moiety, reacts faster with the ester-stabilized
ylide 2c than the s-cis a-oxoketene derived from the diazo 1e,
precluding the isolation of a,a’-bisoxoallene 4g. However, this
situation is reverted in the case of the a,a’-bisoxoallenes
4b,c,e, capable of adopting a s-trans conformation of the a-
oxoallene moieties (Scheme 1). As demonstrated by the model
study, the abnormal Wittig olefination can only be operative
with a,b-unsaturated carbonyl compounds having an s-cis con-
formation. In the cases of a,a’-bisoxoallenes 4b,c,e, the fa-
vored s-trans conformations of their a-oxoallene moieties are
unproductive, slowing down their participation in abnormal
Wittig processes and allowing their isolation under some con-
ditions. Similarly, the a-oxoketenes derived from the acyclic
Experimental Section
General procedure for the synthesis of a,a’-bisoxoallenes 4a–f,
4H-pyran-4-ones 5a–l, and 4H-pyranylidenes 6a–f: A 10 mL seal-
Chem. Eur. J. 2018, 24, 1 – 10
www.chemeurj.org
7
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Beilstein J. Org. Chem. 2005, 1, 5; k) C.-Y. Li, X.-B. Wang, X.-L. Sun, Y.
Tang, J.-C. Zheng, Z.-H. Xu, Y.-G. Zhou, L.-X. Dai, J. Am. Chem. Soc. 2007,
129, 1494–1495.
able oven-dried tubular reaction vessel was charged with diazo
compound 1 (ca. 0.2–1.0 mmol), phosphorous ylide 2 (1.0 or
2.0 equiv), and dry toluene (2 mL). The resulting heterogeneous
mixture was subjected to microwave irradiation to reach 1708C as
fast as possible and held at that temperature for 15 minutes, after
which the reaction mixture was cooled to 558C with an air flow.
The resulting solution was concentrated under vacuum and direct-
ly purified by silica gel flash chromatography to afford the pure
product.
[5] For representative syntheses, see: a) T. Yoshino, F. Ng, S. J. Danishefsky,
J. Am. Chem. Soc. 2006, 128, 14185–14191; b) T. Boisse, B. Rigo, R.
Millet, J.-P. Hꢂnichart, Tetrahedron 2007, 63, 10511–10520; c) M. A.
Calter, N. Li, Org. Lett. 2011, 13, 3686–3689; d) I. Mizota, Y. Matsuda, S.
Kamimura, H. Tanaka, M. Shimizu, Org. Lett. 2013, 15, 4206–4209.
[6] For the previous reports describing isolated examples of an “abnormal
Wittig olefination”, see: a) H. Strzelecka, M. Simalty-Siemiatycki, C. Prꢂ-
vost, C. R. Hebd. Sꢀances Acad. Sci. 1963, 257, 926–928; b) M. von
Strandtmann, M. P. Cohen, C. Puchalski, J. Shavel, Jr., J. Org. Chem.
1968, 33, 4306–4309; c) W. M. Abdou, N. A. F. Ganoub, Heteroat. Chem.
1992, 3, 133–137; d) W. M. Abdou, E.-S. M. A. Yakout, N. A. F. Ganoub,
Tetrahedron 1995, 51, 11411–11420; e) F. Ramirez, O. P. Madan, C. P.
Smith, J. Org. Chem. 1965, 30, 2284–2290; f) F. Ramirez, O. P. Madan,
S. R. Heller, J. Am. Chem. Soc. 1965, 87, 731–734; g) P. Crabbꢂ, E. Dꢃaz, J.
Haro, G. Pꢂrez, D. Salgado, E. Santos, Tetrahedron Lett. 1970, 5069–
5072. In regards of the non-classical Wittig olefination, which corre-
sponds to the olefination of carbonyl compounds other than ketones
and aldehydes (e.g., esters, amides), we propose to name this transfor-
mation “the abnormal Wittig olefination” instead of ”the internal Wittig
reaction” as previously tentatively denominated.
CCDC 1550800(5a), 1550801 (6e) and 1550802 (6 f) contain the
supplementary crystallographic data for this paper. These data can
be obtained free of charge from The Cambridge Crystallographic
Data Centre.
Acknowledgements
We thank Dr. Michel Giorgi (CNRS) for the X-ray structural anal-
yses, Dr. Michel Rajzmann (Aix-Marseille Universitꢂ) for assis-
tance in computational chemistry, and the Centre Rꢂgional de
Compꢂtences en Modꢂlisation Molꢂculaire (Aix-Marseille Uni-
versitꢂ) for computing facilities. Financial support from the
Agence Nationale de la Recherche (ANR-13-JS07-0002-01), Aix-
Marseille Universitꢂ and the Centre National de la Recherche
Scientifique (CNRS) is gratefully acknowledged.
[7] Reviews: a) A. K. Miller, D. Trauner, Synlett 2006, 2295–2316; b) W. Wilk,
H. Waldmann, M. Kaiser, Bioorg. Med. Chem. 2009, 17, 2304–2309; c) P.
Sharma, K. J. Powell, J. Burnley, A. S. Awaad, J. E. Moses, Synthesis 2011,
2865–2892. For recent total syntheses: d) M. Henrot, A. Jean, P. A. Peix-
oto, J. Maddaluno, M. De Paolis, J. Org. Chem. 2016, 81, 5190–5201;
e) L. Hoffmeister, T. Fukada, G. Pototschnig, A. Fꢄrstner, Chem. Eur. J.
2015, 21, 4529–4533; f) J. C. Henrikson, T. K. Ellis, J. B. King, R. H. Ciche-
wicz, J. Nat. Prod. 2011, 74, 1959–1964; g) V. Jeso, C. Yang, M. D. Camer-
on, J. L. Cleveland, G. C. Micalizio, ACS Chem. Biol. 2013, 8, 1241–1252.
[8] For approaches to di- or trisubstituted g-pyrones involving inverse elec-
tron demand hetero Diels–Alder reactions with a-oxoketenes, see: a) G.
Jꢅger, J. Wenzelburger, Liebigs Ann. Chem. 1976, 1689–1712; b) E. Wen-
kert, T. P. Ananthanarayan, V. F. Ferreira, M. G. Hoffmann, H.-S. Kim, J.
Org. Chem. 1990, 55, 4975–4976; c) G. Himbert, L. Henn, Liebigs Ann.
Chem. 1987, 771–776; d) A. Stadler, K. Zangger, F. Belajb, G. Kollenz,
Tetrahedron 2001, 57, 6757–6763. For an approach to a’-methoxy-g-py-
rones based on a desymmetrization strategy, see: e) H. Rosso, M. De Pa-
olis, V. C. Collin, S. Dey, S. M. Hecht, C. Prandi, V. Richard, J. Maddaluno,
J. Org. Chem. 2011, 76, 9429–9437. For approaches based on the cycli-
zation of ynone derivatives, see: f) Y.-F. Qiu, F. Yang, Z.-H. Qiu, M.-J.
Zhong, L.-J. Wang, Y.-Y. Ye, B. Song, Y.-M. Liang, J. Org. Chem. 2013, 78,
12018–12028; g) S. Dong, C. Fang, W. Tang, T. Lu, D. Du, Org. Lett.
2016, 18, 3882–3885. Selected miscellaneous approaches: h) D. L. Oby-
dennov, G.-V. Rçschenthaler, V. Sosnovskikh, Tetrahedron Lett. 2013, 54,
6545–6548; i) D. Hobuß, S. Laschat, A. Baro, Synlett 2005, 123–124; j) J.
Morris, G. P. Luke, D. G. Wishka, J. Org. Chem. 1996, 61, 3218–3220;
k) C. A. B. Rodrigues, A. Misale, F. Schiel, N. Maulide, Org. Biomol. Chem.
2017, 15, 680–683; l) K. Onda, I. Hayakawa, A. Sakakura, Synlett 2017,
28, 1596–1600.
Conflict of interest
The authors declare no conflict of interest.
Keywords: density functional calculations
·
microwave
chemistry · reactive intermediates · synthetic methods · Wittig
reactions
[1] Reviews: a) C. Wentrup, W. Heilmayer, G. Kollenz, Synthesis 1994, 1219–
1248; b) G. Kollenz, S. Ebner in Science of Synthesis: Houben-Weyl Meth-
ods of Molecular Transformations, Vol. 23 (Ed.: R. Danheiser), Georg
Thieme, Stuttgart, 2006, pp. 271–349; c) K. P. Reber, S. D. Tilley, E. J. Sor-
ensen, Chem. Soc. Rev. 2009, 38, 3022–3034; d) A. Ford, H. Miel, A. Ring,
C. N. Slattery, A. R. Maguire, M. A. McKervey, Chem. Rev. 2015, 115,
9981–10080. For representative recent work: e) M. Presset, Y. Coquerel,
J. Rodriguez, J. Org. Chem. 2009, 74, 415–418; f) M. Presset, Y. Coquerel,
J. Rodriguez, Org. Lett. 2010, 12, 4212–4215; g) M. Presset, K. Mohanan,
M. Hamann, Y. Coquerel, J. Rodriguez, Org. Lett. 2011, 13, 4124–4127;
h) J. Galvez, J.-C. Castillo, J. Quiroga, M. Rajzmann, J. Rodriguez, Y. Co-
querel, Org. Lett. 2014, 16, 4126–4129; i) P. Neupane, L. Xia, Y. R. Lee,
Adv. Synth. Catal. 2014, 356, 2566–2574; j) Y. Dudognon, M. Presset, J.
Rodriguez, Y. Coquerel, X. Bugaut, T. Constantieux, Chem. Commun.
2016, 52, 3010–3013.
[2] a) J. A. Hyatt, P. L. Feldman, R. J. Clemens, J. Org. Chem. 1984, 49, 5105–
5108; b) R. Cookson, T. N. Barrett, A. G. M. Barrett, Acc. Chem. Res. 2015,
48, 628–642.
[3] Reviews: a) W. Kirmse, Eur. J. Org. Chem. 2002, 2193–2256; b) Y. Coquer-
el, J. Rodriguez in Molecular Rearrangements in Organic Synthesis (Ed.: C.
Rojas), Wiley, Hoboken, 2015, Chap. 3, pp. 59–84.
[4] a) G. Wittig, A. Haag, Chem. Ber. 1963, 96, 1535–1543; b) H.-J. Best-
mann, H. Hartung, Angew. Chem. 1963, 2, 214; c) H.-J. Bestmann, H. Har-
tung, Chem. Ber. 1966, 99, 1198–1207; d) G. Aksnes, P. Frøyen, Acta
Chem. Scand. 1968, 22, 2347–2352; e) Z. Hamlet, W. D. Barker, Synthesis
1970, 543–544; f) R. A. Ruden, J. Org. Chem. 1974, 39, 3607–3608; g) R.
Gompper, U. Wolf, Liebigs Ann. Chem. 1979, 1388–1405; h) R. W. Lang,
H.-J. Hansen, Helv. Chim. Acta 1980, 63, 438–455; i) K. Diehl, G. Himbert,
L. Henn, Chem. Ber. 1986, 119, 2430–2443; j) S. P. Marsden, P. C. Ducept,
[9] A. B. Marco, P. M. Burrezo, L. Mosteo, S. Franco, J. Garꢃn, J. Orduna, B. Di-
osdado, B. E. Villacampa, J. T. L. Navarrete, J. Casado, R. Andreu, RSC
Adv. 2015, 5, 231–242, and references therein.
[10] Review: R. Pratapand, V. J. Ram, Tetrahedron 2017, 73, 2529–2590.
[11] For examples of Wittig-type olefinations of 4H-pyran-4-one derivatives,
see: a) K. Minami, T. Shirahata, M. Yohji, Chem. Lett. 2013, 42, 1548–
1550; b) E. Azuma, K. Kuramochi, K. Tsubaki, Tetrahedron 2013, 69,
1694–1699.
[12] a) R. Robiette, J. Richardson, V. K. Aggarwal, J. N. Harvey, J. Am. Chem.
Soc. 2006, 128, 2394–2409. Review on the Wittig olefination mecha-
nism: b) P. A. Byrne, D. G. Gilheany, Chem. Soc. Rev. 2013, 42, 6670–
6696.
[13] E. P. A. Couzijn, J. C. Slootweg, A. W. Ehlers, K. Lammertsma, J. Am.
Chem. Soc. 2010, 132, 18127–18140.
[14] See Supporting Information for details. It may be noted that the pres-
ent study is the first realistic computational mechanistic study on the
Wittig reaction with ketenes.
&
&
Chem. Eur. J. 2018, 24, 1 – 10
www.chemeurj.org
8
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
[15] Related dioxatriphenylphosphoranes have been isolated in the past,
see for example: E. L. Clennan, P. C. Heah, J. Org. Chem. 1982, 47, 3329–
3331.
b) D. Bonne, T. Constantieux, Y. Coquerel, J. Rodriguez, Org. Biomol.
Chem. 2012, 10, 3969–3973; c) D. Bonne, T. Constantieux, Y. Coquerel, J.
Rodriguez, Chem. Eur. J. 2013, 19, 2218–2231; d) D. Bonne, Y. Coquerel,
T. Constantieux, J. Rodriguez, Tetrahedron: Asymmetry 2010, 21, 1085–
1109; e) D. Bonne, T. Constantieux, Y. Coquerel, J. Rodriguez in Stereose-
lective Organocatalysis: Bond Formation Methodologies and Activation
Modes (Ed.: R. R. Torres), Wiley, Hoboken, 2013, Chap. 16, pp. 559–585.
[16] From this model study, the calculated half-life of 4h was estimated at
t1/2 =2.5 h (with the pre-exponential factor of the Eyring equation as-
sumed to be 1), which appears over-estimated. Also, an overall barrier
of 159.7 kJmolꢀ1 would be involved in the reaction 4h!5a, which also
appears over-estimated. Given the large approximations of the model
employed, relative errors of up to ꢁ40 kJmolꢀ1 are undoubtedly possi-
ble. For an excellent discussion on the accuracy of computational
models in the case of polar reactions in solution such as the Wittig ole-
fination, see: J. N. Harvey, Faraday Discuss. 2010, 145, 487–505.
Manuscript received: March 27, 2018
Revised manuscript received: May 7, 2018
Version of record online: && &&, 0000
[17] Selected reviews from our laboratory: a) X. Bugaut, D. Bonne, Y. Coquer-
el, J. Rodriguez, T. Constantieux, Curr. Org. Chem. 2013, 17, 1920–1928;
Chem. Eur. J. 2018, 24, 1 – 10
www.chemeurj.org
9
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
FULL PAPER
&
Synthetic Methods
D. Pierrot, M. Presset, J. Rodriguez,
D. Bonne,* Y. Coquerel*
&& – &&
Normal, Abnormal, and Cascade
Wittig Olefinations of a-Oxoketenes
Abnormal Wittig: The reactions of a-
oxoketenes with carbonyl-stabilized
Wittig ylides have been studied in
detail, which has led to the practical
syntheses of functionalized allenes, py-
rones, and pyranylidenes (see scheme).
An abnormal Wittig olefination was re-
vealed.
&
&
Chem. Eur. J. 2018, 24, 1 – 10
www.chemeurj.org
10
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!