A re-evaluation of the electrophilic substitution reactions of the Ramirez ylide

Article abstract of DOI:10.1021/jo7014104

(Chemical Equation Presented) Cyclopentadienylidenetriphenylphosphorane (the Ramirez ylide), unexpectedly and contrary to a number of earlier reports, has been shown to be like pyrrole in undergoing electrophilic substitution on the cyclopentadienide ring

Full text of DOI:10.1021/jo7014104

A Re-Evaluation of the Electrophilic Substitution Reactions of the
Ramirez Ylide
Lee J. Higham,*,† Jimmy Muldoon,§ P. Gabriel Kelly,‡ David M. Corr,§
Helge Mu¨ller-Bunz,§ and Declan G. Gilheany*,§
School of Chemistry and Chemical Biology, Centre for Synthesis and Chemical Biology, Conway Institute
for Biomolecular and Biomedical Research, UniVersity College Dublin, Belfield, Dublin 4, Ireland,
Chemistry Department, National UniVersity of Ireland, Maynooth, Co. Kildare, Ireland, and School of
Natural Sciences-Chemistry, Bedson Building, UniVersity of Newcastle,
Newcastle upon Tyne, NE1 7RU, UK
lee.higham@ncl.ac.uk; declan.gilheany@ucd.ie
ReceiVed July 8, 2007
Cyclopentadienylidenetriphenylphosphorane (the Ramirez ylide), unexpectedly and contrary to a number
of earlier reports, has been shown to be like pyrrole in undergoing electrophilic substitution on the
cyclopentadienide ring at either the 3- or the 2-position, depending on the electrophile. Formylation
under Vilsmeier conditions and addition of tetracyanoethylene occurs at the 3-position, while activated
acetylenes and the nitrosyl electrophile substitute at the 2 position. The 3-formylated product was reduced
to the 3-methyl derivative and it also reacted under Knoevenagel conditions to give a number of novel
condensation products. The results of single-crystal X-ray crystallographic analyses are given for four of
the compounds studied, and a careful 2D NMR analysis of all of the compounds was performed in order
to develop a reliable method for the unambiguous assignment of the regiochemistry of adduct formation.
Introduction
that 3 could be used as a source of further substitution on the
azulene, for example, by its reduction to a methyl group.
However, azulene formation was not observed when adducts
that were synthesized from the formylated product of 1 were
reacted with a further equivalent of alkyne. This, therefore, cast
doubt on the original structural assignment of 3. We now report
that 1 reacts with certain electrophiles to give products
substituted instead at the 3-position. Derivatives prepared from
one such product, triphenylphosphonium 3-(formyl)cyclopen-
tadienylide, are also presented, as well as the characterization
of a number of the key compounds by X-ray crystallography.
An NMR method for the correct structural assignment of both
2- and 3-substituted compounds in the absence of X-ray
crystallography is also demonstrated.
We have recently reported1 that 1,3,5,6-tetrasubstituted azu-
lenes can be prepared by sequential reaction of the Ramirez
ylide2 (1) with 2 equiv of dialkyl acetylenedicarboxylate via
the E-configured vinyl adducts 2 (Scheme 1).
In investigating the scope of this azulene synthesis and as
part of our study of its mechanism, we sought to prepare adducts
analogous to 2, bearing a hydrogen on the CR of the alkene
with geminal substitution on the â-carbon. We proposed to make
these via the key compound triphenylphosphonium 2-(formyl)-
cyclopentadienylide 3 (Scheme 2), a literature compound
reported by Yoshida and co-workers3 as being formed under
Vilsmeier conditions from the Ramirez ylide. We also envisaged
* Corresponding authors. Tel.: +441912225542 (L.J.H.), +35317162308
(D.G.G.). Fax: +35317062127 (D.G.G.).
Results and Discussion
† University of Newcastle.
Our attempts to prepare compound 3 following the literature
method (addition of 1 to an excess of Vilsmeier reagent at -15
to -10 °C) yielded a solid that gave similar spectroscopic data
to that reported.3 However, a product of diformylation was also
routinely found as a contaminant under these conditions
(sometimes as the only product; see below), and we were able
‡ National University of Ireland, Maynooth.
§ University College Dublin.
(1) Higham, L. J.; Kelly, P. G.; Corr, D. M.; Muller-Bunz, H.; Walker,
B. J.; Gilheany, D. G. Chem. Commun. 2004, 684-685.
(2) Ramirez, F.; Levy, S. J. Am. Chem. Soc. 1957, 79, 67-69.
(3) Yoshida, Z.; Yoneda, S.; Murata, Y.; Hashimoto, H. Tetrahedron
Lett. 1971, 1523-1526.
10.1021/jo7014104 CCC: $37.00 © 2007 American Chemical Society
Published on Web 10/19/2007
8780
J. Org. Chem. 2007, 72, 8780-8785

Substitution Reactions of the Ramirez Ylide
SCHEME 1. Azulene Synthesis from Ramirez Ylide (R1/R2 ) CH3, C2H5, and (CH3)3C)
SCHEME 2. Proposed Route to Geminal-Substituted
substituted at the 3-position), with that of triphenylphosphonium
2-(Z-(R,â-dicarbomethoxyvinyl)cyclopentadienylide 7, which we
have previously confirmed as substituted at the 2-position, also
by virtue of an X-ray crystal structure.5 Figure 2 illustrates that,
as expected, the coupling patterns for the cyclopentadienylide
protons of the two compounds are very different. H5 and H3
both couple to H4 for 7, whereas for 5, only H4 and H5 couple
strongly together (see Supporting Information for further
comment).
Adductsa (R/R1 ) Carbonyl, Ester, or Cyano Groups)a
a Reagents and conditions: (i) POCl3/DMF (-15 °C), (ii) activated
methylene compound using Knoevenagel conditions.
When the formyl ylide 4 prepared by us is subjected to the
same analysis, we find that the NMR is consistent with
substitution at the 3- and not the 2-position; by way of
confirmation, a 1H NOE experiment (with variable mixing times;
see Supporting Information) showed a correlation between H4
and H5, and another between H2, H4 and the aldehyde proton.
to show that formylation had in fact given the 3-substituted
isomer 4. This was because the compound formed could be
subsequently reacted (Scheme 3) with dimethyl malonate under
Knoevenagel conditions to give 5, the structure of which was
established by X-ray crystallography (Figure 1). Similar reaction
with methyl cyanoacetate gave the analogous adduct 6 with the
nitrile group syn to the ring, again established by X-ray (Figure
1).4 The structural parameters of compounds 5 and 6 compare
well with our previous work5 and that of others.6
A
1H-31P correlation showed cross-peaks for each of the
cyclopentadienylide protons with the phosphorus resonance at
14.2 ppm, although the correlation is weaker for H4 (these
spectra are reproduced in the Supporting Information). These
additional NMR experiments for structural confirmation were
important, because use of the COSY analysis alone can be
misleading, as it may show coupling between all the ring protons
in certain cases; for instance, in the reaction of 1 with tetrahalo-
p-benzoquinones to give 3-substituted ring products, both 3JHH
and 4JHH coupling were observed.7 Two other intriguing NMR
phenomena are encountered when analyzing these systems: (i)
there are occasionally no observable correlations between certain
5-membered ring protons with any carbon in the HSQC NMR
and (ii) extreme broadening of some resonances in the 13C NMR
hindered their identification.8
To a certain extent it is understandable that the original
workers misassigned the structure of compound 4, because it
was consistent with the results with some other electrophiles
1
(see below) and was not ruled out by the H NMR spectrum.
However, it is notable in hindsight that the presence of 31P
coupling to the three cyclopentadienylide protons serves to
complicate matters and would make such an analysis difficult.
Furthermore, it is important to note that there is occasionally a
4JHH coupling observed between H2 and H4 and H5 in the
3-substituted compoundssa feature noted previously7sdepending
on the threshold value employed in the NMR processing
We also found that the IR spectrum of 4 showed a carbonyl
band at 1632 cm-1 (cf. the 1640 cm-1 reported).3 Freeman and
Lloyd9 have suggested that the carbonyl stretch can distinguish
between 2- and 3-substitution in the acetyl derivatives; the
former leads to a lowering of the V(CO) absorption as a result
of the resonance canonical structure, whereby a negative charge
on the oxygen atom takes part in an intramolecular association
with the heteronium atom. A 3,5-diformylated derivative
(phenyls at the 2- and 4-positions) was reported in the same
paper as having two carbonyl stretches at 1622 and 1645 cm-1
and the 3,4-diformylcyclopentadienylide made by us (see below)
has a V(CO) absorption at 1647 cm-1, so it appears this method
has some validity. It should also be noted that the UV spectra
of 4 determined by us does differ somewhat from that reported
in that we did not find a band at 370 nm. We attribute this to
contamination by decomposition products, well-known in the
case of the parent Ramirez ylide.
1
package. In the H-1H COSY spectra, the magnitude of this
3
coupling appears less than the JHH coupling between H4 and
H5. However, this smaller coupling has the potential to be
misleading if used as the sole method of analysis; the phenom-
enon presumably owes its origins to the high degree of
delocalization in the compounds.
In order to provide an accurate and straightforward method
for the secure assignment of the products obtained when 1 reacts
with a variety of electrophiles (in lieu of samples suitable for
X-ray crystallographic analysis), we compared the 1H-1H
COSY NMR spectra of compound 5 (proven here to be
(4) Crystallographic data (excluding structure factors) for the structures
in this paper have been deposited with the Cambridge Crystallographic
Data Centre as supplementary publication numbers CCDC 272186 (5),
272187 (6), 633415 (8) and 633413 (9). Copies of the data can be obtained,
free of charge, on application to CCDC, 12 Union Road, Cambridge,
CB2 1EZ, UK [fax: +44(0)-1223-336033 or e-mail: deposit@ccdc.cam.
ac.uk].
The misassignment of the structure of compound 4 (and some
other structures; see below) is quite unfortunate, because both
these authors and subsequent workers have relied on it in trying
to understand electrophilic substitution of the Ramirez ylide.
Thus Yoshida and co-workers went to considerable lengths in
(5) Higham, L. J.; Kelly, P. G.; Muller-Bunz, H.; Gilheany, D. G. Acta
Crystallogr. C 2004, 60, o308-o311.
(6) See, for instance, (a) Huy, N. H. T.; Compain, C.; Ricard, L.; Mathey,
F. J. Organomet. Chem. 2002, 650, 57-58. (b) Ammon, H. L.; Wheeler,
G. L.; Watts, P. H., Jr. J. Am. Chem. Soc. 1973, 95, 6158-6163. (c) Holy,
N. L.; Baenziger, N. C.; Flynn, R. M.; Swenson, D. C. J. Am. Chem. Soc.
1976, 98, 7823-7824.
(7) Pons, M.; Pe´rez Pla, F.; Valero, R.; Hall, C. D. J. Chem. Soc., Perkin
Trans. 1996, 2, 1011-1019.
(8) A full listing of these issues is given in the Supporting Information.
(9) Freeman, B. H.; Lloyd, D. Tetrahedron 1974, 30, 2257-2264.
J. Org. Chem, Vol. 72, No. 23, 2007 8781

Higham et al.
FIGURE 1. The X-ray crystal structures of compounds 5 (left) and 6.
FIGURE 2. 1H COSY NMR analysis of 5 and 7. The spectra illustrate the distinct coupling patterns observed for 5 (H4 and H5 coupling, shown
left) and 7 (H3, H4 and H5 couple together).
a subsequent paper10 to understand which is the most susceptible
position on the 5-membered ring to electrophilic substitution.
Among a number of models studied (reaction indices including
superdelocalizabilities, localization energies, and π-electron
densities), electron transfer theory was used to rationalize the
supposed preference for attack at the 2-position. This was despite
steric considerations and HMO calculations on the electron
densities at the different sites deeming the 2-position less
favorable than C1 or C3.3,10 Lloyd and Singer instead speculated
that 2-substitution may be due to a more favored transition state
expected of the linear rather than cross-conjugated system.11
Electrophilic substitution at the 3-position on the cyclopenta-
dienylide ring of 1 has been reported for azasulfonium salts12
and recently with tungsten phosphinidene reagents, with the
latter placement accounted for on steric grounds.6a Addition
reactions at the 3-position with mercury(II) iodide,6c iodine, and
thallium(III) trifluoroacetate13 have also been described, the
reaction outcomes noted as surprising and again attributed to
sterics; a recent theoretical study suggests both the 2- and
3-positions are the preferred sites of protonation.14 Taken with
(12) (a) Schlingensief, K. H.; Hartke, K. Tetrahedron Lett. 1977, 14,
1269-1272. (b) Schlingensief, K. H.; Hartke, K. Liebigs Ann. Chem. 1978,
1754-1764.
(13) Roberts, R. M. G. Tetrahedron 1980, 36, 3295-3300.
(14) Laavanya, P.; Krishnamoorthy, B. S.; Panchanatheswaran, K.;
Manoharan, M. J. Mol. Struct. (THEOCHEM) 2005, 716, 149-158.
(10) Yoshida, Z.; Yoneda, S.; Murata, Y. J. Org. Chem. 1973, 38, 3537-
3541.
(11) Lloyd, D.; Singer, M. I. C. Chem. Ind. 1971, 786.
8782 J. Org. Chem., Vol. 72, No. 23, 2007

Substitution Reactions of the Ramirez Ylide
FIGURE 3. The X-ray crystal structure of 8 and 9.
SCHEME 3. Reaction of the 3-Substituted Isomer 4 with Diethyl Malonate and Methyl Cyanoacetate
SCHEME 4. Verification Method Used by Yoshida et al. to Support Substitution in the 2-Position
our results, it is now clear that ylide 1 is much more like pyrrole
than had been previously thought. We were intrigued as to why
3-substitution occurs under Vilsmeier conditions. One possibility
is that this is a situation of thermodynamic versus kinetic control.
Strong evidence for reversibility of addition at the 3-position
was found by Pons et al. in reactions with tetrahalobenzoquino-
nes.7 Unfortunately, it was difficult to study this possibility.
While no other formylated product was present in the crude
reaction mixture, this does not rule out the possibility that an
initially formed 2-substituted Vilsmeier intermediate rapidly
rearranges to the 3-isomer. The use of milder reaction conditions
merely lowered the yield while the use of even slightly more
forcing conditions gives substantial amounts of a diformylated
compound, which is the exclusive product with excess Vilsmeier
reagent. This had also been reported by Yoshida and co-workers
and assigned as the 2,5-diformylated cyclopentadienylide. The
IR, UV, and melting point data of the compound obtained by
derivatives as well as 2-substituted compounds derived by
reaction with benzaldehyde and maleimide. The authors con-
verted the 2-nitroso compound to triphenylphosphonium (2-
phenylazo)cyclopentadienylide to support their assignment at
the 2-position. The latter compound had been previously
reported by Ramirez and Levy,15 who had made it through
reaction of 1 with benzenediazonium chloride. They had also
established its identity through its treatment with hydrobromic
acid and reduction of the resultant hydrobromide to (2-
phenylhydrazonocyclopentyl)triphenylphosphonium bromide,
whose substitution pattern was in turn authenticated by its
independent preparation from phenylhydrazine and (2-oxocy-
clopentyl)triphenylphosphonium bromide); see Scheme 4.
We repeated the nitrosation reaction according to the reported
experimental procedure and found that this reaction did indeed
lead to the 2-substituted derivative 9, assignable by our described
NMR studies and further characterized by X-ray crystallography
(see Figure 3). Despite repeated attempts, we could not repeat
the nitration reaction of 1, nor the reported reactions of 1 with
benzaldehyde or maleimide.
1
us correlates with that reported. However, again the H NMR
suggests a different substitution pattern with one doublet (3JHP
) 4.47 Hz) for the two 5-membered ring protons, an NOE
between the 5-membered ring protons and a phenyl ring, and
an NOE between the aldehyde protons and the ring protons only.
This strongly suggests the 3,4-substitution pattern of compound
8, which was confirmed by an X-ray crystallographic study
(Figure 3).
Yoshida and co-workers also reported that the reaction
between 1 and diethyl acetylenedicarboxylate goes at the
2-position, which we have previously confirmed and shown to
be a separable mixture of the E- and Z-vinyl adducts 10,16 as
shown by our X-ray crystallographic study on the dimethyl ester
We turned then to other electrophilic substitution reactions.
Yoshida and co-workers10 also reported the 2-nitroso and 2-nitro
(15) Ramirez, F.; Levy, S. J. Org. Chem. 1956, 21, 1333-2036.
J. Org. Chem, Vol. 72, No. 23, 2007 8783

Higham et al.
SCHEME 5. Reaction of 1 with the Original Electrophiles
Used and the Correct Assignmentsa
SCHEME 6. The Derivatives Prepared from the
3-Substituted Formyl Ylide 4a
a Reagents and conditions: (i) POCl3/DMF (-15 °C), (ii) POCl3/DMF
(25 °C), (iii) benzenediazonium chloride (0 °C), (iv) NaNO2 in CH3COOH,
(v) C2H5CO2CCCO2C2H5, (vi) (NC)2CdC(CN)2. Compounds 4 and 11 were
originally assigned as substituted in the 2-position, while 8 was described
as the 2,5-diformylated adduct.
a Reagents and conditions: (i-iv) utilized piperidine/benzoic acid/reflux
reaction conditions; (i) dimethyl malonate, (ii) methyl cyanoacetate, (iii)
diethyl malonate, (iv) 2,4-pentanedione, (v) (a) toluene sulfonhydrazide/
reflux, (b) NaBH4/0 °C.
dienylide. For this and a number of other reactions with
electrophilic reagents, 1H-1H COSY and 1H NOE NMR
analysis can be a very useful tool in the structural assignment
of the products. In addition, we have prepared six new
derivatives of the formylated compound (including the tosyl
hydrazone), two of which were characterized by X-ray crystal-
lography.
Our study shows that care must be taken when making
structural assignments in this system, and papers3,10,11,18,20,21,22
detailing the purported 2-substituted adducts must be treated
with caution.
In reference to our initial objective of preparing geminal-
substituted vinyl adducts suitable as precursors for the prepara-
tion of novel azulenes (Scheme 1), we found that the 3-substi-
tuted compounds do not react with acetylenes to give azulenes,
a now entirely expected result in light of our proposed
mechanism.1
analogues.5 However, reaction of 1 with tetracyanoethylene leads
to the 3-substituted adduct 11, which we have also characterized
by X-ray crystallography.17 This contradicts the 2-substitution
proposed by Yoshida and co-workers10 and also by Rigby et
al.18 The NMR, IR, and melting point data obtained by us
suggest that the compounds are the same.
The authenticated results for electrophilic substitution of
Ramirez ylide are shown in Scheme 5. It appears that the
position of substitution may depend on the bulk of the
electrophile used. Thus in comparing similar strength electro-
philes (alkyne vs alkene and NO+/N2+ vs Vilsmeier) the more
bulky react at the 3-position. However, we cannot say this with
certainty, because these are not large differences in size and it
is very noticeable that in all cases the other isomer is not
detected, which would be expected in the case of small energy
differences between the two alternatives. Furthermore, we have
not been able to rule out the intervention of kinetic versus
thermodynamic effects.
Experimental Section
With the substitution pattern of 4 secure, the structures of
our Knoevenagel condensation derivatives from diethylmalonate
and 2,4-pentanedione are now established as 12 and 13. We
also prepared the methyl compound 14 via the tosyl hydrazone
of 4 (see Scheme 6), after other reduction methods failed.19
In summary, we have addressed the issue of the site of
substitution in the parent formyl compound by showing the
compound to be triphenylphosphonium 3-(formyl)cyclopenta-
Please refer to the Supporting Information for detailed charac-
terization data (full NMR, HRMS, and elemental analysis).
Triphenylphosphonium 3-(formyl)cyclopentadienylide (4).
Phosphorus oxychloride (2.85 mL, 4.69 g, 30.6 mmol) was
added over 10 min to a stirred suspension of 1 (10.0 g, 30.6
mmol) at -10 °C in N,N-dimethylformamide (50 mL, 47.2 g, 0.65
mol). The mixture became a brown solution, and after 20 min was
poured onto ice/water (200 g), precipitating a pink solid. NaOH
(10% w/v) was added until the mixture became alkaline to litmus
and it was then stirred overnight. The pink solid was filtered, washed
with water and cold diethyl ether (20 mL), and dried in vacuo.
The solid can be recrystallized from toluene or ethyl acetate/
chloroform.
(16) Under the reaction conditions employed by Yoshida, we found that
a mixture of both the E and Z adducts was obtained. We have success-
fully modified the reaction conditions for 7 and 10 to facilitate the production
of the Z-isomer only in both cases; see Supporting Information for
details.
1
(17) This assignment, based on the H-1H COSY NMR analysis, was
Yield: 7.9 g (71%). Mp: 223-225 °C (lit.10 mp: 221-222 °C
later verified by an X-ray crystallographic study. Although the substitution
at the 3-position is confirmed, some disorder about the vinyl group itself
precludes its inclusion here. The authors would be happy to supply the
CIF to interested parties.
1
dec). H NMR (dimethyl sulfoxide-d6, 500 MHz): 9.60 (s, 1H,
(18) (a) Rigby, C. W.; Lord, E.; Hall, C. D. Chem. Commun. 1967, 714-
716. (b) Rigby, C. W.; Lord, E.; Naan, M. P.; Hall, C. D. J. Chem. Soc. B
1971, 1192-1197.
(19) No reduction of 4 was observed in Wolf-Kishner, Clemmensen,
or Meerwein-Verley-Pondorff reactions or by employing sodium boro-
hydride in refluxing ethanol.
(20) Yoneda, S.; Watanabe, M.; Yoshida, Z. Bull. Chem. Soc. Jpn. 1978,
51, 2605-2608.
(21) Yoshida, Z.; Yoneda, S.; Murata, Y.; Hashimoto, H.; Murata, Y.
Tetrahedron Lett. 1971, 1527-1529.
(22) Yoshida, Z.; Yoneda, S.; Yato, T. Tetrahedron Lett. 1971, 2973-
2976.
8784 J. Org. Chem., Vol. 72, No. 23, 2007

Substitution Reactions of the Ramirez Ylide
aldehyde), 7.91 (m, 3H, Hpara), 7.79 (m, 6H, Hmeta), 7.74 (m, 6H,
Enterprise Ireland for Scholarships (P.G.K. and D.M.C.) and
the European Commission for a Marie Curie Development Host
Fellowship HPMD-CT-2000-00053 (L.J.H.).
Hortho), 6.72 (m, 1H, cyclopentadienylene H4), 6.68 (m, 1H,
cyclopentadienylene H2), 6.30 (m, 1H, cyclopentadienylene H4)
ppm. 31P NMR (dimethyl sulfoxide-d6, 500 MHz): 6.7 ppm.
HRMS: calcd for [MH]+ (C24H20PO) 355.1252, found 355.1265
(3.7 ppm). IR (KBr): Vmax 1630 (CdO) cm -1. λmax (log ꢀ, CH2-
Cl2): 277 (sh) (3.75), 304 (3.98) nm.
Supporting Information Available: Detailed experimental
procedures, NMR spectra, and selected crystallographic data. This
materialisavailablefreeofchargeviatheInternetathttp://pubs.acs.org.
Acknowledgment. The authors thank the National Univer-
JO7014104
sity of Ireland, Maynooth, University College Dublin and
J. Org. Chem, Vol. 72, No. 23, 2007 8785