P(MeNCH2CH2)3N: A Highly Selective Reagent for Synthesizing trans-Epoxides from Aryl Aldehydes

Article abstract of DOI:10.1021/jo000122+

In contrast to its acyclic analogue P(NMe₂)₃ (1), which in benzene at room temperature reacts with two aryl aldehyde molecules bearing electron-withdrawing groups to give the corresponding diaryl epoxide as an isomeric mixture (trans/cis ratios: 72/28-51/49), P(MeNCH₂CH₂)₃N (2a) under the same reaction conditions is found to be a highly selective reagent that provides epoxides with trans/ cis ratios as high as 99/1. These reactions are faster with 2a, because its phosphorus atom is apparently more nucleophilic than that in 1. Thus, it is found that 2a more easily forms 1:1 and 1:2 adducts with one or two molecules of aldehyde, respectively. These adducts apparently are intermediates in the formation of the product epoxide and the corresponding phosphine oxides of 1 and 2a.

Full text of DOI:10.1021/jo000122+

4560
J . Org. Chem. 2000, 65, 4560-4564
P (MeNCH2CH2)3N: A High ly Selective Rea gen t for Syn th esizin g
tr a n s-Ep oxid es fr om Ar yl Ald eh yd es
Xiaodong Liu and J ohn G. Verkade*,†
Department of Chemistry, Iowa State University, Ames, Iowa 50011
jverkade@iastate.edu
Received J anuary 31, 2000
In contrast to its acyclic analogue P(NMe2)3 (1), which in benzene at room temperature reacts with
two aryl aldehyde molecules bearing electron-withdrawing groups to give the corresponding diaryl
epoxide as an isomeric mixture (trans/cis ratios: 72/28-51/49), P(MeNCH2CH2)3N (2a ) under the
same reaction conditions is found to be a highly selective reagent that provides epoxides with trans/
cis ratios as high as 99/1. These reactions are faster with 2a , because its phosphorus atom is
apparently more nucleophilic than that in 1. Thus, it is found that 2a more easily forms 1:1 and
1:2 adducts with one or two molecules of aldehyde, respectively. These adducts apparently are
intermediates in the formation of the product epoxide and the corresponding phosphine oxides of
1 and 2a .
In tr od u ction
of sterically hindered and deactivated alcohols,16 and for
the acylation of alcohols.17 Proazaphosphatrane 2a is a
much stronger base than DBU,18 a commonly used
nonionic base in organic synthesis. Thus, 2a is a superior
base for the synthesis of porphyrins,19for the dehydro-
halogenation of secondary and tertiary halides,20 and for
the synthesis of a chiral fluorescence agent.21 As a result
of these and other emerging applications, 2a has become
commercially available.22
Epoxides are important starting materials in organic
synthesis.1-3 The most frequently used method to gener-
ate them is to oxidize olefins with peroxides. However,
for the preparation of epoxides with sensitive structural
features, this method is not always applicable.4 For
converting aryl aldehydes possessing electron-withdraw-
ing groups to the corresponding epoxides, Mark et al.
found that P(NMe2)3 (1) is a reagent that provides a
simple and mild approach often leading to high yields of
product.4 Moreover, this transformation tolerates sensi-
tive functional groups (such as pyridyl) that oxidative
methods do not. The use of 1 has therefore constituted a
practical synthetic approach.5-8 Although trans-epoxides
are always major products with reagent 1, stereoselec-
tivity is usually poor since trans/cis ratios between 2.6/1
and 1.1/1 are generally realized,4 depending on the
substrate. Thus, finding a stereoselective synthesis of
epoxides from aryl aldehydes has remained challenging.
In recent years we have been exploring the chemistry
of proazaphosphatranes such as 2a -e,9-14 some of which
are proving to be exceedingly potent catalysts, promoters,
and strong nonionic bases that facilitate a variety of
useful organic transformations. For example, 2a is an
efficient catalyst for the trimerization of aryl and alkyl
isocyanates that function as additives in the manufacture
of Nylon-6,15 for the protective silylation of a wide variety
We report herein a facile, efficient, and highly selective
procedure for the synthesis of trans-epoxides from aryl
aldehydes bearing electron-withdrawing groups. For
(11) Laramay, M. A. H.; Verkade J . G. Z. Anorg. Allg. Chem. 1991,
605, 163.
(12) Schmidt, H.; Lensink, C.; Xi, S. K.; Verkade, J . G. Z. Anorg.
Allg. Chem. 1989, 578, 75.
(13) Wroblewski, A. E.; Pinkas, J .; Verkade, J . G. Main Group Chem.
1995, 1, 69.
(14) Liu, X.; Bai, Y.; Verkade, J . G. J . Organomet. Chem. 1999, 582,
16.
(15) Tang, J .-S.; Mohan, T.; Verkade, J . G. J . Org. Chem. 1994, 59,
4931.
† Ph: (515)294-5023. Fax: (515)294-0105.
(1) Sheldon, R. A. Top. Curr. Chem. 1993, 164, 23.
(2) Simandi, L. I. Catalytic Activation of Dioxygen by Metal Com-
plexes; Kluwer Academic Publishers: Dordrecht, 1992; p 109.
(3) Drago, R. S. In Dioxygen Activation and Homogeneous Catalytic
Oxidation; Sima´ndi, L. I., Ed.; Elsevier: New York, 1991; Vol. 66, p
83.
(4) Mark, V. J . Am. Chem. Soc. 1963, 85, 1884.
(5) Newman, M. S.; Blum, S. J . Am. Chem. Soc. 1964, 86, 5598.
(6) Ramirez, F.; Gulati, A. S.; Smith, C. P. J . Org. Chem. 1968, 33,
13.
(7) Moriarty, R. M.; Dansette, P.; J erina, D. M. Tetrahedron Lett.
1975, 30, 2557.
(8) Agarwal, S. C.; Van Duuren, B. L. J . Org. Chem. 1975, 40, 2307.
(9) Verkade, J . G. Acc. Chem. Res. 1993, 26, 483.
(10) Verkade, J . G. Main Group Chem. Rev. 1994, 137, 233.
(16) D′Sa, B.; Verkade, J . G. J . Am. Chem. Soc. 1996, 118, 832.
(17) D′Sa, Bosco A.; Verkade, J . G. J . Org. Chem. 1996, 61, 2963.
(18) Tang, J .-S.; Dopke, J .; Verkade, J . G. J . Am. Chem. Soc. 1993,
115, 5015.
(19) Tang, J .-S.; Verkade, J . G. J . Org. Chem. 1994, 59, 7793.
(20) Mohan, T.; Arumugam, S.; Wang, T.; J acobson, R. A.; Verkade,
J . G. Heteroatom Chem. 1996, 7, 455.
(21) Tang, J .-S.; Verkade, J . G. J . Org. Chem. 1996, 61, 8750.
(22) Strem Chemical Company.
10.1021/jo000122+ CCC: $19.00 © 2000 American Chemical Society
Published on Web 06/29/2000

P(MeNCH2CH2)3N: A Highly Selective Reagent
J . Org. Chem., Vol. 65, No. 15, 2000 4561
comparison of conversions and stereoselectivity, both 1
and 2a were used in parallel reactions. We also find that
the more bulky proazaphosphatrane 2b does not lead to
epoxides, although color changes do occur.
Thus, for substrates 4a , 6a , and 7a , compound 1 gave
higher isolated yields than with 2a although the conver-
sions were lower.
For all the substrates tested except 13a and 14a (which
did not react), compound 2a gave excellent stereoselec-
tivity (trans/cis ratios: 92/8-99/1) based on 1H NMR
integration of the reaction mixtures, while 1 gave iso-
meric mixtures (trans/cis ratios from 51/49 to 72/28)
under the same conditions. The reactions were also
substrate dependent: for the aryl aldehydes with electron-
withdrawing groups (i.e., 4a -7a ), nearly quantitative
conversions (>99%) were achieved within 1 h using 2a .
For the delocalized polyaromatic aldehydes (i.e., 8a -11a ),
high conversions (95-99%) were realized within 12-14
h with 2a . Epoxide formation was more sluggish for
substrate 12a and 60 h were necessary to obtain a 28%
conversion with 2a while less than 2% conversion was
observed with 1. However, 2a did provide good stereo-
selectivity (trans/cis ratio ) 92/8) despite the low conver-
sion. In the case of ketone 13a and 14a , no detectable
formation of corresponding epoxide 13b and 14b was
observed even after 120 and 24 h, respectively, probably
because of steric hindrance present in the substrate.
Moreover, aryl aldehydes bearing electron-donating groups
such as 2,5-dimethylbenzaldehyde and 2-methoxyben-
zaldehyde yielded no detectable epoxide formation over
2 days under the same reaction conditions. These results
are consistent with earlier work4 wherein it was shown
that relatively electrophilic aldehydes promote epoxide
formation. It should be noted that although the trans/
cis ratios of chromatographically purified epoxides 4b-
8b were the same as those displayed by the correspond-
ing reaction mixtures, we observed that when filtration
was used to purify the product, higher trans/cis ratios
were achieved for the isolated epoxides (9b-11b) than
those observed for the corresponding reaction mixtures.
Here, the trans-epoxides were less soluble than their cis
isomers in benzene.
Two reaction pathways have been considered for 1.4,6
The one put forth by Mark et al. in Scheme 14 involves
phosphorus nucleophilically attacking an aldehyde car-
bonyl carbon to form the 1:1 adduct 15 wherein the
oxygen attacks a second molecule of the aldehyde to form
the 1:2 adduct 16 (for which three resonance forms are
shown). Epoxide formation would occur by carbanion
attack in 16c directly on the opposite benzyl carbon to
give trans-epoxide or by carbanion attack on the benzylic
carbon in an SN2 manner after rotation of the O-CHAr
bond to give the cis-epoxides. However, this mechanism
was questioned by Ramirez6 who proposed an alternative
pathway (Scheme 2) wherein the phosphorus in 1 first
electrophilically attacks the carbonyl oxygen of the aryl
aldehyde to form a 1:1 adduct (17) on the grounds that
the phosphorus of 1 should exhibit an even greater
tendency to electrophilically attack the carbonyl oxygen
than the phosphorus of trialkyl phosphites that were
reported to give isolable adducts of type 18b in Scheme
2.6 Thus, after 17 is formed, a second molecule of the
aldehyde attacks 17 to give a mixture of erythro (18a )
and threo (18b) 1:2 adducts. If the erythro form is
predominant, as might be expected for steric reasons, the
trans-epoxide formed by loss of the oxide of 1 should
predominate over the cis product. However 1 gave rise
to poor stereoselectivity, with trans/cis ratios generally
ranging from 1.1 to 2.6,4 probably owing to similar steric
hindrance in the erythro and threo forms of the adduct
Resu lts a n d Discu ssion
That 2a becomes oxidized to 3 in the reaction of two
aryl aldehyde molecules in benzene to give the corre-
1
sponding epoxide (eq 1) was shown by 31P and H NMR
spectroscopic analysis of a C6D6 solution of a 2.0/1.0 equiv
ratio of 9a to 2a (Table 1). Only one 31P resonance at 20.2
ppm corresponding to oxide 310 was observed after 12 h
at room temperature. In the 1H NMR spectrum, the CHO
proton resonance also disappeared after 12 h and a
singlet at 3.95 ppm corresponding to the oxirane proton
was observed (99% conversion). By comparison of 1H and
13C NMR spectroscopic data in the literature,23 the
epoxide 9b present in the reaction mixture was found to
be almost pure trans (trans/cis ratio ) 98/2). Simply
filtrering the reaction mixture and washing the filtered
1
solid with cool benezene afforded H NMR spectroscopi-
cally pure trans-9b in 75% yield. For comparison, the use
of 1 under the same conditions led to a considerably
slower reaction and lower stereoselectivity (85% conver-
sion with a trans/cis ratio of 69/31). The isolated yield of
epoxide in both bases, however, was virtually the same
(74% with 1 and 75% with 2a ) because a substantial
amount of starting material is converted to intermediate
adducts with 2a while no such adducts survive in the
reaction with 1 (see later). The improved stereoselectivity
realized with 2a prompted us to evaluate this reagent
with the substrates in Table 1. Compound 1 was used in
parallel reactions to obtain conversions, yields, and trans/
cis ratios for comparison with the data realized with 2a ,
and these data are also included in this table.
Table 1 reveals that substrates 12a , 13a , and 14a
afforded no detectable or very slow reaction rates,
respectively, with 0.5 equiv of 1, although a 28% conver-
sion to product was realized with 0.5 equiv of 2a (trans/
cis ratio ) 92/8) in the case of substrate 12a . Faster
reactions were observed for substrates 4a -11a with 0.5
equiv of 2a (95-100% conversions) than those carried out
with 0.5 equiv of 1 (5-95% conversions). However, the
faster conversions with 2a did not necessarily lead to
higher product yields, owing to formation of compara-
tively robust 1:2 adducts of 2a and the aldehyde substrate
(see later), which gave rise to higher conversions of
starting materials but to lower isolated yields of epoxide.
(23) Wong, J . P. K.; Fahmi, A. A.; Griffin, G. W.; Bhacca, N. S.
Tetrahedron 1981, 37, 3345.

4562 J . Org. Chem., Vol. 65, No. 15, 2000
Ta ble 1. Ep oxid a tion of Ar yl Ald eh yd es w ith 1 a n d 2a a
Liu and Verkade
a
b
1
Reactions were carried out in C6D6 or C6H6 at room temperature under Ar. Identification was made by comparing H and 13C NMR
spectroscopic data with those in the references indicated. New compounds were fully characterized by 1H and 13C and HRMS (EI)
spectroscopies. c Conversions were determined by 1H NMR integration of signals of the methine proton in the epoxide product to the
aldehyde proton in the corresponding reactant. The trans:cis ratios were obtained by integrating the relevant 1H NMR signals in the
d
reaction mixture or of the isolated epoxide isomer mixture (see Results and Discussion). e Isolated yields were obtained by the indicated
methods , and purity was determined by 1H NMR spectroscopy. f See ref 4. g Silica gel column chromatography using a mixture of hexanes
h
(95%) and ethyl acetate (5%) as the eluant. Minami, T.; Matsuzaki, N.; Ohshiro, Y.; Agawa, T. J . Chem. Soc, Perkin Trans. 1 1980,
i
j
k
1731. Clark, K. B.; Bhattacharyya, K.; Das, P. K.; Scaiano, J . C.; Schaap, A. P. J . Org. Chem. 1992, 57, 3706. See ref 23. Since the
product is insoluble in benzene, filtration followed by washing with benzene and drying in vacuo was employed to give 1H NMR pure
product. See Supporting Information. Aldrich Library of 13C and 1H FT NMR spectra 1993, 1(2), 222A. Aldrich Library of 13C and 1H
l
m
FT NMR spectra 1993, 1(2), 221C. D’Auria, M.; Mauriello, G. Photochem. Photobiol. 1994, 606, 542. o No observable reaction. No epoxide
n
p
formation was observed,
formed from a variety of substituted aryl aldehydes.4,6 It
was also reported6 that when cyclic 19 was employed
instead of 1 in the reaction with 4-nitrobenzaldehyde, the
corresponding 1:2 adducts containing pentavalent phos-
phorus (21a and 21b in Scheme 3) were isolated as a
mixture in 60% yield in a 1:1 ratio.
NMR spectrum of the reaction mixture exhibited three
resonances (20.6, 22.7, and 22.8 ppm) in a 20:4:1 ratio.
The signal at 20.6 ppm is characteristic of the oxide 3,10
and the other two are believed to be associated with the
1:1 and 1:2 adducts formed as intermediates. Mass
spectroscopy (ESI) of the same reaction mixture revealed
a large peak at 468 Da corresponding to the 1:2 adduct
and a considerably smaller signal at 337 Da attributable
In the present work, when 0.5 equiv of 2a was allowed
to react with 4a in C6D6 at room temperature, the 31P

P(MeNCH2CH2)3N: A Highly Selective Reagent
J . Org. Chem., Vol. 65, No. 15, 2000 4563
Sch em e 1
Sch em e 4
Sch em e 2
owing to peak overlaps. These observations are consistent
with the reported result from the reaction of cyclic 19 in
Scheme 3 with 4-nitrobenzaldehyde to give the 1:2
adduct. However, no evidence for the formation of the
1:1 adduct was reported.6 On the basis of the 31P NMR
chemical shifts of the 1:1 and 1:2 adducts observed in
our work (which are much closer to those of a P-O
bonded adduct24 than to that of a carbon bonded20 adduct
of 2a ) the reaction pathway proposed by Ramirez6
(Scheme 2) seems to be supported by our results. Since
2a like 19 is a cyclic aminophosphine, the reaction
pathway in Scheme 4 is analogous to that of 19 in Scheme
3. We also note that when more bulky substrates such
as 9a were allowed to react with 0.5 equiv of 2a , similar
evidence of the presence of a mono- and diadduct was
obtained from 31P NMR and mass (ESI) spectroscopies.
However, the ratio of 3 to 1:2 adduct to 1:1 adduct was
30:5:1, implying a higher production of 9b from 9a
compared with the production of 4b from 4a (Table 1).
Efforts to isolate a pure adduct of 4a with 2a or of 9a
with 2a in a 1:1 or 2:1 ratio at -78 °C (in toluene) or
room temperature always resulted in a mixture of 3, the
1:2 adduct, and the 1:1 adduct. Attempts to purify the
1:1 or 1:2 adduct by recrystallization also failed.
Sch em e 3
As mentioned earlier, when the cyclic aminophosphine
19 was allowed to react with an aryl aldehyde bearing
an electron-withdrawing group, a 1:2 adduct (trans:cis
ratio 1:1) was isolated.6 The cis isomer isolated from this
mixture refluxing in ethanol gave the trans-epoxide.
However, the isomeric mixture of the 1:2 adducts of 2a
observed in the present work gave no evidence of decom-
position in this manner under the same conditions.
During silica gel column chromatographic purification,
however, these adducts did dissociate to give the starting
aldehyde. Attempts to improve product yields by prolong-
ing the reaction time, changing the reaction temperature,
or using different solvents such as THF, toluene, or CH2-
Cl2 failed. By contrast, the reaction of 1 with these
aldehydes provided no detectable quantities of the cor-
responding 1:1 or 1:2 adducts.
to the 1:1 adduct. Thus, it is reasonable to assign the
large 31P NMR peak at 22.7 ppm to the 1:2 adduct and
the smaller one at 22.8 ppm to the 1:1 adduct. The H
NMR spectrum of the reaction mixture failed to give
unambiguous evidence for the ratio of these two adducts
1
(24) Liu, X.; Verkade, J . G. Inorg. Chem. 1998, 37, 5189.

4564 J . Org. Chem., Vol. 65, No. 15, 2000
Liu and Verkade
Although 1 reacts with 1 equiv of benzaldehyde to
produce a 1:1 adduct,4 aryl aldehydes bearing electron-
withdrawing groups give epoxides as the dominant
products and whose stereochemistry is only somewhat
preferentially trans. On the other hand, when 19 was
employed, an isomeric mixture of trans and cis (1:1), a
1:2 intermediate adduct was obtained from which the
pure cis 1:2 adduct intermediate could be isolated.6 The
cis 1:2 adduct gave pure trans-epoxide in 86% yield but
in <20% overall yield from the aldehyde.6 It was found
in the present work that 1 reacts much more slowly with
aryl aldehydes bearing electron-withdrawing groups than
2a (see Table 1). Owing to the possibility of transannular
interaction from the axial nitrogen to the bridgehead
phosphorus in 2a , this base is stronger and perhaps more
nucleophilic. Thus, 2a was expected to be more reactive
with an aldehyde carbonyl group than 1, and this is
reflected in Table 1. It is conceivable that in the use of
acyclic 1 or cyclic 19 for epoxide formation from alde-
hydes, steric differences between cis and trans 1:2 adduct
intermediates are relatively small, owing to P-N bond
rotation of at least one phosphorus substituent, resulting
in poor stereoselectivity. On the other hand, the structure
of 2a is rigid and the 1:2 adduct intermediate must adopt
a less steric hindered conformation (Scheme 4). Some-
what surprisingly, the present work showed that the cis
1:2 adduct (23b) is apparently the more stable isomer.
It is speculated that the steric hindrance between the
aromatic rings in 23a and the methyl groups in 2a is
higher than that in 23b owing to a folding of the five-
membered ring in 23b away from the methyl group on
the mirror plane to an “envelope” conformation with
phosphorus in the “flap” position. The greater bulk and
rigidity of the intermediate 1:2 adduct of 2a than the
analogue with 1 is thus perhaps responsible for the
greater stereoselectivity of base 2a as shown in Table 1.
However, the cause for the survival of the 1:2 isomeric
intermediate adduct of 2a in refluxing ethanol remains
unclear.
Interestingly, the isolated cis 1:2 adduct 21b in Scheme
3 and the almost exclusively formed analogue 23b in
Scheme 4 give rise to trans-epoxide. This is best under-
stood in terms of heterolytic cleavage of a P-O bond to
form a zwitterion followed by rotation about the ArC-
CAr bond and subsequent nucleophilic attack of the
alkoxide oxygen on the adjacent aryl carbon in an SN2
fashion to give the trans product. By inference the 1:2
adducts 21a and 23a lead to cis-epoxides. This conclusion
coupled with the stereoselectivity for trans-epoxide in our
reaction supports our assignment of the cis configuration
to the greatly predominant 1:2 adduct isomer 23b.
When ketone 14a was allowed to react with 1, no
change was observed in the 1H NMR spectrum of the
reaction mixture. However, when 2a was allowed to react
with 14a in a 1:2 ratio, a 1H NMR spectrum of the
reaction mixture showed that the decrease in intensity
of the pair of aromatic proton doublets for the substrate
was accompanied by the formation of two new doublets
in the aromatic range. This tentatively suggests that a
1:1 adduct is formed with relative ease, but that steric
hindrance inhibits the carbanion in the 1:1 adduct from
nucleophilically attacking the CdO carbon of the sub-
strate to form a 1:2 adduct. When 2a and 14a were mixed
in a 1:1 ratio, the same new aromatic resonance for this
proposed 1:1 adduct was observed. It should be men-
tioned that this species is probably not a 1:2 adduct since
a more complex aromatic resonance in the 1H NMR
spectrum should then have been observed. Upon workup
of the reaction by silica gel chromatography, 14a was
recovered and no 14b was observed.
Compound 2b, a more sterically hindered analogue of
2a , was also allowed to react with substrates 4a -9a
under the same conditions as with 2a . Surprisingly, no
epoxide products were generated according to 1H and 31P
NMR spectroscopy. Instead, a remarkable color change
from colorless to dark red was observed, which may be
associated with a charge-transfer complex between the
aminophosphine phosphorus as the donor and the aryl
aldehyde bearing an electron-withdrawing group as the
acceptor as was postulated to form as an intermediate
in the reaction of 1 with aryl aldehydes.6 The failure of
2b to facilitate epoxide formation may well be due to the
steric hindrance around the phosphorus that prevents it
from nucleophilically attacking the carbonyl group of a
substrate, thus restricting the interaction to charge
transfer.
Exp er im en ta l Section
Benzene (C6H6 and C6D6), toluene, and THF were dried with
sodium. CH2Cl2 was dried with CaH2. All reactions were
carried out under an Ar atmosphere. Chemicals employed were
purchased from Aldrich Chemicals and were used without
purification unless otherwise noted. Compounds 2a 12 and 2b13
were prepared according to our previously published methods.
NMR, MS (ESI), and HRMS (EI) spectroscopic measurements
were performed in the Instrument Services Laboratory of the
Chemistry Department at Iowa State University.
Gen er a l P r oced u r e for Con ver tin g Ar yl Ald eh yd es to
Ep oxid es w ith 1 or 2a . To a solution of an aryl aldehyde
(2.00 mmol) in C6D6 (3.0 mL) at room temperature under an
Ar atmosphere was slowly added a solution of 1 or 2a (1.05
mmol) in C6D6 (1.5 mL). The reaction was monitored by 1H
and 31P NMR spectroscopies. After the reaction time stated in
Table 1, an 1H NMR spectrum was recorded from which the
conversion and the trans/cis ratio of the product were obtained
by integration. For known compounds, these spectra were also
compared with those recorded in the literature. The product
epoxide was then isolated and purified by silica gel column
chromatography, or filtration followed by washing (see Table
1). The identities of the purified products were confirmed by
1H and 13C NMR spectroscopies. In the case of new compounds,
HRMS (EI) data are also given (see Supporting Information).
Ack n ow led gm en t. The authors are grateful to the
donors of the Petroleum Research Fund, administered
by the American Chemical Society, for a grant in
support of this research. The authors thank Dr. Palani-
chamy Ilankumaran for valuable discussions.
Su p p or t in g In for m a t ion Ava ila b le: 1H and 13C NMR
and HRMS (EI) spectroscopic data. This material is available
free of charge via the Internet at http://pubs.acs.org.
J O000122+