On the reaction of acyl chlorides and carboxylic anhydrides with phosphazenes

Full text of DOI:10.1021/jo951951b

5638
J . Org. Chem. 1996, 61, 5638-5643
Sch em e 1
On th e Rea ction of Acyl Ch lor id es a n d
Ca r boxylic An h yd r id es w ith P h osp h a zen es
Imma Bosch, Angel Gonza´lez, Fe`lix Urp´ı,* and
J aume Vilarrasa*
Department of Organic Chemistry, Faculty of Chemistry,
University of Barcelona, 08028 Barcelona, Catalonia, Spain
Received November 2, 1995
Sch em e 2
Recently, Molina et al. reported1 that the reaction
between arylacetyl chorides and triphenylphosphazenes2
gave phosphonium-substituted amides (see Scheme 1) in
25-97% yields. These results are in sharp contrast with
other reports in which acyl chlorides are heated with
triphenylphosphazenes to afford imidoyl chlorides (chlor-
oimines) and, after aqueous treatment, the expected
carboxamides.3,4 The results of Molina et al.1 also seem
to disagree with the reactivity shown by mixed carboxylic
anhydrides when treated with trialkylphosphazenes,
which after the workup afford normal amides (Scheme
1); these reactions are the basis of a method developed
by us for the formation of macrolactams and peptides
under mild conditions.5
tween using simple aliphatic acyl chlorides or the corre-
Why in some cases these unexpected C-phosphonium
salts are the predominant products whereas in many
others standard carboxamides are chiefly isolated is not
clear. It is surprising that such small changes in the
substrates and reagents could give rise to such a dispar-
ate behavior. We present here chemical and spectro-
scopic evidence that accounts for the differences observed.
sponding alkanoyl-aroyl anhydrides.
The yields of N-benzylpropanamide (1) were higher
with Bu3PdNCH2Ph (Table 1, entries 1 and 2) than with
Ph3PdNCH2Ph (Table 1, entries 3 and 4). Apparently,
the former reacted more quickly than the latter; in fact,
we noted that the reaction was complete after 5 h in the
two first cases but not in the other two. These differences
may be accounted for bearing in mind that, in general,
trialkylphosphines are more basic and nucleophilic than
triarylphosphines7 and that the basicity and nucleophi-
licity of the corresponding phosphazenes is not identical
either;4,8 it may then be assumed that some reactive
intermediates are formed and converted into the final
products more rapidly in the former examples than in
the latter. Similarly, the yields of N-benzylisobutyramide
(2) were also higher in entries 5 and 6 than in entries 7
and 8 (Table 1), respectively. What is clear, however, is
that Bu3PdNCH2Ph and Ph3PdNCH2Ph show a parallel
behavior.
Resu lts a n d Discu ssion
In flu en ce of th e Rea ction Con d ition s. Reaction of
three simple acyl chlorides and three aliphatic-aromatic
carboxylic anhydrides with Bu3PdNCH2Ph [(N-benzyl)-
tributylphosphazene] or with Ph3PdNCH2Ph [(N-benzyl)-
triphenylphosphazene] has been systematically investi-
gated in order to determine the percentages of amides
1-3 and C-phosphonium salts (see Scheme 2) that may
be achieved in each case. The results obtained in
benzene, the solvent most commonly used in previous
papers,1,5 are summarized in Table 1.
Table 1 shows that aliphatic carboxyl derivatives,
either acyl chlorides or mixed anhydrides (Table 1,
entries 1-8), gave the normal carboxamides after work-
up; the abnormal amides (C-phosphonium salts) were not
detected.6 Thus, there are no practical differences be-
On the other hand, entries 9-18 of Table 1 indicate
clearly that phenylacetic derivatives are prone to give
phosphonium salts (4a -d ), since both 3 and 4 were
isolated in all runs. Shortening of the reaction time to
30 min (Table 1, entry 13) gave rise to a poor yield of 3
but to almost the same yield of 4a (compare entry 13 to
entry 9 of Table 1). By lengthening the reaction time to
15 h, the sums of yields of 3 and 4 increased with respect
to those after 5 h (compare entries 14 and 15 to entries
9 and 10 of Table 1, respectively),9 but mainly the yield
of 3 improved. The experiments of entries 16-18 (Table
1) were also illuminating, since it was noted that low
temperature and, especially, dropwise addition of the
phosphazene solution to the carboxyl derivative (in order
(1) Molina, P.; Alajar´ın, M.; Lo´pez-Leonardo, C.; Alca´ntara, J .
Tetrahedron 1993, 49, 5153. Molina, P.; Alajar´ın, M.; Lo´pez-Leonardo,
C. Tetrahedron Lett. 1991, 32, 4041.
(2) P,P,P-Triphenylphospha- λ5-azenes; phosphazenes are also called
iminophosphoranes and phosphimines.
(3) For example, see: Zbiral, E.; Stroh, J . Liebigs Ann. Chem. 1969,
29. Bachi, M. D.; Vaya, J . J . Org. Chem. 1979, 44, 4393. Bachi, M. D.;
Klein, J . J . Chem. Soc., Perkin Trans. 1 1983, 1925. Barluenga, J .;
Ferrero, M.; Palacios, F. J . Chem. Soc., Perkin Trans. 1 1990, 2193.
Katritzky, A. R.; Zhang, G.; J iang, J . J . Org. Chem. 1994, 59, 4556.
(4) For reviews on phosphazene chemistry, see: Barluenga, J .;
Palacios, F. Org. Prep. Proced. Int. 1991, 23, 1. Gololobov, Y. G.;
Kasukhin, L. F. Tetrahedron 1992, 48, 1353. Eguchi, S.; Matsushita,
Y.; Yamashita, K. Org. Prep. Proced. Int. 1992, 24, 209. J ohnson, A.
W. Ylides and Imines of Phosphorus; Wiley: New York, 1993.
(5) Bosch, I.; Romea, P.; Urp´ı, F.; Vilarrasa, J . Tetrahedron Lett.
1993, 34, 4671. For a related paper, see: Bosch, I.; Urpı´, F.; Vilarrasa,
J . J . Chem. Soc., Chem. Commum. 1995, 91.
(7) For a review, see: Hudson, H. R. In The Chemistry of Organo-
phosphorus Compounds; Hartley, F. R., Ed.; Wiley: New York, 1990;
Vol. 1, pp 385 and 473. For early studies: Henderson, W. A.; Streuli,
C. A. J . Am. Chem. Soc. 1960, 20, 5791. Henderson, W. A.; Buckler, S.
A. J . Am. Chem. Soc. 1960, 20, 5794. For illustrative examples, from
ourselves: Urp´ı, F.; Vilarrasa, J . Tetrahedron Lett. 1986, 27, 4623.
Urp´ı, F.; Vilarrasa, J . Tetrahedron Lett. 1990, 31, 7497.
(6) Although not indicated in Table 1, the reactions of entries 1 and
5 were repeated at 60 °C, with phosphazene excess, but C-phosphonium
salts were not detected either, while the normal carboxamides were
isolated in 90-92% yields.
(8) Genkina, G. K.; Korolev, B. A.; Gilyanov, V. A.; Kabachnik, M.
I. Zh. Obshch. Khim. 1971, 41, 80; Chem. Abstr. 1971, 75, 129120w.
S0022-3263(95)01951-7 CCC: $12.00 © 1996 American Chemical Society

Notes
entry
J . Org. Chem., Vol. 61, No. 16, 1996 5639
Ta ble 1. Rea ction of Ca r boxyl Der iva tives w ith P h osp h a zen es in Ben zen ea
acid derivative
phosphazene
reactn time (h)
amide
yield (%)
phosp salt
yield (%)
1
2
3
4
CH3CH2COCl
Bu3PdNCH2Ph
Bu3PdNCH2Ph
Ph3PdNCH2Ph
Ph3PdNCH2Ph
5
5
5
5
1
1
1
1
88
91
51
50
e
e
CH3CH2COOCOArb
CH3CH2COCl
CH3CH2COOCOAr
5
6
7
8
(CH3)2CHCOCl
(CH3)2CHCOOCOAr
(CH3)2CHCOCl
Bu3PdNCH2Ph
Bu3PdNCH2Ph
Ph3PdNCH2Ph
Ph3PdNCH2Ph
5
5
5
5
2
2
2
2
87
95
79
71
(CH3)2CHCOOCOAr
9
10
11
12
PhCH2COCl
PhCH2COOCOAr
PhCH2COCl
Bu3PdNCH2Ph
Bu3PdNCH2Ph
Ph3PdNCH2Ph
Ph3PdNCH2Ph
5
5
5
5
3
3
3
3
30
24
30
21
4a
4b
4c
4d
36
40
33
36
PhCH2COOCOAr
13
14
15
16
17
18
PhCH2COCl
PhCH2COCl
PhCH2COOCOAr
PhCH2COCl
PhCH2COCl
PhCH2COCl
Bu3PdNCH2Ph
Bu3PdNCH2Ph
Bu3PdNCH2Ph
Bu3PdNCH2Ph
Bu3PdNCH2Ph
Ph3PdNCH2Ph
0.5
15
3
3
3
3
3
3
17
46
44
40
62
59
4a
4a
4b
4a
4a
4c
34
38
42
29
13
10
15
15c
dropw,d
dropw,d
5
5
a
0.2 M solutions of both reagents. With the exceptions pointed out by footnotes, the carboxyl derivative solution was added to the
b
phosphazene solution at 5 °C, and the mixture was then stirred at rt as indicated. Ar ) 2,4,6-trichlorophenyl, but the results were
practically identical with Ar ) 3,5-dinitrophenyl in several experiments. c Addition at -20 °C in toluene instead of benzene. Dropwise
d
addition, over 45 min, of the phosphazene solution to the acid chloride solution. e Not detected (no precipitate was formed; after the
workup, no species of this kind was present in the crude product or was isolated after column chromatography).
Ta ble 2. Effect of th e Tem p er a tu r e a n d Rea gen t Excess in th e Rea ction of P h en yla cetyl Ch lor id e w ith
(N-Ben zyl)tr ibu tylp h osp h a zen ea
entry
acid chloride
phosphazene (equiv)
solvent
T (°C)
60
time (h)
15
15
15
yield of 3 (%)
yield of 4a (%)
1
2
3
4
5
6
PhCH2COCl
PhCH2COCl
PhCH2COCl
PhCH2COCl
PhCH2COCl
PhCH2COCl
Bu3PdNCH2Ph (2.0)
Bu3PdNCH2Ph (1.0)
Bu3PdNCH2Ph (1.0)
Bu3PdNCH2Ph (1.0)
Bu3PdNCH2Ph (1.0)
Bu3PdNCH2Ph (0.9)
CHCl3
CHCl3
CHCl3
CHCl3
CHCl3
CHCl3
16
30
41
49
64
70
65
49
37
30
16
9
60
rt
-20/rtb
-78/rtb
-78/rtb
15
15
dropw,c 15
7
8
9
PhCH2COCl
PhCH2COCl
PhCH2COCl
PhCH2COCl
Bu3PdNCH2Ph (1.0)
Bu3PdNCH2Ph (2.0)
Bu3PdNCH2Ph (1.0)
Bu3PdNCH2Ph (0.9)
CH3CN
CH3CN
CH3CN
CH3CN
rt
15
15
15
47
61
71
78
32
14
5
-35/rtb
-35/rtb
-35/rtb
10
dropw,c 15
trace
a
b
0.2 M solutions of both reagents. Addition of the acyl chloride to the phosphazene. Addition at the first temperature, and stirring
at the second one for several hours, as indicated.c Dropwise addition over 45 min of the phosphazene to the acyl chloride at the first
temperature (bath temperature) and stirring for ca. 15 h at rt.
to avoid a relative excess of the phosphazene) decreased
the formation of C-phosphonium salts 4.
(Table 2, entry 7). Again, the yield of 4a was at a
minimum at the lower temperature and when a dropwise
addition of a small molar deficiency of phosphazene was
used (Table 2, entry 10).
It should be mentioned that 4a -d are stable com-
pounds that could not be hydrolyzed to 3 by heating in
water (with a few drops of acetonitrile to increase the
solubility) or by refluxing in 2 M HCl for many hours.
This means that the abnormal reaction may be very
detrimental in practice, when the preparation of peptides
and macrolactams5 is sought.
From Tables 1 and 2, it is clear that the pathways
toward 3 and 4 are in competition and that the chance
of obtaining phosphorus-containing amides of type 4 is
much lower when the reaction is performed in the more
polar solvent of the three studied so far, when the
reagents are mixed at low temperature, and/or by avoid-
ing an excess of phosphazene during the reaction. In
short, the undesired products can be largely avoided
under appropriate conditions.
In the light of these last results, we carried out a series
of additional experiments in chloroform and in acetoni-
trile, which are summarized in Table 2. In chloroform,
it turned out (see Table 2, entries 2-5) that the isolated
yields of normal amide 3 and abnormal amide 4a were
strongly dependent on the temperature: 4a was the major
product at 60 °C while 3 predominated at -78 °C. The
highest yield of 4a was obtained in refluxing chloroform
with an excess of Bu3PdNCH2Ph (Table 2, entry 1). On
the other hand, the highest yield of 3 was achieved when
the addition was carried out at -78 °C by dropwise
addition of a small molar deficiency of this phosphazene
(Table 2, entry 6). In acetonitrile, in agreement with
these results, amide 3 predominated largely over 4a at
-35 °C (Table 2, entries 8 and 9) but not so much at rt
(9) After
5 h of reaction (Table 1, entries 9-12), the crudes
contained, in the elution order, 5-10% (w/w) of relatively nonpolar
products (P-lacking byproducts, according to the NMR spectra), 20-
30% of amide 3, ca. 30% of Bu3PO, 33-40% of phosphonium salt 4,
small amounts of phosphazenium salt, and very polar products (not
eluted with 90:10 CH2Cl2-MeOH). At longer reaction times (entries
14 and 15): 5-10% of nonpolar products, ca. 45% of amide 3, ca. 50%
of Bu3PO, ca. 40% of 4, small amounts of phosphazenium salt, and no
polar products at all.
Detection of Rea ctive In ter m ed ia tes. To gain
insight into the mechanisms involved, a few of the
reactions shown in Tables 1 and 2 were monitored by 31
P
and 1H NMR spectroscopy. First, the reaction of
Bu3PdNCH2Ph (freshly prepared in benzene followed by

5640 J . Org. Chem., Vol. 61, No. 16, 1996
Notes
Sch em e 3
type 5 (5b) was much less intense; (ii) the O-phospho-
nium species of type 6 predominated; (iii) afterwards, the
signal at δ 53 decreased and the signal at δ 44.7 due to
Bu3PdO grew more rapidly than in the former experi-
ment. This last fact indicates that intermediate CH3-
CH2C(dNCH2Ph)OPBu3+ ArCOO- (6b) is converted into
acyl imidate CH3CH2C(dNCH2Ph)OCOAr (7b) more
readily than intermediate 6a is converted into imidoyl
chloride 7a . Apparently, the aza-Wittig-like product 7b
is not formed directly either.
It can be seen that only phosphorus-containing inter-
mediates of type 5 and 6 appear to have a half-life long
enough to be detected under the above-mentioned condi-
tions. The differences between using an acid chloride or
an active mixed anhydride are few: at low temperature
5a predominates for a long time in the former case while
it is 6b the most abundant species detected in the latter.
However, the series of events are similar. This agrees
with the fact that, in practice, provided that the reaction
time is over 5 h (compare, e.g., entries 1 and 2 of Table
1), the desired amide (1) is obtained in similar yields.13
Reaction of Bu3PdNCH2Ph with isobutyryl chloride,
(CH3)2CHCOCl, was also monitored at -20 °C in CDCl3.
In the proton-decoupled 31P NMR spectrum, it was
observed that the singlet at δ 22.3 due to the starting
phosphazene disappeared immediately on addition of ca.
1 equiv of the acid chloride, while a new signal at δ 66.7
appeared. Then, this signal diminished quite rapidly and
new peaks grew at δ 53 and, more slowly, at δ 45 (see
the supporting information). Essentially, the reaction
sequence was identical to that shown in Scheme 3.
Also, the reaction of Bu3PdNCH2Ph with isobutyryl
2,4,6-trichlorobenzoyl anhydride, (CH3)2CHCOOCOAr,
was parallel to that of phosphazene and CH3CH2-
COOCOAr.
On the other hand, when the reaction of Bu3PdNCH2-
Ph with PhCH2COCl was monitored by 31P NMR in
CDCl3 at -20 °C as in the preceding cases, significant
differences were noted. Besides two small signals at δ
69.5 and 67.5, a broad peak at δ 53.4 and another peak
at δ 29.8 (which was confirmed independently to cor-
respond to C-phosphonium salt 4a , by addition of a
sample of pure compound) appeared immediately upon
mixing; the last signal did not increase upon time.14 The
reaction was repeated starting at -40 °C, but again a
peak at ca. δ 30 (much smaller than in the previous
experiment) was instantaneously observed; this signal
did not increase at the expense of the large signal at δ
53.4 (nor probably of the smaller signals, although this
turned out to be difficult to see by integration) when the
sample was warmed. At -65 °C, 4a was hardly noted;
apart from the already mentioned signals at δ 69, 67,
and 53, no other peaks (i.e., no other P-containing
intermediates) were detected. On the other hand, when
the reaction was carried out at 50 °C, the peak due to
removal of the solvent in vacuo) with CH3CH2COCl was
monitored at -20 °C in CDCl3. In the proton-decoupled
31P NMR spectrum, the singlet at δ 22.3 (CDCl3 solution
of the starting phosphazene) disappeared immediately
on addition of a CDCl3 solution, previously cooled at -20
°C as well, of the acid chloride (ca. 1 equiv), while an
intense peak at δ 66.7 and a small peak at δ 53 was
observed. The strong peak began to diminish, and the
signal at ca. δ 53 grew slowly (in fact, there were two
peaks, at 53.5 and 53.4, in variable ratios);10 minor
amounts of Bu3PdO (δ 44.7) were also noted. On
warming the probe temperature to rt this series of
consecutive reactions took place more rapidly, as the peak
at δ 66.7 eventually disappeared, that at δ 53.5 dimin-
ished (but not so that at δ 53.4), and the area due to the
Bu3PdO signal became the largest one. Taking also the
1
analysis of the corresponding 1H NMR spectra and a H-
31P HETCOR experiment into account, the series of
events may be explained as follows (see Scheme 3): the
first intermediate detected at low temperature (5a ) is
converted, through a N-to-O migration of the phospho-
nium group, into 6a (only one set of proton signals is
seen), which in turn slowly gives Bu3PdO and imidoyl
chloride 7a .11,12
The reaction of Bu3PdNCH2Ph with propanoyl 2,4,6-
trichlorobenzoyl anhydride, CH3CH2COOCOAr (Ar )
2,4,6-trichlorophenyl), was also investigated at -20 °C.
The proton-decoupled 31P NMR spectra looked like those
of the above-mentioned reaction of Bu3PdNCH2Ph with
CH3CH2COCl, but differences were noted: (i) the signal
that can be attributed to the N-phosphonium species of
(10) The ratio between these two peaks depended on propanoyl
chloride purity (whether freshly distilled or not), on the time elapsed
before registering the first spectrum, and on the addition order of the
reagents and their relative excess.
(11) By adding a drop of water into the NMR tube and shaking, all
the signals attributed to 5a , 6a , and 7a disappeared, while the
formation of the desired amide was clearly seen by 1H NMR at the
same time that the 31P signal of Bu3PO increased. Only a set of signals
(δP at 53.4; double doublet at δΗ 4.15 and NH proton at δΗ 8.07)
remained, which we confirmed to belong to small amounts of phos-
phazenium salt (much more stable to hydrolysis than O-phosphonium
(13) On the other hand, if reactions between Bu3PdNCH2Ph and
CH3CH2COX are quenched with water at much shorter reaction times,
when intermediates of type 5 may predominate, the amide yields are
much lower (more polar products were mainly obtained).
salts) by addition of an authentic sample of PhCH2NHPBu3 Cl-,
+
prepared independently. The percentage of this phosphazenium salt
in the reaction mixture was estimated to be around 10% (by 1H NMR).
(12) When trimethylphosphine derivative Me3PdNCH2Ph (δCH2-
Ph 4.22, J HP ) 26.2 Hz; δMe 1.41, J HP ) 12.6 Hz), instead of
Bu3PdNCH2Ph, was treated with CH3CH2COCl in CDCl3, parallel
results were noted by 1H NMR: (i) instantaneous formation of CH3-
(14) By adding of drop of water to this sample and shaking, the
signal at δ 29.8 did not change either, those at δ 69.5 and 67.5
disappeared, and that at δ 53.4 diminished significantly (but not
completely), while the signal due to Bu3PO became the largest peak,
as expected. The peak remaining at δ 53.4 can be attributed to a
phosphazenium salt, as in the preceding case (see ref 11). In parallel
CH2CON(PMe3 Cl-)CH2Ph (δCH2Ph 5.20, J HP ) 15.0 Hz; δMe 2.50,
+
+
+
J HP ) 14.5 Hz); (ii) slow appearance of CH3CH2C(dNCH2Ph)OPMe3
experiments, the phosphazenium salt PhCH2NHPBu3 Cl- was iso-
Cl- (δCH2Ph 4.15, J HP ) 16.5 Hz; δMe 1.92, J HP ) 14.0 Hz); (iii) even
slower appearance of CH3CH2C(dNCH2Ph)Cl (δCH2Ph 4.68) and Me3-
PO (δMe 1.49, J HP ) 12.6 Hz).
lated in 12% yield by column chromatography. We believe that part of
the phosphazenium salt arises from the presence of acid impurities in
the (water-sensitive) acid chloride.

Notes
J . Org. Chem., Vol. 61, No. 16, 1996 5641
Sch em e 4
4a , was more intense than in previous experiments (see
the supporting information).
By mixing Bu3PdNCH2Ph and PhCH2COCl solutions,
again in a 1:1 molar ratio, in a bath at -78 °C and
1
allowing then the sample to warm to rt in the H NMR
probe, N-phosphonium species 8a , O-phosphonium spe-
cies 9a , and imidoyl chloride 10a (but not 4a ) were clearly
seen. In a subsequent experiment, to an identical reac-
tion mixture, at -78 °C as well, 0.5 molar equiv of
phosphazene was further added before registering the
spectra, in order to check if an excess of phosphazene
when the reaction had already started showed any effect,
but no significant changes could be noted in relation to
the preceding experiment (8a and 9a did not disappear
much more rapidly; only small signals attributable to 4a
could be seen).15 These experiments corroborate the fact
that the abnormal C-phosphonium salts have less chance
by carrying out the reactions at low temperature. They
also point out that a relative excess of phosphazene,
although it is essential at the beginning for the formation
of 4, is not important afterwards and that most 4a does
not come from the reaction of 8a or 9a with phosphazene.
In short, it seems that most of the anomalous product
4a does not arise from the other P-containing species that
are seen by NMR, but it is mainly formed at the very
beginning, from a previous, short-lived reactive interme-
diate. (This agrees with the fact that the isolated yields
of 4a are quite similar at very different reaction times,
as shown in entries 9, 13, and 14 of Table 1.) Accordingly,
a plausible mechanism for this reaction is suggested in
Scheme 4, which shows that the precursor of 4a could
be one of the first reactive intermediates that can be
reasonably written (see, e.g., RI-1 in Scheme 4). In the
presence of small amounts of a base such as a phospha-
zene, RI-1 would lose HCl to afford RI-2, which would
undergo an N-to-C phosphonium-group rearrangement
to give RI-3.
Sch em e 5
Similar results were noted by treating Bu3PdNCH2-
Ph with PhCH2COOCOAr, even though as in the former
experiments with mixed anhydrides the relative areas
of the peaks along the reaction time were not the same
as with acyl chlorides.
Comparison of Schemes 3 and 4 suggests that the main
basis for the differences is the tendency of reactive
intermediates arising from PhCH2COCl to lose HCl,
which may be temporarily trapped by unreacted phospha-
zene present in the reaction flask. There is an obvious
connection between this fact and the higher acidity of
the PhCH2 methylene protons as compared to the RCH2
ones. That the pathway leading to 4a becomes predomi-
nant above rt may be accounted for if the rate of the HCl
elimination step is accelerated by rising the temperature
much more than the formation of the N-phosphonium
intermediate 8a .
Finally, we would like to discuss another possible route
to the abnormal products (4a -d ). Since the tendency of
arylacetic acid derivatives to give arylketenes in the
presence of tertiary amines is higher than that of
standard aliphatic acid derivatives,16 the possibility that
the generation of phenylketene is the first step of an
alternative pathway leading to 4a -d should not be ruled
out. This is shown in Scheme 5 (route i, or indirectly
steps ii + iii), which summarizes only the possible
pathways toward the abnormal product 4a. Phenylketene
would then react with another molecule of phosphazene
(route iv) to give several species such as those suggested
in the roundabout of Scheme 5 (stereoisomers not shown),
outstanding among them is RI-3. This reactive inter-
(15) When these experiments were repeated with triphenylphos-
phazene Ph3PdNCH2Ph, instead of the tributylphosphazene, the
spectra were analogous although the O-phosphonium species (that
related with 9a ) was much less abundant.
(16) For reviews, see: Schaumann, E.; Scheiblich, S. In Houben-
Weyl Methoden der organischen Chemie; G. Thieme-Verlag: Stuttgart,
1993; E 15/Teil 2. The Chemistry of Ketenes, Allenes and Related
Compounds; Patai, S., Ed.; Wiley: New York, 1980.

5642 J . Org. Chem., Vol. 61, No. 16, 1996
Notes
mediate, through a proton exchange with the protonated
phosphazene, could obviously afford C-phosphonium salt
4a .
phosphazene. However, we have shown that 4a is
formed only in small amounts when the reaction is
carried out in the reverse mode (by dropwise addition of
the phosphazene to the PhCH2COCl solution), by mixing
the substrate and reagent at low temperature, and in
acetonitrile.
However, there are some results that argue against a
significant participation of phenylketene as a precursor
of 4 when phosphazenes and acid derivatives are mixed
in equimolar amounts. In several reactions of PhCH2-
COCl summarized in Tables 1 and 2, we isolated small
amounts of a mixture of relatively nonpolar byproducts,9
whose NMR spectra suggested that the well-known
phenylketene dimers and trimers16 had been formed. A
similar mixture was rapidly and quantitatively formed
when PhCH2COCl was treated with DMAP (but not so
with the less basic and nucleophilic 2,6-lutidine) in CDCl3
at -20 °C; addition of 1.1 equiv of Bu3PdNCH2Ph within
5 min did not give 3 and/or 4a at all, even after 5 h at rt,
but a complex mixture of relatively nonpolar products
(apparently, phenylketene dimers as well as their reac-
tion products with phosphazene; dimerization is so rapid
that phosphazene is unable to trap substantial amounts
of monomer under these conditions). When PhCH2COCl
was added to a well-stirred mixture of DMAP (1 equiv)
and Bu3PdNCH2Ph (1 equiv), as in entry 14 of Table 1
but in the presence of DMAP, we obtained ca. 30% (w/w)
of relatively nonpolar byproducts, 28% of 3, and 29% of
4a . Thus, under conditions that may favor the formation
of phenylketene the yield of both 3 and 4a decreased. It
seems that those substrate molecules that react first with
DMAPsthose that thereafter may be converted into
phenylketene through this routesdo not enter into the
pathway leading to 3 and/or 4a . On the other hand,
when PhCH2COCl was added to a well-stirred mixture
of 2,6-lutidine (1 equiv) and Bu3PdNCH2Ph (1 equiv), we
isolated ca. 5% of nonpolar byproducts, 31% of 3, and 49%
of 4a . Thus, the presence of a sterically hindered
pyridine such as 2,6-lutidine did not increase the per-
centages of phenylketene-derived byproducts but in-
creased the yield of 4a at the expense of 3, in the same
way that an excess of Bu3PdNCH2Ph did (i.e., acting as
the base, B:, shown in Schemes 4 and 5). We interpret
that, in the reaction of PhCH2COCl with phosphazenes
in a 1:1 molar ratio, phenylketene is usually a minor
product, as routes i and iii (Scheme 5) are not important
in relation to others; it is expected that phenylketene
reacts instantaneously and unselectively with any nu-
cleophile present in the medium (phenylketene precursor,
remaining phosphazene, etc.) to afford dimers, P-contain-
ing intermediates, etc., but, as it is generated in relatively
small amounts, most 3 and 4a do not arise from it.
In general, Bu3PdNCH2Ph reacts firstly and chiefly
as a nucleophile, with attack at the COX carbon atom of
RR′CHCOX to give RR′CHCON(PBu3+ X-)CH2Ph, rather
than as a base. However, for substrates PhCH2COX
(with a relatively acid R-CH proton, which can give
intermediates with a double bond conjugated with the
aromatic ring), the presence of a relative excess of
phosphazene in the medium catalyzes the anomalous
reaction, which begins with an elimination reaction.
Elimination of HX may occur from PhCH2COX (Scheme
5, route i), from a short-lived reactive intermediate such
as RI-1 (route vi) and/or from N-phosphonium species
such as 8a (route v). We suggest that most HX is lost
from RI-1. The difference is subtlesit concerns only the
timing of the HX eliminationsbut agrees better with the
available experimental results.
Exp er im en ta l Section
Melting points are uncorrected. Crude products were purified
by column cromatography on silica gel of 230-400 mesh (flash
chromatography). 1H and 13C{1H} NMR spectra were obtained
in CDCl3 at 200 and 50.3 MHz, respectively; chemical shifts are
given in ppm with respect to internal TMS, and J values are
quoted in Hz. Variable-temperature 31P and 1H NMR spectra
were recorded in CDCl3 at 121.4 MHz (with respect to an
external standard of 85% H3PO4) and 300 MHz, respectively.
Infrared spectra were obtained in KBr; only the most significant
absorptions, in cm-1, are indicated. Microanalyses were per-
formed by the Serveis Cient´ıfico-Te`cnics (Universitat de Barce-
lona). All the solvents were distilled from an appropriate drying
agent and stored under nitrogen atmosphere. All the reactions
were carried out under an atmosphere of nitrogen or argon, using
dry glassware.
P h osp h a zen e P r ep a r a tion . To a stirred solution of benzyl
azide (133 mg, 1.0 mmol) in benzene (5 mL) was added
tributylphosphine (250 µL, 202 mg, 1.0 mmol) via syringe. The
mixture was stirred for 2 h at rt, and the product was used
without purification.
Rea ction of P h osp h a zen es w ith Acyl Ch lor id es. Typ i-
ca l P r oced u r e. To a stirred solution of (N-benzyl)tributylphos-
phazene [N-(tributylphosphoranylidene)benzylamine] (307 mg,
1.0 mmol), in benzene (5 mL) cooled at 5 °C, under nitrogen,
was added phenylacetyl chloride (130 µL, 155 mg, 1.0 mmol) in
benzene (1 mL). The reaction mixture was allowed to warm to
rt and stirred for 5 h. The solvent was removed in vacuo
(procedure A), and the residue was separated by chromatography
on silica gel (95:5 CH2Cl2-MeOH) to afford N-benzylphenylac-
etamide (3) (68 mg, 30%) and then [R-(N-benzylcarbamoyl)]-
benzyltributylphosphonium chloride (4a ) (168 mg, 36%).9 Al-
ternatively (procedure B), the reaction mixture was poured into
a separatory funnel containing an excess of CH2Cl2 and was
washed with water; drying of the organic layer, evaporation of
the solvent, and separation of the residue by chromatography
as above afforded practically identical results.
Rea ction of P h osp h a zen es w ith Mixed An h yd r id es.
Typ ica l P r oced u r e. To a stirred solution of (N-benzyl)-
tributylphosphazene (307 mg, 1.0 mmol), in benzene (5 mL)
cooled at 5 °C, under nitrogen, was added a solution of phenyl-
acetic 2,4,6-trichorobenzoic anhydride (343 mg, 1.0 mmol),
prepared according to the method of Yamaguchi et al.,17 in
benzene (1 mL). The reaction mixture was allowed to warm to
rt and stirred for 5 h. Separation of the crude product following
any of the two above-mentioned procedures afforded 3 (55 mg,
Con clu sion s
Phosphazenes react similarly with acid chlorides and
mixed anhydrides to give almost identical yields of
standard carboxamides after the workup. Closely related
intermediates are detected by 31P and 1H NMR at low
temperature, although some differences have been noted
in the relative areas of the NMR signals that indicate
that the counterions (Cl-, ArCOO-) may also play a role
in the reaction course.
Trialkylphosphazenes and triphenylphosphazenes re-
act similarly with carboxyl-activated derivatives, al-
though amide yields are systematically better with the
former.
Phenylacetic acid derivatives tend to give anomalous
+
products 4a -d (involving a N-to-C transfer of the PR3
(17) Inanaga, J .; Hirata, K.; Saeki, H.; Katsuki, T.; Yamaguchi, M.
Bull. Chem. Soc. J pn. 1979, 52, 1989.
group), especially when there is a relative excess of

Notes
J . Org. Chem., Vol. 61, No. 16, 1996 5643
24%) and [R-(N-benzylcarbamoyl)]benzyltributylphosphonium
2,4,6-trichlorobenzoate (4b) (258 mg, 40%).
Yields of the amides and C-phosphonium salts obtained in
every case are summarized in Table 1. Physical and spectro-
scopical data are as follows:
7.23 (2 H, s), 6.65 (1 H, d, J HP ) 12.6), 4.49 (1 H, dd, J ) 7.1,
14.8), 4.27 (1 H, dd, J ) 5.3, 14.8), 2.30-2.05 (6 H, m), 1.45-
1.33 (12 H, m), 0.87 (9 H, t, J ) 7.1); 13C NMR δ 169.1, 166.3
(J CP ) 1.9), 138.3, 138.2, 135.0, 134.2, 131.3 (J CP ) 6.0), 129.4
(J CP ) 3.2), 129.0 (J CP ) 2.7), 128.7 (J CP ) 6.6), 127.8, 127.6,
127.3, 127.1, 44.5 (J CP ) 48.9), 43.6, 27.4 (J CP ) 64.8), 23.6 (J CP
) 25.7), 18.4 (J CP ) 45.4), 13.4; 31P NMR δ 29.8; IR 3195, 1670,
1620. Anal. Calcd for C34H43Cl3NO3P: C, 62.72; H, 6.66; N,
2.15. Found: C, 62.36; H, 6.42; N, 2.27.
[ r - ( N - B e n z y l c a r b a m o y l ) ] b e n z y l t r i p h e n y l -
p h osp h on iu m ch lor id e (4c): mp 206.8-208.2 °C (lit.1 mp 207-
208 °C). Spectroscopical data (1H, 13C, and 31P NMR) agree with
those reported by Molina et al. (compound 5i).1
N-Ben zylpr opion am ide [N-(ph en ylm eth yl)pr opan am ide,
1]: mp 50.4-52.1 °C (lit.18 mp 52-53 °C); 1H NMR δ 7.31-7.26
(5 H, m), 5.78 (1 H, br s), 4.44 (2 H, d, J ) 5.8), 2.25 (2 H, q, J
) 7.7), 1.18 (3 H, t, J ) 7.7); 13C NMR δ 174.4, 138.4, 128.7,
127.8, 127.5, 43.6, 29.7, 9.9; IR 3300, 1650.
N-Ben zylisob u t yr a m id e [2-m et h yl-N-(p h en ylm et h yl)-
p r op a n a m id e, 2]: mp 84.1-86.3 °C (lit.19 mp 86.7-87.5 °C);
1H NMR δ 7.32-7.27 (5 H, m), 5.81 (1 H, br s), 4.42 (2 H, d, J
) 5.7), 2.39 (1 H, hept, J ) 6.9), 1.17 (6 H, d, J ) 6.9); 13C NMR
δ 176.8, 138.5, 128.6, 127.7, 127.4, 43.4, 35.6, 19.7; IR 3300, 1650.
N-Ben zylp h en yla ceta m id e [N-(p h en ylm eth yl)ben zen e-
a ceta m id e, 3]: mp 102.8-103.1 °C (lit.20 mp 104 °C); 1H NMR
δ 7.31-7.19 (10 H, m), 5.82 (1 H, br s), 4.40 (2 H, d, J ) 5.8),
3.61 (2 H, s); 13C NMR δ 170.9, 138.0, 134.7, 129.4, 129.0, 128.6,
127.4, 127.2, 126.8, 43.7, 43.5; IR 3300, 1640.
[r-(N-Ben zylca r b a m oyl)]b en zylt r ib u t ylp h osp h on iu m
ch lor id e (4a ): mp 138.4-140.1 °C; 1H NMR δ 10.26 (1 H, dd, J
) 5.5, 7.2), 7.8-7.7 (2 H, m), 7.3-7.2 (8 H, m), 6.64 (1 H, d, J HP
) 12.8), 4.48 (1 H, dd, J ) 7.2, 14.6), 4.26 (1 H, dd, J ) 5.5,
14.6), 2.33-2.02 (6 H, m), 1.44-1.34 (12 H, m), 0.86 (9 H, t, J )
7.3); 13C NMR δ 166.2 (J CP ) 1.8), 138.2, 131.2 (J CP ) 5.8), 129.4
(J CP ) 3.4), 128.9 (J CP ) 2.5), 128.6 (J CP ) 6.5), 127.9, 127.5,
127.2, 44.7 (J CP ) 48.2), 43.7, 27.5 (J CP ) 64.8), 23.4 (J CP ) 25.9),
18.4 (J CP ) 45.4), 13.3; 31P NMR δ 29.8; IR 3190, 1665. Anal.
Calcd for C27H41ClNOP: C, 70.18; H, 8.94; N, 3.03. Found: C,
69.79; H, 8.63; N, 2.78.
[ r - ( N - B e n z y l c a r b a m o y l ) ] b e n z y l t r i p h e n y l -
p h osp h on iu m 2,4,6-tr ich lor oben zoa te (4d ): mp 117.5-119.1
1
°C; H NMR δ 10.48 (1 H, dd, J ) 5.2, 7.2), 8.10 (1 H, d, J HP
)
12.5), 7.8-7.2 (25 H, m), 7.36 (2 H, s), 4.44 (1 H, dd, J ) 7.2,
15.0), 4.17 (1 H, dd, J ) 5.2, 15.0); 13C NMR δ 168.4, 166.1 (J CP
) 1.7), 138.4, 137.9, 135.0, 134.7 (J CP ) 9.5), 134.4 (J CP ) 2.3),
134.1, 131.1 (J CP ) 5.9), 129.6 (J CP ) 12.6), 129.0 (J CP ) 3.1),
128.6 (J CP) 2.5), 128.5 (J CP ) 6.3), 128.2, 127.7, 127.2, 126.7,
117.9 (J CP ) 84.9), 47.7 (J CP ) 51.1), 43.5; 31P NMR δ 20.6; IR
3200, 1675, 1620. Anal. Calcd for C40H31Cl3NO3P: C, 67.57;
H, 4.39; N, 1.97. Found: C, 67.29; H, 4.56; N, 2.08.
Ack n ow led gm en t. This work has been supported
by the Comisio´n Interministerial de Ciencia y Tecnolo-
g´ıa (CICYT) in connection with the Projects FAR90-0349
and SAF93-0201. A doctorate studentship to I.B. from
the Ministerio de Educacio´n y Ciencia (1991-94) is
acknowledged as well. J .V. thanks his colleague, Prof.
Pedro Molina, Universidad de Murcia, for the exchange
of information.
[r-(N-Ben zylca r b a m oyl)]b en zylt r ib u t ylp h osp h on iu m
2,4,6-tr ich lor oben zoa te (4b): mp 62.3-65.6 °C; 1H NMR δ
10.34 (1 H, dd, J ) 5.3, 7.1), 7.8-7.7 (2 H, m), 7.3-7.2 (8 H, m),
Su p p or tin g In for m a tion Ava ila ble: Representative pro-
ton-decoupled 31P NMR spectra of the reactions of Bu3PdNCH2-
Ph with isobutyryl chloride and with phenylacetyl chloride (4
pages). This material is contained in libraries on microfiche,
immediately follows this article in the microfilm version of the
journal, and can be ordered from the ACS; see any current
masthead page for ordering information.
(18) Kushner, S.; Cassell, R. I.; Morton, J .; Williams, J . H. J . Org.
Chem. 1951, 16, 1283. Kita, Y.; Akai, S.; Ajimura, N.; Yoshigi, M.;
Tsugoshi, T.; Yasuda, H.; Tamura, Y. J . Org. Chem. 1986, 51, 4150.
(19) Dermer, O. C.; King, J . J . Org. Chem. 1943, 8, 168. Also see:
Maran, F.; Vianello, E.; D’Angeli, F.; Cavicchioni, G.; Vecchiati, G. J .
Chem. Soc., Perkin Trans. 2 1987, 33. Consiglio, G.; Kolla´r, L.; Ko¨lliker,
R. J . Organomet. Chem. 1990, 396, 375.
(20) Kotera, A.; Shibata, S.; Sone, K. J . Am. Chem. Soc. 1955, 77,
6183. Also see: Matsumoto, K.; Hashimoto, S.; Uchida, T.; Okamoto,
T.; Otani, S. Chem. Ber. 1989, 122, 1357.
J O951951B