Reactive species formed from N-benzyloxycarbonyl α-aminophosphonochloridates and triethylamine: Probable identity and implications for synthesis

Article abstract of DOI:10.1039/b109231f

The sterically congested phosphonochloridate BnOCONHCMe₂P(O)(OMe)Cl reacts rapidly with Et₃N to give what is thought to be an oxazaphospholine oxide 7 (and Et₃NHCl); unhindered BnOCONHCH₂P(O)(OMe)Cl seems to rea

Full text of DOI:10.1039/b109231f

Reactive species formed from N-benzyloxycarbonyl
a-aminophosphonochloridates and triethylamine: probable identity and
implications for synthesis†
Paul M. Cullis and Martin J. P. Harger
Department of Chemistry, University of Leicester, Leicester, UK LE1 7RH
Received (in Cambridge, UK) 10th October 2001, Accepted 18th December 2001
First published as an Advance Article on the web 13th February 2002
The sterically congested phosphonochloridate BnO-
CONHCMe2P(O)(OMe)Cl reacts rapidly with Et3N to give
what is thought to be an oxazaphospholine oxide 7 (and
Et3NHCl); unhindered BnOCONHCH2P(O)(OMe)Cl seems
to react in the same way, in which case the product is not a
phosphonylammonium salt as has been suggested.
limiting step would involve nucleophilic attack at phosphorus
and would be subject to steric hindrance. If there is intra-
molecular nucleophilic catalysis, however, a reactive cyclic
intermediate will be formed in the rate-limiting step and the
sterically-sensitive intermolecular nucleophilic attack at phos-
phorus will occur only in a subsequent fast step (Scheme 2).
As phosphonopeptides have become important synthetic targets
so coupling reactions involving a-aminophosphonic acid deriv-
atives have attracted attention.1–3 Hirschmann, Smith et al.3
have examined the reactivity of the N-protected a-aminophos-
phonochloridate 1 in phosphonylation (alcohols and amines),
with emphasis on the formation of a new and more reactive
phosphonylating agent with Et3N and its reliance on intra-
molecular catalysis by the carbamate group.4
The reaction of 1 with Et3N is accompanied by a rather
dramatic downfield shift of the 31P NMR signal (dP 35.6 ? 44.7
in CDCl3).3 This, and also the change in selectivity for O vs. N
nucleophiles, was attributed to the formation of a phosphonyl-
ammonium salt 2.3 The remarkable ability of chloridate 1 to
form the salt was rationalised in terms of electrophilic catalysis
by the carbamate group, i.e. activation of the PNO group by
hydrogen bonding with the acidic carbamate NH moiety.4 That,
however, would require formation of a five-membered ring
(Scheme 1) and intramolecular hydrogen bonding is generally
rather weak in a ring so small.5 Since intramolecular nucleo-
philic catalysis seems to be responsible for the high reactivity of
the a-amidophosphonate ester 3 towards hydrolysis (acid or
base)6 we felt there might be an alternative explanation. That
possibility has now been explored using the N-benzyloxy-
carbonyl a-aminophosphonochloridates 4 and 5 (X = Cl).‡
Scheme 2
Our chloridate 4 (X = Cl) differs from 1, the substrate
studied by Hirschmann, Smith et al.,3 only in the identity of the
‘spectator’ alkoxy group on phosphorus (OMe in place of OEt)
so it was predictable that it too would react with Et3N and that
the 31P chemical shift would also change quite dramatically (dP
37.3 ? 45.2 in CDCl3) (product A).† By contrast the chemical
shift of the sterically congested chloridate 5 (X = Cl) changed
little with Et3N (dP 47.5 ? 48.3). It might be thought that no
reaction occurs here because of steric hindrance, and that the
small change in dP is merely a medium effect. When less than
one equivalent of Et3N was used, however, two discrete species
(dP 47.4 and 48.2) were clearly present. For the reasons outlined
below we are certain that the new species (product B) is not in
this case a phosphonylammonium salt; rather we think it is
probably a cyclic oxazaphospholine oxide 7 resulting from
nucleophilic participation by the carbamate group (Scheme 2);†
an ammonium salt is formed but only as the by-product
(Et3NHCl). Hirschmann, Smith et al.3 considered a similar
structure for the new phosphonylating agent formed from 1 and
Et3N but rejected it as being inconsistent with their data. It is
indeed difficult to see how the formation of such a cyclic species
could be reliant on intramolecular electrophilic catalysis by the
NH of the carbamate group.4
Scheme 1
Evidence was first sought for nucleophilic participation by
the carbamate group in the absence of base by examination of
the reactions of the chloridates 4 and 5 (X = Cl) with PriOH
(0.08 mol dm23) in CDCl3. The expected esters (X = OPri)
were formed rapidly (!90% completion in 6 min at rt), with rate
enhancements of 103–104 (for 4) or ~ 106 (for 5) relative to
analogous substrates lacking the carbamate group [4 and 5 (X =
Cl) with BnOCONH replaced by Cl]. The fact that the sterically
congested a,a-dimethyl substrate 5 (X = Cl) is as reactive as its
unalkylated counterpart 4 (X = Cl) is mechanistically very
significant. If the BnOCONH group were providing electro-
philic catalysis of substitution (activation of PNO) the rate-
Our reasons for thinking that product B has the structure 7 are
as follows. (i) When < 1 equiv. Et3N is used (in CDCl3) and
5 (X = Cl) is only partially converted into B the only NMR
signal in the region dH 2.0–3.8 is a doublet of quartets (dH 3.11,
dq, J 4 and 7, NCH2Me). The doublet splitting (4 Hz) does not
collapse when the P nucleus is irradiated, implying that the N
atom of Et3N has not formed a bond to phosphorus but has
simply acquired a proton (NH, d 11.4). (When > 1 equiv. Et3N
1
is used the detail of the H NMR signal is obscured by line
broadening, presumably because of exchange between Et3NH+
and Et3N.) (ii) In the 13C NMR spectra of the acyclic
compounds 4 and 5, regardless of the nature of the substituent
on phosphorus (X = OH, Cl, OMe, OPri), the carbamate
carbonyl C atom (dC 155.8 1.1) shows only a small three-bond
coupling to phosphorus (JPC 0–11 Hz). For product B, however,
the corresponding signal (dC 152.9) displays a rather large
coupling (JPC 36 Hz), implying a fundamental change of
structure (although we do not know what value of JPC would be
† Electronic supplementary information (ESI) available: NMR data for
compounds 4 and 5 (X = Cl) and the products formed by them with Et3N.
See http://www.rsc.org/suppdata/cc/b1/b109231f/
458
CHEM. COMMUN., 2002, 458–459
This journal is © The Royal Society of Chemistry 2002

expected for structure 7). There is also a substantial change in
JPC for the a carbon atom, from 152 Hz in 5 (X = OMe) and
134 Hz in 5 (X = Cl) to 107 Hz in B. (iii) In the IR spectrum
(CDCl3) of 5 (X = Cl) the carbamate CNO gives a strong
absorption at 1735 cm21. When Et3N is added and product B is
formed that absorption is replaced by a comparably strong one
at 1660 cm21 attributable to a CNN bond (the N–H peak at 3420
cm21 disappears as well). (iv) The chloridate 5 (X = Cl) also
reacts with Et3N (slight excess) in C6D6 (dP 46.9 ? 47.5); a
solid (Et3NHCl) precipitates and the 1H NMR spectrum shows
that product B is present in solution. However, the NCH2Me
signal for the solution (dH 2.25) is only about a tenth as strong
most unlikely on steric grounds. The other is the failure of their
b-aminophosphonochloridate derivative (structure 31 in ref. 4)
to react with Et3N: it has a fixed conformation about the CO–
NH bond that ensures efficient hydrogen bonding between the
carbamate NH and the PNO group but which precludes
nucleophilic attack at phosphorus by the carbamate CNO
group.
Reinterpretation of the observations of Hirschmann, Smith et
al.3 need not detract from the value of their procedure and
treatment of an a-amidophosphonochloridate with Et3N before
the nucleophile is introduced may become the method of choice.
Our analysis does suggest some limitations however, at least
with sterically hindered chloridates such as 5 (X = Cl).
Intramolecular nucleophilic attack7 is not retarded by the Me
groups on Ca but the product-forming intermolecular reaction
of the resulting cyclic species potentially is. That will be of little
consequence when the intermediate is protonated (as in 6)
because then it is extremely reactive, but when deprotonated (as
in 7) it will be less reactive and product formation may be the
slow step. In that case some of the benefit to be had from
participation of the neighbouring carbamate group will have
been wasted. Thus, for example, the rapid reaction of 5 (X =
Cl) with PriOH is at least an order of magnitude slower if the
chloridate is pretreated with Et3N (!1 equiv). The competing
reaction with traces of moisture tends also to be more serious
when Et3N is used because the less reactive (deprotonated)
intermediate 7 discriminates more strongly against the less
nucleophilic (more hindered) OH group of PriOH.§ If moisture
is rigorously excluded, however, enhanced selectivity could be
a real advantage with phosphonate acceptors containing two or
more competing nucleophilic groups.
+
(integral) as it would be if B contained an NEt3 group and
represents little more than the slight excess of the amine.
Taken as a whole the spectroscopic evidence rules out the
possibility that product B is a phosphonylammonium salt; it is
not so conclusive as evidence for a cyclic oxazaphospholine
oxide structure 7 but that would be the expected outcome of
nucleophilic participation by the carbamate group in the
presence of a base if nucleophilic attack at phosphorus is
impaired by steric hindrance (Scheme 2). With an unhindered
chloridate like 1 or 4 (X = Cl) it is possible to envisage
participation being followed by rapid nucleophilic attack and
formation of a phosphonylammonium salt. Our attempts to
obtain spectroscopic confirmation have been hampered by the
relative instability (reactivity) of the lowfield species (product
A) (dP 45.2) formed from 4 (X = Cl) and Et3N but the following
seem to us significant. (i) The ‘carbonyl’ C atom (dC 155.1) in
product A has a coupling to phosphorus (JPC 42 Hz) as large as
that of the imine C atom in B and the coupling of Ca (JPC 104
Hz) is again substantially reduced relative to the acyclic
compounds 4 (X = OMe or Cl) (JPC 158 or 142 Hz). (ii) The
IR spectrum of product A shows a strong absorption at 1670
cm21; a moderately strong peak at 1730 cm21 (CNO) was also
present in our spectrum but that could well have resulted from
partial hydrolytic regeneration of the carbamate group (a
substantial amount of the pyrophosphonate was evident in the
31P NMR spectrum of the solution). (iii) In C6D6 the addition
of Et3N (slight excess) again produced a lowfield species (dP
46.2), presumably product A, but also a precipitate of Et3NHCl.
The 1H NMR spectrum of the solution was poor because of the
insoluble material but the integrals of the NCH2CH3 signals (dH
2.4 and 0.95) were clearly only about a fifth of what they should
be for a compound having an +NEt3 group.
Notes and references
‡ The phosphonochloridates 4 and 5 (X = Cl) were prepared from the
corresponding dimethyl phosphonates (X = OMe) by partial hydrolysis (2
mol dm23 NaOH, rt, 40 min) (for 4) or demethylation (2 mol dm23 NaI in
acetone, 55 °C, 18 h) (for 5) and treatment of the resulting monoesters (X
= OH) with SOCl2. Compounds 4 and 5 (X = OMe, OH, Cl, OPri) were
characterised by NMR (1H, 13C, 31P) and IR spectroscopy and except for the
chloridates (X = Cl) by mass spectrometry (including accurate mass
measurement for new compounds). The chloridates were obtained as
crystalline solids but they were extremely sensitive to moisture and any
manipulation (in the absence of SOCl2) generally introduced some of the
anhydride (pyrophosphonate) [31P NMR: 2 peaks (diastereoisomers) at ca.
17 or 23 ppm].
§ With a 10+1 mixture of PriOH and MeOH the chloridate 5 (X = Cl) gives
similar amounts of the two possible esters (X = OPri or OMe) in the
absence of base but if the chloridate is pretreated with Et3N the reaction
becomes more selective and hardly any of the isopropyl ester ( < 1%) is
formed.
Rather than confirming product A as a phosphonylammon-
ium salt these observations actually suggest (but do not prove)
a structure analogous to that of product B, i.e. an oxazaphos-
pholine oxide 7 with H atoms in place of the Me groups at Ca.
Moreover the N-methyl derivative of 4 (X = Cl) shows no sign
of reaction with Et3N in the time taken for the parent compound
to react completely ( < 10 min); since the N-methyl derivative
is no less reactive in substitution with PriOH (no base) it would
be surprising if the reaction of 4 (X = Cl) with Et3N were
formation of a phosphonylammonium salt. Smith, Hirschmann
et al.4 likewise noted the failure of the N-methyl derivative of
their chloridate 1 to react with Et3N. They supposed the reason
to be the lack of activation of the PNO group by intramolecular
hydrogen bonding and a consequent loss of the ability to form
the phosphonylammonium salt. We think the carbamate group
actually acts as a nucleophile, and that where possible (4 and 5)
the resulting cyclic intermediate is deprotonated by Et3N (6 ?
7 in Scheme 2) (cf. oxazolone formation from activated
acylamino acids in peptide synthesis). Where deprotonation is
not possible (N-methyl derivatives), the cyclic intermediate
simply returns to the chloridate.
1 Aminophosphonic and Aminophosphinic Acids, ed. V. P. Kukhar and H.
R. Hudson, Wiley, Chichester, 2000.
2 N. A. Jacobsen and P. A. Bartlett, J. Am. Chem. Soc., 1981, 103, 654.
3 R. Hirschmann, K. M. Yager, C. M. Taylor, J. Witherington, P. A.
Sprengeler, B. W. Phillips, W. Moore and A. B. Smith, III, J. Am. Chem.
Soc., 1997, 119, 8177.
4 A. B. Smith, III, L. Ducry, R-M. Corbett and R. Hirschmann, Org. Lett.,
2000, 2, 3887.
5 M. B. Smith and J. March, MarchAs Advanced Organic Chemistry, 4th
edn, Wiley, New York, 2001, p. 99 N. S. Isaacs, Physical Organic
Chemistry, 2nd edn, Longman, Harlow, 1987, p. 73.
6 J. Rahil and R. F. Pratt, J. Chem. Soc., Perkin Trans. 2, 1991, 947;
Intramolecular nucleophilic catalysis has more often been noted for
systems in which the CO–NH sequence (relative to the P = O group) is
reversed: R. Kluger and J. L. W. Chan, J. Am. Chem. Soc., 1976, 98,
4913; R. Kluger, J. F. Chow and J. J. Croke, J. Am. Chem. Soc., 1984,
106, 4017; L. A. Reiter and B. P. Jones, J. Org. Chem., 1997, 62,
2808.
Two other observations of Smith, Hirschmann et al.4 are now
more easily understood. One is the reaction of chloridate 1 with
Pri2NEt (HünigAs base): it proceeds in the same way as with
Et3N even though nucleophilic attack at phosphorus now seems
7 B. Capon and S. P. McManus, Neighbouring Group Participation,
Plenum, New York, 1976, vol. 1, ch. 2.
CHEM. COMMUN., 2002, 458–459
459