Nucleophilic and acid catalysis in phosphoramidite alcoholysis

Article abstract of DOI:10.1039/b104910k

Numerous triflates, mesylates, chlorides, trifluoroacetates and tetrazolides of trialkylammonium, pyridinium and azolium ions have been studied as activators for the reaction of N,N,O,O′-tetraisopropylphosphoramidite with propan-2-ol in acetonitrile. The progress of the reactions was followed by ³¹P NMR spectroscopy, and the pK_a values of the activators were determined by a¹³C NMR spectroscopic method based on competing simultaneous protonation of two bases. The salts promoted the reaction both as acids and nucleophiles, the acidity playing a more important role than the nucleophilicity. The Bronsted α value for the general acid catalysis was observed to range from 0.6 to 0.9 and the β_nucl value to be 0.2 (pK_a of the conjugate acid used as the measure of nucleophilicity). Mixtures of neutral azoles or pyridines and weakly acidic ammonium salts were also shown to be useful activators that allow the acidity and nucleophilicity to be tuned independently of each other.

Full text of DOI:10.1039/b104910k

View Article Online / Journal Homepage / Table of Contents for this issue
Nucleophilic and acid catalysis in phosphoramidite alcoholysis
Erkki J. Nurminen,*a Jorma K. Mattinena and Harri Lönnbergb
a Department of Organic Chemistry, Åbo Akademi University, FIN-20500 Åbo, Finland
b Department of Chemistry, Turku University, FIN-20014 Turku, Finland
2
Received (in Cambridge, UK) 5th June 2001, Accepted 17th August 2001
First published as an Advance Article on the web 11th October 2001
Numerous triflates, mesylates,† chlorides, trifluoroacetates and tetrazolides of trialkylammonium, pyridinium and
azolium ions have been studied as activators for the reaction of N,N,O,OЈ-tetraisopropylphosphoramidite with
propan-2-ol in acetonitrile. The progress of the reactions was followed by 31P NMR spectroscopy, and the pKa
values of the activators were determined by a 13C NMR spectroscopic method based on competing simultaneous
protonation of two bases. The salts promoted the reaction both as acids and nucleophiles, the acidity playing a more
important role than the nucleophilicity. The Brønsted α value for the general acid catalysis was observed to range
from 0.6 to 0.9 and the βnucl value to be 0.2 (pKa of the conjugate acid used as the measure of nucleophilicity).
Mixtures of neutral azoles or pyridines and weakly acidic ammonium salts were also shown to be useful activators
that allow the acidity and nucleophilicity to be tuned independently of each other.
general acid catalysis have been determined.16 In acetonitrile,
Introduction
the most frequently used solvent, not much is known of the
Phosphoramidites (1) are phosphitylating agents utilised in
effect of acid strength on catalytic efficiency. The dissociation
automated solid-support DNA synthesis; a nucleoside 3Ј-phos-
phoramidite (1b) is repeatedly reacted with the 5Ј-hydroxy
constants for only a limited number of catalysts are known in
acetonitrile, and the values referring to aqueous solution do not
allow firm conclusions even at semiquantitative level. In addi-
function of the growing support-bound oligonucleotide
chain.1,2 The reaction requires an acidic activator to promote
tion to being an acid catalyst, the activator also participates as a
the displacement of the amino group by the entering nucleo-
nucleophile, although not necessarily before or during the rate-
phile. 1H-tetrazole3 (TH) has become the most widely used
activator in small-scale oligonucleotide synthesis. Although
good coupling yields have been achieved with this activator, it
also suffers from several disadvantages. Tetrazole is rather
expensive, explosive when heated, hygroscopic and sparingly
soluble in acetonitrile, the solvent commonly used in the coup-
ling reaction. These problems have become more pronounced
on attempting to scale-up the process for the synthesis of modi-
fied oligonucleotides necessitated by antisense drug research.4
limiting step. In the absence of alcohol, tetrazole has been
shown to react readily with phosphoramidites giving tetrazolyl-
phosphonites (2) that then easily phosphitylate alcohols.17,18
Consequently, many alternative catalysts have been screened to
find an optimal activator for the phosphoramidite alcoholysis.
These include azoles, such as 5-(4-nitrophenyl)-1H-tetrazole,5
5-ethylthio-1H-tetrazole6 and 4,5-dicyanoimidazole,7 azolium
salts, such as benzimidazolium,8 imidazolium9 and N-methyl-
imidazolium triflates,10 and pyridinium salts, such as pyridin-
ium hydrochloride,11,12 hydrobromide,12 sulfonates,12 tetrafluoro-
borate13 and trifluoroacetate.14 Pyridinium salts have also been
applied together with imidazole.11,14 However, a systematic
search for alternative catalysts has been hampered by insuffi-
cient understanding of the contributions of protolytic and
nucleophilic processes to the activation.
Intermediary accumulation of 2 has been observed during
the tetrazole-promoted alcoholysis of phosphoramidites, and
kinetic evidence for 2 lying on the reaction pathway has been
obtained.15,19 Accordingly, nucleophilic catalysis cannot be
ignored as a factor that possibly affects the applicability of the
activator. To elucidate this nucleophilic contribution in more
detail we now report on studies with salts that potentially may
participate as both acids and nucleophiles in the activation pro-
cess. Many of these salts are actually viable alternative catalysts
for the phosphoramidite alcoholysis, and a systematic study
on their reactivity facilitates the search for an optimal catalyst.
For this purpose, the reactions of a simple model compound,
O,OЈ-diisopropyl N,N-diisopropylphosphoramidite (1a), with
several ammonium, pyridinium and azolium salts have been
studied both in the absence and presence of propan-2-ol in
acetonitrile, and the observed catalytic efficiency has been com-
pared to the nucleophilicity and acidity of the salt.
Results
Phosphoramidite alcoholysis is known to be catalysed by
acids. No reaction takes place in the absence of acid.15 For
ammonium hydrohalide promoted reactions with tert-butyl
alcohol and methanol, a values of 0.13 and 0.65 indicative of
Table 1 lists the dissociation constants determined in aceto-
nitrile for the protolytes employed. Values not obtained from
the literature were measured by a 13C NMR spectroscopic
method based on competing simultaneous protonation of two
different bases (for details see ref. 19). It is worth noting that in
some cases the relative acidities differ considerably from those
† The IUPAC name for triflate is trifluoromethanesulfonate and that for
mesylate is methanesulfonate.
DOI: 10.1039/b104910k
J. Chem. Soc., Perkin Trans. 2, 2001, 2159–2165
This journal is © The Royal Society of Chemistry 2001
2159

View Article Online
Table 1 The dissociation constants (pKBHϩ or pKHA) in acetonitrile for
the protolytes employed
h
Protolyte
pKBHϩ or pKHA
Triethylammonium iona
N,N-Diisopropylammonium ion
N-Methylmorpholinium ionb
Triallylammonium ion
N,N-Diisopropylanilinium ion
N,N-Dimethylanilinium ion
2,6-Lutidinium ionc
Pyridinium ionc
18.5
16.5
15.6
15.2
14.5
8.6
14.4
12.6
10.0
6.8
17.1
17.1
14.3
14.1
10.0
14.5
12.7
8.4
3-Chloropyridinium ionc
2-Chloropyridinium ionc
Imidazolium ion
N-Methylimidazolium ion
Benzimidazolium ion
N-Methylbenzimidazolium ion
1,2,4-Triazolium ion
Tetrazole
Fig. 1 The reaction of 1a with PriOH in the presence of various
trialkylammonium triflates (᭿), mesylates (•) and trifluoroacetates (ꢀ)
in acetonitrile at 20 ЊC: Brønsted dependence of log t1/2 on ϪpKBHϩ of
ammonium ions.
Trifluoroacetic acidd
Methanesulfonic acide
HClf
8.9
5.5
2.6
HBrf
not quite as fast as most of the other initial displacement
products.
Trifluoromethanesulfonic acidg
The 31P NMR spectroscopic data for the substitution prod-
ucts are given in Table 2. It is worth noting that the chemical
shifts of azolylphosphonites, especially those of the positively
charged N-methylazolylphosphonites, seem to depend on the
identity of the salt anion, probably due to electrostatic ion–
ion or ion–dipole interactions. The products of the initial
nucleophilic displacement (2–12) were observed to be highly
reactive, undergoing a rapid reaction upon addition of PriOH,
consistent with the fact that no sign of their intermediary
accumulation could be detected in the presence of alcohol. The
rate of alcoholysis of 1a depended on the nature of the activ-
ator salt, as discussed below in more detail, but not on the
concentration of the alcohol; no deceleration took place on
using less than 10 equiv. of PriOH. The results are summarized
in Tables 3–5.
a
J. F. Coetzee and G. R. Padmanabhan, J. Am. Chem. Soc., 1965,
b
87, 5005. P. Beltrame, G. Gelli and A. Loi, Gazz. Chim. Ital., 1980,
c
110, 491. D. Augustin-Nowacka and L. Chmurzyñski, Anal. Chim.
d
Acta, 1999, 381, 215. T. Jasinski, A. A. El-Harakany, F. G. Halaka
e
and H. Sadek, Croat. Chem. Acta, 1978, 51, 1. T. Fujinaga and
f
I. Sakamoto, J. Electroanal. Chem., 1977, 85, 185. I. M. Kolthoff,
S. Bruckenstein and M. K. Chantooni Jr., J. Am. Chem. Soc., 1961, 83,
g
3927. T. Fujinaga and I. Sakamoto, J. Electroanal. Chem., 1977, 85,
h
185. All values measured by us have statistical errors less than 0.1.
Notation: Many of the values are included in the compendium:
K. Izutsu, Acid-base Dissociation Constants in Dipolar Aprotic Solvents,
Blackwell scientific, Oxford, 1990.
observed in aqueous medium. For instance, tetrazole is only
2 pKHA units weaker an acid than trifluoroacetic acid, and the
latter is barely capable of protonating pyridine. Methane-
sulfonic acid, in turn, is barely sufficiently strong to protonate
N,N-dimethylaniline.
Ammonium salts as activators
The kinetic runs were performed by mixing equal amounts of
1a and the activator salt to obtain an initial concentration of
0.1 mol dmϪ3 for each, and monitoring the progress of the reac-
tion by 31P NMR spectroscopy. Similar measurements were
carried out in excess (usually 10 equiv.) of propan-2-ol (PriOH).
In the absence of alcohol, displacement of the amino group by
the anion or deprotonated cation of the salt usually took place.
Ammonium trifluoroacetates, for example, gave trifuoroacetyl-
phosphonite, and all azolium salts gave the corresponding
azolylphosphonites. In an equimolar mixture of 1a and the
activator, complete displacement of the amino function is pos-
sible only if the entering nucleophile donates a proton upon
nucleophilic attack (cf. reaction I in Scheme 1). Otherwise,
the departing diisopropylamine abstracts the proton from
the acidic activator, decreasing the concentration of the acid
catalyst (II and III in Scheme 1). Hence, N-methylazolium salts
expectedly resulted in only a 50% substitution, while the non-
methylated azolium cations yielded a 100% conversion (see
Scheme 1).
Regardless of the nucleophilicity of the activator salt anion
or the deprotonated cation, an additional substitution product
was repeatedly observed: a 31P NMR signal at 131.0 ppm
(detected before,15 but then unexplained) refers to diisopropyl-
phosphite anhydride, (PriO)2P–O–P(OPri)2 (13).20 This com-
pound may be formed by a nucleophilic attack of diisopropyl
phosphonate, (PriO)2P(O)H, on 1, as depicted in Scheme 2. The
phosphonate is, in turn, produced by the reaction of 1a with
traces of water present. Hence, both the nature of the activator
and the moisture content of the solution affect the extent of
formation of 13. The anhydride also reacts with alcohol but
Ammonium salts were the least efficient of the activators stud-
ied. The reactions were so slow that the differences in rate could
be quantified by 31P NMR spectroscopy. Accordingly, the key
feature of the catalysis, viz. the possible involvement of both an
acidic and a nucleophilic contribution to the catalysis, could
best be examined with these activators (Table 3). On using
ammonium triflates, mesylates or trifluoroacetates as activators,
the catalytic efficiency was increased with the acidity of the
ammonium ion, the Brønsted α value for the alcoholysis being
0.6 (triflates), 0.7 (mesylates) and 0.9 (trifluoroacetates) (all
values
0.1), as illustrated in Fig. 1. If the cation was too
weakly acidic, no reaction took place, as was the case with the
triethylammonium salts.
Measurements in the absence of alcohol showed that
ammonium triflates did not result in detectable accumulation of
a covalent intermediate. The only reaction observed was slow
formation of 13. By contrast, the mesylates and trifluoro-
acetates of all ammonium ions gave a covalent displacement
product as an intermediate. Among triallylammonium salts, the
mesylate and bromide ions resulted in partial substitution,
while chloride, trifluoroacetate and tetrazolide ions instantly
exhibited the expected 50% displacement. Interestingly, the
level of displacement by the anion also to some extent
depended on the acidity of the cation. Triallylammonium
trifluoroacetate resulted in a 43% displacement, while only 30%
and 16% displacements were achieved with the corresponding
N-methylmorpholinium and N,N-dimethylbenzylammonium
salts, respectively. Even these partial substitutions reached their
equilibrium almost instantly. The tetrazolides of all the cations
studied caused a complete (50%) displacement.
2160
J. Chem. Soc., Perkin Trans. 2, 2001, 2159–2165

View Article Online
Table 2 31P NMR spectroscopic data (δ given in ppm and J in Hz) of the substitution products
Compound
δ
Multiplicity
3JPH
Diisopropyl tetrazolylphosphonite (2)
128.1
t
8.6
Diisopropyl imidazolylphosphonite (4)
Diisopropyl benzimidazolylphosphonite (5)
Diisopropyl triazolylphosphonite (6)
126.3/126.9
128.6/128.5
129.7/124.6
129.4/136.0
129.4/138.1
138.5
168.4
186.2
137.1
131.0
br s
br s
br s
br s
br s
t
Diisopropyl N-methylimidazolylphosphonite (7)
Diisopropyl N-methylbenzimidazolylphosphonite (8)
Diisopropyl trifluoroacetylphosphonite (9)
Diisopropyl phosphorochloridite (10)
Diisopropyl phosphorobromidite (11)
Diisopropyl methanesulfonylphosphonite (12)
Diisopropyl phosphite anhydride (13)
8.6
9.5
t
br s
t
tt
9.1
6.3
5.4a
a Value for 5JPH
.
Table 3 Half-lives for the reaction of 1a (0.1 mol dmϪ3) with propan-2-ol (1.0 mol dmϪ3) in acetonitrile at 20 ЊC on using ammonium salts
(0.1 mol dmϪ3) as activators
Activator
Intermediatea
t1/2/sb
N,N-Dimethylbenzylammonium triflate
N-Methylmorpholinium triflate
Triallylammonium triflate
N,N-Diisopropylanilinium triflate
N,N-Dimethylanilinium triflate
N,N-Dimethylbenzylammonium mesylate
N-Methylmorpholinium mesylate
Triallylammonium mesylate
Triallylammonium bromide
Triallylammonium chloride
Triethylammonium trifluoroacetate
N,N-Dimethylbenzylammonium trifluoroacetate
N-Methylmorpholinium trifluoroacetate
Triallylammonium trifluoroacetate
Triethylammonium tetrazolide
—
—
—
—
13000
4500
1800
950
20
100% (P-protonated 1a)
—
—
1300
400
120
200
20
12% (12)
12% (11)
50% (10)
—
16% (9)
30% (9)
43% (9)
—
c
∞
125
20
10
c
∞
N,N-Dimethylbenzylammonium tetrazolide
N-Methylmorpholinium tetrazolide
Triallylammonium tetrazolide
50% (2)
50% (2)
50% (2)
30
15
15
a The level of substitution in absence of alcohol. The identity of the substitution product is given in parentheses. b Half-lives for the formation of 3.
The half-lives for the formation of the intermediate in the absence of alcohol were always less than 5 s (the lower limit for the detection).
c No reaction detected.
Scheme 1 Displacement of the diisopropylamino group of 1a with a protonated azole (reaction I), protonated N-methylazole (II) and an acid anion
(III). The overall stoichiometry is determined by both the actual displacement and the subsequent protolytic equilibrium.
In the presence of alcohol, the catalytic efficiency of the ter-
tiary ammonium salts depended on the nucleophilicity of the
anion (see Fig. 1): the reaction rate was increased in the order
triflates < mesylates < trifluoroacetates < tetrazolides. If the
basicity of the anion is taken as a measure of its nucleophilicity,
a βnucl value of 0.2 0.1 is obtained for the triallylammonium,
N-methylmorpholinium and N,N-dimethylbenzylammonium
salts.
J. Chem. Soc., Perkin Trans. 2, 2001, 2159–2165
2161

View Article Online
Table 4 Half-lives for the reaction of 1a (0.1 mol dmϪ3) with propan-2-ol (1.0 mol dmϪ3) in acetonitrile at 20 ЊC on using pyridinium and azolium
salts (0.1 mol dmϪ3) as activators
Activator
Intermediateb
t1/2/sa
2,6-Lutidinium triflate
2,6-Lutidinium mesylate
2,6-Lutidinium bromide
2,6-Lutidinium trifluoroacetate
Pyridinium triflate
3-Chloropyridinium triflate
2-Chloropyridinium triflate
N-Methylbenzimidazolium triflate
N-Methylimidazolium triflate
1,2,4-Triazolium triflate
Benzimidazolium triflate
Imidazolium triflate
—
100
10
10
50% (12)
35% (11)
50% (9)
—
<5
<5
<5
25
<5
<5
<5
<5
<5
<5
<5
<5
<5
—
100% (P-protonated 1a)
50% (8)
50% (7)
100% (6)
100% (5)
100% (4)
50% (8)
50% (7)
100% (5)
100% (4)
N-Methylbenzylimidazolium trifluoroacetate
N-Methylimidazolium trifluoroacetate
Benzimidazolium trifluoroacetate
Imidazolium trifluoroacetate
a,b See the corresponding footnotes in Table 3. The half-lives for the formation of the intermediate in the absence of alcohol were less than 5 s, except
for 2,6-lutidinium bromide, where a value of 50 s was observed.
Table 5 Half-lives for the reactions of 1a (0.1 mol dmϪ3) in the
Azolium salts proved to be most efficient among the
presence and absence of propan-2-ol (1.0 mol dmϪ3) in acetonitrile at
activators studied in the present work: all alcoholyses and dis-
placement reactions were instant. In the absence of alcohol,
azolium salts always resulted in quantitative displacement, the
20 ЊC on using mixtures of triethylammonium triflate (TEAHϩ triflate,
0.1 mol dmϪ3) and pyridine or azole as activators
azole, not the anion, becoming always bonded to the phos-
phorus atom. According to 31P NMR spectroscopy, even the
N-methylazoles gave covalent azole–phosphorus adducts at
the expected 50% conversion level. While this manifests the
superior nucleophilicity of the azoles, it is as well to bear in
mind that among the azoles used in this work, only tetrazole is
able to react with phosphoramidites when introduced in neutral
form.
Activator
t1/2/s
3000
3000
7200
>18000
<60
3600
TEAHϩ triflate ϩ imidazole
TEAHϩ triflate ϩ imidazole ϩ propan-2-ol
TEAHϩ triflate ϩ benzimidazole
TEAHϩ triflate ϩ triazole
TEAHϩ triflate ϩ N-methylimidazole ϩ propan-2-ol
TEAHϩ triflate ϩ N-methylbenzimidazole ϩ propan-2-ol
TEAHϩ triflate ϩ 2,6-lutidine ϩ propan-2-ol
TEAHϩ triflate ϩ pyridine ϩ propan-2-ol
180
10800
Cooperativity of activators
Since the high catalytic efficiency of azolium and pyridinium
salts prevented the quantification of their relative reactivities, a
series of experiments with the respective nucleophiles, neutral
azoles and pyridines, were carried out in the presence of a very
weak acid, triethylammonium ion. When triethylammonium
triflate was used as an acid, reactions both in the presence and
in the absence of alcohol could be monitored by 31P NMR
spectroscopy. The results are presented in Table 5.
The catalytic efficiency clearly correlates with the basicity of
the nucleophile, and the tertiary N-methylated azoles are more
reactive than their unsubstituted counterparts. The reactivity of
N-methylimidazole is striking even when used together with
such a weak acid as triethylammonium ion. It is worth noting
that the reaction was enabled even by pyridines, though no
covalent adducts between them and phosphorus could be
detected. Finally, the reaction between 1a and imidazole was
followed in the presence of triethylammonium triflate using
imidazole in excess. The presence of imidazole was indispens-
able for the alcoholysis to take place, but the use of imidazole in
excess did not further accelerate the imidazolysis reaction.
Within experimental error, use of up to 5 equiv. of imidazole
had no effect on the reaction rate.
Scheme 2 The formation of diisopropylphosphite anhydride (13).
Pyridinium and azolium salts as activators
The results obtained with the pyridinium and azolium salts are
summarized in Table 4. Generally, pyridinium salts were more
effective catalysts of the alcoholysis than the ammonium
salts. Among them, 2,6-lutidinium salts were least reactive.
Prolonged treatment of 1a with lutidinium triflate resulted in
formation of unidentified by-products exhibiting 13C NMR
signals at 164.5 ppm and 43.6 ppm (doublets with J = 303.5
Hz). The final product of alcoholysis, however, proved to be
the expected phosphite 3. With the strongly acidic 2-chloro-
pyridinium ion, P-protonation took place in the absence of
alcohol.
As with ammonium salts, all anions except the triflate ion
served as nucleophiles in the absence of alcohol: mesylate,
bromide and trifluoroacetate ions all resulted in displacement
of the diisopropylamino ligand on using 2,6-lutidinium ion as
an acid. No direct evidence of intermediates having a covalent
phosphorus–pyridine was obtained. Two observations are,
however, noteworthy. While 2,6-lutidinium and N,N-diiso-
propylanilinium cations are approximately as acidic, their
mesylates behave in a rather different manner. Firstly, 2,6-
lutidinium mesylate converted 1a quantitatively to mesyl phos-
phonate, whereas only a 10% displacement took place on using
N,N-diisopropylanilinium mesylate. Secondly, the alcoholysis
was considerably faster in the presence of 2,6-lutidinium triflate
than in the presence of N,N-diisopropylanilinium triflate.
Discussion
Acid catalysis
The alcoholysis of 1a in MeCN is subject to general acid
catalysis, proceeding only in the presence of a sufficiently strong
proton donor. On using tertiary ammonium ions as an acid, an
ion having a pKa value less than 18.5 was required to obtain
alcoholysis; no triethylammonium salt was able to catalyse the
process, not even triethylammonium tetrazolide, in striking
2162
J. Chem. Soc., Perkin Trans. 2, 2001, 2159–2165

View Article Online
contrast to tetrazolides of more acidic ammonium ions. The
Brønsted α values observed for the general acid catalysis by
ammonium, pyridinium and azolium ions range from 0.6 to 0.9,
suggesting a rather advanced proton transfer in the transition
state.
The acidity of the catalyst may also affect the level of substi-
tution in the absence of alcohol, as in the case of ammonium
trifluoroacetates. This could be attributed to the protolytic equi-
librium between the acid catalyst and the amine replaced by the
nucleophile (see Scheme 1). If the equilibrium constant of the
substitution is close to unity, a strong acid that effectively traps
the amine may drive the reaction to completion. A weak acid
capable only of partial protonation results in an equilibrium
depending also on the pKa of the catalyst. Ammonium tetra-
zolides, in turn, constantly give complete replacement because
the tetrazolylphosphonite is more stable, and due to the high
equilibrium constant of the substitution reaction it is drawn
to completion regardless of the position of the protolytic
equilibrium.
It has been shown previously21 that phosphoramidites pro-
tonated on phosphorus are rather unreactive, and observable
for instance by 31P NMR spectroscopy. According to the pres-
ent results, P-protonated phosphoramidites are formed only by
acids having a pKa value less than 10 (methanesulfonic acid,
and N,N-dimethylanilinium and 2-chloropyridinium ions as
triflates). Therefore, the acid catalysis by proton donors weaker
than this in all likelihood involves protonation of the nitrogen
atom, which cannot be observed directly. According to molecu-
lar modelling,22 N-protonation lengthens and weakens the
P–N bond and gives a highly unstable structure resembling elec-
tronically that of a phosphenium–amine complex. Protonation
of phosphorus, in turn, strengthens and shortens the P–N
bond.
Nucleophilic catalysis
In addition to acid catalysis, the alcoholysis of 1a also seems to
be susceptible to nucleophilic catalysis, as indicated by the
following observations. 1) Most activators displace the diiso-
propylamino ligand from 1a in the absence of alcohol. 2) These
displacement reactions are usually quantitative and clearly
faster than the alcoholysis in the presence of a poorly nucleo-
philic triflate salt of comparable acidity. 3) The products of
these reactions react rapidly with alcohols. 4) The alcoholysis in
the presence of a nucleophilic salt is considerably faster than
that in the presence of a non-nucleophilic salt of equal acid-
ity. For instance, the alcoholysis of 1a in the presence of
triallylammonium triflate exhibited a half-life of 1800 s while
triallylammonium trifluoroacetate and tetrazolide gave the
phosphite 3 instantly, as also does imidazolium triflate in spite
of its lower acidity. Assistance by a good nucleophile also
appears to enable alcoholysis under conditions that are not
otherwise sufficiently acidic for the reaction: triethylammonium
triflate alone does not catalyse the alcoholysis, but addition of
imidazole promotes the reaction with alcohol, and in the
absence of alcohol a covalent substitution product is formed.
As discussed previously,21 the phosphorus atom is protonated
by strong acids (pKa < 10), since in the ground state of the
molecule it is more basic than the nitrogen. The resulting
phosphonium structure can then react with nucleophiles via
an associative reaction path. Acids that are too weak for
P-protonation may protonate the nitrogen atom of the exited
state of 1: on stretching the P–N bond the electron density on
nitrogen increases until it becomes more basic than the phos-
phorus and basic enough even for acids with pKa = 17.
N-Protonation can take place only in the presence of a suitable
nucleophile that in a more or less concerted manner traps the
developing phosphorus cation resulting in substitution by a
dissociative mechanism (Scheme 3).
Measurements with ammonium salts suggest that the nucleo-
Ϫ
philicity of the anions studied increases in the order CF3SO3
<
CH3SO3Ϫ ≈ BrϪ < ClϪ ≈ CF3COOϪ < tetrazolide ion. Whether
triflate anion is capable of acting as a nucleophile towards 1a at
all is uncertain as no displacement products have been detected,
in contrast to the other anions. Nevertheless, this possibility
cannot be ruled out, since covalent and ionic triflate adducts
with other trivalent phosphorus compounds have been detected
by 31P NMR.23 In the absence of a better nucleophile, reaction
proceeding via dialkylphosphite anhydride after nucleophilic
assistance of hydrogen phosphonate traces must also be con-
sidered as an alternative. Interestingly, all neutral nitrogen bases
seem to be better nucleophiles than the anions. With the weakly
acidic triethylammonium ion as a proton donor, the tetra-
zolide anion is unable to promote the alcoholysis, whereas all
neutral azoles and pyridines, when used together with triethyl-
ammonium triflate, enabled the reaction. Although pyridines
have not been observed to yield displacement products with 1a,
they catalyse the alcoholysis in a way that would be best
explained by a nucleophilic contribution. The high catalytic
efficiency of pyridinium triflate compared to that of almost
equally acidic N,N-diisopropylanilinium triflate is a good
indication of this.
Scheme 3 Discussed associative (higher) and dissociative (lower)
reaction pathways triggered by strong and weak acids, respectively.
Although the P-protonation overwhelmingly predominates
in the presence of strong acids, a minor amount of N-proton-
ation may also take place, and hence the dissociative pathway
may compete with the associative reaction of the predominant
ionic form. The alcoholysis initiated by P-protonation with
N,N-dimethylanilinium (pKa = 8.6) and 2-chloropyridinium
(pKa = 6.8) ions is, for example, considerably faster than that
triggered by triflic acid, which may well refer to competition
between the associative and dissociative mechanisms. Although
N,N-dimethylaniline and 2-chloropyridine present in solution
after the protonation are very weak bases, they still are con-
siderably more efficient proton acceptors than the triflate anion.
On addition of a nucleophile, the bases can mediate the proton
transfer from phosphorus to the nitrogen atom of the depart-
ing amine, thus facilitating decomposition by a dissociative
pathway.
One might expect that since the reaction is subject to
nucleophilic catalysis its rate would be dependent on the nucleo-
phile concentration. In order to study this, one should vary the
concentration of the nucleophile while keeping the acid concen-
tration constant, which cannot be accomplished using salts
alone. A possible approach is to use imidazole, one of the
best nucleophiles for this work, together with a weak acid such
as triethylammonium triflate. Use of one equiv. of imidazole
(0.1 mol dmϪ3) as a nucleophile resulted in a slow imidazolysis,
but additions up to 5 equiv. (0.5 mol dmϪ3) did not further
J. Chem. Soc., Perkin Trans. 2, 2001, 2159–2165
2163

View Article Online
Scheme 4 Proposed reaction mechanism. Attack of the nucleophile takes place only after the proton transfer. Magnitude of the partial charges
developing on nitrogen and phosphorus atoms depends on the strength of the acid and reactivity of the nucleophile, respectively. The rate limiting
step is either the proton transfer alone or together with nucleophilic attack, depending on the identity of the nucleophile.
enhance the reaction. In fact, the nucleophilic substitutions of
phosphoramidites have rarely been detected to depend on the
nucleophile concentration.24 Kinetic studies on alcoholysis have
been performed in pseudo-first order conditions16 and the
transaminolysis of phosphoramidites has been reported to be
independent of the amine concentration.25 Reaction of phos-
phoramidites with tetrazole19 is dependent on tetrazole concen-
tration because of the acidic role of the reactant.
ation of an ammonium ion of similar acidity and N-methyl-
imidazole.
Experimental
The synthesis of 1a has been published previously.15 The
activator salts were generated in situ in an NMR tube by
mixing the bases with tetrazole, trifluoroacetic acid, methane-
sulfonic acid and trifluoromethanesulfonic acid. Hydro-
chlorides and hydrobromides were prepared separately from
desired bases and PCl3 or PBr3, respectively. Alcohols and
amines were distilled and stored over molecular sieves (4 Å)
and KOH pellets, respectively. Liquid acids were stored under
airtight septa. Acetonitrile was dried with CaH2 and stored
over it.
The finding that the reaction rate depends on the identity of
the nucleophile suggests that the nucleophile is present in the
transition state. The observed zero-order dependence of
the imidazolysis rate on the nucleophile concentration is in at
least apparent contradiction to this. One may presume that
the nucleophile attacks the phosphorus atom as soon as the
protonation-induced partial positive charge at this site is suf-
ficiently high. The proton transfer leads to a stable intermediate
only when the electron deficient phosphorus centre is trapped
by the nucleophile, otherwise the amine and phosphenium ion
rapidly reassociate to the starting material. Previously, a pre-
association type mechanism was discussed,21 but this would
explain the observed kinetics only if the association equilibrium
was quantitatively on the side of the associate and hence every
protonation would lead to nucleophilic attack and displace-
ment reaction. A better way to rationalise our observations is
simply to think that at excess concentration of a very good
nucleophile the trapping of the developing phosphorus cation is
so fast that the N-protonation facilitated P–N bond cleavage
alone becomes rate-limiting. Attack of a poor nucleophile
involves an energy barrier comparable to that of the proton
donation, resulting in an energy profile of two slightly separ-
ated peaks of almost similar height. The lower activation
energy of a good nucleophile, in turn, is only a shoulder after
the protonation peak. Hence, if the trapping is quantitative,
i.e. if the departed amine is no longer able to compete for the
phosphenium ion with the nucleophilic activator, the activator
is not present in the transition state but participates after
that (Scheme 4). Accordingly, at high concentrations of the
nucleophilic activator the reaction rate becomes independent of
the activator concentration.
In kinetic runs, 1a, triflic acid, methanesulfonic acid and tri-
fluoroacetic acid were introduced neat, other reagents were used
as solutions that were dried with molecular sieves (4 Å). All
reactions were performed in oven-dried, septum-sealed NMR
tubes, into which the reagents were introduced by syringes. The
31P NMR spectroscopic method for following the reactions15
and the 13C NMR spectroscopic method for determining the
dissociation constants19 have been published previously. The
NMR spectra were recorded on a 500 MHz spectrometer
magnet (202.35 MHz for 31P, 125.65 MHz for 13C).
References
1 S. L. Beaucage and R. P. Iyer, Tetrahedron, 1992, 48, 2223.
2 É. E. Nifant’ev and M. K. Grachev, Russ. Chem. Rev. (Engl.
Transl.), 1994, 63, 575; É. E. Nifant’ev and M. K. Grachev,
Usp. Khim., 1994, 63, 602; E. E. Nifantiev, M. K. Grachev and
S. Yu. Burmistrov, Chem. Rev., 2000, 100, 3755.
3 (a) G. I. Koldobskii and V. A. Ostrovskii, Russ. Chem. Rev. (Engl.
Transl.), 1994, 63, 797; G. I. Koldobskii and V. A. Ostrovskii,
Usp. Khim., 1994, 63, 847; (b) S. J. Wittenberger, Org. Prep. Proced.
Int., 1994, 26, 499.
4 (a) S. T. Crooke, in Therapeutic Applications of Oligonucleotides,
R. G. Landes Co., Austin, 1995; (b) S. T. Crooke, in Burger’s
Medicinal Chemistry and Drug Discovery, John Wiley & Sons,
New York, 1995, vol. 1, 863; (c) D. G. Knorre, V. V. Vlassov,
V. F. Zarytova, A. V. Lebedev and O. S. Federova, Design and
Targeted Reactions of Oligonucleotide Derivatives, CRC Press,
Boca Raton, 1994.
Conclusion
Triflates, mesylates, chlorides, bromides, trifluoroacetates and
tetrazolides of trialkylammonium, pyridinium and azolium
ions serve as activators for the phosphoramidite alcoholysis
promoting the reaction both as acids and nucleophiles. In all
likelihood, the reaction proceeds by rate-limiting proton trans-
fer to the departing amide ion, followed by trapping of the
developing, only marginally stable phosphenium ion with a
preassociated nucleophile. The Brønsted α value for the general
acid catalysis falls in the range of 0.6 to 0.9, while the βnucl value
is only 0.2 (pKa of the conjugate acid used as the measure of
nucleophilicity).
5 (a) B. C. Froehler and M. D. Matteucci, Tetrahedron Lett., 1983, 24,
3171; (b) Y. Hayakawa and M. Kataoka, J. Am. Chem. Soc., 1997,
119, 11758.
6 (a) P. Wright, D. Lloyd, W. Rapp and A. Andrus, Tetrahedron Lett.,
1993, 34, 3373; (b) F. Wincott, A. DiRenzo, C. Schaffer, S. Grimm,
D. Tracz, C. Workman, D. Sweedler, C. Gonzales, S. Scaringe and
N. Usman, Nucleic Acids Res., 1995, 23, 2677.
7 C. Vargeese, J. Carter, J. Yegge, S. Krivjansky, A. Settle, E. Kropp,
K. Peterson and W. Pieken, Nucleic Acids Res., 1998, 26, 1046.
8 Y. Hayakawa, M. Kataoka and R. Noyori, J. Org. Chem., 1996, 61,
7996.
9 Y. Hayakawa and M. Kataoka, J. Am. Chem. Soc., 1998, 120,
12395.
Mixtures of neutral azoles or pyridines and weakly acidic
ammonium salts may also be used as activators. Since in this
case the proton donor (ammonium ion) and nucleophile (azole,
pyridine) are independent of each other, both of them may be
tuned in the desired manner to optimise the activation process.
The advantage is that undesired cleavage of acid labile protect-
ing groups may be avoided by using a weakly acidic activator of
high nucleophilicity, such as imidazolium triflate, or a combin-
10 (a) L. Arnold, Z. Tocik, E. Bradkova, Z. Hostomsky, V. Paces
and J. Smrt, Collect. Czech. Chem. Commun., 1989, 54, 523;
(b) Z. Hostomsky, J. Smrt, L. Arnold, Z. Tocik and V. Paces,
Nucleic Acids Res., 1987, 15, 4849.
11 S. M. Gryaznov and R. L. Letsinger, Nucleic Acids Res., 1992, 20,
1879.
12 M. Beier and W. Pfleiderer, Helv. Chim. Acta, 1999, 82, 879.
13 W. K. D. Brill, J. Nielsen and M. N. Caruthers, J. Am. Chem. Soc.,
1991, 113, 3972.
2164
J. Chem. Soc., Perkin Trans. 2, 2001, 2159–2165

View Article Online
I. F. Lutsenko, Zh. Obshch. Khim., 1977, 47, 479; (c) D. Houalla,
M. Sanchez and R. Wolf, Bull. Soc. Chim. Fr., 1965, 2368.
21 E. J. Nurminen, J. K. Mattinen and H. Lönnberg, J. Chem. Soc.,
Perkin Trans. 2, 2000, 2238.
14 A. Eleuteri, D. C. Capaldi, A. H. Krotz, D. L. Cole and
V. T. Ravikumar, Org. Proc. Res. Dev., 2000, 4, 182.
15 E. J. Nurminen, J. K. Mattinen and H. Lönnberg, J. Chem. Soc.,
Perkin Trans. 2, 1998, 1621.
16 (a) É. E. Nifant’ev, M. K. Gratchev, S. Yu. Burmistrov, L. K.
Vasyanina, M. Yu. Antipin and Yu. T. Struchkov, Tetrahedron, 1991,
47, 9839; (b) S. Yu. Burmistrov, N. M. Shchedrova, M. K. Gratchev,
L. K. Vasyanina and É. E. Nifant’ev, Russ. J. Gen. Chem. (Engl.
Transl.), 1995, 65, 500; S. Yu. Burmistrov, N. M. Shchedrova,
M. K. Gratchev, L. K. Vasyanina and É. E. Nifant’ev, Zh. Obshch.
Khim., 1995, 65, 562.
17 (a) S. Berner, K. Mühlegger and H. Seliger, Nucleic Acids Res., 1989,
17, 853; (b) S. Berner, K. Mühlegger and H. Seliger, Nucleosides
Nucleotides, 1988, 7, 763.
22 (a) A. A. Korkin and E. N. Tsvetkov, Russ. J. Gen. Chem. (Engl.
Transl.), 1987, 57, 1929; A. A. Korkin and E. N. Tsvetkov, Zh.
Obshch. Khim., 1987, 57, 2155; (b) A. A. Korkin, A. M. Mebel and
E. N. Tsvetkov, Russ. J. Gen. Chem. (Engl. Transl.), 1988, 58, 900;
A. A. Korkin, A. M. Mebel and E. N. Tsvetkov, Zh. Obshch. Khim.,
1988, 58, 1015; (c) A. A. Korkin and E. N. Tsvetkov, Bull. Soc. Chim.
Fr., 1988, 335; (d ) A. A. Korkin and E. N. Tsvetkov, Russ. J. Inorg.
Chem. (Engl. Transl.), 1989, 34, 161; A. A. Korkin and E. N.
Tsvetkov, Zh. Neorg. Khim., 1989, 34, 295.
23 (a) O. Dahl, Tetrahedron Lett., 1982, 23, 1493; (b) V. A. Jones,
S. Sriprang, M. Thornton-Pett and T. P. Kee, J. Organomet. Chem.,
1998, 567, 199.
18 (a) B. H. Dahl, J. Nielsen and O. Dahl, Nucleic Acids Res., 1987, 15,
1729; (b) O. Dahl, Phosphorus Sulfur, 1983, 18, 201.
19 E. J. Nurminen, J. K. Mattinen and H. Lönnberg, J. Chem. Soc.,
Perkin Trans. 2, 1999, 2551.
20 Similar structures have been reported in the literature: (a) I. F.
Lutsenko, M. V. Proskurina and A. L. Chekhun, Phosphorus, 1974,
4, 57; (b) V. L. Foss, Y. A. Veits, N. V. Lukashev, Y. E. Tsvetkov and
I. F. Lutsenko, Russ. J. Gen. Chem. (Engl. Transl.), 1977, 47,
439; V. L. Foss, Y. A. Veits, N. V. Lukashev, Y. E. Tsvetkov and
24 E. V. Borisov, A. K. Akhlebinin and E. E. Nifant’ev, Russ. J. Gen.
Chem. (Engl. Transl.), 1984, 54, 2096; E. V. Borisov, A. K.
Akhlebinin and E. E. Nifant’ev, Zh. Obshch. Khim., 1984, 54, 2345.
25 É. E. Nifant’ev, M. K. Gratchev, S. Yu. Burmistrov and L. K.
Vasyanina, Russ. J. Gen. Chem. (Engl. Transl.), 1992, 62, 2013;
É. E. Nifant’ev, M. K. Gratchev, S. Yu. Burmistrov and L. K.
Vasyanina, Zh. Obshch. Khim., 1992, 62, 2439.
J. Chem. Soc., Perkin Trans. 2, 2001, 2159–2165
2165