N-arylbenzonitrilium ions. Photochemical generation and effect of substituents in the phenyl rings on lifetimes in water and reactivity with azide ion

Article abstract of DOI:10.1039/b107028m

Nitrilium ions Ar-C≡N⁺-Ar′ have been studied by laser flash photolysis in aqueous solutions containing 20% acetonitrile. The cations were generated by photoheterolysis of benzimidate esters Ar-CZ=N-Ar′ (Z = -OC₆H₄-4-CN) with 4-cyanophenoxide as the photochemical leaving group. Rate constants for the reaction with water (k_w), azide ion (k_az) and hydroxide (k_OH) were measured. The cation Ph-C≡N⁺-Ph is only 50-fold shorter lived in water compared to Ph-C≡N⁺-iPr (2); thus the effect of replacing an N-alkyl group with N-phenyl is modest. These two cations are also shown to have similar lifetimes to iminium analogs, e.g. Ph-CH=N⁺ (Me)-Ph. Thus, addition of water to analogous sp and sp² hybridized systems occurs at a similar rate, and the increased steric access to the nitrilium plays at most a modest role. For the series where substituents in Ar′ were varied with Ar equal to phenyl, the Hammett plot for log k_w correlated with σ, with ρ = - 1.4. For the series where Ar was varied, the data correlated much better with σ⁺, although ρ⁺ was only -0.6. This remarkably small effect of substituents in Ar constrasts with the effects seen in benzylic carbocations, but is consistent with the nitrilium structure, with most of the positive charge located on the nitrogen. Rate constant ratios k_az:k_w, including the N-alkylnitrilium ion 2, are constant at ～10⁴, with values of k_az well below the diffusion limit, even for the most reactive nitrilium ions. This is very different behaviour from that of carbocations and arylnitrenium ions of similar lifetimes in water. For these cations the rate constants k_az would be at or at least approaching the diffusion limit.

Full text of DOI:10.1039/b107028m

View Article Online / Journal Homepage / Table of Contents for this issue
N-Arylbenzonitrilium ions. Photochemical generation and effect of
substituents in the phenyl rings on lifetimes in water and reactivity
with azide ion
2
Patrick H. Ruane, Abid R. Ahmed and Robert A. McClelland*
Department of Chemistry, University of Toronto, Toronto, Ontario M5S 3H6, Canada
Received (in Cambridge, UK) 2nd August 2001, Accepted 9th November 2001
First published as an Advance Article on the web 3rd January 2002
ϩ
᎐
Nitrilium ions Ar–C᎐N –ArЈ have been studied by laser flash photolysis in aqueous solutions containing 20%
᎐
acetonitrile. The cations were generated by photoheterolysis of benzimidate esters Ar–CZ᎐N–ArЈ (Z = –OC H -4-
CN) with 4-cyanophenoxide as the photochemical leaving group. Rate constants for the reaction with water (kw),
᎐
6
4
ϩ
᎐
azide ion (kaz) and hydroxide (kOH) were measured. The cation Ph–C᎐N –Ph is only 50-fold shorter lived in water
᎐
ϩ
᎐
compared to Ph–C᎐N –iPr (2); thus the effect of replacing an N-alkyl group with N-phenyl is modest. These two
᎐
cations are also shown to have similar lifetimes to iminium analogs, e.g. Ph–CH᎐Nϩ(Me)–Ph. Thus, addition of water
᎐
to analogous sp and sp2 hybridized systems occurs at a similar rate, and the increased steric access to the nitrilium
plays at most a modest role. For the series where substituents in ArЈ were varied with Ar equal to phenyl, the
Hammett plot for log kw correlated with σ, with ρ = Ϫ1.4. For the series where Ar was varied, the data correlated
much better with σϩ, although ρϩ was only Ϫ0.6. This remarkably small effect of substituents in Ar contrasts with the
effects seen in benzylic carbocations, but is consistent with the nitrilium structure, with most of the positive charge
located on the nitrogen. Rate constant ratios kaz : kw, including the N-alkylnitrilium ion 2, are constant at ∼104, with
values of kaz well below the diffusion limit, even for the most reactive nitrilium ions. This is very different behaviour
from that of carbocations and arylnitrenium ions of similar lifetimes in water. For these cations the rate constants kaz
would be at or at least approaching the diffusion limit.
We have recently reported a flash photolysis study of the
N-isopropylbenzonitrilium ion 2.1 This cation was obtained
upon irradiation of the imidate ester 1, a photoheterolysis with
4-cyanophenoxide as the leaving group (Scheme 1). This
Scheme 2
ent variation in both aromatic rings. The cations were obtained
photochemically from imidate precursors 3, with 4-cyano-
phenoxide as a leaving group. One of the objectives in this study
was to evaluate how substituent variation at the two ends of
the nitrilium ion affect reactivity. We also anticipated that
the N-arylnitrilium ions would be more short-lived than the the
N-alkyl derivative 2, as did turn out to be the case. In this
respect we were interested in seeing what would happen with the
rate constants for azide.
Scheme 1
aryloxide had been employed previously in flash photolysis
studies of tri- and diarylmethyl cations.2,3 The nitrilium ion 2
was generated in aqueous solutions and rate constants meas-
ured for its reaction with solvent (kw), hydroxide ion (kOH) and
azide ion (kaz). By carbocation standards,4 the nitrilium ion 2 is
relatively long-lived in water, with a lifetime (1/kw) of 2.5 milli-
seconds. This cation reacts with nucleophiles at the carbon end
Results
᎐
of the C᎐N triple bond, i.e. via the resonance contributor 2Ј.
᎐
There are now extensive studies of structure–reactivity relations
for carbocations such as the tri- and diarylmethyl cations where
the electrophilic center is sp2 hybridized.4–7 Studies of sp-
hybridized examples are, however, more limited.8–16 One of the
unusual features of the reactivity of 2 was its relatively poor
quenching by azide ion. Azide is an excellent nucleophile for
sp2-hybridized carbocations, and has seen extensive use as a
clock, with the assumption that the azide–cation combination is
diffusion limited.17,18 Such an assumption for 2 produces a life-
time that is wrong by three-orders of magnitude.
We start by discussing the evidence that the benzimidates, 3, do
produce nitrilium ions, 4, and 4-cyanophenoxide upon irradi-
ation in aqueous solutions. This takes the form of experiments
with parent 3, 4-cyanophenyl N-phenylbenzimidate. These
experiments (as is true of all of the experiments in this paper)
were conducted in 20% acetonitrile because of solubility prob-
lems in 100% aqueous solutions. The laser flash photolysis also
required that the irradiation be carried out in solutions with pH
greater than the pKa of 4-cyanophenol. The 4-cyanophenoxide
is a product, as will be discussed below. Irradiation in more
acidic solutions produces decays associated with the proton-
ation of the 4-cyanophenoxide (as was noted previously).1 In
In this paper, we report a flash photolysis investigation of a
series of N-arylbenzonitrilium ions 4 (Scheme 2), with substitu-
312
J. Chem. Soc., Perkin Trans. 2, 2002, 312–317
This journal is © The Royal Society of Chemistry 2002
DOI: 10.1039/b107028m

View Article Online
the present case this interfered with the decays due to the nitril-
ium ion.
the benzimidate was irradiated in a Rayonet reactor, and the
spectra recorded at various times. Spectrum A obtained before
irradiation is that of the benzimidate. This has a λmax at 240 nm,
but there is tailing absorbance out past 300 nm. Upon irradi-
ation a new peak grows with a λmax of 274 nm, with isosbestic
points at 224 nm and 254 nm. The λmax at 274 nm corresponds
to the maximum of 4-cyanophenoxide. It can be seen that the
difference between curve D and curve A in Fig. 2 matches the
final spectrum of Fig. 1. Our experiences with 4-cyanopheno-
xide as a photochemical leaving group show that the ion is
generated within the 20 ns laser pulse.1–3 Thus, the absorbance
for 4-cyanophenoxide is also present in the initial spectrum in
Fig. 1. We argue that the additional absorbance present in this
spectrum must be due to the other product of the photo-
heterolysis, the nitrilium ion.
Fig. 1 shows spectra obtained by laser flash photolysis (LFP)
of the benzimidate in a basic solution. There is immediate
absorbance increase in the region 260–320 nm that occurs in the
20 ns laser pulse, and results in a spectrum with an apparent
λmax at 280–285 nm. This is not necessarily a true λmax since the
LFP apparatus only measures a difference in absorbance (∆A),
i.e. the absorbance after the pulse minus the absorbance of the
solution before. The initial ∆A values are shown as the closed
symbols in Fig. 1. This absorbance decays following good
exponential kinetics, with k = 4 × 105 sϪ1 for the system of
Fig. 1. The decay does not go to zero, but to the absorbances
indicated by the open symbols in Fig. 1.
(a) This absorbance is similar to that seen upon irradiation
of compound 1, where the transient was assigned to the
N-isopropylbenzonitrilium 2.1 This comparison is shown in
Fig. 3. The closed squares are the data for 2. As was shown in
Fig. 1 Absorbance changes observed with laser flash photolysis (248
nm irradiation) of 4-cyanophenyl N-phenylbenzimidate (1.5 × 10Ϫ5 mol
dmϪ3) in 0.006 mol dmϪ3 sodium hydroxide in 20% acetonitrile : 80%
water. The closed squares are the readings of ∆A immediately after the
laser pulse. The open squares are the readings at the completion of the
exponential decay process.
Fig. 3 Laser flash photolysis spectra of the N-isopropylbenzonitrilium
ion (closed squares) and the N-phenylbenzonitrilium ion (open
squares). Data for the former were obtained by irradiation of 4-
cyanophenyl N-isopropylbenzimidate, and are the experimental points
from Fig. 5 of ref. 1. The data for the latter are the differences between
initial and final ∆A values in Fig. 1 of this paper.
The analysis of these spectra and spectral changes is com-
plicated since all of the species—the benzimidate starting
material, the nitrilium ion, and the products—have overlapping
absorbances. This can be seen in Fig. 2, which depicts the optical
densities obtained in an experiment where a cuvette containing
Fig. 5 of ref. 1, the points here represent a shoulder in the
overall spectrum of the nitrilium ion 2; the maximum is not real
since there is overlap with other species present. The open sym-
bols in Fig. 1 are the difference between the initial and final
spectra of Fig. 1, i.e. with the contribution from 4-cyano-
phenoxide subtracted out. It can be seen that there is similarity
with the ∆A values obtained for 2. There is some additional
absorbance above 300 nm in the present case. This can be
attributed to a contribution from the additional N-phenyl
chromophore.
(b) HPLC analysis of the products of irradiation of 4-cyano-
phenyl N-phenylbenzimidate in 20% acetonitrile shows that
there is conversion to 4-cyanophenol (or the phenoxide) and
benzanilide, with no other peaks appearing. The reaction is
within experimental error, quantitative even at prolonged
irradiation times. As shown in Scheme 3, these are the two
products expected for photoheterolysis, with the benzanilide the
product of the reaction of the nitrilium ion.
(c) The spectral change from the initial to the final A values in
Fig. 1 follows exponential kinetics with the same rate constant
across the entire spectrum. This points to a single reactive
species as being responsible. This decay is accelerated by azide
ion and hydroxide ion, but not by oxygen. These are character-
istics of carbocation intermediates that react with added nucleo-
philes,4 but are unaffected by O2.
Fig. 2 Spectra (1 cm pathlength) obtained with 4-cyanophenyl N-
phenylbenzimidate (4.5 × 10–6 mol dmϪ3) in 0.002 mol dmϪ3 sodium
hydroxide in 20% acetonitrile : 80% water. Spectra were obtained before
irradiation (A) and after 3 minutes (B), 7 minutes (C) and 16 minutes
(D) irradiation at 254 nm in a Rayonet reactor. Additional irradiation
produced no further change.
J. Chem. Soc., Perkin Trans. 2, 2002, 312–317
313

View Article Online
Table 1 Substituents X and Y in N-(Y-aryl)-X-benzonitrilium ions
and rate constants for the reactions with water, azide ion and hydroxide
ion. Conditions are 20% acetonitrile, ionic strength = 0.1 mol dmϪ3
(NaClO4), 20 ЊC
Table 2 Lifetimes of cations
Cation
(1/kw)/s
(1/ks)/s
4-Me2NC6H4Cϩ(CH3)2 (5)
6 × 10Ϫ3 a
kw × 104 a
kaz × 108 b
kOH × 107 b
Ph–CH᎐ϩN(CH2Ph)Bu (6)
2–10 × 10Ϫ3 b
3 × 10Ϫ3 d
9 × 10Ϫ5 (MeOH)c
2 × 10Ϫ5 (MeOH)c
X
Y
᎐
ϩ
᎐
Ph–C᎐ N–iPr (2)
᎐
H
H
H
H
H
H
H
H
H
2.1
0.67
1.09
1.57
1.95
2.6
2.4
3.2
4.3
5.1
1.02
1.7
1.48
1.9
5.3
7.1
2.6
0.75
1.10
1.91
2.4
2.8
3.2
4.5
4.5
5.8
1.15
1.76
2.0
2.2
5.3
7.1
8.1
11.3
6.4
2.0
Ph–CH᎐ϩN(CH3)Ph (7)
4–20 × 10Ϫ4 b
5 × 10Ϫ5 e
᎐
ϩ
᎐
4-MeO
4-PhO
4-Me
3-Me
3-MeO
4-Cl
3-Cl
3-CF3
4-CF3
H
H
H
H
H
Ph–C᎐ N–Ph (4, X = Y = H)
᎐
CH2᎐᎐ϩN(CH3)C6H4-4-Cl (8)
3 × 10Ϫ7 f
4.4
4-MeOC6H4CHϩ–CH3 (9)
∼3 × 10Ϫ9 g
∼7 × 10Ϫ10 g
∼1 × 10Ϫ10 i
3 × 10Ϫ6 (TFE)h
7 × 10Ϫ7 (TFE)h
4-MeOC6H4–Cϩ᎐CH2 (10)
᎐
᎐
4-Me2NC6H4–Cϩ᎐O (11)
10.0
12.0
a Ref. 16. b Estimated on basis of difference of 20–100 lifetimes between
cations in water and methanol. Ref. 19. c Ref. 20. d Ref. 1. e This work.
f Ref. 21. g Estimated on basis of difference of 1000 lifetimes between
cations in water and TFE. h Ref. 13. TFE = 2,2,2-trifluoroethanol. i Ref.
11.
H
H
4-MeO
4-PhO
4-Me
3-Me
4-Cl
3-Cl
3-CF3
4-CF3
4.7
6.1
the latter has an electron withdrawing benzyl group on the
nitrogen.
10.1
14.2
H
H
H
Phenyl groups are inductively electron withdrawing relative
to alkyl groups. Thus, N-phenyl substituted nitrilium ions and
iminium ions are expected to be more reactive than N-alkyl
analogs. The data in Table 1 show that while this is the case, the
effect is not large. The best comparison here is the N-phenyl-
benzonitrilium ion with the N-isopropyl analog 2—a difference
in lifetime of a factor of only 50.
8.9
16.4
19.2
a Units of sϪ1 b Units of dm3 molϪ1 sϪ1
. .
It can also be seen that benzonitrilium ions are much longer-
lived than α-phenyl substituted vinyl cations and acylium ions
such as 10 and 11. In fact the comparison in Table 2 consider-
ably underestimates the difference, since 10 and 11 bear strong
para π donors. Thus the analogs with just a phenyl ring will be
much more reactive. This is in fact seen in the weakly nucleo-
philic solvent 1,1,1,3,3,3-hexafluoroisopropyl alcohol (HFIP).
Cation 10 with the 4-methoxy substituent is 10000 times
longer-lived than the unsubstituted cation PhCϩ᎐CH .16
᎐
2
Effects of substituents in N-arylbenzonitrilium ions
Scheme 3
The effects of substituents within the N-arylbenzonitrilium ions
of the present study are shown in the form of Hammett plots
in Fig. 4. The data can be divided into two series, an N-series
Table 1 shows the nitrilium ions studied in this work. These
ions were all obtained by photolysis of a 4-cyanophenyl benz-
imidate precursor. Spectral changes in the region 260–320 nm
similar to those of the parent system, with quenching by azide
ion and hydroxide, were evidence that the ions had formed.
Table 1 provides the results of the kinetic experiments. The
parameter kw is the first-order rate constant for the decay of the
cation in solvent alone; kaz and kOH are the second-order rate
constants for the quenching by azide ion and hydroxide ion,
obtained as the slope of plots of the observed rate constants
versus the concentrations of the nucleophile.
Discussion
Nitrilium ion reactivity
Table 2 provides a survey of the lifetimes of relevant carbo-
cations, iminium ions and nitrilium ions that have been meas-
ured or estimated in water. There has been a suggestion that
sp-hybridized cations should be more reactive than sp2 analogs,
because of the decreased steric effect in the former.15 This is,
however, not borne out in the data. The vinyl cation 10 and
the carbenium analog 9 have very similar lifetimes in 2,2,2-
trifluoroethanol, as noted previously.13 A similar conclusion
can be made comparing nitrilium ions and iminium ions. The
best comparison here is between the N-phenyliminium ion 7
and the N-phenylbenzonitrilium ion. While the former was not
studied in water, its lifetime can be estimated, and it is only
about an order of magnitude longer-lived. There is also only a
small difference between the N-alkylnitrilium ion 2 and the
iminium ion 6, although here the analogy is not as close since
Fig. 4 Hammett plots for N-(Y-phenyl)-X-benzonitrilium ions. The
open squares (versus σ) and the crosses (versus σϩ) represent the C-series
where Y = H and the substituent X is varied. The closed squares (versus
σ) represent the N-series where X = H and the substituent Y is varied.
The ‘y’ axis for the C-series is log kw; the ‘y’ axis for the N-series is log
k
w ϩ 0.5.
314
J. Chem. Soc., Perkin Trans. 2, 2002, 312–317

View Article Online
where the substituent Y on the N-aryl ring is varied with the
substituent X on the C-aryl ring equal to hydrogen, and a
C-series where the latter is varied with the former equal
to hydrogen. Within both series the effect of the substituent
variation is modest, particularly in comparison with other
carbocations. A change from the most electron withdrawing
substituent, 4-trifluoromethyl, to the best electron donor,
4-methoxy, only results in a kinetic stabilization of about one
order of magnitude in both series. The N-series produces a
good linear correlation (correlation coefficient = 0.989) when
plotted against σ substituent constants. This correlation with
σ is not surprising since the positive charge in the cation cannot
be delocalized by resonance into the N-aryl ring (without
including unfavorable zwitterionic resonance contributors).
With the C-series however, a plot versus σ shows significant
deviation in the points for the para π donors, especially
4-phenoxy and 4-methoxy. The σϩ correlation is significantly
improved (correlation coefficient = 0.984). Even though there
must only be a small amount of positive charge at the carbon
end of the C–N bond, para donors in the phenyl ring interact in
a manner similar to that in benzylic cations.
the more reactive derivatives react with azide with a constant
rate constant that although slightly dependent on structural
class lies in the range 5–10 × 1010 dm3 molϪ1 sϪ1. This behavior
can be explained by the azide reaction being diffusion con-
trolled, so that changes in structure no longer have any effect on
the rate constant. This behavior is observed once these cations
have a solvent rate constant of ∼105 sϪ1. In other words cations
of these series with lifetimes of 10 microseconds or less in water
react with azide ion with a rate constant that is, at the very least,
close to the diffusion limit.
This behavior is certainly not seen for the benzonitrilium
ions. The N-aryl derivatives studied in the present work have
lifetimes of 10–100 microseconds, and although the rate
constants for azide ion are large, they are still 1–2 orders of
magnitude smaller than the values associated with diffusion
control. Fig. 6 depicts this in a different perspective, through a
The slopes of the correlation lines in Fig. 4 are Ϫ1.4 for the
N-series versus σ, and Ϫ0.6 for the C-series versus σϩ. This
greater sensitivity to the substituent change in the N-aryl ring is
consistent with a greater positive charge at the nitrogen. In
other words, these are nitrilium ions as in resonance contributor
2 of Scheme 1, and the contribution from the vinyl cation con-
tributor 2Ј is small. This conclusion is also reached when one
considers the effects of substituents in other benzylic-type
cations. A pertinent example are the two vinyl cations
4-MeOC H Cϩ᎐CH and C H Cϩ᎐CH where there is a 10000-
᎐
᎐
6
4
2
6
5
2
fold difference in reactivity in the solvent HFIP.16 The nitrilium
ϩ
ϩ
᎐
᎐
ions 4-MeOC H C᎐N Ph and C H C᎐N Ph show only a three-
᎐
᎐
6
4
6
5
fold difference.
Azide ion
Azide ion does effectively quench the nitrilium ions of this
study. However, as observed with the N-alkyl derivative 2 previ-
ously,1 its reactivity is low when compared to what has been
observed with other carbocations. This comparison is depicted
graphically in Fig. 5 in the form of a plot of log kaz versus log
kw. The behaviour of the carbocations and arylnitrenium ions
has been commented upon previously.22,23 Within each series,
Fig. 6 Correlation of azide : water selectivity with azide reactivity.
Symbols have the same meaning as in Fig. 5.
logarithmic plot of azide : water selectivity versus the rate
constant for azide. As originally suggested by Ta-Shma and
Rappoport,24 the simple interpretation of this figure is that
there are two separate behaviors with a short region represent-
ing the changeover from one to the other. For most of the tri-
and diarylmethyl carbocations and the arylnitrenium ions, kaz is
effectively constant because the reaction has reached diffusion
control. A change in structure that increases reactivity has no
effect on kaz, but it does increase kw. This means that the select-
ivity kaz : kw decreases, so that a plot of log kaz : kw versus log kaz
is essentially a vertical line. The other extreme is shown by the
data for the nitrilium ions, where the plot is essentially a hori-
zontal line. In this series a change in structure that increases
reactivity has essentially the same effect on kaz as it does on kw,
and the ratio kaz : kw is constant. This is true within the series of
N-arylnitrenium ions, and also if the point for the N-isopropyl
derivative is included. Constant selectivity with changing
reactivity is a feature of highly stabilized carbocations, as
observed by Ritchie.25,26 The nitrilium ions, although they are
reactive towards water, are still in a constant selectivity region,
at least so far as the azide : water ratio is concerned. The start of
the changeover to diffusion control in azide is just apparent
from the slight downward deviation of the points with the
highest kaz.
Fig. 5 Correlation of log kaz versus log kw for benzonitrilium ions
Fig. 6 shows that the kaz : kw ratio in the constant selectivity
region must vary over a wide range with cation structure. For
the nitrilium ions this ratio is about 104 and for the triarylme-
thyl cations 2–4 × 106. The arylnitrenium ions are just starting
(open circles—this work and ref.
1 (point indicated by ‘1’)),
triarylmethyl cations (closed circles—ref. 22), diarylmethyl cations
(open squares—ref. 22) and arylnitrenium ions (closed squares—ref.
23).
J. Chem. Soc., Perkin Trans. 2, 2002, 312–317
315

View Article Online
Table 3 Melting points and HRMS data for 4-cyanophenyl N-(Y-phenyl)-X-benzimidates
X
Y
Mp/ЊC
Formula
HRMS calc.
HRMS obs.
H
H
H
H
H
H
H
H
H
95–96
119–120
179–181
132–134
64–65
91–93
125–126
97–98
82–83
110–111
71–72
156–157
76–77
81–82
C20H14N2O
298.1106
328.1212
390.1368
312.1263
312.1263
328.1212
332.0716
332.0716
366.0980
366.0980
328.1212
390.1368
312.1263
312.1263
332.0716
332.0716
366.0980
366.0980
298.1113
328.1211
390.1355
312.1268
312.1268
328.1220
332.0722
332.0718
366.0991
366.0974
328.1201
390.1351
312.1256
312.1266
322.0716
332.0711
366.0995
366.0985
4-MeO
4-PhO
4-Me
3-Me
3-MeO
4-Cl
3-Cl
3-CF3
4-CF3
H
H
H
H
H
C21H16N2O2
C26H18N2O2
C21H16N2O
C21H16N2O
C21H16N2O2
C20H13ClN2O
C20H13ClN2O
C21H13F3N2O
C21H13F3N2O
C21H16N2O2
C26H18N2O2
C21H16N2O
H
H
4-MeO
4-PhO
4-Me
3-Me
4-Cl
3-Cl
3-CF3
4-CF3
C21H16N2O
105–106
77–79
41–42
C20H13ClN2O
C20H13ClN2O
C21H13F3N2O
C21H13F3N2O
H
H
H
96–97
to show the changeover to constant selectivity; it appears as if
the constant selectivity ratio will be around 108. There are also
literature data for 9-arylxanthenylium ions that show constant
kaz : kw of 107.27 There is no obvious reason for these differences.
They cannot be related to a steric effect, since the nitrilium
carbon is likely to be the most accessible with the triarylmethyl
carbon the least, and the other two in between. One con-
sequence of these differences relates to the practice of employ-
ing kaz : kw ratios determined from competition experiments to
calculate absolute kw values based on the assumption of diffu-
sion control in azide ion. Whether this assumption is true will
be very dependent on the class of cation. A kaz : kw ratio of 104
for an arylnitrenium ion or a triarylmethyl cation does indeed
represent a situation where the azide reaction is diffusion con-
trolled. The same ratio for a nitrilium ion does not; in fact this is
the ratio of the constant selectivity region.
therefore possible that as the cation becomes more reactive, the
hydroxide reaction does not respond to the same extent because
this reaction is partially limited by the need to desolvate the
nucleophile before bond formation. This was observed previ-
ously for amines reacting with diarylmethyl cations in water.28
For the more reactive cations desolvation of the amine was
clearly a very important component of the reactivity and the
rate constant for the amine showed little increase with increased
cation reactivity.
Experimental
Laser flash photolysis experiments involved ca. 20 ns pulses at
248 nm (60–120 mJ per pulse) from a Lumonics excimer laser
(KrF emission), a pulsed xenon lamp providing monitoring
light. The sample was placed in a 4 × 1 × 1 cm cuvette, irradi-
ated with the laser on the 4 × 1 face, and monitored perpendicu-
larly so that the path length was 4 cm. The cuvette was replaced
with a fresh solution after each irradiation. After passing
through a monochromator, the signal from the photomultiplier
tube was digitized and sent to a computer for analysis.
Hydroxide ion
While the azide : water ratio is constant this is not true for the
hydroxide : water or hydroxide : azide ratios. This is shown in
Fig. 7 in the form of a plot of log kOH versus log kaz. This plot is
The spectra in Fig. 1 were constructed point-by-point by
irradiating solutions of the same concentration at varying wave-
lengths over a short period of time to ensure constant laser
power. Kinetic experiments were performed at a monitoring
wavelength of 290 nm. The ∆A–time data were fitted to the
exponential equation. In general good fits were obtained, with
rate constants from replicate experiments agreeing to 5%. The
quenching experiments with azide ion were carried out by
measuring the observed rate constant in 3–4 solutions of
sodium azide over a concentration range 0–0.003 mol dmϪ3
.
With hydroxide ion concentrations of 0–0.02 were employed.
Plots of k(obs) versus [NaN3] and [NaOH] were linear, with
values of kaz and kOH being obtained as the linear regression
slope.
Product analyses following the irradiation of 4-cyanophenyl
N-phenylbenzimidate were performed with a Waters HPLC
system using a C18 column with 2 cm3 minϪ1 of 50 : 50
acetonitrile : pH 4.5 0.05 mol dmϪ3 acetate buffer as the eluting
solvent and 240 nm detection. The products 4-cyanophenol and
benzanilide were identified by comparison of retention times
with those of authentic samples. Quantitative analyses were
performed by determining the response factor at 240 nm, and
correcting the peak areas of the unknown solutions.
4-Cyanophenyl N-(Y-aryl)-X-benzimidates 3 were prepared
as follows. An appropriately substituted benzanilide (50 mmol),
prepared by reaction of an X-substituted benzoyl chloride with
a Y-substituted aniline, was treated in dry benzene (100 cm3)
with phosphorus pentachloride (52.5 mmol).29 After refluxing
Fig. 7 Correlation of log kOH and log kaz for benzonitrilium ions. The
point marked ‘1’ refers to the N-isopropyl cation 2, data from ref. 1.
reasonably linear, but the slope is only 0.65. Thus the rate con-
stant for hydroxide ion does not increase as rapidly as that for
kaz (or kw). Hydroxide ion is strongly hydrated in water. It is
316
J. Chem. Soc., Perkin Trans. 2, 2002, 312–317

View Article Online
2 R. A. McClelland, V. M. Kanagasabapathy and S. Steenken, J. Am.
Chem. Soc., 1988, 110, 6913.
3 R. A. McClelland, V. M. Kanagasabapathy, N. Banait and
S. Steenken, J. Am. Chem. Soc., 1989, 111, 3966.
4 R. A. McClelland, Tetrahedron, 1996, 52, 6823.
5 H. Mayr and M. Patz, Angew. Chem., Int. Ed. Engl., 1994, 33, 958.
6 C. D. Ritchie, Acc. Chem. Res., 1972, 5, 348; C. D. Ritchie,
Acc. Chem. Res., 1986, 64, 2239.
7 J. Richard, Tetrahedron, 1995, 51, 1535.
8 W. Schnabel, I. Naito, T. Kitamura, S. Kobayashi and H. Taniguchi,
Tetrahedron, 1980, 36, 3229.
9 S. Kobayashi, T. Kitamura, H. Taniguchi and W. Schnabel, Chem.
Lett., 1983, 1117; S. Kobayashi, T. Kitamura, H. Taniguchi and
W. Schnabel, Chem. Lett., 1984, 2101.
10 F. I. M. Van Ginkel, R. J. Visser, C. A. G. O. Varma and G. Lodder,
J. Photochem., 1985, 30, 453.
11 D. B. Song and W. P. Jencks, J. Am. Chem. Soc., 1987, 109, 3160;
D. B. Song and W. P. Jencks, J. Am. Chem. Soc., 1989, 111, 8470.
12 S. Kobayashi, Q. Q. Zhu and W. Schnabel, Z. Naturforsch., Teil B,
1988, 43, 825.
for two hours, the volatiles, benzene and phosphorus oxy-
chloride, were removed in vacuo to give the N-(Y-aryl)-X-
benzimidoyl chloride as an oil or low-melting solid that was
used directly in the next step. The sodium salt of 4-cyanophenol
was prepared by adding a sodium hydride (65 mmol) to a
stirred solution under argon of 4-cyanophenol (60 mmol) dis-
solved in dry THF (100 cm3). The imidoyl chloride was taken
up in a minimum amount of dry THF, and this solution added
to the phenoxide suspension. The mixture was refluxed under
argon overnight. After cooling, water (100 cm3) was added. The
THF layer was washed twice with cold 0.1 mol dmϪ3 NaOH
(100 cm3) and once with water (100 cm3), followed by drying
over magnesium sulfate, filtration and removal of the solvent.
This left a solid that in most cases was purified by recrystalliz-
ation from ethanol or ethanol : water. In a few cases, a pre-
liminary purification using silica gel column chromatography
was necessary.
Table 3 lists melting points and high resolution mass spectral
data obtained for the solids. The low resolution mass spectra
(electron impact) were characterized by a small peak corre-
sponding to the molecular ion, with a large peak at M Ϫ 118
corresponding to the nitrilium ion derived by loss of the 4-
cyanophenoxide radical. The 1H NMR and 13C NMR spectra,
while consistent with the structures, were complex because of
the large number of overlapping signals in the aromatic region.
In addition, both spectra showed broadened resonances. This
13 F. Cozens, R. A. McClelland and S. Steenken, Tetrahedron Lett.,
1992, 33, 173.
14 Y. Chiang, R. Eliason, J. Jones, A. J. Kresge, K. L. Evans and
R. D. Gandour, Can. J. Chem., 1993, 71, 1964.
15 K. N. Dalby and W. P. Jencks, J. Am. Chem. Soc., 1997, 119, 7271.
16 F. L. Cozens, V. M. Kanagasabapathy, R. A. McClelland and
S. Steenken, Can. J. Chem., 1999, 77, 2069.
17 J. P. Richard and W. P. Jencks, J. Am. Chem. Soc., 1982, 104, 4689,
4691.
18 J. P. Richard, M. E. Rothenburg and W. P. Jencks, J. Am. Chem. Soc.,
1984, 106, 1373.
19 R. A. McClelland, V. M. Kanagasabapathy and S. Steenken, J. Am.
Chem. Soc., 1988, 110, 6913.
20 C. Gaebert, C. Seigner, J. Mattay, M. Toubartz and S. Steenken,
J. Chem. Soc., Perkin Trans. 2, 1998, 2735.
suggests that there were two isomers about the C᎐N bond
᎐
present in the CDCl3 solution, and that these were undergoing
slow conformational equilibration on the NMR time scale.
21 S. Eldin and W. P Jencks, J. Am. Chem. Soc., 1995, 117, 4851.
22 R. A. McClelland, V. M. Kanagasabapathy, N. Banait and
S. Steenken, J. Am. Chem. Soc., 1991, 113, 1009.
Acknowledgements
R. A. M. acknowledges the continued financial support of
the Natural Sciences and Engineering Research Council
of Canada.
23 D. Ren and R. S. McClelland, Can. J. Chem., 1998, 76, 78.
24 R. Ta-Shma and Z. Rappoport, J. Am. Chem. Soc., 1983, 105, 6082.
25 C. D. Ritchie, Acc. Chem. Res., 1972, 5, 348.
26 C. D. Ritchie, Can. J. Chem., 1986, 64, 2239.
27 R. A. McClelland, N. Banait and S. Steenken, J. Am. Chem. Soc.,
1989, 111, 2929.
28 R. A. McClelland, V. M. Kanagasabapathy, N. S. Banait and
S. Steenken, J. Am. Chem. Soc., 1992, 114, 1816.
29 R. Ta-Sham and Z. Rappoport, J. Am. Chem. Soc., 1977, 99, 1845.
References
1 P. H. Ruane, R. A. McClelland, A. F. Hegarty and S. Steenken,
J. Chem. Soc., Perkin Trans. 2, 1999, 43.
J. Chem. Soc., Perkin Trans. 2, 2002, 312–317
317