Nucleophilic activation of hydrogen peroxide. On the formation and the reactivity of 1H-1,2,4-triazoleperoxycarboxylic acid and O-p-nitrophenylmonoperoxycarbonic acid

Article abstract of DOI:10.1039/a903852c

Measurements of the infrared phosphorescence of singlet molecular oxygen (¹O₂) at 1270 nm have been employed to demonstrate the formation of ¹O₂ in the system N,N′-carbonyldi-1,2,4-triazole (CDT)-hydrogen peroxide and in the system 1H-1,2,4-triazolecarboxylic acid p-nitrophenyl ester (TCNP)-hydrogen peroxide in tetrahydrofuran. At hydrogen peroxide concentrations of [H₂O₂] ≥ 2 [CDT] or ≥ 2 [TCNP] one molecule of ¹O₂ is generated per molecule of CDT and TCNP, respectively. In both systems a very reactive peroxy-intermediate is formed. In the system CDT-H₂O₂ the 1H-1,2,4-triazoleperoxycarboxylic acid (1) is generated, in TCNP-H₂O₂ the p-nitrophenylmonoperoxy-carbonic acid (4). For the epoxidation of cyclohexene in THF by 1 a rate constant k₁₅ ≈3 x 10⁰ dm³ mol^-1 s^-1 can be estimated. For 4 the corresponding rate constant was found to be k₂₀ ≈ 6 x 10^-2 dm³ mol^-1 s^-1. The Arrhenius parameters of the formation of 1 and 4, respectively, and in addition of their consecutive reactions with hydrogen peroxide were determined. The results are consistent with the assumption that the nucleophilic displacements by hydrogen peroxide at the carbonyl carbon atom are addition-elimination processes, which take the form of a B_AC2-mechanism.

Full text of DOI:10.1039/a903852c

View Article Online / Journal Homepage / Table of Contents for this issue
Nucleophilic activation of hydrogen peroxide. On the formation
and the reactivity of 1H-1,2,4-triazoleperoxycarboxylic acid and
O-p-nitrophenylmonoperoxycarbonic acid†
Carsten Bender and Hans-Dieter Brauer*
Institut für Physikalische und Theoretische Chemie, Universität Frankfurt/M.,
Marie-Curie-Straße 11, 60439 Frankfurt/M., Germany
Received (in Cambridge, UK) 13th May 1999, Accepted 26th August 1999
Measurements of the infrared phosphorescence of singlet molecular oxygen (1O2) at 1270 nm have been employed to
demonstrate the formation of 1O2 in the system N,NЈ-carbonyldi-1,2,4-triazole (CDT)–hydrogen peroxide and in the
system 1H-1,2,4-triazolecarboxylic acid p-nitrophenyl ester (TCNP)–hydrogen peroxide in tetrahydrofuran. At
hydrogen peroxide concentrations of [H2O2] ≥ 2 [CDT] or ≥ 2 [TCNP] one molecule of 1O2 is generated per molecule
of CDT and TCNP, respectively. In both systems a very reactive peroxy-intermediate is formed. In the system CDT–
H2O2 the 1H-1,2,4-triazoleperoxycarboxylic acid (1) is generated, in TCNP–H2O2 the p-nitrophenylmonoperoxy-
carbonic acid (4). For the epoxidation of cyclohexene in THF by 1 a rate constant k15 ≈3 × 100 dm3 molϪ1 sϪ1 can
be estimated. For 4 the corresponding rate constant was found to be k20 ≈ 6 × 10Ϫ2 dm3 molϪ1 sϪ1. The Arrhenius
parameters of the formation of 1 and 4, respectively, and in addition of their consecutive reactions with hydrogen
peroxide were determined. The results are consistent with the assumption that the nucleophilic displacements by
hydrogen peroxide at the carbonyl carbon atom are addition–elimination processes, which take the form of a
B
AC2-mechanism.
CDT–H2O2 postulated by Rebek et al.2 was not studied in
detail. To get more insight into the kinetics of reaction (2) we
have investigated this reaction in tetrahydrofuran (THF) by
measuring the formation of O2 on the basis of the O2 phos-
phorescence at λ = 1270 nm.3 Additionally, employing the same
method we have studied the system CDI–H2O2 and the system
1H-1,2,4-triazolecarboxylic acid p-nitrophenyl ester (TCNP)–
H2O2.
Introduction
Staab1 was the first to demonstrate that the carbonyl group of
diazolides of carbonic acid is strongly electronically activated
on account of the two electron-withdrawing heterocycles.
N,NЈ-carbonyldiimidazole (CDI), e.g., reacts with pure water at
room temperature within a few seconds to give two molecules
of imidazole and one molecule of carbon dioxide.
1
1
Since hydrogen peroxide is more nucleophilic than water1 it
is not surprising that azolides (N-acetylazoles) of the carbonic
acid also react with hydrogen peroxide. Rebek et al.2 have
investigated the system N,NЈ-carbonyldi-1,2,4-triazole (CDT)–
H2O2 in different organic solvents and postulated that 1H-1,2,4-
triazoleperoxycarboxylic acid as a reactive peroxy-intermediate
is produced which effectively epoxidizes olefins like cyclohexene
(vide infra). Moreover, the authors have observed that in this
system, in the absence of olefins, molecular singlet oxygen (1O2)
is quantitatively formed according to eqn. (2).
Experimental
Materials
The following compounds and solvents, respectively, were
purchased from Fluka and used as received: N,NЈ-carbonyl-
diimidazole (CDI) (purum, ~97%, mp 112–115 ЊC), N,NЈ-
carbonyldi-1,2,4-triazole (CDT) (≥98%, mp 145–150 ЊC), N-
(trimethylsilyl)-1,2,4-triazole (puriss. ≥99% (GC)), 4-nitro-
phenyl chloroformate (purum, ≥97%), cyclohexene (puriss. p.a.),
9,10-dimethylanthracene (DMA) (purum, ~97%), THF (for
HPLC), chloroform (CHCl3 for IR-spectroscopy) and diethyl
ether (Et2O, puriss. ≥99.5%; H2O ≤ 0.005%). Benzene (p.a.,
H2O < 0.03%) was obtained from Riedel-de Haën. 30% Hydro-
gen peroxide (pract.) came from Merck. CDI, CDT and the
stabilizer free THF from Fluka were stored under dry nitrogen.
Water free solutions of hydrogen peroxide in CHCl3 and Et2O
were prepared according to literature procedure.4
In contrast to the system CDT–H2O2, the system CDI–H2O2
was found to be ineffective with respect to the epoxidation of
olefins and no 1O2 generation was observed.2
Preparation of 1H-1,2,4-triazolecarboxylic acid p-nitrophenyl
ester (TCNP)
1
Up to now the mechanism of O2 formation in the system
p-Nitrophenyl chloroformate (6.9 g, 34 mmol) was dissolved in
30 mL of dry benzene. A solution of 5 mL of N-(trimethylsilyl)-
† The IUPAC name for carbonic acid is hydroxyformic acid.
J. Chem. Soc., Perkin Trans. 2, 1999, 2579–2587
This journal is © The Royal Society of Chemistry 1999
2579

View Article Online
Table 1 Results of the volumetric measurements of the reaction of
1,2,4-triazole (34 mmol) in 30 mL of dry benzene was added
dropwise to the stirred solution of the chloroformate at room
temperature. After addition was completed the mixture was
stirred for another 15 minutes, while precipitation of the prod-
uct was observed. The product was filtered and washed with dry
benzene and dried in vacuo. The yield of the product was 5.9 g
(72%), mp 144–146 ЊC (Found: C, 46.06; H, 2.69; N, 23.96.
C9H6N4O4 requires: C, 46.13; H, 2.58; N, 23.93%); νmax/cmϪ1
N,NЈ-carbonyldi-1,2,4-triazole (CDT) with 30% hydrogen peroxide in
THF at T = 25 ЊC; [CDT] = 1.0 × 10Ϫ1 mol dmϪ3
———
———
equiv.(H2O2)
equiv.corr
=
———
equiv.a
System
equiv.(H2O)
CDT–H2Ob
CDT–H2O2
0.9 0.1
1.8 0.1
0.9 0.2
1.0 0.1
—
c
2.0
1.0
1.1
3400 (>C᎐N-), 1600, 1580 and 1500 (C᎐C), 1790 (C᎐O), 1530
᎐
᎐
᎐
CDT–H2O2–DMAd
and 1320 (N᎐O), 1290 (C–O). Infrared spectra were taken on a
CDT–H2O2–cyclohexenee
᎐
Philips model IR-PU 9700 infrared spectrometer and were run
as potassium bromide pellets.
a Mean value of 3 measurements. b [H2O] = 100 × [CDT]. c [H2O2] =
20 × [CDT]. d Solution saturated with DMA, [H2O2] = 15 × [CDT].
e [H2O2] = [cyclohexene] = 15 × [CDT].
Gas volumetric measurements
The gas volumetric measurements were carried out according
to the procedures described by Rebek et al.2 For determination
of the purity of CDT the amount of CO2 evolved in the reac-
tion between CDT and double distilled water at T = 25 ЊC was
measured. 95 mg of CDT yielded 12.7 mL of CO2 (90% of
theoretical yield).
Moreover, we have measured the volumes of gases evolved
in the reactions between CDT and 30% hydrogen peroxide
both in the absence and in the presence of cyclohexene and
9,10-dimethylanthracene, respectively.
Singlet oxygen infrared emission measurements
Scheme 1
The near-infrared luminescence (IRL) spectrometer used in our
studies for recording the steady state 1O2 phosphorescence
emission at λ = 1270 nm has been described in detail.3,5–7
step to give most probably the elusive diperoxycarbonic acid
(2). Assuming that 2 is unstable (with respect to the properties
of 2 no prediction was made2) and generates in the third step
very fast finally CO2, H2O and O2, then the rate of O2 form-
ation is determined by the competitive, consecutive second-
order reactions (5) and (6).9
The particular difficulty of such a type of reaction is that
the reaction steps are second-order. However, the solution is
simpler when hydrogen peroxide is in large excess and thus we
have in effect two successive first-order reactions.
1
For the intensity of the O2 phosphorescence emission, IP,
eqn. (3) holds,6 where c is a constant of the IRL spectrometer,
1
1
d[1O2]
IP(t) = ckPτ∆
(3)
dt
1
kp is the rate constant of the O2 phosphorescence emission,
1
1
τ∆ is the lifetime of O2 and d[1O2]/dt denotes the rate of O2
Under first-order conditions for IP(t) eqn. (8) can be derived,
formation.
If the product C = c kp τ∆ is known for a given solvent or a
solvent mixture, the concentration of O2 can be calculated
from the plot of IP(t) versus time according to eqn. (4). The
1
k5Ј
0 k6Ј Ϫ k5Ј
IP(t) = Ck6Ј[CDT]
{exp(Ϫk5Јt) Ϫ exp(Ϫk6Јt)}
(8)
t = ∞
1
where C represents the product c kp τ∆, [CDT]0 is the initial
concentration of CDT and k5Ј = k5[H2O2] and k6Ј = k6 [H2O2]
denote the pseudo first-order rate constants of reactions (5) and
(6), respectively. The rate constant of the slow reaction can be
evaluated from the plot of ln IP(t) versus t which yields at longer
reaction time a straight line with the slope equal to k5Ј or k6Ј
(vide infra). The rate constant of the faster reaction can be
calculated from tmax, i.e. from the time at which IP(t) exhibits
its maximum. tmax is defined in eqn. (9).
[1O2] =
͵
IP(t)dt
(4)
C
t = 0
value of c strongly depends on the sensitivity of the germanium
diode. In our measurements two different germanium diodes of
the type North Coast EO 817L were employed. For the older
one a value of c1 = (3.0 0.2) × 1011 mV dm3 s molϪ1 (ref. 6)
was determined and for the new one a value of c2 = (1.4 0.1)
× 1011 mV dm3 s molϪ1 was obtained.
The values of τ∆ of the solvent mixtures (THF–H2O2 (30%))
were directly measured with our home built IRL spectrometer.3
The values obtained are presented below. For the calculation of
1
k6Ј
lnͩ ͪ
k5Ј
tmax
=
(9)
k6Ј Ϫ k5Ј
1
the amount of O2 evolved in the different systems a value of
kp = (0.45 0.10) sϪ1 was used, which agrees very well within
the error limits with kp values determined for pure THF.8
Reactions were carried out in thermostatted 1 cm quartz
cuvettes. To the THF solutions of CDT, CDI and TCNP 30%
hydrogen peroxide was added by a micro syringe. Vigorous
stirring was necessary. The initial concentrations of CDT and
TCNP used are given in the legends of the figures or in the text.
It must be noted that on the basis of eqn. (8) or eqn. (9) it
cannot be deduced which of the reactions is the faster one.
Results and discussion
A. Investigation of the system N,NЈ-carbonyldi-1,2,4-triazole
(CDT)–H2O2
A.1 Gas volumetric measurements. First we performed gas
volumetric measurements with the system CDT–H2O2 using
30% hydrogen peroxide. The results summarized in Table 1
agree excellently with the data of Rebek et al.,2 who used 98%
hydrogen peroxide for their measurements. In the absence of
Rate law and expression for IP(t)
For reaction (2) Rebek et al.2 have postulated a three step
mechanism, shown in Scheme 1. In the first step 1H-1,2,4-
triazoleperoxycarboxylic acid (1) should be generated which,
moreover, should react with hydrogen peroxide in the second
1
olefins or O2 acceptors 2.0 equiv. of gas are generated. When
2580
J. Chem. Soc., Perkin Trans. 2, 1999, 2579–2587

View Article Online
Fig. 1 IP(t) as a function of time of the system CDT–H2O2 in THF.
The inset shows the corresponding plot of ln IP(t) versus time. Condi-
Fig. 2 Plots of IP(t) versus t of the system CDT–H2O2 in THF deter-
mined at three different temperatures. Conditions: [CDT] = 1.25 × 10Ϫ2
mol dmϪ3; [H2O2] = 1.25 × 10Ϫ1 mol dmϪ3
tions: [CDT] = 1.25 × 10Ϫ2 mol dmϪ3; [H2O2] = 1.25 × 10Ϫ1 mol dmϪ3
;
.
T = 10 ЊC. —— = experimental curve; ---- = theoretically calculated
curve according to eqn. (8) using c1 = (3.0 0.2) × 1011 mV dm3 s molϪ1
kp = (0.45 0.10) sϪ1 and τ∆ = (18 1) µs; kfЈ(CDT) = 7.2 × 10Ϫ3 sϪ1
;
;
ksЈ(CDT) = 3.4 × 10Ϫ3 sϪ1
.
the THF solution contains excess cyclohexene, or is saturated
with DMA only 1.0 equiv. of gas is evolved. The observation
that the results of the gas volumetric measurements are
independent of the percentage of hydrogen peroxide (98% or
30%) is consistent with the fact that hydrogen peroxide is more
nucleophilic than water.1
A.2 1O2 phosphorescence intensity as a function of time. Fig. 1
shows a plot of IP(t) versus time of the system CDT–H2O2. The
shape of this curve is consistent with eqn. (8). In accordance
with a consecutive reaction for the 1O2 formation an induction
period (here about 20 s) is observed. The plot of ln IP(t) versus
time, depicted in the inset of Fig. 1, results in a straight line at
about t ≥ 800 s with a slope equal to ϪksЈ(CDT), the rate con-
stant of the slow reaction (vide supra).
With the values of ksЈ(CDT) and tmax = 253 s the rate con-
stant kfЈ(CDT) of the faster reaction can be estimated accord-
ing to eqn. (9). At T = 10 ЊC the pseudo first-order rate
constants were found to be kfЈ(CDT) = (7.2 1.2) × 10Ϫ3 sϪ1;
ksЈ(CDT) = (3.4 0.4) × 10Ϫ3 sϪ1. With [H2O2] = 1.25 × 10Ϫ1
mol dmϪ3 for the corresponding second-order rate constants
the following values can be calculated: kf = (5.8 1.0) × 10Ϫ2
dm3 molϪ1 sϪ1; ks = (2.7 0.3) × 10Ϫ2 dm3 molϪ1 sϪ1.
Fig. 3 Plots of ln k5 and ln k6 versus 1/T of the system CDT–H2O2 in
THF; with k5 = kf (CDT) and k6 = ks (CDT) {see text}. Conditions:
[CDT] = 1.25 × 10Ϫ2 mol dmϪ3; [H2O2] = 1.25 × 10Ϫ1 mol dmϪ3
.
As already mentioned, a priori it cannot be decided which of
the two rate constants is connected with reaction (5) or reaction
(6). This question will be discussed after the presentation of
further experimental results.
used for the estimation are given in the legend of Fig. 1. As one
can see the experimental and theoretical curves do not com-
pletely cover one another, but the difference between the areas
lies within the error limits of about 15%.
1
According to eqn. (4) the concentration of O2 generated in
the system CDT–H2O2 can be evaluated from the area under
the IP(t)→t curve between t = 20 s and t = 2250 s. The area was
found to be ar = (34100 4%) mV s. With the values of c1, kp
and τ∆ given in the legend of Fig. 1 for the concentration of 1O2
a value of [1O2] = (1.4 0.3) × 10Ϫ2 mol dmϪ3 can be estimated,
whereby the error limit is mainly determined by the uncertainty
of kp. The value of [1O2] obtained agrees very well within the
error limits with the initial concentration of [CDT] = 1.25 ×
10Ϫ2 mol dmϪ3 and confirms the result of the gas volumetric
measurements which have also shown that according to eqn. (2)
one molecule of CDT yields at least one 1O2 molecule.
A.3 Temperature dependence of the second-order rate con-
stants. Fig. 2 shows the IP(t)→t curves observed for the system
CDT–H2O2 as a function of temperature. The maximum of
IP(t) increases and is shifted to shorter times with increasing
temperature. The areas under the different curves are found to
be equal within about 10%.
Measurements were performed in the temperature range
between T = 5 ЊC and T = 20 ЊC. At temperatures T > 20 ЊC the
reproducibility of the IP(t)→t runs becomes worse. The
Arrhenius plots of kf(CDT) and ks(CDT) are given in Fig. 3.
From these plots the following values for the Arrhenius param-
eters are obtained: i) for the fast reaction, ln Aa(f) = 13 3, Eaa
(f) = 37 5 kJ molϪ1; ii) for the slow reaction, ln Aa(s) = 10 1,
Fig. 1 also shows the IP(t)→t curve theoretically estimated
according to eqn. (8). The values of the different parameters
J. Chem. Soc., Perkin Trans. 2, 1999, 2579–2587
2581

View Article Online
Eaa (s) = 33 3 kJ molϪ1; where Aa and Eaa denote apparent
Arrhenius parameters (vide infra).
this respect the results of the measurements carried out by
Rebek et al.2 are of interest. The authors have measured the
epoxidation of cyclohexene by the system CDT–H2O2 in differ-
ent solvents at T = 25 ЊC assuming that the peroxycarboxylic
acid 1 is the epoxidizing agent. Surprisingly, the second-order
rate constant “kep” determined by measuring the initial rates
of epoxidation was found to be insensitive to solvent basicity,
whereas in general for conventional peroxycarboxylic acids
rates of epoxidation decrease as the Lewis basicity of the sol-
vent increases. The values of “kep” determined in CHCl3, THF
and Et2O are about 7.5 × 10Ϫ2 dm3 molϪ1 sϪ1. This result is
interpreted assuming that 1 does not react in the form 1a, i.e. in
a form generally postulated for conventional peroxycarboxylic
acids,11 but in the “more” stabilized form 1b.
For both reaction (5) and reaction (6) relatively small values
of the Arrhenius parameters are found. Especially from the
small values of Aa(f) ≈ 4 × 105 dm3 molϪ1 sϪ1 and of Aa(s) ≈
2 × 104 dm3 molϪ1 sϪ1 it can be deduced that in both reactions
a preequilibrium takes place in which Criegee-intermediates
are involved. The data are in accordance with the rule that the
nucleophilic displacement by hydrogen peroxide at a carbonyl
carbon atom is an addition–elimination process, which takes
the form of a BAC2-mechanism,10 as outlined in Scheme 2.
However this interpretation fails in so far as together with
our results it can be deduced that the second-order rate con-
stant “kep” determined by Rebek et al.2 cannot be identical with
k15, the real rate constant of the epoxidation of cyclohexene
by 1.
As the authors report, at the beginning the epoxidation
occurs according to a second-order reaction. This implies, if 1 is
the real epoxidizing agent, that reaction (15) must occur much
Scheme 2
more quickly than reaction (6) (vide infra). Obviously, Rebek
et al.2 have measured the rate determining step of the epoxid-
ation of cyclohexene. Thus the authors have determined the
rate constant of reaction (5) which could be insensitive with
respect to the Lewis basicity of the solvent. This reaction
should be faster than reaction (6), since the carbonyl group of
CDT may be more electronically activated and therefore more
reactive towards hydrogen peroxide than that of 1.
Moreover, in both reactions the second step (k5b and k6b,
respectively) is rate determining, which is expected since in both
reactions the leaving group (1,2,4-triazole) is more basic than
the entering nucleophile (hydrogen peroxide).10
According to Scheme 2 the rate constants k5 and k6 are
defined by eqns. (10) and (11), where K5a and K6a are the
From the Arrhenius plots depicted in Fig. 3 for the rate con-
stants kf(CDT) and ks(CDT) at T = 25 ЊC the values kf(CDT) =
(1.3 0.5) × 10Ϫ1 dm3 molϪ1 sϪ1 and ks(CDT) = (4.4 0.5) ×
10Ϫ2 dm3 molϪ1 sϪ1 are obtained. Within the error limits the
value of kf(CDT) agrees with the value of “kep”(THF) obtained
by Rebek et al.2 confirming the assumption that reaction (5)
occurs faster than reaction (6). To test whether kf(CDT) = k5 is
insensitive to the basicity of the solvents, we have performed
measurements with the system CDT–H2O2 in CHCl3 and Et2O
at 10 ЊC employing water free solutions of hydrogen peroxide in
CHCl3 and Et2O, respectively. However in none of the systems
IP(t) versus t curves were observed suitable for the evaluation
of the rate constants kfЈ and ksЈ, which probably results from
the inhomogeneity caused by the water produced during the
reaction of CDT with hydrogen peroxide.12
Nevertheless, if reaction (5) is the faster one then the equality
k6 = ks(CDT) holds. With k6 = 4.4 × 10Ϫ2 dm3 molϪ1 sϪ1 and the
experimental conditions used by Rebek et al.2 for the investig-
ation of the epoxidation of cyclohexene: [cyclohexene] = 4.0 ×
10Ϫ2 mol dmϪ3; [CDT] = 8.0 × 10Ϫ2 mol dmϪ3 and [H2O2] =
3.3 × 10Ϫ1 mol dmϪ3 the rate constant k15 can be estimated
according to the inequality k15 [1] [cyclohexene] ӷ k6 [1]
[H2O2] ⇒ k15 ӷ k6 [H2O2]/[cyclohexene] and with [H2O2] ≈
[H2O2]0 Ϫ [CDT] = 2.5 × 10Ϫ1 mol dmϪ3 it follows that k15 ӷ
(4.4 × 10Ϫ2) × (2.5 × 10Ϫ1)/4 × 10Ϫ2 = 2.8 × 10Ϫ1 dm3 molϪ1 sϪ1,
i.e. k15 ≈ 3 × 100 dm3 molϪ1 sϪ1.
k5 = K5a k5b
k6 = K6a k6b
(10)
(11)
constants of the preequilibria (5a) and (5b) and k5b and k6b
denote the first-order rate constants of the decomposition of
the corresponding Criegee-intermediates.
For reaction (6), e.g., the Arrhenius equation (12) holds, where
k6 = Aa(6) exp [ϪEaa(6)/RT]
(12)
Aa(6) and Eaa(6) are defined by eqns. (13) and (14), where A(6b)
Aa(6) = A(6b) exp [∆SЊ(6a)/R]
Eaa(6) = ∆HЊ(6a) ϩ Ea(6b)
(13)
(14)
and Ea(6b) represent the Arrhenius parameters of reaction (6b)
and ∆SЊ and ∆HЊ denote the reaction entropy and the reaction
enthalpy, respectively, of equilibrium (6a).
Since the formation of the Criegee-intermediates is probably
accompanied by negative values of ∆SЊ and ∆HЊ the low
values of the Arrhenius parameters obtained are understand-
able.
A.4 Assignment of kf and ks. Now the question should be
discussed which of the reactions (5) and (6) is the faster one. In
Up to now for peroxycarboxylic acids such a high value of a
rate constant for olefin epoxidation in THF has not been
2582
J. Chem. Soc., Perkin Trans. 2, 1999, 2579–2587

View Article Online
observed. Values of this order of magnitude are known for
epoxidation of olefins by dioxiranes as epoxidizing agents.13
Therefore it must be considered whether an epoxidizing agent
other than 1 is involved. One possible hypothesis is to assume
that 1 decomposes immediately after generation according to
eqn. (16) to give, besides 1,2,4-triazole, carbon trioxide (CO3) 3,
which could react in the absence of an olefin with hydrogen
peroxide to give the diperoxycarbonic acid 2. Obviously CO3,
which should exist in the dipolar structure14 as described in
eqn. (16), may be a powerful epoxidation agent comparable
with dioxiranes.
However the assumption that 1 is unstable and decomposes
very fast into CO3 and 1,2,4-triazole is based on the experience
that the parent 1H-1,2,4-triazolecarboxylic acid is very unstable
and decomposes immediately after formation into 1,2,4-triazole
and CO2.1 But this conclusion is not compelling. In the case of
the N-benzoylperoxycarbamic acid (BPC), first synthesized by
Fig. 4 IP(t) as a function of time of the system TCNP–H2O2 in THF.
The inset shows the corresponding plot of ln IP(t) versus time. Condi-
tions: [TCNP] = 5.2 × 10Ϫ2 mol dmϪ3; [H2O2] = 5.2 × 10Ϫ1 mol dmϪ3
T = 20 ЊC. —— = experimental curve; ---- = theoretically calculated
curve according to eqn. (8) using c2 = (1.4 0.1) × 1011 mV dm3
molϪ1 kp = (0.45 0.10) sϪ1 and τ∆ = (15 1) µs; kfЈ(TCNP) =
5.9 × 10Ϫ4 sϪ1; ksЈ(TCNP) = 3.7 × 10Ϫ4 sϪ1
;
s
;
.
The pseudo first-order rate constants and the corresponding
second-order rate constants of the slow reaction and the fast
reaction, respectively, were found to be: kfЈ(TCNP) = (5.9
1.0) × 10Ϫ4 sϪ1, kf(TCNP) = (1.1 0.2) × 10Ϫ3 dm3 molϪ1 sϪ1;
ksЈ(TCNP) = (3.7 0.4) × 10Ϫ4 sϪ1, ks(TCNP) = (7.1 0.7) ×
10Ϫ4 dm3 molϪ1 sϪ1.
The value of kf(TCNP) is by a factor of 100 smaller than that
of kf(CDT) = 1.1 × 10Ϫ1 dm3 molϪ1 sϪ1 obtained at T = 20 ЊC
(see Fig. 3) and the value of ks(TCNP) is also smaller than that
of ks(CDT) = 3.7 × 10Ϫ2 dm3 molϪ1 sϪ1, but only by a factor of
about 50. It would appear that also in the system TCNP–H2O2
the first addition–elimination process occurs faster than the
second one. Thus the difference in the rate constants indicates
that the carbonyl group of CDT is more electronically activated
than that of TCNP and moreover that the carbonyl group of
the peroxycarboxylic acid 1 is also more activated than that of
the peroxy acid generated in the system TCNP–H2O2 (vide
infra).
Höft and Ganschow,4 the authors have shown that BPC is very
stable at room temperature, whereas the parent N-benzoyl-
carbamic acid is very unstable at room temperature and
decomposes immediately after generation into benzamide and
CO2.4
Moreover, the formation of CO3 in the system CDT–H2O2
may not be compatible with the small difference observed
for the second-order rate constants of the reactions (5) and (6)
(vide supra). It can be assumed that the addition of hydrogen
peroxide to CO3 according to eqn. (17) will occur much more
quickly than the nucleophilic displacement by hydrogen per-
oxide at the carbonyl atom of 1 and even of CDT, respectively.
Consequently, if CO3 is generated a greater difference between
the rate constants of the consecutive second-order reaction
could be expected. Thus the assumption made by Rebek et al.2
that the peroxycarboxylic acid 1 is the epoxidizing agent
appears to be most probable. Considering the value of
ksЈ(CDT) = 3.4 × 10Ϫ3 sϪ1 obtained under pseudo first-order
conditions at T = 10 ЊC (see Fig. 1) for the lifetime of 1 a value
of τ(1) = 1/ksЈ ≈ 5 min can be estimated indicating that 1 is more
stable than its parent 1H-1,2,4-triazolecarboxylic acid.
The area under the experimental IP(t) curve amounts to
ar = (45350 4%) mV s. With the values of c2, kp and τ∆ given
1
in the legend of Fig. 4 for the concentration of O2 evolved a
value of [1O2] = (4.8 0.7) mol dmϪ3 is obtained. Within the
error limits this value agrees very well with the initial concen-
tration of [TCNP] = 5.2 × 10Ϫ2 mol dmϪ3 used in this experi-
ment. This result distinctly shows that as observed for CDT–
H2O2 in the system TCNP–H2O2 one molecule of TCNP yields
one molecule of 1O2 if pseudo first-order conditions are
employed.
To get more information about the in situ generated peroxy-
intermediates in systems like CDT–H2O2 we have additionally
studied the system 1H-1,2,4-triazolecarboxylic acid p-nitro-
In Fig. 4 the theoretically evaluated IP(t)-curve is also
depicted. As already found for the system CDT–H2O2 the area
under the theoretical curve is a little bit smaller than that of
the experimental one. The difference between the areas only
amounts to about 6%.
The experimental data found so far for the system TCNP–
H2O2 are consistent with the following conclusions: i) in the first
addition–elimination process 1,2,4-triazole should be elimin-
ated, since this compound is less basic than the alternative leav-
ing group (p-nitrophenyl anion); ii) consequently in this process
O-p-nitrophenylmonoperoxycarbonic acid 4 should be formed
(see Scheme 3); iii) for the lifetime of 4 a value of τ(4) =
1/ksЈ(TCNP) = 45 min can be estimated. This value confirms the
1
phenyl ester (TCNP)–H2O2 in THF. In this system O2 should
be generated according to a mechanism discussed for CDT–
H2O2.
B. Investigation of the system 1H-1,2,4-triazolecarboxylic acid
p-nitrophenyl ester (TCNP)–H2O2
B.1 1O2 Phosphorescence intensity as a function of time. Fig. 4
shows a run of IP(t) versus t obtained under pseudo first-order
conditions at T = 20 ЊC. The corresponding plot of ln IP(t)
versus t is depicted in the inset of Fig. 4. The form of the IP(t)
curve is consistent with the expectation that TCNP reacts with
hydrogen peroxide in the same manner as CDT.
J. Chem. Soc., Perkin Trans. 2, 1999, 2579–2587
2583

View Article Online
Scheme 3
limits, the values of Eaa determined for reaction (19) and reac-
tion (20) are distinctly greater than the corresponding Eaa
values of reaction (5) and reaction (6), respectively. These
results mean that the difference between the rate constants of
the fast and the slow reaction observed in the different systems
is mainly determined by the different Eaa values.
C. Miscellaneous measurements
C.1 Investigation of the system TCNP–H2O2 and CDT–H2O2
in the presence of cyclohexene. According to Schemes 1, 2 and
3 for both systems it is assumed that in the first addition–
elimination process a reactive peroxycarboxylic acid is formed:
1 in the system CDT–H2O2 and 4 in the system TCNP–H2O2. 1
is found to be, by a factor of 50, more reactive towards hydro-
gen peroxide than 4. Moreover, according to our evaluation 1
should be, by a factor of 50, more reactive towards cyclohexene
than towards hydrogen peroxide. A similar behaviour with
respect to the reactivity should also hold for 4. To test this
hypothesis we have carried out measurements with the system
TCNP–H2O2 under pseudo first-order conditions in the pres-
ence of different concentrations of cyclohexene at T = 20 ЊC.
The results are shown in Fig. 6.
Only for the lowest cyclohexene concentration of [cyclo-
hexene] = [H2O2]/20 = 2.68 × 10Ϫ2 mol dmϪ3 a very small flat
IP(t)-signal is observed after adding hydrogen peroxide to the
TCNP–cyclohexene solution. This signal increases very slowly
but it increases considerably at about 500 s to a narrow IP-
signal. Moreover, it is found that the greater the cyclohexene
concentration the smaller the signal and the later it is detected.
At a cyclohexene concentration of [cyclohexene] = 2.68 × 10Ϫ1
mol dmϪ3 after about 1250 s a small IP-signal is still observed.
For equal concentrations of cyclohexene and hydrogen per-
oxide no IP-signal can be detected.
Fig. 5 Plots of ln k18 and ln k19 versus 1/T of the system TCNP–H2O2
in THF; with k18 = kf(TCNP) and k19 = ks(TCNP) {see text}. Condi-
tions: [TCNP] = 5.2 × 10Ϫ2 mol dmϪ3; [H2O2] = 5.2 × 10Ϫ1 mol dmϪ3
.
observation that O-arylmonoperoxycarbonic acids are more
stable than the corresponding esters of the parent acid.15
In analogy to the system CDT–H2O2 for the formation of 1O2
in the system TCNP–H2O2 the mechanism described in Scheme
3 can be presumed. This mechanism is confirmed by the results
of the measurements at different temperatures.
B.2 Temperature dependence of the second-order rate constants.
Measurements were performed between T = 15 ЊC and T =
In fact, these results confirm the expectation that 4 is more
reactive towards cyclohexene (eqn. (20)) than towards hydrogen
30 ЊC. The Arrhenius plots of k18 = kf(TCNP) and k19
=
ks(TCNP) are presented in Fig. 5. For the apparent Arrhenius
parameters the following values were obtained: ln Aa(18) =
22 7, Eaa(18) = (68 17) kJ molϪ1; ln Aa(19) = 11 3,
Eaa(19) = (46 7) kJ molϪ1.
Again especially the low values of Aa found for reaction (18)
and the slower reaction (19), respectively, indicate that in both
reactions a preequilibrium between the educts and a Criegee-
intermediate takes place according to a BAC2-mechanism.
Whereas the values of Aa obtained for the fast and slow reac-
tions in the different systems agree very well within the error
peroxide. The result that at equal concentrations of cyclohexene
and hydrogen peroxide no 1O2 can be detected distinctly
demonstrates that 4 is practically completely consumed by the
reaction with cyclohexene. The rate constant k20 for this reac-
2584
J. Chem. Soc., Perkin Trans. 2, 1999, 2579–2587

View Article Online
Fig. 6 IP(t) as a function of time of the system TCNP–cyclohexene–
H2O2 in THF at different cyclohexene concentrations. The inset pres-
ents the plot of 1/√IP(t) versus time {see text} for [cyclohexene] =
Fig. 7 IP(t) as a function of time of the system CDT–cyclohexene–
H2O2 in THF at different cyclohexene concentrations. The inset shows
the plot of 1/√IP(t) versus time for the system (c). Conditions:
[CDT] = 2.5 × 10Ϫ2 mol dmϪ3; [H2O2] = 2.5 × 10Ϫ1 mol dmϪ3; T = 10 ЊC.
2.68 × 10Ϫ2 mol dmϪ3. Conditions: [TCNP] = 5.35 × 10Ϫ2 mol dmϪ3
;
[H2O2] = 5.35 × 10Ϫ1 mol dmϪ3; T = 20 ЊC. M denotes the concentration
in mol dmϪ3
M denotes the concentration in mol dmϪ3
.
.
IP(t)-signal of about 4 mV is observed which increases enor-
mously at about 9 seconds. As observed for the system TCNP–
cyclohexene–H2O2 the greater the cyclohexene concentration
the longer the time at which the IP(t)-signal is observed. But in
contrast to the system TCNP–cyclohexene–H2O2 the height of
the narrow IP(t)-signal increases with increasing cyclohexene
concentrations and at equal concentration of cyclohexene and
hydrogen peroxide (system (c)) the highest IP(t)-signal could be
detected.
Moreover, the shape of the IP(t)-signals can only be described
by a second-order reaction as the straight line of the plot of
1/√IP(t) versus time for the system (c) given in the inset of Fig. 7
demonstrates.
tion can be approximately evaluated. For equal concentrations
of TCNP and cyclohexene of 5.35 × 10Ϫ2 mol dmϪ3 and
[H2O2] = 5.35 × 10Ϫ1 mol dmϪ3 (see legend of Fig. 6) 1O2 form-
ation is not observed within the first 800 s. From this result it
can be deduced that within this time the following inequality
holds:
k19[H2O2]
k
20 ӷ
[cyclohexene]
with [H2O2] ≈ [H2O2]0 Ϫ [TCNP] and k19 = 7.1 × 10Ϫ4 dm3
molϪ1 sϪ1 one obtains
These results cannot be explained by the reaction mechan-
isms described so far. But it must be noted that they do not
contradict the observation of Rebek et al.,2 who found that 90%
of cyclohexene is epoxidized by CDT–H2O2. The amounts of
1O2 generated in both systems under the conditions employed
are small (<4%) compared to the amounts expected for the
systems investigated in the absence of cyclohexene.
(7.1 × 10Ϫ4) × (4.81 × 10Ϫ1)
k
20 ӷ
= 6.4 × 10Ϫ3 dm3 molϪ1 sϪ1,i.e.
5.35 × 10Ϫ2
k20 ≈ 6 × 10Ϫ2 dm3 molϪ1 sϪ1.
Thus 4 is, by a factor of 50, less reactive towards cyclo-
hexene than 1, but in comparison with conventional peroxy-
carboxylic acids like peroxybenzoic acid 4 is about a factor of
100 more reactive.2 In this respect it is interesting to recall the
fact that 4 is about a factor of 50 less reactive towards hydrogen
peroxide than 1. Obviously the reactivity of 1 and 4 towards
hydrogen peroxide runs parallel to the reactivity of both
peroxycarboxylic acids towards cyclohexene.
With respect to the shape of the IP(t)-signals shown in Fig. 6
it must be stated that the changes in IP(t) cannot be described
exactly either by a first-order reaction or by a second-order
reaction. For a second-order reaction a plot of 1/√IP(t) versus
time should give a straight line.16 Such a plot of the first IP(t)-
signal is given in the inset of Fig. 6. From this plot it can be
deduced that in the system TCNP–cyclohexene–H2O2, under
the conditions used, a second-order reaction distinctly contrib-
utes to the overall formation of 1O2 (vide infra).
In addition we have performed similar measurements with
the system CDT–H2O2 at different cyclohexene concentrations
in THF at T = 10 ЊC. Surprisingly this system exhibits different
behaviour.
As shown in Fig. 7 for the lowest cyclohexene concentration
of [cyclohexene] = [H2O2]/20 = 1.25 × 10Ϫ2 mol dmϪ3 (system
(a)) about two seconds after addition of hydrogen peroxide an
In connection with the results presented in Figs. 6 and 7 the
behaviour of the system CDI–H2O2 is of interest.
C.2 Investigation of the system N,NЈ-carbonyldiimidazole
(CDI)–H2O2. Measurements with the system CDI–H2O2 in
THF carried out by Rebek et al.2 have shown that this system is
unsuitable for the epoxidation of cyclohexene. Only 1.5% 1,2-
epoxycyclohexane was obtained. We have investigated this
system under pseudo first-order conditions with [CDI] = 2.5 ×
10Ϫ2 mol dmϪ3 and [H2O2] = 2.5 × 10Ϫ1 mol dmϪ3 at T = 10 ЊC.
The result is given in Fig. 8.
1
Surprisingly, O2 formation is observed. But the amount of
1O2 is very small compared with the amount generated in the
system CDT–H2O2 under the same conditions. However it
is comparable to that of system (c) presented in Fig. 7. In
addition the IP(t)-signal is also consistent with a second-order
reaction, which is evidently shown by the plot of 1/√IP(t) versus
t given in the inset of Fig. 8. It must be noted that neither in this
case nor in the two cases discussed above can the second-order
rate constant be evaluated, since neither the exact rate law nor
the concentrations of the corresponding reacting compounds
are known.
As already reported by Rebek et al.2 in the system N,NЈ-
J. Chem. Soc., Perkin Trans. 2, 1999, 2579–2587
2585

View Article Online
Fig. 8 IP(t) as a function of time of the system CDI–H2O2 in THF.
The inset shows the corresponding plot of 1/√IP(t) versus time. Condi-
Fig. 9 IP(t) as a function of time of the system CDT–1,2-epoxy-
cyclohexane–H2O2 in THF. The inset shows the corresponding plot of
tions: [CDI] = 2.5 × 10Ϫ2 mol dmϪ3; [H2O2] = 2.5 × 10Ϫ1 mol dmϪ3
T = 10 ЊC.
;
1/√IP(t) versus time. Conditions: [CDT] = 2.5 × 10Ϫ2 mol dmϪ3; [1,2-
epoxycyclohexane] = 2.5 × 10Ϫ1 mol dmϪ3
dmϪ3; T = 20 ЊC.
;
[H2O2] = 2.5 × 10Ϫ1 mol
carbonyldi-2,4-dimethylpyrazole–H2O2 and in the system CDI–
H2O2 coloured side products are developed. It can be assumed
that in both systems in analogy to the system CDT–H2O2 in the
first addition–elimination process a very reactive peroxycarb-
oxylic acid is formed and that the corresponding acids can
react with the azoles generated in the first process. The assump-
tion that a reactive peroxycarboxylic acid can react for instance
with imidazole is confirmed by the observation that the in situ
generated peroxycarboxylic acid 1 reacts with imidazole. Add-
ition of an excess of imidazole to the system CDT–H2O2 results
immediately in the quenching of 1O2 formation.
In contrast to 1,2,4-triazole both 2,4-dimethylpyrazole and
imidazole contain a carbon–carbon double bond, which prob-
ably can be epoxidized by a reactive peroxycarboxylic acid.
Assuming that among other things the epoxide can react
further with the reactive peroxycarboxylic acid according to
eqn. (21) the IP(t)-signal observed for the system CDI–H2O2 can
be explained.
Apart from this the different behaviour observed for the
systems CDT–cyclohexene–H2O2 and TCNP–cyclohexene–
H2O2 (see Figs. 6 and 7) obviously reflects the different
reactivity of the in situ generated peroxycarboxylic acids 1
and 4. For both systems it can be deduced that in the
presence of a large excess of cyclohexene compared to CDT
and TCNP, respectively, and even at equal concentrations of
cyclohexene and hydrogen peroxide the reactions between the
peroxycarboxylic acids and hydrogen peroxide cannot com-
pete with the corresponding epoxidation reactions. Moreover,
for the system TCNP–cyclohexene–H2O2 this may hold also
for the reaction between 4 and 1,2-epoxycyclohexane, whereas
in the system CDT–cyclohexene–H2O2 the reaction between
1 and 1,2-epoxycyclohexane clearly can compete with the
formation of 1,2-epoxycyclohexane. Consequently in the
system CDT–cyclohexene–H2O2 the rate of the reaction
between 1 and 1,2-epoxycyclohexane is high enough under the
conditions used that 1O2 generated in this reaction can be
detected.6
The IP(t)-signals of the system TCNP–cyclohexene–H2O2
observed at low cyclohexene concentrations can be explained by
1
the contribution of two reactions to the overall O2 formation,
namely by the reaction between 4 and 1,2-epoxycyclohexane
and by the reaction between 4 and hydrogen peroxide.
Further it should be noted that in contrast to 1 and 4 neither
hydrogen peroxide (30%) nor conventional peroxycarboxylic
acidslikem-chloroperoxybenzoicacidreactwith1,2-epoxycyclo-
hexane to generate 1O2.
This possibility has encouraged us to assume that 1,2-epoxy-
cyclohexane produced in the systems TCNP–cyclohexene–
H2O2 and CDT–cyclohexene–H2O2 could react with 4 and 1,
respectively, to give 1O2 according to eqn. (22).
Conclusions
The results of the investigation of the system CDT–H2O2 in
THF are consistent with the assumption that in this system
1H-1,2,4-triazoleperoxycarboxylic acid 1 is formed as first
postulated by Rebek et al.2
The results of the investigation of the system TCNP–H2O2 in
THF are in accordance with the assumption that in this system
O-p-nitrophenylmonoperoxycarbonic acid 4 is generated.
Both peroxycarboxylic acids are powerful epoxidizing agents.
The reactivity of 1 towards olefins is comparable with that of
dimethyldioxirane, e.g., 1 is, by a factor of about ten thousand,
more reactive towards cyclohexene than conventional peroxy-
carboxylic acids like peroxybenzoic acid. The reactivity of 4
Attempts with the systems CDT–1,2-epoxycyclohexane–
H2O2 and TCNP–1,2-epoxycyclohexane–H2O2 confirm this
hypothesis. Fig. 9 shows for example the IP(t)-signal observed
for the system CDT–1,2-epoxycyclohexane–H2O2. In the inset
the plot of 1/√IP(t) versus t is given, which yields a straight line
as expected for a second order reaction (see eqn. (22)).
However, it is evident that only a small amount of 1,2-
epoxycyclohexane reacts with the in situ generated 1 to give
1O2. The same observation is made for the system TCNP–1,2-
epoxycyclohexane–H2O2. As yet we have no plausible explan-
ation for these observations.
2586
J. Chem. Soc., Perkin Trans. 2, 1999, 2579–2587

View Article Online
6 K. Böhme and H.-D. Brauer, Inorg. Chem., 1992, 31, 3468.
towards olefins exceeds after all that of peroxybenzoic acid by a
factor of one hundred.2
7 R. Schmidt and E. Afshari, J. Phys. Chem., 1990, 94, 4377.
8 R. D. Scurlock, S. Nonell, S. E. Braslawsky and P. R. Ogilby,
J. Phys. Chem., 1995, 99, 3521.
The disadvantage of the in situ generated peroxycarboxylic
acids 1 and 4 is that they are also very reactive towards hydro-
gen peroxide. Thus in the case of less reactive olefins (e.g. dec-1-
ene) this reaction successfully competes with the epoxidizing
reaction resulting in a low efficiency of the formation of the
epoxide desired.2
Whereas derivatives of the carbonic acid containing one
hydroxy group are very unstable,17 the lifetimes of 1 and 4 calcu-
lated under the pseudo first-order conditions demonstrate that
both peroxycarboxylic acids (which contain a peroxy hydroxy
group) are more stable than their parent carboxylic acids.
9 A. A. Frost and R. G. Pearson, Kinetics and Mechanism, 2nd edn.,
J. Wiley & Sons, New York, 1961, p. 178.
10 (a) R. Curci and J. O. Edwards, in Catalytic Oxidations with
Hydrogen Peroxide as Oxidant, ed. G. Strukal, Kluwer Academic
Publishers, Dordrecht, 1992, p. 45; (b) C. K. Ingold, Structure and
Mechanism in Organic Chemistry, 2nd edn., Cornell University
Press, Ithaca, NY, 1969, p. 1129.
11 R. Curci, R. A. DiPrete, J. O. Edwards and G. Modena, J. Org.
Chem., 1970, 35, 740 and references cited therein.
12 In this respect it should be noted that measurements of the rate of
1O2 formation are very sensitive to impurities (quenchers of 1O2)
and slight changes in the surroundings. Whereas in general rate
constants are determined by measuring the changes in the
concentrations of the educts and/or products, here the derivation
of the concentration of 1O2 is determined. Already insignificant
changes of [1O2] in a given time can cause distinct changes in the rate
d[1O2]/dt.
Acknowledgements
Financial support by the Deutsche Forschungsgemeinschaft
(Schwerpunktprogramm “Sauerstofftransfer/Peroxidchemie”)
and the Fonds der Chemischen Industrie is gratefully
acknowledged.
13 A. L. Baumstark and C. J. McCloskey, Tetrahedron Lett., 1987, 28,
3311.
14 P. A. LaBonville, R. Kugel and J. R. Ferraro, J. Phys. Chem., 1977,
67, 1477.
15 R. M. Coates and J. W. Williams, J. Org. Chem., 1974, 39, 3054.
16 A. Lange and H.-D. Brauer, J. Chem. Soc., Perkin Trans. 2, 1996,
805.
17 H. R. Christen, Grundlagen der Organischen Chemie, 1. Aufl.,
Sauerländer, Aarau, 1970, p. 271.
References
1 H. A. Staab, Angew. Chem., 1962, 74, 407.
2 J. Rebek, R. McCready, S. Wolf and A. Mossman, J. Org. Chem.,
1979, 44, 1485.
3 R. Schmidt and H.-D. Brauer, J. Am. Chem. Soc., 1987, 109, 6976.
4 E. Höft and S. Ganschow, J. Prakt. Chem., 1972, 314, 156.
5 R. Schmidt, Chem. Phys. Lett., 1988, 151, 369.
Paper 9/03852C
J. Chem. Soc., Perkin Trans. 2, 1999, 2579–2587
2587