Mono, di and trifunctional cyclic organic peroxides: The effect of substituents and ring size on their thermolysis in 1,4-dioxan

Article abstract of DOI:10.1071/CH13171

The thermal decomposition reaction of cyclic organic peroxides was studied in 1,4-dioxan at initial concentrations between ～10-4 and 10-2molL-1 and at a temperature interval between 100 and 170°C, according to the thermal stability of each compound. The kinetic behaviour observed in all systems studied follows a pseudo first order kinetic law up to at least ～86% of peroxide conversion. An important substituent effect is operative on the rate constant values and consequently on the activation parameters of the thermal decomposition reaction. The application of different treatments (compensation affect or a statistical treatment) on the kinetic data shows the existence of two sets of cyclic peroxides with comparable kinetic behaviour. Different peroxide-solvent interaction mechanisms can be considered within each series.

Full text of DOI:10.1071/CH13171

CSIRO PUBLISHING
Aust. J. Chem. 2013, 66, 1080–1087
Full Paper
http://dx.doi.org/10.1071/CH13171
Mono, Di and Trifunctional Cyclic Organic Peroxides:
The Effect of Substituents and Ring Size on their
Thermolysis in 1,4-dioxan
A B C
, ,
B
B
Rosa Nesprias,
Gladys Eyler, and Adriana Canizo
˜
A
Department of Science Basic, Agronomy Faculty, National University of Center of The
Buenos Aires Province (UNCPBA), Italy Republic 780, (PC: 7300), Azul, Buenos Aires,
Argentina.
B
Department of Chemistry Engineering, Engineering Faculty, National University of the
Center of the Buenos Aires Province (UNCPBA), del Valle Avenue 5737, (B7400JWI)
Olavarr´ıa, Buenos Aires, Argentina.
C
Corresponding author. Email: knespria@faa.unicen.edu.ar
The thermal decomposition reaction of cyclic organic peroxides was studied in 1,4-dioxan at initial concentrations
between ,10ꢀ4 and 10ꢀ2 mol Lꢀ1 and at a temperature interval between 100 and 1708C, according to the thermal stability
of each compound. The kinetic behaviour observed in all systems studied follows a pseudo first order kinetic law up to at
least ,86 % of peroxide conversion. An important substituent effect is operative on the rate constant values and
consequently on the activation parameters of the thermal decomposition reaction. The application of different treatments
(compensation affect or a statistical treatment) on the kinetic data shows the existence of two sets of cyclic peroxides with
comparable kinetic behaviour. Different peroxide–solvent interaction mechanisms can be considered within each series.
Manuscript received: 18 April 2013.
Manuscript accepted: 20 May 2013.
Published online: 15 July 2013.
Introduction
free radicals of trioxane. Finally, the damage of key proteins
necessary for the parasite’s survival occurs. The structural
arrangement of the heterocyclic ring in the region of the peroxide
group is essential information to investigate its activation mech-
anism. A strong indication about the importance of the steric
hindrance in the reaction with iron(II) has been reported.[5]
Onthe other hand, cyclic organic peroxides are usually usedas
free radical initiators in free radical polymerization of most
commercial, industrially produced vinyl polymers such as poly-
styrene, poly(methyl methacrylate), polyethylene, and poly(vinyl
chloride). Cyclic organic di- and triperoxides decompose ther-
mally by homolytic cleavage of the O–O bond to give an
intermediate biradical, which in turn initiates the polymerization
of vinyl monomers. The biradical species is incorporated into the
polymerstructureyieldingpolymericspeciescontainingperoxide
groups that decompose later during the polymerization process.
Since the nature of the biradical formed is dependent on the
peroxide structure and has a strong influence on the final polymer
properties, it is important to clarify the initiation mechanism of
organic peroxides, in particular di- and triperoxides, which can be
considered as multifunctional initiators as a consequence of the
presence of two or three active sites in the peroxide molecule.[6]
In this work, a kinetic study of the thermal decomposition of
different mono-, di-, and trifunctional cyclic organic peroxides
1–3, 5–7, and 9 (Figs 1–3) in 1,4-dioxan at different tempera-
tures is reported. Kinetic and activation parameters can be
compared with those previously found for 4,[7] 8,[8] and 10[9]
(Figs 1–3) in the same solvent.
In recent years, cyclic organic peroxides with one or more per-
oxidic bonds (O–O) in the molecule, have gained great impor-
tance worldwide due to the different uses and applications found
for this family of compounds. Among the most important are
substituted 1,2,4-trioxacyclohexanes with interesting pharma-
cological properties. On the other hand, substituted 1,2,4,5-
tetraoxacyclohexanes and 1,2,4,5,7,8-hexaoxacyclononanes are
used in industry as polyfunctional initiators of polymerization.[1]
Currently 300–500 million people worldwide have malaria.
Each year, 1–2 million people, mostly young children in Africa,
Asia, and South America, die as a result of this infectious
disease.[2] There are several drugs that, in the past, were
effective against malaria, spread by the parasite Plasmodium
falciparum. Due to the increasing resistance of this kind of
malaria to traditional drugs such as chloroquine, considerable
effort has been made in the past years to discover more powerful
chemotherapeutics against the disease.[3,4] Artemisinin, a natu-
ral product of the vegetal extract of the Artemisia annual used in
an old Chinese fever-reducing medicine is the prototype of a
new class of anti-malarial drugs.
The artemisinin-like high activity might be attributed to a cis-
fused bicyclic ring of 1,2,4-trioxacyclohexane and reinforced by
an aryl substituent attached to the heterocycle.[3,4] The existence
of a peroxidic bond in the 1,2,4-trioxane ring is essential for the
anti-malarial activity of this class of drugs. These molecules act
onhemereleasedby the parasite. The transference ofone electron
from iron(II) to peroxide leads to the formation of carbon-centred
Journal compilation Ó CSIRO 2013
www.publish.csiro.au/journals/ajc

Cyclic Organic Peroxides: Thermolysis
1081
H
R3
Table 1. Pseudo-first-order rate constant values (kexp) for the thermal
R1
decomposition reaction of cyclic organic peroxides in 1,4-dioxan
O
O
O
Peroxide
T [8C]
kexp (ꢁ 10ꢀ6) [sꢀ1
]
rA
R2
1
110
120
130
140
120
130
140
150
110
120
130
140
135
143
143
150
150
165
140
150
160
170
120
130
140
150
140
150
160
170
140
150
160
170
140
150
160
170
140
145
150
165
172
39.1
117
0.999
1 R1 ꢀ R2 ꢀ ꢁC6H5; R3 ꢀ ꢁ(CH2)4ꢁ
327
2 R1 ꢀ R2 ꢀ ꢁC6H5; R3 ꢀ ꢁ(CH2)2 CO(CH2)2ꢁ
788
3 R1 ꢀ R2 ꢀ ꢁC6H4F; R3 ꢀ ꢁ(CH2)4ꢁ
2
61.7
213
0.982
0.993
0.988
CH3
672
1105
69.9
143
CH3
O
3
O
O
342
4
991
4B
19.8
41.1
31.9
70.7
72.0
158
Fig. 1. Chemical structures of the 1,2,4-trioxacyclohexanes: cis-6-phenyl-
5,6-(2-phenylpropylidene)-3,3-tetramethylene-1,2,4-trioxacyclohexane (1),
cis-6-phenyl-5,6-(2-phenylpropylidene)-3,3-(pentan-3-ona)-1,2,4-trioxa-
cyclohexane(2), cis-6-p-fluorphenyl-5,6-(2-p-fluorphenylpropylidene)-3,
3-tetramethylene-1,2,4-trioxacyclohexane (3), and trans-3,3-dimethyl-
5,6-tetramethylene-1,2,4-trioxacyclohexane (4).
5
16.6
73.0
216
0.997
0.989
0.999
0.990
0.999
0.998
R1
792
6
18.5
70.4
132
O
R2
O
O
O
R2
349
7
4.84
14.8
23.2
68.2
9.00
18.5
38.5
117
R1
5 R1 ꢀ R2 ꢀ ꢁ(CH2)2CH3
6 R1 ꢀ ꢁCH3; R2 ꢀ ꢁC(CH3)3
7 R1 ꢀ R2 ꢀ ꢁCH3
8C
8 R1 ꢀ ꢁC6H5; R2 ꢀ ꢁCH3
9
1.18
6.43
14.1
70.1
45.0
105
Fig. 2. Chemical structures of the 1,2,4,5-tetraoxacyclohexanes:
3,3,6,6-tetra-n-propyl-1,2,4,5-tetraoxacyclohexane (5), 3,6-ditertbutyl-3,
6-dimethyl-1,2,4,5-tetraoxacyclohexane (6), 3,3,6,6-tetramethyl-1,2,4,
5-tetraoxacyclohexane (7), and trans-3,6-dimethyl-3,6-diphenyl-1,2,
4,5-tetraoxacyclohexane (8).
10D
233
O
O
R
R
R
R
1680
2150
O
O
ACorrelation coefficient according to the Arrhenius equation.
BRef. [7].
CRef. [8].
DRef. [9].
O
O
R
R
9 R ꢀ ꢁCH3
systems studied follows a pseudo first order kinetic law up to at
least ,86 % of peroxide conversion. The initial peroxide con-
centration is relatively low to avoid effects of induced second
order decomposition reactions that would accelerate the thermal
decomposition of the peroxides at higher conversions.
10 R ꢀ ꢁCH2CH3
Fig. 3. Chemical structures of the 1,2,4,5,7,8-hexaoxacyclononanes:
3,3,6,6,9,9-hexamethyl-1,2,4,5,7,8-hexaoxacyclononane (9) and 3,3,6,6,9,
9-hexaethyl-1,2,4,5,7,8-hexaoxacyclononane (10).
The temperature effect on the experimental rate coefficient
(kexp) in the investigated solvent for each of the investigated
compounds can be represented by the Arrhenius equation (see
Table 2 for data). The linearity of the Arrhenius equations over a
relatively large temperature interval (308C) suggests that calcu-
lated activation parameters for all the organic cyclic peroxide
thermal decomposition reactions in 1,4-dioxan belong to a simple
process. The rate constant values (Table 1) could be associated
Results and Discussion
The thermal decomposition reaction of organic cyclic peroxides
was studied in 1,4-dioxan at initial concentrations between
,10ꢀ4 and 10ꢀ2 mol Lꢀ1 and at a temperature interval between
100 and 1708C, according to the thermal stability of each
compound (Table 1). The kinetic behaviour observed in all

1082
R. Nesprias, G. Eyler and A. Can˜izo
(a)
(b)
(c)
Table 2. Data for the Arrhenius equations for the experimental rate
R
R
coefficients (kexp) of compounds 1 3, 5 7, and 9 in 1,4-dioxan
]
]
The Arrhenius equation is given by ln kexp (sꢀ1) ¼ ln A – Ea/RT, where Ea is
the activation energy, where conversion factor is 1 cal ¼ 4.19 joul, A is a pre-
exponential factor, R is the gas constant, and T is the temperature (K). The
errors correspond to the standard deviations derived from a treatment of the
kinetic data by the method of minimum squares.[10]
Products
R
R
O
O
O
O
R1
R1
Compound
lnA
Ea [cal molꢀ1
]
R1
R1
1
2
3
5
6
7
9
31.4 ꢃ 0.7
31.9 ꢃ 3.9
28.5 ꢃ 1.7
47.0 ꢃ 1.7
29.3 ꢃ 3.0
26.2 ꢃ 3.1
46.0 ꢃ 4.8
31610 ꢃ 292
O
R2
32391 ꢃ 1264
29158 ꢃ 690
47621 ꢃ 711
31298 ꢃ 1241
31535 ꢃ 1320
48981 ꢃ 2052
O
R2
O
O
O
O
Products
O
O
R2
R2
R1
R1
R1
R2
R1
R2
R2
R1
O
O
O
O
R2
R1
O
O
with the initial O–O bond rupture to form an intermediate
biradical species. Thus, the first step in the thermolysis mecha-
nism of these peroxides in solution is initiated by a process of the
type represented in Scheme 1 for generic molecules of a substi-
tuted mono-, di-, and trifunctional cyclic organic peroxide.
In previous studies,[11] it has been demonstrated that the
initial biradical species undergo a further thermal decomposi-
tion either by C–C, C–O, or O–O bond rupture, to form free
radicals that react with the solvent or combine together to
generate a different final product.
O
O
Products
O
O
O
O
R1
R2
R1 R2
Scheme 1.
Schuster and Bryant[12] in 1978 demonstrated that upon
thermolysis the trioxane 4 undergoes unimolecular cleavage to
form acetone and adipaldehyde and reported that the biradical
formed from homolysis of the O–O bond (Scheme 1a) must
rearrange rapidly by an a-cleavage reaction.
Cyclopentanone was the main reaction product of the decom-
position of 1 and 3 (Scheme 2) as identified by HPLC and
originated from C–O rupture of the initial biradical. Moreover,
the observation of benzene as a minor product from the decom-
position of 1 and 2 evidences the C–C bond rupture to form a
phenyl radical which later extracts hydrogen from the solvent
(Scheme 3 where R ¼ Ph).
(Scheme 4, path a) abstracts a hydrogen atom from the solvent
to form 2-methylpropane (Scheme 3, where R ¼ (CH3)3C) and
2-tertbutyl-1,4-dioxane (Scheme 5, where R1 ¼ (CH3)3C).
The decomposition of 7 gives acetone (Scheme 4, paths a and
b, where R1 ¼ R2 ¼ CH3), methane (Scheme 3), and ethane
(Scheme 4, path f) as the main organic products identified, in
all cases R ¼ CH3.
During the thermal decomposition of 9 (Scheme 1c), meth-
ane (Scheme 3, where R ¼ CH3), ethane (Scheme 4, path f),
´
acetone (Scheme 6, upper), and a solvent derived product (2,2-
bi-1,4-dioxan) (Scheme 6, lower) were formed.
According to the kinetic data reported in Table 1 monofunc-
tional cyclic organic peroxides (Fig. 1) are more reactive than
the di- and trifunctional compounds investigated in this work
(Figs 2 and 3).
Compounds 1–3 have similar structures around the 1,2,4-
trioxacyclohexane ring in a chair conformation, have large
volumes (Table 3), possess aromatic rings or an unsaturated
ring bonded directly to the peroxidic cycle, and a cyclic
substituent of five or six members, and as a consequence they
show analogous kinetic behaviour (Table 1). The high reactivity
of these trioxanes is probably associated with the large size of
the substituents.
Thus, once the ring is opened through the initial O–O bond
(Scheme 1a) the intermediate species formed would adopt
highly stable conformations within a new reorganized solvent
cage that complicates the recyclization process. Consequently,
the reaction progresses with the decomposition of the initial
biradical to form the detected reaction products with high values
of kexp (Table 1).
The biradical species obtained from the decomposition of
compound 5 (Scheme 1b) can undergo either C–O bond cleav-
age, to give 4-heptanone and O2 (Scheme 4, paths a and b, where
R1 ¼ R2 ¼ CH3CH2CH2), or O–O and C–C bond cleavages
occur to give n-propyl and butanoate radicals (Scheme 4, paths
c and d). Butanoate radicals can abstract a hydrogen atom from
the solvent to give butanoic acid and a new radical species (Sꢂ)
derived from the solvent (Scheme 3, where R ¼ R2COO) or take
part in some radical–radical reaction to form n-propylbutanoate
(Scheme 4, path e). A radical–radical coupling reaction between
two n-propyl radicals may explain the n-hexane detected as a
product (Scheme 4, path f). The presence of 4-heptanone as a
peroxide decomposition product was verified through the inter-
pretation of the mass spectrum and also by comparison of
retention times with an authentic sample of this compound.
The presence of hexane, n-propyl butanoate, and propane was
qualitatively identified by interpretation of their mass
spectrum.[13]
3,3-Dimethyl-2-butanone (Scheme 4, paths a and b,
where R1 ¼ (CH3)3C, and R2 ¼ CH3) and 2-methylpropane
(Scheme 3, where R ¼ (CH3)3C), were detected by gas
chromatography–mass spectrometry (GC-MS). The radical
species (2-methylpropyl radical) formed by C–C bond rupture
Studies carried out by Jubert[14] et al. on trioxane 1 have
revealed that there are 18 different conformers geometrically
optimal within the energy range of 14 kcal molꢀ1 corresponding
to a mixture of chair and boat conformations of the molecule

Cyclic Organic Peroxides: Thermolysis
1083
R
R
ꢂ R ꢂ Other products
R
R
O
O
O
O
O
Scheme 2.
O
O
•
•
O
O
R
R
R
R• ꢂ HS
RH ꢂ S•
where S ꢀ
•
R
O
O
O
O
ꢂ
2R2CO
O
O
O
O
Scheme 3.
R
2
R
R
R
R1
O
O
•
R1
R2
O
•
O
•
R2
•
O
O
a
where S ꢀ
S
S2
ꢂ
R1COR2
O
•
O
O
R2
R1
Scheme 6.
c
b
•
R1
R2
O
Table 3. Calculated van der Waals volumes for the different cyclic
organic peroxides according to the expression deduced by Zhao et al.[25]
d
e
R2COOR1
•
R1 ꢂ R2COO
•
2
•
O
3
van der Waals volume [A molecule
ꢀ1
]
˚
Peroxide
f
2R1•
R1R1
1
313.5
336.9
318.4
165.9
273.6
238.9
135.2
245.8
203.9
308.4
Scheme 4.
2
3
4
O
O
5
6
•
R1• ꢂ S
R1S
where S ꢀ
7
8
9
10
Scheme 5.
studied. The existence of different conformations for 1,2,4-
trioxacyclohexane plus the large size of substituents in 1–3
(Fig. 1) gives good reasons to interpret the differences observed
in their reactivities in solution assuming that they have differ-
ences in their solvation mechanisms.
Table 4. First order rate constant values (kexp) for the thermal
decomposition reaction of cyclic peroxides in different solvents
The van der Waals volumes of trioxanes 1–3 approximately
duplicate the size of 4[7] (Fig. 1, Table 3) which behaves as the
most stable member of this family (Table 1). Compound 4 has
methyl groups attached directly to the cycle and decomposes 24
times more slowly than 3, 19 times more slowly than 1, and 16
times more slowly than 2. It can be postulated that the compound
4–solvent interaction mechanism within the solvent cage would
be different from that proposed for trioxanes 1–3.
The rate constant values are influenced by changing substit-
uent as can be clearly observed when the reactivity of diperoxide
7 (methyl substituted) is compared with 5, 6, and 8 in 1,4-dioxan
(Table 1) and in other organic solvents like toluene, acetone, and
propan-1-ol (Table 4). The presence of methyl groups as
Peroxide
kexp (ꢁ 10ꢀ6) [sꢀ1
Acetone (1508C)
]
Toluene (1458C)
Propan-1-ol (1508C)
1A
4A
1170
4
–
–
–
–
5A,B
6A,B
7A,B
8A
9A,B
10A,B
59
183
1178
6.67
–
591
1283
26.7
–
560
2.8
2.2
4.9
66
9.1
255
25
268
ARef. [16].
BRef. [17].

1084
R. Nesprias, G. Eyler and A. Can˜izo
substituents of the tetraoxacyclohexane ring in 7 clearly stabi-
lizes the molecule, probably because it is confined in a compact
solvent cage where the recyclization reaction is highly probable.
Similar kinetic results have been obtained in previous works
while studying the thermal decomposition in benzene,[15] tolu-
ene,[16] acetone, and propan-1-ol,[17] which demonstrates that 7
decomposes at the lowest rates in comparison with those cyclic
peroxides with voluminous substituents (e.g., n-propyl, tert-
butyl, and phenyl) (Table 4).
A comparison between the molecular structures of 9[18,19]
and 10[20] shows that substitution of a methyl group by an ethyl
groups does not introduce additional strain in the cyclononane
ring system. The hexaoxacyclononane core in 10 is stabilized in
a twisted-boat–chair conformation with a local symmetry of
approximately D[3,20] very close to that previously described for
9.[19] Triperoxide 9, in its solid state, has been used as a primary
explosive in terrorist acts[21] such as the 2005 London bomb-
ing,[22] nevertheless, in its crystalline form 10 is not as sensitive
to impact or temperature changes as 9. On the other hand, when
methyl groups are replaced by ethyl groups as substituents on the
hexaoxacyclononane ring (10 in Fig. 3) the reactivity in solution
is substantially increased (Table 1 and 4). In this work, it was
found that 10 is 38 times more reactive than 9 in 1,4-dioxan at
1408C, even so the differences in reactivity are greater at higher
temperatures.
It has been demonstrated that 9 is highly stable in different
organic solvents (Table 4).[23,24] On the other hand, when
comparing acetone cyclic peroxides 7 and 9 with cycles of
different size it is possible to observe that in 1,4-dioxan at 1708C
their reactivities are practically the same, as it was observed in
propan-1-ol at 1508C (Table 4).
The dependence of the rate constant values at 1408C on the
calculated van der Waals volumes for the cyclic organic per-
oxides (Table 3) were in accord with the expression deduced by
Zhao et al.,[25] and is represented in Fig. 4. Compounds 4, 7, and
9 have the smallest volumes in each family, include methyl
substituents, and show the lowest reactivity in solution. It is
likely that the solvent molecules are able to interact more
strongly with a small peroxide molecule generating a close
packed system. Thus, peroxides 4, 7, and 9 in solution would be
confined to a small ‘solvent cage’ with strong interaction
energy, delaying the subsequent decomposition of the critical
state of the reaction (formation of a biradical species) to yield the
corresponding reaction products (Scheme 1).
Diperoxide 6 has shown a very high reactivity not only in 1,4-
dioxan (Table 1) but in aprotic non-polar (toluene),[16] aprotic
mildly polar (acetone), and protic solvents (propan-1-ol)
(Table 4),[17] probably because of the presence of the bulky
tert-butyl groups which causes a particular substituent effect on
the rate constant values assigned to the O–O bond linkage. The
thermal stability of 6 in solution appears close to that observed
for voluminous trioxanes 1–3 rather than diperoxides 7 and 8
with similar van der Waals volumes (Fig. 4). It can be assumed
that the intermediate biradical species resulting from diperoxide
6 (Scheme 1b), would tend to quickly move away from the
‘solvent cage’, rather than reform the cyclic peroxide. There-
fore, the reaction progresses in the direction of the products and
kexp reaches very high values compared with those determined
for the others diperoxides investigated (Table 1 and 4).
Activation Parameters
The peroxidic compounds studied in the present work show
structural differences that typically affect the reactivity of
these molecules in solution, and also generate a significant
variation on the activation parameters (DDHz ¼ 24.1 kcal molꢀ1
and DDSz ¼ 51.8 cal molꢀ1 Kꢀ1 or DDHz ¼ 100.9 kJ molꢀ1 and
DDSz¼ 216.5 J molꢀ1Kꢀ1, where conversion factor is 1cal ¼
4.19 J) assigned to the unimolecular thermal decomposition
reaction. Those variations are large compared with the maxi-
mum probable errors of those parameters (2.1 kcal molꢀ1 and
4.8 cal molꢀ1 Kꢀ1, respectively) (Table 5).
By evaluating the activation parameters (Table 5), it can be
observed that a change in the activation enthalpy (DHz) is
accompanied by a change in the same sense in the corresponding
activation entropy (DSz). This phenomenon is known as the
compensation effect and can be represented by a linear relation-
ship between the activation parameters (Eqn 1) where the slope
b, known as the isokinetic temperature, has an important
physical meaning; it represents a temperature at which all
reactions of the series should proceed at the same rate.[26]
z
z
z
DH ¼ DH0 þ bDS
ð1Þ
ꢁ2
ꢁ3
ꢁ3
ꢁ4
ꢁ4
3
1
2
6
4
10
ꢁ5
5
ꢁ5
7
8
Triperoxide
Diperoxide
Trioxane
ꢁ6
9
ꢁ6
ꢁ7
100
150
200
250
300
350
Van der Waals volume [A°3 moleculeꢃ1
]
Fig. 4. Dependence of log kexp for the unimolecular thermolysis reaction of different cyclic organic peroxides in 1,4-
dioxan at 1408C with the van der Waals volume of each molecule.

Cyclic Organic Peroxides: Thermolysis
1085
50
45
40
35
30
25
20
Table 5. Activation parameters for the unimolecular thermal decom-
position reaction of different cyclic organic peroxides in 1,4-dioxan
DT, change in temperature; DHz, activation enthalpy; DSz, activation
entropy; DGz, free energy of activation
9
5
10
Peroxide
DT
[8C]
DHz
[kcal molꢀ1
DSz
DGzA
[kcal molꢀ1
]
a
]
[cal molꢀ1 Kꢀ1
]
b
1
30
30
30
30
30
30
30
30
30
32
30.9 ꢃ 0.3
31.6 ꢃ 1.6
28.4 ꢃ 0.7
24.0 ꢃ 1.1
46.7 ꢃ 0.7
30.4 ꢃ 1.2
30.7 ꢃ 1.5
30.6 ꢃ 1.3
48.1 ꢃ 2.1
45.5 ꢃ 1.7
1.3 ꢃ 0.7
2.3 ꢃ 3.9
30.3 ꢃ 0.3
30.7 ꢃ 1.6
30.2 ꢃ 0.7
32.6 ꢃ 1.1
33.0 ꢃ 0.7
31.7 ꢃ 1.2
34.6 ꢃ 1.5
34.1 ꢃ 1.3
35.2 ꢃ 2.1
32.2 ꢃ 1.7
2
3
4B
ꢀ4.3 ꢃ 1.7
ꢀ19.8 ꢃ 2.6
32.0 ꢃ 4.8
ꢀ3.2 ꢃ 3.0
ꢀ9.2 ꢃ 3.5
ꢀ8.5 ꢃ 3.1
30.2 ꢃ 4.8
31.4 ꢃ 3.9
7
1
6
5
2
8
6
6
8C
3
9
10D
4
ACalculated at 1408C.
BRef. [7].
CRef. [8].
DRef. [9].
ꢁ30
ꢁ20
ꢁ10
0
10
20
30
40
ΔS‡ ꢀ [cal molꢁ1 Kꢁ1
]
Fig. 5. Representation of ‘isokinetic relationship’ according to the Leffler
treatment (Eqn 1) for the unimolecular thermolysis reaction of different
organic peroxides in 1,4-dioxan.
Two series of reactions can be observed when the relationship
according to Eqn 1 is studied (Fig. 5). One of the lines, line a,
includes peroxides with methyl substituents (4 and 7–9, series
a), and the other, line b, includes peroxides with bulky sub-
stituents and consequently greater volumes (1–3, 5, and 10,
series b). Both lines have excellent correlations coefficients
(line a, r ¼ 0.999 and line b, r ¼ 0.999). As it was previously
discussed, compound 6, with methyl and tert-butyl substituents,
shows an unexpectedly high reactivity in solution, nevertheless,
the activation parameters are similar to those obtained for other
peroxides, so it is not surprising that the corresponding point
appears in the middle of both correlation series (Fig. 5).
The linearity of both lines and the fact that they are nearly
parallel make it possible to consider them as two reaction
series[26] with a common reaction stepwise mechanism initiated
by the homolytic cleavage of one O–O bond. This initial step is
influenced by the dimensions of the substituents, the number of
O–O bonds in the cycle, the size of the cycle, and the different
conformations that these compounds may adopt, so, different
specific interactions mechanism can be considered. The activa-
tion parameter values obtained and the nature of the detected
products support the above proposed mechanism where specific
interactions between each peroxide molecule or biradical and
1,4-dioxan solvent molecules could be postulated. Changes
observed in activation parameters in lines a and b generate a
practically distinguishable variation of the free energy of acti-
vation (series a: DDGz ¼ 2.6 kcal molꢀ1 and series b: DDGz ¼
2.8 kcal molꢀ1). These later values are close to the order of the
calculated experimental error for these parameters.
proposed by Exner,[27] this point of intersection in the Arrhenius
plane, known as the inverse of the isokinetic temperature, has
greater reliability and validity because ln kexp and T are statisti-
cally independent.
The use of a simplified statistical technique based on the
least-squares method allows us to obtain the function Su, the
residual sum of squares.[28,29] In this procedure 1/b corresponds
with the minimum of function, a value coincident with the point
(or region) of intersection of the lines in the Arrhenius plane.
In Fig. 6 it is possible to differentiate two groups of
compounds called series a (containing peroxides with methyl
substituents directly attached to the cycle like 4 and 7–9), and
series b (including compounds with bulky substituents or fused
rings like 1–3, 5, and 10). Once again, diperoxide 6 was not
included in the treatment because its Arrhenius line moves away
from both series.
According to the mathematical treatment[27–29] applied for
the kinetic data of the thermolyses of 4 and 7–9 in 1,4-dioxan,
a
the isokinetic temperature value (b0 ) was 461 K (1888C). The
maximum value of the Su function falls in the range of experi-
mental temperature (Fig. 6) as was desirable. The ba value
a
calculated by Eqn 1 (Fig. 5) and b0 value obtained by the
statistic treatment represented in Fig. 6 are practically coinci-
dent; only a difference of 1.9 % was detected.
The same treatment applied to compounds included in series
b
b gives a value of isokinetic temperature (b0 ) of 508 K (2358C)
Both series represented in Fig. 5 show a slight difference
a
obtained at the minimum value of Su. The discrepancy between
b
(4.3 %) in the isokinetic temperature values: series a, b0 ¼ 470
both values of isokinetic temperature (bb and b0 ) is negligible
K (,1978C for methyl substituted peroxides) and series b:
(3.3 %) compared with the errors in the experimental determina-
tions of this work.
b
b0 ¼ 491 K (,2108C for the group of peroxidic compounds
with bulky substituents). These values fall outside of the
experimental temperature interval selected to carry out the
kinetic study (110–1728C).
Conclusions
The analysis of the reaction products and the kinetic parameters
of the thermal decomposition reaction of some substituted cyclic
organic peroxides belonging to different families (trioxanes,
diperoxides, and triperoxides) in 1,4-dioxan support a general
homolytic stepwise mechanism. The different organic products
The isokinetic relationship is valid for a series of reactions,
when the general plot of the Arrhenius equation (ln kexp versus
1/T, Fig. 6) is represented by linear correlations according to the
data in Table 2 and all lines intersect at a point, or a region, that
corresponds to a reduced interval of temperatures. As it was

1086
R. Nesprias, G. Eyler and A. Can˜izo
20
15
10
5
3
9
5
10
8
4
2
6
1
2
1
0
3
7
0
2
8
10
1
4
3
7
5
ꢁ5
9
6
βbꢁ1
βaꢁ1
2.1 ꢄ 10ꢁ3
Su
Su
ꢁ10
1.9 ꢄ 10ꢁ3
2.3 ꢄ 10ꢁ3
1/T [sꢁ1
2.5 ꢄ 10ꢁ3
2.7 ꢄ 10ꢁ3
]
Fig. 6. Arrhenius equation plot for the thermal decomposition reaction of different cyclic organic peroxide in 1,4-dioxan
in the 110–1708C temperature ambit. Dependence of the residual sum of squares, Su, on the isokinetic temperature.
detected demonstrate the existence of O–O, C–O, and C–O bond
ruptures and solvent derived products obtained by radical attack.
An important substituent effect is operative on the rate
constant values and consequently on the activation parameters
of the thermal decomposition reaction. The application of
different treatments (compensation effect or a statistical treat-
ment) on the kinetic data shows the existence of two sets of
cyclic peroxides with comparable kinetic behaviour: Series a
contains cyclic peroxides with methyl substituents (4 and 7–9),
and series b contains peroxides with voluminous substituents or
condensed rings (1–3, 5, and 10). Different peroxide–solvent
interaction mechanisms can be considered within each series.
Peroxide 6 with two tert-butyl groups attached to the cycle
exhibits anomalous behaviour. Thus, the rate of reaction is
influenced mainly by the substituents attached to the ring, and
in some cases by the size of the cycle.
thoroughly degassed under vacuum at ꢀ1968C and then sealed
with a flame torch using a mixture of natural gas and oxygen. To
perform the runs, the ampoules were submerged in a thermo-
static silicone oil bath stabilized at a chosen temperature
(ꢃ0.18C) and extracted after predetermined times, the peroxide
decomposition reaction was stopped by cooling them in an ice-
water bath (08C).
The quantitative determination of the remaining peroxide 1–
3 in the reaction solution in 1,4-dioxane and the reaction
products were determined by analysis with a reverse phase
column (ODS Hypersil, 5 mm, 100 mm long ꢁ 4.6 mm) installed
in a HPLC Hewlett–Packard instrument series 1050C with UV
detection (l ¼ 254 nm). The mobile phase employed with 1 was
a mixture of methanol/acetonitrile/water (40/30/30), with a flow
of 0.5 mL minꢀ1. The mobile phase used for the corresponding
analysis of 2 and 3 was a mixture of methanol/acetonitrile/water
(40/40/20), with a flow of 0.8 mL minꢀ1. Under these experi-
mental conditions, the retention times were 5.6 min for 1,
8.1 min for 2, and 6.3 min for 3.
Experimental
Materials
The cyclic monofunctional peroxides[30] (1–3, Fig. 1) and the di-
and trifunctional cyclic organic peroxides[13,31–33] (5–7 and 9,
Figs 2 and 3) employed in this work were prepared by methods
described in the literature indicated.
Warning: The preparation and handling of cyclic organic
peroxides are dangerous operations that require standard safety
precautions for handling primary explosives. These compounds
are sensitive to impact, friction, electric discharge, and flame.
The concentration of 5–7 and 9 remaining in the solution was
determined by gas chromatography (GC, internal standard
method, naphthalene, 0.5 % P/V) in a DB-5 capillary column
(5 % biphenyl–95 % dimethyl polysiloxane, 30 m ꢁ 0.32 mm
internal diameter, 0.25 mm film thickness) installed in a Konik-
2000C gas chromatograph, with nitrogen as the carrier gas and
flame ionization detection (2508C). The injection port was fixed
at 1508C in split mode. The experiments were carried out under
programmed conditions: for 5 (808C, 4 min, 258C minꢀ1, 1608C,
15 min), for 6 (608C, 6 min, 158C minꢀ1, 1608C, 15 min), for 7
(608C, 5 min, 108C minꢀ1, 2508C, 15 min), and for 9 (608C,
5 min, 208C minꢀ1, 1608C, 15 min). Under these instrumental
conditions the retention times of 5–7, and 9 were ,10, 12.7, 5.9,
and 8.6 min, respectively.
Solvents
The 1,4-dioxan employed for the kinetic reactions was purified
with the appropriate technique[34,35] and its purity was checked
by GC and high-resolution liquid chromatography (HPLC)
analysis. The solvents employed as the mobile phase were of
HPLC quality.
Product Analysis
Kinetic Methods
The reaction products were identified by GC-MS analysis in a
Rtx-5MS capillary column (5 % biphenyl–95 % dimethyl
polysiloxane, 30 m, 0.25 mm internal diameter, 0.25 mm film
Pyrex glass tubes (7 cm long ꢁ 4–6 mm outer diameter) half
filled with the appropriate cyclic peroxide solution were

Cyclic Organic Peroxides: Thermolysis
1087
thickness), installed in a Thermo Quest Race 2000 CG model gas
chromatograph with helium as carrier gas (0.5mL minꢀ1) and the
injection port at 1508C in split mode (split ratio: 1 : 33). The
detection was carried out with a Finnigan Polaris Ion Trap MS
detector with the transfer line at 2758C and the ion source at
2008C. The analyses were carried out under programmed con-
ditions (508C, 3 min, 208C minꢀ1, 2008C, 10 min). The identifi-
cation of the reaction products was verified through the
interpretation of mass spectra and by comparison of retention
times with authentic samples of each compound analysed by GC.
[14] A. Jubert, R. Pis Diez, L. F. R. Cafferata, J. Mol. Struct. THEOCHEM
2001, 545, 167. doi:10.1016/S0166-1280(01)00415-8
[15] L. F. R. Cafferata, G. N. Eyler, E. L. Svartaman, A. I. Can˜izo,
E. J. Borkowski, J. Org. Chem. 1990, 55, 1058. doi:10.1021/
JO00290A045
[16] A. I. Can˜izo, G. N. Eyler, C. M. Mateo, E. E. Alvarez, R. K. Nesprias,
Heterocycles 2004, 63, 2231. doi:10.3987/COM-04-10128
[17] M. Iglesias, G. Barreto, N. Eyler, A. Can˜izo, Int. J. Chem. Kinet. 2010,
42, 347. doi:10.1002/KIN.20487
[18] A. H. Jubert, R. Pis Diez, E. A. Castro, A. I. Can˜izo, L. F. R. Cafferata,
J. Raman Spectrosc. 1999, 30, 45. doi:10.1002/(SICI)1097-4555
(199901)30 : 1,45::AID-JRS342.3.0.CO;2-C
[19] F. Dubnikova, R. Kosloff, J. Almong, Y. Zeiri, R. Boese, H. Itzhaky,
A. Alt, E. Keinan, J. Am. Chem. Soc. 2005, 127, 1146. doi:10.1021/
JA0464903
Acknowledgements
This study was financially supported by the Engineering Faculty and the
Secretary of Science, Art and Technology of the National University of the
Center of the Buenos Aires Province (SECAT, UNCPBA). A. Can˜izo is a
member of the CONICET (National Council of Scientific and Technological
Research). The authors acknowledge Professor Charles W. Jefford (Uni-
versity of Geneva, Switzerland) for providing compounds 1–3.
[20] J. Cerna, S. Berne´s, A. Can˜izo, N. Eyler, Acta Crystallogr. 2009, 65,
562.
[21] A. J. Bellami, J. Forensic Sci. 1999, 44, 603.
[22] A. J. Pen˜a-Quevedo, J. A. Laramee, H. Dupont Durst, S. P. Herna´ndez-
Rivera, IEEE Sens. J. 2011, 11, 1053. doi:10.1109/JSEN.2010.
2057730
References
[23] G. N. Eyler, J. Phys. Org. Chem. 2006, 19, 776. doi:10.1002/POC.
1078
[1] J. R. Cerna, G. Morales, G. N. Eyler, A. I. Can˜izo, J. App. Polym. Sci.
2002, 83, 1.
[24] J. Pachman, R. Matya´sˇ, Forensic Sci. Int. 2011, 207, 212. doi:10.1016/
J.FORSCIINT.2010.10.010
[25] Y. H. Zhao, M. H. Abraham, A. M. Zissimos, J. Org. Chem. 2003, 68,
[2] G. Posner, P. O‘Neill, Acc. Chem. Res. 2004, 37, 397. doi:10.1021/
AR020227U
[3] C. Borstnik, I. Paik, T. A. Shapiro, G. H. Posner, Int. J. Parasitol. 2002,
7368. doi:10.1021/JO034808O
32, 1661. doi:10.1016/S0020-7519(02)00195-9
[26] J. E. Leffler, J. Org. Chem. 1955, 20, 1202. doi:10.1021/JO01126A009
[27] O. Exner, Nature 1970, 227, 366. doi:10.1038/227366A0
[28] O. Exner, Collect. Czech. Chem. Commun. 1972, 37, 1425.
doi:10.1135/CCCC19721425
[4] P. Ploypradith, Acta Trop. 2004, 89, 329. doi:10.1016/J.ACTATRO
PICA.2003.10.006
[5] A. Robert, F. Benoit-Vical, B. Meunier, Coord. Chem. Rev. 2005, 249,
1927. doi:10.1016/J.CCR.2004.12.022
[29] W. Linert, R. F. Jameson, Chem. Soc. Rev. 1989, 18, 477. doi:10.1039/
CS9891800477
[6] A. I. Can˜izo, G. N. Eyler, G. Morales, J. R. Cerna, J. Phys. Org. Chem.
2004, 17, 215. doi:10.1002/POC.716
[30] C. W. Jefford, S. Kohmoto, D. Jaggi, G. Tima´ri, J. C. Rosier,
M. Kudaz, O. Barbuzzi, D. Gerard, U. Burger, L. Kamalaprija,
J. Mareda, G. Bernardinelli, I. Manzanares, C. J. Ganfield, S. L. Fleck,
R. Robinson, W. Peters, Helv. Chim. Acta 1995, 78, 647. doi:10.1002/
HLCA.19950780312
[7] G. N. Eyler, A. I. Can˜izo, C. M. Mateo, E. E. Alvarez,
L. F. R. Cafferata, J. Org. Chem. 1999, 64, 8457. doi:10.1021/
JO980658J
[8] G. N. Eyler, A. I. Can˜izo, C. M. Mateo, Quim. Nova 2004, 27, 920.
doi:10.1590/S0100-40422004000600016
[31] G. N. Eyler, A. I. Can˜izo, R. K. Nesprias, Quim. Nova 2002, 25, 364.
doi:10.1590/S0100-40422002000300005
[32] G. N. Eyler, A. I. Can˜izo, E. E. Alvarez, C. M. Mateo, Informacio´n
[9] A. I. Can˜izo, Trends Org. Chem. 2006, 11, 55.
[10] S. Huyberechts, A. Halleux, P. Kruys, Bull. Soc. Chim. Belg. 1955, 64,
203. doi:10.1002/BSCB.19550640502
Tecnolo´gica 1997, 8, 153.
[33] L. F. R. Cafferata, G. N. Eyler, M. V. Mir´ıfico, J. Org. Chem. 1984, 49,
2107. doi:10.1021/JO00186A005
[11] L. F. R. Cafferata, G. N. Eyler, A. I. Can˜izo, E. L. Svartman,
E. J. Borkowski, J. Org. Chem. 1990, 55, 1058. doi:10.1021/
JO00290A045
[34] J. A. Y. Riddick, W. B. Bunger, in Organic Solvents (Ed.
A. Weissberger) 1970, Vol. 2, pp. 707–710 (Wiley: New York, NY).
[35] P. P. Perrin, W. L. F. Armarego, in Purification of Laboratory
Chemicals, 3rd edn 1988, pp. 164–165. (Pergamon Press: Oxford).
[12] G. B. Schuster, L. A. Bryant, J. Org. Chem. 1978, 43, 521. doi:10.1021/
JO00397A033
[13] K. Nesprias, M. Iglesias, S. Rivas, E. Alvarez, G. Barreto, N. Eyler,
A. Can˜izo, Int. J. Chem. Kinet. 2011, 43, 657. doi:10.1002/KIN.20600