Process Safety and Environmental Protection 9 1 ( 2 0 1 3 ) 262–268

Contents lists available at SciVerse ScienceDirect

Process Safety and Environmental Protection

journal homepage: www.elsevier.com/locate/psep

Detailed thermal and kinetic modeling of cumene

hydroperoxide decomposition in cumene

Ilaria Di Somma a,∗ , Raffaele Marotta b , Roberto Andreozzi b , Vincenzo Caprio b

a Istituto di Ricerche sulla Combustione (CNR), P.le V. Tecchio, 80, 80125 Napoli, Italy
b Dipartimento di Ingegneria Chimica, Facoltà di Ingegneria, Università di Napoli Federico II, P.le V. Tecchio 80, 80125 Napoli, Italy

a b s t r a c t

The characterization of thermal decomposition of cumene hydroperoxide (CHP) in cumene from kinetic point of
view is of great interest for industrial applications. However, the efforts done so far by the researchers lead only
to the availability of single kinetic laws (first order, autocatalytic, etc.) whose usefulness is strongly limited by the
presence in the starting solutions of some impurities. A detailed model aiming at simulating both, thermal and
kinetic behavior of the system, is now presented. In other words, the proposed model tries to simulate CHP thermal
decomposition process at varying reaction conditions (different initial temperature, in the range 120–160 ◦ C, different
initial CHP concentrations, addition of ␣-methylstyrene to CHP solutions in cumene) and during isothermal, scanning
and adiabatic runs. A good capability of the model is observed to simulate the system behavior by using data collected
under isothermal and adiabatic conditions and during scanning calorimetric runs. It is important to stress that the
use of the model proposed in the present investigation does not require the adoption of adjustable parameters.
© 2012 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Keywords: Detailed kinetic modeling; Cumene hydroperoxide; Thermal decomposition

1. Introduction (as combustion and pyrolysis) is a chemical process of elevated

complexity whose quantitative description may be attempted
Organic peroxides and hydroperoxides are largely used by using different approaches. In fact, it is clear that during its
in chemical and process industry, mainly in the field of development a large number of chemical intermediates and
polymerization initiation (Lu et al., 2010). Most of them products are formed which render very difficult an accurate
are characterized by a strong tendency to undergo vio- simulation of their concentration profiles against times. Fun-
lent, exothermic thermal decomposition, often resulting in damentally, two extreme approaches may be considered to
runaways and thermal explosions (Wu et al., 2008, 2009). simulate the system behavior in these cases: (i) global single-
Therefore, a characterization of their thermal behavior, and step kinetic model and (ii) detailed multi-step kinetic model.
consequently of their kinetic behavior, is of great impor- The first type of models is based on the lumping of all the reac-
tance for the safety of their industrial applications. Due to tions occurring in a single one which describes the conversion
the complexity of thermal decomposition process of solu- of the reagent into a product, disregarding any intermedi-
tions containing organic peroxides or hydroperoxides, it can ates. At the opposite, the second type includes a large number
be expected that an accurate description of system behav- of elementary steps (often hundreds or even thousands) to
ior may be just achieved by the adoption of detailed kinetic account for the presence in the reacting system of many inter-
models. mediates and products. It is important to stress that the latter
To better understand the object of the present investiga- presents a higher capability, with respect to the former, to
tion it is important to make some additional considerations. It describe the system behavior although, in general, it requires
is well known that the thermal degradation of organic species the knowledge of the values of many kinetic parameters which
often are not known a priori and need to be considered in a first
attempt as adjustable parameters. Between these two extreme

Corresponding author. Tel.: +39 081 7682225; fax: +39 081 5936936.
∗

E-mail address: [email protected] (I. Di Somma).

Received 22 February 2012; Received in revised form 31 May 2012; Accepted 1 July 2012
0957-5820/$ – see front matter © 2012 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.psep.2012.07.001
 Process Safety and Environmental Protection 9 1 ( 2 0 1 3 ) 262–268 263

Table 1 – Thermal balance equations adopted in the

Nomenclature model to simulate the system behavior.

Ar pre-exponential factor (s−1 ) or (L mol−1 s−1 ) Conditions Thermal balance

Er activation energy (kJ/mol)
 
dT 1 dCCHP Ua
Isothermal = − ·cp·ϕ · · (−H) + Vliq · (T − Te ) = 0
Coi initial concentration of ith species (mol L−1 ) dt dt
dCCHP 1
C* oxygen concentration at liquid–gas interface Scanning q̇ = cp · ϕ dT =− · (−H) ·
dt dt m0
CHP
(mol L−1 ) dT dCCHP T measured
ad
Adiabatic =− ·
Ci concentration of ith species (mol L−1 ) dt dt C0
CHP
T temperature (K)
R universal constant of gas (J mol−1 K−1 ) 2. The model
kr kinetic constant for nth reaction (s−1 ) or
(L mol−1 s−1 ) Since the ultimate goal of model is that of predicting the
nO2 moles of oxygen in gas phase (mol L−1 ) development of hazardous situations during the use of
n0O initial moles of oxygen in gas phase (mol L−1 ) CHP/cumene solutions at an industrial site, its capability to
2
t time (min) foresee the thermal behavior of the system of interest is of
PO2 oxygen partial pressure (atm) fundamental importance. Therefore, the proposed model was
KL a overall mass-transfer coefficient (min−1 ) developed by coupling an overall thermal balance equation
H Henry constant (atm L mol−1 ) to the material ones for each of the species participating to
Vliq liquid phase volume (L) the process. To write the thermal balance equation a funda-
Vgas gas phase volume (L) mental assumption was done which considered that the heat
yi calculated concentration of ith species evolution was connected only to the disappearance of CHP.
(mol L−1 ) That is, the instantaneous power generated by the reacting
ϕ thermal inertia (dimensionless) system was obtained by multiplying the CHP reaction rate
 density (kg L−1 ) by the overall experimental heat of decomposition. Following
H heat of overall reaction (kJ/mol) this assumption an appropriate equation was derived (Table 1)
cp mean specific heat of reacting mixture whose form varied according to the experimental conditions.
(J K−1 g−1 ) To write the material balance equations, which represent
Tad adiabatic temperature rise (K) the kinetic sub-model, it was preliminarly necessary to pro-
Ua overall heat-transfer coefficient (W K−1 ) pose a possible reaction network summarizing all the reaction
pathways through which the investigated thermal process
develops. According to the observations done during a pre-
vious investigation (Di Somma et al., 2008), it can be stated
approaches, there are many other models which may be devel- that, for the system CHP in cumene, the initiation reactions
oped starting from the global single-step one and increasing are represented by unimolecular or bimolecular homolysis of
its complexity by including in it new steps. O O bond in hydroperoxide molecules (Swern, 1970) (see
Despite the difficulties reported above regarding this sec- Table 2 for the acronyms):
ond type of approach to the problem in some cases it cannot
be avoided. For example, in the case of the thermal decompo- CHP → R1 + R5
sition of cumene hydroperoxide (CHP) solutions in cumene,
in a previous paper some of the authors (Di Somma et al., 2CHP → R1 + R4 + W
2008), following some suggestions in the literature (Hou et al.,
The fate of cumyloxy radicals (R1) formed in these reac-
2001), adopted a kinetic model, based on the most simple
tions is strongly dependent upon the competition between
autocatalytic scheme, for the analysis of the data collected
their tendency to undergo a ␤-scission, to form acetophe-
on CHP solutions in cumene at different initial concentrations
none (ACP) and methyl radicals and the capability to abstract
(80–30%, w/w), to estimate the kinetic parameters regulating
a H-atom to form phenyl-dimethylcarbinol (Baigneé et al.,
the process. During this investigations the characterization of
1983) (DC) (mainly from cumene but also from other species
the intermediate and product distribution at varying reaction
present in the solution as CHP itself, DCP, DC, etc.). If oxy-
time indicated also that the proposed simple global kinetic
gen is admitted to the reacting solution, cumylradicals formed
model is of limited use, the system reactivity being signif-
in H-abstraction from cumene are rapidly converted into
icantly influenced by the addition to CHP solution of small
cumyl-peroxy radicals (Bhattacharya, 2008) otherwise they
quantities of organic species, such as ␣-methylstyrene, often
mainly terminate leading to the formation of 2,3-dimethyl-
present at trace level as impurities in CHP solutions in cumene.
diphenylbutane.
It is clear from these results the need of developing a (more)
The appearance in the reacting solution of DCP molecules,
detailed kinetic model capable of taking into account the pres-
during CHP thermal decomposition, may be explained through
ence in the reacting solution of these impurities.
the occurrence of terminating reactions such as:
The present paper aims at developing and validating a
detailed model capable of simulating the system behavior 2R4 → DCP + OX
when CHP undergoes thermal decomposition in cumene (even
Moreover, since – as previously reported (Jordan et al., 1985)
in the presence of some impurities such as ␣-methylstyrene)
– the addition of ␣-methyl-styrene to the investigated system
under isothermal, scanning and adiabatic conditions which
results into an increase of its reactivity, a direct interaction of
could represent a valid tool to predict hazardous condi-
CHP and this olefin may be foreseen:
tions that may arise in the industrial applications where
CHP/cumene solutions are used (Jordan et al., 1985). AMS + CHP → R7 + R1
264 Process Safety and Environmental Protection 9 1 ( 2 0 1 3 ) 262–268

Table 2 – Acronyms used in Table 1.

Species Abb Species Abb Species Abb

[C6 H5 C(CH3 )2 ]2 DB C6 H5 C(CH3 )2 OH DC CH3 •

R3
C6 H5 C(O)CH3 ACP C6 H5 C(CH3 )2 OOH CHP C6 H5 COO• (CH3 )2 R4
CH3 OOH B [C6 H5 C(CH3 )2 O]2 DCP HO• R5
CH3 CH3 ET C6 H5 C(CH3 )2 H CUM C6 H5 (CH3 )2 (CO)2 (CH3 )(CH2 • )C6 H5 R6
H2 O W C6 H5 C(CH3 ) CH2 AMS C6 H5 C• (CH3 )(CH2 OH) R7
CH3 OOCH3 G C6 H5 C(CH3 )2 OOCH3 F C6 H5 (CH3 )(CH2 • )COOH R8
O2 OX C6 H5 CH(CH3 )CH2 OH AL C6 H5 (CH3 )(CH2 • )COH R9
H2 O2 WX CH2 O FO CH3 OO• R10
C6 H5 C(CH3 )3 M C6 H5 CO• (CH3 )2 R1 HO2 • R11
CH4 MET C6 H5 C• (CH3 )2 R2 C6 H5 (CH3 )(CH2 O• )CH R12

Table 4 – List of reactions involving DCP. Pre-exponential

factor (k), activation energy (Ea ).
Reaction Log10 (k) Ea (kJ/mol)

DCP → 2R1 14.7 144.4 ± 2.1

R4 + DCP → CHP + R6 9.20 ± 0.3 74.09 ± 2.9
R1 + DCP → DC + R6 9.00 ± 0.6 36.00 ± 3.8
R3 + DCP → MET + R6 9.00 ± 0.5 55.25 ± 4.2
R10 + DCP → B + R6 8.20 ± 0.3 51.49 ± 3.7
R12 + DCP → AL + R6 7.20 ± 0.3 89.58 ± 2.5

Table 5 – List of reactions involving radical species.

Pre-exponential factor (k), activation energy (Ea ).
Reaction Log10 (k) Ea (kJ/mol)

R1 → ACP + R3 9.80 ± 0.1 20.51 ± 0.4

2R2 → DB 7.00 ± 0.3 12.56 ± 0.4
Scheme 1 – Reaction network for CHP thermal
R4 + R2 → DCP 8.20 ± 0.3 15.07 ± 0.7
decomposition. 2R3 → ET 9.40 ± 0.1 4.605 ± 0.1
2R5 → WX 8.40 ± 0.5 2.093 ± 0.1
R5 + R2 → DC 8.70 ± 0.1 3.767 ± 0.2
to form R7: R2 + R3 → M 9.10 ± 0.2 2.093 ± 0.1
R6 → AMS + R4 8.20 ± 0.4 63.63 ± 4.2
R8 → AMS + R11 6.60 ± 0.1 76.18 ± 8.4
R9 → AMS + R5 8.20 ± 0.3 94.18 ± 2.1
R4 + R3 → F 9.20 ± 0.2 27.63 ± 0.8
which can lead to the appearance of 2-phenyl-1-propanol 2R4 → DCP + OX 8.40 ± 0.1 20.51 ± 2.5
(AL), an isomer of DC. 2R10 → G + OX 9.20 ± 0.3 4.186 ± 0.1
The main reaction pathways through which the thermal R7 → R12 8.20 ± 0.3 96.28 ± 8.4
R4 → AMS + R11 3.20 ± 0.1 19.67 ± 1.2
decomposition of CHP proceeds are summarized in Scheme 1.
2R11 → WX + OX 8.30 ± 0.6 8.372 ± 0.3
Other reactions (terminating/H-abstraction, etc.) are intro-
duced to complete the description of the reacting system. A taken from the literature (Di Somma et al., in press; Hendry
list of all the reactions included in the model is thus shown et al., 1974; Swern, 1970).
in Tables 3–6 along with the values of the related parameters On the basis of this list, the kinetic sub-model can be sum-
marized by the equations reported in Table 7 (the values of

Table 3 – List of reactions involving CHP. Pre-exponential

factor (k), activation energy (Ea ). Table 6 – List of reactions completing the model.
Pre-exponential factor (k), activation energy (Ea ).
Reaction Log10 (k) Ea (kJ/mol)
Reaction Log10 (k) Ea (kJ/mol)
CHP → R1 + R5 12.3 ± 0.1 139.4 ± 0.8
2CHP → R1 + R4 + W 7.70 108.8 R3 + CUM → MET + R2 8.40 ± 0.5 35.58 ± 4.2
AMS + CHP → R7 + R1 8.20 ± 0.3 90.00 ± 4.2 R1 + CUM → DC + R2 8.40 ± 0.4 15.91 ± 0.8
R1 + CHP → DC + R8 9.30 ± 0.2 47.72 ± 1.7 R4 + CUM → CHP + R2 7.20 ± 0.2 50.65 ± 2.1
AMS + CHP → DCP 5.10 ± 0.1 92.09 ± 1.7 R5 + CUM → W + R2 8.50 ± 0.1 60.70 ± 2.1
R4 + CHP → CHP + R8 7.20 ± 0.2 65.30 ± 5.0 R7 + CUM → AL + R2 8.40 ± 0.1 96.28 ± 4.6
R10 + CHP → B + R8 8.40 ± 0.7 37.25 ± 0.8 R12 + CUM → AL + R2 8.20 ± 0.5 67.81 ± 2.9
R3 + CHP → MET + R8 9.00 ± 0.5 55.25 ± 4.2 R5 + AMS → R7 8.20 ± 0.4 46.46 ± 2.1
R5 + CHP → W + R8 9.10 ± 0.2 59.44 ± 0.8 R10 + AMS → R12 + FO 9.00 ± 0.6 33.49 ± 1.7
R11 + CHP → WX + R8 7.30 ± 0.2 68.65 ± 4.2 R2 + OX → R4 7.20 ± 0.3 0.000
R1 + CHP → DC + R4 8.40 ± 0.2 15.49 ± 1.2 R3 + OX → R10 5.90 ± 0.3 0.000
R3 + CHP → MET + R4 8.50 ± 0.3 28.88 ± 2.1 R4 + DC → CHP + R9 9.30 ± 0.1 72.42 ± 6.3
R5 + CHP → W + R4 8.60 ± 0.5 60.28 ± 5.0 R3 + DC → MET + R9 9.00 ± 0.5 55.25 ± 4.2
R10 + CHP → B + R4 8.10 ± 0.5 15.07 ± 0.8 R1 + DC → DC + R9 9.30 ± 0.2 20.09 ± 0.8
R12 + CHP → AL + R8 8.20 ± 0.3 71.16 ± 1.7 WX → 2R5 13.4 ± 0.2 100.5 ± 6.3
 Process Safety and Environmental Protection 9 1 ( 2 0 1 3 ) 262–268 265

Table 7 – Material balance equations representing the

kinetic sub-model to simulate the system behavior.
dCh
 
Material balance on hth species dt
=± Ar · e−(Er /RT) · Cm
r m
h indicates the hth species
consumed (−) or generated (+),
m indicates the mth species
participating to the rth
reaction, and
h can be also equal to m.
dnO
1 2
Oxygen balance in gas phase Vliq dt
= KL a · (C∗ − CO2 )
dCO
Oxygen balance in liquid phase 2 =
dt
n
dnO

1 2
Vliq dt
− ki · Ci · CO2
i=1
(n0 −nO )·R·T
Fig. 1 – CHP (80%, w/w) thermal decomposition at T = 403 K:
1 1 O2 2
Henry law C∗ = H · PO2 = H · Vgas comparison of experimental (symbols) and calculated
(continuous lines) data. Cumene (CUM), dicumylperoxide
(DCP), acetophenone (ACP), alphamethylstyrene (AMS),
the parameters in the table have been taken from literature,
cumene hydroperoxide (CHP), and total alcohols (DC). Total
Di Somma et al., in press; Emanuel et al., 1967; Hendry et al.,
alcohols indicates the sum of the concentration of
1974; Swern, 1970).
dimethyl-phenyl-carbinol and 2-phenyl-1-propanol.
Since oxygen, when admitted in the reacting system, par-
ticipates to the process, its transfer from the gas phase to the
as received. Samples of CHP when received were preliminar-
liquid solution needs to be accounted for. In the present work,
ily submitted to HPLC analysis to quantitatively evaluate the
this transfer is considered under the hypothesis that the reac-
presence of species such as carbinol and ␣-methyl styrene,
tions which consume oxygen occur mainly in the liquid bulk
which were expected to influence the substrate decomposi-
(and only negligibly in the liquid film) and that the latter is
tion.
present in the reacting solution at a concentration which is
The data under adiabatic and scanning conditions used in
lower than the saturation value.
the present work were collected by means of different calori-
The proposed model, obtained by adding to the thermal
metric devices. For the adiabatic experiments was used an
balance equation (see Table 1) the kinetic sub-model consist-
A.R.C. calorimeter (by Columbia Scientific Industries) while
ing of a set of 31 ordinary differential equations (see Table 7),
for the scanning conditions DSC experiments were performed
needs to be solved according to proper initial conditions. In
by using a Q10 (thermal analysis). The A.R.C. runs were
order to test it, the temperature profiles and those of each
performed in Hastelloy C bombs with the following set param-
species were calculated at varying the reaction time and com-
eters: wait time 10 min, heat step temperature 283 K, and heat
pared with those recorded during experimental runs.
rate threshold value 0.02 K min−1 .

3. Material and methods 4. Results and discussion

For the experiments in the presence of oxygen, in a series of
In Figs. 1 and 2 a comparison of calculated (solid lines) and
sealed, stirred glass tubes of 1.45 × 10−2 L total volume, ini-
experimental (symbols) data is shown for two runs carried
tially containing air at a pressure of 1 atm, 1.5 × 10−4 L of CHP
solutions in cumene with a concentration equal to 0.6 mol L−1
were poured. For runs performed in absence of oxygen, the
glass tubes were previously purged with a nitrogen stream. For
isothermal runs, in both the cases (with and without oxygen),
the tubes were placed into an oil bath at a desired tempera-
ture between 403 and 433 K and their thermal behavior was
checked by means of a small thermocouple contacting the
solution and connected at an appropriate acquisition sys-
tem. The samples were withdrawn from the oil bath after
the desired reaction time and rapidly cooled, were recovered
with acetone, diluted with acetonitrile and submitted to HPLC
analyses by means of a 1100 Hewlett Packard HPLC equipped
with a Synergi 4␮ Fusion RP-80 column and a diode array
detector. The mobile phase was 65% (v/v) of acetonitrile and
35% of a buffer solution (5% CH3 OH, 0.4% H3 PO4 and 94.6%
H2 O), with a flow of 1.0 × 10−3 L min−1 . The signals at wave- Fig. 2 – CHP (80%, w/w) thermal decomposition at T = 433 K:
lengths of 210 and 240 nm were used for quantitative analyses. comparison of experimental (symbols) and calculated
The oven temperature was set at 305 K. Gas analysis was per- (continuous lines) data. Cumene (CUM), dicumylperoxide
formed with a Gaschromatograph HP 5890 equipped with a (DCP), acetophenone (ACP), alphamethylstyrene (AMS),
FID detector and a PPU column (30 m × 0.53 mm) with an oven cumene hydroperoxide (CHP), and total alcohols (DC). Total
temperature of 313 K and an helium flow of 7.0 ml/min. All the alcohols indicates the sum of the concentration of
reagents were purchased from Sigma Aldrich and were used dimethyl-phenyl-carbinol and 2-phenyl-1-propanol.
266 Process Safety and Environmental Protection 9 1 ( 2 0 1 3 ) 262–268

Fig. 3 – CHP (30%, w/w) thermal decomposition at T = 403 K:

Fig. 5 – CHP (8.0%, w/w) thermal decomposition at T = 403 K:
comparison of experimental (symbols) and calculated
comparison of experimental (symbols) and calculated
(continuous lines) data. Cumene (CUM), dicumylperoxide
(continuous lines) data. Cumene (CUM), dicumylperoxide
(DCP), acetophenone (ACP), alphamethylstyrene (AMS),
(DCP), acetophenone (ACP), alphamethylstyrene (AMS),
cumene hydroperoxide (CHP), and total alcohols (DC). Total
cumene hydroperoxide (CHP), and total alcohols (DC). Total
alcohols indicates the sum of the concentration of
alcohols indicates the sum of the concentration of
dimethyl-phenyl-carbinol and 2-phenyl-1-propanol.
dimethyl-phenyl-carbinol and 2-phenyl-1-propanol.

out, on 80% by weight CHP solution in cumene, at two dif-

different from those chosen for the other runs) was adopted. It
ferent constant temperatures (403 K and 433 K). It is evident
is interesting to observe that these conditions are quite similar
from these figures that the model used has a good capabil-
to those prevailing in the oxidation reactors for CHP produc-
ity to simulate the behavior of the system both in terms of
tion from cumene, in which the hydroperoxide has the role of
species concentrations (total percentage standard deviations
starting the oxidation process through its controlled decom-
are of 5.7% and 6.5%) and temperature (constant) profile (data
position and successive generation of cumyl radical (arising
not shown).
from H-abstraction to cumene) which give cumylperoxy one
A similar behavior of the model was registered on the
by reacting with oxygen (see Scheme 1).
data collected at other investigated temperatures (data not
Again, a good performance of the model was observed
reported).
when the modeling of some isothermal runs was attempted
Good results were also observed when the model was
in which a known amount of ␣-methyl styrene was initially
used to simulate the CHP decomposition process at varying
added to a CHP solution in cumene. For example, in Fig. 6 the
its initial concentration for a constant temperature of 403 K
results were reported of a run in which ␣-methyl styrene was
(initial CHP concentrations of 3.94 mol L−1 , 7.11 mol L−1 and
added to a 80% CHP solution in cumene at an initial concen-
0.62 mol L−1 in Figs. 3–5, respectively) (overall percentage stan-
tration of 0.43 M.
dard deviation of 3.9%, 5.4% and 2.4%).
The simulation work was successively extended to the
The results of Fig. 5 are of particular interest since they
results of calorimetric runs previously analyzed by some of the
show that the model is also capable to simulate the system
authors by means of a global kinetic model (Di Somma et al.,
behavior when a very low initial concentration of CHP (quite

Fig. 6 – CHP (80%, w/w) thermal decomposition at T = 403 K

Fig. 4 – CHP (90%, w/w) thermal decomposition at T = 403 K: with the addition of ␣-methylstyrene: comparison of
comparison of experimental (symbols) and calculated experimental (symbols) and calculated (continuous lines)
(continuous lines) data. Cumene (CUM), dicumylperoxide data. Cumene (CUM), dicumylperoxide (DCP), acetophenone
(DCP), acetophenone (ACP), alphamethylstyrene (AMS), (ACP), alphamethylstyrene (AMS), cumene hydroperoxide
cumene hydroperoxide (CHP), and total alcohols (DC). Total (CHP), and total alcohols (DC). Total alcohols indicates the
alcohols indicates the sum of the concentration of sum of the concentration of dimethyl-phenyl-carbinol and
dimethyl-phenyl-carbinol and 2-phenyl-1-propanol. 2-phenyl-1-propanol.
 Process Safety and Environmental Protection 9 1 ( 2 0 1 3 ) 262–268 267

reacting solution of some species such as, for example, ␣-

methyl styrene at trace levels (Fig. 6). This example is of
practical relevance for phenol–acetone production plants in
which the recycling stream of cumene to the oxidation reac-
tor partially originates from the hydrogenation of ␣-methyl
styrene recovered in the purification section. It is evident that
in case of troubles in the hydrogenation section ␣-methyl
styrene may arrive in the oxidation reactor thus affecting CHP
stability.

5. Conclusions

A detailed model has been proposed to predict the thermal

Fig. 7 – Power generated per unit of mass during DSC decomposition of CHP coupling an overall thermal balance
experiments on 80% (w/w) CHP solution in cumene at and a kinetic sub-model. It has been preliminarily demon-
varying the heating rate. strated that the reaction network that can be adopted to
describe the main reaction pathways through which the ther-
mal decomposition of CHP in cumene evolves is the same
2008). The adoption of the detailed model proposed in this
as that previously proposed for DCP in the same solvent.
work allowed the obtainment of good results in the simulation
A kinetic sub-model has been thus developed on the basis
of the experimental data for the runs at the highest investi-
of this network to quantitative describe the decomposition
gated CHP concentration (80%, w/w). A mean experimental
of CHP. A good capability of the proposed model to sim-
heat of decomposition of 1219 J/g was considered for scanning
ulate the system behavior has been observed for almost
runs at varying the heating rate (Fig. 7) and an experimental
all the experimental conditions, which have been inves-
adiabatic temperature rise of 238.7 K was used for those car-
tigated (different initial temperature, 120–160 ◦ C, different
ried out under adiabatic conditions (Fig. 8). It is noteworthy
initial CHP concentrations, addition of ␣-methylstyrene to
to observe that the onset temperature of this run is 363 K thus
CHP solutions in cumene, isothermal and scanning condi-
demonstrating the capability of the model to well simulate the
tions) without the use of any adjustable kinetic parameter.
system behavior also in a temperature range larger than that
For the data collected under adiabatic conditions, the model
adopted in the most part of the present experimental inves-
demonstrated to be capable of well describing those at the
tigations. This is particularly important if one considers that,
highest CHP concentration (80%, w/w) whereas poor results
according to Weber (2006), commercial processes for cumene
have been obtained at lower concentration (30%, w/w). This
hydroperoxide production are operated also at temperatures
discrepancy may be probably ascribed to the adoption in
lower than 373 K. For calorimetric runs with a CHP concen-
the thermal balance equation of an overall heat of reaction
tration of 30% (w/w) the results of the model were poor. This
instead of using those related to each step included in the
result can be explained by considering that the thermal bal-
model.
ance equations (Table 1) assume that the power generated
during a single run is only due to the reactions involving CHP
molecules. This assumption clearly fails for diluted CHP solu- References
tions.
At first sight the results in Fig. 8 appear as good as Baigneé, A., Howard, J.A., Scaiano, J.C., Stewart, L.C., 1983.
those in Fig. 3 of the paper previously published by the Absolute rate constants for reactions of cumyloxy in solution.
authors (Di Somma et al., 2008). However, it cannot be for- J. Am. Chem. Soc. 105, 6120–6123.
gotten that a global model as that proposed by the authors Bhattacharya, A., 2008. Kinetic modeling of liquid phase
in Ref. (Di Somma et al., 2008) does not allow to predict autoxidation of cumene. Chem. Eng. J. 137, 308–319.
how the system behavior is affected by the presence in the Di Somma, I., Andreozzi, R., Canterino, M., Caprio, V., Sanchirico,
R., 2008. Thermal decomposition of cumene hydroperoxide:
chemical and kinetic characterization. AIChE J. 54,
1579–1584.
Di Somma, I., Marotta, R., Andreozzi, R., Caprio, V. Dicumyl
peroxide thermal decomposition in cumene: development of
a kinetic model. Ind. Eng. Chem. Res., in press.
Emanuel, N.M., Denisov, E.T., Maizus, Z.K., 1967. Liquid-Phase
Oxidation of Hydrocarbons. Plenum Press, New York.
Jordan, W., van Barneveld, H., Gerlich, O., Kleine-Boyman, M.,
Ullrich, J., 1985. Phenol-production. In: Elvers, B., Hawkins, S.,
Schulz, G. (Eds.), Ullmann’s Encyclopedia of Industrial
Chemistry. , sixth ed. Wiley-VCH, Germany, pp. 299–312.
Hendry, D.G., Mill, T., Piszkiewicz, L., Howard, J.A., Eigenmann,
H.K., 1974. A critical review of H-atom transfer in the liquid
phase: chlorine atom, alkyl, trichloromethyl, alkoxy, and
alkylperoxy radicals. J. Phys. Chem. Ref. Data 3 (4), 937–978.
Hou, H., Shu, C., Duh, Y., 2001. Exothermic decomposition of
cumene hydroperoxide at low temparature conditions. AIChE
J. 47, 1893–1896.
Fig. 8 – Thermal behavior of 80% (w/w) solution of CHP in Lu, K.T., Chu, Y., Chen, T., Hu, K., 2010. Investigation of the
cumene under adiabatic conditions. decomposition reaction and dust explosion characteristics of
268 Process Safety and Environmental Protection 9 1 ( 2 0 1 3 ) 262–268

crystalline dicumyl peroxide. Process Saf. Environ. Prot. 88, Wu, S.H., Wang, Y., Wu, T.C., Hu, W.N., Shu, C.M., 2008. Evaluation
356–365. of thermal hazards for dicumyl peroxide by DSC and VSP2. J.
Swern, D., 1970. Organic Peroxides, vol. I. Wiley-Interscience, New Therm. Anal. Calorim. 93, 189–194.
York. Wu, S.H., Shyu, M., Chi, Y.I., Shu, J.C., 2009. Evaluation of runaway
Weber, M., 2006. Some safety aspects on the design of Sparger reaction for dicumyl peroxide in a batch reactor by DSC and
systems for the oxidation of organic liquids. AIChE J. 25 (4), VSP2. J. Loss Prev. Process Ind. 22, 721–727.
326–330.