Subscriber access provided by University of Newcastle, Australia

Article
Zero-order versus intrinsic kinetics for the determination of
TMRad: Application to the decomposition of hydrogen peroxide
Lamiae VERNIERES-HASSIMI, Amine dakkoune, lokmane
abdelouahed, Lionel Estel, and Sebastien Leveneur
Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.7b01291 • Publication Date (Web): 24 Jun 2017
Downloaded from http://pubs.acs.org on June 25, 2017

Just Accepted

“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Industrial & Engineering Chemistry Research is published by the American Chemical

Society. 1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.
Page 1 of 43 Industrial & Engineering Chemistry Research

1
2
3
4
5
Zero-order versus intrinsic kinetics for the
6
7
8
9 determination of TMRad: Application to the
10
11
12
13 decomposition of hydrogen peroxide
14
15
16
17
18
Lamiae Vernières-Hassimi1, Amine Dakkoune1, Lokmane Abdelouahed1, Lionel Estel1, Sébastien
19
20 Leveneur*1,2
21
22
23 1
Normandie Univ, INSA Rouen, UNIROUEN, LSPC, EA4704, 76000 Rouen, France, E-mail :
24
25
26 sebastien.leveneur@insa-rouen.fr
27
28
2
29 Laboratory of Industrial Chemistry and Reaction Engineering, Johan Gadolin Process
30
31 Chemistry Centre, Åbo Akademi University, Biskopsgatan 8, FI-20500 Åbo/Turku, Finland.
32
33
34
35 KEYWORDS: safety criteria, adiabatic reactor, TMRad, zero-order model, kinetic modeling.
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
1
 Industrial & Engineering Chemistry Research Page 2 of 43

1
2
3
GRAPHICAL ABSTRACT.
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
2
Page 3 of 43 Industrial & Engineering Chemistry Research

1
2
3
ABSTRACT.
4
5
6
7 Thermal safety of chemical processes requires the knowledge of safety parameters quantifying
8
9
10 the probability like time-to-maximum rate under adiabatic conditions (TMRad) and the severity
11
12 like adiabatic temperature rise under adiabatic conditions (∆Tad). The zero-order approximation
13
14
is used to ease the determination of TMRad values at different process temperature, but how to be
15
16
17 sure that this approximation is acceptable compared to the use of an intrinsic kinetic model? In
18
19 the literature, there are not such studies that compare the values of TMRad by using zero-order
20
21
22
and intrinsic kinetic models. For that, decomposition of hydrogen peroxide in the presence and in
23
24 the absence of copper sulfate was studied in an advanced reactive system screening tool
25
26 (ARSST) unit. This calorimeter works under near-adiabatic conditions based on heat loss
27
28
29 compensation principle and by using a background heating rate (β). In a first stage, a kinetic
30
31 model was built to estimate the intrinsic kinetic constants. Then, a comparison between the
32
33 values of TMRad from the zero-order and the intrinsic kinetic model was performed. It was found
34
35
36 that the difference of TMRad values obtained by these two models can be significant. The
37
38 influence of β and reactant concentrations were found to play an important role on this
39
40
difference. A good practice in case of kinetic and thermodynamic data missing, a user should test
41
42
43 different background heating rates to verify their influence on TMRad values obtained from the
44
45 zero-order model.
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
3
 Industrial & Engineering Chemistry Research Page 4 of 43

1
2
3
1. INTRODUCTION
4
5
6
7 Safety assessment of a chemical plant is complex because it requires the knowledge of the
8
9 different process units, chemicals reactivity and toxicology.1 Furthermore, one should define a
10
11
12 clear methodology to rank the different risks (toxic, thermal, environmental, etc.). The
13
14 determination of safety parameters for thermal risk can be cumbersome because the reactivity of
15
16 the chemical system and reactor characteristic should be known. This risk is important. For
17
18
19 example, Balasubramanian and Louvar2 have revealed that 26% of the major petrochemical plant
20
21 accidents are due to runaway.
22
23
24
Thermal runaway occurs when the heat flow-rate due to chemical reactions is higher than the
25
26
27 one removed by the cooling system. Before ranking this risk for an exothermic reaction system,
28
29 the kinetics and thermodynamics of the reaction system should be characterized at some extent.
30
31
32
Saada et al.,3 have shown that runaway reaction accidents in chemical industries are important
33
34 due to a non-systematic assessment of chemical hazards.
35
36
37 Several causes can lead to a thermal runaway situation: wrong operating conditions (reactants
38
39
40 loading or thermal conditions), malfunction of the cooling system (low heat exchange surface
41
42 area, failure of the automate system, cooling failure...), presence of exothermic secondary
43
44 reactions, etc. 3
In the first stage of this accident, there is a shift from a controlled reaction
45
46
47 temperature to an uncontrolled reaction temperature. In the worst-case scenario, the uncontrolled
48
49 reaction temperature can lead to an adiabatic thermal mode. In the second stage, there is a fast
50
51
52
temperature increase due to the exothermic reactions which could trigger secondary reactions,1
53
54 i.e., decomposition reactions producing non-condensable products. The worst final consequence
55
56 is the overpressure in the reactor leading to the explosion of the reactor structure.
57
58
59
60
ACS Paragon Plus Environment
4
Page 5 of 43 Industrial & Engineering Chemistry Research

1
2
3
Depending on the operating conditions (initial temperature, reactants loading, concentrations,
4
5
6 etc.), the consequences vary from loss and disruption of production to the irreversible damage of
7
8 the unit process and plant worker fatalities.4
9
10
11
12 The main challenges to rank the thermal risk of a chemical process are to determine some
13
14 safety criteria and to measure some safety parameters. Thermal risk is the product of the
15
16 probability of occurrence by the severity.
17
18
19
20 The parameter describing the severity of the thermal risk is the adiabatic temperature rise
21
22 (∆Tad), which represents the temperature increase due to chemical reactions under adiabatic
23
24
conditions. The severity parameter is linked to the thermodynamic constant: reaction enthalpy.
25
26
27 From this parameter, it is possible to distinguish two other severity parameters:
28
29
30 - Maximum temperature for synthesis reactions (MTSR), which is the temperature increase due
31
32
33
to the synthesis reactions under adiabatic conditions,
34
35
36 - Final temperature (Tfinal), which is the temperature increase due to the synthesis and secondary
37
38 reactions under adiabatic conditions.
39
40
41
42 The parameter describing the probability of the thermal risk is the time-to-maximum rate
43
44 under adiabatic conditions at a process temperature Tp (TMRad(Tp)). The probability parameter is
45
46
linked to the kinetics of the reaction system. For a thermal risk assessment, it is important to
47
48
49 determine two particular probability parameters:
50
51
52 - TD24 which represents the initial process temperature to reach the maximum reaction rate in 24
53
54
55 hours. Stoessel1 defines this temperature as the one at which the thermal stability of the reaction
56
57 mixture is not a problem.
58
59
60
ACS Paragon Plus Environment
5
 Industrial & Engineering Chemistry Research Page 6 of 43

1
2
3
- TD8 which represents the initial process temperature to reach the maximum reaction rate in 8
4
5
6 hours.
7
8
9 The determination of MTSR can be measured directly by calorimetry5-6 or by knowing the
10
11
12 reaction enthalpies.7 The determination of the final temperature can be difficult because the
13
14 thermodynamics of secondary reactions could be unknown. The experimental determination of
15
16 MTSR and Tfinal in adiabatic reactors requires that these calorimeters have a low Φ-factor. The
17
18
19 Φ-factor represents the thermal inertia of the reactor. As this factor is close to one, the thermal
20
21 system is close to adiabatic mode and the measurement is reliable.8
22
23
24
The safety parameter time-to-maximum rate under adiabatic conditions TMRad(Tp) can be
25
26
27 measured by calorimetry or by knowing the kinetic and thermodynamic constants. Nevertheless,
28
29 the detailed kinetic model for a reaction system including synthesis and secondary reactions can
30
31
32
be time-consuming and cumbersome. For example, secondary reactions can involve radicals,
33
34 which are not easy to follow and model.9 The experimental determination of TD24 can also be
35
36 cumbersome because it depends on the temperature sensitivity of the calorimeter and its Φ-
37
38
39 factor. In case this parameter cannot be directly measured by an adiabatic calorimeter, then the
40
41 determination of TD24 is done by extrapolation. Usually, the zero-order approximation is applied,
42
43 allowing to determine rapidly the values of TMRad at different process temperatures.5 By using
44
45
46 this assumption, TMRad(TP) can be expressed as
47
48
 . . .
TMR  =
49
50   .
(1)
51
52

where mR is the reaction mixture mass, C is the specific heat-capacity of the reaction mixture,
53
54
55
56 R is the gas constant, TP is the initial process temperature, qr(TP) is the heat-flow rate due to
57
58
59
60
ACS Paragon Plus Environment
6
Page 7 of 43 Industrial & Engineering Chemistry Research

1
2
3
chemical reactions and Ea is the apparent activation energy. The use of this approximation to
4
5
6 determine the probability parameters can be found in Supporting Information and in the articles
7
8 of our group.5,10
9
10
11 The safety community uses the zero-order approximation because it gives rapidly the safety
12
13
14 parameters (TMRad) in the worst-case conditions. Nevertheless, this method does not consider
15
16 the concentrations of any compounds. Hence, the TMRad values obtained by this method are
17
18
19
correct for specific operating conditions. How is reliable the determination of TMRad(Tp) or TD24
20
21 and TD8 by using this approximation? To the best of our knowledge, there are no studies that
22
23 compare the probability parameters obtained from the intrinsic kinetic model and the zero-order
24
25
26 kinetic model. The goal of this manuscript is to fill this gap by studying the decomposition of
27
28 hydrogen peroxide.
29
30
31 Several articles describe the peroxide stability or decomposition by calorimeters due to the
32
33
34 high reactivity of the chemical bond O-O. The study of peroxide compounds decomposition as
35
36 test system for calorimeter is frequent .11 For instance, the oxidation of sodium thiosulfate by
37
38
hydrogen peroxide, which is a fast and exothermic reaction, is used as a model system in the
39
40
41 field of heat transfer or thermal safety. 5,12-15 Furthermore, the use of hydrogen peroxide as an
42
43 oxidizing agent in industry is growing because it is eco-friendly.16-17 Hydrogen peroxide is used
44
45
46
in different fields such as waste/water/effluent treatment, paper/pulp/textile bleaching, chemical
47
48 synthesis, mining/metallurgy or detergents. In 2006, the annual production was around 2.2
49
50 million tons18 and 4.5 million tons in 2014.17
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
7
 Industrial & Engineering Chemistry Research Page 8 of 43

1
2
3
Decomposition of hydrogen peroxide can occur in the presence of some ppm of metals leading
4
5
6 to thermal runaway accidents.19 In hydrometallurgical extraction, dissolution of metallic copper
7
8 is done in an acidic hydrogen peroxide solution.20 During this dissolution, hydrogen peroxide is
9
10
11
used as an oxidizing agent producing in fine copper (II) sulfate.21 For an economic reason, the
12
13 decomposition of hydrogen peroxide by copper (II) sulfate should be lowered.22
14
15
16 In this manuscript, the decomposition of hydrogen peroxide by copper (II) sulfate was used as
17
18
19 a test system. A kinetic model was built to estimate the intrinsic kinetic parameters based on the
20
21 work of Perez-Benito.23 The establishment of such kinetic model for an adiabatic reactor is
22
23 scarce in the literature.24,25 Then, a comparison is done between the values of TMRad(Tp), TD24
24
25
26 and TD8 at different process temperatures obtained from the intrinsic kinetic model and the zero-
27
28 order kinetic model.
29
30
31
32
Different calorimeters can be used to determine these safety parameters: accelerating rate
33
34 calorimeters (ARC),5,24,26-27 vent sizing package (VSP),28-30 Phi-Tech II,6,8 differential scanning
35
36 calorimetry,5,26-28,30 or advanced reactive system screening tool (ARSST). 10,31-35,25 Experiments
37
38
39 performed with this calorimeter can be done under adiabatic or near-adiabatic conditions with a
40
41 low Φ-factor.1,36
42
43
44 Decomposition study of compounds in liquid phase can be difficult because temperature
45
46
47 increase can lead to evaporation and hence to the diminution of the reaction mass. Fauske and
48
49 Associates have developed the Advanced Reactive System Screening Tool (ARSST), with a low
50
51
52
Φ-factor (ca. 1.04), where it is possible to work under high inert pressure to limit the
53
54 evaporation. The other benefit of ARSST compared to ARC or VSP calorimeters is to provide
55
56
57
58
59
60
ACS Paragon Plus Environment
8
Page 9 of 43 Industrial & Engineering Chemistry Research

1
2
3
fast screening of exothermic reactions.19,37 The ARSST system was used for this study to
4
5
6 measure the reaction temperature.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
9
 Industrial & Engineering Chemistry Research Page 10 of 43

1
2
3
2. EXPERIMENTAL SECTION
4
5
6
7 The decomposition of hydrogen peroxide (33 wt.% in water, VWR) in a sulfuric acid (95-98
8
9 wt.%, ITW companies) solution in the presence of copper sulfate (99 wt.%, VWR), was
10
11
12 investigated in the ARSST unit. The ARSST unit, as shown in Figure 1, is used to characterize
13
14 chemical system that could lead to rapid pressure and temperature rise.
15
16
17 The reactor is essentially composed of two compartments. The first compartment is a test cell
18
19
20 assembly including a glass test cell for the reaction mixture with a volume capacity of 10 mL. In
21
22 the first step, a solution of copper (II) sulfate and sulfuric acid was added into the test cell, then
23
24
hydrogen peroxide at room temperature. The glass test cell is surrounded by a heater and
25
26
27 insulated by foil wrap and glass fiber. Inside the glass test cell, there are a thermocouple (T1) in
28
29 contact with the reaction mixture, a stir bar and a fill tube. Stirring of the reaction mixture is
30
31
32
ensured by a magnetic stirrer system.
33
34
35 The test cell assembly is inserted into the second compartment, which is a 450 mL containment
36
37 vessel made of stainless steel. A second thermocouple (T2) is placed in the upper part of the
38
39
40 second compartment to measure the temperature in the headspace and a pressure transducer is
41
42 used to follow the evolution of the pressure in the headspace. Thus, experiments were carried out
43
44 in the open test cell in closed containment.
45
46
47
48 Nitrogen was used, as an inert gas, to work under mid-pressure (20-50 bar) to minimize the
49
50 evaporation of the liquid mixture. Different initial pressures were tested to investigate the effect
51
52
53
on the kinetics of decomposition. The pressure in the headspace increases because of the
54
55 production of non-condensable products like oxygen and the temperature increase.
56
57
58
59
60
ACS Paragon Plus Environment
10
Page 11 of 43 Industrial & Engineering Chemistry Research

1
2
3
The time to charge the test cell and to start the heater is less than 10 minutes, and during that
4
5
6 time there is no significant decomposition of hydrogen peroxide.
7
8
9 Different background heating rates were tested (0.6-4°C/min). ARSST can perform
10
11
12 experiments under near-adiabatic conditions, working on the basis of heat loss compensation
13
14 principle.25,32 The Φ-factor values for the different experiments carried out varied between 1.04
15
16 and 1.05. The boiling point of water at 35 bar was estimated to be ca. 242°C by using Aspen plus
17
18
19 software v9.0 (Aspen Technology, Inc.) and by following the method described in the book of
20
38
21 Jana using the thermodynamic method of Wilson. The reaction mixture used during these
22
23 experiments was mainly composed of water. For that reason, the stop criterion for the different
24
25
26 experiments was fixed to 200°C, i.e., before the boiling point.
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
11
 Industrial & Engineering Chemistry Research Page 12 of 43

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27 1- Inert gas supply (N2).
28
29 2- 450 mL containment vessel made of stainless steel.
30
31 3- Test cell assembly made of aluminum.
32
33
4- Glass fiber.
34
35
36 5- Magnetic stir plate.
37
38 6- Pressure transducer.
39
40 7- Control and Data Acquisition.
41
42 8- Heater belt.
43
44
45
9- T1 and T2- Thermocouples.
46
47 10- Glass test cell.
48
49 11- Magnetic stirring bar.
50
51 Figure 1. Schematical representation of the experimental setup of ARSST.
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
12
Page 13 of 43 Industrial & Engineering Chemistry Research

1
2
3
3. RESULTS AND DISCUSSION
4
5
6
7 The first chapter focuses on the main parameters influencing the decomposition of hydrogen
8
9 peroxide in the ARSST unit by showing the evolution of the measured temperature versus time.
10
11
12 Based on these preliminary results and the literature, an intrinsic kinetic model was built and
13
14 described in the second chapter. The methodology to determine Tonset and the safety parameters
15
16 (TMRad, TD24 and TD8) by using the zero-order assumption is explained in the Supporting
17
18
19 Information. The standard deviation for Tonset was found to be 0.11°C showing the repeatability
20
21 of the experiments. The last chapter compares the TMRad values obtained from the zero-order
22
23 approach and the ones obtained by simulation using the intrinsic kinetic constants under
24
25
26 adiabatic conditions.
27
28
29 3.1. Parameters influencing the decomposition of hydrogen peroxide
30
31
32
33
From a kinetic viewpoint, the decomposition of hydrogen peroxide is influenced by temperature
34
35 and the concentrations of catalyst, hydrogen peroxide and sulfuric acid.22,39-40 Calorimeter
36
37 parameters such as the background heating rate and the initial pressure of nitrogen (inert gas)
38
39
40 could also have an influence on the kinetics of decomposition. During this study, we have varied
41
42 these different parameters to measure their influence on the temperature increase and the value of
43
44 TMRad. The pressure information provided by the ARSST is more qualitative than quantitative
45
46
47 due to the high volume of the gas phase compared to the liquid phase. For that reason, the
48
49 evolution of pressure was not used to determine the kinetic and thermodynamic constants during
50
51
52
the parameter estimation stage. Due to space limitation, the evolution of pressure in the gas
53
54 phase can be found in Supporting Information.
55
56
57
58
59
60
ACS Paragon Plus Environment
13
 Industrial & Engineering Chemistry Research Page 14 of 43

1
2
3
3.1.1 Calorimeter parameters
4
5
6
7 Different background heating rates were tested in the range 1-4°C/min. As the background
8
9 heating rate increases, the kinetics of decomposition is faster (Figure 2). The pressure evolution
10
11
12 for experiments displayed by Figure 2 is displayed in Figure S3 (Supporting Information).
13
14
15 200
16
17 180
18 4°C/min
19 160 2°C/min
20
21 140 1°C/min
Temperature (°C)

22 Tonset=81.1°C
23 120
24 Tonset=80.3°C
25
100
26
80
27
28 60
29 Tonset=79.1°C
30 40
31
32 20
33
34 0
35 0 10 20 30 40 50 60 70 80 90 100
36
37 Time (min)
38
39
40
41
42 Figure 2. Influence of the background heating rate on reaction temperature at an initial pressure
43
44 of ca 36 bar and with [H2O2]0= 10.96 mol/L, [CuSO4]0= 0.04 mol/L and [H2SO4]0= 0.72 mol/L.
45
46
47
48
As the background heating rate increases, the time to reach Tonset is lower. When the background
49
50 heating rate increases, it is logical to reach faster Tonset. The results of TD24 and TD8 by using the
51
52 zero-order approach are presented and compared to the ones obtained from the intrinsic model in
53
54
55 the last chapter (Table 1). Chapter 3.3 shows the evolution of TMRad(TP) obtained from the
56
57 intrinsic kinetic and zero-order model corresponding to this set of experiments.
58
59
60
ACS Paragon Plus Environment
14
Page 15 of 43 Industrial & Engineering Chemistry Research

1
2
3
By varying the initial pressure of nitrogen between 27.5 and 41 bar, the kinetics of hydrogen
4
5
6 peroxide decomposition was similar. Thus, one can conclude that the initial pressure does not
7
8 have a significant influence on TMRad(Tp) within the initial pressure range 27.5-41 bar. Casson
9
10
11
et al.39 have also made this observation by using the Phi-TEC II.
12
13
14 3.1.2 Reaction conditions parameters
15
16
17 Decomposition of hydrogen peroxide is sensitive to pH. Hydrogen peroxide is stable in acidic
18
19
20
environment.40 In the absence of sulfuric acid in the reaction mixture, the decomposition is
21
22 faster. The influence of sulfuric acid on the mechanism of decomposition can be very complex to
23
24 take into account due the different degrees of oxidation of the metals. For that reason, the
25
26
27 comparison between the safety parameters obtained from the zero-order and the intrinsic kinetic
28
29 models was done by using one concentration of sulfuric acid, i.e., [H2SO4]= 0.72 mol/L. At
30
31 lower sulfuric acid concentration, the decomposition might be too fast for the acquisition system.
32
33
34
35 As the concentration of CuSO4 increases, the kinetics of H2O2 decomposition is faster (Figure
36
37 3). Hence, when the concentration of copper increases, the values of TD24 and TD8, by using the
38
39
zero-order model, decrease (Table 2). In the absence of copper (II) sulfate, the value of TD24 is
40
41
42 ca. 3.5 higher than in the presence of this salt (Table 2). These experiments show the catalytic
43
44 effect of copper (II). A comparison between the safety parameters obtained from the intrinsic
45
46
47 kinetic model and zero-order approach is discussed in the last chapter. The evolution of pressure
48
49 for experiments displayed in Figure 3 was inserted in Supporting information (Figure S4). The
50
51 trend is similar for the evolution of temperature and pressure.
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
15
 Industrial & Engineering Chemistry Research Page 16 of 43

1
2
3
200
4
5
6 180
7
8 160
Tonset=136.4°C
9
10 140
Temperature (°C)

11
12 120
13
14 Tonset=78.2°C Tonset=81.5°C
100
15 Tonset=69.9°C
16 80
17
18 60
19
20
40
21 [CuSO4]=0.22 mol/L [CuSO4]=0.05 mol/L
22
23 20
[CuSO4]=0.03 mol/L [CuSO4]=0 mol/L
24
25 0
26 0 10 20 30 40 50 60 70 80 90 100 110 120 130
27 Time (min)
28
29
30
31 Figure 3. Effect of CuSO4 concentration on reaction temperature with the following initial
32
33
34 experimental conditions: background heating rate of 2°C/min, [H2O2]0= 10.96 mol/L, and
35
36 [H2SO4]0= 0.72 mol/L.
37
38 The concentration of hydrogen peroxide has an influence on the values of TD24 and TD8, by
39
40
41 using the zero-order model (Figure 4). When the concentration of H2O2 increases, the values of
42
43 TD24 decrease (Table 3). This observation is logical because the rate of hydrogen peroxide
44
45
decomposition depends on the concentration of hydrogen peroxide. The safety parameters
46
47
48 obtained from the intrinsic kinetic model and zero-order approach is discussed in the last chapter.
49
50 The evolution of pressure for experiments displayed in Figure 3 was inserted in Supporting
51
52
53
Information (Figure S5). The trend is similar for the evolution of temperature and pressure. As
54
55 the concentration of hydrogen peroxide increase, the pressure increase is faster and higher.
56
57
58
59
60
ACS Paragon Plus Environment
16
Page 17 of 43 Industrial & Engineering Chemistry Research

1
2
3
180
4
5
6 160
7 Tonset=81.2°C
8 140
9 Tonset=76.3°C
Temperature (°C)

10 120
11
12 100 Tonset=69.9°C
13
14
80
15
16 [H2O2]=10.96 mol/L
17 60
18 [H2O2]=7.30 mol/L
19 40
20 [H2O2]=5.48 mol/L
21 20
22
23 0
24
0 5 10 15 20 25 30 35
25
26 Time (min)
27
28
29 Figure 4. Effect of H2O2 concentration on reaction temperature with the following initial
30
31
32 experimental conditions: background heating rate of 2°C/min, [CuSO4]0= 0.22 mol/L, and
33
34 [H2SO4]0= 0.72 mol/L.
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
17
 Industrial & Engineering Chemistry Research Page 18 of 43

1
2
3
3.2 Intrinsic kinetic model for hydrogen peroxide decomposition
4
5
6
7 The zero-order model was explained in Supporting Information. In this section, the kinetics of
8
9 the reaction system, the mass and energy balance and the parameter estimation are presented. To
10
11
12 estimate the intrinsic kinetic and thermal parameters for the decomposition of hydrogen peroxide
13
14 by copper (II) sulfate, a series of 9 experiments with the ARSST was performed (Table S1,
15
16 Supporting Information).
17
18
19
20 3.2.1 Kinetics
21
22
23 One should distinguish two routes of hydrogen peroxide decomposition: thermal or spontaneous
24
25
decomposition without the aid of copper and the catalyzed decomposition by copper.
26
27
28
29 The overall reaction can be written as
30
31
32 (1)
33 H2O 2 H2O + 1/2 O 2
34
35
36 The kinetics of hydrogen peroxide decomposition can be written as
37

  =  &!%$'%(" +  *'$'+,-. /, *((11)

38
39 !"#$#% (2)
40
41
42
A first order reaction was assumed for the spontaneous decomposition of hydrogen peroxide
43
44

R 345676859: = k 345676859: . <H> O> @

45
46 (3)
47
48
49 Decomposition of hydrogen peroxide by copper (II) undergoes a complex mechanism. Based on
50
51
the article of Perez-Benito,23 the detailed kinetics of decomposition catalyzed by copper (II)
52
53
54 R Catalyzed by Cu(II) can be expressed as:
55
56
R = 2. k G . <Cu>I @. <H> O> @> + 2. k J . <Cu>I @K/> . <H> O> @
57
58 7ABC8 DB 9(EE) (4)
59
60
ACS Paragon Plus Environment
18
Page 19 of 43 Industrial & Engineering Chemistry Research

1
2
3
where, kA and kB are two rate constants. These parameters were estimated with their associated
4
5
6 activation energies during the modeling stage.
7
8
9 3.2.2 Mass and energy balance
10
11
12 Experiments were performed in an open cell where decomposition reaction produces non-
13
14
15 condensable product, i.e., O2. Thus, molar balance of an arbitrary compound (i) in the liquid
16
17 phase can be expressed by
18
19
6P
rN . VAN = + nR N,597
20
21 7
(5)
22
23
24 where, ri is the kinetics of formation or disappearance of compound (i), Vliq is the volume of the
25
6P
is the accumulation of compound i in the liquid phase and nR N,597 is the outlet
26
27 liquid phase, 7
28
29
interfacial component flux.
30
31
32
33 As mentioned in the experimental section, the stop criterion was 200°C, which is lower than the
34
35 boiling point at 35 bar. Thus, the evaporation of the reaction mixture can be assumed to be
36
37
negligible. Hence, the term nR N,597 is equal to zero for water and hydrogen peroxide, i.e.,
38

nR T U ,597 = nR T U,597 = 0 mol/s.

39
40
41
42
43 The variation of the mass is due to the release of oxygen in the headspace of the ARSST unit.
44
6] ,^P_
45
46 The liquid-gas mass transfer for oxygen is assumed to be fast, then nU ,AN → 0 mol ⇔ 7
=
47

0 mol/s.
48
49
50
51
52 6] ,`a
The accumulation of oxygen in the headspace can be written as =nR U ,597 .
53 7
54
55
56
57
58
59
60
ACS Paragon Plus Environment
19
 Industrial & Engineering Chemistry Research Page 20 of 43

1
2
3
The number of moles in the gas phase was expressed by using the ideal gas law. Thus, the mass
4
5
6 balance for oxygen in the gas phase can be expressed as
7
8
4] cdefgfaPhPfi .j^P_ . .
9
7
= +0.5. j`a
(6)
10
11
12
13 where, p O 2 is the partial pressure of oxygen in the headspace. The temperature (T2) in the gas
14
15
16 phase can be assumed constant (Figure S1b, Supporting Information).
17
18
19 The mass balances for water and hydrogen peroxide can be written as
20
21
 k ] p.q.r] . cdefgfaPhPfi .j^P_
22
= −R m8n545:N7N56 − . o− t
k ]
23
7 j^P_ s^P_
(7)
24
25
26
 k ] p.q.r] . cdefgfaPhPfi .j^P_
= R m8n545:N7N56 − . o− t
27 k ]
7 j^P_ s^P_
(8)
28
29
30
31 A detailed explanation of the equation derivation is explained in Supporting Information.
32
33
34
Energy balance was expressed by
35
36

umAN . C^P_ + mN6:8v7 . CPiadh w . 7x = q v + q 8A8n7vNnA


37
38
39
40

⇔ Φ. mAN . C^P_ .
x
= q v + q 8A8n7vNnA
41
7
42
43
44
x 
⇔ + β8A8n7vNnA
45
7
= {. (9)
^P_ .
46
^P_
47
48
49
50 where, Φ is the thermal inertia of the system, a typical value is 1.04 and βelectrical is the
51
52 background heating rate. The term qr is the heat flow-rate due to chemical reactions. Heat
53
capacity of the reaction mixture C^P_ was calculated as C^P_ (TK ) = ∑N wN . CP (TK ). The evolution
54
55
56
57 of CP for water and hydrogen peroxide with temperature was determined from Aspen plus
58
59
60
ACS Paragon Plus Environment
20
Page 21 of 43 Industrial & Engineering Chemistry Research

1
2
3
software v9.0 (Aspen Technology, Inc.) using the Wilson thermodynamic model. (Figure S6,
4
5
6 Supporting Information).
7
8
u^P_ . IPiadh . w
, where CPiadh is the heat-capacity of the test
9 ^P_ Piadh
10 The Φ-factor is equal to ^P_ .
11 ^P_

12
13 cell and equal to 0.83 kJ.kg-1.K-1 (value given by the manufacturer).
14
15
16 Heat-flow rate due to chemical reactions was expressed as
17
18
q v (TK ) = −R 345676859: . ΔH −R 7ABC8 DB 9(EE) . ΔH , 7ABC8 DB 9(EE) . VAN
19
20 ,345676859: (10)
21
22
23 According to several authors,41-42 the enthalpy of hydrogen peroxide decomposition without
24
25 catalyst is of ca. -98 kJ/mol. Thus, the value of -98 kJ/mol was used for the spontaneous
26

decomposition reaction enthalpy ΔH

27
28 ,345676859: . The reaction enthalpy for the catalyzed
29
30 decomposition, ΔH , 7ABC8 DB 9(EE) , was estimated.
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
21
 Industrial & Engineering Chemistry Research Page 22 of 43

1
2
3
3.2.3. Kinetic model: parameter estimation
4
5
6
7 ModEst software43 based on Fortran code was used to estimate: kspontaneous, Easpontaneous , kA, EaA,
8
9 ΔH , 7ABC8 DB 9(EE) , kB and EaB.
10
11
12
The temperature dependences of the rate constants (kA and kB) was described by a modified
13
14
15 Arrhenius equation,
16
17
 K K
k = k €8 . exp „ .u − w‡
18
(11)
 †d
19
20
21
22
‰Š
u w
where, k €8 = A. e .‹†d
23
24
, and Tave is the average temperature of the set of experiments. This
25
26 modification was done to decrease the correlation between the activation energy and the pre-
27
28 exponential factor.
29
30
(B BP )

The coefficient of determination of the model was expressed as: R> = 1 − (BP
31
BP )
, where y was
P Ž
32
33
34
35 the mean value of the experimental observables, ŷ i was the observable simulated by the model
36
37
38
and yi was the experimental observable.
39

The objective function ω was calculated as = ∑( − ‘ )> .

40
41
42
43
44 The objective function was minimized by using simplex and Levenberg-Marquardt algorithms.
45
46 Ordinary differential equations 6-9 were solved out by backward difference method. The
47
48
49 reaction temperature (T1) was used as an observable. As discussed in paragraph 3.1.1, the
50
51 pressure measurement is not accurate due to the high volume of the gas phase compare to the
52
53
liquid phase. Thus, the pressure rise was not used as an observable.
54
55
56
57
58
59
60
ACS Paragon Plus Environment
22
Page 23 of 43 Industrial & Engineering Chemistry Research

1
2
3
The kinetic modeling strategy and the statistical results are presented in the Supporting
4
5
6 Information (Figures S7-S9, Tables S2-S3). To minimize the number of estimated parameters,
7
8 two models were built: one in the absence of copper (II) to determine the spontaneous
9
10
11
decomposition kinetic parameters and one in the presence of copper (II) to determine the
12
13 catalyzed decomposition kinetic parameters. Based on this model, it is possible to determine
14
15 TMRad at any process temperature and also TD24 and TD8.
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
23
 Industrial & Engineering Chemistry Research Page 24 of 43

1
2
3
3.3 Evolution of TMRad from the zero-order and the intrinsic kinetic model.
4
5
6
7 By using the estimated kinetic and thermodynamic constants (Tables S2-S3, Supporting
8
9 Information), we have simulated the evolution of the reaction temperature under real adiabatic
10
11 conditions, i.e., Φ=1 and without electrical heating. Based on this simulation stage, it is possible
12
13
14 to determine the values of TMRad(TP), TD24 and TD8 from the intrinsic kinetic model and by using
15
16 the initial concentrations of the experiments described in chapters 3.1.1 and 3.1.2.
17
18
19 Tables 1-3 propose a comparison between the values of TD24 and TD8 obtained by using the
20
21
22 zero-order and the intrinsic kinetic model. The difference between the values of these safety
23
24 parameters provided by the zero-order approximation and the intrinsic kinetic model is
25
26
introduced.
27
28
29
30 The use of the zero-order approximation is widely accepted by the safety community because it
31
32 overestimates the risk. In this study, one should notice that the zero-order model does not
33
34 overestimate all the time the safety parameter. In other words, the values of TD24 or TD8 obtained
35
36
37 from the zero-order model are not always lower than the ones obtained from the intrinsic kinetic
38
39 model. The difference on TD24 or TD8 from these two models vary between ca. -9°C to ca. +23°C.
40
41
42 Figure 5 and Table 1 compare the safety parameters obtained from the zero-order at different
43
44
45 background heating rates and from the intrinsic kinetic model. One can notice that for a β equal
46
47 to 2°C/min, TMRad values obtained from both models are similar. From Table 1, one can notice
48
49
that the background heating rate has a significant influence on the values of TD24 and TD8
50
51
52 obtained from the zero-order approximation. For these three experiments performed under the
53
54 same experimental conditions but at different background heating rates, the values of TD24 and
55
56
57
58
59
60
ACS Paragon Plus Environment
24
Page 25 of 43 Industrial & Engineering Chemistry Research

1
2
3
TD8 should be similar. However, the absolute difference is higher than 5°C for a background
4
5
6 heating rate of 1 and 4°C/min.
7
8
9
10
11
12 At low value of background heating rate, the determination of Tonset is difficult because we do
13
14
15 not observe a clear temperature increase due to chemical reactions (Figure 2). For example, when
16
17 β is of 1°C/min the time to reach Tonset is 85 minutes (Table 1).
18
19
20 At higher value of β, there is an overestimation of the safety parameter by the zero-order model,
21
22
23 which do not lead to a dangerous situation. This overestimation is due to the fact that the
24
25 electrical part is not negligible on the temperature increase, making difficult the detection of
26
27 Tonset.
28
29
30
31
For this reaction system, a background heating rate of 2°C/min gives similar results for both
32
33 models as shown by Figure 5.
34
35
36 Table 1. Influence of the background heating rates on Tonset, tonset, TD24 and TD8 by using the
37
38
zero-order model (experiments illustrated by Figure 2) and comparison with the values obtained
39
40
41 from the intrinsic kinetic model.
42
43
44
45
ZERO- INTRINSIC
46
47
ORDER KINETIC
48 Background Tonset tonset TD24 TD8 TD24 TD8 Difference Difference
49 heating
50 rate(°C/min) (°C) (min) (°C) (°C) (°C) (°C) TD24 (°C) TD8 (°C)
51
1 79.1 85.3 31.7 41.6 -8.7 -4.6
52
53 2 80.3 33.1 25.7 35.2 23 37 -2.7 1.8
54 4 81.1 17.1 16.6 26.8 6.4 10.2
55
56
57
58
59
60
ACS Paragon Plus Environment
25
 Industrial & Engineering Chemistry Research Page 26 of 43

1
2
3
4 140
5
6
7
8 120
9 Zero-order: Rate 4°C/min
10
11
Zero-order: Rate 2°C/min
12 100 Zero-order: Rate 1°C/min
13
Process temperature (°C)

14 Intrinsic kinetic model

15
16 80
17
18
19
20 60
21
22
23
24 40
25
26
27
28 20
29
30
31
32
33
0
34 0 500 1000 1500 2000
35 TMRad (min)
36
37
38 Figure 5. Evolution of TMRad(TP) by using zero-order approximation for different background
39
40
41 heating rates and the intrinsic kinetic model.
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
26
Page 27 of 43 Industrial & Engineering Chemistry Research

1
2
3
Figure 6 and Table 2 compare the safety parameters obtained from the zero-order and from the
4
5
6 intrinsic kinetic model at different copper (II) sulfate concentrations. The zero-order model tends
7
8 to overestimate the risk. Besides, one can notice that as the concentration of copper decreases,
9
10
11
the difference between the safety parameters is more pronounced.
12
13
14 When the concentration of catalyst increases, the kinetics of decomposition is faster and the
15
16 values of TD24 and TD8 are shorter. In the absence of copper, the difference between the two
17
18
19
model is higher than 20°C. The time to reach the Tonset is 115 minutes in absence of catalyst
20
21 (Figure 3). As noticed in the previous chapter, when the time to reach Tonset is relatively long,
22
23 then the determination of this parameter is less accurate.
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
27
 Industrial & Engineering Chemistry Research Page 28 of 43

1
2
3
4 160
5
6
7 140
8 Zero-order: [CuSO4]=0.22 mol/L
9
10 Intrinsic: [CuSO4]=0.22 mol/L
11 120
12 Zero-order: [CuSO4]=0.0 mol/L
13
Intrinsic: [CuSO4]=0.0 mol/L
Process temperature (°C)

14
15 100
16
17
18
19 80
20
21
22
23
60
24
25
26 40
27
28
29
30 20
31
32
33
34 0
35 0 500 1000 1500 2000
36 TMRad (min)
37
38
39
40 Figure 6. Evolution of TMRad(TP) by using zero-order approximation and intrinsic kinetic model
41
42 for different copper (II) sulfate.
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
28
Page 29 of 43 Industrial & Engineering Chemistry Research

1
2
3
Table 2. Influence of copper (II) sulfate on Tonset, tonset, TD24 and TD8 by using the zero-order
4
5
6 model (experiments illustrated by Figure 3) and the intrinsic kinetic model.
7
8
9
10
11
ZERO- INTRINSIC
12 ORDER KINETIC
13 Background [CuSO4] Tonset tonset TD24 TD8 TD24 TD8 Difference Difference
14 heating
15 rate(°C/min) (mol/L) (°C) (min) (°C) (°C) (°C) (°C) TD24 (°C) TD8 (°C)
16
17 0.22 69.9 18.9 16.1 25 13 26 -3.1 1.0
18 0.05 78.2 25.4 19.3 29.1 22 35 2.7 5.9
19 2
20 0.03 81.5 28.8 21.7 31.8 25 39 3.3 7.2
21 0 136.4 115.2 69.8 81.7 93 102 23.2 20.3
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
29
 Industrial & Engineering Chemistry Research Page 30 of 43

1
2
3
Figure 7 and Table 3 compare the safety parameters obtained from the zero-order and from the
4
5
6 intrinsic kinetic model at different hydrogen peroxide concentrations. Experiments were
7
8 performed by using a background heating rate of 2°C/min and a catalyst concentration of 0.22
9
10
11
mol/L (Figure 4). Under these experimental conditions, i.e., β=2°C/min, [CuSO4]=0.22 mol/L
12
13 and [H2SO4]=0.72 mol/L, the values of TD24 and TD8 obtained by the two models are similar in
14
15 the hydrogen peroxide concentration range of 5.48-10.96 mol/L.
16
17
18
19 140
20
21
22
23 120
24 Zero-order: [H2O2]=10.96 mol/L
25
26
27 100 Intrinsic: [H2O2]=10.96 mol/L
28
Process temperature (°C)

29 Zero-order: [H2O2]=5.48 mol/L

30
31 80 Intrinsic: [H2O2]=5.48 mol/L
32
33
34
35 60
36
37
38
39
40
40
41
42
43
44 20
45
46
47
48 0
49 0 500 1000 1500 2000
50 TMRad (min)
51
52
53
54 Figure 7. Evolution of TMRad(TP) by using zero-order approximation and intrinsic kinetic model
55
56 for different hydrogen peroxide concentration.
57
58
59
60
ACS Paragon Plus Environment
30
Page 31 of 43 Industrial & Engineering Chemistry Research

1
2
3
Table 3. Influence of hydrogen peroxide concentrations on TD24 and TD8 by using the zero-order
4
5
6 model (experiments illustrated by Figure 4) and the intrinsic kinetic model.
7
8
9 ZERO- INTRINSIC
10 ORDER KINETIC
11
12 Background [H2O2] Tonset tonset TD24 TD8 TD24 TD8 Difference Difference
13 heating
14 rate(°C/min) (mol/L) (°C) (min) (°C) (°C) (°C) (°C) TD24 (°C) TD8 (°C)
15 10.96 69.89 18.9 16.1 25.0 13 26 -3.1 1.0
16
17 2 7.3 76.29 22.25 20.8 29.8 19 32 -1.8 2.2
18 5.48 81.18 25.03 23.7 33.1 24 36 0.3 2.9
19
20
21 One can notice that when the time to reach the Tonset (tonset) is longer than 90 minutes, then the
22
23 difference between the safety parameters obtained from the zero-order and the intrinsic kinetic
24
25
26 model is more pronounced for this reaction system. When the kinetics of the reaction system is
27
28 slow, then, the determination of Tonset is less obvious leading to erroneous safety parameters by
29
30 using the zero-order approximation.
31
32
33
34 The good practice is to test different background heating rates to verify the influence on TD24 and
35
36 TD8 obtained by using the zero-order approximation. If the values of these safety parameters are
37
38 very different, more investigation on the kinetics and thermodynamics are needed.
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
31
 Industrial & Engineering Chemistry Research Page 32 of 43

1
2
3
4. CONCLUSIONS
4
5
6
7
8 The determination of the TMRad values at different process temperature for the decomposition of
9
10
11
hydrogen peroxide in the absence and presence of copper sulfate was investigated. The advanced
12
13 reactive system screening tool was used to perform experiments under near adiabatic conditions
14
15 and under mid-pressure to limit the evaporation. By using the zero-order approximation, we have
16
17
18 found that the initial pressure of nitrogen does not significantly influence the safety criteria TD24
19
20 and TD8 within the pressure range 27.5-41 bar. However, the values of the background heating
21
22 rate (β) significantly influence the different values for these safety criteria for similar
23
24
25 experimental conditions.
26
27 An intrinsic kinetic model was built to estimate the intrinsic values of TMRad by taking into
28
29
account the reactant concentrations.
30
31
32 A comparison between zero-order and kinetic model for the determination of the safety
33
34 parameter, TMRad, TD24 and TD8, was done. We have noticed that the zero-order approximation
35
36
37 does not always overestimate these safety parameters. The difference between the safety criteria
38
39 obtained from the zero-order approximation and the intrinsic kinetic model vary from -9 to
40
41 +21°C for this reaction system.
42
43
44 The benefit of ARSST unit is to obtain rapidly the values of TMRad by using the zero-order
45
46 approximation, but these estimated values are sensitive to the determination of Tonset. A good
47
48 practice with ARSST is to vary the background heating rate (β) to verify its influence on TMRad
49
50
51 (based on zero-order approximation).
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
32
Page 33 of 43 Industrial & Engineering Chemistry Research

1
2
3
In that work, decomposition of hydrogen peroxide was used a test system. More investigation
4
5
6 should be done by testing different chemical system with different kinetics, to have a general
7
8 trend regarding the influence of the background heating rate.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
33
 Industrial & Engineering Chemistry Research Page 34 of 43

1
2
3
ASSOCIATED CONTENT
4
5
6
7 Supporting Information.
8
9
10 Methodology for the determination of TMRad based on the zero-order assumption; Kinetic model
11
12
13 and Modeling results.
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
34
Page 35 of 43 Industrial & Engineering Chemistry Research

1
2
3
AUTHOR INFORMATION
4
5
6 Corresponding Author
7
8
9 Sébastien Leveneur*1,2
10
11
1
12 Normandie Univ, INSA Rouen, UNIROUEN, LSPC, EA4704, 76000 Rouen, France, E-mail :
13
14
sebastien.leveneur@insa-rouen.fr
15
16
17 2
Laboratory of Industrial Chemistry and Reaction Engineering, Johan Gadolin Process
18
19
20 Chemistry Centre, Åbo Akademi University, Biskopsgatan 8, FI-20500 Åbo/Turku, Finland.
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
35
 Industrial & Engineering Chemistry Research Page 36 of 43

1
2
3
ACKNOWLEDGMENTS
4
5
6 The authors express their gratitude to Jean-Pierre Hébert for his technical assistance. The authors
7
8 thank Morgan Jones, the head of international office of INSA Rouen, to make possible the
9
10
11
master thesis of Amine Dakkoune. This paper is dedicated to Professor Tapio Salmi.
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
36
Page 37 of 43 Industrial & Engineering Chemistry Research

1
2
3
ABBREVIATIONS
4
5
6
7 )
8 CpCell Specific heat-capacity of the cell [J.kg-1.K-1]
9
10 )
11 CpR Specific heat-capacity of the reaction mixture [J.kg-1.K-1]
12
13
14
Ea Activation energy [J.mol-1]
15
16 ∆H Reaction enthalpy [J.mol-1]
17
18 k Rate constant
19
20
21 mCell Cell mass [kg]
22
23 mR Reaction mixture mass [kg]
24
25 qel Electrical heating-rate [°C.min-1]
26
27
28 qr(T) Heat-flow rate due to chemical reactions at temperature T [J.s-1]
29
30 R Gas constant [J.mol-1.K-1]
31
32 R2 Coefficient of explanation [%]
33
34
35 RCatalyzed by Cu(II) Kinetic rate for the decomposition of hydrogen peroxide by copper (II) [mol.L-1.s-1]
36
37 Rdecompositon Kinetic rate for the decomposition of hydrogen peroxide [mol.L-1.s-1]
38
39
40
RSpontaneous Kinetic rate for the spontaneous decomposition of hydrogen peroxide [mol.L-1.s-1]
41
42 ri Rate of formation or disappearance of compound i [mol.L-1.s-1]
43
44 ∆Tad Adiabatic temperature rise [°C]
45
46
47 TD8 Initial process temperature at which TMRad is 8 hours [°C]
48
49 TD24 Initial process temperature at which TMRad is 24 hours [°C]
50
51 Tfinal Final temperature under adiabatic conditions [°C]
52
53
54 tmax Time at Tmax [min]
55
56 Tmax Maximum temperature [°C]
57
58
59
60
ACS Paragon Plus Environment
37
 Industrial & Engineering Chemistry Research Page 38 of 43

1
2
3
T0 Initial reaction temperature [°C]
4
5
6 tonset Time at Tonset [min]
7
8 Tonset Onset temperature [°C]
9
10
11
VP Vapor pressure [bar]
12
13 wi Weight percentage of compound i
14
15
16
Greek letters
17
18
19
20 β Background heating rate [°C.min-1]
21
22 ω Objective function
23
24
25 mR CPR + mcell C Pcell
Φ Thermal inertia factor Φ =
26 mR C PR
27
28
29 Abbreviations
30
31
32 ARSST Advanced Reactive System Screening Tool
33
34
35
MTSR Maximum Temperature for Synthesis Reaction [K]
36
37 MTT Maximum Temperature for Technical reasons [K]
38
39 TMRad Time to Maximum Rate under Adiabatic conditions [min]
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
38
Page 39 of 43 Industrial & Engineering Chemistry Research

1
2
3
REFERENCES
4
5
6 (1) Stoessel, F. Thermal Safety of Chemical Processes; Wiley-VCH Verlag GmbH & Co. KGaA/
7
8 Weinheim, 2008.
9
10
11
(2) Balasubramanian, S. G.; Louvar, J. F. Study of Major Accidents and Lessons Learned.
12
13 Process Saf. Prog. 2002, 21, 237.
14
15 (3) Saada, R.; Patel, D.; Saha, B. Causes and Consequences of Thermal Runaway incidents—
16
17
18 Will They Ever Be Avoided? Process Saf. Environ. Prot. 2015, 97, 109.
19
20 4) Westerterp, K. R.; Molga, E. J. No More Runaways in Fine Chemical Reactors. Ind. Eng.
21
22 Chem. Res. 2004, 43, 4585.
23
24
25 (5) Leveneur, S.; Estel, L.; Crua, C. Thermal Risk Assessment of Vegetable Oil Epoxidation. J.
26
27 Therm. Anal. Calorim. 2015, 122, 795.
28
29
(6) Valdes, O. R.; Moreno, V. C.; Waldram, S.; Véchot, L.; Mannan, M. S. Runaway
30
31
32 Decomposition of Dicumyl Peroxide by Open Cell Adiabatic Testing at Different Initial
33
34 Conditions. Process Saf. Environ. Prot. 2016, 102, 251.
35
36
37 (7) Leveneur, S.; Vernieres-Hassimi, L.; Salmi, T. Mass & Energy Balances Coupling in
38
39 Chemical Reactors for a Better Understanding of Thermal Safety. Educ. Chem. Eng. 2016, 16,
40
41 17.
42
43
44 (8) Valdes, O. J. R.; Moreno, V. C.; Waldram, S. P.; Véchot, L. N.; Mannan, M. S. Experimental
45
46 Sensitivity Analysis of the Runaway Severity of Dicumyl Peroxide Decomposition Using
47
48 Adiabatic Calorimetry. Thermochim. Acta 2015, 617, 28.
49
50
51 (9) Leveneur, S.; Salmi, T.; Musakka, N.; Wärnå, J. Kinetic Study of Decomposition of
52
53 Peroxypropionic Acid in Liquid Phase through Direct Analysis of Decomposition Products in
54
55
56
Gas Phase. Chem. Eng. Sci. 2007, 62, 5007.
57
58
59
60
ACS Paragon Plus Environment
39
 Industrial & Engineering Chemistry Research Page 40 of 43

1
2
3
(10) Leveneur, S. Thermal Safety Assessment through the Concept of Structure–Reactivity:
4
5
6 Application to Vegetable Oil Valorization. Org. Process Res. Dev. 2017, 21, 543.
7
8 (11) Casson, V.; Maschio, G. Screening Analysis for Hazard Assessment of Peroxides
9
10
11
Decomposition. Ind. Eng. Chem. Res. 2012, 51, 7526.
12
13 (12) Grau, M. D.; Nougués, J. M.; Puigjaner, L. Batch and Semibatch Reactor Performance for
14
15 an Exothermic Reaction. Chem. Eng. Process. 2000, 39, 141.
16
17
18 (13) Vernieres-Hassimi, L.; Assoudi-Baikari, R. E.; Abdelghani-Idrissi, M.-A.; Mouhab, N.
19
20 New Analytical Method for Maximum Temperature Assessment in an Exothermic Tubular
21
22 Chemical Reactor. Chem. Eng. Commun. 2016, 203, 174.
23
24
25 (14) Vernières-Hassimi, L.; Leveneur, S. Alternative Method to Prevent Thermal Runaway in
26
27 Case of Error on Operating Conditions Continuous Reactor. Process Saf. Environ. Prot. 2015,
28
29
98, 365.
30
31
32 (15) Anxionnaz, Z.; Cabassud, M.; Gourdon, C.; Tochon, P. Transposition of an Exothermic
33
34 Reaction From a Batch Reactor to an Intensified Continuous One. Heat Transfer Eng. 2010, 31,
35
36
37 788.
38
39 (16) Noyori, R. Pursuing Practical Elegance in Chemical Synthesis. Chem. Commun. 2005, No.
40
41 14, 1807.
42
43
44 (17) Ciriminna, R.; Albanese, L.; Meneguzzo, F.; Pagliaro, M. Hydrogen Peroxide: A Key
45
46 Chemical for Today’s Sustainable Development. ChemSusChem 2016, 9, 3374.
47
48 (18) Hage, R.; Lienke, A. Applications of Transition-Metal Catalysts to Textile and Wood-Pulp
49
50
51 Bleaching. Angew. Chem., Int. Ed. 2006, 45, 206.
52
53 (19) Fauske, H. K. Managing Chemical reactivity—Minimum Best Practice. Process Saf. Prog.
54
55
56
2006, 25, 120.
57
58
59
60
ACS Paragon Plus Environment
40
Page 41 of 43 Industrial & Engineering Chemistry Research

1
2
3
(20) Kokes, H.; Morcali, M. H.; Acma, E. Dissolution of Copper and Iron from Malachite Ore
4
5
6 and Precipitation of Copper Sulfate Pentahydrate by Chemical Process. Engineering Science and
7
8 Technology, an International Journal 2014, 17, 39.
9
10
11
(21) Adebayo, A. O.; Ipinmoroti, K. O.; Ajayi, O. Dissolution Kinetics of Chalcopyrite with
12
13 Hydrogen Peroxide in Sulphuric Acid Medium. Chem. Biochem. Eng. Q. 2003, 17, 213.
14
15 (22) Mlasi, B.; Glasser, D.; Hildebrandt, D. Kinetics of the Decomposition of Hydrogen
16
17
18 Peroxide in Acidic Copper Sulfate Solutions. Ind. Eng. Chem. Res. 2015, 54, 5589.
19
20 (23) Perez-Benito, J. F. Copper(II)-Catalyzed Decomposition of Hydrogen Peroxide: Catalyst
21
22 Activation by Halide Ions. Monatsh. Chem. 2001, 132, 1477.
23
24
25 (24) Bhattacharya, A. A General Kinetic Model Framework for the Interpretation of Adiabatic
26
27 Calorimeter Rate Data. Chem. Eng. J. 2005, 110, 67.
28
29
(25) Marco, E.; Cuartielles, S.; Peña, J. A.; Santamaria, J. Simulation of the Decomposition of
30
31
32 Di-Cumyl Peroxide in an ARSST Unit. Thermochim. Acta 2000, 362, 49.
33
34 (26) Ait Aissa, K.; Zheng, J. L.; Estel, L.; Leveneur, S. Thermal Stability of Epoxidized and
35
36
37 Carbonated Vegetable Oils. Org. Process Res. Dev. 2016, 20, 948.
38
39 (27) Duh, Y.-S.; Kao, C.-S.; Lee, C.; Yu, S. W. Runaway Hazard Assessment of Cumene
40
41 Hydroperoxide From the Cumene Oxidation Process. Process Saf. Environ. Prot. 1997, 75, 73.
42
43
44 (28) Huang, C.-C.; Peng, J.-J.; Wu, S.-H.; Hou, H.-Y.; You, M.-L.; Shu, C.-M. Effects of
45
46 Cumene Hydroperoxide on Phenol and Acetone Manufacturing by DSC and VSP2. J. Therm.
47
48 Anal. Calorim. 2010, 102, 579.
49
50
51 (29) Véchot, L.; Bigot, J.-P.; Testa, D.; Kazmierczak, M.; Vicot, P. Runaway Reaction of Non-
52
53 Tempered Chemical Systems: Development of a Similarity Vent-Sizing Tool at Laboratory
54
55
56
Scale. J. Loss Prev. Process Ind. 2008, 21, 359.
57
58
59
60
ACS Paragon Plus Environment
41
 Industrial & Engineering Chemistry Research Page 42 of 43

1
2
3
(30) Chi, J.-H.; Wu, S.-H.; Shu, C.-M. Thermal Explosion Analysis of Methyl Ethyl Ketone
4
5
6 Peroxide by Non-Isothermal and Isothermal Calorimetric Applications. J. Hazard. Mater.
7
8 2009, 171, 1145.
9
10
11
(31) Veedhi, S.; Sawant, A. Designing a Safer Process for the Reaction of TFA with Sodium
12
13 Borohydride in THF by Calorimetric Technique. J. Therm. Anal. Calorim. 2013, 111, 1093.
14
15 (32) Tang, W.; Sarvestani, M.; Wei, X.; Nummy, L. J.; Patel, N.; Narayanan, B.; Byrne, D.;
16
17
18 Lee, H.; Yee, N. K.; Senanayake, C. H. Formation of 2-Trifluoromethylphenyl Grignard Reagent
19
20 via Magnesium−Halogen Exchange: Process Safety Evaluation and Concentration Effect. Org.
21
22 Process Res. Dev. 2009, 13, 1426.
23
24
25 (33) Veedhi, S.; Mishra, V.; Kulkarni, S.; Gorthi, R. Incident Investigation on Thermal
26
27 Instability of an Intermediate Using Adiabatic Calorimeter. J. Therm. Anal. Calorim. 2014, 115,
28
29
909.
30
31
32 (34) Shimizu, S.; Imamura, Y.; Ueki, T. Incompatibilities between N-Bromosuccinimide and
33
34 Solvents. Org. Process Res. Dev. 2014, 18, 354.
35
36
37 (35) Theis, A. E.; Burelbach, J. P.; Askonas, C. F. Safely Scale-up Processes and Accommodate
38
39 Recipe Changes. Process Saf. Prog. 2009, 28, 135.
40
41 (36) Sarge, S. M., Höhne, G. W. H., Hemminger, W. In Calorimetry: Fundamentals,
42
43
44 Instrumentation and Applications; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, 2014.
45
46 (37) Mannan, M., Lees’ Process Safety Essentials; Butterworth-Heinemann: Oxford, 2014.
47
48 (38) Jana, A.K. Process Simulation and Control using AspenTM; PHI Learning: New Delhi, 2012.
49
50
51 (39) Casson, V.; Battaglia, E.; Maschio, G. Hydrogen Peroxide Decomposition Analysis by
52
53 Screening Calorimetry Technique. Chem. Eng. Trans. 2012, 26, 27.
54
55
56
(40) Koskinen, H.; Leveneur, S.; Sundquist, A.; Musakka, N.; Salmi, T.; Renvall, I.
57
58
59
60
ACS Paragon Plus Environment
42
Page 43 of 43 Industrial & Engineering Chemistry Research

1
2
3
Functionality of Poly(alpha-Hydroxyacrylic Acid) as H2O2 Stabilising Agent. Oxid. Commun.
4
5
6 2010, 33, 258.
7
8 (41) Tatsuoka, T.; Koga, N. Energy Diagram for the Catalytic Decomposition of Hydrogen
9
10
11
Peroxide. J. Chem. Educ. 2013, 90, 633.
12
13 (42) Marzzacco, C. J. The Enthalpy of Decomposition of Hydrogen Peroxide: A General
14
15 Chemistry Calorimetry Experiment. J. Chem. Educ. 1999, 76, 1517.
16
17
18 (43) Haario, H., 2001. MODEST-User’s Guide; ProfmathOy: Helsinki, 2001.
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
43