A new reaction calorimeter for screening purposes during process development

Article abstract of DOI:10.1021/op025575q

Calorimetry, in combination with thermal analysis, is an essential tool for a data-based assessment of thermal risks linked with the performance of chemical reactions at industrial scale. The energies of synthesis reactions or of decomposition reaction as well as the heat capacities of reaction masses can be measured by these techniques. The performance of the differential reaction calorimeter (DRC) from SETARAM in the determination of essential safety data was demonstrated using two example reactions. The differential reaction calorimeter was found to be a powerful screening tool in an organic synthesis laboratory or in a development laboratory, and it is especially well suited for a fast and low-cost determination of the thermal parameters of chemical reactions, even when only a few raw materials are available.

Full text of DOI:10.1021/op025575q

Organic Process Research & Development 2002, 6, 915−921
A New Reaction Calorimeter for Screening Purposes during Process
Development
R e´ mi Andr e´ , Leila Bou-Diab, Pablo Lerena, Francis Stoessel,* Maxime Giordano, and Christophe Mathonat§
†
†
†
,†,‡
§
Swiss Institute for the Promotion of Safety & Security, WKL 32-322, CH-4002 Basel, Switzerland, Swiss Federal Institute
of Technology, Institute of Process Sciences, CH-1015 Lausanne, Switzerland, and SETARAM, 7 rue de l’Oratoire,
69300 Caluire, France
Abstract:
Calorimetry, in combination with thermal analysis, is an
essential tool for a data-based assessment of thermal risks linked
with the performance of chemical reactions at industrial scale.
The energies of synthesis reactions or of decomposition reaction
as well as the heat capacities of reaction masses can be measured
by these techniques. The performance of the differential
reaction calorimeter (DRC) from SETARAM in the determi-
nation of essential safety data was demonstrated using two
example reactions. The differential reaction calorimeter was
found to be a powerful screening tool in an organic synthesis
laboratory or in a development laboratory, and it is especially
well suited for a fast and low-cost determination of the thermal
parameters of chemical reactions, even when only a few raw
materials are available.
Figure 1. Cooling failure scenario.
the temperature. The maximum temperature reached during
the reaction loss of control is called maximum temperature
of the synthesis reaction (MTSR). At this point, a secondary
decomposition reaction may be triggered and leads to a
further temperature increase.
Introduction
One of the tasks of the chemist in charge of the
development of an industrial process is to ensure the process
safety. This means that he has to develop a process, which
is robust against deviations from normal operating condi-
A systematic examination of the phenomena occurring
after a cooling failure allows drawing a so-called runaway
scenario (Figure 1) from which the required thermal data
1
tions. Thus, at early stages of the process development,
questions will be asked that require safety data to be able to
evaluate thermal hazards, (i.e., the capability of a system to
enter into a runaway reaction).
Thermal runaways are the cause of many incidents in
process chemical industry, and some of them resulted in
3
can be drawn: (1) the heat of the desired reaction,
corresponding to the first step of the cooling failure scenario,
(2) the heat of decomposition corresponding to the second
step of the diagram, and (3) the heat capacity of the mixtures.
These data enable the evaluation of the time available
before the beginning of the decomposition reaction starts at
the MTSR. The MTSR depends on the accumulation of
nonconverted reagents, the heat of reaction, and the heat
capacity of the reaction mass.
2
major accidents. Thermal runaways are due to disturbances
of the heat balance of a reactor, resulting in a fast increase
of temperature in the reactor, which may result in a sharp
increase of the pressure. Therefore, the chemist has to know
the capability of a system to enter into a runaway reaction.
Considering a chemical reactor equipped with a jacket in
which a cooling fluid circulates, if either the cooling system
or the stirrer fails, heat removal is stopped and the reaction
cannot be controlled anymore. If an important quantity of
reagent, which has not been converted yet, remains in the
reactor, its reaction will lead to a noncontrolled increase of
The time to maximum rate under adiabatic conditions
(TMRad), or time left before the explosion occurs, can also
be calculated from the heat-release rate, the heat capacity,
and the energy of activation of the reaction. This implies a
kinetic study of the reaction, which can be restricted to a
zero-order approximation when used for safety purposes
only.
One way of obtaining this essential data about tempera-
ture, pressure, and energy is the experimental determination
by thermal analysis and calorimetry.
*
Corresponding author. E-mail: francis.stoessel@epfl.ch.
Swiss Institute for the Promotion of Safety & Security.
Swiss Federal Institute of Technology.
SETARAM.
†
‡
§
(1) Stoessel, F. What Is Your Thermal Risk? Chem. Eng. Progr. 1993, October,
6
8-75.
(
2) Maddison, N.; Rogers, R. L. Chemical Runaways, Incidents and Their
Causes. Chem. Technol. Eur. 1994, 11-12, 28-31.
(3) Gygax, R. Chemical Reaction Engineering for Safety. Chem. Eng. Sci. 1988,
43, 1759-1771.
1
0.1021/op025575q CCC: $22.00 © 2002 American Chemical Society
Vol. 6, No. 6, 2002 / Organic Process Research & Development
•
915
Published on Web 11/15/2002

experiment is easy and fast. Reagents are placed in the
crucible and are submitted to a temperature ramp. Due to
the small size of the sample to be introduced into the crucible,
it is difficult to study heterogeneous mixtures quantitatively.
Due to the small volume, special care is required to obtain
representative samples of the reaction mass. Therefore, DSC
is mainly used to study decomposition reactions in which
the heat exchanged is very high and the high sensitivity of
measurement is not needed. It is only rarely used to study
the desired reaction.
•
The Calvet calorimeter based on the Tian-Calvet
principle has a furnace provided with two cylindrical cells
in which the sample and reference are placed. Samples are
surrounded by thermopiles enabling the integration of the
energy released during the experiment. A few grams of
product can be analyzed. Certain versions of the Calvet
calorimeter enable the simulation of a chemical reaction in
batch operation. Up to 12 mL of product can be analyzed. It
is a very useful tool for measuring heat capacity and heat of
reaction. For example, stirring is possible; therefore, it can
be used to study the heat of reaction of heterogeneous
mixtures with a good accuracy. The pressure can also be
measured during a reaction.
Figure 2. Temperature in °C and sensitivity ranges in W‚kg-1
of commonly used calorimetric methods in safety laboratories.
2. Calorimeters for Safety Studies
There is a broad choice of various calorimeters based on
different calorimetric principles available on the market, but
safety studies impose specific requirements that only some
calorimeters meet. The criteria for choosing the appropriate
calorimetric method can be summarized as follows:
•
The amount of sample available: on one hand, at early
stages of the development of a process, only a few milligrams
of samples may be available or may be “sacrificed” for safety
studies. On the other hand, studying a runaway reaction will
preferably not be performed at large scale.
•
The purpose of adiabatic calorimeters is to simulate
runaway conditions. In fact, true adiabatic conditions are
difficult to carry out experimentally. Certain methods are
based on Dewar calorimetry, avoiding heat losses by thermal
insulation, whereas other methods avoid heat losses by
adapting the temperature of the sample surroundings to the
sample temperature. The problem is that, whichever tech-
nique is used, a part of the heat of reaction is used to heat
up the crucible. Therefore, the so-called φ correction is of
primary importance. With most of these calorimeters, a
simultaneous measure of the pressure is also possible.
•
The sensitivity of the calorimeter is important, if only
small heat-release rates have to be measured. This is
especially important in the study of stability at storage or
transportation, where even a heat release of a few milliwatts
per kilogram may be critical. On the other hand, for tracking
the steep part of a runaway reaction as required for vent-
sizing purposes, there is no need for high sensitivity.
•
The temperature and pressure range must be adapted to
the process temperature range and also to the range in which
secondary decomposition reactions may be triggered. Both
parameters, temperature range and sensitivity, can be rep-
resented in a graphical form allowing the illustration of the
ranges as an Arrhenius diagram (Figure 2).
•
Reaction calorimeters are instruments making possible
the performance of chemical reactions under conditions close
to those of industrial ones, while making possible quantitative
measurements of the heat release rate. Traditional reaction
calorimeters use only one reaction mass having a volume of
about one liter.
Recently, SETARAM has commercialized a new reaction
calorimeter. This calorimeter developed in collaboration with
Aventis and their Security laboratory in Neuville/ Sa oˆ ne will
be presented in more detail following a study, which has
been made at the Swiss Institute of Safety.
•
The operating conditions: the dynamic or scanning
mode is well adapted for screening purposes in matters of
thermal stability, whereas the capability of reproducing
adequate process conditions is essential for the study of a
synthesis reaction.
Therefore, four main kinds of calorimetric devices are
commonly used in a safety testing laboratory: differential
scanning calorimeters (DSC), Calvet calorimeters, adiabatic
calorimeters, and calorimetric reactors.4
3
. Description of the Differential Reaction Calorimeter
.1. General Description of the Calorimeter. The new
reaction calorimeter called differential reaction calorimeter
DRC) enables an easy and rapid determination of the most
-6
3
•
DSC is a technique in which the change of the heat-
flow rate difference between the sample and a reference
sample is analyzed while they are subjected to a temperature
(
7
important thermodynamics data such as the heat of reaction
and heat capacity of the reaction mass. It is very simple to
use; its design is very similar to classical organic chemistry
laboratory equipment. The instrument allows studying chemi-
cal reactions in the same conditions as those required by
scan. DSC is a low-cost method using small amounts of
sample: that is, a few milligrams of product. Setting up the
(
4) Heemskerk, A. H. Chemical ReactiVity EValuation and Application to
Process Design; American Institute of Chemical Engineers, Center for
Chemical Process Safety: New York, 1995.
(5) Steinbach, J. Safety Assessment for Chemical Processes, 1999 ed.; VCH:
Weinheim, 1999.
(7) Brown, M. E. Principles and Practice. Handbook of Thermal Analysis and
(6) Rogers, R.; Barton, A. Chemical Reaction Hazards; Institution of Chemical
Engineers: Rugby, UK, 1997.
Calorimetry; Gallagher, P. K., Ed.; Elsevier: The Netherlands, 1998; Vol.
1.
916
•
Vol. 6, No. 6, 2002 / Organic Process Research & Development

Figure 4. Calibration of the differential reaction calorimeter
by Joule effect.
c
calibration (q ) can be set up to 10 W. If, during the
calibration, no reaction takes place and the reactor is not
fed, the heat balance may be written:
Figure 3. Schematic view of the differential reaction calorim-
eter DRC from SETARAM.
classical laboratory procedures because of the use of stirrers,
a thermostatic jacket, and the possibility of adding liquids
or gases in the reactor even during measurement.
d(TR1 - TR2)
qc ) UA(TR1 - T ) + (m c + cpi)
(2)
R2
R pR
dt
3
.2. Measurement Principles. The reaction calorimeter,
The heat-transfer characteristics are obtained by integration
of the signal over time:
DRC, continuously measures a temperature difference be-
tween the working reactor and the reference reactor (Figure
3
). The reaction to be studied is performed in the working
∞
∫
q dt
c
reactor, whereas the reference reactor only contains a solvent
with physical properties close to those of the reacting
mixture. The reference reactor enables the correction of the
perturbations of the system to make possible an accurate
determination of the heat of reaction and heat capacity. All
phenomena that are not directly related to the reaction, such
as temperature fluctuations of heat carrier, heat losses through
the top of the reactor, and heat related to the stirring, can be
corrected by using the reference reactor.
According to the temperature control of its surrounding,
the DRC is classified as an isoperibolic calorimeter. Both
reactors are double-enveloped spherical reactors connected
in parallel. A heat-carrier fluid flows through the jackets,
maintaining the reactor surroundings at a constant temper-
ature with a (0.01°C stability. The overall heat balance can
be written:
0
UA )
(3)
∞
∫
(TR1 - TR2)dt
0
The specific heat capacity of the reactor’s contents can be
obtained by evaluating the thermal relaxation after the
calibration heater has been switched off. eq 2 can be
integrated:
qc
[1 - e-(t-t )/τ]
0
(4)
(
TR1 - T ) ) (T - T ) +
R2 (t )
0
R2 (t)
R1
UA
with the time constant,
τ )
mrcpR + cpi
(5)
UA
d(TR1 - TR2)
The heat capacity of the inserts (cpi) is determined by a
calibration using a solvent with a known specific heat
capacity.
qR ) UA(T - T ) + (m c + cpi)
+
R1
R2
R pR
dt
m˘ cpdos(TR1 - Tdos) (1)
3.3. Calorimeter Features. Reactors are made up of glass
Measuring the temperature difference between both reactors
R1 - TR2) enables the determination of the heat-release
rate of the reaction, if the heat transfer (UA) and the heat
capacities of the reactor contents and the inserts (m
pi) are known. U is the overall heat-transfer coefficient
and use a glass stirrer with PTFE blades and metal inserts
(temperature probe and Joule effect probe). The temperature
in the working reactor and the one in the reference reactor
are measured using a platinum-tantalum-cased probe. Three
kinds of reactors can be adapted on the reaction calorimeter.
Two double-enveloped spherical reactors having respectively
a volume of 250 and 500 mL, are available. The double-
enveloped Keller-type reactor has a volume of 100 mL. Such
small-volume reactors allow working with small quantities
of reagent and solvent, rending the calorimeter useful for
studies carried out on expensive products such as those in
the pharmaceutical industry.
(
T
R
c
pR
+
c
through the reactor wall to the heat-carrier fluid flowing in
the jacket. A is the exchange area that depends on the quantity
of solvent in the reactor. The term mcpdos (TR1 - Tdos
)
corresponds to the heat due to the feed, which may be at a
different temperature (Tdos) than that of the reactor.
A Joule effect probe made up of a special alloy is used
to calibrate the calorimeter (Figure 4). The power used for
Vol. 6, No. 6, 2002 / Organic Process Research & Development
•
917

Reactors have a central port, which receives the agitation
bearing through which the stirrer passes. Agitation speed
inside the reactors is fixed by stirring motors. Reactors also
have four connection pipes, which enable the introduction
of the temperature and calibration probes as well as any other
accessory (a pH probe, for example). Reagents are added in
the calorimeter using a dosing funnel or a driven syringe
pump. All the accessories are held in place using glass pipes.
Therefore, installing and starting up the DRC are very easy
operations.
Figure 5. Reaction scheme of the hydrolysis of acetic anhy-
dride.
same temperature as that of the measure reactor, avoiding
the perturbation by the sensible heat of addition. This
procedure allows for checking the effect of the sole viscosity
on calorimetric measurement. From the heat input of 15 kJ
the integration of the signal using a linear interpolation of
UA rends 13.5 kJ (-10.0%). As a comparison, a simulta-
neous addition of 50 mL of the PEG 3000 solution to both
reactors was performed: the integration of the signal gives
13.2 kJ (-12.0%). The results can be improved by taking
into account the nonlinear variation of calorific sensitivity
(UA) using a spreadsheet: the first experiment gives 14.7
kJ (-2.0%), and the second, 14.2 kJ (-5.3%).
These results are within the commonly accepted tolerance
of (15% for safety screening, showing that the approxima-
tion made in eq 1 is valid in this frame.
To test the calorimeter under more realistic conditions,
two reactions have been studied in the DRC, using the
isothermal batch and semi-batch operation.
Since this calorimeter is operated as a differential
calorimeter, the question of symmetry and especially of loss
of symmetry during a reaction arises.
3.4. Effect of Loss of Symmetry on the Performance
of the Differential Calorimeter. Equation 1 assumes the
same calorific sensitivity (UA) in both measure and reference
reactors, but during performance of a reaction this term may
vary either due to the change of the heat-exchange area of
the measure reactor resulting from addition of a reactant in
semi-batch operation or due to a viscosity change in the
measure reactor resulting from the chemical reaction.
To study the effect of a volume increase in the sole
measure reactor, a semi-batch reaction was simulated by
charging 100 mL of water into both measure and reference
reactors. Then 100 mL (doubling of volume) of water was
added to the measure reactor only with a simultaneous heat
input from the calibration heater. The added water was
maintained at the same temperature as that of the measure
reactor, avoiding the perturbation by the sensible heat of
addition. From the heat input of 15 kJ the integration of the
signal using a linear interpolation of UA rends 13.1 kJ
4. Example 1: Hydrolysis of Acetic Anhydride
4
.1. Reaction. The first reaction studied is a classical
“
calibration” reaction for calorimeters: the hydrolysis of
acetic anhydride (Figure 5). It is a fast reaction well suited
for checking the dynamic response of a calorimeter. It has
been studied in batch mode using the DRC and the 250-mL
reactors.
Acetic anhydride was obtained from Fluka (Fluka 45830,
puris. p.a. ACS g 99.5%). Deionized water was used.
(-12.7%). As a comparison, a simultaneous addition of 50
mL of water to both reactors was also performed: the
integration of the signal gives 13.4 kJ (-10.7%). The results
can be improved by taking into account the nonlinear
variation of the heat-exchange area (A) using a spreadsheet:
the first experiment gives 14.5 kJ (-3.3%), and the second,
4
.2. Experimental Procedure. The reaction has been
performed as an isothermal batch operation at three different
temperatures 10, 25, and 40 °C. The reactor is thermally
equilibrated at working temperature and only contains the
solvent: 150 g (8.33 mol) of water. A heated dosing funnel,
maintained at the temperature of the experiment, was used
to introduce 12 g (0.12 mol) of acetic anhydride in one
portion. The reference reactor contains 160 g of water, which
is almost equal to the final reaction mass in the working
reactor. In these conditions, symmetry is maintained between
both reactors; hence, U and A are very similar in both
reactors, and UA may be measured only once at the end of
the experiment.
14.2 kJ (-5.3%).
The other common cause for loss of symmetry is viscosity
change. A series of calibrations was performed with water
in the measure reactor and poly(ethylene glycol) (PEG 3000)
-
1
solutions 500 and 700 g‚L in water into the reference
reactor. These concentrations represent a change of the
viscosity by ca. 140 mPa‚s. The calorific sensitivities (UA)
-1
increased by 0.37 and 0.47 W‚K , but this had no significant
effect on the determination of the specific heat capacities,
-1
-1
which remained constant at 4.10 ( 0.03 J‚g ‚K . The PEG
solution was added into the reference reactor to maintain
the specific heat capacity of the measure reactor contents
constant.
4.3. Results. The evolution of the working reactor
temperature as a function of time allows following the
reaction progress qualitatively: that is, the temperature returns
to its initial value (baseline) after about 45 min at 40 °C. In
the example shown in Figure 6, the maximum heat release
To ensure conditions that are closer to practical ones, a
semi-batch reaction was simulated by charging 100 mL of a
-
1
5
00 g‚L solution of poly(ethylene glycol) in water into
-
1
both measure and reference reactors. Then 50 mL of water
was added to the measure reactor and within the same time
was added 50 mL of the same PEG solution into the
reference reactor with a simultaneous heat input from the
calibration heater. The added solution was maintained at the
rate was found ca. 90 W‚kg , which would require a rather
high cooling capacity for an industrial reactor. The reaction
enthalpy is obtained by integration of the temperature
difference with time using the Joule effect calibration
performed at the end of the experiment with a horizontal
918
•
Vol. 6, No. 6, 2002 / Organic Process Research & Development

Table 2: Results of esterification reaction in the batch mode
temperature in °C
conversion at the end of the
experiment
60
0.868
70
0.934
80
0.968
-
1
enthalpy of reaction in kJ‚mol
-62.5
-58.1
-63.3
5
.2. Reaction Performed as Batch Reaction. The study
was first performed in the batch mode at three different
temperatures: 60, 70, and 80 °C. Seventy-five grams (ca.
0
.56 mol) of propanoic anhydride were charged into the
working reactor, and 43 g (ca. 0.58 mol) of 2-butanol were
added; 130 g of water was placed in the reference reactor.
To calculate the heat of reaction, it was necessary to know
which quantity of reagent had reacted. Therefore, after
reaction, the degree of conversion was obtained by chemical
analysis of the final reaction mass by gas chromatography.
The results are summarized in Table 2. At 60 and 70 °C,
the reaction does not reach completion after 20 h (time at
which the study was stopped). At 80 °C, after 20 h, the
conversion is practically complete.
Figure 6. Thermogram of the hydrolysis of acetic anhydride
with water at 40 °C.
Table 1: Measured enthalpy of reaction for the hydrolysis of
acetic anhydride
temperature in °C
enthalpy of reaction in kJ‚mol
10
57.6
25
57.7
40
58.3
-
1
These values were corrected by the heat of mixing
∆Hmix ) + 4.2 kJ‚mol ).
baseline also drawn from the end of the experiment. The
-1 12
(
results are summarized in Table 1.
The average value of the enthalpy of reaction obtained
1
The average value is -∆H
R
) 57.9 kJ‚mol- with a
-1
for the three temperatures is -∆H ) 61.63 kJ‚mol with
R
standard deviation of 0.67%. This value compares well with
a standard deviation of 4.6%. The heat of reaction is
-
1
values from literature: -∆H ) 60.4 kJ‚mol between 15
independent of the temperature and compares well with
and 35°C and -∆H ) 58.3 kJ‚mol at 30°C.9
For this relatively fast reaction, the differential reaction
calorimeter gives reproducible results with a good accuracy
when compared to literature data.
8
-1
-1 13,14
literature values: -∆H
R
) 60 kJ‚mol
-∆H ) 64
R
-1 15
-1 16
kJ‚mol , -∆H ) 63.7-69.5 kJ‚mol .
R
This shows that the differential reaction calorimeter allows
obtaining reliable results for enthalpies of reaction even for
slow reactions, which requires a good thermal sensitivity
especially at the end of the experiment, where the heat release
rate is low and becomes difficult to distinguish from the
measurement noise. In such cases, it is essential to complete
5
2
. Example 2: Esterification of Propanoic Anhydride by
-Butanol
1
7
the measurement by a chemical analysis.
.3. Reaction Performed as Semi-Batch Reaction. As
5
described in the Introduction, the accumulation of reagents
in a semi-batch reactor is essential information for the safety
assessment. Therefore, this relatively slow reaction was also
performed in the semi-batch mode, that is, with an addition
of one of the reagents.
Figure 7. Reaction scheme of the esterification of acetic
anhydride by 2-butanol.
5.1. Reaction. As a second example, a slow reaction was
chosen: the esterification of propanoic anhydride by 2-bu-
tanol (Figure 7). This reaction is weakly exothermic, its
kinetics may be influenced by catalysis, and there is no
secondary decomposition reaction. Therefore, it is often used
About 100 g (ca. 1 mol) of propanoic anhydride was
initially charged into the working reactor, 57 g of 2-butanol
(0.77 mol) was added at a constant rate during 100 min using
as a test reaction for calorimetry.1
0,11
a pump and a balance. The reagent was added in the working
reactor only; 170 g of propanoic anhydride was placed in
the reference reactor. The variation of reaction volume is
about 70 mL in the working reactor.
In the conditions we
used, it is slow and well adapted to the study of reactant
accumulation a in semi-batch reactor and to test the thermal
sensitivity of the calorimeter.
Propanoic anhydride was obtained from Fluka (Fluka
1942, purum 98%) as well as butanol-2 (Fluka 19440, puris.
p.a. g99.5%).
(12) Ubrich, O.; Srinivasan, B.; Lerena, P.; Bonvin, D.; Stoessel, F. The Use of
Calorimetry for On-Line Optimisation of Isothermal Semi-Batch Reactors.
Chem. Eng. Sci. 2001, 56, 5147-5156.
13) Cronin, J. L.; Nolan, P. F.; Barton. J. A. Strategy for the Thermal Hazard
Evaluation of Chemical Reactions, Illustrated by an Analysis of the Nitration
of Toluene. In Int. Symp. Runaway React. 1989, 633-659.
8
(
(
8) Martin, H. W a¨ rmeflusskalorimetrie unter pr a¨ paratiVen Bedingungen und
ihre Anwendung zur Verfolgung der Isomerisierungskinetik Von Trimeth-
ylphosphit. Basel, 1973.
(14) Riesen, R.; Grob, B. Reaktionskalorimetrie in der chemischen Prozess-
Entwicklung. Swiss Chem. 1985, 7(56).
(
9) Smith, T. L. J. Phys. Chem. 1955, 59, 385-389.
(15) Wright, T. K.; Rogers, R. L. Adiabatic Dewar calorimeter. In Hazards IX;
Institution of Chemical Engineers: Rugby, UK, 1986; Vol. 97, pp 121-
132.
(16) Steele, C. H.; Nolan, P. F. Int. Symp. Runaway React. 1989, 191-231.
(17) Regenass, W. The Development of Stirred-Tank Heat Flow Calorimetry
as a Tool for Process Optimization and Process Safety. Chimia 1997, 51,
189-200.
(
10) Snee, T. J; Barcons, C.; Hernandez, H.; Zaldivar, J. M. Characterization
of an Exothermic Reaction Using Adiabatic and Isothermal Calorimetry.
J. Therm. Anal. 1992, 38, 2729-2747.
11) Galvan, I. M.; Zaldivar, J. M.; Hernandez, H.; Molga, E. The Use of Neural
Networks for Fitting Complex Kinetic Data. Comput. Chem. Eng. 1996,
(
2
0, 1451-1465.
Vol. 6, No. 6, 2002 / Organic Process Research & Development
•
919

Figure 8. Feed, conversion, and accumulation in function of time during the esterification of propanoic anhydride with isobutanol
at 70 °C as semi-batch reaction.
Table 3: Results of esterification reaction in the semi-batch
mode
Table 4: Results of esterification reaction in the batch mode
temperature in °C 60 70 80
measured rate constant in L‚mol 1‚h-1 0.031 0.088 0.226
-
temperature in °C
conversion at the end of the experiment
enthalpy of reaction in kJ‚mol
70
0.934
-56.8
80
0.968
-57.6
rate constant from literature in
0.041 0.093 0.219
-
1
-1 -1
L‚mol ‚h
The determination of the reaction enthalpy was performed
using eq 1, that is, the sensitive heat of the feed was taken
into account. The results are summarized in Table 3.
The average of the measured values of the reaction
sion is identical to the chemical conversion. Thus, the kinetics
can be evaluated from the experiments performed in the
batch-mode by using:
enthalpy in the semi-batch mode is -∆H
R
) 57.2 ( 0.4
t
-1
∫
qR(t′)dt′
kJ‚mol . This value is slightly inferior to the value obtained
in the batch mode but remains in good agreement with
literature values.
The reagent addition as a function of time makes possible
the quantitative determination of the instantaneous thermal
conversion (Xth). In turn, the determination of the thermal
conversion, linked with the measure of the percentage of
addition enables obtaining the thermal accumulation of
reagent:
0
Xth(t) )
(7)
∞
∫
q (t′)dt′
R
0
Thus, the heat release rate curve (expressed in W/kg) as a
function of time can be transformed into conversion curves
as a function of time, which content can, in turn, be directly
used for the kinetic evaluation. The esterification of pro-
panoic anhydride with 2-butanol is known to be a second-
order reaction. The rate constant k was calculated from the
experiments performed at 60, 70, and 80 °C respectively and
compared to values from a previous work. The results are
11
t
∫
q (t′)dt′
R
0
18
Xacc ) 1 - X (t) ) 1 -
(6)
th
∞
qR(t′)dt′
summarized in Table 4.
∫
0
An activation energy of 84.4 kJ‚mol- was calculated
from these rate constants and agrees well with the value
1
The comparison of the feed with thermal conversion for the
1
8
70 °C reaction is represented in Figure 8.
found in the previous work.
Since butanol is only present in stoichiometric default,
the accumulation is maximum (ca. 80%) at the end of the
addition and corresponds to an adiabatic temperature rise of
Conclusions
Calorimetry, in combination with thermal analysis, is an
essential tool for a data-based assessment of thermal risks
82 °C. This means that a temperature of MTSR ) 152 °C
could be reached in case of a cooling failure at this instant.
In these conditions, the reaction is by far not feed-controlled.
(
18) Ubrich, O.; Srinivasan, B.; Lerena, P.; Bonvin, D.; Stoessel, F. Optimal
Feed Profile for a Second-Order Reaction in a Semi-Batch Reactor under
Safety Constraints, Experimental Study. J. Loss PreV. Process Ind. 1999,
12, 485-493.
5.4. Kinetic Analysis of the Esterification. For single
reactions such as the present esterification, thermal conver-
920
•
Vol. 6, No. 6, 2002 / Organic Process Research & Development

linked with the performance of chemical reactions at
industrial scale. The energies of synthesis reactions or of
decomposition reaction, as well as the heat capacities of
reaction masses, can be measured by these techniques. The
performance of the Differential Reaction Calorimeter DRC
from SETARAM was demonstrated using two example
reactions.
The first and fastest reaction (hydrolysis of acetic
anhydride) allowed showing the calorimeter performance in
terms of tracking a fast reaction performed in the batch mode.
The values of reaction enthalpy are in good agreement with
values from the literature.
The second and slowest reaction could be studied in the
semi-batch mode as well as in the batch mode. With this
example, the performance of the calorimeter for the deter-
mination of the reaction enthalpy requiring a good caloric
sensitivity was shown. In the semi-batch mode, the ac-
cumulation of nonconverted butanol could be quantitatively
measured, allowing the establishment of the failure scenario
on which the safety study will be based.
Furthermore, since the heat-release rate of a reaction is
proportional to its rate, calorimetric methods give access to
kinetic data. The examples shown in this paper illustrate the
performance of the differential reaction calorimeter in the
determination of kinetic data.
The differential reaction calorimeter was found to be a
powerful screening tool in an organic synthesis laboratory
or in a development laboratory and is especially well suited
for a fast and low-cost determination of the thermal
parameters of chemical reactions, even when only a few raw
materials are available.
Acknowledgment
We thank Mr. H. R. R u¨ egg and Mr. M. Glor from the
Swiss Institute for the Promotion of Safety & Security and
Mr. J. Viry and Mr. P. Martin from SETARAM who made
this study possible.
Received for review July 30, 2002.
OP025575Q
Vol. 6, No. 6, 2002 / Organic Process Research & Development
•
921