You are on page 1of 30

Advanced Kinetics-based simulation method for

determination of the
thermal aging, thermal runaway TMRad and SADT
using DSC
Analysis Report (example)

Dr. Bertrand Roduit

AKTS AG Advanced Kinetics and Technology Solutions, http://www.akts.com


TECHNOArk 3, 3960 Siders, Switzerland, e-mail address: b.roduit@akts.com

June 11th 2008

Advanced Kinetics and Technology Solutions

AKTS AG Phone: +41 - 848 - 800221


TECHNOArk 3 Fax: +41 - 848 - 800222
3960 Siders URL: http://www.akts.com
Switzerland e-mail: info_contact@akts.com

We recommend SWISSI
as your test laboratory

http://www.bag.admin.ch http://www.eif.ch http://www.nitrochemie.com http://www.swissi.ch


http://www.armasuisse.ch
1. INTRODUCTION - PROCESS ANALYSIS

1.1 Application of thermokinetics for the determination of the behaviour of materials

In the present report, it is discussed how advanced numerical techniques can be


applied for the interpretation of DSC signals for the prediction of the thermal behaviour of
one energetic substance. If DSC monitors the evolution of the reactions, the signals can be
used not only for the qualitative and quantitative analysis but for the kinetic description as
well. The main challenge is the prediction of the thermal stability of the substance both in
extended temperature ranges and at the temperature conditions for which experimentation is
difficult or impossible. These difficulties are prevalent at low temperatures (requiring very
long investigation times), as well as under specific temperature fluctuations. The goal of this
advanced numerical approach is:
- a deeper insight into the reaction course of any material
- and an early detection of the stability/reactivity of any composition for fast screening

More generally, the main goal of AKTS-Thermokinetics Software Package [1] is to facilitate
kinetic analysis of any type of thermoanalytical data (DSC, DTA, TGA, TG-MS or TG-FTIR)
for the study of raw materials and products within the scope of research, development and
quality assurance. The technique provides a means to infer additional characteristics and
behaviour of examined substances based on conventional thermoanalytical measurements.
The method begins with the determination of the kinetic parameters for a given substance.
These parameters are then used to predict reaction progress under various temperature ranges
and conditions. By comparison, direct investigation of such reactions would be very difficult
at low temperatures (requiring very long times), as well as under complex temperature
profiles. Using AKTS-Thermokinetics Software, the rate and the progress of the reactions can
be predicted for the following temperature profiles: isothermal, non-isothermal, stepwise,
modulated temperature or periodic temperature variations, rapid temperature increase
(temperature shock), real atmospheric temperature profiles (up to 7000 climates) for low
temperature decomposed substances and even for adiabatic conditions.

1.2 Analysis Process

A full kinetic analysis of a solid state reaction has at least three major stages:
(1) experimental collection of data
(2) computation of kinetic parameters using the data from stage 1
(3) predictions of the reaction progress for required temperature profiles applying determined
kinetic parameters.

Using experimental DSC measurements performed on the examined samples (stage 1) AKTS-
Thermokinetics software determines the kinetic characteristics of the reaction (stage 2). The
calculated kinetic parameters are subsequently employed to predict the reaction progress of
the investigated samples under any given temperature mode (stage 3).

2. EXPERIMENTAL PART
2
2.1 Selection of the evaluation range and determination of the baseline (Stage 1)

Company … supplied AKTS AG thermoanalytical data of a material. For this


examination the DSC technique was used. The measured data were subsequently exported in
ASCII format for further thermokinetic interpretation with AKTS-Thermokinetics Software.
Temperature programs run from about 30°C to 400°C. Experiments were performed in high
pressure sealed crucibles at different heating rates ranging from 0.5 to 8 K/min with a sample
mass of about 3 mg. Figures 2.1 show the DSC signals at 0.5, 1, 2, 4 and 8 K/min of the
examined material used for the thermokinetic evaluation.

3
4
Figure 2.1
(Top: 5 first Figs) DSC curves for the reaction of the examined material recorded under different heating rates.
(Middle) Zoom of the DSC curve at 8 K/min. This curve denotes a self adiabatic effect shortly after 14 min of
measurement. This self-adiabatic effect is visible on the temperature and its derivative as well. It can sometimes
occur during DSC measurements of very strong autocatalytic reactions at higher heating rates. As the
temperature is not longer strictly controlled by the thermoanalyzer, such curves should not be used for the
determination of the thermokinetics.
(Bottom) Selection of the evaluation range (without the curve at 8 K/min).

The selection of the evaluation range should contain the signal before and after the
occurrence of the measured thermal event as depicted in figure 2.1 for the examined substance.

5
Generally the application of straight-line form for the baseline is incorrect [2]. The recorded
signal results not only from the heat of the reaction but is additionally affected by the change
of the specific heat of the mixture reactant-products during the progress of the reaction.

With:
⎡ mW ⎤
B(t) ⎢ ⎥ the baseline
⎣ mg ⎦
⎡ mW ⎤
and S(t) ⎢ ⎥ the differential signal,
⎣ mg ⎦

dα ⎡ 1 ⎤
the reaction rate =
dt ⎢⎣ s ⎥⎦
and the reaction progress α(t) [−] can be expressed as

⎡ ⎛ mW mW ⎞ ⎤ t
⎡ ⎛ mW mW ⎞ ⎤
dα (S(t) − B(t))
⎢⎜ ⎜
⎝ mg

mg




⎡ 1 ⎤
∫ (S(t) − B(t))dt ⎢ ⎜⎜
⎝ mg

mg
⎟⎟ s ⎥
⎠ ⎥ = [−]
= tend =⎢ ⎥=
⎢ ⎥ α (t) = tend
to
=⎢
dt ⎢ ⎛ mW mW ⎞ ⎥ ⎣s⎦ ⎢ ⎛ mW mW ⎞ ⎥

to
(S(t) − B(t))dt ⎢ ⎜

⎣⎢ ⎝ mg
− ⎟ s
mg ⎟⎠ ⎦⎥
⎥ ∫
to
(S(t) − B(t))dt ⎢ ⎜⎜
⎣⎢ ⎝ mg
− ⎟s ⎥
mg ⎟⎠ ⎥⎦

with

(0 < α(t) < 1) and B(t) = (1- α(t))*(a1+b1*t) + α(t)*(a2+b2*t)


⎡ mW ⎤ ⎡ mW mW mW mW ⎤
⎢ mg ⎥ = ⎢(−) * ( mg + mgs * s) + (−) * ( mg + mgs * s)⎥
⎣ ⎦ ⎣ ⎦

where

(a1+b1*t) : tangent at the beginning of the signal S(t).


(a2+b2*t) : tangent at the end of the signal S(t).

The tangential area-proportional baseline is the most universal type because of its correction
possibilities. It is created at α(t) => 0 and at α(t) => 1 by the appropriate tangents at the
beginning or the end of the measured DSC signal. It allows compensation of not only changes
in the size of cp of the reactant and product, but also of changes in their temperature
dependency. This type of baselines can be described by the following equation:

B(t) = (1- α(t))*(a1+b1*t) + α(t)*(a2+b2*t)


with
(a1+b1*t) : tangent at the beginning of the signal S(t).
(a2+b2*t) : tangent at the end of the signal S(t).

B(t) can be calculated iteratively. The convergence is achieved as soon as the relative average
deviations between two iterations are smaller than an arbitrarily chosen value (for example
1e-6). Presented in Fig.2.2 an area-proportional baseline has been calculated using arbitrarily
chosen 300 iteration loops.

6
Figure 2.2 DSC curve of the examined substance and the baseline calculation illustrated for a heating rate of 4
K/min.

It is obvious that the baseline determination can significantly influence the determination of
the kinetic parameters of the reaction. Moreover, the correct baseline determination should be
intimately combined with the computation of the kinetic parameters for the investigated
reaction. Advanced mathematical procedures are therefore necessary for an objective
calculation of the most appropriate baseline for each DSC signal.

Figure 2.3 DSC curve of the examined material at 4 K/min after baseline subtraction and normalization.

7
3. DETERMINATION OF THE KINETIC CHARACTERISTICS (STAGE 2)

3.1 Single kinetic triplet

The noticeable weakness of the ‘single curve’ methods (determination of the kinetic
parameters from single run recorded with one heating rate only) has led to the introduction of
the ‘multi curve’ methods over the past few years, as discussed in the International ICTAC
Kinetics project [3-6]. Only series of non-isothermal measurements carried out at different
heating rates can give a data set which generally contains the necessary amount of
information required for full identification of the complexity of a process [7-8].

If the decomposition follows a single mechanism then the reaction can be described in terms
of a single pair of Arrhenius parameters and the commonly used set of reaction models. In
such cases the dependence of the logarithm of the reaction rate over 1/T is linear with the
same slope of m = E/R for all conversion degrees αi. The reaction rate can be described by
only one value of the activation energy E and one value of the pre-exponential factor A by the
following expression:

dα ⎛ E ⎞
= A exp⎜⎜ − ⎟⎟ f( α )
dt ⎝ RT(t) ⎠

However, decomposition reactions are often too complex to be described in terms of a single
pair of Arrhenius parameters and the commonly applied set of reaction models (table 3.1).

Table 3.1 The forms of the f(α) function dependent on the reaction model.

Autocatalytic : (1-α)^n α^m A1.5 : 1.5 (1-α) [-ln(1-α)]^(1/3)


A2 : 2 (1-α) [-ln(1-α)]^(1/2)
F1 : 1-α An : n (1-α) [-ln(1-α)]^(1-1/n)
F2 : (1-α)^2
F3 : (1-α)^3 R2 : 2 (1-α)^(1/2)
Fn : (1-α)^n R3 : 3 (1-α)^(2/3)
Rn : n (1-α)^(1-1/n)
P1 : α^0
P2 : 2 α^(1/2) D1 : 1/(2α)
P3 : 3 α^(2/3) D2 : [-ln(1-α)]^-1
P4 : 4 α^(3/4) D3 : 1.5 [1-(1-α)^(1/3)]^-1 (1-α)^(2/3)
Pn : n α^(1-1/n) D4 : 1.5 [(1-α)^(-1/3)-1]^-1

As a general rule, decomposition reactions demonstrate profoundly multi-step characteristics.


They can involve several processes with different activation energies and mechanisms. In
such situation the reaction rate can be described only by complex equations, where the
activation energy term is no more constant but is dependent on the reaction progress α
(E ≠ const but E=E(α), see Fig.3.2). Thus a simplified kinetic analysis can no more lead to an
accurate description of the experimental data. For multistage overlapped reactions the
prediction of the thermal behavior under any new temperature profile, without taking into
account the dependence of the activation energies E(α) on the conversion degree α, is of little
value.

8
3.2 Differential isoconversional concept (basics)

Differential isoconversional concept (basics)

The kinetic parameters can be evaluated by the isoconversional method. This is a


numerical method which involves determination of temperatures corresponding to a certain,
arbitrarily chosen values of the conversion extent α recorded in the experiments carried out at
e.g. different heating rates β.

In fact, the thermoanalytical data set usually contains:

- the relationship between specific conversion, αi, and temperatures for different heating rates
(non-isothermal mode).

- the relationship between specific conversion, αi, and time for different temperatures
(isothermal mode).

Commonly applied are the following three isoconversional methods known as: Friedman [9],
Ozawa-Flynn-Wall [10-11] and the ASTM E698 analysis [12].

Friedman Analysis:

The Friedman method is a so-called differential isoconversional method. Based on the


Arrhenius equation
dα ⎛ E ⎞
= Aexp⎜⎜ − ⎟⎟ f (α )
dt ⎝ RT(t) ⎠
with
f(α) : the model function
A: the preexponential factor
E: the activation energy
T : the temperature
t: the time

Friedman proposed to apply the logarithm of the conversion rate dα/dt as a function of the
reciprocal temperature at any conversion α:
dα ⎛ E (α ) ⎞ dα E (α )
= A( α )exp⎜⎜ − ⎟⎟ f (α ) ln( ) = ln( A(α )) − + ln( f (α ))
dt ⎝ RT(t) ⎠ dt RT (t )
As f(α) is a constant in the last term of at any fixed α, the logarithm of the conversion rate
dα/dt over 1/T shows a straight line with the slope m = E/R.

By extension
dα E (α )
ln( ) = ln( A' (α )) −
dt RT (t )
with A' (α ) = A(α ) f (α )
So, having determined A' (α ) and E (α ) we can predict the reaction rate or reaction progress
using the following expression:
dα ⎛ E (α ) ⎞
= A' ( α )exp⎜⎜ − ⎟⎟
dt ⎝ RT(t) ⎠

9
at any temperature profile:
• isothermal, non-isothermal, stepwise
• modulated temperature or periodic temperature variations
• rapid temperature increase (temperature shock)
• real atmospheric temperature profiles for investigating properties of e.g. low-
temperature decomposed substances under different climates (yearly temperature
profiles with daily minimal and maximal fluctuations. 50 climates available in the
default version).
• STANAG 2895 temperature profile: Zones A1, A2, A3, B1, B2, B3, C0, C1, C2, C3,
C4, M1, M2, M3.
• customized temperature profiles
• possibility to compare the reaction progress of substances at any temperature profile
• confidence interval of the prediction

The extended features of AKTS-Software applied to High Sensitivity Isothermal Heat Flow
Calorimetry enable
• the calculations of the thermokinetics from long term isothermal Heat Flow
Calorimetry data for very precise lifetime prediction on the first percent of degradation
(quality control)

Concerning the Friedman analysis, additional information and applications examples are
available at the following link: http://www.akts.com/faq/info.html

Ozawa-Flynn-Wall Analysis:

The Ozawa-Flynn-Wall analysis is a so-called integral isoconversional method. The


activation energies obtained as a function of the reaction progress are therefore less precise
than with the differential isoconversional method of Friedman.
Independent on each other Ozawa, Flynn and Wall developed a method for the determination
of the activation energy, which is based on several curves measured at different, but constant
heating rates into the analysis. Starting from the Arrhenius equation:
dα ⎛ E ⎞
= Aexp⎜⎜ − ⎟⎟ f (α )
dt ⎝ RT(t) ⎠
with constant heating rate β
dT/dt = β = constant

the integration form leads to:


α T (α )
dα A ⎛ E ⎞
G(α ) = ∫ = ∫ exp⎜⎜ − ⎟⎟dt
α = 0 f( α ) β To ⎝ RT(t) ⎠
with
f(α) : the model function, A: the preexponential factor, E: the activation energy, β : the
heating rate, T : the temperature and t: the time.

If ‘To’ lies below the temperature at which the reaction is noticeable, then one can set the
lower limit of integration to To = 0, so that the following equation is obtained after integration
and using a logarithm expression form:
AE
lnG(α ) = ln( ) − ln(β ) + ln( p( z ))
R

10
z
exp(− z ) exp(− z ) E
with : p( z ) = −∫ dz and z =
z −∞
z RT

By using the approximation given by DOYLE [C. D. Doyle: J. Appl. Anal., 27 (1962) 639]
ln p(z) = -5.3305 + 1.052·z and transposing, ones obtains:

AE E
ln(β ) = ln( ) − lnG(α ) − 5.3305 + 1.052
R RT

It can be seen from the above equation that during a series of measurements with the heating
rates β at a fixed degree of conversion α the graph ln(β) versus 1/T shows straight lines with a
slope m = - 1.052·E/R (see figure below). The temperatures T are those at which the
conversion α is reached at a heating rate β. As a result, the slope of the straight lines enables
to calculate the activation energy as a function of the reaction progress α.

ASTM E698:

The analysis according to ASTM E698 is based on the assumption that the maximum
(for example maximum of the DSC or DTG curve) of a single step reaction is reached at the
same conversion degree independent on the heating rate. Although this assumption is only
partly right, the resulting errors are low [ J.P.Elder: J. Thermal Anal., 30 (1985) 657 ]. In this
method, the logarithm of the heating rate is plotted over the reciprocal temperature of the
maximum. The slope of the yielded straight line is proportional to the activation energy, just
as in the Ozawa-Flynn-Wall analysis.
General information about the isoconversional methods

A detailed analysis of the various isoconversional methods (i.e. the isonconversional


differential and integral methods) for the determination of the activation energy has been
reported in the literature by Budrugeac [13]. The convergence of the activation energy values
obtained by means of a differential method like Friedman method [9] with those resulted from
using integral methods with integration over small ranges of reaction progress α comes from
the fundamentals of the differential and integral calculus. In other words, it can be
mathematically demonstrated that the use of isoconversional integral methods (for example:
Ozawa-Flynn-Wall [10-11]) can yield systematic errors when determining the activation
energies. These errors depend directly on the size of the small ranges of reaction progress ∆α
over which the integration is performed. These errors can be avoided by using infinitesimal
ranges of reaction progress ∆α. As a result, isoconversional integral methods turn back to the
differential isoconversional methods formerly proposed by Friedman [9].

The differential methods for the calculation of the kinetic parameters are based on the use of
the following reaction rate equation:

dα ⎛ E( α ) ⎞
β = A( α ) exp ⎜ − ⎟ f( α )
dT ⎝ RT ⎠

where β is the heating rate, T the temperature, E(α) the activation energy, A(α) the
preexponential factor and f(α) is the differential conversion function.

11
As far as isoconversional integral methods are concerned, these techniques are based on the
equation:

α
dα A( α ) T ⎛ E( α ) ⎞
g (α ) = ∫ = ∫ exp ⎜ − ⎟dT where g(α) is the integral conversion function.
0
f( α ) β Tα = 0 ⎝ RT ⎠

The isoconversional integral methods with the integration over low ranges of the degree of
conversion and respectively temperature, are based on the equation:

α T
dα A( α ) α ⎛ E( α ) ⎞
g (α − ∆α , α ) = ∫
α − ∆α f( α )
= ∫
β Tα - ∆α
exp ⎜ −
⎝ RT ⎠
⎟dT

which is derived by supposing that in the range of the variation of the conversion degree ∆α,
the activation energy E can be assumed constant. Obviously, the use of such an approach
leads to a plot of E versus the degree of conversion α. However, the activation energy as a
function of the conversion progress looks like a stair function in which the low ranges of ∆α
where E keeps a constant value are clearly marked. The number of stairs depends directly on
the size of ∆α.

In order to evaluate the integrals from the previous equation, one can use the theorem of the
average value, we obtain:

1 ⎛ E( α ) ⎞
A( α )
∆α = exp ⎜ − ⎟∆T
f( αζ ) β ⎜ RT ⎟
⎝ ζ ⎠
where (α − ∆α ) < α ζ < x, (Tα − ∆α − Tα ) < Tζ < Tα and ∆T = Tα − Tα −∆α

Since the number of stairs (where the activation energy E is assumed constant in the
isoconversional integral methods) depends directly on the range of chosen ∆α, then an
unlimited number of stairs can be reached by making ∆α infinitesimal. For ∆α => 0, we have
Tξ => T and f(αξ) => f(α). As a consequence, the previous equation turn back into its
differential form:

dα ⎛ E( α ) ⎞
β = A( α ) exp ⎜ − ⎟ f( α )
dT ⎝ RT ⎠

that grounds the isoconversional differential methods which corresponds to the Friedman
approach. More generally, the conversion rate expression can be adapted to an arbitrary
variation of temperature (as well as to isothermal conditions) by replacing β(dα/dT) with
dα/dt. Friedman analysis, based on the Arrhenius equation, applies the logarithm of the
conversion rate dα/dt as a function of the reciprocal temperature at different degrees of the
conversion α.

dα Ei
ln = ln(Ai f( α i, j )) −
dt αi RTi, j

with i: index of conversion, j: index of the curve and f(αi,j) the function dependent on the
reaction model that is constant for a given reaction progress αi,j for all curves j. As f(α) is

12
constant at each conversion degree αi, the method is so-called ‘isoconversional’ and the
dependence of the logarithm of the reaction rate over 1/T is linear with the slope of m = E/R
and the intercept A as presented in figures 3.1 and 3.2.

Figure 3.1. Friedman analysis of the examined material

Figure 3.2. Activation energy and pre-exponential factor as a function of the reaction progress for
decomposition of the examined material.
The reaction rate can now be described as following:
dα ⎛ E( α ) ⎞ dα ⎛ E( α ) ⎞
= A( α )f( α ) exp ⎜⎜ − ⎟⎟ or = A' ( α ) exp ⎜⎜ − ⎟⎟
dt ⎝ RT(t) ⎠ dt ⎝ RT(t) ⎠

13
Figure 3.3. Reaction rates (normalized DSC-signals after correctly calculated baselines and kinetics) for the
decomposition of the examined material. Experimental data are represented as symbols, solid lines represent the
calculated signals. The values of the heating rates are marked on the curves.

3.3 Short note on baseline optimization

The accurate determination of the kinetic parameters under experimental conditions


applied which enables the correct fit of the experimental data is a prerequisite for prediction
of the reaction progress under any new temperature profile. When solving the complicated
interrelation between the baseline, the kinetic parameters of the reaction and reaction progress,
two important points have to be considered:
(1) The reaction rate should be of Arrhenius type.
(2) When measuring the progress of a reaction one tries to eliminate the systematic errors, so
that only accidental errors have to be taken into account. In that case the measured values will
spread around the average value for each heating rate, in a form of the Gaussian-type curve.
The Gaussian distribution results from a summation of several events e.g. overlapping
reactions, noise, drift, artefact, uncertainties in the baseline construction, etc. During the
optimization, the true information has to be extracted by fulfilling the criterion ‘1’ and the
whole optimization becomes more and more stable. Once the stability is reached, the
optimization is finished and the reliability of the predictions depends on how big was the sum
of all possible sources of errors. Therefore, under consideration of all heating rates the
mathematical approach has to determine for each heating rate the 'best value' or 'central
tendency' of the signal, for which the chance of the good reproducibility on subsequent
measurements is maximal.
Fulfilling both above conditions (I: Arrhenius type reaction rate and II: Gaussian-type
distributed errors) makes possible the iterative calculation and objective determination of the
correct baseline for each signal measured under different heating rates. This objective
determination is carried out by the iterative calculation of all tangent parameters for each
heating rate: ai,β, bi,β, aj,β, bj,β for each heating rate β with

14
i = indices of the slope and intercept of the tangent at the beginning of the signal S(T) with the
heating rate 'β'.
j= indices of the slope and intercept of the tangent at the end of the signal S(T) with the
heating rate 'β '.
The baselines are no more arbitrarily chosen by the users but objectively optimized taking
into account:
- statistics, for the consideration of the experimental noise and shape of the signals
- the kinetic parameters, for the consideration of reaction rates following Arrhenius
relationship.

15
4. KINETICS AND MILLIGRAM SCALE - THERMAL STABILITY PREDICTIONS

4.1 Prediction of the reaction progress under any temperature mode (Stage 3)

The DSC data collected during non-isothermal reaction of the examined materials with
different heating rates were used for the determination of the kinetic parameters used later for
the prediction of the reaction extent. More generally, the kinetic parameters calculated from
the non-isothermal experiments make possible the prediction of the reaction progress for any
other heating rate and more generally for any temperature mode [14-18] such as:
• isothermal
• non-isothermal
• stepwise
• modulated temperature or periodic temperature variations
• rapid temperature increase (temperature shock)
• real atmospheric temperature profiles for investigating properties of e.g. low-
temperature decomposed substances under different climates (yearly temperature
profiles with daily minimal and maximal fluctuations. 50 climates available in the
default version).
• NATO norm STANAG 2895 temperature profile: Zones A1, A2, A3, B1, B2, B3, C0,
C1, C2, C3, C4, M1, M2, M3.
• customized temperature profiles
• possibility to compare the reaction progress of substances at any temperature profile
• combination of mass loss TG, heat flow signal e.g. DSC/DTA and MS data in multi-
projects for simultaneous comparison of the mass loss, heat flow and volatiles species
evolution at any temperature profile
• confidence interval of the prediction
• viewing data in form of overall conversion alpha(T(t)) and conversion rate
dalpha(T(t))/dt
• viewing data in natural form Q(T(t)), dQ(T(t))/dt, P(T(t)), dP(T(t))/dt, M(T(t)),
dM(T(t))/dt
• integration according Runge-Kutta, important for stiff systems of differential
equations
• extended features for High Sensitivity Isothermal Heat Flow Calorimetry (HFC):
Ability to calculate the thermokinetics from long term isothermal HFC data for very
precise lifetime prediction on the first percent of degradation (quality control)

16
4.2 Isothermal, modulated, stepwise and real atmospheric temperature modes

The kinetic parameters calculated from the non-isothermal experiments allow


prediction of the reaction progress at any temperature mode: isothermal, non-isothermal and
intermediate intervals in the heating rate, expressed, e.g. in oscillatory temperature modes.
The prediction of the reaction progress in various temperature modes is for example given
below.

Figure 4.1 Top: Calculated reaction progress (normalized signals) of the decomposition of the examined
material as a function of time under isothermal conditions. The values of the temperature in °C are marked on
the curves. Bottom: Calculated reaction rate at 80°C. The nature of the curve reflects an ‘autocatalytic reaction
type’.

17
Figure 4.2 Calculated reaction progress (normalized signals) of the decomposition of the examined substance as
a function of time under isothermal (30°C) and oscillatory (30°C±10°C, 30°C±20°C and 30°C±30°C, 24 h
period) temperature conditions. Presented figure clearly illustrates the influence of the oscillatory temperature
mode on the reaction progress.

18
Figure 4.3 Stepwise mode. Reaction progress α (DSC, normalized signals) of the examined substance as a
function of the time for combined isothermal and non-isothermal temperature modes (stepwise mode).

19
Figure 4.4
Top: Average daily minimal and maximal temperatures recorded for each day of the year between 1961 and
1990 (Tokyo and Hong Kong).
Bottom: Reaction progress α (DSC, normalized signals) of the examined substance as a function of time for the
Tokyo and Hong Kong temperature profiles.

20
Figure 4.5 (Top) Selected shock temperature modes. Prediction of the reaction progress α (Bottom) during rapid
temperature increase.
Starting temperature before thermal shock: 20 °C
Function used to simulate the shock temperature mode (Gaussian with asymmetry):
f(T) = A*exp(-ln(2)*(ln(1+2*D*(T-B)/C)/D)^2)
With T the temperature and
Amplitude ‘A’: from 90°C up to 100 °C
Position ‘B’: 1200 s
Halfwidth ‘C’: 1200 s
Asymmetry ‘D’: 1.1 -

21
5. KINETICS AND SCALE UP - SAFETY ANALYSIS
Calculation of adiabatic thermal transformation and heat accumulation from non-
isothermal DSC measurements

5.1 Introduction

The precise prediction of reaction progresses in adiabatic conditions is necessary for


the safety analysis of many technological processes [14-26]. Calculations of an adiabatic
temperature-time curve for the reaction progress can also be used to determine the decrease of
the thermal stability of materials during storage at temperatures near the threshold
temperature for triggering the reaction. Due to insufficient thermal convection and limited
thermal conductivity, a progressive temperature increase in the sample can easily take place,
resulting in an explosion.

Several methods have been presented for predicting the reaction progress of exothermic
reactions under heat accumulation conditions [19-25]. However, because decomposition
reactions usually have a multi-step nature, the accurate determination of the kinetic
characteristics strongly influences the ability to correctly describe the progress of the reaction.
The use of simplified kinetic models for the assessment of runaway reactions can, on one
hand, lead to economic drawbacks, since they result in exaggerated safety margins. On
another hand, it can cause dangerous situations when the heat accumulation is underestimated.
For adiabatic self-heating reactions, incorrect kinetic description of the process is usually the
main source of prediction errors.

5.2 The kinetic based approach for the determination of the Time to Maximum Rate
under adiabatic conditions (TMRad)

5.2.1 Kinetic based approach - Concept:

The kinetic parameters calculated from DSC measurements are used for the
description of the thermal behavior of larger amount of substance because the kinetics of any
reaction is the same for ten milligrams of substance and for one ton. However, during up-
scaling two important factors have to be considered:
(i) the application of advanced kinetics, which properly describes the complicated, multistage
course of the decomposition process,
(ii) the effect of heat balance in the energetic system, as the sample mass is increased by a few
order of magnitude compared to the thermoanalytical experiments.

Because decomposition reactions usually have a multi-step nature, the isoconversional


analysis enables a more accurate determination of the kinetic characteristics compared to
simplified kinetic based approach that are essentially based on simplified kinetics assumption
like ‘let’s assume that the reaction is zeroth order etc…’

5.2.2 Kinetic based approach - Determination of the TMRad and adiabatic calorimeter
behaviour

There are several issues that need to be examined for the examination of the adiabatic
conditions or reconstructing an ARC experiment starting from the kinetic based approach.
Let’s consider below different aspects.

22
It is possible to determine the behaviour of an adiabatic calorimeter coming from the kinetic
based approach. For this purpose, one has to take into account the thermal inertia (or phi
factor) to consider the effect of the vessel’s inertia i.e. the heat lost into the bomb or into
possible solvent when working with diluted solution.

Heat balance over the sample inside the bomb:


dT dα dT
UA(Tc − Ts ) = M s C p , s s + M s ∆H r + M x C p, x x
dt dt dt
with U: heat transfer coefficient, CP: specific heat, T: temperature, M: mass, indices c:
container (or bomb in the ARC experiment), indices s: sample, indices x: solvent,

Heat balance over the container (bomb):


dT
UA(Ts − Tc ) = M c C p,c c
dt

The combination of the above equations under consideration of


dT dTc dTs dTx
T (t ) = Tc (t ) = Ts (t ) = Tx (t ) => = = =
dt dt dt dt
for the achievement of the adiabatic conditions leads to the following expression:
dT dα
( M c C p ,c + M s C p , s + M x C p , x ) = M s ∆H r
dt dt
that gives after reformulation:
dT M s C p,s ∆H r dα
=
dt M c C p ,c + M s C p , s + M x C p , x C p, s dt

and can be rewritten as


dT 1 dα
= ∆Tad ,real
dt φ dt

with:
∆H r
- the adiabatic temperature rise: ∆Tad ,real =
C p,s
M c C p ,c + M s C p , s + M x C p , x
- the phi factor φ =
M s C p,s

- the kinetic expression :
dt
In case of isoconversional analysis we have :
dα ⎛ E (α ) ⎞
= A' ( α )exp⎜⎜ − ⎟⎟ with A' (α ) = A(α ) f (α )
dt ⎝ RT(t) ⎠

Note, for comparison, in case of simplified zeroth order kinetic assumption we would have :
dα ⎛ E ⎞
= A exp⎜⎜ − ⎟⎟
dt ⎝ RT(t) ⎠

23
One can now use the kinetic based approach for prediction of the reaction progress α(t) and
rate dα/dt as well as the development of the temperatures T(t) and dT/dt and adiabatic
induction times at any selected starting temperatures.

It comes from the above equations that the φ factor influences the simulation of an ARC
experiment. The φ factor influences:
1
- the ∆Tad ,measured because it comes from above description that ∆Tad ,measured = ∆Tad ,real
φ
- and the TMRad in different level depending on the type of decomposition kinetics.

Usually, the standard method of correcting the time for the reaction under adiabatic conditions
TMRad ,measured
is TMRad ,real = . This is however an approximation.
φ

Note:
To solve this problem we can also combine the advanced kinetic description of the process
with a heat balance as following:

⎧ dα
⎪ dt = reaction rate expression
⎨ dT dα
⎪ M Rcp = V (−∆H R )Co + UA(TC − T )
⎩ dt dt
After reformulation we can write
⎧ dα
⎪⎪ dt = reaction rate expression
⇒ ⎨ dT UA dα
⎪ = (TC − T ) + ∆Tad
⎪⎩ dt M R c p dt

Since we are dealing with adiabatic conditions, we just need to set the heat transfer coefficient
U=0 and solve both above presented differential equations which leads basically to the same
solution as before
dT 1 dα
= ∆Tad ,real with φ = 1.
dt φ dt

The solution of the problem leads to a thermal stability diagram, i.e. for each starting
temperature the Time to Maximum Rate under Adiabatic conditions (TMRad) can now be
easily obtained.

24
5.3 Construction of a thermal safety diagram: runaway time as a function of process
temperature under adiabatic conditions (TMRad = f(T)) for the examined substance

The decomposition of the examined substance follows an exothermal process. Using


the reaction heat (∆Hr = 606.1±20.73 J/g) and the heat capacity (CP = 1.2 J/g/K as indicated
by Company …), one can calculate the reaction progress due to self-heating for ∆Tad (with
∆Tad = ∆Hr/CP/Φ). One can now vary the heat transfer parameter U or φ for reproducing:
> strictly adiabatic conditions (U=0 or Φ=1)
> non-adiabatic conditions (U>0 or Φ>1 in the ARC experiments)
> isothermal conditions (∆Tad ->0°C and U=0). Isothermal conditions can be retrieved by
setting an exceptionally low value for ∆Tad. If ∆Tad ->0°C and U=0 then
dT
T (α = 0) ≅ T (α = 1) ≅ Tisothermal or ≅0
dt
For a reaction rate expression with differential isoconversional methods we would obtain:
⎧ dα ⎛ − E (α ) ⎞
⎪⎪ ≅ A ' (α ) exp⎜⎜ ⎟⎟
dt ⎝ RTisothermal ⎠

⎪ dT ≅ 0
⎪⎩ dt
As long as thermal safety analysis is concerned, the adiabatic induction time is defined as the
time which is needed for self-heating from the start temperature to the time of maximum rate
(TMRad) under adiabatic conditions. Depending on the decomposition kinetics and ∆Tad, the
choice of the starting temperatures strongly influences the time to runaway and the rate of the
temperature evolution under adiabatic conditions. Figure 5.1 presents the starting temperature
and corresponding adiabatic induction time TMRad relationship. The confidence interval was
determined for 95% probability.

Figure 5.1 Thermal safety diagram: Starting temperature and corresponding adiabatic induction time TMRad
relationship of the examined material. The choice of the starting temperatures strongly influences the adiabatic
induction time. ∆Hr =606.1±20.73J/g and ∆Tad=∆Hr/CP/Φ=505±17.2°C for Φ =1 and CP = 1.2 J/g/°C.
(Confidence interval: 95%).

25
∆Hr =606.1 ± 20.73 J/g
CP = 1.2 J/g/°C
∆Tad=∆Hr/Cp=505 ± 17.2°C
TMRad TMRad
Temperature Lower Mean
[°C] Limit Value Confidence interval
40 43.16 hours 52.02 hours +- 17.03 %
41 38.8 hours 46.73 hours +- 16.98 %
42 34.9 hours 42.01 hours +- 16.93 %
43 31.42 hours 37.8 hours +- 16.88 %
44 28.3 hours 34.03 hours +- 16.82 %
45 25.52 hours 30.66 hours +- 16.77 %
(*) 46 23.02 hours 27.64 hours +- 16.72 %
(*) 47 20.78 hours (*) 24.94 hours +- 16.67 %
(*) Means that if Φ=1 and CP=1.2 J/g/°C the determined TMRad is about 24 hours for a
temperature of 47°C (a more conservative value of 46°C is given by the lower limit of the
confidence interval)
48 18.78 hours 22.52 hours +- 16.62 %
49 16.98 hours 20.35 hours +- 16.57 %
50 15.36 hours 18.4 hours +- 16.52 %
51 13.9 hours 16.65 hours +- 16.47 %
52 12.6 hours 15.07 hours +- 16.42 %
53 11.42 hours 13.65 hours +- 16.37 %
54 10.36 hours 12.38 hours +- 16.32 %
55 9.4 hours 11.23 hours +- 16.27 %
56 8.54 hours 10.2 hours +- 16.22 %
(**) 57 7.76 hours 9.26 hours +- 16.17 %
58 7.06 hours 8.42 hours +- 16.12 %
(**) 59 6.43 hours 7.66 hours +- 16.07 %
(**) Means that if Φ=1 and CP=1.2 J/g/°C the determined TMRad is about 8 hours for a
temperature of 59°C (a more conservative value of 57°C is given by the lower limit of the
confidence interval)
60 5.85 hours 6.97 hours +- 16.02 %
61 5.33 hours 6.35 hours +- 15.97 %
62 292.12 min 5.79 hours +- 15.92 %
63 266.55 min 5.28 hours +- 15.87 %
64 243.38 min 289.12 min +- 15.83 %
65 222.36 min 264 min +- 15.78 %
66 203.28 min 241.2 min +- 15.73 %
67 185.94 min 220.51 min +- 15.68 %
68 170.19 min 201.72 min +- 15.63 %
69 155.87 min 184.64 min +- 15.59 %
70 142.83 min 169.11 min +- 15.54 %
71 130.96 min 154.97 min +- 15.5 %
72 120.15 min 142.1 min +- 15.45 %
73 110.29 min 130.37 min +- 15.41 %
74 101.31 min 119.68 min +- 15.36 %
75 93.1 min 109.93 min +- 15.31 %
76 85.61 min 101.03 min +- 15.27 %
77 78.77 min 92.91 min +- 15.22 %
78 72.51 min 85.48 min +- 15.18 %
79 66.79 min 78.7 min +- 15.13 %
Table 5.1 Starting temperatures and corresponding TMRad for the examined material under adiabatic heat
accumulation conditions. ∆Hr =606.1±20.73J/g and ∆Tad=∆Hr/CP/Φ=505±17.2°C for Φ =1 and CP = 1.2 J/g/°C.

26
Figure 5.2 illustrates the applications of the above equations for the simulation of the thermal
behaviour under adiabatic conditions. The simulated T-time relationships are presented with a
starting temperature of 47°C for ∆Tad=∆Hr/CP/Φ = 505±17.2°C for Φ =1 and CP = 1.2 J/g/°C.

Figure 5.2 Adiabatic runaway curves for the substance (isochoric conditions) showing the confidence interval
for the prediction: Tbegin=47°C, ∆Hr =606.1±20.73J/g and ∆Tad=∆Hr/CP/Φ=505±17.2°C for Φ =1 and CP = 1.2
J/g/°C. The confidence interval was determined for 95% probability.

The self heat rate curves under adiabatic conditions can be calculated similarly for different
starting temperatures as presented in figures 5.3.

Figure 5.3 Self heat rate curves of the examined materials under adiabatic conditions. The different starting
temperatures are displayed in the legends. ∆Hr =606.1±20.73J/g and ∆Tad=∆Hr/CP/Φ=505±17.2°C for Φ =1 and
CP = 1.2 J/g/°C.

27
Similar calculations can be performed for different φ factors (see figures 5.4) with a starting
temperature of 47°C corresponding to a TMRad of 24 h under strict adiabatic conditions
(corresponding to φ =1 and a CP = 1.2 J/g/°C).

Figure 5.4
Thermal behaviour of the examined material under adiabatic conditions for different φ factors.
Adiabatic runaway curves (top) and self heat rate curves (bottom)
Tbegin=47°C, ∆Hr =606.1±20.73J/g, CP = 1.2 J/g/°C and
∆Tad=∆Hr/cp/Φ=505±17.2°C for φ = 1 (strict adiabatic conditions)
∆Tad=∆Hr/cp/Φ=336±11.5°C for φ = 1.5 (peudo- adiabatic conditions)
∆Tad=∆Hr/cp/Φ=168±5.7 °C for φ = 3 (peudo- adiabatic conditions)
∆Tad=∆Hr/cp/Φ=84±2.8 °C for φ = 6 (peudo- adiabatic conditions)

28
6. SELF-ACCELERATING DECOMPOSITION TEMPERATURE (SADT)

The Self-Accelerating Decomposition Temperature (SADT) is an important parameter


that characterizes thermal hazard under transport conditions of self-reactive substances. The
SADT has been introduced into the international practice by the regulations of the United
Nations presented in “Recommendations on the Transport of Dangerous Goods, Manual of
Tests and Criteria” (TDG) [27]. The Globally Harmonized System (GHS) [28] has inherited
the SADT as a classification criterion for self-reactive substances. According to the
Recommendations on TDG the SADT is defined as “the lowest temperature at which self-
accelerating decomposition may occur with a substance in the packaging as used in transport”.
Important feature of the SADT is that it is not an intrinsic property of a substance but “…a
measure of the combined effect of the ambient temperature, decomposition kinetics,
packaging size and the heat transfer properties of the substance and its packaging” [27].

As the SADT determination was not requested for the examined substance, it has not
been introduced into the present report. But the manuscript
http://www.akts.ch/faq/self-accelerating-decomposition-temperature-sadt.pdf
presents the approach for determination of the SADT by Advanced Kinetic Elaboration
of DSC Data.

7. CONCLUSIONS

The present study aimed at the thermokinetic analysis of DSC data supplied by
Company …. The study focuses on prediction of the reactivity of a substance both in
extended temperature ranges and under temperature conditions for which experimental data
collection is difficult. Adequate predictive examination of the investigated reactions requires
about four DSC measurements carried out with different heating rates, generally in the range
of 0.5 to 8 K/min. More generally, applying the results obtained by Differential Scanning
Calorimetry (DSC) advanced numerical techniques enable prediction of the reactivity of
various in broad temperature range. In fact, numerical simulations can be used to replace
experiments in situations, which are not directly accessible to the experiment for timing
reasons. The examples of such modeling analysis can be helpful for guiding screening and
development activities of candidate materials. If modeling proceeds in parallel with
experimental studies, then it should result in lower costs in the development phase of a project.
The proposed method has therefore several advantages:
• It is fast: Reliable screening of candidate energetic materials by revelation of potentially
unstable mixtures at prolonged times and temperature exposures can be done within few hours
• It is economical: The determination of the kinetics offers very significant time/expertise
savings compared to real time analysis which can extend over prolonged periods, even
months.
• It is convenient: DSC devices are widely available and require only a small amount of
material.
• It is versatile: With one set of measurements different scenarios can be calculated. The
method can be used to predict the rate of the reaction progress for any temperature profile.

‘Safety through design not by Accidents’


For more information download other studies from the following link:
http://www.akts.com/faq/info.html

29
REFERENCES

[1] Advanced Kinetics and Technology Solutions: http://www.akts.com (AKTS-Thermokinetics software and
AKTS-Thermal Safety software)
[2] Hemminger W. F., Sarge, S. M., J. Therm. Anal., 37 (1991), 1455.
[3] M.E. Brown et al. Computational aspects of kinetic analysis. The ICTAC Kinetics project data, methods and
results. Thermochim. Acta, 355 (2000) 125.
[4] M. Maciejewski, Computational aspects of kinetic analysis. The ICTAC Kinetics Project - The
decomposition kinetics of calcium carbonate revisited, or some tips on survival in the kinetic minefield.
Thermochim. Acta, 355 (2000) 145.
[5] A. Burnham, Computational aspects of kinetic analysis. The ICTAC Kinetics Project - multi-thermal-history
model-fitting methods and their relation to isoconversional methods. Thermochim. Acta, 355 (2000) 165.
[6] B. Roduit, Computational aspects of kinetic analysis. The ICTAC Kinetics Project - numerical techniques
and kinetics of solid state processes. Thermochim. Acta, 355 (2000) 171.
[7] B. Roduit, Thermochim. Acta, 388 (2002) 377.
[8] B. Roduit, Ch. Borgeat, B. Alonso, J.N. Aebischer, P. Pollien, A. Raemy and I. Blank, 32nd NATAS
CONFERENCE, Williamsburg Marriott, Williamsburg, VA October 4-6, 2004 (in progress).
[9] H. L. Friedman: J. Polymer Lett., 4 (1966) 323.
[10] T. Ozawa: Bull. Chem. Soc. Japan, 38 (1965) 1881.
[11] J.H. Flynn, L.A. Wall, J. Res. Nat. Bur. Standards, 70A (1966), 487.
[12] http://www.astm.org
[13] P. Brudugeac, J. Therm. Anal., Vol. 68 (2002) 131.
[14] B. Roduit, Ch. Borgeat, B. Berger, P. Folly, B. Alonso, J.N. Aebischer and F. Stoessel, J. Therm. Anal. Cal.,
ICTAC special issue, 80 (2005) 229–236.
[15] B. Roduit, Ch. Borgeat, B. Berger, P. Folly, B. Alonso, J.N. Aebischer, J. Therm. Anal. Cal., ICTAC special
issue, 80 (2005) 91–102.
[16] U. Ticmanis, G. Pantel, S. Wilker, M. Kaiser, Precision required for parameters in thermal safety
simulations, 32nd Internationl Annual Conference of ICT July, (2001), 135.
[17] B. Roduit, Ch. Borgeat, U. Ticmanis, M. Kaiser, P. Guillaume, B. Berger, P. Folly, 35th International
Annual Conference of ICT, June 29 - July 2, 2004, 37-1.
[18] P. Folly, Chimia, 58 (2004), 394.
[19] F. Stoessel, J. Steinbach, A. Eberz: Plant and process safety, exothermic and pressure inducing chemical
reactions, In: Ullmann's encyclopedia of industrial chemistry. Weise E (Eds), VCH, Weinheim (1995):343-354.
[20] A. Keller, D. Stark, H. Fierz, E. Heinzle, K. Hungerbuehler: Estimation of the TMR using dynamic DSC
experiments. Journal of Loss Prevention in the Process Industries (1997) 10(1):31-41.
[21] D.A. Frank-Kamenetskii, Diffusion and Heat Transfer in Chemical Kinetics, Plenum Press, New York,
London, 1969.
[22] J.M. Dien, H. Fierz, F. Stoessel, G. Killé: The thermal risk of autocatalytic decompositions: a kinetic study.
Chimia (1994) 48(12):542-550.
[23] D.W. Smith, Assessing the hazards of runaway reactions, Chem. Eng., 14, (1984) 54.
[24] T. Grewer, Thermochim. Acta, 225 (1993) 165.
[25] R. Gygax, International Symposium on Runaway reactions, March 7-9, 1989, Cambridge, Massachusetts,
USA, 52.
[26] F. Stoessel, Chem. Eng. Progress, October (1993) 68.
27] 2003, Recommendations on the Transport of Dangerous Goods, Manual of Tests and Criteria, 4 revised
edition, United Nations, ST/SG/AC.10/11/Rev.4 (United Nations, New York and Geneva).
[28] 2003, Globally Harmonized System of Classification and Labelling of Chemicals (GHS), United Nations,
New York and Geneva

30

You might also like