Fast and calibration free determination of first order reaction kinetics in API synthesis using in-situ ATR-FTIR

Article abstract of DOI:10.1016/j.ejpb.2017.09.013

In early stages of drug development only sparse amounts of the key substances are available, which is problematic for the determination of important process data like reaction kinetics. Therefore, it is important to perform experiments as economically as

Full text of DOI:10.1016/j.ejpb.2017.09.013

Accepted Manuscript
Research paper
FAST AND CALIBRATION FREE DETERMINATION OF FIRST ORDER
REACTION KINETICS IN API-SYNTHESIS USING IN-SITU ATR-FTIR
Moritz C. Rehbein, Sascha Husmann, Christian Lechner, Conrad Kunick,
Stephan Scholl
PII:
S0939-6411(17)30178-9
https://doi.org/10.1016/j.ejpb.2017.09.013
EJPB 12600
DOI:
Reference:
To appear in:
European Journal of Pharmaceutics and Biophar-
maceutics
Received Date:
Revised Date:
Accepted Date:
21 February 2017
17 July 2017
25 September 2017
Please cite this article as: M.C. Rehbein, S. Husmann, C. Lechner, C. Kunick, S. Scholl, FAST AND
CALIBRATION FREE DETERMINATION OF FIRST ORDER REACTION KINETICS IN API-SYNTHESIS
USING IN-SITU ATR-FTIR, European Journal of Pharmaceutics and Biopharmaceutics (2017), doi: https://
doi.org/10.1016/j.ejpb.2017.09.013
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

FAST AND CALIBRATION FREE DETERMINATION OF FIRST ORDER REACTION
KINETICS IN API-SYNTHESIS USING IN-SITU ATR-FTIR
1
,3
1
2
2,3
1,3*
Moritz C. Rehbein , Sascha Husmann , Christian Lechner , Conrad Kunick , Stephan Scholl
1
TU Braunschweig, Institute for Chemical and Thermal Process Engineering, Langer Kamp 7, 38016
Braunschweig, Germany (*Corresponding author: s.scholl@tu-braunschweig.de)
2
TU Braunschweig, Institute for Medicinal and Pharmaceutical Chemistry, Beethovenstraße 55, 38106
Braunschweig
3
TU Braunschweig, Center for Pharmaceutical Engineering, Franz-Liszt-Straße 35a, 38106 Braunschweig
ABSTRACT: In early stages of drug development only sparse amounts of the key substances are available,
which is problematic for the determination of important process data like reaction kinetics. Therefore, it is
important to perform experiments as economically as possible, especially in regards to limiting compounds. Here
we demonstrate the use of a temperature step experiment enabling the determination of complete reaction
kinetics in a single non-isothermal experiment. In contrast to the traditionally used HPLC, the method takes
advantage of the high measuring rate and the low amount of labor involved in using in-situ ATR-FTIR to
determine time-dependent concentration-equivalent data.
Keywords: API; synthesis; in-situ FTIR; reaction kinetics
1
. INTRODUCTION
During the process development for new API (active pharmaceutical ingredients), different properties of the
substances and processes involved have to be determined. These are for example solubility (solvent selection),
reaction kinetics (reaction characterization and optimization) and caloric data. Especially in the early stages of
drug substance development, only sparse amounts of the required substances are available to perform the above
tasks, hindering rapid process development [1].
For the determination of reaction kinetics, usually sampling and subsequent HPLC analysis is used to derive
time-dependent concentration profiles of the reactant with good accuracy. However, this method has some
limitations due to slow and manual sampling and the time- and labor-demand for sample preparation. These
disadvantages can be overcome by in-situ ATR-FTIR. The low amount of labor involved and the high sampling
rates possible with this technique can enable information rich and material saving experiments.
Kinetic analysis of chemical reactions has been under investigation for many years now. For first order reactions
the standard procedure is to plot the logarithm of the reactant concentration or concentration-equivalent physical
property over time and determine the reaction rate constant as the slope of the resulting straight line [2,3].
Various researchers have extended this principle approach by the development of specialized and more accurate
methods, applicable for example, when the initial reactant concentration cannot be determined [2,4]. For more
complex reactions, Mauser proposed a method to determine kinetics from spectroscopic data [5,6]. However, the
straightforward linearized plotting and determination of the slope is the standard procedure [3] and might still be
practical for first order reactions in most cases.
The determination of first order reaction kinetics directly from FTIR-signals using the standard plotting method
has also been demonstrated before. Krappitz et al. [7] and Xiao et al. [8] calculate reaction rate constants directly
from absorbance signals. In this contribution, a similar method for the determination of first order reaction
kinetics utilizing in-situ ATR-FTIR is demonstrated. However, compared to the isothermal approaches described
by Krappitz and Xiao, our method enables the computation of Arrhenius parameters (E
carefully designed, non-isothermal experiment.
A 0
, k ) from just one
As a model reaction (Figure 1, (iii)), a dealkoxycarbonylation [9–11] was used to synthesize the
pharmaceutically relevant scaffold 3,4-dihydro-1H-1-benzazepine-2,5-dione (3) from the enolized 3-
oxocarboxylic ester 2 as starting material. The procedure is carried out under neutral conditions in wet DMSO at
elevated temperatures [12]. 3,4-Dihydro-1H-1-benzazepine-2,5-dione (3) and its derivatives are used as platform
molecules for the synthesis of various protein kinase inhibitors and anticancer agents [13–18].
In the following, the development of the method as well as its application on the model system and a validation
with data derived from conventional HPLC measurements is presented.
1

1
2
3
Figure 1: Synthesis of 3,4-dihydro-1H-1-benzazepine-2,5-dione (3): (i) ethyl succinyl chloride, pyridine,
toluene, -10 °C  80 °C, 2h (ii) KO-tert-Bu, toluene, DMF, -10 °C  80 °C, 3.5 h (iii) DMSO, H O, N , 110 °C
2
2
–
160 °C, 1-3 h
2
2
. MATERIALS
.1. Chemistry
Solvents: Acetonitrile HPLC grade (Fisher Scientific UK, Loughborough, UK), DMSO anhydrous 99.8+ %
under argon (ThermoFisher (Kandel) GmbH, Karlsruhe, Germany), DMSO HPLC grade (ThermoFisher
(Kandel) GmbH, Karlsruhe, Germany), toluene HPLC grade (Fisher Scientific UK, Loughborough, UK),
pyridine for synthesis > 99 % (Carl Roth GmbH + Co. KG, Darmstadt, Germany) and N,N-dimethylformamide
pure (Applichem GmbH, Darmstadt, Germany) were used as received without further purification. Reactants:
Potassium tert-butoxide 98+ % (Acros Organics, Geel, Belgium), ethyl 2-aminobenzoate 99 % (Acros Organics,
Geel, Belgium) and ethyl 4-chloro-4-oxobutyrate 97 % (ThermoFisher (Kandel) GmbH, Karlsruhe, Germany)
were used. Precursors 1 and 2 and product 3 have been synthesized in house, based on published methods. See
supplementary for detailed information.
2
.2. Equipment and instruments
All syntheses for the determination of reaction kinetics of 3 where performed using an EasyMax™ 102 synthesis
workstation (Mettler Toledo Intl. Inc., Gießen, Germany). Reactor size was 50 mL and the reactor temperature
was monitored via a PT100 thermal element which was calibrated prior to use. HPLC measurements were
conducted on an UltiMate™ 3000 UHPLC system (Thermo Fisher Scientific GmbH, Dreiech, Germany)
equipped with a RP18 (3 µm, 100 mm x 3.0 mm) column (Phenomenex Ltd., Aschaffenburg, Germany) at 35
°
C. Absorption was detected with a RS Diode Array Detector at 252 nm. Isocratic elution with 40 % ACN and
6
1
0 % water at 0.4 ml/min was used for the determination of reaction kinetics (retention times: t
.81 min). Samples where diluted with HPLC grade DMSO and the injection volume was 3 µL.
2 3
= 6.17 min, t =
In-situ ATR-FTIR: A ReactIR™ 15 base unit (Mettler Toledo Intl. Inc., Gießen, Germany) equipped with a 9.5
mm DiComp™ probe tip and an AgX Fiber Conduit™ and cooled by liquid N has been used for in-situ FTIR
measurements. Backgrounds where collected prior to each experiment with 256 scans against air. Kinetic
2
-
1
measurements where performed in 15 s intervals with 30 scans and a resolution of 4 cm . Spectral data was
treated and analyzed using the ReactIR™ software by Mettler Toledo.
3
3
. METHOD DEVELOPMENT
.1. In-situ ATR-FTIR: In order to determine reaction kinetics, concentration over time data has to be
collected. Following Beer´s Law, the absorbance of a component in a mixture linearly depends on its
concentration. To identify characteristic absorbance peaks for each component, especially reactants, in the
spectrum of the reaction mixture, pure component FTIR spectra of solvent, reactant and product are recorded and
-1
-1
plotted from 1300 to 1100 cm in Figure 2 (left). The peak at 1257 cm is only present in the reactant spectrum
and associated with the C-O-C stretching vibration of the ester group of reactant 2 that is removed from the
molecule due to reaction. Therefore, this functional group is not present in the product molecule and thus the
-
1
spectrum of the product does not contain this absorbance band. Consequently, the absorbance band at 1257 cm
is used as measurement parameter for the reactant concentration.
2

0
0
0
0
0
.4
.3
.2
.1
.0
1.25
3
2
C-O-C
@
R² = 0.9988
-
1
1257 cm
1
0
0
0
.00
.75
.50
.25
DMSO
b(T)
0.00
1
300
1250
1200
Wavenumber [cm-1
1150
1100
0
10
20
30
40
50
60
]
Concentration via HPLC [mg mL-1]
Figure 2: left: Untreated pure component spectra of solvent DMSO, reactant 2 and product 3; right: correlation between the
-1
-1
reactant (2) concentration determined by HPLC and the area of the peak at 1257 cm (integration from 1260 – 1252 cm )
calculated from treated FTIR spectra at 110 °C
-1
For the calculation of reaction kinetics (see below) A1257, the area of the peak at 1257 cm , has to reflect the
concentration of the reactant meaning that the peak integral is zero at zero concentration and that it increases
linearly with increasing reactant concentration. Therefore, the treatment of the FTIR spectra collected during the
reaction comprised of the subtraction of the pure solvent spectrum (DMSO) recorded at 110 °C. Pure solvent
spectra have been recorded at temperatures between 110 °C and 150 °C, however, differences were found to be
-1
insignificant. Baseline correction was carried out by subtracting the absorbance value at 1130 cm from the
-
1
whole spectrum, i.e. shifting every spectrum in a way, that the absorbance at 1130 cm is zero. The peak was
-
1
then integrated from 1260 to 1252 cm and a trend of the peak area over time was generated. Solvent
TM
subtraction, baseline-correction, peak integration and trend generation where performed in the ReactIR
software by Mettler Toledo. As seen in Figure 2, right, the spectra treatment and integration procedure described
-1
above results in a linear dependency of the absorbance at 1257 cm and the concentration of the reactant in the
mixture, enabling the kinetic analysis described in the following section.
3
.2. Temperature step experiments: The EasyMax™ reactor system was set to a specific temperature profile in
order to derive reaction rate constants for different temperatures from the same experiment. After filling the
reactor with 15 mL of anhydrous DMSO, the reactor temperature was brought to 110 °C and approx. 1 g of
reactant 2 was added. From preliminary results it was concluded, that there is no detectable reaction at this
temperature. 10 equivalents (0.73 mL) of water where added to start the reaction. After 15 min at 110 °C the
reactor was subjected to a specific temperature ramp (see also Figure 3): 5 min heating to the next temperature
and 15 min reaction at constant temperature for 120, 130, 140 and 147.5 °C, respectively. After performing the
-
1
temperature profile, the reaction mixture was kept at 147.5 °C until the peak at 1257 cm had disappeared to
assure complete conversion. After cooling to 20 °C, 15 mL of water where added, the mixture was cooled to 5
°
C, the precipitate was filtered off and dried without further purification.
-
1
In Figure 3, the absorbance at 1257 cm during the temperature step experiment is plotted against the reaction
time. Upon adding reactant 2 after about 2000 s the signal increases to about 1.3 A.U./cm. After the addition of
water a slight decrease of the signal is observed due to dilution. The signal stays constant until the temperature is
increased to 120 °C indicating no noticeable reaction at T ≤ 110 °C. In general, a decrease in signal strength is
seen during each temperature increase. This is due to the temperature dependence of the FTIR signal, which
decreases with increasing temperature. During periods of constant temperature, the absorbance signal decreases.
The slope of the decrease increases with increasing temperature, which is to be expected since the reaction rate is
accelerated with temperature.
3

1
1
1
0
0
0
0
0
.4
.2
.0
.8
.6
.4
.2
.0
160
150
140
A1257(t=0) for T = 120
and 130 °C
0.0
0.1
0.2
-
-
A1257(t=0) for T = 140 130
and 147.5 °C
120
k(T)
120 °C
30 °C
110
100
90
1
2 Area
T [°C]
140 °C
147.5 °C
-0.3
0
3600
7200
time [s]
10800
14400
0
200
400
600
800
1000
time [s]
-1
Figure 3: left: Temperature profile (---) and area of the reactant 2 absorbance peak (-) at 1257 cm (integration
-
1
from 1260 – 1252 cm ) over the reaction time right: Logarithm of the ratio of the time dependent absorbance
and the initial absorbance at the given temperature according to Eq. 2
-
1
To obtain quantitative information, the time dependent absorbance peak area at 1257 cm (A1257) was then used
to calculate reaction kinetic data based on reactant concentration, i.e. absorbance change over time t. A1257 is
linearly dependent on the concentration of the absorbing component in the mixture at constant temperature T
(
Figure 2, right) in the form of, with b being a the temperature dependent proportionality factor:
ꢅꢆꢇꢈꢉꢊꢋꢁ .
(1)
ꢀꢁꢂꢃꢄ
For a first order reaction, Eq. (2) can be used to determine the reaction rate constant k(T) from the ratio of the
actual (c (t)) over initial (c (t=0s)) reactant concentration. When Eq. (1) is valid, the term c (t) / c (t=0s) can, for
2
2
2
2
constant temperature conditions, be replaced by the ratio of the peak integrals A1257(t) / A1257(t=0s). This enables
the calculation of the reaction rate constant k(T) for first order reactions directly from the time dependent peak
area according to Eq. (2) without the need to calibrate the FTIR signal and concentration at each temperature.
ꢋꢁꢇꢏꢉ
ꢀꢁꢂꢃꢇꢏꢉ
ꢌꢍꢎ
ꢓꢅꢔꢇꢈꢉꢊꢏꢄꢕꢖꢄꢅꢌꢍꢎ
ꢓ
(2)
ꢋꢁꢇꢏꢐꢑꢒꢉ
ꢀꢁꢂꢃꢇꢏꢐꢑꢒꢉ
Plotting the logarithm of the normalized peak area data over the reaction time results in a straight line (Figure 3,
right) that can be fitted with Eq. 2, where the slope represents the reaction rate constant k(T). Theoretically, the
line should run through the origin. However, due to measurement noise, normalization may lead to a slight
vertical shift of the data. Therefore, the correction parameter S is introduced, which enables the correct
TM
calculation of the slope and thus k. Peak area over time data is exported out of ReactIR and the procedure
described above is applied for each period of constant temperature.
Since different temperatures have been selected, the k-values for different temperatures can be derived from one
single experiment. Finally, standard Arrhenius analysis may be applied to determine reaction rate constant
frequency factor k
0 A
and activation energy E , with R being the universal gas constant:
ꢚ
ꢀ
ꢌꢍꢗꢔꢇꢈꢉꢘꢅꢌꢍꢇꢔ ꢉꢙ ꢊ .
(3)
ꢑ
ꢛ
ꢈ
ln[k(T)] and corresponding 1/T values have been fitted by Eq. 3 and Arrhenius parameters are calculated from
slope and intercept accordingly.
4
. RESULTS AND DISCUSSION
4

The temperature step experiment was conducted four times with identical settings in order to evaluate
repeatability and statistical deviations between the different runs. Reaction rate constants where determined from
spectroscopic data according to Figure 3 right. In addition, and to validate the temperature-step method using
FTIR, the reaction rate constants were also determined during four isothermal experiments (124.3, 134.5, 144.3
and 159.1 °C, respectively), where manual sampling and conventional HPLC analysis were employed. For this
validation method, the reaction was conducted for 2 hours at the respective temperature, samples were collected
every 15 minutes and analyzed via HPLC to derive concentration over time data, which was then used to
calculate reaction rate constants. All resulting reaction rate constants k(T) for the different temperatures and
experimental methods are plotted according to Eq. (3) in Figure 4 (◊ ≙FTIR with T-step, ○ ≙isothermal with
HPLC) .
-
-
-
-
-
-
-
-
6.0
6.5
7.0
7.5
8.0
8.5
9.0
9.5
HPLC Validation
R² = 0.9993
FTIR temperature step
Linear (HPLC Validation)
R² = 0.9981
Linear (FTIR temperature step
without values @ 1/T =0,00254)
-
-
-
10.0
10.5
11.0
0
.0023
0.00235
0.0024
0.00245
0.0025
0.00255
0.0026
-
1
1
/T [K ]
Figure 4: Arrhenius plot of k-values derived from FTIR-temperature-step experiments (◊, mean values; n = 4)
and HPLC validation measurements (○)
-
1
For all values 1/T ≤ 0.00252 K , corresponding to T ≥ 123.7 °C, a satisfactory correlation of ln(k) over 1/T is
obtained for FTIR temperature step experiments and HPLC validation experiments, respectively.
-
1
However, for the lowest temperature (120 °C, 1/T ~ 0.00255 K ), the calculated reaction rate constants derived
from FTIR measurements show a pronounced scattering as indicated by the error bars. A combination of
evaporation effects and the overall slow reaction rate at that temperature are assumed to corrupt the
measurement. Due to the evaporation of water in the beginning of the temperature step experiment (e.g. at the
low temperatures 110 °C and 120 °C), the concentration of the reactant 2 increases, which is counteracting the
relatively slow decrease in concentration due to reaction. Since at the higher temperatures most of the water has
already evaporated and since the reaction rate is larger at higher temperatures anyway, this problem occurs only
in the beginning of the temperature step experiment. Consequently, all of the reaction rate constants for 120 °C
are assumed to be defective and therefore excluded from the calculation of the regression.
Still, the remaining k-values for temperatures 130, 140 and 147.5 °C show a low standard deviation and can be
fitted very well (R² = 0.9981) using the linearized form of the Arrhenius-equation. The Arrhenius parameters
calculated from FTIR temperature step experiments and HPLC validation measurements are displayed in Table
1
. It can be seen that FTIR and HPLC derived data for the reaction rate constants are in good agreement.
Table 1: Comparison of Arrhenius parameters calculated from isothermal HPLC measurements and
temperature-step experiments employing in-situ FTIR
HPLC Validation
FTIR without
values @ 120 °C
-
1
E
A
[kJ mol ]
140.38
- 1.636
139.76
- 1.304
14
-1
k  10 [s ]
0
Considering the error bars of the k-values derived from the FTIR temperature step experiments at the specific
temperatures, it can be concluded that the deviation between the four experiments is very low (coefficient of
variation ≤ 3 %). Therefore it might be possible to describe the temperature dependent reaction kinetic of the
given reaction by only conducting a single non-isothermal temperature step experiment. In contrast, four
isothermal HPLC experiments where necessary involving much more labor, reactant and also resources for
HPLC analysis itself.
5

4
.1. Modelling
The determination of the Arrhenius parameters k
0
A
and E from temperature step experiments using FTIR allows
to compare simulated reaction kinetics to the reference method: experiments at T = const. and offline HPLC
analysis. For a first order reaction the relative concentration profile of the reactant over time is given by:
ꢠꢚ
ꢋꢁꢇꢈꢜꢏꢉ
ꢋꢑꢜꢁ
ꢎ
ꢓ
ꢅꢝꢞꢟꢄꢗꢔ ꢊꢝ ꢛꢈ ꢊꢏꢘ.
(4)
ꢑ
In Figure 5, the normalized concentration profiles for the four validation experiments are plotted over the
reaction time and compared with the calculated profiles using Eq. (4) and the Arrhenius parameters derived only
from FTIR measurements (Table 1, right). The results are generally in good agreement, however slight
deviations are found. A closer look at the concentration profiles reveals that concentrations are overestimated by
the model for 124.3, 144.3 and 159.1 °C and underestimated slightly for 134.5 °C.
1
0
0
0
0
0
0
0
0
0
0
.0
.9
.8
.7
.6
.5
.4
.3
.2
.1
.0
0
1000
2000
3000
4000
5000
6000
Reaction time [s]
Figure 5: Comparison of HPLC concentration data (data points) with the calculated concentrations using the
Arrhenius-model (curves) with model parameters derived only by FTIR measurements
Reconsideration of Figure 4 shows that the k-values calculated from HPLC measurements also do not follow the
linear regression perfectly. The value for 144.3 °C is slightly above while the value for 134.5 °C is slightly
below the regression line in the Arrhenius plot. This indicates that also the HPLC measurements show some
measurement scatter and that the FTIR-derived concentration profiles therefore may not fit perfectly to the
discrete HPLC measurements. For an additional comparison of both methods, k-values, calculated from the E
A
and k
0
values in Table 1, are displayed in Table 2 for the considered temperature range.
Table
k-values
Arrhenius
from HPLC-
measurements
2:
Comparison of
calculated with
parameters
-5
-1
-5
-1
T [°C]
HPLC: k ꢊ10 [s ]
FTIR: k ꢊ10 [s ]
3.51
k
FTIR / kHPLC [-]
0.965
1
1
1
1
1
20
30
40
50
60
3.64
10.57
29.13
76.52
192.26
and
FTIR-
10.15
27.85
72.84
182.25
0.960
0.956
0.952
0.948
Over the temperature range of interest, the k-values derived from FTIR are always lower than those from HPLC,
indicating that the combination of FTIR and temperature step experiments slightly underestimates the Arrhenius
parameters in comparison to the standard method HPLC. However, the deviation is small and the model is still
able to predict the concentration profiles with a good accuracy. Regarding the modelled profiles for the higher
temperatures (170 °C and 190 °C, displayed Figure 5) it can be concluded that an increase of reaction
temperature above the original 150 °C can greatly improve reaction time. For full conversion (99 %), this
extrapolation predicts that reaction time can be decreased from about 105 min (150 °C) to 3.5 min (190 °C).
6

However, this will only give rough estimations for the potential reduction of reaction time and will certainly
need experimental validation.
5
. CONCLUSION
In early stages of process synthesis fast and material-saving methods are required to determine fundamental data
for process analysis and optimization. Reaction kinetics information is of special interest. This contribution
presented a method combining a consecutive series of temperature step experiments with in-situ FTIR analysis.
This is demonstrated for the reaction kinetics of a pharmaceutically relevant scaffold. The high sampling rate and
the low amount of labor involved in the use of in-situ ATR-FTIR generates sufficient data in just a single
experimental run to fully describe the time- and temperature-depended concentration profile of the reactant in the
given temperature range.
The presented method can be used for the fast determination of reaction kinetics during drug substance process
development to design and scale appropriate processes and process equipment. It is of special relevance in the
context of transferring batch to continuous synthesis, as it allows for fast screening experiments with minimum
material consumption
6
. ACKNOWLEDGEMENT
Tꢀiꢁ ꢁtudꢂ ꢃaꢁ ꢄartiaꢅꢅꢂ ꢆundꢇd ꢈꢂ tꢀꢇ ꢉiꢇdꢇrꢁꢊꢋꢀꢁiꢁꢋꢀꢇꢁ ꢌiniꢁtꢇriuꢍ ꢆꢎr ꢏiꢁꢁꢇnꢁꢋꢀaꢆt und Kuꢅtur (MWK) in
the joint research project µ-Props oꢆ tꢀꢇ Cꢇntꢇr oꢆ ꢐꢀarꢍaꢋꢇutiꢋaꢅ ꢑnꢒinꢇꢇrinꢒ ꢓꢐꢔꢕ) oꢆ tꢀꢇ Tꢇꢋꢀniꢁꢋꢀꢇ
ꢖnivꢇrꢁitꢊt ꢗraunꢁꢋꢀꢃꢇiꢒ – Processing of poorly soluble drugs at small scale.
7
. SYMBOLS
[A.U. cm ]
-
1
-1
A
1257
area of the absorbance peak at 1257 cm
-
1
-1
b(T)
[A.U. mL cm mL ]
temperature dependent proportionality factor
concentration of compound 2
activation energy
reaction rate constant
reaction rate frequency factor
universal gas constant
correction parameter
temperature
-
1
c
2
[mg mL ]
-1
E
k
A
[kJ mol ]
-1
[s ]
-
1
k
0
[s ]
-
1
-1
R
S
T
t
[J mol K ]
[-]
[K], [°C]
[s]
time
8
. REFERENCES
[
1] A.E. Cervera-Padrell, T. Skovby, S. Kiil, R. Gani, K.V. Gernaey, Active pharmaceutical ingredient (API)
production involving continuous processes--a process system engineering (PSE)-assisted design framework,
Eur. J. Pharm. Biopharm. 82 (2012) 437–456.
[
2] E.A. Guggenheim, On the determination of the velocity constant of a unimolecular reaction, Phil. Mag. 2
(1926) 538–543.
[
[
3] O. Levenspiel, Chemical Reaction Engineering, 3rd ed., Wiley, New York, 1999.
4] E.S. Swinbourne, Method for obtaining the Rate Coefficient and Final Concentration of a First-order
Reaction, J. Chem. Soc. (1960) 2371–2372.
[
[
5] J. Polster, H. Mauser, Kinetic Analysis of second order Reactions using UV-VIS Spectroscopy, Talanta 10
(1992) 1355–1359.
6] H. Mauser, U. Hezel, Zur kinetischen Analyse mit Hilfe der formalen Integration: Die theoretischen
Grundlagen und ihre Anwendung auf die spektralphotometrische Untersuchung der Photoisomerisation
von Azobenzol, Z. Naturforsch. 26 b (1971) 203–208.
[
[
[
7] T. Krappitz, D. Brauer, P. Theato, Synthesis of poly(allyl 2-ylidene-acetate) and subsequent post-
polymerization modification via thiol–ene reaction, Polym. Chem. 7 (2016) 4525–4530.
8] L. Xiao, F. Li, Y. Li, X. Jia, L. Liu, Kinetic study of carbene polymerization of ethyl diazoacetate by
palladium and rhodium catalysts, RSC Adv 4 (2014) 41848–41855.
9] A.P. Krapcho, J.F. Weimaster, J.M. Eldridge, E.G.E. Jahngen, A.J. Lovey, W.P. Stephens, Synthetic
applications and mechanism studies of the decarbalkoxylations of geminal diesters and related systems
effected in dimethyl sulfoxide by water and/or by water with added salts, J. Org. Chem. 43 (1978) 138–147.
10]A. Krapcho, G.A. Glynn, B.J. Grenon, The decarbethoxylation of geminal dicarbethoxy compounds,
Tetrahedron Letters 8 (1967) 215–217.
[
7

[
11]A. Krapcho, E. Jahngen, A.J. Lovey, F.W. Short, Decarbalkoxylations of geminal diesters and β-keto esters
in wet dimethyl sulfoxide. Effect of added sodium chloride on the decarbalkoxylation rates of mono- and di-
substituted Malonate esters, Tetrahedron Letters 15 (1974) 1091–1094.
[
[
12]C. Kunick, Synthese [b]-kondensierter azepindione durch dealkoxycarbonylierung, Arch. Pharm. Pharm.
Med. Chem. 324 (1991) 579–581.
13]A. Becker, S. Kohfeld, A. Lader, L. Preu, T. Pies, K. Wieking, Y. Ferandin, M. Knockaert, L. Meijer, C.
Kunick, Development of 5-benzylpaullones and paullone-9-carboxylic acid alkyl esters as selective
inhibitors of mitochondrial malate dehydrogenase (mMDH), Eur. J. Med. Chem. 45 (2010) 335–342.
14]A.-M. Egert-Schmidt, J. Dreher, U. Dunkel, S. Kohfeld, L. Preu, H. Weber, J.E. Ehlert, B. Mutschler, F.
Totzke, C. Schächtele, M.H.G. Kubbutat, K. Baumann, C. Kunick, Identification of 2-anilino-9-methoxy-
[
[
5
,7-dihydro-6H-pyrimido5,4-d1benzazepin-6-ones as dual PLK1/VEGF-R2 kinase inhibitor chemotypes by
structure-based lead generation, J. Med. Chem. 53 (2010) 2433–2442.
15]H. Falke, A. Chaikuad, A. Becker, N. Loaec, O. Lozach, S. Abu Jhaisha, W. Becker, P.G. Jones, L. Preu, K.
Baumann, S. Knapp, L. Meijer, C. Kunick, 10-iodo-11H-indolo3,2-cquinoline-6-carboxylic acids are
selective inhibitors of DYRK1A, J. Med. Chem. 58 (2015) 3131–3143.
[
[
16]C. Kunick, Fused azepinones with antitumor activity, Curr. Pharm. Des. 5 (1999) 181–194.
17]C. Kunick, C. Bleeker, C. Prühs, F. Totzke, C. Schächtele, M.H.G. Kubbutat, A. Link, Matrix compare
analysis discriminates subtle structural differences in a family of novel antiproliferative agents, diaryl-3-
hydroxy-2,3,3a,10a-tetrahydrobenzobcycylopentaeazepine-4,10(1H,5H)-d iones, Bioorg. Med. Chem. Lett.
1
6 (2006) 2148–2153.
18]N. Tolle, C. Kunick, Paullones as Inhibitors of Protein Kinases, Curr. Top. Med. Chem. 11 (2011) 1320–
332.
[
1
8

Graphical abstract
9