(88c) FT-IR & Spectra Deconvolution: A Fruitful Combination for in-Situ Process Monitoring in Homogeneous Catalysis

FT-IR & Spectra Deconvolution: A
fruitful combination for in-situ Process Monitoring in Homogeneous
Catalysis

K.U. Kuennemann,
J.M. Dreimann*, D. Vogt

Keywords:
Chemometric techniques, Infrared spectroscopy, Band-targeted Entropy
Minimization, Homogeneous Catalysis

Chemical and
allied industries manufacture products that are essential for creating and
sustaining modern societies. The chemical transformations necessary to make
these essential products often involve the use of catalysts. The development, selection,
and investigation of the right catalyst on the molecular level, therefore, can
make a substantial impact on process viability and economics. In order to improve
existing and future processes in terms of catalytic productivity and energy
efficiency, new ways of process intensification, computational models, and
real-time process monitoring and control, amongst other things, are necessary.^[1]
 Todays chemical industry is still mostly limited on the measurement and control of
physico-chemical parameters like temperature, pressure, density, conductivity,
pH or  for process control. Additional information
on a molecular level typically provided by gas chromatography (GC) or
high-performance liquid chromatography (HPLC) is mostly only available in
off-line mode which takes typically 30–60 minutes for the result to be
available.^[2]
Consequently, obtaining information on a molecular
level in real time from a process stream by non-invasive methods is highly preferable
for a proper process control. For that different non-invasive
and real-time analytical tools with a high information density are necessary. Nowadays different spectroscopic methods are
able to allow this, such as nuclear magnetic resonance, Raman, UV-Vis and
IR-spectroscopy. Unfortunately, having only trace components in a reaction
mixture or a mixture consisting of various components, these non-invasive
techniques are often not able provide a clear discrimination between components
or even reflect their molecular interaction. For that, chemometrics emerged as
science that uses mathematical, statistical, and other methods to provide
maximum relevant chemical information by analysing chemical data.^[3]
Different combinations in chemometrics are known e.g. the fourier transformation
IR (FT-IR) with computer-based mathematical methods like band-target-entropy-minimization
(BTEM), FacPack or e.g. non-negative matrix factorization (NMF).^[4]

Trace compontens
in a reaction mixture can have a huge influence on the peformance of espically
homogeneous chemical processes in terms of forming by-products, shifting the
equlibrium of a reaction or acceerating catalyst deactivtion. First sucessful attemps to in-situ monitor a catalyzed
chemical reaction in homogenous catalysis were latley shown by Dreimann et al. for the hydroformylation
with focus in particular on the detection of the active catalytic species using
transmission infrared measurements in combination with advanced chemometric
analysis by BTEM to obtain pure component spectra.^[5]
Since the hydroformylation is one of the major homogenously catalyzed reactions
in terms of world production, homogeneous catalysis
offers also an enormous potential for the synthesis of long chain amines. In
comparison to the common industrial routes, amines herein can be directly
synthesized from olefins by hydroaminomethylation (HAM), for instance. HAM
consists of initial hydroformylation of an alkene to an aldehyde followed by
reductive amination of the intermediate to the desired amine. Both reactions take
place in the same reaction mixture in a tandem catalytic manner. In order to
set up a potential process, two different approaches are imaginable. A single
step tandem-catalytic HAM process or a two-step process using a reasonable
combination of hydroformylation and reductive amination. The literature
contains studies on hydroformylation, reductive amination and
hydroaminomethylation reactions. A detailed comparison of the efficiency of the
one-step and the consecutive process has not yet been carried out. This
comparison is heart of the german collaborative
research centre/Transregio 63 “Integrated Chemical Processes in Liquid
Multiphase Systems” (InPROMPT), and thus, reductive amination and
its limitations needs to be investigated in a very detailed way.

Consequently
this work will focus on the monitoring of the consecutive reductive amination for
a continous miniplant process. The reductive amination
itself also consists of two reaction steps, which increase the difficulty of an
effective online monitoring: a) non-catalysed condensation of aldehyde A
to enamine B and b) hydrogenation of the double bond to the saturated
amine C. However, analysis of the reaction network of reductive
amination (Figure
1) shows other potential side reactions leading to, for
instance, alcohol E by hydrogenation of the aldehyde substrate and aldol
condensates D. The equilibrium of the aldehyde condensation has a big
influence on the by-product formation. If the concentration of the aldehyde is
too high, side reactions are promoted and yields of the by-products increase
significantly. This research presents the rhodium catalysed reductive amination
of long-chain aldehydes. As model reaction the reductive amination of n-decanal
with diethylamine to the tertiary amine N,N-diethyldecylamine was
chosen.

Figure 1:
Reaction network of the reductive amination.

Espacilly
the evolving water as by-product from the condensation reaction is highly
interesting in the manner of process control since it not only shifts the
equilbirium B to A if starts to accumalate in the reaction
system, but also on the possible spectroscopy measurements. In this work the
transmission based FT-IR is used to investigate and to control the reaction
system. In a test set-up the influence of water and the amount of components in
the system was not to neglect due to the influence on the spectrocopy data.
Based on published literature by Garland
et al. the BTEM algorithm was implemeted at our research laboratory.^[6,7] However, the
implemented BTEM algorithm has shown its limitations, so that a suifficient
spectra deconvulution was not possible,as Dreimann
et al. have shown previously for the hydroformylation. Therefore, this
work does not aim to show the molecular behaviour of the catalyst but rather a
way to customize the published BTEM algorithm for multicomponent systems. In
this manner the BTEM code was scattered in its main components and for each component
their influence on the deconvolution quality was shown. With this information,
new optimization ideas, such as a highpass-filter from the information
technology, came up, which were implemented to optimize this algorithm.^[8] So, that in
the end a successful deconvultion of the model reaction with an automized BTEM algorithm
was partly achieved. Since the stated influence parameters are not only limited
to the model reaction system, the information gathered in this work can be used
for the optimization for various multicomponent systems. When also combined
with a real-time and non-invasiv anlytic tool the deconvolution algorithm can
have huge impact on the process control of complex mulitcomponent systems.

Therefore, the
present contribution has five goals, including

(i)                 
the in-situ spectroscopic study of the reductive
amination using a customized high-pressure infrared cell

(ii)               
the use of the BTEM algorithm to deconvolute and
untangle each set of data

(iii)              
showing general influence parameters within the
BTEM to adapt a the algorithm to new reaction system

(iv)              
general optimization strategies for the BTEM
algorithm for complex multicomponent mixture

(v)               
comparing limitation and potential of
chemometric for the homogeneous reductive amination

Reference:

[1]         S. S. Joshi, V. V. Ranade
(Eds.), "Industrial catalytic processes for fine and specialty
chemicals", Elsevier, Amsterdam, Netherlands, 2016.

[2]         D. Martoccia, H. Lutz, Y.
Cohen, T. Jerphagnon, U. Jenelten, "FT-NIR: A Tool for Process Monitoring
and More", Chimia 2018, 72, 139–145.

[3]         D. L. Massart, L. Buydens,
"Chemometrics in pharmaceutical analysis", Journal of
Pharmaceutical and Biomedical Analysis 1988, 6, 535–545.

[4]         J. Wang, Feng Tian, X. Wang, H.
Yu, C. Hong Liu, L. Yang, "Multi-Component Nonnegative Matrix
Factorization" in Proceedings of the Twenty-Sixth International
Joint Conference on Artificial Intelligence (IJCAI-17), 2017.

[5]         J. M. Dreimann, E. Kohls, H.
Warmeling, M. Stein, L. Guo, M. v. Garland, T. N. Dinh, A. Vorholt,
"In-situ infrared spectroscopy as a tool for monitoring molecular catalyst
for hydroformylation in continuous processes", ACS Catal. 2019.

[6]         H. Zhang, M. Garland, Y. Zeng,
P. Wu, "Weighted two-band target entropy minimization for the
reconstruction of pure component mass spectra: Simulation studies and the
application to real systems", Journal of the American Society for
Mass Spectrometry 2003, 14, 1295–1305.

[7]         E. Widjaja, C. Li, W. Chew, M.
Garland, "Band-Target Entropy Minimization. A Robust Algorithm for Pure
Component Spectral Recovery. Application to Complex Randomized Mixtures of Six
Components", Anal. Chem. 2003, 75, 4499–4507.

[8]         H. Stögbauer, Dissertation,
Bergische Universität, Wuppertal, 2005.