Introduction
In recent years, the field of organic synthesis has developed rapidly. Flow chemistry
is a revolutionary achievement in the field of organic chemistry, which has advanced
batch chemical synthesis and devices. The most significant features of flow chemistry
are the efficiency improvement and increased safety of synthesis processes. Furthermore,
common devices like microreactors are superior to batch reactors in terms of mixing
time, temperature control, material consumption, and number of intermediate purification
steps required for continuous experiments in both academia and industry.[1 ] However, early applications required manual sampling and transportation of samples
to a central analytical laboratory. This process is unsafe and costly, often resulting
in an inaccurate and untimely representation of process conditions and delays in analyzing
results, which hinder the real-time tracking of process dynamics. Furthermore, the
data-utilization insufficiency during this process leads to long development times.
Process analysis techniques (PATs) are essential for collecting the results of experiments
in any auto-operating platform. PAT devices include mass-flow controllers, as well
as temperature, pressure, pH, and conductivity sensors, and they can provide valuable
information about the state of the reactor system. Collecting real-time data using
immersion probes will improve our understanding of chemical reactions and crystallization.
This also eliminates the need to analyze physical interference with the reaction by
sampling/quenching and to use time-consuming offline methods, such as high-performance
liquid chromatography (HPLC). This is particularly beneficial for water- and/or oxygen-sensitive
reactions, the formation of transient intermediates, or reactions for which safe and
accurate sampling is difficult, which are often heterogeneous, pressurized, or performed
at elevated temperatures.[2 ]
A PAT device can be integrated as an inline, online, or at-line sensor. In terms of
process types, the PAT can be based on mass spectrometry or spectroscopy techniques,
such as nuclear magnetic resonance (NMR), infrared (IR), Raman, and ultraviolet–visible
(UV–vis) spectroscopies.[3 ] The sampling frequency of PAT needs to be sufficiently high to capture all the variations
during the dynamic experiment. Combined with real-time analysis during compound synthesis,
real-time information on intermediates and by-products can be obtained, providing
significant material, time, and cost savings. Therefore, combining reactors with inline
monitors is particularly important.[4 ]
[5 ] For example, the combination of microreactors with inline monitors promotes the
safe and efficient synthesis of compounds. The integration of flow chemistry into
a fully automatic microreactor platform will enable the convenient and efficient observation
of the reaction progress. The establishment of kinetic models through the multi-dimensional
optimization of reaction parameters can replace the existing labor- and cost-intensive
inspection procedures, reduce waste generation, improve the efficiency of the production
process, ensure overall sustainability, and guide the reaction to achieve the ideal
conditions. Additionally, manual intervention no longer becomes a requirement. There
are many kinds of process-analysis techniques[5 ]
[6 ]
[7 ]
[8 ]
[9 ]
[10 ]
[11 ]
[12 ]; however, this review only discusses the application of inline IR and inline NMR
in flow chemistry.
Introduction of Inline IR and NMR Technology
Real-time process-analysis systems include sampling systems, analyzers, sensor technologies,
and data-analysis computers with specialized algorithms. These systems are typically
located in the laboratory and are expected to operate unattended. The analysis information
is usually converted into relevant processing information after its quality is ascertained
in the onboard computer system before being transmitted to the process control system.
As reliability increases, this information is employed more for closed-loop control
rather than the historical open-loop approach, where researchers can view the analysis
results to determine if the process control needs to be changed or the data need to
be archived for future reference.[13 ]
Researchers can use inline IR spectroscopy to monitor the distinguishable characteristic
peaks formed during the reaction process and obtain the structures of different compounds.
Thus, this technique can help us rapidly detect the target compound and can be exploited
for qualitative and quantitative compound analyses (analysis time ≤ 1 minute).[14 ] Furthermore, it can be used for kinetic model prediction because of its high efficiency
in collecting kinetic data. Regarding highly exothermic processes, accurate enthalpy
and kinetic data are necessary to find safe and optimal operating conditions. This
is usually achieved by batch experiments, often requiring large quantities of reactants.
To address this problem, a microreactor combined with inline IR was applied to achieve
continuous flow in a thermodynamic measurement platform.[15 ] Inline IR can also be combined with other analysis methods, such as mass spectrometry,
to form an inline analysis platform where IR is used to monitor the main components
of the reaction. The highly sensitive mass spectrometry technique provides insights
into the formation of by-products, enabling the study of complex optimization problems
and the full discovery of the potential of inline analysis.[16 ]
NMR spectroscopy has many advantageous features: it is nondestructive and can provide
accurate quantitative and structural information. Thus, it is a valuable tool for
collecting accurate experimental data for complex processes. Such data are essential
for a comprehensive understanding of the process and for the development of reliable
process models. NMR can often distinguish structurally similar but distinct compounds
in multicomponent mixtures. Furthermore, quantitative results, that is, the composition
of the mixtures being studied, can be obtained with little or no calibration.[17 ] Quantitative NMR spectroscopy has been widely employed to monitor reactions in a
variety of devices.[18 ]
[19 ]
[20 ]
[21 ]
[22 ] NMR spectroscopy can also be combined with other methods, such as HPLC, to identify
complex mixture components.[23 ]
Introduction to Inline IR and NMR Applications
Kinetic and Thermodynamic Studies
Designing chemical processes requires reliable data on the reaction kinetics. Using
the imine syntheses of benzaldehyde and benzylamine as examples ([Fig. 1 ]), Fath et al obtained and modeled the kinetic data using inline Fourier transform
infrared (FTIR) spectroscopy, unsteady state conditions, and self-modeling curve resolution.[24 ] This automated microreactor system combined with continuous inline measurements
allowed flow experiments to be performed with continuous changes in the flow rate
([Fig. 2 ]). The availability of analytical results in real-time significantly reduced the
reagent consumption and the time required to collect sufficient kinetic data. Scaling
up from laboratory to pilot or production levels is important for industrial applications.
Scaling requires a good understanding of mass and heat transfer, as well as the kinetic
model itself. Fath et al used inline FTIR spectroscopy to achieve model-based upscale
predictions of reactions in a high-heat-release model from the laboratory to the pilot
plant.[24 ] Based on the study of the laboratory-scale dynamics, they successfully scaled the
experimental model using FTIR measurements. They increased the channel diameter of
the microreactor (inner diameter = 0.5 mm) to fit the nanoscale pilot reactor (diameter = 2 mm),
considering the mixing efficiency, residence time distribution, and heat transfer.[25 ]
[26 ]
Fig. 1 Imine synthesis.
Fig. 2 Fath et al's microreactor setup for kinetic experiment.[24 ]
Bazzoni et al demonstrated the inline monitoring of a flow photochemical reaction
using one-dimensional (1D) and ultrafast two-dimensional (2D) NMR methods in a high
magnetic field.[27 ] The reaction mixture exiting the flow reactor flowed through the NMR spectrometer
and was directly analyzed. For simple substrates, suitable information could be obtained
through 1D 1 H spectra; however, for molecules of higher complexity, the use of 2D experiments
was key in addressing signal overlaps and assignment issues. The experimental setup
used in this study is represented schematically in [Fig. 3 ]. The custom-made flow reactor is connected to a commercial flow tube (InsightMR,
Bruker) inserted into a 500 MHz spectrometer and an HPLC pump is used to control the
sample flow. The flow tube comprises a 5 mm NMR tube tip and two 7 m long 0.5 mm ID
peek capillaries that run to and from the tube tip through a thermostatic line. The
entire flow tube has a volume of ca. 4 mL. The reaction mixture is introduced through
the 5 mL injection loop, and it flows through the reactor to the NMR detection system.
The flow stops during the NMR experiments. Note that for photochemical reactions,
the reactivity is restrained to the section of the capillary exposed to light, and
this allows the accurate estimation of the residence time even with a long capillary
connecting the flow reactor to the analytical system ([Fig. 3 ]). Reaction monitoring provides information on the reaction progress and kinetics
and is crucial for mechanistic understanding and optimization.
Fig. 3 Bazzoni et al's experimental setup.[27 ]
For rapid reactions, microreactor NMR probes can be used to determine reaction kinetics
at different temperatures.[21 ] Scheithauer et al used a novel microreactor NMR probe that combines the advantages
of online flow 1 H NMR spectroscopy and microreaction techniques.[28 ] The kinetic model of the reaction was developed, and the rate constants were fitted
to the new experimental data. The kinetic model developed in this study described,
for the first time, the kinetic effects of acetaldehyde and water over a wide range
of temperature and pH values.[28 ]
[29 ]
[30 ]
Sagmeister et al demonstrated the use of inline benchtop NMR to optimize a complex
nitration reaction in flow, using a multivariate analysis statistical approach for
data processing.[31 ] They accurately quantified four overlapping species, which enabled the generation
of a robust design for an experiment model with the accurate evaluation of dynamic
experiments ([Fig. 4 ]). Even if the data-acquisition period is very short (2.0 seconds), the advantages
of benchtop NMR in flow-chemistry automation optimization, mechanism experiment, and
process control were realized.
Fig. 4 Sagmeister et al's inline benchtop NMR in a complex nitration reaction in flow.[31 ] (A ) Reaction scheme and the representative low-field NMR spectrum. (B ) Continuous flow chemistry. NMR, nuclear magnetic resonance.
Rubens et al developed an automated polymer-synthesis platform based on an inline
low-field NMR spectrometer.[32 ] By monitoring the monomer conversion rates over the residence time range of the
continuous reactor, the platform could construct accurate and efficient polymer kinetic
curves. The machine-assisted self-optimization procedure allowed the reaction to be
stopped at any given preselected conversion rate, resulting in unprecedented polymer-synthesis
reproducibility. Additionally, they proposed an automated synthesis platform based
on continuous flow that can rapidly screen polymerization reactions. The platform
used inline monitoring to obtain real-time analysis data. The software was developed
to guide data acquisition and, most importantly, to enable the autonomous performance
of reactions and analysis. Further algorithms could automatically detect errors in
the experiment and purge the wrong data. The data were aggregated and made directly
available in a machine-readable manner, enabling the creation of “big data” for dynamic
information that was not subject to individual user biases and system errors ([Fig. 5 ]).[33 ]
Fig. 5 Automated polymer screening platform.[33 ]
Monitoring and Controlling the Chemical Synthesis Process
Inline IR can also be applied for the precise addition of reagents in continuous flow.
Lange et al developed a method to precisely control the addition of reagents in a
multistep operation.[34 ] Using inline IR monitoring and new LabVIEW software, additional pumps could be controlled
to distribute more reagents in real-time, depending on the concentration of the reaction
intermediates. This enabled accurate mixing at perfect times (which significantly
improved the product quality) and the piecewise chemical-flow processing of extended
reaction sequences.[34 ]
[35 ]
The biggest problem with esterification is the reversibility of the reaction in most
cases. Schnoor et al used inline IR for calibration during continuous esterification.[36 ] The full potential of inline attenuated total reflectance–FTIR spectroscopy of continuous
reactor units was revealed by introducing a new calibration method that bridges the
gap among continuous synthesis, measurement, calibration, and analysis on a bench
scale ([Fig. 6 ]). Further, fully automatic reaction mixture calibration under process conditions
was achieved. The use of different components highlights the robustness of the calibration
method in determining the kinetic parameters. Transform-based calibration significantly
improves the precision of the analysis. The difference is that the new method requires
only two formulated calibration mixtures. In this way, this method can also be applied
to large-scale processes, and it allows the use of a single measurement setup from
the laboratory to the industrial scale.[36 ]
Fig. 6 Schnoor et al's inline IR for calibration in continuous esterification reaction.[36 ] IR, infrared.
Marchand et al utilized a fast and mobile-compatible diffusion NMR experiment that
enabled the collection of accurate diffusion data for samples with flow rates of up
to 3 mL/min.[19 ] This assay was employed to monitor the continuous-flow synthesis of Schiff bases
with a temporal resolution of approximately 2 minutes. Wu et al conducted operational
NMR studies of anthraquinone/ferrocyanide-based redox flow batteries (RFBs) on a low-cost
and compact 43 MHz desktop system,[37 ] revealing the reactive intermediates and crossovers in reduction–oxidation cells.
The operating-table NMR approach is expected to be widely used for flow electrochemical
studies in different applications, including RFBs, carbon dioxide capture and utilization,
ammonia synthesis, desalination, and organic electrochemical syntheses.
Reaction Optimization of Macro- and Microreactors
Since combining inline monitoring with a platform for automated reaction optimization
can improve efficiency, Fath et al developed an enhanced autonomous microfluidic reactor
platform for organolithium and epoxide reactions that continuously combined inline
FTIR spectroscopy with inline mass spectrometry ([Fig. 7 ]).[16 ] The self-optimization platform achieved model-free autonomous optimization without
manual intervention and was employed to determine the optimal reaction conditions
for organic synthesis involving complex reaction mechanisms. Inline FTIR measurements
were used to monitor the main components of the reaction, whereas the highly sensitive
mass spectrometry provided insights into the formation of by-products. Unlike the
case in previous studies, no chromatographic separation was performed before the mass
spectrometry; thus, greatly accelerating the analytical process. This research presented
a new method for solving complex multidimensional optimization problems and maximizing
product yield and purity.[16 ]
Fig. 7 Fath et al's microreactor setup for organometallic synthesis with n -butyllithium.[16 ]
Simon et al improved the efficiency of the multiphase catalytic hydrogenation reaction
using inline IR spectrometry.[38 ] One of the main potential disadvantages of heterogeneous catalytic hydrogenation
is substrate failure, which leads to catalyst deactivation. Functional/heterocyclic
tolerant multiphase hydrogenation reactions could be evaluated efficiently and rapidly
using automated continuous flow and real-time analysis platforms. Simon et al used
inline FTIR and inline ultra-HPLC (UHPLC) as orthogonal analytical methods to rapidly
acquire and quantify major chemicals (substrates, products, and all additives) using
a mobile platform ([Fig. 8 ]). Therefore, changes in the reaction results and the stability of additives to the
reaction conditions were evaluated. In particular, real-time quantification of chemicals
by FTIR using advanced data-analysis models (partial least squares regression [PLS])
helped them to generate quantitative data early in development and understand the
sensitivity to reaction conditions with different functional groups and heterocycles.
The method was applied to the reduction of nitrobenzene compounds with a heterogeneous
catalyst (Pd/Al2 O3 ) and hydrogen ([Fig. 9 ]). The experimental workload was significantly reduced.[38 ]
Fig. 8 Simon et al's flow reactor and analytics setup in the hydrogenation reaction.[38 ]
Fig. 9 Simon et al's continuous flow and real-time analytics platform for catalytic reaction
screening.[38 ]
Van den Broek et al used a microreactor with inline IR analysis for the Vilsmeier–Haack
formylation ([Fig. 10 ]) to produce thermally labile intermediates in continuous flow ([Fig. 11 ]).[39 ] The main drawback encountered in using UV to measure the formation of unstable intermediates
was that real-time analysis could not be conducted without using additional solvents,
such as acetonitrile. This drawback was overcome by connecting the Mettler Toledo
FlowIR IR flow cell to the microreactor outlet. The data obtained from this inline
IR analysis provided insights not available from other sources, where the reaction
time, temperature, and molar ratio were easily optimized as process parameters. Thus,
the reaction scale was expanded, and the continuous production of 2-formyl pyrrole
was successful, with a yield of 5.98 g/h. Similarly, inline IR can be used to monitor
the preparation of standard reagents, which is easier than titration or gas chromatography/NMR
techniques. In particular, inline IR combined with flow chemistry can optimize the
reaction conditions and expand the reaction scale for the formation of format reagents.[40 ]
[41 ]
Fig. 10 Vilsmeier–Haack formylation of pyrrole.
Fig. 11 Schematic drawing of the microreactor setup.
Henry et al performed the thermal decomposition of alkoxy alkyne under flow conditions
and subsequently captured amide and ester with amine and alcohol.[42 ] Using inline IR monitoring of temperature and flow-rate changes, the process of
alkoxy alkyne conversion into amide was monitored. The conversion rates of EtO, iPrO,
and tBuO alkoxy alkyne were greater than 95% in the reaction time of 10 minutes.
Rueping et al studied the asymmetric organocatalytic hydrogenation of benzoxazine,
quinoline, and hydrogen in italic compounds in a continuous-flow microreactor.[43 ] The combination of the ReactIR flow cell and microreactor as the inline monitoring
equipment optimized the reaction parameters rapidly and easily. The reduction effect,
yield of the separated product, and enantiomer selectivity were all improved.
Schotten et al optimized the continuous-flow synthesis of difluoromethyl trimethyl
silane (TMSCF2 H) with the Ruppert–Prakash reagent (TMSCF3 ) using inline real-time NMR and temperature measurements ([Fig. 12 ]).[44 ] These measurements were used to maximize the spatiotemporal yield to ensure the
safety of the heat-release process. Thus, the spatiotemporal yield was three times
that of the reported batch procedure, with 25 g of pure TMSCF2 H isolated after 105 minutes.
Fig. 12 TMSCF2 H synthesis. TMSCF2 H, difluoromethyl trimethyl silane.
Qualitative and Quantitative Analysis of Compounds
In addition, to its qualitative function in continuous-flow synthesis, inline IR can
be used for quantitative analyses in continuous-flow synthesis.[45 ] Rao et al used IR spectroscopy to understand the processes of ketonization and sulfonation
to form tetrasubstituted acyclic olefins.[46 ] Ketene formation and sulfonation were monitored with ReactIR in a two-step one-pot
reaction. A univariate quantitative model was established based on the peak height
of the ketone–carbonyl IR spectrum of raw materials. The model was used to determine
the starting material consumption rate and the endpoint for various reaction conditions.
For the second step of the reaction, the vinyl sulfonate formation step, the entire
IR spectrum was analyzed, and offline HPLC data were collected to measure the ratio
of the E to Z tetrasubstituted methyl ethyl sulfonate products. A multivariate PLS model was developed
using principal component analysis and PLS regression to quantitatively predict the
relative contents of the E and Z stereoisomers in the new reaction.
The use of NMR spectroscopy to obtain both quantitative and qualitative information
was particularly beneficial for the study of complex reaction systems containing multiple
components. Maiwald et al used 13 C NMR quantitative spectroscopy to study the chemical equilibrium of a formaldehyde–water–methanol
ternary liquid mixture.[47 ] Their study considerably expanded the range of data on the composition of these
complex mixtures, such as the distribution of formaldehyde among different species.
Experiments conducted independently in two laboratories showed that NMR spectroscopy
could provide systematically reliable quantitative data, with an absolute error generally
less than 0.5%.
Conclusion
The combination of continuous-flow and real-time process analyses considerably improves
chemical synthesis; however, this approach is fraught with challenges, such as the
summit overlap generated in IR or Raman spectra of different compounds with similar
structures. Overlapping peaks hinder the quantitative analysis of the spectra, and
extensive and cumbersome calibration procedures are often required for all components
over a wide concentration range. The ability of inline desktop NMR is occasionally
limited by its low spectral resolution, which also leads to the abovementioned problems.
The disadvantages of NMR spectroscopy are its low sensitivity (i.e., low detection
limit) and low signal-acquisition rate. In most cases, the increase in the signal-acquisition
rate comes at the expense of sensitivity. To minimize the influence of the detector
on the concentration distribution, the benchtop NMR instruments currently in the market
are limited to operating within a narrow temperature window close to the operating
temperature of the magnet, and the instruments are used at 28°C. Therefore, active
temperature control of compact NMR probes is very promising in the field of reaction
monitoring; however, it is yet to be studied and developed. In addition, to continuous-flow
reaction monitoring, inline IR and inline NMR spectroscopy can be adopted to monitor
photoelectric reactions in complex biological fluids[48 ] and material changes[49 ] or to safely detect[50 ] amino acid activity for obtaining proof of polymerization behavior,[51 ] among many areas.
Typically, fast process analyses, such as spectroscopy techniques (NMR, FTIR, and
Raman) are used for dynamic experiments. HPLC, which runs for 10 to 30 minutes, is
usually too slow to capture the dynamics of the system. However, it can be used to
confirm and validate experimental points quickly. In addition, HPLC instruments have
been used as offline PAT devices in combination with a fraction collector or as an
inline process-analysis device with special interfaces that allow samples to clump
together before analysis. UHPLC (run time ≤5 minute) may offer a good balance between
fast data acquisition and sensitivity to trace impurities. These instruments can provide
fast results to control systems (often <5 seconds per data point) to make data-driven
(automated) decisions and control processes in real-time. Although online or inline
chromatography provides information on impurity profiles and trace impurities, long
measurement times between mixing and first data acquisition are required due to the
separation of species. Therefore, more appropriate data-processing methods are required.
The introduction of algorithm-assisted synthesis and machine learning is currently
driving fundamental evolution in the way laboratory procedures are conducted. Although
human intuition can only partially predict optimal conditions, these computer-based
methods can significantly improve the accuracy and precision of chemical syntheses.
Automation and digitalization provide opportunities to respond more rapidly to future
challenges. However, chemists cannot accomplish this transformation alone; therefore,
we must be open to collaborating with other disciplines to make the leap and go digital.
