Introduction
The synergy between continuous flow reactors and benchtop Nuclear Magnetic Resonance (NMR) spectroscopy has attracted significant attention in recent years due to its transformative impact on chemical synthesis and analysis. Continuous flow reactors offer distinct advantages, including precise control over reaction parameters, enhanced safety, and improved reaction efficiency. Conversely, benchtop NMR spectroscopy provides non-destructive, quantitative insights into reaction kinetics, intermediates, and final products. This integration enables real-time monitoring of the conversion as a function of the reactor parameters, facilitating rapid and even automated reaction optimization. The benefits of this hybrid approach are evident in the accelerated processes of reaction discovery and optimization. Researchers can utilize real-time NMR data to make on-the-fly adjustments to reaction conditions, resulting in enhanced yields and selectivity. Furthermore, in-line analysis reduces the need for laborious sample preparation and offline analysis, thus streamlining workflows and minimizing waste.
In this application note, we have integrated a Spinsolve 80 ULTRA benchtop NMR spectrometer from Magritek at the outlet of an H-Cube® Pro flow reactor from ThalesNano. The Spinsolve ULTRA model is a compact spectrometer that offers high-resolution capabilities, crucial for acquiring spectra of products dissolved in protonated solvents. The H-Cube® Pro, on the other hand, offers multiple advantages, including enhanced safety, precise reaction control, high efficiency, reduced catalyst loadings, and broad substrate compatibility. Its automation capabilities make it possible to streamline hydrogenation reactions in a versatile manner, making it a valuable tool in chemical synthesis and research. To showcase the benefits of combining these technologies, we optimized the reaction parameters (temperature, pressure, equivalents of H2, and liquid flow rate) to maximize the hydrogenation of Cinnamyl alcohol (Figure 1). The results demonstrate that by integrating a rapid online analytical tool, such as NMR spectroscopy, the full spectrum of reaction parameters can be thoroughly explored within a few hours to optimize the reaction.
Figure 1: Hydrogenation of the double bond of Cinnamyl alcohol.
Experimental Setup
In Figure 2, a photograph of the setup installed within a fume hood is presented. For these experiments, an H-Cube® Pro from Thales Nano was equipped with an HPLC pump to precisely set the flow rate of the starting material solution. The catalyst cartridge used was the 10% Pd/C cartridge from ThalesNano. Hydrogen gas was generated via controlled electrolysis by the H-Cube® and subsequently mixed with the substrate solution before entering the catalyst. Any unreacted hydrogen gas remaining after passing through the reactor was efficiently separated at the outlet of the H-Cube® using a Zaiput “SEP-10” gas-liquid phase separator. The reaction kinetics were continuously monitored using 1H-NMR spectroscopy with solvent suppression using a Spinsolve Ultra 80 MHz Multi-X Benchtop NMR spectrometer equipped with a reaction monitoring kit 2 (RMK2). This kit includes a glass flow cell that spans the spectrometer from top to bottom enabling the seamless pumping of the reaction mixture through the Spinsolve’s sensitive volume in either continuous or stop-flow mode. In this study, we connected the output of the reactor to the flow cell and monitored the H-Cube Pro in continuous mode, with the flow rate being easily controlled using the HPLC pump. As starting material, we prepared a solution of cinnamyl alcohol in methanol at a concentration of 0.5 Molar (22.487 g / 167.59 mmol in 335 ml). The reaction mixture was monitored every 2 minutes by acquiring a 1D 1H WET sequence with 4 scans and a 10-second repetition time. The WET sequence was configured to suppress the two methanol signals at 3.3 ppm and 4.9 ppm, and it included the capability to collect the signal in the presence of carbon decoupling, effectively eliminating carbon satellites from the spectrum.
Figure 2: Picture of the setup that combines an H-Cube® Pro flow reactor with a Spinsolve 80 ULTRA for online reaction monitoring. The Spinsolve is equipped with a reaction monitoring kit that includes a glass flow cell (1) that makes it possible to circulate the reaction mixture through the NMR spectrometer to measure it on-line. The flow reactor produces the H2 required for the hydrogenation in a safe mode by electrolysis and mixes it with the solution at (2). A bubble detector (3) confirms the presence of H2. Then the starting material finally enters in the column heater containing a CatCart® catalyst at (4).
Results
The success of online monitoring by NMR spectroscopy relies on the high performance of the solvent suppression method, which attenuates the large solvent signals that would otherwise overlap with the smaller signals from the starting material and products in the sample. Figure 3 compares the spectra acquired in continuous flow mode, showing the standard proton spectrum alongside a spectrum obtained using the WET solvent suppression. The overlaid spectra clearly demonstrate how the WET suppression significantly reduces the methanol peaks, eliminating overlap with the CH2 signal at 3.5 ppm and simultaneously removing the 13C satellites of the CH3 of methanol.
Figure 3: Comparison of the spectra acquired for the product with a standard pulse and acquire sequence (blue) and a WET suppression with carbon decoupling sequence (red). The measurements were performed under continuous flow conditions with a liquid flow rate of 1 mL/min.
To optimize the reaction conditions, various parameters (temperature, pressure, hydrogen amount, and liquid flow rate) were systematically varied, and spectra were continuously acquired to identify the optimal parameter combination. Figure 4 offers a subset of spectra collected during incremental variations in temperature and pressure. The signal regions corresponding to specific protons are color-coded. Protons from the double bond and the CH2 group of the starting material are marked in red and yellow, respectively, while CH2 protons of the product are highlighted in light blue, dark blue, and green. The graph indicates that, with increasing pressure and temperature, the signals of the starting material decrease while the signals of the product increase. To quantify reagent concentrations, the signal regions were integrated and multiplied by a single response factor to obtain absolute concentration values.
Figure 4: Stack plot of the spectra acquired with the Spinsolve RMX software during an optimization run of the double bond reduction of cinnamyl alcohol. The data was collected with a WET sequence set to suppress the signals of methanol at 3.3 and 4.9 ppm. To eliminate the carbon satellites, the signal was acquired in the presence of carbon decoupling.
The concentrations obtained for the marked regions in Figure 4 are plotted against time in Figure 5. The left graph displays concentrations for the starting material, while the right graph shows those for the product. Each point in these curves corresponds to a spectrum in the stack plot of Figure 4, and different curves within each plot represent concentrations independently obtained from different integral regions. The excellent agreement between these curves demonstrates the accuracy and robustness of the quantification method. The minimal scatter of the points collected during each steady-state phase underscores the precision of the measurements. After altering a reaction parameter, the new steady state is typically reached within 10 minutes. Whenever a new pressure value is set on the H-Cube® (marked with a vertical line in the graphs), the signal of the starting material briefly increases before reaching a new steady state. This temporary increase occurs because the hydrogen flow is paused while the system adjusts the reactor pressure to the new set value. Once the reaction pressure stabilizes at the new set value, hydrogen flow is resumed, and the reactor quickly attains a steady state characterized by constant conversion. The void volume of tubing between the reaction zone and the NMR spectrometer is a couple of milliliters, therefore all of the above data is accessible in real-time during the reaction, within a couple of minutes after the mixture exits the column, enabling us to directly ascertain from the NMR data whether the steady state representative of each specific setting has been achieved or not.
Figure 5: Screenshot of the concentration plots of the starting material (left) and product (right) as they are displayed in real time on the Spinsolve reaction monitoring software (RMX). The different curves on each plot correspond to the concentration calculated using the signal of different chemical groups available for both, starting material and product. It can be seen that the values are in very good agreement, showing how linear the NMR response is. It should be noticed that the calibration of the system is done using an external standard with a known concentration. The same response factor is used to quantify the concentration of both reactant and product using the different signals available in the spectrum.
Figure 6 displays a subset of representative spectra collected during steady states for different temperature and pressure combinations from the optimization run depicted in Figures 4 and 5. These results indicate that, for the examined reaction, an increase in pressure significantly enhances conversion, whereas changes in temperature do not lead to a substantial increase in conversion.
Figure 6: Set of spectra collected with the reactor in steady state for different values of temperature and pressure. These measurements were performed with a hydrogen amount of 42 NmL/min and a liquid flow rate of 1 mL/min.
Figure 7: Optimization of the hydrogenation reaction run as a function of the hydrogen content for a temperature of 60 °C and a pressure of 40 bar.
In a similar manner, the impact of the conversion dependency on other reaction parameters was determined. Figure 7 shows the spectra collected as a function of hydrogen content for a temperature of 60 °C and a pressure of 40 bar. The curves reveal that a hydrogen amount of 42 NmL/min is required to achieve full conversion. Additionally, the impact of the liquid flow rate on the conversion was investigated, as depicted in Figure 8. Here, it is observed that an increase in flow rate, while keeping all other parameters constant, results in a decrease in conversion. This is an expected phenomenon due to the decreased residence time of the substrate on the column resulting from the higher liquid flow rate.
Figure 8: Conversion as a function of the liquid flow rate for a temperature of 40°C, a pressure of 40 bar, and a hydrogen amount of 42 NmL/min.
Conclusions
Overall, the advantages of a continuous flow reactor, combined with an on-line NMR spectrometer for immediate reaction monitoring were successfully demonstrated. This technique offers various advantages compared to conventional batch protocols and off-line analysis – since the steady state is achieved within 10-15 minutes after inserting new reaction conditions, a research chemist can easily analyze at least 30 different combinations of parameters in a single day, significantly facilitating and speeding up the optimization process.
We see the combination of benchtop NMR and H-Cube® as a time- and cost-saving advantage as well as an efficient and safe way to optimize a hydrogenation reaction. It is also worth highlighting that the H-Cube® and the Spinsolve spectrometer behaved very stable and robust during the whole reaction run, delivering consistent conversion values every time the same settings were chosen.
Read the Full Application Note