Continuous-Flow Chemistry: From the Wavelength Dependence of Photoreactions to Spatial Confinement Effects in Olefin Metathesis

Continuous-flow chemistry investigations are becoming increasingly important for organic synthesis, as they offer several advantages over classical synthesis in a flask. In order to investigate reactions in continuous-flow, a setup consisting of two linked HPLC systems has been established, whereby...

Full description

Saved in:
Bibliographic Details
Main Author: Roider, Thomas
Contributors: Tallarek, Ulrich (Prof. Dr.) (Thesis advisor)
Format: Doctoral Thesis
Published: Philipps-Universität Marburg 2022
Online Access:PDF Full Text
Tags: Add Tag
No Tags, Be the first to tag this record!
Summary:Continuous-flow chemistry investigations are becoming increasingly important for organic synthesis, as they offer several advantages over classical synthesis in a flask. In order to investigate reactions in continuous-flow, a setup consisting of two linked HPLC systems has been established, whereby the first HPLC system is used for reaction control (first dimension) and the second for the analysis of the reaction solution (second dimension). The reactor on which the reaction takes place is placed in the first dimension. In this dimension, the flow rate and thus the reaction time on the reactor can be adjusted. A 2-position/6-port valve located behind the reactor connects the two systems and allows to transfer a part of the reaction solution from the first to the second dimension. In the second dimension, the separation of the substances takes place and their quantification is made possible via an external calibration. Therefore, a higher reproducibility is obtained and the employee is freed from labor-intensive, repetitive steps, which means he may use the time to evaluate and interpret the results or plan further investigations. In this thesis, two areas are presented in which continuous-flow can be used efficiently. The first area deals with photochemical catalysis as well as photochemical reactions in general. Here, a catalyst (or the substrate) is excited by light of a certain wavelength, which starts a chemical reaction. Synthesis in continuous-flow allows to use reactors which have a very small channel diameter and thus a low penetration depth for the light. This results in an almost homogeneous irradiation of the reaction solution, which makes it possible to better control the reaction and minimize the formation of by-products. Due to the progress in the development of LED technology, photochemical reactions have also gained in importance, as versatile light sources are now available that require low energy and are inexpensive to purchase. Many photochemical reactions can also be carried out with sunlight, which has made them very attractive in terms of green chemistry. In photochemical investigations, however, the extent to which the wavelength of the emitted light affects the reaction, or how the reaction can be influenced by irradiation with light of a different wavelength, is often neglected when investigating photochemical reactions. Chapters 1 and 3 deal with the investigation of photochemical reactions regarding their wavelength dependence, whereby chapter 2 is a direct follow-up project from chapter 1, which was made possible by the good interaction between reaction control and instrumental analysis. The second area which is very well suited for continuous-flow investigations concerns heterogeneous catalysis. Here, a catalyst is immobilized on a carrier material (usually silica particles) and the material is packed into a column. This column represents the reactor. Since no further reaction takes place once the solution has left the reactor, heterogeneous catalysis is very well suited for investigations in continuous-flow. In this case, the reaction does not have to be quenched, nor does the catalyst have to be separated, as it is only stationary in the reactor, which also allows the reaction time to be controlled very well via the flow rate applied to the reactor. A major problem in heterogeneous catalysis lies in catalyst poisoning, whereby the activity of the catalyst and thus the conversion of the reaction decreases when the catalyst is used several times. In continuous-flow, different reaction conditions can be investigated relatively quickly in succession with the same reactor and the influence of catalyst poisoning can be minimized. In chapters 4 and 5, olefin metathesis reactions with a Grubbs catalyst are investigated in continuous-flow. Here, the catalyst is immobilized only within the pores (60 Å inner diameter) of the silica, allowing the influence of the spatial confinement of the pore on the chemoselectivity of the reaction to be investigated. The individual chapters of this thesis are briefly summarized below. Chapter 1 lays the foundation for the systematic investigation of photochemical reactions on the dependence of the wavelength of the irradiating light. For this project, 16 identical LED arrays are being developed, with 12 LEDs on each array emitting light in the same wavelength range. These arrays are used to cover the range between 365 and 670 nm in the smallest possible steps to achieve the highest possible resolution when studying the wavelength dependence of the reaction. In addition, a microreactor is being developed that is well matched to the LED arrays so that the light can be used efficiently and investigations at high light intensities are possible. Furthermore, a small channel depth (0.5 mm) was chosen in order to obtain the most homogeneous irradiation possible and to keep the reaction times as short as possible. As an example reaction, the perfluoroalkylation of 2-methylindole with eosin Y as photoredox catalyst and diazabicycloundecene (DBU) as base is shown. The reaction has only been investigated with white light in the literature. In this project, the entire wavelength range of visible light between 365 and 700 nm is investigated to determine possible differences in the reaction. An important point of these investigations, which also sets this work apart from other wavelength investigations, is the calibration of the different LED arrays. A method is developed which makes it possible to normalize the number of photons hitting the reactor using a spectroradiometer. In this way, only the wavelength of the photons changes during the examination, but the number of photons hitting the photoreactor remains almost the same (deviation of approx. 3%). This excludes the possibility that an LED array delivers better or worse results due to a higher or lower number of photons. In the range between 430 and 550 nm, the conversion of the reaction correlates with the absorption spectrum of eosin Y. However, it is noticeable that the highest conversions are obtained in the wavelength range below 400 nm, in which eosin Y shows no absorption. In a further investigation without a catalyst, the results could be reproduced, although the conversions in the range in which eosin Y absorbs were absent. In a third investigation, which is carried out without a base, the same trend can be observed, albeit with significantly lower conversions. In addition, there is a change in the regioselectivity of the reaction (between the 3 and 4 position), which suggests that different reaction mechanisms are involved. Thus, three different reaction mechanisms can be selectively controlled by the choice of the reactants and the wavelength. In the wavelength range between 430 and 550 nm, the reaction takes place via photoredox catalysis of the eosin Y catalyst. In the wavelength range below 430 nm, the electron donor acceptor (EDA) complex between the base DBU and perfluorobutyl iodide dominates, which generates free perfluoroalkyl radicals by photoactivation. The third mechanism is an EDA complex between 2-methylindole and perfluorobutyl iodide, which is much weaker than the EDA complex between DBU and perfluorobutyl iodide. Here, photoactivation generates a radical pair in the solvent cage, which also influences the regioselectivity of the reaction. Thus, three different reaction mechanisms were revealed by the systematic investigation of the wavelength dependence of the reaction. The second chapter arose directly from the thematic of the first chapter. It was already observed in the first chapter that if the product mixture of 1-methyl-3-(perfluorobutyl)-1H-indole and 1-methyl-4-(perfluorobutyl)-1H-indole is stored for a longer period, hydrolysis of 1-methyl-3-(perfluorobutyl)-1H-indole occurs at the α-position of the perfluorinated group, whereas 1-methyl-4-(perfluorobutyl)-1H-indole is stable. This observation is very unusual, as the C–F bond is a very stable bond that can usually only be activated by harsh reaction conditions. Another great advantage of the interaction between reaction control and instrumental analysis is that all data are recorded and stored. Thus, with the help of the data from chapter 1, it was discovered that the hydrolysis product was already formed in a very small proportion (less than 1%) in all investigations with DBU as the base. The low proportion is explained by the fact that HPLC solvents were used in the investigations, which have only a low water content. The reaction could be reproduced with a base (sodium hydride) and with the addition of water in a flask, whereby a yield of 37% was obtained over two steps. The electron-withdrawing group on the 2-methylindole allows the amino function to be deprotonated, resulting in the cleavage of F– and the addition of a nucleophile at the α-position of the perfluorinated group. By using sodium ethanolate as a base, the ethanolate group can also be introduced into the molecule. Analogous reactions can also be carried out with 2-phenylindole instead of 2-methylindole, with slightly higher yields of 44% to 54% over two steps. The reaction can also be carried out with heptafluoro-2-iodopropane, a secondary perfluorinated iodoalkane (49% over two steps). In this case, however, the alcohol and not the ketone is obtained as the hydrolysis product. A method could be developed which allows the indirect C–F activation at the α-position of a perfluorinated side chain at the C(3) position of 1H-indoles under mild reaction conditions. Here, the 3-substituted 2-phenyl indoles should be highlighted, which represent promising structures for medicinal chemistry. In the third chapter, which is linked thematically to the first chapter, the setup for the investigation of photochemical reactions regarding their wavelength dependence and the intensity of the light source is completed. For this purpose, the setup is extended by a self-developed module that enables the LED arrays and thus the wavelength irradiating the reactor as well as the intensity of the light source to be changed automatically. The module consists of a linear drive on which a carriage is mounted that can carry up to 10 LED arrays. However, the LED arrays can be interchanged without any problems. The LED array that is in operation is always located directly under the microreactor. A software is used to control which LED array is active at any given time for any given duration. The intensity of the LED array can then be changed, or another LED array can be used for the next examination. Thus, in combination with the HPLC equipment, both the reaction control and the analysis as well as the quantification of the compounds formed from photochemical reactions can now be investigated in a fully automated manner regarding their wavelength dependence and the intensity of the light source, resulting in a very time-efficient recording of reaction data. The reaction of an arylazo sulphone with the solvent (water/acetonitrile 1/9 (v/v)) as well as with a dioxaborolane as an additive is used as a test reaction for the system. The reaction is very promising as it has already been studied in the literature for its wavelength dependence and a different ratio between the products has been obtained at different wavelengths. However, the reaction has only been studied at two different wavelengths (366 nm and 450 nm) and with white light at different intensities and irradiation durations. In this chapter, the reaction is systematically investigated in the wavelength range from 373 to 522 nm (12 LED arrays). Since three different products are to be formed by irradiating the substrate in the solvent, the additive was initially omitted in order to keep the reaction simpler. In the investigation under inert gas, only methoxybenzene is formed as a product over all wavelengths (max 60% at the shortest wavelength). However, during the reaction in air atmosphere, another product (4-methoxybenzenediazonium methanesulfonate) can be observed, which is formed by the reaction with air. This compound is difficult to observe because the absorption spectrum of the product is very similar to that of the substrate and it is also converted to the same products. The absorption spectrum of the intermediate product is shifted with a maximum of about 50 nm into the shorter wavelength range. This makes it possible to stop the reaction at the intermediate stage by using an LED array with which only the substrate but not the intermediate is excited. This enables to use the intermediate as substrate and to investigate the reaction for the dependence of the wavelength. Here, a significantly higher proportion of 4-methoxyacetanilide was produced (methoxybenzene 45%, 4-methoxyacetanilide 15%). The 4-methoxyacetanilide is thus only formed from the intermediate and not directly from the substrate. The question which arises subsequently is whether the reaction with the dioxaborolane is also influenced by the intermediate. The investigation under inert gas and in air atmosphere lead to very similar results (45% product, 22-28% methoxybenzene), which is due to the fact that the reaction with the additive proceeds much faster than the formation of the intermediate. With the pre-generated intermediate, on the other hand, only 38% product, 17% methoxybenzene and 3-4% of the 4-methoxyacetanilide are obtained. However, the conversion is also somewhat lower because the reaction proceeds more slowly. In addition, it is very noticeable that the yield decreases at higher wavelengths across all investigations. However, the yield decreases much more slowly when the intermediate is used. At a wavelength of 430 nm, 30% product is still generated with the intermediate, whereas only 17% product is generated with the substrate at 430 nm. Thus, the path via the intermediate is particularly advantageous for reactions in which the higher energy of light at short wavelengths leads increasingly to by-products or to the decomposition of the substrate. In summary, this study presents a setup that can investigate a photochemical reaction under three different reaction conditions (inert gas, in air atmosphere, with pre-generated intermediate product) within nine hours in a fully automated manner for its dependence on the wavelength of the incoming light. During these investigations, an intermediate which shows a completely different reactivity than the substrate was selectively generated by carefully choosing the wavelength used. It could thus be shown that the wavelength of the light source used should be given significantly more attention and that even a small difference in the wavelength of the light used can have a major influence on the reaction. Chapter 4 leaves the field of photochemical investigations and enters the field of heterogeneous catalysis. For this purpose, an olefin metathesis reaction using a Hoveyda–Grubbs catalyst is investigated, in which a diene is used as the reactant. The desired product results from the intramolecular reaction in which a ring is formed. However, an intermolecular reaction can also occur as a competing reaction, resulting in a dimer, trimer, or oligomer from the substrate. It is an established method to work with very low concentrated solutions to favor the intramolecular reaction over the intermolecular reaction. However, this has the disadvantage that a large amount of solvent must be used to produce a small amount of product. In this project, a different approach is taken to prioritize ring closing over oligomerization. Since the dimer requires significantly more space to be formed, the catalyst is immobilized only within the pores of silica particles (60 Å inner diameter), thus spatially confining the reaction and favoring the formation of the ring. The silica particles are packed into a column on which the reactions are studied in continuous-flow, at different temperatures and different flow rates, which lead to different reaction times. Similar trends can be seen at all temperatures. As expected, the conversion decreases with lower reaction times. The course of the selectivity, which indicates the ratio between the product of ring closing metathesis and the oligomers, is clearly more complex. Both ring closing metathesis and oligomerization release ethene, which could quickly escape if the catalyst were freely accessible in a flask and thus would not have much influence on the equilibrium. In this case, however, the catalyst is present in the pores and the column is sealed (except for the inlet and outlet), which means that the ethene releases much more slowly and must be included in the equilibrium. The ethylene can reopen the ring that has already been formed, which leads to a lower selectivity. In addition, the dimer can react with itself, whereby the desired product is also formed. However, this reaction takes place relatively slowly, since the dimer must first be formed. These processes can also explain the course of the selectivity in the experiment. In the range where the yields are higher (above 2%), the selectivity increases with increasing flow rates. This is because at higher flow rates the ethene can be removed more quickly, limiting the back reaction (the ring opening). At low conversion rates (below 2%), only small amounts of ethylene are generated, so the back reaction does not have much influence on the selectivity. In this range, the selectivity increases with decreasing flow rate (increasing reaction time), as the dimer now has enough time to react with itself and form the desired product. Under spatial confinement, higher selectivities can also be achieved compared to the reaction in the flask (selectivity of 57% instead of 47%). In addition, significantly smaller oligomers are formed in spatial confinement. With MALDI-TOF it was shown that only the dimer and trimer are formed (95% to 5%), whereas with the freely accessible catalyst in the flask oligomers up to the hexamer are observed. In summary, the study showed that the spatial restriction has advantages for the selectivity of the reaction. Furthermore, the influence of the ethylene formed on the mechanism was observed and reaction conditions were found at which the selectivity of the reaction is maximal (approx. 60% selectivity at 2% conversion). The fifth chapter also deals with olefin metathesis under spatial confinement. In this project, however, the focus is not on the selectivity between the product from ring closing metathesis and the oligomers, but on the investigation of another side reaction, the formation of isomers. For this investigation, the same catalyst as in chapter 4 is used (Grubbs–Hoveyda), which is also known to form isomers from the reactant. This project investigates which isomers can be formed during the reaction and how the formation of the isomers can be influenced. In contrast to chapter 4, this time a reactant is used that contains an aromatic system. Thus, both the reactant and all the resulting products can be tracked and quantified online with a DAD. In addition, a mass spectrometer (ESI-MS) was connected, which helps to obtain an overview of the by-products formed. Catalyst poisoning plays an important role for isomerization. First, the active ruthenium catalyst reacts with ethylene to form hydrides, which in turn are responsible for isomerization. For isomerization to take place, enough ethylene must first be produced in order to form the hydride species. Therefore, no isomers are observed at higher flow rates and lower reactant concentrations. In order to determine the exact influence of catalyst poisoning on isomerization, a reactor is examined for 200 minutes at a reactant concentration of 100 mM and the lowest flow rate (0.01 mL/min or 18.6 minutes reaction time), which can be set by the HPLC pumps. During this time, 18 measuring points are recorded and three different phases of the reactor are detected. In the first phase up to 100 minutes, the selectivity is very high at the beginning and decreases slowly. The reason for this is that the hydride species must first be formed before the isomerization reaction can take place. In the second phase of the reactor between 100 and 145 minutes, the selectivity is constant. In this range, the formation and consumption of hydride species is in equilibrium. After 145 minutes, the selectivity increases rapidly because the conversion is lower due to catalyst poisoning and not enough ethylene is formed to form enough hydride species. These results fit very well with the results from chapter 4, where the best selectivities were also achieved when the activity of the catalyst was very low. In summary, this project has shown that spatial confinement has not only advantages, but also that a higher isomerization rate can be expected. Furthermore, the isomerization products are very versatile, as the isomer of the substrate can also undergo ring-closing metathesis or the formation of oligomers. Finally, with the help of NMR experiments, it was shown that NMR analysis from reaction mixtures of olefin metathesis, is not well suited for the determination of conversions and selectivities, since the signals of the isomers can overlap with the signals of the product, and NMR analysis from a reaction mixture thus provides false results. Nevertheless, this is currently a widely used method, which should be reconsidered. The 2D-LC/MS system shown in chapter 5 would be much better suited for this purpose. In conclusion, this work has shown how great the influence of wavelength on photochemical reactions can be. Even small differences in wavelength can lead to significantly different results. In order to ensure the reproducibility of a photochemical investigation, it is therefore important to sufficiently characterize the light sources used. For this purpose, an emission spectrum should be provided at least for each light source used. It is not sufficient to specify the color of the light, as different results can be obtained depending on the wavelength range. In the field of olefin metathesis with Grubbs catalysts, it was shown that more attention should be paid to catalyst poisoning. It does not only influence the conversion of the reaction, but directly interferes with the equilibrium of the mechanism. Therefore, it is also not trivial to record kinetics, because they would be falsified by catalyst poisoning if it cannot be suppressed or significantly limited. In addition, with long reaction times in the flask, a result is always obtained over all phases of catalyst poisoning and no phase can be specifically targeted. Overall, it was shown that investigations in continuous-flow do have their advantages. Especially in the area of photochemical investigations and heterogeneous catalysis, measurement points can be recorded much more efficiently, which means that results can be obtained more quickly and mechanisms can be resolved that would have remained hidden in the classical synthesis using a flask.
Physical Description:240 Pages