Tudományos folyóirat címlapján a RICHTER munkatársainak összefoglaló tanulmánya

A Reaction Chemistry and Engineering tudományos folyóirat szeptemberi számának címlapjára a RICHTER munkatársainak összefoglaló tanulmánya került a gyógyszerhatóanyagok korszerű, folyamatos üzemű előállítási lehetőségeiről.

The dynamic progress in the multi-step continuous-flow synthesis of active pharmaceutical ingredients is part of the current industrial revolution. The recently implemented examples are the forerunners of safer and more sustainable pharmaceutical manufacturing. These systems utilize interconnected flow apparatuses, integrated with in-line work-up, purification and analysis. Challenging targets call for customized, flow-oriented designs to establish novel, telescoped continuous-flow sequences. In this review we focus on these efforts towards the development of uninterrupted continuous-flow systems.

I. Introduction

Manufacturing with the highest efficiency is a fundamental objective of modern-day industry. In the pharmaceutical sector, this is motivated by the basic human desire to improve quality of living through its core value, health. The currently unfolding fourth industrial revolution (Industry 4.0) stimulates the integration of technologies to create smart, interconnected networks that meet these requirements on a global scale. Inspired by these trends, the pursuit of automation and digitalization in chemical production can lead to a more efficient, multidisciplinary approach, referred to as “Synthesis 4.0”.1 This could give spark to a “molecular industrial revolution” in the manufacturing of active pharmaceutical ingredients (APIs).2 Continuous-flow synthesis3,4 is a key enabling technology to achieve such breakthroughs,5 with great potential for safe and sustainable6–12 industrial applications.13–16 These trends are also recognized by regulatory agencies, such as the USA's Food and Drug Administration (FDA), encouraging the application of continuous-flow technologies in API manufacturing.17,18

The end-to-end continuous-flow production of APIs19–28 is based on multi-step sequences consisting of consecutive steps conducted in telescoped flow reactors. The resulting network of fluidic devices can be operated as a closed system. The flowing stream of an intermediate directly enters the next reactor, without interrupting the system with time consuming, manual work-up and intermediate purification operations. These automated systems can be operated on-demand,29,30 aided by final product purification techniques (such as continuous crystallization,31–34 simulated moving bed chromatography35,36 and centrifugal partition chromatography37), and formulation methods38–40 to produce the final dosage forms.

There is no doubt that virtually any synthetic organic transformation can be conducted in flow reactors.41 However, the implementation of a pre-existing batch synthetic route into a multistep continuous-flow processes is hardly straightforward, because telescoping is limited by the complex interplay of the several different reaction conditions, most of which was optimized to provide the best results as standalone batch steps. Innovative solutions, including the selection of reagents and reaction conditions, reactor design, as well as the strategic application of in-line analytics, work-up and purification help overcoming these issues during the integration of flow steps and the design of the continuous reactor system.42 We argue that more advanced multi-step continuous-flow systems can be achieved, if the chemical steps are designed without the constraints of the previously explored synthetic route, with only the continuous-flow realization in mind. In this review, we intend to present the current status of the flow chemical toolbox for the multi-step preparation of APIs, as well as highlight the design features, which enable the construction of these systems.

II. Flow-oriented design

The knowledge and experience gained during the decade long progress toward more and more advanced multi-step continuous-flow API syntheses can be summarized as flow-oriented design (FOD). These principles can be applied to the construction of each segment of a flow system to facilitate the prospective telescoping, both on the level of the individual reactions and the entire synthetic route.

During the implementation of a multi-step flow system, the conditions of each reaction step should be planned in advance with an eye kept on the entire synthesis. The reagents, solvents and other parameters have to be selected not only to maximize yield and purity in that particular transformation, but also to provide adequate compatibility with the other connected reactions. Moreover, the design needs to address the application of excess reagents and the formation of different byproducts, and their effects on the remaining parts of the system.

In a truly FOD approach, the original batch synthetic route may need to be revised, even by using an unprecedented route consisting of novel intermediates. Any process that hinders telescoping (by requiring long residence times, incompatible solids or off-line workup and purification procedures) should be avoided. On the other hand, there are multiple commonly utilized reactions that are well suited for telescoped flow sequences, and therefore should be preferred when designing such sequences. The safe utilization of various reactive gases can lead to efficient flow transformations.43 Continuous-flow hydrogenation offers convenient possibilities for reductions and reductive aminations as part of multi-step systems, because the crude mixtures are usually free from by-products and the excess reducing agent is simply removed by degassing the hydrogen.44–47 Amide bond formation is among the most frequent reactions in medicinal chemistry.48 Therefore, the direct transformation of amines and acids to amides using various coupling reagents in flow reactors is highly desirable for the synthesis of pharmaceuticals.49

Compatibility issues in telescoped sequences are effectively prevented by using traceless transformations,41 which do not require the use of reagents in a conventional sense. Decompositions, eliminations and intramolecular rearrangements can be achieved by thermolysis under high temperature and pressure conditions50 in flow operation. Alternatively, the emerging applications of continuous-flow photochemistry offer unique reactivity towards diverse structures.51,52 Improvements in light source and reactor design, as well as understanding of photochemical process intensification are pointing in the direction of sustainability.53,54 Similarly, continuous-flow electrochemistry is able to achieve reagentless redox processes.55,56 Depending on the electrodes, electrons can act both as oxidizing and reducing agents, replacing the commonly used, wasteful reagents. The latter two methodologies are still relatively new, only elaborated in detail as single-step processes. However, these are particularly useful as intermediate steps in long multistep flow sequences and have the potential for implementation in the multistep flow synthesis of pharmaceuticals.

Shorter routes are also advantageous, since the controllability of a system decreases with the increasing number of reaction steps and reactor modules. As a result, not only atom-economy, but also process-efficiency57 becomes priority in FOD. For instance, the safer application of high temperature and pressure conditions50 or hazardous reagents58 in flow reactors are often useful to reduce the number of reaction steps or to avoid steps that are unfeasible in flow.59,60

The issues preventing plug flow operation can also be bridged by introduction of miniaturized continuous stirred tank reactor (CSTR) cascades61,62 into the continuous-flow sequence. Finally, the inclusion of batch steps can also be considered in the rare cases, when flow conditions are not favorable. Preferably, these are placed at the beginning or the final step of the synthetic route, not to interrupt a telescoped continuous-flow sequence.

Similarly to the aims of FOD presented here, Kobayashi and co-workers have previously classified continuous-flow systems into categories based on the reagents and catalysts used.63 This system advocated the use of supported reagents and catalysts, which make removal of their excess and the formed by-products unnecessary. This way, reagent compatibility issues are eliminated, and simple telescoping is enabled. This approach was applied to the synthesis of multiple APIs, including (R)- and (S)-rolipram64 and the anti-Alzheimer's drug donepezil.65 However, the use of supported reagents generally leads to semi-continuous processes with low productivity, and therefore not generally applicable. In our opinion, FOD covers a wider range of possibilities and tools for successful telescoping, by not being restricted to limited reaction conditions.

III. Contemporary tools of telescoping multiple flow steps

In most of the multi-step continuous-flow systems, the reactions are conducted in a linear fashion, however convergent approaches are more useful, when separately constructed building blocks can be combined to afford the target compound.66–69 Divergent methods enable the production of multiple structurally similar APIs in “chemical assembly systems”.70–73 Modularity is a unique feature of such systems, which enables developing flexible process routes by the combination of pre-optimized flow units. Recently, a radial approach was demonstrated, in which the individually accessible flow reactors are arranged around a central switching station, and operated sequentially to realize multiple synthetic steps.74

In uninterrupted (or so called one-flow28) systems, the continuous-flow reactors are telescoped without the need for manual work-up and purification, or other batchwise operations. This can be achieved directly, if the pressures in the connected units are the same. Otherwise, the exiting stream of the former reactor can be directed into a small volume buffer flask, from which the unchanged mixture is continuously pumped into the next reactor. Buffer flasks play an important role in continuous production of APIs under cGMP conditions as sampling points for off-line quality control.75 Furthermore, buffer flasks may be applied for the purpose of an automated solvent exchange,76 or removal of excess gas from the liquid reaction media.77–79

If the composition of a crude mixture (containing excess reagents or incompatible by-products, which would be detrimental to the next steps) restricts telescoping, in-line work-up and purification units can be introduced between the reactions,80 in order to preserve the uninterrupted sequence of chemical steps. Several in-line methods have been developed for this purpose, such as liquid–liquid phase separation,81 evaporation82 or filtration.38 Advanced nanofiltration methods enable the effective removal and recycling of solvents and reagents which is advantageous in terms of sustainability too.83 Similar to in-line separations, scavenging columns can also be used to remove unwanted components for the reaction stream,84,85 but their operation is semi-continuous, since these columns need to be regenerated or replaced from time to time. In the aspect of in-line purification, FOD prefers the selection of reagents and solvents, which facilitate purification by providing easily removable by-products or simplifying isolation of the desired product.

Real-time analysis of the reaction mixture at multiple points of the continuous-flow system86–88 enables efficient optimization and better understanding the process during development, together with continuous control and quality monitoring during its operation. For this reason, in-line analytical methods are preferred, such as FT-IR,89–92 NIR,93,94 Raman95,96 and NMR spectroscopy.97–101

IV. Analysis of flow-oriented design features in the multistep flow syntheses of APIs

In the context of this review, multistep continuous-flow sequences of APIs are discussed (published in the last five years, listed in order of their first flow synthesis milestone), in which successful telescoping of scalable flow reactors led to an uninterrupted closed system (or can lead to in theory). The earlier flow approaches of these APIs are also discussed for comparison. Telescoped sequences (in which intermediates are not isolated) are depicted with a special reaction arrow (Fig. 1; B to C to D), in order to clearly distinguish from the ones with isolated intermediates (Fig. 1; D to E). The innovative FOD solutions enabling the telescoping of the flow steps and construction of an uninterrupted continuous-flow system are highlighted as “FOD features” at the end of each synthesis.

The complexity and productivity of the end-to-end continuous-flow approaches have been gradually rising during the last decade (Fig. 2). This can not only be attributed to the technical progress of flow reactors, but also the developments in the underlying chemistry. Similarly, the ambitions aiming for higher and higher peaks by researchers in the growing field of flow chemistry are in a healthy balance with the ever-growing interest for these innovative approaches from the pharmaceutical industry and regulatory agencies.

The flow chemistry community (consisting of researchers with both academic and industrial backgrounds) has wisely put the most urgent needs of patients to the priority, as most of the multistep flow synthesis aim the preparation of drugs on the WHO’s List of Essential Medicines102 or the top 200 best selling drugs,103 these features are also highlighted on the respective schemes. The high prevalence of drugs for infectious diseases (antivirals, antibiotics and other antimicrobials) among the discussed examples is notable. Flow chemistry efforts can spearhead the secure supply of these important drugs. In case of local outbreaks, the on-site production could be quickly established using already optimized, modular automatic flow systems, similarly to the rapid installation of temporary hospitals near the affected population.104

Ciprofloxacin (flow synthesis milestones: 2005, 2017, 2019)

Ciprofloxacin (9) is a broad-spectrum antibiotic, from the family of second-generation fluoroquinolone derivatives. It is one of the most frequently used drugs for the treatment of urinary- and respiratory tract infections.105

An early attempt for the flow synthesis of ciprofloxacin (9) was described by Schwalbe and his group in 2005, with the goal of the automated library synthesis of fluoroquinolone analogues.106 The one-pot batch procedure developed by Bayer's researchers107,108 was adapted to continuous-flow operation, with modifications aiming to avoid clogging caused by precipitation. This was achieved by using tributylamine in the first step instead of triethylamine (which forms a non-soluble hydrochloride salt), and preheating the solution of intermediate 8 (to counteract its limited solubility).

In the first step, 2,4,5-trifluorobenzoylchloride (1) was reacted with ethyl 3-(dimethylamino)acrylate (2) in the presence of tributylamine (Scheme 1). The resulting intermediate (3) was subjected to substitution by cyclopropylamine, followed by removal of dimethylamine (formed as a by-product) by evaporation, in order to avoid the undesired dimethyl-amino side-product (10). 1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU) catalyzed ring closure of the cyclopropyl derivative (4) was followed by nucleophilic aromatic substitution of 7 using piperazine in a separate step, to give the ester derivative of ciprofloxacin (8). Basic hydrolysis in the last flow step and subsequent pH adjustment gave the final product (9).

In Schwalbe's procedure, each step is conducted in different solvents and requires individual work-up. Therefore, it was not possible to link the steps. The telescoped continuous-flow synthesis of ciprofloxacin (9) without solvent switches was achieved by Jensen and co-workers, by using some key modifications of the previous route.109 The screening of solvents and bases revealed that the combination of acetonitrile and N,N-diisopropylethylamine (DIEA) is not only ideal for the first step, but also compatible throughout the whole process.

Compared to the previous approach, an additional step was introduced for the solution phase scavenging of the dimethylamine by-product formed in the second step. Acylation with acetyl chloride in the third reactor gave the innocuous dimethylacetamide, while intermediate 4 was not affected. This unusual operation allowed the removal of dimethylamine without breaking the continuous operation.

Clogging caused by the low solubility of the fluoroquinolone intermediate 8 was avoided by thermal insulation of the related parts, to prevent the solution from cooling down. Further optimization of the key steps and utilizing higher reaction temperatures led to a more effective process. The sodium salt of ciprofloxacin (9·Na) was successfully prepared in 5 telescoped flow reactors with only 9 min of total residence time. Off-line acidic treatment of the salt gave the ciprofloxacin hydrochloride (9·HCl) with 60% yield and 1.95 g h−1 productivity. Furthermore, a standalone unit was also developed for the end-to-end manufacturing of 9·HCl from 4, which involved downstream processes as well, including precipitation, filtration and antisolvent crystallization.30 Finally, the resulting crystal suspension was formulated into a solution (for ophthalmological use), which met the United States Pharmacopeia (USP) standards.

More recently, Gupton and co-workers have streamlined the process by choosing a different starting material.110 The vinylogous cyclopropyl amide (6) was prepared from the more affordable, commercially available vinyl ether (5) using a high-temperature substitution, conducted as a standalone batch operation. The following flow sequence differs from the previous examples in two main aspects. A chemoselective C-acylation of 6 is required in the key step, which can be achieved by the application of the bulky lithium bis(trimethylsilyl)amide (LiHMDS) base. The excess of this base is also effective in the consecutive ring closure, which readily leads to the quinone intermediate (7), allowing the consolidation of the acylation and cyclization into a single unit operation. The final two steps (substitution by piperazine and the final hydrolysis) are conducted analogously to the previously described routes, apart from using a stirred buffer tank to preheat the reaction mixture before entering the high temperature reactor.

As by-product 10 cannot be formed in this case, the scavenging step is not required, leading to shorter sequence (4.7 min total residence time), along with higher yield and productivity (83% and 15.8 g h−1).

FOD features. Using bases with good solubility (NBu3, DBU, and DIEA) to prevent clogging caused by precipitation.

  • Solvent-screening for optimal solubility and compatibility throughout the whole telescoped process.
  • Keeping the solutions of precipitation–prone intermediates at higher than ambient temperature between the reactors, to avoid clogging.
  • Removing the dimethylamine side-product with an acylation reaction (solution phase scavenging) instead of the evaporation, which doesn't break the continuous operation.
  • Preparing starting materials or early intermediates in batch, when more appropriate than a flow procedure.
  • Optimized chemoselective transformation (C-acylation of the vinylogous amide).
  • Consolidating two or more steps requiring similar conditions into one.

Ibuprofen (flow synthesis milestones: 2009, 2015, 2016, 2019)

Ibuprofen (14) is a long-established nonsteroidal anti-inflammatory drug. A previously known three-step batch route111 served as a foundation for the pioneering continuous-flow synthesis developed by the McQuade group.112

In the first step, Friedel–Crafts acylation of isobutylbenzene (11) to the corresponding ketone (12) could be carried out using AlCl3 with high yield (Scheme 2), however the by-products were found to be incompatible with the downstream steps. Using triflic acid (TfOH) together with propionic acid not only proved to be an effective method to synthesize the ketone (12), but it was also compatible with the next steps.

Scheme 2 Flow syntheses of ibuprofen (14) via different approaches (TfOH: triflic acid, TMOF: trimethyl orthoformate, MeOTf: methyl triflate).

During the second step, 1,2-aryl migration leading to an ester intermediate (13) was achieved by the addition of trimethyl orthoformate (TMOF) and PhI(OAc)2 to the crude mixture exiting the first step. An acid is also required, the TfOH already present in the reaction stream proved to be appropriate acid for this purpose.

In the final step, saponification of the ester (13) with KOH resulted in the desired product (14) with 51% yield and 405 mg h−1 productivity after recrystallization.

While revisiting this synthesis, Snead and Jamison made the Friedel–Crafts acylation possible with AlCl3 under flow conditions.113 They used the propionyl chloride reagent as solvent, since it forms a stable, homogeneous solution if mixed with AlCl3 in 1[thin space (1/6-em)]:[thin space (1/6-em)]1 ratio. The reaction afforded the ketone (12) in quantitative yield. However, in-line aqueous work-up and liquid–liquid separation had to be incorporated at this point, to eliminate the aluminium containing waste formed in the Friedel–Crafts reaction.

The aryl migration was conducted using iodine monochloride (ICl) and TMOF in propanol. Additional N,N-dimethylformamide (DMF) prevented clogging caused by I2 formed during the reaction. The remaining ICl had to be quenched with acetone, in order to avoid the oxidation of the gold spring inside the back-pressure regulator (BPR). Since acetone was not compatible with the next step (the hydrolysis did not proceed in the presence of chloroacetone) 2-mercaptoethanol was used instead. Otherwise, the last step required similar conditions to the previous method. Adding a second in-line separation step after the hydrolysis gave pure ibuprofen (14). The authors reported 83% overall yield and 8.09 g h−1 productivity.

These methods both use small amounts of solvents, resulting in increased productivity, and low amounts of waste. McQuade's approach maximizes reagent compatibility between the steps, but the total residence time is relatively long (10 min), and the product is obtained with low yield and productivity. The Jamison group managed to achieve a much higher yield in only 3 min, but the high productivity is presumably the result of the use of aggressive reagents, which required elaborate in-line quenching and work-up.

Baxendale and co-workers introduced a “green” way of producing this API under flow conditions.114 In this approach, the key chloropropionyl intermediate (15) was transformed into ibuprofen (14) through photo-Favorskii reaction using a continuous-flow photoreactor. The formation of by-products was controlled by the residence time and the irradiation power. The advantages of this method include the fewer number of steps, and the absence of aggressive reagents. Green solvents (acetone, water) were applied and there was no need for hydrolysis or protecting groups, which makes the process atom efficient. However, long residence time (15 min) was needed to reach complete conversion in the photoreactor, and some by-products were unavoidable, which made purification necessary.

An alternative flow route was developed by the Kim group,115 in which ibuprofen (14) was produced via three chemoselective benzylic C–H metalation steps, utilizing superbases generated in situ from tBuOK and alkyl lithium reagents.

The mixture of p-xylene (16) and tBuOK was combined with nBuLi in a micromixer using a very short residence time, followed by the addition of methyl triflate (MeOTf) in a further micromixer to obtain the ethyl derivative (17). The second reaction step leading to the isobutyl derivative (18) was conducted in a similar way, although longer residence time was necessary due to the lower reactivity of isopropyl iodide. In the last step, the API (14) was formed in a biphasic reaction with gaseous CO2, from an anionic intermediate generated by tBuOK and tBuLi. Diluting the stream with 2-methyltetrahydrofuran (2-MeTHF) before addition of the gas was beneficial to ensure dissolution of CO2, and to prevent clogging.

Extensive optimization (which would not be possible using batch methods) of the reaction temperature, composition and concentration was carried out to ensure high selectivity in each step, in which maintaining efficient mixing was found to be crucial. The three continuous-flow metalation reactions (which are two step processes in themselves) were performed separately with off-line aqueous work-up. The last module produced ibuprofen (14) with 57% isolated yield and a productivity of 13.8 g h−1.

FOD features. Using uncommon reagent systems (TfOH or AlCl3 dissolved in the acid chloride for the Friedel–Crafts reaction) for high compatibility or high effectivity.

  • Reusing reagents from the previous step which are present in the reaction stream (like TfOH in McQuade's method).
  • Using in-line separation to enable more efficient, higher throughput process.
  • Carefully selected quenching agents used for harmful reagents, in order to resolve compatibility problems (ICl quench with mercaptoethanol).
  • Reagentless transformation (photochemistry).
  • Extensive flow optimization for fit-to-purpose chemoselectivity (CH metallations).

Imatinib (flow synthesis milestones: 2010, 2016, 2019)

The Bcr–Abl tyrosine kinase inhibitor imatinib (24) is used for the treatment of Philadelphia chromosome-positive chronic myeloid leukaemia, gastrointestinal stromal tumors, and other types of cancer. The original batch process route developed by Novartis AG involves a handful of steps with insoluble intermediates and several purifications.116

In order to minimize manual intervention, a flow-based synthetic route was established by the Ley group based on the known batch mode syntheses.76 The commercially available 4-(chloromethyl)benzoyl chloride (19) was reacted with 3-bromo-4-methylaniline (20) to form the central amide bond of intermediate 21 (Scheme 3). In the next step, SN2 displacement of the chlorine atom led to a piperazine derivative (22). The final C–N coupling reaction yielded the API (24). Most of these steps utilized solid supported catalysts and scavenging reagents.

In a later approach, Buchwald and co-workers followed the same sequence, but only used solution phase reactions. Employing biphasic conditions and an amphiphilic solvent led to improvements in the key C–N coupling reaction.117

In the first step of Ley's procedure, the acid chloride (19) was preloaded onto polystyrene-supported 4-(dimethylamino)pyridine (DMAP) inside a glass column to activate it, then a solution of the amine (20) was pumped through the column to form the product (21) and facilitate its release. A further column of polymer-supported dimethylamine was added to scavenge the acid formed as a by-product by hydrolysis of the starting material. In Buchwald's alternative approach, this step was performed in a 2-MeTHF/H2O biphasic system with KOH as the base, enabling the use of a coil reactor.

Preliminary batch experiments indicated poor solubility of the product of the SN2 reaction (22) in dichloromethane. Thus, the Ley group used DMF as the solvent, after automated solvent exchange by the addition of DMF in a buffer flask, while bubbling nitrogen gas through the solution to evaporate dichloromethane. An UV spectrometer in combination with an autosampler was also necessary at this point, to minimize the dispersion of the reaction stream. The mixture of crude 21 and N-methylpiperazine was pumped through a column of CaCO3. A column of polystyrene-supported isocyanate served for scavenging the unreacted piperazine after the reaction. On the other hand, the usage of biphasic conditions by the Buchwald group enabled the addition of the piperazine as an aqueous solution. Direct connection of the steps was possible by avoiding the solvent switch. In this case, the reaction was performed in a packed bed reactor filled with inert material, which ensured adequate mixing of the 2-MeTHF/water biphasic solvent.

In the earlier study, the final step was precluded by a catch-and-release purification of intermediate 22, which was eluted from the silica column with DBU in dioxane/tBuOH. This mixture entered the Buchwald–Hartwig coupling, in which the reagents were chosen according to their solubility, while avoiding clogging by the formation of insoluble Pd black. The salt by-product was also taken into account. Employing BrettPhos Pd precatalyst and NaOtBu as base gave satisfactory results. In the other approach, the addition of N,N-dimethyloctanamide as an amphiphilic solvent addressed the precipitation of the inorganic by-products during the C–N cross-coupling. Moreover, the poor solubility of the 2-aminopyrimidine derivative (23) was addressed by converting it into its more soluble conjugate acid. The high selectivity of the BrettPhos catalyst made the removal of the excess N-methylpiperazine unnecessary, consequently a direct connection was possible. No solvent change and purification steps were needed during the whole process, while 2-MeTHF, a solvent from renewable resources was used.

Recently, Jamison and co-workers described a different approach for the modular assembly of imatinib (24) in a single uninterrupted stream.118 An aryl nitrile (25) was selected as starting point, since it is considered to be more accessible and stable than the aniline and acid chloride. The hydration of the nitrile to the corresponding amide (26) was complete in 15 minutes with Cs2CO3 in dioxane/water, while minimal amount of waste is generated. The same solvent mixture was also suitable for the subsequent chemoselective amidation reaction, since it dissolved the inorganic salts. In this directly connected step, the crude solution of the amide (26), the streams of catalyst (BrettPhos Pd G4) premixed with the aryl-halide (27) and the base were mixed in a cross mixer. The resulting near-homogeneous mixture contained aqueous microdroplets, which increase interfacial contact. Hence, intermediate 28 could be prepared in a stainless-steel coil, instead of using a packed bed reactor. A stream of iPrOH was introduced after the reaction to prevent clogging of the BPR. The final C–N coupling step of this sequence required similar conditions, but it was burdened by the poor solubility of the aminopyrimidine (23), which was solved by finding the right placement of reagent addition points along the flow path and optimizing the start-up procedure of the telescoped three-step system.

Regarding the overall yield, the Ley group's pioneering work managed to reach 32% yield, while the Buchwald-group isolated the API (24) with 56% yield and 149 mg h−1 productivity, with a total residence time of 33.1 min. The Jamison-group obtained imatinib (24) in 58% overall yield with a total residence time of 48 min and a production rate of 327 mg h−1.

FOD features. Using various general conditions (solid-supported with solvent switches, biphasic or near-homogeneous solvent system) in the various approaches determines productivity.

  • Optimization for selectivity and reagent compatibility helps avoiding interruptions and the need for supported catalysts and scavengers.
  • Complex systems require in-line analytical solutions (Ley's approach) and elaborate start-up procedures (Jamison's method) for robust operation.

Artemisinin derivatives (flow synthesis milestones: 2012, 2013, 2014, 2015, 2018)

Artemisinin (32) and especially its derivatives (33 and 34) are the most important antimalarial drugs nowadays, but unfortunately they are not accessible to the most affected 3rd world population. Artemisinin (33) is extracted from the plant Artemisia annua, which contains only a small amount of this drug. Alternatively, it can be produced by semi-synthesis from artemisinic acid (29) or dihydroartemisinic acid (30) that have higher concentrations in the plant (and can be produced in genetically engineered yeast), making them ideal starting points of the semi-synthetic route.119,120 The efforts of several groups led to significant advancements towards the continuous-flow production of these drugs at the point of their use.104

The reduction of artemisinic acid (29) is possible with in situ formed diimide (Scheme 4). The photo-oxidation of the resulting dihydroartemisinic acid (30) requires well-designed flow equipment to achieve a scalable operation. Addition of a photosensitizer is also necessary, which can be tetraphenylporphyrin (TPP), 9,10-dicyanoanthracene (DCA) or the chlorophyll from the crude plant extract. The oxidizing agent is the in situ photochemically prepared singlet oxygen. The resulting key intermediate (31) can be transformed into the artemisinin (32) by Hock-cleavage and oxidation by triplet oxygen. The reduction of the artemisinin (32) leads to dihydroartemisinin (33), which can be derivatized via various O-alkylations.

Scheme 4 Flow synthesis of artemisinin (32), and its derivatives (33, 34) (TPP: tetraphenylporphyrin, DCA: 9,10 dicyanoanthracene, TFA: trifluoroacetic acid).

The initial reduction of artemisinic acid (29) to dihydroartemisinic acid (30) was described by Kappe and co-workers.121 Diimide (formed in situ from hydrazine hydrate and oxygen) was successfully applied for the reduction in flow environment despite its low reactivity. This reduction agent shows high selectivity against unsaturated C–C bonds, and it is considered an environment-friendly method (as nitrogen and water are the only side-products). However, its use in batch is limited to low amounts of oxygen for safety reasons.122 In the flow method, oxygen can be applied in higher concentration, and the reactivity of the diimide could be improved with the continuous application of the fresh hydrazine hydrate in a multi-injection system. This method affords dihydroartemisinic acid (30) with excellent yield and diastereomer purity (yield ≥ 93%, d.r. ≥ 97[thin space (1/6-em)]:[thin space (1/6-em)]3) and 127 mg per day productivity.

The first continuous-flow synthesis of artemisinin (32) was described by Lévesque and Seeberger.123,124 The solution of dihydroartemisinic acid (30), DCA and trifluoroacetic acid (TFA) in toluene was mixed with oxygen, and reacted at −20 °C, while being irradiated by a LED module. After the photoreaction, the mixture was passed through a second reactor at 10 °C and a third reactor at 25 °C, resulting the desired product (32) with an overall yield of 65%, and a productivity of 6.88 g h−1. The careful choice of technical parameters played an important part to achieve a good yield. The emitted light needs to be of 420 nm wavelength to avoid the side-reactions, therefore the use of a monochromatic LED or a Hg lamp with a filter is necessary. The tubing in the photoreactor was made from fluorinated-ethylene propylene copolímer (FEP), because of its high transmittance and stability against UV-vis lights and chemicals. During the reaction the lamp emitted a significant amount of heat, so proper cooling of the photoreactor was also important. With in-line FTIR analytics, the reaction was conveniently monitored, and it could even detect a simulated “lamp failure”.

Further flow steps could also be coupled to this system, to facilitate the transformation of artemisinin (32) to its derivatives (33 and 34). Reduction was accomplished in a prefilled NaBH4/Li2CO3/LiCl/celite column resulting in dihydroartemisinin (33), which was further functionalized by O-alkylation to β-artemether (34a), β-arteether (34b) and α-artesunate (34c), important first-line antimalarial drugs.125

Gilmore and co-workers achieved the photo-oxidation by using the crude extract of the Artemisia annua. Depending on the extraction procedure, the crude toluene extract can contain enough chlorophyll to utilize it as a photosensitizer, therefore only TFA was added to the feed in this “greener” approach. In fact, the mixture of the different chlorophyll derivatives present in the extract outperformed the synthetic DCA as a photosensitizer.126

FOD features. Optimization for selectivity by using monochromatic LED or filtering the irradiation source, to prevent side-reactions caused by undesired wavelengths.

  • Designing for longevity by using fluorinated-ethylene propylene (FEP) tubing, because of its high transmittance and stability against UV-vis lights and chemicals.
  • Divergent synthesis of artemisinin derivatives.
  • Utilizing the chlorophyll of the crude extract as a readily available photosensitizer.

Rufinamide (flow synthesis milestones: 2013, 2014, 2016)

Rufinamide (43) is an anticonvulsant and antiepileptic pharmaceutical, used for the treatment of the Lennox–Gastaut syndrome, a severe type of epilepsy.127 The key step in its synthesis is the formation of the 1,2,3-triazole moiety, via Huisgen cycloaddition (Scheme 5) utilizing an organic azide (37). In an efficient, one-pot batch method, the non-toxic, inexpensive methyl-3-methoxyacrylate (38) was selected as the dipolarophile. The ester intermediate (39) formed in the cycloaddition can be transformed into the API (43).128 The conventional batch production requires prolonged heating, which is problematic for the large-scale production involving organic azides.

Similar conditions were applied for the continuous-flow synthesis of the ester precursor (39) of the API (43), by the Hessel group.31 First, the use of dilute solutions was attempted in order to avoid clogging caused by the poorly soluble product, however this resulted in poor yields. Thus, neat conditions were preferable to obtain higher reaction rate. To avoid uncontrolled crystallization, the reactor was heated above the melting point of the product (39), and a stream of diluent was introduced in a heated T-mixer. The crude mixture was collected in a holding vessel, where crystalline product formed upon cooling, also serving as an integrated purification step. During optimization of this regioselective, catalyst-free reaction, the authors had to balance the conditions between the decomposition of the azide (37) and the rapid formation of the triazole ester (39) at elevated temperatures.

Later, the Hessel group developed an improved, integrated process, which took green chemistry aspects into consideration.129 Based on a life-cycle assessment, 2,6-difluorobenzyl alcohol (35) was selected as the preferred staring material. This was transformed into a benzyl chloride (36) in a solvent-free reaction using pure hydrogen chloride gas, which produces water as the only by-product. Then, substitution using NaN3 gave the azide (37), during which the formation of the highly toxic and explosive hydrazoic acid was prevented by addition of NaOH base to neutralize the HCl containing crude mixture of the first step. The azide (37) is a liquid under ambient conditions, which enabled its facile separation from the aqueous waste using a liquid–liquid separator unit.

Finally, the azide (37) was combined with neat methyl-3-methoxyacrylate (38) at 210 °C in a Hastelloy reactor. The immediate consumption of the azide minimized the risk of its decomposition. The product stream was diluted with MeOH to facilitate purification by crystallization. An overall 82% yield was achieved for the rufinamide precursor (39) with a productivity of 9 g h−1. Following this procedure, a pilot-scale multistep solvent-free synthesis was developed, which provided the rufinamide precursor (39) with an overall 47% yield and a productivity of 47 g h−1.130

The Jamison group reported the total synthesis of rufinamide (43) under flow conditions from the commercially available benzyl bromide (42) and methyl propiolate (40).67 The synthesis of 2,6-difluorobenzylazide (37) using sodium azide was complete within 1 min of residence time at room temperature, supposedly as a result of the efficient mixing. Propiolamide (41) was prepared by treating 40 with aqueous ammonia. Interestingly, a more economic reaction temperature of 0 °C proved to be suitable in the flow reactor, instead of the cryogenic temperature used in the batch method.

The cycloaddition was attempted by combining 37 and 41 in a high-temperature flow reactor made of fluoropolymer or stainless steel. However, these experiments gave poor yields, and the 1,5-regioisomer was found in significant amounts besides the desired product (43). Significant improvement was observed, when copper tubing was used for construction of the reactor, suggesting that copper served as a catalyst. Temperature had to be optimized between the release of insufficient amounts of copper from the tubing and the decomposition of 37 at lower temperatures. Likewise, the importance of pressure became evident, when a significant drop in yield was observed at higher pressure, with no observable off-gassing at the end of the reaction. Presumably the increased solubility of ammonia at elevated pressure interfered with the copper catalyst. At lower than optimal pressure, reduced residence time caused by the considerable gas generation lowered the yield.

The entire flow synthesis required 11 min total residence time, and it was achieved without solvent change. Rufinamide (43) was isolated with 92% yield, corresponding to 217 mg h−1 productivity. Based on this method, the same group established a continuous-flow platform for the end-to-end manufacturing of the API (43), including the downstream and liquid formulation steps.30 After precipitation, filtration and continuous antisolvent crystallization, the crystal suspension was directed to the formulation unit, where the pre-determined final concentration was adjusted.

FOD features. Cycloaddition under neat conditions, and heating the reactor to resolve solubility issues.

  • Using hazardous reagents (HCl gas and NaN3) for process intensification.
  • Using copper as a reactor and catalyst.

Efavirenz (flow synthesis milestones: 2015, 2019)

Efavirenz (47) is an antiretroviral medication used in combinations to treat and prevent AIDS. Two notable batch routes were developed to produce the API (47), a 5-step synthesis by Merck,131 and a 4-step synthesis by Lonza.132 Since the latter method avoids wasteful amine protection and deprotection, Seeberger and co-workers chose this route as the starting point in the development of the flow synthesis,85 in which the stepwise formation of the cyclic carbamate core is replaced by a one-step process, via an isocyanate intermediate (Scheme 6).

The key propargylic alcohol intermediate (46) was prepared through two organometallic-mediated continuous-flow steps. High reactivity of the lithiated intermediates required careful control of temperature: quick warming of the reaction streams had to be avoided, as it could result in clogging.

First, 1,4-dichlorobenzene (44) was reacted with n-butyllithium, then a trifluoroacylation agent was added. The morpholine amide of trifluoroacetic acid was used for this purpose, instead of the commonly used ester, because the resulting morpholine by-product could be removed in-line during quenching the reaction stream in a column filled with anhydrous silica.

Next, the crude ketone (45) obtained in the first step was utilized in the subsequent step in a similar set-up. Deprotonation of cyclopropylacetylene followed by its nucleophilic addition to the carbonyl compound in a separate flow reactor rapidly produced the required quaternary alcohol (46). The product was isolated with a yield of 73% for the telescoped sequence starting from 1,4-dichlorobenzene (44).

The final flow step consisting of carbamate formation and cyclization utilized copper catalysis, which was incompatible with the by-products from the previous steps, so the purified alcohol (46) was used as starting material. The NaOCN reagent and the Cu0 catalyst was filled into a column together with celite. Different diamine ligands were tested, and trans-N,N′-dimethyl-1,2-cyclohexanediamine (CyDMEDA) showed the best results. Although CuI is the active catalyst, the use of CuII catalysts were more effective and robust, while Cu0 was necessary to speed up the in situ reduction of the CuII to the active CuI. Pure efavirenz (47) was isolated with 62% yield and 76.8 mg h−1 productivity.

Shibata and co-workers reported a formal synthesis of this API (47) through an alternative approach.133 Trifluoromethylation of an appropriate aryl alkynyl ketone (48) in a biphasic flow system using fluoroform gas in the presence of potassium bis(trimethylsilyl)amide (KHMDS) base gave the tertiary alcohol intermediate (46) with moderate yield.

FOD features. Novel formation of the cyclic carbamate core via a one-step process involving an isocyanate intermediate.

  • Using a morpholine-based trifluoroacylation agent to enable facile in-line side-product removal with a dry silica column.
  • Using packed column with NaOCN, because its low solubility.
  • Novel direct trifluoromethylation enables an alternative route.

Gabapentin and pregabalin (flow synthesis milestones: 2015, 2017)

Gabapentin (53a) and pregabalin (53b) are commonly used drugs in the treatment of neuropathic pain and seizure disorders (such as epilepsy). Their structures are closely related the CNS neurotransmitter γ-aminobutyric acid (GABA).134

Seeberger and his group has successfully achieved the modular continuous-flow synthesis of γ-amino acids using the combination of four self-contained units, including an olefination, a Michael addition, a hydrogenation and a saponification module (Scheme 7).70 Each module and the necessary in-line work-up was optimized individually before linking them in this specific order.

In the first module, the preformed phosphonate carbanion was reacted with a ketone (49a) or an aldehyde (49b) in a Horner–Wadsworth–Emmons olefination reaction to give the corresponding olefins (50a and b). A mixture of toluene and methanol was used as solvent, to prevent the precipitation of the side-product phosphate salt. The reaction mixture was quenched with HCl, and the aqueous phase was separated with an in-line membrane liquid–liquid extractor. This removes the methanol co-solvent, which would be disadvantageous in the next step.

The stream containing the olefin (50a and b) was transferred to the second module, where it was reacted with the solution of tetrabutylammonium fluoride and nitromethane in toluene. After similar in-line work-up, the γ-nitroesters were obtained (51a and b).

In the final two modules, these intermediates (51a and b) were hydrolyzed to the corresponding acids (52a and b) with aqueous LiOH, and after in-line separation, the organic phase was subjected to hydrogenation in an H-Cube® reactor. The crude products were recrystallized, resulting in 49% and 68% total yield of gabapentin (53a) and pregabalin (53b), respectively, and 250–333 mg h−1 productivity. The variation of the order of the same modules, together with the choice of the substrates and reagents made possible the divergent preparation of several other APIs.

In an alternative approach developed by Kobayashi and co-workers,135 the flow synthesis of pregabalin (53b) was initiated by the Knoevenagel reaction of dimethyl malonate (54) and isovaleraldehyde in a packed column reactor containing primary amine-modified silica (Chromatorex NH, Scheme 8). In order to reach high conversion and long-term stability in the condensation, 4 Å molecular sieves (MS) had to be added to the packed bed.

The following Michael addition of nitromethane to the olefin (55) leading to intermediate 56 was catalyzed by an anion-exchange resin. Catalytic hydrogenation of the nitro compound (56) on polysilane-supported palladium catalyst provided the final, cyclic amide intermediate (57) with ca. 92 mg h−1 productivity. The best results were obtained in the presence of propanol co-solvent. A pre-column filled by 5 Å MS and silica gel had to be introduced before this step, to maintain good yields for longer periods. Off-line hydrolysis and decarboxylation using hydrochloric acid yielded pregabalin (53b) in 67% yield.

The application of the immobilized catalysts in the latter approach helps avoiding the multiple in-line work-up units needed in the earlier method. However, ensuring appropriate residence times in the filled columns requires low flow rates, which limits the scalability of the process.

Using a similar platform, but a different order of reactions, the same group reported the enantioselective synthesis of a precursor of baclofen, an important aminobutyric acid receptor agonist.136

FOD features. Using continuous in-line work-up procedures (Seeberger's divergent system).

  • Using immobilized catalyst to simplify the work-up procedure (Kobayashi's method).
  • Strategic use of co-solvents.

Lidocaine (flow synthesis milestones: 2016, 2018)

Two continuous-flow syntheses were described for the well-known local anesthetic lidocaine (60), both of which start with the acylation of 2,6-dimethylaniline (58) with chloroacetyl chloride, then the resulting amide (59) is reacted with diethylamine (Scheme 9). Both reactions are performed in the presence of base, which is used for acid scavenging. The previously described batch methods utilized inorganic bases or the excess of the basic reagent for this purpose, which can lead to precipitation.137

In the earlier method developed by Adamo and co-workers,29 the initial step is achieved in N-methyl-2-pyrrolidone (NMP) at high temperature, without basic catalysis. In the second step, the stream of intermediate 58 is mixed with Et2NH and KOH in aqueous MeOH and reacted at 130 °C. Precipitation was prevented by the polar solvent mixture. The resulting API (60) was purified by hydrochloride salt formation and recrystallization. The whole telescoped system containing both the upstream and downstream process afforded the hydrochloride salt of lidocaine (60·HCl) with 88% yield and a 675 mg h−1 productivity.

Precipitation of the hydrochloride salts was prevented by formation of ionic liquids from the corresponding bases in the approach described by Newman and his group.138 The melting point of the hydrochloride salts derived from Bu3N and DBU are under 70 °C, therefore no crystallization occurs during the high-temperature transformations, which are conducted similarly to the previous method, in a telescoped fashion (Scheme 9). The use of two different bases was motivated by the different basicity required in the two steps. This method gave the lidocaine (60) with 66% isolated yield and a productivity of 370 mg h−1.

FOD features. Preventing the precipitation of salts either by using polar protic solvents or forming ionic liquids, under high temperature conditions.

(−)-Oseltamivir (flow synthesis milestones: 2016, 2020)

(−)-Oseltamivir (68) is a neuraminidase inhibitor for the treatment of influenza. In combination with other antivirals, it was considered as a possible treatment for the novel coronavirus infection (COVID-19) causing a global pandemic in early 2020.139

The Hayashi group developed a one-pot batch procedure to prepare (−)-oseltamivir (68), which was further optimized in order to minimize the number of synthetic steps and shorten the reaction time to 170 minutes, or 60 minutes with microwave (MW) heating.140,141 This strategy was implemented to a telescoped continuous-flow approach by the same group.142 In contrast to the batch procedure, alternative solvent systems and bases had to be used to avoid solubility issues.

The flow route utilized the asymmetric Michael reaction between a (Z)-nitroalkene (61) and an aldehyde (62, Scheme 10). The chiral product (64) was formed with the required stereochemistry and reasonable reaction rate in the presence of Schreiner's thiourea and diphenylprolinol silyl ether (63).

Next, consecutive Michael and intermolecular Horner–Wadsworth–Emmons reactions took place in the second reactor, furnishing a cyclic intermediate (66). The specific order of addition was important, to avoid by-product formation involving EtOH in the second step. Protonation of the nitronate anion in a subsequent reactor using in situ formed HCl from trimethylsilyl chloride (TMSCl) gave the nitro compound (67).

The fourth flow unit was responsible for the epimerization of the 5R-isomer using tetrabutylammonium fluoride (TBAF) in EtOH. The last step consisted of the reduction of the nitro group by utilizing solid zinc filled into a column reactor and TMSCl. The concentration of the reagent had to be optimized to suppress epimerization. Off-line work-up and chromatographic purification afforded (−)-oseltamivir (68). The overall productivity of the system was 4 mg h−1, which corresponds to 13% yield with a total residence time of 310 min.

Sagandira and Watts reported the continuous semi-synthesis of (−)-oseltamivir phosphate (68·H3PO4) from ethyl shikimate (69, Scheme 11),143 the flow synthesis of which was also described from the naturally occurring (−)-shikimic acid.144 Trimesylation afforded an activated shikimate (70) which undergoes nucleophilic substitution in the allylic position with an azidating agent to give 71. Although aqueous NaN3 proved to be the most efficient, diphenylphosphoryl azide (DPPA) was used instead, since the succeeding aziridination step using (EtO)3P requires water-free conditions. Thus, telescoping of these steps could be accomplished. Stereoselective ring opening of 72 and a subsequent tandem N–P bond cleavage and acetylation gave intermediate 73. In a second azidation step using a method developed by the authors,145 the acetamide (73) was transformed into 74. (−)-Oseltamivir phosphate (68·H3PO4) was obtained after chemoselective azide reduction and treatment with H3PO4 with 48% semi-telescoped yield for the process.

FOD features. Changing solvents and bases to form homogeneous mixtures.

  • Control of stereochemistry by highly optimized reagent systems.
  • Controlled addition and premixing of solvents.
  • Changing azidating agent to facilitate telescoping (Sagandira and Watts method).

Pyrazole derivatives (flow synthesis milestone: 2017)

AS-136A (79) is a potent measles virus RNA-dependent RNA polymerase complex (RdRp) inhibitor.146 Celecoxib (80a) and mavacoxib (80b) are COX-2 selective nonsteroidal anti-inflammatory drugs (NSAIDs). The key step of the synthesis of these three APIs is the formation of the pyrazole ring.

The synthesis of the 3-trifluoromethylpyrazole (77) core has been reported using a batch procedure via the Ag2O and NaOAc-catalyzed cycloaddition of the corresponding diazomethane and a terminal alkyne. This method uses super-stoichiometric amounts of additives, and the hazardous diazomethane derivative has to be transferred between the vessels, carrying considerable risk.147 A unified flow synthetic route towards these drugs was reported by Britton and Jamison, to address these issues.73,148

The trifluoromethyl diazomethane (76) was formed in situ from trifluoroethylamine (75) and tBuONO, then reacted with a terminal alkyne in two sequential reactors, yielding the corresponding 3-trifluoromethylpyrazole derivatives (77, Scheme 12). During the rapid cycloaddition under flow conditions, only a small amount of the diazo compound is present in the system at any time, making this method inherently safer than the batch procedure. Furthermore, the closed system allows the use of dichloromethane at 60 °C under pressure, consequently the high amounts of additives can be omitted.

In a connected flow reactor, the N-atom is methylated in the presence of DBU base leading to intermediate 78. The following direct amidation using LiHMDS base rapidly affords AS-136A (79). The telescoped system provided 34% isolated yield, 72% purity, and a productivity of 1.73 g h−1.

The arylation of the N-atom leading to the “coxibs” (80a and b) is possible via the Ullmann coupling with the corresponding protected aryl iodides. This was achieved in batch, because of the required long reaction times. The final deprotection afforded celecoxib (80a) and mavacoxib (80b) with 71% and 43% yields, respectively.

The Ley group described an alternative continuous-flow method to produce N-arylated pyrazoles including celecoxib (80a).149 The process involved four continuous steps, beginning with the diazotization of the appropriate sulfanilamide, followed by a reduction using ascorbic acid. Eventually, hydrolysis and condensation with the corresponding dione afforded the API (80a) with 48% yield.

FOD features. Safe handling of the diazonium compound in a flow reactor enables a facile route, using lower amounts of catalyst.

  • Divergent route to three APIs, through a common intermediate.
  • Using organic base (DBU) to prevent clogging caused by precipitation.

Eflornithine (flow synthesis milestones: 2017, 2018)

Eflornithine (83) is an API used to treat the 2nd stage of sleeping sickness (African trypanosomiasis). The batch synthesis routes rely on chlorodifluoromethane (an industrial gas known as R-22) for inserting the difluoromethyl moiety.150 This reagent is a chlorofluorohydrocarbon with ozone depleting potential and will be completely phased out under the Montreal protocol. The scalable continuous-flow synthesis of the eflornithine (83) was achieved by Kappe and co-workers. Instead of R-22,151 it uses the more environmentally friendly fluoroform as a difluoromethyl source.152,153

The starting material was an ornithine ester (81) protected as bis-imine using 4-chlorobenzaldehyde (Scheme 13). In the first step, deprotonation by LiHMDS base was followed by addition of fluoroform leading to the protected intermediate (82). Due to the lower reactivity of fluoroform, intense mixing between the gas and liquid phases was necessary. Also, precise back-pressure regulation was crucial to prevent flow fluctuations and thus allow the use of fluoroform in nearly equivalent amount. This is important for reducing environmental impact, as fluoroform is still a greenhouse gas. An independent study used in-line FTIR and 19F NMR monitoring of the stream exiting the flow reactor, in order to ensure complete consumption of this reagent.154

The protecting groups and the ester were cleaved with aqueous HCl in a connected flow reactor, providing the desired product (83). High flow rate (resulting in faster mixing) was necessary to prevent precipitation during the hydrolysis. Using this 3-step telescoped route eflornithine (83) was obtained with 86% isolated yield and 4.37 g h−1 productivity after purification.

FOD features. Due to the intense mixing between the phases in the flow system the use of the less reactive and greener fluoroform is possible.

Fluconazole (flow synthesis milestones: 2017, 2018, 2019)

Two distinct, multi-step routes have been published for the continuous-flow preparation of the antifungal fluconazole (88). Gupton and co-workers described a semi-continuous approach,155 following a known route.156 Their strategy is based on the previously established continuous-flow organomagnesium chemistry using turbo Grignard reagents.90,157 In the reaction of 1-bromo-2,4-difluorobenzene (84) and iPrMgCl·LiCl, the organometallic intermediate (85) is formed at ambient temperature (Scheme 14). Next, the commercially available dichloroketone (86) is introduced in a second reactor at room temperature. In order to reach high conversion, 85 had to be used in excess relative to 86. Although the apparent conversion in the first step was adequate, it was found that introducing a continuous stirred tank reactor between the steps was beneficial in terms of the tertiary alcohol's (87) yield. The transformation of this key intermediate to fluconazole (88) via the alkylation of 1,2,4-triazole had to be conducted in batch, because the long reaction times of this final step couldn't be adapted to a flow reactor.

More recently, Collins and co-workers developed a four-step flow route, which could be entirely accomplished in flow reactors.158 The synthetic route begins with the Friedel–Crafts acylation of 1,3-difluorobenzene (89), which was conducted in nitromethane solvent in the presence of aluminum chloride (Scheme 15). Neat chloroacetyl chloride was used in slight access to ensure adequate conversion with high selectivity. Solvent incompatibility and the need for off-line aqueous work-up precluded the integration of this step with the following reactors.

The α-chloroketone intermediate (90) was used in the alkylation of 1,2,4-triazole in a high-temperature flow reactor. The amount of the over-alkylated side-product was successfully minimized by using a large excess of the triazole. In the final step, the Corey–Chaykovsky epoxidation of the ketone (91) using a sulfur ylide reagent provides an epoxide (92). This ultimate intermediate is immediately consumed in the presence of the residual triazole to give fluconazole (88). This way, two chemical transformations are accomplished in a single flow reactor. Key to achieving a reliable flow process was the utilization of the trimethylsulfoxonium chloride instead of the initially used, insoluble iodide. The crude product mixture was passed through an in-line celite/charcoal column to remove by-products, after which fluconazole (88) was collected in high purity, with 26% overall yield. Compared to the former route, this approach is more balanced in terms of reaction times, consequently, each of the transformations could be achieved in flow reactors.

In a recent addition, Kappe and co-workers employed in situ generated bromomethyl lithium for the formation of a novel epoxide type intermediate (93) directly from the α-chloroketone (90, Scheme 16). The batch transformation of this epoxide (93) using 1,2,4-triazole directly leads to fluconazole (88), via a shorter route. This approach is alternative to the Corey–Chaykovsky method, which is incompatible with α-chloroketones.159

FOD features. Application of turbo Grignard chemistry, which is well suited to flow methods (Gupton's method).

  • Utilizing excess of the 1,2,4-triazole both to ensure selectivity, and as a reactant in the subsequent step, that is already present in the reaction mixture (Collins's method).
  • Switching reagent to avoid clogging issues.
  • Balanced reaction times for end-to-end flow realization (Collins's method).
  • Shorter route with rapid transformations via instable metalloorganic intermediates (Kappe's method).

Mepivacaine and its derivatives (flow synthesis milestone: 2017)

Mepivacaine (97a), ropivacaine (97b) and bupivacaine (97c) are local anesthetics containing different alkyl groups attached to the piperidine ring.160 The Kappe group reported the continuous-flow synthesis of these 3 drugs through a common amide intermediate (96).161 This intermediate was produced in the coupling of α-picolinic acid (94) and 2,6-dimethylaniline (95), under batch conditions with PCl3 in a microwave reactor (Scheme 17). Due to the precipitation formed during this reaction, implementation of flow technology was not feasible.

In contrast with the preexisting batch methods, continuous-flow hydrogenation conditions were used for both saturating the pyridine ring and installing the N-alkyl groups by reductive amination. This catalytic method avoided wasteful metal hydride reducing agents. The authors realized, that these two steps can be conducted in a single H-Cube® Pro device, starting from intermediate 96 and the corresponding aldehydes. For appropriate conversion and recovery, the reagents were applied in low concentrations and methanol was used together with acetic acid (AcOH) co-solvent. However, after one hour, the activity of the catalyst decreased rapidly, therefore optimization was necessary. Decreasing both the reaction temperature and the amount of co-solvent proved to be effective for long term stability. With the optimized methods, mepivacaine (97a) was obtained with 83% isolated yield and 71 mg h−1 productivity. Ropivacaine (97b) and bupivacaine (97c) were isolated after work-up with 80 and 89% yields respectively.

FOD features. Consecutive steps under similar conditions can be conducted in the same reactor in single operation (ring saturation followed by reductive amination).

  • Greener flow alternatives for problematic steps (reductive amination).
  • Reaction optimization in order to increase catalyst activity.

Dolutegravir (flow synthesis milestone: 2018)

Dolutegravir (106) is an HIV integrase inhibitor, used in combinations to treat AIDS. The chemical complexity of the drug molecule warrants several competing batch routes.162,163 Its continuous-flow synthesis reported by Roper's and Jamison's groups164 is based on a highly efficient batch synthetic route to the related API cabotegravir.165,166

The flow synthesis of the chiral tricyclic structure consists of seven steps, arranged in three-step telescoped sections, followed by a final, single-step flow transformation. In the first part, the condensation of methyl 4-methoxyacetoacetate (98) and N,N-dimethylformamide dimethyl acetal (DMF-DMA) led to intermediate 99 (Scheme 18). Next, substitution of the dimethylamino group by aminoacetaldehyde dimethyl acetal gave an intermediate (100), which was prone to crystallization. Its solubility was increased by raising the reaction temperature to avoid clogging. To this point, no solvents were used, all reagents were applied in their neat forms. The subsequent step for the formation of the pyridone (101) employs solid reagents, thus the introduction of a solvent was necessary. Methanol was optimal for the dissolution of the NaOMe base. A smaller reactor was employed for premixing dimethyl oxalate with intermediate 100 at room temperature, before the addition of the base and entering the heated reactor. The total yield of this section was 56% (3.4 g h−1 productivity), with 74 min residence time.

The second telescoped section began with a chemoselective direct amidation between the ester group at the 5-position of intermediate 101 and a substituted benzylamine (102). Instead of the lengthy procedure involving hydrolysis and coupling with the amine employed during the batch route, the desired amide (103) was obtained in a single step, catalyzed either by bases (i.e. LiOMe) or acids (i.e. AcOH). Although the basic method gave shorter reaction times, only the latter one proved to be compatible with the downstream steps.

The deprotection of the acetal group and subsequent cyclization of the aldehyde (104) using a chiral amino alcohol was realized as a two-step process, which gave better results compared to the initial attempts to conduct these reactions in a single unit. This way, the tricyclic intermediate (105) was formed under less harsh conditions with full conversion (diastereomer ratio 7[thin space (1/6-em)]:[thin space (1/6-em)]1). Furthermore, p-toluenesulfonic acid (pTsOH) could be used instead of the high amounts of formic acid required in the single-step approach.

Connecting the acid mediated amidation with the deprotection–cyclization sequence gave an overall yield of 48% in 190 minutes residence time. The major diastereomer was separated using silica gel chromatography before forwarding it to the last step.

In the final flow step, demethylation of 105 using LiBr in aqueous THF gave dolutegravir (106) with 89% yield in 31 min residence time. Concentration proved to be the key issue in this reaction, as the solubility of dolutegravir (106) was limited in THF. Besides chromatography, this was the reason why the authors could not connect this step with the previous part. The total 7-step flow process gave an overall yield of 24%. Solubility problems, the necessary solvent switches and purifications between the sections prohibited the connection of these already complex parts.

FOD features. Heating, using low concentration and choosing the ideal basis and solvents to prevent clogging caused by precipitation.

  • Replacing the lengthy two-step amidation with a chemoselective single-step reaction, in which the catalyst is chosen according to the telescoping requirements.
  • Separation of the deprotection–cyclization reactions into two steps for milder conditions.

Ivacaftor (flow synthesis milestone: 2018)

Ivacaftor (110) is used for the treatment of cystic fibrosis, an autosomal recessive genetic disorder. The flow synthesis described by Kulkarni and co-workers78 follows a novel synthetic route, which was previously developed by the same authors in batch.167 In contrast to the literature syntheses, which usually employ harsh conditions, this approach relies on the oxidative transformation of an indole-3-acetic acid to a 4-quinolone-3-carboxylic acid derivative.

The three-step sequence utilizes an indole-3-acetic amide (107) prepared in batch as starting material (Scheme 19). The oxidative cleavage of the indole double bond was accomplished by ozonolysis, leading to intermediate 108. Ozone gas is a greener oxidizing agent than the alternative metal-based oxidants, and the associated hazards are manageable in flow reactors, making ozonolysis well suited for continuous processing. The quenching of the transient ozonide by Me2S was rather slow, which led the authors to using acetone–water solvent mixture,168 a method already proven in context of multistep flow processing.169 This way, the carbonyl oxide intermediates are captured in situ, followed by decomposition to the desired carbonyl compound (108) within seconds, without the need for any additives. Excess O3 was degassed in a buffer flask.

The following cyclization of the β-keto ester (108) to obtain the quinolone (109) was achieved in the presence of high excess of DMF–DMA, because the concentration of the reagent was crucial to allow reasonably low residence time. The first step's water containing solvent system was incompatible with this reaction. Telescoping could be accomplished using an extraction (by gravity-based liquid–liquid separation) to allow the in-line solvent switch to 2-MeTHF before the ring closure. Finally, removal of the protecting group from 109 with sodium methoxide solution gave ivacaftor (110). Total yield of the telescoped reactions was 60%, with 300 mg h−1 productivity.

FOD features. Sustainable and safe continuous-flow ozonolysis (O3 is a traceless oxidizer) enabled a novel route to the quinolone core of the API, and the adequate quenching method ensured rapid reaction.

  • Connection of the aqueous and water-free steps was facilitated by an in-line extraction.

Pyrazinamide and isoniazid (flow synthesis milestone: 2018)

Pyrazinamide and isoniazid (113) are antibiotics for the treatment of tuberculosis, which are often used in combination. The former API was prepared in a single-step MnO2-catalyzed hydration, integrated with continuous crystallization.170 The synthesis of isoniazid in an integrated, three-stage process was described by the same group,171 in which a chromatographic separation was utilized between two chemical steps.

First, the catalytic hydration of a 4-pyridinecarbonitrile (111) in a manganese dioxide containing packed bed reactor172–174 afforded the corresponding amide (112, Scheme 20). Although obtaining full conversion is possible in this setup, the authors adjusted the conditions in a way that the crude product contained 91% of 112, together with 9% of the starting material. This models the effects of different upstream disturbances and the eventual deactivation of the catalyst, which need to be handled by the system in order to provide robust operation. An in-line infrared (IR) detector was installed after the first step, which detects the presence of the crude mixture during start-up and then monitors the steady state operation.

The crude solution was injected to four supercritical fluid chromatography (SFC) columns operated in parallel, at an average rate that matched the flow rate of the intermediate. Methanol was used as a modifier solvent alongside the supercritical CO2 mobile phase. Overlap of the product peaks between columns resulted in an uninterrupted stream of the intermediate at the output of the SFC system, where outgassing of the CO2 in cyclone vessels gave the solution of pure 112 in methanol. This mixture was directed into the second reactor, where displacement of the ester using hydrazine monohydrate took place. After purification, isoniazid (113) was collected with 309 mg h−1 productivity. Extensive automation throughout the system allowed operation with minimal manual intervention.

FOD features. Efficient purification technique, which is seamlessly integrated.

  • Intermediate purification can bring considerable robustness to a complex system.
  • Using overall automation.

Flibanserin (flow synthesis milestone: 2019)

The aryl piperazine-type serotonin receptor modulator drug flibanserin (119) was originally developed for the treatment of depression. After repurposing, it was approved in the indication of female hypoactive sexual desire disorder (HSDD).175 Greiner and co-workers developed an unprecedented route for its synthesis, designed for efficient operation in an uninterrupted sequence of continuous-flow rectors.79

In the first step, tert-butyl (2-aminophenyl)carbamate (114) was reductively alkylated using 2,2-dimethoxyacetaldehyde, which serves as the precursor of the two-carbon linker (Scheme 21). This way, the use of potentially genotoxic alkylating agents and wasteful bases can be avoided at both of the C–N bond formation steps. Hydrogenation in an H-Cube Pro™ reactor using 10% Pd/C catalyst afforded the monoalkylated product (115) with high selectivity. Excess hydrogen gas was removed in a buffer flask.

Next, intermediate 115 was directly transformed in a ring closure reaction to form the benzimidazolone core (116), by incorporating the carbonyl group of the carbamate into the resulting heterocycle. In a high-temperature flow reactor, DBU was used as base at 200 °C. The next step was directly connected, in which deprotection of the dimethoxy intermediate (116) led to the corresponding aldehyde (117) under biphasic conditions with aqueous hydrochloric acid. This point of the sequence offers an ideal opportunity for the acidic transformation, as the molecule has no basic functions (the formation of potentially insoluble salts is avoided, and subsequent neutralization is not necessary). Afterwards, the gravity-based separation the non-miscible liquid phases in a buffer flask also served as an integrated in-line purification to remove DBU from step 2, as well as impurities and traces of previous intermediates.

In the final step, the aldehyde (117) was utilized in a second reductive amination with 1-(3-trifluoromethylphenyl)piperazine (118) under catalytic hydrogenation conditions, to yield flibanserin (119). Selective salt formation and filtration was employed offline, to obtain crystalline 119·HCl with 31% isolated yield (184 mg h−1 in terms of the base form). The telescoped flow synthesis used iPrOAc as solvent, together with tBuOH and MeOH produced as by-products, which served as co-solvents to prevent the precipitation of low solubility intermediates.

FOD features. Application of two reductive amination steps is not only favorable to flow chemistry, but also makes avoidance of genotoxic alkylating agents possible.

  • Careful placement of acidic conditions to avoid salt formation and the need for subsequent neutralization.
  • Incorporation of in-line purification to facilitate the telescoping of steps.
  • Use of in situ formed co-solvents to prevent precipitation.

Ketamine (flow synthesis milestone: 2019)

The NMDA receptor antagonist ketamine (123) was originally marketed as an anesthetic, but its human use was burdened with side effects. It has recently found use as a rapid-acting treatment for depression.176 Its three-step flow synthesis was described by Monbaliu and co-workers,177 following an economically and environmentally favorable alternative new route based on a known strategy.178,179 The readily available cyclopentyl phenyl ketone precursor (120) was hydroxylated using molecular oxygen to give intermediate 121 (Scheme 22). Next, reaction with methyl-amine generated an imine (122), which was thermally rearranged to racemic ketamine (123).

During the development of the continuous-flow procedure, undesirable solvents were replaced by ethanol as a single solvent throughout the telescoped procedure, without intermediate solvent swaps.

In the first step, instead of the introduction of a bromine atom used in the original procedure (which would be converted to a hydroxyl group in a later stage), a novel direct hydroxylation180 was adapted to this purpose. Oxygen gas was selected as the primary oxidant, leading to an intermediate hydroperoxide, which is reduced to the hydroxyketone (121) using triethyl phosphite. The exothermic nature of the reaction was handled by thermostating both the reactor and the T-mixer. In order to avoid unwanted rearrangement of the product, an extensive optimization identified the ideal combination of potassium hydroxide and PEG-400. The authors also attempted the use of supported bases, but the rapid decline of their performance made these impractical. It was shown, that the phosphite reductant and its oxidized form, as well as the PEG-400 additive are unreactive under the conditions of the following steps, and their removal is unnecessary from the crude mixture. The hydroxylation was scaled-up using a mesofluidic reactor working with kg per day productivity, while retaining safe handling of the oxygen gas in the flow system.181

In the imine formation step, transforming the sluggish batch conditions to flow required triisopropyl borate additive, which helped keeping the temperature low for a clean product mixture. Before the last step, a surge vessel was installed for venting excess oxygen and dilution of the stream with ethanol.

The last step, a reagentless thermolysis catalyzed by montmorillonite K10 at 180 °C was conducted in a packed bed reactor. The presence of potassium hydroxide and other basic species in the reaction mixture deactivated the acidic sites of the catalyst after 20 minutes, which limited the achievable productivity. Pure racemic ketamine (123) could be isolated either by crystallization of the base after concentration and cooling, or by crystallization of the hydrochloride salt, with 36% and 62% yield, respectively. In both cases, reagents and additives remained in the mother liquor. The authors reported a 2.96 g h−1 productivity for the free base. The three-step uninterrupted reactor network was also capable for the preparation of two analogs of 123.

FOD features. Using only one solvent and economically favorable, non-toxic additives for the entire procedure.

  • Using intensified conditions in the imine formation and the reagentless thermal rearrangement.
  • Applying hydroxylation step designed for this purpose, which uses molecular oxygen as a traceless reagent.

Linezolid (flow synthesis milestone: 2019)

Linezolid (132) is an antibiotic against multi-drug resistant Gram-negative bacteria. Its seven-step, convergent and protecting group-free flow synthesis was realized by Russell and Jamison, which is the longest integrated flow chemical synthesis known to date (Scheme 23).69 In line with the flow oriented design principles, the authors utilized retrosynthetic analysis to identify a novel strategy based on the late-stage formation of the oxazolidinone ring. Flexible modifications of the route were crucial to enable flow operation.

In the initial form of the first step, a Ritter-type reaction between (+)-epichlorohydrin (124) and acetonitrile using BF3·OEt2 as a Lewis acid provided an amide (133) after quenching the intermediate 125 with aqueous NaHCO3 (Scheme 24). However, the authors had to overcome two different issues to realize this reaction in flow mode. First, a side-product (134) was formed by the nucleophilic attack of the diethyl ether containing reagent. This could be avoided using the more hindered BF3·OBu2 complex under the same conditions. Secondly, aqueous quench led to the formation of insoluble boric acid and clogging. This issue was solved by quenching with 2-propanol at −35 °C, while instead of 133, an imidate (126) was formed by the addition of 2-propanol. As a consequence of the unplanned masking of the reactive amide group, side-reactions were minimized during the epoxidation, which was conducted in the presence of lithium tert-butoxide leading to the epoxide bearing a masked amide (127). For the solubilization of the resulting lithium chloride, 1,2-dichloroethane was necessary in the solvent mixture, which was also compatible with the cryogenic temperatures.

In a separate stream, the high-temperature SNAr reaction of 3,4-dinitrofluoronitrobenzene (128) and morpholine, followed by catalytic hydrogenation of the nitro group (129) over solid supported Pd0 catalyst provided the substituted aniline (130) with high selectivity in both steps, allowing direct telescoping towards the API (132). The 1,4-dioxane and DMF solvent mixture ensured dissolution of all reaction components, enabled high SNAr reactivity and it was also compatible with the supported Pd0 catalyst. Hydrogen was degassed in a buffer flask.

It is worth mentioning the related flow preparations starting from the 2,4-isomer of the difluoroaniline, which yielded the mixture of regioisomers. In these cases, continuous-flow separation techniques, such as simulated moving-bed chromatography (SMB)35 or centrifugal partition chromatography (CPC)37 were implemented to provide the pure regioisomers.

In the concluding section of the convergent route, coupling of the epoxide (127) and aniline (130) led to the acyclic precursor (131). Interestingly, the water formed during this reaction also hydrolyzed the masked amide. The authors noted the sensitivity of 130 and 131 towards oxidation, which was prevented by their immediate consumption in the next flow reactors. In the final ring closure, 1,1′-carbonyldiimidazole was utilized, which proved to be compatible with the water containing multi component solvent system. Addition of HCl and in-line extraction provided linezolid (132), which was purified off-line. A remarkable 73% isolated yield, and 816 mg h−1 throughput was achieved.

FOD features. Using retrosynthetic analysis to identify a novel strategy more ideal to flow synthesis.

  • Flexible modifications of the route were crucial in order to enable flow operation.
  • Utilizing an unplanned masking to avoid side-reactions.
  • Several different solvents (and solvent mixtures) specifically chosen for each step, but compatible with each other and the subsequent steps.
  • Applying both cryogenic and high-temperature transformations for the best throughput.

Lomustine (flow synthesis milestone: 2019)

Thompson and co-workers described the two-step, telescoped, continuous-flow synthesis of lomustine (138), an anti-cancer chemotherapy drug.182 The process involves highly toxic alkylating agents such as the starting material (136), the intermediate (137) and even the API (138) itself (Scheme 25). The application of the flow system and in-line work-up at the intermediate stage minimizes the manual handling of these toxic compounds.

The urea intermediate (137) was produced in the reaction of cyclohexylamine (135) and a chloroalkyl isocyanate (136) in the presence of catalytic TEA. However, TEA decreased the yield of the second step, therefore dichloromethane and water were added to the reaction mixture and TEA was removed using in-line extraction. After separation of the phases, the base containing aqueous solution was directed to the waste, while the organic phase entered the next step, where nitrosation with tert-butyl nitrite provided lomustine (138) with 63% yield and 110 mg h−1 productivity.

FOD features. • Using in-line extraction to facilitate the connection of the steps (to remove detrimental reagents).

  • The closed system flow realization prevents human exposure to highly toxic API and intermediates.

Paroxetine (flow synthesis milestone: 2019)

The selective serotonin reuptake inhibitor (−)-paroxetine (143) is used for the treatment of depression, anxiety and panic disorder. The published batch procedures consist of 10–15 reaction steps, including classic resolution methods or chiral auxiliaries. In most of these, the key chiral piperidine intermediate (142) or its derivatives are constructed, and then converted into the API (143, Scheme 26).183 Enantioselective catalytic transformations, which require less steps, have also been reported, however the productivity of these are usually low.184

Kappe and co-workers pioneered the use of polystyrene-supported organocatalysts for enantioselective transformations under solvent-free continuous-flow conditions. In contrast to the previous flow approaches using commercially available chiral building blocks to chiral products, achiral starting materials were used in this work.77

In the first step, the conjugate addition of dimethyl malonate to fluorocinnamaldehyde (139) was performed in the presence of a resin-bound chiral organocatalyst. As the preliminary experiments indicated that the increase in concentration results in improved yield, the transformation was conducted under neat conditions. This offers significant advantages in terms of process costs, sustainability and productivity.

A continuous-flow heterogeneous catalytic method was used for the tandem reductive amination–lactamization of intermediate 140. Complete conversion and high selectivity were achieved, when the hydrogenation was conducted in a concentrated 2-MeTHF solution, using 5% Pt/C catalyst at 100 °C. This approach is considered more environmentally friendly, as the previous batch methods used borohydride based reducing agents.

Following a safe and highly productive borane mediated continuous-flow method that was studied in detail by the authors,185 neat BH3·DMS was used for the reduction of both lactam and ester moieties of intermediate 141, in a single operation. High conversion and selectivity could be reached, without affecting the enantiomeric purity.

Telescoping the last two steps was feasible, since no solvent exchange was needed. After the reductive amination–lactamization, the released water had to be removed, therefore the mixture was directed through a cartridge packed with 4 Å MS. The excess H2 was separated in a buffer flask, then the dried and degassed feed was reintroduced to the flow system, and combined with the stream of neat BH3·DMS to yield the key chiral intermediate (142). This can be converted into the API (143) upon etherification and removal of the benzyl group.183 The productivity of the entire flow process to 142 was 2.97 mg h−1 with excellent 96% ee.

FOD features. • Use of solid-supported organocatalyst for enantioselective transformation from achiral starting materials.

  • Application of neat conditions to improve sustainability and productivity.
  • Using tandem reactions (reductive amination–lactamization and simultaneous reduction of lactam and ester moieties) to simplify the process.
  • Introducing a multimodal separation unit (both water and hydrogen gas) to allow telescoping.

Lesinurad (flow synthesis milestone: 2020)

The urate anion exchange transporter 1 (URAT1) inhibitor lesinurad (149) is a pharmaceutical for treating high uric acid levels in blood, associated with gout. Various batch routes have been reported to produce this API, which often involved lengthy reaction times, several purification steps, and the use of hazardous reagents.186–188

The Pastre group has developed a flow platform to produce 3-thio-1,2,4-triazoles, via condensation of hydrazides and isothiocyanates, followed by cyclization (Scheme 27).189 Based on this key intermediate, the five-step continuous-flow synthesis of lesinurad (149) was reported by the same group in three separate flow operations.190

In the initial approach, the cyclopropyl group was planned to be introduced in a Suzuki reaction, but this step could not be integrated into the telescoped flow system, owing to the need for work-up and solvent exchange. Hence, a commercially available starting material (144) was utilized (which already contained the cyclopropyl group), and the coupling reaction was omitted.

Condensation of the isothiocyanate (144) and formic hydrazide in a coil reactor was combined with the subsequent cyclization of the intermediate (145) by directly introducing a stream of aqueous NaOH. The concentration of NaOH was crucial to prevent by-product formation. Telescoping this step with the S-alkylation of the thiol (146) produced the key intermediate (147) in 55 minutes with 88% yield.

The bromination of 147 was performed under flow conditions with N-bromosuccinimide, which is considered to be a safer option, than the highly corrosive and toxic elemental bromine. Since the best yield was obtained in a different solvent than the previous reactions, the telescoping of the halogenation was not attempted. The same issue occurred in case of the last step, as the hydrolysis of 148 required aqueous LiOH, therefore the connection of this step was not possible either. The overall yield of the three separate flow operations leading to lesinurad (149) was 68%, with ca. 2 h of total residence time.

FOD features. • Design of a flow platform to 3-thio-1,2,4-triazoles using telescoped condensation and cyclization.

  • Choosing an alternative starting material, to avoid breaking the flow sequence (Suzuki coupling could not be applied using flow chemistry).
  • Avoiding the use of highly corrosive reagents, if possible (using NBS instead of bromine).

Conclusions and outlook

Flow-oriented design (FOD) features could be identified in all of the discussed multi-step flow approaches. This means that the effective planning of the synthetic conditions and the applied flow technology is key in achieving continuous API production. In multiple cases, FOD is represented as a novel synthetic route, specifically designed for flow-operation. This creates unprecedented possibilities by the application of highly effective flow techniques, instead of translating pre-existing batch routes to flow reactors.

The toolbox of FOD clearly broadened over time, due to the evolution of flow equipment and expanding experience at hand. The solid-supported methods tend to be replaced by solution-phase scavenging or biphasic systems to facilitate continuous operation. Reagentless transformations (e.g. photochemistry) and the use of neat conditions became preferred, which are also in compliance with the green chemical trends.

The FOD features highlighted in this work can be utilized in the design of more complex, future approaches for other drug molecules. It can also be argued, that effective telescoping leads to highly productive systems, consequently these principles should be applied during the development of industrial scale flow processes.

The multistep continuous-flow procedures presented in this review were mostly demonstrated on laboratory-scale, with a few exceptions of pilot plant scale operations reported by industrial groups. Nonetheless, these systems are already capable of producing drug substances for small populations, even in an automatic mode without local human interaction, supervised over the internet. This offers an effective strategy to establish on the spot drug supply near the affected areas. For example, global collaboration proved that flow chemistry can be a useful tool in the hand of the chemists to replace the current, non-reliable extraction methods in the production of antimalarial drugs with the continuous-flow photochemical production of the first-line artemisinin derivatives.

In the near, future flow chemistry based systems could provide drugs where other methods are not, and save lives in the most isolated places of Earth, such as research stations on Antarctica.191 Possible applications beyond Earth include advanced technologies providing essential medicines on space stations or long duration space expeditions of the future.192,193

Conflicts of interest

There are no conflicts to declare.


This work was performed in the frame of FIEK_16-1-2016-0007 project, implemented with the support provided from the National Research, Development and Innovation Fund of Hungary, financed under the FIEK_16 funding scheme. Zs. F. and P. Sz. thank the Gedeon Richter Talentum Foundation for financial support. Authors would like to thank György I. Túrós for valuable discussions during the preparation of the manuscript.

Notes and references

  1. R. A. Bourne, K. K. Hii and B. J. Reizman, React. Chem. Eng., 2019, 4, 1504–1505 RSC.
  2. M. Trobe and M. D. Burke, Angew. Chem., Int. Ed., 2018, 57, 4192–4214 CrossRef CAS PubMed.
  3. Flow Chemistry: Volume 1 Fundamentals, ed. F. Darvas, V. Hessel and G. Dormán, de Gruyter, Berlin, Germany, 2014 Search PubMed.
  4. T. Glasnov, Continuous-Flow Chemistry in the Research Laboratory, Springer International Publishing, Cham, 2016 Search PubMed.
  5. N. G. Anderson, Org. Process Res. Dev., 2012, 16, 852–869 CrossRef CAS.
  6. S. G. Newman and K. F. Jensen, Green Chem., 2013, 15, 1456–1472 RSC.
  7. L. Vaccaro, D. Lanari, A. Marrocchi and G. Strappaveccia, Green Chem., 2014, 16, 3680–3704 RSC.
  8. J. A. M. Lummiss, P. D. Morse, R. L. Beingessner and T. F. Jamison, Chem. Rec., 2017, 17, 667–680 CrossRef CAS PubMed.
  9. F. Fanelli, G. Parisi, L. Degennaro and R. Luisi, Beilstein J. Org. Chem., 2017, 13, 520–542 CrossRef CAS PubMed.
  10. D. Dallinger and C. O. Kappe, Curr. Opin. Green Sustain. Chem., 2017, 7, 6–12 CrossRef.
  11. L. Rogers and K. F. Jensen, Green Chem., 2019, 21, 3481–3498 RSC.
  12. J. A. Bennett, Z. S. Campbell and M. Abolhasani, Curr. Opin. Chem. Eng., 2019, 26, 9–19 CrossRef.
  13. S. A. May, J. Flow Chem., 2017, 7, 137–145 CrossRef CAS.
  14. D. L. Hughes, Org. Process Res. Dev., 2018, 22, 13–20 CrossRef CAS.
  15. V. Hessel, Chem. Int., 2018, 40, 12–16 CAS.
  16. M. Baumann, T. S. Moody, M. Smyth and S. Wharry, Org. Process Res. Dev., 2020 DOI:10.1021/acs.oprd.9b00524.
  17. S. L. Lee, T. F. O'Connor, X. Yang, C. N. Cruz, S. Chatterjee, R. D. Madurawe, C. M. V. Moore, L. X. Yu and J. Woodcock, J. Pharm. Innov., 2015, 10, 191–199 CrossRef.
  18. Quality Considerations for Continuous Manufacturing Guidance for Industry, https://www.fda.gov/regulatory-information/search-fda-guidance-documents/quality-considerations-continuous-manufacturing, (accessed 28 May 2020) Search PubMed.
  19. B. Gutmann, D. Cantillo and C. O. Kappe, Angew. Chem., Int. Ed., 2015, 54, 6688–6728 CrossRef CAS PubMed.
  20. M. Baumann and I. R. Baxendale, Beilstein J. Org. Chem., 2015, 11, 1194–1219 CrossRef CAS PubMed.
  21. S. Kobayashi, Chem. - Asian J., 2016, 11, 425–436 CrossRef CAS PubMed.
  22. R. Porta, M. Benaglia and A. Puglisi, Org. Process Res. Dev., 2016, 20, 2–25 CrossRef CAS.
  23. J. Britton and C. L. Raston, Chem. Soc. Rev., 2017, 46, 1250–1271 RSC.
  24. R. Gérardy, N. Emmanuel, T. Toupy, V.-E. Kassin, N. N. Tshibalonza, M. Schmitz and J.-C. M. Monbaliu, Eur. J. Org. Chem., 2018, 2018, 2301–2351 CrossRef.
  25. D. L. Riley, I. Strydom, R. Chikwamba and J.-L. Panayides, React. Chem. Eng., 2019, 4, 457–489 RSC.
  26. A. R. Bogdan and A. W. Dombrowski, J. Med. Chem., 2019, 62, 6422–6468 CrossRef CAS PubMed.
  27. A. Gioiello, A. Piccinno, A. M. Lozza and B. Cerra, J. Med. Chem., 2020, 63, 6624–6647 CrossRef PubMed.
  28. V. R. L. J. Bloemendal, M. A. C. H. Janssen, J. C. M. van Hest and F. P. J. T. Rutjes, React. Chem. Eng., 2020, 5, 1186–1197 RSC.
  29. A. Adamo, R. L. Beingessner, M. Behnam, J. Chen, T. F. Jamison, K. F. Jensen, J. C. M. Monbaliu, A. S. Myerson, E. M. Revalor, D. R. Snead, T. Stelzer, N. Weeranoppanant, S. Y. Wong and P. Zhang, Science, 2016, 352, 61–67 CrossRef CAS PubMed.
  30. P. Zhang, N. Weeranoppanant, D. A. Thomas, K. Tahara, T. Stelzer, M. G. Russell, M. O'Mahony, A. S. Myerson, H. Lin, L. P. Kelly, K. F. Jensen, T. F. Jamison, C. Dai, Y. Cui, N. Briggs, R. L. Beingessner and A. Adamo, Chem. – Eur. J., 2018, 24, 2776–2784 CrossRef CAS.
  31. S. Borukhova, T. Noël, B. Metten, E. de Vos and V. Hessel, ChemSusChem, 2013, 6, 2220–2225 CrossRef CAS PubMed.
  32. P. L. Heider, S. C. Born, S. Basak, B. Benyahia, R. Lakerveld, H. Zhang, R. Hogan, L. Buchbinder, A. Wolfe, S. Mascia, J. M. B. Evans, T. F. Jamison and K. F. Jensen, Org. Process Res. Dev., 2014, 18, 402–409 CrossRef CAS.
  33. B. Wood, K. P. Girard, C. S. Polster and D. M. Croker, Org. Process Res. Dev., 2019, 23, 122–144 CrossRef CAS.
  34. K. Tacsi, H. Pataki, A. Domokos, B. Nagy, I. Csontos, I. Markovits, F. Farkas, Z. K. Nagy and G. Marosi, Cryst. Growth Des., 2020, 20, 4433–4442 CrossRef CAS.
  35. A. G. O'Brien, Z. Horváth, F. Lévesque, J. W. Lee, A. Seidel-Morgenstern and P. H. Seeberger, Angew. Chem., Int. Ed., 2012, 51, 7028–7030 CrossRef PubMed.
  36. Z. Horváth, E. Horosanskaia, J. W. Lee, H. Lorenz, K. Gilmore, P. H. Seeberger and A. Seidel-Morgenstern, Org. Process Res. Dev., 2015, 19, 624–634 CrossRef.
  37. R. Örkényi, J. Éles, F. Faigl, P. Vincze, A. Prechl, Z. Szakács, J. Kóti and I. Greiner, Angew. Chem., Int. Ed., 2017, 56, 8742–8745 CrossRef PubMed.
  38. S. Mascia, P. L. Heider, H. Zhang, R. Lakerveld, B. Benyahia, P. I. Barton, R. D. Braatz, C. L. Cooney, J. M. B. Evans, T. F. Jamison, K. F. Jensen, A. S. Myerson and B. L. Trout, Angew. Chem., Int. Ed., 2013, 52, 12359–12363 CrossRef CAS PubMed.
  39. A. Balogh, A. Domokos, B. Farkas, A. Farkas, Z. Rapi, D. Kiss, Z. Nyiri, Z. Eke, G. Szarka, R. Örkényi, B. Mátravölgyi, F. Faigl, G. Marosi and Z. K. Nagy, Chem. Eng. J., 2018, 350, 290–299 CrossRef CAS.
  40. A. Domokos, B. Nagy, M. Gyürkés, A. Farkas, K. Tacsi, H. Pataki, Y. C. Liu, A. Balogh, P. Firth, B. Szilágyi, G. Marosi, Z. K. Nagy and Z. K. Nagy, Int. J. Pharm., 2020, 581, 119297 CrossRef CAS PubMed.
  41. M. B. Plutschack, B. Pieber, K. Gilmore and P. H. Seeberger, Chem. Rev., 2017, 117, 11796–11893 CrossRef CAS PubMed.
  42. P. Bana, R. Örkényi, K. Lövei, Á. Lakó, G. I. Túrós, J. Éles, F. Faigl and I. Greiner, Bioorg. Med. Chem., 2017, 25, 6180–6189 CrossRef CAS PubMed.
  43. C. J. Mallia and I. R. Baxendale, Org. Process Res. Dev., 2016, 20, 327–360 CrossRef CAS.
  44. M. Irfan, T. N. Glasnov and C. O. Kappe, ChemSusChem, 2011, 4, 300–316 CrossRef CAS PubMed.
  45. P. J. Cossar, L. Hizartzidis, M. I. Simone, A. McCluskey and C. P. Gordon, Org. Biomol. Chem., 2015, 13, 7119–7130 RSC.
  46. D. Riley and N. Neyt, Synthesis, 2018, 50, 2707–2720 CrossRef CAS.
  47. T. Yu, J. Jiao, P. Song, W. Nie, C. Yi, Q. Zhang and P. Li, ChemSusChem, 2020, 13, 2876–2893 CrossRef CAS PubMed.
  48. D. G. Brown and J. Boström, J. Med. Chem., 2016, 59, 4443–4458 CrossRef CAS PubMed.
  49. B. Li, G. A. Weisenburger and J. C. McWilliams, Org. Process Res. Dev., 2020 DOI:10.1021/acs.oprd.0c00112.
  50. T. Razzaq and C. O. Kappe, Chem. – Asian J., 2010, 5, 1274–1289 CAS.
  51. D. Cambié, C. Bottecchia, N. J. W. Straathof, V. Hessel and T. Noël, Chem. Rev., 2016, 116, 10276–10341 CrossRef PubMed.
  52. F. Politano and G. Oksdath-Mansilla, Org. Process Res. Dev., 2018, 22, 1045–1062 CrossRef CAS.
  53. J. D. Williams and C. O. Kappe, Curr. Opin. Green Sustain. Chem., 2020 DOI:10.1016/j.cogsc.2020.05.001.
  54. T. H. Rehm, ChemPhotoChem, 2020, 4, 235–254 CrossRef CAS.
  55. M. Atobe, H. Tateno and Y. Matsumura, Chem. Rev., 2018, 118, 4541–4572 CrossRef CAS PubMed.
  56. D. Pletcher, R. A. Green and R. C. D. Brown, Chem. Rev., 2018, 118, 4573–4591 CrossRef CAS PubMed.
  57. T. Newhouse, P. S. Baran and R. W. Hoffmann, Chem. Soc. Rev., 2009, 38, 3010–3021 RSC.
  58. M. Movsisyan, E. I. P. Delbeke, J. K. E. T. Berton, C. Battilocchio, S. V. Ley and C. V. Stevens, Chem. Soc. Rev., 2016, 45, 4892–4928 RSC.
  59. L. Kupracz and A. Kirschning, Adv. Synth. Catal., 2013, 355, 3375–3380 CrossRef CAS.
  60. A. R. Bogdan, M. Charaschanya, A. W. Dombrowski, Y. Wang and S. W. Djuric, Org. Lett., 2016, 18, 1732–1735 CrossRef CAS PubMed.
  61. Y. Mo and K. F. Jensen, React. Chem. Eng., 2016, 1, 501–507 RSC.
  62. M. R. Chapman, M. H. T. Kwan, G. King, K. E. Jolley, M. Hussain, S. Hussain, I. E. Salama, C. González Niño, L. A. Thompson, M. E. Bayana, A. D. Clayton, B. N. Nguyen, N. J. Turner, N. Kapur and A. J. Blacker, Org. Process Res. Dev., 2017, 21, 1294–1301 CrossRef CAS.
  63. W.-J. Yoo, H. Ishitani, Y. Saito, B. Laroche and S. Kobayashi, J. Org. Chem., 2020, 85, 5132–5145 CrossRef CAS PubMed.
  64. T. Tsubogo, H. Oyamada and S. Kobayashi, Nature, 2015, 520, 329–332 CrossRef CAS PubMed.
  65. B. Laroche, Y. Saito, H. Ishitani and S. Kobayashi, Org. Process Res. Dev., 2019, 23, 961–967 CrossRef CAS.
  66. I. R. Baxendale, J. Deeley, C. M. Griffiths-Jones, S. V. Ley, S. Saaby and G. K. Tranmer, Chem. Commun., 2006, 2566–2568 RSC.
  67. P. Zhang, M. G. Russell and T. F. Jamison, Org. Process Res. Dev., 2014, 18, 1567–1570 CrossRef CAS.
  68. S. Borukhova, T. Noël and V. Hessel, ChemSusChem, 2016, 9, 67–74 CrossRef CAS PubMed.
  69. M. G. Russell and T. F. Jamison, Angew. Chem., Int. Ed., 2019, 58, 7678–7681 CrossRef CAS PubMed.
  70. D. Ghislieri, K. Gilmore and P. H. Seeberger, Angew. Chem., Int. Ed., 2015, 54, 678–682 CAS.
  71. T. Nobuta, G. Xiao, D. Ghislieri, K. Gilmore and P. H. Seeberger, Chem. Commun., 2015, 51, 15133–15136 RSC.
  72. B. Pieber, K. Gilmore and P. H. Seeberger, J. Flow Chem., 2017, 7, 129–136 CrossRef CAS.
  73. J. Britton and T. F. Jamison, Eur. J. Org. Chem., 2017, 2017, 6566–6574 CrossRef CAS.
  74. S. Chatterjee, M. Guidi, P. H. Seeberger and K. Gilmore, Nature, 2020, 579, 379–384 CrossRef CAS PubMed.
  75. K. P. Cole, J. M. Groh, M. D. Johnson, C. L. Burcham, B. M. Campbell, W. D. Diseroad, M. R. Heller, J. R. Howell, N. J. Kallman, T. M. Koenig, S. A. May, R. D. Miller, D. Mitchell, D. P. Myers, S. S. Myers, J. L. Phillips, C. S. Polster, T. D. White, J. Cashman, D. Hurley, R. Moylan, P. Sheehan, R. D. Spencer, K. Desmond, P. Desmond and O. Gowran, Science, 2017, 356, 1144–1150 CrossRef CAS PubMed.
  76. M. D. Hopkin, I. R. Baxendale and S. V. Ley, Chem. Commun., 2010, 46, 2450–2452 RSC.
  77. S. B. Ötvös, M. A. Pericàs and C. O. Kappe, Chem. Sci., 2019, 10, 11141–11146 RSC.
  78. N. Vasudevan, M. K. Sharma, D. S. Reddy and A. A. Kulkarni, React. Chem. Eng., 2018, 3, 520–526 RSC.
  79. P. Bana, Á. Szigetvári, J. Kóti, J. Éles and I. Greiner, React. Chem. Eng., 2019, 4, 652–657 RSC.
  80. N. Weeranoppanant and A. Adamo, ACS Med. Chem. Lett., 2020, 11, 9–15 CrossRef CAS PubMed.
  81. A. Adamo, P. L. Heider, N. Weeranoppanant and K. F. Jensen, Ind. Eng. Chem. Res., 2013, 52, 10802–10808 CrossRef CAS.
  82. B. J. Deadman, C. Battilocchio, E. Sliwinski and S. V. Ley, Green Chem., 2013, 15, 2050–2055 RSC.
  83. T. Fodi, C. Didaskalou, J. Kupai, G. T. Balogh, P. Huszthy and G. Szekely, ChemSusChem, 2017, 10, 3435–3444 CrossRef CAS PubMed.
  84. M. D. Hopkin, I. R. Baxendale and S. V. Ley, Org. Biomol. Chem., 2013, 11, 1822–1839 RSC.
  85. C. A. Correia, K. Gilmore, D. T. McQuade and P. H. Seeberger, Angew. Chem., Int. Ed., 2015, 54, 4945–4948 CrossRef CAS PubMed.
  86. S. V. Ley, D. E. Fitzpatrick, R. J. Ingham and R. M. Myers, Angew. Chem., Int. Ed., 2015, 54, 3449–3464 CrossRef CAS PubMed.
  87. M. Baumann, Org. Biomol. Chem., 2018, 16, 5946–5954 RSC.
  88. P. Sagmeister, J. D. Williams, C. A. Hone and C. O. Kappe, React. Chem. Eng., 2019, 4, 1571–1578 RSC.
  89. C. F. Carter, H. Lange, S. V. Ley, I. R. Baxendale, B. Wittkamp, J. G. Goode and N. L. Gaunt, Org. Process Res. Dev., 2010, 14, 393–404 CrossRef CAS.
  90. T. Brodmann, P. Koos, A. Metzger, P. Knochel and S. V. Ley, Org. Process Res. Dev., 2012, 16, 1102–1113 CrossRef CAS.
  91. S. T. R. Müller, A. Murat, D. Maillos, P. Lesimple, P. Hellier and T. Wirth, Chem. – Eur. J., 2015, 21, 7016–7020 CrossRef PubMed.
  92. A. Perro, G. Lebourdon, S. Henry, S. Lecomte, L. Servant and S. Marre, React. Chem. Eng., 2016, 1, 577–594 RSC.
  93. A. E. Cervera-Padrell, J. P. Nielsen, M. Jønch Pedersen, K. Müller Christensen, A. R. Mortensen, T. Skovby, K. Dam-Johansen, S. Kiil and K. V. Gernaey, Org. Process Res. Dev., 2012, 16, 901–914 CrossRef CAS.
  94. A. Mitic, A. E. Cervera-Padrell, A. R. Mortensen, T. Skovby, K. Dam-Johansen, I. Javakhishvili, S. Hvilsted and K. V. Gernaey, Org. Process Res. Dev., 2016, 20, 395–402 CrossRef CAS.
  95. T. A. Hamlin and N. E. Leadbeater, Beilstein J. Org. Chem., 2013, 9, 1843–1852 CrossRef PubMed.
  96. G. Chaplain, S. J. Haswell, P. D. I. Fletcher, S. M. Kelly and A. Mansfield, Aust. J. Chem., 2013, 66, 208–212 CrossRef CAS.
  97. J. Bart, A. J. Kolkman, A. J. Oosthoek-de Vries, K. Koch, P. J. Nieuwland, H. Janssen, J. van Bentum, K. A. M. Ampt, F. P. J. T. Rutjes, S. S. Wijmenga, H. Gardeniers and A. P. M. Kentgens, J. Am. Chem. Soc., 2009, 131, 5014–5015 CrossRef CAS PubMed.
  98. C. M. Archambault and N. E. Leadbeater, RSC Adv., 2016, 6, 101171–101177 RSC.
  99. B. Ahmed-Omer, E. Sliwinski, J. P. Cerroti and S. V. Ley, Org. Process Res. Dev., 2016, 20, 1603–1614 CrossRef CAS.
  100. M. V. Gomez and A. de la Hoz, Beilstein J. Org. Chem., 2017, 13, 285–300 CrossRef CAS PubMed.
  101. T. H. Rehm, C. Hofmann, D. Reinhard, H.-J. Kost, P. Löb, M. Besold, K. Welzel, J. Barten, A. Didenko, D. V. Sevenard, B. Lix, A. R. Hillson and S. D. Riegel, React. Chem. Eng., 2017, 2, 315–323 RSC.
  102. World Health Organization Model List of Essential Medicines, https://apps.who.int/iris/bitstream/handle/10665/325771/WHO-MVP-EMP-IAU-2019.06-eng.pdf, (accessed 28 May 2020) Search PubMed.
  103. Top 200 Pharmaceuticals by Retail Sales in 2019, https://njardarson.lab.arizona.edu/sites/njardarson.lab.arizona.edu/files/Top 200 Drugs By Retail Sales in 2019_0.pdf, (accessed 28 May 2020) Search PubMed.
  104. J. L. Howard, C. Schotten and D. L. Browne, React. Chem. Eng., 2017, 2, 281–287 RSC.
  105. L. A. Mitscher, Chem. Rev., 2005, 105, 559–592 CrossRef CAS PubMed.
  106. T. Schwalbe, D. Kadzimirsz and G. Jas, QSAR Comb. Sci., 2005, 24, 758–768 CrossRef CAS.
  107. K. Grohe and H. Heitzer, Liebigs Ann. Chem., 1987, 1987, 29–37 CrossRef.
  108. R. Zerbes, P. Naab, G. Franckowiak and H. Diehl, EP0657448A1, 1995.
  109. H. Lin, C. Dai, T. F. Jamison and K. F. Jensen, Angew. Chem., Int. Ed., 2017, 56, 8870–8873 CrossRef CAS PubMed.
  110. N. P. Tosso, B. K. Desai, E. De Oliveira, J. Wen, J. Tomlin and B. F. Gupton, J. Org. Chem., 2019, 84, 3370–3376 CrossRef CAS PubMed.
  111. Y. Tamura, T. Yakura, Y. Shirouchi and J. Haruta, Chem. Pharm. Bull., 1985, 33, 1097–1103 CrossRef CAS.
  112. A. R. Bogdan, S. L. Poe, D. C. Kubis, S. J. Broadwater and D. T. McQuade, Angew. Chem., Int. Ed., 2009, 48, 8547–8550 CrossRef CAS PubMed.
  113. D. R. Snead and T. F. Jamison, Angew. Chem., Int. Ed., 2015, 54, 983–987 CrossRef CAS PubMed.
  114. M. Baumann and I. R. Baxendale, React. Chem. Eng., 2016, 1, 147–150 RSC.
  115. H. Lee, H. Kim and D. Kim, Chem. – Eur. J., 2019, 25, 11641–11645 CrossRef CAS PubMed.
  116. J. Zimmermann, US5521184A, 1996.
  117. J. C. Yang, D. Niu, B. P. Karsten, F. Lima and S. L. Buchwald, Angew. Chem., Int. Ed., 2016, 55, 2531–2535 CrossRef CAS PubMed.
  118. W. C. Fu and T. F. Jamison, Org. Lett., 2019, 21, 6112–6116 CrossRef CAS.
  119. D. K. Ro, E. M. Paradise, M. Quellet, K. J. Fisher, K. L. Newman, J. M. Ndungu, K. A. Ho, R. A. Eachus, T. S. Ham, J. Kirby, M. C. Y. Chang, S. T. Withers, Y. Shiba, R. Sarpong and J. D. Keasling, Nature, 2006, 440, 940–943 CrossRef CAS PubMed.
  120. N. J. White, Science, 2008, 320, 330–334 CrossRef CAS PubMed.
  121. B. Pieber, T. Glasnov and C. O. Kappe, Chem. – Eur. J., 2015, 21, 4368–4376 CrossRef CAS.
  122. M. P. Feth, K. Rossen and A. Burgard, Org. Process Res. Dev., 2013, 17, 282–293 CrossRef CAS.
  123. F. Lévesque and P. H. Seeberger, Angew. Chem., Int. Ed., 2012, 51, 1706–1709 CrossRef.
  124. D. Kopetzki, F. Lévesque and P. H. Seeberger, Chem. – Eur. J., 2013, 19, 5450–5456 CrossRef CAS PubMed.
  125. K. Gilmore, D. Kopetzki, J. W. Lee, Z. Horváth, D. T. McQuade, A. Seidel-Morgenstern and P. H. Seeberger, Chem. Commun., 2014, 50, 12652–12655 RSC.
  126. S. Triemer, K. Gilmore, G. T. Vu, P. H. Seeberger and A. Seidel-Morgenstern, Angew. Chem., Int. Ed., 2018, 57, 5525–5528 CrossRef CAS PubMed.
  127. S. Hakimian, A. Cheng-Hakimian, G. D. Anderson and J. W. Miller, Expert Opin. Pharmacother., 2007, 8, 1931–1940 CrossRef CAS PubMed.
  128. W. H. Mudd and E. P. Stevens, Tetrahedron Lett., 2010, 51, 3229–3231 CrossRef CAS.
  129. S. Borukhova, T. Noël, B. Metten, E. de Vos and V. Hessel, Green Chem., 2016, 18, 4947–4953 RSC.
  130. M. Escribà-Gelonch, G. A. de Leon Izeppi, D. Kirschneck and V. Hessel, ACS Sustainable Chem. Eng., 2019, 7, 17237–17251 CrossRef PubMed.
  131. M. E. Pierce, R. L. Parsons, L. A. Radesca, Y. S. Lo, S. Silverman, J. R. Moore, Q. Islam, A. Choudhury, J. M. D. Fortunak, D. Nguyen, C. Luo, S. J. Morgan, W. P. Davis, P. N. Confalone, C. Y. Chen, R. D. Tillyer, L. Frey, L. Tan, F. Xu, D. Zhao, A. S. Thompson, E. G. Corley, E. J. J. Grabowski, R. Reamer and P. J. Reider, J. Org. Chem., 1998, 63, 8536–8543 CrossRef CAS.
  132. D. Dai, X. Long, B. Luo, A. Kulesza and Y. Guo, WO2012/097510A1, 2012.
  133. K. Hirano, S. Gondo, N. Punna, E. Tokunaga and N. Shibata, ChemistryOpen, 2019, 8, 406–410 CrossRef CAS.
  134. J. S. Bryans and D. J. Wustrow, Med. Res. Rev., 1999, 19, 149–177 CrossRef CAS.
  135. H. Ishitani, K. Kanai, Y. Saito, T. Tsubogo and S. Kobayashi, Eur. J. Org. Chem., 2017, 2017, 6491–6494 CrossRef CAS.
  136. H. Ishitani, Y. Furiya and S. Kobayashi, Chem. - Asian J., 2020, 15, 1688–1691 CrossRef CAS.
  137. T. J. Reilly, J. Chem. Educ., 1999, 76, 1557 CrossRef CAS.
  138. S. Kashani, R. J. Sullivan, M. Andersen and S. G. Newman, Green Chem., 2018, 20, 1748–1753 RSC.
  139. C. Harrison, Nat. Biotechnol., 2020, 38, 379–381 CrossRef PubMed.
  140. T. Mukaiyama, H. Ishikawa, H. Koshino and Y. Hayashi, Chem. – Eur. J., 2013, 19, 17789–17800 CrossRef CAS PubMed.
  141. Y. Hayashi and S. Ogasawara, Org. Lett., 2016, 18, 3426–3429 CrossRef CAS PubMed.
  142. S. Ogasawara and Y. Hayashi, Synthesis, 2017, 49, 424–428 CAS.
  143. C. R. Sagandira and P. Watts, Synlett, 2020 DOI:10.1055/s-0039-1690878.
  144. C. R. Sagandira and P. Watts, J. Flow Chem., 2019, 9, 79–87 CrossRef CAS.
  145. C. R. Sagandira and P. Watts, Beilstein J. Org. Chem., 2019, 15, 2577–2589 CrossRef CAS.
  146. A. Sun, N. Chandrakumar, J. J. Yoon, R. K. Plemper and J. P. Snyder, Bioorg. Med. Chem. Lett., 2007, 17, 5199–5203 CrossRef CAS.
  147. F. Li, J. Nie, L. Sun, Y. Zheng and J. A. Ma, Angew. Chem., Int. Ed., 2013, 52, 6255–6258 CrossRef CAS.
  148. J. Britton and T. F. Jamison, Angew. Chem., Int. Ed., 2017, 56, 8823–8827 CrossRef CAS PubMed.
  149. J.-S. Poh, D. L. Browne and S. V. Ley, React. Chem. Eng., 2016, 1, 101–105 RSC.
  150. H. Mettler and E. Brig-Gils Greth, CH672124A5, 1978.
  151. B. Gutmann, P. Hanselmann, M. Bersier, D. Roberge and C. O. Kappe, J. Flow Chem., 2017, 7, 46–51 CrossRef CAS.
  152. M. Köckinger, C. A. Hone, B. Gutmann, P. Hanselmann, M. Bersier, A. Torvisco and C. O. Kappe, Org. Process Res. Dev., 2018, 22, 1553–1563 CrossRef.
  153. M. Köckinger, T. Ciaglia, M. Bersier, P. Hanselmann, B. Gutmann and C. O. Kappe, Green Chem., 2018, 20, 108–112 RSC.
  154. B. Musio, E. Gala and S. V. Ley, ACS Sustainable Chem. Eng., 2018, 6, 1489–1495 CrossRef CAS.
  155. S. Korwar, S. Amir, P. N. Tosso, B. K. Desai, C. J. Kong, S. Fadnis, N. S. Telang, S. Ahmad, T. D. Roper and B. F. Gupton, Eur. J. Org. Chem., 2017, 2017, 6495–6498 CrossRef CAS.
  156. S.-T. Chen and J.-M. Fang, J. Chin. Chem. Soc., 2003, 50, 927–930 CrossRef CAS.
  157. Z. Qian, I. R. Baxendale and S. V. Ley, Chem. – Eur. J., 2010, 16, 12342–12348 CrossRef CAS PubMed.
  158. J. Szeto, V.-A. Vu, J. P. Malerich and N. Collins, J. Flow Chem., 2019, 9, 35–42 CrossRef CAS.
  159. T. von Keutz, D. Cantillo and C. O. Kappe, Org. Lett., 2019, 21, 10094–10098 CrossRef CAS PubMed.
  160. Y. Ruetsch, T. Boni and A. Borgeat, Curr. Top. Med. Chem., 2001, 1, 175–182 CrossRef CAS PubMed.
  161. N. S. Suveges, R. O. M. A. de Souza, B. Gutmann and C. O. Kappe, Eur. J. Org. Chem., 2017, 2017, 6511–6517 CrossRef CAS.
  162. D. L. Hughes, Org. Process Res. Dev., 2019, 23, 716–729 CrossRef CAS.
  163. Y. Aoyama, T. Hakogi, Y. Fukui, D. Yamada, T. Ooyama, Y. Nishino, S. Shinomoto, M. Nagai, N. Miyake, Y. Taoda, H. Yoshida and T. Yasukata, Org. Process Res. Dev., 2019, 23, 558–564 CrossRef CAS.
  164. R. E. Ziegler, B. K. Desai, J. A. Jee, B. F. Gupton, T. D. Roper and T. F. Jamison, Angew. Chem., Int. Ed., 2018, 57, 7181–7185 CrossRef CAS PubMed.
  165. S. N. Goodman, M. D. Kowalski, D. M. Mans and H. Wang, US8889877B2, 2014.
  166. H. Wang, M. D. Kowalski, A. S. Lakdawala, F. G. Vogt and L. Wu, Org. Lett., 2015, 17, 564–567 CrossRef CAS.
  167. N. Vasudevan, G. R. Jachak and D. S. Reddy, Eur. J. Org. Chem., 2015, 2015, 7433–7437 CrossRef CAS.
  168. R. Willand-Charnley and P. H. Dussault, J. Org. Chem., 2013, 78, 42–47 CrossRef CAS PubMed.
  169. R. J. Ingham, C. Battilocchio, D. E. Fitzpatrick, E. Sliwinski, J. M. Hawkins and S. V. Ley, Angew. Chem., Int. Ed., 2015, 54, 144–148 CrossRef CAS PubMed.
  170. C. D. Scott, R. Labes, M. Depardieu, C. Battilocchio, M. G. Davidson, S. V. Ley, C. C. Wilson and K. Robertson, React. Chem. Eng., 2018, 3, 631–634 RSC.
  171. D. E. Fitzpatrick, R. J. Mutton and S. V. Ley, React. Chem. Eng., 2018, 3, 799–806 RSC.
  172. C. Battilocchio, J. M. Hawkins and S. V. Ley, Org. Lett., 2014, 16, 1060–1063 CrossRef CAS PubMed.
  173. D. E. Fitzpatrick, C. Battilocchio and S. V. Ley, Org. Process Res. Dev., 2016, 20, 386–394 CrossRef CAS.
  174. C. Battilocchio, S.-H. Lau, J. M. Hawkins and S. V. Ley, Org. Synth., 2017, 94, 34–44 CrossRef CAS.
  175. A. Mullard, Nat. Rev. Drug Discovery, 2015, 14, 669 Search PubMed.
  176. L. Li and P. E. Vlisides, Front. Hum. Neurosci., 2016, 10, 612 Search PubMed.
  177. V.-E. H. Kassin, R. Gérardy, T. Toupy, D. Collin, E. Salvadeo, F. Toussaint, K. Van Hecke and J.-C. M. Monbaliu, Green Chem., 2019, 21, 2952–2966 RSC.
  178. C. L. Stevens, US3254124A, 1966.
  179. S. A. Chambers, J. M. DeSousa, E. D. Huseman and S. D. Townsend, ACS Chem. Neurosci., 2018, 9, 2307–2330 CrossRef CAS PubMed.
  180. Y.-F. Liang and N. Jiao, Angew. Chem., Int. Ed., 2014, 53, 548–552 CrossRef CAS PubMed.
  181. C. A. Hone and C. O. Kappe, Top. Curr. Chem., 2019, 377, 2 CrossRef.
  182. Z. Jaman, T. J. P. Sobreira, A. Mufti, C. R. Ferreira, R. G. Cooks and D. H. Thompson, Org. Process Res. Dev., 2019, 23, 334–341 CrossRef CAS.
  183. C. De Risi, G. Fanton, G. P. Pollini, C. Trapella, F. Valente and V. Zanirato, Tetrahedron: Asymmetry, 2008, 19, 131–155 CrossRef CAS.
  184. A. Porey, S. Santra and J. Guin, J. Org. Chem., 2019, 84, 5313–5327 CrossRef CAS PubMed.
  185. S. B. Ötvös and C. O. Kappe, ChemSusChem, 2020, 13, 1800–1807 CrossRef PubMed.
  186. A. Halama, J. Stach, S. Rádl and K. Benediktová, Org. Process Res. Dev., 2018, 22, 1861–1867 CrossRef CAS.
  187. J. Wang, W. Zeng, S. Li, L. Shen, Z. Gu, Y. Zhang, J. Li, S. Chen and X. Jia, ACS Med. Chem. Lett., 2017, 8, 299–303 CrossRef CAS PubMed.
  188. Q. Meng, T. Zhao, D. Kang, B. Huang, P. Zhan and X. Liu, Chem. Cent. J., 2017, 11, 86 CrossRef PubMed.
  189. M. C. F. C. B. Damião, R. Galaverna, A. P. Kozikowski, J. Eubanks and J. C. Pastre, React. Chem. Eng., 2017, 2, 896–907 RSC.
  190. M. C. F. C. B. Damião, H. M. Marçon and J. C. Pastre, React. Chem. Eng., 2020, 5, 865–872 RSC.
  191. D. J. Lugg, JAMA, J. Am. Med. Assoc., 2000, 283, 2082 CrossRef CAS PubMed.
  192. G. Sipos, T. Bihari, D. Milánkovich and F. Darvas, J. Flow Chem., 2017, 7, 151–156 CrossRef CAS.
  193. Novel Platform for Flow Chemistry Facilitation is Launching to the ISS, https://www.issnationallab.org/blog/boston-university-flow-chemistry-facilitation-spacexcrs20/, (accessed 29 June 2020) Search PubMed.