flow reactor networks for integrated synthesis of active … · flow reactor networks for...

Flow reactor networks for integrated synthesis of activepharmaceutical ingredientsCitation for published version (APA):Borukhova, S. (2016). Flow reactor networks for integrated synthesis of active pharmaceutical ingredients.Technische Universiteit Eindhoven.

Document status and date:Published: 09/11/2016

Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne

Take down policyIf you believe that this document breaches copyright please contact us at:[email protected] details and we will investigate your claim.

Download date: 01. Feb. 2021

https://research.tue.nl/en/publications/flow-reactor-networks-for-integrated-synthesis-of-active-pharmaceutical-ingredients(2f600aed-537e-44a2-a8a2-1675bc2a141c).html

Flow Reactor Networks for integrated synthesis of active pharmaceutical ingredients

Svetlana Borukhova Semenovna

This research was funded by the European Research Council (ERC) advanced grant on

“Novel Process Windows- Boosted Micro Process Technology” under grant agreement

number 267443.



Eindhoven University of Technology, 2016

A catalogue record is available from the Eindhoven University of Technology Library

ISBN 978-90-386-4169-0

Printed by Ipskamp Printing

Cover design Svetlana Borukhova and Ilse Stronks, Persoonlijk Proefschrift


PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de rector magnificus prof.dr.ir. F.P.T. Baaijens, voor een commissie aangewezen door het College voor Promoties, in het

openbaar te verdedigen op woensdag 9 november 2016 om 16:00 uur

door


geboren te Tashkent, Oezbekistan

SUMMARY

The current thesis presents the realization of the Novel Process Windows (NPW) concept,

which relies on two intensification principles. The first is the chemical intensification of a

single reaction step with the purpose of intensifying the intrinsic kinetics. This is usually

achieved by confining the reactants within a system under high temperature, high pressure

and high concentration. The second principle is process-design intensification, which aims

at integrating reaction steps and simplifying the overall process. In this context, the

thesis describes the development of flow reactor networks with the goal of intensifying

single reaction steps and subsequently constructing a multi-step process to deliver active

pharmaceutical ingredients. In order to deliver sustainable processes, special attention

is given to the chemical reagents used and their characteristics such as toxicity, chemical

stability and cost. Green chemistry and green engineering principles serve as a guideline

in the decision making process.

Organochlorides represent a class of valuable intermediates in the pharmaceutical

industry. The biggest concern associated with the way organochlorides are prepared is

the use of chlorinating agents. They are used in stoichiometric or excessive amounts

leading to a low atom economy and a substantial generation of waste. To address this

concern, chapter 2 presents a process where hydrochloric acid is used to convert alcohols

to the corresponding chlorides. Hydrochloric acid presents a valid alternative to the

toxic and wasteful chlorinating agents since only water is generated as a by-product. The

synthesized chlorides are further used in a micro flow network consisting of multi-step

synthesis of antiemetic drugs and antihistamines. Cinnarizine and a buclizine derivative

were constructed from bulk alcohols in a 6-stage 4-reaction step processes. The whole

sequence was realized in 90 minutes with good overall yields (>80%) at a production rate

of 2 mmol of final product per hour. The drawbacks of the process were the excessive

amount of hydrochloric acid required and the use of a reagent loop, leading to only

a partial continuity of the process. Chapter 3 presents a solution to aforementioned

drawbacks, in the form of a fully continuous process with the use of 1.2 equivalents of

hydrogen chloride. The process was superior due to the switch from hydrochloric acid to

pure hydrogen chloride gas. Furthermore, water use minimization was achieved due to

the switch from hydrochloric acid, where hydrogen chloride gas is dissolved in water, to

pure gas. Safe pressurization in micro flow allowed the maximization of hydrogen chloride

solubility in alcohol and its local concentration even further. To the best of our knowledge,

this was the first documented use of pure hydrogen chloride gas in a continuous process.

High temperature, pressure and concentration are the main tools in chemical

intensification as mentioned above. Chapter 4 aims to investigate the effects of pressure

Dit proefschrift is goedgekeurd door de promotoren en de samenstelling van de promotiecommissie is als volgt: voorzitter: prof.dr.ir. J.C. Schouten 1e promotor: prof.dr.ir. V. Hessel copromotor(en): dr. T. Noël leden: prof.dr. R.P. Sijbesma prof.dr. C.O. Kappe (University of Graz) prof.dr. T. Wirth (Cardiff University) prof.dr. F.P.J.T. Rutjes (Radboud University) adviseur: dr. B. Metten (Ajinomoto OmniChem N.V.)

Het onderzoek dat in dit proefschrift wordt beschreven is uitgevoerd in overeenstemming met de TU/e Gedragscode Wetenschapsbeoefening.

on chemical reaction kinetics and regioselectivity of 1,3-dipolar cycloaddition. Increasing

pressure within the reactor from 500 bar to 1800 bar favored the selectivity demonstrated

by a 30% increase of the 1,4-regioisomer of cycloadduct. Meanwhile, only a marginal

increase of 8% in conversion leading to a final conversion of 78% was observed. Bringing

the reaction system into a continuous-flow micro reactor did not improve the conversion

when high pressure was used. Instead, it was shown that high pressure was a good tool

to achieve superheated conditions allowing a greater extent of chemical intensification

by temperature. Chapter 5 concentrates on a further intensification of 1,3-dipolar

cycloaddition to produce a rufinamide precursor in a continuous mode using inexpensive

and relatively green dipolarophile. Due to the low reactivity the reaction in batch takes

more than 24 hrs. When the cycloaddition was performed in a micro flow capillary reactor

under neat conditions at 210 ̊C and 70 bar, reaction time decreased to 10 min. The

rufinamide precursorposed a problem of blockage within the reactor of micro dimensions,

as it is in its solid state under atmospheric conditions. In order to avoid crystallization of

the cycloadduct, an anti-solvent was introduced into the hot product flow in its molten

state. Upon cooling in the collection vessel, the product precipitated in its crystalline

form. Such an approach allowed to merge the reaction and separation stages, simplifying

the process.

Chapter 6 describes the combination of the single-step intensifications described earlier

to develop a continuous process of a rufinamide precursor and realize process-design

intensification. 2,6-dilfuorobenzyl alcohol was converted to the corresponding chloride

through the use of hydrogen chloride gas. Subsequently, without intermediate separation,

the chloride was converted to 2,6-dilfuorobenzyl azide. The latter was then separated in

an in-house developed inline L-L separator and further fed into a final reactor to undergo

the aforementioned cycloaddition to an inactive, but green dipolarophile. Upon collection

cycloadduct crashed out of the anti-solvent in 82% overall yield. The entire process relied

on no solvent, nor catalyst, and a minimum amount of water. Thus, an own momentum of

process-design intensification was identified, which is more than the added momenta of

chemical intensification.

Chapter 7 presents an evaluation of the developed 5-stage 3-reaction step process to

produce a rufinamide precursor. Batch and flow process models were developed to be

compared in terms of material and energy requirements. In the batch process, large

amounts of organic solvents are used. Moreover, intermediate separations require

a significant amount of water. Thus, when compared against the batch process, the

continuous flow process is more efficient in terms of material use and waste generation.

The cost of the equipment was calculated to be similar, while operating expenditures are

expected to be lower for continuous flow process than for batch. A life cycle assessment

showed the environmental profile of process in terms of environmental, toxicity and health

aspects. The human toxicity factor was found to be marginal in the case of the continuous

flow process as opposed to the batch process. The global warming potential of the batch

process was found to be remarkably higher than that of the flow process. This is mainly

attributed to the air emissions which account for 53% of the total GWP profile of the batch

process. The next contributory factor is the energy needed for solvent production. In

overall, the continuous-flow process was found to be more favorable in terms of economic

and environmental impacts. Finally, chapter 8 gives an overview of the research impacts

achieved in each area related to continuous manufacturing of pharmaceuticals in micro

flow.

The current thesis represents one branch of a concerted research cluster, supported by

the ERC Advanced Grant “Novel Process Windows – Boosted Micro Process Technology”

(no. 267443). As mentioned above, it covers comprehensive investigations of NPWs using

the classical chemical-engineering activation means, such as high temperature, pressure

and concentration. The activation means are exploited within the exploration of new

chemical transformations taking place in constructed flow reactor networks for which the

achievement of process simplification was crucial. This connects to the other principle of

NPW, process design intensification, explored in the thesis of I. Vural-Gursel (ISBN: 978-90-

386-3957-4). The overall concerted NPW approach is fully realized when investigation of

green solvents’ use for NPWs (thesis S. Stouten, ISBN: 978-94-6233-355-0), exploration of

new activation means via electromagnetic waves (thesis of E. Shahbazali, to be published

in March 2017), and assurance of intrinsic safety provided in micro-flow (thesis of M.

Shang, ISBN: 978-90-386-4061-7) are demonstrated.

To my men, Mirotje and Marktje.

ACKNOWLEDGEMENTS

While still living in Uzbekistan, I remember learning that our tall and handsome neighbor

was a PhD student, a term I never heard before at the age of 11. After inquiring about the

prerequisites, as in having completed bachelors and masters, benefits, as in how high was

the salary, and future prospects, as in whether a job was guaranteed, I concluded that our

neighbor was a naive lunatic with no future prospects. Yet, here I am at the end of my PhD

years needing to say a bittersweet good bye. Over the past four years, I had the pleasure

to work among talented and inspiring people who enabled me in multiple aspects.

Foremost, I would like to thank my best friend and partner in each and every crime, Mark.

Thank you for motivating me through the times of multiple clogging incidents and pump

malfunctions, for being my wall when I needed to throw my thoughts around and endlessly

listening about novel process windows and micro flow chemistry. You kept me healthy by

bringing food into the office when I wanted to work late. You assured my safety when I

worked in the lab in the evenings and weekends. You proofread every paper I wrote and

celebrated their acceptance with numerous chocolate fondues.

I was lucky to have several mentors that indulged me with their utmost care, broad

knowledge and profound interest. Dear Volker, thank you for all your work and time

invested in acquiring the ERC grant. Thanks to your wise management, there was never

lack of resources to realize our ideas. I appreciate your trust in my capabilities that

allowed me to work and develop freely. Thanks to your vast network and excellent

reputation, I benefited from useful discussions with numerous experts in the times of

stagnation. You taught me how to construct and tell a story about research. Your drive

to publish might have appeared as too demanding at first, but now I realize that it made

my life so much easier towards the end of my PhD. Thank you for sending me to various

conferences and giving me feedback on every abstract and presentation despite your

busy schedule.

I would also like to express my special gratitude to Tim for our discussions regarding

experiments and for polishing our papers. Furthermore, thank you for keeping me updated

about group and department dynamics, for more than just a glimpse on one year spent in

MIT and for showing me how efficient progress meetings can be.

Dear Bert, I consider myself the luckiest PhD to have an advisor like you. Throughout

the years you gave me lots of ideas to work on and your input into our papers was

indispensable. I enjoyed our weekly calls and hopefully will be working with you in the

future. Thank you for always being there!

Dear Erik, thank you for not only building my very first experimental rig, but also teaching

me how to assemble the reactor networks. You helped me to become independent in the

design of experimental setups. I enjoyed our multiple discussions and your positive outlook.

My special gratitude goes to the Flow Chemistry community, especially Klavs Jensen, Oliver

Kappe, Timothy Jamison, Andrea Adamo, Steven Ley, Peter Seeberger, Norbert Kockmann

and Dominique Roberge. I enjoyed following your work over the years. I would also like to

thank my committee members. I appreciate your time and comments regarding the work

described in this thesis.

I would also like to express my immense gratitude to our group’s secretaries. Dear Agnes,

you were there right from the beginning. Your warm and patient attitude helped me in

meeting the deadlines. Dear Denise, thank you for breaking all the required procedures

down into handleable tasks. I loved the apple cakes you made for your birthdays. Dear

Jose, thank you for being patient with my inability to fill out the time sheet right.

Dear Carlo, thank you for making sure that our work environment was a safe place. Dear

Marlies and Peter, I appreciate your efforts in keeping everything running smoothly.

Bart van Heugten, thank you for sharing your expertise and helping me in the design of

hydrogen chloride setup.

I would also like to thank Prof. Markus Busch for making it possible to work on high

pressure experiments in his lab. Andreas Seeger, thank you for building high pressure

setup and assuring our safety during the experiments in those bunkers.

I deeply appreciate the support of my colleagues. Dear John, thank you for numerous

discussions not only regarding my project, but also personal life. You challenged my way

of thinking, my approach and contributed to my personal development. You are fun to

be around. Dear Jan Meuldijk, thank you for joining every group meeting where I had to

present. Your questions helped me in writing papers. Martin Timmer, thank you for your

humor and making me question some of my assertions. I hope one day we both will live

in Portugal.

Super strong Slavisa, even though I was tough to warm up to, by sitting in my hoodie with

headphones on or for example leaving a piece of cake on your desk on my birthday, while

saying that there was no special occasion, you were like a brother to me as soon as we

got to know each other. Thank you for listening to all my frustrations about clogging and

unsuccessful experiments. Thank you for all the high fives when things did work. Thank

you for all your support.

Lovely Catherine, your moral support and a constant supply of Nutella-filled muffins was

indispensable. I miss our talks already. I know that you will land a great job in UK and

climb up career ladder with a speed of a tachyon.

Nico! Thank you for listening to my ideas and trust whenever I had a new plan of action.

Hannes! Thank you for being the most cheerful guy in the group and certainly for coming

to my spinning classes. Sweet Iris, it was always a pleasure to chat with you and I am happy

that we maintain our friendship even after you left the group. I would also like to thank

Leticia, Smitha, Elnaz, Ceci, Minjing, Emila, Violetta, Fernanda, Dario, Stefan, Myrto,

Carlos, Nathan, Shohreh, Jannelies, Paola, Michiel, Alvaro and Wim for your contributions

into making SCR a cosy place to work. I will miss your smiles!

I could not have done it without my supportive friends and family. Karina, Jx, Anya,

Maryam, Sander and Frederico I appreciate your presence during all the fun things we

did. Sander, I will miss our water breaks and your cookies. I would also like to thank Lara

for not only being my friend, but also for giving me an opportunity to work as a spinning

instructor in Eindhoven Student Sports Center.

Finally, I thank my little wonder, Miro, who has been the biggest motivation to stay

focused.

TABLE OF CONTENTS

1. Introduction 19

1.1. Continuous manufacturing in the pharmaceutical industry

1.2. Micro flow chemistry

1.2.1 Classification of reactions in micro flow

1.2.2 Classification of micro flow reactors

1.3. Green chemistry and green engineering

1.4. Chemical and Process-Design Intensification via Novel Process Windows

1.5. Multi-step synthesis of active pharmaceutical ingredients in micro flow

1.5.1 Aliskiren

1.5.2 Artemisinin

1.5.3 Gleevec

1.5.4 Ibuprofen

1.5.5 Rufinamide

1.6. Multi-step syntheses of active pharmaceutical ingredients in meso and

large-scale flow reactors

1.6.1 Hydroxypyrrolotriazine

1.6.2 2,2-Dimethylchromenes

1.6.3 Fused-Bicyclic Isoxazolidines

1.6.4 7-Ethyltryptophol on the way to etodolac

1.6.5 6-Hydroxybuspirone

1.6. Scope and outline of the thesis

References

2. Continuous-flow multi-step synthesis of Cinnarizine, Cyclizine and a Buclizine derivative from bulk alcohols

45

2.1. Introduction

2.2. Synthesis of alkyl chlorides

2.3. Utilization of benzyl chloride

2.4. Applications in API synthesis

2.4.1. Synthesis of Cyclizine

2.4.2. Synthesis of meclizine and buclizine derivatives

2.4.3. Synthesis of cinnarizine

2.5. Conclusion

References

3. Hydrogen chloride gas in solvent-free continuous conversion of alcohols to chlorides in micro flow

59

3.1. Introduction

3.2 Experimental platform

3.3 Results and Discussion

3.3.1 Expansion of scope

3.4 Conclusion

References

4. Pressure accelerated azide-alkyne cycloaddition 75

4.1. Introduction

4.2. Operating platforms

4.2.1 Process windows of reactors under study

4.3. Results and Discussion

4.3.1 Exploring the pressure and temperature windows in the standard

and autoclave batch reactors

4.3.2 Exploring the pressure, temperature and concentration windows in

the micro capillary reactor

4.3.3 Exploring the concentration, pressure and temperature windows in

the micro capillary reactor with catalyst

4.4. Conclusion

4.5. Outlook

References

5. Solvent- and catalyst-free Huisgen cycloaddition towards Rufinamide 95

5.1. Introduction

5.2. Results and Discussion

5.3. Conclusion

References

6. From alcohol to 1,2,3-triazole via multi-step continuous-flow synthesis of rufinamide precursor

107

6.1. Introduction

6.2. Background

6.3. Step 1- Chlorodehydroxylation

6.4. Step 2- Synthesis of azide

6.5. Step 1+2+ separation

6.6. Step 1+2+separation+3+crystallization

6.6. Conclusion

References

7. Economic and life-cycle assessment of integrated continuous and batch manufacturing of rufinamide

121

7.1. Introduction

7.2. Process model development

7.2.1. Starting point- lab-scale continuous-flow process

7.2.2. Scale-up of continuous-flow process

7.2.3. Batch process

7.3. Energy requirements

7.4. Materials consumption

7.5. Equipment costs

7.6. Environmental analysis

7. 6.1. Process Mass Intensity

7. 6.2. Life cycle assessment

7.7. Conclusion

References

8. Research impact and further recommendations 141

8.1 Continuous manufacturing of active pharmaceutical ingredients via multi-

step continuous-flow processes

8.2 Micro flow process technology

8.3 Green chemistry and green engineering

8.4 Novel Process Windows

List of publications 149

CHAPTER 1Introduction

This chapter is based on :

Borukhova, S., Hessel, V. (2017). Continuous manufacturing of active pharmaceutical ingredients in micro flow. In J. Khinast, P. Kleinbudde & J. Rantanen (Eds.), Continuous Manufacturing of Pharmaceuticals Wiley-VCH.

Borukhova, S., Hessel, V. (2013). Micro process technology and novel process windows : giving three intensification fields. In K. Boodhoo & A. Harvey (Eds.), Process intensification for green chemistry : engineering solutions for sustainable chemical processing (pp. 91-157). Wiley-VCH.

20

CHAPTER 1 Introduction

21

1

1.1. CONTINUOUS MANUFACTURING IN THE PHARMACEUTICAL INDUSTRY

The pharmaceutical industry continuously investigates, discovers, develops and delivers

new treatments to existing and emerging diseases1. Meanwhile, it faces the challenges

such as slower success rate, expirations of patents, intense price competition with generics

and tighter constraints due to environmental concerns2. According to a recent Pharma

Profile report less than 12% (Figure 1) of the candidate medicines that are successful in

phase I clinical trials are approved by the FDA3. Developing a new medicine on average

takes at least 10 years and costs $2.6 billion. The economic pressure has been on since

early 2000s, when the number of approved medicines has essentially decreased, while

research and development expenditures soared as shown in Figure 24.

Malet-Sanz et al. described a usual sequence of events in drug development carried out

in Pfizer2. From the beginning of the development, a large number of hits is identified,

followed by the lead discovery stage, where the structures are varied to deliver the desired

target activity, selectivity, metabolism and pharmacokinetics. At this stage only 5-10 mg

of the compound is needed for primary testing. Selected compounds are categorized

into 1-3 series based on their structural similarity. At this point less than 100 compounds

are prepared in the amounts of 500 mg to 1 g for pharmacokinetic studies. Within the

investigated compounds a particularly interesting compound will then be synthesized to

yield 1-2 g of solid material for pharmacokinetic and toxicity studies. If the study results

in a promising fate for a candidate, a first exploratory toxicology study will require 5-30

g of the compound, followed by yet another scale-up to deliver 30-500 g for second

species toxicity studies. In case of a positive outcome the commercial route to drug will

be designed and kilograms of the drug will be needed for the clinical trials. As the product

moves along the development process, production batches get larger as more material

is required. Figure 3 shows how the process development of a drug consists of a series

of scale-ups. Consequently, process that can operate at various scales with a constant

quality becomes one of the necessities in the pharmaceutical industry. Continuous

processing offers a variable production volume that serves as a solution at every stage

of the development process. Moreover, early investigation of the continuous processing

leads to a better understanding of process and of the critical product attributes, which

in turn contributes to an easy transition from medicinal chemistry to commercial route.

In pharmaceutical industry chemical reactions are run continuously in either continuous

flow stirred-tank (CSTR) or plug flow reactors (PFR)5. Despite their wide applicability and

mechanical simplicity, CSTRs operate in highly nonlinear mode in terms of the degree of

mixing and fluid motion. The complexity grows with the increasing number of phases,

presenting a great challenge during the switch from lab to a larger scale. As opposed to

CSTRs, PFRs provide a more uniform environment for the reactants and a more efficient

use of a reactor volume. Uniform environment due to the enhanced mixing on a molecular

level can be achieved when static mixers6,7 or mechanical elements, such as spinning

disc or rotating tube-in-tube, are incorporated8–12. In addition, packed bed reactors

Figure 1. The timeline and intermediate stages of drug development3.

Figure 2. Research and development expenditures adjusted for inflation. A 3-year moving average for approved new chemical entities. Source: Tufts CSDD; PhRMA, 2014 Industry Profile.

22


23

1

with immobilized catalyst13,14 or enzymes15,16 benefit from decreased diffusion path and

increased contact area leading to higher reaction rates per given volume. Further control

over operating parameters and outcomes can be achieved when micro flow reactors17,18

are used, as discussed in the following section.

1.2. MICRO FLOW CHEMISTRY

Micro reactors came into play within organic synthesis since the advancement of

microfabrication techniques in 1990s20. Microfabrication allowed the production of reactors

with channel size ranging from submicrometers to submillimeters. The main advantage

of the decreased channel dimensions, under otherwise same operating parameters, is

the increase in the Reynolds number, which leads to a decrease in the diffusion path and

increase in the specific heat and mass transfer areas, resulting in a high control over

the temperature and reactant distribution. The biggest drawback, however, is the large

pressure drop along the narrower channels, which leads to the requirement of lower flow

rates resulting in laminar flows. Laminar flow is a constant unidirectional flow of fluid

elements that move with a constant velocity with respect to the stream axis. Viscous

forces dominate over inertial forces, resulting in an undesired well-defined parabolic flow

that results in wider residence time distribution. Thus, the convective mass transfer takes

place along the direction of the fluid only, limiting the mixing of species to diffusional

transport.

Discovery Rapid Process Development

Scale-up Scale-up Scale-up Scale-up

Make product

Make product

Make product

Make product

Validation Technology Relocation

Production

mg kg

Figure 3. Representation of intermediate scale-up steps within the drug development process19.

Similarly to the large scale industrial PFRs, diffusional transport of mass and heat can

be enhanced in micro reactors when secondary flows are created by active or passive

methods21,22. Active methods rely on an external source of energy, such as ultrasound,

microwave or piezoelectricity21. Certain mechanical tools can be incorporated as well,

such as actuators or pulsating walls21,23,24. However, due to the simplicity and lower risk

of mechanical malfunction passive methods are preferred over active. Passive methods

rely on the flow energy gained from the pressure provided by an upstream pump and

on the geometry of the micro channels. Based on the geometry and size of the channel

multiple lamellae can be formed leading to a drastic decrease in the diffusion path25. Flow

focusing, split-and-recombine, colliding jets and recirculation techniques can further be

used to decrease the diffusion path and increase heat and mass transfer22. As a result,

plug flow behaviour and narrow residence time distibution are attained.

Given the advantages of operation within the micro channel, micro flow environment

on its own provides two classes of intensification: mass transfer intensification and

heat transfer intensification26. Mass transfer intensification provides a more uniform

distribution of reactants and products due to a better mixing, which leads in some cases to

higher selectivity and faster reaction times27. Decreasing channel size increases the heat-

transfer coefficient, which in combination with increased surface area maximizes the heat

transfer. In this way, the formation of hot spots is minimized and thermal overshooting

is avoided, leading to a safer operation and in some cases higher selectivity28,29. Finally,

due to the fact that micro flow reactors can be used in continuous processes, the process

can not only be intensified due to the mass and heat transfer intensification, but also by

establishing a continuous process when batch chemistry is transferred into continuous31–33.

1.2.1 Classification of reactions in micro flowDespite the intensification brought by micro flow reactors, not every reaction can be

carried out in micro reactors. Roberge and coworkers analyzed 22 large scale processes

employed in fine chemicals and pharmaceuticals synthesis in Lonza Exclusive Synthesis34.

Based on the analysis, 50% of the reactions can benefit when carried out in micro flow.

The reactions are categorized based on their kinetics into four classes, A, B, C and D.

Class A represents extremely fast and exothermic reactions that occur within the mixing

zone and completed within a seconds range35–38. Tight control over heat transfer minimizes

the formation of hot spots leading to higher selectivity in parallel-competitive reactions.

Meanwhile, better mixing and accurate stoichiometry minimizes consecutive-competitive

reactions. Class A reactions represent reaction usually carried out at cryogenic conditions.

Class B reaction are exothermic reactions that are completed in less than 10 min time

and benefit from better mixing and better heat transfer, while still requiring some time

to react due to slower reaction kinetics. Class C encompasses slower reactions that are

24


25

1

kinetically limited and pose some safety concerns, such as autocatalysis and thermal

accumulation. Finally, class D reactions proceed very slowly the in hours and day range,

and need an intensification of intrinsic kinetics by harsh process conditions.

1.2.2 Classification of micro flow reactorsReactions can be carried out in various flow reactors that are fabricated from various

materials depending on the particular application. Chemical compatibility and physical

robustness are two main factors determining the choice of the material. Reactors

can be constructed from glass, quartz, diamond, polymethylmethacrylate (PMMA),

Figure 4. Flow rector types: Right-chip or plate type reactor; left- tubular coil reactor30.

Figure 5. Left: Alfa Laval ART® Plate Reactor 49 processes up to 500 l/h; Right: Custom-made tubular coil reactor of 73 L internal volume.

polydimethylsiloxane (PDMS), polytetrafluoroethylene (PTFE), perfluoroalkoxy alkane

(PFA), fluorinated ethylene propylene (FEP), ethylene tetrafluoroethylene (ETFE),

ceramic, silicone, copper, stainless steel, nickel, and Hastelloy. Based on the phase of

process development and/or production rate requirements, reactors can range in scale:

‒ Micro scale flow reactors are based on a channel of dimensions lower than a

millimeter. They can be fabricated as chips, plate reactors, straight tube, coiled

tube or packed bed reactors. Micro scale flow reactors are mainly used for process

optimization and/or to prepare small amounts of material. Macchi and Roberge et

al. presented a toolbox approach in transferring batch processes to continuous,

where reactors are selected based on the reaction types and reacting phases30.

Knowledge of the kinetics of a particular reaction determines the choice of a

particular reactor. Very fast, exothermic (Class A and B) reactions benefit from

smaller dimensions, high mass and heat transfer, and therefore are usually

performed in chips or plate-type reactors39. Meanwhile, slower reactions are

carried out in coil or tubular reactor (Class C and D) of larger dimensions and

volumes40. Reactions requiring heterogeneous catalysts are performed in packed

bed reactors41.

‒ Mesoscale flow reactors have higher production rate when compared to micro scale

flow reactors. An increase in productivity can be achieved via stacking of multiple

micro scale flow reactors or parallelization, due to increase in flow rates and/or

an increase in a channel diameter. Furthermore, combination of the techniques is

possible. It is essential that fluid dynamics and mixing characteristic remain the

same throughout the scale-up of the process.

‒ Large scale flow reactors (Figure 5) can deliver more than 100 megatons per year

and can be fabricated based on the same techniques as mesoscale flow reactors.

Companies such as Chemineer, Alfa Laval, Chemtrix and Sulzer Chemtech produce

ready-to-use large scale flow reactors.

1.3. GREEN CHEMISTRY AND GREEN ENGINEERING

In 2005, continuous processing and process intensification were defined as key green

engineering research areas for sustainable manufacturing by the American Chemical

Society (ACS) Green Chemistry Institute (GCI) Pharmaceutical Roundtable42. Figure 6 shows

the main drivers to implement continuous manufacturing43,44. Logistics and quality aspects

can be improved through the advantages brought by flow chemistry in production capacity,

speed of implementation and minimization of inventory and validation stocks. Production

capacity is limited in batch by limited throughput when slow dosing, low temperature

26


27

1

and high dilution are required. Continuous processing can accelerate the slowest steps

and contribute to faster deliveries of the material based on the medicinal chemistry

route. Speed of implementation can be significantly increased since there is no need for a

sequential scale-up and construction of different scale facilities. Use of the same equipment

to produce both clinical and commercial products decreases the total development time.

Moreover, the construction of flow plants can be seen as operational expenditures rather

than capital costs. Inventory and validation stocks can be minimized due to the precise

overlap between supply and demand due to the variable operational volume.

Flow technology offers a larger playground in terms of process conditions. Reaction

temperatures below -40 ̊C or above 200 ̊C are outside the limits of standard batch reactors,

however safely reachable in micro flow. Moreover, cooling or heating the whole installation

to those limits takes a longer time and requires more energy input than in the case of

flow reactors. Flow reactors allow safe operation in that thermal range along with the

creation of novel environments for the molecules to react in the most efficient way with

minimization of energy input and reaction time18,45. Tight control over process conditions

improves selectivity and thus the yield of a reaction46. Finally, in downstream processing

continuous countercurrent extractions have been shown to simplify the separation in case

of unfavorable extraction coefficients47.

Safety concerns associated with highly exothermic reactions can be minimized due to the

efficient heat removal and temperature control available in flow reactors48. Moreover,

Drivers to implement continuous processing

Logistics & Quality Chemistry and Process Safety

Production capacity

Implementation speed

Inventory

Thermal limits

Reaction Yield

Downstream processing

Heat Removal

High pressure

Head space

Manual handling

Figure 6. Main driver to implement continuous manufacturing of pharmaceuticals43.

due to a smaller operating volume only limited amounts of chemicals engage in a failure

scenario as opposed to a whole batch contents in batch procedures. High pressure facilities

require special precautions during the construction. Meanwhile, operating with hazardous,

flammable, explosive or toxic gases presents higher risk in case of over-pressure and/

or leakages. Again due to a smaller hold-up of the flow reactors pressurization can be

achieved with a smaller amount of gas leading to a limited amount available at a time.

The absence of headspace in liquid-filled reactors prevents condensation of low-boiling

hazardous substances that normally can lead to an explosion (e.g., hydrazoic acid)50,51.

Manual handling can be minimized through automatization. Moreover, used disposable

flow reactors can be implemented if highly potent or cytotoxic chemicals are used to

avoid contamination.

1.4. CHEMICAL AND PROCESS-DESIGN INTENSIFICATION VIA NOVEL PROCESS WINDOWS

In 2008, Hessel introduced the concept of Novel Process Windows (NPW) into the field

of micro reaction technology26,49,52,53. Novel Process Windows reside on two mutually

connected pillars of process design and development as shown in Figure 7. First pillar

is the chemical intensification, where harsh operating conditions intensify the intrinsic

reaction kinetics with the target of maximizing the productivity. Chemical intensification

has been observed in multiple cases when temperature, pressure and/or concentration

are increased54–59. All of the parameters affect reaction rate and thus present a potential

to speed-up a reaction. Temperature affects reaction rate according to the well-known

NPW

TCA p

New chemical transformations

Safe explosiveregime

Process simplification& integration

Figure 7. Schematic representation of Novel Process Windows49.

28


29

1

Arrhenius equation. An increase in temperature increases energy of the system and

facilitates the shift from reactants to products. Micro reactors provide an opportunity

for rapid heat transfer and thus safe operation under non-classical conditions, such as

elevated temperature and pressure. Heating the reaction medium beyond its boiling point

demands an increase in pressure (high-p) and results in superheated conditions. Apart

from facilitating superheated conditions, high pressure affects the reaction by influencing

the reaction equilibrium and rate constants. In case volume occupied by product of the

reaction is lower than that of the sum of two separate reactants, the overall volume

change is negative. Thus, increase in pressure will have an acceleration effect on the

reaction rate constant at constant temperature. Affected reactions are cycloadditions,

condensations, reactions proceeding via cyclic transition state, such as Cope and Claisen

rearrangements, reactions involving formation of dipolar transition states, such as

electrophilic aromatic substitutions, and reactions with steric hindrance60.

Second pillar of NPW is the process-design intensification, which forces integration of

intensified step changes into an existing process or to rethink the whole sequence of

process steps with the aim to simplify and intensify the process as a whole61,62.

New chemical transformations occur at the interface of chemical and process-design

intensification. Examples are multi-step synthesis comprising a flow reactor network and

one-pot processes translated into telescoped one-flow processes63. Based on the large-

scale process analysis performed by Roberge and coworkers the average production

campaign involves 2.7 work-up unit operations for 2.1 reaction unit operations34. Thus,

in order to be industrially relevant multi-step flow processes have to be developed with

connected reaction steps omitting intermediate separations when possible.

1.5. MULTI-STEP SYNTHESIS OF ACTIVE PHARMACEUTICAL INGREDIENTS IN MICRO FLOW

Multiple exemplary uninterrupted multi-step processes to deliver active pharmaceutical

ingredients have been published64,17. Uninterrupted in this case means that a particular

intermediate does not undergo any batch unit-operations or manual handling, apart from

being collected in a reservoir with an inserted feed line for the next reactor on its way to

the subsequent reaction. Moreover, the reactants are introduced into a reaction platform

continuously and not based via loops, which contain only limited amount of reactants and

thus result in a limited amount of operating time in between the refills.

1.5.1 Aliskiren The most extensive and complete end-to-end synthesis performed in micro flow reactors

was developed by Trout et al. in collaboration with Novartis International AG65. The target

API in this process was aliskiren hemifumarate, shown in Figure 8, existing under the

brand name of Tekturna (US) or Rasilez (UK). The API is used in the management of

hypertension.

An end-to-end integrated continuous manufacturing plant was built to manufacture tablets

containing 112 mg of the free base form of aliskiren with a total nominal throughput of

45 g/h, which corresponds to 2.7 million tablets per year. The capacity ranges from 20

to 100 g/h. The plant fits into a conventional university building and occupies a 2.4 by

7.3 m2 area. The process, shown in Figure 8 (right), starts off with the introduction of

a melted intermediate at 100 ̊C into a tubular reactor to undergo an amide formation

under neat conditions. Performing the step in flow intensified the reaction, which led to

a decrease in reaction time from 72 h needed in batch down to 4 h in flow. The amide

is extracted in the in-line membrane based liquid-liquid extractor with the addition of

water and ethyl acetate. The typical recrystallization and filtration is translated by the

addition of heptane with a subsequent cooling of the suspension to 5 ̊C. The slurry was

Figure 8. Left: Synthetic steps from intermediate 1 to aliskiren hemifumarate. Right: Process flow diagram including the major unit operations. R- reactor, S- separation, Cr-crystallization, W- filter/wash, D- dilution tank, E- extruder, MD- mold. (Reproduced from Ref. 65 with permission of John Wiley & Sons, Inc.)

30


31

1

filtered on a rotating porous plate while vacuum was applied to the back of the plate. The

slurry is then washed with ethanol and ethyl acetate, prior to being scraped off and sent

to another vessel. A density flow cell was integrated to adjust the flow rate of the ethyl

acetate and control the concentration of 4 at the outlet. The stream is then sent to a

second reactor for an acid-catalyzed deprotection of the amine. The reaction is quenched

with sodium hydroxide which proceeds in an isothermal manner in a flow reactor. The

product stream is worked up by the addition of ethyl acetate to dilute the product to

the concentration of 6 wt%. The sodium chloride formed upon the neutralization of

hydrochloric acid precipitates within the flow, which is than filtered via a microfiltration

membrane and the concentration is measured with an inline UV flow cell. Finally the

flow passes through a column packed with molecular sieves in order to remove traces of

water. The final product is obtained by performing reactive crystallization with fumaric

acid. The addition of fumaric acid is controlled by an inline UV detector to maintain 0.55

equivalents of the acid with respect to the amine.

The crystallization of 6 takes place in two stages, the first crop is obtained at 20 ̊C,

followed by crystallization at -10 ̊C. The crystals are then washed and mixed with the

Figure 9. Synthetic pathway towards artemisinin performed in continuous flow reactor. For a detailed description of the setup components, please refer to the original reference. (Reproduced from Ref.66 with permission of John Wiley & Sons, Inc.)

first excipient (SiO2). The mixed slurry is then dried in two convection-heated drums,

scraped off and sent into a vacuum chamber. Further, the dried flakes are transported

through the heated tubes with the aim of final solvent removal. Later, the dried matter

is mixed with PEG and melted at 60 ̊C to provide a good mixing, and create tablets with

a predefined geometry. The final tablets underwent a quality testing. One major impurity

was detected, but was kept within specification limits.

The described process employs custom-built equipment with incorporated PAT tools.

The characteristic dimensions of the reactors are in the range of millimeters, while

the reaction times are in the range of hours. The process represents a bridge in terms

of reactor and time dimensions between lab scale flow chemistry and not yet widely

accessible production scale.

1.5.2 Artemisinin Malaria poses a serious threat on undeveloped regions, where underpaid infected are

in desperate need for a cure67. Currently, the most effective treatment is offered by

artemisinin. It is extracted from the plant Artemisia annua (sweet wormwood), however

the high demand and limited supply results in an elevated cost of the drug68. The

precursor to artemisinin, artemisinic acid is a simpler structure and can be extracted

from the same wood in higher yields. The challenge is to convert available artemisinic

acid to highly desired artemisinin. Seeberger et al. developed a simple, fast, efficient and

inexpensive continuous process of converting the reduction product of artemisinic acid,

dihydroartemisinic acid, into the target compound, artemisinin66,69, as shown in Figure 9.

By wrapping transparent FEP tubing around a Schenk photochemical reactor containing a

450W mercury lamp that was cooled to 25 ̊C, the authors afforded the generation of singlet

oxygen in the presence of tetraphenylporphyrin. The ene reaction of artemisinic acid and

singlet oxygen resulted in generation of tertiary allylic hydroperoxide (91% conversion,

75% yield). Hock cleavage was afforded by treating the product with trifluoroacetic acid

in the first 16 mL of reactor at room temperature and then raising the temperature of

the last 10 mL to 60 ̊C. Hock cleavage was followed by the addition of triplet oxygen

and condensation reactions to completion, and afforded 39% yield after chromatographic

purification. The authors calculated that at the current capacity 1500 of photochemical

flow setups will be needed to meet the estimated annular demand of 225 million of

artemisinin doses.

The process represents a brilliant approach to access a highly important pharmaceutical.

The employed equipment has a small foot-print, which can lead to worldwide distribution

of artemisinin producing stations around the areas affected by the disease. Moreover, such

32


33

1

1

2

23

3

4

5

Figure 10. Synthetic pathway towards Gleevec performed in continuous flow reactor network. (Reproduced from Ref.70)

stations could prevent a wide-spread problem of fake medicines circulating in the afore-

mentioned areas. However, before the implementation the final yield of the compound

needs to be increased and chromatography-based separation to be circumvented, if

possible.

1.5.3 Gleevec Gleevec (Imatinib mesylate) was developed by Novartis AG and is used for the treatment

of chronic myeloid leukemia and gastrointestinal stromal tumours71. The synthesis of the

target molecule in flow by Ley et al. started with the formation of the amide core via a

reaction of acid chloride and aniline (Figure 10). Polystyrene-supported DMAP within a glass

column was followed by a column with supported dimethylamine, in order to facilitate

the reaction and neutralize the hydrochloric acid formed as a by-product. Connecting

the outlet stream to the next step of nucleophilic substitution of chloride by N-methyl

piperazine turned out to be problematic due to the dispersion of the product within the

stream, and changing concentration, as was indicated by in-line monitoring through a UV

spectrometer. An automated fraction collector was then attached to the UV spectrometer

and programmed to collect the output of the reaction with a pre-defined UV absorption

threshold. The most concentrated fraction of 4 mL (75% isolated yield) was then premixed

with N-methyl piperazine. The second reaction was optimized when the concentration of

the collected product was increased, by evaporating the DCM from the collected 4 mL of

the product solution. The output of the first reaction was combined with the N-methyl

piperazine in DMF, after that nitrogen was aspirated over the solution heated up till 50 ̊C

and the evaporated DCM was removed from the headspace to an exhaust. The evaporation

of the volume took 30 min, and was followed by the injection of the reactant mixture into

a column containing calcium carbonate heated to 80 ̊C. To scavenge unreacted N-methyl

piperazine on polystyrene-supported isocyanate followed the calcium carbonate column.

In order to catch the product silica-supported sulfonic acid was used. Later, the product

was released into the mixture of ammonia in methanol resulting in 80% yield and >90%

purity. The final step on the way to the target molecule is Buchwald-Hartwig amination,

which was achieved by using 10 mol% of a BrettPhos Pd precatalyst and 4.0 equivalents of

amine and sodium tert-butoxide. Thus, three-step synthesis of Gleevec was afforded with

32% overall yield and >95% purity.

The current process represents a proof-of-concept approach and requires extensive

improvement to be translated to a larger scale. The main requirements are minimization

of cartridge use for the sake of entrapment of the unreacted reagents. In addition,

minimization of equivalents of reagents and catalyst used is needed, especially in

Buchwald-Hartwig amination. Finally, the yield of 32% needs to be increased to be

considered a potential alternative to the batch process.

1.5.4 IbuprofenJamison et al. reduced the total synthesis of Ibuprofen down to 3 minutes with 72% overall

yield and initial productivity of 8.1 g/h72. The first step of the Fridel-Crafts acylation

takes place under neat conditions in 1 min yielding 95% yield of the product. Quenching

of the reaction occurs when HCl is introduced into the reacting stream (Figure 11). The

organic and aqueous phases are then separated within a membrane based liquid-liquid

separator. The organic outlet is immersed into a reservoir, where the product is collected

and pumped into the subsequent reactor with a separate pump (144 µl/min) to mix with

trimethylorthoformate in DMF (630 µl/min). Iodine chloride used in the following step

was pumped in its neat melted form (109 µl/min). The absence of a solvent prevented

1

2

3 4

5

6

Figure 11. Multi-step synthesis of Ibuprofen performed in continuous flow reactor network. (Reproduced from Ref.70)

34


35

1

the formation of iodine precipitate. Finally, the product stream of the second reactor

is diluted by a factor of 4 upon meeting with the stream of sodium hydroxide and a

2-mercaptoethanol in water-methanol mixture (3 ml/min).

This process demonstrates an improvement to a previously developed process to deliver

ibuprofen73. The current process minimizes the use of solvent and increases the yield from

51% to 72%, while decreasing the total residence time and increasing productivity. This

improvement demonstrates how significant process-design intensification can be attained

when the process is rethought in a holistic manner.

1.5.5 RufinamideRufinamide (Figure 12, (1)), used in the treatment of Lennox-Gastaut syndrome, exists

under the brand names of Inovelon and Banzel and is produced by Eisai. Recently the total

synthesis of rufinamide in flow was published by Jamison et al. 2,6-difluorobenzyl bromide

(2) dissolved in DMSO was mixed with sodium azide (3) solution to yield a final 0.29 M

concentration of 2,6-difluorobenzyl bromide within the reactor, with 1.3 equivalents of

sodium azide74. Nucleophilic substitution took place in 1 min at room temperature. The

dipolarophile, needed to construct 1,2,3-trizole moiety in rufinamide, was synthesized

from neat methyl propiolate (4) and aqueous ammonium hydroxide (5) in 5 min and 0 ̊C.

Finally, 2, 6-difluorobenzyl azide and propiolamide were combined in copper tubing to

facilitate a catalyzed [3+2]-Huisgen cycloaddition at 110 ̊C and given 6.2 min residence

Figure 12. Multi-step synthesis of Ibuprofen performed in continuous flow reactor network. (Reproduced from Ref.70)

time. Thus, the total synthesis took 11 min and resulted in 92% overall yield. The product

was purified by addition of 2 equivalents of water with respect to reaction mixture,

filtering off the precipitate and drying. The productivity of the sequence was 217 mg/h.

The current approach is an example of lowering the chemical barriers for the sake of making

a final molecule and presents large room for improvement. Nucleophilic substitutions

are known to proceed tremendously fast in non-protic polar solvents, such as DMSO.

However, the separation of DMSO and its use are not favorable in industrial applications.

Moreover, after iodine, substitution of bromide is one of the easiest, however it leads

to large waste generation due to its large molecular weight. The methyl propiolate, a

dipolarophile, used in 1, 3-dipolar Huisgen cycloaddition is highly electron-deficient, thus

very reactive. It is also an expensive reagent. During cycloaddition, regioselectivity is

directed by leached-out copper species from copper tubing, which brings about a concern

of toxic metal separation from the product mixture.

1.6. MULTI-STEP SYNTHESES OF ACTIVE PHARMACEUTICAL INGREDIENTS IN MESO AND LARGE-SCALE FLOW REACTORS

While micro reactors largely and to a somewhat fewer extent also flow chemistry is a

European development, the recently seen strong push towards continuous manufacturing

in pharmaceutical manufacturing has been strongly enforced by large US companies, US

academics, the industrially guided ACS Green Chemistry Institute Roundtable, and the

FDA. This kind of joined initiative is active in publishing the success of their emerging

technologies and this will be described in particular in the following. It is expected that

similar investigations are done in Europe and possibly in the Far East, yet publications

hereabout are rare.

1.6.1 Hydroxypyrrolotriazine (Bristol-Myers-Squibb)LaPorte et al. described the design of the process for manufacturing of an intermediate

in the synthesis of brivamib alaninate75. Hydrogen peroxide is used as the oxidant in a

benzylic hydroperoxide rearrangement. However, its combination with an acid catalyst

brings about a thermal runaway potential at ambient conditions. A chemical hazard

analysis concluded that the reaction was unsafe for operating at more than 0.5 kg of

starting material in batch. Alternative chemistries were not sufficient for the quality

expected, therefore the engineering solution of transferring batch process into continuous

was devised. In order to prevent the decomposition of hydrogen peroxide and minimize

side reactions, the mixing section of the compounds was kept at -5 ̊C as shown in Figure

13. A process simulation showed that operating the reactor in two temperature regimes

36


37

1

Figure 13. Top) Reaction scheme for the synthesis of hydroxypyrrolotriazine.; Bottom Left) Lab scale continuous setup.; Bottom Right) Pilot plant assembly on a portable rack. (Reproduced from Ref.75 with permission of American Chemical Society)

was the safest, i.e. first low temperature (0 ̊C) prevented the thermal runaway, while a

second higher temperature zone (12-15 ̊C) drove the reaction to completion. The lab scale

setup proves to operate in a stable manner and therefore was further scaled-up to kilo-lab

and pilot-plant scale. Figure 13 (bottom right) shows a portable pilot plant rack reactor,

which operated at 400 mL/min with a productivity of 22.4 kg/day.

1.6.2 2,2-Dimethylchromenes (Bristol-Myers-Squibb)2,2-Dimethylchromenes represent a highly interesting class of compounds due to their

biological acitivity in plants and animals76. Multi-step synthesis of the compound involves

thermal Claisen rearrangement. Its execution at large scale in batch poses a safety risk,

in addition to decomposition of the product. Therefore, Polomski and Soundararajan

et al. performed a DSC study, and subsequently developed a mathematical model that

could identify the key parameters in the optimization of the reaction outcome as well as

maximizing the safety76. A stainless steel capillary reactor (80 ml) of 1/8” ID dimensions

was heated within an oil bath to 195-200 ̊C. A flow of 20 mL/min allowed 4 min residence

time and resulted in 98% yield with a productivity of 7 kg/h.

1.6.3 Fused-Bicyclic Isoxazolidines (Eli Lilly and Company)Process intensification in synthesizing bicyclic isoxazolidines via 1,3-dipolar cycloaddition

of nitrone was achieved because a flow reactor could be heated up to much higher

temperatures than a conventional batch reactor77. The process was realized in a stainless

steel tubular coil (1/8”OD, 343 mL) at 210 ̊C in 10 min residence time. The concentration

of the reactant in toluene was 0.24 M, which was pumped with 34.3 ml/min and resulted

in a productivity of 120 g/h (50% yield).

1.6.4 7-Ethyltryptophol on the way to etodolac Indole moiety is often found in natural products and biologically active compounds78.

7-ethyltryptopol is a key intermediate for the anti-inflammatory drug etodolac. Etodolac

proved to be safe with respect to gastrointestinal and renal tract and to slow down the

progression of skeletal changes in rheumatoid arthritis. The benchmark method consists

of hydrozone formation, followed by the [3+3]-sigmatropic rearrangement. The instability

of hydrozones poses a safety concern on a large scale calling for one-pot procedures. Su et

al. developed a continuous process to synthesize 7-ethyltryptopol79. Two reacting streams

of 2-ethylphenylhydrazine and 4-hydroxybutyraldehyde were mixed at a total flow rate of

230 mL/min within a mixing T (ID 7 mm). A tubular reactor was heated within an oil bath.

It was observed that decomposition of hydrozone could be minimized by shortening the

reaction time to 20 s, after which sulphuric acid was added to promote [3+3]-sigmatropic

rearrangement to the final product. In total, the one-pot process took 280 s, including

cooling of 20 s. On a 10 kg-scale the developed procedure afforded 74 % yield of the final

Figure 14. Top) Reaction scheme for the synthesis of 6-hydroxybuspirone. Bottom left) Lab-scale continuous setup and bottom of the Quad oxidation reactor; Bottom right) Internal reaction tubes and extra-tubular baffles of the Quad trickle-bed reactor. (Reproduced from Ref.70)

38


39

1

product with 83% purity. The authors state that during the manufacturing the isolation of

7-ethyltryptopol would not be necessary and the process could go on to the synthesis of

etodolac.

1.6.5 6-Hydroxybuspirone (Bristol-Myers-Squibb)LaPorte et al. designed a process in line with PAT to deliver a safer and more efficient

continuous process on production scale80. The process consisted of converting buspirone

to 6-hydroxybuspirone via hydroxylation in three steps, first enolization with sodium

bis(trimethylsilylamide) in THF at or below – 70 ̊C in the presence of excess of triethylphosphite

(Figure 14). Once enolization is complete, air is introduced to afford oxidation, which is

followed by the reduction of formed hydroperoxy anion to hydroxide. Equivalents of base

govern the formation of the by-product, diol. Moreover, premature introduction of oxygen

leads to the formation of residuals of unreacted sodium bis(trimethylsilylamide) and

unreacted buspirone enolate, which may result in the deprotonation of the intermediate

that can be oxidized to diol impurities. A high excess (3.5—5 eq.) of triethylphosphite is

needed to maintain sufficient progress in the reduction and prevent formation of amino

acids and other impurities. Apart from possible waste generation, there is a danger coming

from the exothermic oxidation within flammable solvent in the presence of oxygen.

Therefore, the headspace of the reactor is continuously diluted with nitrogen. The

complications listed above lead to the design of a continuous process. The reactant and

base were mixed within a T-mixer with internal dimensions of 3/8”, followed by one static

mixer (0.5” ID with 27 elements) at room temperature to avoid a sudden precipitation

of base and a second mixer that was cooled with a jacket containing glycol to -15 ̊C.

The enolation was allowed to reach completion within the heat exchanger. A React-IR

DiComp probe was installed at the outlet of the enolatation reactor to perform on-line

FTIR measurements. The product stream was then split in four streams that entered a

trickle bed reactor containing four columns. Oxygen was fed in a counter-current fashion,

since it lead to a higher selectivity. The oxidation reactor (600 mL) was cooled down to

-37 with the same fluid used to cool the heat exchanger used in enolation. Unreacted

oxygen flowed out the top of the reactor, where it was diluted with nitrogen and sent to

the plant’s thermal oxidizer, while the product flowed by gravity along the reactor into a

tank with a level control system. When a certain level was reached the level output signal

would make the valve on the outlet of a piston pump open and allow the product stream

to flow into the quench tank. The product flow rate into the quench tank was 168 ml/

min and the pH was kept constant by continuous addition of HCl. In total, the continuous

process took 4 min as total residence time, while the batch process took 8 h in lab, and

16-24 h on a pilot plant scale. In three pilot-plant campaigns with the continuous process

more than 100 kg of API was produced.

Thus, the current process benefited from speeded-up kinetics, higher safety due to no

accumulation and limited reactive species in handling, higher selectivity and thus less

generation of waste.

1.6. SCOPE AND OUTLINE OF THE THESIS

The current thesis presents the development of flow reactor networks with the goal of

intensifying single reaction steps and subsequently constructing a multi-step process to

deliver active pharmaceutical ingredients. When a single reaction step is investigated,

attention is first given to the chemical reagents used and their characteristics, such

as toxicity, stability and cost. Green chemistry and green engineering principles serve

as guidelines in the decision making process. The principles of Novel Process Windows,

presented above, are followed in the two subsequent stages, first during single-

step chemical intensification and later process-design intensification is realized while

connecting several steps. As a result, sometimes synergistic effects arise benefitting the

process in overall.

Organohalides represent a class of valuable intermediates in the pharmaceutical industry.

Among several synthetic route alternatives is the conversion of alcohols to chlorides by

the use of chlorinating agents. The biggest concern associated with the latter is a low

atom economy and stoichiometric generation of waste. Chapter 2 presents a process

where hydrochloric acid was used instead of toxic and wasteful chlorinating agents. The

synthesized chlorides were further used in a flow network consisting of 4-step multi-step

synthesis of antiemetic drugs and antihistamines. The drawbacks of the process were

the requirement of 3 equivalents of hydrochloric acid and use of a reagent loop, which

resulted in only a partial continuity. Chapter 3 presents a further improvement of the

process. The continuity and the decrease in equivalents was achieved by switching from

hydrochloric acid to pure hydrogen chloride gas. Safe pressurization in micro flow allowed

the maximization of hydrogen chloride solubility in alcohol. To our best knowledge, it was

the first documented use of pure hydrogen chloride gas in a continuous process. Finally,

less water was involved in the synthesis due to the switch from acid to pure gas.

High temperature, pressure and concentration are the main tools in chemical

intensification as mentioned above. Chapter 4 investigated the influence of pressure on

the regioselectivity of 1,3-dipolar cycloaddition in an autoclave reactor with an operating

pressure reaching 1800 bar. The selectivity towards the 1,4-regioisomer increased by 30%

when operating pressure was increased from 500 bar to 1800 bar. In flow experiments

pressure effects were not observed due to a lower pressure limit imposed by the available

40


41

1

equipment. However, a clear activation by temperature was observed. The reaction

proceeded at a higher speed and with a greater productivity in flow micro capillary

reactors. Chapter 5 concentrates on a further intensification of 1,3-dipolar cycloaddition.

The aim was to use organoazide and the highly inactive, but inexpensive and relatively

green dipolarophile to synthesize a key precursor to rufinamide. The reaction time

decreased from >24 hrs required in batch to 10 min when the reaction was performed in

a micro flow capillary reactor under neat conditions at 210 ̊C and 70 bar.

Chapter 6 describes the combination of single-step intensifications to develop a

continuous process of a rufinamide precursor and realize process-design intensification.

2,6-dilfuorobenzyl alcohol was converted to the corresponding chloride through the use of

hydrogen chloride gas. Immediately after the reaction a sodium azide stream was injected

into the product stream to synthesize 2,6-dilfuorobenzyl azide. The product was then

separated in an in-house developed inline L-L separator and further pumped to undergo

the aforementioned cycloaddition. In order to avoid crystallization of the cycloadduct, a

methanol stream was injected into the product stream. Upon collection the cycloadduct

crashed out of the mother liquor in 82% overall yield. The entire process relied on no

solvent, nor catalyst, and water use was minimized via a combination of a quenching

agent, sodium hydroxide, and sodium azide in one stream. In 95 min 3 reaction steps and

2 separations were carried out resulting in 9 g/hr productivity. Thus, an own momentum

of process-design intensification was identified, which is more than the added momenta

of chemical intensification. Finally, chapter 7 presents an evaluation of the developed

5-stage 3-reaction step process to produce a rufinamide precursor. Batch and flow process

models were developed to be compared in terms of material and energy requirements.

Due to the use of organic solvents in and required intermediate separations in the batch

process, which employed water, a significant difference in material requirements was

observed. Meanwhile, in the continuous-flow process 50% of employed materials were

reagents. Further, an economic and environmental evaluation was performed. Once

again, the continuous-flow process was found to be favourable in terms of economic and

environmental impacts.

REFERENCES

1. Abou-Gharbia, M.; Childers, W. E. J. Med. Chem. 2014, 57 (13), 5525–5553.

2. Malet-Sanz, L.; Susanne, F. J. Med. Chem. 2012, 55, 4062–4098.

3. Biopharmaceutical Research & Development www.phrma.org/sites/default/files/pdf/rd_brochure_022307.pdf.

4. DiMasi, J. A.; Grabowski, H. G.; Hansen, R. W. J. Health Econ. 2016, 47, 20–33.

5. Anderson, N. G. 2012.

6. Kopach, M. E.; Singh, U. K.; Kobierski, M. E.; Trankle, W. G.; Murray, M. M.; Pietz, M. A.; Forst, M. B.; Stephenson, G. A.; Mancuso, V.; Giard, T.; Vanmarsenille, M.; DeFrance, T. Org. Process Res. Dev. 2009, 13 (2), 209–224.

7. Brechtelsbauer, C.; and Francois Ricard*. Org. Process Res. Dev. 2001, 5 (6), 646–651.

8. Vicevic, M.; Boodhoo, K. V. K.; Scott, K. Chem. Eng. J. 2007, 133 (1–3), 43–57.

9. Clemens Brechtelsbauer; Norman Lewis, *; Paul Oxley; and Francois Ricard; Ramshaw, C. Org. Process Res. Dev. 2001, 5 (1), 65–68.

10. Natthinee Anantachoke, †; Mohamed Makha *, ‡; Colin L. Raston *, ‡; Vichai Reutrakul, †; Nigel C. Smith, ‡; and Martin Saunders§. J. Am. Chem. Soc. 2006, 128 (42), 13847–13853.

11. Gonzalez, M. A.; Ciszewski, J. T. Org. Process Res. Dev. 2009, 13 (1), 64–66.

12. Hampton, P. D.; Whealon, M. D.; Roberts, L. M.; Yaeger, A. A.; Boydson, R. Org. Process Res. Dev. 2008, 12 (5), 946–949.

13. Frost, C. G.; Mutton, L. Green Chem. 2010, 12 (10), 1687–1703.

14. Singh, B. K.; Kaval, N.; Tomar, S.; der Eycken, E. Van; Parmar, V. S. Org. Process Res. Dev. 2008, 12 (3), 468–474.

15. Yu, R. H.; Polniaszek, R. P.; Becker, M. W.; Cook, C. M.; Yu, L. H. L. Org. Process Res. Dev. 2007, 11 (6), 972–980.

16. Truppo, M. D.; Hughes, G. Org. Process Res. Dev. 2011, 15 (5), 1033–1035.

17. Wegner, J.; Ceylan, S.; Kirschning, A. Chem. Commun. (Camb). 2011, 47 (16), 4583–4592.

18. Hartman, R. L.; McMullen, J. P.; Jensen, K. F. Angew. Chem. Int. Ed. Engl. 2011, 50 (33), 7502–7519.

19. Suresh, P.; Basu, P. K. J. Pharm. Innov. 2008, 3 (3), 175–187.

20. Geyer, K.; Gustafsson, T.; Seeberger, P. Synlett 2009, 2009 (15), 2382–2391.

21. Hessel, V.; Löwe, H.; Schönfeld, F. Chem. Eng. Sci. 2005, 60 (8–9), 2479–2501.

22. Nguyen, N.-T. Micro mixers: fundamentals, design and fabrication.; William Andrew, Norwich, NY, 2008.

23. Kuhn, S.; Noël, T.; Gu, L.; Heider, P. L.; Jensen, K. F. Lab Chip 2011, 11 (15), 2488–2492.

24. Rahbar, M.; Shannon, L.; Gray, B. L. J. Micromechanics Microengineering 2014, 24 (2), 25003.

25. Nagy, K. D.; Shen, B.; Jamison, T. F.; Jensen, K. F. Org. Process Res. Dev. 2012, 16 (5), 976–981.

26. Borukhova, S.; Hessel, V. In Process Intensification for Green Chemistry; John Wiley & Sons, Ltd, 2013; pp 91–156.

27. Pelleter, J.; Renaud, F. 2009, 13 (4), 698–705.

28. Gunther, A.; Jensen, K. F. Lab Chip 2006, 6 (12), 1487–1503.

29. Marre, S.; Adamo, A.; Basak, S.; Aymonier, C.; Jensen, K. F. 2010, 11310–11320.

30. Plou, P.; Macchi, A.; Roberge, D. M. 2014.

31. Target, O.; Ways, M.; Industry, T. 2002, 1920–1924.

32. Charpentier, J.-C. Chem. Eng. Technol. 2005, 28 (3), 255–258.

33. Illg, T.; Löb, P.; Hessel, V. Bioorg. Med. Chem. 2010, 18 (11), 3707–3719.

34. Roberge, D. M.; Ducry, L.; Bieler, N.; Cretton, P.; Zimmermann, B. Chem. Eng. Technol. 2005, 28 (3), 318–323.

35. Nagaki, A.; Kenmoku, A.; Moriwaki, Y.; Hayashi, A.; Yoshida, J. Angew. Chemie 2010, 122 (41), 7705–7709.

36. Nagaki, A.; Moriwaki, Y.; Haraki, S.; Kenmoku, A.; Takabayashi, N.; Hayashi, A.; Yoshida, J.-I. Chem. Asian J. 2012, 1–9.

37. Kawaguchi, T.; Miyata, H.; Ataka, K.; Mae, K.; Yoshida, J.-I. Angew. Chem. Int. Ed. Engl. 2005, 44 (16), 2413–2416.

38. Nagaki, A.; Matsuo, C.; Kim, S.; Saito, K.; Miyazaki, A.; Yoshida, J. Angew. Chemie Int. Ed. 2012, 51 (13), 3245–3248.

39. Yoshida, J.; Nagaki, A.; Yamada, T. Chemistry 2008, 14 (25), 7450–7459.

40. Krasberg, N.; Hohmann, L.; Bieringer, T.;

42


43

1

Bramsiepe, C.; Kockmann, N. Processes 2014, 2, 265–292.

41. Irfan, M.; Glasnov, T. N.; Kappe, C. O. 2011, 300–316.

42. Jiménez-González, C.; Poechlauer, P.; Broxterman, Q. B.; Yang, B.-S.; am Ende, D.; Baird, J.; Bertsch, C.; Hannah, R. E.; Dell’Orco, P.; Noorman, H.; Yee, S.; Reintjens, R.; Wells, A.; Massonneau, V.; Manley, J. Org. Process Res. Dev. 2011, 15 (4), 900–911.

43. Poechlauer, P.; Colberg, J.; Fisher, E.; Jansen, M.; Johnson, M. D.; Koenig, S. G.; Lawler, M.; Laporte, T.; Manley, J.; Martin, B.; O’Kearney-McMullan, A. Org. Process Res. Dev. 2013, 17 (12), 1472–1478.

44. Bryan, M. C.; Dillon, B.; Hamann, L. G.; Hughes, G. J.; Kopach, M. E.; Peterson, E. A.; Pourashraf, M.; Raheem, I.; Richardson, P.; Richter, D.; Sneddon, H. F. J. Med. Chem. 2013, 56 (15), 6007–6021.

45. Hartman, R. L.; Jensen, K. F. Lab Chip 2009, 9 (17), 2495–2507.

46. Yoshida, J.; Kim, H.; Nagaki, A. ChemSusChem 2011, 4 (3), 331–340.

47. Adamo, A.; Heider, P. L.; Weeranoppanant, N.; Jensen, K. F. Ind. Eng. Chem. Res. 2013, 52, 10802–10808.

48. Roberge, D. M.; Bieler, N. 2008, No. 8, 1155–1161.

49. Hessel, V. Chem. Eng. Technol. 2009, 32 (11), 1655–1681.

50. Palde, P. B.; Jamison, T. F. Angew. Chemie Int. Ed. 2011, 50 (15), 3525–3528.

51. Gutmann, B.; Roduit, J.-P.; Roberge, D.; Kappe, C. O. Angew. Chemie Int. Ed. 2010, 49 (39), 7101–7105.

52. Illg, T.; Hessel, V.; Löb, P.; Schouten, J. C. Chem. Eng. J. 2011, 167 (2–3), 504–509.

53. Hessel, V.; Kralisch, D.; Kockmann, N.; Noël, T.; Wang, Q. ChemSusChem 2013, 6 (5), 746–789.

54. Razzaq, T.; Kappe, C. O. Chem. Asian J. 2010, 5 (6), 1274–1289.

55. Reichart, B.; Tekautz, G.; Kappe, C. O. 2013.

56. Razzaq, T.; Glasnov, T. N.; Kappe, C. O. Chem. Eng. Technol. 2009, 32 (11), 1702–1716.

57. Damm, M.; Glasnov, T. N.; Kappe, C. O. Org. Process Res. Dev. 2010, 14 (1), 215–224.

58. Kawanami, H.; Sato, M.; Chatterjee, M.; Otabe, N.; Tuji, T.; Ikushima, Y.; Ishizaka, T.;

Yokoyama, T.; Suzuki, T. M. Chem. Eng. J. 2011, 167 (2–3), 572–577.

59. Snead, D. R.; Jamison, T. F. Chem. Sci. 2013, 4 (7), 2822–2827.

60. Benito-López, F.; Egberink, R. J. M.; Reinhoudt, D. N.; Verboom, W. Tetrahedron 2008, 64 (43), 10023–10040.

61. Hessel, V.; Vural Gürsel, I.; Wang, Q.; Noël, T.; Lang, J. Chem. Eng. Technol. 2012, 35 (7), 1184–1204.

62. Vural-Gürsel, I.; Wang, Q.; Noël, T.; Hessel, V.; Tinge, J. T. Ind. Eng. Chem. Res. 2013, 52 (23), 7827–7835.

63. Ott, D.; Borukhova, S.; Hessel, V. Green Chem. 2016, 18 (4), 1096–1116.

64. Webb, D.; Jamison, T. F. Chem. Sci. 2010, 1 (July 1999), 675.

65. Mascia, S.; Heider, P. L.; Zhang, H.; Lakerveld, R.; Benyahia, B.; Barton, P. I.; Braatz, R. D.; Cooney, C. L.; Evans, J. M. B.; Jamison, T. F.; Jensen, K. F.; Myerson, A. S.; Trout, B. L. Angew. Chem. Int. Ed. Engl. 2013, 52 (47), 12359–12363.

66. Lévesque, F.; Seeberger, P. H. Angew. Chem. Int. Ed. Engl. 2012, 51 (7), 1706–1709.

67. Booker-Milburn, K. Nat. Chem. 2012, 4 (6), 433–435.

68. Kopetzki, D.; Løvesque, F.; Seeberger, P. H. 2013, 5450–5456.

69. Kopetzki, D.; Lévesque, F.; Seeberger, P. H. Chem. – A Eur. J. 2013, 19 (17), 5450–5456.

70. Baumann, M.; Baxendale, I. R. Beilstein J. Org. Chem. 2015, 11, 1194–1219.

71. Hopkin, M. D.; Baxendale, I. R.; Ley, S. V. Chem. Commun. (Camb). 2010, 46 (14), 2450–2452.

72. Snead, D. R.; Jamison, T. F. Angew. Chemie Int. Ed. 2015, 54 (3), 983–987.

73. Bogdan, A. R.; Poe, S. L.; Kubis, D. C.; Broadwater, S. J.; McQuade, D. T. Angew. Chem. Int. Ed. Engl. 2009, 48 (45), 8547–8550.

74. Zhang, P.; Russell, M. G.; Jamison, T. F. Org. Process Res. Dev. 2014.

75. LaPorte, T. L.; Spangler, L.; Hamedi, M.; Lobben, P.; Chan, S. H.; Muslehiddinoglu, J.; Wang, S. S. Y. Org. Process Res. Dev. 2014, 18 (11), 1492–1502.

76. Bogaert-Alvarez, R. J.; Demena, P.; Kodersha, G.; Polomski, R. E.; Soundararajan, N.; Wang, S. S. Y. Org. Process Res. Dev. 2001, 5, 636–645.

77. Rincón, J. A.; Mateos, C.; García-Losada, P.; Mergott, D. J. Org. Process Res. Dev. 2015, 19 (2), 347–351.

78. Yu, Z.; Lv, Y.; Yu, C. Org. Process Res. Dev. 2012, 16, 1669–1672.

79. Lv, Y.; Yu, Z.; Su, W. Org. Process Res. Dev. 2011, 15 (2), 471–475.

80. LaPorte, T. L.; Hamedi, M.; DePue, J. S.; Shen, L.; Watson, D.; Hsieh, D. Org. Process Res. Dev. 2008, 12 (5), 956–966.

CHAPTER 2Continuous-flow multi-step synthesis of Cinnarizine, Cyclizine and a Buclizine derivative from bulk alcohols

This chapter is based on:

Borukhova, S., Noël, T, Hessel, V. (2015). Continuous-flow multi-step synthesis of Cinnarizine, Cyclizine and a Buclizine derivative from bulk alcohols, ChemSusChem, 9(1), 67-74.

46

CHAPTER 2 Continuous-flow multi-step synthesis

47

2

ABSTRACT

Cinnarizine, cyclizine, buclizine and meclizine belong to a family of antihistamines that

resemble each other in terms of 1-diphenylmethylpiperazine moiety. Current chapter

presents the development of 4-step multistep continuous process to generate the final

antihistamines using bulk alcohols as starting compounds. Hydrochloric acid is used to

synthesize the intermediate chlorides in short reaction time and excellent yields. The

methodology presented offers a great way to synthesize intermediates to be used in

drug synthesis. Inline separation allows collection of pure products and their immediate

consumption in the following steps. Overall isolated yields for cinnarizine, cyclizine and

a buclizine derivatives are 82, 94 and 87%, respectively. The total residence time for four

steps is 90 min with productivity of 2 mmol/hr.

2.1. INTRODUCTION

Functional group interconversion is one of the most used techniques in drug synthesis1.

The conversion of hydroxyl groups to their corresponding halides is arguably one of the

most versatile transformations, which is often followed by a nucleophilic substitution.

Therefore, it is one of the key research areas in the synthesis of active pharmaceutical

ingredients (API)2. Chlorides are very interesting due to their abundance in nature and

their low molecular weight. Synthesis of chloride derivatives from alcohols usually

requires use of chlorinating agents such as thionyl chloride3, phosphorus chlorides4,

pivaloyl chloride5, Vilsmeier reagent6, tosyl chloride7, 2,4,6-trichloro-[1,3,5]triazine with

DMF8, oxalyl chloride9 and phosgene10. Unfortunately, these chlorinating agents constitute

major concerns due to their high toxicity11. Moreover, since only a part of the chlorinating

agent molecule is used in the reaction, waste is generated in stoichiometric amounts

leading to a process that is not atom efficient nor green12.

The ideal process to convert primary alcohols to corresponding chlorides would be based

on neat alcohol reacting with hydrogen chloride. This will maximize atom efficiency and

result in water as sole by-product. Although possible, handling hydrogen chloride gas is

challenging, especially on a large scale. Hydrochloric acid can be used as an alternative.

However, low to average temperatures need to be maintained to keep the solubility of

hydrogen chloride in water sufficiently high13. Thus, reaction rates can only be increased

to a limited extent in batch reactors. In contrast to batch reactors, micro flow technology

offers a great platform to enable this sort of processes14–17. The absence of headspace

and safe pressurization within micro channels assists in keeping hydrogen chloride

dissolved at high reaction temperatures. Multiple examples of reaction rate acceleration

in micro reactors under high temperature and pressure exist18–21. In addition, modularity

of continuous-flow reactors allows constructing reaction networks to deliver natural

products22,23 and APIs24–29 in a safe, quick and efficient way.

Several milestones have been reached on conversion of alcohols to chlorides, i.e.

chlorodehydroxylation. Tundo et al. have described a solvent-free synthetic method,

i.e. gas-liquid phase-transfer catalysis (GL-PTC), in which gaseous reagents flow through

a molten phase-transfer catalyst supported on solid30,31. Microwave heating technology

showed high efficiency in the field of chlorodehydroxylations32. Reid et al. combined

microwave heating of ionic liquids and aqueous hydrochloric acid for a range of alcohols

with and without added catalysts32. Short chain alcohols have been converted in high yields

and selectivity within 10 minutes under pressurized microwave conditions. The reaction

rates were measured to be in the order of 100 times faster than non-microwave high

temperature chlorodehydroxylations. Kappe et al. have extended the microwave heating

48


49

2

technology with hydrochloric acid to a continuously operating microchip-based flow setup33.

Full conversion of n-butanol was obtained with 3 equivalents of hydrochloric acid, within

15 minutes residence time at 160 C̊ and 20 bar. When applying the optimal conditions to

n-hexanol and n-decanol, the reactivity decreased with the increased chain length.

In the present chapter bulk alcohols are converted into corresponding chlorides that

are subsequently consumed in continuous multi-step synthesis of antiemetic agents and

antihistamines such as, cyclizine, meclizine and buclizine derivative, and cinnarizine.

The progress in the field of chlorodehydroxylation is extended by applying milder

conditions on a wider scope of substrates in the capillary reactor. This offers a robust

and more accessible alternative to microchips, as well as higher versatility regarding the

production volume and rate. Due to the high abundance of the benzyl moiety in active

pharmaceutical ingredients, we started with optimizing the conversion of benzyl alcohol

to benzyl chloride. Integrating an in-line liquid-liquid separator with the reactor afforded

pure benzyl chloride, which was readily coupled to subsequent reactions. Finally, a step

by step procedure to carry out four-step continuous synthesis to prepare relevant APIs,

such as cinnarizine, cyclizine and a buclizine derivative is demonstrated.

2.2. SYNTHESIS OF ALKYL CHLORIDES

Synthesis of chlorides started from similar conditions applied by Kappe et al., where 3 eq

of hydrochloric acid was used with 15 min as a residence time33. The results of the initial

screening at 120 ̊C are shown in Table 1. Already 88% of 1-chlorobutane was obtained in

Table 1. Results of initial screening of synthesis of primary chlorides and optimization of benzyl chloride synthesis.[a]

Entry Substrate Residence time HCl (eq) GC Yield (%)[b]

1 1-butanol 15 3 88

2 1-butanol 30 3 77

3 1-hexanol 15 3 46

4 1-hexanol 30 3 82

5 Benzyl alcohol 15 3 100

6 Benzyl alcohol 15 2 97 [c]

7 Benzyl alcohol 30 2 98 [c]

[a] The experiments were performed at 120 ̊C in the micro capillary based setup (Scheme 1) with aqueous HCl (36 wt%). [b] The yields were calculated based on GC-FID response with respect to 1,3-dimethoxybenzene as an internal standard. Calibrations were performed with corresponding chlorides purchased from Sigma-Aldrich. [c] Presence of unconverted alcohol and di-benzyl ether.

15 min residence time. Extending the residence time to 30 min increased the formation

of di-butyl ether and resulted in a decrease in yield to 77%. For hexanol, the reaction was

slower and longer residence times improved the yield. nor any other by-products present

in GC chromatogram in 15 min residence time.

Due to the fact that benzyl moiety is widely used in API synthesis, our target compound

was benzyl chloride. At the same conditions, 100% yield of benzyl chloride was obtained

with no di-benzyl ether,me. We speculated that the partial solubility of benzyl alcohol in

water minimizes the mass transfer limitations. Moreover, due to the lower solubility of

benzyl chloride in water than that of benzyl alcohol, the equilibrium of the reaction shifts

to the product side. Decreasing the equivalents of hydrochloric acid, while allowing more

residence time leads to incomplete conversion along with formation of di-benzyl ether.

Subsequently, the optimal conditions found for the reaction with benzyl alcohol were

applied to other alcohols. Scheme 1 shows the chlorides synthesized at 120 ̊C within

15 min residence time, along with their yields. Chloride derivatives of alcohols are

usually more volatile due to weaker intermolecular forces, therefore those products

were collected into a sealed vial with deuterated chloroform containing an external

standard (1,3-dimethoxybenzene) and later analysed as such. In case of terminal

alkyne, 4-pentyn-1-ol, the formation of dark viscous agglomerates with low yield of the

chloride was observed. The reaction with tetrahydrofurfuryl alcohol resulted in partial

cleavage of ether and gave a mixture of mono- and di-substituted chloropentanes. When

the same conditions were applied to 1,5-pentanediol, full conversion was observed

and 19% of di-chloropentane was formed. Aromatic compounds included in the scope

differed in the solubility in aqueous phase, which can explain differences in yields in the

substitution on benzylic position. For solid compounds such as dodecanol, cyclohexanol,

napthalenemethanol, diphenyl methanol and cinnamyl alcohol, a toluene:acetone (2:1)

solvent combination was used to afford solutions of 2.2 M.

2.3. UTILIZATION OF BENZYL CHLORIDE

Micro flow technology allows the connection of subsequent reactions to afford truly

continuous operation with a minimum of manual handling34,35. To see the potential in

application of continuous chloride synthesis, we connected the product stream of

the benzyl chloride forming reaction to a second reactor, where another nucleophilic

substitution would take place. The connection relied on continuous inline liquid-liquid

separation that occurred in a Teflon membrane-based separator. Similar separators have

been previously used in multi-step continuous syntheses36–38. The design and operation of

50


51

2

this separator is described in a greater detail within the supporting information of the

publication current chapter is based on. For a complete separation, the right pressure

difference should be applied over the membrane to allow permeate to pass through the

membrane without causing a breakthrough of the retained phase, or alternatively, without

carrying the permeate within the retentate. Using a 5 psi back pressure regulator at the

aqueous outlet afforded perfect separation with no loss of benzyl chloride. A loop of a

larger volume (35 ml), filled with hydrochloric acid, was used in this two-step synthesis

to allow longer uninterrupted operation. After letting 3 volumes of reacting mixture pass

through the reactor, separated benzyl chloride was collected in an Erlenmeyer flask (10

ml). The flowrate of benzyl chloride at the organic outlet of liquid-liquid separator was

40 µl/min (21 mmol/hr). Two hour of collection time was allowed prior to pumping the

product into a second reactor via another HPLC pump as shown on Scheme 2.

Our choice of nucleophiles for the subsequent reaction fell upon nitrogen-containing

substrates, due to their presence in API structures. Thus piperazine, n-methyl piperazine,

morpholine and ethyl isonipecotate were used as nucleophiles. When combined with

benzyl chloride, piperazine was the only insoluble substrate. The other substrates, even

though miscible at first, resulted in precipitate at the end of the reaction. Relatively

high volumes of ethanol dissolved the products indicating the need for larger amounts of

solvent and lower concentrations. Diethylene glycol monomethyl ether (methyl carbitol)

afforded homogeneous reaction mixtures under relatively high concentrations of products

(1.1 M). For piperazine the addition of water (methyl carbitol: water-2:1) was needed

to maintain the product soluble. The homogeneity of the reactant and product mixtures

Scheme 1. Top: Schematic representation of the setup used for the current investigation. Bottom: Chlorides synthesized with 3 equivalents of HCl (aq) at 120 ̊C given 15 min residence time. Yields are determined from 1H NMR with respect to 1,3-dimethoxybenzene added as an external standard.

allowed the translation of batch to continuous conditions. The reaction of benzyl chloride

and n- methyl piperazine (1.2 eq) at 120 ̊C, given 15 min residence time, resulted in

81% conversion of benzyl chloride, and 100% selectivity. The conversion increased up to

>99%, when the temperature was increased to 160 ̊C, residence time to 30 min and 1.5

equivalents of n-methyl piperazine was used. More information on optimization is given

within the suppoting information of the publication current chapter is based on. The

isolated yields obtained at those conditions are given in Scheme 2.

2.4. APPLICATIONS IN API SYNTHESIS

The diphenylmethyl moiety is widely found in antiemetic, antimigraine and antihistamine

drugs. The diphenyl piperazine substrate is a key constituent of cyclizine, cinnarizine,

meclizine, cetirizine, hydroxyzine, buclizine, dotarizine and other drugs of the same family.

The ease of connecting the synthesis of chloride with a subsequent reaction motivated

expanding the scope to commercially available APIs, such as cyclizine, cinnarizine and 1-

diphenylmethyl-4-benzylpiperazine, which resembles meclizine and buclizine as shown in

Scheme 3. Therefore, synthesis along with inline separation of diphenyl methyl chloride

was optimized to pave a way to these final compounds.

2.4.1. Synthesis of CyclizineCyclizine is an antihistamine used in treatment of nausea, vomiting and dizziness

associated with motion sickness, vertigo and consequences of anaesthesia and use of

opioids. Its chemical name is 1-diphenylmethyl-4-methyl piperazine and it is constructed

via nucleophilic substitution of diphenylmethyl chloride with n-methyl piperazine. The two-

Scheme 2. Schematic representation of two-step synthesis (top). Synthesis of benzyl chloride is connected to its separation from aqueous phase, taking place in liquid-liquid membrane based separator. Separated neat benzyl chloride is mixed with secondary amines to undergo another nucleophilic substitution. Collected and isolated products are shown along with the corresponding yield values.

52


53

2

step setup shown in Scheme 2 was used to connect the formation of chlorodiphenylmethane

to the stream of N-methyl piperazine to form cyclizine (Scheme 4). The solubility of diphenyl

methanol was re-evaluated and acetone proved to be a better solvent, resulting in a high

concentration of 3.1 M. High concentrations are desired to minimize the use of solvent.

However, upon mixing the stream of diphenylmethanol in acetone with hydrochloric acid

diphenylmethanol precipitated in the T-mixer, due to the change in solution composition.

The relatively low melting point (69 C̊) of diphenylmethanol allowed the circumvention of

clogging by immersing the T-mixer into an oil bath. Due to the high volatility of acetone,

a lower temperature of 100 C̊ was set as reaction temperature. Due to the miscibility of

acetone with water, the mass transfer barrier between diphenylmethanol and hydrochloric

acid was minimized leading to a shorter time required for reaction completion. At 100 ̊C and

10 min, full conversion of diphenyl methanol was reached. Evaporation of acetone upon

collection resulting in 97% isolated yield of diphenylmethyl chloride. In order to carry out a

second nucleophilic substitution to synthesize cyclizine, we pumped neat piperazine into a

product stream and quenched with a NaOH solution. However, this resulted in clogging, thus

we diluted n-methyl piperazine with methyl carbitol. As a result, substantial formation of

bis- diphenylmethyl ether was observed due to the presence of water. Maximum selectivity

of 80% towards cyclizine was reached at full conversion when a wide range of operating

conditions was screened. Alternatively, as in case with benzyl chloride we proceeded with

connecting the product stream of diphenylmethyl chloride to a liquid-liquid separator.

Applying 5 psi on the aqueous outlet resulted in the breakthrough of retentate at the

permeate outlet. Applying a pressure of 2 psi via an ultra-low volume adjustable BPR on

the aqueous outlet, while attaching tubing of 20 cm with 250 µm ID to organic side outlet

gave a perfect qualitative separation. A continuous run on a 7 mmol/h scale was carried out

during 8 h resulted in the loss of 0.2% of organic phase into the aqueous phase. Karl Fischer

titration indicated the presence 1.78% of water within the separated product.

Scheme 3. Diphenylmethyl piperazine based generic APIs used in treating nausea, vomiting, cerebral apoplexy and dizziness.

The product was pumped subsequently into a T-mixer where it was mixed with n-methyl

piperazine (1.5 eq) in methyl carbitol, with a final concentration of 1.5 M of diphenylmethyl

chloride. At high temperatures due to the presence of water, partial hydrolysis of

diphenylmethyl chloride was observed. Therefore, we decreased the temperature to

100 ̊C and increased the residence time to 45 min to reach full conversion. Transparent

crystals precipitated upon cooling. The first crop resulted in 92% yield of cyclizine with

high purity (>99% GC-MS). Evaporation of acetone and extraction of the rest of the

mother liquor afforded 2% more of cyclizine. Impurities detected in mother liquor were

diphenylmethanol and bis-diphenylmethyl ether.

2.4.2. Synthesis of meclizine and buclizine derivativesMeclizine and buclizine are alike in terms of their chemical structures and are built

from 1-chloro-4-phenylmethyl benzene, piperazine and benzyl moieties. 1-chloro-4-

phenylmethyl benzene resembles diphenyl methyl moiety, therefore we decided to

build 1-diphenylmethyl-4-benzylpiperazine as a common derivative. Scheme 5 shows

the schematic representation of the overall synthesis. More information on assembly

of the setup and its operation is provided in the supporting information. The synthesis

and separation of diphenyl methyl and benzyl chlorides took place as described above.

The synthesis of 1-(diphenylmethyl)piperazine started with repeating the conditions

for benzyl chloride (in methyl carbitol: water (2:1)). However, at full conversion of

chlorodiphenylmethane, only a minor formation of 1-(diphenylmethyl) piperazine (<15 %

yield) was formed, while major products were ((2-(2-methoxyethoxy)ethoxy) methylene)

dibenzene and diphenylmethanol. Similar results were obtained when ethanol and

methanol were used as solvents. We concluded that due to the higher steric hindrance

of diphenylmethyl chloride, compared to benzyl chloride, nucleophilic substitution by

smaller molecules, such as water or alkoxides were faster. Alternative organic solvents

such as toluene, heptane, dimethyl formamide, acetonitrile, dimethoxyethane,

methyl tetrahydrofuran, 4-methyl pentanone, acetone, dichloromethane and various

combinations could not dissolve piperazine. Combination of dichloromethane and

Scheme 4. Synthesis of cyclizine in continuous flow starting from diphenylmethanol.

54


55

2

acetone (4:1) dissolved piperazine with a maximum concentration of 1.26 M. However,

the mixture was metastable and lead to precipitationof piperazine when pumped through

a reactor tubing. Therefore, we switched to tetrahydrofuran, which dissolved piperazine

with concentrations only up to 0.5 M. Nevertheless, precipitation within the reactor

was observed. Switching to a tubular reactor of a larger internal diameter of 1.58 mm

allowed visual observation of the precipitate formation. Stopping the reaction, followed

by pushing out the precipitate indicated that it was unreacted piperazine. One of the

causes for the precipitation could be bubble formation within the reactor, followed by

decrease of available solvent due to evaporation, causing the precipitation of piperazine.

Thus, we increased the pressure from the theoretically required 100 psi to 250 psi. As

a result less precipitation was observed. The fact that the crystals formed towards the

end of the reactor lead us to believe that those might be piperazine hydrochloride or

1-(diphenylmethyl piperazine hydrochloride. Mixing water into the product stream

dissolved the unreacted reactant, and allowed optimization of the reaction. When

conversion was incomplete, diphenyl methanol was formed upon mixing with water at

the outlet. Due to a lower concentration and non-polar solvent the reaction was slower.

Scheme 5. Schematic representation of four-step synthesis with tabulated conditions. Synthesis and separation of chlorides are connected to subsequent reactions. 2 steps of buclizine derivative and cinnarizine were the same. To differentiate the rest of the sequence cinnarizine reagents are highlighted in dashed rectangles.

Full conversion with 92% selectivity towards 1-(diphenylmethyl)piperazine was observed

at 150 ̊C given 45 min of residence time and with 1.5 equivalents of piperazine. Finally, no

precipitation was observed when water addition stopped after 2 volumes of the reactants

have passed.

Subsequently, in order to synthesize buclizine and the meclizine derivative,

1-(diphenylmethyl) piperazine needed to react with synthesized and separated benzyl

chloride. Prior to the flow experiments, batch investigations showed the precipitation

of the product from the reacting mixture. Methanol, used in an equivolumetric amount

with respect to the product mixture, was sufficient to afford homogeneity. Thus,

methanol along with 2.0 eq of benzyl chloride were injected into a product stream of

1-(diphenylmethyl)piperazine via a cross as shown in Scheme 5. At 100 ̊C and 15 min

residence time, 67% yield was observed. This result could be further increased to 71%

when the temperature was raised till 120 ̊C. Increasing temperature to 150 ̊C resulted

in full conversion of 1-(diphenylmethyl)piperazine and a yield of 89% based on diphenyl

methanol was obtained. Benzyl piperazine, dibenzylpiperazine, benzyl alcohol and

benzyl methyl ether were formed as by products in the final product mixture. Isolated

yield experiments were performed without the addition of internal standard. In order

to separate the product, all volatile solvents were removed from the collected sample.

Water was added to the product and pH was adjusted to <3. Ethyl acetate was then used

to extract excess of benzyl chloride and formed benzyl methyl ether. Subsequently, the

pH of the aqueous phase was adjusted to 4 with acetic acid/sodium acetate trihydrate

buffer. At this point secondary amines were extracted into an aqueous phase. Evaporation

of ethyl acetate followed by recrystallization of 1-diphenylmethyl-4-benzylpiperazine

from EtOH afforded 87% of isolated yield.

2.4.3. Synthesis of cinnarizineIn order to synthesize cinnarizine, 1-diphenylmethyl piperazine needed to react with

synthesized and separated cinnamyl chloride. The optimal conditions for benzyl chloride

synthesis were not optimal for the synthesis of cinnamyl chloride. Lower temperatures,

60 ̊C gave higher selectivity and maximum yield of 95% yield at 5.0 M concentration

in toluene. Batch investigations showed that homogeneous conditions were afforded,

when the final reaction was carried out in the solution of equal volumes of acetone and

25 wt% NaOH in water. Therefore, the solution and 2.0 eq of cinnamyl chloride were

injected into a product stream of 1-(diphenylmethyl)piperazine via a cross without its

intermediate separation. At 100 ̊C and 15 min residence time 77% yield was observed,

increasing time to 30 min lead to full conversion of 1-(diphenylmethyl)piperazine and

85% yield of cinnarizine based on diphenyl methanol. Isolated yield experiments were

performed without addition of internal standard in the similar manner to the synthesis

56


57

2

of 1-diphenylmethyl-4-benzylpiperazine. However, MeOH was used for recrystallization

of cinnarizine and afforded 82% of isolated yield.

2.5. CONCLUSION

We have demonstrated a simplification and acceleration of chemical reactions with the

help of micro flow technology. Bulk alcohols were used in a sequence of nucleophilic

substitution reactions to build meclizine and a buclizine derivative, and commercially

available cyclizine and cinnarizine. Our continuous-flow protocol delivered excellent yield

of benzyl chloride in 15 min residence time, under superheated conditions of 120 C̊ and

100 psi using hydrochloric acid as chlorinating agent. The same conditions were applied to

a wide variety of aliphatic and aromatic alcohols.

In order to connect chlorodehydroxylation reactions to a subsequent reaction step, a

membrane-based liquid-liquid separator was designed and integrated in the reactor

network. This allowed separation of intermediate chlorides and their consumption in

building target molecules. Overall 4-steps were realized in 90 minutes with good yield at

a production rate of 2 mmol/hr of final products.

The pumps used within the study were corroded upon their contact with hydrochloric

acid, despite the fact that pump heads were made of titania. As a solution corrosive

hydrochloric acid was delivered via a loop, which has to be refilled, while continuous

operation of chlorination steps is paused. Therefore, the greatest limitation of the

current method is the supply of hydrochloric acid. HCl gas supply could revolutionize

the synthesis, however its supply into micro flow environment is not yet developed. The

methodology demonstrated herein allows easy and a more accessible way to chlorides.

REFERENCES

1. Carey, J. S.; Laffan, D. In Pharmaceutical Process Development: Current Chemical and Engineering Challenges; The Royal Society of Chemistry, 2011; pp 39–65.

2. Carey, J. S.; Laffan, D.; Thomson, C.; Williams, M. T. Org. Biomol. Chem. 2006, 4, 2337–2347.

3. Larock, R., C. Compr. Org. Transform. 2nd ed. John Wiley Sons 1999, 689.

4. Weiss, R. G.; Snyder, E. I. Chem. Commun. (London) 1968, No. 21, 1358–1359.

5. Dubey, A.; Upadhyay, A. K.; Kumar, P. Tetrahedron Lett. 2010, 51 (4), 744–746.

6. Kimura, Y.; Matsuura, D. Ijoc 2013, 2013 (November), 1–7.

7. Ding, R.; He, Y.; Wang, X.; Xu, J.; Chen, Y.; Feng, M.; Qi, C. Molecules 2011, 16, 5665–5673.

8. De Luca, L.; Giacomelli, G.; Porcheddu, A. Org. Lett. 2002, 4 (4), 553–555.

9. Denton, R. M.; An, J.; Adeniran, B. Chem. Commun. 2010, 46 (17), 3025–3027.

10. Ayala, C. E.; Villalpando, A.; Nguyen, A. L.; McCandless, G. T.; Kartika, R. Org. Lett. 2012, 14 (14), 3676–3679.

11. Henderson, R. K.; Jiménez-González, C.; Constable, D. J. C.; Alston, S. R.; Inglis, G. G. a.; Fisher, G.; Sherwood, J.; Binks, S. P.; Curzons, A. D. Green Chem. 2011, 13, 854.

12. Adams, J. P.; Alder, C. M.; Andrews, I.; Bullion, A. M.; Campbell-Crawford, M.; Darcy, M. G.; Hayler, J. D.; Henderson, R. K.; Oare, C. A.; Pendrak, I.; Redman, A. M.; Shuster, L. E.; Sneddon, H. F.; Walker, M. D. Green Chem. 2013, 15 (6), 1542–1549.

13. Gerrard, W.; MacKlen, E. J. Appl. Chem. 1956, 6 (6), 241–244.

14. Vaccaro, L.; Lanari, D.; Marrocchi, A.; Strappaveccia, G. Green Chem. 2014, 16, 3680.

15. Gutmann, B.; Cantillo, D.; Kappe, C. O. Angew. Chemie Int. Ed. 2015, 54 (23), 6688–6728.

16. Newman, S. G.; Jensen, K. F. Green Chem. 2013, 15, 1456.

17. Hessel, Volker, Kralisch, Dana, Kockmann, N. Novel Process Windows: Innovative Gates to Intensified and Sustainable Chemical Processes; 2015.


19. Borukhova, S.; Seeger, A. D.; Noël, T.; Wang, Q.; Busch, M.; Hessel, V. ChemSusChem 2015, 8 (3), 504–512.

20. Murphy, E. R.; Martinelli, J. R.; Zaborenko, N.; Buchwald, S. L.; Jensen, K. F. Angew. Chemie 2007, 119 (10), 1764–1767.

21. Marre, S.; Adamo, A.; Basak, S.; Aymonier, C.; Jensen, K. F. 2010, 11310–11320.

22. Baxendale, I. R.; Deeley, J.; Griffiths-jones, C. M.; Ley, S. V; Saaby, S.; Tranmer, G. K. 2006, No. 1, 2566–2568.

23. Pastre, J. C.; Browne, D. L.; Ley, S. V. Chem. Soc. Rev. 2013.



26. Webb, D.; Jamison, T. F. Chem. Sci. 2010, 1 (July 1999), 675.


28. Hartman, R. L.; Sahoo, H. R.; Yen, B. C.; Jensen, K. F. Lab Chip 2009, 9 (13), 1843–1849.

29. Snead, D. R.; Jamison, T. F. Chem. Sci. 2013, 4 (7), 2822–2827.

30. Tundo, P.; Moraglio, G.; Trotta, F. Ind. Eng. Chem. Res. 1989, 28 (7), 881–890.

31. Tundo, P.; Selva, M. Green Chem. 2005, 7 (6), 464.

32. Reid, M. C.; Clark, J. H.; Macquarrie, D. J. Green Chem. 2006, 8 (5), 437.

33. Reichart, B.; Tekautz, G.; Kappe, C. O. Org. Process Res. Dev. 2013, 17 (1), 152–157.



36. Adamo, A.; Heider, P. L.; Weeranoppanant, N.; Jensen, K. F. Ind. Eng. Chem. Res. 2013, 52, 10802–10808.

37. Cervera-Padrell, A. E.; Morthensen, S. T.; Lewandowski, D. J.; Skovby, T.; Kiil, S.; Gernaey, K. V. Org. Process Res. Dev. 2012, 16, 888–900.

38. Snead, D. R.; Jamison, T. F. Angew. Chemie Int. Ed. 2015, 54 (3), 983.

CHAPTER 3Hydrogen chloride gas in solvent-free continuous conversion of alcohols to chlorides in micro flow


Borukhova, S., Noël, T, Hessel, V. (2016). Hydrogen chloride gas in solvent-free continuous conversion of alcohols to chlorides in micro flow, Org. Process Res. Dev., 20(2), 568-573.

60

CHAPTER 3 Hydrogen chloride gas in micro flow

61

3

ABSTRACT

Chlorides represent a class of valuable intermediates that are utilized in the preparation

of bulk and fine chemicals. An earlier milestone to convert bulk alcohols to corresponding

chlorides was reached when hydrochloric acid was used instead of toxic and wasteful

chlorinating agents. This chapter presents the development of an intensified solvent-

free continuous process by using hydrogen chloride gas only. The handling of corrosive

hydrogen chloride became effortless when the experimental setup was split into dry and

wet zones. The dry zone is used to deliver gas and prevent corrosion, while the wet zone is

used to carry out the chemical transformation. The use of gas instead of hydrochloric acid

allowed a decrease in hydrogen chloride equivalents from 3 to 1.2. In 20 min residence

time, full conversion of benzyl alcohol yielded 96wt% of benzyl chloride in the product

stream. According to green chemistry and engineering principles, the developed process

is of an exemplary type due to its truly continuous nature, no use of solvent and formation

of water as a sole by-product.

3.1. INTRODUCTION

Sustainable manufacturing of active pharmaceutical ingredients (APIs) encompasses several

aspects such as continuous processing, process intensification, minimization of solvent use

and advances in bioprocesses1.Currently, continuous processing is one aspect within process

intensification, which targets the reduction of equipment size, costs, energy consumption,

solvent utilization and waste generation2. Micro reactor technology is an actively studied

platform aiming at implementing continuous processing, achieving process intensification

and finally assisting in delivering sustainable processes for the production of APIs3.

In case of biologically active compound synthesis, the final target is constructed from

intermediates. The quality of the final product and its cost are inevitably dependent on

the manner intermediates are produced. Chlorides serve as good intermediates in the

synthesis of APIs, usually in nucleophilic substitutions exemplified on Scheme 1, where a

chloride atom is substituted by a nucleophile in the subsequent step4. Due to the lower

molecular weight of chloride, when compared to other halogens, substitution of chloride

generates less waste.5 Unfortunately, synthesis of chlorides from alcohols requires highly

toxic and waste-intensive chlorinating agents such as thionyl chloride6, phosphorus

chlorides7, pivaloyl chloride8, Vilsmeier reagent9, tosyl chloride10, 2,4,6-trichloro-[1,3,5]

triazine with DMF11, oxalyl chloride12 and phosgene13. Mostly, chlorinating agents are used

in stoichiometric or excessive amounts that lead to high generation of waste14.

Therefore, the ideal process would involve conversion of neat alcohols to chlorides by

hydrogen chloride (HCl). This will minimize waste generation since the sole by-product

is water. As stated in Chapter 2, Kappe et al. showed utilization of 30 wt% aqueous

hydrochloric acid in chlorination of 1-butanol, 1-hexanol and 1-decanol within a micro

reactor15. In 15 min residence time, with 3 equivalents of 30 wt% aqueous HCl and at

elevated temperatures quantitative yields were obtained. Chapter 2 presents the

development of a process utilizing 37 wt% aqueous HCl to afford a wider scope of aliphatic

and benzylic substrates in the same time range16. Furthermore, prepared chlorides were

coupled with piperazine derivatives to synthesize cinnarizine, cyclizine and buclizine

derivatives in multi-step continuous synthesis. The main drawback of both processes was

the need for 3 equivalents of hydrochloric acid to prevent synthesis of by-products, such

as ethers. In addition, continuity of the process was limited by the need to stop to refill

the hydrochloric acid loop, which was used to circumvent corrosion of the pumps.

The limited solubility of hydrogen chloride in water leads to a limitation in its concentration.

The maximum available concentration in hydrochloric acid is 37wt%, meaning that in

the nucleophilic substitution case chloride competes with water as nucleophile. In order

62


63

3

to maximize the concentration of hydrogen chloride within a medium, pure hydrogen

chloride gas can be used. Using hydrogen chloride gas can also solve the problem of

the aforementioned limited continuity and allows truly continuous processing. Large

pressurized reaction vessels need to be under constant observation when such toxic and

corrosive gases are used, requiring special precautions such as dedicated high-pressure

facilities to avoid any leakages and allow moderate solubility of gaseous reagents in the

reaction media. Gas-liquid reactions in large batch reactors are usually performed at

lower temperatures to increase gas solubility and minimize associated risks. Finally, due

to the low interfacial areas in batch reactors and low temperatures, reactions take place

in extended times, which most of the times does not justify the effort.

Meanwhile, microreactors offer a great platform for gas-liquid reactions due to the

formation of distinct, regular flow patterns with high periodicity or symmetry17. These offer

a high surface-to-volume ratio leading to high heat and mass transfer rates18. Moreover,

due to the low operating volumes of micro reactors only small volumes are pressurized

at a time, while continuously performing the reaction. The higher safety associated with

micro reactors allows investigation of a wide range of process conditions, usually leading

to process intensification. A number of examples utilizing gases as reagents within micro

flow reactors, such as CF3I,19 C2H2,

20 C2H4,21 CH2O,22 CH2N2,

23 Cl2,24 CO,25 CO2,

26 F2,27 H2,

28

NH3,29 O2

30 and O331 have been published.

Herein, we report for the first time the use of hydrogen chloride gas as a reagent for a

continuous synthesis of chloroalkanes in micro flow. Highly corrosive in the presence of

Scheme 1. Possible scaffolds that can be synthesized from chloroalkanes.

moisture, hydrogen chloride requires special precautions during the design of the process.

We present related challenging aspects and corresponding solutions. Finally, the merit of

the use of hydrogen chloride along with its limitations are presented.

3.2 EXPERIMENTAL PLATFORM

In order to control the flow rate of hydrogen chloride, a gas mass flow controller is

needed. In general, mass flow controllers are made of stainless steel and hastelloy, which

are susceptible to corrosion. Hydrogen chloride in its pure state is harmless to stainless

steel and hastelloy. However, the moment the moisture content rises above 10 ppm,

severe corrosion starts to take place. Therefore, absolute dry conditions are needed to

prevent the corrosion of the device. By splitting the experimental setup into two zones,

dry and wet, first as a hydrogen chloride gas delivery unit and second as a reaction rig,

the corrosion was circumvented.

Hydrogen chloride delivery unitFigure 1 (top) shows the assembly of the hydrogen chloride supply unit. In order to keep

moisture out of the unit all connections were of VCR type from Swagelok. Moreover, due

to the fact that polymer based tubing is permeable to moisture, stainless steel tubing of

¼” size was used. One out of three nitrogen bottles was set to 40 bar for startups and

shutdowns of the system. The other two were set to 15 bar pressure to be used in constant

purging of the system in between experiments to prevent diffusion of moisture into the

mass flow controller. Despite purging, a diffusion front is still present, which takes place

in the direction opposite to the flowing nitrogen. Therefore, a stainless steel 2 m long

‘pig tail’ with 250 µm internal diameter was added following the last valve of the delivery

unit. In our experience, inline non-metallic check valves sometimes fail at high pressures

and allow liquid to enter the gas stream. In case of such an unexpected failure, the liquid

shall destroy the mass flow controller upon reaching it due to corrosion.

In order to visually verify if any liquid ever was moving towards the mass flow controller,

transparent ethylene tetrafluoroethylene (ETFE) tubing of 250 µm was added after

stainless steel tubing. A polyether ether ketone (PEEK) valve was attached, which can be

shut in case liquid flow was detected. Another 2 m of ETFE tubing of 750 µm was added

after, followed by an inline check valve.

In order to enhance desorption of water molecules from the surface of the tubing used in

construction of the setup, a vacuum line was installed. A cyclic vacuum-purge procedure

was applied before the start of the operation and before disassembling the setup. It is

64


65

3

important to mention that in case of no vacuum, and sole purging, components of the setup

corrode upon disassembly and exposure to air. A by-pass around the mass flow controller

was installed to assist vacuuming of the mass flow controller from both ends. Before

introducing hydrogen chloride into the unit for the first time, a dew point transmitter was

used to the measure moisture content. A moisture content of 0.8 ppm was typical for our

system due to the dry nitrogen that was used for purging after purge-vacuum procedure.

Pictures with more detailed information regarding the setup are enclosed within the

supporting information of the publication the current chapter is based on.

Chlorodehydroxylation rigAlcohols that are liquid under atmospheric conditions were pumped with a Knauer HPLC

pump. A gas-liquid slug flow initiated in a Y-mixer and continued into the ETFE reactor

as shown in Figure 1 (bottom). NOTE: a T-mixer was not suitable due to the liquid slugs

appearing on the gas feed line prior to the mixer. As a result, these penetrations brought

Figure 1. Hydrogen chloride gas delivery unit (top, dry zone) and chlorodehydroxylation rig (bottom, wet zones), where reaction takes place.

about fluctuations in the gas flow rate observed on the mass flow controller. The reactor

was made of ETFE tubing of 762 µm internal diameter. When 1 mm inner diameter tubing

was used instead, escape of gas into the heating media upon operation was observed,

due to the thinner wall thickness. A hot product stream was allowed to flow through 30

cm long tubing prior to entering the Equilibar back pressure regulator (BPR) to allow

cooling. For present application, an Equilibar BPR demonstrated a unique ability to apply

pressure and keep gas-liquid flows stable up to 16 bar pressure. Its use circumvented the

need for gas-liquid separation prior to the pressurization unit. NOTE: Inline cartridge

based BPRs could not provide constant flow, and resulted rather in stop-flow behavior,

which affected the operation of the mass flow controller, thus causing inaccuracy in the

gas flow rate. Table 1 demonstrates the window of operating conditions of the operating

platform.

3.3 RESULTS AND DISCUSSION

Benzyl chloride is used as an intermediate in the preparation of pharmaceuticals,

flavorants, plasticizers and perfumes. According to Gerrard et al., hydrogen chloride

dissolves in benzyl alcohol under atmospheric pressure to a significant extent as shown in

Figure 232. Significant absorption of gas, as shown in Figure 3, was observed when benzyl

alcohol and hydrogen chloride were mixed in the tubing prior to entering the reactor

coil (at room temperature under 5 bar pressure). In theory, the solubility of gas in liquid

increases with pressure and decreases with temperature. In addition, throughout the

reactor the gas is consumed as the reaction progresses. With an increase in temperature,

the extent of gas expansion, and thus residence time as well, are hard to quantify due to

its substantial expansion and faster consumption. Therefore, the sole measure of success

of a reaction was based on the yield of synthesized chloride, while residence time was

estimated based on flow behavior. In-line absorption measurements to determine the

residence time were avoided due to the corrosive nature of the product stream.

Table 1. Operating parameters of chlorodehydroxylation flow setup

Parameter Range

Reactor volume 4 ml

Reactor internal diameter 762 µm

Gas mass flow rate 0.001-0.600 g/min (0.1-80 ml/min)

Reactant flow rate 0.09-0.15 ml/min

Temperature 25-120 ̊C

Pressure 16 bar

66


67

3

One of the goals in using gas was to minimize excess of HCl used, thus generated waste.

In our previous investigations with 3 equivalents hydrochloric acid, in 15 min residence

time at 120 ̊C >99% yield of benzyl chloride was obtained. Decreasing equivalents of HCl

gas to one and allowing the same residence time resulted in 80wt% at 60 ̊C and 89wt% at

100 ̊C. Higher temperatures were not investigated due to a significant expansion of gas

slugs, leading to significantly reduced residence time. Dibenzyl ether was formed as a

sole side-product in 3wt% at 60 ̊C and 5wt% at 100 ̊C. Important to stress is that no gas

slugs were observed at the inlet of BPR, and minor amount of gas slugs was observed at

the outlet.

0

0,2

0,4

0,6

0,8

1

0 10 20 30 40 50

mol

HC

l/m

ol R

-OH

Temperature, (̊C)

Figure 2. Solubility of hydrogen chloride gas in 1-butanol and benzyl alcohol (redrawn with permission of John Wiley and Sons from).32

Figure 3. Gas slug size with respect to distance from the mixing point of gas and liquid, Y-mixer.

0

1

2

3

4

0 10 20 30

Gas

slu

g si

ze (

cm)

Distance from Y-mixer (cm)

To see whether benzyl ether formation could be minimized, while maximizing the yield of

benzyl chloride, the effect of hydrogen chloride excess was studied. Gradually, increasing

equivalents from 1.0 to 2.0 led to no change in side-product formation at 100 ̊C. We

therefore decided to screen different reaction temperatures at 1.1 and 1.5 equivalents.

The results tabulated in Table 2 show that the selectivity does not improve with excess

of hydrogen chloride. With the increase of equivalents, gas- hold up in the reactor

increased, which led to a marginal decrease in residence time. Assuming that dibenzyl

ether is formed due to insufficient hydrogen chloride, prompted us to increase the system

pressure to allow a higher concentration of hydrogen chloride within the liquid phase.

Figure 4 shows the pressure effect on the reactant, product and side-product weight

distribution at 80 ̊C. Benzyl chloride production increases with pressure from 79wt% at

5 bar to 93wt% at 16 bar, while the formation of byproduct stays the same and equal to

3-4wt%. Thus, higher concentration of hydrogen chloride increased the conversion, while

showing no effect on selectivity. No change in conversion was observed with increasing

pressure at 90 ̊C and 100 ̊C. Dibenzyl either formed in 4wt% at 90 ̊C with 92wt% of benzyl

chloride, and in 5wt% at 100 ̊C with 95wt% of benzyl chloride. We proceeded further

with increasing equivalents at higher pressure (10, 12, 16 bar) at 100 ̊C. The minimum

concentration of dibenzyl ether was 4% at 10 bar at 1.2 and more equivalents. Therefore,

those conditions, i.e. 100 ̊C, 1.2 equivalents, 20 min residence time and 10 bar were set

as optimal, yielding full conversion and 96wt% of benzyl chloride.

3.3.1 Expansion of scopeOptimized conditions for benzyl alcohol were applied to a range of aliphatic and benzylic

alcohols. Scheme 2 shows corresponding yields at 100 ̊C, 1.2 equivalents and 10 bar.

0

20

40

60

80

100

5 10 15

%w

t

pressure (bar)

Figure 4. Pressure effect on weight distribution among benzyl chloride (red), benzyl alcohol (blue) and dibenzyl ether (green).

68


69

3

Table 2. Effect on benzyl chloride and dibenzyl ether formation at various temperatures and equivalents of hydrogen chloride.

Entry Eq HCl T ( ̊C) BenzCl (wt%) DBE (wt%)

1 1.1 80 89 3

2 1.1 90 92 5

3 1.1 100 95 5

4 1.5 80 87 3

5 1.5 90 96 4

6 1.5 100 95 5

7 2.0 100 95 5

When, aliphatic alcohols were used a significant decrease in gas solubility was observed

that lead to large gas slugs both at the Y-mixer and at the BPR outlet. Increase in gas slugs

drastically decreased the residence time to <5 min in the reactor (4 ml) used for benzyl

alcohol. In order to have a similar residence time as for benzyl alcohol experiments, a

10 ml reactor was used instead, which led to residence time ranging from 15-20 min

depending on the substrate used. The difference in residence time is due to the different

extent of gas absorption within the liquid phase.

The rate of nucleophilic substitution reactions depends on the type of nucleophile,

electrophile, solvent and leaving group. In order to assist nucleophilic substitution, either

polar protic (SN1) or aprotic (SN2), solvents are used as solvents. One of the frequently

used reaction systems in solvent-free nucleophilic substitutions is solvolysis, when the

solvent is used as a nucleophile. The current investigation comprises a reverse system,

where electrophile is a solvent. While the nucleophile is a chloride anion resulting from

dissociation of hydrogen chloride.

Primary alkanes undergo SN2 type of nucleophilic substitution, which occurs via a back-

attack of an electrophile33. Due to insufficient polarity and relatively low hydrogen bonding

in primary aliphatic alcohols, dissociation to release a free chloride anion as a nucleophile

is not favorable. In contrast, due to the higher polarity and/or stronger hydrogen bonding

present within pentanediol, both solubility of HCl gas and conversion of the alcohol were

significantly increased. Conversion of pentanediol increased to 98%, resulting in the

formation of 1,5-dichloro pentane as a sole side product. Secondary alkanes can react via

SN2 or SN1 depending on the steric hindrance of an electrophilic carbon or stabilization of

the cationic intermediate. Higher yields in isopropyl alcohol and cyclohexyl alcohol cases

can be explained by this additional path being available. Among benzylic species 3-methoxy

benzyl alcohol resulted in the largest gas consumption and the highest yield with no traces

of a dibenzyl ether derivative forming. Conversion decreased when 2,6-difluorobenzyl

alcohol was used, while no side product was formed. In case of 2-phenylethanol absolutely

no conversion was observed. These observations indicate that resonance stabilization of

the benzyl cation is responsible for better yields when compared to aliphatic substrates

and that SN1 is a main path for the reactions to take place.

In case with secondary and benzylic alcohols dissociation of protonated alcohols results in

water and carbocation. As the reaction proceeds, more of the water is formed promoting

dissociation of hydrogen chloride, increasing not only the absorption of HCl, but also

the rate of SN1 reactions. Looking back at the optimization of benzyl alcohol, it can be

concluded that at the start of the reaction when limited water is formed as a by-product,

resulting in a limited amount of chloride anion, benzyl cation associates with both benzyl

alcohol and chloride. Once a certain threshold concentration of water is reached, the

chloride prevails in the reaction to combine with the benzyl cation. The SN1 reaction

pathway is also supported by the fact that no effect on the selectivity was observed when

an excess of hydrogen chloride was used.

3.4 CONCLUSION

A continuous synthesis of chlorides from bulk alcohols via use of hydrogen chloride gas

instead of toxic and wasteful chlorinating agents is demonstrated. The most challenging

aspect of safe handling of this corrosive gas was its continuous controlled delivery into the

reacting system. The elimination of moisture by continuous purging of the setup with dry

nitrogen solved any corrosion issues.

Scheme 2. Prepared chloroalkanes at optimal for benzyl chloride conditions, i.e. 120 °C, 10 bar with 1.2 equivalents of hydrogen chloride. *-reactions carried out in 10 ml reactor.

70


71

3

The use of a micro flow reactor allowed the application of process conditions that are

beyond the limits of conventional batch technology. High temperature and pressure, easily

applicable within the reacting system allowed reaching the initial objective of minimizing

the excess of HCl used from 3 equivalents to 1.2. However, the developed process is

more beneficial for SN1 type reactions than SN2. As a result, further investigations are

needed for the synthesis of specific target chloroalkanes. One of the possible steps is

minimal addition of polar aprotic solvents to promote SN2. Based on green chemistry

and engineering principles, a significant improvement is demonstrated due to the truly

continuous nature of the process, no use of solvent and formation of only water as a by-

product.

ACKNOWLEDGMENT

We would like to thank Erik van Herk from Eindhoven University of Technology, Bert

Metten from Omnichem and Bart van Heugten from TSCC Technology for their help during

different stages of our investigation. Funding by the Advanced European Research Council

Grant “Novel Process Windows – Boosted Micro Process Technology” (Grant number:

267443) is kindly acknowledged.

REFERENCES

1. Jiménez-González, C.; Poechlauer, P.; Broxterman, Q. B.; Yang, B.-S.; am Ende, D.; Baird, J.; Bertsch, C.; Hannah, R. E.; Dell’Orco, P.; Noorman, H.; Yee, S.; Reintjens, R.; Wells, A.; Massonneau, V.; Manley, J. Org. Process Res. Dev. 2011, 15, 900–911.

2. Poechlauer, P.; Colberg, J.; Fisher, E.; Jansen, M.; Johnson, M. D.; Koenig, S. G.; Lawler, M.; Laporte, T.; Manley, J.; Martin, B.; O’Kearney-McMullan, A. Org. Process Res. Dev. 2013, 17, 1472–1478.

3. (a) Vaccaro, L.; Lanari, D.; Marrocchi, A.; Strappaveccia, G. Green Chem. 2014, 16, 3680–3704; (b) Hessel, V.; Kralisch, D.; Kockmann; N., Noël, T.; Wang, Q.; ChemSusChem 2013, 6, 746-789; (c) Tambarussi Baraldi, P.; Hessel, V. Green Process Synth 2012, 1, 149-168; (d) Borukhova, S. and Hessel, V. 2015, in Continuous Manufacturing of Pharmaceuticals (eds. P. Kleinebudde, J. Khinast and J. Rantanen), Wiley Inc.

4. (a) Carey, J. S.; Laffan, D.; Thomson, C.; Williams, M. T. Org. Biomol. Chem. 2006, 4, 2337–2347; (b) Carey, J. S.; Laffan, D. In Pharmaceutical Process Development: Current Chemical and Engineering Challenges; The Royal Society of Chemistry, 2011; pp. 39–65.

5. Anastas, P.; Eghbali, N. Chem. Soc. Rev. 2010, 39, 301–312.

6. Chaudhari, S. S.; Akamanchi, K. G. Synlett 1999, 1763–1765.

7. Weiss, R. G.; Snyder, E. I. Chem. Commun. (London) 1968, 1358–1359.

8. Dubey, A.; Upadhyay, A. K.; Kumar, P. Tetrahedron Lett. 2010, 51, 744–746.

9. Kimura, Y.; Matsuura, D. Ijoc 2013, 2013, 1–7.

10. Ding, R.; He, Y.; Wang, X.; Xu, J.; Chen, Y.; Feng, M.; Qi, C. Molecules 2011, 16, 5665–5673.

11. De Luca, L.; Giacomelli, G.; Porcheddu, A. Org. Lett. 2002, 4, 553–555.

12. Denton, R. M.; An, J.; Adeniran, B. Chem. Commun. 2010, 46, 3025–3027.

13. Ayala, C. E.; Villalpando, A.; Nguyen, A. L.; McCandless, G. T.; Kartika, R. Org. Lett. 2012, 14, 3676–3679.

14. Adams, J. P.; Alder, C. M.; Andrews, I.; Bullion, A. M.; Campbell-Crawford, M.; Darcy, M. G.;

Hayler, J. D.; Henderson, R. K.; Oare, C. A.; Pendrak, I.; Redman, A. M.; Shuster, L. E.; Sneddon, H. F.; Walker, M. D. Green Chem. 2013, 15, 1542–1549.

15. Reichart, B.; Tekautz, G.; Kappe, C. O. Org. Process Res. Dev. 2013, 17, 152–157.

16. Borukhova, S.; Noël, T.; Hessel, V. ChemSusChem 2016, 9, 67–74.

17. (a) Hessel, V.;Angeli, P.; Gavriilidis, A.; Löwe, H.; Ind. Eng. Chem. Res. 2005, 44, 9750–9769. (b) Gunther, A.; Khan, S. A.; Thalmann, M.; Trachsel, F.; Jensen, K. F. Lab Chip 2004, 4, 278–286. (c) Dencic, I.; Hessel, V.; De Croon, M. H. J. M.; Meuldijk, J.; Van Der Doelen, C. W. J.; Koch, K. ChemSusChem 2012, 5, 232–245. (d) Dencic, I.; Hessel, V. In Microreactors in Organic Chemistry and Catalysis; 2013; pp. 221–288. (e) Chen, G.; Yue, J.; Yuan, Q. Chinese J. Chem. Eng. 2008, 16, 663–669. (f) Sobieszuk, P.; Aubin, J.; Pohorecki, R. Chem. Eng. Technol. 2012, 35, 1346–1358.

18. (a) Mallia, C. J.; Baxendale, I. R. Org. Process Res. Dev. 2015. (b) Su, Y.; Hessel, V.; Noël, T. AIChE J. 2015, 61, 2215–2227.

19. (a) Straathof, N. J. W.; Gemoets, H. P. L.; Wang, X.; Schouten, J. C.; Hessel, V.; Noël, T. ChemSusChem 2014, 7, 1612–1617. (b) Straathof, N. J. W.; Tegelbeckers, B. J. P.; Hessel, V.; Wang, X.; Noel, T. Chem. Sci. 2014, 5, 4768–4773. (c) Straathof, N.; Osch, D.; Schouten, A.; Wang, X.; Schouten, J.; Hessel, V.; Noël, T. J. Flow Chem. 2014, 4, 12–17.

20. Deng, Q.; Shen, R.; Ding, R.; Zhang, L. Adv. Synth. Catal. 2014, 356, 2931–2936.

21. (a) Samuel, C.; Bourne, L.; Koos, P.; Brien, M. O.; Martin, B.; Schenkel, B.; Baxendale, I. R.; Ley, S. V. 2011. (b) Terao, K.; Nishiyama, Y.; Tanimoto, H.; Morimoto, T.; Oelgemöller, M.; Morimoto, T. J. Flow Chem. 2012, 2, 73–76.

22. Buba, A. E.; Koch, S.; Kunz, H.; Löwe, H. European J. Org. Chem. 2013, 2013, 4509–4513.

23. (a) Maurya, R. A.; Park, C. P.; Lee, J. H.; Kim, D.-P. Angew. Chemie Int. Ed. 2011, 50, 5952–5955. (b) Pinho, V. D.; Gutmann, B.; Miranda, L. S. M.; de Souza, R. O. M. A.; Kappe, C. O. J. Org. Chem. 2014, 79, 1555–1562.

24. Hrich, H.; Linke, D.; Morgenschweis, K.; Baerns, M.; Jähnisch, K. Chim. Int. J. Chem. 2002, 56, 647–653.

25. (a) Gong, X.; Miller, P. W.; Gee, A. D.; Long, N. J.; de Mello, A. J.; Vilar, R. Chem. – A Eur. J.

72

CHAPTER 3

2012, 18, 2768–2772. (b) Miller, P. W.; Long, N. J.; de Mello, A. J.; Vilar, R.; Audrain, H.; Bender, D.; Passchier, J.; Gee, A. Angew. Chem. Int. Ed. Engl. 2007, 46, 2875–2878. (c) Miller, P. W.; Long, N. J.; de Mello, A. J.; Vilar, R.; Passchier, J.; Gee, A. Chem. Commun. 2006, 546–548. (d) Murphy, E. R.; Martinelli, J. R.; Zaborenko, N.; Buchwald, S. L.; Jensen, K. F. Angew. Chem. Int. Ed. Engl. 2007, 46, 1734–1737. (e) Koos, P.; Gross, U.; Polyzos, A.; O’Brien, M.; Baxendale, I.; Ley, S. V. Org. Biomol. Chem. 2011, 9, 6903–6908. (f) Gross, U.; Koos, P.; O’Brien, M.; Polyzos, A.; Ley, S. V. European J. Org. Chem. 2014, 2014, 6418–6430. (g) Mercadante, M. a; Leadbeater, N. E. Org. Biomol. Chem. 2011, 9, 6575–6578. (h) Takebayashi, Y.; Sue, K.; Yoda, S.; Furuya, T.; Mae, K. Chem. Eng. J. 2012, 180, 250–254.

26. (a) Wu, J.; Yang, X.; He, Z.; Mao, X.; Hatton, T. A.; Jamison, T. F. Angew. Chemie Int. Ed. 2014, 53, 8416–8420. (b) Pieber, B.; Glasnov, T.; Kappe, C. O. RSC Adv. 2014, 4, 13430–13433. (c) Kupracz, L.; Kirschning, A. Adv. Synth. Catal. 2013, 355, 3375–3380. (d) van Gool, J. J. F.; van den Broek, S. A. M. W.; Ripken, R. M.; Nieuwland, P. J.; Koch, K.; Rutjes, F. P. J. T. Chem. Eng. Technol. 2013, 36, 1042–1046. (e) Sci, C.; Brien, M. O.; Taylor, N.; Polyzos, A.; Baxendale, I. R.; Ley, S. V. 2011, 1250–1257.

27. McPake, C. B.; Sandford, G. Org. Process Res. Dev. 2012, 16, 844–851.

28. Cossar, P. J.; Hizartzidis, L.; Simone, M. I.; McCluskey, A.; Gordon, C. P. Org. Biomol. Chem. 2015, 13, 7119–7130.

29. (29) Cranwell, P. B.; O’Brien, M.; Browne, D. L.; Koos, P.; Polyzos, A.; Pena-Lopez, M.; Ley, S. V. Org. Biomol. Chem. 2012, 10, 5774– 5779.

30. (a) Gemoets, H. P. L.; Hessel, V.; Noël, T. Org. Lett. 2014, 16, 5800–5803. (b) Vanoye, L.; Aloui, A.; Pablos, M.; Philippe, R.; Percheron, A.; Favre-Réguillon, A.; de Bellefon, C. Org. Lett. 2013, 15, 5978–5981. (c) Chaudhuri, S. R.; Hartwig, J.; Kupracz, L.; Kodanek, T.; Wegner, J.; Kirschning, A. Adv. Synth. Catal. 2014, 356, 3530–3538. (d) Sipos, G.; Gyollai, V.; Sipőcz, T.; Dormán, G.; Kocsis, L.; Jones, R. V; Darvas, F. J. Flow Chem. 2013, 3, 51–58. (e) Pieber, B.; Kappe, C. O. Green Chem. 2013, 15, 320–324. (f) Gemoets, H. P. L.; Su, Y.; Shang, M.; Hessel, V.; Luque, R.; Noel, T. Chem. Soc. Rev. 2016, 45, 83–117. (g) Talla, A.; Driessen, B.; Straathof, N. J. W.; Milroy, L.-G.; Brunsveld, L.; Hessel, V.; Noël, T. Adv. Synth.

Catal. 2015, 357, 2180–2186.

31. (a) Wada, Y.; Schmidt, M.A.; Jensen, K.F. Ind. Eng. Chem. Res. 2006, 45, 8036–8042. (b) O’Brien, M.; Baxendale, I. R.; Ley, S. V. Org. Lett. 2010, 12, 1596–1598. (c) Irfan, M.; Glasnov, T. N.; Kappe, C. O. Org. Lett. 2011, 13, 984–987. (d) Zak, J.; Ron, D.; Riva, E.; Harding, H. P.; Cross, B. C. S.; Baxendale, I. R. Chem. – A Eur. J. 2012, 18, 9901–9910.

32. Gerrard, W.; MacKlen, E. J. Appl. Chem. 1956, 6, 241–244.

33. Streitwieser, A. Chem. Rev. 1956, 56, 571–752.

CHAPTER 4Pressure accelerated azide-alkyne cycloaddition


Borukhova, S., Seeger, A. D., Noël, T, Wang, Q., Busch, M., Hessel, V. (2015). Pressure accelerated azide-alkyne cycloaddition: micro capillary vs. autoclave reactor performance in yielding Rufinamide precursor. ChemSusChem, 8(3), 504-512.

76

CHAPTER 4 Pressure accelerated azide-alkyne cycloaddition

77

4

ABSTRACT

Pressure effects on regioselectivity and yield of cycloaddition reactions have been

shown to exist. Nevertheless, high pressure synthetic applications with subsequent

benefits in the production of natural products are limited by the general availability of

the equipment. Meanwhile, the virtues and limitations of micro flow equipment in the

standard environment are well established. Herein, we apply Novel Process Windows (NPW)

principles, such as intensification of intrinsic kinetics of a reaction via high temperature,

pressure and concentration, on azide-alkyne cycloaddition towards Rufinamide precursor.

We applied the three main activation modes on uncatalyzed and catalyzed azide-alkyne

cycloaddition. We compare performances of two reactors, a specialized autoclave batch

reactor for high-pressure operation up to 1800 bar and a capillary flow reactor (up to

400 bar). A differentiated and comprehensive picture is given for the two reactors and

three modes of activation, i.e. uncatalyzed batch, uncatalyzed flow and catalyzed flow.

Reaction speed-up and consequent increases in space-time yields were reached, while

widening process windows maps of favorable operation to selectively produce Rufinamide

precursor in good yields. The best conditions were applied to several azide-alkyne

cycloadditions to widen the scope of the presented methodology.

4.1. INTRODUCTION

High temperature, pressure, concentration and/or addition of catalyst can maximize

reaction kinetics of most chemical reactions1. Provided that activation energy of a

particular reaction is positive (Ea>0), its reaction rate constant will increase at higher

temperature. Similarly, as long as activation volume (∆V≠< 0) of a reaction is negative,

increasing pressure will facilitate the reaction2,3. Thus, application of relatively high

temperature and pressure on a reactive system may lead to a chemical intensification4–7.

Novel Process Windows (NPWs) is a concept that embraces the opportunities presented by

chemical and process-design intensification4. Reducing the size of equipment to a micro

scale leads to a process intensification due to the enhanced heat- and mass-transfer8.

Continuous micro flow operation is a very desirable mode in pharmaceutical industry due

to a possible high degree of control over reaction parameters, a higher safety, a reduced

manual handling, a flexibility of production volume, an easier reproducibility, a possibility

of reaction telescoping, and a potential for an integrated purification9,10. Even though

high-temperature processing in micro flow reactors resulted in numerous successful

cases6,11–15, high pressure operations, having a potential of accelerating reactions and

directing selectivity, still constitute to a mystery nowadays. Up to now, pressure effects

in micro flow have been observed (i) under sub- or supercritical conditions and (ii) in

enhancement of interfacial mass transfer for a gas-liquid reactions16–21. In the case of

pressure impact on reaction with negative activation volume as envisaged in this study,

notable impact with capillaries or microchips was mostly achieved under non-continuous,

stop-flow, conditions17. Most powerful high-pressure experimental installations can

sustain the pressure of up to 3 GPa (30,000 bar) and are based on a diamond anvil cell or

piston-cylinder type reactors22–24. Such reactors supply new kind of information on physical

properties of matter as well as on the reaction dynamics and mechanism. High fabrication

cost, limited volume, high energy cost and constrained safety are however the main reasons

for not being widely applicable on a larger scale. To our best knowledge industrially used

high pressure reactors range in 5-2000 bar, and require special precautions, advanced

safety control installation, regular check of tightened areas for leaks and constant

monitoring during the operation. In micro flow, high pressure applications fall into a much

narrower space of only 5-600 bar20,25–27. Pressure of a few hundred bars can be applied

and sustained within a micro flow reactor of any desired volume by HPLC pumps with an

appropriate back pressure regulators. It is expected that the handling of flow reactors

at high pressures is generally easier17. Flow reactors allow high pressures to be easily

reached due to their lower inventory, the low share of tightening areas and they provide

a variable volume for production.

78


79

4

As mentioned earlier, reactions with negative activation volume constitute to a class

of reactions facilitated by high pressure2,28. Thus, reactions such as cycloadditions,

condensations, reactions proceeding via cyclic transition state, such as Cope and Claisen

rearrangements, reactions involving formation of dipolar transition states, such as

electrophilic aromatic substitutions, and reactions with steric hindrance can be influenced

by high pressure. Moreover, based on the difference in volume occupied by a product,

distribution of the reaction products can be altered. One of the most extensively studied

class of reactions under high pressure is [4+2] Diels-Alder cycloadditions due to their

wide applications in general, and second most negative change in activation volume,

-25— - 50 mL/mol29–34. High pressure was shown to direct regioselectivity of cycloaddition

due to the difference in volume of regioisomers, changes in electronic demand and steric

hindrance35. Moreover, the combination of catalysis and high pressure was demonstrated

to have a synergistic effect, when Lewis acid was used to catalyze cycloaddition of

pyrrole derivative with an electron-rich diene36–39. Finally, high pressure affects the

reaction medium by affecting its physical properties, such as boiling and melting points,

density, viscosity, dielectric constant, compressibility, conductivity and surface tension,2,22

however, this is out of the scope of the present study.

Several reactions have been performed in micro reactors under high-pressure and stop-

flow regime. Nucleophilic aromatic substitution reaction among p-halonitrobenzenes

with cyclic amines has been investigated in micro capillary under batch conditions

under the pressures up to 600 bar17. Rate enhancements by a factor of 2.7, 1.7, and

1.5 were observed for pyrrolidine, piperidine and morpholine, respectively. Diels-Alder

reaction of 2- and 3-furylmethanol with maleimides, performed under elevated pressure,

demonstrated that pressure increases the rate of the 2-furylmethanol, which is less

reactive than 3-furylmethanol under atmospheric conditions. Larger negative change in

the reaction volume of exo-product when compared to endo-product formation resulted

in a slight increase in the formation of exo-product. An increase of the reaction rate of

the Diels-Alder reaction, between cyclopentadiene and phenylmaleimide, by a factor of

14 was observed upon increasing the pressure to 150 bars in high-pressure glass micro

reactor. Kappe et al. reported multiple high-pressure, high-temperature accelerations in

reaction rates of Newman-Kwart and Claisen rearrangements, a Fischer indole synthesis

and nucleophilic substitution40,41.

Due to their lower activation volumes, [3+2] Huisgen cycloadditions are less popular

class of reactions in being studied under high pressure42–46. [3+2] Huisgen cycloaddition

takes place when 1,3-dipole reacts with a dipolarophile to form 5-membered cyclic

compounds. Azides represent one class of 1,3-dipoles and their reaction with terminal

alkynes results in a mixture of 1,4- and 1,5-cycloadducts, unless selectivity is directed by

a catalyst in favor of a single cycloadduct42. Azide-alkyne cycloaddition, when catalyzed

by copper, serves as one of the best examples of a ‘perfect’ reaction termed as ‘click’

reaction by Kolb and Sharpless in 200147–49. In the last decade, click reaction became a

synthetic target tool with a special accent in the use of combinatorial chemistry to yield

natural products on the way of drug development50. Copper-catalyzed cycloaddition of

alkyl azides and terminal alkynes results in 1,4-substituted 1,2,3-triazole, which is the

building block of many natural products51–55. One of the best-selling 200 drugs of recent

years is an antiepileptic drug, 1,2,3-triazole-Rufinamide (Scheme 1)56. The production

process was initially developed by Novartis and now realized by Eisai Ltd., under the

commercial names of Inovelon and Banzel. The anticonvulsant is used in the treatment

of seizures associated with Lennox-Gastaut syndrome of patients older than 4 years old57.

Total continuous flow synthesis of Rufinamide starting from 2,6-difluorobenzyl bromide

and methyl propiolate was recently reported by Jamison et al58.

Here in, we focus on optimization of 1,3-dipolar cycloaddition of 2,6-dilfuorobenzyl azide

and methyl propiolate, which leads to 4-substituted 1,2,3-triazole, Rufinamide precursor

(Scheme 2). Separation of 1,5-cycloadduct is a requirement in the industrially applied

process, when performed without Cu-catalyst. We investigate the effect of pressure on

the regioselectivity to maximize the yield of the desired 1,4-cycloadduct. Moreover, we

look into a synergy between high pressure and catalyst in affecting the reaction outcome.

Additionally, the performances of the high-pressure autoclave reactor and the flow

reactor, two specialised apparati built for the current study, are compared in the light

of the herein mentioned advantages and disadvantages. Finally, best flow conditions are

applied on a wider scope of azide-alkyne cycloadditions.

Scheme 1. Rufinamide structure

Scheme 2. 1,3-dipolar cycloaddition to Rufinamide precursor

80


81

4

4.2. OPERATING PLATFORMS

4.2.1 Process windows of reactors under studyThe reaction of interest was studied using three systems different in operating pressure

and temperature limits, as well as in operating mode. The first system is the standard

round bottom flask under atmospheric conditions, that is limited in terms of temperature

by the boiling point of the solvent.

The second system is the high pressure autoclave reactor setup having operating limits of

2500 bar and 300 ̊C. The reactor of 14 mL reactor volume was constructed according to

a schematic representation shown in Figure 1 (left). No copper or ruthenium-containing

material was used in the construction of the reactor. Two heating jackets surrounded

upper and lower halves of the reactor. Thermocouple was inserted in one of the four

inlets, with the tip located in the middle of the reactor. Two pumps were needed to deliver

the reacting mixture and generation of a higher pressure. One of the pumps was auxiliary

and was used to fill the syringe pump and the reactor with a reacting mixture, while the

syringe pump was used to apply pressure. Pressure was measured by a transducer, located

between the syringe pump and the reactor. Rupture disc with the upper limit of 2500 bar

was inserted at another outlet of the reactor as a barrier with an emergency outlet. In

case of pressure build-up higher than 2500 bar, reactor contents would be sucked in from

the reactor interior, minimizing the risk of the experiments. The components of the setup

were connected in a manner demonstrated by Figure 1 (right).

Figure 1. Left: Autoclave reactor with 14 ml of internal volume and operating limits of 2500 bar and 300 ̊C. Four outlets: inlet, sample collection outlet, thermocouple and connection to the safety line with rupture disk, withstanding 2500 bar. Right: Autoclave batch high-pressure setup consists (from left to right) of feed tank, auxiliary manual pump (250 bar), three valves, manual syringe pump (3000 bar), pressure transducer and autoclave reactor, surrounded with electrical heating jackets.

Figure 2. High pressure, high temperature (HPHT) micro capillary based flow system with operating limits of 400 bar and 300 ̊ C. The setup consists of stainless steel (SS) capillary (500 µm ID, 10 m), two HPLC pumps of Knauer 1000 series, heating and cooling Lauda oil baths, sample loop connected to a 6-port Vici Valco valve, and one Bronkhorst back-pressure regulator (BPR).

High temperature, high pressure (HPHT) flow setup was assembled from the commercially

available equipment and is shown in Figure 2. HPLC pumps can operate at pressures up

to 400 bar, while BPR keeps the system under the set pressure. SS micro capillary reactor

was heated in the oil bath with upper temperature limit of 300 ˚C. Second bath was used

to cool down the reacting mixture with the goal of quenching the reaction and preparing

the stream for the safe sample collection. The valve was used as a sample collection loop

so that dead volume of BPR would not contribute to residence time of reactants. The BPR

was selected to be combined with a control valve applicable at relatively low flow rates,

2 – 20 ml/min. Residence time was manipulated by change in flow rate and length of the

capillary tubing. Due to the lower internal volume (2 mL) and faster existing p, T process

window of the standard batch reactor could be considerably widened by the use of more

advanced, such as autoclave and micro flow, reactors.

4.3. RESULTS AND DISCUSSION

We performed our investigations in three stages:

1. Batch experiment in a stirred glass round bottom flask under uncatalyzed conditions;

2. High-pressure and -temperature experiments in a non-stirred autoclave reactor

under uncatalyzed conditions;

3. High-pressure and -temperature experiments in HPHT flow reactor under un- and

catalyzed conditions.

82


83

4

4.3.1 Exploring the pressure and temperature windows in the standard and autoclave batch reactorsWe performed cycloaddition in a stirred round bottom glass flask at 90 C̊ and 0.25 M

concentration of 2,6-difluorobenzyl azide. After 24 hrs the yield of 55% of 1,4-cycloadduct

was obtained, leading to 60% yield in 6 days. The product distribution remained the

same throughout the reaction with the ratio of 1,4-:1,5-cycloadducts of 2.8:1. In order

to determine the activation volume and pressure effect on 1,3-dipolar cycloaddition of

2,6- difluorobenzyl azide to methyl propiolate, we performed high-pressure experiments in

the non-stirred autoclave batch reactor. Performing the reaction at 500 bar and otherwise

same conditions resulted in a yield of 72% of the 1,4-cycloadduct with increased preference

to a desired product, giving ratio of 3.6:1 of 1,4-:1,5-cycloadducts, as shown in Figure 3. as

reference and calculating relative multiplication factor of subsequent reaction rates lead

to a calculation of activation volume of -21.13 mL/mol, which is consistent with published

literature results for 1,3- dipolar cycloadditions2b. Calculations of activation volume are

described within the supporting information of the publication the current chapter is based

on. Switching our focus to temperature effect we performed the reaction at 1200 bar in

batch reactor allowing again 24 hrs as a reaction time, under diluted conditions (0.25 M).

Lower yields due to the decomposition of 2,6-difluorobenzyl azide were obtained at higher

temperatures, i.e. 66% and 2% at 175 ̊C and 250 ̊C, respectively. Thus, the high-temperature

processing has its intrinsic limitations and motivated the study reported hereon.

0

1

2

3

4

5

6

7

0

20

40

60

80

100

0 500 1000 1500 2000

Ratio 1,4/1,5Yiel

d, (

%)

Pressure, (bar)

1,4

1,5

Figure 3. Pressure effect on the yield of uncatalysed azide-alkyne cycloaddition of 2,6-difluorobenzyl azide and methyl propiolate (2 eq.), when experiments were performed in high pressure autoclave reactor at 90 ̊C, 0.25 M and 24 hrs reaction time. Blue line represents yield ratio of 1,4- to 1,5 cycloadduct.

4.3.2 Exploring the pressure, temperature and concentration windows in the micro capillary reactorBased on the availability of equipment, namely pumps and BPRs, maximum pressure

reachable in our home-built micro flow setup was 400 bar. Experience with batch

experiments showed that reaction kinetics at relatively low temperature is relatively slow.

Although possible, long reaction times in flow are not desirable. Thus, the uncatalyzed

azide-alkyne cycloaddition under interest was studied in the HPHT micro capillary flow

setup with a shorter reaction time than in batch. Residence time of 30 min was allowed

for the reaction at 90˚C and pressures up to 400 bar.

Under atmospheric pressure and same diluted, 0.25 M, conditions as in batch 48% yield of

1,4-cycloadduct was obtained in 30 min. Increase of pressure to 400 bar resulted in 58%

yield of desired regioisomer. No significant change in product distribution was observed

with increasing pressure, the product distribution raised from 3.5 at 1 bar to 3.6 at 400

bar. Increasing temperature under same diluted conditions as in batch showed that the

highest yield of 80% of 1,4-cycloadduct under the given the conditions could be obtained.

Next, we investigated effect of residence time at various reaction temperatures under

diluted conditions of 0.25 M. Figure 4 shows that time effect is diminished at temperatures

higher than 140 ̊C, resulting in a full conversion of 2,6-difluorobenzyl azide at all

0

20

40

60

80

100

0 100 200 300 400

Yiel

d of

1,4

cyc

load

duct

, (%

)

Pressure (bar)

1.0 M0.5 M

0.25 M

Figure 4 Temperature effect on the yield of 1,4-cycloadduct in [3+2] cycloaddition, when performed at 400 bar, 0.25 M and allowed 5, 10, 20 and 30 min as residence time.

84


85

4

investigated residence times. Decrease in the yield is observed at higher temperatures

than 140 ̊C and longer residence time than 10 min. This observation can be explained by

decomposition of 2,6-difluorobenzyl azide at high temperatures and longer contact times,

similar to the case in our previous study3b.

Reactions when performed at higher concentrations, provided with better mixing and

higher temperatures proved to proceed in a safer manner in flow than in batch with

an additional benefit of having faster kinetics. In this case, however, temperature was

kept constant at 90 ̊C for the sake of comparison with previously performed batch

experiments and reducing the competing pressure and temperature effects. Increasing

the concentration of 2,6-difluorobenzyl azide to 0.5 and 1.0 M was possible and speeded

up the kinetics as shown in Figure 5. Throughout the whole pressure range investigated

a yield increase of approximately 10% was found for each value of concentration. The

highest yield of 81% of desired 1,4-cycloadduct was obtained at 1.0 M, 90 ̊C and 400 bar

given 30 min residence time.

4.3.3 Exploring the concentration, pressure and temperature windows in the micro capillary reactor with catalystAs mentioned in the introduction section, combination of catalysis and high pressure

may have a synergistic effect.[12] When catalyzed by copper the reaction of azide-alkyne

0%

20%

40%

60%

80%

100%

60 100 140 180

Yiel

dof

1,4

cyc

load

duct

, (%

)

Temperature, (̊C)

5 min

10 min

20 min

30 min

Figure 5. Pressure and concentration effect on yield of 1,4-substituted 1,2,3-triazole at different concentrations 0.25 M, 0.5 M and 1.0 M when experiments performed within the pressure range of 1-400 bar in high pressure flow setup at 90 ̊C and 30 min reaction time.

cycloaddition constitutes to a class of Click reactions. According to mechanistic studies,

copper directs regioselectivity of cycloaddition. The directing power of copper varies for

different combination of catalyst and reactants based on their individual reactivity.[15]

Thus, 100% regioselective cycloaddition is not always guaranteed. In order to check the

merit of reaction activation and regioselectivity control by a catalyst we moved on to

investigate copper catalyst in HPHT micro capillary based flow system. We investigated

copper catalyzed cycloaddition by using 1.0 mol% of homogeneous copper catalyst-

(1,10-phenanthroline) bis(triphenylphosphine) copper (I) nitrate dichloromethane adduct.

The choice of the catalyst was based on its recently exhibited superior performance59, that

resulted in 96% yield of 1,4-cycloadduct in 3 min reaction time when phenyl acetylene

and phenyl azide reacted under solvent-free conditions at room temperature in batch.

We performed cycloaddition of methyl propiolate with 2,6-difluorobenzyl azide under

diluted batch conditions (0.25 M) and 90 ̊C. Only 5% of 1,4-cycloadduct was obtained after

90 min reaction time, while no 1,5-cycloadduct was observed. In order to perform the

reaction in flow, an extra pump was used to introduce the Cu-catalyst into the stream of

2,6-difluorobenzyl azide before mixing with methyl propiolate. Pre-mixing was reported

to result in the formation of copper-alkyne aggregated complexes53. Methyl propiolate

was introduced into the mixed stream via an HPIMM mixer (Re= 55), constructed for

high pressure and based on flow lamination/hydrodynamic focusing (Figure 6). Increasing

the concentration to 0.5 M under otherwise same conditions with no pressure applied,

given 1 min as residence time, 7% yield of 1,4-cycloadduct was obtained. We studied

the reaction with increased to 0.5 M azide concentration in the same range of process

Figure 6. High pressure, high temperature (HPHT) micro capillary based flow system with added stream of homogeneous [Cu(phen)(PPh3)2]NO3 catalyst. The setup consists of stainless steel (SS) capillary (500 µm ID, 10 m), three HPLC pumps of Knauer 1000 series, T- and high pressure IMM mixers, heating and cooling Lauda oil baths, sample loop connected to a 6-port Vici Valco valve, and one Bronkhorst back-pressure regulator (BPR).

86


87

4

conditions as in uncatalyzed case. Figure 7 shows rapid decrease in regioselectivity with

increasing temperature, which speeds up the reaction as demonstrated by the increase of

the 1,4-cycloadduct yield. The highest yield obtained was 77% of the desired cycloadduct,

with 1,4:1,5 cycloadduct ratio of 4.2, at 160˚C in 5 min and in the presence of the Cu-

catalyst. A slight increase in yield due to pressure is observed at moderate temperatures

which is less pronounced as found for the uncatalyzed reaction in flow.

0

20

40

60

80

100

0

20

40

60

80

100

50 100 150 200

1,4/

1,5-

cycl

oadd

uct

rati

o

Temperature, (̊C)

400 bar

1 bar

Yield of 1,4-cycloadduct, (%)

Figure 7. Temperature effect on the yield of 1,4-cycloadduct and regioselectivity expressed as a ratio of 1,4- to 1,5-cycloadduct is monitored in the presence of [Cu(phen)(PPh3)2]NO3 catalyst at various temperature (60- 200 ̊C) and pressure (1-400 bar) values.

400

800

1200

1600

2000

50 100 150 200

Temperature, (°C)

Pres

sure

, (b

ar)

Autoclave

Flow Cat flow

24 hr

30 min 5 min

84 %

81 %

77 %

Figure 8. Process windows for the yields of desired regioisomer above 70% in high pressure autoclave reactor, uncatalyzed and catalyzed processes in flow reactor.

Figure 8 gives an overview about all the findings reported above with the boundaries of

the process windows resulting in higher than 70% yield of the desired 1,4-cycloadduct.

The pressure-temperature window for the uncatalyzed autoclave reactor is large showing

the good capability of this reactor and in general the wide flexibility which are commonly

acknowledged for batch operation. It allows operation in pressure regions, which the flow

reactor cannot exploit at this point of time due to technological limitations. Moreover,

a change in regioselectivity in favor of desired regioisomer can be achieved. Yet, long

processing times are needed which results in a drop in productivity, meaning also more

energy per given unit manufactured product. Here the much shorter processing times of

the uncatalyzed flow reactor provide good alternative, and even more if safety under high

pressure and energy minimization is an issue. Here activation by pressure is somewhat

helpful, yet the big activation boost comes from the temperature and concentration

flexibility. Yet, the small pressure effect is not intrinsic to flow, but the pressure range

utilizable simply is lower at this time of technological development for flow reactors.

Figure 8 also shows the process window coordinates for the best yields obtained for the

three processing types.

The non-catalytic presents a case where both modes are used and this presents the core

information of this chapter. Additional difference is the time needed for the obtained

yield, a 48-fold reduction in residence time is given for the non-catalyzed flow reactor as

compared to the uncatalyzed autoclave reactor.

Catalyst use is justified at lower temperatures and longer residence times, since it affords

only 1,4-cycloadduct. Regioselectivity reduces rapidly when temperature increases,

being highest at 25 ̊C when no formation of 1,5-cycloadduct is detected, decreasing the

ratio of 1,4:1,5 to 64 at 60 ̊C, and rapidly falling to 4.6 at 120 ̊C. Overview of the best

conditions for flow in terms of the yield of 1,4-cycloadduct, at 160 ̊C and 400 bar given

5 min reaction time, shows that merit of using catalyst is only 1% more of a desired

product. Use of catalyst adds an additional downstream operation within the production

process. Separation of toxic metal is required prior to the final stage of API production.

It is evident that the flow operation demands the use of higher temperatures to achieve

its best yields.

Although our primary synthetic target was Rufinamide precursor, we applied the best

conditions of 140 ̊C and 400 bar to other substrates. The results shown in Table 1 imply

that the more electron-deficient the dipolarophile is, the higher is its activity towards

1,3-dipolar cycloaddition. The opposite is true for the dipole, azidobenzene is the most

active among investigated azides, due to the higher conjugation and higher electron

density over the azide dipole.

88


89

4

4.4. CONCLUSION

Novel Process Windows principles were applied to the 1,3-dipolar cycloaddition to yield the

Rufinamide precursor. The activation means of the reaction and regioselectivity towards

the 1,4-cycloadduct, the desired precursor, served as two foci of the study. Concerning

both aspects, merits of high pressure, high temperature, high concentration and catalyst

Table 1. 1,2,3-triazoles synthesized under catalyst-free conditions in micro capillary flow reactor in 30 min residence time at 140 ̊C and either 10 or 400 bar.

Entry Azide Alkyne Product(1,4-regioisomer)

Total yield (%) 10 bar 400 bar

1,4/1,5 10 bar 400 bar

1 N3

O

ONN N

O

O

99 99 2.5 2.5

2 N3 NN N

50 62 1.2 1.4

3 N3 OH NN N

OH 46 59 1.1 0.9

4 N3

O

O

NN N

O

O

99 99 3.6 3.5

5 N3

NN N

58 65 1.1 1.3

6 N3 OHNN N

OH

72 78 1.5 1.4

7 N3

NNN

41 49 2.1 2.1

8 N3

O

O

NN N

O

O

99 99 3.2 2.1

9 N3

NN N

33 42 1.4 1.8

10 N3 OHNN N

OH

23 24 1.5 1.5

[a] Reaction conditions: azide (1 eq, 0.25 M), alkyne (ene) (2 eq, 0.5 M) in n-methyl pyrrolidone, at 140 ̊C given 30 min residence time. [b] Yields were calculated by using 1H NMR spectroscopy with the use of 1,3,5-trimethoxy benzene as an internal standard.

were compared (4 of 6 NPW means as given in ref. 3). Moreover, the comparison was made

in between home-built high pressure autoclave and high pressure and high temperature

micro capillary flow reactors The reaction was on the order of several days when carried

out in a round bottom flask under atmospheric conditions, resulting in 60% yield after 6

days.

When the high-pressure autoclave reactor was used, speed-up of the reaction kinetics

and improvement in regioselectivity were observed. Yield of the desired precursor rose

to 84% in 24 hrs, and resulted in a 1,4:1,5-cycloadduct ratio of 6.3 at 1800 bar with no

catalyst present. Increasing concentration could speed up the reaction further, however

due to the possible decomposition of the azide and subsequent pressure build-up we did

not pursue this option which also would never be viable as industry’s production way. Use

of such a reactor on a production scale even within given limits is still questionable, due

to the fabrication costs and safety issues.

In turn, the flow reactor can be scaled-up by external numbering-up and smart scale-

out (small widening of characteristic dimensions) without a major loss of performance.

Operation in a capillary flow reactor allowed to increase temperatures up to the

superheated range, make use of pressure up to 400 bar, and increase the reactants’

concentration values. The combined action of these three activation modes resulted in

a yield of 81% of the desired 1,4-cycloadduct in 30 min residence time. Copper-based

catalysis was investigated as an additional activation mode for the cycloaddition in the

flow reactor. 76-77% yield at various individual temperature-pressure combinations, given 5

min residence time, was obtained. Thus, although not optimized, the investigated process

window showed that 5 times acceleration is possible when compared with noncatalyzed

flow operation. However, formation of the undesired 1,5-cycloadduct was present even in

the presence of the catalyst.

To give a differentiated, comprehensive picture, process windows maps of favorable

operation (pure 1,4 isomeric-yield >70%) were given. The best conditions were applied

to a wider selection of azides and alkynes. The results obtained in the current study

also have given some insight on the often claimed easiness of pressure operation with

micro flow reactors. Indeed, data collection process was much faster with the micro flow

setup due to the faster heating and shorter reaction times, and easier sampling due to

the installed sample loop. In addition, the much larger volume used for the autoclave

reactor restricted exploration of temperatures above a certain limit. Thus, the range

of information, as related to the expansion of the process windows, was better for the

micro flow reactor. Moreover, combination of catalysis and harsh conditions was slightly

advantageous when compared to the uncatalyzed process. Presence of catalyst requires

90


91

4

extra downstream operation, thus overall benefit is higher when no catalyst is used with

harsh conditions applied.

4.5. OUTLOOK

The design of continuous micro flow-based processes enables either new types of process

integration or leads to process simplification. Maximum impact is here typically gained

for entirely new chemical transformations, which are only realizable in flow. Those when

combined in a sequence, as in multi-step synthesis, aim at compactness and therefore

bring a special attention to slow reactions that need the highest activation. In our case,

the cycloaddition to synthesize 1,2,3-triazole precursor of Rufinamide constitutes to such

a reaction within its multi-step synthesis.

A key to process-design intensification is high space-time yield, allowing to make reactors

very compact which facilitates or even enables system integration likewise in electronics

industry. Based on this motivation we calculated space-time yields for the three types of

operation (and two types of reactors) investigated (see Table 2). In reactor engineering,

comparison of STYs is only possible at same conversion. Yet, we were after order-of-

magnitude changes and had to accept the fact that conversions were somewhat different.

In addition, there is a never-completed discussion among the micro reactor engineering

community, if the inner or outer volume of the reactor is to be taken as reference.

We used the inner volume (i.e. the fluid reaction volume) as reference to calculate the

space-time yield. It is evident in both projections that the flow reactor is more productive

than the autoclave reactor, which needs a substantial outer mass to enable its high-

pressure operation. Yet, it can compensate by larger volume partly what is lost by

lower activation. Naturally, our data are lab-based and production reactors will behave

somewhat different; yet this will not change the overall message. The compactness of

the flow reactor resembles Ramshaw’s first definition of process intensification, ‘to shrink

down the plant’60.

Table 2. Space-time yield for the three types of operation (and two types of reactors).

Operation type Space-time yield (mol L-1 h-1)

Non-catalyzed autoclave reactor 0.0084

Non-catalyzed flow reactor 4.56

Catalyzed flow reactor 4.62

ACKNOWLEDGEMENTS

We would like to thank Flowid B.V. for lending us HPIMM mixer to be used in our

investigations, Jan Volkers and Guus Honcoop from Knauer for their help with setting up

the pumps, and Erik van Herk from Eindhoven University of Technology for helping with

building the HPHT flow setup. Funding by the Advanced European Research Council Grant

“Novel Process Windows – Boosted Micro Process Technology” (Grant number: 267443) is

kindly acknowledged.

92


93

4

REFERENCES

1. Chorkendorff, I.; Niemantsverdriet, J. W. In Concepts of Modern Catalysis and Kinetics; Wiley-VCH Verlag GmbH & Co. KGaA, 2005; pp 23–78.

2. Matsumoto, K.; Hamana, H.; Iida, H. ChemInform 2005, 36 (48).

3. Bini, R.; Ceppatelli, M.; Citroni, M.; Schettino, V. Chem. Phys. 2012, 398, 262–268.



6. Razzaq, T.; Glasnov, T. N.; Kappe, C. O. European J. Org. Chem. 2009, 2009 (9), 1321–1325.




10. Noël, T.; Kuhn, S.; Musacchio, A. J.; Jensen, K. F.; Buchwald, S. L. Angew. Chem. Int. Ed. Engl. 2011, 50 (26), 5943–5946.

11. Borukhova, S.; Noël, T.; Metten, B.; de Vos, E.; Hessel, V. ChemSusChem 2013, 6 (12), 2220–2225.

12. Kobayashi, H.; Driessen, B.; van Osch, D. J. G. P.; Talla, A.; Ookawara, S.; Noël, T.; Hessel, V. Tetrahedron 2013, 69 (14), 2885–2890.

13. Illg, T.; Hessel, V.; Löb, P.; Schouten, J. C. Chem. Eng. J. 2011, 167 (2-3), 504–509.

14. Illg, T.; Löb, P.; Hessel, V. Bioorg. Med. Chem. 2010, 18 (11), 3707–3719.

15. Kockmann, N.; Gottsponer, M.; Zimmermann, B.; Roberge, D. M. Chem. – A Eur. J. 2008, 14 (25), 7470–7477.

16. Benito-Lopez, F.; Tiggelaar, R. M.; Salbut, K.; Huskens, J.; Egberink, R. J. M.; Reinhoudt, D. N.; Gardeniers, H. J. G. E.; Verboom, W. Lab Chip 2007, 7 (10), 1345–1351.

17. Verboom, W. Chem. Eng. Technol. 2009, 32 (11), 1695–1701.

18. Zhao, Y.; Chen, G.; Ye, C.; Yuan, Q. Chem. Eng. Sci. 2013, 87, 122–132.

19. Leclerc, A.; Alamé, M.; Schweich, D.; Pouteau, P.; Delattre, C.; de Bellefon, C. Lab Chip 2008, 8 (5), 814–817.

20. Trachsel, F.; Hutter, C.; von Rohr, P. R. Chem. Eng. J. 2008, 135, Suppl (0), S309–S316.

21. Kobayashi, J.; Mori, Y.; Kobayashi, S. Chem. Commun. (Camb). 2005, No. 20, 2567–2568.

22. Grochala, W.; Hoffmann, R.; Feng, J.; Ashcroft, N. W. Angew. Chem. Int. Ed. Engl. 2007, 46 (20), 3620–3642.

23. Sharma, A.; Scott, J. H.; Cody, G. D.; Fogel, M. L.; Hazen, R. M.; Hemley, R. J.; Huntress, W. T. Science 2002, 295 (5559), 1514–1516.

24. Rulev, A. Y.; Kotsuki, H.; Maddaluno, J. Green Chem. 2012, 14 (2), 503.

25. Keybl, J.; Jensen, K. F. 2011, 11013–11022.

26. Tiggelaar, R. M.; Benito-López, F.; Hermes, D. C.; Rathgen, H.; Egberink, R. J. M.; Mugele, F. G.; Reinhoudt, D. N.; van den Berg, A.; Verboom, W.; Gardeniers, H. J. G. E. Chem. Eng. J. 2007, 131 (1-3), 163–170.

27. Lorber, N.; Sarrazin, F.; Guillot, P.; Panizza, P.; Colin, A.; Pavageau, B.; Hany, C.; Maestro, P.; Marre, S.; Delclos, T.; Aymonier, C.; Subra, P.; Prat, L.; Gourdon, C.; Mignard, E. Lab Chip 2011, 11 (5), 779–787.

28. Schettino, V.; Bini, R. Chem. Soc. Rev. 2007, 36 (6), 869–880.

29. Rao, H. S. P.; Murali, R.; Taticchi, A.; Scheeren, H. W. 2001, 2869–2876.

30. Pie, L. De. 1999, c, 619–625.

31. Shi, Z.; Liang, W.; Luo, J.; Huang, S.; Polishak, B. M.; Li, X.; Younkin, T. R.; Block, B. a.; Jen, A. K.-Y. Chem. Mater. 2010, 22 (19), 5601–5608.

32. Martin, R. E.; Morawitz, F.; Kuratli, C.; Alker, A. M.; Alanine, A. I. European J. Org. Chem. 2012, 2012 (1), 47–52.

33. Minuti, L.; Temperini, A.; Ballerini, E. J. Org. Chem. 2012, 77 (18), 7923–7931.

34. Chen, H.; Ni, B.-B.; Gao, F.; Ma, Y. Green Chem. 2012, 14 (10), 2703.

35. Ku, E.; Sedelmeier, G.; Dyson, P. J. 2007, No. February, 109–114.

36. Antony Chrétien, Isabelle Chataigner, S. R. P. Tetrahedron 2005, 61 (33), 7907–7915.

37. Misumi, Y.; Matsumoto, K. Angew. Chem. Int. Ed. Engl. 2002, 41 (6), 1031–1033.

38. Kwiatkowski, P.; Asztemborska, M.; Jurczak, J. Tetrahedron: Asymmetry 2004, 15 (20), 3189–3194.

39. Matsuo, J.; Sasaki, S.; Tanaka, H.; Ishibashi, H. J. Am. Chem. Soc. 2008, 130 (35), 11600–11601.

40. Illg, T.; Hessel, V.; Löb, P.; Schouten, J. C. ChemSusChem 2011, 4 (3), 392–398.

41. Tilstam, U.; Defrance, T.; Giard, T.; Johnson, M. D. 2009, 13 (2), 321–323.

42. Huisgen, R. Angew. Chemie Int. Ed. English 1963, 2 (10), 565–598.

43. Anderson, G. T.; Henry, J. R.; Weinreb, S. M. J. Org. Chem. 1991, 56 (24), 6946–6948.

44. Melai, V.; Brillante, A.; Zanirato, P. J. Chem. Soc. Perkin Trans. 2 1998, No. 11, 2447–2450.

45. Fan, J.-C.; Liang, J.; Wang, Y.; Shang, Z.-C. J. Mol. Struct. THEOCHEM 2007, 821 (1-3), 145–152.

46. Elamari, H.; Jlalia, I.; Louet, C.; Herscovici, J.; Meganem, F.; Girard, C. Tetrahedron: Asymmetry 2010, 21 (9-10), 1179–1183.

47. Kolb, H. C.; Sharpless, K. B. Drug Discov. Today 2003, 8 (24), 1128–1137.

48. Pachón, L. D.; van Maarseveen, J. H.; Rothenberg, G. Adv. Synth. Catal. 2005, 347 (6), 811–815.

49. Moses, J. E.; Moorhouse, A. D. Chem. Soc. Rev. 2007, 36 (8), 1249–1262.

50. Wang, J.; Sui, G.; Mocharla, V. P.; Lin, R. J.; Phelps, M. E.; Kolb, H. C.; Tseng, H.-R. Angew. Chem. Int. Ed. Engl. 2006, 45 (32), 5276–5281.

51. Smith, C. D.; Baxendale, I. R.; Lanners, S.; Hayward, J. J.; Smith, S. C.; Ley, S. V. Org. Biomol. Chem. 2007, 5 (10), 1559–1561.

52. Bogdan, A. R.; James, K. Chemistry 2010, 16 (48), 14506–14512.

53. Hein, J. E.; Fokin, V. V. Chem. Soc. Rev. 2010, 39 (4), 1302–1315.

54. Varas, A. C.; Noël, T.; Wang, Q.; Hessel, V. ChemSusChem 2012, 5 (9), 1703–1707.

55. Danence, L. J. T.; Gao, Y.; Li, M.; Huang, Y.; Wang, J. Chemistry 2011, 17 (13), 3584–3587.

56. Baumann, M.; Baxendale, I. R.; Ley, S. V; Nikbin, N. Beilstein J. Org. Chem. 2011, 7, 442–495.

57. Lemmon, M. E.; Kossoff, E. H. Curr. Treat. Options Neurol. 2013, 15 (4), 519–528.


59. Wang, D.; Zhao, M.; Liu, X.; Chen, Y.; Li, N.; Chen, B. Org. Biomol. Chem. 2012, 10 (2), 229–231.

60. Rosenfeld, C.; Serra, C.; Brochon, C.; Hessel, V.; Hadziioannou, G. Chem. Eng. J. 2008, 135, Suppl, S242–S246.

CHAPTER 5Solvent- and catalyst-free Huisgen cycloaddition towards Rufinamide


Borukhova, S., Noël, T, Metten, B, de Vos, E., Hessel, V. (2013). Solvent- and catalyst-free Huisgen cycloaddition towards Rufinamide in flow with decision on a greener and less expensive dipolarophile. ChemSusChem, 6(12), 2220-2225.

96

CHAPTER 5 Solvent- and catalyst-free Huisgen cycloaddition towards Rufinamide

97

5

ABSTRACT

Novel Process Windows promotes chemical intensification via operating at high

temperature, pressure and concentration. Herein, we describe the development of a

continuous-flow process for the formation of the 1,2,3-triazole precursor Rufinamide.

The precursor is synthesized via a 1,3-dipolar Huisgen cycloaddition of 2,6-difluoro-

benzylazide and greener, less expensive dipolarophile under catalyst-free and solvent-

free reaction conditions. It was found that the use of elevated temperatures boosted

the reaction rate significantly, allowing the synthesis of the 1,2,3-triazole precursor in

only 10 minutes residence time as opposed to >24 hr required in batch. Clogging was

efficiently avoided by working at reaction temperatures above the melting point of the

target compound. In addition, by introduction of recrystallization solvent purification was

realized upon product collection.

5.1. INTRODUCTION

The 1,2,3-triazole moiety constitutes an interesting class of heterocycles that displays

many useful biological activities, such as anti-convulsant1, anti-HIV2, anti-allergic3, and

anti-microbial activity against Gram positive bacteria4. One of the best-selling five-

membered heterocyclic pharmaceuticals of recent years is an antiepileptic drug which

contains a 1,2,3-triazole moiety, called Rufinamide 15. It prevents seizures by deactivating

sodium ion channels, and is currently used in the treatment of Lennox-Gastaut syndrome6,7.

Rufinamide is manufactured by Eisai under commercial names of Inovelon and Banzel.

Various industrial synthetic routes have been disclosed in the patent literature over the

past three decades8–10. The formation of 1,2,3-triazole precursor via a 1,3-dipolar Huisgen

cycloaddition of 2,6-difluorobenzylazide (2) with an appropriate dipolarophile is a key

step in the synthesis of Rufinamide (Scheme 1). Table 1 lists the various commercially

available dipolarophiles, which have been utilized so far. 2-chloroacrylonitrile results in a

regioselective addition yielding the desired 1,4-cycloadduct in good yield (Table 1, Entry

1)8. However, it is a highly toxic and flammable substance which, in combination with the

explosive nature of organic azides, causes significant safety issues for scale up. Propiolic

acid is less toxic and is, due to the higher electronic deficiency, more reactive (Table 1,

Entry 2)9.

Nevertheless, a transition metal catalyst is required to induce high regioselectivity. In

addition, the benefits of catalyst use becomes controversial when high purity standards for

active pharmaceutical ingredients (APIs) are taken into consideration11. Similar arguments

Scheme 1. General synthetic sequence towards Rufinamide 1.

Scheme 2. Proposed mechanism for 1,3-dipolar cycloaddition accompanied by elimination of methanol.

98


99

5

hold for methyl propiolate (Table 1, Entry 3)10. Apart from toxicity and regioselectivity

issues, the high cost of the starting material is a hurdle for selection of this synthetic route

for production scale. Recently, Mudd et al. reported a greener route utilizing a non-toxic

and inexpensive (E)-methyl 3-methoxyacrylate (E-MMA) as a dipolarophile (Table 1, Entry

4)12. 100% regioselectivity towards desired 1,4-cycloadduct is afforded due to the presence

of leaving methoxy group. Scheme 2 shows the cyloaddition accompanied by elimination to

regain aromaticity of 1,2,3-triazole ring. Thus, no catalyst, nor its subsequent separation

are needed to favor cycloadition to yield desired regioisomer. However, this route requires

28 hours of reaction time for a completion when performed at 135 C̊ under solvent-free

conditions. Apart from the long operating time issue, the reaction temperature is very high

for a conventional batch operation in the pharmaceutical industry. This is especially true

when the exothermic nature of the reaction is taken into account. Moreover, solvent-free

large scale batch processing of organic azides at high temperature is not recommendable

in view of the limited stability of these compounds. This might lead to a runaway

decomposition with gas formation and concomitant pressure build-up.

Extended reaction times can be shortened significantly by intensifying the intrinsic kinetics

of the reaction in a microreactor environment13. Such intensification can be achieved by so-

called Novel Process Windows (NPW), which addresses this issue via a chemical intensification

through elevated temperatures, pressures and increased reaction concentration14,15. In

addition, the NPW concept introduces a process-design intensification stage, where the

goal is to simplify the process at its initial design by integration of several stages within the

Table 1. Overview and comparison of patented (Entry 1-3) and alternative ( Entry 4) syntheses with respect to the choice of dipolarophile used in 1,3-dipolar Huisgen cycloaddition to afford 1,2,3-triazole, precursor of Rufinamide.

Entry Dipolarophile -R Toxicity (NFPA)

Cost* ($/mol)

Time (h)

Yield (%) T (ºC) Equivalents Solvent/

Additives

1[7a]Cl

CN-CN 4 12 24 72 80 1.5 Neat

2[7b]

O

OH -COOH 3 13 2 80 25 1Water-tBuOH/ Ascorbic acid/

CuSO4

3[7c]

O

O -COOMe 2 45 5 48 65 1 Water

4[9]

O O

O-COOMe 0 2 28 85 135 1.2 Neat

*2-chloroacrylonotrile- 5 kg for $680 (AOKBIO), propiolic acid- 10 kg for $1840 (Beta Pharma Scientific), methyl propiolate- 1000 kg for $530/kg (D-L Chiral Chemicals), (E)-methyl 3-methoxyacrylate- 10kg for $160 (Apichemical (Shanghai))

process16,17. Continuous micro flow processing is a very desirable feature in pharmaceutical

industry since it allows a high degree of control over reaction parameters, a higher safety,

a reduced manual handling, a flexibility of production volume, an easier reproducibility, a

possibility of reaction telescoping, and a potential for in-line purification18–20. Reducing the

size of equipment to a micro scale allows to increase heat- and mass-transfer and to safely

operate superheated, runaway and other harsh process conditions21,22.

Current chapter presents a collaboration between academia and industry to realize a new

process that embraces green chemistry and green engineering principles. It starts off with

a combined cost-sustainability guided choice of the dipolarophile and proceeds with micro

flow-process intensification to accelerate the reaction to make the process productive

and safe for industrial implementation. While cost arguments are considered by the raw

material price and the potential for a patent-free process, environmental arguments

relate to toxicity, raw material and solvent use, in addition to generated waste.

Herein, we present an alternative, intensified method to produce 1,2,3-triazole ester,

a key intermediate for the production of Rufinamide 1. The novelty of the proposed

synthesis lays in its continuous nature, the decrease in the reaction time from hours to

minutes, and the utilization of a greener and cheaper starting compound under solvent-

and catalyst-free conditions. In addition, purification is integrated within the reaction

step, demonstrating an application of process-design intensification.

5.2. RESULTS AND DISCUSSION

We commenced our investigations by verifying the batch procedure described by Mudd et

al.12. The original procedure required two intermediate additions of E-MMA, at 2 and 18

hours after start, to reach full conversion. The desired product is reported to be purified

by recrystallization from methanol (MeOH)12. In our hands, a single addition of E-MMA

(3) resulted in 43% yield after one hour and 78% isolated yield after 12 hours of reaction

time, as shown in Figure 1. We observed that upon cooling the reaction mixture to room

temperature the target compound crystallized rapidly.

To avoid clogging issues within the micro capillary reactor and, consequently, facilitate the

transition from batch to continuous-flow, we followed the reaction progress in batch under

more diluted conditions using N-methyl-2-pyrollidone (NMP) as a highly polar solvent, which

could solubilize all the reactants and products. However, poor conversions were observed

under these diluted reaction conditions; only 13 and 16% yield was reached within the first

hour of reaction with 0.5 and 1 M concentration of 2,6-difluorobenzyl azide (2), respectively.

100


101

5

Based on our observations in batch, we have built a micro capillary assembly as shown in

Figure 2. One of the very first requirements for continuous-flow operations is to ensure

a complete homogeneity of the reaction mixtures19,23,24. Due to the rapid decrease in

reaction rate under diluted homogeneous reaction conditions as observed in batch, the

use of solvent-free reaction conditions was preferred to attain a higher reaction rate.

However, it was observed that target compound 4 was sparingly soluble under these neat

reaction conditions, therefore, requiring special handling when 100% conversion is desired.

Nevertheless, we envisioned that continuous-flow processing was feasible when the entire

reactor coil (Stainless Steel capillary tubing, 750 µm ID, 4.7 m length) was maintained

at a temperature above the melting point of the product (mp 4 = 136-137 C̊). Moreover,

purification could be integrated by introducing a stream of acetonitrile (ACN) (1/10 v/v)

or MeOH (1/15 v/v), which yielded crystallized product in a collection tank upon cooling.

Analytical pure compound can be collected by simple filtration. Notably, it was found that

the mixing tee, used to merge the reaction and ACN or MeOH, needed to be heated to the

reaction temperature to avoid uncontrollable crystallizations. The combined diluent and

reaction stream was subsequently cooled down to 60 C̊ prior to passing it through a back

pressure regulator (BPR, 1000 psi). This BPR ensured that a steady continuous flow was

maintained within the capillary microreactor and prevented boiling of E-MMA (bp = 155 ̊C).

One of the key parameters to accelerate the reaction rate is the use of elevated temperatures.

Microreactor technology enables a safe and reliable use of such harsh reaction conditions,

Figure 1. Progress of the [3+2]-cycloaddition of 2,6-difluorobenzyl azide 2 with (E)-methyl 3-methoxyacrylate 3 at 135 ̊C under solvent-free or diluted conditions in batch.

0

20

40

60

80

100

0 2 4 6 8 10 12

HPL

C Yi

eld

[%]

Time [hour]

neat

1.0 M

0.5 M

which are otherwise difficult to attain on a batch scale25–29. Figure 3 depicts the temperature

dependence on the catalyst- and solvent-free Huisgen cycloaddition to yield 4 in continuous-

flow with a residence time of one hour. At 140 C̊, 56% yield could be obtained within one

hour. The optimal reaction temperature was 200 C̊ which resulted in the formation of 4 in

82% yield. Further increase of the reaction temperature resulted in a decrease in yield.

Although one hour residence time is still acceptable for basic investigations in continuous-

flow chemistry, it is far from ideal from a production standpoint. Thus, we decided to study

the reaction behaviour in more detail via Differential Scanning Calorimetry (DSC). As an

additional advantage, solvent-free conditions afford more accurate modelling due to the

absence of solvent-reactant interactions. Based on the energy evolution, DSC predicted

that while one hour is required to obtain full conversion at 200 C̊, only 5 min might be

necessary to afford the same result at 240 ̊C (Figure 4). It should be noted that the observed

energy evolution is not only observed for the exothermic [3+2] cycloaddition, but can also

be due to the decomposition of starting materials.

It is generally known that azides are heat sensitive and decompose rapidly at elevated

temperatures with concomitant evolution of nitrogen gas30.

In order to study the decomposition rate of 2,6-difluorobenzyl azide (2), pure 2 was

introduced into the micro reactor assembly and subjected to elevated reaction

Figure 2. Schematic representation of the micro capillary assembly for the solvent- and catalyst-free Huisgen cycloaddition of 2,6-difluorobenzyl azide 2 with (E)-methyl 3-methoxyacrylate 3. More details are given within the supporting information of the publication the current chapter is based on. Picture shows the precipitation of product 4. (BPR = back pressure regulator.)

102


103

5

Figure 4. Differential Scanning Calorimetry predictions for the [3+2]-cycloaddition of 2,6-difluorobenzyl azide 2 with (E)-methyl 3-methoxyacrylate 3 at different reaction temperatures under solvent-free reaction conditions.

Figure 3. Temperature influence on the [3+2]-cycloaddition of 2,6-difluorobenzyl azide 2 with (E)-methyl 3-methoxyacrylate 3 under solvent-free and catalyst-free reaction conditions in continuous flow.

0

20

40

60

80

100

120 150 180 210 240

HPL

C Yi

eld

[%]

Temperature [̊C]

temperatures (Figure 5). At a reaction temperature of 200 ̊C, only 4% of the azide was

decomposed after a residence time of 10 minutes. However, at 230 ̊C, already 34% of 2

was decomposed within 10 minutes residence time as was evident from the increased

nitrogen gas evolution. Analysis of the reaction sample with GC-MS, revealed the

formation of 2,6-difluorotoluene and 2,6-difluorobenzyl amine together with some other

minor decomposition products. These results demonstrate the existence of a sensitive

balance between decomposition rate of the reagents and acceleration of reaction rate to

form the desired compound at elevated reaction temperatures. In order to fine-tune the

operating conditions, we investigated the influence of the residence time on the yield of

4 at the temperatures ranging from 200 to 240 ̊C, and residence time ranging from 2.5

min to 15 min. HPLC results suggested that 210 ̊C was the optimal reaction temperature.

Isolated yield experiments on a large scale (20 mmol) revealed that the highest yield was

obtained after a residence time of 10 minutes, resulting in 83% yield (4.2 g product, 30

min collection time) in the first crop (Table 2, Entry 2). A second crop obtained from the

mother liquor increased the total yield to 86%.

5.3. CONCLUSION

We have investigated the potential of the Novel Process Windows concept on an industrially

relevant example, i.e. the synthesis of a key intermediate for the production of Rufinamide.

A continuous-flow process was developed for the formation of the 1,2,3-triazole precursor

via a 1,3-dipolar Huisgen cycloaddition of 2,6-difluoro-benzylazide and (E)-methyl

3-methoxyacrylate under catalyst-free and solvent-free reaction conditions. Moreover,

we have made a switch to a greener and less expensive dipolarophile possible, while

adhering to green chemistry and green engineering principles. It was found that the use

of elevated temperatures boosted the reaction rate significantly, allowing the synthesis

of the 1,2,3-triazole precursor in only 10 minutes residence time. Clogging was efficiently

avoided by working at reaction temperatures above the melting point of the target

0

20

40

60

80

100

0 2 4 6 8 10 12

2,6-

difl

uoro

benz

yl a

zide

, re

mai

ning

[%

]

Residence Time [min]

230 ⁰C

A B CD

E

F

200 ⁰C

Figure 5. Above: Decomposition rate of 2,6-difluorobenzyl azide 2 at different reaction temperatures in continuous-flow. Below: Color change observed upon decomposition of 2. The numbering on the vials is matched with the one in the above graph.

104


105

5

compound. In addition, by introduction of recrystallization solvent purification was

realized upon product collection. Finally, it was shown that continuous-flow processing,

as shown herein, constitutes an environmentally benign alternative for the currently

employed procedures in the industry.

ACKNOWLEDGEMENTS

We would like to thank Peter Bracke from OmniChem for his contribution with DSC study.

Funding by the Advanced European Research Council Grant “Novel Process Windows –

Boosted Micro Process Technology” (Grant number: 267443) is kindly acknowledged. T.

Noël would like to acknowledge financial support from the Dutch Science Foundation for a

VENI Grant (No 12464) and from the European Union for a Marie Curie CIG Grant.

REFERENCES

1. Kelley, J. L.; Koble, C. S.; Davis, R. G.; McLean, E. W.; Soroko, F. E.; Cooper, B. R. J. Med. Chem. 1995, 38 (20), 4131–4134.

2. Alvarez, R.; Velazquez, S.; San-Felix, A.; Aquaro, S.; Clercq, E. De; Perno, C.-F.; Karlsson, A.; Balzarini, J.; Camarasa, M. J. J. Med. Chem. 1994, 37 (24), 4185–4194.

3. Buckle, D. R.; Rockell, C. J. M.; Smith, H.; Spicer, B. A. J. Med. Chem. 1986, 29 (11), 2262–2267.

4. Genin, M. J.; Allwine, D. A.; Anderson, D. J.; Barbachyn, M. R.; Emmert, D. E.; Garmon, S. A.; Graber, D. R.; Grega, K. C.; Hester, J. B.; Hutchinson, D. K.; Morris, J.; Reischer, R. J.; Ford, C. W.; Zurenko, G. E.; Hamel, J. C.; Schaadt, R. D.; Stapert, D.; Yagi, B. H. J. Med. Chem. 2000, 43 (5), 953–970.

5. Baumann, M.; Baxendale, I. R.; Ley, S. V; Nikbin, N. Beilstein J. Org. Chem. 2011, 7, 442–495.

6. Lemmon, M. E.; Kossoff, E. H. Curr. Treat. Options Neurol. 2013, 15 (4), 519–528.

7. Suter, M. R.; Kirschmann, G.; Laedermann, C. J.; Abriel, H.; Decosterd, I. Anesthesiology 2013, 118 (1), 160–172.

8. R. Portmann. U.S. Patent No. 6,156,907, 2000.

9. Attolino, E., Colombo, L., Mormino, I., Allegrini, P. U.S. Patent No. 8,198,459 B2, 2012.

10. Kankan, R. N. U.S. Patent No. 8,183,269 B2, 2012.

11. Lewen, N. J. Pharm. Biomed. Anal. 2011, 55 (4), 653–661.

12. Mudd, W. H.; Stevens, E. P. Tetrahedron Lett. 2010, 51 (24), 3229–3231.

13. Yoshida, J.; Kim, H.; Nagaki, A. ChemSusChem 2011, 4 (3), 331–340.


15. Hessel, V.; Vural Gürsel, I.; Wang, Q.; Noël, T.; Lang, J. Chem. Eng. Technol. 2012, 35 (7), 1184–1204.


17. Newman, S. G.; Jensen, K. F. Green Chem. 2013, 15, 1456.


19. Noël, T.; Kuhn, S.; Musacchio, A. J.; Jensen, K. F.; Buchwald, S. L. Angew. Chem. Int. Ed. Engl. 2011, 50 (26), 5943–5946.

20. Browne, D. L.; Wright, S.; Deadman, B. J.; Dunnage, S.; Baxendale, I. R.; Turner, R. M.; Ley, S. V. Rapid Commun. Mass Spectrom. 2012, 26 (17), 1999–2010.



23. Hartman, R. L.; Naber, J. R.; Zaborenko, N.; Buchwald, S. L.; Jensen, K. F. 2010, No. 12, 1347–1357.

24. Kuhn, S.; Noël, T.; Gu, L.; Heider, P. L.; Jensen, K. F. Lab Chip 2011, 11 (15), 2488–2492.

25. Kobayashi, H.; Driessen, B.; van Osch, D. J. G. P.; Talla, A.; Ookawara, S.; Noël, T.; Hessel, V. Tetrahedron 2013, 69 (14), 2885–2890.

26. Varas, A. C.; Noël, T.; Wang, Q.; Hessel, V. ChemSusChem 2012, 5 (9), 1703–1707.

27. Murphy, E. R.; Martinelli, J. R.; Zaborenko, N.; Buchwald, S. L.; Jensen, K. F. Angew. Chemie 2007, 119 (10), 1764–1767.



30. Abramovitch, R. A.; Kyba, E. P. In The Azido Group (1971); John Wiley & Sons, Ltd., 1971; pp 221–329.

CHAPTER 6From alcohol to 1,2,3-triazole via multi-step continuous-flow synthesis of rufinamide precursor


Borukhova, S., Noël, T, Metten, B, de Vos, E., Hessel, V. (2016). From alcohol to 1,2,3-triazole via a multi-step continuous-flow synthesis of a rufinamide precursor. Green Chem, DOI: 10.1039/C6GC01133K

108

CHAPTER 6 From alcohol to 1,2,3-triazole via multi-step continuous-flow

109

6

ABSTRACT

Rufinamide is an antiepileptic drug used to treat the Lennox-Gastaut syndrome. It

comprises a relatively simple molecular structure. Rufinamide can be synthesized from

an organohalide in three steps. Recently we have shown that microreactor flow networks

have better sustainability profiles in terms of life-cycle assessment than the respective

consecutive processing in batch. The analysis was based on results of a single step

conversion from batch to continuous mode. An uninterrupted continuous process towards

rufinamide was developed, starting from an alcohol precursor, which is converted to the

corresponding chloride with hydrogen chloride gas. The chloride is then converted to

the corresponding organoazide that yields the rufinamide precursor via cycloaddition to

the greenest and cheapest dipolarophile available on the market. The current process

demonstrates chemical and process-design intensification aspects encompassed by Novel

Process Windows. Single reaction steps are chemically intensified via a wide range of

conditions available in a microreactor environment. Meanwhile, the connection of

reaction steps and separations results in process-design intensification. With two in-

line separations the process consists of five stages resulting in a total yield of 82% and

productivity of 9g h-1 (11.5 mol h-1 L-1). The process minimizes the isolation and handling of

strong alkylating or energetic intermediates, while minimizing water and organic solvent

consumption.

6.1. INTRODUCTION

In nature, complex molecules are constructed into one entity via series of reactions such as

bio-assembly lines1. According to Fujita, a bio-assembly line represents an energy-driven

conveyer-belt, where a product undergoes a series of modifications on its way to the

end1. Within a cell, single reactions are carried out continuously within the bio-assembly

line under simple operating conditions, such as a single set of solvent type, temperature

and pressure values. The activation of reactions is realised via use of enzymes, while

compartmentalization takes place in supramolecular structures or organelles2. On the

contrary, in fine chemical and active pharmaceutical ingredient (API) synthesis it is

common to build molecules from intermediates that are gathered from various sites in

the world or produced on a site in quantities governed by the capacity of the equipment

available. We believe that by mimicking nature we can increase the sustainability and

maximize the efficiency in chemical processes. A first technological equivalent to a bio-

assembly line is a micro flow reactor network based on reactor cascades operated under

non-classical conditions with integrated separation units3.

Our aim was to design and realize such a flow microreactor network consisting of a multi-

step synthesis in compartmentalized continuous manner in micro flow reactors. The biggest

advantage behind the use of micro flow reactors is the access to a larger playground in

terms of process conditions without compromising safety4. Chemical reactions can be

kinetically accelerated via high temperature, concentration and pressure5–8. In addition,

a combination of those operating parameters may lead to new chemical transformations

and operating regimes that are not accessible in standard batch technology9–12. The

aforementioned operating conditions accompanied with a fresh attitude on process design

with the aim of intensification, comprise Novel Process Windows6,13–15. Substantial numbers

of reactions and processes have been shown to benefit from the concept16–19.

In collaboration with Ajinomoto OmniChem N.V., we aimed at developing a continuous

process to synthesize a rufinamide precursor, while leaving amidation, the last step, to

be handled according to a conventional batch procedure. Rufinamide is a sodium-channel

blocker that is used in treatment of a type of epilepsy called Lennox-Gastaut syndrome.

Herein, we describe a five-stage multi-step process with minimized use of organic solvent

and catalyst. The process consists of gas-liquid, biphasic liquid-liquid and solid forming

reactions and two in-line separation steps.

110


111

6

6.2. BACKGROUND

Rufinamide was developed by Novartis Pharma, AG in 2004 and currently is marketed

by Eisai. The initial synthetic path developed by Novartis relied on constructing the

1,2,3-triazole precursor via 1,3-dipolar cycloaddition of 2,6-difluorobenzyl azide to

2-chloroacrylonitrile (1) as shown in Scheme 120. Alkynes (2)-(4) have been shown to be

readily reactive, however, regioselectivity needs to be directed by a copper catalyst,

otherwise a mixture of 1,4- and 1,5-cycloadduct products is formed21,22. Mudds et al.

proposed a solventless batch sequence using (E)-methyl 3-methoxyacrylate (5) as an

alternative23. The alkene is a significantly safer and cheaper alternative to all previously

mentioned dipolarophiles. In addition, due to the presence of the methoxy leaving

group no catalyst is needed to direct regioselectivity. However, the biggest

drawback is its relatively low reactivity requiring high concentrations. Furthermore

solventless reactions with such potentially energetic compounds are not scalable

in batch.

Following batch procedure developments, continuous-flow processes have been

developed. First, Borukhova et al. showed how the reactivity of (5) can be enhanced

in a micro flow capillary reactor under neat and superheated conditions24. The

reaction time was reduced from >24 hrs to 10 min when the temperature was

increased from 135 ̊C in batch to 210 ̊C in flow. Moreover, continuous synthesis was

combined with crystallization of the rufinamide ester precursor upon collection and

Scheme 1. Retrosynthetic sequence of rufinamide.

cooling of the product. 86% of isolated yield with a productivity of 9 g/h (16.6 mol

h-1 L-1) was obtained. Subsequently, Jamison et al. demonstrated a total synthesis of

rufinamide using 2,6-difluorobenzyl bromide and methyl propiolate (3) as starting

reagents25. 2,6-difluorobenzyl bromide was converted to 2,6-difluorobenzyl azide,

while (3) was converted to the corresponding amide (4) after reacting with aqueous

ammonia. Without any intermediate separation, the intermediates reacted to form

rufinamide in the copper tubing, where the reaction was catalysed by the leached

ionic copper species within the reaction stream. The overall process concentration

was 0.25 M in DMSO and resulted in 92% yield in 11 min with a productivity of 217

mg/h (2.1 mol h-1 L-1) .

Novel Process Windows, apart from concentrating on chemical and process-design

intensification, advocate a holistic approach to the process design to yield green processes

and sustainable manufacturing. Recently, we performed a life-cycle assessment26 and saw

a potential within the synthetic route based on 2,6-difluorobenzyl alcohol and (5) as

key reagents, as shown in Scheme 2. It is proposed that in order to minimize waste,

instead of high-weight halides, like bromides, chlorides should be used in the synthesis of

2,6-difluorobenzyl azide. Another green alternative is the route to the chloride itself, by

using pure hydrogen chloride gas and generates water as the sole by-product. The chloride

is converted into the azide. Finally, due to the immediate consumption of synthesized

2,6-difluorobenzyl azide in cycloaddition to yield triazole (Scheme 2), minimal amounts

are available at any point in time, minimizing the risk associated with detonation of

organic azides in general.

Scheme 2. Proposed alternative reagents to prepare rufinamide precursor.

112


113

6

6.3. STEP 1- CHLORODEHYDROXYLATION

The first step of the 5-stage multi-step synthesis is chlorodehydroxylation of

2,6-difluorobenzyl alcohol to afford 2,6-difluorobenzyl chloride. Recently, we realized

the continuous synthesis of chloroalkanes from alcohols using either concentrated

hydrochloric acid27 (HCl (aq)) or hydrogen chloride28 (HCl) gas as greener alternatives to

more wasteful chlorinating agents. Valuable intermediates were formed with water as a

sole by-product. The intermediates were further continuously utilized in the construction

of antiemetic drugs and antihistamines. In order to prepare the chlorides, HCl (aq) was

introduced into a continuous flow reactor via a reagent loop, which was refilled when

needed. Rapid conversion of neat alcohols was afforded with 3 equivalents of HCl (aq)

in 15 min residence time at 120 ̊C. Despite the great practicality of the process, the

use of HCl gas proved to be an even greener alternative to the use of aqueous HCl due

to the possibility of truly continuous operation (without any need for reagent loop) and

of less equivalents needed for a full conversion of investigated alcohols. Benzyl alcohol

served as a model compound, which was fully converted to a corresponding chloride in 20

min residence time at 100 ̊C with 1.2 equivalents of HCl gas. When the same conditions

were applied to 2,6-difluorobenzyl alcohol 87% yield was obtained. The lower yield was

attributed to the presence of fluorine atoms, that slow down the dissociation of the

protonated hydroxyl group to yield the benzylic cation. Scheme 3.a shows a schematic

representation of the experimental platform, where further optimization was performed.

When the temperature was increased from 100 ̊C to 110 ̊C, a temporary decrease in

flowrate at the outlet was observed. The decrease in flow rate can be explained by the

higher consumption of HCl gas, which results in a larger volume available for the liquid

phase to occupy. The constant flow of liquid phase was once again observed after 40 min

reaction time. The resulting product contained >99wt% of 2,6-difluorobenzyl chloride and

no dibenzyl ether by-product was observed.

6.4. STEP 2- SYNTHESIS OF AZIDE

Organoazides are often high energetic substances that are prone to detonate upon

the release of nitrogen under minimal energy input, such as friction, pressure, heat or

light29. Thus, the production and storage of large quantities of organoazides present

high safety risks. Previously performed safety assessment of the 2,6-difluorobenzyl

azide indicated that the reagent was relatively stable and not shock sensitive24.

Borukhova et al. demonstrated safe cycloaddition of organoazides with a variety of

dipolarophiles under high temperature and pressure in micro flow reactors30. The

most straightforward synthesis of azide is the use of organohalide and azide source,

such as sodium azide. Upon solvation of sodium azide (NaN3) in water the highly

toxic and explosive hydrazoic acid (HN3) is formed as shown in (1).

NaN3 + H2O ⇌ HN3 + NaOH (1)

In order to force equilibrium to the reactant side and prevent the formation of the

acid, sodium hydroxide (NaOH) can be added. However, to see if there was any effect

of the base on conversion and reaction, the reaction was first run without the base.

Scheme 3.b shows the experimental platform used along with the conditions applied

and corresponding results. It should be noted that 2,6-difluorobenzyl chloride is a solid

under atmospheric conditions. Since the aim was to work with as concentrated conditions

as possible, 2,6-difluorobenzyl chloride was melted and 10 wt% of toluene was added.

However, this lead to crystallization upon cooling. Finally, 25 wt% of toluene was needed

to afford homogeneous reactant solution.

Scheme 3. a. Schematic representation of 2,6-difluorobenzyl chloride synthesis and optimization results. b. Schematic representation of 2,6-difluorobenzyl azide synthesis and optimization results. c. Schematic representation of 5-stage multi-step process to deliver rufinamide precursor.

114


115

6

2,6-difluorobenzyl chloride and sodium azide solutions were mixed within an ETFE T-mixer

to yield a two-phase slug flow and heated 140 ̊C. The starting operating conditions were the

ones previously investigated in the optimization of benzyl azide synthesis. As expected,

the addition of base increased the hydrolysis of 2,6-difluorobenzyl chloride and decreased

the conversion, as tabulated in scheme 3.b. Decreasing equivalents of base increased the

selectivity towards 2,6-difluorobenzyl azide. However, more equivalents of NaN3 were

needed to obtain full conversion in 30 min residence time.

6.5. STEP 1+2+ SEPARATION

In order to connect the synthesis of 2,6-difluorobenzyl chloride to the synthesis of azide, the

chloride had to be separated from the acidified water phase, formed in the course of the

reaction. However, crystallization of the product occurred at the outlet of the inline liquid-

liquid separator. Moreover, 2,6-difluorobenzyl chloride is a lachrymator, which is less stable

and more toxic than the corresponding azide. Therefore, in order to minimize the handling

of the chloride it was decided to connect two steps by the neutralization of the excess of HCl

(aq), without any intermediate separation, as shown in Scheme 3.c. A significant difference

in results was observed between the optimization results for azide synthesis only and the

azide synthesis in 2-in-1 flow. The results of the optimization are tabulated in Table 1. The

same concentration of 0.4 M NaOH, after the neutralization of excess HCl from reaction

1, was maintained in reactor 2, which led to partial hydrolysis of the chloride. Decreasing

the concentration of base improved the results, again increasing the selectivity towards

azide. Higher temperatures and NaN3 equivalents were required to reach full conversion of

2,6-difluorobenzyl chloride, indicating higher mass transfer barrier due to the stronger phase

separation at higher concentration of sodium chloride formed during the neutralization.

Table 1. Intermediate results for optimization of 2-step 2,6-difluorobenzyl azide synthesis from 2,6-difluorobenzyl alcohol.

Entry T ( ̊C) Res. time (min) NaN3 (eq) NaOH (M) Conv. (%) Yield (%)

1 140 30 - 0.4 28 0

2 140 30 1.1 0.4 73 60

3 140 30 1.5 0.4 77 66

4 150 30 1.5 0.4 83 72

5 150 30 1.5 0.2 91 86

6 150 40 1.5 0.2 98 92

7 160 30 1.5 0.2 98 91

8 160 40 1.6 0.2 >99 95

2,6-difluorobenzyl azide is liquid under atmospheric conditions and immiscible with

water. Previously, Borukhova et al. described the design and fabrication of an inline Teflon

membrane-based liquid-liquid separator27. In order to afford a perfect separation, the

‘right’ pressure difference should be applied over the membrane to allow permeate,

organic phase, to pass through the membrane without causing a breakthrough of the

retained, aqueous, phase or alternatively, without carrying the permeate within the

retentate. An ultra-low volume adjustable BPR was attached to the aqueous outlet

of the separator and a pressure of 2 psi was set. Meanwhile, 2,6-difluorobenzyl azide

was collected at the organic side via a tubing of 20 cm with 250 µm ID. After reaching

steady-state, 2,6-difluorobenzyl azide was collected for 4 hr resulting in 93% isolated

yield. Meanwhile, 1.2% of total organic phase was lost based on the collected volume

measurements. Karl Fischer titration indicated the presence 0.8% of water within the

separated organic phase.

By connecting two steps, the separation of the hazardous chloride intermediate and the

complication due to its precipitation that would require the addition of a solvent or

a phase transfer catalyst were circumvented. Thus, a decrease in the number of unit

operations and solvent use was achieved. Total residence time of 80 min was needed for

the two-step synthesis of 2,6-difluorobenzyl azide via two nucleophilic substitutions. If

desired, the rate of the reactions could be increased by the addition of a polar solvent,

which would accelerate the reactions. However, due to the fact that our goal was to

deliver a solvent-free process, we avoided the addition of any solvent.

6.6. STEP 1+2+SEPARATION+3+CRYSTALLIZATION

The intensified cycloaddition of 2,6-difluorobenzyl azide to (5) greatly benefited from being

carried out in a continuous manner under superheated conditions24. The most important

benefit is the increased safety due to the circumvention of any significant decomposition

of organic azide. Similarly to a previously published process, 2,6-difluorobenzyl azide

was mixed with (5), heated to 210 ̊C and allowed 10 min reaction time. In order to carry

out continuous synthesis and purification, a stream of methanol was introduced at the

outlet of the reactor. The streams were allowed to mix within a 1 ml mixing zone. Upon

collection needle-like crystals crashed out of the solution. However, the analysis of the

mother liquor indicated the presence of unconverted azide. We attribute the difference

in the results to the difference in the starting composition of the azide feed. Therefore,

we increased the residence time to 15 min, which gave full conversion of the azide. GC-

MS showed that the main side-products were benzyl alcohol, and decomposition products

of 2,6-difluorobenzyl azide. Further cooling, filtration and drying under vacuum resulted

116


117

6

in 88% isolated yield based on 2,6-difluorobenzyl azide, leading to an overall 82% yield

based on 2,6-difluorobenzyl alcohol used. As a result, a productivity of the rufinamide

precursor of 9 g/h was afforded (11.5 mol h-1 L-1).

6.7. CONCLUSION

Continuous flow and solvent-free processing comprise two eminent issues of Green

Chemistry and Green Engineering. Yet, these two processing issues are difficult to combine

due to the high susceptibility of micro flow reactors to clog. Nevertheless, we believe that

the combination should be the next step in the synthesis of fine chemicals to minimize

solvent waste and energy used within a given process. Moreover, incorporation of Novel

Process Windows as a concept to assist in chemical and process-design intensification

proves to be an important stepping stone on the way to sustainable and efficient processes.

Herein, we demonstrated a 5-stage 3-step continuous synthesis with integrated separation

steps. A total yield of 82% of rufinamide precursor with productivity of 9 g/h (11.5 mol h-1

L-1) was delivered. The total residence time needed for the whole process is 95 min, which

is relatively long when compared to other continuous-flow 5-stage processes31. This stems

from the absence of solvent. In this case nucleophilic substitutions could proceed at much

higher rates if diluted with polar solvents. However, when considering the separation of

the solvent and its recycling or disposal, and the purity of the final product, prevention

of the waste generation surpasses the benefits of higher reaction speed. The aim of

minimizing solvent use and, thus, energy needed for its later separation was realized.

Similarly, water use was minimized by combining introduction of a quenching reagent

(NaOH) and a reactant for the subsequent step (NaN3) at once. Moreover, isolation of

toxic and unstable lachrymator, alkylating chloride intermediate, was circumvented

by connecting its synthesis to the synthesis of the corresponding azide. Finally, in the

isolation of 2,6-difluorobenzyl azide no organic extractant was added, instead the azide

was separated in its neat form and used as it is in the subsequent step. All of the mentioned

factors contribute to a better sustainability profile of the individual steps and the flow

microreactor network as a whole.

Finally, if compared to conventional batch process:

‒ Operating time is drastically decreased from days to hours;

‒ Solvent use is circumvented without compromising the safety usually afforded by

dilution;

‒ Overall isolated yield is higher;

‒ Production capacity is variable based on demand;

‒ Inline separation minimizes manual handling.

‒ Meanwhile, if the process is compared to the existing continuous total synthesis:

‒ Synthetic route starts off with a cheaper, greener and more accessible starting

compound;

‒ No solvent is used, as opposed to DMSO use;

‒ 9 times longer residence times are needed to make a precursor;

‒ No catalyst is used, thus no leached metal in the final product;

‒ Inline separation affords more concentrated intermediate;

‒ 6 times higher space time yield is obtained.

ACKNOWLEDGEMENTS

We would like to thank Erik van Herk from Eindhoven University of Technology for his work

on mechanical aspect of the process. Funding by the Advanced European Research Council

Grant “Novel Process Windows – Boosted Micro Process Technology” (Grant number:

267443) is kindly acknowledged.

118

CHAPTER 6

REFERENCES

1. Fujita, M. J. Theor. Biol. 1982, 99 (1), 9–13.

2. Tietze, L. F.; Beifuss, U. Angew. Chemie Int. Ed. English 1993, 32 (2), 131–163.

3. Webb, D.; Jamison, T. F. Chem. Sci. 2010, 1 (6), 675–680.

4. Poechlauer, P.; Colberg, J.; Fisher, E.; Jansen, M.; Johnson, M. D.; Koenig, S. G.; Lawler, M.; Laporte, T.; Manley, J.; Martin, B.; O’Kearney-McMullan, A. Org. Process Res. Dev. 2013, 17 (12), 1472–1478.




8. Gutmann, B.; Cantillo, D.; Kappe, C. O. Angew. Chemie Int. Ed. 2015, 54 (23), 6688–6728.

9. Baumann, M.; Baxendale, I. R.; Martin, L. J.; Ley, S. V. Tetrahedron 2009, 65 (33), 6611–6625.

10. Ducry, L.; Roberge, D. M. Angew. Chemie 2005, 117 (48), 8186–8189.

11. Roberge, D. M.; Ducry, L.; Bieler, N.; Cretton, P.; Zimmermann, B. Chem. Eng. Technol. 2005, 28 (3), 318–323.

12. Roberge, D. M.; Bieler, N. 2008, No. 8, 1155–1161.



15. Hessel, Volker, Kralisch, Dana, Kockmann, N. Novel Process Windows: Innovative Gates to Intensified and Sustainable Chemical Processes; 2015.

16. Damm, M.; Glasnov, T. N.; Kappe, C. O. 2010, 14 (1), 215–224.

17. Benito-Lopez, F.; Tiggelaar, R. M.; Salbut, K.; Huskens, J.; Egberink, R. J. M.; Reinhoudt, D. N.; Gardeniers, H. J. G. E.; Verboom, W. Lab Chip 2007, 7 (10), 1345–1351.

18. Razzaq, T.; Glasnov, T. N.; Kappe, C. O. European J. Org. Chem. 2009, 2009 (9), 1321–1325.

19. Cantillo, D.; Sheibani, H.; Kappe, C. O. J. Org. Chem. 2012, 77 (5), 2463–2473.

20. R. Portmann, U. C. Hofmeier, A. Burkhard, W. S. and M. S. Crystal modification of 1-(2,6-difluorobenzyl)-1H-1,2,3-triazole-4-carboxamide and its use as antiepileptic. US6,740,669, 2004.

21. Kolb, H. C.; Sharpless, K. B. Drug Discov. Today 2003, 8 (24), 1128–1137.

22. Smith, C. D.; Baxendale, I. R.; Lanners, S.; Hayward, J. J.; Smith, S. C.; Ley, S. V. Org. Biomol. Chem. 2007, 5 (10), 1559–1561.

23. Mudd, W. H.; Stevens, E. P. Tetrahedron Lett. 2010, 51 (24), 3229–3231.

24. Borukhova, S.; Noël, T.; Metten, B.; de Vos, E.; Hessel, V. ChemSusChem 2013, 6 (12), 2220–2225.



27. Borukhova, S.; Noël, T.; Hessel, V. ChemSusChem 2016, 9 (1), 67–74.

28. Borukhova, S.; Noël, T.; Hessel, V. Org. Process Res. Dev. 2016, 20 (2), 568–573.

29. Abramovitch, R. A.; Kyba, E. P. In The Azido Group (1971); John Wiley & Sons, Ltd., 1971; pp 221–329.

30. Borukhova, S.; Seeger, A. D.; Noël, T.; Wang, Q.; Busch, M.; Hessel, V. ChemSusChem 2015, 8 (3), 504–512.

31. Snead, D. R.; Jamison, T. F. Angew. Chemie Int. Ed. 2015, 54 (3), 983–987.

CHAPTER 7Economic and environmental evaluation of integrated continuous and batch manufacturing of rufinamide precursor


Borukhova, S., Anastasopoulou, A., Hessel, V. (2017). Economic and environmental evaluation of integrated continuous and batch manufacturing of rufinamide precursor. Chem. Eng. Res. Des., (submitted)

122

CHAPTER 7 Economic and environmental evaluation of continuous manufacturing

123

7

7.1. INTRODUCTION

Pharmaceutical industry is facing tighter economic constraints which lead to consideration

of alternative to conventional ways of manufacturing1. Continuous manufacturing is

becoming increasingly appealing due to the potential of decreasing the production costs

and improving the product quality2. Throughout lifetime of a drug development the

production scale increases from micro grams needed for initial pharmokinetic studies to

kilograms needed for clinical trials3. Within the scale-up stages quality of the product may

vary leading to the waste of chemicals among other resources. Meanwhile, a plethora of

lab scale continuous syntheses of active pharmaceutical ingredients performed in micro

flow reactors has been described4–6. Common benefits observed from converting batch

processes to continuous-flow are tighter control over operating conditions, higher safety,

better mass and heat transfer leading to higher selectivity and thus productivity, and a

possibility to use alternative reagents leading to greener processes2,7–9.

The greenness of continuous processes have been evaluated for several examples10–12.

Conceptual process modelling, economic evaluation and environmental analysis have been

performed for continuous processes of ibuprofen and artemisinin13. Trout et al. performed

an economic analysis of integrated continuous and batch processes of Aliskiren tablet

production and concluded that overall cost savings of 9 to 40% are possible at equal to batch

process yields, and higher in case of better yields14. In terms of environmental assessment

production of pharmaceutical chemicals poses a more challenging task due to the structural

complexity of the reagents and solvents involved. It has been shown that production of

pharmaceuticals results in significantly greater environmental impacts per kilogram when

compared to bulk chemicals. For instance, Cumulative Energy Demand (CED) and Global

Warming Potential (GWP) have been shown to be 20 and 25 times greater, respectively15.

Several mass-based green metrics, such as environmental factor (E-factor), mass efficiency,

reaction mass efficiency, atom economy, and process mass intensity (PMI) exist and are

used in comparison of the process alternatives16. The American Chemical Society Green

Chemistry Institute Pharmaceutical Roundtable evaluated the aforementioned metrics

and presented PMI as a key mass-based green measure in sustainability of pharmaceutical

processes17. Meanwhile, the Roundtable stressed that PMI did not address concerns

associated with health and safety of the raw materials or waste generated. Alternatively, it

was proposed to carry out Life-cycle Assessment (LCA) to evaluate the processes further18.

LCA comprises a methodology evaluating expanded environmental impact, which is known

as ‘cradle-to-grave’ assessment19. It evaluates the processes based on extraction of raw

materials, production, transportation, distribution, sale and final fate of the product.

Moreover, resource consumption, pollutants emitted and waste generated are also taken

into account. ‘Cradle-to-grave’ approach can be split into ‘cradle-to-gate’ and ‘gate-to-

cradle’ assessments. The former option is used when two chemical routes or processes are

compared, such as in the current analysis.

Recently, Ott, Borukhova and Hessel performed a simplified LCA on several alternative

batch and continuous-flow processes to produce rufinamide20. The scenarios evaluated

were a mixture of published and envisioned processes. Herein, we perform a conceptual

scale-up of a recently developed lab-scale continuous-flow process to deliver a precursor

to rufinamide21. In addition, we perform economic and LCA analyses based on the

developed process and compare it to the batch process described within the patents.

7.2. PROCESS MODEL DEVELOPMENT

7.2.1. Starting point- lab-scale continuous-flow processThe flowsheet development for a continuous-flow process is based on the lab-scale

process developed by Borukhova et al21. Figure 1 shows process flow diagram (PFD) for

the continuous flow process. In the first reactor, neat 2,6-difluorobenzyl alcohol (DFB-OH)

reacts with pure hydrogen chloride gas (HCl) (1.2 equivalents) at 140°C and 7 bar to yield

>99% of 2,6-difluorobenzyl chloride (DFB-Cl). At the end of the first reactor an aqueous

stream of sodium hydroxide (NaOH) and sodium azide (NaN3) solution is introduced. NaOH

is used to quench the excess of 0.2 equivalents of HCl used in the previous step, while

NaN3 is a reactant for a following reaction where 2,6-difluorobenzyl azide (DFB-N3) is

formed. Thus, intermediate separation of unstable lachrymator, DFB-Cl, is avoided and

the product flows directly into the second continuous reactor. In the end of the second

reaction, isolation of DFB-N3 takes place within the ETFE-membrane-based liquid-liquid

separator. The process benefits from a sufficient difference among surface tension of the

product and aqueous phase and does not require any addition of the organic extracting

solvent. Isolated product is collected and later pumped into a third reactor from an open

vessel. DFB-N3 is then pumped into a third reactor along with methyl trans-3-methoxy

acrylate (EMMA) to produce ester precursor of rufinamide in 100% regioselectivity and 88%

isolated yield, resulting in 82% overall yield. The final separation of the precursor takes

place when it is mixed in its molten phase with methanol (MeOH) and precipitates upon

cooling within a product collection vessel. Lab-scale process produced 9 g h-1 (11.5 mol h-1

L-1) of a precursor corresponding to 72 kg yr-1.

7.2.2. Scale-up of continuous-flow processBased on the annual sale and the price of rufinamide tablets containing 400 mg dose, we

estimated a yearly production of the compound to be 5 tons. Mass balance calculations

124


125

7

Flow

s g/

h

Line

no.

St

ream

co

mpo

nent

1 Al

coho

l fe

ed

2 H

Cl g

as

feed

3 N

aOH

& N

aN3

feed

4 EM

MA

feed

5 M

eOH

fe

ed

6 R

eact

or I

feed

7 R

eact

or I

outle

t

8 R

eact

or II

fe

ed

9 R

eact

or II

ou

tlet

10

Org

anic

si

de

11

Aque

ous

Side

- Was

te

12

Azid

e fe

ed

13

Rea

ctor

III

feed

14

Rea

ctor

III

outle

t

15

Col

lect

ion

Vess

el

DFB

-OH

10

-

- -

- 10

-

- -

- -

- -

- -

HC

l -

3 -

- -

3 0.

5 -

- -

- -

- -

- N

aOH

-

- 0.

72

- -

- -

0.2

0.2

Trac

e 0.

2 Tr

ace

Trac

e Tr

ace

Trac

e N

aN3

- -

7.2

- -

- -

7.2

2.9

Trac

e 2.

9 Tr

ace

Trac

e Tr

ace

Trac

e W

ater

-

- 19

.3

- -

- 1.

2 20

.8

20.8

Tr

ace

20.8

Tr

ace

Trac

e Tr

ace

Trac

e EM

MA

- -

- 7.

3 -

- -

- -

- -

- 7.

3 2.

4 2.

4 M

eOH

-

- -

- 15

2 -

- -

- -

- -

- 1.

3 1.

3 D

FB-C

l -

- -

- -

- 11

.3

11.3

0.

5 0.

5 Tr

ace

0.4

0.4

0.4

0.4

DFB

-N3

- -

- -

- -

- -

11.1

11

0.

1 7.

4 7.

4 0.

4 0.

4 ES

TER

- -

- -

- -

- -

- -

- -

- 10

.5

0.8

ESTE

R*

- -

- -

- -

- -

- -

- -

- -

9.7

NaC

l -

- -

- -

- -

0.8

4.6

Trac

e 4.

6 Tr

ace

Trac

e Tr

ace

Trac

e

To

tal

10

3 27

.2

7.3

152

13

13

40.2

40

.2

11.6

28

.6

7.8

15.1

15

.1

15.1

Pr

ess

bar

1 32

1

1 1

7 7

7 7

1 1

1 69

69

1

Tem

p °C

20

20

20

20

20

20

11

0 11

0 16

0 20

20

20

20

21

0 20

*

Cry

stal

line

1,2,

3-tr

iazo

le e

ster

pre

curs

or o

f ru

fina

mid

e

Fi

gure

1.

Top:

Pro

cess

flow

dia

gram

(PF

D)

for

the

lab-

scal

e co

ntin

uous

flow

syn

thes

is o

f ru

finam

ide

prec

urso

r. B

otto

m:

Mas

s flo

w b

alan

ce o

f th

e la

b-sc

ale

proc

ess

depi

cted

on

top.

showed the lab-scale needs to be multiplied by the factor of 46 to meet the required

productivity in 8000 hrs/year of operation. Higher flowrates result in faster intermediate

collection of DFB-N3, allowing faster flow rates into the consequent reactor. Therefore,

flowrate of stream 12 was multiplied by a factor of 1.4. Due to the large consumption of

methanol within the process, its separation with a recycle stream entering the crystallizer

feed stream was incorporated within the process. The calculations were based on the

steady-state operation. Operating conditions within the reactors and stoichiometry of the

feed streams were assumed to be the same as in the lab-scale process. Figure 2 shows the

PFD of the continuous-flow large scale process.

7.2.3. Batch processBatch process described herein is based on the information compiled from available

patents. The process starts off with chlorination of DFB-OH with thionyl chloride (2 eq.)

(SOCl2) in dichloromethane (DCM) at 0°C. The reaction is completed within 30 min, which

is followed by addition of water to quench SOCl2. Gaseous products of the reaction, i.e.

hydrogen chloride gas (HCl) and sulfur dioxide (SO2) are separated in their gaseous form,

while biphasic mixture is transferred into a settling tank to extract DFB-Cl from water

and DCM. The amount of water added is based on required amount for quenching and full

solubility of DCM.

Separated in 97% yield DFB-Cl is then mixed with sodium azide in DMSO to produce

DFB-N3. The product is then extracted into toluene, which also serves as a solvent for

the next step. Consequently, methyl propiolate is added to DFB-N3 in toluene and heated

up to 60-65°C for 4-5 hrs. The resulting mixture contains both 1,4- and 1,5- regiosomers.

After the final separation, 1,4-regioisomer is isolated in 55% yield. No information has

been provided on separation of 1,5-regioisomer and therefore is not included within the

calculations performed. In order, to reach the same capacity of annual production of 5

tons, it was assumed that 10 batches are carried out each time delivering 500 kg of the

API. Another assumption was that every campaign is completed in four stages. Figure 3

shows the PFD and amounts used per one stage.

Table 1. Annual energy requirement for continuous and batch processes to produce 5 tons of rufinamide precursor.

Energy Requirements (MJ/yr) Continuous* Batch

Heating requirement 33.4 7433

Cooling requirement 27.7 1616.2

Total 61.1 9049.2

*excluding distillation column needed for the recycle of methanol.

126


127

7

Flow

s g/

h

Line

no.

St

ream

co

mpo

nent

1 Al

coho

l fe

ed

2 HC

l ga

s fe

ed

3 N

aOH

&

NaN

3 fe

ed

4 EM

MA

feed

5 M

eOH

fe

ed

6 R

eact

or I

feed

7 R

eact

or I

outle

t

8 R

eact

or II

fe

ed

9 R

eact

or II

ou

tlet

10

Org

anic

si

de

11

Aque

ous

Side

- W

aste

12

Azid

e fe

ed

13

Rea

ctor

III

fe

ed

14

Rea

ctor

III

outle

t

16

Cry

stal

lizer

ou

tlet

17

18

19

20

DFB

-OH

45

9 -

- -

- 45

9 -

- -

- -

- -

- -

- -

- -

HC

l -

138

- -

- 13

8 22

-

- -

- -

- -

- -

- -

- N

aOH

-

- 33

-

- -

- 9

9 Tr

ace

9 Tr

ace

Trac

e Tr

ace

Trac

e -

Trac

e Tr

ace

- N

aN3

- -

330

- -

- -

330

133

Trac

e 13

2 Tr

ace

Trac

e Tr

ace

Trac

e -

Trac

e Tr

ace

- W

ater

-

- 88

9 -

- -

57

957

957

1 95

6 1

1 1

1 -

1 1

- EM

MA

- -

- 33

4 -

- -

- -

- -

- 33

4 24

24

-

24

24

- M

eOH

-

- -

- 35

4 -

- -

- -

- -

- 85

70

80

- 70

80

354

6726

D

FB-C

l -

- -

- -

- 51

8 51

8 26

26

Tr

ace

24

24

24

24

- 24

24

-

DFB

-N3

- -

- -

- -

- -

512

506

6 47

5 47

5 24

24

-

24

24

- ES

TER

- -

- -

- -

- -

- -

- -

- -

- -

- -

- ES

TER

* -

- -

- -

- -

- -

- -

- -

675

675

(626

*)

626

50

50

- N

aCl

- -

- -

- -

- 35

21

2 2

210

2 2

2 2

- 2

2 -

Tota

l 45

9 13

8 12

52

334

6995

59

7 59

7 18

49

1849

53

5 13

14

502

836

836

7831

6

26

7205

48

0 67

26

Pres

s ba

r 1

32

1 1

1 7

7 7

7 1

1 1

69

69

1 1

1 1

1 Te

mp

°C

20

20

20

20

20

20

110

110

160

20

20

20

20

210

5 20

20

65

20

*

Cry

stal

line

1,2,

3-tr

iazo

le e

ster

pre

curs

or o

f ru

fina

mid

e

Fi

gure

2.

Top:

Pro

cess

flow

dia

gram

(PF

D)

for

larg

e sc

ale

cont

inuo

us fl

ow s

ynth

esis

of

rufin

amid

e pr

ecur

sor.

Bot

tom

: M

ass

flow

bal

ance

of

the

larg

e sc

ale

proc

ess

illus

trat

ed o

n to

p.

kg/s

tage

Line

no.

St

ream

co

mpo

nent

1 R

eact

or I

feed

2 N

aN3

feed

3 To

luen

e LL

E fe

ed

4 Pr

opio

late

fe

ed

5 W

ater

fe

ed

6 O

ff-ga

s st

ream

7 LL

E in

let

8 LL

E w

aste

9

LLE

outle

t 10

R

eact

or II

ou

tlet

11

Wat

er fe

ed

12

LLE

was

te

13

LLE

outle

t 14

R

eact

or II

I ou

tlet

DFB

-OH

13

8 0

0 0

0 0

0 0

0 0

0 0

0 0

SOC

l2

227

0

0 0

0 0

0 0

0 0

0 0

0 D

CM

13

75

0

0 0

0 13

75

1375

0

0 0

0 0

0 W

ater

0

0 0

0 53

728

0 53

728

5372

8 0

0 37

61

3761

0

0 D

FB-C

l 0

0 0

0 0

0 15

0 0

146

0 0

0 0

0 N

aN3

0 60

0

0 0

0 0

0 0

0 0

0 0

0 D

MSO

0

1881

0

0 0

0 0

0 0

1881

0

1881

0

0 To

luen

e 0

0 15

04

0 0

0 0

0 0

0 0

0 15

04

1504

D

FB-N

3 0

0 0

0 0

0 0

0 0

152

0 0

152

0 Pr

opio

late

0

0 0

75

0 0

0 0

0 0

0 0

0 0

HC

l 0

0 0

0 0

35

0 0

0 0

0 0

0 0

SO2

0 0

0 0

0 61

0

0 0

0 0

0 0

0 N

aCl

0 0

0 0

0 0

0 0

0 52

0

52

0 0

1,4-

Este

r 0

0 0

0 0

0 0

0 0

0 0

0 0

125

1,5-

Este

r 0

0 0

0 0

0 0

0 0

0 0

0 0

102

To

tal

1740

19

41

1504

75

53

728

96

5525

4 55

103

146

2085

37

61

5694

16

56

1732

Pres

s ba

r 1

1 1

1 1

1 1

1 1

1 1

1 1

1 Te

mp

°C

20

20

20

20

20

20

0 20

20

20

20

20

20

65

Figu

re 3

. To

p: P

roce

ss fl

ow d

iagr

am (

PFD

) fo

r la

rge

scal

e ba

tch

proc

ess

to p

rodu

ce r

ufina

mid

e pr

ecur

sor.

Bot

tom

: M

ass

flow

bal

ance

of

the

batc

h pr

oces

s sh

own

abov

e.

128


129

7

7.3. ENERGY REQUIREMENTS

Heating and cooling requirements were calculated based on the heat needed to be added

or removed to the reagents within the reactors based on the optimized procedures. The

results should be considered as a theoretical minimum since actual energy consumed by

the heating or cooling equipment was not considered here. Specific heat capacities under

constant pressure of reagents not listed within the Ullmann encyclopedia were calculated

based on the estimation methodology proposed by Rihani et al.22. The approximation is

based on equation 1, where ni is the number of occurrences of functional group i within

a specific molecule. ai, bi, ci and di are given based on the functional group type, while T

is the reference temperature.

𝐶𝐶# = 𝑛𝑛&𝑎𝑎& +&

𝑛𝑛&𝑏𝑏&𝑇𝑇 + 𝑛𝑛&𝑐𝑐&𝑇𝑇, + 𝑛𝑛&𝑑𝑑&𝑇𝑇.&&&

(1)

Table 1 shows the energy requirements for large scale continuous-flow and batch

processes. Due to the intensive use of the solvents in case of batch process significant

energy consumption is required, even though the temperature differences are lower.

7.4. MATERIALS CONSUMPTION

Raw material requirements were calculated based on the large scale flow and batch

processes. The price for each chemical was acquired from vendors providing the chemicals

in the bulk quantities (>50 kg). Table 2 shows the results of the calculations. Water

contributes to more than 90% of the consumed materials in batch process. As mentioned

earlier, it is used to remove DCM and DMSO from intermediate reagents. In terms of the

costs, batch process requires >8 times larger investment for raw materials to produce the

same amount of the final product.

Table 2. Annual raw materials requirements and associated costs for continuous and batch processes to produce 5 tons of rufinamide precursor.

Materialsraw materials requirements (kg/yr) raw materials costs (€/yr)

Continuous Batch Continuous Batch

Organic reagents 6,345 8,519 961,261 2,990,016

Inorganic reagents 4,009 11,487 120,922 931,674

Organic solvents 2,800 190,404 81,165 5,005,023

Water 7,110 2,299,574 711 229,957

Total 20,264 2,509,984 1,164,059 9,156,669

7.5. EQUIPMENT COSTS

Based on the required productivity, we selected similar equipment employed in continuous-

flow process, but with a higher capacity. The quotations were obtained from the providers

of the corresponding equipment. Meanwhile, the price of the distillation column required

for the recycle of methanol was estimated in ASPEN. Table 3 shows the equipment cost

for the continuous-flow process.

Costs for batch equipment were obtained from an online equipment cost calculator23. For

the reactors it was assumed that 80% of the reactor is filled in order to account for vortex

formation during mixing. Moreover, due to the corrosive nature of HCl formed in the first

reactor, and use of sodium azide in the following reactors, glass-lined stainless reactors

were chosen. It was assumed that reactors could also be used for extraction of a product.

Mixer costs were not included in batch equipment cost calculation due to the lack of

information regarding the power input needed.

Total equipment costs for continuous process are three times more expensive than those

of batch. Capital Expenditures (CapEx) are proportional to the equipment cost, thus

the CapEx calculations were not further carried out. Operating Expenditures (OpEx) are

expected to be substantially lower for continuous process due to the following factors:

‒ twice as many operators would be required for batch process than continuous14;

‒ materials handling and storage can be estimated to be 40% of batch;

‒ off-spec product is minimal for continuous process;

‒ quality assurance and control is cheaper in case of continuous process due to the

inline analytical measurements;

‒ utility expenditures are less in case of continuous process;

‒ waste disposal is lower for continuous process due to the lower use of water and

solvents and higher material efficiency.

7.6. ENVIRONMENTAL ANALYSIS

7. 6.1. Process Mass IntensityA PMI evaluation was carried out to provide an overview for improvements brought by

the switch from batch to continuous process in terms of material consumption. PMI is

defined as the total mass of materials consumed to deliver a specified mass of product.

Ideally PMI equals 1 when no waste is generated and all of the materials are incorporated

within the final product. The Roundtable developed a PMI calculator tool, which allows

130


131

7step and cumulative calculations of PMI18. Figure 4 shows PMI values for flow and batch

processes based on the total materials consumed in single reaction step calculated by

the developed calculator tool. Figure 5 shows the breakdown into contribution made

by reagents, solvents and water used in PMI calculations. The distribution published by

the Roundtable for processes implemented by several pharmaceutical companies shows

that solvents contribute to 56% of PMI, water to 32%, while reagents only to 7%17. In this

context, the batch process is similar to a typical pharmaceutical process. The comparison

of batch and flow processes shows how more mass-efficient flow process is, mainly due to

the fact that the process does not use the major contributor to PMI, i.e. solvent.

7.6.2. Life cycle assessmentAs stated in the introduction, PMI only shows the mass intensity of the process leaving

out concerns regarding the environment, health and safety of the materials employed or

produced. Therefore, LCA needs to be performed to shine the light on the aforementioned

concerns.

System Boundaries

The LCA study was carried out using Umberto NXT LCA software. The system boundaries

were defined as “cradle to gate”, meaning the inclusion of all activities commencing from

Table 4. Cost of the equipment required for batch process assembly.

Equipment Volume (m3) Cost (€)

Reactor 1 2,2 30,844

Reactor 2 2,61 34,747

Reactor 3 2,16 30,844

Total 96,435

Table 3. Cost of the equipment required for continuous-flow process assembly.

Equipment Quantity Vendor Cost (€)

Mass Flow Controller 1 Bronkhorst 4,729

Pump 7 Knauer 4,995

L-L Separator 1 Zaiput 85,000

Reactors 3 Syrris 24,000

Vessels 2 VWR 2,400

Filter 1 Custom-made 65,000

Distillation column 1 Custom-made 61,000

Kenics Mixers 4 Chemineer 1,600

Total 248,724

the extraction of raw materials to the production of the final product. In Figure 6 the

system boundaries of both examined flow and batch systems are presented with respect

to their elementary material and energy exchanges. For both processes, the energy costs

linked to the recycle of the methanol and toluene solvents have not been considered and,

thus, the final product is perceived to be the produced ester dissolved in the respective

amount of solvent. In the given LCA study, the waste and emissions generated from both

processes have been treated in the same way as presented in Figure 6. Moreover, the

storage and transportation of the final product has also not been taken into account.

Inventory Analysis

The functional unit of the present LCA study is defined to be 1 kg ester dissolved/

suspended in the respective solvent- methanol and toluene for flow and batch processes,

Figure 4. PMI values for flow and batch processes based on sum of reagents, solvents and water consumed within each synthetic step.

0

2

4

6

8

10

12

14

16

1 2 3

Step 1 Step 2 Step 3

FLOW

BATCH

Figure 5. PMI values for flow and batch processes in terms of reagents, solvents and water consumed within each synthetic step.

1-Flow1-Batch

2-Flow2-Batch

3-Flow3-Batch

0

2

4

6

8

10

12

14

1 2 3Reagents Solvents Water

132


133

7

respectively. Regarding the raw material extraction and the energy provision to the flow

and batch production systems, unit-operations embedded in the Umberto software have

been employed for all material and energy inputs as shown in Figure 6. The relevant

mass and energy flows for the flow and batch processes have been acquired from the

Figures 2 and 3 and are expressed per functional unit in Tables 4 and 5, respectively.

Furthermore, as far as the life cycle inventory (LCI) data is concerned, the Ecoinvent 3.0

(v3.1) database incorporated in Umberto NXT LCA software has been deployed. Inventory

data of the energy exchanges and the majority of the chemical components has been

directly available from the database. However, for those components with no available

entries in the database and no complete background information, LCI data of similar

chemical components from the database has been utilized instead. The list of all the

inventory data employed in the given LCA study is presented in Table 6. Considering the

Figure 6. System boundaries of flow and batch processes for the production of 1 kg ester (in solution).

Table 4. Material and energy input for the flow process expressed per kg of ester.

Material/Energy Input Output

DFB-OH 0.73 kg -

HCl 0.22 kg -

NaOH 0.05 kg 0.01 kg

NaN3 0.53 kg 0.21 kg

EMMA 0.53 kg 0.04 kg

MeOH 11.31 kg 10.74 kg

MeOH (waste) - 0.57 kg

DFB-Cl - 0.04 kg

Water 1.42 kg 1.53 kg

DFB-N3 - 0.05 kg

Ester - 1.00 kg

Ester (waste) - 0.08 kg

NaCl - 0.34 kg

Cooling Energy 5.54E-03 MJ -

Heat 6.68E-03 MJ -

Table 5. Material and energy input for the batch process expressed per kg of ester.

Material/Energy Input Output

DFB-OH 0.61 kg -

SOCl2 1.00 kg -

DCM 6.06 kg 6.06 kg

Water 253.26 kg 253.26 kg

DFB-Cl - -

NaN3 0.26 kg -

DMSO 8.29 kg 8.29 kg

Toluene 6.63 kg 6.63 kg

DFB-N3 - -

Methyl Propiolate 0.33 kg -

Ester - 1.00 kg

HCl - 0.15 kg

SO2 - 0.27 kg

NaCl - 0.23 kg

Cooling Energy 0.32 MJ

Heat 1.49 MJ

134


135

7

Table 6. Inventory data incorporated in the LCA study.

Material/Energy Exchanges Inventory Data (Ecoinvent 3.0)

MeOH (input) methanol production [GLO]*

NaOH (input) market for sodium hydroxide, without water, in 50% solution state [GLO]*

EMMA (input) methyl acrylate production [GLO]*

NaN3 (input) atrazine production [RER]*

DFB-OH (input) benzyl alcohol production [RER]*

HCl (input) hydrochloric acid production, from the reaction of hydrogen with chlorine [RER]*

SoCl2 (input) thionyl chloride production [RER]*

Toluene (input) toluene production, liquid [RER]*

DMSO (input) dimethyl sulfoxide production [RER]*

DCM (input) dichloromethane production [RoW]*

Methyl propiolate (input) methyl acrylate production [GLO]*

Water (input) water production, deionised, from tap water, at user [CH]*

Cooling Energy (input) cooling energy, from natural gas, at cogen unit with absorption chiller 100kW [CH]*

Heat (input) heat production, natural gas, at industrial furnace >100kW [Europe without Switzerland]*

Ester (product) Esters of versatic acid [intermediate exchange]

Methanol (product stream) Methanol [intermediate exchange]

NaCl (waste) Sodium chloride, in ground [natural resource/in ground]

NaOH (waste) Hydroxide [water/ground-]]

Water (waste) Wastewater,average

EMMA (waste) Methyl acrylate [water/ground-]

Methanol (waste) Methanol [water/ground-]

DFB-Cl (waste) Methane, dichlorofluoro-, HCFC-21 [water/ground-]

DFB-N3 (waste) Methane, dichlorofluoro-, HCFC-21 [water/ground-]

NaN3 (waste) Atrazine [soil/agricultural]

HCl (emissions) Hydrogen chloride [air/unspecified]

SO2 (emissions) Sulphur trioxide [air/unspecified]

Toluene (product stream) Toluene, liquid [intermediate exchange]

DCM (waste) Methane, dichloro-, HCC-30 [water/ground-]

DMSO (waste) Dimethyl ether [water/ground-]

*Refers to activity datasets embedded in the Ecoinvent 3.0 database

fact that the final product comprises two components, ester and solvent, in order to

avoid the allocation problem of the distribution of the generated emissions among the

defined products posed by the multi-functionality of the studied systems, the material

flow of methanol and toluene in the product streams has been treated in the LCA study

as “neutral” which denotes no contribution to the overall environmental performance of

the systems. In addition to that, the NaCl and ester components considered as waste in

this study are elementary and intermediate exchanges, respectively, in the Ecoinvent 3.0

with a “good” material type, meaning that no environmental burden is allocated to them.

Since these components are in the product stream they are considered to be co-products.

Based on this fact, in order to eliminate the allocation issue, the same approach as to

the methanol and toluene flows has been also applied to the respective flows of those

components.

Impact Assessment

The impact categories against which the environmental performance of the flow

and batch processes will be assessed in this LCA study have been selected from the

ReCiPe Midpoint (Hierarchist) and are the following: climate change (GWP100), fossil

depletion (FDP), freshwater eutrophication (FEP), human toxicity (HTPinf), metal

depletion (MDP), natural land transformation (NLTD), ozone depletion (ODPinf),

photochemical oxidant formation (PDFD), terrestrial acidification (TAP100) and

terrestrial ecotoxicity (TETPinf).

Interpretation

The interpretation of the generated LCA results has been established upon the individual

environmental impact of the elementary material and energy exchanges. More precisely,

the distribution of the selected impact categories values among the unit-operations linked

to the material and energy exchanges, as well as, the emissions/waste directly produced

from the process is reflected in a normalized manner for both flow and batch processes

in a pivot graph shown in the results section. In that graph the components, which have

been replaced by similar ones with respect to the inventory data, are presented with both

components names. However, in the discussion of the LCA results their environmental

contribution will be expressed only in terms of the initial component names so as to avoid

any potential ambiguity.

Regarding the replacement of the missing components in the Ecoinvent 3.0 database

with similar ones, there has been a given group of available choices for some of these

components, and more precisely for the EMMA as an input material and the NaN3 as an

input and output exchange flows. In order to evaluate the possible influence of those

alternatives on the overall environmental profile of the batch and flow processes, a

sensitivity analysis has been conducted. The results from this analysis have demonstrated

no remarkable fluctuation in the values of the majority of the impact categories as to the

ones presented below.

136


137

7

Results

In Figure 7 the normalized impact categories of the flow and batch processes are presented

against the process with the higher nominal value. As it can be deduced, the GWP of the

batch process is remarkably higher than that of the flow process. This is mainly attributed

to the air emissions directly given off by the batch process which account for 53% of the

total GWP profile. The next contributory factors are the DCM production with a share of

21% and the use of toluene and DMSO with a corresponding share of 10%.

In the case of the flow process, the GWP is mainly dominated by the impact of the

methanol and sodium azide production which accounts for 40% and 29%, respectively. The

use of DFB-OH is the third factor influencing the overall GWP value of the flow process by

20%. The use of HCl and NaOH along with the cooling and heat requirements contribute

the least to the overall emissions of the process by a total percentage of 2%.

In terms of the FDP, FEP and MDP impact categories, the batch process remains inferior to

the flow process but to a lesser extent and with different material/energy distribution as

compared to the GWP profile. To elaborate, the certain impact categories for the batch

process are majorly prevailed by the impact of the toluene and DMSO. On the other hand,

for the flow process the same trend is followed as in the GWP profile with respect to the

individual impact of materials and energy exchanges.

As far as the HTPinf is concerned, the contribution of the emissions in the batch process

reaches the percentage of 85%. On the contrary, the profile of the flow process, with a

nominal HTPinf value lower by 91%, is mainly controlled by the MeOH and NaN3 input flows

whose shares are 36% and 28%, correspondingly. Additionally, the flow process exhibits

also markedly lower value as to the batch one for the ODPinf impact category, with the

profile of the latter process being dominated by the DCM material flow.

The environmental performance of the flow process seems to be inferior to that of

the batch process for the NTLP and TETPinf impact categories. More specifically, the

methanol production and the generated waste are the main factors influencing those

impact categories with a respective share of 75% and 100%. Regarding the remaining

category of POFP, the profile of both production systems follows a similar trend as to that

of the TAP100 with certain fluctuations in the individual share of the major contributory

materials.

7.7. CONCLUSION

Continuous processes are developed as alternative to batch processes due to economic

and environmental benefits. In the last decade, multiple continuous flow processes

have been developed on a lab-scale in academia. Their accurate comparison to existing

batch process is usually difficult due to the lack of information disclosed by industry.

The approach, followed herein, was based on the assembly of overall process based on

process conditions acquired from available patents. From the comparison, it was found

that continuous flow process based on existing lab-scale process requires less energy and

less materials to deliver the final product. The cost of the equipment was calculated to

higher for continuous flow process. Meanwhile, operating expenditures are expected to

be lower for continuous flow process than for batch.

Environmental analysis was based on the process mass intensity (PMI) and life cycle

assessment (LCA). Continuous process is based on solvent-free reaction steps, which

leads to substantially lower PMI than that of batch process. Finally, LCA analysis

showed environmental profile of process in terms of environmental, toxicity and health

Figure 7. Normalized LCIA profiles of flow and batch processes for the for the production of 1 kg ester (in solution).

0

0,2

0,4

0,6

0,8

1

Batch

Flow

Batch

Flow

Batch

Flow

Batch

Flow

Batch

Flow

Batch

Flow

Batch

Flow

Batch

Flow

Batch

Flow

Batch

Flow

GWP FDP FEP HTPinf MDP NTLP ODPinf POFP TAP100 TETPinf

Toluene ThionylchlorideDCM DMSOAtrazine/NaN3 Benzylalcohol/DFB-OHWater Methylacrylate/MethylpropiolateNaOH Methylacrylate/EMMAMethanol HClCoolingenergy HeatEmissions/Waste

138


139

7

aspects. Human toxicity factor was found to be marginal in the case of continuous flow

process as opposed to the batch. In the remaining categories, continuous flow process

outweighs batch in terms of negative impact in terrestrial ecotoxicity and natural land

transformation. Methanol production and the generated waste were found to be the main

factors influencing those impact categories.

REFERENCES



3. DiMasi, J. A.; Grabowski, H. G.; Hansen, R. W. J. Health Econ. 2016, 47, 20–33.

4. Hopkin, M. D.; Baxendale, I. R.; Ley, S. V. Chem. Commun. (Camb). 2010, 46, 2450–2452.

5. Pastre, J. C.; Browne, D. L.; Ley, S. V. Chem. Soc. Rev. 2013.

6. Rodrigues, T.; Schneider, P.; Schneider, G. Angew. Chemie - Int. Ed. 2014, 53, 5750–5758.


8. Haswell, S. J.; Watts, P. Green Chem. 2003, 5 (2), 240–249.

9. Wegner, J.; Ceylan, S.; Kirschning, A. Adv. Synth. Catal. 2012, 354 (1), 17–57.

10. LaPorte, T. L.; Spangler, L.; Hamedi, M.; Lobben, P.; Chan, S. H.; Muslehiddinoglu, J.; Wang, S. S. Y. Org. Process Res. Dev. 2014, 18 (11), 1492–1502.

11. Laporte, T. L.; Hamedi, M.; Depue, J. S.; Shen, L.; Watson, D.; Hsieh, D. 2008, 12 (5), 956–966.


13. Jolliffe, H. G.; Gerogiorgis, D. I. Comput. Chem. Eng. 2016, 91, 269–288.

14. Schaber, S. D.; Gerogiorgis, D. I.; Ramachandran, R.; Evans, J. M. B.; Barton, P. I.; Trout, B. L. Ind. Eng. Chem. Res. 2011, 50 (17), 10083–10092.

15. Wernet, G.; Conradt, S.; Isenring, H. P.; Jiménez-González, C.; Hungerbühler, K. Int. J. Life Cycle Assess. 2010, 15 (3), 294–303.

16. Jimenez-Gonzalez, C.; Constable, D. J. C.; Ponder, C. S. Chem. Soc. Rev. 2012, 41 (4), 1485–1498.

17. Jimenez-Gonzalez, C.; Ponder, C. S.; Broxterman, Q. B.; Manley, J. B. Org. Process Res. Dev. 2011, 15 (4), 912–917.

18. Jiménez-González, C.; Ollech, C.; Pyrz, W.;

Hughes, D.; Broxterman, Q. B.; Bhathela, N. Org. Process Res. Dev. 2013, 17 (2), 239–246.

19. Cespi, D.; Beach, E. S.; Swarr, T. E.; Passarini, F.; Vassura, I.; Dunn, P. J.; Anastas, P. T. Green Chem. 2015, 17 (6), 3390–3400.


21. Borukhova, S.; Noel, T.; Metten, B.; de Vos, E.; Hessel, V. Green Chem. 2016, -.

22. Rihani, D. N.; Doraiswamy, L. K. Ind. Eng. Chem. Fundam. 1965, 4 (1), 17–21.

23. Peters, M.S., Timmerhaus, K.D., West, R. E. Plant Design and Economics for Chemical Engineers http://highered.mheducation.com/sites/0072392665/student_view0/cost_estimator.html.

CHAPTER 8Research Impact and Further Recommendations

142

CHAPTER 8 Research Impact and Further Recommendations

143

8

8.1. CONTINUOUS MANUFACTURING OF ACTIVE PHARMACEUTICAL INGREDIENTS VIA MULTI-STEP CONTINUOUS-FLOW PROCESSES

HighlightsMicro flow chemistry as mentioned throughout the current thesis is an exceptional

operating platform to deliver APIs at different scales with consistent quality. The current

thesis presented the development of a process to deliver antiemetic and antihistaminic

piperazine-based drugs. Cinnarizine and a buclizine derivative were constructed from

bulk alcohols in 6-stage 4-reaction step processes. The whole sequence was realized in

90 minutes with good overall yields (>80%) at a production rate of 2 mmol/hr of final

products. If so desired, the scale could be increased by using wider and/or longer reactors

and/or the employment of several parallel reactors. Furthermore, a precursor for an

antiepileptic drug, rufinamide, was synthesized from alcohol in a 5-stage 3-reaction step

process. A total yield of 82% with productivity of 9 g h-1 (11.5 mol h-1 L-1) within a total

residence time of 95 min was achieved.

RecommendationsIn order to improve the continuous manufacturing of these particular APIs the secondary

production process would be the next aspect to look into. Continuous separation of solid

APIs in order to prepare the final solid tablets would complete the production sequence.

Luckily, the secondary API production stage is already continuous or semi-continuous.

The missing link is bringing down the scale to meet the one micro- or meso-scale is

capable of delivering. As a step closer, the research described herein served as a base

for a submission of a Proof-of-Concept project proposal, which passed the evaluation

threshold and is currently in a waiting list. Finally, the study also stimulated a bigger scale

EU proposal such as Fast-Track-to-Innovation. The proposal is concentrated on a flow

multi-step pilot study and pilot-scale flow equipment development.

8.2. MICRO FLOW PROCESS TECHNOLOGY

HighlightsBased on our knowledge, we were the first to construct and successfully operate a

continuous process utilizing the highly corrosive pure hydrogen chloride gas. Usually,

when the highly oxidative gas is used less expensive and less accurate valves are used to

minimize the equipment cost in case of failure. Within the current design, the operation

was possible with no harm to the equipment fabricated from stainless steel. The highly

accurate mass flow controller employed saw no harm due to the division of the setup

into dry and wet zones. The dry zone was employed as a gas delivery unit, while the

wet zone was used to carry out the reaction. After operating for more than 6 months

no corrosion was observed within the dry zone. The use of a micro flow reactor allowed

the application of process conditions that are beyond the limits of conventional batch

technology. Operation at high pressures increased the solubility of hydrogen chloride gas

within the reagent increasing the rate of the reaction, while not requiring as much excess

as when hydrochloric acid was used as an alternative.

A membrane based liquid-liquid separator was designed and constructed to deliver

integrated processes, comprising micro flow networks.

In order to investigate high pressure effects on the intrinsic kinetics of 1,3-dipolar

cycloaddition, a high-pressure micro flow operating platform was built. The upper

threshold of operating pressure was limited by the availability of HPLC pumps that could

operate only up to 400 bar. Even though the pressure was not sufficient to affect kinetics

by itself, it allowed operation under superheated conditions that are not attainable in

conventional batch reactors.

The continuous synthesis of 1,2,3-triazole involved organoazide requires special precaution

when carried out in batch due to its temperature sensitive nature and predisposition to

decompose to yield nitrogen, which leads to explosions. Carrying out the reaction at 210

°C and 70 bar in micro flow presented no explosion concerns.

Operation under superheated conditions above the melting point of the solid API precursor

prevented the common problem of precipitation, leading to clogging of micro reactors.

In addition, by introduction of recrystallization solvent purification was realized upon

product collection.

RecommendationsDuring operation under superheated conditions polymer-reactors were observed to have a

limited life-time due to the harsh conditions as a consequence of using high temperature

and pressure. More robust, stainless steel capillaries coated with polymer on the inside

would present a better alternative and increase the safety. In addition, when the gas-

liquid reaction was connected to the liquid phase reaction and separated by a check

valve, the latter would fail in case the reagents were pumped into the second part of the

combination. We attributed that due to the creation of a low pressure region due to the

consumption of the gas, thus a decrease of the volume within the first reactor. Meanwhile,

superheated operation within the second reactor required high pressure. Therefore, the

flow of reagents fed into the second reactor would flow in the opposite to expected

144

CHAPTER 8 Research Impact and Further Recommendations

145

8

direction through the check valve. The problem was solved by allowing pressure within

the first reactor to build up, while gradually increasing the flow rate of the reagents for

the second reactor. While the solution was good enough at the lab-scale, a more robust

solution has to be created for the larger scale to avoid failure of the operation.

The residence time was estimated based on the internal volume and the flow rate of the

reagents. However, a more accurate way to measure the actual residence time would lead

to a deeper understanding of the flow and reaction dynamics.

Finally, the consumption of the chemicals during the optimization stage could be decreased

if inline analytics were integrated. The best option for the current project would be

an installation of the inline FLOWIR made available by METTLER TOLEDO. Monitoring of

alcohol, organohalide and organoazide production and consumption could benefit the

overall process optimization.

8.3. GREEN CHEMISTRY AND GREEN ENGINEERING

HighlightsCircumvention of employing wasteful chlorinating agents in a continuous synthesis of

chlorides from bulk alcohols was afforded via use of hydrogen chloride gas instead. The

initial need for an occasional stop of the continuous operation in the case hydrochloric

acid was used, was gone when hydrogen chloride was utilized. High temperature and

pressure, easily applicable within the reacting system, allowed minimizing the excess

of HCl used from 3 equivalents to 1.2. Moreover, no use of solvent was needed and the

formation of only water as a by-product was observed.

The isolation of the toxic and unstable lachrymator, an alkylating chloride intermediate,

was circumvented by connecting its synthesis to the synthesis of the corresponding azide.

When the separation was initially needed, a liquid-liquid separator required no addition

of extracting solvent to isolate organohalides and organoazides. The aim of minimizing

solvent use and, thus, energy needed for its later separation was realized. Similarly,

water use was minimized by combining the introduction of a quenching reagent (NaOH)

and a reactant for the subsequent step (NaN3) at once.

When the high-pressure autoclave reactor was used, speed-up of the reaction kinetics

and improvement in regioselectivity were observed. Yield of the desired precursor rose

to 84% in 24 hrs, and resulted in a 1,4:1,5-cycloadduct ratio of 6.3 (30% increase) at 1800

bar with no catalyst present.

A continuous-flow process developed for the formation of the 1,2,3-triazole precursor via

a 1,3-dipolar Huisgen cycloaddition was carried out under catalyst-free and solvent-free

reaction conditions.

RecommendationsAs mentioned above intensive use and waste of materials took place during the optimization

and the development of the process. Incorporation of the inline process analytical tools

would minimize the waste. Moreover, automatization of the equipment control would

minimize the manual labor even further.

8.4. NOVEL PROCESS WINDOWS

HighlightsNPW is relying on several tools introduced in Chapter 1 with the aim of reaching chemical

and process design intensification. When applied at the same time the outcomes are

synergistic, benefitting the process on several levels. Within the carried out research we

observed;

‒ simplification and acceleration of chemical reactions when carried out under high

temperature, pressure and concentration;

‒ when the high-pressure autoclave reactor was used, speed-up of the reaction

kinetics and improvement in regioselectivity were achieved;

Table 1 shows examples of findings due to the application of NPW tools.

RecommendationsIn order to minimize the use of resources, it is advised to first check the potential of

chemical intensification by carrying a particular reaction under superheated conditions

in microwave. This approach will also provide initial understanding of solvent use and

separation requirements.

146

CHAPTER 8

Table 1. Outcomes of application of NPWs principles.

Novel Process Windows Findings Examples

Chem

ical

Inte

nsifi

cati

on High T Reaction rate acceleration Cycloaddition in 10 min vs >24 hrs

High p Reaction rate acceleration Increase in regioselectivity

Increased solubility of HCl gas in organic phase at high temperatures

30% increase in regioselectivity towards 1,4-cycloadduct

High c Reaction rate acceleration Process simplification

Cycloaddition in 10 min vs >24 hrsCircumvention of intermediate

separation and use of organic solvents in extraction

Proc

ess-

desi

gn in

tens

ifica

tion New chemical

transformationProcess simplification HCl gas used as a reagent

Process simplification and integration

Process simplification Circumvention of intermediate separation and use of organic solvents in

extractionConversion of bulk alcohols to complex

APIsMinimization of water use by combining

the quench of one reaction and introduction of the reagent for the

subsequent reaction.

LIST OF PUBLICATIONS

150

List of publications List of publications

151

LIST OF PUBLICATIONS

Journal Articles1. Borukhova, S., Noël, T, Metten, B, de Vos, E., Hessel, V. (2016). From alcohol to

1,2,3-triazole via a multi-step continuous-flow synthesis of a rufinamide precursor.

Green Chem, DOI: 10.1039/C6GC01133K

2. Borukhova, S., Anastasopoulou, A., Hessel, V. (2017). Economic and environmental

evaluation of integrated continuous and batch manufacturing of rufinamide precursor.

Chem. Eng. Res. Des., (submitted)

3. Ott, D., Borukhova, S., Hessel, V. (2016). Life Cycle Assessment of multi-Step rufinamide

synthesis – from isolated reactions in batch to a continuous reaction cascade. Green

Chem, 18(4), 1096-1116.

4. Borukhova, S., Noël, T, Hessel, V. (2016). Hydrogen chloride gas in solvent-free

continuous conversion of alcohols to chlorides in micro flow, Org. Process Res. Dev.,

20(2), 568-573.

5. Borukhova, S., Noël, T, Hessel, V. (2015). Continuous-flow multi-step synthesis of

Cinnarizine, Cyclizine and a Buclizine derivative from bulk alcohols, ChemSusChem,

9(1), 67-74.

6. Borukhova, S., Seeger, A. D., Noël, T, Wang, Q., Busch, M., Hessel, V. (2015). Pressure

accelerated azide-alkyne cycloaddition: micro capillary vs. autoclave reactor

performance in yielding Rufinamide precursor. ChemSusChem, 8(3), 504-512.

7. Borukhova, S., Noël, T, Metten, B, de Vos, E., Hessel, V. (2013). Solvent- and catalyst-

free Huisgen cycloaddition towards Rufinamide in flow with decision on a greener and

less expensive dipolarophile. ChemSusChem, 6(12), 2220-2225.

Books8. Borukhova, S., Hessel, V. (2018). Microreactor networks. De Gruyter (To be published

in February 2018).

Book Chapters 9. Borukhova, S. and Hessel, V. (2016). Continuous synthesis of active pharmaceutical

ingredients in micro flow, in Continuous Manufacturing of Pharmaceuticals (eds. P.

Kleinebudde, J. Khinast and J. Rantanen), Wiley Inc.

10. Borukhova, S. and Hessel, V. (2013). Micro Process Technology and Novel Process

Windows – Three Intensification Fields, in Process Intensification for Green Chemistry:

Engineering Solutions for Sustainable Chemical Processing (eds K. Boodhoo and A.

Harvey), John Wiley & Sons, Ltd, Chichester, UK.

Conference ContributionsOral Presentations

11. Borukhova, S., Noël, T., Hessel, V. (2015). Continuous conversion of alcohols to

chlorides: an efficient way to prepare active pharmaceutical ingredients, Proceedings

of ECCE 2015, September 31-October 2nd 2015, Nice, France

12. Borukhova, S., Hessel, V. (2015). Applications of Novel Process Windows in micro flow

process technology, CHAINS 2015, December 1-2, 2015, Veldhoven, The Netherlands.

13. Borukhova, S., Noël, T, Metten, B, de Vos, E., Hessel, V. (2014). Rufinamide in flow

– from processing to process design, Proceedings of IMRET13, June 23-25, 2014,

Budapest, Hungary

14. Borukhova, S., Noël, T., Hessel, V. (2013). Chemical and Process-Design Intensification

towards greener synthesis of Rufinamide in flow Proceedings of Microwave and Flow

Chemistry Conference, 20-23 July 2013, Napa, CA, U.S.

15. Borukhova, S., Wang, Q., Hessel, V. (2012). Chemical Intensification, particularly the

Impact of Pressure, on the 1,3-Dipolar Huisgen Cycloaddition to Rufinamide in Flow.

Proceedings of the 20th International Congress of Chemical and Process Engineering,

CHISA 2012, 25-29 August 2012, Prague, Czech Republic

Poster Presentations

16. Borukhova, S., Varas, A. C., Hessel, V., Wang, Q., Watts, P., Wiles, C. (2012).

1,3-Dipolar Huisgen Cycloaddition to Rufinamide and the Impact of Pressure on

Reaction Environment and Kinetic Parameters. Proceedings of the 12th International

Conference on Microreaction Technology, IMRET12, 20-22 February 2012, Lyon, France

17. Borukhova, S., Varas, A. C., Hessel, V., Wang, Q., Watts, P., Wiles, C (2011). Huisgen

cycloaddition to Rufinamide: pressure impact on reaction environment and kinetics.

Proceedings of the 3rd Symposium on Continuous Flow Reactor Technology for Industrial

Application, 2-4 october 2011, Lake Como, Italy

flow reactor networks for integrated synthesis of active … · flow reactor networks for...

Documents