HAZARDOUS REACTIONS

BARTHOLOMÄUS PIEBER, C. OLIVER KAPPE* *Corresponding author Institute of Chemistry, University of Graz, NAWI Graz, Heinrichstrasse 28, A-8010 Graz, Austria

Bartholomäus Pieber Oliver Kappe Taming “forbidden“ olefin reductions using hydrazine and oxygen by continuous flow technology

KEYWORDS: Flow chemistry, diimide, hydrazine, oxygen, olefin reduction.

One of the rare alternative reagents for the reduction of carbon−carbon double bonds is diimide which Abstract can be generated in situ from hydrazine hydrate and oxygen. While this selective method is extremely clean and powerful it is rarely used as the rate-determining oxidation of hydrazine in the absence of a catalyst is relatively slow using conventional batch protocols. A continuous high-temperature/high-pressure methodology dramatically enhances the initial oxidation step, at the same time allowing for a safe and scalable processing of the hazardous reaction mixture. The continuous strategy not only allows the reduction of simple alkenes but can also be applied in semi-synthetic drug manufacturing as shown for the production of the artemisinin precursor dihydroartemisinic acid and a synthetic protocol towards hydrocodone.

INTRODUCTION ineffi cient using standard batch conditions (8). Another drawback of this synthetic strategy is the fact that diimide The reduction of unsaturated carbon-carbon bonds is is prone to several undesired side-reactions (Scheme 1).

arguably among the most important synthetic transformation On one hand, an over-oxidation process of the unstable N2H2 known. In general, the majority of these reactions are carried intermediate with either O2 or H2O2 - generated during the out by using hydrogen gas and heterogeneous transition initial oxidation - results in the formation of nitrogen and water

metal catalysts such as e.g. Pd/C, PtO2, Rh/C or Raney (9). On the other hand, disproportionation of N2H2 leads to Nickel (1). Alternatively, homogeneous transitions metal the formation of hydrazine and nitrogen. Thus, the diimide complexes can be utilized which also accesses asymmetric precursor is usually added in high excess for a quantitative hydrogenation reactions (2). However, in certain cases these consumption of the unsaturated starting material. common strategies are accompanied by severe selectivity problems as several undesired side-reactions such as e.g. hydrogenolysis of protecting groups, reduction of other functionalities, alkene migration or racemization can occur.

One of the few alternatives is the use of diimide (N2H2) which has been used as transfer hydrogenation agent in synthetic organic chemistry for more than a century (3). This extremely labile hydrogen donor predominantly reduces unpolarized carbon−carbon double bonds avoiding the above mentioned side-reactions. Diimide is usually prepared in situ by decarboxylation of dipotassium azodicarboxylate or from sulfonylhydrazides via thermal decomposition or a base induced elimination process (3). A more straightforward and Scheme 1. Olefi n reduction by in situ generation of diimide atom economic pathway is the oxidation of hydrazine using, from N2H4 and O2.

e.g., H2O2, NaIO4, K3[(FeCN)6] or even molecular oxygen (3). In the latter case several catalytically active species such as Cu (3), or Fe salts (4), guanidine derivatives (5), fl avin-based A key feature of the olefin reduction via in situ generation

catalysts (6), or even visible light (7) were studied to enhance of diimide from N2H4 and O2 is the fact that theoretically the reaction rate of the initial oxidation step which is very H2O, N2 and H2O2 are the only byproducts.

38 Chimica Oggi - Chemistry Today - vol. 34(3) May/June 2016 Since the latter is a relatively strong oxidation agent it can and molecular oxygen. Our basic hypothesis was that the be suspectedthat it quantitatively reacts with hydrazine or typical high surface-to-volume areas in a biphasic reaction diimide resulting in negligible amounts in the fi nal reaction mixture in continuous fl ow should dramatically enhance the mixture. It should be kept in mind that H2O2 could also lead oxidation rate of hydrazine hydrate (13, 14). Furthermore, we to selectivity problems by oxidizing other functionalities hypothesize that the additional use of a high-temperature/ of the starting material. A catalyst-free protocol would high-pressure protocol would push the diimide generation to therefore ultimately result in a virtually work-up free transfer its limits offering the possibility for an effi cient and catalyst- hydrogenation process allowing for connecting subsequent free reduction process. The fi rst generation reactor (Scheme 2) downstream processes without tedious purifi cation consisted of a mass fl ow controller (MFC) to regulate the procedures. The economic and environmental advantages oxygen stream and a pump delivering a stream of the using these cheap and readily available reagents in an respective olefi n olefi n and hydrazine hydrate in an organic almost waste-free process therefore represent an interesting solvent (16). After mixing of the liquid and gaseous streams, alternative for transition metal catalyzed processes. However, the resulting segmented fl ow pattern entered a heated it has to be pointed out that hydrazine is rather toxic and residence time unit made of perfl uoroalkoxy polymer. The dangerously unstable especially in its anhydrous form. Even reaction mixture was fi nally cooled in a heat exchanger and the less hazardous hydrazine hydrate should not be used depressurized by passing a backpressure regulating unit. without stringent hazard assessments and proper safety precautions. Oxidations using molecular oxygen or air are generally associated with severe safety risks and process challenges since they are exothermic and the heat of the reaction can be diffi cult to dissipate. As the oxidation of hydrazine is relatively slow at ambient conditions a process utilizing elevated temperatures and pressures in organic solvents would be required. To avoid spontaneous ignition of such reactions mixtures, large scale applications in conventional batch reactors have to be carried out below the limiting oxygen concentration (LOC) by mixing the gaseous oxidant with an inert gas as, e.g. N2 to dilute the oxygen/solvent vapour (10). In 1974, a feasibility study for diimide reductions in the syntheses of pharmaceuticals resulted in the utilization Scheme 2. Two feed continuous fl ow set-up for the reduction of of tosylhydrazide as N H precursor as the hydrazine/ oxygen 2 2 olefi ns with N2H2. route “…is hazardous on an industrial scale since mixtures of hydrazine vapor and oxygen are potentially explosive…”(11). An initial study on the catalyst-free reduction of allylbenzene in various alcohols showed that the solvent choice is - in DIIMIDE GENERATION IN CONTINUOUS FLOW particular at elevated temperatures - of crucial importance for enhancing the rate determining oxidation step. Given the Continuous fl ow technology offers the unique possibility to low boiling nature of some potential target molecules n-PrOH address these safety hazards, concurrently working at high (b.p. 97 °C) was the solvent of choice being a compromise temperature/pressure regimes (“Novel Process Windows”) between reaction rate at higher temperatures and the ability feasible for effi cient oxidation protocols (12-14). Exothermic for smooth solvent evaporation. A subsequent, in-depth reactions are easily controlled by the excellent mass and optimization study resulted in a fi nal, totally catalyst-free heat transfer making this technology an ideal tool to harness transfer hydrogenation process at 100-120°C and a system hazardous chemical processes (13, 14). Importantly, the small pressure of 20 bar within 10-30 minutes by using 4-5 equivalents volumes and channel dimensions minimize the possibility hydrazine monohydrate and 8 equivalents of oxygen. Notably, of propagation of an explosion inside the reactor thereby substrates bearing nitro- functionalities, protecting groups and tremendously broadening the possible operation range (15). aryl bromides can be selectively reduced under the applied Another advantage of this enabling technology is that scaling conditions. In most cases evaporation of the solvent and is generally considerably easier for a continuous process than careful drying of the crude material resulted in the desired for a batch process. Flow routes developed and optimized in saturated products in analytical purity (16,17). Only in case the laboratory can often be scaled to production quantities of allyl phenyl sulfi de a non-selective transfer hydrogenation with minimal re-optimization and/or without major changes occurred resulting in signifi cant amounts of oxidation products in the synthetic path (13). Numbering-up of fl ow devices or (sulfoxides and sulfones) of both, the olefi n and its saturated scaling-up of the reactor volume increases the throughput, derivative (18). Moreover it has to be stressed that the use of while the performance of the reactor can be largely aldehydes or ketones as starting materials is troublesome as conserved by keeping certain characteristics of the system these functionalities immediately undergo hydrazine and azine constant (“smart dimensioning”). Alternatively, simply running formation in the presence of hydrazine. a reactor for extended periods of time to generate the desired quantities of pharmaceutical intermediates or fi nal products is often an acceptable strategy. APPLICATION 1: SEMI-SYNTHESIS OF ARTEMISININ Based on these safety and process benefi ts we evaluated the feasibility of continuous fl ow technology for the reduction Subsequently, we aimed to apply this concept for an of olefi ns by generating diimide from hydrazine hydrate industrially relevant example to demonstrate the high

Chimica Oggi - Chemistry Today - vol. 34(3) May/June 2016 39 potential of our continuous methodology. A potential hydrazine stoichiometry of 5 equivalents (2+1+1+1) and just application of diimide chemistry is the diastereoselective two equivalents of O2 are suffi cient for a complete and reduction of artemisinic acid (AA) resulting in highly selective reduction of AA within 37 minutes at 60 °C. dihydroartemisinic acid (DHAA), which can be further Moreover, crystallization afforded the desired artemisinin converted to the antimalarial drug artemisinin (19-21). precursor in high yield (≥ 93 %) containing only 2 % of the Importantly, researchers from Sanofi -Aventis recently over-reduced tetrahydroartemisinic acid and a diastereomeric presented a batch procedure using diimide, generated in situ ratio of ≥97:3. A comparison of the continuous methodology from hydrazine hydrate and O2 (Scheme 3) (21). Due to safety with the batch process demonstrates that the major reasons, the highest admissible oxygen concentration allowed advantage is the relatively short reaction time of less than 40 in the presence of a fl ammable solvent (i-PrOH) was 5% O2 in minutes instead of 11 h with similar yield and selectivity. N2 (v/v). At 40°C full conversion and a high diastereoselectivty (≥97:3) was achieved after 11 h using only 3 equivalents of Due to the relatively clean reaction and the inert nature of

N2H4 ∙ H2O (22). the main byproducts (H2O, N2), we expect that the crude reaction mixture can be directly converted to the antimalarial artemisinin since also a crude plant extract from Artemisia annua could be processed in the continuous drug synthesis (20b). In an ideal case the reduction process could be directly coupled with the photochemical methodology as both transformations require the same gaseous reagent (20).

Scheme 3. Diastereoselective reduction of Artemisinic Acid using N2H2 in batch.

Inspired by this protocol, we hypothesized that the diimide reduction of AA to DHAA could be dramatically enhanced by our continuous strategy in a safe manner (23). By applying similar conditions as in the fi rst continuous study ~90% consumption of the starting material were obtained. Neither variations of the oxygen fl ow, nor longer residences times, higher reaction temperatures or more excess of hydrazine monohydrate increased this value. It appears that the competitive over-oxidation and disproportionation of diimide consumes comparably high amounts of the reactive intermediate resulting in insuffi cient reduction rates. We concluded that a multiple injection, mimicking a traditional dropping funnel in batch experiments, could drive the reaction to completion by continuously adding fresh hydrazine hydrate. This methodology would possibly reduce the amount of disproportionation due to a reduced hydrazine/diimide concentration along the continuous flow reactor. Since we realized that under the continuous high-temperature/high-pressure conditions most of the Table 1. Reduction of artemisinic acid in fl ow using multi injection of hydrazine hydrate. N2H4 ∙ H2O is consumed within less than 10 min this methodology also enables the possibility to increase the effective reaction time. The fi rst multi-injection experiment was carried out using two times two equivalents of hydrazine APPLICATION 2 – SEMI-SYNTHESIS OF HYDROCODONE hydrate and two residence time units (2 x 10 mL) at 100°C resulting in an overall reaction time of 18 min (Table 1, entry The semi-synthesis of the non-natural opiod hydrocodone, 2). We ultimately observed full consumption of the starting one of the most prescribed drugs with a steadily increasing material whilst a comparison experiment with a single trend in manufacture and consumption quantities over addition of 4 equivalents of N2H4 ∙ H2O showed an incomplete the past 20 years, is another process which can potentially reaction (entry 1). However, further analysis of the reaction benefi t from the continuous diimide process. The typical mixture revealed that the reaction was not as selective precursor for hydrocodone preparation is codeine which as the batch protocol due to the formation of signifi cant can be easily hydrogenated by transition metal catalyzed amounts of the undesired diastereomer and over-reduced procedures subsequently transformed into the active substrate (22).We hypothesized that this could be a result of pharmaceutical ingredient by an Oppenauer-type the comparably high reaction temperature and stepwise oxidation as illustrated in Scheme 4 (24). Another attractive reduced it to 60°C resulting in an signifi cantly improved starting material for hydrocodone production which can selectivity but lower conversion (entries 3 and 4). Encouraged be directly extracted from the poppy plant is thebaine. The by this promising result we further optimized the multi-injection key transformation in this production route is a selective reactor aiming for a quantitative reaction (entries 5-9). In double bond reduction followed by simple hydrolysis the fi nal process, a total of four liquid feeds using an overall (Scheme 4).

40 Chimica Oggi - Chemistry Today - vol. 34(3) May/June 2016 Scheme 4. Semi- synthetic pathways for hydrocodonesynthesis.

Unfortunately, common hydrogenation procedures suffer from severe selectivity problems in the reduction step due to over-reduction of the diene moiety and hydrogenolysis of the dihydrofuran (25). Therefore, the current method of choice is a transfer hydrogenation procedure utilizing diimide (26, 27). As a result of the severe safety hazards associated with the oxidation of hydrazine, sulfonyl hydrazides such as p-toluenesulfonyl hydrazide are currently used in combination with stoichiometric amounts of a weak base to generate N2H2 for a quantitative and selective reduction (27). Consequently, we decided to study the possible application our safe continuous fl ow strategy in the selective reduction of thebaine (28). In this case, initial experiments using just one liquid feed at 100°C resulted in relatively low conversions for the double bond reduction (~50%). Moreover, more than 50% of both, the reduced and the unreduced alkaloids were further transformed into the corresponding N-oxides. We suspected that the conversion issue can be again solved by multi-injection of hydrazine hydrate and therefore put a focus on the undesired oxidation. Apparently, this side reaction is a result of hydrogen peroxide formation which is in good agreement with the above described observation using sulfi de substrates (Scheme 2). A selective reduction process is theoretically possible at lower temperatures (50°C) but this also resulted in conversion of less than 10%.

Since N-oxides can be easily reduced to amines by e.g. NaBH4, we tried to intensify this process in order to fully convert thebaine to the reduced N-oxide. At temperatures above 120°C the reaction mixture almost exclusively contained the oxidized compounds but the overall amount of double bond reduction did not increase signifi cantly. Experimental as well as computational investigations on the transfer hydrogenation of thebaine N-oxide additionally indicated that its reduction is considerably slower than for thebaine itself. We considered that we may be able to trap the formed hydrogen peroxide in a simultaneous oxidation rather than decomposing the oxidation agent by a catalyst. By screening of several potential anti-oxidants we realized that addition of simple dimethyl sulfi de –which is oxidized to DMSO- in combination with a reduced amount of 2O and a somewhat lower system pressure results in a highly selective reduction process in moderate conversions. As anticipated, a subsequent combination of these fi ndings with the above described multi-injection strategy resulted in ≥95 % conversion by applying four liquid feeds and residence time units with a total of 12 equivalents of the diimide precursor.

Scheme 5. Continuous transfer hydrogenation and batch hydrolysis for the synthesis of hydrocodone from thebaine.

Chimica Oggi - Chemistry Today - vol. 34(3) May/June 2016 41 We could isolate the crude material containing ~8% thebaine 10. Osterberg P. M., Niemeier J. K., Welch C. J., Hawkins J. M., Martinelli by evaporation of the solvent and precipitation in water in J. R., Johnson T. E., Root T. W., Stahl S. S., Org. Process Res. Dev. 19 ~90% yield. As we were neither able to separate the resulting 1537-1543 (2015) compound from the starting material by chromatography nor 11. Grew E. L., Robertson A. A., Reduction of thebaine. US 3812132 (1974). by recrystallization, we decided to directly hydrolyze the crude 12. For an excellent discussion on novel process windows, see: Hessel material in batch mode in order to obtain hydrocodone in up to V., Kralisch D., Kockmann, N., Noël T., Wang Q., ChemSusChem, 6, 81 % overall isolated yield (Scheme 5). 746-789 (2013) 13. For recent reviews on flow chemistry, see: (a) Pastre J. C., Browne D. L., Ley S. V., Chem. Soc. Rev., 42, 8849-8869 (2013); (b) McQuade D. CONCLUSION AND OUTLOOK T., Seeberger P. H., J. Org. Chem., 78, 6384-6389 (2013); (c) Newman S. G., Jensen K. F., Green Chem., 15, 1456-1472, (2014); (d) Wiles C., The in situ generation of diimide for the transfer hydrogenation Watts P., Green Chem., 16, 55-62, (2014); (f) Baumann M., Baxendale of olefins from hydrazine and oxygen can be efficiently carried I. R., Beilstein J. Org. Chem., 11, 1194-1219 (2015). (g) Gutmann B., out in a catalyst-free procedure using a gas-liquid continuous Cantillo D., Kappe C. O., Angew. Chem. Int. Ed., 54, 6688-6728 (2015) flow approach. As illustrative example for potential applications 14. For detailed discussions about aerobic oxidations utilizing continuous flow technology, see: (a) Gemoets H. P. L., Su Y., Shang M., Hessel in API synthesis, the selective reduction of artemisinic acid V., Luque R., Noël T., Chem. Soc. Rev., 45, 83-117 (2016). (b) yielding the direct precursor molecule for the antimalarial drug Pieber B., Kappe C. O., Top. Organomet. Chem., in press, DOI: artemisinin and a transformation of thebaine into the narcotic 10.1007/3418_2015_133 painkiller hydrocodone could be successfully accomplished. 15. (a) Jahnisch K., Hessel V., Lowe H., Baerns M., Angew. Chem. Int. Ed., The main drivers for the continuous approach are on the 43, 406-446 (2004); (b) Inoue T., Schmidt M. A. , Jensen K. F., Ind. Eng. one hand that safety concerns can be significantly reduced Chem. Res., 46, 1153-1160 (2007). when working with a continuous flow (micro-) reactor and, 16. Pieber B., Martinez S. T., Cantillo D., Kappe C. O., Angew. Chem. Int. on the other hand, working at elevated temperatures and, Ed., 52, 52, 10241-10244 (2013). more importantly, at high pressures for improved and highly 17. It has to be mentioned that simultaneously a different continuous intensified oxidation protocols. Thus, the atom efficient diimide route for the generation of diimide by electrophilic amination of generation which was previously “forbidden” can be operated hydroxylamine was published: Kleinke A. S., Jamison T. F., Org. Lett. 15, 710-713 (2013). with this enabling technology with little risk on various scales. 18. It was previously shown that the hydrazine/O2 system can oxidize sulfides and amines under ambient conditions in the presence of a flavin catalyst: (a) Imada Y., Iida H., Ono S., Murahashi S.-I., J. Am. ACKNOWLEDGEMENTS Chem. Soc., 125, 2868-2869 (2003); (b) Iamda Y., Iida H., Ono S., Masui Y., Murahashi S.-i., Chem. Asian J., 1, 136-147 (2006). (c) Imada Research on continuous flow chemistry in our laboratories Y., Tonomura I., Komiya N., Naota T., Synlett, 24, 1679-1682 (2013). for the past decade has been generously supported by the 19. (a) Turconi J., Griolet F., Guevel R., Oddon G., Villa R., Geatti A., Christian Doppler Research Association (CDG) and a variety Hvala M., Rossen K., Göller R., Burgard A., Org. Process. Res. Dev., 18, of industrial partners including Lonza, DPx, Microinnova, 417-422 (2014), (b) Paddon C. J., et al., Nature, 496, 528–532 (2013). ThalesNano, Anton Paar, Eli Lilly, AstraZeneca, Noramco, 20. For continuous flow procedures to convert DHAA into artemisinin and its derivatives, see: (a) Levesque F., Seeberger P. H., Angew. BayerPharma, BASF and Clariant. Chem. Int. Ed., 51, 1706-1709 (2012); (b) Kopetzki D., Levesque F., Seeberger P. H., Chemistry Eur. J., 19, 5450-5456 (2013); (c) Gilmore K., Kopetzki D., Lee J. W., Horváth Z., McQuade D. T., Seidel-Morgenstern REFERENCES AND NOTES A., Seeberger P. H., Chem. Commun., 50, 12652-12655 (2014); (d) Amara Z., Bellamy J. F. B., Horvath R., Miller S. J., Beeby A., Burgard 1. Nishimura S., Handbook of Heterogeneous Catalytic Hydrogenation A., Rossen K., Poliakoff M., George M. W., Nature Chem., 7, 489-495 for Organic Synthesis, Wiley-Interscience, New York, USA (2001). (2015). 2. The Handbook of Homogeneous Hydrogenation, Vol. 1-3, Edited by 21. Feth M. P., Rossen K., Burgard A., Org. Process Res. Dev.,17, 282-293 de Vries J. G., Elsevier C. J, Wiley-VCH, Weinheim, Germany (2007) (2013). 3. (a) Hünig S., Müller H. R., Thier W., Angew. Chem. Int. Ed., 4, 271-280 22. For a detailed mechanistic investigation, see: Castro B., Chaudret R., (1965); (b) Pasto D. J., Taylor R. T., Org. React., 40, 91-155 (1991). Ricci G., Kurz M., Ochsenbein P., Kretzschmar G., Kraft V., Rossen K., 4. Lamani M., Ravikumara G. S., Prabhu K. R., Adv. Synth. Catal., 354, Eisenstein O., J. Org. Chem., 13, 5939-5947 (2014) 1437-1442 (2012). 23. Pieber B., Glasnov T., Kappe C. O., Chem. Eur. J., 21, 4368-4376 (2015) 5. Lamani M., Guralamata R. S., Prabhu K R., Chem. Commun., 48, 24. (a) Black T. H., Forsee J. C., Probst D. A., Synth. Commun., 30, 3195- 6583-6585 (2012). 3201 (2000); (b) Rapoport H., Naumann R., Bissell E. R., Bonner R. M., 6. (a) Imada Y., Iida H., Naota T., J. Am. Chem. Soc., 127, 14544-14545 J. Org. Chem., 15 1103-1107 (1950); (c) Pfister K., Tishler M., Process of (2005); (b) Smit C., Fraaije M. W., Minnaard A. J., J. Org. Chem., 73, preparing dihydrocodeinone. US 2715626 A (1955). 9482-9485 (2008); (c) Marsh B. J., Heath E. L., Carbery D. R., Chem. 25. (a) Leisch H., Carroll R. J., Hudlicky T., Cox D. P., Tetrahedron Lett., 48, Commun., 48, 280-282 (2011); (d) Teichert J. F., den Hartog T., 3979-3981 (2007); (b) Carroll R. J., Leisch H., Rochon L., Hudlicky T., Hanstein M., Smit C., ter Horst B., Hernandez-Olmos V., Feringa B. L., Cox D. P., J. Org. Chem., 74, 747-752 (2009); (c) Carroll R. J., Leisch Minnaard A. J., ACS Catal., 1, 309-315 (2011); (e) Imada Y., Iida H., H., Hudlicky T., Conversion of thebaine to morphine derivatives., US Kitagawa T., Naota T., Chem. Eur. J., 17, 5908-5920 (2011). 7928234 B2, (2011) 7. Leow D., Chen Y.-H., Su Y., Lin Y.-Z., Eur. J. Org. Chem., 7343- 26. Eppenberger U., Warren M. E., Rapoport H., Helv. Chim. Acta, 51, 7352 (2014). 381-397 (1968) 8. (a) Chen H., Wang J., Hong X., Zhou H.-B., Dong C., Can. J. Chem., 27. (a) Grew E. L., Robertson A. A., Reduction of thebaine, US 90, 758-761 (2012); (b) Menges N., Balci M., Synlett, 25, 671-676 (2014); 3812132, (1974); (b) Orr B., Dummitt W. E., Methods for producing (c) Santra S., Guin, J., Eur. J. Org. Chem. 7253-7257 (2015). hydrocodone, hydromorphone or a derivative thereof, US 8399671 9. For a recent mechanistic investigation on the oxidation of hydrazine, B2, (2013). see: Banerjee A., Ganguly G., Roy L., Pathak S., Paul A., Chem Eur. J., 28. Pieber B., Cox P. D., Kappe C. O., Org. Process Res. Dev., 20, 376-385 22, 1216-1222 (2016). (2016)

42 Chimica Oggi - Chemistry Today - vol. 34(3) May/June 2016