Continuous flow reduction of artemisinic acid utilizing multi-injection strategies - Closing the gap towards a fully continuous synthesis of antimalarial drugs

Article abstract of DOI:10.1002/chem.201406439

One of the rare alternative reagents for the reduction of carbon-carbon double bonds is diimide (HN=NH), which can be generated in situ from hydrazine hydrate (N₂H₄·H₂O) and O₂. Although 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. Simple alkenes can be selectively reduced within 10-20 min at 100-120°C and 20 bar O₂ pressure. The development of a multi-injection reactor platform for the periodic addition of N₂H₄·H₂O enables the reduction of less reactive olefins even at lower reaction temperatures. This concept was utilized for the highly selective reduction of artemisinic acid to dihydroartemisinic acid, the precursor molecule for the semisynthesis of the antimalarial drug artemisinin. The industrially relevant reduction was achieved by using four consecutive liquid feeds (of N₂H₄·H₂O) and residence time units resulting in a highly selective reduction within approximately 40 min at 60°C and 20 bar O₂ pressure, providing dihydroartemisinic acid in ≥93% yield and ≥95% selectivity.

Full text of DOI:10.1002/chem.201406439

DOI: 10.1002/chem.201406439
Full Paper
&
Flow Chemistry |Hot Paper|
Continuous Flow Reduction of Artemisinic Acid Utilizing Multi-
Injection Strategies - Closing the Gap Towards a Fully Continuous
Synthesis of Antimalarial Drugs
Bartholomꢀus Pieber, Toma Glasnov, and C. Oliver Kappe*[a]
Abstract: One of the rare alternative reagents for the reduc-
tion of carbon–carbon double bonds is diimide (HN=NH),
which can be generated in situ from hydrazine hydrate
(N2H4·H2O) and O2. Although this selective method is ex-
tremely clean and powerful, it is rarely used, as the rate-de-
termining oxidation of hydrazine in the absence of a catalyst
is relatively slow using conventional batch protocols. A con-
tinuous high-temperature/high-pressure methodology dra-
matically enhances the initial oxidation step, at the same
time allowing for a safe and scalable processing of the haz-
ardous reaction mixture. Simple alkenes can be selectively
reduced within 10–20 min at 100–1208C and 20 bar O2 pres-
sure. The development of a multi-injection reactor platform
for the periodic addition of N2H4·H2O enables the reduction
of less reactive olefins even at lower reaction temperatures.
This concept was utilized for the highly selective reduction
of artemisinic acid to dihydroartemisinic acid, the precursor
molecule for the semisynthesis of the antimalarial drug arte-
misinin. The industrially relevant reduction was achieved by
using four consecutive liquid feeds (of N2H4·H2O) and resi-
dence time units resulting in a highly selective reduction
within approximately 40 min at 608C and 20 bar O2 pressure,
providing dihydroartemisinic acid in ꢀ93% yield and ꢀ95%
selectivity.
Introduction
The advantages of continuous-
flow processing are increasingly
appreciated by
a
growing
number of scientists, from re-
search chemists in academia to
process chemists and chemical
engineers in pharmaceutical
companies. Thus, continuous-
flow technologies are nowadays
more and more used on a rou-
tine basis on laboratory as well
Scheme 1. Production routes to Artemisinin and its derivatives.
as industrial scales for fine
chemical production.[1–3] An intri-
guing recent example of industrial importance is the develop-
ment of a continuous photochemical reactor for the synthesis
of artemisinin from dihydroartemisinic acid.[4]
Artemisinin, a sesquiterpene endoperoxide, is the key com-
ponent of so called artemisinin-based combination therapies
(ACTs), which are currently the standard treatment for malar-
ia.[5] The pharmaceutically active ingredient can be extracted in
low amounts (typically ꢁ1 wt%) from Artemisia annua (sweet
wormwood) but unfortunately, environmental and economic
factors render the amount of artemisinin obtained to be un-
predictable, causing a steadily fluctuating price situation.[6]
These issues can be tackled by semisynthetic approaches start-
ing from potential precursor molecules like artemisinic acid
(AA) or dihydroartemisinic acid (DHAA), which can be also
found in the plant extract (Scheme 1).[7] Furthermore, recent
breakthroughs in synthetic biology have shown that glucose
can be used to generate amorphadiene or even AA in geneti-
cally modified yeast, offering a plant-independent production
strategy.[8,9]
[a] B. Pieber, Dr. T. Glasnov, Prof. Dr. C. O. Kappe
Christian Doppler Laboratory for Flow Chemistry (CDLFC) and Institute of
Chemistry, University of Graz, NAWI Graz
Heinrichstrasse 28, 8010 Graz (Austria)
Fax: (+43)316 3809840
The semisynthetic route to artemisinin is based on a biomim-
etic, photochemical reaction of DHAA which was discovered in
the late 1980s.[10] Central in this chemical transformation is an
E-mail: oliver.kappe@uni-graz.at
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201406439.
Chem. Eur. J. 2015, 21, 1 – 10
1
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
ene reaction of dihydroartemisinic acid with singlet oxygen
(1O2). The photooxidation is subsequently followed by a Hock
cleavage, addition of triplet oxygen, and a series of spontane-
ous condensation reactions. Unfortunately, the photochemical
generation of singlet oxygen proved to be extremely challeng-
ing to scale. In fact, it took industrial scientists more than two
decades to overcome these limitation and recently researchers
from Sanofi–Aventis were able to adapt this chemical route for
industrial artemisinin production.[11] Simultaneously, Seeberger
and co-workers showed that the complete reaction sequence
can be ultimately performed as a single, fully continuous
chemical process.[4] In addition, the same group recently com-
municated a module-based continuous-flow strategy for the
synthesis of dihydroartemisinin, b-arthemeter, b-artemotil, and
artesunate from AA, all ingredients of ACT antimalarials.[12]
Since plant extraction, as well as synthetic biology, provides
AA, a simple and efficient strategy for the diastereoselective re-
duction to DHAA is essential for a productive process. Batch
hydrogenation in the presence of Wilkinson’s catalyst results in
quantitative conversion and good diastereoselectivity (94:6)
after 19 h at 808C at 47 bar hydrogen pressure.[9b] The im-
proved protocol by Sanofi–Aventis provides similar results
within 6 h at room temperature and 22 bar using a ruthenium
catalyst in the presence of triethylamine.[11] Importantly, the
same group recently presented a catalyst-free protocol using
diimide (N2H2), generated in situ from hydrazine hydrate and
O2.[13] Due to safety reasons, the highest admissible oxygen
concentration allowed in the presence of a flammable solvent
(iPrOH) was 5% O2 in N2 (v/v). At 408C, full conversion was
achieved after 11 h using only 3 equivalents of N2H4·H2O. The
surprisingly high diastereomeric ratio (ꢀ97:3) achieved in this
reduction process can be explained mechanistically.[14]
Results and Discussion
Continuous flow concept
The reduction of olefins is traditionally carried out by metal-
catalyzed processes involving hydrogen gas, typically at elevat-
ed pressures in dedicated hydrogenation devices. However, in
certain cases this strategy is accompanied by severe selectivity
problems as several undesired side-reactions, such as hydroge-
nolysis of protecting groups, reduction of other functionalities,
alkene migration or racemization, can occur.[19b] In contrast, in
situ generated diimide is an extremely selective reagent for
the reduction of unsaturated carbon–carbon bonds.[15] The
generation of this highly unstable and reactive azo compound
from simple hydrazine hydrate and oxygen/air is of particular
interest, as both reagents are readily available and of low cost.
In addition, a catalyst-free protocol ultimately results in a virtu-
ally workup-free methodology, as only water and nitrogen gas
are generated as benign chemical byproducts. However, since
this reaction is predominantly carried out in the presence of
a catalyst,[15–19] the oxidation of hydrazine is a rather slow and
inefficient process under standard batch conditions.[21] Notably,
diimide generation using this oxidation protocol results not
only in the reduction of olefins; an over-oxidation process of
the unstable N2H2 intermediate results in the formation of ni-
trogen and water. Furthermore, a disproportionation can cause
the formation of hydrazine and nitrogen. Thus, the diimide
precursor has to be usually added in (high) excess to guaran-
tee a quantitative consumption of the unsaturated starting
material.
Based on this knowledge, we started to design a continu-
ous-flow reactor for the reduction of olefins using this method-
ology. Our initial hypothesis was that the typical high surface-
to-volume areas in a biphasic reaction mixture in continuous
flow should dramatically enhance the oxidation rate of hydra-
zine hydrate.[2] Furthermore, we assumed that the additional
use of a high-temperature/high-pressure protocol (Novel Pro-
cess Windows)[23] would push the diimide generation to its
limits. We thus assembled a two-feed continuous-flow reactor
(Figure 1).
In situ generated diimide has been used as transfer hydroge-
nation agent for more than one century.[15] Several methods
for its generation, such as decarboxylation of dipotassium azo-
dicarboxylate or the thermal decomposition of sulfonylhydra-
zides, are known.[15] However, from an (atom-)economic point
of view, the aerobic oxidation of N2H4 is the most attractive
source. Since this oxidation is generally very slow, several cata-
lysts, such as Cu[15] or Fe salts,[16,17] guanidine derivatives,[18]
flavin-based catalysts,[19] or even visible light,[20] were studied.
The generation of diimide in batch by oxidation of hydrazine
with O2 in the absence of a catalyst is also possible, even
though significantly longer reaction times in combination with
a high excess of hydrazine are required to reduce simple ole-
fins.[21]
As shown in Figure 1, an HPLC pump (P) was used to contin-
uously pump the olefin and hydrazine hydrate dissolved in
a proper solvent. Oxygen was delivered from a standard com-
pressed-gas cylinder and the flow controlled by a mass-flow
controller (MFC). The two reagent streams were combined in
a glass static mixer (GSM) resulting in a segmented flow pat-
Recently, our group reported a catalyst-free continuous-flow
protocol for the selective reduction of olefins to alkanes based
on the highly efficient in situ generation of N2H2 from
N2H4·H2O and O2.[22] Herein, we present the application of this
methodology towards the stereoselective reduction of artemi-
sinic acid to dihydroartemisinic acid under relatively mild con-
ditions, applying a multi-injection continuous-flow platform.
Figure 1. Two-feed continuous-flow setup for the reduction of olefins with
diimide.
&
&
Chem. Eur. J. 2015, 21, 1 – 10
www.chemeurj.org
2
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
tern, which was then allowed to pass a heated residence time
unit (RTU). For that purpose, perfluoroalkoxy tubing (PFA, i.d.=
0.8 mm, o.d.=1.6 mm) was used, which has lower gas permea-
bility than, for example, PTFE. Concerning gas permeability,
stainless steel would be a more suitable option but, from
a chemical point of view, we were concerned that over time
iron oxide could be generated in such a corrosive regime. This
material was recently shown to catalyze the reduction of nitro
and azide groups in the presence of hydrazine and could
therefore cause selectivity problems in certain cases.[17b,24] The
reaction mixture was finally cooled in a heat exchanger and
depressurized by passing a back-pressure-regulating unit (BPR).
choice is of crucial importance for enhancing the rate-deter-
mining oxidation step.
Since this catalyst-free methodology generates only water
and nitrogen as byproducts, finding a convenient workup pro-
tocol was one of the ultimate goals of this study. Given the
low-boiling nature of some potential target molecules, nPrOH
(b.p.=978C) was the solvent of choice, being a compromise
between reaction rate at higher temperatures and the ability
for smooth solvent evaporation. The model reaction was fur-
ther optimized aiming for quantitative consumption of the
starting material (Table 1).
Table 1. Optimization of the reduction of allylbenzene using in situ gen-
erated diimide.[a]
Process intensification, scope, and limitations
Initially, the reduction of allylbenzene was studied using the
above described continuous-flow setup. Since most protocols
involving the generation of diimide from hydrazine and
oxygen use polar, protic solvents, such as ethanol, we decided
to start the process intensification by screening various alco-
hols as reaction media. For that purpose, the model substrate
(0.1m) and 5 equivalents of hydrazine hydrate where dissolved
in different alcohols (methanol, ethanol, isopropanol, n-propa-
nol, n-butanol, n-pentanol) and pumped at a flow rate of
0.4 mLminÀ1. The oxygen stream was set at 40 mLN minÀ1, re-
sulting in a suitable segmented flow pattern at a back pressure
of 20 bar. By using a 10 mL PFA coil, a residence time of
10 min was observed, which appeared to be adequate for
comparing different solvents for the chosen model reaction.
The assessment of the different alcohols at 808C showed
a nearly linear trend with respect to the reaction rate
(Figure 2). This observation seems to directly relate to the
Entry Concentration
N2H4·H2O
[equiv]
Oxidant
T
[8C]
Conversion
[%][b]
[m]
1
2
3
4
5
6
7
8
9
0.1
0.1
0.2
0.4
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
5
5
5
5
5
5
4
3.5
3
3
O2
O2
O2
O2
O2
air
O2
O2
O2
O2
O2
N2
80
60
94
98
>99
>99
53
>99
98
97
98
<1
8
100
100
100
100
100
100
100
100
120
100
100
10
11
12
0
4
[a] Conditions: Allylbenzene and hydrazine hydrate in nPrOH (2 mL),
0.4 mLminÀ1 liquid flow rate, 40 mLN minÀ1 gas flow rate, 20 bar back
pressure. [b] Conversion was determined as GC peak area percent.
When the reaction temperature was increased to 1008C,
a
significantly higher conversion was obtained (Table 1,
entry 2). We further noticed that better conversions could be
achieved by using a higher olefin concentration (Table 1, en-
tries 3–5). The use of a 0.5m solution resulted in an oxygen
stoichiometry of approximately 8 equivalents. Changing the
oxidizing agent to synthetic air (ca. 1.6 equivalents O2) caused
a dramatic reduction in conversion (Table 1, entry 6). Neverthe-
less, it was possible to reduce the amount of hydrazine hy-
drate to 4 equivalents, resulting in a selective and quantitative
reduction of allylbenzene within just 10 min (Table 1, entry 7).
Further attempts to reduce the amount of the diimide precur-
sor gave incomplete reactions, even at higher temperatures,
corroborating the need for a relatively high excess of hydrazine
owing to over-oxidation and disproportionation of diimide
(Table 1, entries 8–10). In the absence of hydrazine hydrate, no
conversion was observed at all (Table 1, entry 11). Since a cata-
lytic hydrazine-mediated reduction of double bonds was also
reported to proceed under inert conditions,[17] a control experi-
ment using N2 instead of O2 was carried out, resulting in a con-
version of only 8%. This can be rationalized by small amounts
of oxygen dissolved in nPrOH. The optimized conditions were
further tested for a range of simple olefins to evaluate the
Figure 2. Solvent screening for the continuous reduction of allylbenzene
using 5 equivalents of hydrazine hydrate in presence of oxygen.
oxygen solubility in this series of alcohols.[25] However, when
the same experiments were carried out at a somewhat lower
temperature (608C) a significant reduction of this phenomenon
was observed. Whereas methanol and ethanol resulted in
roughly 10% conversion of allylbenzene to propylbenzene, re-
actions using the higher-boiling homologues showed starting
material consumptions between 30% and 40%. Therefore, we
assume that, in particular at elevated temperatures, the solvent
Chem. Eur. J. 2015, 21, 1 – 10
www.chemeurj.org
3
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
clearly more a result of the high-temperature/high-pressure
concept than of the high surface-to-volume area obtained by
the segmented flow pattern.
Table 2. Reduction of olefins in continuous flow using hydrazine hydrate
and oxygen.[a]
Efficiency of hydrazine oxidation
The above-described method is limited to olefins of relatively
high reactivity. When less reactive substrates, such as menthol
(2m), were passed through the continuous-flow reactor under
optimized conditions only moderate conversions were ob-
tained (Table 3). Even at higher temperatures (1208C) and
a longer residence time (ca. 30 min, 26 mL coil volume) only
70% of the starting material was consumed. Similar reactivates
could be observed for the carbazole derivative 2n, camphene
(2o) and a stilbene analog (2p).
Table 3. Conversion of olefins of lower reactivity in continuous flow
using hydrazine hydrate and oxygen.[a]
[a] Conditions: Olefin (0.5m) and hydrazine hydrate (4 equiv) in nPrOH,
0.4 mLminÀ1 liquid flow rate, 40 mLN minÀ1 O2 flow rate, 20 bar back pres-
sure, 10 mL PFA coil. [b] 5 equivalents of hydrazine hydrate were used.
[c] conc.=0.33 m. [d] nPrOH/H2O (1:1 v/v) as solvent. [e] Reaction was car-
ried out at 1208C in a 16 mL coil. [f] Reaction was carried out at 1208C in
a 26 mL coil.
scope and robustness of the continuous methodology
(Table 2).
The majority of the chosen simple olefins resulted in quanti-
tative conversion under the optimized reaction conditions
(2a–e, g, i, j). However, in some cases it was necessary to in-
crease the amount of hydrazine hydrate to drive the reaction
to completion (2 f, h). For some less reactive examples an in-
creased reaction temperature in combination with a somewhat
longer reaction time successfully led to the saturated deriva-
tives (2k, l). Noteworthy, neither a nitro-reduction (2c) nor a de-
protection (2e, g) as side reaction was observed. The only ex-
ample where a nonselective transfer hydrogenation occurred
was by using sulfide 2j. In this case, oxidation products (sulfox-
ides and sulfones) of both the olefin and its saturated deriva-
tive were observed. We argue that this is not a result of an
aerobic oxidation, but more likely due to the formation of hy-
drogen peroxide during the hydrazine oxidation process. A
similar finding was made by Imada and co-workers, who re-
ported that the hydrazine/O2 system can oxidize sulfides and
amines under ambient conditions in the presence of a flavin
catalyst.[26] The main limitation of this protocol was the use of
aldehydes or ketones as starting materials, as these functionali-
ties immediately undergo hydrazine and azine formation in the
presence of hydrazine. It is somewhat surprising that ethyl cin-
namate (2k) could be selectively reduced, even though the
ester moiety is prone to hydrazide formation. In fact, the same
reaction in batch under atmospheric conditions (O2 balloon)
resulted mainly in the unreduced hydrazide after 1–2 h at
1208C. An autoclave experiment resulted in selectivity similar
to the continuous experiment under more or less identical
conditions (1208C, 20 min, 10 bar). We can therefore conclude
that the oxidation efficiency in the continuous protocol is
[a] Conditions: Olefin (0.5m) and hydrazine hydrate (4 equiv) in nPrOH,
0.4 mLminÀ1 liquid flow rate, 40 mLN minÀ1 O2 flow rate, 20 bar back pres-
sure, 10 mL PFA coil. Conversions determined as GC-FID peak area per-
cent. [b] Reaction was carried out at 1208C in a 26 mL coil. [c] Toluene/
nPrOH (3:7 v/v) was used as solvent. [d] 0.1 m in toluene/nPrOH (3:2 v/v).
It appears that the competitive over-oxidation and dispro-
portionation of diimide consumes comparably high amounts
of the reactive intermediate resulting in insufficient reduction
rates. This would imply that a total reduction of such com-
pounds would theoretically need an enormous excess of hy-
drazine and therefore also the amount of oxygen gas would
have to be increased dramatically. To overcome this severe lim-
itation, a deeper understanding of the initial oxidation step is
clearly of great importance.
To get a general idea of the efficiency of the initial hydrazine
oxidation and the unwanted (side) reactions of the highly reac-
tive diimide species, we decided to carry out a range of trap-
ping experiments. The underlying idea was that, in a first coil,
O2 and N2H4·H2O are allowed to react and remaining hydrazine
can be subsequently trapped by the addition of benzaldehyde.
If unreacted N2H4·H2O is still in solution after a defined resi-
dence time, this will ultimately lead to the formation of the
corresponding hydrazone, which subsequently undergoes
a condensation to form benzaldehyde azine. Therefore, we
&
&
Chem. Eur. J. 2015, 21, 1 – 10
www.chemeurj.org
4
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Figure 3. Continuous-flow setup for the hydrazine trapping experiments
using benzaldehyde.
modified the original flow set up (Figure 3) by installing
a second liquid feed and an additional residence time unit
(RTU 2) connected by a T-mixer (T).
Initially, we mounted a 2.5 mL coil for the oxidation step at
1008C using a hydrazine hydrate concentration of 0.5m. The
liquid flow rate was set at 0.4 mLminÀ1 and the oxygen stream
at 40 mLminÀ1. After passing the heated reaction zone, a 1m
solution of benzaldehyde was added via the mixing unit at
a flow rate of 0.2 mLminÀ1 and the mixture passed the second
heated zone (1008C, 6.5 mL PFA coil) before cooling and de-
pressurization. GC-MS analysis of the resulting reaction mixture
showed a 38% yield of benzaldehyde and the desired benzal-
dehyde azine in 52% yield. In addition, we identified an azine
formed from benzaldehyde and propanal (10%), which likely
resulted from solvent oxidation (Figure 4A). When the same
experiment was carried out with a 5.5 mL coil (RTU 1), the ma-
jority of hydrazine was consumed as GC analysis showed ben-
zaldehyde as the main analyte (>80%; Figure 4B).
Figure 4. GC-MS chromatograms from the benzaldehyde trapping experi-
ments using a 2.5 (A) or a 5.5 mL coil (B) for the oxidation of hydrazine and
diimide oxidation/disproportionation.
In addition, the same experimental approach was carried
out with 3-nitrostyrene instead of benzaldehyde to estimate
how much of the remaining hydrazine initializes the reduction
of the alkene moiety. The experiment using a 2.5 mL coil re-
sulted in a consumption of only around 10% of the olefin.
Since the corresponding benzaldehyde experiment indicated
significant amounts of residual hydrazine, it can be concluded
that the majority of hydrazine entering the second coil is not
contributing to the olefin reduction. This observation could be
confirmed in the second example (RTU 1: 5.5 mL), where just
trace amounts (ca. 1%) of the reduction product could be de-
tected, although the aldehyde trapping experiments indicated
remaining hydrazine. However, the values have to be taken
with caution, as all experiments contained small amounts of
propanal azine, which was most likely formed in the first coil.
In addition, the aldehyde experiments also showed significant
amounts of benzaldehyde propanal azine, which is probably
a result of propanal hydrazone formed in RTU 1.
well as the subsequent transformation to its pharmaceutically
active derivatives, according to a continuous protocol was al-
ready reported,[4,12] the selective AA to DHAA reduction can be
regarded as the “missing link” towards a fully continuous strat-
egy for the synthesis of such antimalarial drugs. Inspired by
the previous work by Sanofi–Aventis,[13] we hypothesized that
diimide reduction of AA to DHAA could be dramatically en-
hanced by our continuous strategy. Safety concerns in batch
methods prompted the Sanofi–Aventis team to use 5% O2 in
N2 at atmospheric pressure, which is most likely the reason for
the comparably long reaction time (11 h at 408C). The small
volumes and channel dimensions in a continuous-flow (micro)-
reactor minimize possible flame propagation, leading to an in-
herently safer processing of such explosive mixtures, even
under relatively harsh reaction conditions.[27]
We started our optimization study using a slightly modified
protocol (Table 4). The glass static mixer described above was
replaced by a simple T-mixer since we realized that a) the flow
pattern is similar without the active mixing unit, and b) that
the efficiency of the hydrazine oxidation is essentially a result
of the elevated temperature/pressure combination and not of
the mixing or flow pattern achieved.
Reduction of artemisinic acid
The ultimate target of our studies on the in situ generation of
diimide in continuous flow was to establish an efficient proto-
col for the reduction of artemisinic acid to the artemisinin pre-
cursor DHAA. Since the conversion of DHAA to artemisinin, as
By applying similar conditions to those in the previous
study, we obtained 82% conversion of AA, as analyzed by
Chem. Eur. J. 2015, 21, 1 – 10
www.chemeurj.org
5
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Table 4. Reduction of artemisinic acid in continuous flow.[a]
Table 5. Reduction of artemisinic acid in flow using multi-injection of hy-
drazine hydrate.[a]
Entry N2H4 O2 flow rate T [8C] RTU [mL] t [min] Conversion [%][b]
[equiv] [mLminÀ1
]
Entry
N2H4 [equiv]
T [8C]
RTU [mL]
t [min]
Conversion [%][b]
1
2
3
4
4
4
4
4
40 (4 equiv) 100
20 (2 equiv) 100
20 (2 equiv) 120
20 (2 equiv) 100
10
10
10
20
7
11
11
25
82
91
90
92
1
2
3
4
5
6
7
8
9
10
2+2
2+2
2+2
3+3
2+2+2
2+1+1
2+1+1
2+1+1
4+0+0
2+1+1+1
100
80
60
60
60
60
40
25
60
60
2ꢂ10
2ꢂ10
2ꢂ10
2ꢂ10
3ꢂ10
3ꢂ10
3ꢂ10
3ꢂ10
3ꢂ10
4ꢂ10
18
18
18
18
27
27
27
27
27
37
>99(78)[c]
>99(79)[c]
89
95(89)[c]
96(91)[c]
97(91)[c]
80
[a] Conditions: Artemisinic acid (1 mmol) and hydrazine hydrate dissolved
in nPrOH (1 mL), liquid flow rate 0.4 mLminÀ1. [b] Determined as HPLC
peak area percent at 215 nm; relative response factor of AA/DHAA
(4.6:1).
67
83
>99(80)[d]
HPLC-UV/Vis (Table 4, entry 1).[28] However, decreasing the
oxygen flow to 20 mLminÀ1 resulted in a longer residence
time maintaining an excess of the oxidizing agent. Unfortu-
nately, neither an increased residence time nor the use of
a higher reaction temperature gave quantitative formation of
DHAA (Table 4, entries 2–4).
[a] Conditions: Artemisinic acid (1 mmol) and hydrazine hydrate dissolved
in nPrOH (1 mL), liquid flow rate 0.4 mLminÀ1. [b] Determined as HPLC
peak area percent at 215 nm; relative response factor of AA/DHAA
(4.6:1). [c] Determined by 1H NMR spectroscopy using pyridine as internal
standard. [d] Yield of isolated product.
Taking the previous experiments on oxidation efficiency into
consideration we concluded that a multiple injection, mimick-
ing a traditional dropping funnel in batch experiments, could
drive the reaction to completion by continuously adding fresh
hydrazine hydrate. This strategy was originally used in continu-
ous processes involving highly exothermic reactions to im-
prove thermal control.[29,30] We argued that a multiple injection
of hydrazine hydrate would possibly reduce the amount of dis-
proportionation, due to a reduced hydrazine/diimide concen-
tration along the reactor. In addition, this methodology ena-
bles the possibility to increase the effective reaction time,
since we have already shown that, under the continuous high-
temperature/high-pressure conditions, most of the N2H4·H2O is
consumed within less than 10 min. However, from a technical
point of view, this setup is similar to the trapping experiment
described above (Figure 3). Since low flow rates for the addi-
tional hydrazine feeds were required, syringe pumps were
used instead of standard HPLC pumps. The pressure limit of
these devices necessitates a maximum back pressure of 20 bar.
In addition all further optimization experiments were carried
out with a liquid flow rate of 0.4 mLminÀ1 at an AA concentra-
the DHAA yield was only 78%. GC-MS analysis after silylation
revealed that the reaction was not as selective as the batch
protocol,[13] showing a purity of 85% with 7% of the undesired
dihydroartemisinic acid diastereomer (DHAA 2) and 8% of the
over-reduced tetrahydroartemisinic acid (THAA). We assumed
that this selectivity problem is a result of the comparably high
reaction temperature. Therefore, a similar experiment was car-
ried out at 808C, which unfortunately led to similar values
(Table 5, entry 2). A further reduction of the temperature to
608C decreased the reaction rate and an even higher excess of
hydrazine was not enough to fully consume the substrate
(Table 5, entries 3 and 4). However, at this temperature
1
a higher selectivity was obtained according to H NMR spec-
troscopy. Encouraged by this promising result, a third hydra-
zine feed was added and almost quantitative conversion was
obtained (Table 5, entry 5). In addition, the amount of hydra-
zine could be decreased to an overall value of 4 equivalents
(Table 5, entry 6). A further reduction of the reaction tempera-
ture caused a significant reduction of the reaction rate
(Table 5, entries 7 and 8), indicating that the optimal tempera-
ture for high conversions and sufficient selectivity is 608C. Im-
tion of approximately 0.8m and a gas flow rate of 20 mLminÀ1
,
portantly, a control experiment also demonstrated that
corresponding to 2 equivalents of O2 (Table 5).
a single hydrazine addition at the beginning results in compa-
rably low conversion (Table 5, entry 9). Finally, by adding
a fourth liquid feed, AA could be quantitatively reduced within
37 min at 608C using an overall hydrazine stoichiometry of
5 equivalents (2+1+1+1) and just two equivalents of O2
(Table 5, entry 10, and Figure 5). Isolation by crystallization af-
forded DHAA in 80% yield. GC-MS analysis confirmed a highly
selective reduction with small amounts of THAA (ca. 2%) and
DHAA 2 (ca. 1%) as the only byproducts.
The first multi-injection experiment was carried out using
2ꢂ2 equivalents of hydrazine hydrate and two identical resi-
dence time units (2ꢂ10 mL) at 1008C, resulting in an overall
reaction time of approximately 18 min (Table 5, entry 1). Inter-
estingly, we ultimately observed complete consumption of the
starting material, whereas a similar experiment with 4 equiva-
lents of N2H4·H2O and a 20 mL residence time led to an incom-
plete reaction (Table 4, entry 4). This result already strongly
supports our initial hypotheses on the benefits of the multi-in-
jection approach. However, NMR spectroscopy revealed that
Long-run experiments using 8.5 mmol of artemisinic acid re-
sulted in an identical selectivity with slightly lower conversion
&
&
Chem. Eur. J. 2015, 21, 1 – 10
www.chemeurj.org
6
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Figure 5. Optimized multi-injection setup for the reduction of artemisinic acid towards in situ generation of diimide from hydrazine hydrate and oxygen.
(ca. 97% according to HPLC). However, on this scale, crystalliza-
tion afforded the desired artemisinin precursor in significantly
higher yield (ꢀ93%) with only 2% of THAA present and a dia-
stereomeric ratio of ꢀ97:3.
with the photochemical methodology as both transformations
require the same gaseous reagent.[4]
Conclusion
A comparison of the continuous methodology with other
published batch strategies for the reduction of artemisinic acid
demonstrates that the major advantage of this novel protocol
is the relatively short reaction time of less than 40 min
(Table 6). A more detailed inspection reveals that the diastereo-
meric ratio in the diimide protocols is slightly higher compared
to transition metal-catalyzed hydrogenation procedures. From
an economic point of view, it is noteworthy that no precious
metal catalyst is required when N2H2 is used as reducing
agent. Due to the lack of values for isolated yields from the re-
ported procedures, we calculated the space–time-yield based
on quantitative reactions for the ruthenium-catalyzed hydroge-
nation protocol. A comparison of this value shows that the
continuous methodology described herein is clearly superior
over the batch protocols for the reduction of artemisinic acid.
Due to the relatively clean reaction and the inert nature of
the main byproducts (H2O, N2), we assume that the crude reac-
tion mixture can be directly converted to the antimalarial arte-
misinin, since a crude plant extract from Artemisia annua could
also be processed in the continuous drug synthesis.[4a] In an
ideal case, the reduction process could be directly coupled
The in situ generation of diimide for the transfer hydrogena-
tion of olefins from hydrazine and oxygen can be efficiently
carried out in a catalyst-free procedure using a gas–liquid con-
tinuous-flow approach. Simple alkenes can be selectively re-
duced within 10 min in a virtually work-up free procedure.
It could be shown that the oxidation of hydrazine hydrate,
a time consuming step under conventional batch conditions,
can be dramatically enhanced using this enabling technology.
The obtained kinetic information led to the development of
a multi-injection principle applying periodic additions of fresh
hydrazine hydrate. This methodology enables the possibility
for increased effective residence times and can be applied in
cases where less reactive olefins need to be reduced.
As an illustrative example, the selective reduction of artemi-
sinic acid yielding the direct precursor molecule for the anti-
malarial drug artemisinin was successfully accomplished. This
industrially relevant reduction was achieved by using four con-
secutive liquid feeds and residence time units with 2 equiva-
lents of O2, a total amount of 5 equivalents of N2H4·H2O, and
an overall reaction time of approximately 40 min. A compari-
Table 6. Comparison of different strategies for the reduction of artemisinic acid.
Amyris[a]
Sanofi–Aventis[b]
Sanofi–Aventis[c]
This work
Technology
Reducing agent
Catalyst
batch
H2
Batch
H2
batch
N2H2
–
continuous flow
N2H2
–
[e]
[RhCl(PPh3)3] [0.05 mol%]
[RuCl2{(R)-dtbm-Segphos}](dmf)2 [0.01 mol%]
T [8C]
80
25
40
60
P [bar]
t [h]
Yield [%]
d.r.
47
19
22
atm.
11
>90
ꢀ97:3
0.023
20
6
quantitative (not isolated)
95:5
ca. 0.6
ꢀ93
ꢀ97:3
0.56
quantitative (not isolated)
94:6
0.023
Space–time yield [mmolLÀ1 hÀ1
]
–
[d]
[a] Data taken from ref. [9b]. [b] Data taken from ref. [10]. [c] Data taken from ref. [13]. [d] Apace–time-yield cannot be calculated as no reactor volume was
reported. [e] dtbm = (R)-dtbm-Segphos = (R)-(À)-5,5’-bis[di(3,5-di-tert-butyl-4-methoxyphenyl)phosphino]-4,4’-bi-1,3-benzodioxole.
Chem. Eur. J. 2015, 21, 1 – 10
www.chemeurj.org
7
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Manley, B. Martin, A. O’Kearney-McMullan, Org. Process Res. Dev. 2013,
17, 1472; c) L. Malet-Sanz, F. Susanne, J. Med. Chem. 2012, 55, 4062.
[2] For extensive treatises on microreactor and continuous-flow technolo-
gy, see: a) Flow Chemistry, (Eds.: F. Darvas, V. Hessel, G. Dorman), De
Gruyter, Berlin, 2014; b) Microreactors in Preparative Chemistry, (Ed.: W.
Reschetilowski), Wiley-VCH, Weinheim, 2013; c) Microreactors in Organic
Synthesis and Catalysis, 2n Ed. (Ed.: T. Wirth), Wiley-VCH, Weinheim,
2013; d) Handbook of Micro Reactors (Eds.: V. Hessel, J. C. Schouten, A.
Renken, Y. Wang, J.-i. Yoshida), Wiley-VCH, Weinheim, 2009; e) Chemical
Reactions and Processes under Flow Conditions (Ed.: S. V. Luis, E. Garcia-
Verdugo), RSC Green Chemistry, 2010.
[3] For recent reviews on flow chemistry, see: B. Gutmann, D. Cantillo, C. O.
Kappe, Angew. Chem. Int. Ed. 2015, DOI: 10.1002/anie.201409318;
Angew. Chem. 2015, 127, 10.1002/ange.201409318; a) K. F. Jensen, B. J.
Reizmana, S. G. Newman, Lab Chip 2014, 14, 3206; b) C. Wiles, P. Watts,
Green Chem. 2014, 16, 55; c) S. G. Newman, K. F. Jensen, Green Chem.
2013, 15, 1456; d) I. R. Baxendale, L. Brocken, C. J. Mallia, Green Process.
Synth. 2013, 2, 211;e) I. R. Baxendale, J. Chem. Technol. Biotechnol. 2013,
88, 519; f) N. G. Anderson, Org. Process Res. Dev. 2012, 16, 852; g) C.
Wiles, P. Watts, Green Chem. 2012, 14, 38.
son with other published procedures for this reduction shows
a >20 fold-higher space–time yield for the continuous process
described herein.
Experimental Section
Reduction of simple olefins in a single liquid-injection reactor
(Table 2): Feed A consisted of the respective olefin and hydrazine
monohydrate dissolved in n-propanol, whereas feed B was oxygen
(purity 5.0). The liquid stream (0.4 mLminÀ1) and the gaseous
stream (the flow was set at 40 mLminÀ1 (standard conditions) at
the instrument, resulting in a calculated flow of 2 mLminÀ1 at
a back pressure of 20 bar) were mixed together in a glass static
mixer. The resulting segmented flow stream was passed through
a PFA reactor coil (0.8 mm inner diameter, 10 mL reactor volume, if
not stated otherwise) at 1008C (if not stated otherwise). Subse-
quently, the mixture was cooled in a heat exchanger with water as
cooling agent. After passing a back pressure regulator (20 bar) the
solution was collected. For isolating the title compounds the sol-
vent was evaporated under reduced pressure and the product was
dried overnight (workup A). In certain cases, the reaction mixture
was concentrated, filtered through a plug of silica, and eluted with
copious amounts of CHCl3. The solvent was evaporated under re-
duced pressure and the product was dried overnight using a desic-
cator with CaCl2 (workup B).
[4] a) D. Kopetzki, F. Levesque, P. H. Seeberger, Chem. Eur. J. 2013, 19, 5450;
b) F. Lꢃvesque, P. H. Seeberger, Angew. Chem. Int. Ed. 2012, 51, 1706;
Angew Chem. 2012, 124, 1738.
[5] World Health Organization. World Malaria Report 2013, (WHO, Geneva,
2013).
[6] R. van Noorden, Nature 2010, 466, 672.
[7] For recent reviews and highlight articles about the semisynthesis of ar-
temisinin, see: a) M. A. Corsello, N. K. Garg, Nat. Prod. Rep. 2015, DOI:
10.1039/C4NP90042A; b) D. S. Barnett, R. K. Guy, Chem. Rev. 2014, 114,
11221; c) A. Kumar, D. Paliwal, D. Saini, A. Thakur, S. Aggarwal, D. Kau-
shik, Eur. J. Med. Chem. 2014, 85, 147; d) M. Peplow, Science 2013, 494,
160; e) K. B. Booker-Milburn, Nat. Chem. 2012, 4, 433; f) K. Kupfer-
schmidt, Science 2012, 336, 798.
[8] P. J. Westfall, D. J. Pitera, J. R. Lenihan, D. Eng, F. X. Woolard, R. Regentin,
T. Horning, H. Tsuruta, D. J. Melis, A. Owens, S. Fickes, D. Diola, K. R. Ben-
jamin, J. D. Keasling, M. D. Leavell, D. J. McPhee, N. S. Renninger, J. D.
Newman, C. J. Paddon, Proc. Natl. Acad. Sci. USA 2012, 109, E111.
[9] a) D.-E. Ro, E. M. Paradise, M. Ouellet, K. J. Fisher, K. L. Newman, J. M.
Ndungu, K. A. Ho, R. A. Eachus, T. S. Ham, J. Kirby, M. C. Y. Chang, S. T.
Withers, Y. Shiba, R. Sarpong, J. D. Keasling, Nature 2006, 440, 940;
b) C. J. Paddon, P. J. Westfall, D. J. Pitera, K. Benjamin, K. Fisher, D.
McPhee, M. D. Leavell, A. Tai, A. Main, D. Eng, D. R. Polichuk, K. H. Teoh,
D. W. Reed, T. Treynor, J. Lenihan, H. Jiang, M. Fleck, S. Bajad, G. Dang,
D. Dengrove, D. Diola, G. Dorin, K. W. Ellens, S. Fickes, J. Galazzo, S. P.
Gaucher, T. Geistlinger, R. Henry, M. Hepp, T. Horning, T. Iqbal, L. Kizer, B.
Lieu, D. Melis, N. Moss, R. Regentin, S. Secrest, H. Tsuruta, R. Vazquez,
L. F. Westblade, L. Xu, M. Yu, Y. Zhang, L. Zhao, J. Lievense, P. S. Covello,
J. D. Keasling, K. K. Reiling, N. S. Renninger, J. D. Newman, Nature 2013,
496, 528.
Reduction of artemisinic acid in a multi-injection flow reactor:
Feed A consisted of n-propanol, whereas feed B was oxygen
(purity 5.0). Artemisinic acid (8.5 mmol, 1.99 g) and hydrazine hy-
drate (17 mmol, 0.825 mL) were dissolved in nPrOH (8.5 mL) result-
ing in a total volume of approximately 10.2 mL and injected in
a sample loop which was connected to feed A via a 6-way valve.
The liquid stream (400 mLminÀ1
) and the gaseous stream
(40 mLN minÀ1) were mixed together in a T-mixer. The resulting seg-
mented flow stream was passed through a PFA reactor coil
(0.8 mm inner diameter, 10 mL reactor volume) at 608C. After
coil 1, a T-mixer connected another feed C adding hydrazine hy-
drate Feed in nPrOH (3.33m in nPrOH) at
a flow rate of
100 mLminÀ1. The combined stream passes another 10 mL of PFA
tubing at 608C (coil 2). This addition principle is repeated two
times (feed D, coil 3; feed E, coil 4) resulting in three additional hy-
drazine hydrate feeds and an overall reactor volume of 40 mL. Sub-
sequently, the mixture was cooled in a heat exchanger with water
as cooling agent. After passing a back-pressure regulator (20 bar)
the solution was collected. To isolate the title compounds, the sol-
vent was evaporated under reduced pressure and the product was
precipitated by adding water (3–5 mL) and concentrated H3PO4
until pH <2 was attained. The solids were filtered and dried over-
night, resulting in ꢀ93% yield of a yellow solid with a purity of
ꢀ95%.
[10] a) R. J. Roth, N. Acton, J. Nat. Prod. 1989, 52, 1183; b) R. K. Haynes, S. C.
Vonwiller, J. Chem. Soc. Chem. Commun. 1990, 451.
[11] J. Turconi, F. Griolet, R. Guevel, G. Oddon, R. Villa, A. Geatti, M. Hvala, K.
Rossen, R. Gçller, A. Burgard, Org. Process Res. Dev. 2014, 18, 417.
[12] K. Gilmore, D. Kopetzki, J. W. Lee, Z. Horvꢄth, D. T. McQuade, A. Seidel-
Morgenstern, P. H. Seeberger, Chem. Commun. 2014, 50, 12652.
[13] M. P. Feth, K. Rossen, A. Burgard, Org. Process Res. Dev. 2013, 17, 282.
[14] B. Castro, R. Chaudret, G. Ricci, M. Kurz, P. Ochsenbein, G. Kretzschmar,
V. Kraft, K. Rossen, O. Eisenstin, J. Org. Chem. 2014, 79, 5939.
[15] a) S. Hꢅnig, H. R. Mꢅller, W. Thier, Angew. Chem. Int. Ed. Engl. 1965, 4,
271; Angew. Chem. 1965, 77, 368; b) D. J. Pasto, R. T. Taylor in Organic
Reactions, Vol. 40: Reduction with Diimide, Wiley, Hoboken, 1991, p. 91.
[16] M. Lamani, G. S. Ravikumara, K. R. Prabhu, Adv. Synth. Catal. 2012, 354,
1437.
Acknowledgement
This work was supported by a grant from the Christian Dopp-
ler Research Association (CDG).
[17] For reports on olefin reductions under argon atmosphere employing
N2H4 in the presence of Fe3O4 nanoparticles or Rh-Fe3O4 heterodimer
nanocrystals without further oxidizing agents, see: a) E. Kim, S. Kim,
B. M. Kim, Bull. Korean Chem. Soc. 2011, 32, 3183; b) Y. Jang, S. Kim,
S. W. Jun, B. H. Kim, S. Hwang, I. K. Song, B. M. Kim, T. Hyeon, Chem.
Commun. 2011, 47, 3601.
Keywords: antimalarial drugs
design · reduction · synthetic methods
· flow chemistry · reactor
[1] For a discussion on continuous manufacturing in pharma, see: a) A. M.
Thayer, Chem. Eng. News 2014, 92, 13; b) P. Poechlauer, J. Colberg, E.
Fisher, M. Jansen, M. D. Johnson, S. G. Koenig, M. Lawler, T. Laporte, J.
[18] M. Lamani, R. S. Guralamata, K. R. Prabhu, Chem. Commun. 2012, 48,
6583.
&
&
Chem. Eur. J. 2015, 21, 1 – 10
www.chemeurj.org
8
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
[19] a) Y. Imada, H. Iida, T. Naota, J. Am. Chem. Soc. 2005, 127, 14544; b) C.
Smit, M. W. Fraaije, A. J. Minnaard, J. Org. Chem. 2008, 73, 9482; c) B. J.
Marsh, E. L. Heath, D. R. Carbery, Chem. Commun. 2011, 47, 280; d) J. F.
Teichert, T. den Hartog, M. Hanstein, C. Smit, B. ter Horst, V. Hernandez-
Olmos, B. L. Feringa, A. J. Minnaard, ACS Catal. 2011, 1, 309; e) Y. Imada,
H. Iida, T. Kitagawa, T. Naota, Chem. Eur. J. 2011, 17, 5908.
[20] D. Leow, Y.-H. Chen, Y. Su, Y.-Z. Lin, Eur. J. Org. Chem. 2014, 7347.
[21] a) N. Menges, M. Balci, Synlett 2014, 671; b) H. Chen, J. Wang, X. Hong,
H.-B. Zhou, C. Dong, Can. J. Chem. 2012, 90, 758.
[22] B. Pieber, S. T. Martinez, D. Cantillo, C. O. Kappe, Angew. Chem. Int. Ed.
2013, 52, 10241; Angew. Chem. 2013, 125, 10431.
[23] V. Hessel, D. Kralisch, N. Kockmann, T. Noel, Q. Wang, ChemSusChem
2013, 6, 746.
[24] For the continuous reduction of nitro groups by iron oxide and hydra-
zine, see a) D. Cantillo, M. Baghbanzadeh, C. O. Kappe, Angew. Chem.
Int. Ed. 2012, 51, 10190; Angew. Chem. 2012, 124, 10337; b) D. Cantillo,
M. M. Moghaddam, C. O. Kappe, J. Org. Chem. 2013, 78, 4530; c) M. M.
Moghaddam, B. Pieber, T. Glasnov, C. O. Kappe, ChemSusChem 2014, 7,
3122.
[25] R. Battino, T. R. Retich, T. Tominaga, J. Phys. Chem. Ref. Data 1983, 12,
163.
[26] Y. Imada, H. Iida, S. Ono, S.-I. Murahashi, J. Am. Chem. Soc. 2003, 125,
2868.
[27] M. Hamano, K. D. Nagy, K. F. Jensen, Chem. Commun. 2012, 48, 2086.
[28] It has to be noted that HPLC analysis just shows DHAA and AA, we
could neither separate DHAA from its diastereomer (DHAA 2), nor
detect the over-reduced tetrahydroartemisinic acid (THAA).
[29] J. Haber, M. N. Kashid, A. Renken, L. Kiwi-Minsker, Ind. Eng. Chem. Res.
2012, 51, 1474.
[30] a) P. Barthe, C. Guermeur, O. Lobet, M. Moreno, P. Woehl, D. M. Roberge,
N. Bieler, B. Zimmermann, Chem. Eng. Technol. 2008, 31, 1146; b) D. M.
Roberge, N. Bieler, M. Mathier, M. Eyholzer, B. Zimmermann, P. Barthe, C.
Guermeur, O. Lobet, M. Moreno, P. Woehl, Chem. Eng. Technol. 2008, 31,
1155.
Received: December 11, 2014
Published online on && &&, 0000
Chem. Eur. J. 2015, 21, 1 – 10
www.chemeurj.org
9
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
FULL PAPER
&
Flow Chemistry
B. Pieber, T. Glasnov, C. O. Kappe*
&& – &&
Continuous Flow Reduction of
Artemisinic Acid Utilizing Multi-
Injection Strategies - Closing the Gap
Towards a Fully Continuous Synthesis
of Antimalarial Drugs
The missing link for a totally continu-
ous transformation of artemisinic acid
to a range of antimalarials is the initial
olefin reduction in artemisinic acid. We
demonstrate how this transformation
can be selectively carried out in contin-
uous flow using in situ generated dii-
mide. Key to the success was the devel-
opment of a multi-injection reactor for
the consecutive addition of the diimide
precursor hydrazine hydrate.
&
&
Chem. Eur. J. 2015, 21, 1 – 10
www.chemeurj.org
10
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!