Facile stoichiometric reductions in flow: An example of artemisinin

Article abstract of DOI:10.1021/op200373m

Stoichiometric reduction of artemisinin to dihydroartemisinin (DHA) has been successfully transferred from batch to continuous flow conditions with a significant increase in productivity and an increase in selectivity. The DHA space-time-yield of up to 1.6 kg h^-1 L^-1 was attained which represents a 42 times increase in throughput compared to that of conventional batch process.

Full text of DOI:10.1021/op200373m

Article
pubs.acs.org/OPRD
Facile Stoichiometric Reductions in Flow: An Example of Artemisinin
Xiaolei Fan,† Victor Sans,† Polina Yaseneva,† Dorota D. Plaza,† Jonathan Williams,‡ and Alexei Lapkin*,†
†School of Engineering, University of Warwick, Coventry CV4 7AL, U.K.
‡Department of Chemistry, University of Bath, Bath BA2 7AY, U.K.
ABSTRACT: Stoichiometric reduction of artemisinin to dihydroartemisinin (DHA) has been successfully transferred from batch
to continuous flow conditions with a significant increase in productivity and an increase in selectivity. The DHA space-time-yield of
up to 1.6 kg h−1 L−1 was attained which represents a 42 times increase in throughput compared to that of conventional batch
process.
multistep workup procedure required to obtain a stable
formulation of DHA. These factors were assembled by authors
from heuristic data on end-user experiences with scaling up
artemisinin reduction in UK, India, and China, the earlier project
on artemisinin derivatisation sponsored by Medicines for Malara
Ventures,12 and the material safety data sheets information. On
the basis of our earlier studies of flow processes13−15 we under-
took to develop a flow protocol for the synthesis of dihydro-
artemisinin, which is reported in this paper.
1. INTRODUCTION
Stoichiometric reductions are ubiquitous in organic synthesis
due to their synthetic utility and high selectivity. Many reductions
are performed at low temperatures for the reasons of safety and
maintaining selectivity in the highly exothermic reactions. They
also involve handling of solids. Both factors complicate scale-up of
reductions. In this respect flow chemistry offers potential signif-
icant advantages for the generic class of stoichiometric reductions.
Here we present an example of a typical stoichiometric reduction
of a current drug precursor transferred into a flow process with a
significant gain in productivity and selectivity.
2. EXPERIMENTAL SECTION
Reduction of Artemisinin 1 with NaBH4. Artemisinin
(200 mg, 0.71 mmol) was suspended in methanol (10 mL)
under moderate stirring speed and cooled in an ice−water bath
to ∼4 °C. Sodium borohydride (67 mg, 1.77 mmol, 2.5 equiv)
was added in portions to the suspension over a period of 5 min.
The reaction mixture was stirred vigorously under N2 until
TLC showed no 1 left in the reaction mixture (∼90 min). Then
the reaction mixture was neutralised (pH 5−6) with 50% v/v of
a mixture of acetic acid/methanol (added by portion, 50 μL
each time). The reaction mixture was evaporated to dryness
under vacuum (at 40 °C). This standard procedure was
developed by Buzzi et al.12
The routes to the active ingredients in the artemisinin
combination therapies (ACTs) for the treatment of Plasmodium
falciparum malaria, e.g., artesunate is via reduction of artemisinin
1 into dihydroartemisinin (DHA) 2 using sodium borohydride
in methanol or ethanol, see Scheme 1.1−8 This synthetic
Scheme 1. Reaction scheme of artemisinin reduction using
conventional batch protocol
Workup Procedure. Dry residue was extracted using ethyl
acetate 2, three times (10 mL each time) for transferring the
product completely (monitored by TLC) into ethyl acetate.
The combined ethyl acetate extracts were dried with Na2SO4
(for 6 h), filtered, and evaporated to dryness to give a white
flakelike product.
protocol involves batch reaction at low temperatures (0−5 °C)
with a suspension of NaBH4, followed by a multistep workup
procedure.1−5,7,8 Diisobutylaluminium hydride (DIBAL-H)9−11
is another reductant reported in the literature, requiring dry
dichloromethane as solvent, lower reaction temperature −78 °C
and with a smaller yield of 2. Both protocols require long reaction
times (0.75−3 h) at low temperatures due to the exothermic
nature of the reaction. Scale-up of this reaction is problematic for
four reasons: (i) artemisinin is pyrophoric due to the presence of
a highly unstable endoperoxide function, (ii) batch protocol
requires low temperatures to avoid thermal runaway, (iii) batch
reaction with suspended solids is not easily scalable due to high
sensitivity to variations in the feedstocks and the physical form of
the reactants (size of particles, crystal allotrope, etc.), and (iv) the
Reduction of Artemisinin with LiAlH(OtBu)3. A 1 M
solution of LiAlH(OtBu)3 in THF (2.2 mL) was added drop-
wise (with a syringe) to a solution of 1 (200 mg, 0.71 mmol) in
dry THF (5 mL) stirred under N2 at ∼3 and 40 °C. After two
hours of reaction time the reaction was quenched to pH 5−6
with 20%v/v acetic acid solution in THF. Ethyl acetate and
distilled water were added into the mixture, and the phases
were separated. The aqueous layer was extracted with ethyl
acetate several times. The combined ethyl acetate extracts were
Special Issue: Continuous Processes 2012
Received: December 16, 2011
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op200373m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Table 1. Summary of results of reduction of artemisinin under batch reaction conditions
a
b
entry
reductant
NaBH4
mol equiv/−
T/°C
t/min
solvent
MeOH
conversion of 1/%
yield of 2/%
1
2
2.5
3
4
4
90
100
100
120
60
10
10
10
10
10
10
5
96
98
87
90
89
42
67
46
0
c
3
3
4
97
4
LiAlH(OtBu)3
3
3
THF
65
5
3
40
3
81
6
RedAl
3
Toluene
THF
93
7
LiBHEt3
1
2
96
8
2
2
100
100
100
100
100
100
100
100
100
75
87
94
92
80
81
83
75
46
9
2.5
3
2
10
2
d
11
3
2
12
13
14
2
18
18
19
20
20
2.5
3
5
5
e
15
3
15
5
16
3
a
b
X - conversion determined by HPLC. isolated yield determined by HPLC (the total amount of α- and β-dihydroartemisinin epimers), ratio of α-2
c
d
e
to β-2 was ∼40%/60% for fresh sample by H NMR. Synthesis with 1 g of substrate 1 200 mg of 1 in 5 mL THF. Synthesis with 1.412 g of
substrate 1.
dried with Na2SO4 (for 6 h), filtered, and evaporated to dryness
under reduced pressure to give a white residue.
(a)
Analytical Protocols. TLC, eluents: CH2CL2 +
MeOH (v/v = 20/0.5). Developing agent: phosphomo-
lybdic acid.
Reduction of Artemsinin with Red Al. Red-Al (≥65 wt %
in toluene, 0.65 mL) was added dropwise (with a syringe) to a
solution of 1 (200 mg, 0.71 mmol) in dry toluene (20 mL) stirred
at ∼3 °C under N2. The solution was stirred at 3 °C for 10 min,
and then a 0.01 M aqueous NaOH solution was added dropwise
to stop the reaction. This was followed by the addition of ethyl
acetate. The phases were separated, and the aqueous phase was
back-extracted with ethyl acetate. The combined organic layer was
dried over Na2SO4 (for 6 h), filtered, and evaporated to dryness
under reduced pressure to give a combination of oily residue and
white residue.
Reduction of 1 with LiBHEt3. LiBHEt3 solution (1 M in
THF) was added (with a syringe) dropwise to the solution of 1
in dry THF (200 mg, 0.71 mmol; 20 mL) stirred under N2
(reactions were performed at ∼2 °C and ∼19 °C). The stirring
was continued until TLC showed no 1 left in the reaction mixture
(5−10 min). The temperature of the reaction mixture was
maintained constant by either using water bath or ice−water bath.
An acetic acid solution in THF (20% v/v) was added by portions
(100 or 50 μL each time) to quench the reaction mixture to pH
5−6. The reaction mixture was vacuum evaporated to dryness.
Ethyl acetate and distilled water were used to extract the product
from the dry residue. The two phases were separated, and the
aqueous layer was extracted with ethyl acetate twice. The
combined ethyl acetate extracts were dried with Na2SO4 (for 6 h),
filtered, and evaporated to dryness under reduced pressure to give
a white flake-like product.
(b) H NMR, Bruker ICONNMR. 1H NMR (300 MHz,
CDCl3, 25 °C, TMS): δ = 0.94 (d, J = 6.8 Hz, 6H, CH3),
1.24 (s, 3H, CH3), 1.65 (dd, J1 = 3.41 Hz, J2 = 13.23 Hz,
2H) 1.8−1.9 (m, 4H), 2−2.1 (m, 1H, H), 2.33 (m, 1H),
4.73 (d, J =9.2 Hz, 1H), 5.26 (t, J =3.2 Hz, 1H), 5.37 (s,
1H), 5.58 (s 1H).
(c) HPLC, Shimadzu Prominence instrument equipped with
ELSD-ET (low temperature−evaporative light-scattering
detector), cell temperature 40 °C, polarity +, response
time 1.5 s; column: Thermo hypersil-keystone, 250 mm ×
4.6 mm, 5 μm, BETASIL C18, column oven temperature
45 °C. Mobile phase: acetonitrile/H2O (65/35, %v/v,
flow rate 0.8 mL min−1. Sample: ∼2 mg in 1 mL
acetonitrile each time, 20 μL injection volume.
(d) LC−MS, Dionex 3000RS UHPLC coupled with Bruker
MaXis Q-TOF mass spectrometer, A Sigma Ascentis
Express column (C18, 150 mm × 2.1 mm, 2.7 μm) was
used. Mobile phases consisted of A (water with 0.1%
formic acid) and B (as acetonitrile with 0.1% formic
acid). A gradient of 30% B to 100% B in 15 min was
employed with flow rate at 0.2 mL min−1, UV was set at
220 nm. Mass spectrometer was operated in electrospray
positive mode with a scan range 50−2000 m/z. Source
conditions are: end plate offset at −500 V; capillary at
−4500 V; nebulizer gas (N2) at 1.6 bar; dry gas (N2) at
8 L min−1; dry temperature at 180 °C. Ion tranfer
conditions as, ion funnel RF at 200 Vpp; multiple RF at
200 Vpp; quadruple low mass set at 55 m/z; collision
energy at 5.0 ev; collision RF at 600 Vpp; ion cooler RF
at 50−350 Vpp; transfer time set at 121 μs; prepulse
storage time set at 1 μs. Calibration was done with
sodium formate (10 mM) through a loop injection of
20 μL of standard solution at the beginning of each run.
Flow Synthesis Procedure. The reactor was an XXL-ST-03
by The Little Things Factory GmbH, with reactor internal
volume of 3 mL and incorporating an internal cross-flow heat
exchanger. Solution of artemisinin was pumped with a Kontron
42 HPLC pump. The LiBHEt3 solution was prepared under inert
atmosphere and pumped employing a Knauer 100 HPLC pump.
After the reaction a Y-shaped micromixer was used to quench the
reaction with an acetic acid solution. The pH of the solution was
controlled in the collected aliquots.
3. RESULTS AND DISCUSSION
Batch experiments were carried out first to screen the best
reducing agents for the continuous-flow processes. Due to the
B
dx.doi.org/10.1021/op200373m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
labile nature of the peroxy bond in 1 and the requirement of the
continuous-flow experiments, i.e. high solubility of a reducing
agent, only three candidates were selected on the basis of the
literature,16 namely, lithium tri-tert-butoxyaluminum hydride
(LiAlH(OtBu)3), sodium bis(2-methoxyethoxy)aluminum hy-
dride (NaAlH2(OCH2CH2OMe)2, Red-Al), and lithium triethyl-
borohydride (LiBHEt3).
high chemoselectivity. Further advantages of using LiBHEt3 for
the synthesis of 2 are short reaction time and tolerance to
reaction temperature.
We first examined the published protocol of NaBH4
reduction of 1.12 The highest yield (90%) of 2 was obtained
at 4 °C after 100 min (Table 1, entry 2). LiAlH(OtBu)3 was
reported to be able to reduce a peroxy ester into the cor-
responding alcohol without breaking the peroxy bond.16 A 1 M
solution of LiAlH(OtBu)3 in THF was used in this study. The
best result obtained for LiAlH(OtBu)3 reduction was 67% yield
of 2 within 60 min (entry 5, 40 °C, 81% conversion of 1).
Reaction of 1 with Red-Al was found to be fast at 3 °C:
disappearance of 1 was confirmed by TLC after 10 min. The
yield of 2 was only 46% by HPLC (93% conversion of 1 by
HPLC, entry 6). The mass imbalance was most likely caused by
the formation of byproduct. These byproducts (Scheme 2)
were identified by H NMR and LC−MS analysis.
Figure 1. Schematic diagram of the continuous flow rig for
stoichiometric reductions.
Experiments under flow conditions were performed in a rig
shown schematically in Figure 1. Results are summarized in
Table 2. Flow experiments showed very high conversion and
Scheme 2. Identified byproducts in the reduction of 1 to 2
Table 2. Summary of results of reduction of artemisinin
under flow conditions
residence
time/min
conversion of
yield of
b
a
entry
solvent
THF
T/°C
1/%
2/%
1
2
5
5
99
99
99
99
99
99
98
97
96
97
97
98
98
98
98
98
97
95
93
95
94
94
2
1
3
2
25
25
25
15
0
4
0.5
1
5
6
1
7
1
The two byproducts (3 and 4) have been well-documented
for the process of reducing 1 to 2. The glycal 3 (anhydro-
dihydroartemisinin, MS: m/z (% intensiy) 267.1588 [MH+]
8
2-MeTHF
0.5
0.5
0.33
0.33
25
5
9
10
11
25
5
(calcd 267.1591); 284.1855 [MNH4 ] (calcd 284.1856))17,18 is
+
a
b
the dehydration product of 2. By-product 4 (MS: m/z (%
intensiy) 297.2046 [MH+] (calcd 297.2036) is a product of
fragmentation of 2 under reductive conditions.5,18 The two side
reductions may be attributed to bad mixing (uneven distri-
bution of reactants and distribution of residence time) and poor
temperature control in the batch reactor. Another byproduct
(trace amount) with a molecular mass of 143.1073 (MS
calculated formula: C8H15O2) was also detected by LC−MS.
This byproduct has previously been identified by us during
the accelerated stressing of 1 and is yet to be structurally
characterised.
LiBHEt3 was found to be more effective in terms of reaction
time and yield. The substrate was found to be fully consumed
after 10 min of reaction at 2 °C. The yield of 2 was found to
depend on the molar equivalent of LiBHEt3 (entries 7−10).
The product 2 could be produced in 94% yield by using 3 mol
equiv of LiBHEt3. The tolerance of synthesis of 2 with LiBHEt3
to temperature was also examined. Lower yield of 2 at ∼20 °C
was obtained after 5 min reaction (entries 15). Apart from the
workup procedure (described in ESI), a direct water precip-
itation method was also tested for quenching the reaction
(entry 16). However, only 46% yield of 2 was recovered after
the workup. Among the reductants examined, LiBHEt3
demonstrated the advantages of excellent reducing power and
Calculated from NMR and HPLC data. Yield means the total
amount of α- and β-dihydroartemisinin epimers, 2, was confirmed by
MS and H NMR; yields were determined by HPLC, α-2/β-2 ≈ 45%/
55% by H NMR. Reaction conditions: artemisinin 0.033 M, LiBHEt3
0.1 M, acetic acid 20%.
selectivity under all experimental conditions tested. The residence
times as low as 30 s were enough to produce 2 quantitatively
when THF was employed as a solvent. This represents a
reduction of at least 1 order of magnitude in reaction time
compared to batch experiments. The reaction was performed at
room temperature, thus reducing the overall energy intensity
compared to the traditional reduction protocols, requiring
cooling.
The replacement of traditional solvents by less toxic and
more environmentally benign solvents is a key principle of
green chemistry.19 We studied the replacement of THF by a
biomass-derived alternative solvent.20 2-Methyltetrahydrofuran
is synthesized in two steps from 2-furaldehyde, a chemical
obtained from agricultural waste.21 It has Lewis base properties
and polarity between that of THF and dimethyl ether. LiBHEt3
showed a good solubility in MeTHF, and the reaction was
conducted in flow (Table 2, entries 8−11). Conversion was
found to be slightly lower than that with THF (compare entries
C
dx.doi.org/10.1021/op200373m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(7) Galal, A. M.; Gui, W.; Slade, D.; Ross, S. A.; Feng, S. X.;
Hollingshead, M. G.; Alley, M. C.; Kaur, G.; ElSohly, A. A. Bioorg. Med.
Chem. 2009, 17 (2), 741−751.
8 and 4). Traces of artemisinin were found under these
conditions by HPLC and NMR, even though conversion was
found to be very high (>96%). An even lower residence time of
20 s (entries 10, 11) showed similar degrees of conversion and
selectivity, indicating fast kinetics of the reaction. A small
amount of side products from rearrangement or over-reduction
(<1%, 3 and 4) was found by HPLC. Therefore, one can see
that, in comparison with batch reactors, the well-controlled
reaction conditions in the microflow reactors can significantly
reduce the possibility of forming byprodcuts.
On the basis of the reaction data we evaluated basic reaction
mass metrics of the benchmark batch reaction and of the flow
reaction using Me-THF solvent. Atom economy is better for
the process using sodium borohydride due to its lower mass,
i.e., 0.89 against 0.73 for the flow process using LiBHEt3.
However, energy intensity and the life cycle data are required
for more detailed comparison. Full life cycle assessment of the
new process as well as further optimisation of the flow protocol
(solvent replacements and new reducing agents) are underway.
̀
(8) Jaziri, M.; Dialloa, B.; Vanhaelena, M.; Homesb, J.; K., Y.;
Shimomura, K. Phytochemistry 1993, 33 (4), 821−826.
(9) Avery, M. A.; Mehrotra, S.; Johnson, T. L.; Bonk, J. D.; Vroman,
J. A.; Miller, R. J. Med. Chem. 1996, 39 (21), 4149−4155.
(10) Avery, M. A.; Bonk, J. D.; Mehrotra, S. J. Labelled Compd.
Radiopharm. 1996, 38 (3), 249−254.
(11) Posner, G. H.; Paik, I.-H.; Sur, S.; McRiner, A. J.; Borstnik, K.;
Xie, S. J.; Shapiro, T. A. J. Med. Chem. 2003, 46 (6), 1060−1065.
(12) Buzzi, S.; Presser, A.; Freyhold, M. Determining a viable protocol
for the derivatisation of artemisinin into dihydroartemisinin; A report
commissioned by MMV, 2007, available from www.mmv.org.
(13) Lapkin, A.; Plucinski, P. K., Engineering factors for efficient flow
processes in chemical industries. In Chemical Reactions and Processes
under Flow Conditions, Luis, S. V., Garcia-Verdugo, E., Eds.; Royal
Society of Chemistry: Cambridge, 2010; pp 1−43.
(14) Bavykin, D. V.; Lapkin, A. A.; Kolaczkowski, S. T.; Plucinski, P. K.
Appl. Catal., A 2005, 288, 165−174.
(15) Fan, X.; Plucinski, P. K.; Lapkin, A. A. Catal. Today 2009, 147S,
S313−S318.
(16) Jin, H. X.; Liu, H. H.; Zhang, Q.; Wu, Y. K. J. Org. Chem. 2005,
70 (11), 4240−4247.
4. CONCLUSIONS
We demonstrated a new flow protocol for stoichiometric reduc-
tions using an example reaction of reduction of artemisinin to
dihydroartemisinin.22 DHA was obtained in high yields using
LiBHEt3 at room temperature. Short residence time and full
conversion attained result in high overall productivity ∼1.60
kg h−1 L−1. A biomass-derived solvent, Me-THF, was successfully
employed to substitute THF in this protocol. The developed
flow protocol for reduction using stoichiometric reducing agents
is likely to have broader applicability in organic synthesis as well
as synthesis of ligands.
(17) Haynes, R. K.; Chan, H.-W.; Cheung, M.-K.; Chung, S. T.; Lam,
W.-L.; Tsang, H.-W.; Voerste, A.; Williams, I. D. Eur. J. Org. Chem.
2003, 11, 2098−2114.
(18) Boehm, M.; Fuenfschilling, P. C.; Krieger, M.; Kuesters, E.;
Struber, F. Org. Process Res. Dev. 2007, 11 (3), 336−340.
(19) Anastas, P. T.; Warner, J. C., Green Chemistry: Theory and
Practice. Oxford University Press: Oxford, 1998.
(20) Aycock, D. F. Org. Process Res. Dev. 2007, 11 (1), 156−159.
(21) Kottke, R. H., Furan Derivatives. In Kirk-Othmer Encyclopedia of
Chemical Technology, 4th ed.; Kroschwitz, J. I., Howe-Grant, M., Eds.;
Wiley: New York, 1998; Vol. Suppl.
(22) Lapkin, A.; Fan, X.; Sans, V.; Williams, J. M. J. The process of
production of artemisinin based APIs in flow. Patent application GB
1115988.6, 2011.
AUTHOR INFORMATION
■
Corresponding Author
*Email: a.lapkin@warwick.ac.uk. Fax: (+44) 24764 18922.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
A sample of artemisinin was kindly provided by Malcolm
Cutler, FSC Development Ltd, UK; the HPLC instrument was
donated by Medicine for Malaria Ventures (MMV, Switzerland).
This study was partially funded by EPSRC Grants EP/F023456,
EP/G028141 and an Impact Grant from University of Warwick.
We thank Dr Lijiag Song (Department of Chemistry, University of
Warwick) for LC−MS analysis and interpretation. The paper is
dedicated to the memory of Dr. Ian Bathurst who, in his role as
Director of Research at MMV, was instrumental in bringing many
malaria-related projects to fruition.
REFERENCES
■
(1) China Cooperative Research Group on quighaosu and its
derivatives. J. Tradit. Chin. Med. 1982, 2 (1), 9−16.
(2) Lin, A. J.; Klayman, D. L.; Milhous, W. K. J. Med. Chem. 1987, 30
(11), 2147−2150.
(3) Brossi, A.; Venugopalan, B.; Dominguez Gerpe, L.; Yeh, H. J. C.;
Flippen-Anderson, J. L.; Buchs, P.; Luo, X. D.; Milhous, W.; Peters, W.
J. Med. Chem. 1988, 31 (3), 645−650.
(4) Ferreira, J. F. S.; Janick, J. Phytochemistry 1996, 41 (1), 97−104.
(5) Sy, L.-K.; Hui, S.-M.; Cheung, K.-K.; Brown, G. D. Tetrahedron
1997, 53 (22), 7493−7500.
(6) Bhakuni, R. S.; Kahol, A. P.; Singh, T.; Khanuja, S. P. S. 2004.
D
dx.doi.org/10.1021/op200373m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX