Integrating medicinal plants extraction into a high-value biorefinery: An example of Artemisia annua L.

Article abstract of DOI:10.1016/j.crci.2013.10.023

A principle of biorefining is extended to medicinal plants with the view of developing a more sustainable business model for biomass producers and extractors. This is demonstrated for Artemisia annua L. currently cultivated or harvested in the wild for ex

Full text of DOI:10.1016/j.crci.2013.10.023

C. R. Chimie 17 (2014) 232–241
Contents lists available at ScienceDirect
Comptes Rendus Chimie
www.sciencedirect.com
´
Full paper/Memoire
Integrating medicinal plants extraction into a high-value
biorefinery: An example of Artemisia annua L.
b
c
d
Alexei Lapkin a, , Eba Adou , Benhilda N. Mlambo , Smain Chemat ,
*
John Suberu b, Alana E.C. Collis e, Andrew Clark e, Guy Barker b,
*
a Department of Chemical Engineering and Biotechnology, University of Cambridge, Cambridge, CB2 3RA, UK
b School of Life Sciences, University of Warwick, Coventry, CV4 7AL, UK
c Department of Chemical Engineering, University of Bath, Bath, BA2 7AY, UK
d Centre de Recherches en Analyses Physico-Chimiques (CRAPC), Alger, Algeria
e Department of Chemistry, University of Warwick, Coventry, CV4 7AL, UK
A R T I C L E I N F O
A B S T R A C T
Article history:
A principle of biorefining is extended to medicinal plants with the view of developing a
more sustainable business model for biomass producers and extractors. This is
demonstrated for Artemisia annua L. currently cultivated or harvested in the wild for
extraction of a single compound, artemisinin, comprising on average 1 wt% dry weight of
the plant biomass. We scaled extraction of artemisinin by a non-toxic to bacterial
fermentation solvent tetrafluoroethane to a 5 L pilot scale. We identified a number of co-
metabolites that could be extracted from the plant along with artemisinin and describe the
multi-step extraction-fractionation sequence that potentially could be transferred to a
large-scale multi-step extraction process. We also show possible routes to higher-value
compounds on the basis of A. annua secondary metabolites, exemplified by the conversion
of flavonoids to monomers.
Received 15 July 2013
Accepted after revision 31 October 2013
Available online 8 January 2014
Keywords:
Biorefinery
High-value biorefinery
Artemisia annua
Artemisinin
Hydrofluorocarbons
Extraction
´
ß 2013 Academie des sciences. Published by Elsevier Masson SAS. All rights reserved.
Bio-pharmaceuticals
1. Introduction
increasing demand for water. It is prudent to accept that at
present we are unable to predict all potential conse-
Efficient use of biomass as a sustainable feedstock for
production of chemicals is an increasingly important topic
of academic and industrial research due to uncertainty
over long-term accessibility of traditional petrochemical
feedstocks and the political drive towards CO2-neutral
products and processes, helping to reduce the emissions of
greenhouse gases. At the same time, the use of biomass as a
feedstock for fuels and chemicals has already caused
considerable controversy through unintentional secondary
effects, such as volatility in food prices (Mexico), an
apparent reduction in biodiversity (Malaysia), as well as an
increased use of fertilizers, affecting nitrogen and phos-
phorus cycles, and the increased demand for irrigation,
quences of a biofeedstocks-based economy, especially
with regards to system-level environmental, social and
economic effects, such as biodiversity, climate change,
food and chemicals supply chains, etc. Within this context
it is highly attractive to consider non-food biomass and
waste biomass as a source of chemical structural diversity
for the production of platform molecules.
An increasing number of demonstration projects are
being initiated targeting the development of integrated
biorefineries (see e.g., www1.eere.energy.gov/biomass/
integrated_biorefineries.html for an interactive map of
different biorefining projects in the US), and several good
reviews and reports were published on the general
principles of biorefining [1,2]. There were also earlier
examples of large-scale integrated biorefineries. Most
notably the acetone-butanol complex in Soviet Union
based on fermentation of agricultural waste [3]. All these
examples are targeting large-scale processes, aimed at
*
Corresponding authors.
E-mail addresses: aal35@cam.ac.uk (A. Lapkin),
guy.barker@warwick.ac.uk (G. Barker).
1631-0748/$ – see front matter ß 2013 Acade´mie des sciences. Published by Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.crci.2013.10.023

A. Lapkin et al. / C. R. Chimie 17 (2014) 232–241
233
production of platform molecules and high-tonnage
products.
blocks could be obtained from biomass. This may also
require a different extraction technology, amenable to
multi-product separation and using cleaner solvents which
leave the residual biomass suitable for further fermenta-
tion. In the biorefining schemes published to-date [1,2],
following the extraction of high-value components, the
residual lignocellulose is subjected to fractionation into
lignin, cellulose and hemicellulose. Further upgrading of
hemicellulose and cellulose is likely to proceed via
fermentation. Fermentation of lignin or direct fermenta-
tion of lignocellulose is also becoming feasible, given the
emergence of high-yielding lignin degraders [6] and
intensive work on dissolution and fermentation of
cellulose [7,8]. However, fermentation of biomass contain-
ing residues of organic solvents, typically used in extrac-
tion, may be prohibitively slow due to their toxicity to bio
cultures. Thus, the use of clean extraction technologies
based on supercritical carbon dioxide and other similar
solvent systems is preferred [9].
In parallel, there exist a well-established industry of
manufacturing high-value products from bio-derived
feedstocks for specific niche applications, such as bio-
pharmaceuticals, flavour and fragrances additives, etc.
Within this industry typically either the whole bio-
material is used (e.g., micro algae as a health food or
component of cosmetic products), or a single compound is
obtained from a plant/organism that is either harvested
wild or cultivated. Waste from these processes is typically
landfilled, used as animal feed or as a fuel in the best case.
The future of this industry critically depends on main-
taining a sustainable supply of biomass and on the market
demand for the bio-derived high-value chemicals. The
latter can be affected by many factors, most notably the
increasing frequency of extreme weather in many agri-
cultural regions, affecting food commodities prices and
availability, an increasing awareness of impact of business
and technology on biodiversity and climate, affecting
purchasing choices.
All interdependent influences described above are
brought together schematically into a hypothetical project
workflow in Fig. 1. Thus, the development of a high-value
biorefinery is a complex interdisciplinary problem requir-
ing involvement of several industries, including some that
are not connected at present.
In the case of bio-pharmaceuticals, typically the content
of the desired metabolite is in the order of 1–30% dry
weight at best and thus between 99 and 70%wt of biomass
(dry weight) is either being put back into the fields as a
source of nutrients, or burned.
A
specific example
In this paper we describe results of an interdisciplinary
project that looked at bringing together some of the pieces
of technology required to build a high-value biorefinery.
The adopted target was to maximise production of the
main target metabolite (P1 in Fig. 1) whilst achieving two
more objectives: (i) isolating several more molecules of
commercial value and (ii) leaving biomass suitable for
downstream fermentation. Thus, the residual biomass
from high-value biorefining should be a feedstock for
large-scale biorefinery, producing bulk chemical inter-
mediates and fuels. The main target product P1 was
artemisinin, whereas additional medium and low-value
products of choice were various monomers.
considered in this study is the extraction of artemisinin
for production of artemisinin-based antimalarial drugs [4].
A typical plant content of artemisinin is between 0.7–
1.5%wt. The main process of extraction is using petroleum
ether or hexane-ethyl acetate mixed solvent [5]. Residual
biomass from this process always contains traces of
hydrocarbon solvents even after prolonged evaporation
and cannot be put back into soil. Its best use is as a fuel for
boilers to reduce the energy intensity of the extraction
process, which is done commercially at some artemisinin
extraction facilities in China.
The single-use of biomass in production of bio-pharma-
ceuticals is a significant drawback of the current processes
from the point of view of process sustainability. Specifically,
in the case of artemisinin a significant price volatility is
induced by unpredictable demand driven by the availability
of public subsidies. An unstable business environment
drives farmers to abandon cultivation of Artemisia annua L.,
which creates further uncertainty over the future avail-
ability of the drug precursor and over its prices. Availability
of the semi-synthetic artemisinin produced by Sanofi from
2012 in part resolves this issue. However, there is still a
considerable demand for artemisinin obtained through
extraction. The hypothesis of this paper is that by increasing
the range of high-value products manufactured from such
plants as A. annua, a dampening effect on prices due to the
increased diversity of the end-products markets will be
observed, resulting in an increased profitability of biomass
production and processing.
The use of plant metabolites to make monomers,
polymers or composite materials is a growing area of
research [10,11]. Recent research has focused upon the use
of vegetable oils [12], carbohydrates [10] and lignin [13]
among others. These natural products may be polymers
themselves (e.g. lignin) or molecules which can be made
into polymers (e.g. vegetable oils). While some natural oils
(e.g. castor oil) can be used directly to produce polymeric
materials [14], others often need chemical manipulation
before they are suitable. For example, epoxidised soybean
oil can be chemically manipulated to hydroxylated
monomers useful in the preparation of polyurethanes
and polyesters [15]. Similarly, acrylates and methacrylates
can be chemically incorporated into vegetable oils to
provide useful surface coatings [1].
The use of flavonoids as potential building blocks for
new monomers has not been extensively studied. Flavo-
noids are important secondary metabolites, often exhibit-
ing anti-oxidant activity [16]. Some flavonoids e.g.,
quercitin, have been suggested to show anti-inflammatory
or anti-cancer activity [17]. However, their use as building
blocks for fine and speciality chemicals is not well studied.
During this study a range of flavonoid derivatives could be
In order to achieve this, it is necessary to identify the
entry points for new high- and medium-value bio-derived
molecules into chemicals supply chain: what new
products could be obtained, what existing products could
be replaced or produced starting from complex
bio-derived molecules, or what new generic synthetic

234
A. Lapkin et al. / C. R. Chimie 17 (2014) 232–241
Fig. 1. A schematic depiction of a high-value biorefinery project development.
obtained from the biorefinery process and one of the
simpler flavonoids was used to show the possible
derivatization to novel monomers.
follows: ratio of solvent to dry biomass was always 10:1
(v:wt), extraction was performed on a 1 L scale in a
Buchi double-jacket glass vessel and in 1 L round-bottom
flasks–employing either of two methods, involving heating
at 40 8C for three hours with mechanical stirring, or
extraction in a sonication bath (PUL 125, Kerry Ultrasonics,
UK) operated at 220 V, 50 Hz for one hour. The ratio of
solvent-to-biomass is taken deliberately large to avoid
equilibrium limitation. The temperature of extraction was
previously optimized and corresponds to typical industrial
conditions [5]. In the US-assisted extraction the duration
was optimized.
Exhaustive extraction of artemisinin to determine its
amount in biomass was performed as follows. A sample of
10 g biomass was extracted with 100 mL ethyl acetate at
50 8C over 6 hours three times with fresh portions of
solvent. Three extracts were combined and evaporated to
dryness in a rotary evaporator. The residue was dissolved
in 20 mL of acetonitrile. This results in precipitation of
waxes. Solution was filtered through a syringe filter
The aim of the present work was to evaluate the
feasibility of switching from conventional petrochemical
extraction of artemisinin to an extraction process using a
clean solvent, leaving no residues that could adversely
affect downstream lignocellulose fermentation. Also, to
evaluate the range of other molecules that could be
extracted in a sequential process from A. annua and to
show some exemplar chemical transformations allowing
to convert some of the new potential chemical precursors
to polymer end-products. This approach exemplifies the
potential future development of high-value biorefineries,
linked with conventional large-scale biorefineries.
2. Experimental
2.1. Chemicals and materials
Hexane, ethyl acetate, acetonitrile were purchased from
Sigma Aldrich and used without further purification.
Tetrafluoroethane (R134a) was purchased from Ineos
Fluor, UK.
(0.2 mm) and injected into HPLC.
Extraction of artemisinin using tetrafluoroethane
(R134a) was performed in a pilot rig consisting of a 30 L
chilled solvent storage tank, a diaphragm pump equipped
with a pressure damper, a Coriolis flow meter, a 5 L
extraction vessel equipped with an electrical tape heater,
2.2. Plant samples
a
45 L jacketed evaporation vessel and two heat
Dried A. annua biomass was obtained from CHEMO
(Argentina), GlaxoSmithKline (Tasmania, Australia), REAP
(Kenya) and SensaPharm Ltd (UK). These materials were
stored in a cool dark condition until required for analysis
and extraction. These plant lines are commercial varieties
obtained as crushed leaves. The companies hold voucher
specimens. However, DNA has also been taken from each
plant in order to confirm by genotyping if required at any
time.
exchangers for condensing the solvent into the storage
vessel. All equipment was rated to 40 bar. The biomass
was packed into the extraction vessel in a nylon mesh bag.
Solvent was fed from the storage tank. Extraction vessel
was initially cooled to condense liquid R134a solvent prior
to starting the pump. Once pumping started the extractor
was heated to the required temperature. At the set
pressure (between 5–18 bar, depending on the extractor
temperature) a throttle valve was open and solvent
circulated through the evaporator, heat exchangers and
condensed back to the solvent storage vessel. Overall
duration of extraction experiments was between 1–8 h.
Experiments with the pumped flow were performed at 25
and 40 8C. Due to experimental scatter associated with
2.3. Extraction of artemisinin and co-metabolites
Extraction of artemisinin from A. annua by conventional
hexane-ethyl acetate solvent mixture was performed as

A. Lapkin et al. / C. R. Chimie 17 (2014) 232–241
235
batch-to-batch variability in the packing density and in the
flow rates the data were averaged. Following the extraction
and complete evaporation of the solvent, the precipitate in
the evaporator vessel was washed with 300 mL of acetoni-
trile. A 0.2 mL aliquot was made up to 1 mL with acetonitrile
and analysed by HPLC. A sample of residual biomass was
quickly washed with acetonitrile at ambient temperature
and the amount of artemisinin in the wash was determined
by HPLC. This comprised the amount of artemisinin
extracted, but not transported out of the biomass bed and
it was added to the amount of artemisinin found in the
evaporator for the purpose of calculating the extraction
efficiency. After this wash, a sample of the residual biomass
was exhaustively extracted with acetonitrile at 60 8C over
12 h to determine the amount of non-extracted artemisinin
in the biomass, and to close the material balance. Material
balance was closed within Æ 5% of the overall extraction
efficiency. Results are expressed as extraction efficiency
defined as the percent of artemisinin recovered after extrac-
tion from the amount of artemisinin in the biomass. SEM
images of the leaves following extraction were obtained using
JEOL 6480 instrument.
ethyl acetate (solvent C) in three separate bottles was used.
Separation was achieved on an Eclipse Zobax C18 column
(150 Â 4.6 mm, 0.5
m
m) at a flow rate of 1 mL minÀ1 over a
30 min run time. Detection was at 470 nm on a photodiode
array detector (PAD) attached to an Agilent 1100 series
HPLC with a quart-pump, auto-sampler and a degasser.
2.4.4.
b-Carotenoid analysis
Extracts were analysed on a Shimadzu Prominence HPLC
equipped with an auto-sampler, degasser, and photodiode
array detector using a literature method [21]. Separation
was achieved on a Phasesep-Partisil C18 (250 Â 4.6 mm,
5 mm) column attached to a C18 guard column maintained
at 30 8C. A solvent system composed of eluent A (acetoni-
trile:methanol, 95:5 v/v) and eluent B (acetonitrile:metha-
nol:ethylacetate, 60:20:20 v/v/v). The eluents were
modified with 0.1% butylated hydroxytoluene (BHT) and
0.05% triethylamine (TEA) respectively. A flow rate of
1 mL minÀ1 was used on a gradient elusion which was as
follows: 100% A 0-5, 100% B 13 min, 100% B 30 min, 100% A
45 min. Detection was at 450 nm on the PDA.
Fractionation of A. annua total extract was performed as
follows. A sample of dry leaves was soaked in ethyl acetate
overnight using two portions of fresh solvent (biomass to
solvent ratio 10:1 wt:v with each fresh solvent portion).
The crude extract was evaporated, re-dissolved in 20%
aqueous ethanol and partitioned first with hexane and
then with dichloromethane. Each solvent fraction was
evaporated to dry residue.
2.4.5. Mass spectrometry analysis
Metabolites in the extracts were profiled based on our
method [22]. Briefly, the system consists of an Acquity
liquid chromatograph (LC) (Waters Corp., Milford, MA,
USA) coupled to an Acquity tandem quadrupole detector
(TQD). The column heater was set at 30 8C and a Genesis1
Lightn C18 column (100 Â 2.1 mm; 4
mm) (Grace, IL, USA).
The mobile phase consisted of A: 0.1% formic acid in water
and B: 0.1% formic acid in acetonitrile used in the following
gradient: 0–7.0 min, 25–98% B; 7–9.5 min, 98% B; 9.5–
10 min, 98–25% B; 10–15 min, 25% B; at a flow rate of
0.4 mL minÀ1. Weak wash solvent was 10% acetonitrile,
2.4. Analytical procedures
2.4.1. Artemisinins analysis
Artemisinin and related metabolites were quantified
using published HPLC protocol [18]. HPLC instrument
(Shimadzu Prominence) equipped with a UV-vis diode
array (SPD-M20A, DAD) and evaporative light scattering
(ELSD, LTII, 350 kPa N2, nebulizer at 40 8C) detectors in line
were used. Mobile phase 65:35% v/v acetonitrile:water
strong and needle wash solvent was
acetonitrile, propan-2-ol, methanol
a
mixture of
and water
(30:30:30:10 v/v/v/v). The TQD was operated with an
ESI interface in positive ionization mode (ESI+). The cone
and desolvation gas flow rates were set at 45 L hÀ1 and
800 L hÀ1, respectively. MS parameters were automatically
defined using Waters IntelliStart1 software for the tuning
and calibration of the TQD and subsequently manually
optimized for artemisinin and related metabolites.
mobile phase. Column: Betasil C18 5
m
m, 250 Â 4.6 mm,
flow-rate 0.8 mL minÀ1, column temperature 45 8C.
2.4.2. Flavonoid analysis
An Agilent 1100 series HPLC instrument (Agilent
Technologies, UK) equipped with a quaternary pump,
auto-sampler, a degasser and a diode-array detector was
used. A modification of the method by Bilia et al. [19] was
employed. Briefly, the isocratic mobile phase consisted of
water adjusted to pH 3.2 by acetic acid (eluent A) and
acetonitrile (eluent B) operated at 50% A and 50% B at a
flow rate of 1 mL minÀ1 for 30 minutes. Separation was on
2.5. Synthesis of higher-value chemical intermediates
Various simple flavone derivatives were synthesized as
potential novel monomers for polyurethane, polyester,
epoxy resin and polyglycerol synthesis. One specific
example was chrysin methacrylate 6. This was co-
polymerized with methyl methacrylate (1:1 molar ratio)
in the presence of AIBN in toluene to furnish a novel
methacrylate polymer.
a
Zobax Eclipse C18 column (150 Â 4.6 mm,
5 mm)
protected by a Zobax C18 guard column and detection
was at 280 nm with an injection volume of 20
m
L.
3. Results and discussion
2.4.3. Xanthophylls analysis
3.1. Conventional extraction and optimisation of extraction
by tetrafluoroethane (R134a)
Extracts were analysed for their xanthopyll content
using a modification of the HPLC method [20]. Briefly, a
mobile phase of 5% methanol in acetonitrile (solvent A),
100% methanol (solvent B) and 0.05% triethylamine in
Table 1 shows extraction efficiencies obtained with a
hexane: ethyl acetate (95:5) solvent mixture using two

236
A. Lapkin et al. / C. R. Chimie 17 (2014) 232–241
Table 1
different extraction methods, conventional soaking with
mechanical stirring and ultrasound-assisted extraction.
The obtained efficiencies of extraction are very high at 91
and 95% for the two methods used. This is considerably
higher than the efficiency of extraction attained normally
in industrial-scale operations, when a static biomass bed is
soaked in hot solvent, or hot solvent is percolated through
a biomass bed [5]. Both, the mechanical agitation and
ultrasound provide a considerably higher mass transfer in
comparison with soaking or percolation-type extractions.
The use of ultrasound marginally increases the efficiency of
extraction. The reason for high efficiency of extraction
under mechanical stirring conditions and, consequently,
only a small improvement due to sonication is due to
effective disruption of trichomes containing artemisinin,
observed in SEM images of leaves post extraction (Fig. 2,
left image). Thus, the poor extraction at larger scales is
mainly due to poor access of solvent to leaves and very
slow mass transfer through a large bed.
Extraction efficiencies of artemisinin from biomass using hexane-ethyl
acetate (95:5) solvent mixture using two extraction methods.
Method of extraction
Artemisinin
in planta
Artemisinin Extraction
extracted
efficiency
(%)
biomass (%wt) (%wt)
Heat for three hours (40 8C) 0.78
with mechanical stirring
0.71
0.74
91
95
Sonication for one hour
0.78
a
Values determined by exhaustive extraction.
terms of running and capital costs. There remained a
question whether sophisticated pumped flow extractor is
required for this process or a more basic and energy
efficient gravity-flow extraction is sufficient. The latter
process has been operated commercially [23]. Thus,
tetrafluoroethane as
technical solution.
a solvent remains an attractive
In this study extraction of artemisinin by R134a was
performed in a 5 L scale vessel in three different modes of
operation: pumped down- and up-flow, and gravity-driven
down-flow. These three cases represent typical potential
industrial extraction modes. The observed trend (Table 2)
correlates well with the expected improvement in mass
transfer. The gravity down-flow is likely to be character-
ized by incomplete wetting of biomass and slow solvent
flow-rate, thus resulting in poor recovery of artemisinin.
The maximum attained efficiency of extraction is for
pumped up-flow, which offers the most intensive mass
transfer within the three flow regimes studied. The
obtained efficiency at 84% is well above what is typically
obtained in hexane (petroleum ether) extraction on the
large scale [5], but is below what we obtained in the small-
scale extraction in hexane with intensive mixing (Table 1).
We have also observed that many of the trichomes were
intact following extraction with R134a (Fig. 2, right image).
This may indicate inhomogeneous wetting of biomass by
R134a solvent in the static biomass bed and sampling in
SEM of a region of a leaf that has seen little contact with the
However, as was mentioned above, extraction by
hydrocarbon solvents leaves biomass unsuitable to fer-
mentation processes due to toxicity of the solvent residues
to fermentation culture. An earlier study assessed eco-
nomic and environmental potential of extracting artemi-
sinin
by
ethanol,
scCO2,
ionic
liquids
and
tetrafluoroethane, as compared to hexane-ethyl acetate
extraction [5]. Extraction by tetrafluoroethane was found
to be the most promising technology among the con-
sidered new alternative processes.
There is a valid concern over the use of tetrafluor-
oethane as a solvent, due to its significant green house gas
potential. The substance is to be withdrawn from the use as
propellant in inhalers and as
a refrigerant, due to
uncontrolled release of the significant green house gas.
However, the volume of solvent used in extraction is
minuscule compared to the volume used in refrigeration
and near-to-complete recovery of used solvent is feasible
(300 pm residual concentration in the biomass is attain-
able with vacuum degassing of spent biomass), which
gives a relatively low environmental impact, comparable
with that of petrochemical solvents. Low operating
temperature and pressure of this process compared to
scCO2 extraction make it significantly more attractive in
solvent. An alternative explanation would imply
a
mechanism of extraction that enhances mass transfer of
artemisinin out of trichomes without disrupting the walls
of trichomes, which we have no evidence for. In any case,
Fig. 2. SEM images of A. annua leaves following extraction with hexane:ethyl acetate (left) and with R134a solvents (right).

A. Lapkin et al. / C. R. Chimie 17 (2014) 232–241
237
Table 2
there clearly is room for further optimisation of extraction
by R134a.
Comparison of extraction efficiency of three different operating regimes
of R134a extraction.
Entry
Extraction mode
Extraction efficiency/%
3.2. Total extraction and fractionation; profile of co-
metabolites
1
2
3
Gravity down-flow
Pumped down-flow
Pumped up-flow
31
46
84
Metabolic profile of A. annua has been extensively
studied [4,24]. However, it varies significantly with the
specific growth conditions and local climatic and soil
conditions. We have investigated this variation by
analysing several groups of metabolites either quantita-
tively or qualitatively in A. annua grown in four different
countries with very different soil and climatic conditions:
Argentina, UK, East Africa and Tasmania.
A significant variation in the relative amounts of
different metabolites in the biomasses grown in different
geographic regions is observed (Table 3). Chemical
structures of various compounds listed in Table 3 are
given in Scheme 1. As far as we have been able to identify,
the four biomasses were grown from genetically identical
materials. Therefore, the differences in metabolic profiles
are due to environmental factors. We did not have access to
the same biomasses over several years of cultivation and
therefore cannot quantify variability within the same
growth region between years. However, we anticipate
such variation also to be significant. Based on this
variability we may speculate that each biomass would
require re-optimization of the extraction-purification
process, since particular composition of all the metabolites
would influence solubilities and separation effectiveness.
Specifically, there are statistically significant variations in
the amounts of some flavonoids and pigments.
Different metabolites could be grouped according to
their polarity and the corresponding type of solvent in
which they are preferentially extracted. Thus, the labora-
tory-scale extraction-fractionation of A. annua was per-
formed with conventional solvents of different polarities.
The representative solvents reflect the broad groups of
compounds of low, intermediate and high polarity, as
shown in Table 4.
In a typical extraction sequence, from 329 g of starting
dry leaf material (Kenyan biomass was used) 29.7 g of
residue was obtained after exhaustive extraction with
ethyl acetate. This was re-dissolved in aqueous ethanol and
partitioned first with hexane, yielding after evaporation
7.5 g of dry residue, and then with dichloromethane,
yielding after evaporation 21.7 g of dry residue. The final
water soluble fraction was 1.02 g.
The hexane fraction contains the primary compound of
interest, artemisinin,aswellasothersimilarco-metabolites,
waxes, oils, carotenoids. The largest group of compounds in
this fraction are waxes which are difficult to separate for
detailed analysis and further use. However, within artemi-
sinin analogues, artemisinic acid, artemisitene and dihy-
droartemisinin were identified and quantified (Table 3).
Deoxiartemisinin is a known impurity which is believed to
be a product of decomposition of artemisinin, whereas
arteannuin H and B maybe of interest as bio-actives. Hexane
fraction also contains a large amount of pigments.
The data in Table 3 also indicate that some metabolites
are either present or not in the plants grown in different
countries. This does affect the universal optimization of
any high-value biorefining system. Specifically, para-
coumaric acid is a potentially useful source of aromatics
that could be used in polymerisation and high-value
consumer products.
Within the dichloromethane fraction the class of
compounds of interest is flavonoids. A large number of
flavonoids were identified in A. annua in the literature [4].
Table 3
Major artemisinin related metabolites and other compounds found in Artemisia annua samples of different geographical origin.
Compounds
Origin of plant sample
Argentina
UK
East Africa
Tasmania
Sesquiterpenes
Artemisinin (mg gÀ1
)
3.28
3.29
4.09
9.69
Artemisinic acid (mg gÀ1
)
0.35
0.63
0.53
1.14
Artemisitene (mg gÀ1
)
0.44
0.22
0.16
0.05
Dihydroartemisinin (mg gÀ1
Deoxyartemisinin
Arteannuin H
)
0.22
0.21
0.48
1.04
Present
Present
Present
Present
Present
Present
Present
Present
Not detected
Present
Present
Not detected
Arteannuin B
Pigments (m )
gÀ1
Xanthophylls
Beta-carotene
2.31
4.92
1.29
4.08
5.75
4.61
30.73
149.89
Flavonoids (mg gÀ1
)
Para-coumaric acid
Casticin
Not detected
1.47
Present
0.42
Not detected
1.45
Present
1.80
Artemetin
0.06
0.03
0.22
0.09
Retusin
0.17
0.09
0.49
0.39

238
A. Lapkin et al. / C. R. Chimie 17 (2014) 232–241
Scheme 1. Molecular structures of compounds identified in A. annua extracts.
We have quantified three flavonoids due to their potential
impact on purification of artemisinin [25].
its pure form and it is structurally similar to some of the
flavonoids identified in A. annua, enabling a parallel to
extraction research programme. Initial studies concen-
trated on directly polymerizing the flavonoids isolated
with di-isocyanates (MDI, TDI and HDI) making use of the
reactive phenolic functionality. However, the relative
insolubility of the flavonoid derivatives under the poly-
merization reaction conditions1 and the differential
reactivity of the phenolic hydroxyl groups precluded the
efficient formation of polyurethane polymers. Conse-
A potential alternative extraction sequence would
involve the initial separation of terpenes by, for example,
steam distillation. In this study we have not investigated
this option, due to the risk of degrading artemisinin, which
is rather unstable in aqueous environment. A rapid vacuum
assisted or MW-assisted steam distillation may, however,
be a viable option to recover volatile terpenes prior to
extraction and fractionation of structurally more complex
compounds.
3.3. Potential for new products from A. annua metabolites
1
Polymerisation was attempted by heating the flavanoids at 110 8C
with the isocyanates in a mixed solvent system of toluene and dry DMF.
Addition of either pyridine or triethylamine as catalyst was also
attempted.
For the chemical derivatization study a simple flavo-
noid chrysin was selected, as it is commercially available in

A. Lapkin et al. / C. R. Chimie 17 (2014) 232–241
239
Table 4
The types of compounds identified in different fractions after sequential extraction of A. annua.
Fraction
Compound category
Specific compounds identified
Hexane
Oils, waxes, carotenoids, triterpenoids
Flavonoids
Artemisinin, arteannuin B, artemisinic acid
Casticin, artemetin, retusin
Tannins
Dichloromethane
Water soluble
Biopolymers, polyphenols
quently, we explored the chemical derivatization of
flavonoids to improve both their solubility and reactivity.
We illustrate here the chemistry with reference to chrysin.
Selective functionalization of the C-7 OH group was
possible with a range of C-4 organic fragments. Alkylation
with K2CO3/1,4-dibromobutane furnished bromide 1 in
86% yield. Hence, reaction of chrysin (10.1 g) with 1.05 eq
of K2CO3 (5.8 g) and 1,4-dibromobutane (9.0 g) in dry DMF
(130 mL, 0.3 M) at room temperature for 24 h gave 1
(13.4 g) as a white solid. The X-ray structure of 1 clearly
shows the intramolecular H-bond between C9-OH and the
C11 C = O which decreases the reactivity of C9-OH
and alkylation using commercially available tosylate 9
(derived from glycerol). [1H NMR (400 MHz, CDCl3)
d 12.72
(1H, s), 7.89 (2H, dd, J = 14.3, 7.7 Hz), 7.60–7.50 (3H, m),
6.67 (1H, s), 6.53 (1H, d, J = 2.2 Hz), 6.39 (1H, d, J = 2.2 Hz),
4.51 (1H, app quin, J = 5.8 Hz), 4.19 (1H, dd, J = 8.5, 5.8 Hz),
4.12 (1H, dd, J = 9.6, 5.8 Hz), 4.03 (1H, dd, J = 9.6, 5.8 Hz),
3.92 (1H, dd, J = 8.5, 5.6 Hz) 51.48 (3H, s), 1.42 (3H, s)].
Simple acid catalysed deprotection of 7 with acetic acid
furnished 8 in 87% overall yield from chrysin. While the
other derivatives 2-7 were soluble in chloroform, glycerol
derivative was less soluble but dissolved readily in
methanol or ethanol. Selective methacryloylation of 8 at
the primary hydroxyl was possible using the method for
formation of 6. Finally, epoxide 11, a potential monomer
useful for polyether synthesis was prepared from tosylated
glycidol. It was found that the thermal stability of the
functionalised monomers was lower than chrysin itself
with thermal gravimetric analysis indicating that the
temperature at which 10% of the monomer decomposed to
be 1 (382 8C), 2 (330 8C), 3 (303 8C), 4 (320 8C), 6 (333 8C), 7
(350 8C), 8 (309 8C), 9 (304 8C) respectively (Scheme 4).
Most methacrylates can be polymerized under free
radical conditions. Flavonoids however are known to be
anti-oxidants and will inhibit free radical reactions [30]. It
was therefore not clear that monomers such as 6 would
readily undergo polymerization. Using the TEAC (Trolox
equivalent anti-oxidant activity) ABTSꢀ+ assay [31] it was
possible to measure the relative anti-oxidant abilities of a
range of C-7 functionalised derivatives with respect to
respective to the C-7 OH. Spectral details for 1: IR (film)/
À1
cm
y
3061, 2963, 2934, 2865, 1649, 1601, 1586; 1H
max
NMR (300 MHz, CDCl3)
d 12.71 (1H, s), 7.90–7.87 (2H, m),
7.58–7.50 (3H, m), 6.68 (1H, s), 6.49 (1H, d, J = 2.2 Hz), 6.36
(1H, d, J = 2.2 Hz), 4.08 (2H, t, J = 6.0 Hz), 3.50 (2H, t,
J = 6.5 Hz), 2.12–2.04 (2H, m), 2.04–1.95 (2H, m); 13C NMR
(75.5 MHz, CDCl3)
d 182.5, 164.8, 164.0, 162.2, 157.8,
131.8, 131.3, 129.1, 126.3, 105.9, 105.7, 98.5, 93.1, 67.5,
33.2, 29.3, 27.6; ESI Found (MH)+, 389.0383, C19H17Br79O4
requires (MH)+, 389.0388 (Scheme 2).
Repeating the reaction with only 0.5 eq of dibromobu-
tane and K2CO3 furnished the dimer 2. After three hours a
precipitate was formed which was re-dissolved in THF and
the solution left overnight. This solution was neutralized
with 2 N HCl and the solid precipitate washed with water
to give 2 [1H NMR (300 MHz, CDCl3)
d 12.71 (2H, br s),
7.89–7.86 (4H, m), 7.55–7.49 (6H, m), 6.66 (2H, s), 6.50
(2H, d, J = 2.2 Hz), 6.38 (2H, d, J = 2.2 Hz), 4.15 (4H, m), 2.05
(4H, m)]. A range of other C-4 fragments ultimately derived
from succinic or maleic acids (themselves renewable
chemicals derived from fermentation of carbohydrates
[26], potentially part of this biorefinery) were incorpo-
rated. In this way it was possible to prepare a range of
symmetrical diol derivatives as potential monomers for
polyurethane or polyester synthesis 2-4. Reaction of
chrysin with succinic anhydride (1.0 eq) in a 1:1 v/v
THF/Et3N mixture furnishes 5, a potential monomer for
polyester synthesis (Scheme 3) [27].
chrysin itself (23.6
phenolic groups lowered the anti-oxidant activity sig-
nificantly, [7 (15.9 M), 8 (14.2 M) and 12 (14.5 M)]
mM). As expected, removal of one of the
m
m
m
and it was possible to polymerise methacrylate 6. Hence,
reaction of a 1:1 molar ratio of methyl methacrylate and
monomer 6 with AIBN in toluene at 70 8C furnished a
polymethylmethacrylate 12 (Mn 7.3 kDa, Mw 14.8 kDa,
PDI = 2.03) incorporating the renewable derived monomer.
Fig. 3 shows the traces using both ultraviolet (UV) and
infrared (IR) detection. The UV trace indicates that
monomer 6 has been incorporated into the polymer.
We also explored the feasibility of incorporating
methacrylate 6, 10 and glycerol groups 7 and 8 into the
flavonoid skeleton. Methacrylate derivatives can be poly-
merized via a free radical process [28], while polyglycerols
have attracted significant attention in recent years [29].
Methacrylate 6 was prepared by acylation of chrysin with
1.0 eq of methacryloyl chloride in pyridine overnight. After
acidification (pH 2) the mixture was extracted with ethyl
acetate to furnish 6 in 89% yield [1H NMR (300 MHz, CDCl3)
4. Conclusions
In conclusion, we have shown that a conventional
extraction of a high-value bio-pharmaceutical could be
effectively performed using a clean solvent extraction,
leaving residual biomass suitable for downstream fermen-
tation in bulk biorefineries. The metabolic profile of the
chosen example biomass was highly variable depending on
the geographic origin of the biomass. This affects
optimization of each extraction-separation process as well
as target compounds for isolation in a potential high-value
biorefinery. We have identified a number of co-metabo-
d
12.73 (1H, s), 7.90 (2H, m), 7.59–7.51 (3H, m), 6.91 (1H, d,
J = 1.1 Hz), 6.74 (1H, s), 6.62 (1H, d, J = 1.1 Hz), 6.39 (1H, s),
5.83 (1H, s), 2.08 (3H, s)]. Derivative 7 was prepared from
chrysin in 98% yield by selective deprotonation with K2CO3

240
A. Lapkin et al. / C. R. Chimie 17 (2014) 232–241
Scheme 2. The product of functionalization of chrysin via C7 position.
Scheme 3. Potential monomers for polyurethane and polyester synthesis.
lites of potential commercial interests and have shown
that one flavonoid output from the biorefinery can be
transformed into a large number of novel building blocks.
These new feedstocks are potential monomers for poly-
urethane, polyester, polymethylmethacrylate, polygly-
cerol and polyether synthesis. Despite the known anti-
oxidant, and radical inhibiting properties of flavonoids, it
was possible to polymerise one derivative using a free
radical strategy.
Acknowledgements
This project was funded by Engineering and Physical
Sciences Research Council (EPSRC) via a Sandpit initiative
Scheme 4. Potential monomers for polymethacrylate, polyglycerol and polyether synthesis.

A. Lapkin et al. / C. R. Chimie 17 (2014) 232–241
241
[3] V.V. Zverlov, O. Berezina, G.A. Velikodvorskaya, W.H. Schwarz, Appl.
Microbiol. Biotechnol. 71 (2006) 587–597.
[4] C.W. Wright, Artemisia, Taylor & Francis, London and New York, 2002.
[5] A. Lapkin, P.K. Plucinski, M. Cutler, J. Nat. Prod. 69 (2006) 1653–1664.
[6] T.F.H. Bugg, M. Ahmad, E.M. Hardiman, R. Singh, Curr. Opin. Biotechnol.
22 (2011) 394–400.
[7] L.R. Lynd, P.J. Weimer, W.H.V. Zyl, I.S. Pretorius, Microbiol. Mol. Biol.
Rev. 66 (2002) 506–577.
[8] H. Wang, G. Gurau, R.D. Rogers, Chem. Soc. Rev. 41 (2012) 1519–1537.
[9] F. Chemat, M.A. Vian, G. Cravotto, Int. J. Mol. Sci. 13 (2012) 8615–8627.
[10] A. Corma, S. Iborra, A. Velty, Chem. Rev. 107 (2007) 2411–2502.
[11] P. Gallezot, Green Chem. 9 (2007) 295–302.
[12] M.A.R. Meier, J.O. Metzger, U.S. Schubert, Chem. Soc. Rev. 36 (2007)
1788–1802.
[13] M.N.S. Kumar, A.K. Mohanty, L. Erickson, M. Misra, J. Biobased Mater.
Bioenergy 3 (2009) 1–24.
[14] D.P. Pfister, Y. Xia, R.C. Larock, ChemSusChem 4 (2011) 703–717.
[15] G. Lligadas, J.C. Ronda, M. Galia, V. Cadiz, Biomacromolecules 11 (2010)
2825–2835.
[16] J.K. Lin, S.H. Tsai, S.Y. Lin-Shiau, Drugs Future 26 (2008) 145–152.
[17] A. Murakami, H. Ashida, J. Terao, Cancer Lett. 269 (2008) 315–325.
[18] A.A. Lapkin, A. Walker, N. Sullivan, B. Khambay, B. Mlambo, S. Chemat, J.
Pharm. Biomed. Anal. 49 (2009) 908–915.
[19] A.R. Bilia, P.M.D. Malgalhaes, M.C. Bergonzi, F.F. Vincieri, Phytomedi-
cine 13 (2006) 487–493.
[20] D.B. Rodrigues-Amaya, M. Kimura, Handbook for carotenoid analysis,
International Food Policy Research Institute, 2004.
[21] D. Marinova, F. Ribarova, J. Food Composition Anal. 20 (2007) 370–374.
[22] J. Suberu, L. Song, S. Slade, N. Sullivan, G. Barker, A. Lapkin, J. Pharm.
Biomed. Anal. 84 (2013) 269–277.
[23] P.F. Wilde, Batch records of artemisinin extraction; design of contin-
uous R134a extractor, Personal Communication, 2006.
[24] R.S. Bhakuni, D.C. Jain, R.P. Sharma, Phytochemistry of Artemisia annua
and the development of artemisinin-derived antimalarial agents, in:
C.W. Wright (Ed.), Artemisia, Taylor & Francis, London and New York,
2002, pp. 211–248.
[25] J.O. Suberu, P. Yamin, K. Leonhard, L. Song, N. Sullivan, G. Barker, A.
Lapkin, J. Biotechnol. 171 (2014) 25–33.
[26] D. Du, S. Ki, C. Lin, A. Koutinas, R. Wang, C. Webb, App. Microbiol.
Biotechnol. 76 (2007) 1263–1270.
Fig. 3. GPC traces for the polymerisation of 6 with MMA at 70 8C.
‘‘Processing Industries 2020’’. The authors are grateful for
excellent collaboration within this project with the groups
of Prof. Ian Thompson (University of Oxford), Prof. Yulong
Ding (University of Leeds) and Prof. Adam Harvey (New-
castle University). J.S. was funded via EPSRC CASE student-
ship jointly with Sensapharm Ltd (UK). A. annua biomass
for this project was kindly provided by Fundacion Mundo
Sano (Argentina), Sensapharm Ltd (UK), GSK (Tasmania)
and REAP (Kenya). A.L. is grateful to Medicine for Malaria
Ventures (MMV) for funding the HPLC instrument and
extraction plant, and to Ineos Fluor for technical advice on
tetrafluoroethane extraction. Specifically, the pilot plant
study would not have been possible without the support
by late Dr Ian Bathurst (MMV) to whose memory this paper
is dedicated.
[27] R.G. Schmeltzer, K.E. Uhric, Polym. Bull. 57 (2006) 281–291.
[28] G. Odian, Principles of polymerisation, 4th ed, J. Wiley and Sons, New
Jersey, 2004.
[29] D. Wilms, S. Salah-Eddine, F. Holger, Acc. Chem. Res. 43 (2010)
129–141.
References
[30] J.L. Zhang, R.A. Stanley, A. Adaim, L.D. Meton, M.A. Skinner, Mol. Nutr.
Food Res. 50 (2006) 996–1005.
[31] R. Re, N. Pellegrini, A. Proteggente, A. Pannala, M. Yang, C. Rice-Evans,
Free Radical Biol. Med. 126 (1999) 1231–1237.
[1] M. FitzPatrick, P. Champagne, M.F. Cunningham, R.A. Whitney, Bior-
esour. Technol. 101 (2010) 8915–8922.
[2] W. Smith, Mapping the Development of UK biorefinery complexes, UK
National Non Food Crops Centre (NNFCC), 2007.