Bioresource Technology 102 (2011) 178185

Contents lists available at ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Oil extraction from microalgae for biodiesel production

Ronald Halim *, Brendan Gladman, Michael K. Danquah, Paul A. Webley
Bio Engineering Laboratory (BEL), Department of Chemical Engineering, Monash University, Clayton, Victoria, 3800, Australia

a r t i c l e i n f o a b s t r a c t

Article history: This study examines the performance of supercritical carbon dioxide (SCCO2) extraction and hexane
Received 25 March 2010 extraction of lipids from marine Chlorococcum sp. for lab-scale biodiesel production. Even though the
Received in revised form 23 June 2010 strain of Chlorococcum sp. used in this study had a low maximum lipid yield (7.1 wt% to dry biomass),
Accepted 25 June 2010
the extracted lipid displayed a suitable fatty acid prole for biodiesel [C18:1 (63 wt%), C16:0 (19
Available online 23 July 2010
wt%), C18:2 (4 wt%), C16:1 (4 wt%), and C18:0 (3 wt%)]. For SCCO2 extraction, decreasing tempera-
ture and increasing pressure resulted in increased lipid yields. The mass transfer coefcient (k) for lipid
Keywords:
extraction under supercritical conditions was found to increase with uid dielectric constant as well as
Microalgae
Biodiesel
uid density. For hexane extraction, continuous operation with a Soxhlet apparatus and inclusion of iso-
Lipid extraction propanol as a co-solvent enhanced lipid yields. Hexane extraction from either dried microalgal powder or
wet microalgal paste obtained comparable lipid yields.
2010 Elsevier Ltd. All rights reserved.

1. Introduction for microalgal biodiesel production needs to be not only lipid

The rate of depletion of fossil fuels and the effect of greenhouse
gas emissions on global climate change are creating much interest
in biodiesel (Lang et al., 2001; Lee et al., 2010). Commercial pro-
duction of biodiesel or fatty acid methyl ester (FAME) usually in-
volves alkaline-catalysed transesterication of triglycerides found
in oilseed crops (mainly rapeseed in Europe and soybean in US)
with methanol. However, escalating demand for these crops as a
food source coupled with the nite availability of arable land
makes their cultivation for biofuels unsustainable (Chisti, 2007).
Microalgae have been recognized as a promising alternative
source for biodiesel-convertible lipids. They are a group of
diverse photosynthetic organisms that can accumulate substantial
amounts of lipids up to 50% of dry cell weight in certain species
(Chisti, 2007; Sheehan et al., 1998). Many microalgal species can
grow in brackish water or seawater, thereby avoiding demand for
fresh water, a limited resource in many parts of the world (Umdu
et al., 2009).
Despite the routine use of lab-scale extraction to determine
microalgal lipid content, the variables affecting lipid extraction
from microalgae are not well understood, making scale-up for
commercial production of microalgal biodiesel difcult. Further-
more, most studies investigating microalgal lipid composition fail
to assess the suitability of the extracted lipids for biodiesel produc-
tion as they instead focus on nutraceutical/maricultural applica-
tions (De Angelis et al., 2005; Herrero et al., 2006; Ramadan
et al., 2008; Sajilata et al., 2008). An ideal lipid extraction process
* Corresponding author. Tel.: +61 3 9905 3420; fax: +61 3 9905 5686.
E-mail address: ronald.halim@eng.monash.edu.au (R. Halim).
specic in order to minimize the co-extraction of non-lipid con-
taminants but also selective towards desirable lipid fractions (neu-
tral lipids containing mono-, di-, and trienoic fatty acid chains)
(Fajardo et al., 2007; Medina et al., 1998). Microalgae are cultivated
in an aqueous environment and removing water content beyond a
paste consistency (typically 1030 wt% dry biomass) is energy
intensive. As a consequence, the selected lipid extraction technol-
ogy needs to be effective when applied directly to a wet feedstock.
Supercritical carbon dioxide (SCCO2) extraction is a promising
green technology that can potentially displace the use of tradi-
tional organic solvents for lipid extraction. Advantages associated
with SCCO2 extraction include tuneable solvating power, low tox-
icity of the supercritical uid, favourable mass transfer equilibrium
due to intermediate diffusion/viscosity properties of the uid, and
the production of solvent-free extract (Macias-Sanchez et al., 2007;
Mendes et al., 2006; Pourmortazavi and Hajimirsadeghi, 2007;
Taylor, 1996; Thana et al., 2008). The main disadvantage of the pro-
cess is high cost associated with its infrastructure and operation.
Even though the classic Folch chloroform-based lipid extraction
protocol is effective for the majority of microalgal lipid analyses, an
alternative organic solvent method that is more user-friendly is
needed for scale-up. Hexane, despite being reported to be less ef-
cient than chloroform when extracting from microalgae, is less
toxic, has minimal afnity towards non-lipid contaminants, and
apparent higher selectivity towards neutral lipid fractions (Lee
et al., 1998; Medina et al., 1998).
In this study, the performance of SCCO2 extraction and hexane
extraction was evaluated based on the yield and the fatty acid
composition of the lipids extracted from microalgae under differ-
ent conditions. The suitability of the extracted microalgal lipids

0960-8524/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biortech.2010.06.136
 R. Halim et al. / Bioresource Technology 102 (2011) 178185
for biodiesel production was assessed and the extraction kinetics of
the two systems was compared. Preliminary studies that evaluated
the feasibility of applying each extraction system to wet miroalgal
paste were also conducted. Finally, a mass transfer model was t-
ted to the SCCO2 extraction data to understand the effect of process
variables on extraction kinetics.
2. Methods
2.1. Chemicals and reagents
FAME standards and methylation catalysts (10 wt% H2SO4 in
methanol and 25 wt% KCH3O in methanol) were obtained from
SigmaAldrich Pty. Ltd. (Australia). All organic solvents (n-hexane,
isopropanol, and methanol) were analytical grade.
2.2. Strain and cultivation
The microalgal strain used for this study was a Chlorococcum sp.
cultivated in outdoor bag bioreactors. The species had a high
growth rate, survived well in outdoor conditions (Danquah et al.,
2009), and displayed wide temperature tolerances over the period
of the study.
A modied F medium was used in this study. The medium was
composed of 150.0 mg/L NaNO3, 22.7 mg/L Na2SiO35H2O,
11.3 mg/L NaH2PO42H2O, 9.0 mg/L C6H8O7xFe, 9.0 mg/L C6H8O7,
0.360 mg/L MnCl24H2O, 0.044 mg/L ZnSO47H2O, 0.022 mg/L
CoCl26H2O, 0.020 mg/L CuSO45H2O, 0.013 mg/L Na2MoO42H2O,
trace vitamin B12, biotin, and thiamine dissolved in synthetic
seawater. Each bioreactor held approximately 100 L of microalgal
culture and was aerated with compressed air. Temperature and
illumination intensity were dependent on day-to-day weather
conditions.
2.3. Dewatering
Microalgal cultures from multiple bioreactors (1200 L with a
concentration of 0.7 g dried microalgae/L) were harvested at the
same time and dewatered using a Westfalia disk stack centrifuge.
The concentrate (10 L with a concentration of 80 g dried microal-
gae/L) was thoroughly mixed to average out any potential discrep-
ancies in lipid content between cultures harvested from the
different bioreactors.
A portion of the concentrate (2.5 L) was dewatered further
using a bench top centrifuge (Heraeus Multifuge 3 S-R, Kendro,
Germany) at 4500 rpm for 10 min. The centrate was discarded
and the resulting microalgal paste (0.67 L with a concentration
of 300 g dried microalgae/L) was rinsed once with deionised water
to remove residual salts. The microalgal paste provided all the
biomass needed for the study (200 g of dried microalgae).
2.4. Extraction pre-treatment
In experiments where dried microalgae was used, the microal-
gal paste was dried at 85 C in an oven (Model UNE 500 PA,
Memmert GmbH+Co., Germany) for 16 h. A ring mill was then used
to grind the dried biomass into powder. In experiments involving
wet microalgae, the microalgal paste obtained from bench top cen-
trifugation was used directly without further treatment. The solid
concentration of the paste was 30 wt%. This was determined by
drying a portion of the paste and comparing its pre-dried mass
to that of the corresponding dried biomass. Both microalgal
powder and paste were stored at 5 C for no longer than
1.5 months before they were used for lipid extraction.
179
2.5. Supercritical CO2 lipid extraction
Speed SFE-2 from Applied Separations, Inc. (USA) was used for
SCCO2 extraction (Fig. 1). The extraction unit can be divided into
three separate parts. Part I consists of a feed pump which com-
presses and delivers liquid CO2 to the extraction vessel. Part II com-
prises a 24 mL stainless steel extraction vessel installed inside an
oven module. Once the oven is heated, the compressed CO2 enters
the vessel in a supercritical state and performs lipid extraction
from the microalgae. Part III of the unit consists of a heated
micrometering valve to depressurize incoming SCCO2. Once com-
pletely decompressed, CO2 evaporates as gas to the ambient, forc-
ing the extracted lipid to precipitate out in the adjoining lipid
collection glass vial (Sajilata et al., 2008).
Microalgal powder (20 g) or paste (8 g) was mixed with inert
diatomaceous earth (d.e.) at specic ratios (powder: d.e. =
2:1 w/w and paste: d.e. = 1:2 w/w). The addition of d.e. or celites
(diameter = 35 lm, specic surface area = 30720 cm2/cm3) as an
anchoring co-matrix enhanced contact surface area between the
microalgal cells and the eluting supercritical uid. The resulting
homogeneous mixture was tightly packed into the extraction ves-
sel and the vessel was sealed with metal frits at both ends to avoid
entrainment of the microalgal biomass. Once the vessel was placed
in the oven, the extraction unit was operated as previously
described.
All SCCO2 extractions were carried out in dynamic mode with a
constant decompressed CO2 ow rate of 400 mL/min (residence
time between 4.9 and 14.1 min depending on uid density). The
pressure was varied from 10 to 50 MPa between experiments
and the temperature was set to either 60 or 80 C. Extractions
lasted between 80 and 120 min. For each extraction, a new pre-
weighed glass vial for lipid collection was used every 20 min. The
crude lipid in each vial was gravimetrically quantied and ana-
lysed for FAME proles.
2.6. Static hexane lipid extraction
For straight hexane extraction, 300 mL of n-hexane was added
to either 4 g of microalgal powder or 13.3 g of microalgal paste
(equivalent to 4 g of microalgal powder). For extraction evaluating
the addition of alcohol on hexane extraction, a 3/2 v/v single-phase
mixture of n-hexane and isopropanol (also 300 mL) was added to
4 g of microalgal powder. The conical asks were sealed with alu-
minium foil to reduce solvent evaporation and all extraction mix-
tures were agitated at 800 rpm at ambient conditions for 7.5 h. Cell
residue was removed by ltering through Whatman GF/C paper.
Fig. 1. Schematic of SCCO2 extraction unit.
180 R. Halim et al. / Bioresource Technology 102 (2011) 178185
In the case of straight hexane extraction from microalgal paste,
the ltrate was transferred into a separating funnel to partition the
hexane from water. For hexane/isopropanol extraction from dried
microalgae, the ltrate was transferred into a separating funnel
and sufcient hexane and water (approximately 40 mL each) were
added to induce biphasic layering. After settling, the solvent mix-
ture partitioned into two distinct phases: a top dark-green hexane
layer containing most of the extracted lipids and a bottom light-
green aqueous-isopropanolic layer containing most of the co-ex-
tracted non-lipid contaminants. The hexane phase from each
extraction was collected in a pre-weighed ask before it was
heated to dryness in the oven (60 C) to enable gravimetric quan-
tication of the lipid extract. The crude lipid was re-dissolved in
hexane (approximately 20 mL) and transferred into a sealed glass
vial for storage.
2.7. Dynamic hexane lipid extraction
To compare the performance of hexane extraction with SCCO2, a
Soxhlet apparatus was used. Description of a typical Soxhlet unit
can be found elsewhere (Wang and Weller, 2006). Microalgal pow-
der (4 g) was packed in a cellulose thimble inside the extraction
chamber of the Soxhlet unit. Pure n-hexane (300 mL) was used to
extract the lipid for 7.5 h at the rate of 10 reuxes per hour. The ex-
tracted lipid was gravimetrically quantied at 5.5 and 7.5 h. The
crude lipid was then re-dissolved in hexane (approximately
20 mL) and transferred into a sealed glass vial for storage.
2.8. Storage of lipid extracts
The vials containing the dry lipid extracts or the lipid extract
solutions were stored in the dark at 5 C for less than 1 month.
2.9. Fatty acid methylation
Dried lipid from SCCO2 extraction was re-dissolved in 20 mL of
hexane in order to enhance the reaction kinetics during methanolic
alkylation. A two-step protocol was used for the methylation pro-
cess of all extracted lipids (from both SCCO2 extraction and hexane
extractions) and all reagents were added in stoichiometric excess.
The rst step used an acid catalyst to methylate free fatty acids
(FFA) and to transmethylate acylglycerols (Christie, 2007). Twenty
millilitre methanol and 10 wt% H2SO4 in 1 mL methanol were
added to the lipid in hexane solution stored in each vial. The mix-
ture was transferred into a ask, heated to 50 C, and moderately
agitated for 2 h. Evaporated methanol was frequently replenished.
In the second step, 25 wt% KCH3O in methanol was added drop-
wise to the gently stirred reaction mixture until a pH value of 13
was attained. The inclusion of an alkaline catalyst was intended
to serve four purposes (Lang et al., 2001): (1) to neutralize acids
from the previous step; (2) to quench any interfering water con-
taminants; (3) to further transmethylate acylglycerols; and (4) to
transmethylate phospholipids. The mixture was heated to 55 C
and moderately agitated for 2 h. Evaporated methanol was fre-
quently replenished. The mixture was evaporated in the oven
(60 C) to obtain the dried post-methylated lipid extract. When
the lipid was re-dissolved in 20 mL of hexane for FAME analysis,
some fractions had become insoluble and formed precipitates
(comprising mainly glycerol and chlorophyll).
2.10. Analytical methods
Analyses of the FAME compositions of the post-methylated lipid
extracts were carried out using an HP 5973 gas chromatography
unit equipped with a mass spectrometry detector (HewlettPack-
ard Company, USA). One microlitre sample was injected into a
DBWax polyethylene glycol capillary column (30 m  0.25 mm id
and 0.25 lm lm thickness). Helium was used as the carrier gas
at a constant ow rate of 1.5 mL/min and a split ratio of 20:1. A sol-
vent delay period of 1.5 min was assigned. Both the injector and
detector temperatures were maintained at 150 C. The oven tem-
perature was raised from 140 C at 5 C/min to 240 C and main-
tained for 5 min. Each sample was analysed in triplicates.
Individual FAME was identied by comparison of its retention time
with that of mixed FAME standards as well as positive mass spec-
trum identication from the GC-MSD library. Concentrations of
FAMEs (g/mL) in the injected hexane solution were quantied by
comparing their peak areas with those obtained from the stan-
dards of known concentration. Mass of FAME in the post-methyl-
ated lipid extract (g) was calculated as the product of the sum of
concentrations of individual FAMEs (g/mL) and the volume of hex-
ane used in re-dissolution. FAME yield (g FAME extract/g dried
microalgae) measured the actual amount of biodiesel produced
from a unit mass of dried microalgae.
Extraction performance was evaluated by two key indicators:
lipid yield and FAME composition of the post-methylated lipid
extract. Lipid yield measured the extraction efciency and was
calculated by dividing mass of crude lipid extract with mass of
dried microalgae. Where a wet paste was used, the corresponding
amount of dried biomass was determined. FAME composition is
derived by dividing mass of individual FAME with collective FAME
mass, and is an indicator of the selectivity of the extraction process.
2.11. Mass transfer modelling for SCCO2 extraction from dried
microalgae
In order to describe the mass transfer behaviour of lipid mole-
cules during SCCO2 extraction, the following kinetic model (Ozkal
et al., 2005) was used:
dme
kms;t  ms;t 1
dt
where me is the lipid yield at time t (g lipid extract/g dried microal-
gae), t is the extraction time (min), k is the lipid mass transfer coef-
cient from the microalgal cells into the SCCO2 (min1), ms,t is the
amount of extractable intracellular lipid that remains unextracted
at time t (g lipid/g dried microalgae), and m*s,t is the amount of
unextracted intracellular lipid when there is equilibrium in the lipid
concentration between the microalgal cells and the SCCO2 (g lipid/g
dried microalgae). This model is a lumped version of Ficks Law of
Diffusion in which spatial proles have been averaged by lumping
these into the mass transfer coefcient. The difference between
ms;t and ms;t serves as the driving force for diffusion of lipid mole-
cules from the solid cellular phase into the bulk uid phase and rep-
resents the departure from equilibrium. However, since fresh SCCO2
was continuously swept across the extraction vessel, the equilib-
rium concentration of lipid molecules ms;t theoretically approaches
0.
dme
kms;t  0 k  ms;t 2
dt
Since ms;t is equal to the difference between the original amount of
extractable intracellular lipid, ms;0 (g lipid extract/g dried micro-
algae), and the amount of lipids that have been extracted at time
t, me:
dme
kms;0  me 3
dt
Eq. (3) is a rst-order linear ODE. Together with the initial condition
me(t = 0) = 0, it can be solved to give (Andrich et al., 2005, 2006):
me ms;0 1  ekt 4
 R. Halim et al. / Bioresource Technology 102 (2011) 178185 181

ms;0 is assumed to be equal to the highest nal lipid yield obtained
in all SCCO2 extractions from dried microalgae (0.060 g extract/g
dried microalgae). Through rearrangement of Eq. (4), k is deter-
mined as the slope of the straight line passing through the origin
of the plot of ln(ms,0/(ms,0 me)) against t.
 
ms;0
ln kt
ms;0  me
3. Results and discussions
3.1. Supercritical CO2 lipid extraction
3.1.1. Effects of extraction time and temperature
Lipid extractions were conducted for 80 min at 60 and 80 C.
The pressure was constant at 30 MPa and 20 g of dried microalgal
powder was used in each experiment. Table 1 (under Experimen-
tal lipid yield) shows that lipid was extracted most rapidly in the
beginning of the extraction (before 20 min) and that the rate of li-
pid extraction decreased with experimental time. The asymptotic
prole obtained was characteristic of most SCCO2 extraction re-
ported in other microalgal studies (Cheung, 1999). The model cor-
related well with the experimental data for lipid evolution
(compare Theoretical lipid yield with Experimental lipid yield,
r 260 C 0:93, r 280 C 0:92) and was consistent with the expected dif-
fusion-driven nature of the SCCO2 process, where the extraction
rate was directly proportional to the amount of unextracted intra-
cellular lipids (Ozkal et al., 2005).
The nal experimental lipid yields (0.058 g lipid extract/g dried
microalgae for 60 C and 0.048 g lipid extract/g dried microalgae
for 80 C) are in agreement with those reported for Chlorococcum
sp. in previous studies (Ota et al., 2009). However, the lipid content
of the strain of Chlorococcum sp. used in this study is considerably
lower than reported for other oleaginous microalgal species, such
as Nannochloropsis sp. (reported at 0.250 g lipid extract/g dried
microalgae) and Botryococcus braunii (reported at 0.286 g lipid
extract/g dried microalgae) (Andrich et al., 2005; Lee et al.,
1998). As such, the use of this particular strain of Chlorococcum
5 sp. for commercial biodiesel production appears unattractive.
An increase in the temperature of SCCO2 extraction was
observed to decrease the mass transfer coefcient (k60C = 0.052/
min versus k80C = 0.019/min). A temperature increase during
supercritical extraction results in two simultaneous competing ef-
fects. The decrease in uid density reduces uid transport rate
which retards extraction kinetics while the increase in solute vol-
atility enhances mass transfer (Cheung, 1999; Taylor, 1996). Since
the extractions were carried out at a relatively low pressure where
the supercritical uid was highly compressible (q60 = 0.83 g/mL
while q80 = 0.74 g/mL), the experimental temperature increase
lead to a considerable density reduction that more than offset
the increase in lipid volatility resulting in a lower overall lipid mass
transfer rate.
For both extractions, the amounts of FAME after methylation of
the crude lipids were found to be roughly between 0.23 and 0.44 of
the masses of extracted crude lipids (Table 1 under FAME yield/
experimental lipid yield). Even though this result agreed with
the analysis reported by Cheung (1999), the majority of previous
work on microalgal lipid extraction did not report on the purity
of the extracted lipids. Despite the low afnity of SCCO2 for polar
compounds, the high concentration of polar lipids known to make
up the cell membrane (up to 90% of lipids) made concurrent leach-
ing during supercritical extraction of neutral lipids inevitable. As
such, polar lipids that were unsuccessfully esteried with the
applied methylation techniques (complex phospholipids and

Table 1
Effects of extraction time and temperature during SCCO2 extraction on lipid yield and on FAME composition of the post-methylated lipids. Mass of dried microalgae = 20 g,
P = 30 MPa, T = 60 and 80 C. Lipid yields (both experimental and theoretical) and FAME yields are shown as functions of time (020 min lists the total yield of lipid and of FAME
obtained from 0 to 20 min, while 040 min lists the total yield of lipid and FAME obtained from 0 to 40 min). Theoretical lipid yield is the lipid yield predicted by the kinetic
model. FAME composition, on the other hand, is shown as a function of 20-min time interval rather than as a function of time (2040 min lists the FAME composition of only the
lipid extracted from 20 to 40 min). FAME composition is reported as weight percent of all FAMEs that can be analysed in the lipid extracted within the given time interval and is
an average of three analyses.

60 C 80 C
020 040 060 080 020 040 060 080
min min min min min min min min
Experimental lipid yield 0.035 0.056 0.058 0.058 0.025 0.032 0.035 0.048
(g lipid extract/g dried microalgae)
Theoretical lipid yield 0.038 0.052 0.057 0.059 0.018 0.031 0.040 0.046
(g lipid extract/g dried microalgae)
FAME yield (g FAME/g 0.010 0.013 0.013 0.014 0.011 0.013 0.014 0.015
dried microalgae)
FAME yield/experimental 0.29 0.23 0.23 0.24 0.44 0.40 0.40 0.31
lipid yield
020 2040 4060 6080 080 020 2040 4060 6080 080
min min min min min min min min min min
(overall) (overall)
FAME (wt% of FAME) C14:0 0.7 0.0 0.0 0.0 0.5 0.8 0.0 0.0 0.0 0.6
C16:0 19.0 16.8 25.9 21.6 18.8 19.2 16.8 20.8 20.4 19.1
C16:1 4.0 1.8 0.0 0.0 3.3 5.2 1.8 0.0 0.0 4.1
C16:1t 3.3 0.0 0.0 0.0 2.4 3.5 0.0 0.0 0.0 2.6
C16:2 3.1 0.0 0.0 0.0 2.3 3.5 0.0 0.0 0.0 2.6
C17:0 0.3 1.4 0.0 0.0 0.5 0.4 1.4 0.0 4.6 0.8
C18:0 1.8 5.0 8.9 5.9 2.8 1.2 5.0 4.0 5.7 2.2
C18:1 66.3 56.2 52.7 60.5 63.6 64.3 56.2 62.1 58.2 62.8
C18:2 0.0 18.7 12.4 12.0 4.7 0.0 18.7 13.1 11.1 3.8
C20:5 1.5 0.0 0.0 0.0 1.1 1.9 0.0 0.0 0.0 1.4
No. of double bonds in 0 21.8 23.2 34.8 27.5 22.6 21.6 23.2 24.8 30.7 22.7
FAME (wt% of FAME) 1 73.6 58.1 52.7 60.5 69.4 73.0 58.1 62.1 58.2 69.5
2 3.1 18.7 12.4 12.0 6.9 3.5 18.7 13.1 11.1 6.4
3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
P4 1.5 0.0 0.0 0.0 1.1 1.9 0.0 0.0 0.0 1.4
182 R. Halim et al. / Bioresource Technology 102 (2011) 178185

glycolipids) were assumed to be major constituents of the contam-
inants in the extracted lipids. The unidentied portion of the crude
lipids might also comprise non-saponiable neutral lipids (hydro-
carbons, alcohols, and aldehydes) and non-lipids (protein and car-
bohydrate). The dark green appearance of the extracted lipids also
suggested chlorophyll presence. As such, without further rening
to remove these unusable components, the extracted lipids cannot
be considered a viable biodiesel feedstock.
(a)
(b) 70
30 MPa 40 MPa
60
Wt% of total FAME
In terms of overall FAME compositions (Table 1 under 0
80 min (overall)), C18:1 (63.2 wt% of FAME), C16:0 (19.0 wt%
of FAME), C18:2 (4.3 wt% of FAME), C16:1 (3.7 wt% of FAME),
and C18:0 (2.5 wt% of FAME) were identied as the principal fatty
acids in SCCO2-extracted lipids. Other fatty acids, such as C14:0,
C16:1t, C16:2, C17:0, and C20:5, were also present in smaller
quantities. The extracted lipids appeared to have an appropriate
fatty acid composition for biodiesel. The lipids contained only
modest amounts of polyunsaturated fatty acids (PUFA) with 4 or
more double bonds (C20:5 being the only one at 1.3 wt% of
FAME). Highly unsaturated PUFAs are known to be responsible
for the poor volatility, low oxidation stability, and tendency for
gum formation observed in some oilseed-derived biodiesel. Addi-
tionally, the relatively low proportion of saturated fatty acids
(22.7 wt% of FAME) together with a minimal presence of trans-
conjugated unsaturated fatty acids (C16:1t being the only one at
2.5 wt% of FAME) was preferable. The crystal packing ability of
these fatty acid classes produces biodiesel with undesirable cold
ow properties such as high cloud point and high pour point
(Christie, 2007; Lang et al., 2001). The fatty acid composition dis-
played in Table 1, however, is different to the prole presented
by Ota et al. (2009). Despite similar results regarding the respective
proportions of C16:0, C16:1, and C18:2, the authors reported Chlo-
rococcum sp. to contain particularly high levels of C16:4 (>20 wt%
of FAME) and C18:3 (>30 wt% of FAME), with a corresponding
reduction in C18:1 (<13 wt% of FAME). Fatty acid prole of a mic-
roalgal species is known to be a function of its culturing conditions
(Andrich et al., 2005; Guzman et al., 2010; Rao et al., 2007).
The collection of extracted lipids into a fresh vial every 20-min
intervals facilitated the evaluation of FAME composition as a func-
tion of time interval (Table 1). The observed changes in the FAME
50 MPa
proportions of the post-methylated lipids as a function of
extraction duration (the absence of C18:2 before 20 min and the

50 disappearance of C16:1 after 40 min) suggest a possible time-

facilitated variation of SCCO2 selectivity (Ozkal et al., 2005; Pour-
40 mortazavi and Hajimirsadeghi, 2007; Soares et al., 2007). However,
the variation in fatty acid solubilities appeared rather fraction-
30
specic and no correlation based on chain length or degree of
20 unsaturation can yet be identied. In agreement with Andrich
et al. (2006) who performed SCCO2 extraction on Spirulina platensis,
10 no meaningful difference was found in the overall FAME composi-
tion of lipids extracted at 60 and 80 C.
0
3.1.2. Effect of pressure and dielectric constant
:0

:1

:2

:0

:0
:0

:1

:2

:5
t
:1
14

16

16

17

18

18

18

20
16
16

In order to evaluate the effect of pressure on the efcacy of

C

C
C

SCCO2 extraction, two different stepped pressure congurations

70 (30, 40, and 50 MPa against 10, 20, and 30 MPa) were trialed. For
10 MPa 20 MPa 30 MPa the higher-pressure extraction, the initial pressure of 30 MPa was
60 increased to 40 MPa at 80 min, followed by a further increment
to 50 MPa at 100 min. The lower-pressure extraction was carried
Wt% of total FAME

50
out in a similar arrangement, although pressure increments
40 occurred at different times: 40 and 80 min (see Fig. 2a). Each
extraction was performed at 60 C for 120 min using 20 g of dried
30 microalgae.
Fig. 2a shows the positive effect of pressure on extraction
20 performance, where the nal lipid yield of higher-pressure

10
Table 2
0 Effect of SCCO2 dielectric constant on lipid mass transfer coefcient along an isotherm
(T = 60 C). Density and dielectric constant values extracted from Zhang et al. (2005).
:5
:0
:0

:0

:1

:2
:2

:0
t
:1

:1

20
18
14

16

16

18

18
16

17
16

Extraction SCCO2 density SCCO2 dielectric k (min1) r2

C
C

C
C

pressure (MPa) (g/mL) constant

Fig. 2. Effect of pressure during SCCO2 extraction (a) on lipid yield; (b) on overall 10 0.32 1.17 0.004 0.98
FAME composition of the post-methylated lipids. j higher-pressure experiment, 20 0.73 1.43 0.013 0.92
N lower-pressure experiment. Mass of dried microalgae = 20 g, T = 60 C, 30 0.83 1.49 0.052 0.93
duration = 120 min.
 R. Halim et al. / Bioresource Technology 102 (2011) 178185 183

extraction (0.060 g lipid extract/g dried microalgae) was almost This behaviour was anticipated since an increasing solvent polarity
twice the nal lipid yield of lower-pressure extraction (0.034 g would enhance the disruption of membrane-lipid bilayer and facil-
lipid extract/g dried microalgae). This is expected as the dielectric itate the extraction of membrane-bound lipids.
constant or the solvent polarity of a supercritical uid directly de-
pends on the uid density which increases with pressure elevation 3.1.3. Performance on wet microalgal paste
at a constant temperature (refer to Table 2) (Cheung, 1999; Taylor, For SCCO2 extraction technology to have any chance of being
1996). The pronounced increase in the lipid yields of both congu- practically implemented for industrial-scale microalgal biodiesel
rations upon pressure ramping demonstrated how a sudden shift production, it has to extract lipids effectively from dewatered mic-
in the uid solvation power could trigger the solubilisation of lipid roalgal paste (30 wt% solid) just as it does from dried biomass. In
fractions which were insoluble at reduced pressure. The rst pres- this section, an identical protocol to the dry extraction was used
sure change (after 80 min for higher-pressure extraction and after to study wet SCCO2 extraction process (T = 60 C, P = 30 MPa).
40 min for lower-pressure extraction) was only carried out when However, the immiscibility of water and CO2 resulted in highly
the lipid yield at the original pressure had reached an asymptote. compacted particle bed which restricted ow of the uid through
Only slight variations in fatty acid compositions of lipids ex- the column. Different ratios of paste to d.e. were trialled and
tracted at different pressure levels were observed (Fig. 2b). The increasing amount of d.e. in the mixture was observed to improve
same dominant fatty acid molecules were identied in all lipid ex- CO2 ow. The d.e. adsorbs water and provides the additional poros-
tracts, with C18:1 and C16:0 being the most abundant. The slight ity needed for CO2 permeation. At a paste-to-d.e. ratio of 1:2, a CO2
changes in FAME compositions allude to the differential selectivity ow rate equal to that used for dry extraction (400 mL/min) was
that SCCO2 is known to display towards specic lipid fractions at achieved. At this ratio, the vessel contained a paste equivalent to
different pressures (Cheung, 1999). 2.4 g of dried microalgae, substantially less than the 20 g used pre-
The dielectric constant or the solvent polarity of a supercritical viously for dry extraction.
uid measures the degree of interaction between the solvent and From Fig. 3a, extraction from the wet paste obtained a higher
the organic molecules (Zhang et al., 2005). Table 2 shows the expo- nal lipid yield than from dried biomass (lipid yieldwet = 0.071 g
nential dependence of lipid mass transfer coefcient (k) during our lipid extract/g dried microalgae, while lipid yielddry = 0.058 g lipid
SCCO2 extractions on uid dielectric constant along an isotherm. extract/g dried microalgae). Despite the common view that resid-
ual water in the cells acts as a barrier which impedes mass transfer
(a) of the analyte molecules into the bulk uid, a few studies on SCCO2
0.08 extraction of miscellaneous analytes from moist samples (caffeine
(g lipid extract / g dried microalgae)

from green coffee beans and essential oil from Origanum vulgare
0.07 leaves) have demonstrated that residual water in the feedstock
does not reduce extraction performance. Water was found to aid
0.06
extraction through its swelling of the cellular matrix and its natu-
0.05 ral role as a polar co-solvent (Pourmortazavi and Hajimirsadeghi,
Lipid yield

2007; Schwartzberg, 1997). The lipid extracts from the wet paste
0.04 comprised higher quantities of C16:0, C16:1, C18:0, and C18:2 than
0.03 the corresponding extracts from the dried biomass (Fig. 3b). The
increased co-extraction of polar lipids due to the presence of water
0.02 in the cells could have potentially contributed to this composi-
tional discrepancy.
0.01

0.00 3.1.4. Comparison with dynamic hexane lipid extraction (Soxhlet)

0 1 2 3 4 5 Fig. 3a shows SCCO2 extraction to be more efcient than
Time (hour) dynamic hexane extraction. Eighty minutes of SCCO2 extraction
achieved a higher lipid yield than 5.5 h of Soxhlet extraction (lipid
(b) 80 yield of SCCO2 after 80 min = 0.058 g lipid extract/g dried micro-
SCCO2 - dried microalgae algae, lipid yield of Soxhlet after 5.5 h = 0.032 g lipid extract/g
70
SCCO2 - wet paste dried microalgae). This observation was expected due to the
60 hexane (soxhlet) - dried microalgae
favourable physicochemical properties of the supercritical uid
Wt% of total FAME

(Cheung, 1999; Mendes et al., 2003). Despite demonstrating that

50 both extractions eventually obtained equivalent lipid yields from
40 Nannochloropsis sp., Andrich et al. (2005) also measured a lower
mass transfer coefcient for extraction with hexane than with
30 SCCO2. Future comparison between the two methods should take
into account the energy requirement for each respective extrac-
20
tion. Fluid compression represents the largest energy requirement
10 for SCCO2 extraction (Taylor, 1996), while downstream extract
purication remains the primary energy obstacle for large-scale
0
organic solvent extraction.
:5
:1

:1

:2
:0

:0

:2

:0

:0
:1

18

20
14

16

16

18
16

17

18
16

C
C

C
C

3.2. Static hexane lipid extraction

Fig. 3. Comparison between different extraction systems. (a) lipid yield; (b) overall
FAME composition of post-methylated lipids. j SCCO2 extraction from dried The effects of operation mode and addition of alcohol on the
microalgae: mass of dried microalgae = 20 g. s SCCO2 extraction from wet micro- performance of hexane extraction from dried microalgae were as-
algal paste: mass of paste = 8 g, solid concentration = 30 wt%. For both SCCO2
extractions, T = 60 C, P = 30 MPa. N Hexane extraction from dried microalgae using
sessed. The feasibility of applying hexane extraction straightaway
Soxhlet apparatus: mass of dried microalgae = 4 g, P = ambient, number of on wet microalgal paste was also evaluated. The nal lipid yield
reuxes = 55. for each of the above hexane extraction system is discussed in
184
80
A B C1
70
60
Wt% of total FAME
50
40
30
20
10
0 4. Conclusions
:1
:0
:0
:2
:0
t
:1
14
16
16
16
18
18
16
C
C
Fig. 4. Comparison of the overall FAME composition of post-methylated lipids
extracted by different hexane extraction systems. (A) hexane extraction from dried
microalgae. (B) hexane extraction using Soxhlet apparatus from dried microalgae.
(C) hexane/isopropanol (3:2 v/v) extraction from dried microalgae (C1: hexane
phase, C2: isopropanol phase). (D) hexane extraction from wet microalgal paste. For
(A, B, C) mass of dried microalgae = 4 g. For (D) mass of paste = 13.3 g, solid
concentration = 30 wt%. For (A, B, C, D) volume of solvent = 300 mL, T = ambient,
P = ambient, duration = 7.5 h. (A, C, D) are static extractions.
the text. Fig. 4 shows the fatty acid proles of the lipids extracted
by the different hexane extraction systems together with the
operating conditions of each system.
3.2.1. Effect of dynamic operation
Operating the extraction system in a dynamic mode (B) rather
than a static mode (A) increased the lipid yield by roughly 280% (-
nal lipid yield of A = 0.015 g lipid extract/g dried miroalgae, nal li-
pid yield of B = 0.057 g lipid extract/g dried microalgae). This
improvement was expected since Soxhlet operation, through sol-
vent reuxing, constantly exposed a fresh batch of hexane to the
cellular matrix and enabled continuous re-establishment of mass
transfer equilibria (Wang and Weller, 2006). Despite its enhanced
yield, Soxhlet procedure is unlikely to be implemented in an indus-
trial scale due to the high energy requirement for continuous
distillation.
3.2.2. Effect of alcohol addition
At the end of hexane/isopropanol extraction, the solvent mix-
ture partitioned into two distinct phases: a top dark-green hexane
layer (C1) and a bottom light-green aqueous-isopropanolic layer
(C2). Even though most of the lipid seemed to be transferred to
the hexane phase, both phases were collected and gravimetrically
quantied. Only the hexane phase (C1) was, however, analysed for
FAME composition.
It was anticipated that nal lipid yield would substantially in-
crease with the addition of isopropanol as a co-extractant. The
threefold increase in the hexane lipid yield (nal lipid yield of
A = 0.015 g lipid extract/g dried microalgae, nal lipid yield of
C1 = 0.048 g lipid extract/g dried microalgae) was further aug-
mented by an additional lipid yield from the dried aqueous-isopro-
panolic phase (nal lipid yield of C2 = 0.020 g lipid extract/g dried
microalgae). Since polar lipids (phospholipids, glycolipids, and cho-
lesterols) are mostly bound to protein molecules in the cell mem-
brane via hydrogen or strong electrostatic bonds, they require the
presence of a polar solvent (such as isopropanol) to disrupt the
complex lipidprotein interactions before they are extracted out
by hexane. Additionally, even though non-polar hexane readily
R. Halim et al. / Bioresource Technology 102 (2011) 178185
interacts with neutral lipid molecules, the micelle-type structures
D that triacylglycerols frequently form in their natural state prevent
quantitative lipid extraction and often require the assistance of an
alcohol for their rapid destruction (Medina et al., 1998).
3.2.3. Performance on wet microalgal paste
Hexane extraction from wet paste (D) resulted in 33% less lipid
yield than its dry counterpart (A) (nal lipid yield of A = 0.015 g li-
pid extract/g dried microalgae, nal lipid yield of D = 0.010 g lipid
extract/g dried microalgae). Despite its reduced wet yield, hexane
extraction remains a promising option for microalgal biodiesel par-
ticularly since its wet extraction protocol, unlike SCCO2 process,
does not require additional cost-intensive pre-treatment step.
:2
:1
:3
:5
:6
t
:1
18
18
20
22
The Chlorococcum sp. used in this study was found to have a low
18
C
maximum lipid yield (7.1 wt% to dry biomass) compared to other
lipid-rich microalgal species, e.g. Nannochloropsis sp. (reported at
25.0 wt% to dry biomass) and B. braunii (reported at 28.6 wt% to
dry biomass). For SCCO2 extraction, lipid yield was found to de-
crease with temperature and to increase with pressure. No signif-
icant differences were found in the FAME compositions of lipid
extracts obtained under different SCCO2 extraction parameters.
The success of the proposed model in correlating lipid evolution
during SCCO2 extraction (r2 > 0.92) enabled further analysis that
demonstrated lipid mass transfer coefcient (k) to increase with
solvent polarity. Soxhlet extraction using hexane was found to be
signicantly less efcient than SCCO2 extraction, achieving a com-
parable lipid yield (0.058 g lipid extract/g dried microalgae) in
5.6 the required time.
Acknowledgements
This work was supported by an Australian Research Council
(ARC) Linkage grant between the Department of Chemical Engi-
neering in Monash University (Victoria, Australia) and Bio-Fuel
Pty Ltd (Victoria, Australia). The authors also thank Dr. Craig Davis
from the Department of Primary Industries and Fisheries (Queens-
land, Australia) for use of SCCO2 apparatus.
References
Andrich, G., Nesti, U., Venturi, F., Zinnai, A., Fiorentini, R., 2005. Supercritical uid
extraction of bioactive lipids from the microalgae Nannochloropsis sp.. European
Journal of Lipid and Science Technology 107, 381386.
Andrich, G., Zinnai, A., Nesti, U., Venturi, F., Fiorentini, R., 2006. Supercritical uid
extraction of oil from microalga Spirulina (Arthrospira) platensis. Acta
Alimentaria 35, 195203.
Cheung, P., 1999. Temperature and pressure effects on supercritical carbon dioxide
extraction of n3 fatty acids from red seaweed. Food Chemistry 65, 399403.
Chisti, Y., 2007. Research review paper: biodiesel from microalgae. Biotechnology
Advances 25, 294306.
Christie, W.W., 2007. Methylation of fatty acids a beginners guide. Available
from: http://www.lipidlibrary.co.uk/topics/methests/index.htm.
Danquah, M.K., Gladman, B., Moheimani, N., Forde, G.M., 2009. Microalgal growth
characteristics and subsequent inuence on dewatering efciency. Chemical
Engineering Journal 151, 7378.
De Angelis, L., Rise, P., Giavarini, F., Galli, C., Bolis, C.L., Colombo, M.L., 2005. Marine
microalgae analyzed by mass spectrometry are rich sources of polyunsaturated
fatty acids. Journal of Mass Spectrometry 40, 16051608.
Fajardo, A.R., Cerdan, L.E., Medina, A.R., Fernandez, F.G.A., Moreno, P.A.G., Grima,
E.M., 2007. Lipid extraction from the microalga Phaedactylum tricornutum.
European Journal of Lipid Science and Technology 109, 120126.
Guzman, H.M., Valido, A.J., Duarte, L.C., Presmanes, K.F., 2010. Estimate by means of
ow cytometry of variation in composition of fatty acids from Tetraselmis
suecica in response to culture conditions. Aquaculture International 18, 189
199.
Herrero, M., Cifuentes, A., Ibanez, E., 2006. Sub- and supercritical uid extraction of
functional ingredients from different natural sources: plants, food-by-products,
algae and microalgae, a review. Food Chemistry 98, 136148.
 R. Halim et al. / Bioresource Technology 102 (2011) 178185
Lang, X., Dalai, A.K., Bakhshi, N.N., Reaney, M.J., Hertz, P.B., 2001. Preparation and
characterization of bio-diesels from various bio-oils. Bioresource Technology
80, 5362.
Lee, J.Y., Yoo, C., Jun, S.Y., Ahn, C.Y., Oh, H.M., 2010. Comparison of several methods
for effective lipid extraction from microalgae. Bioresource technology 101, S75
S77.
Lee, S.J., Yoon, B.D., Oh, H.M., 1998. Rapid method for the determination of lipid
from the green algae Botryococcus braunii. Biotechnology Techniques 553, 556.
Macias-Sanchez, M.D., Mantell, C., Rodriguez, M., de la Ossa, E.M., Lubian, L.M.,
Montero, O., 2007. Supercritical uid extraction of carotenoids and chlorophyll
a from Synechococcus sp.. Journal of Supercritical Fluids 39, 323329.
Medina, A.R., Grima, E.M., Gimenez, A.G., Ibanez, M.J., 1998. Downstream processing
of algal polyunsaturated fatty acids. Biotechnology Advances 16, 517580.
Mendes, R.L., Nobre, B.P., Cardoso, M.T., Pereira, A.P., Palavra, A.F., 2003.
Supercritical carbon dioxide extraction of compounds with pharmaceutical
importance from microalgae. Inorganica Chimica Acta 357, 328334.
Mendes, R.L., Reis, A.D., Palavra, A.F., 2006. Supercritical CO2 extraction of gamma-
linolenic acid and other lipids from Arthrospira (Spirulina) maxima: comparison
with organic solvent extraction. Food Chemistry 99, 5763.
Ota, M., Kato, Y., Watanabe, H., Watanabe, M., Sato, Y., Smith, R., Inomata, H., 2009.
Fatty acid production from a highly CO2 tolerant alga, Chlorocuccum littorale, in
the presence of inorganic carbon and nitrate. Bioresource Technology 100,
52375242.
Ozkal, S.G., Salgin, U., Yener, M.E., 2005. Supercritical carbon dioxide extraction of
hazelnut oil. Journal of food engineering 69, 217223.
Pourmortazavi, S.M., Hajimirsadeghi, S.S., 2007. Supercritical uid extraction in
plant essential and volatile oil analysis review. Journal of chromatography A
1163, 224.
Ramadan, M.F., Asker, M.H.S., Ibrahim, Z.K., 2008. Functional bioactive compounds
and biological activities of Spirulina platensis lipids. Czech Journal of Food
Science 26, 211222.
185
Rao, A.R., Dayananda, C., Sarada, R., Shamala, T.R., Ravishankar, G.A., 2007. Effect of
salinity on growth of green alga Botryococcus braunii and its constitutents.
Bioresource Technology 98, 560564.
Sajilata, M.G., Singhal, R.S., Kamat, M.Y., 2008. Supercritical CO2 extraction of c-
linolenic acid (GLA) from Spirulina platensis ARM 740 using response surface
methodology. Journal of Food Engineering 84, 321326.
Schwartzberg, H.G., 1997. Mass transfer in a countercurrent, supercritical extraction
system for solutes in moist solids. Chemical Engineering Communications 157,
122.
Sheehan, J., Dunahay, T., Benemann, J., Roessler, P., 1998. A look back at the US
department of energys aquatic species program: biodiesel from algae. NREL/TP-
580-24190 ed. In: Laboratory, N.R.E. (Ed.), National Renewable Energy
Laboratory. US Department of Energy, pp. 1100.
Soares, B.M.C., Gamarra, F.M.C., Paviani, L.C., Goncalves, L.A.G., Cabral, F.A., 2007.
Solubility of triacylglycerols in supercritical carbon dioxide. Journal of
Supercritical Fluids 43, 2531.
Taylor, L.T., 1996. Supercritical Fluid Extraction. John Wiley & Sons Inc,
New York.
Thana, P., Machmudah, S., Goto, M., Sasaki, M., Pavasant, P., Shotipruk, A., 2008.
Response surface methodology to supercritical carbon dioxide extraction of
astaxanthin from Haematococcus pluvialis. Bioresource Technology 99, 3110
3115.
Umdu, E.S., Tuncer, M., Seker, E., 2009. Transesterication of Nannochloropsis oculata
microalgas lipid to biodiesel on Al2CO3 supported CaO and MgO catalysts.
Bioresource Technology 100, 28282831.
Wang, L., Weller, C.L., 2006. Recent advances in extraction of nutraceuticals from
plants. Trends in Food Science and Technology 17, 300312.
Zhang, Y., Yang, J., Yu, Y.-X., 2005. Dielectric constant and density dependence of the
structure of supercritical carbon dioxide using a new modied empirical
potential model: a Monte Carlo simulation study. The Journal of Physical
Chemistry B 109, 1337513382.