The Indian mackerel (local name: kembung) is the most popular

marine fish in Malaysia due to its abundance, year-round availabil-

ity, low cost, high polyunsaturated fatty acid (PUFA) content, and
suitable size. In Sarawak, Indian mackerel is the highest landing
species (12.5%) than any other species while east coast peninsular
Malaysia is 7.6% (Gambang et al., 2003). It is known as fatty fish,
having fat content well distributed throughout the body (Adeniyi
and Bawa, 2008). Fish oils are rich in PUFA especially those of
the x-3 family, mainly eicosapentaenoic acid (EPA; C20:5n-3)
and docosahexaenoic acid (DHA; C22:6n-3). The importance of
PUFA in human health and nutrition is well recognised. Many clin-
ical and epidemiologic studies have shown positive roles for x-3
PUFA in infant development; cancer; cardiovascular diseases; and
more recently, in various mental illnesses, including depression,
attention-deficit hyperactivity disorder, and dementia Fish skin is the by-product/waste of fish
processing industry.
Some of the by-product is utilized, but the main bulk is dumped
to waste, creating both disposal and pollution problems (Nuraini
et al. 2008). Fish by-product has a lot of unexploited potential for
value adding. Mackerel skins are usually discarded during process-
ing and they have high oil content. Zuta et al. (2003) extracted
PUFA from mackerel waste, because it is an underutilized fish,
most likely due to its highly perishable nature. They also reported
that, skins are a prime material for producing PUFA because they will generate comparatively
high yields (about 38% compared to
9% for viscera and muscle) of oil from small inputs.
The conventional fish processing methods for oil extraction,
fractionation, and purification include hydraulic pressing, vacuum
distillation, urea crystallization, hexane extraction, and conven-
tional crystallization, which all involve high-temperature process-
ing or flammable or toxic solvents (Staby and Mollerup, 1993).
These methods can contribute to the loss, denaturation, or decom-
position of the precious nutritional fish oil. SFE is a promising pro-
cess for the extraction and fractionation of edible oils containing
labile PUFA, which can be carried out under mild operating condi-
tions (Létisse et al., 2006). Supercritical extraction and fraction-
ation of fish oil fatty acids have been studied by many
researchers (Esquível et al., 1997; Riha and Brunner, 2000; Létisse
et al., 2006; Perretti et al., 2007). Besides, extraction with carbon
dioxide (CO2) as solvent offers new opportunities for the solution
of separation problems as it is nontoxic, non-flammable, inexpen-
sive, and clean solvent. Supercritical carbon dioxide (SC-CO2)
extraction may influenced by the moisture from the matrix (Atlan-
tic mackerel) which affects both the yield and changes the sarco-
plasmatic proteins in the matrix.

Materials and methods

2.1. Materials
Indian mackerels were collected from the fish landing centre,
located at Kuala Selangor, Malaysia. Commercial liquefied CO2
(purity, 99.9%), nitrogen (as a carrier gas with a purity of 99.9%),
auxiliary gases, hydrogen (purity, 99.9%) and compressed air (free
from organic impurities), were purchased from Malaysian Oxygen
Ltd., Kuala Lumpur. n-Hexane, AR Grade (Merck, Germany) and
petroleum ether, AR Grade (Merck, Germany) was obtained from
Kuala Lumpur. Menhaden oil (PUFA-3), which was used as a refer-
ence standard fish oil (purity 99%), and sodium methoxide were
obtained from Sigma–Aldrich Co., USA. An Omegawax-320 capil-
lary column (30 m -
0.32 mm -
0.25 lm film thickness) was also
purchased from Sigma–Aldrich Co., USA.
2.2. Sample preparation for experiment
The fresh fish samples were kept in plastic bags and transported
in an insulated icebox to the laboratory. The samples were imme-
diately de-headed, gutted, and washed with copious amounts of
cool water, and then the skins were separated using a de-Bonner
(model- FD 6, Safe World Food-Tech Pvt. Ltd., Kuala Lumpur).
The skins were stored at 20 -
C and freeze dried (Model: LAB-
CONCO, USA) at a drying temperature of 47 -
C and vacuumed
at 0.133 bar. The dry skins were kept in desiccators until use. For
experimental analysis, the dried skins were ground into particle
sizes ranging from 0.2 to 0.5 mm by sieving.
2.3. Determination of moisture content
The moisture content of the ground fish skin was determined
using the PORIM test method (No. p5.2), 1995.
2.4. Extraction of total oil using the Soxhlet method
The Soxhlet extraction method was used as a comparison to the
supercritical extraction. It was carried out in triplicate for 5 g (dry
basis) of skin powder with 200 ml of petroleum ether for 6 h. The
extracted oil was evaporated under vacuum at 65 -
C using a rotary
evaporator and then dried in the oven at 45 ± 1 -
C for 1 h and
cooled in desiccators before reweighing.
2.5. SC-CO2 extraction with various techniques
The experimental setup for the supercritical fluid extraction
(SFE) process is shown in Fig. 1. The SFE apparatus consisted of
two intelligent HPLC pumps (model PU-1580, Jasco Corporation,
Tokyo) i.e., CO2 pump and cosolvent pump. The CO2 pump was fit-
ted with a cooling jacket to deliver CO2, and the cosolvent pump
was fitted to deliver cosolvents as modifiers/entrainers. CO2 from
the supply tank was cooled using a cooling jacket attached to the
Intelligent HPLC pump. In order to cool the pump head, an ethylene
glycol-deionised water mixture (50:50, v/v) was circulated through
the cooling jacket using a low temperature bath circulator (model
631D, Tech-Lab Manufacturing Sdn. Bhd., Selangor, Malaysia)
which can deliver coolant down to 20 -
C. In each experiment,
10 g of sample (dry basis) was loaded into a 50 ml extraction vessel
(model EV-3, Jasco Corporation, Tokyo), and the vessel was placed
in a column oven (model CO-1560, Jasco Corporation, Tokyo) at the
desired temperature. The extractions were performed at tempera-
tures ranging from 45 to 75 -
C and at pressures from 20 to 35 MPa
for all of the techniques. The liquid CO2 was compressed to the de-
sired pressure and then continuously pumped through the extrac-
tor at 2 ml min 1 for all of the extractions under various
conditions. Carbon dioxide flowed through a long tube preheating
coil for 4 m to ensure that it reached the extraction temperature
before contacting the samples in the extraction vessel. A back pres-
sure regulator (BPR) (model BP-1580-81, Jasco Corporation, Tokyo)
was used to control the extraction pressure. At supercritical condi-
tions, the oil that was dissolved in the SC-CO2 was separated at the
back pressure regulator. The back pressure regulator was used to
reduce the solvent pressure and to separate the fish oil solute from
the solvent in the trap. The components were separated from the
supercritical phase by a pressure reduction through an expansion
valve. A 100 ml blue cup schott bottle was used as a yield trap.
The yield trap was cooled with an ice-water mixture. The proce-
dures of various extraction techniques for the experiments were
as follows; the system was pressurized over 5 min and CO2 was al-
lowed to flow through the extractor for 6 h continuously at a given
temperature and pressure, which is defined as ‘‘continuous tech-
nique” in this study. The oil was collected at the end of the exper-
iment. The volume of CO2 that followed through the system was
calculated by the Eq. (2) at the end of the run. The extractions were
also performed with CO2 and ethanol (as a cosolvent) at 2 ml min 1
(1.6 ml CO2 and 0.4 ml ethanol/min, v/v), for 6 h continuously at a
given temperature and pressure, which is denoted as ‘‘cosolvent
technique” in this study. For the cosolvent technique, both the
CO2 and cosolvent pumps were connected to the extraction vessel,
which was placed inside the column oven, and were conducted to
maintain the extraction flow rate of the fluid. The back pressure
regulator was also used to control the extraction pressure. The
oil trap containing the extracted oil with ethanol, as a residue of
cosolvent, were evaporated under vacuum at 65 -
C using a Hei-
dolph WB/VV 2000 rotary evaporator (Heidolph, Germany), and
then placed in the oven at 45 -
C for 30 min before being transferred
into the desiccators. The yield was collected at the end of the
experiment. The volume of CO2 and ethanol that followed through
the system was converted to weight and calculated by the Eq. (2)
and (3), respectively at the end of the extraction. For the ‘‘soaking
technique”, the sample was soaked with CO2 in the vessel and held
for 10 h at a desired temperature (45, 55, 65 and 75 -
C) and pres-
sure (20, 25, 30 and 35 MPa) and then extracted continuously for
5 h at same temperature and pressure. The yield was collected at
the end of the extraction. The volume of CO2 that followed through
the system was calculated by the Eq. (2) at the end of the experi-
ment. The sample was pressurized with CO2 and held for 1 h and
then depressurized within 18 min. The pressurization–depressur-
ization treatment is denoted as ‘‘pressure swing technique”. The
experimental design of the pressure swing technique is shown in
Fig. 2. At a given pressure and temperature, the extractor contain-
ing the sample was pressurized (step P1, Fig. 2) with CO2 within
5 min, held for 1 h (step H1, Fig. 2), and then depressurized (step
D1, Fig. 2). The pressure swing technique was applied 10 times
from P1–H1–D1 to P10–H10–D10 (Fig. 2). Thus, the total extrac-
tion period was considered as 10 depressurizations of 18 min each
for a total of 3 h (18 min -
10) extraction time. Oil was collected
during the depressurization steps from D1 to D10. The CO2 con-
sumption after each depressurization was calculated. As CO2 was
not consumed in the holding steps (H1 to H10, Fig. 2), only the
time relating to the depressurization steps counted towards the to-
tal extraction time (D1 to D10, Fig. 2). In other words, CO2 did not flow through the vessel
during the holding steps (H1 to H10,
Fig. 2). Thus, the volume of CO2 that followed through the system
was calculated by the Eq. (2) at the end of the 3 h run. The yield of
total oil and the amount of CO2 used were defined in percent (100 g
of ground fish skin on dry basis) and in g min 1
, respectively, as de-
scribed below:
2.6. Solubility of oil from the skin of Indian mackerel in SC-CO2
Solubility of oil from the skin of Indian mackerel in SC-CO2 was
calculated based on the total oil weight extracted period per
weight of CO2 used as shown below. 2.7. Analysis of fatty acid (FA) constituent by gas
chromatography (GC)
The fatty acid (FA) constituents of the total oil yield extracted by
the Soxhlet and various techniques of SC-CO2 extraction at 35 MPa
and 75 -
C was analysed to determine the fatty acid profile using
gas chromatography (GC) (Model, 4D, Shimadzu, Japan). A 100 ll
aliquot of the test sample was thoroughly mixed by dissolving
0.05 g of sample into 0.8 ml of n-hexane, and 0.2 ml of 1 M sodium
methoxide (30% methanol in sodium methoxide) was added to
prepare fatty acid methyl esters (FAME). The mixture was then
shaken vigorously using an auto-vortexer (Stuart, U.K.) for 30 s
and stored for another 5 min so it formed a bilayer. The clear upper
layer containing the FAMEs (0.5 ll) was pipetted off and injected
into a GC using an external standard method, the PORIM test meth-
od No. p3.4, 1995 (AOCS, 1989). The oven temperature was set at
150 -
C, held for 2 min, increased with a heating rate of 5 -
C/min
up to the final temperature of 250 -
C, and then held for 10 min
at 250 -
C. Analysis was carried out with triplicates. The FA compo-
nents were identified based on the elution order of the reference
standard mixture of the FAs in Menhaden oil; C14:0, C16:0,
C16:1x7, C16:2x4, C16:3x4, C18:0, C18:1x9, C18:1x7,
C18:2x6, C18:3x4, C18:3x3, C18:4x3, C20:1x9, C20:4x6,
C20:4x3, C20:5x3, C22:5x3, and C22:6x3. The FA components
are expressed as a percentage by mass of FAMEs.
2.8. Statistics
Experiments were performed in triplicate and each set of yields
were averaged. The averages and Duncan t-test were computed to
measure the variations in the total oil yield extracted at various
conditions and also to measure the variations in fatty acid constit-
uents of the oil yield using various techniques. The least significant
difference at the 95% confident level was calculated for each
parameter.
2.8. Statistics
Experiments were performed in triplicate and each set of yields
were averaged. The averages and Duncan t-test were computed to
measure the variations in the total oil yield extracted at various
conditions and also to measure the variations in fatty acid constit-
uents of the oil yield using various techniques. The least significant
difference at the 95% confident level was calculated for each
parameter.
3. Results and discussion
3.1. Characterisation of yield
The moisture content of dry ground fish skin was 6.33% (dry ba-
sis), and the oil content as determined by Soxhlet extraction meth-
od was 53.6 ± 0.4/100 g dry ground fish skin. At this moisture level,
the extractability of oil may not be affected; Devittori et al. (2000)
have shown that moisture content ranging from 1% to 8% did not
influence the extractability of oil from ground millet bran.
3.2. Effect of SC-CO2 techniques on the yield
Table 1 shows the effect of the extraction techniques at the var-
ious temperatures ranging from 45 to 75 -
C and pressures ranging
from 20 to 35 MPa on the yield extracted using SC-CO2. The extractability of oil increased with
pressure and temperature in
all extraction modes (Table 1), and it was highest at 35 MPa and
75 -
C. Kil-Yoon et al. (2005) extracted squid oil using SC-CO2,
where the extractability of oil increased with pressure. The extract-
ability of oil was lowest with the continuous elution mode, and it
was the highest with cosolvent extraction but very closely fol-
lowed by both soaking and pressure swing techniques. The highest
oil extractability achieved at 35 MPa and 75 -
C were 53.2, 52.8,
52.3 and 24.7/100 g dry sample by cosolvent extraction, soaking,
pressure swing and continuous extraction, respectively, compared
with the extractability of 53.6/100 g dry sample by the Soxhlet
method. Among the extraction techniques, cosolvent, soaking and
pressure swing techniques were able to almost quantitatively ex-
tract (cf. Soxhlet extraction) oil from the dry skin of Indian mack-
erel at 35 MPa and 75 -
C; whereas the extractability was less
than 50% of the maximum extractable oil by continuous extraction.
This suggests that the solubility of oil in SC-CO2 and/or the mass
transfer of oil from the solid matrices are enhanced by those tech-
niques in relation to continuous SC-CO2 technique. The poor
extractability in the latter presumably due, in part, to solubility
of oil in SC-CO2, since closer to maximum extractability was
achieved with additional of cosolvent; and to a certain extent
due to limitation of diffusion of oil within the solid matrix. More-
over, with continuous elution the extraction of oil appears to have
taken place mainly from the particle surfaces, and the extraction of
oil from the interior of the sample matrices may not have been
complete due to mass transfer limitations within the solid matrix.
The enhanced extractability by cosolvent extraction can be attrib-
uted to enhanced solvent power. Kil-Yoon et al. (2005) observed
marked increase in squid oil solubility with the use of ethanol as
cosolvent at 1.5% (by weight). The enhanced extractability by soak-
ing and pressure swing techniques is presumably due to rupture of
the networks in the solid matrix leading to improved mass transfer
of the oil into SC-CO2. CO2 is soluble in the oil phase, which in-
creases with pressure and temperature. During soaking the dis-
solved CO2 causes swelling of the matrix, which improves mass
transfer. Multiple depressurization steps in the pressure swing
mode may lead to rupture of the matrices, in addition to solubility
of CO2 in the oil phase causing its swelling. Thus the combined ef-
fect makes pressure swing the most effective mode in the extrac-
tion of oil from solid matrices. However, certain minimum
holding time before depressurization would be necessary to allow
penetration of CO2 into the matrix and its dissolution in the oil
phase. Therefore, when applying soaking and pressure swing tech-
niques, some holding time of CO2 before depressurization seemed
to be very effective to allow penetration of CO2 into the sample
matrices. Zaidul et al. (2007) successfully extracted almost double
the amount of palm kernel oil using the pressure swing technique
than the continuous technique from the whole and untreated palm
kernel with a similar or even less amount of CO2. The pressure
swing technique required lower amount of CO2 than any other
technique, and the consumption of CO2 in soaking technique required almost similar amount of
a cosolvent technique for yielding
a similar amount of oil. EPA and DHA are labile molecules that
could be altered by high-temperatures required for solvent evapo-
ration (Létisse et al., 2006). Based on these observations, SC-CO2
extraction without using any organic modifiers is optimal to isolate
these valuable molecules from the waste of the canned fish
industry.
3.3. Effect of SC-CO2 techniques on CO2 consumption
Table 2 shows the total amount of CO2 consumption required to
extract the amount of oil using the various techniques. The total
amount of CO2 consumption decreased significantly with temper-
ature and increased with pressure in all extraction modes using
SC-CO2. A higher amount of CO2 was needed for the continuous
technique, compared to the techniques of cosolvent, soaking and
pressure swing at all pressures and temperatures. The quantity of
CO2 used in the continuous technique was about 1.1, 1.2 and 2.0
folds higher than cosolvent extraction, soaking technique and pres-
sure swing technique, respectively. Maximum extractability of oil
with least amount of CO2 consumption was achieved with pressure
swing mode at 35 MPa and 75 -
C. Thus, pressure swing technique
would make economic sense while providing an optimal final
product. Higher yields using lower amounts of CO2 are economi-
cally better and provide an optimal final product for human con-
sumption (Saldana et al., 2002).
3.4. Solubility of oil from skin of Indian mackerel in SC-CO2
The solubility of oil from skin of Indian mackerel in SC-CO2 in-
creased with both temperature and pressure in continuous (Fig. 3),
cosolvent (Fig. 4), soaking (Fig. 5), and pressure swing (Fig. 6) tech-
niques. The significant differences were observed in the solubility
of the oil in SC-CO2 at any temperature and pressure within contin-
uous, cosolvent and pressure swing techniques. The solubility be-
tween cosolvent and soaking techniques were not significantly
different at any temperature and pressure. With continuous tech-
nique the solubility of the oil in SC-CO2 increased from 1.98 to
4.25 g oil/100 g of CO2 as the pressure and temperature were in-
creased from 20 to 35 MPa and 45 to 75 -
C, respectively (Fig. 3);
whereas, with cosolvent technique the solubility of the oil in SC-
CO2 increased from 5.97 to 10.78 g oil/100 g of CO2 (Fig. 4) that
was of about three times higher than the continuous technique
as the pressure and temperature were increased from 20 to
35 MPa and 45 to 75 -
C, respectively. Similar observation was
made in the soaking technique where the solubility increased from
6.54 to 10.89 g oil/100 g of CO2 at 20 to 35 and 45 to 75 -
C, respec-
tively (Fig. 5). A greater increase in the solubility of the oil in SC-
CO2 was observed in pressure swing technique. The solubility
was increased from 9.53 to 17.98 g oil/100 g of CO2 at 20 to 35
and 45 to 75 -
C, respectively (Fig. 6).
Solubility of substances in supercritical fluids generally in-
creases with pressure. However, temperature affects solubility in a more complicated way, i.e.,
solubility decreases with tempera-
ture up to certain level of pressure, but it increases above that pres-
sure (inversion pressure). Above the inversion pressure, isothermal
solubility increases with pressure; and isobaric solubility increases
with temperature. A similar solubility trend was observed in the extraction of oil from Indian
mackerel skin in this work. At higher
pressures the solubility of the oil increased due to the increase in
density of CO2. As the density increases, the distance between mol-
ecules decreases and the interaction between oil and CO2 in-
creases, leading to greater oil solubility in CO2 (deCastro et al.,
1994). The density of SC-CO2 as a function of pressure and temper-
ature plays a major role in affecting the yield. When there is an in-
crease in pressure and a concomitant decrease in the temperature,
this allows greater contact between the solute and the solvent
leading to enhanced solubility of the solute thus increasing the
yield. The cosolvent interacts strongly with the solute and signifi-
cantly increases the solubility (Dixon and Johnston, 1997; Akger-
man and Giridhar, 1994). For example, addition of a small
amount (0.4 mol%) of ethanol to CO2 increased the solubility of
the oil by an order of magnitude in this study. This is attributable to increased polarity of SC-
CO2 by the addition of ethanol, which
effectively increases the solvent capacity leading to enhanced
extraction of relatively more polar unsaturated triglycerides. A
swelling phenomenon of oil has been reported by a number of
authors (Patel et al., 2006; Suzuki et al., 1997; Smith et al.,
1998). Suzuki et al. (1997) and Smith et al. (1998) found that the
volume expansion of oils in contact with supercritical CO2 could
be as high as 200% for n-decane or close to 50% for fish oil methyl
esters. These authors also confirmed that cashew nut liquid shell
(CNSL) swelled by examining the CNSL after a single depressuriza-
tion (Smith et al., 2003). They examined that the CNSL swelling in-
creased the available surface area of the CNSL to the bulk CO2
phase and would account for some of the higher extraction yields
observed in their studies. Thus it would seem appropriate, in the
pressure swing technique, to hold CO2 for certain time to allow
penetration of CO2 into the solid matrices and its dissolution in
the oil phase.
3.5. Fatty acid composition of the oil recovered by different modes of
extraction
Table 3 shows the fatty acid constituents of the fish oil ex-
tracted by the Soxhlet and other SC-CO2 techniques at 35 MPa
and 75 -
C. While almost quantitative extraction of oil was achieved
with pressure swing, soaking and co-solvent techniques, only a
partial extraction was possible with continuous elution due to low-
er solvent capacity of SC-CO2 compared with SC-CO2 containing
ethanol as well as mass transfer limitations. From the fatty acid
composition of the oil obtained by continuous elution, it is evident
that more saturated triglycerides are preferentially extracted by
SC-CO2 compared with polyunsaturated triglycerides. In other
words, the oil was fractionated to some extent in the continuous
mode, unlike the other extraction methods. The extraction of tri-
glycerides containing x-3 fatty acids (EPA, DPA and DHA) was
not fully completed in the continuous extraction as seen by the
lower extractability achievable with this technique compared with
either cosolvent, soaking or pressure swing method. On the other
hand, higher levels of triglycerides containing saturated fatty acids
(C14:0, C16:0 and C18:0) were extracted in the continuous elution compared with the other
methods. In the latter methods, no frac-
tionation of the oil was apparent as seen by the high extractability
comparable to that of Soxhlet extraction; and the fatty acid compo-
sition of the oil obtained by those techniques, including PUFA con-
stituents, were also more or less identical to that of the oil obtained
by Soxhlet method. Osman et al. (2007) extracted the total oil from
Indian mackerel of Malaysian water and analyzed the fatty acid
constituents using GC. The authors reported a 16.1, 19.3, 4.2,
36.5 and 13.6% of EPA, DHA, AA, PUFA (x-3) and PUFA (x-6),
respectively, which was similar to our observations. Létisse et al.
(2006) extracted oil from sardines using SC-CO2 at 30 MPa and
75 -
C without an organic modifier and achieved the highest yield.
Authors also reported that a high amount of PUFA, comprising of
10.9% EPA and 13.0% DHA, could be obtained under the same
extraction conditions. These results are in agreement with our
findings.
4. Conclusion
Various techniques such as continuous, cosolvent, soaking and
pressure swing of SC-CO2 were applied to extract the oil from
the skin of Indian mackerel (Rastrelliger kanagurta), and the yield
from these techniques were compared to the yield from applying
the Soxhlet method. The yield from the Soxhlet extraction was
53.6/100 g sample (dry basis) was closer to the yields of the cosol-
vent (53.2/100 g sample dry basis), soaking (52.8/100 g sample dry
basis) and pressure swing techniques (52.3/100 g sample dry basis)
extracted at 35 MPa and 75 -
C. The PUFA constituents of cosolvent,
soaking and pressure swing techniques were also similar to the
Soxhlet method. The highest recoveries of AA, DHA and EPA were
also obtained from the soaking and pressure swing techniques.
The consumption of CO2 in the continuous elution mode was 1.1,
1.2 and 2.0 folds higher than cosolvent extraction, soaking and
pressure swing methods, respectively. Both pressure swing and
soaking methods are recommended as they do not involve organic
solvent entrainers. Between them, pressure swing technique is
more attractive from the standpoints of extractability, CO2 con-
sumption as well as recovery of AA, EPA and DHA from Indian
mackerel skin.
Acknowledgements
This work was financially supported by the Research University
Grant Scheme (RUGS) of Universiti Putra Malaysia and the Third
World Organization for Women in Science (TWOWS).