supercritical fluid extraction of organochlorines from fish muscle with different sample preparation
TRANSCRIPT
Supercritical fluid extraction of organochlorines from fishmuscle with different sample preparation
Paulo Antunes a, Odete Gil b, M. Gabriela Bernardo-Gil a,�a Centre for Biological and Chemical Engineering, IST-DEQ, Av. Rovisco Pais, 1049-001 Lisbon, Portugal
b Instituto de Investigacao das Pescas e do Mar, Av. Brasılia, 1449-006 Lisbon, Portugal
Received 2 January 2002; received in revised form 14 May 2002; accepted 23 May 2002
Abstract
Supercritical carbon dioxide was used to extract PCBs, p ,p ?-DDE, p ,p ?-DDD, p ,p ?-DDT and dieldrin from fillets of
black scabbardfish (Aphanopus carbo ). Using standards, the optimal extraction conditions of pressure and temperature
were determined using the statistical method of central composite surface design. The studied temperatures ranged from
309 to 337 K and pressure from 10 to 24 MPa. It was observed that temperature do not affect significantly the
extraction and, on the contrary, a significant effect was recorded for pressure. The best extraction efficiencies were
obtained at pressures near 14 MPa, for all the studied compounds. To study the matrix effect on the yields, three types
of raw materials were selected: fresh fillet, fresh fillet with anhydrous sodium sulphate and freeze-dried fillet. It was
observed that supercritical carbon dioxide can extract efficiently the organochlorine compounds from freeze-dried
fillets, but very low recoveries were obtained from fresh fillets. The pressure effect on extraction was studied for the
freeze-dried samples. Three pressures were tested: 18, 22 and 26 MPa, maintaining the temperature at 328 K. Analysing
the initial extraction velocities it was observed that 22 MPa was the optimal extraction pressure. The concentration of
organochlorine compounds obtained by supercritical fluid extraction was, in some cases, higher than those obtained by
Soxhlet extraction, using n -hexane.
# 2002 Elsevier Science B.V. All rights reserved.
Keywords: Supercritical extraction; PCB; DDT; Dieldrin; Black scabbardfish
1. Introduction
Polychlorinated biphenyls (PCBs), and chlori-
nated pesticides, such as dichlorodiphenyltrichlor-
oethane (DDT) and its breakdown products
dichlorodiphenylethane (DDE) and dichlorodi-
phenyldichloroethane (DDD), and dieldrin are
persistent and toxic environmental pollutants.
Due to their lipophilic nature they concentrate in
fatty tissues of aquatic organisms such as fish [1]
and undergo bioaccumulation through food web
[2]. Consumers of this fish, including human, may
also be affected. Therefore, there are increasing
public concern over the presence of these com-
pounds in the environment and food supply.
� Corresponding author. Tel.: �/351-21-841-7582; fax: �/
351-21-841-9176
E-mail address: [email protected] (M.G. Bernardo-
Gil).
J. of Supercritical Fluids 25 (2003) 135�/142
www.elsevier.com/locate/supflu
0896-8446/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 8 9 6 - 8 4 4 6 ( 0 2 ) 0 0 1 4 1 - 9
Although analytical approaches have improved,
most studies rely on time-consuming procedures.
These require large volume of expensive and toxic
solvents, much of which must eventually be
disposed of as hazardous waste.
Multiple solvent concentration steps are com-
mon, providing ample opportunity for introduc-
tion of laboratory contaminants, loss of target
compounds, and worker exposure. Time and
labour to perform these and other necessary tasks,
such as cleaning large amount of glassware to
remove trace levels of contaminants, are signifi-
cant and result in considerable analysis costs and
delay in obtaining results.
Supercritical fluid extraction (SFE) has received
increasing attention as the solution for these
problems [3]. A commonly used extracting solvent
in SFE is CO2.
Recently several researchers have examined the
suitability of SFE for the extraction of organo-
chlorines from soil [4], other natural products [5,6]
and fish tissues [7�/9]. However, most of these
studies have been made in spiked samples.
The aim of this work was to determine and
optimise the experimental conditions, in terms of
pressure, temperature and sample preparation for
SFE of organochlorine compounds from fish
tissues.
2. Materials and methods
2.1. Materials
Solvents */Acetone and n-hexane distilled in the
laboratory. Dichloromethane (Merck). Sodium
sulphate */Anhydrous (Merck), heated at 440 8Covernight. Standards */PCB congeners were ob-tained, as crystals, from Promochem, and p ,p ?-DDT, p ,p ?-DDD, p ,p ?-DDE and dieldrin were
purchased from Supelco. Florisil */60-100 mesh
(Merck). Carbon dioxide */N48 grade (99.998% of
purity), supplied by AirLiquide (Portugal), using
activated charcoal filter.
2.2. Extraction
SFE was performed on an apparatus build in-
house similar to the one described by Esquıvel and
Bernardo-Gil [10] with a fixed-bed tubular extrac-
tor of 0.028 L (section area of 1.13 cm2) and with
some modifications in the decompression zone inorder to minimise losses and deposits of the
extracts in valves and piping. A schematic diagram
is shown in Fig. 1. Carbon dioxide was delivered at
a pressure of about 6 MPa and compressed until
extraction conditions by an air driven liquid pump
(B1) model MCP-71 (Haskel Inc., USA) after
Fig. 1. Schematic representation of the extraction apparatus. V1, cut valve; V2, safety valve; V3, cut valve; V6, cut valve; V8,
micrometric valve; P1, pressure in CO2 bottle; P2, pressure before extraction cell; B1, CO2 pump; C, extraction cell; L, opening for
wash; R1, collector; F, flow meter; T, thermometer; DTM, dry test meter.
P. Antunes et al. / J. of Supercritical Fluids 25 (2003) 135�/142136
cooling in a bath at about 277 K. The CO2 was
then heated by a heat exchanger to reach extrac-
tion temperature before entering the extractor (C).
The first valve (V6) (Hoke Inc., Cresskill, USA)
was totally open during extraction, and the second
valve (V8) is used for decompression and flow
control. A dry test meter (American Meter Com-
pany, Philadelphia, USA, DTM-200A) with a
accuracy of 9/0.05 L was used to measure the
delivered CO2.
Before each set of determinations, the extraction
cell was manually filled with the sample. When the
temperature of extraction cell was raised to a
predetermined value, CO2 was pumped into the
extractor up to the desired extraction pressure
value. After ensuring that there was no leak in the
equipment, the expansion valve (V8) was opened
and a steady stream of the CO2 was allowed to
pass upward through the bed. In order to obtain
intermediate points to an extraction curve after a
given extraction period, the first valve (V6) was
closed, the expansion valve and pipes leading from
the extractor to the collector were washed with
hexane to remove any oil trapped in this region,
and then sucked out with a vacuum pump in order
to get all the extracted compounds in a stainless
steel tube of about 10 mL (R1), to be analysed.
Each cleaning step takes about 10 min.
Using a solution of pure standards of PCB
congeners (IUPAC No. 101, 153, 105, 138, 183,
187, 128 and 194) and p ,p ?-DDE, ten runs weredone to determine the optimal conditions of
pressure and temperature based on a response
surface methodology (RSM) [11] using a central
composite design, presented in Table 1. The ranges
of pressure and temperature used were, respec-
tively, 9.9�/24.1 MPa, and 309�/337 K. The super-
ficial velocity was controlled around 0.08 cm s�1.
Muscle of black scabbardfish (Aphanopus carbo )captured in coastal zone near Sesimbra (Portugal),
were extracted by using supercritical carbon diox-
ide. Fresh and lyophilised muscle tissues ground in
a mechanical grinding device were examined. SFE
was performed at selected conditions of pressure
and temperature.
Samples of raw material were also Soxhlet
extracted with hexane for 6 h. Aliquots of theextracts were evaporated in a rotary evaporator
and fat content was determined gravimetrically.
Both Soxhlet and SFE extracts were concen-
trated and passed through a Florisil column for
separation of the compounds. The first fraction
eluted with hexane contained PCBs and p ,p ?-DDE
and the second fraction eluted with dichloro-
methane:hexane (30:70) contained dieldrin, p ,p ?-DDD and p ,p ?-DDT. A further clean-up with
sulphuric acid was also made. Recovery of the
Florisil column was evaluated with a standard
solution and more than 85% of each compound
was obtained.
2.3. Gas chromatography
Soxhlet and SFE extracts were analysed using aHewlett-Packard 5890 Series II gas chromato-
graph equipped with an electron capture detector
and a DB-5 (J&W Scientific) capillary column (60
m�/0.25 mm i.d.). The column was held at 60 8Cfor 1 min, then programmed in three levels: at a
rate of 20 8C min�1�/210 8C (8 min); 2 8Cmin�1�/250 8C (17 min) and 4 8C min�1 to a
final temperature of 260 8C (15 min). The injectortemperature was kept at 270 8C and the detector
was maintained at 320 8C. Helium and argon:-
methane (90:10) were used as the carrier and the
make-up gases, respectively.
Concentrations of CBs and pesticides were
quantified from the peak heights using a 6-point
Table 1
Central composite design of temperature and pressure condi-
tions for optimisation of SFE, using pure standards
Run Coded independent vari-
able
Uncoded independent vari-
able
T P T (K) P (MPa)
A 0 0 323 17.0
B �/1.414 0 309 17.0
C �/1 �/1 313 12.0
D �/1 1 313 22.0
E 0 �/1.414 323 9.9
F 0 1.414 323 24.1
G 0 0 323 17.0
H 1 1 333 22.0
I 1 �/1 333 12.0
J 1.414 0 337 17.0
P. Antunes et al. / J. of Supercritical Fluids 25 (2003) 135�/142 137
Table 2
Experimental conditions of SFE of a standard solution of PCB congeners and p ,p ?-DDE and extraction times to achieve 90% of the final recovery (t90)
Run T (K) P (MPa) Superficial velocity (cm s�1) t90 (min)
PCB congener DDE
101 153 105 138 187 183 128 194
A 323 17.0 0.101 3.8 4.1 5.3 3.1 2.6 2.7 7.2 3.1 3.5
B 309 17.0 0.081 5.4 4.2 5.5 4.9 4.3 4.0 5.9 4.0 4.4
C 313 12.0 0.078 6.9 7.5 7.8 7.4 7.1 7.2 7.6 7.2 7.2
D 313 22.0 0.070 22.1 19.4 22.6 19.3 17.1 15.2 20.6 18.5 23.1
E 323 9.9 0.080 3.0 3.2 6.6 4.3 3.5 3.2 5.1 3.8 3.1
F 323 24.1 0.070 27.8 25.1 23.6 22.5 23.3 20.0 19.8 14.6 27.0
G 323 17.0 0.079 4.3 3.6 4.1 3.6 3.0 3.1 3.5 2.4 4.9
H 333 22.0 0.075 12.7 11.9 13.0 11.5 10.4 10.0 12.0 7.6 11.4
I 333 12.0 0.077 6.0 5.8 7.9 6.0 5.4 5.6 7.2 7.2 5.3
J 337 17.0 0.074 5.5 4.9 5.8 4.8 4.4 4.0 5.2 4.1 4.7
P.
An
tun
eset
al.
/J
.o
fS
up
ercritical
Flu
ids
25
(2
00
3)
13
5�
/14
21
38
multilevel calibration curve. A mixture of indivi-
dual PCB congeners, p ,p ?-DDT, p ,p ?-DDD, p ,p ?-DDE and dieldrin was used for quantification.
3. Results and discussion
3.1. SFE at different conditions with standards
Ten conditions were selected by using thecentral composite technique (Table 1) to determine
the optimal extraction conditions of pressure and
temperature. Extraction curves were conducted by
using a standard solution of eight pure PCB
congeners (IUPAC No 101, 153, 105, 138, 183,
187, 128, 198) and DDE. To obtain these curves,
samples were collected at 3, 6, 10 and 25 min.
In order to normalise errors and to proceedcalculations, the extraction curves were fitted to a
kinetic Eq. (1) [12], and final recoveries were
determined.
Y �a�(a(1�c)�bTime(1�c))(1=1�c) (1)
where Y is the recovery of extraction at a certain
extraction time (Time), and a , b and c are adjusted
parameters.
The extraction times to achieve 90% of thesevalues, in minutes (t90) (Table 2) were used as
dependent variables in the RSM treatment [11]. It
was verified that temperature had a small influence
in extraction (factors of T and T2 were not
significant, as p �/0.2 [13]). Therefore, the re-
sponse surfaces were obtained using only the
pressure (P ), in MPa:
t90 �a0�a1P�a2P2 (2)
In Table 3 are presented the values of the
coefficients a0, a1 e a2, and the statistical para-
meters: p represents the p-level, which is a measure
of the statistical significance; R2 can be interpretedas the proportion of the variability in the data, and
R2Adj is the adjusted R -square, which depends on
the R -square and the number of degrees of free-
dom [11,13]. The response surfaces were identical
for all the analysed compounds. Fig. 2 illustrates
the response surface for CB153, as an example.
Table 3
Results of RSM for SFE of a standard solution, factors according to the expression: t90�/a0�/a1P�/a2P2, statistical parameters and the
optimal pressure for the analysed compounds
PCB congeners DDE
101 153 105 138 187 183 128 194
a0 45.37 44.27 48.57 43.73 44.49 38.55 34.93 32.44 45.88
a1 �/6.14 �/5.87 �/6.11 �/5.66 �/5.83 �/4.96 �/4.33 �/3.96 �/6.20
a2 0.22 0.21 0.21 0.20 0.20 0.17 0.16 0.14 0.22
R2 0.88 0.88 0.86 0.87 0.88 0.87 0.81 0.62 0.85
R2Adj 0.85 0.85 0.81 0.83 0.84 0.83 0.75 0.51 0.81
p B/0.02 B/0.01 B/0.01 B/0.01 B/0.01 B/0.01 B/0.04 B/0.1 B/0.03
Optimal pressure (MPa) 13.8 14.1 14.4 14.3 14.4 14.4 13.9 14.5 13.9
Fig. 2. Response surface for optimisation of extraction times
for 90% of final recoveries, for CB 153.
P. Antunes et al. / J. of Supercritical Fluids 25 (2003) 135�/142 139
The surfaces showed a minimum and an abrupt
increase at 24.1 MPa. This may be due to back-
diffusion originated by the increase of diffusivity
of the analysed compounds, combined with the
low superficial velocity. To obtain the optimal
pressure, it was determined the minimum of theextraction time, by using the first derivative. A
value of about 14 MPa was found as the optimal
pressure for the extraction of the studied com-
pounds (Table 3).
3.2. SFE with fish samples
The fish extractions were carried out with
muscle samples of black scabbardfish. Humidity
and lipid content of samples ranged from 76 to
82% and 1.2�/14%, respectively. For all the extrac-
tions, the temperature was kept constant at 328 K,
because it was verified a small influence of this
parameter on SFE with standards.
3.2.1. Extractability as a function of matrix sample
The influence of matrix sample was evaluated by
using three different sample preparations: fresh
fillet, fresh fillet with anhydrous sodium sulphate
and freeze-dried fillet.
3.2.2. SFE of fresh fillet
The experiments with fresh fillet were performed
at 328 K, pressures of 10, 14, and 18 MPa and an
extraction time of 60 min. SFE recoveries were
calculated comparing the results with the amounts
obtained by Soxhlet extraction (Table 4). Very low
recoveries were obtained. On the other hand,
Soxhlet extractions of solid residues showed high
content of the analysed compounds. These results
indicate that SFE do not extract efficiently the
organochlorines from fresh fish.
As very low organochlorine recoveries were
obtained at 10 and 14 MPa, and a slight increase
Fig. 3. Comparison of SFE at 18 MPa and 328 K of different
sample preparations of fish with Soxhlet extraction.
Table 4
Concentration of organochlorines in muscle of black scabbardfish, fresh fillet SFE recoveries, and Soxhlet recoveries of SFE solid
residues
Compound Concentration (ng g�1) SFE of fresh fillet recovery (%) Soxhlet of SFE solid residues recovery (%)
Soxhlet 10 MPa 14 MPa 18 MPa 10 MPa 14 MPa 18 MPa
CB101 1.6 2 2 4 65 45 43
CB153 5.4 1 1 3 60 47 41
CB105 1.5 2 4 9 71 54 41
CB138 4.7 1 1 4 51 53 42
CB187 2.1 0 1 4 62 52 41
CB183 0.73 1 4 6 64 54 45
CB128 0.72 2 5 13 64 53 42
CB194 0.27 4 9 13 65 54 52
DDE 34 0 0 2 57 48 38
P. Antunes et al. / J. of Supercritical Fluids 25 (2003) 135�/142140
was observed at 18 MPa (Table 4), the other
sample preparations were tested only at 18 MPa.
3.2.3. SFE of fresh fillet with anhydrous sodium
sulphate
An extraction of fresh fillet with anhydrous
sodium sulphate was performed at 328 K, 18 MPa
and an extraction time of 60 min. An increase of
the efficiency was obtained, in relation to freshfish, however, with very low recoveries also (about
30%).
3.2.4. SFE of freeze dried fillets
An extraction was carried out with freeze-dried
muscle sample at 328 K, 18 MPa and 60 min of
extraction time. Higher recoveries were observed
in relation to fresh samples, being obtained values
of 70�/110% for PCBs and 130% for DDE.
In Fig. 3 comparisons of the CB recoveries, in
fish samples are made. The analysis of the results
showed that the presence of water decreased theextraction efficiency. The amount of the studied
congeners extracted from lyophilised samples was
similar to Soxhlet extracts, and ten to 20 fold or
nearly triple of the extracted from samples of fresh
fish or fish dried with anhydrous sodium sulphate,
respectively. These results suggest that, to a certain
degree, at the studied pressure, the presence of
water hinders the diffusion of CO2 in the whole
cell, and inhibits the contact between CO2 and the
extracting compounds, therefore, the yields are
low.
3.2.5. Extractability as a function of pressure
Lee et al. [8] observed higher efficiencies of SFE
of PCBs, at 34 MPa than at 15 MPa, using fish
samples with anhydrous sodium sulphate. With
lyophilised fish, Bøwadt et al. [9] obtained good
recoveries at 22 and 38 MPa. So, unlike the results
for the pure standards, pressures of 18, 22 and 26MPa were selected for the study of the influence of
pressure on the SFE of PCB congeners, DDE,
DDD, DDT and dieldrin. As the SFE of fresh fish,
at the studied conditions, were not efficient,
lyophilised fish tissues were selected for this study.
Three different fish samples were used for each
set of pressure conditions, at 328 K and extraction
times of 10, 20, 30 and 60 min. For eachcompound, initial extraction velocities were calcu-
lated based on the extraction curves obtained for
each run. Fig. 4 shows the variation of the initial
extraction velocity with pressure. An increase of
pressure from 18 to 22 MPa enhanced the initial
extraction velocity, and, in general, no significant
Fig. 4. Initial extraction velocities for different organochlorines vs. pressure: j, sample 1; ^, sample 2; ", sample 3.
P. Antunes et al. / J. of Supercritical Fluids 25 (2003) 135�/142 141
variation between 22 and 26 MPa was observed.The pressure of 22 MPa is, therefore, recom-
mended for this analysis.
4. Conclusions
The SFE method, at the studied conditions, can
extract organochlorine compounds from freeze-
dried fish fillets, but do not extract efficiently
organochlorines from fresh fish, even when added
anhydrous sodium sulphate. Pressure has a sig-
nificant effect on extraction, on the other hand,temperature does not affect significantly the effi-
ciency of the extraction. At 328 K the optimum, in
terms of recoveries, was found at 22 MPa.
The method was less time consuming, required
less organic solvents and clean glass material than
Soxhlet extraction, but produced comparable
results.
Acknowledgements
This work was partially supported by the grant
Praxis XXI/BM/20517/99 of Fundacao para aCiencia e Tecnologia (Portugal).
References
[1] P. Antunes, O. Gil, P. Sobral, J. Charrao, Nıveis de
compostos organoclorados em peixes capturados na Ria
de Aveiro, in: F.X. Malcata, F.J. Carballo (Eds.), Proceed-
ings of the 5o Encontro de Quımica dos Alimentos, 2001,
pp. 189.
[2] D. Pastor, J. Boix, V. Fernandez, J. Albaiges, Bioaccumu-
lation of organochlorinated contaminants in three estuar-
ine fish species (Mullus barbatus , Mugil cephalus and
Dicentrarchus labrax ), Mar. Pollut. Bull. 32 (1996) 257.
[3] S.B. Hawthorn, Analytical-scale supercritical fluid extrac-
tion, Anal. Chem. 62 (1990) 633A.
[4] B.E. Berg, H.S. Lund, A. Kingstad, A.L. Kvernheim,
Routine analysis of hydrocarbons, PCB and PAH in
marine sediments using supercritical CO2 extraction,
Chemosphere 38 (1999) 587.
[5] W. Fiddler, J.W. Pensabene, R.A. Gates, D.J. Donoghue,
Supercritical fluid extraction of organochlorine pesticides
in eggs, J. Agric. Food Chem. 47 (1999) 206.
[6] Y.C. Ling, H.C. Teng, C. Cartwright, Supercritical
fluid extraction and clean-up of organochlorine pesticides
in Chinese herbal medicine, J. Chromatogr. A 835 (1999)
145.
[7] R.C. Hale, M.O. Gaylor, J.F. Thames, C.L. Smith, R.F.
Mothershead, II, Robusteness of supercritical fluid extrac-
tion (SFE) in environmental studies: analysis of organo-
chlorinated pollutants in tissues from the osprey (Pandion
haliaetus ) and several fish species, Intern. J. Environ. Anal.
Chem. 64 (1996) 11.
[8] H.B. Lee, T.E. Peart, A.J. Niimi, C.R. Knipe, Rapid
supercritical carbon dioxide extraction method for deter-
mination of polychlorinated biphenyls in fish, J. AOAC
Int. 78 (1995) 437.
[9] S. Bøwadt, B. Johansen, P. Fruekilde, M. Hasen, D. Zilli,
B. Larsem, J. de Boer, Supercritical fluid extraction of
polychlorinated biphenyls from lyophilised fish tissue, J
Chromatogr. A 675 (1994) 189.
[10] M.M. Esquıvel, M.G. Bernardo-Gil, Extraction of olive
husk oil with compressed carbon dioxide, J. Supercrit.
Fluids 6 (1993) 91.
[11] D.C. Montgomery, Design and Analysis of Experiments,
fourth ed., Wiley, New York, 1997.
[12] P.W. Atkins, Physical Chemistry, fourth ed., Oxford, 1992.
[13] StatSoft, Inc., STATISTICA for Windows (Computer pro-
gram manual), 1995.
P. Antunes et al. / J. of Supercritical Fluids 25 (2003) 135�/142142