THE CHEMICAL

ENjXW~hfG
The ChemicalEngineering
Journal 61 (1996) 227-231

Short communication
Recovery of grape seed oil by liquid and supercritical carbon dioxide
extraction: a comparison with conventional solvent extraction
A. Molero Gbmez, C. Pereyra L6pez, E. Martinez de la Ossa
Department of Chemical Engineering, University of Cddiz, Apdo. 40, E-11510 Puerto Real (Cddiz), Spain

Received29 November 1994;accepted20 June 1995

Abstract

In this work the extraction of grape seed oil by means of liquid and supercritical carbon dioxide as solvent is described. The operating
conditions to determine the maximum extraction yield were studied.The efficiency of supercriticalfluidextraction (WE) was similar to that
obtained by conventional liquid extraction, but the quality of the supercriticallyextracted oil was higher, equivalent to a degununed, liquid-
extracted oil. It is considered that SFE is competitive with conventional liquid extraction, because the solvent distillation and oil refining
stages can be omitted.

Keywords: Grape seed oil; Liquid/supercriticalcarbon dioxideextraction;Conventionalsolvent extraction

1. Introduction The operating conditions for effective extraction of grape

seed oil using liquid and supercritical carbon dioxide have
been studied both for the process itself (pressure, temperature
The application of supercritical fluid extraction (SFE),
and flow rate of solvent) and for the pretreatment of the grape
particularly the use of liquid and supercritical carbon dioxide,
seeds (humidity and size). Finally the physicochemical char-
has received much attention in the food industry in the last
acterization and fatty acid composition of the supercritically
few years. This separation technique offers extraction yields
extracted oil are discussed.
comparable with those obtained by conventional extraction
methods using organic solvents. Moreover, in contrast with
organic solvents, carbon dioxide is non-toxic, non-flamm- 2. Materials and methods
able, non-corrosive, cheap and readily available in large
quantities with high purity. Since CO* also has a relatively The grape seeds (type airen) used in this study had pre-
low critical pressure (73.8 atm) and critical temperature viously been treated in the wine fermentation industry with
(3 1.1 “C) , it can be considered an ideal solvent for the treat- hot water in order to recover all the remaining sugars, since
ment of natural products. this wash water is normally fermented and distilled to obtain
A variety of processes involving extraction with supercrit- alcohols. As a consequence of the hot water treatment, the
ical fluids (SFs) have been developed as promising alterna- total oil content of treated grape seeds is always lower than
tives to the current separation processes, and industrial that of untreated grape seeds [ 6-101.
applications of SE using carbon dioxide have increased in The seeds were treated under a range of conditions
the last few years, e.g. the decaffeination of coffee [ 1,2], the whereby their grinding size and subsequent drying time were
extraction of hops [ 1,2] and spices [ 31, etc. modified. The milling was carried out in a Futurmat model
Grape seed is a byproduct of the wine fermentation indus- FP coffee mill and the drying in a Heraeus heater at 70 “C.
try and is generally disposed of by burning, although it is To determine the conditions leading to the optimum extrac-
sometimes used for cattle feed. However, grape seed oil has tion yields, a size range of 0.35-2.83 mm and a humidity
many advantages for human consumption owing to its high range of 0.35%-6.50% were studied. .
level of unsaturated fatty acids [ 4,5]. In this work we aim to Conventional extraction was carried out using hexane in a
show that it is possible and economically attractive to extract Soxhlet apparatus for 20 h (with a fraction grape seed size
oil from grape seeds by using supercritical carbon dioxide as of 0.75 mm and humidity less than 0.35%) to guarantee the
solvent. maximum extraction efficiency.

0923~0467/96/$15.000 19% Elsevier Science S.A. All rights reserved

SSDIO923-0467(95)03040-9
228 A. Molero et al. /The Chemical Engineering Joumal61(1996) 227-231

is expressed as the ratio of the amount of extracted oil to the

amount of grape seed placed in the extractor vessel.
To characterize the extracted oils, the following parameters
were determined: refractive index (refractometer, Atago
model 88 141) , density (densimeter, Paar model DMA 48))
viscosity (drop ball viscosimeter, Haake model B/BH),
!. / ” +J v-7
/ &
“-6 :
absorption at 290 nm (W-visible spectrophotometer, Shi-
! ‘A--
E madazu model MPS-2000)) unsaponifiable and free fatty acid
!
I
fractions and iodine, peroxide and saponification indices. All
these parameters were determined according to the Associa-
tion of Official and Analytical Chemists’ Society procedures
[Jll.
The fatty acid composition of the oils after saponification
Fig. 1. Schematic diagram of supercriticaJ CO, extraction apparatus: B, [ 123 was determined by gas chromatography (GC).A Hew-
liquid CO* cylinder; J, cooler; C, check valve; P, pump; PRV, pressure lett-Packard chromatograph, model 5890, equipped with a
regulating valve; E, extraction vessel; S, separator; EHJ, electrical heating flame ionization detector and a Carbowax 20M fused silica
jacket; FMV, Row micrometer valve; T/C-1,2, thermocouples; V-1.8, shut-
capillary column of 0.2 mm internal diameter (i.d.) and 30
off valves; PC-1,2, pressure gauges; F-l, coalescing filter; F-2, carbon
absorption filter; FM, flowmeter. m length was used.
The water content (humidity) of the grape seeds was deter-
The flow diagram of the equipment used for supercritical mined using a model MicroKF 2025 Crison titration appa-
extractions is shown in Fig. 1. Liquid CO* was cooled (J) to ratus by Karl-Fisher.
prevent its gasification and introduced into a pump (P) of All products and chemicals employed were “analytical
46-460 ml h-’ capacity with a cooling head. The extraction reagent” grade. Commercial grade (99.95%) CO* was used.
vessel (E) was a 75 ml capacity, 316 stainless steel (SS)
high pressure cylinder with an electrical heating jacket
(EHJ). A sample of about 40 g mass was placed in the 3. Results and discussion
extractor vessel. The pressure of the CO* cylinder was main-
tained throughout the system by opening valves V-l, V-2, V-
3. I. Operating conditions for extraction processes
3 and V-4 and closing the other valves (V-5, V-6 and V-7).
Before the pump was switched on, valves V-l, V-2, V-3 and
V-4 were closed. Then the pump was switched on and its Fig. 2 shows the effect of pressure on the extraction yield
output pressure increased by the pressure regulating valve of grape seed oil at lo,40 and 60 “C. Sharp changes in the
(PRV) to reach the required processing pressure. Subse- amount of oil extracted were observed around the critical
quently the compressed CO2 was introduced into the extractor pressure of carbon dioxide. The changes seem to correspond
vessel by opening valves V-3 and V-2. When the desired to changes in the physical properties of CO* such as density,
pressure had been reached by adjusting the pressure regulat- which closely relates to its dissolving capacity. At pressures
ing valve, the heating jacket was used to reach the processing below 150 bar the amount of oil extracted was higher with
temperature. When both the desired pressure and temperature liquid carbon dioxide, whereas above 150 bar the amount of
had been reached, the extraction was started by opening valve oil extracted was higher with supercritical carbon dioxide.
V-7. The flow rate of carbon dioxide through the extractor
vessel was regulated by the flow micrometer valve (FMV) . 0.07

The oil dissolved in the supercritical CO* was separated from nm0.06

the carbon dioxide and collected in the separator (S) at ambi- \

~0.05
ent temperature and pressure. The CO* was passed through
two filters (F-l and F-2) to remove the entrained oils and _D 0.04 E,
.a,
then through an in-line volumetric flowmeter (FM) which x
0.03
controlled the quantity and flow rate of COZ used. The flow- c
0 Temperature
meter used was a model FC-70 by EG&G Flow Technology. 15 002

The operating condition ranges studied to optimize the g 0 40 c

,’ 0.01
extraction processes were: pressure, 50-350 bar; temperature, n 60 C
6
10-60 “C; solvent flow rate, 0.5-2.0 1 min-’ (at standard / d , I
0 100 200 300 4
temperature and pressure (SIP) ) . For liquid extractions at Pressure (bar)
10 “C the heating jacket was replaced by a brass mantle
Fig. 2. Effect of pressure on extraction yield of grape seed oil using liquid
through which a cooling liquid was passed. and supercritical carbon dioxide at three different temperatures (operating
The amount of extracted oil was determined gravimetri- conditions: extraction time, 5 h; STP solvent flow rate, 2.0 1mm-‘; grape
tally after separation from the solvent. The extraction yield seed size, 0.75 mm; drying time, 6.5 h)
 A. Molero et al. /The Chemical Engineering Journal 61(19%) 227-231 229

Similar results have been reported by other authors working 0.05 ,

with other seeds [ 13,141.

From Fig. 2 it can be seen that the best extraction yield is 3.04
0
reached at an operating temperature of 40 “C. At this tem-
perature some 90% of the highest oil recovery yield is 7J 0.03
a,
achieved at 200 bar, much greater pressure being needed to ._
x
increase the yield significantly. Consequently, 40 “C and 200 c 0.02
bar are considered to be the efficient operating conditions. ._0

Fig. 3 shows the effect of supercritical CO:! flow rate on go01

the extraction yield at the above temperature but at a consid- > 0 6.3 hr. 0.3%
lz
erably higher pressure. The amount of grape seed oil extracted 0.00 I , I
per unit of carbon dioxide used was closely dependent on the 3
Exiraction timi (hr.)
flow at rates up to 1.5 1min- ‘. The yields at flow rates of 1.5
and 2.0 1 min-’ appeared to be very similar, reaching 96% Fig. 4. Effect of grape seed humidity (by drying time) on oil extraction yield
of their maximum levels within 3 h of operation. Lower (operating conditions: pressure, 350 bar; temperature, 40 “C; STP solvent
flow rate, 2.0 1 mm-‘; grape seed sire, 0.75 mm).
solvent flow rates needed longer operating times to reach
lower maximum extraction yields. From these figures and
taking account of the probable process economics, 1.5 1 0.10 - 43
I3

min- ’ (at STF’) should be considered the minimum efficient

supercritical carbon dioxide flow rate. \” cnO.08 -
V
Fig. 4 shows the effect of grape seed humidity (measured
u
by length of drying time) on the extraction yield. As can be ._
-
x
seen in the figure, the extraction yield is not significantly
affected by the relative grape seed humidity. The longer the
grape seed drying time, the lower is the yield, owing to the
evaporation of constituents during the drying process. -
2.83mm
Fig. 5 shows the effect of grape seed size on the extraction
yield. Milling of the raw materials greatly improved the
extraction efficiency, as can be seen in this figure, and the
smaller the particle size, the greater is the yield. Therefore
the milled grape seed size should be 0.35 mm or smaller for 1 min-‘;
process efficiency. The particle size modification by milling .

of the physical structure of the grape seed is considered to

Table 1
affect the extraction efficiency by supercritical carbon diox- Comparison
ide very significantly. These results are similar to those for
supercritical carbon dioxide extraction from soy bean
[ 13,151 and rice bran [ 161 but different from wheat germ
[ 141, which can be satisfactorily extracted without any prior
milling. The presence of skin with the grape seeds and dif-

[ 141

[ 161

[ 38.4
[

Flow rate

Characterization of extracted oils

‘O” Table 1 gives a comparison of the SFE and conventional

Fig. 3. Effect of supercritical carbon dioxide flow rate on extraction yield of hexane extraction yields from grape seeds together with sim-
grape seed oil (operating conditions: pressure, 350 bar; temperature.. 40 “C; ilar yield comparisons for other types of seeds. The following
grape seed size, 0.75 mm; drying time, 6.5 b) . were the operating conditions for the SFJZprocess: pressure,
230 A. Molero et al. /The Chemical Engineering Journal 61(1996) 227-231

Table 2
Comparison of properties of extracted grape seed oils obtained using hexane and supercritical carbon dioxide

Physicochemical parameter Extraction method Standard for

refined oil [ 171
CO,, 350 bar, 40 “C, 3 h Hexane. Soxhlet, 20 h

Refractive index 1.475 1.474 1.473-1.475 (25°C)

Density (g ml-‘) 0.924 0.928
Viscosity (cP) 69 66
Absorbance at 290 nm ( IgE) 2.730 3.177
Free fatty acid (%) 3.4 33.8 G 0.2 (oleic acid)
Iodine index (g per 100 g) 98 124 125-150
Saponification index 259 289 185-196
Unsaponifiable fraction (%) 0.27 2.89
Peroxide index (meq per kg) 383 101 610

Table 3
Fatty acid compositionof oil extractedusing hexane and supercriticalcarbon dioxide

Oil Fatty acid composition (%)

Palmitic C16:O Palmitoleic C16: 1 Stearic C18:O Oleic C18:l Linoleic C18:2 Linolenic C18:3

Hexane, Soxhlet, 20 h 8.12 0.15 5.60 19.59 66.16 0.37

CO*, 350 bar, 40 “C, 3 h 8.03 0.15 5.07 19.06 67.39 0.30
Refined oil [ 171 5-10 g1.2 3-5 12-26 58-77 <l

350 bar; temperature, 40 “C; solvent flow rate at STP, ( 1) Supercritical carbon dioxide is selective for trigly-
2.0 1min-‘; extraction time, 3 h, milled seed size, 0.75 mm; cerides and does not extract the free fatty acids in the grape
seed humidity, 0.35%. seed.
The best SFE yield for grape seeds was 6.9%, some 92% (2) The grape seeds used as raw material were previously
of the best hexane extraction yield of 7.5%. Similar results washed in hot water. The high temperature involved in this
have been found by other authors for other vegetable oils process leads to an increase in the free fatty acids in the grape
[ 13-161. In all extractions the yield was higher for extraction seeds, because triglycerides break down into fatty acids.
by hexane irrespective of the physical nature of the seeds. Hence the amount of free fatty acids in the washed grape
This is due to the fact that hexane, unlike carbon dioxide, is seeds was higher and that of triglycerides lower than in the
non-selective for triglycerides, extracting free fatty acids, unwashed seeds. For this reason the SFE yield for the washed
phospholipids, pigments and unsaponifiable substances grape seeds is always slightly lower than that for the
together with triglycerides. Hence the amount of matter unwashed seeds and the acidity of the hexane-extracted oil is
higher than that of the oil extracted by SFE. These effects
yielded by using organic solvents will be always higher than
have also been reported by other authors [ 6-101.
that by using supercritical fluids [ 13-161.
(3) The operating temperatures of the two extraction
In Table 2 the properties of the grape seed oil extracted by
processes were significantly different: 40 “C for SFE and 69
SFE using carbon dioxide (at the selected operating condi-
“C for hexane extraction using a Soxhlet apparatus. The
tions) and by conventional extraction using hexane are given.
higher temperature of hexane extraction resulted in consid-
The oils are not significantly different when the main physi-
erable free fatty acid formation from triglycerides during
cochemical parameters are considered, but saponification, extraction, which was not the case with SFE.
peroxide and iodine indices show a high concentration of When free fatty acids were removed from both oils, both
triglycerides in the oil extracted by SFE owing to the higher the iodine and saponification indices showed similar amounts
selectivity of the supercritical solvent of triglycerides. Grape seeds were totally extracted by super-
The main differences between oils are related to the free critical carbon dioxide over 3 h. The low value of the unsa-
fatty acid concentration and the unsaponifiable fraction, ponifiable fraction shows that the supercritical carbon
whose values are much lower for carbon-dioxide-extracted dioxide, being more highly selective, extracts only a minimal
oil than for hexane-extracted oil. Considering both the phys- fraction of unsaponifiable compounds.
ical nature of grape seeds and the characteristics of the two Finally, on the basis of the characteristics and composition
extractive processes, these differences could be due to the of the oil described, the supercritical carbon dioxide extrac-
following reasons. tion process is simpler than the conventional extraction pro-
 A. Molero et al. /The Chemical Engineering Journal 61(1996) 227-231 231

Table 4 The quality of grape seed oil extracted by SFE is similar

Fatty acid composition of oil extracted using supercritical carbon dioxide to that of oil extracted by organic solvent and then refined.
according to extraction time of process
Consequently, the refining process is not needed with WE.
Time Fatty acid composition (96 ) It is considered that, in spite of the high operating costs,
(mm) SFE of grape seed oil could be more economical than con-
Palmitic Palmitoleic Stearic Oleic Linoleic Linolenic ventional liquid extraction, because the last two stages of oil
C16:O Cl61 C18:O C18:l C18:2 C18:3 refining and solvent distillation, which consume most of the
30 10.52 0.22 5.32 27.60 56.34
energy, can be eliminated. An industrial scale plant design
60 9.94 0.19 6.15 26.53 56.84 0.36 and costing comparison are required to confirm this last
120 7.86 0.12 7.54 24.00 60.22 0.34 conclusion.
180 8.03 0.15 5.07 19.06 67.39 0.30

Acknowledgments

The authors with to thank the CAICYT for financing this

cess, because the final refining stage of edible oil processing
work through the R&D project “Aprovechamiento integral
can be omitted [ 14-161.
de 10s subproductos de vinificacidn por extraccidn supercrf-
tica de compuestos de inter&” (ref. ALI94-0473).
3.3. Fatty acid composition of extracted oils

As can be seen in Table 3, there were no significant differ- References

ences between the oils extracted by supercritical carbon diox- [l] M. Borkand J.P. Kiimer,Z.F.L., 42 (1991) 732.
ide and by hexane. However, as can be seen in Table 4, in [2] E. Stahl, K. Quirin and D. Gerard, Dense Gases for Extraction and
which the fatty acid composition of grape seed oil extracted Refining, Springer, New York, 1988, Chap. 2.
by supercritical carbon dioxide at 30,60, 120 and 180 min is 131 RAPS Informationsheet, Kulmbach, 1993.
[4] D. Horrobin and M. Manku, Lipids, 18 (1983) 558.
given, SFE causes a fractionation of the oil by depressuriza-
[5] F.A. Kummerow, J. Food Sci., 40 (1975) 12.
tion, which permits the production of various grades of oil [6] P. Ramel, M.T. Lanteaume, A.M. Le Clerc, J. Ranau and E. Morel,
with different fatty acid contents. Rev. Fr. Corps Gras, I2 (1965) 517.
[7] P. Ramel, A.M. Le Clerc, J. Dumain and D. Fauqumbcrgue, Rev. Fr.
Corps Gras, 12 (1965) 153.
[S] P. Ramel, M.T. Lanteaume, A.M. Le Clerc and J. Rannaud, Rev. Fr.
4. Conclusions COTS Gras, 14 (1967) 505.
[9] M.T. Lanteaume, P. Ramel, A.M. Le Clerc and J. Rannaud, Rev. Fr.
Corps Gras, 13 (1966) 603.
As detailed in the above discussion, the following conclu- [lo] A.M. Gattuso, G. Faxio and V. Cilluffo, Riv. Sot. Ital. Sci. Alim., 12
sions have been reached. (1983) 47.
Supercritical carbon dioxide gives higher extraction yields [ 111 Standard Methods of Analysis, Association of Official and Analytical
than liquid carbon dioxide. The efficient operating conditions Chemists’ Society, Washington, DC, 13th edn., 1980.
proposed for the carbon dioxide WE of grape seed oil are: [ 121 I. BarceM and F. Bercel6, Grasas y Aceites, 36 (1985) 269.
[ 131 E. Stahl, E. Schtitz and K. Mangold, J. Agric. Food Chem., 28 (1980)
pressure, 200 bar; temperature, 40 “C; solvent flow rate, 1.5 1153.
1min- ’ (at SIP) ; milled seed size, 0.35 mm; initial humidity [ 141 M. Tan&u&i, M. Kamihira, T. Tsuji and T. Kobayashi, World Congr.
of seed, 0%-6.5%; process time, 2 h. of Chemical Engineering, Tokyo, 1986, p. 1040.
Under these operating conditions, SFE yields are similar [ 151 J.P. Friedrich and G.R. List, J. Agric. Food Chem.. 30 (1982) 192.
to those of conventional extraction with hexane. Since the oil [ 161 W. Zhao, A. Shisshiiura, K. Fujimoto, K. Arai and S. Saito, Agric.
Biol. Chem., 51 (1987) 1773.
extracted by SFE is solvent free, the distillation process to [ 171 Real Decreto 308/1983, January 25th. Sobre la Reglamentaci6n-
recover solvent, which is necessary in conventional extrac- T&mico-Sanitariade Aceites Vegetates Comestibles (B.O.E. February
tion, is not needed. 21st. 1983).