p•luid Phase Equilibria, 49 (1989) 115—126

115
Elsevier Science Publishers B.V., Amsterdam — Printed in The Netherlands

EXTRACTION OF PERFUI•IES AND FLAVOURS FRON PLANT MATERIALS /ITH

LIQUID CARBON DIOXIDE UNDER LIQUID-VAPOR EQUIL1BRIUN CONDITIONS

S.N. NASK, H. LENTZ

Universitat-GH Siegen, Fach bereich 8, Postfach \ 01240, D-5900 Siegen (F.R.G.)
and
R.C. NAHESHWARI
Indian Institute of Technology, Centre for Rural Development and Appropriate
Technology, Hauz Khas, New Del hi 110016, tndia

Dedicated to the 65 b irthday of Fried rich Kohler

(Received December 22, 1988)

ABSTRACT
Nal k, S.N., Lentz, H. and Nahesh sari, R.C., 1989. Extraction of Perfumes
and Flavours from Plant Naterials with Liquid Sarbon Dioxide under liquid-
vapor equiIibrium conditions.
Dlfferent plant materlals includIng cardamom, clove, cumin, fennel, glnger,
parsley and sandel wood were extracted with liquid carbon dioxide under liquid-
vapor conditions. The novel apparatus used is described in some detail. The
extraction with C0z is compared to the conventional steam diatlllatlon. The
yiel ds of the extraction and of the distillation were determined quantitati vel y
and the extracts were analysed by thin layer and by gas chromatography. The
yiel ds of the COz extractlons were JO s to 360 s larger than the yiel ds of the
steam distillations, • hile the extraction time Is only I/2 to 1/10 Of the time
needed for distil lation, The yield of the extractions as function of time can be
described by a simple equation. The energy consumption of the extraction
process is approximately a factor of three Io¥ter than the energy required for
steam distillation.

INTRODUCTION
The use of supercritlcal fluids (1-6) and also of liquids With vapor
pressures higher than normal pressure as extractants has opened up a new
dimension in separatlon science (Schneider et at., 1980; Brunner and Peter,
1981; Discussion Heeting of Bunsengesellschaft, 1984; NCHugh and K rukonis,
1986; Stahl et al., 1987; International Symposium on Supercritical FI uids,
1988). Attempts to extract natural products by carbon dioxide have been
reported in the literature e.g. for apples, pears, orange J uice, roasted coffee,
anise, caraway, coriander, parsl ey, eucal yptus, py rethrum, papri ka and
ginger (reviews by Rothmay r, 1975 and Caragay, 1981), and have been
especiall y described in patents, e.g. for hop, spices, p yretherin, flavours from
frults, papri ka, vanilla and camomile. However, there are still promlsing
chances for the application of the elevated pressure extraction process to raw
materials as pharmaceutlcal and aromatic plants. Conventional process as
distillation, solvent

0378-3812/89/$03.50 0 1989 Elsevier Science Ptibliahera B.V.

116

extraction, enflurage etc. often require additlonal steps e.g. separating the
extractant or have a hi gher energy consumption (a. g. steam distillation) and
are usuall y inferior to COz w ith respect to selectivity. If the selecti vity is the
main concern, liquid COz extraction is even more favourable than supercritical
extraction. In addition, the lower temperature is an advantage for the separa-
tion of thermall y lab ile materials.
Solubility data for 261 oompounds in C0z for a limited temperature and
pressure range are reported by Francis (1954). Solubilities in a wider
temperature and pressure interval are reported for essentlal oils (Stahl et al.,
1987 and literature quoted there), caffel ne (Lentz et al., 1986 and literature
quoted there) and organic chemical s (Schnei der, 1984).
In the paper presented here, xe have extracted raw materials •Ith llquid
CO2 under Its vapor pressure and for comparlson also by conventional steam
distillation (Guenther, 1948).

APPARAT US
For the COz extraction an apparatus of the Jennin gs-type (1980) has been
used. However, the apparatus has been modlfled in some important features, the
main modification being the construction of vindo•s (Lentz, 1988) to control the
correct function of the apparatus through visual observation of the condenser
and the siphon. In addition the vapor llne of the Soxhlet apparatus has been
removed, since this line is necessary onl y on an open laboratory bench but not
in a sealed autocl ave. By omittin g this line, the space available w ithin the
autoclave can be used more efficiently.
The high pressure part of the apparatus (fig.1) consists of a cylindrical
autoclave (i.d. 63 mm, o.d. 80 mm, hei ght of the cylinder 340 mm, volume
935 cm’) from stainless ateel (German oode 1.4571) with closures at bottom
and top. In the cylinder calls sapphi re ulndous were placed in order to allow the
observation of the condenser, the siphon and the flask. The light of a small
electric lamp wac passed through one of the \vindous not used for the actual
visual observations. T he u indows are from synthetic sapp hire (h 10 mm,
d 12 mm) sealed by an o-rlng. The top closure contains the cal ing finger and
the high pressure tubes connectin g a gauge and a valve. The valve is used for
loading and discharging the COz.
The inner extraction apparatus is built completel y from glass. It consists of
an extraction thimble \¥ith a slphon for the perlod Ical removal of the extractant
and extract and a separate glass veesel servi ng as reservoi r for the evaporat-
ing extractant and for the storage of the extract. This glass vessel must have
a volume larger than the volume of the extractant delivered periodicall y by the
slphon.
 117

Fig. 1: Apparatus for extractlon by liquid carbon dioxide

The COz content in the autoclave was around 180 g. The (free) volume in
the extraction thimble to be filled with the liquid COz Was 56 cm* t10 8
to 43 g. Hence around 24 ¥ of the total com used Was
recycled in our experiments. This of course depends on the part of the volume
of the autoclave filled ulth the plant material. In a technical process this part
will be much larger and hence the percentage of the COC recycled sould also
be greater.
The apparatus has to operate in the two phase region beIo\• the critical
temperature. Hence temperature and pressure have to be below the crltlcal
values of 31.06 °C and 73.825 bar (critical density 0.466 g cm-^) (IUPAC
tnternat. Tables, g83j. The working temperature is most easily detected and
controlled b y the operatin g pressure. The temperature correeponding to the
operating pressure can be read off from the well kno in vapor pressure curve
of COz. The temperature is mainl y influenced by the temperature of the coolant
clrculating in the cooling finger. tn Germany the water available from the
normal water tape can be used as coolant. Small changes in this temperatures
118

will also change the extraction conditions. For using lower temperatures cooling
equipment or a ice-water bath have to be used.
The bottom of the apparatus was kept in a water bath at 40 •C in order to
ensure the evaporation of the extractant.

TABLE 1

Yields and extraction time for dlfferent plant materials extracted

by liquid carbon dioxide and prepared by steam distlllatlon.

Material
liquld carbondloxide steam destillation
yleld time yleld time
(x \d/yf) (h) (9 fl/\f) (h)

Cardamom
(Elettaria Cardamum) 9.4 2.5 9.0 5

ChIIIi
(Capsicum annum Linn.) 4.9 2.5

Clove
(Eu genia Caryophyll us) 20.0 2.0 15,0 6

Cumin
(Cumini um cyminum Linn.) 4.5 2.0 2.5 4

Cumin
(Nigella Sativa Linn.) 4.5 2.0 1,0

Fennel
(Foeniculum dulce NIII.) 5.8 2.0 3.0

Ginger
(Zlnglber officlnale Llnn.) 3.7 2.5 1.6

Nace
(Hyristlca fragrance) 1&5 2.5 13,1

Parsley
(Petrosellnum crisbum) 9.8 2.0 2.7 5

Poppy
(Papaver somnlferum) 2.5

Sandeluood
(SantaJum album) 4.8 2.5 10 24

Turmerlc
(Curcuma Longa Linn.) 3.4 2.0

Vatlver 0.9 8
1.0 2.0
(Vernonia Zlzanloides)
 119

HAT ERfALS
T he plant material s used were purchased at a sp ice mar ket in New Del
hi (India) and for clove, fennel and mace in Siegen (Germany). For most plants
the parts normall y used as spices were extracted. However, for parsley and
poppy the seeds, for veti ver and perui nkle the rat and for sandel wood the
stem were used. The onl y preparatory treatment of the materials was a raw
milli ng yiel ding a powder of the grain size of milled coffee. Hence, the materials
contain thei r natural contents of humidity. The xeight of the samples were in
most casee 16 g With the exception of vetiver and sandel wood where 19.5 g
and
22 g were used respecti vel y.
COz (Buse, Germany ) of 99.97 ¥ purity was transfered as a li qui d from the
storage vessel to the autoclave. T he amount Nas controlled by Rei ghin g the
whole autoclave on a normal scales.
For the chromatog rap hles limolene, cineole, carrone, eu genol, anethol
and geranlol were purchased from FI uka (Bucks, Switzerland ) and used as
standards.

RESU LTS
a) Products
Table 1 shows the yield by weight and the extraction time for different
plant materials extracted w ith liqui d COz and - in part - al so processed by
steam distillation. The apparatus was operated at pressures of 56 to 60 bar
corresponding to the vapor pressure curve of COz to a temperature of 19 to
22 °C. In all cases a remarkable increase of the yield can be observed. In
addition the preparation time is shortened significantl y. The yield is only a
crude measure of the quality of the extract. Hence, the products were anal ysed
by thin layer chromatog raphy (TLC) using silica gel 254 (Merck, Germany) and
standard substances and standard reactions for the identification of the spots
( Lew is et at. 1966; Attaway et al. 1965 a, 1965 b; Stahl 1967, 1986). Table 2
gives qualitative results of the different compounds detected by TLC.
For cardamom, ginger, parsley, sandel wood and vetiver extracts from steam
distillation Were also anal ysed by TLC. However, the results were c‹xnparabIe
and are not separatel y I isted in table 2.
For the extracts and distillates of 7 substances a gaschromatografic anal ysis
was made using a Carlo E rba 6000 GC instrument with a flame ionisation
detector and a column filled with DV 17 (10 x) on cromosorb \¥A\¥. The carrier
gas was nitrogen (75 cm1in-’) and the temperature reachl ng from 80 to
220 °C. As an example the chromatogramms for the COz extract and the
steam distillate from sandel wood is show n in fig. 2.
120

TABLE 2

Composition of extracts from plant materials detected by TLC.

ytggj gt Constituents
Car damom
Cineole, g-Terpinyl acetate, Linalool,
Geraniol, Terpenes
Chilli

Clove
Eugenol, Terpenes
Cumin
Cuminal deh y de,
Cumin
Terpenes Carvone,
Fennel
Terpenes Anethol,
Gin ger
Terpenes
Nace
Cineole, Li nalool, Geraniol, Terpenes
Parsley
My rlstici ne, Terpenes
Turmeric
Carvone, Terpenes
Veti ver
Terpenes, Xanthorhl zol
Sandel wood
Khusinol, Khusi lal, Terpenes

Santalol, Terpenes

0. 059 37. 43.24

3.631

FIG.: 2: Gaschromatog rams of extract and of distillate from sandel lead

 121

TABLE 3

Percentage (b y area) of ma)or comDonents and number of GC-detectable

components (in brackets) from extracts and steam distillates of plant materials.

plant
component extract distillate

Cardamom Ci neole 33.3 (23) 40.4 ( 17

a-Terpinyl - 20.9 12.0
acetate

Clove Eu genol 71.2 (19) 63.0 (17)

Cumin Cuminaldehy de 20.3 (19) 20.5 (21)

Cymol 26.5 23.9

Fennel Anethol 86.6 (15) 82.7 (15)

Ginger Zing iberene 57.1 (29) 44.7 (34)

Curcumene 20.2 18.4

Parsley Carvone 22.2 (19) 27.0 (22)

Sandel uCod a- Santalol 55.9 (16) 57.6 (13)

8-SantaIoI 25.3 24.3

In table 3 the percentage of the major components and the number of

detectabl e oomponents (in brackets) are listed for comparison of the extracts
and distillates. The number of components is a possible measure for the quality
of an extract. The number of components extracted can often be increased if
the process is carried out at lower temperature, e.g. for parsley seeds at 14 •C
the number increases to 30 detected components.
The reported results show that the COz extraction is compared to stem dis-
til lation superior in y iel d (i ncreases bet \veen 10 x and 360 ¥). In addition the
preparation ti me is remark ably shorter (1/2 to 1/10) and the quality of the
extraction products is at least equal to the distillates.

b) process
For the comparison of the time needed for steam distillation or liquld COz
extraction, the duration of the extraction ie used. However, the time interval
for the periodical drainin g of the siphon sould be an even more relevant
measure. This time increases slightly from 10 minutes at the beginnlng of the
experiments to 12 to 13 minutes at the end. Thus in an extraction time of 2
hours, the extractant is circulated app roximatel y 10 times. In independent
122

1&

50 1I¥I 150 fmin.

FIG.3: Yield (¥ w/w) aa function of extractlon time, experlmentaJ point

indicated, lines calculated by equation (1)

measurements the yield {¥ w/q) as a function of the extraction time was

determined. The results are given in flg. 1 For clarity we omlt parsley (similar
to cardamxn), fennel (similar to sandel¥iood) and cufnin (slightly lover than
sandel-wood). A first inspection leads to the statement that the kinetics of
process are mainly determined by the final yield. If this yield is high, then the
solublllty will be high and hence the driving force of the mass trana(er
process will be large. This statement can be quantitatlvely correct only if the
transport fficients for all materials are Identical.
A better Insight into the kinetics of the extractlon proceee Is obtained froc
a plot at the yields divided by the hlghest yield, m given In fig. 4. The points
for different materials do not fall on the same curve, although in host casec
hlgh solubilities cause fast extractlon. Ginger and vetlver antradict thla rule,
because for vetiver w lth the smallest final yield the process is faster than for
glnger u ith a much larger final yiel d.

0 50 190 I/min.
FIG.4: Yield divided by the highest yield as function of extraction time
x Clove, o Cardamom, a Glnger, O Vetlver

The yield as functlon of time is an Important function for attempts at

process optimisation. Hence, even a formal description of thls functlon would be
valuable. Attempts of a dascrl ption by a slngle exponential functlon (e.g.:y=
1-a exp-kt) fall. From inspection of fi gure 3 a function of the type of a
Langmui r gas adsorption isothem seems to be proclslng, since at long times
the yield does become constant, hile at short times the Increase of ylelds
could be assumed to be llnear. Also the success to descrl be 6oIubillties in
compressed gases as functlon of the density by a similar equation ( Lentz, 987)
enoourages such an attempt.
(1)
'. b + I

y is the yield In wet ght fraction or percent, I Is the extractlon time, jé, and b
are constants. yo la the yleld after Infinite extractlon time (the flnal level In
fig. 3) and yr/b Is the initial slope of the y leld as function of tlme (fig. 3).
Obvlousl y yr u III differ for the varlous substances. However, If the inltlal slope
Is proportional to y than the oonstant b should show only small varlations If
the different substances are compared.
y and b are llsted In table 4. These constants are detemlned from the
experimental points shown in fl g. 3 by Ilnearlslng equation 1
124

Hence a plot of the inverse yield against the inverse extractlon time should
gl ve straight lines with the intersect yr' and a slope of b/y . The substances
listed in table 4 indeed show this linear relationship in the limit of the
precision of the experlmental points. However, for fennel and parsley the
suggested fit is very poor. JI could deflnitel y be Improved by fittlng the short
and the Ion g time data separatel y to equations giving two linear regions. The
failure of equation 1 for fennel and parsley is perhaps due to the fact that
two groups of substances are extracted, for instance fi rst etherlc or essential
oils and than fatty or crystal line I parsley) substances. In such cases of course
no single function can represent the yield as function of time. The validity of
equation 1 is demonstrated in fl g. 3 there the curves have been calculated
with the constants of table 4. Equation 1 will permit the estimation of the final
yield even if the experimental points are still far from saturation. A comparison
of y and the hlghest yield of the experlments Indlcates the dlfferent
ap p roaches to saturation.
ft can be seen from fig. 3 that the extraction times gl ven in table 2 can be
reduced in commercial applications for rnoet of the substances. Thus for clove,
parsley and vetiver an extractlon time of 1 h corresponding to 5 circulations of
the C02 seems to be sufficient.
For eugenol in clove, for clneole and Ilmonene in cardamom and for santalol
in sandel ¥/Cc/d vie have registered the peak area6 from gaschromatographic
anal ysis. For eugenol we find a small increase between 1 and 2 hours even
though the total yield was constant. The contents of clneole and llmonene are
constant after 15 min extraction time of cardamom. Due to the slight increase of
the yield the relatlve contents of these components are lowered with lncreaslng

TABLE 4

The oonstants of equatlon (1) for dlfferent plant materials.

Hateria 1 yp yp b
Cardarnorn O.OB0 0.095 1.04 0.1019
CIOV6 0.2(A 0.20 0.16 0.0326
Cumln 0.M4Z 0.OF &55 0.464
GInger 0.0417 0.037 6.20 0.2585
Sandel wood 0.0549 0.049 6.28 0.3437
Vetlver 0.0119 0.010 31.5 0.3749
 t25

extraction time. The final santalol content Is almost obtained within 60 min, In
splte of a further increase of the yield. Hence for the exploitatlon of special
components, the extraction time could be even shorter than estimated from the
total yiel d.
T he knowledge of the amount of Coz circulated and the perlodicity of the
circulation enables us to estimate rou ghly the total requirement of COz and
hence of the energy consumption. The 43 g COC cycled with a heat of evap-
oration of 149 I g-’ correspond to an energy consumption of the apparatus of
6.41 kJ per cycle. If we assume that •e will use 8 cycles to exploit a plant
material and the mass of this material is ° 17 g then the energy oonsumption
for 17 g w ill be 6.41 kJ/cycle 8 cycles= 51.28 kJ and thus for 1 kg the
energy requl red will be 3016 kJ. If less circulations of the COz are necessary
the energy consumption for the treatment of 1 kg plant material will be
smaller, for Instance for 4 cycles 1508 kJ and so on. This has to be compared
u ith the energy requirements of the steam distillation. We have determined the
quantity of steam requl red for 6 substances by weighing the amount of water
used in the process. This is for cardamom 3.9 kg per kg plant material, for
fennel 5.0, for ginger 4.3, for parsley 3.0, for sandel wood 14.4 and for vetiver
3.8. If we do not take Into account the extremely hlgh value for sandel wood,
then an average value of 4 kg water per 1 kg plant material has to be taken.
With a heat of vaporisation of 22S0 kJ per kg water the energy requirement to
evaporate 4 kg xlll be 9000 kJ per kg plant material, a factor of 3 higher than
required for a COz extraction with 8 cycles of the extractant.

ACKN0¥/LEDONENT
We would like to thank Dr. U. Nihm (University of Slegen) for his valuable
help In the analytical ¥/ork.
The financial support by the Volkswagen Foundatlon and also by the Fonds
der Chemlschen lndustrle Is gratefully acknowledged.

REFERENCES
Attaway, T.A., Barabas, T.A. and Holford, W.R, 1965.Analysis of terpene
hydrocarbons by thln layer chromatography. Anaiyt.Chem., 37: 1289-1290.
Attauay, T.A., Holford, \¥.R. and Edwards, G.J., 1905. Deteml nation of esters
by thin layer chromatography. Analyt. Chem., 37: 74-76
Brunner, G. and Peter, S., 1981. Zum Stand der Extraktlon mit komprlnlerten
Gaeen. Chem. Ing. Tech., 53: 529-542
Caragay, A.B., 1981. Supercrltical fl uids for extraction of flavors and fragrances
from natural products. Perfumer and Flavourlst, 6: 43-53.
Dlscusslon N.eeting, Supercrltical Fluid Solvents, 1984, Kdnl gsteln, Germany. Ber.
Bunsenges. Phys. Cheai., 88: 9, 783-926
t26

F rancis, A.W., 1954. Ternary systems of liquld carbon dioxide. J. Physical

Chemistry, 58: 1099-1114
Guenther, E., 1948. The Essential Oils, Van Nostrand
lnternational Symposium on Supercrltical fluids, Nlce 1988
IUPAC International Tables of the Fluid State, y, 1ga3. Carbon Dioxide (Ed.
S. Angus et.al.), Pergamon Prees
Jennings, ¥/.G., Wohleb, R.H. and \¥ohIers, N.¥/., 1980. Hi gh preseure soxhlet
extractor. US Patent 4.265.860 (1981, filed 1980)
Lentz, H., 1989. Solubilities of solids in compressed gases a new attempt to
discribe and to calculate. 11.AIRAPT Intern. Conf. on High Pressure
science and Technology, Kiev, Naukova Dumka, 1: 52-56
Lentz, H., 1988. Vorrlchtung zur Extraktion durch Fliissi gkelten unter hohen
Dampfd rucken. German utility patent G 8810 807.4
Lentz, H., Gehri g, N. and Schulmey er, J., 1986. Dy namic solubiIity measurements
of caffei ne in carbon dioxide and in carbon dioxide saturated with water.
Physics 139 & 140 B: 70-72
Plc Hug h, H.A. and Kru konis, V.J., 1986. Supercritical FI uld Extraction,
Butterworth
Rothmay r, ¥/.W., 1975. Fcod process engineering. Proc. R. Soc. London B191,
71-86
Schneider, G.M., 1984. Phase equilibria in fl uid systems. Ber. Bunsenges. Phys.
Chem., 83: 841-B48
Sch neider, G.N., Stahl, E. and \'/ilke, G. (E ditors), 1980. Extraction w ith super-
critical gasas, Verlag Chemle
Stahl, E., 1967. Dunnschicht-Chromatographie. Springer- Verlag
Stahl, E., 1986. E xtraktion von Naturstoffen. Gustav Fischer Verlag Stuttgart
Stahl, E., Qul rl n, K.\¥. and Gerard, D., 1987. verdichtete Gase zur E xtraktion
und Raffinatlon, Springer-Verlag