You are on page 1of 11

J.

of Supercritical Fluids 25 (2003) 45 /55


www.elsevier.com/locate/supflu

Supercritical fluid extraction of ethanol from aqueous solutions


M. Budich 1, G. Brunner 
Technical University of Hamburg-Harburg, Thermische Verfahrenstechnik, Eissendorferstrasse 38, 21071 Hamburg, Germany
Received 4 September 2001; received in revised form 22 April 2002; accepted 30 April 2002

Abstract
The recovery of ethanol from aqueous solutions was studied by using supercritical carbon dioxide. At 333.2 K and
10.0 MPa, vapor /liquid equilibrium data of the mixture CO2/ethanol/water were determined. No azeotrope was
observed. Theoretical calculation of equilibrium stages was performed and compared with countercurrent column
experiments. Separation of extract and solvent was optimized by multistage solvent distillation. The height of one
theoretical stage was found to depend on the ethanol content of the liquid phase. Moreover, flooding point
measurements were carried out with ethanol/water mixtures of different composition.
# 2002 Elsevier Science B.V. All rights reserved.
Keywords: Azeotropic mixtures; Countercurrent extraction; Flooding; Phase equilibria; Stage calculations; HETS

1. Introduction
In the field of supercritical fluid extraction
(SFE), various researchers proposed the use of
supercritical carbon dioxide (CO2) for ethanol
recovery and the separation of other volatile
organic components from aqueous solutions.
Among other applications, literature covers the
enrichment of flavor fractions from fruit juices [1]
or wine [2], ethanol production [3], separation of
impurities from fermentation processes [4,5] and
dealcoholization of beverages [6]. Only a few
papers cover process evaluation from an engineer-

 Corresponding author. Tel.: /49-40-42878-3040; fax: /


49-40-42878-4072
E-mail address: brunner@tu-harburg.de (G. Brunner).
1
Present address: BASF AG, D-67056, Ludwigshafen,
Germany.

ing point of view. Therefore, it was the objective of


this study to investigate the mixture CO2/
ethanol/water, to compare literature data to the
present measurements, and to perform thermodynamic stage calculations to demonstrate performance and limits of this technology.
Up until the present, vapor /liquid equilibrium
(VLE) data of CO2/ethanol/water and its binary mixtures were published over a wide range of
temperature and pressure. From an economic
point of view, gas extraction of ethanol/water
mixtures should yield ethanol of high purity to
compete with conventional processes. However, in
order to obtain a high solubility of ethanol in the
vapor phase, many studies were carried out at
conditions of complete miscibility of ethanol and
CO2 [7,8]. At these conditions, the phase behavior
of the ternary mixture CO2/ethanol/water is of
type I [9], as illustrated in the left triangular

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 0 9 1 - 8

46

M. Budich, G. Brunner / J. of Supercritical Fluids 25 (2003) 45 /55

Fig. 1. Schematic phase behavior of the ternary mixture CO2/ethanol/water.

diagram of Fig. 1. With reference to the phase


diagram it is easy to understand that anhydrous
ethanol cannot be produced at a type I phase
behavior. Nevertheless, some researchers concluded that it is impossible to break the azeotrope
by using supercritical CO2.
Lim et al. [10] assumed that ethanol can be
concentrated above azeotropic composition whenever the pressure in the ternary mixture CO2/
ethanol/water is below the critical pressure of
the binary mixture CO2/ethanol, equivalent to a
phase behavior of type II (right triangular diagram
of Fig. 1). One of the first studies reporting the
possibility to produce anhydrous ethanol by
means of CO2 without adding any entrainer was
by Nagahama et al. [11]. Experiments were carried
out at conditions of type II phase behavior.
Further VLE measurements of the mixture
CO2/ethanol/water were performed at Kobe
Steel Ltd. in Japan at 10.1 MPa and 313, 323,
and 333 K [12].
A solvent-free representation of different VLE
data is shown in Fig. 2 in comparison with data at
ambient pressure [13]. According to data of Furuta
et al. [12] and Horizoe et al. [14], it was not
possible to enrich ethanol in the vapor phase
above azeotropic composition at 10.1 MPa and
either 313 or 323 K, conditions of type I phase
behavior. However, no azeotrope was observed at
10.1 MPa and both 333 and 383 K, where the
phase behavior of the ternary mixture is of type II.
At 10.1 MPa and 333 K, solubility of pure ethanol
in CO2 is around 5 wt.% [10]. At 333 K, higher
pressures are not recommended due to approach-

ing the point of complete miscibility of CO2/


ethanol around 10.7 MPa.
Due to scattering in data at 333 K and 10.0 or
10.1 MPa, respectively, and diverging opinions
about the existence of an azeotrope, VLE measurements of the ternary mixture CO2/ethanol/
water were carried out prior to stage calculations
and countercurrent column experiments.

2. Experimental set-up and procedure


2.1. Equilibrium apparatus
Fig. 3 shows the equilibrium apparatus used in
this study that comprised two autoclaves connected in series. The apparatus was designed by
Bunz [15] with the objective to enable large vapor
phase sampling at conditions of very low vapor
phase load.

Fig. 2. McCabe /Thiele diagram of CO2/ethanol/water,


solvent-free representation.

M. Budich, G. Brunner / J. of Supercritical Fluids 25 (2003) 45 /55

47

Fig. 3. Phase equilibrium apparatus.

The apparatus was thermostated inside a stirred


water bath, covered with insulation balls. Temperature inside the water bath held constant to 9/
0.2 K. Pressure was adjusted to 9/0.05 MPa.
Liquid feed was charged to an evacuated autoclave
of 500 cm3 that was separated from a second
autoclave (300 cm3) by means of high pressure
valves. Carbon dioxide was charged into the
apparatus by a compressor, and phase equilibrium
was achieved by circulating the vapor phase
through both autoclaves for at least 10 h by using
an air-operated piston pump (designed and built at
Hoffmann-LaRoche, Kaiseraugst, Switzerland).
Afterwards, the larger autoclave contained a vapor
and a liquid phase in equilibrium, whereas the
smaller one contained only the vapor phase.
After closing all valves, three liquid samples of
approximately 5 g were taken in series from the
large autoclave via valve V9. Liquid was separated
from gaseous CO2 by means of three cooling traps
connected in series. Noncondensable gas was
transferred to a burette system by a vacuum pump.
The entire content of the vapor phase autoclave
(ca. 100 g) was withdrawn via valves V4, V6, and

V7 and solidified in an evacuated flask cooled with


liquid nitrogen. The total mass of vapor phase
sample was determined by weighing the sampling
flask. Sublimation of CO2 was carried out in a
freezer at approximately 253 K to obtain a
solvent-free vapor-phase condensate at the bottom
of the flask. Gas was forced to flow through
methanol filled traps to determine the loss of small
quantities of ethanol and water from the flask
during sublimation of CO2.
Experimental results for the liquid phase were
reproducible by maximum 9/5% with respect to
the CO2 content, and for the vapor phase by
maximum 9/10% with respect to the condensed
liquid. At least two repeated measurements were
carried out to reproduce vapor phase composition.
Details on the experimental procedure can be
found elsewhere [16].
2.2. Countercurrent extraction apparatus
Countercurrent multistage extraction was carried out in an extraction column of 6 m total
height (25 mm ID, equipped with 4 m of Sulzer EX

48

M. Budich, G. Brunner / J. of Supercritical Fluids 25 (2003) 45 /55

packing). Packing elements (specific surface area


1710 m2/m3 [17], voidage 0.86% [17], each element
70 mm high, made of stainless steel 1.4404) were
fitted at a 908 turned position in relation to the
preceding element.
The experimental set-up is shown in Fig. 4.
Extraction conditions were set to 333.2 K and 10
MPa. Feed was charged to the middle section of
the column by a piston pump. Carbon dioxide
entered the column at the bottom. Both feed and
CO2 were preheated to extraction conditions
before entering the column. Loaded solvent was
withdrawn from the top. Solvent and extract were
separated by pressure reduction down to 5 MPa.
Moreover, solvent distillation was established as
proposed by DeFilippi and Vivian [18] and reported by Ikawa et al. [3].
The distillation column used for extract separation had a height of 1.5 m and 35 mm ID and was
filled with stainless steel mesh packings (diameter 8
mm). The loaded solvent entered the separator at
the bottom. The lower section of the column (ca.
0.2 m) was heated to 303.2 K. By refluxing liquid
CO2 to the top of the separator with a reflux ratio
of 0.5, the residual ethanol content of the regen-

erated solvent was greatly reduced and ethanol


recovery was enhanced.
Extract and raffinate were withdrawn continuously from the apparatus. Gas coolers were used
to reduce the loss of condensable product with
gaseous CO2 at ambient conditions. Extract was
partly refluxed to the top of the extraction column.
Fresh CO2 was added to replace lost solvent.
Samples were repeatedly taken after some hours
of constant process conditions. Usually, it took 2 /
3 h to obtain steady state conditions with respect
to sample composition after establishing constant
mass flow.

Fig. 4. Countercurrent extraction apparatus.

Fig. 5. Flooding point apparatus.

2.3. Flooding point apparatus


By means of a flooding point apparatus, maximum liquid and vapor phase cross-section capacity was determined while both phases were in
equilibrium. The apparatus was designed by
Meyer [17] and is illustrated in Fig. 5. It was
equipped with 2 m of Sulzer EX packing with 25
mm ID. Flooding points were determined visually
via sapphire windows and by measuring pressure
drop. Both phases were charged countercurrently

M. Budich, G. Brunner / J. of Supercritical Fluids 25 (2003) 45 /55

49

to the column by gear pumps and collected in a


receiver for the pumps. The entire apparatus was
built inside a hot-air cabinet.

3. Materials and analytical procedure


The carbon dioxide used in this study had purity
higher than 99.95 wt.% (Hydrogas, Bad Honning,
Germany). Ethanol and methanol had a purity
higher than 99.8 wt.% (grade Lichrosolv, Merck,
Darmstadt, Germany). Deionized water was supplied by the universitys utility station.
The water content of a sample was determined
by Karl /Fischer titration. Additionally, samples
with an ethanol content below 10 wt.% and
samples from the methanol filled traps were
analyzed by gas chromatography to minimize
analytical error. The capillary gas chromatograph
berlingen, Germany)
(type 8320, Perkin/Elmer, U
with flame-ionization detector was equipped with
a polar GC column (Stabilwax, 60 m, 0.25 mm ID,
0.25 mm df by Restek, Bad Soden, Germany) and a
guard column (Hydroguard FS, 5 m, 0.25 mm ID;
Restek). Helium was used as carrier gas, and the
split ratio was set to 1:50 for 0.5 ml of injected
sample. All analyses were performed in triplicate.

4. VLE measurements of CO2/ethanol/water


Fig. 6 shows VLE data of this study for the
ternary system CO2/ethanol/water at 333.2 K
and 10 MPa compared with literature data.
Experimental results are shown in Table 1. The
Janecke diagram [19] was used to illustrate the
influence of the solvent-free phase composition on
mutual solubility of liquid and solvent.
Several papers reported difficulties in reproducing VLE data of this ternary mixture accurately
by equations of state [10]. Therefore, empirical
relationships were derived to represent the vaporphase (Eq. (1)) and liquid-phase line (Eq. (2)) of
the Janecke diagram to ensure reliable stage
calculations.

Fig. 6. VLE data of CO2/ethanol/water.

yCO2
100  yCO2
1:5
2
 4261:29Yethanol
0:0884Yethanol

(1)

xCO2
100  xCO2


1
2:5
23:6  0:367Xethanol  0:000137Xethanol

(2)

with xCO2, yCO2 are the weight percentage of CO2


in liquid and vapor phase and Xethanol, Yethanol are
the weight percentage of ethanol in CO2-free
liquid- or vapor-phase sample.
Liquid phase data are in good agreement with
literature data, except from measurements at high
ethanol concentrations. Vapor-phase data by Suzuki et al. [20] show a higher solubility of extract in
CO2 at 333.7 K and 10.1 MPa, whereas data at
333.6 K are in good agreement to the present
measurements. Furuta et al. [21] reported a lower
solubility of extract in CO2. Scattering of literature
data is due to different experimental techniques
and probably due to difficulties in separating
ethanol and gaseous CO2 during sampling. However, good agreement was observed with literature
data of the binary mixtures CO2/ethanol [22] and
CO2/water [23].
Fig. 7 illustrates the separation factor of ethanol
to water as a function of the concentration of
ethanol in the solvent-free liquid phase. The
separation factor decreased from around 30 at
infinite dilution of ethanol in water to approxi-

M. Budich, G. Brunner / J. of Supercritical Fluids 25 (2003) 45 /55

50

Table 1
VLE data of CO2/ethanol/water mixtures at 333.15 K and 10.0 MPa
Liquid phase (wt.%)
xCO

3.821
4.851
4.091
3.583
3.882
4.057
5.914
11.345
16.378
14.921
42.573
43.328
58.868
60.048
60.574

Vapor phase density (kg/m3)

Vapor phase (wt.%)

xEtOH

xH O

yCO

0.000
0.276
0.845
0.852
1.604
7.310
24.403
41.522
49.165
52.244
50.858
50.362
39.925
39.843
39.334

96.179
94.873
95.064
95.565
94.514
88.634
69.684
47.133
34.457
32.834
6.569
6.311
1.207
0.109
0.091

99.757
99.768
99.713
99.723
99.630
99.294
98.424
98.046
97.800
97.906
96.853
96.833
96.083
95.575
95.094

yEtOH

yH O

0.000
0.018
0.056
0.053
0.110
0.451
1.314
1.695
1.940
1.826
2.957
2.976
3.847
4.415
4.896

0.243
0.214
0.231
0.224
0.260
0.256
0.262
0.260
0.260
0.269
0.190
0.190
0.070
0.009
0.010

293.5
294.1
292.3
293.6
297.3
299.7
312.2
313.9
319.4
317.7
332.4
333.2
343.8
350.8
357.7

Data by Furuta et al. [21] are larger, whereas


those by Lim et al. [10] are smaller compared with
results of this study. Good agreement was found to
results of Furuta et al. [12].
By sampling the entire volume of the vaporphase autoclave, vapor-phase density was easily
determined during VLE measurements. Vaporphase density is required for hydrodynamic evaluation of a countercurrent flow process. Results
are shown in Fig. 8 and Table 1. Eq. (4) represents
a simple relationship between load (in g/kg) and
vapor-phase density at 333.2 K and 10.0 MPa.
Fig. 7. Separation factor of ethanol to water.

mately 1.25 at infinite dilution of water in ethanol.


No azeotrope was formed at the conditions
investigated. Separation factors are larger compared with data at atmospheric conditions. The
line in Fig. 7 was calculated by Eq. (3).
aethanol=H2 O 

yethanol =xethanol
ywater =xwater

0:7328:1 exp(0:0265Xethanol ) (3)


with aethanol/H2O is the separation factor of ethanol
to water.

Fig. 8. Vapor-phase density of the mixture CO2/ethanol/


water.

M. Budich, G. Brunner / J. of Supercritical Fluids 25 (2003) 45 /55

rV rCO2 1:3L

(4)

with rCO2 (333.2 K, 10 MPa)/290.2 kg/m3; L is


the load in grams of extract per kilogram CO2.
Equilibrium data at conditions of low ethanol
solubility in CO2 are needed to evaluate feasible
conditions for solvent regeneration. VLE data of
CO2/ethanol at lower pressures are available in
literature [24 /26] and are shown in Fig. 9. The
lines illustrate the shapes of the phase envelopes. It
becomes obvious that the load of gaseous CO2 at
lower pressures cannot be neglected. Ethanol
solubility below 0.2 wt.% can only be achieved at
pressures close to the vapor pressure of liquid
CO2. Unfortunately, large amounts of CO2 will
dissolve in the ethanol-rich phase at the same time.
Therefore, separation of ethanol and CO2 was
carried out by solvent distillation.

5. Calculation of theoretical stages


The number of theoretical stages for a predefined separation task were calculated by the
Ponchon /Savarit method [27,28]. Details on the
method are reported elsewhere [9]. VLE data for
the ternary mixture CO2/ethanol/water were
represented by the above-mentioned equations.
The residual load of CO2 after regeneration was
not taken into account by the stage calculation
model.

Fig. 9. VLE data of CO2/ethanol at conditions of solvent


regeneration.

51

Fig. 10 illustrates calculated results that are


based on a feed mixture of 10 wt.% ethanol that is
separated into an ethanol-rich extract (99.0 wt.%
ethanol) and a water-rich raffinate (0.1 wt.%
ethanol). The number of theoretical stages was
calculated as a function of the extract reflux ratio.
Solvent-to-feed ratio increases linearly with increasing reflux ratio. The minimum reflux ratio for
this specific separation task was calculated to be
6.2, equal to a minimum solvent-to-feed ratio of
around 16, while the minimum number of stages is
10.
The minimum number of stages required to
achieve a certain extract composition increases
with increasing ethanol purity as illustrated in Fig.
11. The required number of theoretical stages
decreased rapidly up to a solvent-to-feed ratio
around 20, whereas a further increase of the
solvent-to-feed ratio above 30 did not reduce the
number of theoretical stages significantly. Therefore, in this example working with a feed mixture
of 10 wt.% ethanol, ethanol separation should be
carried out at solvent-to-feed ratios between 20
and 30. A detailed economic evaluation has to take
equipment costs and energy costs for solvent
recycling into account.
Solvent-to-feed ratios are relatively small compared with other countercurrent gas extraction
processes [16]. This is due to large separation
factors and a relatively high solubility of pure
ethanol in CO2 of around 5 wt.% at the conditions
investigated. Instead of the solvent-to-feed ratio,
the ratio of solvent flow rate to extract flow rate

Fig. 10. Calculation of the theoretical number of stages.

52

M. Budich, G. Brunner / J. of Supercritical Fluids 25 (2003) 45 /55

to the solvent-to-extract ratio when using a feed


with less than 20 wt.% ethanol. Solvent-to-extract
ratios remain almost constant at a larger ethanol
content of the feed and decrease when the feed
composition approaches the desired extract composition.

6. Countercurrent and flooding point experiments


6.1. Ethanol recovery

Fig. 11. Influence of extract quality on process requirements.

should be used as an indicator for operating costs.


Since the raffinate is a waste product, the extract is
the only product that incorporates all process
costs. For the above mentioned separation task,
the solvent-to-extract ratio becomes 300 kg/kg at a
solvent-to-feed ratio of just 30 kg/kg.
Fig. 12 demonstrates the influence of feed
composition and the number of theoretical stages
on the solvent-to-extract ratio. When the desired
ethanol contents of raffinate and extract are set
constant, extract flow increases with increasing
ethanol content of the feed. Thus, the changing
slope of operating lines induces a change in the
required reflux ratio for a given number of
theoretical stages. The calculation reveals that
the production of pure ethanol by means of
supercritical CO2 is most expensive with respect

Fig. 12. Influence of feed composition on the solvent-to-extract


ratio.

By using feed mixtures of approximately 10, 40,


and 94 wt.% ethanol, countercurrent column
extraction was carried out to compare experimental with calculated results. During first experiments, conditions in the separator were set to
303.2 K and 5 MPa. For different solvent-to-feed
ratios and reflux ratios, the amount of residual
ethanol in the raffinate remained always above 4
wt.%. After establishing liquid solvent reflux to the
top of the separation column, residual ethanol
content of the raffinate phase dropped below 2
wt.%. However, product composition was limited
by the limited number of stages achieved in both
the separation and extraction column.
Zobel [30] proposed to use either an adsorbing
material like activated carbon or water to absorb
residual ethanol after pressure reduction. Activated carbon was also used by Perrut [6] to remove
volatile components from distilled beverages.
Other suggestions to improve the separation of
volatile components include the use of a trapping
substance such as glycerol [31]. This procedure is
limited to very few applications when the trapping
substance is of further use for the product itself.
Distillation of CO2 should be preferred to absorption or adsorption to avoid further regeneration
steps or additional waste products.
Some researchers did not recirculate the solvent
due to the high residual load after separation of
extract and solvent by means of pressure reduction
[32]. Countercurrent extraction without solvent
regeneration is very expensive and will never be
established in production scale.
Experiments by Ikawa et al. [3] were carried out
with a feed mixture of 92.83 wt.% ethanol.
Solvent-to-feed ratio was set to 26 kg/kg. Extract

M. Budich, G. Brunner / J. of Supercritical Fluids 25 (2003) 45 /55

purity above 99.6 wt.% was achieved at a reflux


ratio of 12, thus obtaining a raffinate phase with
92.15 wt.% ethanol. This is very close to the
composition of the feed mixture itself. Although
the study of Ikawa et al. [3] demonstrated the
potential of supercritical CO2 to produce pure
ethanol, the limit of ethanol recovery and the
benefits of solvent distillation with respect to
raffinate purity were not fully exploited.
6.2. Evaluation of HETS values
When a feed with 94 wt.% ethanol was used,
extract with 99.5 wt.% ethanol was produced at a
reflux ratio of 4 and a solvent-to-feed ratio of 60
(using 9 kg CO2 per h and 150 g feed per h).
According to calculations, 12 equilibrium stages
were achieved, equal to a height equivalent to one
theoretical stage (HETS) of 0.33 m. Liquid solvent
reflux was not required during this experiment
because the raffinate was still very rich in ethanol
(87 wt.%). During experiments with feed mixtures
of low ethanol content, HETS values were found
to be in the range of 1 m.
HETS values reported by Ikawa et al. [3] were
calculated by an empirical model and decreased
from 0.75 to 0.48 m with increasing reflux ratio.
This was probably due to an improved mass
transfer at a higher cross-section capacity. However, it is not likely that the cross-section capacity
alone accounts for the change of HETS.
Similar HETS values were found by other
research groups. Bernad et al. [33] worked with a
1.4 m column of 54 mm ID equipped with Sulzer
BX packing. Experiments were carried out at 313
K and 10 MPa with 30 wt.% of ethanol in aqueous
solution, used as a continuous phase as well as a
dispersed phase. Feed flow was varied at constant
solvent flow rates, and HETS decreased from 1.5
m at a solvent-to-feed ratio of 2 /0.5 m at a
solvent-to-feed ratio of 10. Extract composition
did not exceed 90 wt.% ethanol because the phase
behavior was of type I.
Lim et al. [34] also worked at 313 K and 10 MPa
with a feed mixture of 8.5 wt.% ethanol. HETS
values were around 0.45 m when the liquid phase
was dispersed and around 0.3 m when CO2 was
dispersed in the liquid phase. In contrast to Bernad

53

et al. [33], only a very small influence of the


solvent-to-feed ratio on HETS was observed at a
constant liquid flow rate.
The opposite effect was reported by Dondoni et
al. [35]. This study was carried out at 313 K and 10
MPa with a column of 1.7 m effective height and
28 mm ID packed with Raschig rings of 4 mm
diameter. A feed mixture with 7 wt.% ethanol was
charged to the top of the column as a dispersed
phase. With increasing the solvent-to-feed ratio
from 10 to 25, HETS increased from 0.1 to 0.5 m.
Concluding, HETS is influenced by the type of
packing, cross-section capacity, the method to
evaluate the number of stages, and transport
properties of both phases, e.g. viscosity and
interfacial tension, that influence mass transfer
and backmixing.
6.3. Flooding point measurements
When running a countercurrent column at its
maximum hydrodynamic capacity, flooding will be
observed. Several studies on flooding points of
SFE processes involving non-aqueous systems are
available in literature [17]. Interesting phenomena
can be observed for the CO2/ethanol/water
system using a flooding point apparatus as mentioned above. Results are shown in Fig. 13.
Maximum cross-section capacity was found to
be a function of the ethanol content of the solventfree liquid phase. Changes in flooding behavior at
high ethanol concentrations between 100 and 70
wt.% are mainly due to the influence of phase

Fig. 13. Flooding point data for CO2/ethanol/water.

54

M. Budich, G. Brunner / J. of Supercritical Fluids 25 (2003) 45 /55

composition on the density of the phases. With a


further decrease in ethanol content of the liquid
phase, the influence of viscosity and surface
tension of the liquid phase starts to become
significant. In fact, flooding occurred almost
instantly when pure water circulated in the column
and a small vapor phase flow entered the bottom
of the column. This happened during experiments
carried out with a Sulzer EX packing.
A few additional flooding experiments were
carried out with a Sulzer CY packing (35 mm
diameter) that is characterized by a larger distance
of the wire mesh layers. In contrast to measurements with Sulzer EX packing, no flooding
occurred at lower cross-section capacities when
investigating mixtures of low ethanol content.
Therefore, the observed change of flooding behavior was found to be a typical scale-down problem.
Furthermore, large HETS values for aqueous
solutions may also be due to an unsuitable type
of packing that causes backmixing. Further measurements should be carried out using a Sulzer CY
packing.
No flooding was observed at the extraction
tower besides few experiments where the raffinate
level went beyond the CO2 inlet. Whenever flooding occurred, the experiment could not be restarted
by simply removing some raffinate, but the entire
hold-up of the packing had to be withdrawn.

7. Conclusions
A reliable technique was developed to determine
VLE data of volatile components and supercritical
CO2. Furthermore, the need of an improved
solvent regeneration was pointed out for continuous SFE processes. Solvent distillation should be
established for SFE processes that require raffinate almost free of volatile components. Stage
calculation helps to understand possibilities and
limits of a process.
HETS values were found to be a function of the
percentage of water in the liquid phase. Flooding
point measurements of aqueous mixtures need to
be carried out very carefully due to scale-down
problems observed when using close meshed
packing material.

The application of supercritical countercurrent


extraction to aqueous solutions is limited whenever foaming is observed. Chemical reaction of
organic components with CO2 should also be
taken into account. Besides the production of
pure ethanol, the removal of organic fractions of
high market value such as flavor components from
fruit juices or of pharmaceutic agents from aqueous plant extracts are promising applications of
SFE processes.

Acknowledgements
The support of this work by Deutsche Forschungsgemeinschaft (DFG) under grant Br 846/
15-1 is gratefully acknowledged.

References
[1] J.M. Randall, W.G. Schulz, A.I. Morgan, Extraction of
fruit juices and concentrated essences with liquid carbon
dioxide, Confructa 16 (1971) 10.
[2] D.R.P. Jolly, Process of enhancing the flavour of wines,
Australian Patent No. 489834 1975.
[3] N. Ikawa, Y. Nagase, T. Tada, S. Furuta, R. Fukuzato,
Separation process of ethanol from aqueous solutions
using supercritical carbon dioxide, Fluid Phase Equilib.
83 (1993) 167.
[4] S. Hirohama, T. Takatsuka, S. Miyamoto, T. Muto, Phase
equilibria for the carbon dioxide /ethanol /water system
with trace amounts of organic components, J. Chem. Eng.
Japan 26 (1993) 247.
[5] R. Fukuzato, N. Ikawa, Y. Nagase, Development of new
processes for purification and concentration of ethanol
solution using supercritical carbon dioxide, in: D.C.
Shallcross, R. Painmin, L.M. Prvcic (Eds.), Value Adding
Through Solvent Extraction, vol. 2, The University of
Melbourne, Australia, 1996, p. 1011.
[6] M. Perrut, Aromas from fermented and distilled beverages
by liquid /fluid fractionation, in: Proceedings of the Forth
International Symposium on Supercritical Fluids, vol. C,
Sendai, Japan, 1997, p. 845.
[7] K. Kreim, Zur Trennung des Gemisches Ethanol-Wasser
mit Hilfe der Gasextraktion, Dissertation, TU HamburgHarburg, Germany, 1983.
[8] M.L. Gilbert, M.E. Paulaitis, Gas /liquid equilibrium for
ethanol-water-carbon dioxide mixtures at elevated pressures, J. Chem. Eng. Data 31 (1986) 296.
[9] G. Brunner, Gas Extraction, Springer, Berlin, 1994.

M. Budich, G. Brunner / J. of Supercritical Fluids 25 (2003) 45 /55


[10] J.S. Lim, Y.Y. Lee, H.S. Chun, Phase equilibria for carbon
dioxide /ethanol /water systems at elevated pressures, J.
Supercrit. Fluids 7 (1994) 219.
[11] K. Nagahama, J. Suzuki, T. Suzuki, High pressure vapor /
liquid equilibria for the supercritical CO2/ethanol/water
system, in: Proceedings of the First International Symposium on Supercritical Fluids, vol. 1, Nice, France, 1988, p.
143.
[12] S. Furuta, N. Ikawa, R. Fukuzato, N. Imanshi, Extraction
of ethanol from aqueous solutions using supercritical
carbon dioxide, Kagaku Kogaku Ronbunshu 15 (1989)
519.
[13] E. Kirschbaum, Destillier- und Rektifiziertechnik,
Springer, Berlin, 1969.
[14] H. Horizoe, T. Tanimoto, I. Yamamoto, Y. Kano, Phase
equilibrium study for the separation of ethanol /water
solution using subcritical and supercritical hydrocarbon
solvent extraction, Fluid Phase Equilib. 84 (1993) 297.
[15] A.P. Bunz, Hochdruckphasengleichgewichte in Mehrkomponentensystemen aus Kohlenhydraten, Wasser, Alkoholen und Kohlendioxid, VDI-Fortschrittbericht 3/406, VDIVerlag, Dusseldorf, 1995.
[16] M. Budich, Countercurrent extraction of citrus aroma
from aqueous and nonaqueous solutions using supercritical carbon dioxide, VDI-Fortschrittbericht 3/606,
VDI-Verlag, Dusseldorf, 1999.
[17] J.T. Meyer, Zur Hydrodynamik von Gegenstromkolonnen
bei der Gasextraktion, Dissertation, TU Hamburg-Harburg, Germany, 1998.
[18] R.P. DeFilippi, J.E. Vivian, Process for separating organic
liquid solutes from their solvent mixtures, US Patent
4349415, 1982.
[19] E. Janecke, Z. Anorg. Chemie 51 (1906) 132.
[20] T. Suzuki, N. Tsuge, K. Nagahama, Supercritical extraction of alcohol from aqueous solution using only carbon
dioxide, in: T. Sekine (Ed.), Solvent Extraction, Elsevier,
New York, 1990, p. 1701.
[21] S. Furuta, N. Ikawa, R. Fukuzato, N. Imanshi, Extraction
of ethanol from aqueous solutions using compressed
carbon dioxide, in: Proceedings of the Second International Symposium on High-Pressure Chemical Engineering, Erlangen, Germany, 1990, p. 345.
[22] K. Suzuki, H. Sue, M. Itou, R.L. Smith, H. Inomata, K.
Arai, S. Saito, Isothermal vapor- /liquid equilibrium data
for binary systems at high pressures: carbon dioxide /
methanol, carbon dioxide /ethanol, carbon dioxide /1propanol,
methane /ethanol,
methane /1-propanol,
ethane /ethanol, and ethane /1-propanol systems, J.
Chem. Eng. Data 35 (1990) 63.

55

[23] R. Wiebe, The binary system carbon dioxide /water under


pressure, Chem. Rev. 29 (1941) 475.
[24] S. Hirohama, T. Takatsuka, S. Miyamoto, T. Muto,
Measurement and correlation of phase equilibria for the
carbon dioxide /ethanol /water system, J. Chem. Eng.
Japan 26 (1993) 408.
[25] S. Takishima, K. Saiki, K. Arai, S. Saito, Phase equilibria
for CO2 /C2H5OH /H2O system, J. Chem. Eng. Japan 19
(1986) 48.
[26] Y.S. Feng, X.Y. Du, C.F. Li, Y.J. Hou, An apparatus for
determining high pressure fluid phase equilibria and its
applications to supercritical carbon dioxide mixtures, in:
Proceedings of the First International Symposium on
Supercritical Fluids, vol. 1, Nice, France, 1988, p. 75.
[27] M. Ponchon, Technol. Mod. 13 (1921) 20, 55. (cited from
Treybal [29]).
[28] R. Savarit, Arts Metiers, 1922, pp. 65, 142, 178, 241, 266,
307. (cited from Treybal [29]).
[29] E.T. Treybal, Mass-Transfer Operations, third ed.,
McGraw-Hill, New York, 1980.
[30] R. Zobel, Supercritical carbon dioxide extraction: Its
application to the food and flavour industries, in: Proceedings of the Second International Symposium on HighPressure Chemical Engineering, Erlangen, Germany, 1990,
p. 271.
[31] J.L. Lorne, J. Adda, Improvement of the extraction yield
of volatile flavour components, in: Proceedings of the First
International Symposium on Supercritical Fluids, vol. 2,
Nice, France, 1988, p. 815.
[32] G. Bunzenberger, R. Marr, Counter-current high-pressure
extraction in aqueous systems, in: Proceedings of the First
International Symposium on Supercritical Fluids, vol. 2,
Nice, France, 1988, p. 613.
[33] L. Bernad, A. Keller, D. Barth, M. Perrut, Separation of
ethanol from aqueous solutions by supercritical carbon
dioxide */comparison between simulations and experiments, J. Supercrit. Fluids 6 (1993) 9.
[34] J.S. Lim, Y.W. Lee, J.D. Kim, Y.Y. Lee, H.S. Chun,
Mass-transfer and hydraulic characteristics in spray and
packed extraction columns for supercritical carbon dioxide /ethanol /water system, J. Supercrit. Fluids 8 (1995)
127.
[35] A. Dondoni, P. Colombo, A. Stassi, A. Schiraldi, Assemblaggio di una colonna per lestrazione e il frazionamento
di matrici liquide con FSC, in: Proceedings of the Third
Italian Conference on Supercritical Fluids, Trieste, Italy,
1995, p. 95.

You might also like