Perrut, 2019. Supercritical Fluids in The Food Industry PDF
Perrut, 2019. Supercritical Fluids in The Food Industry PDF
Perrut, 2019. Supercritical Fluids in The Food Industry PDF
7.7
Supercritical Fluid Applications in the
Food Industry
Michel Perrut, Vincent Perrut
Atelier Fluides Supercritiques, Nyons, France
Gases in Agro-food Processes 483 # 2019 Elsevier Inc. All rights reserved.
https://doi.org/10.1016/B978-0-12-812465-9.00020-7
484 7.7. SUPERCRITICAL FLUID APPLICATIONS IN THE FOOD INDUSTRY
selectivity and reaction rates than in liquid called the critical point (Pc Tc). Beyond this point
solvents. (P > Pc et T > Tc), only one phase exists, called
- Formulation (SFFO): Considerable effort has supercritical fluid (SCF) while the liquid phase
been undertaken to dry and formulate pressurized beyond the critical pressure (P > Pc
ingredients by carrier impregnation and particle et T < Tc) is called subcritical liquid (SCL).
“engineering,” mainly for designing drug- Most compounds exhibit a critical pressure in
delivery systems but also for preparing novel the range of 3–8 MPa, except highly polar mole-
food, cosmetics, and nutraceutical products. cules such as water and ammonia. The critical
- Sterilization and biological applications temperature increases with the complexity of
(Biotech): The unique biocide properties of the molecule and very few “current” compounds
supercritical fluids, and especially of CO2 exhibit a critical temperature between 0°C and
possibly added with various additives, open 50°C (ethane, ethylene, CO2, N2O, CHF3).
new opportunities for pest elimination, Mixtures behave in a more complex way,
sterilization, and virus inactivation in “mild” depending on their composition, but “critical”
conditions (near room temperature) phenomena are also observed.
preserving substrate quality. Obviously, CO2 (PC ¼ 7.38 MPa; TC ¼ 31°C) is
the most attractive supercritical fluid for many
reasons:
7.7.2 PRINCIPLES OF
• Very inexpensive and abundant in pure form
SUPERCRITICAL FLUID PROCESSES
(food grade) worldwide.
• Not flammable, not toxic, and
7.7.2.1 What Is a Supercritical Fluid? environmentally friendly.
All pure compounds can be found in three • Critical temperature 31°C, permitting
states: solid, liquid, and vapor (or gas). On the operations at near-ambient temperature.
(pressure, temperature) diagram as shown in
As shown in Fig. 7.7.2, CO2 density varies in a
Fig. 7.7.1, the three regions corresponding to
wide range versus pressure and temperature,
these three states are separated by curves that
especially around the critical point, but the fluid
meet at the triple point. Surprisingly, the vapor-
becomes almost incompressible at pressure
ization/liquefaction curve presents an end point
beyond 50 MPa.
Although they do not present the advantages
of carbon dioxide, other types of fluids are con-
sidered for specific applications:
• Light hydrocarbons, especially liquefied
propane (PC ¼ 4.25 MPa; TC ¼ 96.7°C), appear
to be much stronger solvents than carbon
dioxide vis-à-vis lipids, but they present a
high explosion hazard.
• Hydrofluorocarbons (HFCs) are
environmentally acceptable at the present
time but very expensive; neither toxic nor
flammable, they may decompose in highly
toxic gases when submitted to a flame.
• Dimethyl ether, used as liquefied gas,
FIG. 7.7.1 General pressure-temperature diagram for
pure compounds. behaves as a “polar” solvent and is able to
FIG. 7.7.2 Carbon dioxide density versus pressure at different temperatures. From NIST http://webbook.nist.gov/chemistry/
fluid/.
dissolve a very wide range of compounds, monitoring pressure and temperature, particu-
including many polymers (Lemert and De larly on the possibility of varying their solvent
Simone, 1991). power over a wide range.
• Nitrous oxide (N2O) exhibits similar critical Carbon dioxide always behaves as a rather
properties as CO2; although it is commonly weak nonpolar solvent that selectively dissolves
used in anesthesia, it should be considered as lipids such as vegetable oils, butter, fats, hydro-
an oxidant that may lead to explosion when it carbons, and essential oils, but has a weak affinity
contacts flammable solutes. with oxygenated or hydroxylated molecules. It
• Water exhibits very attractive properties at does not dissolve any hydrophilic compound
subcritical or supercritical conditions that are such as sugars and proteins or mineral species
in a completely different range (PC ¼ 22.1 MPa, such as salts or metals. However, CO2 solvent
TC ¼ 374°C). Great variation of its dielectric power and polarity can be significantly increased
constant results in polarity and solvent by adding a polar cosolvent that is generally cho-
properties varying from the exceptionally sen between short-chain alcohols, esters, or
high polarity of common liquid water to a ketones. For obvious reasons, ethanol is often pre-
nonpolar fluid dissolving organics and ferred as it is abundant and cheap in pure forms
precipitating salts in supercritical conditions. (food grade, pharmacopeia grade) while being
nonhazardous to the environment and not very
toxic. Moreover, it is to be noticed that although
7.7.2.2 Solubility in Supercritical
water is only slightly soluble in supercritical CO2
Solvents (1–3 g/kg), it plays a very important role as a
Most applications of supercritical fluids (and “cosolvent” for many polar molecules; in fact,
subcritical liquids) are related to their “tunable” water is present in most applications, especially
properties that can be easily changed by when natural products are processed.
Fluid phase equilibria of mixtures are very com- different organic compounds (Billoni et al.,
plex, and many types of phase diagrams can be 1988) without any complicated calculations:
found. After the basic work of FRANCIS in 1954
(Francis, 1954, 1955), who measured solubilities C5ρk : exp ½a=T + b (7.7.1)
in liquid CO2 of hundreds of components and
where C is the solute concentration and a, b, and
phase equilibria of tens of ternary systems
k the empirical constants. In fact, this simple
including liquid CO2, thousands of articles have
equation shows that:
dealt with high-pressure fluid phase equilibria,
covering a very wide range of compounds and • Solubility is strongly dependent on the fluid
operating conditions. The reader could start on specific gravity ρ as k is always positive and
books focusing on natural products (Stahl in the range of several units; it drastically
et al., 1987; King and Bott, 1993; Rizvi, 1994; decreases when the fluid is depressurized at
McHugh and Krukonis, 1994; King and List, constant temperature below its critical
1996) and symposium proceedings (ISASF, pressure, with a solubility variation of several
1988–2016). Recent progress in thermodynamic orders of magnitude permitting fluid-solute
modeling permits predicting the behavior of separation.
many mixtures; however, some measurements • Solubility increases with pressure at constant
that are difficult to perform are yet required to temperature but solubility may increase or
set interaction parameters that cannot be calcu- decrease when temperature is raised at
lated, especially for polar liquids (for example constant pressure. In fact, experimental
in the case of strong hydrogen bonding). But values on various substances show that
happily, it is not always necessary to handle below a certain “inversion” pressure,
detailed thermodynamic data of the processed solubility decreases when temperature is
mixtures to design SCF processes. increased while beyond this pressure,
For “simple” systems and relatively low solu- solubility increases with temperature. For
bility, the empirical correlation proposed by example, vanillin solubility curves versus
Chrastil (Chrastil, 1982) can be used to interpret pressure at various temperatures intersect at
experimental results with a good reliability that an “inversion” pressure of 140 bar
satisfactorily fits the solubility data of 104 (Fig. 7.7.3); meanwhile, the inversion
VANILLIN solubility
10
Mass fraction (g/kg)
4 42°C
51°C
2
56°C
0
60 80 100 120 140 160 180 200 220
Pressure (bar)
FIG. 7.7.3 Vanillin solubility in CO2. Data from Billoni, N., Jose, J., Merlin, J.C. 1998. Solubility of heavy components in super-
critical CO2 using directly coupled supercritical fluid extraction-HPLC. In: Perrut, M. (Eds.), Proceedings of International Symposium on
Supercritical Fluids, Nice, France, ISBN 2-905267-13-5, Tome 1, pp. 373–380.
FIG. 7.7.4 CO2 viscosity versus pressure at different temperatures. From NIST http://webbook.nist.gov/chemistry/fluid/.
transfer (and similarly heat transfer) is fast in SCF • Extraction of nonpolar or low-polarity
in comparison with liquid solvents or water. compounds present in the raw materials in
Moreover, SCF rapidly diffuses into porous conditions where the fluid is a powerful
media, easing either extraction from solid mate- solvent (supercritical or subcritical with a high
rials or impregnation of solutes (NIST, n.d.; specific gravity); for solid raw materials, this is
ISASF, 1988–2016; Reid et al., 1986; Brunner, 1994). operated in batch mode as there are no means
Another important property of SCF solvents is available to introduce and withdraw solid
related to the drastic viscosity reduction of a liquid materials to and from a high-pressure vessel
phase contacted with a SCF solvent that partly (except in the very special case of coffee
dissolves in the liquid, even at a pressure below beans). In very large-scale plants, feed loading
critical. This is of use for fluidifying viscous oils and spent material unloading are operated by
and waxes in order to ease processing (filtering, pneumatic transport; in lab, pilot, and
reaction, extraction). Similarly, many polymers medium-scale plants, feed is introduced in a
are “swollen” and “plasticized” by compressed “basket” consisting of a cylinder closed by
carbon dioxide, with a decrease of the glass tran- two filters that is installed in the pressure
sition temperature by several tens of degrees Cel- extractor.
sius, permitting an easy processing (forming and • Fluid-extract separation in conditions where
atomizing, mixing, grafting, foaming), as shown the fluid is a weak solvent, by
in Fig. 7.7.5 for PEG6000 (Deschamps, n.d.). depressurization and heating to operate
almost isothermally, avoiding formation of
liquid CO2, or, in some rare cases, by
7.7.3 SUPERCRITICAL FLUID heating or fluid scrubbing at constant
EXTRACTION (SFE) pressure.
• Fluid recycle either by gas liquefaction, liquid
7.7.3.1 SFE Concept pumping and heating to extraction
temperature, or by gas compression and
The drastic variation of solvent power of
cooling to extraction temperature. In very
supercritical fluids with pressure and tempera-
large plants, the second cycle is preferred
ture permits designing extraction processes
while the first one is operated in most
based on the following concepts:
lab, pilot, and medium-scale extraction
plants.
65 This batch mode imposes costly operation of
pressure cycling and material handling, what-
Binary system
Melting point (°C)
35
0 40 80 120 160 7.7.3.2 Extraction Kinetics
Pressure (bar)
FIG. 7.7.5 Melting point decrease of PEG6000 in presence
Thousands of works on thousands of differ-
of pressurized CO2 (Deschamps, n.d.). ent materials, mostly natural products, have
been published with or without modeling
3 1: Us = 10 kg/h
Diffusion
2 2 : Us = 15 kg/h
3 : Us = 30 kg/h
1 Solubility :
y/yT = a. S
FIG. 7.7.7 Oil extraction from sunflower seeds with pure CO2 at various fluid flow rates (Perrut et al., 1997).
Spices
100
80
Yield/Total yield ratio (%)
caraway, a = 3
60
curcuma, a = 35
paprika, a = 4
40 pimento, a = 4
black pepper, a = 8
20 coriander, a = 7
nutmeg, a = 6
0
0 100 200 300 400 500
S *a (kg/kg)
FIG. 7.7.8 Spices extraction with supercritical CO2 (250–300 bar; 40–60°C).
0
0 5 10 15 20
Solvent/feed ratio (kg/kg)
FIG. 7.7.9 Extraction of Piper cubeba by pure CO2 (90 bar; 40°C; 3.86 kg CO2/kg feed/h).
teaches valuable information related to feed when the extract is readily soluble in water like
preparation that is of major importance for scal- caffeine or vanillin.
ing up to commercial scale (see Section 7.7.3.3). In several processes operated at a very large
scale for long periods, the feed must be hydrated
before processing. Coffee beans are saturated in
7.7.3.2.3 Humid Feeds water prior to supercritical CO2 decaffeination;
But, another parameter can modify the extrac- similarly, cork is humidified before treatment
tion curve shape: The moisture content in the in order to remove trichloroanisole to make
processed material. In fact, most “dry” natural “safe” wine stoppers and ginseng powder is trea-
products always contain some moisture. This ted for pesticide elimination.
water has a positive effect on extraction kinetics, This effect is illustrated in Fig. 7.7.10, where
as shown by many works, when its concentration we compare extraction curves of “crude” and
is limited (generally below 10% m/m), moreso humidified vanilla powder.
60%
40%
Powder
Powder + water
20%
0%
0 20 40 60 80 100 120
Solvent ratio (kg CO2/kg feed)
Black-current
25
20
10
Seeds 1 700 µm
5 Seeds 2 700 µm
Seeds 1 1,500 µm
0
0 50 100 150 200 250
Solvent ratio (kg CO2/kg feed)
FIG. 7.7.11 Total extract of black-current seeds by pure CO2 (200 bar; 55°C; 10–14 kg CO2/kg feed/h).
Black-current oil
But, water is also extracted by supercritical 100
CO2 as, at the difference with hexane to which
4.0
3.0
2.0
1.0
0.0
0 2 4 6 8 10 12 14
Solvent ratio (kg CO2/kg feed)
4.0
3.0
2.0
1.0
0.0
0 2 4 6 8
Solvent ratio (kg CO2/kg feed)
4
properties.
Moreover, a polarity modifier could be added
3 after this step-wise extraction with pure CO2,
2 DM 67% leading to higher-polarity compounds, as exem-
DM 32% plified in the extraction/fractionation of Ortho-
1
siphon (Java tea) presented in Fig. 7.7.16. The
0 first step is at 120 bar and the second step at
0 5 10 15 20 25
280 bar with pure CO2, followed by a third step
Solvent ratio (kg CO2/kg dry matter)
at 280 bar with the addition of ethanol, leading to
FIG. 7.7.15 Extraction of rosemary by pure CO2 (300 bar; three different fractions of growing polarity.
44°C; 13.3 kg CO2/kg dry matter/h).
Orthosiphon stamineus
4.0
2.5
2.0
1.5
1.0
0.5
0.0
0 5 10 15 20 25 30 35
Solvent ratio (kg CO2/kg feed)
FIG. 7.7.16 Extraction of Java tea leaves (Orthosiphon stamineus) (120 bar then 280 bar then 280 bar +9.5% EtOH; 40°C;
3–3.5 kg CO2/kg feed/h).
(A)
(B)
FIG. 7.7.17 (A) Linalool Thyme: Total extract (left) before and after water decantation by centrifugation. (B) Linalool
Thyme: Total extract (left) leads to wax powder (center) and aroma oil (right) after ethanol/water reprocessing.
filtering at low temperature). The sticking crude slow rate, and at 280 bar leading to an unpleas-
extract (“concrète”) with a yield of 4% leads to a ant smelling dark extract. But, by fractionating
wax powder with a yield of 1.8% and an aromatic the extract through a two-step separation
oil with a yield of 1.5%; water and losses repre- (65 bar—20°C and 55 bar—40°C), a viscous and
sent the complement (Fig. 7.7.17B). unpleasant-smelling fraction is obtained in the
Another typical example is presented in first separator and an extract very similar to
Fig. 7.7.18, which shows vanilla bean extract the first one is collected in the second separator
after decantation of the three phases: an oil at a much higher rate than at 100-bar extraction
phase on top and an intermediate aqueous (intermediate curve). By this way, a much lower
phase rich in vanillin. amount of fluid is required.
Hemp
4.5
3.0
Mass yield (%)
280 bar
(fractionation)
2.5
2.0
1.5
1.0
0.5
0.0
0 5 10 15 20 25 30 35
Solvent ratio (kg CO2/kg of feed)
FIG. 7.7.19 Extraction of hemp by pure CO2 with or without extract fractionation (120 and 280 bar–40°C: 6–10 kg CO2/kg
feed/h).
seldom operated in a unique extractor at the com- depressurization (spent feed unloading; fresh
mercial scale, as it is far from optimum except in feed loading) and recompression (Fig. 7.7.20).
very special high-value product processing. In In order to reduce the quantity of CO2 needed
most cases, several extractors are operated in for a given feed mass, it is valuable to simulate a
series in order to minimize CO2 losses and pro- countercurrent flow between feed and fluid.
cess cycle duration. At least three extractors Supposing that three extractors are implemen-
(Del Valle et al., 2014; Nunez and Del Valle, ted. CO2 is first contacted with a partially
2014; Del Valle, 2015) are employed, two or more extracted feed and then is saturated by percola-
ones being in extraction while one is in the tion through the “fresh” feed extractor.
Extractors Separators
Condensor
Extract
CO2
CO2 reservoir
Heater
CO2 pump
Co-solvent
pump
Then, when the first extractor feed is exhausted, profitable in light of pilot-scale kinetics.
this extractor is depressurized and unloaded. In fact, many natural products must be
Meanwhile, the second one becomes the first to pretreated before SFE. To avoid “caking,”
be percolated and the newly loaded third one hops are pelletized while waxy and oil-rich
becomes the second to be contacted. By this feeds often need to be mixed with an inert
way, the fluid exiting the extractor battery is packing such as food-grade cellulose
almost always saturated in extract and, conse- fibers. Great care must also be taken to
quently, is “optimally” used. Of course, with four avoid the formation of very fine particles
or five extractors, this countercurrent implementa- during raw material milling. Dust may
tion is facilitated, but global cost suffers of supple- be responsible of filter blockage, leading
mentary investment (Del Valle et al., 2014; Nunez to safety issues during basket
and Del Valle, 2014; Del Valle, 2015). depressurization, filter collapse entraining
The following points must also be empha- transport of feed powder throughout the
sized when scale up is to be implemented: plant, or basket deformation requiring a
costly maintenance.
• Feed preparation is of key importance as feed • Pilot-scale optimization is of key importance
agglomeration and fluid channeling may with the analysis of extraction kinetic curves
jeopardize an operation that could appear and extract composition in various conditions
1.40%
1.20%
1.00%
% carotenoid
P = 600 bar
0.60% P = 300 bar
0.40%
0.20%
0.00%
0 20 40 60 80 100
Solvent ratio
FIG. 7.7.21 Sea buckthorn (Hippophae rhamnoides) extraction: carotenoid content in extract by pure CO2 (Clavier et al.,
2014).
deodorization and decoloration; polyunsatu- separations are not at all efficient enough in this
rated fatty acid ester fractionation to obtain case, even with high performance cyclonic sep-
enriched cuts of ω3-esters (essentially EPA or/ arators; adsorption of the organics carried by
and DHA); mono-, di-, and tri-glyceride frac- gaseous CO2 appears to be the most convenient
tionation; phospholipid and glycol-lipid purifi- process, even if it requires a periodic system for
cation; concentration of squalene and aroma recovery by desorption with supercritical
tocopherols; etc. (Brunner, 1994; Riha and CO2.
Brunner, 2000; Fiori et al., 2014; Perrut et al., Schultz and Randall (1970), Schultz et al.
2007). From these numerous works, it appears (1974) founded the basic knowledge of selective
that SFF of oils is very selective, either based fractionation of aroma compounds from alco-
on fatty acid chain length (but not at all on the holic aqueous mixtures, using liquid CO2. They
unsaturation number) or on polarity differences established simple relations between the carbon
(for minor polar molecules concentration) as for numbers of the alcohols or esters and their dis-
isolation of pure fractions of digalactosyl- tribution coefficient between water and liquid
diglycerides from wheat gluten lipids (Perrut CO2 (Francis, 1954). In the late 1970s and the
et al., 2007). 1980s, numerous investigators worked on etha-
nol extraction from fermentation broth in order
7.7.4.2.2 Alcoholic Beverage Fractionation to obtain “dry” ethanol to be used as fuel in
Alcoholic beverage fractionation is of high motors. However, it rapidly appeared that
economical interest, as the manufacture of CO2 is not selective enough to “break” the azeo-
low-alcohol beverages is of increasing interest trope, which ruins any interest in this process.
for ethical, religious, and/or dietetic motiva- Concerning aroma-ethanol separation, attrac-
tions, and improving their quality will lead to tive data (Di Giacomo et al., 1991) evaluate phase
growing profitable markets. Alcoholic bever- equilibria at various pressures and temperatures
ages contain mainly water and ethanol (3– between CO2, water, ethanol, and a few aro-
10 wt% for fermented ones and 40–60 wt% for matic compounds that are present at significant
distilled ones). Meanwhile, aromas are present concentrations in wine and beer (Table 7.7.2),
at a total concentration between 500 and showing that, as expected, selectivity (defined
5000 ppm comprising hundreds of components as α ¼ ðaroma wt=ethanol wtÞ in extract
ðaroma wt=ethanol wtÞ in feed ) decreases when
at the trace level, with the main ones being very
ethanol concentrations in the liquid phase and
similar to ethanol (ethyl acetate, other alcohols).
optimal conditions can be found for “mild” con-
Hydrophilic components are also present
ditions (100 bar, 40°C), although selectivity
(sugars, proteins, colorants) that often cause liq-
seems maximum with liquid CO2 (60 bar, 20°C).
uid foaming.
Wine and wine must (Perrut and Nunes da
• Process concept Ponte, 1997; Ruiz-Rodriguez et al., 2010, 2012;
Macedo et al., 2007), beer (Perrut and Nunes
In fact, such fractionation represents a very
da Ponte, 1997), and cider (Medina and
difficult challenge as ethanol increases the fluid
Martinez, 1997) dealcoholization have been
solvent power, reducing its selectively vis-à-vis
investigated for many years.
water and aroma products. High selectivity
between ethanol and aroma extract requires a
• Wine fractionation
high reflux in very “mild” conditions (low fluid
density and solvent power). Moreover, aroma SFF was proved to be a valuable process to
collection is extremely complex (high dilution fractionate wine, leading to a tasteless raffinate
and high volatility). Classical mechanical and a global extract yield near 0.6 wt%
From Di Giacomo, G., Brandini, V., Del Re, G., Martinez de la Ossa, E. On the feasibility of dense carbon dioxide based extraction-recovery process for alcohol-
reduced beer and wine production. In: Perrut, M. (Ed.), Actes du 2ème Colloque sur les Fluides supercritiques, Paris, France, 1991, ISBN 2-905267-17-8,
pp. 63–68.
comprising a first pink ethanol-rich fraction a tasteless raffinate. This process was validated
smelling wine, and an aromatic-rich extract frac- at the semiindustrial scale (several tens of tons
tion concentrated in heavy components with a of feed) on different feeds: 0.12 yield on cognac
much smaller yield of 1 to 3.104. As tested by (53 wt% ethanol), 0.7% yield on whiskey (58 wt%
a panel of wine experts, the organoleptic quality ethanol), and 1.65% yield on Rum (49 wt% etha-
of these aromatic extracts is strongly dependent nol). Extracts are considered as excellent by a
on the recovery process used and is very satis- taste panel and either comparable or better than
factory when optimized (Perrut and Nunes da those obtained by dichloromethane extraction,
Ponte, 1997). especially for rum where the aromatic power
and spectrum are judged as exceptional
• Beer dealcoholization
(Perrut and Nunes da Ponte, 1997).
For a long time, most ethanol-free beer (<0.5
or 1% vol. ethanol) has been considered of poor 7.7.4.2.3 Essence and Concrete
organoleptic quality. So, selective fractionation Fractionation
of beer aromas is of key interest, either for rein- For 30 years, citrus peel oil deterpenation
corporation in the aroma-free raffinate after has received wide attention as it requires a
ethanol depletion (distillation, reverse osmo- very selective process due to the fact that lim-
sis, or pervaporation) or for aroma reinforce- onene is readily soluble in CO2. Also, this mix-
ment of low-alcohol beers. The best aroma ture is a very strong solvent from which it is
concentrates obtained by SFF correspond to α difficult to separate the oxygenated terpenes
values as high as 30. However, extraction effi- selectively (Stahl et al., 1987). For the separa-
ciency varies for the different aroma com- tion of such hydrocarbon-oxygenated com-
pounds and the global aroma profile might pounds, a combination of SFF and
be different from the original beer one adsorption (frontal chromatography) is very
(Perrut and Nunes da Ponte, 1997). efficient (Barth et al., 1994).
SFF is also efficient for flower concrete repro-
• Fractionation of distilled beverages
cessing in order to obtain a high-value extract,
The aroma of distilled beverages has been similar to an absolue by selective elimination of
currently extracted by hazardous organic sol- waxy compounds widely extracted by organic
vents (pentane and dichloromethane). Mean- solvents such as hexane. This process was
while, SFF in mild conditions (120–140 bar, described on model molecules consisting of lim-
30–50°C) leads to a very high extract yield and onene and canola oil (Yasumoto et al., 2014).
FIG. 7.7.23 Kava-Kava root (left), CO2 extract (center) and impregnated powder (right) (Majewski & Perrut, 2000).
in a stirred vessel in which the extract is depos- 7.7.6.3 Particle Design Using
ited into a porous excipient. This permits avoid- Supercritical Fluids
ing the handling of a very viscous and sticky
extract and readily obtaining a free-flowing Many processes are being developed for
powder that is easy to handle for commerciali- manufacturing various types of particles for
zation (Fig. 7.7.23). applications such as drug-delivery systems, per-
Other applications were also reported fumes and cosmetics, or aromas in food prod-
with impregnation of polymers by APIs (i.e., ucts. Supercritical fluids (mainly CO2) are used
antibiotics in soft lens), colorants in polycarbon- in different ways, leading either to nano- or
ate view glasses, or perfumes in beauty microparticles with the possibility of tuning
accessories. their crystal polymorphism or to complex
microparticles or capsules ( Jung and Perrut,
2001). To summarize, we can cite the basic
concepts:
7.7.6.2 Foodstuff Drying
Although CO2 is a weak solvent of water, it - Solvent power modulation: The substance is
can be used to dry natural products at the condi- dissolved in fluid at high pressure and
tion that a high solvent/feed rate is acceptable depressurized though a nozzle for very fine
and the fluid itself is dried after contacting the particle precipitation according to RESS (rapid
product. According to a recent patent (Agterof expansion of supercritical solutions). There
et al., 2007), a demonstration plant is now operat- are two main issues at the commercial scale:
ing to dry a food product by CO2 in an innovative Heating the fluid to avoid liquid CO2
pressure vessel incorporating a high flow-rate formation and microparticle collection for
turbine pump. The water-loaded fluid is perco- which innovative systems are proposed such
lated through a specific adsorbent bed (i.e., 3A- as deep filtration of microparticle-loaded CO2
zeolite) prior to being recirculated. This process through a bed of excipient macroparticles, as
is claimed to lead to high-quality dried products shown in Fig. 7.7.24 (Perrut et al., 2005).
with better preservation of valuable compounds - Antisolvent: The substance is dissolved in a
that are often destroyed by heat in spray drying, polar solvent (particles, organic, or water) that
and with less energy consumption. is pulverized into a stream of CO2 (possibly
FIG. 7.7.24 Capture of microparticles (right) of lovastatin generated by RESS in CO2 onto lactose particles (left) (Perrut
et al., 2005).
added with ethanol) that dissolves into the United States. Most of them are processing
solvent and diminishes its polarity so as to natural raw materials for food ingredients,
precipitate the substance while the solvent is nutraceuticals, pharmaceuticals, and cos-
carried out by the fluid. Solvent elimination metics. This number is now increasing follow-
from the processed solid is often very difficult, ing the demand boost for organic ingredients
particularly when “heavy” solvents are used manufactured by “sustainable” techniques.
(such as DMSO, DMF, or NMP).
- Plasticizer agent: CO2 dissolves in lipids and
various polymers and “plasticized” these
decreasing their viscosity and their melting 7.7.7.2 New Trends for New
points, even at moderate pressures (40 to 80 bar) Applications
as shown earlier in Fig. 7.7.16. The innovative
processes known as PGSS (particles from gas- Increasing extraction pressure appears as
saturated suspensions) or FAME (fluid-assisted a major trend for SFE development, although
microencapsulation) rely on this phenomenon: this possibility to increase fluid solvent
Dispersion of an active-ingredient powder in an power is not always a plus. It is to be empha-
excipient liquefied by CO2 saturation followed sized that optimization of the global processing
by rapid depressurization through a nozzle chain, including material pretreatment,
down to atmospheric pressure, leading to core- extraction, and final extract processing, is the
shell microcapsules. key for reaching high-value, easy-to-use
products.
7.7.7 PRESENT AND FUTURE SFF presents great potential for several food
INDUSTRIAL DEVELOPMENT fractionation challenges concerning dairy prod-
ucts, various oils, polar lipids, flower concretes,
7.7.7.1 Hundreds of Plants Now Operate alcoholic beverages, etc.
in the World, Mainly for Food Moreover, the combination of SFE or SFF with
online formulation of active ingredients is also
Applications
of major industrial interest. Meanwhile, pest
As many as 400–500 large-scale plants are elimination and “CO2 pasteurization” are now
now operating worldwide, mainly in Europe, considered favorably as green alternatives to
Asia (China, South Korea, and Japan) and the “classical” treatments.