oil extraction and analysis phần 5 pps

Chapter 5
Analytical Supercritical Fluid Extraction
for Food Applications
Tracy Doane-Weideman and Phillip B. Liescheski
Isco Incorporated, Lincoln, NE 68504
Abstract
In this review, we explore the fundamental concepts of supercritical fluids and
supercritical fluid extractions. Carbon dioxide and other solvents are discussed; the
solubility theory is introduced together with the calculation of the density of carbon
dioxide. The state-of-the-art instrumentation is presented in terms of fundamental
components. The most widely used application of analytical SFE is in the food
industry and this review includes fats, oils, vitamins, and pesticides in research and
routine applications.
Introduction
Supercritical fluid extraction (SFE) is becoming an important sample preparation
method in the chemical analysis of food products, especially for fats and fatty oils.
SFE has been used successfully for over a decade in analyses of food samples
(1,2). The most popular SFE solvent is carbon dioxide (CO
2
). Triglycerides, cho-
lesterol, waxes, and free fatty acids are quite soluble in supercritical CO
2
. The sol-
ubility of polar lipids, such as phospholipids, can be improved by augmenting the
supercritical CO
2
with a small addition of ethanol or other polar modifier solvent.
Even though CO
2
is considered a “green-house” gas, it is ubiquitous in nature and
can be retrieved from the environment and returned clean (3). As a result, SFE can

still contribute positively to “Green Chemistry.” CO
2
has the additional advantage
of being nonflammable and less toxic than most organic solvents. For example,
petroleum ether, which is commonly used in fat extractions, can be easily detonat-
ed by static electricity, and diethyl ether can form explosive peroxides. On the
other hand, some fire extinguishers use CO
2
, which is also commonly found in
foods and drinks such as bread and carbonated drinks. Finally, several common
chlorinated solvents are banned by law, and supercritical CO
2
can be an alternative
to these solvents. All of these factors make SFE attractive.
What Is a Supercritical Fluid?
A supercritical fluid is a dense gas (4). It is compressible and thus expands to com-
pletely fill its container. A liquid, on the other hand, takes the shape of its container
Copyright © 2004 AOCS Press
but does not expand to fill the container. Instead it settles at the bottom. Supercritical
fluids, unlike the air we breathe, have densities comparable to liquids. As a result,
these fluids have solvating power. A supercritical fluid can be defined as a form of
matter in which the liquid and gaseous phases are indistinguishable (5).
The three most common phases of matter on earth are solid, liquid, and gas.
The phase of a pure simple substance depends on the temperature and pressure. A
plot showing a substance's phase for a given temperature and pressure is called a
phase diagram. Figure 5.1 is a phase diagram for CO
2
. In a phase diagram, the
solid, liquid, and gas regions are divided by branches or equilibrium curves. These
curves represent valuable information concerning the substance's melting, boiling,

or sublimation temperatures at given pressures. These curves characterize the tem-
peratures and pressures at which two phases coexist in equilibrium. For example,
the liquid-gas equilibrium curve divides the liquid and gaseous phase regions. On
this curve, the substance coexists as both a liquid and gas (vapor) in equilibrium.
When the temperature and pressure change so that the substance leaves the liquid
phase region and crosses the equilibrium curve into the gas phase region, the sub-
stance boils. As its state crosses this curve, there is an entropy change. In the case
of boiling, its entropy increases and absorbs heat, known as heat (enthalpy) of
vaporization. In the case of condensation, its entropy decreases and liberates heat,
known as heat of condensation. An obvious physical change is seen in the sub-
stance as its state crosses one of these curves.
Fig. 5.1. Phase diagram of carbon dioxide.
Temperature (°C)
Pressure (atm)
Copyright © 2004 AOCS Press
The three equilibrium curves (solid-gas, solid-liquid, liquid-gas) intersect at a
common point, called the triple point. At this point, the substance coexists in equi-
librium with all three phases. Each substance has only one triple point. The solid-
liquid equilibrium curve radiates from the triple point to infinity. The solid-gas
equilibrium curve radiates from the triple point and eventually terminates at
absolute zero and vacuum. The liquid-gas equilibrium curve does not radiate indef-
initely from the triple point but terminates at another important point, called the
critical point. This point is the critical temperature and critical pressure of the sub-
stance. Beyond the critical point, there is no longer an equilibrium curve to divide
the liquid and gaseous regions; thus, the liquid and gas phases are no longer distin-
guishable. There are no physical changes observed as the substance's state crosses
over this region. This region of the phase diagram is sometimes called the super-
critical fluid region.
The critical temperature is the temperature above which the substance can no
longer be condensed into a liquid. Increasing the pressure will not induce conden-

sation. For a liquid that partially fills a tube, the liquid's meniscus disappears when
it is heated above the critical temperature. The critical pressure is the vapor pres-
sure of the substance at its critical temperature. It is also the maximum vapor pres-
sure of the substance because at a higher temperature, the liquid phase cannot be
distinguished from its vapor. Based on the critical point, a supercritical fluid can
also be defined as the state beyond the critical temperature and critical pressure of
the substance. The critical temperature for CO
2
is 31.1°C, and its critical pressure
is 72.84 atm.
Supercritical fluids can be considered on the molecular level. Molecules have
both kinetic and potential energies. The kinetic energy is related to the motion of
the molecules, which depends on the temperature. The potential energy is related
to the Van der Waal force, the close proximity attractive interaction between mole-
cules. The potential energy of molecules depends on how close they are to each
other. This attractive force between solvent molecules and solute molecules allows
for solvation, the dissolving process. It also allows solvent molecules to “stick”
together into clusters, thus forming a liquid. The molecules aggregate locally but
there is no long-range order as observed in solid crystals. In the liquid phase, exter-
nal pressure is not required to keep the molecules close together because they
already “stick” together. However, these “sticky” molecules also give rise to high-
er surface tension, viscosity, and slower diffusion. Such properties can hinder an
extraction process. In supercritical fluids, the temperature is above the solvent's
critical temperature. At these higher temperatures, molecules move more quickly
and thus have a higher kinetic energy. This higher kinetic energy reduces the sig-
nificance of the potential energy to a point at which the molecules no longer
“stick” together. As a consequence, lowered surface tension, viscosity, and faster
diffusion allow supercritical fluids to perform better during extraction. Lower sur-
face tension allows the fluid to “wet” surfaces better and to penetrate more deeply
into small pores and features. However, higher pressure is required to keep the

Copyright © 2004 AOCS Press
molecules close together to maintain the molecular attraction necessary for solva-
tion. The pressure must be at least above the critical pressure. Higher pressure
yields higher density at a constant temperature. In turn, higher density yields
greater solvating power.
Advantages and Disadvantages
The most popular SFE solvent is carbon dioxide. There are several reasons for its
popularity. First, CO
2
is inexpensive and commercially available even at high puri-
ty. Second, it is nonflammable, unlike many organic solvents, and is used in some
fire extinguishers. It also does not support combustion, except in the extraordinary
case of burning magnesium. Third, CO
2
is relatively nontoxic, especially in com-
parison to many organic solvents; it is actually present in air, foods, and drinks.
Some caution must be followed with the use of CO
2
. Because CO
2
does not sup-
port combustion or human respiration, it can be an asphyxiant at high concentra-
tions. Fourth, its critical temperature is low, allowing it to be used to extract ther-
mally liable analytes. Its critical temperature and pressure are easily attainable. As
a comparison, the critical temperature of water, 374.0°C, is a challenge for many
materials. Finally, CO
2
is environmentally compatible. Even though it is consid-
ered a “green-house” gas, it is ubiquitous in nature.
Other solvents have been used in SFE, but they have serious drawbacks.

Nitrous oxide has been used in extracting environmental samples, but it is a strong
oxidizing agent. An explosion was reported due to its use with organic materials
(6). Nitrous oxide, also known as “laughing gas,” has narcotic properties. Propane
has been used in the extraction of fats from food samples, but it is highly flamma-
ble and a small leak in the extractor plumbing could be a disaster. Ammonia is
polar and has a practical critical point, but it is a strong, corrosive base and is toxic
and obnoxious. Fluoroform is also polar and has a practical critical point, but it is
expensive and not readily available (7). It may also damage the environment.
Freons have proven to be excellent SFE solvents, but they are suspect in damaging
the environment, especially the ozone layer. As a result, they are being banned.
Water, which is environmentally friendly, has too high a critical temperature to be
practical. Also, its pK
w
approaches 1 at this temperature, making it quite corrosive
to steels and other metals.
Carbon dioxide does have a few disadvantages. First, it is practically the only
solvent for SFE, as shown in the previous paragraph. Even though supercritical flu-
ids offer flexible solubility depending on pressure, CO
2
still has limited solvating
power. As a rule of thumb, its solvent strength is comparable to that of hexanes.
Because it is nonpolar, extracting polar analytes can be a challenge. Fortunately,
the solvent strength of CO
2
can be enhanced by the addition of a small amount of
polar modifier solvent or a surfactant agent. The extraction of polar analytes can be
improved by the addition of a small amount of ethanol. In the area of supercritical
fluid cleaning, DeSimone and colleagues (8,9) developed fluorinated dendritic sur-
Copyright © 2004 AOCS Press
factants that significantly enhanced the cleaning performance of CO

2
when added
in small amounts. Second, the high pressure necessary for SFE is a concern
because some people are still uncomfortable with such pressures. Current technol-
ogy makes such pressure safe to use in the laboratory; however, it renders the
equipment expensive. To reduce the cost of the equipment, sample size is restricted
because smaller high-pressure vessels are safer and less expensive. Smaller sam-
pling size can be a disadvantage though for nonhomogeneous sample matrices.
Smaller sampling sizes can also reduce the detection sensitivity of the analytical
method and increase measurement error. These disadvantages must be addressed to
develop a successful SFE application method.
Giddings-Hildebrand Solubility Theory
The solubility of analytes in a supercritical fluid can be treated by thermodynamics.
The Gibbs free energy of the mixing process can be described by the Gibbs-
Helmholtz equation:
∆G
mix
= ∆H
mix
– T∆S
mix
For the process of solvation to proceed spontaneously, the value of the Gibbs free
energy of mixing ∆G
mix
must be negative. Because the solvation (dissolution)
process increases the disorder of the analyte-solvent system, the entropy of mixing
∆S
mix
is expected to have a large positive value. The spontaneity of the solvation
process ultimately depends on the heat of mixing ∆H

mix
. A smaller value predicts
greater solubility (10).
Using the earlier work of van Laar and Lorenz on the vapor pressure of binary
liquid mixtures according to the Van der Waal equation, Hildebrand and Scott
showed that the heat of mixing ∆H
mix
can be expressed as follows:
∆H
mix
= ϕ
s
ϕ
n
(a
s
1/2
/V
s
– a
n
1/2
/V
n
)
2
where ϕ
s
ϕ
n

is the partial volume factor, V
s
and V
n
are the molar volumes for the
solvent and analyte, and a
s
and a
n
are the Van der Waal intermolecular attraction
parameters for the solvent and analyte. They further showed that
a
1/2
/V = (∆E
v
/V)
1/2
where ∆E
v
is the energy of vaporization of either the liquid solvent or liquid analyte.
The heat of mixing ∆H
mix
is produced from the breaking and reformation of attractive
forces between solvent-solvent, analyte-analyte, and solvent-analyte molecules.
Intuitively, it should be related to their energy of vaporization. Seeing the physical sig-
nificance of this formula in reference to solvation, they defined it as the solubility
parameter δ, also known as the Hildebrand solubility parameter:
Copyright © 2004 AOCS Press
δ = a
1/2

/V = (E
v
/V)
1/2
so that
∆H
mix
= ϕ
s
ϕ
n
(δ
s
– δ
n
)
2
where δ
s
and δ
n
are the Hildebrand solubility parameters for solvent and analyte
(11). To keep the heat of mixing as small as possible, the solubility parameter of
the solvent should be similar in value to the solubility parameter of the analytes. A
smaller difference gives higher solubility.
The solubility parameter for a supercritical fluid cannot be determined from
the energy of vaporization as with liquids because the liquid and vapor phases cannot
be distinguished (12). Giddings and colleagues assumed that a supercritical fluid sol-
vent can be described qualitatively by the Van der Waal state equation. The intermole-
cular attraction parameter can be related to the critical values of the solvent as

a = 3P
c
V
c
2
where P
c
and V
c
are the critical pressure and critical molar volume, respectively
(13). Upon substitution, they obtained
δ
s
= (3P
c
)
1/2
V
c
/V
The volumetric ratio can be written in terms of the reduced density as
V
c
/V = ρ/ρ
c
= ρ
r
giving
δ
s

= (3P
c
)
1/2
ρ
r
From experimental data, they were led to the better estimate
δ
s
= 0.47 P
c
1/2
ρ
r
where δ
s
is in units of (cal/cm
3
)
1/2
, and P
c
is in atmospheres (14). King and
Friedrich showed that the reduced solubility parameter ∆, defined as
∆ = δ
n
/δ
s
is a good indicator of analyte solubility in a supercritical fluid. Solubility improves
as the reduced solubility parameter approaches 1. They correlated solubility data

using solubility parameters for analytes estimated by Fedors’ method (15). One
advantage of the reduced solubility parameter is that it is unitless.
Figure 5.2 relates the density of supercritical CO
2
with its corresponding
Hildebrand solubility parameter based on Giddings’ formula. The solubility para-
Copyright © 2004 AOCS Press
meters for a few common solvents are also included for comparison. Table 5.1
contains the Hildebrand solubility and reduced solubility parameters for a few
common lipid analytes as reported in the literature and estimated by Fedors’
method (16). Fedors’ group contribution method estimates an analyte’s solubility
parameter solely from information about its molecular structure (17). Table 5.2
contains an example of a calculation for estimating the Hildebrand solubility para-
meter by Fedors’ method. It can be incorporated into a spreadsheet for calculating
other fatty acids and their corresponding glycerides.
Even though higher temperatures at a given pressure would lower the density of
the supercritical fluid, the overall extraction performance should be enhanced. First,
supercritical fluids can solvate liquids better than solids. Performing an extraction at a
temperature above the melting point of the analyte should improve the recovery.
Second, temperature affects the solubility parameter of the analytes. The values pre-
sented in Tables 5.1 and 5.2 are for 25°C; higher temperatures tend to reduce the ana-
lyte’s solubility parameter. According to King, a temperature increase of 60°C can
reduce an analyte’s solubility parameter by 1.0–1.5 cal
1/2
/cm
3/2
(18). As a final point,
CO
2
Density

(g/mL)
Solubility Parameter
(cal/cc)
1/2
Fig. 5.2. Carbon dioxide
density vs. Hildebrand
solubility.
Copyright © 2004 AOCS Press
solubility is less of a concern for trace-level analytes, such as pesticides in foods. The
above discussion applies mainly to analytes present at high levels in the sample where
solubility saturation could be an issue. Much lower pressures may actually be suffi-
cient for analytes on the ppb level.
CO
2
Density Calculations
The solvent strength of supercritical CO
2
can be determined from its density as
shown previously. Its density is related to its pressure and temperature. Unfortunately,
the ideal gas law is useless because a supercritical fluid is far from being an ideal gas.
The Van der Waals equation predicts certain qualitative properties of a supercritical
fluid, but it is not quantitatively accurate. A better state equation is required for dense
gases.
TABLE 5.1
Hildebrand Solubility and Reduced Solubility Parameters for Lipid Analytes
a
Hildebrand solubility parameter
Reduced solubility parameter
Reported
Fedors’ method in CO

2
at 80°C and
Analyte MPa
1/2
MPa
1/2
(cal/cm
3
)
1/2
200 atm 400 atm 600 atm
Pentane 14.5 14.5 7.1 0.71 0.99 1.11
Hexane 14.9 14.9 7.3 0.69 0.96 1.08
Heptane 15.3 15.2 7.4 0.68 0.95 1.07
1-Butanol 23.1 23.2 11.3 0.45 0.62 0.70
1-Octanol 20.9 21 10.3 0.49 0.68 0.77
Stearyl alcohol 19.3 9.4 0.54 0.75 0.84
Hexyl acetate 17.3 17.7 8.7 0.58 0.81 0.91
Methyl oleate 17.7 8.6 0.59 0.82 0.92
Stearyl stearate 17.6 8.6 0.59 0.82 0.92
Tripalmitin 18.6 18.3 9.0 0.56 0.78 0.88
Triolein 1 8.5 18.3 8.9 0.57 0.79 0.89
Tristearin 17.9 18.2 8.9 0.57 0.79 0.89
Distearin 19.3 9.5 0.53 0.74 0.83
Monostearin 22 10.8 0.47 0.65 0.73
Glycerol 36.1 40.9 20 0.25 0.35 0.39
Acetic acid 21.4 22.8 11.2 0.45 0.63 0.70
Butyric acid 18.8 21.2 10.3 0.49 0.68 0.77
Palmitic acid 18.8 9.2 0.55 0.77 0.86
Oleic acid 18.7 9.1 0.55 0.77 0.87

Stearic acid 18.7 9.1 0.55 0.77 0.87
Cholesterol 20.7 19.6 9.6 0.53 0.73 0.82
a
Source: References 15,16.
Copyright © 2004 AOCS Press
According to the Law of Corresponding States, two gases with the same
reduced temperature and reduced pressure are in corresponding states. Both gases
should have the same reduced density. A reduced state parameter is the state para-
meter divided by its corresponding critical value. For example, the reduced tem-
perature is the temperature divided by the critical temperature of the gas. Even
though quantitatively inaccurate, the Van der Waal equation predicts the Law of
Corresponding States (19).
For dense gases, the deviations from the ideal gas law can be treated by the
compressibility factor Z which is defined as
Z = M
w
P/(ρRT)
where, M
w
, P, T and ρ are the molecular weight, pressure, temperature, and weight
density of the gas, and R is the ideal gas constant. For ideal gases, Z = 1. The com-
pressibility factor can be expressed as a function of reduced temperature and
reduced pressure: Z(T
r
,P
r
). According to the Law of Corresponding States, gases in
corresponding states should have the same compressibility factor Z value. In the
engineering literature, there are isotherm plots of Z vs. reduced pressure at various
reduced temperatures, which are universal for all gases (20). Using these graphs,

along with knowledge of the critical point for a gas, the weight density of any
dense gas can be determined as follows:
ρ = M
w
P/[RT Z(T/T
c
, P/P
c
)]
Even though the Law of Corresponding States is not exact, it is sufficiently accu-
rate for practical engineering calculations.
TABLE 5.2
Example for Estimating the Hildebrand Solubility Parameter by Fedors’ Method
Diolein: CH
2
–COO–(CH
2
)
7
–HC=CH–(CH
2
)
7
–CH
3
CH–COO–(CH
2
)
7
–HC=CH–(CH

2
)
7
– CH
3
CH
2
–OH
in
i
U
i
(cal/mol) n
i
⋅U
i
V
i
(cm
3
/mol) n
i
⋅V
i
CH
3
2 1125 2250 33.5 67
CH
2
30 1180 35400 16.1 483

CH 1 820 820 –1 –1
HC= 4 1030 4120 13.5 54
OH 1 7120 7120 10 10
COO 2 4300 8600 18 36
COOH 0 6600 0 28.5 0
Σ n
i
⋅U
i
= 58310 Σ n
i
⋅V
i
= 649
δ = (Σ n
i
⋅U
i
/Σ n
i
⋅V
i
)
1/2
9.48 (cal/cm
3
)
1/2
|
|

Copyright © 2004 AOCS Press
Pitzer and colleagues improved upon the Law of Corresponding States by intro-
ducing an acentric factor ω for each gas. The acentric factor is a measure of the devia-
tion of the entropy of vaporization from that of a simple fluid (21). Pitzer's work is
based on a virial expansion treatment of dense gases (22). Using the acentric factor ω,
the compressibility factor Z can be determined more accurately by
Z(T
r
,P
r
) = Z
0
(T
r
,P
r
) + ω Z
1
(T
r
,P
r
)
where Z
0
and Z
1
are obtained from their published tables through linear interpola-
tion. Using this improved compressibility factor Z, a more accurate value for the
density of a gas can be calculated using the formula in the previous paragraph.

Figure 5.3 is a contour plot of the density of CO
2
for various temperatures and
pressures based on Pitzer’s work.
As noted earlier, supercritical CO
2
can be augmented with another modifying
solvent to enhance its solubility for challenging analytes. These binary solvent
mixtures have different critical values and solubility parameters. Their new values
can be calculated using the modified Handinson-Brobst-Thomson equations:
V
b
=
1/4
[(x
s
V
s
+ x
m
V
m
) + 3(x
s
V
s
2/3
+ x
m
V

m
2/3
)(x
s
V
s
1/3
+ x
m
V
m
1/3
)]
T
cb
= [x
s
2
x
s
T
cs
+ 2x
s
x
m
(V
s
V
m

T
cs
T
cm
)
1/2
+ x
m
2
V
m
T
cm
]/V
b
Fig. 5.3. Contour plot of CO
2
density.
Temperature (°C)
Pressure (atm)
Copyright © 2004 AOCS Press
where V
b
, V
s
, and V
m
are the characteristic molar volumes for the binary mixture,
principal solvent, and modifying solvent, respectively; x
s

and x
m
are the mole frac-
tions for the solvent and modifier; and T
cb
, T
cs
, and T
cm
are the critical tempera-
tures of the binary mixture, solvent, and modifier, respectively. The acentric factor
ω
b
and the molecular weight M
wb
of the binary mixture are treated as weighted
averages using the mole fractions as weights
ω
b
= x
s
ω
s
+ x
m
ω
m00
M
wb
= x

s
M
ws
+ x
m
M
wm
where ω
s
and ω
m
are the acentric factors for the solvent and modifier, and M
ws
and M
wm
are the molecular weights of the solvent and modifier. The critical pres-
sure P
cb
of the binary mixture is first calculated by determining the value of the
compressibility factor Z
cb
at the critical point using Pitzer’s tables:
Z
cb
= 0.291 – 0.08ω
b
The critical pressure is then determined using the definition of the compressibility
factor Z:
P
cb

= Z
cb
RT
cb
/V
b
Using these critical values for the binary mixture, the compressibility factor Z can
be calculated from Pitzer’s tables for a given temperature and pressure (23). The den-
sity and Hildebrand solubility parameter can then be calculated for the new mixture.
Table 5.3 lists the critical values for a few binary mixtures. Isco Incorporated offers a
Microsoft Windows–based program that calculates the state parameters and solubility
parameters for CO
2
and a few binary mixtures. As a final note, this theory does not
adequately treat polarity and hydrogen bonding. This treatment is only an approxima-
tion but has been successfully applied to a wide range of solvent systems.
TABLE 5.3
Critical Parameters for Pure Fluids and Mixtures
V T
c
P
c
Fluid (cc/mol) (°C) (atm) ω MW
CO
2
93.8 31.1 72.8 0.239 44.01
Ethanol 175.2 243.1 62.2 0.644 46.07
5 mol% ethanol in CO
2
97.5 42.9 71.9 0.259 44.11

15 mol% ethanol in CO
2
104.9 66.1 70.8 0.299 44.32
Acetone 208 235.1 47.6 0.304 58.08
5 mol% acetone in CO
2
98.8 43 71.3 0.242 44.71
15 mol% acetone in CO
2
109 66.2 69.2 0.249 46.12
Copyright © 2004 AOCS Press
Instrumentation
The instrumentation required to perform a successful SFE is commercially avail-
able. The process begins with a clean source of fluid, which in most cases is a
high-pressure cylinder of CO
2
. A pump is used to increase the pressure of the fluid
above its critical pressure. The working extraction pressure is determined by the
density required to dissolve the target analytes from the sample. The sample is con-
tained in the extraction chamber, which is heated to the desired extraction tempera-
ture above the critical point. The pressurized fluid is brought to temperature by the
chamber and allowed to flow through the sample matrix to extract the analytes.
After the sample, the analyte-laden fluid flows to a restrictor, which controls the
flow rate of the fluid. The restrictor maintains the high pressure of the fluid in the
chamber. At the restrictor, the supercritical fluid loses its solvating strength as its
pressure drops to atmosphere. After the restrictor, the analytes can be collected for
analysis. Figure 5.4 shows a block diagram of a complete SFE system.
Fluid Source
In SFE, an organic extraction solvent is replaced with CO
2

. Just as with organic
solvents, the CO
2
source must be clean enough so as not to contaminate the sample
being analyzed. As with most solvents, high-pressure CO
2
is commercially avail-
able in different grades of purity. For trace-level analyses, such as extracting pesti-
cides at ppb levels, the CO
2
must be at its highest grade of purity. Most gas ven-
dors can supply an SFE-grade that is intended for such applications, generally in
Fig. 5.4. Block diagram of a typical SFE system.
Copyright © 2004 AOCS Press
either stainless steel or aluminium cylinders. The next level of purity is SFC-grade,
which is slightly less pristine, but also less expensive. This grade should be suffi-
cient for extracting fat from food products intended for fatty acid profile (FAME)
analysis. In less demanding applications, such as a simple gravimetric fat determi-
nation, welding-grade CO
2
is satisfactory. In our experience, welding-grade CO
2
has been used successfully in more demanding applications; however, this grade is
not carefully controlled for purity and there may be variation of purity from cylin-
der to cylinder. If in doubt, always include blanks in the analysis to check the puri-
ty of the fluid supply. This is good laboratory practice.
CO
2
is supplied in high-pressure cylinders. The large cylinders generally con-
tain 40 pounds (18 kg) of liquid CO

2
. At room temperature, the head (vapor) pres-
sure is ~900 psi or 60 atm. For the pumps to operate efficiently, liquid CO
2
must
be supplied. Cylinders must be equipped with an eductor or dip tube to deliver the
liquid fraction at the bottom instead of the gas head at the top. SFE-grade cylinders
should automatically be supplied with dip tubes. CO
2
cylinders with helium-head
pressure are also available to improve the performance of HPLC-style reciprocat-
ing piston pumps. The helium-head pressure is usually charged between 130 and
200 atm. However, there are concerns about the helium diluting the solvent
strength of the CO
2
because a significant amount of helium dissolves in liquid
CO
2
. It has been reported that helium-tainted CO
2
can reduce extraction perfor-
mance, but the extraction had to be performed well below optimal conditions to
detect a difference (24). Helium-charged cylinders are not necessary for syringe
pumps or head-cooled reciprocating pumps especially designed for SFE. Most SFE
vendors supply pumps that do not require helium-charged cylinders.
The solvating power of CO
2
can be enhanced by the addition of a small
amount of modifying solvent. The purity of the modifier solvent must be ensured.
For trace analyses, GC-grade solvent is satisfactory, whereas HPLC-grade should

be sufficient for most other applications. If cost is an important issue, then blanks
Fig. 5.5. Block diagram of an online SFE-FTIR system.
Copyright © 2004 AOCS Press
should be included in the analysis to ensure purity. In the past, bottled gas vendors
have supplied CO
2
mix cylinders containing a modifier solvent. There have been
concerns about delivered mixture consistency throughout the life of the cylinder.
Most SFE vendors now supply equipment with special modifier pumps that pre-
cisely meter and mix the modifier solvent with the CO
2
. Modifier pumps should
render CO
2
mix cylinders unnecessary.
Pumps
The vapor pressure of CO
2
at room temperature is below its critical pressure. Even
with cylinders charged with helium, the pressure is still too low for most applica-
tions, and a pump is required to increase and control the pressure of the CO
2
. As
shown earlier, higher pressure produces higher density fluid, which offers greater
solvent strength. The solvent strength of a supercritical fluid can be tuned by its
pressure (see Table 5.1).
Both syringe pumps and reciprocating piston pumps have been used in SFE
applications. Syringe pumps offer smoother pressure control, i.e., no pulsation,
higher flow rates, and more precise volumetric delivery. Syringe pumps, on the
other hand, may have to interrupt an extraction to be refilled if the extraction time

is lengthy. They generally are refilled before the start of an extraction to reduce
such interruptions. Reciprocating pumps are smaller and less expensive than
syringe pumps but can be limited in maximum flow rate. The pump head must be
cooled to prevent vapor lock at higher flow rates. Reciprocating pumps are also
less precise in volumetric delivery due to variable fill efficiency. The controlling
software for a reciprocating pump must be sophisticated enough to determine and
compensate for refill efficiency for each stroke of the piston. Delivered volume
becomes an important issue in modifier solvent addition.
Most SFE vendors offer pumps that can deliver up to 680 atm. In most appli-
cations, 500 atm pressure is sufficient. The pumps must also maintain these pres-
sures with both accuracy and precision at maximum flow rate. Most extractions
can be performed in a reasonable time with pressurized fluid flows of >4 mL/min.
If several extractions are performed in parallel, then this number must be consid-
ered in determining the maximum flow rate required. Performing four extractions
in parallel may demand up to 16 mL/min of 680 atm CO
2
. At this flow rate, the
pump must maintain the programmed extraction pressure in all chambers contain-
ing samples.
Most SFE vendors also offer separate pumps to meter and mix modifier sol-
vents volumetrically. Volumetric control is essential in this application to ensure a
consistent mix. Both the volumes of the CO
2
and modifier solvent must be accu-
rately known to maintain the correct percentage. Modified CO
2
can be delivered by
two syringe pumps, in which one delivers the high pressure CO
2
and the other

delivers the pressurized organic solvent. In this scheme, the two fluids are volu-
metrically metered and mixed at the final pressure. With syringe pumps, the vol-
Copyright © 2004 AOCS Press
umes of both the CO
2
and organic solvent are exactly known at the final pressure.
As a result, there is less concern about the compressibility (degree of volume
change due to a pressure change) of the organic solvent, which can be volumetri-
cally significant at 500 atm. Two reciprocating pumps can also deliver modified
CO
2
. With reciprocating pumps, the two fluids can be easily mixed at the lower
cylinder pressure. This scheme puts lower performance demands on the pump
hardware, thus allowing for a less expensive modifier pump. Unfortunately, the
volumetric metering calculations become more complicated. The pump-controlling
software must be sophisticated enough to compensate for the compressibility of the
CO
2
and the organic solvent at the final SFE pressures. To further complicate the
situation, the main pump delivers a mixture instead of pure CO
2
. This can affect
the refill efficiency of the pump, which in turn affects the total volume delivered.
The software supplied by most SFE vendors takes these factors into consideration.
Finally, during modifier mixing, most pumps report a volume percent (%vol)
instead of a mole fraction (%mol).
Extraction Chamber
The extraction chamber serves three functions. First, it controls the temperature of
the extraction by maintaining the temperature of the fluid and sample. Second, it
contains the high pressure of the supercritical fluid. This function places high

demands on the material and design of the chamber. Third, it contains the sample
to be analyzed and allows the supercritical fluid to flow through the matrix without
allowing the sample to extrude.
Because the critical temperature of most fluids is above room temperature, the
extraction chamber for SFE must be heated. This can be accomplished with electri-
cal heating elements embedded in the heating block surrounding the extraction
chamber. These heating elements are powered and controlled by a proportional
temperature controller. The chamber temperature is usually measured with a ther-
mocouple, which serves as feedback for the proportional controller. The heating
block typically includes heat-exchanging coils, which serve to equilibrate the tem-
perature of the supercritical fluid before it comes in contact with the sample. The
typical temperature range for an SFE system should be between 40 and 150°C with
an accuracy and precision of ± 2°C.
The chamber should also be constructed to withstand the maximum allowable
extraction pressure with a factor of four safety margin. In other words, if the maxi-
mum operating pressure is 680 atm, the chamber should be designed to withstand
2720 atm at the maximum operating temperature (typically 150°C). This margin
ensures that SFE instruments are extremely safe. The chamber should also be
equipped with safety devices such as a rupture disk or a pressure relief valve that
will gracefully relieve the pressure if it should rise above 150% of the maximum
operating pressure. Although obvious, it should be noted that a supercritical fluid
under high pressure is more dangerous than a liquid at the same pressure. The high
Copyright © 2004 AOCS Press
compressibility of gases and supercritical fluids allows for the storage of a danger-
ous amount of mechanical energy at high pressures. On the other hand, the volume
for most liquids does not change much with high pressure, thus little mechanical
energy is stored. Vessels originally designed for HPLC applications may not be
rated safe for SFE applications.
Stainless steels, such as 304 and 316, offer both the strength and chemical
resistance required for most SFE applications. Alloys, such as Nitronics 50, can be

used in more demanding applications. All stainless steel fittings and tubing used in
the plumbing must also be properly rated for high pressure and be resistant to
chemical corrosion. The same is true for valves that control the flow of the fluid.
The cycle life and cleanliness of the packing material are other issues for valves,
especially at the maximum operating temperature. Valves and fittings are commer-
cially available from several vendors (e.g., Valco, SSI, and HPE).
Because the chamber must be opened to remove spent samples and introduce
fresh samples, high-pressure seals are required that are reliable, reusable, chemical-
ly resistant, clean, and durable at the maximum operating temperature. O-rings are
popular for many high-pressure applications; however, the elastomers can cause
problems. Under supercritical conditions, CO
2
penetrates and absorbs into the elas-
tomers. The CO
2
can extract oligomers from the O-ring, which may interfere with
the analysis of the analytes. After depressurization, the absorbed CO
2
is rapidly
liberated, causing the elastomer to suffer from the “bends.” This destroys the O-
ring by reducing it to “popcorn.” Teflon lip seals are a better choice for SFE appli-
cations because they are reliable, reusable, and durable at elevated temperatures. In
addition, Teflon is chemically resistant and clean, and does not suffer from the
“bends.” Metal washer seals, made from copper or gold, are appropriate for more
permanent connections.
Finally, the chamber should allow the supercritical fluid to flow easily through
the sample matrix without extruding the sample. Assuming that the sample has
been reduced to a powdered form, this can be accomplished with the aid of frits.
Frits, like filter paper, are porous enough to allow the fluid to flow through but pre-
vent the sample from penetrating. The sample vessel should have frits on both the

inlet and outlet sides. An inlet frit would prevent the sample from flowing back
into the pump in the event of pump failure. Check valves can also prevent back
flow.
In the early days of analytical SFE, empty HPLC column vessels were used as
extraction vessels. Even though readily available in a typical laboratory, they may
not be the safest approach because they are designed for pressurized liquids instead
of gases as noted above. They are also not convenient because tools are required to
open and close these vessels during sample introduction.
Thar, Inc. is a vendor of high-pressure vessels for SFE applications. Their ves-
sels are designed for pressurized gases, making them safer to use for SFE. They
have also been designed to be opened and closed without tools, making them more
convenient to use. Isco, Inc. has a patented design that allows the sample cartridges
Copyright © 2004 AOCS Press
to be made from light-weight polymeric material or aluminum instead of stainless
steel. The light weight aids in sample weighing, and these cartridges can be easily
opened and closed by hand. Their polymer cartridges are durable enough to be
reused and are chemically clean and resistant. The polymer was specially selected
not to absorb CO
2
; therefore, contamination and the “bends” are less likely. The
Isco design allows for light-weight cartridges because the pressure of the supercrit-
ical fluid is applied to both the inside and outside of the cartridge during extrac-
tion. The sample cartridge is inserted into a stainless steel chamber, which is also
easy to open and close at hand (25).
Restrictor
The restrictor controls the flow of the supercritical fluid after passing through the
sample. The restriction to the flow maintains the high pressure in the extraction
chamber. It is the most technologically demanding component of an SFE system.
Even though it can be as simple as a length of fused silica capillary, this simplicity
can be a source of frustration. Silica capillaries are inexpensive, but they also break

or plug easily, which can ruin an extraction and waste the operator's time.
At the restrictor, the CO
2
expands rapidly to atmospheric pressure. As in any
adiabatic expansion, a large amount of heat is absorbed to increase entropy. This
expansion produces very low temperatures; without external heating, extracted
water will freeze and plug a capillary. Water is frequently present in food samples.
In addition, the supercritical fluid rapidly loses its solvent strength during expan-
sion, allowing analytes to precipitate out of solution, which may plug a simple cap-
illary. Heating can help to prevent such problems.
Isco offers a patent restrictor, which is based on a stainless steel capillary. The
stainless steel is less brittle than fused silica and is not easily attacked by modifiers
such as methanol or water. The restrictor is heated from end to end. A controlled
electric current is applied along the entire length of the capillary, so that the capil-
lary serves as an electrical heating element. Because the electrical resistance of the
capillary is dependent on temperature, its temperature can be determined from
measuring its ohmic resistance. This measurement can be used as feedback to pro-
portionally control the temperature of the restrictor. The stainless steel capillary
serves to control the flow of the fluid, to heat the fluid, and to measure the temper-
ature of the restrictor (26). Most SFE vendors offer some form of heated fixed flow
restrictor.
Isco and most other SFE vendors also offer heated variable flow restrictors (27).
Most variable restrictors are based on a needle valve design. To increase the flow, the
needle is withdrawn from the seat, thus opening the orifice. The valve body can be
heated to reduce the cooling effects from the adiabatic expansion. By incorporating a
motor drive to the valve needle, the flow rate of the fluid can be automatically adjust-
ed during an extraction. Unfortunately, the most sophisticated restrictor can still occa-
sionally plug, especially during method development on new samples.
Copyright © 2004 AOCS Press
Extract Collection

The final step in SFE is to transfer the target analyte from the CO
2
to another
medium for further analysis. There are several options available depending on the
analysis. The simplest collection scheme involves trapping the analyte on glass
wool. This is particularly useful for gravimetric fat analysis of food products. In a
gravimetric analysis, the collection vial is weighed before and after the extraction
to determine the weight of the extracted analyte. Fats are generally viscous and
nonvolatile; thus they are easily trapped by the wool in the collection vial. Any
extracted water or modifier solvent (e.g., ethanol), that could interfere with the
final weighing, is poorly trapped by the glass wool. After the extraction, the collec-
tion vial can be dried in a microwave or vacuum oven to remove any residual
water or modifier solvent. This simple scheme has the added elegance of being
completely solvent free, assuming that no modifier was used during the extraction.
For analysis methods based on GC or HPLC, it may be convenient to trap the
analytes in a liquid solvent. Because SFE is selective and offers cleaner extracts,
the collected solution can generally be injected directly into a GC or HPLC after
including an appropriate internal standard. The collected solution can be diluted to
volume or evaporated to dryness under a nitrogen stream in a heated water bath to
isolate the neat analytes. The collected solution may have to be derivatized by
saponification or transesterification as in a FAME analysis before proceeding to
the GC or HPLC. To enhance trapping performance, the collection solvent can be
cooled or maintained under a mild back pressure. A back pressure of 30 psig
reduces the size of the expanding CO
2
bubbles, which enhances analyte transport
from the CO
2
to the liquid solvent. It also reduces the evaporation of the solvent
during the extraction or solvent aerosol formation (28). Certain solvents, such as

acetone, can also dissolve a significant amount of CO
2
during extraction. After the
extraction, the collection solution may have to be sonicated gently to remove any
dissolved CO
2
. Finally, caution must be exercised with makeshift collection sys-
tems because pressurized glass vials can explode.
The extracted analytes can also be trapped on solid-phase cartridges. The
expanding CO
2
is allowed to pass through a C18 silica bed. As the CO
2
loses its
solvent strength, the analytes transfer onto the C18 beads. After the extraction, the
target analytes can be washed off the beads with a few milliliters of an appropriate
solvent. Most C18 solid-phase cartridges have to be preconditioned before the
extraction by rinsing with a few milliliters of methanol. The trapping performance
of the solid-phase cartridge can be enhanced by cooling the cartridge. After the
analyte is washed off the cartridge, the collection solution can be treated as dis-
cussed above. Most SFE vendors offer a range of collection options.
Online Techniques
SFE is quite amenable to direct coupling with other analytical instruments, such as
GC or FTIR. These online techniques offer higher sensitivity and eliminate the
Copyright © 2004 AOCS Press
intermediate solvent collection step. CO
2
could be the only solvent used in the
analysis.
In online SFE-GC, the target analytes after the restrictor are allowed to deposit

at the front of the separation column. During the extraction, the column tempera-
ture is kept low so that the analytes focus on the front of the column. After the
extraction, the normal column temperature program is executed to elute and sepa-
rate the analytes on the column (29).
In online SFE-FTIR, the target analytes, while still in solution in the supercritical
fluid, are analyzed directly by a FTIR spectrometer (30). After the extraction chamber,
the analyte-laden fluid is transferred immediately to a heated, high-pressure IR flow
cell through a heated transfer line. The temperatures of the extraction chamber, transfer
line, flow cell, and restrictor should be the same. Some analytes, such as waxes, can
precipitate out of solution at lower temperatures. In this application, zinc sulfide is an
appropriate IR window material. It is transparent to infrared light and strong enough to
contain the high pressure. The high pressure flow cell is mounted in the optical bench
of the FTIR spectrometer so that the spectra of the target analytes can be monitored
during the extraction. After the flow cell, the supercritical fluid is allowed to flow to a
restrictor where the target analytes can be collected for additional analysis (Fig. 5.5).
FTIR is a nondestructive analysis method. CO
2
is an excellent IR solvent
because it does not contain hydrogen atoms. It is transparent in the C–H stretch
band (3000 cm
–1
); thus, it can substitute for CS
2
, freon, and perchlorinated sol-
vents in many IR applications. Finally, the spectrum is collected real time during
the extraction, making this technique useful in extraction kinetic studies (31,32).
Applications of Supercritical Fluid Extraction
This review is intended to be a limited summary of applications and research
involving methods development, verification, and validation using the technology
of SFE in the agricultural and food industries. Early leaders in research had the

challenge of determining the physical properties of supercritical fluids, properties
both chemical and mechanical in nature. Leading research came from chemical
engineers and scientists who were interested in achieving the full utility of a chem-
istry that has the advantage of being environmentally safe.
SFE is generally used as an alternative primary method for Soxhlet type
extractions. For fats and oils, quantification is generally performed gravimetrically.
As with any gravimetric technique, the question of extract composition was
addressed. For all other SFE from food matrices, postextraction quantification
includes a variety of accepted techniques such as gas chromatography with fatty
acid methyl esterification (GC-FAME), gas chromatography/mass spectroscopy
(GC/MS), thin-layer chromatography (TLC), supercritical fluid chromatography
(SFC), and high-performance liquid chromatography (HPLC).
Method development for any analytical method requires that critical control
points be determined. For SFE applications, these parameters may include sample
Copyright © 2004 AOCS Press
preparation, instrument method parameters, and trapping techniques. Once a robust
SFE method has been proposed, method validation or verification through the use
of informal and/or formal collaborative studies is performed. Methods using SFE
are currently approved by the USEPA, AOAC, and AOCS.
Supercritical Fluid Extraction Followed by Gravimetric Determination
There are numerous areas that require the development of new analytical methods,
including the reduction of solvents, improved recoveries, faster analytical results, or
scale up to industrial processes. The development of SFE as an alternative to solvent-
solvent extraction for the recovery of supercritical carbon dioxide (SCCO
2
)-soluble
lipids (triglycerides, cholesterol, waxes, and free fatty acids) merited attention from
numerous researchers. Applications such as extracting oils from Antarctic krill
attempted to determine the potential of SFE as an industrial process used SCCO
2

at
40–80°C and 245–392 bar to extract valuable oils that are high in unsaturated fatty
acids and phospholipids. The oil extracts, using traditional extraction techniques,
required the removal of extraction solvents (33).
Supercritical CO
2
extraction followed by characterization of cocoa extracts for
the purpose of determining triglycerides, fatty acid composition, unsaponifiable mat-
ter, and aromatics is another broad application. Investigation of SF extracts from
cocoa liquor, nibs, and shells under a variety of SFE conditions (50–80°C and
300–400 bar) was investigated for the purpose of determining industrial applicability
of SFE in cocoa processing. Total fat is determined gravimetrically, whereas the com-
position is determined using fatty acid methyl esterification (FAME) followed by GC,
with the aromatic fraction analyzed by HPLC for pyrazines. This application showed
that the extract obtained from cocoa liquor met the definition of cocoa butter; howev-
er, the portion of triglycerides extracted from the shell contained unsaponifiables that
exceeded the limit allowed. Additionally, it was determined that SFE maintained an
even distribution of aromatics between the cocoa butter and the cocoa powder, thus
maintaining the aromatic properties of the products (34).
The most popular SFE solvent is CO
2
, and triglycerides, cholesterol, waxes,
and free fatty acids are quite soluble in supercritical CO
2
. The solubility of polar
lipids, such as phospholipids, can be improved by augmenting the supercritical
CO
2
by the addition of a small amount of ethanol or other polar modifier solvents.
Phospholipid (PL) content is useful for the investigation of storage conditions of

processed foods (35), whereas lecithin, a mixture of PL and vegetable oil, is a by-
product of the edible oils industry and commercially important for use in the food
and cosmetic industries. Process scale supercritical carbon dioxide extraction of
oilseeds leaves behind commercially valuable lecithin in the meal. However,
extracting with CO
2
first to remove the oil followed by SFE with ethanol modifier
was shown to remove most of the valuable components from oilseeds (36,37).
Some analytical results require PL inclusion as part of the total fat or oil reported
and therefore require the use of a polar modifier to meet the requirement.
Copyright © 2004 AOCS Press
Solubility is one required parameter for successful extraction of analytes inde-
pendent of the extraction method used. Another is the accessibility of the solvent to
the analyte. Some lipids may be bound to proteins. In a study using neat SCCO
2
,
Boselli and Caboni (38) successfully extracted total lipids from dried egg yolk.
These researchers altered the density of the SCCO
2
to >1 g/mL and employed sev-
eral depressurization steps during the extraction process. They suggested that the
freezing-thawing mechanism that occurred in the extraction vessel resulted in the
modification of the protein structure, thus leading to the destruction of lipid-protein
aggregates. They theorized that the PL become readily available to the SCCO
2
, and
thus more extractable. Postextraction analysis and identification of lipid classes
included HPLC-ELSD and TLC (38).
Other lipids may be contained within cell wall components such as cellulose
and not be accessible to SCCO

2
. Studies have shown that the SFE may cause
chemical alterations when sufficient quantities of oil are removed. This alteration
allows subsequent extractions to remove additional oil (39). During sample prepa-
ration for oilseeds, such as grinding, the surface-to-volume ratio is increased, thus
exposing more oil to the extraction solvent. During the process, seeds are uninten-
tionally milled to different particle sizes and the amount of oil extracted from each
fraction (based on size) produces different results (40,41). These studies all indi-
cate that proper sample preparation before SFE has an effect on the quantity of
analyte recovered.
In 1996, the method using SFE was approved by the AOCS. Method Am3-96,
SFE Determination of Oil in Oilseeds, was subsequently adopted by AOAC in
2000 as method number 999.02. The method is based on gravimetric analysis from
a set of oilseeds determined to be representative of the oilseed industry and encom-
passes soybeans, canola, sunflower, safflower, and sunflower. This method refer-
ences AOCS sample preparation methods as determined for each type of oilseed
and allows the user to choose between two variations of SFE (CO
2
alone or CO
2
with a 15% EtOH modifier). The SFE parameters are 100°C, 7500 psi, with or
without a 15% modifier at 2 mL/min with a total extraction time of 30 min (CO
2
only) or 45 min (CO
2
+ 15% EtOH) (42).
The application of SFE for the determination of total fat in meats was investi-
gated by numerous researchers in response to the Nutrition and Labeling Education
Act (NLEA) of 1990, which defined fat as the sum of all fatty acids obtained from
a total lipid extract expressed as triglycerides (43). Snyder et al. (44) investigated

the possibility of determining fat content for the purpose of nutritional labeling of
meats and meat products using SFE (50°C, 12.2 MPa with <1% MeOH continuous
modifier addition) with a unique, on-line lipase catalyzed reaction for GC analysis.
Eller and King (45) continued this research by comparing the SFE-GC-FAME
method with a standard acid hydrolysis (AH) organic extraction/GC-FAME
method. The sample was mixed with diatomaceous earth (1:1.5, w/w), placed into
the extraction vessel, and 1 mL of 100% EtOH was added. SFE was performed at
100°C and 9000 psi for 25 min after an initial 5-min static hold. The results
Copyright © 2004 AOCS Press
showed that the two methods (for meats) were statistically equivalent, indicating
that SFE quantitatively removed and accurately determined the fat from ground
beef samples.
Supercritical fluid extractions of six meat samples and a canned meat refer-
ence material SMRI 94-1 (Swedish Meat Research Institute, Kävlinge, Sweden)
were compared with the Bligh and Dyer (B&D) method (46). Both techniques
were followed by TLC to determine the lipid classes contained in the extracts. The
samples for SFE were mixed in a 1:2 ratio (sample:diatomaceous earth), placed in
the extraction vessel followed by the addition of 1 mL cyclohexane, and extracted
by SFE at 50°C, 370 bar, 8% EtOH modifier addition, at a flow rate of 4.0 mL/min
for 30 min. SFE produced results that showed no statistical difference from the
B&D method in the respective lipid classes (47).
Supercritical fluid extraction has been compared with the Schmid, Bondzynski
and Ratzlaff (SBR) method (48) using gravimetric analysis to determine total fat.
The fatty acid profiles were determined by GC after hydrolysis and methylation of
the extract. Twenty-seven meat and meat products ranging in fat content from
2–37% were assayed using two commercially available SFE instruments. The sam-
ples were mixed with diatomaceous earth (1:1.5, w/w), placed in the extraction
vessel, and 1 mL EtOH was added. The samples were extracted at 100°C, 9000 psi
at a flow rate of 2 mL/min for 25 min after a 5-min static step. The SFE method
correlation with SBR showed that there was no significant difference in the gravi-

metric results (P = 0.246) using a paired t-test. Additionally it was shown that fatty
acid composition can be determined by GC analysis (after hydrolysis and derivati-
zation) by using an aliquot of the same extract (49).
The application of gravimetric SFE as a direct method for the determination of
crude fat (defined as the components of meat that are extractable with petroleum
ether, without digestion of the sample) from meat resulted in an AOAC peer-verified
method (PVM) 3:2000. The study used nine meat matrices including raw ground beef,
raw pork sausage, raw veal sausage, smoked ham, kielbasa, braunschweiger, sweet
Italian sausage, smoked sausage, and bologna with fat ranging from 6–28% and
involved two peer laboratories. The results were compared with AOAC Method
960.39. Samples for SFE were prepared by mixing 1.0–1.5 g of sample with 2.2 g of
diatomaceous earth. The mixture was transferred to the extraction vessel and extracted
at 100°C, 9000 psi for 45 min. To remove any subsequent co-extracted moisture, the
collection vessels were dried after extraction and before reweighing. This method has
a mean accuracy of +0.22 to –1.41 and mean repeatability and reproducibility of <3.0.
The data suggest that the method should perform well for all meats with fat content in
the range of 6–28% (50).
Infant formula is intended to act as a substitute for human milk. Therefore,
due to health concerns and NLEA labeling requirements, the composition and
quantity of fat in infant formula must be disclosed. Traditional methods for the
analysis of milk-type products include solvent extraction techniques that require
sample pretreatment such as acid or base hydrolysis, e.g., AOAC Method 905.02
Copyright © 2004 AOCS Press
(Roese-Gotttlieb). The application of SFE for milk products that employ gravimetric
analysis compared with accepted gravimetric methods were reported (51–53).
Further investigation and direct correlation between traditional hydrolysis/solvent
extraction techniques and SFE was performed using gravimetric and GC/MS
analysis. Method development utilized a powdered infant formula standard refer-
ence material NIST SRM 1846 (the National Institute of Science and Technology,
Gaithersburg, MD) and nine commercially available infant formulas. Sample

preparation for powdered samples entailed mixing 2.0 g of sample with 4.0 g of
deionized water followed by sonciation for 10 min to homogenize the sample. A
portion of the mixture was mixed with diatomaceous earth (1:3, w/w) and trans-
ferred to the extraction vessel. The mixture was allowed to stand at room tempera-
ture for 10 min after which a 1-mL spike of MeOH was added to the vessel.
Supercritical fluid extraction was performed at 100°C, 465 atm, 15% ethanol
(EtOH) modifier addition with a total flow rate of 2 mL/min for 25 min after an
initial 10-min static hold.
Determination of total fat using gravimetric analysis revealed that both meth-
ods reported results that had strong agreement for the SRM and for the nine sam-
ples. However, when the extracts were analyzed for total fat as determined by
GC/MS, the data showed that although the profiles were similar, the gravimetric
results for both methods were significantly higher suggesting that extraneous, non-
fat materials were being extracted from the samples (54).
The application of SFE for infant formula was accepted as AOAC PVM
2:2002 for the determination of total fat in milk and soy-based infant formula pow-
der using SFE. These researchers developed the method to meet the demand for a
rapid, accurate, high-volume gravimetric determination of total fat content of
infant formula powders. This PVM utilized a Data Quality Objectives (DQO)
approach (U.S. EPA and the 1994 Guidance for the Data Objective Process;
Bulletin EPA/600R-96/055). The study used a NIST standard reference material
(SRM 1846) and commercial milk- and soy-based infant formula powders (8 sam-
ples total) and involved 2 collaborating laboratories. Variables identified were
extraction time, flow rate, and sample preparation. The optimized SFE parameters
were 100°C, 9000 psi, 3.0–3.5 mL/min flow rate, and the following extraction
sequence: 15 min hold time followed by 35 min dynamic extraction. Sample
preparation included mixing the sample with diatomaceous earth in a ratio of 1:2
(w/w) followed by the addition of 1 mL of 50% (vol/vol) 2-propanol. The
researchers concluded that the SFE method performance characteristics were quan-
titative and that the experimentally determined values addressed the issues of

assessing the suitability of the analytical method for any given purpose. They fur-
ther suggest that this technique could be applied to other food matrices (55).
The application of SFE for the determination of fat and oil in food and agricul-
ture products is by no means limited to the matrices discussed in this review.
However, there is a growing body of validated and/or approved methods utilizing
SFE as the method of choice for fat and oil determination.
Copyright © 2004 AOCS Press
Fat-Soluble Vitamins and Other Nutritional Components in Foods
Vitamins are essential to the health and nutrition of all members of the animal king-
dom. Animals cannot synthesize vitamins and instead require dietary intake to meet
the nutritional demands for development, growth, and maintenance during their life
cycle. Vitamins are classified on the basis of their solubility in nonpolar organic sol-
vents (fat-soluble vitamins) and polar solvents (water-soluble vitamins) (56). Lipid-
soluble vitamins are the topic of this section and include A, D, E, and K.
Vitamin A (retinol) in its various forms functions as a hormone and is of
importance in protein metabolism of cells that develop from the ectoderm. Retinol
in the form of 11-cis-retinal (II) is the chromophore component of the visual cycle
chromoproteins in the cone and rod cells of the retina. Deficiency of this compo-
nent produces a variety of conditions such as dryness and thickening of the skin,
retarded development and growth, and night blindness.
Vitamin D (calciferol) is comprised of two compounds, Vitamin D
2
and D
3
.
Although both are absorbed from the diet, D
3
(cholecalciferol) is also synthesized
biologically from cholesterol in the skin by exposure to UV radiation. Similarly,
vitamin D

2
(ergocalciferol, II) is formed from ergosterol. Vitamin D
3
metabolites
are of great importance in several bone disorders including childhood rickets,
osteoporosis, and osteomalacia.
Vitamin E is the collective name for a group of closely related lipids that are
biological antioxidants called tocopherols. The various tocopherols differ in the
number and position of the methyl groups on the ring and retard or prevent lipid
oxidation, but also stabilize other active agents such as vitamin A, ubiquinone, hor-
mones, and enzymes. The mechanism of action is not fully understood, and
Vitamin E deficiency in humans is very rare; however, the principal symptom is
fragile erythrocytes and the slowing of aggregation of blood platelets.
The K-group vitamins (phytomenadione) are naphthoquinone derivatives that
differ in their side chains and are an essential nutrient for humans and animals.
Because vitamin K
1
is involved in biosyntheses of the blood clotting factors such
as prothrombin and proconvertin, it is required for the function of blood clotting
and deficiency results in hemorrhage platelets.
Although vitamin deficiency has a negative effect, overconsumption of vita-
mins can also be harmful (57–60), which has led to strict regulations for the use of
vitamins. These regulations require the need for continuous analysis to determine
the levels of fat-soluble vitamins in foods and pharmaceutical preparations.
The sensitivity of fat-soluble vitamins to extreme pH, oxygen, light, and heat
presents challenges in the techniques of extraction and analysis. Conventional
methods for isolation of fat-soluble vitamins depend on the sample matrix and usu-
ally involve solvent extraction with organic solvents such as methanol, ethanol,
acetone, tetrahydrofuran, hexane, or petroleum ether (61). Finished human food
products and animal feeds usually require a step that saponifies triacylglycerols to

glycerol and soaps of the free fatty acids. Postextraction analysis includes normal
Copyright © 2004 AOCS Press
or reversed-phase liquid chromatography (LC) with UV-vis or fluorescence detec-
tion, GC with flame ionization detection (FID), or MS.
Traditional extraction methods expose vitamins to potentially degradating
environments and require the use of large volumes of organic solvents. Supercritical
fluid extraction is an alternative to the use of organic solvents in that it minimizes
the risk of exposure to oxygen and heat, while having the added benefit of using car-
bon dioxide as a replacement for organic solvents. Although carbon dioxide is the
most commonly used extraction solvent due to its low critical parameters, relative
abundance, and low toxicity, it is relatively nonpolar. However, most commercial
SFE instrumentation offers the ability to add a polar modifier to enhance the
extraction capabilities of SCCO
2
.
Turner et al. (62) provided an excellent review of the literature for the supercriti-
cal extraction of vitamins. They presented the fundamental research for the extraction
of fat-soluble vitamins from matrices that included milk powder, infant formula, dairy
products, liver, processed meats, meat products, cereal products, oilseeds, sweet pota-
toes, carrots, tomato paste, broccoli, collard greens, corn, and zucchini followed by
postextraction analysis using supercritical fluid chromatography (SFC).
Mathiasson et al. (63) went on to complete a collaborative study that demon-
strated the ability of SFE using CO
2
as a replacement technology for liquid extraction
solvent methods for the determination of vitamin A, E, and β-carotene from
processed foods. The validation included 10 laboratories, and the data suggested
that as with any analytical method, additional information was required to deter-
mine control points and that further validation was required. However, these
researchers showed that SFE is a suitable replacement technology for the solvent

extraction of fat-soluble vitamins.
Supercritical Fluid Extraction of Pesticides from Foods
Pesticides include both natural and man-made chemicals used to control weeds,
insects, fungi, nematodes, and rodents. Pesticides have been widely used in agri-
cultural applications to increase crop yields, and the pressure to produce food crops
has magnified their use in Third World countries. Although the selective use of
pesticides to control nuisances has played a positive role in increasing food pro-
duction, global monitoring of pesticide build-up through the food chain is under
constant surveillance by local and national regulatory agencies.
Traditional solvent extractions use large volumes of environmentally hazardous
organic solvents. SFE is an alternative to solvent extraction methods, and its methods
are approved by the United States Environmental Protection Agency (USEPA) and
AOAC. USEPA methods 3560, 3561, and 3562 use supercritical CO
2
for the extrac-
tion of pesticides from environmental matrices such as soil, sediment, and sludge, and
AOAC 2002.03 is a multiple-residue extraction method in nonfatty foods (64,65).
In foods, there are more than 750 pesticides, herbicides, and related compounds
that require analytical detection, with the majority analyzed by GC/MS (66). Food
Copyright © 2004 AOCS Press