SUPERCRITICAL CO2
Extraction using supercritical carbon dioxide has been an established industrial-
scale technique for many years. High-pressure CO2extraction is already widely
used, for example, for dealcoholization; decaffeination of coffee and tea; pro- cessing of tobacco, hops, spices, and fats and oils from both vegetable and animal sources; and also to extract specific compounds or active ingredients for the food, beverage, and tobacco sector and in the chemical and pharmaceutical industries, as well as in the fields of cosmetics, leather, textile, paints, and beverages.
In recent years, the elegant and gentle high-pressure extraction method us- ing supercritical gases has also been introduced in many research laboratories as a technique for isolating and analytically investigating flavor and fragrance chemicals from natural products. The resulting flood of scientific publications is beyond the scope of detailed discussion in this chapter. More detailed information
and theoretical data about extraction with compressed CO2, may be found in
recent review articles and several monographs (72–86).
High-pressure extraction with compressed gases is an effective process for separating flavor and fragrance chemicals from complex matrices without the use of organic solvents. The principle of high-pressure extraction is based on the well-known fact that in the supercritical region, gases can acquire solvent proper- ties that are superior to those of liquid solvents.
In principle, a variety of gases that can be rendered supercritical in terms of temperature and pressure could be used as extraction media. In practice, super-
critical CO2 has proven optimal for the extractive processing of natural sub-
stances. Carbon dioxide—a ubiquitous natural material—is physiologically harmless and inexpensive; is available in high purity; is chemically stable and inert, nonflammable, and noncorrosive; and has a low critical point. Because of
its low boiling point, CO2is easy to remove from both product residues and the
target extracts. An important feature regarding solubility is the polarity of CO2
under supercritical conditions: CO2possesses almost exclusively lipophilic sol-
vent properties, and falls between diethyl ether and methylene chloride in terms of polarity. It is a highly selective solvent whose dissolving power can be con- trolled by modifying the temperature and pressure of the extraction process.
When pressure is applied to gaseous CO2, the result (depending on the
present system temperature) is either liquid or solid carbon dioxide. If a tempera- ture above the critical temperature is used, however, the gas cannot be liquefied regardless of how much pressure is applied. The p(T) phase diagram for carbon dioxide is shown inFig. 15.Two points on this diagram are of particular interest: The triple point Tp, at which all three phases are present together in equilibrium,
and the critical point Kp, which separates the liquid and gaseous phases. This
FIGURE15 p(T) phase diagram for carbon dioxide.
above the critical point—is referred to as ‘‘supercritical’’ or ‘‘fluid.’’ Supercriti- cal or fluid carbon dioxide is considered to exist at pressures above 7.3 MPa and temperatures above 31.2°C. Supercritical gases have excellent dissolving proper- ties for many classes of chemical substances. The solubility of supercritical CO2,
for example, is two to three times better than that of subcritical liquid CO2. The
advantage of compressed gases over organic solvents is that although their den- sity is very similar to that of liquids, they have a dynamic viscosity that approxi- mates that of a normal gas. This low viscosity results in rapid mass transport behavior (high diffusion capability), which allows the supercritical medium to penetrate particularly easily into the solid material being extracted and dissolve constituents from it.
With conventional solvents, the solubility of substances can be influenced primarily by way of temperature; normally, it rises with increasing temperature. The solubility of organic compounds in supercritical fluids, however, is a function of temperature and pressure, and can be controlled very easily by modifying the temperature and pressure. The dissolving power of supercritical fluids generally rises with increasing pressure (i.e., increasing density). The extraction rate for natural substances often also depends not only on pressure and temperature but also on temperature-related diffusion, within the natural matrix, of the constit- uents being extracted. If the gas density is reduced by modifying the process parameters (pressure and temperature), the dissolving power decreases so that dissolved substances are completely or partly separated again. This effect is uti- lized in the high-pressure extraction separation process. The gas circulates through the system, which is divided into higher-pressure and lower-pressure
regions. In the loading step (at high gas density), specific substances are dissolved from the natural matrix. In the demixing step (at lower gas density), the sub- stances are separated and removed from the circulation loop. High-pressure ex-
traction with supercritical CO2 avoids elevated temperatures that can lead to
thermal decomposition or rearrangement of labile constituents. Because high- pressure extraction also avoids any oxidative processes, the method yields high- quality genuine extracts that are similar to the natural product, retaining almost the entire native chemical composition of the ingredients. In the field of flavors
and essential oils, extraction with supercritical CO2 therefore offers extremely
attractive parameters. Under conditions of low thermal stress, the method can produce concentrates that are usually far superior to conventionally produced
extracts. By adding modifying agents (such as acetone) to supercritical CO2, it
is possible to vary the dissolving power of the fluid phase over a broad range, allowing the extraction profile to be modified or extraction yields to be signifi- cantly enhanced, for example.
The principle and procedure of high-pressure extraction with supercritical gases is illustrated in Fig. 16. The extractor contains the solid natural material
being processed. Once the entire system has been flooded with CO2, the pre-
selected supercritical pressure and temperature conditions are established in the extraction medium using a pump or compressor and a heat exchanger. The gas, loaded with extracted compounds, is then transported into the separation vessel where it undergoes changes in pressure and/or temperature, thus demixing and removing the constituents dissolved in the fluid phase. After separation of the water that is also extracted (since water becomes increasingly soluble in supercrit-
FIGURE16 Schematic layout of a high-pressure system for extracting solids with super- critical gases.
ical CO2as the temperature is raised), the method yields flavor extracts with very
good organoleptic properties that can then be analyzed by gas chromatography. The extract-free gas is drawn off, brought back to the supercritical state, and returned to the extraction vessel. The process is continuous in terms of the sol- vent, but discontinuous (batch process) in terms of the solid starting material that is used, because the introduction of solids into the high-pressure area is problem- atic.
Liquid products can, of course, also be processed with this extraction tech- nique; in this case the liquid raw material is fed into a distillation column and slowly falls to the bottom in countercurrent against the rising supercritical gas. At the bottom of the column, the residue is discharged from the sump by a level regulation system, and the extracted components are once again precipitated in
the separator. With this procedure, the extractor inFig. 16must be replaced by
a countercurrent extraction column. This process can be run continuously because liquid can be metered into the high-pressure area without difficulty.
SFE has already been used for a number of years for the extraction of food constituents. There are numerous articles that deal with the advantages, applica- tions, and possibilities of SFE in flavor analysis (87–94).
An extraction procedure performed on kiwi fruits and leek with supercriti- cal carbon dioxide in our pilot plant (SITEC-Sieber Engineering AG, CH-Maur/ Zu¨rich) at 9.0 MPa and 40°C (gas density approximately 0.5 g/cm3
) yielded flavor concentrates that were judged, in an organoleptic evaluation, to be of much better quality than extracts produced with conventional methods. The capillary gas chro-
matograms for both total flavor extracts are shown inFigs. 17and 18.Figure 18
shows that numerous sulfur-containing flavor compounds are particularly sig- nificant in the leek flavor concentrate, contributing to the typical and characteris- tic overall flavor impression. Extraction using compressed carbon dioxide makes it possible to recover fruit and vegetable flavor compounds or spice ingredients under mild process conditions at good yield and with outstanding quality.
On the other hand, extraction with supercritical CO2is also a powerful tool
in essential oil and fragrance research, especially because the isolation of volatile components from flowers using traditional methods such as steam distillation or dynamic headspace sampling produce fragrance extracts that do not reflect the organoleptic properties of the natural material. Due to the mild and gentle extrac- tion conditions, supercritical fluid extraction is most suitable to isolate essential oils from spices, flowers, blossoms, herbs, leaves, seeds, and roots (95–102). These extracts are generally considered to be organoleptically superior to concen- trates produced by traditional techniques. Because most of the essential oils from such natural products are susceptible to thermal degradation, SFE is especially advantageous for extracting critical and thermally labile fragrance components. In our laboratory, for example, extraction of roses have been investigated as follows: 150 g of rose leaves was filled into the extractor and extracted for 2
FIGURE17 Gas chromatographic separation of a carbon dioxide kiwi-fruit flavor concen-
trate.
hours with a CO2flow rate of 20 kg/h at 9 Mpa and at a temperature of 40°C.
The flavor and fragrance of the carbon dioxide rose extract was very close in quality to actual rose leaves compared with rose extracts isolated by conventional
techniques. A CO2extract was obtained that quite faithfully reflected nature be-
cause hydrolysis, oxidation, or thermal changes are completely avoided. The FID chromatogram of the supercritical carbon dioxide rose extract is illustrated in Fig. 19. GC/MS analysis has been used to identify components responsible for rose fragrance. The major compounds were citronellol, geraniol, and nerol. More detailed analytical results are given in Fig. 19. Furthermore, it
is worth mentioning thatβ-damascenone could not be detected in the CO2sample.
According to Surburg et al., this is due to the extremely mild extraction condi- tions, thus minimizing the formation of artifacts (103). A certain disadvantage
of the CO2 extraction, however, is the presence of many compounds that do
not contribute to the rose fragrance impression, such as paraffins or long-chain paraffinic alcohols.
In conclusion, supercritical CO2offers considerable advantages as extrac-
FIGURE19 Capillary GC-FID pattern of rose petals extracted with supercritical CO2.
Key to compounds identified: 1, citronellyl acetate; 2, geranyl acetate; 3, citronellol; 4, nerol; 5, 3,5-dimethoxytoluene; 6, geraniol; 7, nonadecane; 8, (Z)-9-nonadecene; 9, 7,8- dihydro-β-ionol; 10, eugenol.
tion solvent for food and plant samples. It is apparent that high-pressure extrac- tion with compressed gases expands the scope of conventional extraction pro- cesses. It combines the principles of the two conventional separating methods (extraction and distillation), but exhibits none of their disadvantages.
High-pressure extraction—also called ‘‘destraction’’—using natural car- bon dioxide thus constitutes a true alternative to conventional extraction tech-
niques using organic solvents. The CO2 extracts consist entirely of the native
constituents and represent the complete spectrum of volatile compounds present in the natural product.
IV. SOLVENT-ASSISTED FLAVOR EVAPORATION (SAFE)
Through the years, analytical chemists have made great efforts to develop effi- cient, reliable methods to extract volatile components of interest from a variety of complex food matrices. Various extraction methods exist. Each technique has its particular advantages and disadvantages. The method of choice strongly de- pends on the variety and complexity of the matrix and the intended analytical investigations.
Solvent-assisted flavor evaporation (SAFE)—a new and versatile technique for the careful and direct isolation of aroma compounds from complex food matri- ces—was developed in 1999 by W. Engel et al. (104). Technical details and the design of the SAFE apparatus are exhaustively described by the authors. SAFE includes a compact distillation unit in combination with a high vacuum pump (Fig. 20).The developers of this new equipment pointed out the following advan- tages:
Higher yields of volatiles compared with the previously used high vacuum transfer technique
Higher yields of more polar flavor substances
Higher yields of odorants from fat-containing matrices
Direct distillation of aqueous samples such as milk, beer, orange juice, fruit pulps
Recovery of really authentic flavor extracts—i.e., the new extraction proce- dure provides a flavor sample with organoleptic properties as close as possible to the natural product
Reliable quantification of polar and labile trace volatiles in complex matri- ces compared with many other highly sophisticated modern isolation methods
In this section, the application of SAFE for the analysis of cheese aroma and the analysis of volatile fragrance components in washing powder is demonstrated.
FIGURE20 Schematic diagram of the basic components of the SAFE technique.
Furthermore, the direct distillation of grapefruit juice as well as of a soft drink is described.
A. Extraction of Cheese
The cheese sample was frozen in liquid nitrogen, then broken into smaller pieces and ground in a Grindomix blender. Powdered material was suspended in diethyl ether and the suspension was stirred for 18 hours at room temperature. The etheral
solution was dried over anhydrous Na2SO4and concentrated to a volume of 200
ml by means of a Vigreux column. Separation of volatile components from non- volatile cheese materials was performed by using the SAFE technique. Subse- quently, the total flavor extract was separated in basic, neutral, and acidic frac- tions. Nitrogen-containing components were extracted with aqueous HCl and carboxylic acids were removed by extraction with aqueous sodium carbonate.
Figure 21shows a comparison of the neutral cheese fractions isolated not only by SAFE but also by the SDE technique. There were differences in both the qualitative and quantitative composition. Many heat-induced alterations of the aroma profile in the SDE sample were observed (data not presented). Identifi- cation of cheese volatiles was based on GC/MS. We found, based on our spectro- scopic investigation of the acidic fraction, that the SAFE technique was more effective in extracting less volatile and polar constituents such as 4-hydroxy- 2,5-dimethyl-3(2H)-furanone, 4-hydroxy-5-methyl-3(2H)-furanone, and 5-ethyl- 4-hydroxy-2-methyl-3(2H)-furanone. As a result, it should be possible to quantify these organoleptically important trace constituents by special techniques—e.g.,
isotope dilution assay using labeled components as internal standards. Further- more, it is worth mentioning that the SAFE sample revealed the more authentic and representative aroma of the original product.
B. Direct Distillation of Beverages
Commercially available grapefruit juice and a soft drink were directly distilled by means of the SAFE technique. In this way, aqueous distillates free from any nonvolatile material were obtained. After extraction of the aqueous distil- lates with pentane/diethyl ether 1 : 1, flavor concentrates ready for direct HRGC
and GC/MS were produced. The resulting chromatograms are shown inFig. 22.
We assume that all important volatile compounds were recovered by the aid of the SAFE distillation and the following extraction step because the aromatic ex- tracts were truly representative. The soft drink extract was characterized by typical citrus oil notes, and the grapefruit extract provided characteristic bitter, fruity, green, tropical, juicy, citrus-like, and clear grapefruit-like aroma impres- sions.
InFig. 23,both SAFE extracts are depicted again but have been separated by means of a relatively new GC technique: EZ-Flash, an assembly that can be used to upgrade existing gas chromatographs. EZ-Flash is an innovative chro- matographic system that accomplishes in a few minutes what traditional GC does in an hour or more—i.e., Flash GC is over 20 times faster than conventional GC. The Flash column consists of a conventional column inserted into a metal tube, which is resistively heated by a precision power supply. The combination of a short capillary column, a high gas flow rate, and fast temperature programming significantly decreases analysis times. The application of EZ-Flash, however, causes some loss of separation efficiency—i.e. in most cases, resistive heating cannot be used to reduce the analysis time of extremely complex flavor or fra- grance samples without causing some loss of separation efficiency. Nevertheless, in spite of this disadvantage, the approach is ideally suited for rapid screening as illustrated in Fig. 23. The fundamental principle and some specific applications of this new technique have recently been published (105–108).
As we have already emphasized in this chapter, aroma profiles depend on the extraction method used. There is no universal extraction procedure in food flavor analysis, and different methods will influence aroma profiles. The new SAFE technique presented above represents an attractive approach in flavor re- search and a new tool for isolating and concentrating volatile components from solid or aqueous food materials. For that reason, the SAFE technique is expected to be widely applied in the future for highly efficient extraction of flavor compo- nents from various food samples.
FIGURE22 Direct distillation of beverages by means of the SAFE technique (30 m⫻
0.25 mm I.D. DB-WAX; 0.25µm df; 60°C-3°C/min ⫺ 230°C). Key to components identi-
fied in soft drink: 1, limonene; 2,γ-terpinene; 3, α-terpinolene; 4, nonanal; 5, linalool; 6, fenchol; 7, 1-terpinen-4-ol; 8,α-terpineol; 9, cinnamaldehyde; 10, myristicin. Key to components identified in grapefruit juice: 1, limonene; 2, cis-linalool oxide (f); 3, trans- linalool oxide (f); 4,β-caryophyllene; 5, α-terpineol; 6, trans-carveol; 7, dihydronootka- tone; 8, nootkatone.
FIGURE23 10 m EZ Flash GC-FID chromatograms of soft drink and grapefruit juice
extracts (10 m⫻ 0.1 mm I.D. DB-WAX; 0.2 µm df; 50°C–90°C/min ⫺ 230°C). Identifi-
TABLE1 Recovery Data Obtained by Distilling a Solvent Solution of 17 Perfume Oil Components as well as a Pentane/Diethyl Ether Extract from Washing Powder
Recovery (%)
Model mixture in Model mixture extracted Compound pentane/diethyl ether from washing powder
cis-3-Hexenyl acetate 73.6 50.5 2,4-Dimethyl-3-cyclohexene-1- 76.6 62.9 carboxaldehyde (Vertocitral) Linalool 75.6 57.4 β-Citronellyl acetate 77.9 54.9 Styrallyl acetate 77.1 77.0 Benzyl acetate 76.4 48.6 β-Geranonitril 78.9 65.6 β-Citronellol 77.3 69.8 Benzyl acetone 79.9 51.4 Lilial 80.4 46.6 Isocylemone E 78.9 64.9 Sandolene 80.7 66.9 Hexyl salicylate 79.5 76.4 Acetyl cedrene 79.5 59.5 α-Hexylcinnamaldehyde 79.8 45.1 Coumarin 79.5 64.6
Methylβ-naphthyl ketone 79.6 53.9
C. Extraction of Washing Powder
The analysis of fragrance volatiles in washing powder is a further excellent exam- ple of the potential of the SAFE technique. In order to test the efficiency of the SAFE technique with regard to perfume oil components, a model mixture of fragrance compounds dissolved in pentane/diethyl ether was distilled by means of the new apparatus. In addition, to study the influence of the washing powder matrix on the yields of the fragrance volatiles, the model mixture was incorpo- rated in the washing powder, extracted with pentane/diethyl ether, and distilled again using the SAFE equipment. As shown in Table 1, no fragrance component was completely recovered from the organic solution. As expected, investigating the influence of nonvolatile washing powder materials, the yield of each fragrance substance was decreased once again compared to the pure pentane/diethyl ether