1990, 3, 15-19
Extraction of Flavors from Milk Fat with Supercritical Carbon Dioxide A. B. de Haan* and J. de Graauw [email protected]
University of Technology, Laboratory for Process Equipment, Leeghwaterstraat 44, 2628 CA DELFT, The Netherlands J. E. Schaap and H. T. Badings NIZO (Dutch Institute for Dairy Research), P.O. Box 20, 6710 BA EDE, The Netherlands Received September 11, 1989; accepted in revised form January 15, 1990
Next to triglycerides, milk fat contains a large number of components (lactones, ketones, aldehydes) that provide milk fat with its characteristic flavor. In this study, supercritical carbon dioxide has been found to be a good solvent for the extraction of these flavor components from milk fat. Concentration factors varying from 20 to 50 have been measured at carbon dioxide densities of 600 to 700 kg/m3 and extraction temperatures between 40 to 50 “C. It has been observed that, if the the flavor extract is produced in two steps, these flavor components could be concentrated 500 to possibly 1000 times. The height equivalent to a theoretical equilibrium stage in the extraction column used, filled with 5-mm steel raschig rings, appearedto be approximately 20 cm. Keywords:
supercritical extraction, flavors, milk fat, carbon dioxide, mass transfer
INTRODUCTION About 98% of the constituents of milk fat are triglycerides with a carbon number varying from 26 to 54.’ In the remaining 2%, a large number of other components (lactones, ketones, aldehydes), that contribute to the characteristic flavor of milk fat, are present in small concentrations (i.e., ppm level).2 These natural flavor components are of especially great interest to the flavor and food industry. This paper presents the results of a study on the extraction of flavor components from milk fat with supercritical carbon dioxide. Carbon dioxide has been chosen as the solvent since its critical temperature (T, = 3 1.1 “C) makes it an ideal solvent for extracting thermally labile materials. Also, carbon dioxide is nontoxic, nonflammable, inexpensive and has a high purity. For these reasons, the properties of the flavor components and the milk fat will be preserved and the extracted milk fat will retain its natural value.3%4 The objective of this study was to produce a flavor extract from milk fat in which the flavor components are concentrated 500 to 1000 times. The process conditions for
the production of a flavor extract were determined from experiments in which the influence of the carbon dioxide density, temperature, and solvent-to-feed ratio on the extract content, the concentration factor, and the fraction of flavor components extracted from the milk fat were measured. Hereafter, the height equivalent to a theoretical stage was calculated by modelling the extraction process with the Kremser equation, and two extraction steps were used to produce a flavor extract. EXPERIMENTAL A simplified flow diagram of the experimental equipment, suited for extractions at pressures between 100 and 600 bar and temperatures from 25 to 90 “C, is shown in Figure 1. The extraction column (4) used in the experiments, with a length of 1 m and an inside diameter of 35 mm, was packed with 5-mm metal raschig rings. The feed (1) entered the top of the column at extraction pressure (2) and extraction temperature (3). After extraction, the remaining raffinate, containing a considerable amount (20-40 wt %) of carbon dioxide, was drawn off from the bottom of the column.
0896-8446/90/030 l-00 15$4.00/O
0 1990 PRA Press
16 Haan et al.
The Journal of Supercritical tltkj-up rolrent
Fluids, Vol. 3, No. I, 1990
300 IL t 200 -
0; do Flowsheet of the supercritical extraction equipment. (1) feed supply vessel; (2) pump; (3) heat exchanger; (4) column; (5) separator; (6) condenser; (7) CO, supply vessel; (8) pump; (9) heat exchanger.
The supercritical carbon dioxide was fed to the bottom of the column at extraction pressure (8) and extraction temperature (9) and left the top of the column with the dissolved extract. From experiments in which solvent flow was varied, it appeared that carbon dioxide was always saturated with the milk fat components. A turbine meter was used to measure the carbon dioxide feed to the column. After extraction, the pressure was reduced below the critical pressure of carbon dioxide (W-60 bar) to condense the extract in the separator (5). This extract was periodically drawn off. Fresh carbon dioxide was added to the evaporating carbon dioxide from the separator to compensate for the carbon dioxide loss in the raffinate. Hereafter, the gaseous carbon dioxide was condensed before increasing the pressure again to the extraction pressure with a membrane pump (8). Two steps were used to analyze the flavor components in the feed, the extract, and the raffinate.5 First, the flavor components were separated from the triglycerides with HPLC. Subsequently, a quantitative analysis of the flavor fraction was made with capillary gas chromatography. To check upon the reliability of the analytical procedure, all analyses were performed twice. The accuracy of the extraction experiments including the analytical procedure appeared to be approximately 10%. RESULTS The solubility in and selectivity of a supercritical solvent are mainly determined by the solvent density and the extraction temperature. 3,4 Figure 2 illustrates that the pressure needed to obtain a desired carbon dioxide density increase with higher temperatures. The effect of the solvent density on the solubility of the milk fat components (extract content) in supercritical carbon dioxide is shown in Figure 3. It can be seen that the extract content increases dramatically at higher carbon dioxide densities. This solubility behavior can be explained by the stronger interactions between the carbon dioxide molecules and the milk fat components that occur at higher solvent densities. In addi-
d90 pro, lka/mJl
2. Pressureas a function of the carbon dioxide density and the extraction temperature.6
3. Extract content (wt %) as a function of carbon dioxide density and the extraction temperature at a solvent-tofeed ratio of 6 (kg/kg).
tion to this density effect, it can be seen from Figure 3 that the extract content also increases at higher temperatures as a result of the higher vapor pressures of the milk fat components. The difference in vapor pressure between the volatile flavor components and the nonvolatile triglycerides offers good possibilities for their separation because components with a higher vapor pressure will be concentrated in the extract since they are more soluble in supercritical carbon dioxide. To illustrate this vapor pressure effect, the concentration factor of the dl2-lactone (6dodecalactone) was plotted in Figure 4 as a function of the carbon dioxide density and the extraction temperature. It can be seen that the dl2-lactone concentration factor decreases dramatically at higher carbon dioxide densities and increasing temperatures. This means that at higher carbon dioxide densities and higher temperatures, the triglyceride fraction extracted from the milk fat is increased markedly as shown by Figure 3. On the other hand it can be concluded from Figure 5 that the faction dl2-lactone extracted from the milk fat increases
Extraction of Flavors from Milk Fat 17
Vol. 3, No. 1, 1990
30. WCF 1 20 -
0-l 500 ,
4. Concentration factor of the dl2-lactone as a function of the carbon dioxide density and the extraction temperature at a solvent-to-feed ratio of 6 (kg/kg).
6. Concentration factor of the dl0, d12, d14, and dl6-lactone as a function of the carbon dioxide density at 60 “C and a solvent-to-feed ratio of 6 (kg/kg).
5. Fraction dl2-lactone extracted from the milk fat as a function of the carbon dioxide density and the extraction temperature at a solvent-to-feed ratio of 6 (kg/kg).
much less because the flavor gets exhausted. As a result of the vapor pressure effect, the flavor components present in milk fat will be concentrated to a different extent depending on their molecular weight. This explains that the concentration factor decreases when the extracted flavor components possess a higher molecular weight, as illustrated in Figure 6. MODELLING OF EXTRACTION PROCESS Figure 7 illustrates that the fraction extracted and the concentration factor of the dl2-lactone are coupled by the solvent-to-feed ratio. A maximum in the concentration factor of the dl2-lactone is attained at a small solvent-to-feed ratio. This maximum indicates that the concentration of the dl2-lactone in the solvent stream leaving the column is in equilibrium with the concentration of the d 12-lactone in the feed. The distribution coefficient of the d 12-lactone between the milk fat and the supercritical carbon dioxide is now calculated from Kd,2 = c, X %E = 40 X 0.0029 CF
7. Fraction extracted from the milk fat (B) and concentration factor (e) of the dl2-lactone as a function of the solvent-to-feed ratio (kg/kg) at 60 “C and a carbon dioxide density of 640 kg/m3 (P = 170 bar). The curves have been calculated with the Kremser equation. K,,, = 0.12, %E = 0.0029 (kg/kg), N = 2.4.
It can be seen from Figure 7 that at higher solvent-to-feed ratios the concentration factor of the dl2-lactone decreases while the fraction extracted increases. Since only a small fraction of triglycerides is extracted from the milk fat and the supercritical carbon dioxide contains only a few weight percent extract, it was assumed that the milk fat as well as the solvent stream will remain constant throughout the column. The distribution coefficient of the dl2-lactone between the milk fat and the supercritical carbon dioxide will not depend on the concentration of the dl2-lactone in the milk fat and the supercritical carbon dioxide because these concentrations will always be very low. For these reasons, the Kremser equation7 can be used to model the extraction process. The Kremser equation relates the fraction dl2-lactone extracted from the milk fat (f) to the extraction factor (E) and the number of equilibrium stages (N) .f=
The Journal of Supercritical
18 Haan et al.
TABLE I Process Conditions and Results for the Production of the Flavor Extract
Fluids, Vol. 3, No. I, 1990
Extract 1 Equation 2 was used to calculated the concentration factor and the fraction extracted of the d 12-lactone as a function of the solvent-to-feed ratio. The calculated curves in Figure 7 are in good agreement with the experiments in the extraction column when the packed part of the column (H = 50 cm) contains 2.4 equilibrium stages with a height of 20 cm. This result is comparable with the height of an equilibrium stage in an extraction column, filled with metal raschig rings of 6.4 mm, measured by Rathkamp8 for the system carbon dioxide/isopropanol/water. PRODUCTION OF FLAVOR EXTRACT The experiments described in the previous paragraphs were used to determine the process conditions for the production of a flavor extract. To achieve concentration factors between 500 and 1000, the flavor extract was produced in two stages with the equipment described in the experimental section. The main objective of the first extraction stage has been the separation of the flavor components from the triglycerides. A high carbon dioxide density was used to extract the major part of the flavor components present in milk fat. The flavor components were concentrated as much as possible by choosing a low extraction temperature and a small solvent-to-feed ratio. In this first extraction stage, a carbon dioxide density of 810 kg/m3 (P = 200 bar, T = 45 “C) and a solvent-to-feed ratio of 4 kg/kg was used. From Table I, it is seen that 33 to 75 % of the lactones were extracted from the milk fat. Reducing the solvent-tofeed ratio from 6 to 4 kg/kg increased the concentration factor of the dl2-lactone from 10 to 19. The second extraction stage was used to obtain a maximum concentration factor. Therefore, we have maintained a solvent-to-feed ratio of 4 kg/kg and reduced the carbon dioxide density to 650 kg/m3 (P = 170 bar, T = 60 “C). Table I shows that after two extraction stages, concentration factors varying from 100 to 650, depending on the molecular weight of the lactones, have been achieved. CONCLUSIONS The results of the experiments and the flavor extract produced have shown that supercritical carbon dioxide can be used for the extraction of flavor components from milk fat. Concentration factors varying from 20 to 50 were measured at carbon dioxide densities from 600 to 700 kg/m3 and extraction temperatures between 40 and 50 “C. From these experiments, it was concluded that the large difference in vapor pressure between the flavor and the other milk fat components (triglycerides) is the main reason for achieving these large concentration factors.
200 45 810 4
P (bar) T (“C) pg2 (Wm3)
d12 d14 d16
Extract 2 170 60 640 4
23 19 14 10
75 64 50 34
650 460 220 100
We have modelled the extraction process with the Kremser equation since only small amounts of flavor and other milk fat components (triglycerides) were extracted from the milk fat. The height equivalent to a theoretical stage in the extraction column used, filled with 5-mm steel raschig rings, appeared to be approximately 20 cm. At least two extraction stages are required to concentrate the flavor components approaching 1000 times. The concentration factor in the first extraction stage can be increased by reducing the carbon dioxide density. Also, higher concentration factors can be achieved in the second extraction stage by reducing the extraction temperature. The fraction of flavor components extracted from the feed will be enlarged when the solvent-to-feed ratio and the number of equilibrium stages in the column is increased. ACKNOWLEDGMENTS We would like to thank A. de Leeuw and F.W.R. Soeterbroek for performing the experiments reported in this paper during their graduation period at the NIZO (Dutch Institute for Dairy Research). Great support was received from C. de Jong and C. Olieman for the analysis of the flavor components in milk fat for which we are very grateful. List P %E G CF
d10 d12 d14 d16 F H K P
of Symbols density (kg m-3) extract content (kg/kg) concentration lactone in extract (solvent free) (ppm) concentration lactone in feed (solvent free) (ppm) d-decalactone d-dodecalactone d-tetradecalactone d-hexadecalactone extraction factor (K x S/F) fraction extracted height of the packing distribution coefficient pressure (bar) solvent-to-feed ratio (kg/kg) temperature (“C)
Vol. 3, No. 1, 1990
REFERENCES (1) Mulder, H.; Walstra, P. In The Milk Fat Globule; Pudoc: Wageningen, 1984. (2) Badings, H. T. In Dairy Chemistry and Physics; Walstra, P.; Jenness, R., Eds.; John Wiley & Sons: New York, 1984, p 336. (3) McHugh, M. A.; Krukonis V. J. In Super-critical Fluid Extraction: Principles and Practice; Butterworths: Boston, 1986. (4) Stahl, E.; Quirin K.-W.; Gerard, D. In Verdichtete Gase zur Extration und Raffination; Springer-Verlag: Berlin, 1987.
(5) (6) (7) (8)
Badings, H. T.; Olieman, C.; de Jong, C. To be published. Angus, S.; Armstrong, B.; de Reuck, K. M. In International Thermodynamic Tables of the Fluid State. Vol. 3. Carbon Dioxide; Pergamon: Oxford, 1976. Pratt, H. R. C. In Handbook of Solvent Extraction; Lo, T. C.; Baird, M. H. I.; Hanson, C., Eds.; John Wiley & Sons: New York, 1983, p 169. Rathkamp, P. J.; Bravo, J. L.; Fair, J. R. Solv. Extr. Ion Exch. 1987,5, 367.