Stevia (Stevia rebaudiana Bertoni) leaves pretreatment with pressurized CO 2 : an evaluation of the extract composition Antonio Pasquel 1 ; Marcia O. M. Marques 2 ; M. Angela A. Meireles 1,* 1 LASEFI-DEA-FEA-UNICAMP, Cx. Postal 6121, 13083-970 Campinas, SP, Brazil, meireles@fea.unicamp.br ; fax: 55.19.289.1513 2 LPN, Unidade de Genética, Biologia Molecular e Fitoquímica - IAC, Cx. Postal 28, 13001-970 Campinas, SP, Brazil To effectively use stevia glycosides as a sugar substitute its bitter taste has to be removed. The bitter taste is associated to some specific terpenic compounds as well as to the glycosides themselves. Therefore a two step process can be considered where some of the bitter compounds are removed previously to the glycoside extraction. Presently, the pretreatment of stevia leaves with carbon dioxide followed by glycoside extraction with a CO 2 /ethanol, CO 2 /water or low pressure water extraction process is been considered. The objective of this work was to investigate the pretreatment of stevia leaves using supercritical carbon dioxide. The kinetics of extraction and the chemical characterization of the extract were studied. Dried stevia leaves (Maringá, Paraná, Brazil) from de 1995 crop were used. The solvent was CO 2 at the following conditions: 200 and 80 bar, 25 and 40 C. The solid particle size varied from mesh 8 to 24. The chemical composition was established using GC-MS and thin-layer chromatography. Sesquiterpenes, fatty acids, aliphatic hydrocarbons, steroids, and triterpenes were identified. Keywords: CO2; Stevia; Supercritical extraction; Chemical composition INTRODUCTION The Stevia rebaudiana (Bertoni) Bertoni aqueous extract is commonly used as a sweetener. It is widely used for a long time in Asian and Latin American countries as a sugar substitute. Recently, stevia leaves or their extracts were allowable for import and commerce in the USA, if they are explicitly labeled as a dietary supplement, or as an ingredient for a dietary supplement. Its distribution in the American continent goes from Southwest United States to the Northwest of Argentina, going to Mexico, Central America, the South America Andes and the mountains of Brazil. Stevia is also cultivated in the Asian Southeast, Japan, China, and Israel. The aqueous stevia extract contains several chemicals called glycosides, which taste sweet, but do not contribute to the human caloric intake. Six sweet-tasting compounds ha ve been reported in stevia leave extracts: stevioside, rebaudiosides A, D, and E, and dulcosides A and B, the major compound being the stevioside. Stevia extracts have been used in Japan since the beginning of the seventies as sweetener agent, flavor modif ier, and sugar substitute. The glycosides from stevia leaves belong to the extensive variety of naturally sweet compounds discovered on the last half of this century. 501
MATERIAL AND METHODS Dried stevia leaves from the 1995 crop was purchased at Maringá (P araná, Brazil). Their humidity was 7% as determined by Jacobs method (1). The dried solid was cleaned, selected, packed in plastic bags and kept at room conditions (15 to 30 o C). Extractions were conducted using the fixed bed extractor shown in Figure 1. The highpressure pump was from Thermo Separations Products (constametric 3200) all other components were purchased locally. The extraction required a period of 12 hours. No detailed information was found in the literature concerning the pre-treatment of stevia leaves with carbon dioxide (2). Therefore, a set of preliminary experiments was done in order to select the experimental conditions. The process variable investigated were temperature, pressure, solvent flow rate, and solid particle size distribution. To analyze the effects of the process variables and keep the number of experiments within reasonable range a fractional factorial experimental design was employed (Table 1). To determine the phytochemical profile of the extracts thin layer chromatography associated with GCMS (Shimadzu, model QP-5000) analysis was employed. A fused silica capillary column DB-1 (25 m x 0.25mm, 0.25? m) was used. The electron impact technique (70ev) was utilized. The carrier gas was helium (1.7mL/min.) and 1?L of sample was injected. Temperature programming started at 70 o C and was raised to 280 o C at 10 o C/min, and held at this value for 35 min. Detector temperature was 230 o C and that for the injector was 240 o C. The identification of chemical constituents was based on i) comparison of substance mass spectrums with GCMS system data bank (Wiley 139 Library); ii) comparison of mass spectrums with data in literature (5); iii) co-injection of standards. 3 6a 9 2g 2f 11 6c 12 2a 4 2b 5 2c 10 13 14 1 8 7 9 3 2d 4 2e 6b 1: carbon dioxide reservoir 2: valves 3: retention valves 4: heat exchanger 5: jacketed surge tank 6: manometers 7: on-line filter 8: HPLC pump 9: thermocouples 10: jacketed extraction column 11: micrometering valve 12: expansion coil 13: extract collectors 14: flow totalizer Figure 1: Schematic view of the experimental unit developed at Campinas. 502
RESULTS AND DISCUSSION The influence of operating variables on the yield Table 1 shows the experimental conditions used and the yield obtained after 12 hours of extractions. The bed density varied from 1.3650 to 1.3447 g/cm 3 for fine (mesh 10/+24)) and coarse (mesh -8/+10) particles, respectively. The analysis of variance perf ormed using data on Table 1 showed that the effect of solvent flow rate and particle size were not significant (P=0.3552, 0.5285, respectively) while that of pressure and temperature were (P=0.0012, 0.0433, respectively). The interaction between temperature and pressure was also significant (P=0.0695). From Table 1 it can be observed that the largest yield was obtained at 200 bar and 40 o C. At 80 bar the larger yields were obtained at 25 o C. On the other hand, at 200 bar the effect of temperature is less clear. At this pressure the average yields were 0.75 and 0.72% (m/m) at 25 and 40 o C, respectively. This result indicates that the best selection for the pre-treatment conditions would be 200 bar and 25 o C, considering the energy saving of working at the lower temperature. The characterization of the extracts According to KINGHORN and SOEJARTO (4) the content of water-soluble substances in dried stevia leaves is about 42%. The major part of these const ituents being formed by the substances responsible for the sweet taste of stevia extracts. A considerable amount of work has been done in literature to elucidate the composition of the aqueous stevia extract (4), (5). Table 1: Extraction Conditions and Yield Run Temperature? 1 ( C) Pressure? 0.5 (bar) Solvent flow rate (L/min) Particle size Yield (mass %) (mesh) 1 25 200 1.47 Fine 0.62 2 40 80 1.45 Fine 0.08 3 40 200 1.46 Fine 0.69 4 40 200 1.05 Fine 0.56 5 40 200 1.00 Coarse 0.44 6 25 80 1.48 Coarse 0.37 7 40 80 1.06 Fine 0.00 8 40 200 1.48 Coarse 1.19 9 25 200 1.02 Coarse 0.84 10 25 80 1.46 Fine 0.42 11 25 80 1.04 Fine 0.75 12 40 80 1.02 Coarse 0.03 13 25 200 1.03 Fine 0.61 14 25 200 1.55 Coarse 0.94 15 40 80 1.48 Coarse 0.04 16 25 80 1.03 Coarse 0.44 503
Additionally to the aqueous soluble material or the diterpenic glycosides the plant also has essential oil with monoterpenes, diterpenes, triterpenes, labdanie diterpenes, steroids, flavonoids, and so on. A preliminary analysis of the chemical composition of the stevia extracts detected the presence of substances belonging to the following classes: mono and sesquiterpenes, fatty acids, etc. besides high molecular weight substances not yet identified. The phytochemical profiles of the extracts were similar for all experimental conditions. Table 2 shows the chemical composition for the extracts obtained in run number 8. The identification of the steroids was done using co-injection of standards. The presence of substances belonging to the class of triterpenes pentacyclic and aliphatic hydrocarbons was suggested by the fragmentation of their mass spectrums. The comparison of the mass spectrums of these substances with data in literature (3), (6), (7), (8) allowed to detect the presence of stigmasterol (molecula r mass 412),? -sitosterol (molecular mass 414),? /?-amyrine (molecular mass 426),? -amyrine acetate (molecular mass 468), lupeol (molecular mass 426). Table 2. Chemical composition of stevia leaves extracts Class Substance Composition (mass %) Sesquiterpenes Spathulenol 1.49 Fatty acids Decanoic acid 3.35 8,11,14-eicosatrienoic acid 3.05 Aliphatic hydrocarbons 2-methyl octadecane 2.55 Pentacosane 8.54 Octacosane 8.42 Steroids Stigmasterol 1.63?-sitosterol 1.26 Triterpenes? /?-amyrine 0.68 Lupeol 5.94?-amyrine acetate 4.85 Triterpene pentacyclic 14.10 Non identified 44.14 CONCLUSIONS The stevia extract is a very viscous fluid that hardly flows through the equipment pipelines. During extraction the outlet was constantly clogged. To avoid the clogging heating up to 100 o C was required. Unfortunately, this procedure resulted in the degradation of the material as visually detected by the darkening of the extract. Pressure and temperature, as well as their interaction significantly affected the yield. The effects of solvent flow rate and particle size on yield were not significant. The effect of pressure was more important than the effect of temperature. The effect of temperature was more pronounced at 80 bar than at 200 bar. At 200 bar the average yields were 0.75 and 504
0.72% (m/m) at 25 and 40 o C, respectively. This result indicates that the best selection for the pre-treatment conditions would be 200 bar and 25 o C, considering the energy saving working at the lower temperature. The substances present in stevia extracts belong to the classes of sesquiterpenes (spathulenol), fatty acids (decanoic acid, 8,11,14-eicosatrienoic acid), aliphatic hydrocarbons (2-methyl octadecane, pentacosane, and octacosane), steroids (stigmasterol and?-sitosterol), triterpenes (? /?-amyrine, lupeol,?-amyrine acetate, and pentacyclic triterpene), and other components not identified. Additional work is being done at LASEFI and LPN to confirm the adequacy of the pretreatment. This includes sensor analysis of stevia extracts obtained from stevia leaves, with and without pretreatment, by CO 2 /ethanol, CO 2 /water and low -pressure water extractions. ACKNOWLEDGMENT The authors want to express their gratitude to Dr. Ademir J. Petenate (DE- IMECC/UNICAMP) for his assistance with the experimental design and the statistical analysis. This research was financially supported by FAPESP (Grant No. 95/5262-3). Antonio Pasquel has a Ph.D. scholarship from FAPESP (95/3390-4). REFERENCES (1) JACOBS M. B. (1973). The chemical analysis of foods and food products. 3rd. Ed., Robert Krieger Publishing Co. New. Yor k, p. 21-23 (2) KIENLE, U. Method of making a natural sweetener based on Stevia rebaudiana, and use thereof. Int. Cl. A61K 35/78. U.S. Cl. 424/195.1, 586690 Sep. 24, 1990; may 12, 1992. Chemical Abstracts n.111, 22442z. (3) McLAFFERTY F.W., STAUFFER D.B. The Wiley/NBS registry of mass spectral data. Vol.1-7, John Wiley & Sons, New York, 1989. (4) KINGHORN D.A., SOEJARTO D.D. Current status of stevioside as a sweetening agent for human use. In: H. Wagner, H. Hikino, N.R. Farnsworth (Eds.) Economic and Medicinal Plant Research. Academic Press, Orlando, Florida, 1985 (5) NIKOLOVA-DAMYANOVA B., BANKOVA V., POPOV S. Separation and quantification of stevioside and rebaudioside A in plant extracts by normal-phase high performance liquid chromatography and thin-layer chromatography: A comparison. Phytochemical Analysis 5, 81-85, 1994 (6) BUDZIKIEWICZ H., DJERASSI C., WILLIAMS D.H. Structure Elucidation of Natural Products by Mass Spectrometry, Holden-Day, San Francisco, 1964. 505
(7) BUDZIKIEWICZ H., WILSON J.M., DJERASSI C. Mass spectrometry in structural and stereochemical problems. XXXII. Pentacyclic triterpenes. Journal of the American Chemical Society 85 (22) 3688-3699, 1963. (8) DJERASSI C., BUDZIKIEWICZ H., WILSON J.M. Mass spectrometry in structural and stereochemical problems. Unsaturated pentacycclic triterpenoids. Tetrahedron Letters (7) 263-270, 1962 506