《现代食品工程高新技术》课程教学资源（文献资料）食品分离新技术 Effect of cosolvents（ethyl lactate, ethyl acetate and ethanol）on the supercritical CO2 extraction of caffeine from green tea.pdf_大学文库

J. of Supercritical Fluids 107 (2016) 507–512 Contents lists available at ScienceDirect The Journal of Supercritical Fluids journ al h om epage: www.elsevier.com/locate/supflu Effect of cosolvents (ethyl lactate, ethyl acetate and ethanol) on the supercritical CO2 extraction of caffeine from green tea David Villanueva Bermejo, Elena Ibánez, ˜ Guillermo Reglero, Tiziana Fornari ∗ Institute of Food Science Research CIAL (CSIC-UAM). CEI UAM+CSIC. C/Nicolás Cabrera 9, Campus de Cantoblanco, 28049 Madrid, Spain a r t i c l e i n f o Article history: Received 16 May 2015 Received in revised form 5 July 2015 Accepted 6 July 2015 Available online 13 July 2015 Keywords: Green tea Caffeine Cosolvent Ethyl lactate Supercritical fluid extraction a b s t r a c t This paper reports experimental data to analyze the effect of different green cosolvents, namely ethyl lactate, ethyl acetate and ethanol, on the supercritical carbon dioxide (SCCO2) extraction of caffeine from green tea leaves. The experiments were carried out in a pilot-scale plant using two different approaches: a static procedure in which the cosolvent was introduced in the extraction cell soaking the vegetal material and then SCCO2 was pumped, and a dynamic assay in which the cosolvent was pumped and mixed with SCCO2 before introduction into the extraction cell. The overall caffeine recovery from plant matrix was determined for all experiments at the same extraction conditions (30 MPa and 343K). Additionally, the overall extraction curves of the static experiments were determined at the same process conditions, and the mass transfer model of Sovová was utilized to adjust the kinetic data and to determine the mass transfer coefficients. The highest caffeine yield was obtained with ethyl lactate in both static and dynamic extractions (13.0 and 14.2 mg/g of tea, respectively), followed by ethanol (10.8 mg/g with the static method and 8.8 mg/g with the dynamic method). The yield obtained with ethyl acetate in both extraction approaches was lower than 7 mg/g. These data reinforce previous results obtained by the authors regarding the competence of ethyl lactate in the extraction of caffeine from natural materials. © 2015 Elsevier B.V. All rights reserved. 1. Introduction Green tea (Camellia sinensis) is one of the most popular beverages in the world [1] and contains around 2–5% mass of caffeine (1,3,7-trimethylxanthine) in the dry leaves [2]. Some adverse effectsderivedfromthe excessive consumptionof caffeine are wellknown [3–5] and thus, the removal of caffeine is desired by several consumers. Extraction using organic solvents is currently the most extended commercial method for decaffeinating green tea leaves. Despite the highselectivity towards caffeine,theuse of solvents like chloroform or methylene chloride has progressively decline due to their high toxicity [6]. Another commercial solvent utilized is ethyl acetate which presents much lesser toxicity than chlorinated solvents. Supercritical fluid extraction (SFE) with CO2 is a viable alternative to organic solvents, offering many advantages such as non-toxicity, solvent-free products and high selectivity in caffeine ∗ Corresponding author: Fax: +34 910017905. E-mail address: tiziana.fornari@uam.es (T. Fornari). removal. Several works aboutthe use of supercritical CO2 for decaffeination of different vegetal matrix, such as coffee beans [7], cocoa butter [8], guarana seeds [9] and mate tea leaves [10], have been reported. In fact, the extraction of caffeine with supercritical CO2 is one ofthe most well-known commercial examples of SFE processes [11]. Many patents have been published about decaffeination oftea and coffee where extraction of caffeine is carried out by means of different layouts [12–15]. Moreover, in different countries, several large-scale plants have been designed for processing coffee, tea, and hops, and for spices and flavor extraction [16]. Regarding the extraction of caffeine from green tea leaves, in most cases, green tea was soaked previously with a liquid solvent, namely ethanol or water. For example, Park et al. [17,18] studied the decaffeination of green tea using CO2 and ethanol as cosolvent at 30 MPa and 343K, attaining a recovery of caffeine around 92–96%. In another contribution [19], the authors employed the same extraction conditions with pure CO2 and water as cosolvent and caffeine recoveries were lower. Similarly, Kim et al. [20] attained a 59% of caffeine recovery at 40 MPa, 323K and using 7% of water as cosolvent. Chang et al. [21] tested water, ethanol and http://dx.doi.org/10.1016/j.supflu.2015.07.008 0896-8446/© 2015 Elsevier B.V. All rights reserved

508 D.V. Bermejo et al. / J. of Supercritical Fluids 107 (2016) 507–512 different water/ethanol ratios at 31 MPa and 333K, obtaining the best results with 70% aqueous ethanol. On the other side, Sun et al.[22] studied the supercritical extraction of caffeine also using ethanol and water, but the cosolvent was pumped and mixed with SCCO2 in continuous flow (the vegetal material was not soaked previously). At 30 MPa and 333K the recovery of caffeine using ethanol, 50% aqueous ethanol and water was, respectively, 67.4%, 92.8%, 78.0%. Huang et al. [23] combined soaking and continuous CO2+cosolvent pumping. Complete decaffeination was attained at 30 MPa and 353K, but it should be taken into account the small content of caffeine (9.92 mg/g) contained in the green tea utilized in their experiments. Thus, to the best of our knowledge, only water and ethanol have been studied as cosolvents in the supercritical extraction of caffeine from green tea leaves. In this work, the kinetic behaviour of the caffeine extraction when ethyl lactate and ethyl acetate are employed as cosolvents is presented for the first time. For comparison, also caffeine extraction using ethanol was analyzed and reported. Furthermore, the two most common approaches, i.e., previous soaking ofthe vegetal material and continuous CO2 and cosolvent pumping, were investigated. 2. Material and methods 2.1. Samples and reagents “Gunpowder” green tea (C. sinensis) leaves were acquired in a Spanish market. The green tea leaves were ground (particle size smaller than 500 m) in a cooled knife mill (Grindomix GM 200. Retsch GmbH, Haan, Germany) using liquid nitrogen. Ethyl lactate (≥98% purity) and caffeine standard (≥99% purity) were obtained from Sigma–Aldrich (St. Louis, MO, USA). Ethyl acetate and acetonitrile (HPLC grade) were obtained from Lab-Scan analytical sciences (Gliwice, Polonia). Formic acid (≥98% purity) was obtained from Merck (Darmstadt, Alemania). Ethanol (≥99% purity) and sea sand were purchased from Panreac Química S.A.U. (Barcelona, Espana). ˜ CO2 was provided by Air Products (Allentown, PA, USA). The total content of caffeine in the vegetal material was measured. For this purpose, 200 mg of green tea leaves were extracted at 343K with 20 mL of an aqueous ethanol solution (30% v/v) [17] in a Stuart Orbital S150 shaker apparatus (Bibby Scientific Limited Stone, UK) during 4 h. Then, the solvent was renewed and successive extraction cycles of 4 h were accomplished until the extraction yield in the corresponding cycle was lower than 2% of total yield. The content of caffeine determined was 27 mg per g of green tea leaves. 2.2. Supercritical extraction method Extractions were carried out in a pilot-scale supercritical fluid extractor (Thar Technology, Pittsburgh, PA, USA, model SF2000) comprising a 2 L cylinder extraction cell and two different separators (S1 and S2), each of 0.5 L capacity, with independent control of temperature and pressure. A detailed explanation of the experimental device can be found elsewhere [24]. The extractions were carried out using pure CO2 and CO2 with three different cosolvents, namely ethyl lactate (CO2 + EL), ethanol (CO2 + ETOH) and ethyl acetate (CO2 + EAC). Based on the experimental information available in the literature [17,18,25,26], the extraction pressure and temperature were selected to be 30 MPa and 343K and were kept constant for all experimental assays. At these conditions high caffeine solubility in SCCO2 is attained [25]. Furthermore, a supercritical procedure Table 1 Single Reaction Monitoring (SRM) transitions and the corresponding collision energies selected for the analysis of the main catechins and caffeine of green tea. Compound Precursor ion Daughter ion Collision energy (V) (+)−catechin 290.971 138.987 20 Epicatechin 290.971 138.987 20 Epigallocatechin 306.932 138.989 15 Epicatechin gallate 443.029 123.034 32 Epigallocatechin gallate 459.035 138.999 23 Caffeine 195.066 138.055 19 for the decaffeination of moistened green tea leaves has been early described at 293–353K and pressure up to 30 MPa [26]. The extraction cell was loaded with 0.7 kg of green tea leaves and CO2 flow rate was set to 9 kg/h. For the extractions using cosolvent, the cosolvent/CO2 ratio (kg/kg) was 0.022 (around 2% mass). These extractions were conducted in two different modes: static and dynamic modes. In the first case, green tea was soaked with the cosolvent and was loaded into the extraction cell. The soaked leaves were heated up to 343K,then CO2 was pumped untilthe extraction pressure was attained. The mixture was allowed to stand during 30 min and finally, CO2 was pumped afterward during 3.5 h (210 min). In the dynamic mode, the dry grounded leaves were loaded into the extraction cell and heated up to 343K, then CO2 and the cosolvent were continuously pumped and mixed in the desired ratio previous to introduction into the cell. Decompression of the extract up to ambient pressure was accomplished in a separator at 303K. Extraction yield was calculated as the difference between the mass of green tea leaves loaded in the extraction cell and the mass of leaves remained in the cell after the extraction. Samples were collected at different intervals of time. Samples were stored at 253K in the dark until analysis. 2.3. Identification and quantification of caffeine and catechins Analysis was performed by HPLC-MS/MS with an Accela (Thermo Electron Corporation; San Jose, CA) equipped with an ACE 3C18-AR column (150 × 4.6 mm, 3 m particle size)(Advanced Chromatography Technologies, Aberdeen, UK) equipped with a DAD detector and triple quadrupole mass spectrometer (TSQQuantum, Thermo Electron Corporation, San Jose, CA) with an ESI (Electrospray Ionization) interface. Based on the method of Pelillo et al. [27] the composition of the mobile phase was (A) 0.5% (v/v) formic acid in water (B) 0.3% (v/v) formic acid in acetonitrile. The column temperature was maintained at 308K, with a flow rate of 0.6 mL/min. The mobile phase gradient employed was as follows: initial 100% A, 30 min 0% A, 32 min 100% A, 37 min 100% A. Spray voltage and sheath gas pressure was set in 5000 and 35 psi respectively and the capillary temperature was 623K. Mass analyzer was set simultaneously in full scan and SRM (Single Reaction Monitoring) modes. In this case, SRM experiments were done using 1 precursor ion and 1 daughter ion, operating in positive mode. Table 1 shows the SRM transitions selected automatically among the most abundant ions and the corresponding collision energy. The amount of caffeine and catechins in the different samples was calculated by triplicate from a calibration curve of standard. 3. Results and discussion Table 2 shows the total extraction yield (g of extract/g oftea) and the caffeine yield (mg caffeine extracted/g tea) obtained for both extraction approaches (static and dynamic) and for each cosolvent studied. The dynamic and static extractions with ethyl lactate were carried out by duplicate. The mean values of the total extraction yields are reported in Table 2 and the standard deviations were,

D.V. Bermejo et al. / J. of Supercritical Fluids 107 (2016) 507–512 509 Table 2 Extraction yield (g of extract/g of tea), caffeine concentration (g of caffeine/g of extract), caffeine yield (mg of caffeine/g of tea) and caffeine recovery (g of caffeine extracted/g of caffeine in tea) in the static and dynamic SFE of green tea leaves using pure CO2, CO2 + EL, CO2 + ETOH and CO2 + EAC. Extraction conditions: 30 MPa and 343K. Extraction mode Solvent Extraction yield (%) Caffeine concentration (%) Caffeine yield (mg/g) Caffeine recovery (%) Pure CO2 3.7 15.9 5.9 21.9 Static CO2 + EL 9.1a 14.3 13.0 48.3 CO2 + ETOH 10.9 9.9 10.8 40.2 CO2 + EAC 8.1 8.1 6.6 24.6 Dynamic CO2 + EL 7.8a 18.2 14.2 52.6 CO2 + ETOH 8.7 10.1 8.8 32.6 CO2 + EAC 7.5 5.5 4.1 15.2 a Mean value between duplicate experiments. respectively, 0.3 and 0.2%. Standard deviation obtained in the quantification of caffeine was lower than 0.1 mg/g. Despite the extraction type, the effect of the cosolvent on the recovery of caffeine was as follows: ethyl lactate > ethanol > ethyl acetate. This behavior may be explained considering that hydrogen bonding would constitute one of the most important forces taking into account the chemical structure of the cosolvents and solute studied in this work [18,22,25,28]. In this respect, ethyl lactate can develop hydrogen bonding owing to its hydroxyl and carbonyl group [29] while ethanol only possesses a hydroxyl group. On the other side, ethyl acetate is apolar aprotic solventthatdonotpossess donor hydrogen atoms and cannot form strong hydrogen bonds. Thus, in comparison with pure CO2 (see Table 2) caffeine recovery was higher using ethanol or ethyl lactate cosolvents, but with ethyl acetate recoveries were similar and even lower (dynamic mode). Additionally, solubility behaviour of caffeine in the liquid solvents also supports the observed behavior. The solubility of caffeine in ethyl lactate at 323K is 5.1% mass [30], while solubility in ethanol and ethyl acetate atthe same temperature is, respectively, 2.3% and 2.2% mass [31]. The highest caffeine extraction yield was obtained with CO2 + EL and the recovery was slightly higher in the case of the dynamic approach (13.0 mg/g and 14.2 mg/g in the static and dynamic modes, respectively). Nevertheless, in the case of ethanol and ethyl acetate cosolvents, recoveries were considerably higher in the static mode. Taking into account the total content of caffeine in the green tea leaves (27.0 mg/g), the recovery of caffeine (mass of caffeine in the extract/mass of caffeine in green tea leaves × 100) at 343K, 30 MPa and 210 min of extraction were 48.3% and 52.6% using CO2 + EL in the static and dynamic modes respectively, while recoveries lower than 40% were obtained with the other cosolvents (see Tables 2 and 3). This result is in agreement with previous results reported by the authors [32] regarding the effectiveness of ethyl lactate in comparison with ethanol or ethyl acetate in the pressurized liquid extraction of caffeine from green coffee beans. The selectivity of the cosolvents to extract caffeine was valued considering the concentration of caffeine in the extracts (see Table 2).As expected,pureCO2 producedextracts withhighcontent of caffeine (15.9% mass) demonstrating high selectivity to extract this alkaloid, although recovery was the lowest (21.9%). Among the cosolvents investigated, ethyl lactate produced the highest caffeine concentrations, with values similar (14.3% mass in static mode) or even higher (18.2% mass in dynamic mode) than pure CO2 and the highest caffeine recoveries as mentioned previously. Assessment of the co-extraction of substances other than caffeine is also important since these substances may affect the sensorial quality and/or bioactivity functions of green tea. In this respect, the content of catechins was determined by HPLC in the whole extract of the static assays, and resulted in values of 0.2, 0.8 and 6.7 mg catechins/g tea, respectively, for EA, EL and ETOH cosolvents. The caffeine/catechins selectivity can be pondered considering the ratio caffeine/catechins extracted per g of green tea leaves, which resulted 33 for EA, 16.3 for EL and 1.6 for ETOH. The cosolvent polarity is influencing the co-extraction of catechins and thus the caffeine/catechins selectivity. Fig. 1 shows the overall extraction curve (OEC) obtained using pure CO2 and the three cosolvents in the static-SFE. The highest extraction rates of caffeine at the early extraction stages were obtained with CO2 + EL, while the lower extraction rates resulted with pure CO2. The early extraction rate of caffeine were calculated on the basis of the first kinetic data collected (after 14 min of CO2 flow) and were 0.063 mg/min in the case of pure CO2 and 0.098 mg/min, 0.289 mg/min and 0.486 mg/min, respectively, for CO2 + EAC, CO2 + ETOH and CO2 + EL. Then, ethyl lactate was the most efficient cosolvent and can produce a caffeine recovery enhancement close to 7 in comparison with pure CO2. The OECs of caffeine static-SFE are depicted in Fig. 1 for the case of pure CO2 (Fig. 1a), CO2 + EAC (Fig. 1b), CO2 + ETOH (Fig. 1c) and CO2 + EL (Fig. 1d). The OECs were adjusted using the model of Sovová [33] which is based on assumption that Xp mass of solute is easy accessible to the supercritical solvent (due to cell wall disruption) while the rest (Xk) remains inside cell walls. Thus, three steps are considered in the SFE process: (i)the constant rate period, where only the easily accessible solute is removed and thus, is controlled by convection in the fluid phase; (ii) the falling rate period, where both convection and internal mass transfer are important; (iii)the internal diffusion controlled rate period, where the remaining solute is only inside the cell walls. The corresponding model equations describing each extraction step are given in theAppendix A. Furthermore, itis considered thatthe supercritical solvent flows axially through a cylindrical extraction bed, the solvent is solutefree at the bed inlet and particle size distribution is homogeneous throughout the extraction cell. The solid particle density (s) is 1046 kg/m3, bed porosity () was determined on the basis of the corresponding apparent density (411.8 kg/m3) and resulted = 0.6, and CO2 density () at 30 MPa and 343K was determined from thermodynamic tables ( = 788.6 kg/m3) [34]. Due to the lack of experimental solubility data of caffeine in the CO2 + EL and CO2 + EAC supercritical solvents, the solubility (Y*) was estimated as the slope of a theoretical linear behavior of the OEC between t = 0 and t = 14 min (Table 4). For comparison,the estimated caffeine solubility in pure CO2 was 0.029% mass, which is quite in agreement with the experimental solubility measured at 343K and 28.1 MPa (0.023% mass) [35]. These estimated solubilities presume that, as in the case of the solubility of caffeine in the liquid solvents [30,31], the solubility of caffeine in CO2 with ethyl lactate cosolvent is higher than in the case of ethanol or ethyl acetate cosolvents. Global caffeine yield (Xo) was assessed on the basis of the maximum yield attained in each OEC; 1% above the total amount of extract for each experiment as reported in Table 4. The extractable

510 D.V. Bermejo et al. / J. of Supercritical Fluids 107 (2016) 507–512 Table 3 Caffeine recovered (mg of caffeine in the extract/mg of caffeine in green tea leaves × 100) at different extraction times corresponding to the static-SFE of green tea leaves using different cosolvents. Extraction conditions: 30 MPa, 343K, 210 min. t(min) CO2 t(min) CO2 + EL t(min) CO2 + ETOH t(min) CO2 + EAC 14 3.1 14 24.3 14 14.4 14 4.9 60 15.6 37 29.6 47 28.5 37 10.5 120 19.6 86 41.0 71 32.4 96 17.8 180 21.3 210 48.3 94 34.3 210 24.6 210 21.9 – – 148 38.1 – – – – – – 210 40.2 – – 0 2 4 6 8 10 12 14 0 50 100 150 200 250 yield (mg caffeine / g tea) time (min) 0 2 4 6 8 10 12 14 0 50 100 150 200 250 yield (mg caffeine / g tea) time (min) 0 2 4 6 8 10 12 14 0 50 100 150 200 250 yield (mg caffeine / g tea) time (min) 0 2 4 6 8 10 12 14 0 50 100 150 200 250 yield (mg caffeine / g tea) time (min) (a) (c) (d ) (b ) Fig. 1. Sovová’s model fitting of the overall extraction curves obtained in the static-SFE of caffeine from green tea leaves at 30 MPa and 343K. (a) CO2; (b) CO2 + EAC; (c) CO2 + ETOH; (d) CO2 + EL. Dashed lines: falling rate period. Table 4 Parameters of the model of Sovová corresponding to the green tea leaves static-SFE at 30 MPa, 343K and using different cosolvents. Ya (kg/kg) Xo (kg/kg) kYA (s−1) kXA (s−1) AARDa CO2 0.00029 0.0060 0.025 0.00026 6.67 CO2 + EAC 0.00043 0.0067 0.025 0.00013 3.72 CO2 + ETOH 0.00134 0.0109 0.045 0.00019 2.48 CO2 + EL 0.00220 0.0132 0.065 0.00017 8.92 a Absolute average relative deviation = 100 N |Yexp−Ycal| Yexp . caffeine (Xp) was considered an adjustable parameter, butthe same Xp/Xo ratio was considered for all OECs, since the same vegetal material (crushing and cell disruption) was employed and thus, it is presumed that the same percentage of caffeine is easy accessible in the constant rate period. Then, the Xp/Xo ratio and the mass transfer coefficients in the fluid and solid phase (kYA and kXA) were adjusted to reproduce the experimental OECs. A suitable Xp/Xo ratio resulted to be 0.2 and the optimal kYA and kXA were adjusted to each kinetic data set. The values are given in Table 4 together with the average absolute relative deviation between calculated and experimental caffeine yield. The optimal Xp/Xo ratio represents a Xk (amount of caffeine inside cell walls) around 4 times higher than Xp (readily accessible caffeine), denoting that caffeine is strongly bond in the vegetal

D.V. Bermejo et al. / J. of Supercritical Fluids 107 (2016) 507–512 511 matrix. Furthermore, this is in accordance with the fact that a very short constant extractionrateperiod was observedinthe extraction curves (tCER values were 0.96, 0.16 and 0.30 min, respectively, for EA, EL and ETOH) and thus, mass transport within the solid phase dominated extraction rate almost from the beginning of the process. In this respect, the solubility value (Y*) calculated as the slope of a theoretical linear behavior up to 14 min of extraction should be considered as an estimated value. The quality ofthe model regression can be observed in Fig. 1a–d, in which the falling extraction rate period (FER) is depicted with dashed lines. A comparison ofthe OECs corresponding to the different supercriticalfluids shows thatif no cosolvents are employed the extraction entered the FER period at 54.5 min. Using the cosolvents, the caffeine extraction rate is considerably higher at the beginning of the extraction and thus, the FER period started at 23.6 min for CO2 + EAC, 11.3 min for CO2 + ETOH and 7.5 min for CO2 + EL. The solid phase mass transfer coefficients (kXA) resulted rather similar in the fitting of all OECs (see Table 4). In general, kXA values were two orders of magnitude lower than kYA values, indicating a strong limitation of caffeine mass transfer in the solid phase. This limitation become evidentfrom the beginning ofthe extraction and thus, it is possible that the saturation of the supercritical phase was not attained. Furthermore, kXA values resulted reasonably similar for all cosolvents used, representing similar mass transfer in the solid phase. Regarding fluid phase mass transfer coefficient (kYA), values were 1.8 and 2.6 times higher for, respectively, CO2 + ETOH and CO2 + EL, with respect to pure CO2 or CO2 + EA. 4. Conclusions The SFE of green tea leaves was studied at 343K and 30 MPa, using pure CO2 and different green cosolvents for food processing. In comparison with ethyl acetate and ethanol cosolvents, ethyl lactate resulted in superior capacity for caffeine extraction. The highest yields of caffeine were obtained with ethyl lactate both in SFE-static mode (the cosolvent soaking the vegetal material before CO2 pumping) and SFE-dynamic mode (the mixture of CO2 + cosolvent was continuously pumped into the extraction cell). The general trend of cosolvent effect was ethyl lactate > ethanol > ethyl acetate, which corresponds with the behaviour observed in the pressurized liquid extraction of caffeine from green coffee beans [32]. The analysis of the overall extraction curves obtained in the SFE-static approach indicate that extraction velocity in the early extraction stages is around 7 times higher using CO2 + EL than with pure supercritical CO2. Consequently, the convective mass transfer coefficients resulted for the model of Sovová, indicate the highest value for the CO2 + EL supercritical solvent (kYA = 6.5 × 10−2 s−1) around 2.6 times higher than for supercritical CO2. Thus, ethyl lactate is a suitable green alternative to be employed as cosolventin SFE to remove caffeine from natural matrices, reducing the extraction time and/or the amount of CO2 employed. Acknowledgements D.Villanueva acknowledges the predoctoral contract(JAE) given by the Consejo Superior de Investigaciones Científicas (CSIC) of Spain. This work was financed thanks ALIBIRD, S2013/ABI-2728 (Comunidad de Madrid) project. Appendix A. The mass extracted (m) as a function of extraction time (t) is calculated according to the following equations: Constant extraction rate period: m = QY ∗ [1 − exp(−Z)]t (1) Falling extraction rate period: m = QY ∗ [t − tCER exp(Zw − Z)] (2) Diffusion controlled extraction rate period: m = mSI Xo − Y∗ W ln 1 + exp WXo Y∗ − 1 exp WQ(tCER − t) mSI Xk Xo (3) where: Z = mSIkYA Q(1 − )s (4) W = mSIkXA Q(1 − ) (5) ZW = ZY∗ WXo lnXo exp WQ(t − tCER)/mSI − Xk Xo − Xk (6) tCER = mSI(Xo − Xk) Y ∗ ZQ (7) mSI = XoF (8) Being the bed porosity, F the mass of solid loaded in the extraction cell, s the solid density, Q the CO2 mass flow rate, the CO2 density, Y* the solubility of the solute CO2 and Xo the global yield. Model parameters which are adjusted according to the experimental OEC are the intra-particle solute ratio (Xk) and the fluid phase and solid phase mass transfer coefficients (kYA and kXA). Eq. (7) defines the time at which the constant extraction rate ends (tCER). The time at which the falling extraction rate period begins (tFER) is given by: tFER = tCER + mSI WQ ln Xk + Xp exp WXo/Y∗ Xo (9) References [1] X. Wan, D. Li, Z. Zhang, Green tea and black tea. Manufacturing and consumption, in: C.T. Ho, J.K. Lin, F. Shahidi (Eds.), Tea and Tea Products. Chemistry and Health-promoting Properties, CRC Press, Boca Raton, USA, 2009, pp. 1–8. [2] U.H. Engelhardt, Chemistry of tea, in: L. Mander, H.B. Liu (Eds.), Comprehensive Natural Products II: Chemistry and Biology, 3, Elsevier Applied Science Publishers Ltd, London, 2010, pp. 999–1032. [3] T.R. Hartley, B.H. Sung, G.A. Pincomb, T.L. Whitsett, M.F. Wilson, W.R. Lovallo, Hypertension risk status and effect of caffeine on blood pressure, Hypertension 36 (2000) 137–141. [4] A. Nehlig, J. Daval, G. Debry, Caffeine and the central nervous system: mechanism of action, biochemical, metabolic and psychostimulant effects, Brain Res. Rev. 17 (1992) 139–170. [5] M. Giannellia, P. Doyle, E. Roman, M. Pelerin, C. Hermon, The effect of caffeine consumption and nausea on the risk of miscarriage, Paediatr. Perinat. Epidemiol. 17 (2003) 316–323. [6] E. Lynge, A. Anttila, K. Hemminki, Organic solvents and cancer, Cancer, Causes Control 8 (1997) 406–419. [7] A.B.A. de Azevedo, P. Mazzafera, R.S. Mohamed, S.A.B. Vieira de Melo, T.G. Kieckbusch, Extraction of caffeine, chlorogenic acids and lipids from green coffee beans using supercritical carbon dioxide and co-solvents, Braz. J. Chem. Eng. 25 (2008) 543–552. [8] R.S. Mohamed, M.D.A. Saldana, ˜ P. Mazzafera, Extraction of caffeine, theobromine, and cocoa butter from Brazilian cocoa beans using supercritical CO2 and ethane, Ind. Eng. Chem. Res. 41 (2002) 6751–6758. [9] C.B. Mehr, R.N. Biswal, J.L. Collins, Supercritical carbon dioxide extraction of caffeine from Guaraná, J. Supercrit. Fluids 9 (1996) 185–191. [10] M.D.A. Saldana, ˜ R.S. Mohamed, P. Mazzafera, Supercritical carbon dioxide extraction of methylxanthines from Maté tea leaves, Braz. J. Chem. Eng. 17 (2000) 251–260. [11] G.I. Burgos-Solórzano, J.F. Brennecke, M.A. Stadtherr, Solubility measurements and modeling of molecules of biological and pharmaceutical interest with supercritical CO2, Fluid Phase Equilib. 220 (2004) 57–69. [12] Zosel, Process for the decaffeination of coffee, US Patent 4247570.

512 D.V. Bermejo et al. / J. of Supercritical Fluids 107 (2016) 507–512 [13] S. Peter G. Brunner Process for decaffeinating coffee, US Patent 4322445. [14] S.N. Katz, J.E. Spence, M.J. OıBrien, ´ R.H. Skiff, G.J. Vogel, R. Prasad, Decaffeination of coffee, EU Patent 0424579B1. [15] O.G. Vitzthum, A.S. Clauis, N.A. El-Hag, V.N. Kapoor, Decaffeination of fermented unfired tea, EU Patent 0167399A2. [16] M.A. MacHugh, V.J. Krukonis, Supercritical Fluid Extraction, in: Principles and Practice, 2nd ed., Butterworth-Heinemann series in chemical engineering, Stoneham, USA, 1994, pp. 1–16. [17] H.S. Park, H.K. Choi, S.J. Lee, K.W. Park, S.G. Choi, K.H. Kim, Effect of mass transfer on the removal of caffeine from green tea by supercritical carbon dioxide, J. Supercrit. Fluids 42 (2007) 205, 212. [18] H.S. Park, H.J. Lee, M.H. Shin, K.W. Lee, H. Lee, Y.S. Kim, K.O. Kim, K.H. Kim, Effects of cosolvents on the decaffeination of green tea by supercritical carbon dioxide, Food Chem. 105 (2007) 1011–1017. [19] H.S. Park, N.G. Im, K.H. Kim, Extraction behaviors of caffeine and chlorophylls in supercritical decaffeination of green tea leaves, Food Sci. Technol. 45 (2012) 73–78. [20] W. Kim, J. Kim, J. Kim, S. Oh, Y. Lee, Selective caffeine removal from green tea using supercritical carbon dioxide extraction, J. Food Eng. 89 (2008) 303–309. [21] C.J. Chang, K.L. Chiu, Y.L. Chen, C.Y. Chang, Separation of catechins from green tea using carbon dioxide extraction, Food Chem. 68 (2000) 109–113. [22] Q.L. Sun, S. Hua, J.H. Ye, J.L. Lu, X.Q. Zheng, Y.R. Liang, Decaffeination of green tea by supercritical carbon dioxide, J. Med. Plants Res. 4 (2010) 1161–1168. [23] K. Huang, J. Wu, Y. Chiu, C. Lai, C. Chang, Designed polar cosolvent-modified supercritical CO2 removing caffeine from and retaining catechins in green tea powder using response surface methodology, J. Agric. Food Chem. 55 (2007) 9014–9020. [24] M.R. García Risco, G. Vicente, G. Reglero, T. Fornari, Fractionation of thyme (Thymus vulgaris L.) by supercritical fluid extraction and chromatography, J. Supercrit. Fluids 55 (2011) 949–954. [25] U. Kopcak, R.S. Mohamed, Caffeine solubility in supercritical carbon dioxide/co-solvent mixtures, J. Supercrit. Fluids 34 (2005) 209–214. [26] E. Lack, T. Gamse, R. Marr, Extraction from solids, in: A. Bertucco, G. Vetter (Eds.), High Pressure Process Technology: Fundamentals and Applications, Industrial Chemistry Library, 9, Elsevier, Amsterdam, The Netherlands, 2001, pp. 378–395. [27] M. Pelillo, B. Biguzzi, A. Bendini, T. GallinaToschi, M. Vanzini, G. Lercker, Preliminary investigation into development of HPLC with UV and MS-electrospray detection for the analysis of tea catechins, Food Chem. 78 (2002) 369–374. [28] Y. Iwai, H. Nagano, G.S. Lee, M. Uno, Y. Arai, Measurement of entrainer effects of water and ethanol on solubility of caffeine in supercritical carbon dioxide by FT-IR spectroscopy, J. Supercrit. Fluids 38 (2006) 312–318. [29] S. Aparicio, R. Alcalde, The green solvent ethyl lactate: an experimental and theoretical characterization, Green Chem. 11 (2009) 65–78. [30] M.S. Manic, D. Villanueva, T. Fornari, A.J. Queimada, E.A. Macedo, V. Najdanovic-Visak, Solubilities of high-value compounds in ethyl lactate: measurements and modelling, J. Chem. Thermodyn. 48 (2012) 93–100. [31] A. Shalmashi, F. Golmohammad, Solubility of caffeine in water ethyl acetate ethanol, carbon tetrachloride, methanol, chloroform, dichlorometane, and acetone between 298 and 323K, Latin Am. Appl. Res. 40 (2010) 283–285. [32] D. Villanueva Bermejo, P. Luna, M. Manic, V. Najdanovic-Visak, G. Reglero, T. Fornari, Extraction of caffeine from natural matter using a bio-renewable agrochemical solvent, Food Bioprod. Process. 91 (2013) 303–309. [33] H. Sovová, Rate of the vegetable oil extraction with supercritical CO2. I. Modeling of extraction curves, Chem. Eng. Sci. 49 (1994) 409–414. [34] National Institute of Standards and Technology (NIST). Available from: http:// webbook.nist.gov/chemistry/fluid/ [35] A.B.A. de Azevedo, T.G. Kieckbusch, A.K. Tashima, R.S. Mohamed, Supercritical CO2 recovery of caffeine from green coffee oil: new experimental solubility data and modeling, Quim. Nova 31 (2008) 1319–1323