Recovery of Wine-Must Aroma Compounds by Supercritical CO 2

Food Bioprocess Technol (2008) 1:74 81 DOI 10.1007/s11947-007-0002-5 Recovery of Wine-Must Aroma Compounds by Supercritical CO 2 Sofia Macedo & Susana Fernandes & José A. Lopes & Hermínio C. de Sousa & Paulo J. Pereira & Paulo J. Carmelo & Carlos Menduiña & Pedro C. Simões & Manuel Nunes da Ponte Published online: 25 July 2007 # Springer Science + Business Media, LLC 2007 Abstract The fractionation of wine musts (the fermentation broth that leads from grape juice to wine) of highly aromatic grape varieties was performed with supercritical carbon dioxide at pressures between 11 and 18 MPa and temperatures between 308 and 318 K. Different configurations of the separation devices were tested after the extraction step to improve the fractionation process. Headspace chromatography and Karl Fisher analyses were used to measure the composition of the extracts. A significant enrichment of the supercritical extracts on aromatic compounds was observed. The recovery of the aromatic compounds from the high pressure solvent stream poses technical problems that must be solved before the process can be feasible in an industrial scale. Keywords Supercritical extraction. Carbon dioxide. Wine musts. Aromas S. Macedo : S. Fernandes : J. A. Lopes : P. J. Pereira : P. J. Carmelo : P. C. Simões (*) : M. Nunes da Ponte REQUIMTE, Departamento de Química, Universidade Nova de Lisboa, 2829516 Caparica, Portugal e-mail: pcs@dq.fct.unl.pt H. C. de Sousa Departamento de Engenharia Química, FCTUC, Pólo II Pinhal de Marrocos, 3030290 Coimbra, Portugal C. Menduiña Departamento de Química Física, Facultad de Ciencias Químicas, Universidad Complutense de Madrid, Ciudad Universitaria, 28040 Madrid, Spain Introduction Supercritical fluid extraction (SFE) uses fluids at supercritical conditions to selectively extract substances from solid or liquid mixtures. The extraction of valuable products by means of supercritical carbon dioxide (SCCO 2 ), for example, can be an attractive alternative to conventional extraction with solvents. It is, therefore, a key enabling technology for clean chemistry, for at the end of the extraction process, one is left with recyclable CO 2 and the desired product, and no organic solvent waste. Much research (Brunner 1994; Bertucco and Vetter 2001) has been carried out in this area in recent years, leading to large-scale applications, including the decaffeination of coffee and tea, hops extraction, fatty acid alkyl esters fractionation from vegetable oils, and SCCO 2 removal of trichloroanisole from cork stoppers. Interest on the application of compressed carbon dioxide to the recovery of aroma compounds from fermented beverages is more than 30 years old (Schultz and Randall 1970; Schultz et al. 1974; Medina and Martinez 1994; Perrut and Nunes da Ponte 1996, 1997; Gamse et al. 1999; Senorans et al. 2001, 2003). The use of liquid carbon dioxide to extract aromatic compounds from wines was first proposed in 1981 by Jolly (1981). An Australian dry white wine was processed with liquid carbon dioxide (6.0 to 6.5 MPa, 296 K) in a Scheibel extraction column. An extract containing 45% v/v of ethanol and practically all the flavor components of the wine was obtained. Perrut and Nunes da Ponte (1996) applied SCCO 2 to the extraction of aromas from red wine and coupled the extractor with several separators in series to optimize the aroma fractionation process. Later, Del Re and Di Giacomo (1998) studied the continuous fractionation of wine with SCCO 2 at a pilot plant scale. Finally, Senorans et al. (2003) investigated the

Food Bioprocess Technol (2008) 1:74 81 75 recovery of brandy aroma extracts by countercurrent SFE. Several patents were registered on this subject, dealing with liquid or supercritical carbon dioxide extraction of wines (Berger et al. 1981; Kolb et al. 1989). So far, however, no successful application at industrial scale is known. The recovery of aromas from wine musts has been scarcely studied. Yet, it is during the fermentation process that the main part of the wine aromas is produced. A meaningful proportion of these aromas are lost to the atmosphere because of their inherent volatility and to the entrainment with evolved carbon dioxide. Many compounds thus lost have positive sensory attributes and may be appropriate to produce aromatic concentrates. Fernandes et al. (1994) have carried out a preliminary work where samples from grape-must fermentation broths were taken at regular intervals during the fermentation process and later extracted by SCCO 2 at 18 MPa and 313 K. A very important enrichment of the aromatic compounds appeared in the extracts in comparison to the original must. Moreover, the proportions of the major components present in the extracts changed as the fermentation progresses. The concentration of the two main compounds identified in the SCCO 2 extracts (besides ethanol), isoamyl alcohol and phenylethanol, seemed to be correlated as the concentration sum remained constant during the fermentation, the first decreasing while the second increased. In this paper, we intend to define the best operating conditions for supercritical dearomatization process of wine musts with the purpose of obtaining high-quality aromatic extracts. Material and Methods This work was divided in two parts. First, batch extraction of wine musts was accomplished in a small-scale apparatus suitable for high-pressure liquid-vapor equilibrium measurements. In the second part of the work, a set of experiments were carried out in a continuous countercurrent packed column. Materials The batch extraction tests were performed with a Loureiro Camarate wine must (José Maria da Fonseca Sucrs., Portugal). The continuous countercurrent extraction experiments were carried out with a Muscatel wine must and a Muscatel wine (José Maria da Fonseca Sucrs., Portugal). In the case of the wine musts, samples were taken from an industrial fermentation tank and processed immediately. When prolonged sample storage (up to 24 h) was inevitable, samples were kept frozen at 25 C in a commercially available plastic bottle (food proof) and defrosted at 4 C before the experiment. Samples to be stored for a few hours only were stored at 4 C until needed. Carbon dioxide was supplied by Air Liquide with a purity of 99.995%. Batch Extraction of Wine Musts The batch SCCO 2 extraction experiments were carried out in a laboratory-scale high-pressure extraction plant. The apparatus consisted of a 100-ml capacity stainless steel extraction vessel (designed and made at Universidade Nova de Lisboa) located in a thermostatized bath. In average, ca. 40 to 50 ml of must was introduced in the extraction cell per extraction. Details of the batch apparatus can be found elsewhere (Fernandes et al. 1994). The extraction was performed by continuously passing a relatively constant flow rate (1 g/min) of CO 2 through the high-pressure cell. The carbon dioxide + extract mixtures were decompressed into a cold trap through a high pressure valve. The extracts were collected in headspace chromatography glass vessels, half-filled with glass beads to increase the surface area available for condensation from the CO 2 stream. Headspace chromatography was used to characterize the extracts and to compare them with the original musts. This type of chromatography is especially well suited to the analysis of organic volatiles in liquid and gas samples (Reis Machado et al. 1993; Noble et al. 1980). The experiments were run at the temperatures of 308, 313, and 318 K and in a pressure range of 11 to 18 MPa. These conditions were chosen, taking into consideration work previously done on the fractionation of fermented or distilled beverages by SCCO 2 (Medina and Martinez 1994; Perrut and Nunes da Ponte 1996, 1997). Continuous Countercurrent Extraction of Wine Musts The continuous extraction experiments were performed in a pilot-scale plant for SFE of liquid mixtures (designed and made at Universidade Nova de Lisboa). A schematic diagram of the apparatus is shown in Fig. 1. The original apparatus is described in detail elsewhere (Simoes et al. 1995). The main element of the apparatus is the highpressure extraction column, EC, with an internal diameter of 2.4 cm and 1-m height. The column is packed with CY Laboratory Sulzer gauze packing. The surface area of the packing is 890 m 2 /m 3 with a void fraction of 0.90. The column is operated in a countercurrent way, with the liquid feed entering at the top of the column, by means of a Lewa (model ELM 1V) high-pressure metering pump, MP L, and the less dense carbon dioxide at the bottom. The solvent stream is pumped by a diaphragm compressor C (Nova Swiss, model 5542121) to the desired extraction pressure and preheated before entering the extraction column by

76 Food Bioprocess Technol (2008) 1:74 81 Fig. 1 Flowsheet of the supercritical fluid extraction apparatus. EC Extraction column, SC separation column, HE heat exchangers, PG pressure gauges, C gas compressor, G gas cylinder, MP L liquid feed metering pump, MFM mass flow meters flowing through a heat exchanger HE. Contact between the two fluid phases is achieved throughout the whole packed bed of the extraction column; the components of the liquid feed with higher solubility in SCCO 2 are preferentially extracted into the solvent-rich fluid phase. The extract phase, with the dissolved solutes, leaves the extraction column by the top. The remaining raffinate, with a considerable amount of solubilized carbon dioxide, is drawn off from the bottom of the column. The separation of the solubilized compounds from carbon dioxide and the consecutive regeneration of the solvent are achieved by expanding the extract phase into a second stainless steel column, labeled as separation column, SC, and packed with the same packing as the extraction column. This apparatus was modified for the fractionation of the wine musts. The main modifications were associated with the regeneration of the solvent and the recovery of the solutes. A high-pressure cyclone, CYC, was installed after the SC to improve the recovery efficiency of the aromatic fraction by fractionating the extract stream. In addition, a filter was installed in the gas recirculating stream before the gas compressor, consisting of several packing sections of molecular sieves, charcoal, and silica gel. In this way, it guaranteed the complete regeneration of the carbon dioxide stream before reentering the extraction column. An improved trap system was also designed to recover the extract samples from the bottom of the SC and of the cyclone. Samples were withdrawn through a capillary 1/16-in. tube directly to a glass trap immersed in a cooling bath where condensation of the aromatic compounds took place. Samples of the feed, extract, and raffinate streams were analyzed by headspace chromatography; the ethanol and water contents of the extract and raffinate samples were also analyzed by high-performance liquid chromatography (HPLC) and Karl Fisher titration, respectively. Some of the extracts collected at the bottom of the high-pressure cyclone and of the SC were evaluated as to their organoleptic quality by a sensory panel of wine tasters (José Maria da Fonseca Sucrs., Portugal). The extraction conditions were the same for both feed supplies (the Muscatel wine must and the Muscatel wine): 18 MPa/313 K for the extraction column, 8 MPa/315 K for the SC (hereafter named as first separator), and 6.5 MPa/ 301 K for the cyclone (hereafter named as second separator). For the Muscatel wine case, a set of additional extraction experiments were performed where the solventto-feed flow ratio and the solvent density were changed to evaluate their influence on the aromatic fraction of extracts. The respective operating conditions are shown in Table 1. Chromatographic Analysis of Samples Headspace Chromatography Samples were analyzed by a gas chromatograph (Hewlett- Packard 5890, series II) connected to an automatic static headspace sampler (Hewlett-Packard 19395A) and equipped with an Chrompack FFAP capillary column (25 m 0.32 mm, 0.25-μm film thickness). The analytical conditions were as follows: injector temperature, 200 C;

Food Bioprocess Technol (2008) 1:74 81 77 Table 1 Extraction conditions for the muscatel wine experiments Run Extractor First separator Second separator S/F P (MPa) T ( C) P (MPa) T ( C) P (MPa) T ( C) Ratio a 1 13.5 43 7.7 47 6.3 28 14 2 15.9 39 8.9 47 6.0 28.5 5 3 13.1 41 7.6 46 5.5 27 5 4 10.9 42 7.6 44 5.0 26 2 a Solvent-to-feed mass flow ratio FID detector temperature, 230 C; initial column temperature, 35 C (held for 5 min), with an oven temperature rise rate of 10 C/min until 220 C. The carrier gas used was nitrogen, the column head pressure was 45 kpa, and the split flux was 20 ml/min. The samples were collected in headspace chromatography glass vessels and immediately sealed with crimp top vial caps and septa and placed in the headspace bath, kept at a constant temperature of 80 C. The equilibration time of the samples was 50 min. Injection time was 1 min. ethyl caprate (C 10 ), phenylethyl acetate, and 2-phenylethanol. The widest peak at short retention times is ethanol, the dominant substance in the headspace chromatograms of both musts and extracts. As shown clearly in Fig. 2, there is HPLC Analysis In the case of the continuous countercurrent extraction experiments, the ethanol content of the raffinate and extract samples was analyzed by HPLC (MerckHitachi) with an Aminex HPX87H ion-exchange column maintained at 60 C (300 7.8 mm, BioRad Laboratories, Amadora, Portugal) and an IR detector (Merck, Germany). The mobile phase was H 2 SO 4, 0.01 N, and the flow rate was 1 ml/min. Results and Discussion Batch Experiments The main purpose of this part of the work was to evaluate the influence of the extraction conditions pressure and temperature in the aroma composition of the extracts. In Fig. 2, the headspace chromatograms of the original wine must and of the corresponding SCCO 2 extract at 313 K and 18 MPa are compared. A great number of peaks appear at similar retention times, corresponding to the same compounds on both the must and the extract, but in strikingly different amounts. Identification of some of the most important peaks was attempted by examination of retention times of samples of known substances, injected under the same conditions as the supercritical extract sample. The following six major compounds were identified in the high pressure extracts (presented by increasing order of retention time): 3-methyl-1-butanol (isoamyl alcohol), ethyl caproate (C 6 ), ethyl caprylate (C 8 ), and Fig. 2 Headspace chromatograms of the original wine must (a) and the corresponding extract obtained by supercritical CO 2 at 313 K and 18 MPa (b). The large peak on the left is due to ethanol. The arrows indicate isoamyl alcohol (left) and 2-phenylethanol (right)

78 Food Bioprocess Technol (2008) 1:74 81 a very important enrichment of the extract in the less volatile compounds (corresponding to the higher retention times) and a relative decrease of the ethanol quantity. Although a quantitative evaluation of the aroma profile of the extracts was not possible to carry out with the present headspace chromatographic method, a general trend could be established with the extraction conditions. A comparison of the extract chromatograms obtained at a constant temperature of 313 K indicates that ethanol and isoamyl alcohol are the main components but their content decreases with an increase of the pressure. In Fig. 3, we present the peak area percentage (in an ethanol free basis) of each identified component as a function of the pressure at 313 K. The concentration of the less volatile aromas in the extracts increases with the pressure in detriment of the two main compounds. The influence of the operating conditions in the selectivity of supercritical carbon dioxide can be assessed by calculating the separation factors between the aroma compounds and ethanol from the relative areas in the extract and in the raffinate liquid chromatograms. The separation factor between aroma i and ethanol is defined by the following equation: ð α ¼ A aroma i =A ethanol Þin extract ð1þ ða aromai =A ethanol Þin raffinate where A stands for the peak area (in percentage) of each substance. Figure 4 shows the calculated separation factors as a function of the pressure at 313 K (Fig. 4a) and as a function of the temperature at 16 MPa (Fig. 4b). By increasing the extraction pressure from 14 to 16 MPa at a constant temperature of 313 K, the separation factor for isoamyl alcohol decreases from 17 to 10, whereas the separation factors for the less volatile aromas ethyl caprylate and ethyl caprate increase from 0.7 to 5 and 0.1 to 2, respectively. The use of higher temperatures seem to favor the more volatile components; at 16 MPa, for the temperatures of 313 and 318 K, the separation factor for 3-methyl-1-butanol is 8 in both conditions, whereas for phenylethanol, the respective separation factor is 18 and 12.5, respectively. This follows the usual trend of supercritical fluid solubility against selectivity on fractionation of aromas from natural raw materials (Brunner 2005; Reverchon and De Marco 2006): higher densities of the solvent (i.e., higher pressures and/or lower temperatures) favor the selectivity of less volatile components. Countercurrent Extraction Experiments A series of extraction experiments were performed with the SFE countercurrent column. An initial series of runs were made with the original configuration of the SFE apparatus, that is, with just one separation step (the regeneration packed column). With this configuration, the majority of the aromas extracted from the wine must feed were not efficiently recovered at the bottom of the SC. The recovery system was, therefore, improved by inserting a high-pressure cyclone after the SC and by using a new trap system. The water and ethanol content of the supercritical extracts obtained from the Muscatel must and wine feeds are presented in Fig. 5. Water and ethanol compositions are expressed in relative weight percentages, as calculated from the Karl Fisher and HPLC analyses, respectively. The results clearly show that the inclusion of the cyclone in the Fig. 3 Relative peak-area percentages of aromatic compounds in supercritical batch extracts as a function of the pressure

Food Bioprocess Technol (2008) 1:74 81 79 Fig. 4 Separation factor between aromatic compound i and ethanol as a function of the pressure at 313 K (a) and as a function of the temperature at 16 MPa (b) SFE apparatus led to a fractionation of the extracted compounds: Water was mainly collected in the first separator and ethanol in the second separator. The ethanol content of the Muscatel wine feed was reduced by half, indicating that supercritical carbon dioxide can be a feasible solvent for a wine dealcoholization process (Perrut and Nunes da Ponte 1997; Gamse et al. 1999). The influence of the extraction pressure on the extraction yield of aromas and the fractionation efficiency was evaluated for the Muscatel wine feed. Figure 6 presents the peak area percentages (in an ethanol-free basis) of five identified aromatic compounds at 13.5 MPa and 316 K on the two separation vessels (graph on the left), and as the ratio of peak area percentages of the aromatic compounds

80 Food Bioprocess Technol (2008) 1:74 81 Fig. 5 Water and ethanol content of muscatel wine and must SFE extracts in the first separator over the second separator at 11.0, 13.5, and 16.0 MPa (graph on the right). Only at the lowest extraction pressure condition, the fractionation of the aromatic compounds occur in an appreciated extent. With higher extraction pressures, the major part of the aromas were recovered in the first separator; the total amount of aromatic compounds, and in particular, the less volatile aromas, decreased. According with these results, it seems preferable to use lower extraction pressures to increase the selectivity of carbon dioxide towards the aromas and the fractionation efficiency of the aromas between the two separators. A wine tasting panel proved the samples collected in the set of experiments carried out in this work. With few exceptions, the majority of the extract samples did not present a remarkable flavor that suggests that the recovery system is still inefficient. Another conclusion from the tasting panel was that the temperature of the separation units should be kept at a lower temperature, ca. 303 K, to avoid aroma degradation as some of the samples had a burning taste, which was attributed to a relatively high temperature condition in the two separators. Conclusions The possibility of extract aromatic compounds from wine musts by SFE was assessed in this work. Batch high pressure experiments were carried out where significant enrichment Fig. 6 Relative peak-area percentages of aromatic compounds in muscatel wine SFE extracts

Food Bioprocess Technol (2008) 1:74 81 81 of the supercritical extracts on aromatic compounds was observed. Subsequent high-pressure experiments in a countercurrent SFE column were carried out with muscatel wine and must feeds. The composition of the supercritical extracts confirmed the previous batch experiments; yet, the recovery of the aromatic compounds from the high-pressure solvent stream poses significant technical problems that must be solved before this process can be feasible in an industrial scale. Acknowledgment The authors thank José Maria da Fonseca and Sucrs., Portugal, for providing the wine must samples and the sensory panel evaluations. References Berger F, Sagi F, Cerles B. Procédé d Extraction de l Arôme de Boissons Alcoolisées obtenues à partir de Fruits ou de Produits Assimilés. European patent no. 0129459, Rhône-Poulenc; 1981. Bertucco A, Vetter G, editors. High pressure process technology: Fundamentals and applications. Industrial Chemistry Library, volume 9. Elsevier, The Netherlands; 2001. Brunner G. Gas extraction. Springer, Berlin, Germany; 1994. Brunner G. Supercritical fluids: technology and application to food processing. J Food Eng 2005;67:2133. Del Re G, Di Giacomo G. Continuous fractionation of wine with dense carbon dioxide. In: Perrut M, Subra P, editors. Proceedings of the 5th meeting on supercritical fluids 1998; 2:425430. Nice, France, ISBN 2905267283. Fernandes S, Lopes JA, Reis Machado AS, Nunes da Ponte M, Reves JC. Supercritical CO 2 extraction from wine fermentation broths. In: Brunner G, Perrut M, editors. Proceedings of the 3rd international symposium on supercritical fluids, October 1994; 2:491495. Strasbourg, France, ISBN 2905267238. Gamse T, Rogler I, Marr R Supercritical CO2 extraction for utilization of excess wine of poor quality. J Supercrit Fluids (1999);14:1238. Jolly DRP. Wine flavour extraction with liquid carbon dioxide. Process Biochemistry, Aug./Sept., 3640, (1981). Kolb E, Marr R, Schildmann, JA, Weisrock, R, Wiesenberger, A. Process for producing alcoholreduced or alcoholfree beverages made by natural fermentation. US Patent No 4867997, (1989). Medina I, Martinez JL. Dealcoholation of Cider By Supercritical Extraction With Carbon Dioxide. In: Brunner G, Perrut M (eds) Proceedings of the 3rd International Symposium on Supercritical Fluids, October 1994; 2:401406. Strasbourg, France. ISBN 2905267238. Noble AC, Flath RA, Forrey RR. Wine headspace analysis. Reproducibility and application to varietal classification. J Agric Food Chem (1980);28:34653. Perrut M, Nunes da Ponte M. Fraccionement LiquidFluide Application A L extraction Supercritique Des Arômes De Boissons Fermentees Et Distillees. In: Pellerin P & Perrut M (eds) Proceedings of the 3eme Colloque sur les Fluides Supercritiques, Grasse, France; 1996, pp 4560. ISBN 2905267259. Perrut M, Nunes da Ponte M. Aromas from fermented and distilled beverages by liquid fluid fractionation. In: Proceedings of the fourth international symposium on supercritical fluids 1997; C:845852. Sendai, Japan. Reis Machado AS, Gomes de Azevedo E, Sardinha RMA, Nunes da Ponte M. High pressure carbon dioxide extraction from geranium plants. J Essent Oil Res 1993;5:185. Reverchon E, De Marco I. Supercritical fluid extraction and fractionation of natural matter. J Supercrit Fluids 2006;38:14666 Senorans FJ, RuizRodriguez A, Ibañez E, Tabera J, Reglero G. Optimization of countercurrent supercritical fluid extraction conditions for spirits fractionation. J Supercrit Fluids 2001;21:4149. Senorans FJ, RuizRodríguez A, Ibánez E, Tabera J, Reglero G. Isolation of brandy aroma by countercurrent supercritical fluid extraction. J Supercrit Fluids 2003;26:12935. Schultz WG, Randall JM. Liquid carbon dioxide for selective aroma extraction. Food Technol 1970;24:12826 Schultz WG, Schultz TH, Carlson RA, Hudson JS. Pilot plant extraction with liquid CO 2. Food Technol 1974;28:32. Simoes PC, Matos HA, Carmelo PJ, Gomes de Azevedo E, Nunes da Ponte M. Mass transfer in countercurrent packed columns application to supercritical CO 2 extraction of terpenes. Ind Eng Chem Res 1995;34:6138.