SciELO - Scientific Electronic Library Online

vol.21 issue2A study of a wire-plate eletrostatic precipitator operating in the removal of polydispersed particlesA new approach to the joined estimation of the heat generated by a semicontiunuous emulsion polymerization Qr and the overall heat exchange parameter UA author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links


Brazilian Journal of Chemical Engineering

Print version ISSN 0104-6632On-line version ISSN 1678-4383

Braz. J. Chem. Eng. vol.21 no.2 São Paulo Apr./June 2004 



Use of reverse osmosis membranes for the separation of lemongrass essential oil and supercritical CO2



L.A.V.SarmentoI; R.A.F.MachadoII; A.BolzanII; C.B.SpricigoIII; J.C.C. PetrusII

ICentro Federal de Educação Tecnológica do Paraná, Av. Brasil 4232, CEP 85884-000, Phone +(55) (45) 264-2762, Fax +(55) (45) 264-4669, Medianeira - PR, Brazil, E-mail:
IIDepartamento de Engenharia Química e Engenharia de Alimentos, Universidade Federal de Santa Catarina, Florianópolis - SC, Brazil
IIICentro de Ciências Exatas e de Tecnologia, Pontifícia Universidade Católica do Paraná, Curitiba - PR, Brasil




Although it is still used very little by industry, the process of essential oil extraction from vegetable matrices with supercritical CO2 is regarded as a potentially viable technique. The operation of separating the extract from the solvent is carried out by reducing the pressure in the system. Separation by membranes is an alternative that offers lower energy consumption and easier operation than traditional methods of separation. Combining the processes essential oil extraction with supercritical CO2 and separation by membranes permits the separation of solvent and oil without the need for large variations in extraction conditions. This results in a large energy savings in the case of solvent repressurisation and reuse. In this study, the effectiveness of reverse osmosis membranes in separating lemongrass essential oil from mixtures with supercritical CO2 was tested. The effects of feed oil concentration and transmembrane pressure on CO2 permeate flux and oil retention were studied for three membrane models.

Keywords: supercritical extraction, reverse osmosis membranes, essential oils.




The extraction of substances from solid materials with supercritical carbon dioxide (SCCO2) occurs when the materials are brought into close contact with a flux of dense CO2 (Brunner, 1994). Generally, the solid material constitutes a fixed bed in the form of a column, over which the solvent is passed to extract and draw with it the solute which, up on leaving the extractor, is precipitated by simple expansion of the solvent. Since there is no solvent residue to contaminate the extract, the extraction of natural products with SCCO2 is considered to be an alternative to traditional extraction processes (Reverchon, 1992; Moyler, 1993). According to Reverchon (1994), a volatile mixture of terpenes and their derivatives is generally responsible for the aroma of vegetable material. The extraction of essential oils from flowers and leaves represents an attempt to obtain a mixture of components in which the original composition of the extract is maintained intact. Unfortunately, the techniques generally adopted, such as distillation and extraction with liquid solvent, have various limitations in relation to preservation of thermolabile compounds, permitting hydrolysis of part of the solute, and contamination of the product by the solvent.

Of the separation processes that employ membranes microfiltration (MF), ultrafiltration (UF), nanofiltration (NF), and reverse osmosis (IO) are prominent. In the field of separation technologies, processes involving membranes are frequently preferred in over classical processes for economic reasons. Many of these processes can be developed under moderate conditions of temperature and pressure. Scaling up is easy and energy consumption is generally low. Numerous studies have reported the use of membranes in the processes of separation of SCCO2 and a solute: ethanol (Semenova et al., 1992), light fractions of petroleum (Higashijima et al., 1995), fractionation of triglycerides in supercritical medium (Sarrade et al., 1999), and nutmeg essential oil (Spricigo et al., 2001), all achieving different degrees of separation in which a linear relationship was found between the permeate flux and the applied pressure differential.

Sartorelli and Brunner (2000) observed a characteristic common to many of the processes involving membranes that appears in some separations of SCCO2 and certain types of extract: high selectivity of the membranes is characterised by low permeate flux or high permeate flux combined with low selectivity. This results from the phenomenon known as plasticisation by CO2 (the CO2 causes the plasticisation of the membrane, which increases its flux, mainly when high pressures are used). Plasticisation causes ''swelling'' of the polymeric structure of the membrane through which a large quantity of SCCO2 as well as other components may pass, resulting in a poorly selective flux through the membrane. Even so, the association of the processes of separation by membranes and of supercritical extraction of essential oils from vegetable matrices with dense CO2 could be an alternative in order to reduce the costs of recompression and reuse of the solvent. Sarrade et al. (2002) present a review of supercritical fluid and membrane coupled processes, such as CO2 recovery by membranes in supercritical extraction processes, nanofiltration plus supercritical extraction processes applied to the purification of low molecular weight compounds, and supercritical fluid extraction assisted ultrafiltration.

The objective of this study was to investigate the use of commercial reverse osmosis membranes in the separation of lemongrass essential oil and SCCO2.




The experiments in this study were carried out in an experimental unit schematised in Figure 1. This diagram shows part of a pilot extraction and separation unit in which two cells, (6) and (7), encased in and constructed of stainless steel with a unit volume of 30 cm3 and an internal diameter of 2.3 cm were arranged in series. The membranes to be tested were positioned in one cell (7), whilst the other cell (6) contained quantities of essential oil corresponding to 5%, 10%, and 15% wt.% in relation to the CO2 that occupied the two cells. The temperature of the cells was maintained by a thermostatic bath (11). The membranes in the cell (6) were placed on a perforated metal support held in place by polytetrafluorethylene rings. The filtration area of the membranes was determined to be 3.14 cm2 and the dead-end flux regime was used during the experiments. The working pressure of the cells was monitored by a pressure transducer (5) (Model RTP12/BE53R, AEP, Italy) and controlled by a pneumatic valve (4) (Model 807, Badger Meter, USA). The working pressure in the surge tank (3) and in the cells was maintained by a booster (2) (Model DLE 15-1, Maxpro, Germany). The temperature in the surge tank was maintained the same as that of the cells through a thermostatic bath (11).




The reverse osmosis membranes denoted CG, SG, and AG, were purchased from Osmonics (USA). The average retention of NaCl, as indicated by the manufacturer, was 85, 98.2, and 99.4%, respectively.

Carbon Dioxide

Carbon dioxide (99.95% purity) was obtained from White Martins (Brazil) and used as solvent.

Essential Oil

The lemongrass essential oil was obtained in this laboratory through extraction of vegetable matrix with supercritical CO2 at 120 bar and 40ºC.

Experimental Procedure

The tests for characterisation of the membranes in relation to the flux of CO2 were carried out when valves (17) and (14) were closed and valves (15) and (16) were open to produce equal pressurisation between the two faces of the membrane. After reaching equilibrium pressure, valve (16) was closed and the pressure differential across the membrane was applied using a back pressure valve (9). The pressure differential was monitored by a pressure transducer (18), and the release of CO2 was obtained via a flow meter (13). Transmembrane pressure differentials (DP) of 10, 20, 30, and 40 bar were applied and the corresponding CO2 flux for each DP was measured until a constant value was obtained. The working pressure was maintained at 120 bar by a booster (2), monitored by a pressure transducer (5) and controlled by a pneumatic valve (4).

The tests for characterisation of the membranes in relation to their essential oil retention capacities and permeability to CO2 were carried out with the equipment used in the pure CO2 flux tests. Quantities of essential oil sufficient to maintain the concentrations of 5, 10, and 15% wt.% were placed in cell (6). Valve (15) was closed and valves (14), (16), and (17) were open. Once a pressure equilibrium had been established between the faces of the membranes, valve (16) was closed, and similarly to the procedure for the tests of pure CO2 flux, the pressure differentials were established by the back pressure valve (9) and monitored by the transducer (18). The quantities of essential oil permeate were measured by the mass collected in the separator (10). These measurements were taken until a constant flux had been achieved through the membranes. The concentration of essential oil retained by the membrane in the cell (7) was determined by the mass of oil collected in the separator (10) at intervals of approximately 30 minutes through periodic opening of the micrometer valve (8).

All the experiments were carried out at a feed pressure of 120 bar and 40°C. These conditions were used for the extraction of the essential oil that was employed in the experiments designed for the characterisation tests for membrane permeability and retention. Under these conditions, even after establishing a pressure differential of 40 bar, the solvent was found to be in conditions of pressure and temperature above its critical state (critical pressure = 73.8 bar and critical temperature = 31.06°C) on both sides of the membrane.

Analytical Procedure

Composition of the oil was determined by gas chromatography-mass spectrometry (HP 5971 MSD, Hewlett-Packard, USA) with a fused silica capillary column DB-5 (20 m x 0.188 mm internal diameter, 0.4 mm film thickness). The injector temperature was 280ºC and the flame ionization detector (HP 5890, Hewlett-Packard, USA) was maintained at 320ºC. Oven temperature increased at a rate of 3ºC/min from 60 to 310ºC. The sample volume injected was 0.15 ml of a 10% solution in CH2Cl2, with a split rate of 1:50. The carrier gas was hydrogen at a flow rate of 1 ml/min.



Figure 2 presents the results of the characterisation tests for the SG, CG, and AG membranes in relation to flux of pure CO2 at a feed pressure of 120 bar and transmembrane pressure differentials of 10, 20, 30, and 40 bar. The results demonstrate a linear variation in flux with the pressure differential and indicate that the AG membrane had the best flux rates with the lowest pressure differentials.



The permeabilities of the membranes, determined by the slope of the flux curves with the pressure differential in Figure 2, were 2.158, 1.048, and 3.048 kg h-1 m-2 bar-1, respectively, for the SG, CG, and AG models.

In Table 1 a comparison between the CO2 permeate fluxes arising from the CO2 and essential oil mixture, through the membranes tested at a feed pressure of 120 bar and transmembrane pressure differentials of 10, 20, 30, and 40 bar is presented From this table, it can be seen that the CG membrane had the highest flux rates with the lowest pressure differentials.



The retention tests were carried out by varying the concentration of essential oil in the feed as well as the applied pressure differential. The retention index (RI) was determined by Equation (1):

where Cp is the permeate concentration of essential oil and Ci is the feed concentration of essential oil.

In Table 2 a comparison of the essential oil retention indexes for the membranes tested, when exposed to pressure differentials of 10 to 40 bar and feed oil concentrations of 5 to 15% wt.% is presented. The results in Table 2 indicate that the SG membrane showed the best retention index at small pressure differentials, thus being the most suitable for the separation of lemongrass essential oil from supercritical CO2, even though the fluxes maintained by this membrane were lower than those observed with the CG membrane. The latter, in turn, had the lowest retention index for lower pressure differentials.



Fractionation of the essential oil did not occur with passage through the membranes, and Table 3 presents the lemongrass essential oil composition as determined by gas chromatography-mass spectrometry.



The retention of the essential oil by the membranes was affected only by the pressure differential and did not show significant variation with the concentration of oil in the feed, as can be seen in Table 2. This indicates that, even when increasing the inlet essential oil concentration, membrane selectivity is not altered. The same results were obtained by Spricigo et al. (2001).

Figure 3 shows the behaviour of the SG membrane in relation to the flux of dense CO2 before and after permeation by essential oil. The results indicate a reduction in the flux that may be the result of the formation of a film of oil over the filter face of the membranes, causing a greater resistance to the mass flux. Alternatively, it may have been caused by fouling, a common phenomenon when using a reverse osmosis membrane in dead-end filtration regime. The permeability of the SG membrane to pure CO2 was 2.158 kg h-1 m-2 bar-1. During the tests with 5, 10, and 15% wt.% essential oil in the feed, the values were 0.127, 0.167, and 0.181 kg h-1 m-2 bar-1. There was no significant influence of feed oil concentration on membrane permeability. After permeation of the oil, the permeability was 0.903 kg h-1 m-2 bar-1, indicating a low degree of reversibility in the processes of resistance to mass flux that act upon the membranes.



In the tests of flux and of permeability to supercritical CO2 and to the solvent-solute mixture, the pressure differential was increased from 10 to 40 bar, causing a linear variation in the flux before being reduced once again to the initial value of 10 bar. For each DP, the fluxes of pure CO2 or of CO2 arising from a mixture of CO2 and essential oil were determined for at least 3 hours, until stationary values were achieved.

In Figure 4 the results for the flux of pure CO2 during compression and decompression of the SG membrane are presented. Hysteresis was not observed, showing that there was not a significant influence of membrane compaction on permeate flux after submission to a transmembrane pressure of 40 bar.




The results presented in this study demonstrate the possibility of the use of commercial reverse osmosis membranes for the separation of lemongrass essential oil and supercritical CO2 in the place of the commonly used depressurization step.

The SG membrane showed good resistance to the severe pressure conditions, resulting in a retention index of up to 88% of the essential oil, maintaining an average flux of 5.75 kg h-1 m-2 and an average permeability of 0.16 kg h-1 m-2 bar-1 for a difference in pressure of 10 bar. Under these conditions a large reduction in energy consumption is possible for the recompression and reuse of CO2 as solvent, since it is not flashed to the gas phase, but rather is maintained as a supercritical fluid. Also, under these conditions the behaviour of the SG membrane was best with regard to the essential oil retention index.

Although the process of coupling membranes to supercritical extraction has been still, studied very little it was shown to have potential, on account of the resulting economy of extraction, mainly for products of high aggregate value.



Brunner, G., Gas Extraction. An Introduction to Fundamentals of Supercritical Fluids and the Application to Separation Processes. Steinkopoff Darmstadt Springer – New York (1994).        [ Links ]

Higashijima, T., Ohya, H., Tsuchiya, Y., Tokunaga, H., Aibara, M., and Negish, Y., Separation of Supercritical Mixtures of CO2 and Petroleum Components with an Asymmetric Polyimide Membrane. J. Membr. Sci., 93, 165 (1995).         [ Links ]

Moyler, D.A., Extraction of Essential Oil with Carbon Dioxide. Flavour and Frag. J., 8, 235 (1993).        [ Links ]

Reverchon, E., Fractional Separation of SCF Extracts from Marjoram Leaves: Mass Transfer and Optimization. J of Supercr. Fluids., 5, 256 (1992).         [ Links ]

Reverchon, E. and Osséo, L.S., Modeling the Supercritical Extraction of Basil Oil. Proceedings of the 3rd International Symposium on Supercritical Fluids, Perrut and Brunner, 2, 189 (1994).         [ Links ]

Sarrade, S., Carlés, M., Perre, C., and Vignet, P., Process and Installation for the Separation of Heavy and Light Compounds by Extraction Using a Supercritical Fluid and Nanofiltration, U.S. Patent 5,961,835 (1999).         [ Links ]

Sarrade, S., Guizard, C., and Rios, G.M., Membrane Technology and Supercritical Fluids: Chemical Engineering for Coupled Processes. J. Membr. Sci., 144, 137 (2002).        [ Links ]

Sartorelli, L., and Brunner, G., Separation of Extracts from Supercritical Carbon Dioxide by Means of Membranes, In: Proceedings of the 5th International Symposium on Supercritical Fluids, Atlanta, 2000.         [ Links ]

Semenova, S.I., Ohya, H., Higashijima, T., and Negishi, Y., Separation of Supercritical CO2 and Ethanol Mixtures with Asymmetric Polyimide Membrane. J. Membr. Sci., 74, 131 (1992).         [ Links ]

Spricigo, C.B., Bolzan, A., Machado, R.F.A., Carlson, L.H.C., and Petrus, J.C.C., Separation of Nutmeg Essential Oil and Dense CO2 with a Cellulose Acetate Reverse Osmosis Membrane. J. Membr. Sci. 188, 173 (2001).         [ Links ]



Received: February 13, 2003;
Accepted: January 5, 2004



* To whom correspondence should be addressed

Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License