2.1 Plant material and chemicals

The herbs: sage (Salvia officinalis), basil (Ocimum basilicum), oregano (Origanum vulgare), lovage (Levisticum officinale) were grown in Estonia and collected in July 2002. The raw material was dried at room temperature (approximately 22 °C) and protected from direct light.

The carbon dioxide (Eesti AGA, Estonia) used in the experiments was 99.5% (w/w) pure. The ethanol as modifier used was spectroscopic grade (Liviko, Estonia) and added according to weight of the sample (from 2.5 to 7%).

2.2 Extraction of herbs

Supercritical fluid extractions from dried leaves or/and flowers of plants were performed on Sample Preparation Accessory apparatus (Milton Roy SPA, USA). Extractions were carried out at 45 °C, at pressures from 17.2 to 25.5 MPa. The flow rate of supercritical CO₂ was 1 ml min^–1 and extraction time for all experiments was 60 min. The extract was collected in chloroform. Extraction time (1 h) was chosen on the basis of previous studies [11]. There was estimated, that at this pressure range used the yield was ~85% of total yield able to achieve during 4-h extraction.

For comparison, a conventional steam distillation using a Clevenger apparatus for 3 h was utilized for the isolation of the essential oils. After steam distillation, the essential oil was isolated and kept refrigerated.

2.3 Gas chromatography (GC)

Chemical composition of extracts was determination by gas-liquid chromatography method using a fused silica capillary column with a bonded stationary phase NB 30 (HNU Nordion, Finland), and flame ionisation detector. The identification was performed by comparing their Kovats relative retention indexes (RI) with those of authentic compounds and from database collected from literature.

2.4 Assay for antimicrobial activity

The microorganisms: Bacillus mesentericus, Staphylococcus albus, Escherichia coli, Aspergillus flavus and Penicillium sp. were used in this study. Laboratory strains, kept on the solid agar at 4 °C, were used. Antimicrobial activity was tested by the filter paper disc diffusion method [12]. A 5.5-mm sterile paper disc was impregnated with a test material and set on agar using a micropipette. Plates were then inverted and incubated for bacteria – at 30 °C for 48 h, for moulds – at 25 °C for 96 h. Following incubation, the zones of inhibition were measured (in mm s).

3.1 Effect of pressure on the extract yield

Preferable extraction conditions were 17.2 MPa for sage and lovage, and 25.5 MPa for basil and oregano (Table 1). The total maximum yield of extract for 1-h extraction time for sage was 13.2 mg g^–1, for lovage 5.2 mg g^–1, for basil 4.4 mg g^–1, and for oregano 4.1 mg g^–1. A decrease in yields of the extracts for sage and lovage at increased pressures is hard to explain. It was found in all experiments made at the same conditions but in different time. The similar results, where system behaviour cannot be solely attributed to the effects of either temperature or pressure, have been reported in the previous study on SCCO₂ extraction of herbs [13], where it was found to be specific for sage only. Later from the literature, the same kind of unusual behaviour was found in case of extraction of stevia leaves [14]. Possibly, here could be some structure changes or textural transformations in the plant matrix at the higher extraction pressure, which would hinder the diffusion, and practically it means that longer extraction times are needed. Another explanation could be in some cumulative effect of extract components solubility changes, but this needs further study on solubility data in supercritical CO₂ of the main components in lovage extract, and estimation of phase equilibrium data is necessary for the system.

The important conclusion is that one must be very careful with extrapolating data from other experiments, and must check the yield dependence from extraction time at higher pressures for certain plants.

3.2 Effect of ethanol as modifier

Although carbon dioxide is a relatively good solvent for the extractions, it has some limitations for the extraction of polar substances from plant matrix. Ethanol was used as modifier in this study, because it is environmentally benign and relatively safe to human health. The amount of ethanol added initially composed 2.5, 5 and 7.5% of the total sample weight.

Fig. 1 shows that an increased ethanol amount increases the extraction yield from the plants, however, for lovage again, the yield decreased with increase the amount of the modifier. It points that processes during the extraction at higher pressures and with solvent mixtures are complicated, and the interaction of the co-solvent with the cellulosic structure or solid matrix of the plant cannot be neglected [14], but more likely, the components in lovage extract cumulatively might not be as soluble in the mixture of SCCO₂ and ethanol at higher pressures and modifier concentrations.

3.3 Comparative analysis of plant components at different extraction pressures

The chemical composition of the SCCO₂ extracts from these plants was determined by GC. The total yield of extracts and relative content of main compounds are presented in Table 1. It confirms that every plant has its own unique extract composition.

GC analyses revealed that the main constituent of the extract from sage at 17.2 MPa is α-thujone, but its relative content decreases at 25.5 MPa, as well as its β-pinene content. Thymol and carvacrol are the main components in the oregano extract. Lovage extract is characterized by the relatively high content of α-terpinyl acetate and β-phellandrene. Linalol is the main constituent in the basil extract. Despite the higher total yield of basil at 25.5 MPa compared to pressure 17.2 MPa, the percentage of the major components (except 1.8-cineole), is less. It means that at higher pressures the additional components are extracted and content of extract is changing substantially. Some of them are difficult to analyse by GC methods normally used for essential oils, and additional chromatographic methods might be instructive.

In conclusion, the compositions of these extracts depends on the extraction conditions, hence SCCO₂ extracts cannot be identified with the essential oils from the plants. Some researchers have recommended temperatures between 40 and 50 °C and extraction pressures lower than 100 bar (9.9 MPa) for obtaining extracts that are similar to essential oils from plant materials [15]. Our results confirm that for each aromatic plant, specific extraction conditions are needed to obtain, especially for an extract similar to that essential oil by conventional methods, and higher extraction pressures and addition of modifier increase the yield of the extract with changed proportion of components.

3.4 Antimicrobial activity of supercritical extract

There is not very much data about the antimicrobial activity of supercritical extracts from herbs, and those that exist concern extracts obtained using traditional methods, such as hydrodistillation [10]. In this study, the antimicrobial activity was tested with extracts from sage; the studies with extracts from other plants are ongoing. Here sage showed the highest inhibitory effect against gram-positive bacteria and a weakly inhibitory effect against moulds. Sage showed no effect against gram-negative E. coli.

For sage it was possible to compare the activity of SFE extract and that from the essential oil obtained using steam distillation/extraction method (Fig. 2a and b). One explanation for this different behaviour is the difference in composition of extracts, as shown for sage in Table 2.

There is more α- and β-pinene, and also α-thujone in the sage SCCO₂ extract, than in the essential oil. It may be supposed that these compounds have higher antimicrobial activity against gram-positive bacteria, or that there is some synergy between the extract components resulting in the observed antimicrobial activity.