The yield of the lipophilic extractives of deciduous bark extracted with dichloromethane was 2.1% (wt), which is in good agreement with previous results for this biomass fraction of E. globulus [21]. A GC-MS chromatogram of this extract (as TMS derivatives) is shown in Figure 2. In Table 1 the corresponding retention times, composition and concentrations are listed. It may be observed that the extract is mainly composed of several triterpenic acids with lupane, oleanane and ursane skeletons (Figure 1), mostly ursolic acid and its acetyl derivative, 3-acetylursolic acid, accounting for 2.77 g/kg and 2.64 g/kg, respectively, in a total of 10.74 g/kg quantified compounds. Betulonic (0.80 g/kg), oleanolic (0.71 g/kg), 3-acetyloleanolic (0.69 g/kg) and betulinic (0.62 g/kg) acids are also abundant in this extract, being the main components of the triterpenoid family of compounds, which also includes minor amounts of β-amyrin. Several fatty acids (C₁₄ to C₂₈, accounting globally for 0.48 g/kg), mainly hexadecanoic, tetracosanoic and hexacosanoic acids, some long chain aliphatic alcohols (C₁₆ to C₂₈, accounting globally for 0.44 g/kg), as hexacosan-1-ol, and β-sitosterol (0.24 g/kg) are also present in considerable amounts in the extract.

In this section, the results obtained for the five supercritical extractions carried out in one step are presented and discussed. On the whole, the extraction yields ranged from 0.48% to 1.76% (wt), a slight increase from 100 bar to 160 bar being observed (Figure 3), but a considerable jump of 103% was obtained when passing from 160 bar to 220 bar. This behavior is due to the direct effect of pressure upon the density of CO₂, which determines solubility at constant temperature.

The densities of CO₂, calculated with the Pitzer and Schreiber equation of state [31], for pressures of 100, 160, and 220 bar, are 633.1, 796.8 and 858.7 kg/m³, respectively, which shows the significant impact of density in the region around 800 kg/m³, a value typical of current organic solvents. An analogous trend was found for the extractions carried out at constant temperature and pressure (40 °C, 160 bar) when the co-solvent (ethanol) introduced was increased from 0% to 5%, and then to 8% (wt) (Figure 3). In this sequence, the densities of the CO₂ and ethanol mixtures are very similar: 796.8, 794.2 and 795.0 [32]—but the extraction yields show two large increments (154% and 240%). Concerning the entrainer effect, which is defined as an increase in both solubility and selectivity [33], the polarity modification imparted by ethanol to the non-polar CO₂, with final positive effect on solvent power, reveals the intermolecular interactions between ethanol and the extract components. It is worth noting that the yield achieved at 160 bar/40 °C/8% (wt) ethanol is close to that obtained by Soxhlet extraction with dichloromethane (1.8% versus 2.1%).

Regarding the composition of those extracts, they exhibit considerable differences, both quantitatively and qualitatively. In Figure 4, the abundances of the most important chemical families detected are plotted together with the Soxhlet extraction results for comparison. The numerical values are listed in Table 2, where the individual concentrations of triterpenoids are specified, given their interest in this work.

In the assays without co-solvent addition, the extracts show similar compositions. The main components are triterpenoids, particularly 3-acetylursolic, 3-acetyloleanolic and betulonic acids, also as β-amyrin and β-sitosterol, followed by fatty acids, (from which palmitic and oleic are the most representative), and minor fractions of long chain aliphatic alcohols. β-sitosterol was included in the triterpenoids group. Besides, considering the recognized biological activities of this molecule [34,35], its eventual exploitation can be of relevant interest for the upgrade of E. globulus biomass residues. In comparison to the dichloromethane extract, it is noteworthy that the three assays without co-solvent have low content levels (or even an absence of content) of ursolic and oleanolic acids (Table 2), two of the main components of E. globulus deciduous bark (see Table 1). This is due to the polarity gap between them and CO₂, which implies that higher pressure or, alternatively, a modifier is necessary to improve their solubility. This is clear from the successful removal of their acetylated forms (i.e., 3-acetylursolic, 3-acetyloleanolic), since after esterification the polarity imparted by the hydroxyl group is much attenuated.

In experiments with co-solvent addition, a significant increase (151%) of the amount of triterpenoids extracted with 5% (wt) ethanol (5.17 g/kg at 160 bar) was observed in comparison to the extract obtained with pure carbon dioxide at the same pressure and temperature (2.06 g/kg)—see Table 2. The recovery of triterpenoids was once again enhanced (in this case, 26%) after raising the ethanol content to 8% (wt), attaining 6.53 g/kg of bark. These results emphasize the chief role played by solvent polarity, since such increments can be attributed mainly to the extraction of non-acetylated triterpenic acids. In fact, at 160 bar and 40 °C, their evolution along with ethanol percentage in the supercritical solvent may be taken from Table 2: ursolic acid, 0.073, 1.74, 1.78 g/kg, and oleanolic acid, 0.050, 0.66, 0.69 g/kg. In conjunction, these variations account for 52.5% of the global increment of triterpenoids. In contrast, the two acetylated acids (acetylursolic and acetyloleanolic), justify only 18.6% of triterpenoid extraction enhancement.

Compared with the dichloromethane extract, the quantities of triterpenoids obtained by SC-CO₂/ethanol extraction reached about 70.7% of its total potential. Nonetheless, such yields may/should be optimized by adjusting operating conditions, such as extraction time, pressure, temperature, and co-solvent percentage in the SC-CO₂ stream.

On the whole, the increased amount of triterpenoids extracted is, essentially, the main difference between the extracts obtained with and without co-solvent, the remaining composition being similar.

The individual cumulative curves for each family of compounds obtained in the stepwise extraction assay are plotted in Figures 5 and 6 as functions of the mass of the CO₂ spent per unit mass of bark. In Figure 5, they are plotted as absolute weight (mg) of solute in the extract, while in Figure 6 they are normalized by their maximum extractable values (taken to be equal to that of the Soxhlet dichloromethane extract).

In the first step (120 bar, 40 °C), one extracts mostly long chain aliphatic alcohols and other compounds (41% and 66% of its total potential in the bark, respectively) and low quantities of triterpenoids and fatty acids. Of the triterpenoids, 3-acetylursolic acid is the predominant component due to its higher lipophilic character, at least in comparison with parent ursolic acid. Moreover, it should be noted that it represents ca. 30% of the triterpenoids extracted by Soxhlet with dichloromethane (see Table 1).

In the second step (180 bar, 40 °C), the improved solvent power of SC-CO₂ explains the extraction of higher amounts of triterpenoids, which increased from ca. 55 to 72 mg (see Figure 5). It is worth noting that the extraction rate of triterpenoids and fatty acids increases instantaneously at the beginning of this step (see Figure 6 also), and then diminishes continuously with time, as in the first step. The remaining families of components do not exhibit this trend. That jump is essentially justified by the effect of pressure upon solubility, as already observed and discussed in section 2.2, whereas the deceleration is due to the fact that the driving force to mass transfer attenuates along the extraction. At higher pressures the proximity between CO₂ and solute molecules is shortened, making possible interactions with CO₂ quadrupole that are almost absent at low densities. Increasing density leads to the creation of substantial dipole (induced or not)–quadrupole interactions that favor the solubility of non-acetylated triterpenic acids, from which the ursolic acid must be detached because it is the most abundant (ca. 68%; see Table 1). As has been mentioned above, this enhancement can be largely incremented using ethanol as co-solvent, which has been accomplished in the following step.

Figure 6 illustrates that the introduction of ethanol (5%, wt) in step 3 contributed to the removal of higher amounts of triterpenoids and fatty acids, which doubled from 72 to 142 mg and from 10.8 to 22.2 mg, respectively, the variation of other families being much smaller. Nevertheless, this figure points out that we are still far from exhausting the total amount of triterpenoids present in eucalyptus deciduous bark; therefore an optimization work on this topic is still needed to improve the extraction yield. This will be the main target of future researches, in order to achieve selective and quantitative triterpenoid extraction using supercritical CO₂ (modified or not) at economically tractable pressures.

Considering the results of this essay, one may expect that a multistep process based on different operating conditions may generate extracts enriched in triterpenoids by firstly removing part of the aliphatic matrix (fatty acids, long chain aliphatic alcohols and other compounds) with supercritical CO₂, since numerous substances are CO₂-philic.

Nonacosan-1-ol (98% purity) and β-sitosterol (99% purity) were purchased from Fluka Chemie (Madrid, Spain); ursolic acid (98% purity), betulinic acid (98% purity) and oleanolic acid (98% purity) were purchased from Aktin Chemicals (Chengdu, China); betulonic acid (95% purity) was purchased from CHEMOS GmbH (Regenstauf, Germany); palmitic acid (99% purity), dichloromethane (99% purity), pyridine (99% purity), bis(trimethylsilyl)trifluoroacetamide (99% purity), trimethylchlorosilane (99% purity), and tetracosane (99% purity) were supplied by Sigma Chemical Co (Madrid, Spain). Carbon dioxide was supplied with a purity of 99.95% from Praxair (Porto, Portugal).

Deciduous bark of E. globulus was randomly harvested from a 20-year-old clone plantation cultivated in the Eixo (40°37′13.56″N, 8°34′08.43″W) region of Aveiro, Portugal, air dried until a constant weight was achieved, and milled to granulometry lower than 2 mm prior to extraction. Deciduous bark was selected as substrate, since it is mostly outer bark (very similar to the last one in terms of triterpenic acids composition) and avoids felling a large number of trees to ensure a continuous supply of raw material of controlled origin for a long-term study.

Samples of E. globulus deciduous bark (15 g) were Soxhlet-extracted with dichloromethane for 7 h. The solvent was evaporated to dryness, the extracts were weighed and the results were expressed as a percent of dry bark. Dichloromethane was chosen because it is a fairly specific solvent for lipophilic extractives and was used as a reference to evaluate the efficiency of the SFE extractions.

Before each GC-MS analysis, nearly 20 mg of dried sample were converted into trimethylsilyl (TMS) derivatives according to the literature [23]. GC-MS analyses were performed using a Trace Gas Chromatograph 2000 Series equipped with a Thermo Scientific DSQ II mass spectrometer, using helium as carrier gas (35 cm·s⁻¹), equipped with a DB-1 J&W capillary column (30 m × 0.32 mm i.d., 0.25 μm film thickness). The chromatographic conditions were as follows: initial temperature: 80 °C for 5 min; temperature rate of 4 °C·min⁻¹ up to 260 °C and 2 °C min⁻¹ until the final temperature of 285 °C; maintained at 285 °C for 10 min; injector temperature: 250 °C; transfer-line temperature: 290 °C; split ratio: 1:50. The MS was operated in the electron impact mode with electron impact energy of 70 eV and data collected at a rate of 1 scan·s⁻¹ over a range of m/z 33–700. The ion source was maintained at 250 °C.

For quantitative analysis, the GC-MS instrument was calibrated with pure reference compounds, representative of the major lipophilic extractives components (namely, palmitic acid, nonacosan-1-ol, β-sitosterol, betulinic acid, ursolic acid and oleanolic acid), relative to tetracosane, the internal standard used. The respective multiplication factors needed to obtain correct quantification of the peak areas were calculated as an average of six GC-MS runs. Compounds were identified, as TMS derivatives, by comparing their mass spectra with the GC-MS spectral library, with data from the literature [23,36–41] and, in some cases, by injection of standards. Two aliquots of each extract were analyzed. Each aliquot was injected in triplicate. The results presented are the average of the concordant values obtained for each part (less than 5% variation between injections of the same aliquot and between aliquots of the same sample).