Obtaining Aromatic Extracts from Portuguese Thymus mastichina L. by Hydrodistillation and Supercritical Fluid Extraction with CO2 as Potential Flavouring Additives for Food Applications

Humans often respond to sensory impulses provided by aromas, and current trends have generated interest in natural sources of fragrances rather than the commonly used synthetic additives. For the first time, the resulting aroma of a selected culture of Thymus mastichina L. was studied as a potential food ingredient. In this context, dried (DR) and fresh (FR) samples were submitted to carbon dioxide (CO2) supercritical extraction (SFE) and hydrodistillation (HD) methods. The extracts were characterised according to their volatile composition by GC-MS, cytotoxicity against a non-tumour cell culture, and sensory attributes (odour threshold and olfactive descriptors). The most abundant aromas were quantified, and the analysis performed by GC-MS revealed an abundance of terpenoids such as thymol chemotype, followed by the precursors α-terpinene and p-cymene. DR and FR extracts (EX) obtained from SFE-CO2 show the highest content of thymol, achieving 52.7% and 72.5% of the isolated volatile fraction. The DR essential oil (EO) contained the highest amount of terpenoids, but it was also the most cytotoxic extract. In contrast, SFE-CO2 products showed the lowest cytotoxic potential. Regarding FR-OE, it had the lowest extraction yield and composition in aroma volatiles. Additionally, all samples were described as having green, fresh and floral sensory notes, with no significant statistical differences regarding the odour detection threshold (ODT) values. Finally, FR-EX of T. mastichina obtained by SFE-CO2 presented the most promising results regarding food application.


Introduction
Lipophilic volatile compounds play a notable influence on the olfactory marketing strategy in the food, cosmetic and fragrance industries. Selective oils from natural origins are recognised as one of the main approaches to product improvement, whether to extend shelf life, in alternative medicine treatments or for sensory purposes [1,2]. The compounds extracted from plants, namely essential oils, have distinct chemical compositions related not only to the species itself but also to agriculture practices, soil properties, climate conditions can be revealed. Thus, exploring these perceptions constitute a potential strategy for marketing purposes. Table 1. Dataset of the main extraction results of Thymus species performed with HD and SFE-CO 2 methods over the last ten years.
This work compares the performance of DR and FR T. mastichina extracts obtained from different extraction methods, namely SFE-CO 2 and HD. Descriptive and quantitative sensory assays (QDA) determine the fragrances attributes and odour detection threshold of samples and, finally, the cytotoxicity potential is defined within a safe concentration range for the addition of extracted oils in food products.
To the authors' best knowledge, this is the first time that the T. mastichina extracts from SFE-CO 2 and the HY fraction obtained in the HD extraction have been studied. Finally, this work aims to present a proposal for the use of T. mastichina EX as a compound rich in odours and which can be used as a flavouring additive for food products.

Plant Material
T. mastichina herb was cultivated under controlled conditions at Deifil Technology (Póvoa de Lanhoso, Braga, Portugal). After the development of micropropagated seedlings from the same mother plant, the aerial material of the plant was collected in May 2020 and subjected to an air-drying process at 40 • C (Venticell, MMM Medcenter, Planegg, Germany) until constant weight, achieving 82.06 ± 1.14% (w/w) of water content.

Essential Oil and Hydrolate Isolation
The HD procedure was conducted with 75 g of FR and DR raw plant in duplicate and independent assays. The extraction was carried out using a conventional extraction apparatus. In this methodology, the essential oils of Thymus are evaporated by heating a mixture of water and plant materials followed by the liquefaction of the vapours in a condenser [25]. The extraction was carried out for 3 h with deionized water, as suggested by the European Pharmacopeia [26]. The experimental apparatus also comprises a condenser and a decanter to collect the condensate and to separate essential oils from water. Both products, essential oil and hydrolate, obtained in each extraction experiment were properly separated by their density differences. The HY fraction was kept under ultra-fast freezing as a preparation stage for the lyophilization process (55C, CoolSafe, Beverwijk, The Netherland) under vacuum at −50 • C for 72 h. All solid fractions obtained were kept under −20 • C, while the oil fraction was stored at 4 • C until further analysis.

SFE-CO 2 T. mastichina L. Extraction
The supercritical extractor developed by Gomes et al. [27] was used for the SFE-CO 2 process, and following this work, the volatile compounds were obtained under fixed conditions of pressure and temperature, namely 80 bar and 50 • C. A stainless-steel extraction cell capable of containing 1 L of CO 2 and designed to perform experiments in static mode was inserted into the system with 30 g of T. mastichina plant using FR and DR samples. To guarantee full recovery of the extracted content after 2 h, a CO 2 re-alimentation cycle was performed. The products were collected through the regulation valves (VR1 operated at 40 bar and VR2 at 6 bar) inside 2 separators (S1 and S2). The recovered EX was stored at 4 • C for further characterization.
The extraction yields for both extraction methods was obtained by (1)

Volatile Composition Determined by GC-MS Assay
The volatile compounds of T. mastichina products were analysed by gas chromatography (GC) with an ion-trap mass spectrometer (MS) (TQ8040 NX Triple Quadrupole, Shimadzu, Japan) using a cross-bonded fused column (30 m × 0.25 mm, 0.25 µm film thickness) for low-polarity phases (Rxi-5Sil MS, Restek, Bellefonte, USA). The GC-MS instrument operated with a splitless injector and automatic sampler (AOC-20s+i). All samples were diluted in n-hexane (GC grade), and a volume of 1 µL was injected at 290 • C, carried at 1 mL·min −1 by ultrapure He.
The isothermal oven program was initially set at 40 • C for 1 min, then raised to 200 • C for 2 min at a rate of 7 • C·min −1 , followed by an increase to 250 • C for 2 min at 15 • C·min −1 and finally to 280 • C for 1 min at 20 • C·min −1 . The mass scanning range was kept at m/z 40-500, while the ion and interface temperature were at 250 • C and 260 • C, respectively.
The resulting aroma fractions are presented in relative concentration (%), obtained by the GC peak areas for each identified compound. Molecules were discriminated based on the National Institute of Standards & Technology (NIST 21, 27, 107, 147, Gaithersburg, MD, USA) mass spectra. To confirm the identification, the retention time of the homologous series of alkanes (C8-C40), analysed under the same chromatographic conditions, was used to calculate the linear retention indices (LRI) by the Kovats retention index equation [28]. The comparison between the obtained LRI and those reported in the literature guarantees the precision of the development chromatographic method and allows for the quantification of the selected molecules according to the calibration curves of the respective analytical standards.

Odour Perception and Description of T. mastichina L. Volatile Compounds
Olfactory evaluations are affected by vapour pressure, relative composition in the gas and liquid phases, atmospheric pressure and temperature conditions, in addition to the psychophysics background [9].
In accordance with the recommendations of the International Standard ISO 8586:2012 [29], a panel of 12 judges was trained to discriminate the most promising samples of T. mastichina on the perception of minimum odour concentration, ODT, and index their respective sensory notes by Quantitative Descriptive Analysis (QDA). Subsequently, the commercial standard was incorporated from this study phase to evaluate the achieved products.
To carry out the experiments, a clean, ventilated and well-lighted room was used with a controlled temperature of 20 • C. This room was also free from annoying noises. Furthermore, in the face of the SARS-CoV-2 crises, social and hygienic precautions against the virus were also adopted.

Odour Detection Threshold (ODT)
Odour analysis was performed for the individual products following the instructions of ISO 13301:2018 [30]. Thus, samples in a concentration range of 1.0 × 10 −5 to 1.0 × 10 −2 µg·mL −1 covered with cotton and conditioned in polypropylene flasks were presented to each panellist. The odour provided by the solutions was properly inhaled, and the minimal odour perception was identified.

Quantitative Descriptive Analysis (QDA)
Aqueous samples with a concentration range of 1.0 × 10 3 µg·mL −1 were all evaluated individually by the panel members. The descriptors used in the sensory analysis were carefully selected in the literature according to the main compounds of T. mastichina. An in-depth discussion indicated woody, green, fresh, citrus, floral, sweet, spicy, fruity and oily as the appropriate odour notes. The methodology was based on ISO 11035:1994 [31], and panellists scored the odour notes on an unstructured scale between 0 "not very intense" and 9 "very intense" to record their perceptions.

Cytotoxicity Analysis
All aqueous and lipophilic-based products were subjected to a cytotoxic assay using a monkey non-tumour cell line, Vero (kidney cells, ATCC CCL81.4, Gaithersburg, MD, USA). Kidneys play an important role in vertebrate organisms due to their osmoregulatory and excretion functions. Therefore, as the envisioned application is for food purposes, the potential cytotoxic effect of the extracts was evaluated through these cells. The samples were dissolved in aqueous DMSO (50%, v/v) at a final concentration of 8 mg·mL −1 and further diluted in the range of 400 to 6.25 µg·mL −1 . Following the protocol previously described by Barros et al. [32], the sulforhodamine B assay was performed, where ellipticine was used as a positive control, while the negative control was represented by a suspension of cells. The tests were performed independently (three independent assays) and in triplicate. The obtained results were expressed in GI50 values (the concentration that inhibited 50% of cell proliferation).

Statistical Analysis
The extraction yields, sensory ODT values and cytotoxicity analysis were carried out using ANOVA and Tukey tests (α = 0.05) performed in Statistica StatSoft (version 64, Tulsa, OK, USA).

Extraction Yields of T. mastichina L. Product
Significant differences were observed in the values of extraction yields between the two methods (SFE-CO 2 and HD) and between types of samples (FR or DR) ( Figure 1). The SFE-CO 2 assays resulted in a mix of extractable compounds, collected only in the S1 separator. FR-EX obtained from SFE-CO 2 showed the highest extraction yield (16.25 ± 0.23%), followed by DR-EX with 3.02 ± 0.39%, determined in DW (Dry Weight). These results are an effect of the high water content in the raw material [33], which reached 82.06 ± 1.14% (w/w); this value was considered in the extraction yield calculations. Additionally, the differences can be explained by three possible phenomena: (i) the dissolution of water in the supercritical solvent increased the solubility of the Thymus compounds; (ii) the hydration of the samples decreased the mass transfer resistance by the diffusion mechanism of intracellular compounds to the particle surface of the Thymus; and (iii) there were significant losses of matter during the drying process [34]. Statistically, all HD products presented similar extraction yield values, ranging from 0.04 ± 0.0004% (EO-FR) to 0.62 ± 0.02% (EO-DR). Previous works report EO of T. mastichina of 1.3% [19], 2.3% [20] and 1.4 to 3.5% [18], far from the percentages achieved in the present study. Regarding the HY obtained for FR and DR plant samples, traces of solid fraction were obtained after the lyophilization process: 196.45 mg and 222.30 mg in FR-and DR-HY, corresponding to 0.36 ± 0.05% and 0.06 ± 0.001%, respectively.

Chemical Composition of T. mastichina L. Product
The chemical variation of the Thymus family goes beyond the extraction method and offers a remarkable diversification according to the species and origin of growth [3]. However, as expected, the SFE-CO2 method showed the highest selectivity in the extracted

Chemical Composition of T. mastichina L. Product
The chemical variation of the Thymus family goes beyond the extraction method and offers a remarkable diversification according to the species and origin of growth [3]. However, as expected, the SFE-CO 2 method showed the highest selectivity in the extracted compounds in relation to the EO-HD results. The EO had a large source of different molecules, with a total of 39; this is as compared to 25 of the FR-EX from SFE-CO 2 . Similar results were obtained in the DR raw material, with 30 molecules for the EO in comparison to 22 compounds from the DR-EX obtained by SFE-CO 2 . Usually, HD extractions show a lower selectivity than SFE-CO 2 . For example, Costa et al. [12] identified 44 and 20 compounds in T. lotocephalus extracts obtained by HD and SFE-CO 2 methods, respectively. Table 2 and Figure 2 present the chemical composition of T. mastichina samples obtained in this work, and Table 3 presents the mass quantification of the main molecules. Thymol is the main molecule found in SFE-CO 2 -EX, representing 72.4% and 52.7% or 1511.3 ± 602.3 µg·g −1 and 282.1 ± 7.5.7 µg·g −1 of FR and DR samples, respectively. In relation to the lyophilized HY, a relative composition of 98.6% to FR-HY and 11.5% to DR-HY was obtained. The high relative concentration of thymol was previously reported for T. vulgaris and T. zygis HY, 77.1% and 61.9%, respectively [35].
The precursors γ-terpinene and p-cymene also showed significant values, mainly in the EO samples. The FR sample was composed of 30.6% of γ-terpinene and 10.3% of p-cymene, while the DR-EO presented 35.5% and 17.6%, respectively. In contrast, the representativeness of both molecules was lower in SFE-CO 2 -EX, reaching 13.6% (FR-EX) and 28.8% (DR-EX). Following the chemical route, the relative influence of thymol increases as the contribution of γ-terpinene and p-cymene decreases. Tohidi et al. [8] observed a positive correlation between the formation molecules, mentioning that p-cymene does not present itself as a final product.
As previously mentioned, the most abundant molecules present in the obtained extracts were quantified, and the results are shown in Table 3. The richest extract was the DR-EO, revealing 2565.5 ± 208.7 µg·g −1 of thymol, 609.4 ± 28.3 µg·g −1 of γ-terpinene and 337.0 ± 20.3 µg·g −1 of p-cymene molecules. In total, the sample presented an amount of 3700.7 µg·g −1 . As observed through the yield results, FR-HY had a higher total mass concentration (2429.1 µg·g −1 ) than the FR-EO fraction (524.8 µg·g −1 ), which suggests a considerable amount of the oil fraction solubilized in hydrophilic fraction. On the other hand, the HY of the DR plant was composed only of 4.3 µg·g −1 . Furthermore, the methodology to concentrate the organic products in HY is limited in terms of energy and time costs, although it is easy and cheap to produce them [35].
Different volatile compounds were previously referenced in Portuguese T. mastichina herbs. Eucalyptol, linalool and camphor stand out in EO obtained by the HD method [18,19,21,22]. β-pinene and limonene were also described as main molecules in samples cultivated in the North and South regions [19,20]. Although the present sample contains these compounds in smaller proportion, the chemotype corresponds to T. vulgaris plants collected in Algeria [44], Italy [45], Portugal [11], Romania [46] and Spain [15,24], or even to T. zygis species from Portugal [9].      Table 2.
As previously mentioned, the most abundant molecules present in the obtained extracts were quantified, and the results are shown in Table 3. The richest extract was the  Table 2.  The thymol chemotype is a natural phenolic monoterpene group synthesized by the transformation of γ-terpinene and p-cymene precursors. Both thymol and carvacrol isomers are indicated as the end of the biosynthetic chain catalysed by the two-step oxidation of γ-terpinene under the action of the cytochrome P450 enzyme (γ-terpinene synthase), which is able to modify the polarity of the compounds and make them more water-soluble [47]. Geranyl diphosphate (GPP) plays an important role in the initial chemical route, followed by the α-terpineol intermediate [48]. Through a series of steps, the molecules are converted into γ-terpinene, which, by cyclization and carbonisation reactions, form the p-cymene molecule [3]. p-cymene is a by-product of the premature release of substrate from the active site [8,47], and, by terpene synthase (TPS) activity, it produces thymol or carvacrol molecules when it reaches the last step of biosynthetic conversion. The TPS offers a wide range of ideal characteristics to modify the metabolic profile of terpenes [3]. At the same time, the final products are formed directly via γ-terpinene hydroxylation [47,48].
Moreover, α-thujene, β-myrcene, 3-carene and β-caryophyllene had some influence on the chemical profile of T. mastichina plant, mainly in DR samples. S-verbenone was only found in FR-EX (0.4% or 61.8 ± 21.1 µg·g −1 ), while notable amounts of eucalyptol were measured in DR-OE (0.5% or 123.8 ± 9.7 µg·g −1 ) compared to the other lipophilic fractions. The resulting mass composition demonstrates that even a low relative concentration can indicate the presence of a compound rich in the molecules of interest.

Sensory Odour Evaluation of T. mastichina L. Products
The ODT results are expressed in Table 4, and according to the panellist's sensitivity, all the evaluated values were similar with 95% of confidence. The standard EO was the most easily detected product at a concentration of 1.0 × 10 −4 µg·mL −1 , while the DR-EO had the highest concentration value (5.0 × 10 −3 µg·mL −1 ). FR-and DR-EX showed a concentration of 6.3 × 10 −4 µg·mL −1 and 3.0 × 10 −4 µg·mL −1 as minimum values to reflect some sensory perception, respectively.
The human nose is a powerful tool to predict the dominant odours in a mixture [50], and, as already indicated above, the T. mastichina herb is comprised of volatile terpene compounds. Its influence on odour description was evaluated under controlled conditions to ensure the accuracy of human measurement, and the results were compared with the literature. The most pronounced attributes mentioned by the mean scores of 12 panellists were green, fresh and floral ( Figure 3). The DR-EO was the only one with an oily note (0.7 points), with an emphasis on the green fragrance (4.4). The commercial EO has a strong green

Sensory Odour Evaluation of T. mastichina L. Products
The ODT results are expressed in Table 4, and according to the panellist's sensitivity, all the evaluated values were similar with 95% of confidence. The standard EO was the most easily detected product at a concentration of 1.0 × 10 −4 µg·mL −1 , while the DR-EO had the highest concentration value (5.0 × 10 −3 µg·mL −1 ). FR-and DR-EX showed a concentration of 6.3 × 10 −4 µg·mL −1 and 3.0 × 10 −4 µg·mL −1 as minimum values to reflect some sensory perception, respectively.
The human nose is a powerful tool to predict the dominant odours in a mixture [50], and, as already indicated above, the T. mastichina herb is comprised of volatile terpene compounds. Its influence on odour description was evaluated under controlled conditions to ensure the accuracy of human measurement, and the results were compared with the literature. The most pronounced attributes mentioned by the mean scores of 12 panellists were green, fresh and floral ( Figure 3). The DR-EO was the only one with an oily note (0.7 points), with an emphasis on the green fragrance (4.4). The commercial EO has a strong green  EX of T. vulgaris plant were related with spicy, pungent, herbaceous and phenol-like odour characteristics. The product was obtained with supercritical CO2 under 167 bar and 40 °C. Changing the pressure to 104 bar, the descriptors' intensity decreased, while the EO extracted by HD presented lower note values plus the burn attribute [7]. These findings suggest the thymol molecule as the key to the thyme aroma [7,51]. Furthermore, this distinct odour profile has been pointed out for three different subspecies of Algeria Thymus, ranging from a sweet herbaceous fragrance, phenolic herbaceous odour and fresh camphoraceous notes [52]. EX of T. vulgaris plant were related with spicy, pungent, herbaceous and phenol-like odour characteristics. The product was obtained with supercritical CO 2 under 167 bar and 40 • C. Changing the pressure to 104 bar, the descriptors' intensity decreased, while the EO extracted by HD presented lower note values plus the burn attribute [7]. These findings suggest the thymol molecule as the key to the thyme aroma [7,51]. Furthermore, this distinct odour profile has been pointed out for three different subspecies of Algeria Thymus, ranging from a sweet herbaceous fragrance, phenolic herbaceous odour and fresh camphoraceous notes [52].
Moreover, using validated models with ODT results, relative odour intensities and resulting odour values (OV), Teixeira et al. [9] classified three Thymus families. According to the data, herbaceous nuances prevailed in the samples, in addition to citrus, woody, floral and phenolic sensory attributes.
In fact, the herbaceous term codifies the aromatic characteristics provided by most Lamiaceae species [7,51] and offers a subjective interpretation of the odour description. Therefore, although it does not represent a fingerprint of the T. mastichina, the description of the odour presented is akin to the profile of the compounds previously mentioned. Furthermore, the obtained products revealed a promising opportunity to enhance the flavouring trajectory of processed foods in a safe range of cytotoxicity, especially to the FR-EX.
Scanning all samples and features analysed in this work, FR-EX obtained by SFE-CO 2 appears most promising. The sample reached a balance between the extraction yields, mass composition of the main compounds and toxicity value, and it has been rarely studied previously [21]. To the best of the authors' knowledge, this is the first time that the fresh T. mastichina species has been sensorially described.

Conclusions
Terpinene-rich extracts were isolated from T. mastichina herb by means of green technologies, namely HD and SFE-CO 2 . The predominant thymol chemotype and its precursors, γ-terpinene and p-cymene, were the main molecules responsible for the sensory profile described as green, fresh and floral fragrance.
Additionally, no ODT results showed significant differences, and both SFE-CO 2 products achieved the lowest toxicity among the lipophilic products. Moreover, the FR-HY fraction showed a higher selectivity of the compound, although with a low yield on a DW. The analysed characteristics of the extracted products suggested the FR-EX obtained through SFE-CO 2 as the most promising sample to use as a flavouring additive for food applications.
According to the authors' best knowledge, the species T. mastichina was studied for the first time under supercritical conditions as well as regarding its sensory profile. Hydrodistillation and supercritical fluid extraction were applied to the respective final extraction yields for fresh and dry samples of thymus plant material. The extracts' chemical composition, sensorial attributes and cytotoxic potential were evaluated. Safe concentration ranges of the most promising T. mastichina products were defined by cytotoxicity measurements and ODT values, and the significant differences between the extraction yields, sensory ODT values and cytotoxicity analysis were achieved using statistical analysis, namely ANOVA and Tukey tests.
The resulting research presented an unexplored point of view for one of the most studied aromatic plants in recent decades and gives a suggestion for the use of T. mastichina extract as compound rich in odours and which can be used as a flavouring additive for food products.