Carotenoids Recovery Enhancement by Supercritical CO₂ Extraction from Tomato Using Seed Oils as Modifiers

Abstract

The food, cosmetic and pharmaceutical industries have strong demands for lycopene, the carotenoid with the highest antioxidant activity. Usually, this carotenoid is extracted from tomatoes using various extraction methods. This work aims to improve the quantity and quality of extracts from tomato slices by enhancing the recovery of the carotenoids from the solid matrix to the solvent using 20 w/w% seeds as modifiers and supercritical CO₂ extraction with optimal parameters as the method. Tomato (TSM), camelina (CSM) and hemp (HSM) seeds were used as modifiers due to their quality (polyunsaturated fatty acids content of 53–72%). A solubility of ~10 mg carotenoids/100 g of oil was obtained for CSM and HSM, while, for TSM, the solubility was 28% higher (due to different compositions of long carbon chains). An increase in the extraction yield from 66.00 to 108.65 g extract/kg dried sample was obtained in the following order: TSM < HSM < CSM. Two products, an oil rich in carotenoids (203.59 mg/100 g extract) and ω3-linolenic acid and a solid oleoresin rich in lycopene (1172.32 mg/100 g extract), were obtained using SFE under optimal conditions (450 bar, 70 °C, 13 kg/h and CSM modifier), as assessed by response surface methodology. A recommendation is proposed for the use of these products in the food industry based on their quality.

1. Introduction

Tomatoes are a rich source of natural bioactive compounds, such as carotenoids (in peels, pulp and the whole fruit) and polyunsaturated fatty acids (PUFA) (in seeds), that can be used in the food, cosmetic and pharmaceutical industries due to their natural antioxidant, preservative and colourant abilities [1].

Carotenoids are non-polar compounds, soluble in non-polar organic solvents and oils. Incorporating carotenoids into vegetable oils enhances their bioavailability and favours their absorption in the body [1]. Carotenoids dissolved in vegetable oils can be used as such in functional food products and cosmetic formulations [2]. The quality of the oils into which the carotenoids are incorporated is also important. The PUFA content and the ω6/ω3 ratio between ω6-linoleic acid (C18:2 ω6) and ω3-linolenic acid (C18:3ω3) are important characteristics of vegetable oils owing to their ability to treat and prevent several health conditions [1,3]. The only essential PUFA that cannot be synthesized by the human body are ω6-linoleic acid (C18:2ω6) and ω3-linolenic acid (C18:3ω3). However, these are found in vegetable oils and can be introduced into the human body through the diet [3]. Moreover, it has been observed that oils protect carotenoids from degradation, isomerization and oxidation reactions [4].

The food, cosmetic and pharmaceutical industries all have a strong demand for lycopene, the carotenoid with the highest antioxidant activity. Lycopene is one of the ten most-commercialized carotenoids because of its capacity to quench singlet molecular oxygen [5], with a global economic value estimated at 1.5 USD billion in 2017 and a compound annual growth rate between 2017 and 2022 of 2.3% [1]. Lycopene and β-carotene are used in food and supplements and are produced on an industrial scale. Due to rising demand for natural carotenoids as food colourants and advancements in carotenoid extraction, the carotenoid market is anticipated to increase from USD 1.5 billion in 2019 to USD 2.0 billion in 2026 [6].

For the food, cosmetic and pharmaceutical industries, it is desirable that the extraction process can ensure environmentally friendly carotenoid extraction with minimal loss of bioactivity [7]. Supercritical fluid extraction (SFE) is a green extraction method with many advantages over conventional methods that use organic solvents [6]. Supercritical carbon dioxide (scCO₂) is an ideal solvent in the health industry due to its green nature and because produces extracts with a high degree of purity, free of traces of toxic solvents [8]. Concerning the operating parameters, the extraction yield increases with pressure, temperature and CO₂ flow rate [9,10]. For the extraction temperature, it has been found that an increase in temperature is associated with a decrease in the scCO₂ density and in the quality of the extract. However, high temperatures facilitate the release of target compounds from the plant matrix due to the increase in the carotenoids’ volatility and diffusivity and tomato structure breakage releasing the solute in the extraction solvent [7,9,11,12].

The extraction pressure influences the scCO₂ properties. An increase in pressure is associated with an increase in scCO₂ density, which favours the solubility and the extraction of solutes due to the improved solvating power of the solvent. Additionally, by using greater pressures, it may be possible to achieve the same extraction yields with less CO₂ [9,10]. Additionally, high extraction pressure and temperature increase the solubility of triglycerides and carotenoids in scCO₂ [13]. However, very high conditions could cause the degradation and isomerization of carotenoids. The optimal conditions for carotenoid extraction using the SFE process are represented by extraction pressures of 200–450 bar and extraction temperatures of 40–70 °C [6,14,15].

The CO₂ flow rate ensures the solvent velocity that influences the residence time between the solvent and the pant matrix and the mass transfer. Higher extraction yields can be obtained at high flow rates because the external resistance to mass transfer decreases and the internal diffusion increases. However, very high flow rates do not improve the extraction process, but only increase the operating costs [9].

The extraction yield in carotenoid recovery from tomatoes can be improved when co-solvents or modifiers are added to the vegetable matrix. Tomatoes contain seeds that are an endogenous source of oil that participates as a co-solvent during the extraction process [5]. Tomato seeds ensure a certain amount of oil in the extraction process, improving the extraction efficiency [16]. Modifiers are added directly to the plant sample and their action decreases with extraction time, their concentration being high at the beginning of the process [11]. Their purpose is to enhance the solubility of target compounds in scCO₂ and consequently improve the extraction efficiency. Additionally, edible oil modifiers do not require separation from the extract and can be used as a product due to their quality [9,13]. Several studies have been concerned with the extraction of carotenoids from plant matrices using different types of vegetable oils as solvents. Vegetable oils are capable of extracting non-polar compounds, such as carotenoids, and represent an ecologic, economic and green alternative to traditional organic solvents [2].

Studies concerning carotenoid extraction with SFE and seeds or vegetable oils from different types of tomato samples (pulp, peels and pomace) used tomato oil [5,12], hazelnut oil [8,17,18], avocado oil [19], canola oil [7], soybean oil [13,18] and linseed, corn, sesame, sunflower, rapeseed and olive oils [18]. The quality of these oils (triglyceride and carotenoid contents) influence SFE extraction. Thus, the evaluation of different seeds was the first step in this study. Three types of seeds were evaluated as possible sources of oil to be used in the SFE process to improve the extraction efficiency.

Tomato seeds (Lycopersicum esculentum) contain 17–23% oil with around 55% ω6-linoleic acid (C18:2ω6) and 24% ω9-oleic acid (C18:1ω9) as major compounds [3,20,21]. The carotenoid content of this oil varies with the amount of peel and pulp residues found in the seed sample subjected to extraction.

Hemp seeds (Cannabis sativa L.) are free of psychoactive substances and can be used in the food, pharmaceutic and cosmetic industries [22]. Hemp seeds contain 28–37.3% edible oil [23,24] that contains ω6-linoleic acid (C18:2ω6) and ω3-linolenic acid (C18:3ω3) PUFA as majority compounds. Hemp seed oil is unique and rare among the common plant oils due to its low ω6:ω3 ratio (3:1) and due to the presence of ω6-linolenic acid (C18:3ω6) [25]. The hemp seed oil ω6:ω3 ratio falls within the ideal range of 1:1–5:1, which has been claimed to be optimal for human consumption [22,26], while the consumption of ω6-linolenic acid is associated with the maintenance of hormonal balance [24]. Regarding the carotenoids, hemp seed oil presents β-carotene, its content ranging between 6.22–12.65 mg/kg oil [22].

Camelina seeds (Camelina sativa L.) contain 28–40% oil [26], with 50–60% PUFA, from which 35–40% consists of ω3-linolenic acid (C18:3ω3) and 15–20% ω6-linoleic acid (C18:2ω6) [27]. Moreover, it has a high content of ω9-gondoic acid (C20:1ω9), which plants very rarely possess, and a low content, below 5%, of ω9-erucic acid (C22:1ω9), which is undesirable in oils [28]. The ω6:ω3 ratio value is below 1, which shows that camelina oil is beneficial in the human diet [26].

The aim of this work was to investigate the impact on the extraction yield and carotenoid content of oils extracted in the dynamic mode by supercritical CO₂ extraction from tomatoes. Powders of tomato, camelina and hemp seeds were added to the initial tomato samples to be used as sources of oils, increasing the carotenoids’ solubility in supercritical fluid. The solubility of carotenoids from tomato samples was checked before supercritical extraction to determine the oil contribution. The extracts obtained by extraction with supercritical CO₂ and oil extracted from the tomato sample with seed powders were analysed and their effects were presented. The optimal effect of modifiers combined with the effects of the pressure and CO₂ flow rate was assessed by the design of the experiments. Four quadratic models (yield of oil and solid extracts and their compositions in carotenoids and lycopene) were formulated based on 15 experimental runs and the optimal values of the extraction parameters were identified. The qualities of the extracts obtained from tomatoes by supercritical extraction under optimal conditions were compared with extracts from tomato pomace and their potential applications were highlighted.

2. Materials and Methods

2.1. Chemicals and Standards

The Supelco 37 Component FAME Mix, 2,2-diphenyl-1-picrylhydrazyl (DPPH), 6-hydroxy-2,5,7,8-tetra methylbroman-2-carboxylic acid (Trolox), sodium hydroxide, boron trifluoride–methanol (BF₃-MeOH) (10–14%) complex solution, anhydrous magnesium sulphate, acetone, methanol, n-heptane and hexane used in this study were of analytical grade and obtained from Sigma-Aldrich (Munich, Germany). CO₂ with 99.9% purity was acquired from Linde Gaz (Bucharest, Romania).

2.2. Plant Material

Rila tomatoes, harvested in June 2022 and farmed in Colibași, Giurgiu County, camelina and hemp seeds were purchased from a local market. For the preparation of tomato slices (TS), ripe tomatoes were washed of soil traces, dried with paper tissues and manually cut into around 5 mm-thickness slices. For the preparation of tomato seeds (TSM), clean tomatoes were manually pressed with a squeezer (Ertone, model MN503) to isolate the pomace from the juice. The seeds from the pomace were separated with a sieve. Tomato slices (TS), tomato seeds (TSM), camelina seeds (CSM) and hemp seeds (HSM) were dried in a food dehydrator (Hendi Profi Line, model 229026) at 50 °C for 48 h and ground before extraction using a grinder (Tarrington House, model KM150S). The initial seed content of the TS sample in this study varied from 6–8%.

2.3. Soxhlet Extraction (SE)

The Soxhlet extraction method was used to recover the oils from TSM, CSM and HSM seeds. Tomato seed oil (TSO), camelina seed oil (CSO) and hemp seed oil (HSO) were extracted using acetone/hexane (v/v, 1:1) (AH) for 6 h. For each extraction experiment, 25 g of dried and ground seeds and 250 mL of solvent were necessary. At the end of each extraction experiment, the solvent was separated from the extract using a rotary evaporator (Hahnvapor, model HS-2000NS) to isolate the oil and to recover the solvent. The collected vegetable oils were weighed and stored in the freezer at −20 °C until analysis. The results were expressed in g oil/100 g dried seeds.

2.4. Maceration (M)

Maceration was used to extract carotenoids from tomato slices using three different solvents, which were vegetable oils (TSO, HSO and CSO). The carotenoids’ solubilities in these oils were determined and compared. For each extraction experiment, 0.5 g of dried and ground TS and 3 mL of solvent (TS:oil ratio of 1:6) were introduced to a 15 mL centrifuge tube and vortexed for 15 min at 3000 rpm (Velp Scientifica, model ZX4). Next, the obtained mixtures were centrifuged for 30 min at 8000 rpm (Hettich centrifuge, model EBA 200S) to isolate the extract (the supernatant phase) from the TS sample. The extracts were analysed by the UV–VIS method to determine the composition of carotenoids. The collected extracts were stored in the freezer at −20 °C until analysis. The results are presented in mg carotenoids/100 g extract.

2.5. Supercritical CO₂ Extraction (SFE)

The supercritical CO₂ extraction method was used to perform the extraction experiments to evaluate the influence of modifiers on mass transfer and for the determination of the optimal parameters. The extraction experiments were carried out in a laboratory-scale High-Pressure Extraction Unit (HPE-CC 500, Eurotechnica GmbH). The values used for the operating parameters were a pressure of 350–450 bar, CO₂ flow rate of 9–13 kg/h, temperature of 70 °C and an extraction time of 600 min to obtain the full extraction curve. For each experiment, 180 g of sample (150 g of dried and ground TS and 30 g of dried and ground TSM, HSM or CSM) was loaded into the extractor and the air from the system was purged with CO₂. Next, the heating was switched on, the extraction temperature was set and the extractor was pressurized with CO₂ using the high-pressure CO₂ pump at the desired value. The pressure and CO₂ flow rate were monitored and kept constant using the pressure control valve and by adjusting the pump stroke. The CO₂ flow rate was measured using a mass flow meter. During the experiment, the extract was collected every 30 min. The extracts were centrifuged for 30 min at 8000 rpm to isolate the SFE-A and SFE-B fractions according to a previous study [21]. The collected extracts were stored in a freezer at −20 °C until analysis. The results were expressed in g extract/100 g dried tomato sample.

2.6. GC–MS Analysis of Vegetable Oils

The vegetable oils (TSO, HSO and CSO) were transesterified to obtain fatty acid methyl esters (FAME) using the BF₃–MeOH complex as a catalyst in an acid-catalysed procedure [29]. The gas chromatography coupled with mass spectrometry (GC–MS) method, using a gas chromatograph (Agilent Technologies 7890A) equipped with a mass spectrometer detector (Agilent Technologies 5975C), was used to determine the FAME components in the analysed oils according to the procedure described in a previous study [30]. The results were expressed as the percentage of FAMEs.

2.7. UV–VIS Analysis of the Extracts

The UV–VIS spectrometry method was used to analyse the carotenoid contents of the extracts with a Helios UV–Visible spectrophotometer (Helios beta, Thermo Spectronic). The extracts’ spectra were recorded in the 325–575 nm wavelength range. The quantification of carotenoids was performed using the IPM-II-WG6 method [31] based on the sample absorbances and specific extinction coefficients of lycopene and β-carotene in acetone/hexane (AH) at the isosbestic point (461 nm) a = 0.194 and lycopene maximum wavelength (504 nm) a = 0.261. The results were expressed in mg carotenoid/100 g extract.

2.8. Antioxidant Activity of the Extracts

The total antioxidant activity of tomato extract samples was assessed by the DPPH (2,2-diphenyl-1-picrylhydrazyl) free radical-scavenging assay using Trolox (6-hydroxy-2,5,7,8-tetramethylbroman-2- carboxylic acid) as a reference antioxidant compound according to the procedure described in a previous study [21]. The results were expressed as the percentage inhibition of free radicals, inhibition % DPPH.

2.9. Statistical Analysis

Extraction (SE, M and SFE), centrifugation and transesterification experiments were performed in duplicate. The UV-VIS and GC-MS analyses of extracts for the quali-quantitative determination of carotenoids and FAME were carried out in triplicate. The results were presented as the mean values ± standard deviation (SD). All experimental datasets were subjected to ANOVA analysis to determine the variability between the samples and within the samples. The extraction yields, carotenoid concentrations and FAME compositions of the extracts were validated using the homogeneity of variances for individual sets, applying Hartley’s Fmax test [32].

2.10. Design of Experiments

The response surface methodology (RSM) was used as a tool to determine the effects of different operating parameters used for supercritical CO₂ extraction and their interactions on some productivity parameters (extraction yield and compounds’ concentrations) based on a number of experiments. The number of experiments was influenced by the design type, the number of factors and the number of replicates in the central point according to Equation (1) [33,34].

N = 2 k (k − 1) + C₀,

(1)

where N is the total number of experiments, k is the factor number and C₀ is the number of replicates at the central point. The selected factors were related to the operating parameters and to the composition of the vegetal matrix, while the evaluated responses were related to the quali-quantitative profiles of the extracts. The Box–Behnken surface experimental design p^k (BBD) was used as the design type in this study to evaluate the effects of three selected factors (k = 3) with three levels (p = 3), including the extraction pressure (350, 400 and 450 bar), seed type (TSM, CSM and HSM) and CO₂ flow rate (9, 11 and 13 kg/h). Four responses, such as the SFE-A (oily extract) yield, SFE-B (solid extract) yield, total carotenoid content in SFE-A and lycopene content in SFE-B, were proposed to evaluate the effects of these three factors. Considering three replicates in the central point and BBD design, the number of experiments was 15, with each experiment being performed in duplicate. The selected factors and responses with their coded and uncoded levels are presented in

Table 1. Independent variables (factors) for the BBD experimental design.

Independent Variables (Factors)	Symbol	Range of Coded Levels of Variables
		Low	Medium	High
		−1	0	+1
Extraction pressure/(bar)	X1	350	400	450
Seed type	X2	TSM	CSM	HSM
CO2 flow rate/(kg/h)	X3	9	11	13

and

Table 2. Dependent variables (responses) for the BBD experimental design.

Dependent Variables (Responses)	Symbol
SFE-A yield/(g/100 g dried sample)	Y1
SFE-B yield/(g/100 g dried sample)	Y2
SFE-A carotenoids/(mg/100 g extract)	Y3
SFE-B lycopene/(mg/100 g extract)	Y4

.

To predict the effect of each factor that affects the chosen responses of the SFE process using tomato slices, four second-order polynomial models (Equation (2)) for k independent variables (X₁, X₂ and X₃) were developed:

$$\begin{array}{l}{\mathrm{Y}}_{\mathrm{i}}={\mathsf{\beta }}_{0,\mathrm{i}}+{\mathsf{\beta }}_{1,\mathrm{i}}\cdot {\mathrm{X}}_1+{\mathsf{\beta }}_{2,\mathrm{i}}\cdot {\mathrm{X}}_2+{\mathsf{\beta }}_{3,\mathrm{i}}\cdot {\mathrm{X}}_3+{\mathsf{\beta }}_{4,\mathrm{i}}\cdot {\mathrm{X}}_1\cdot {\mathrm{X}}_2+{\mathsf{\beta }}_{5,\mathrm{i}}\cdot {\mathrm{X}}_2\cdot {\mathrm{X}}_3+{\mathsf{\beta }}_{6,\mathrm{i}}\cdot {\mathrm{X}}_1\cdot {\mathrm{X}}_3\\ +{\mathsf{\beta }}_{7,\mathrm{i}}\cdot {\mathrm{X}}_1^2+{\mathsf{\beta }}_{8,\mathrm{i}}\cdot {\mathrm{X}}_2^2+{\mathsf{\beta }}_{9,\mathrm{i}}\cdot {\mathrm{X}}_3^2\end{array}$$

(2)

where X₁, X₂ and X₃ are the selected factors and β_ji are the intercept (β_0,i), linear (β_1,i, β_2,i and β_3,i), quadratic (β_4,i, β_5,i and β_6,i) and interaction (β7_,i, β_8,i and= β_9,i) coefficients that described the factors’ effects on the chosen responses (Y_i). Statistical validation of the model was performed using analysis of variance (ANOVA) to verify that the model correctly described the relationship between factors and responses. The accuracy of the model was checked through the coefficient of determination (R²) and lack-of-fit test. Finally, the determination of the optimal conditions was accomplished using the desirability function and response surface plots. Response surface plots were generated using the function of two factors and keeping the third factor constant [34].

3. Results and Discussion

3.1. Extraction of Carotenoids with Vegetable Oils

Vegetable oils used for the extraction of carotenoids from TS were obtained by the SE method with organic solvents (AH) using three types of seeds: TSM, CSM and HSM. The characteristics of the seeds before and after grinding (A and B), the extracts mixed with the solvent (C) and the isolated oils (D) are illustrated in Figure 1.

These oils were chosen due to their particular compositions of ω6-linoleic acid (C18:2ω6) and ω3-linolenic acid (C18:3ω3). Additionally, they contain compounds that provide different colours. Due to their different compositions of carotenoids and chlorophyll, TSO is orange, CSO is yellow and HSO is green.

The oil contents of the seeds are presented in

Table 3. Extraction yields of vegetable oils (g oil/100 g dried seeds ± SD).

Extract ID	Extraction Method *	Extraction Solvent *	Extraction Parameters	Extraction Yield **
TSO	SE	AH	msample = 25 Vsolvent = 250	19.17 ± 0.21
CSO				41.85 ± 0.16
HSO				30.85 ± 0.29

* SE—Soxhlet extraction, AH—acetone:hexane (1:1, v/v), m_sample/(g)—mass of the sample, V_solvent/(mL)—volume of the solvent. ** there were no statistically significant differences between extraction yields for sets variances according to Hartley’s Fmax test (p < 0.05) at the α = 0.05 significance level.

. The extraction yield for TSO was 19.17 ± 0.21 g TSO/100 g dried seeds, being in the range of 17–23%, as reported in the literature [20]. From HSM seeds, 30.85± 0.29 g HSO/100 g dried seeds was extracted, higher than that from TSO, with 61%. The highest amount of oil, 41.85± 0.16 g CSO/100 g dried seeds, was obtained from CSM seeds, twice the yield from TSM seeds. Similar yield values of 28–35% for HSO [23] and 28–40% for CSO [26] were presented in other studies.

The composition of fatty acids of the vegetable oils extracted with AH is presented in

Table 4. Composition of fatty acids of the vegetable oils (% ± SD).

Fatty Acid Profile	Composition *
	TSO	CSO	HSO
Palmitic acid (C16:0)	15.06 ± 0.03 a	6.42 ± 0.05 b	6.43 ± 0.02 c
Stearic acid (C18:0)	6.36 ± 0.09 a	2.32 ± 0.04 b	3.40 ± 0.03 c
Oleic acid (C18:1ω9)	23.52 ± 0.22 a	18.70 ± 0.09 b	16.59 ± 0.06 c
Linoleic acid (C18:2ω6)	55.06 ± 0.17 a	19.29 ± 0.10 b	55.93 ± 0.10 c
Linolenic acid (C18:3ω6)	-	-	2.37 ± 0.03 c
Linolenic acid (C18:3ω3)	-	33.90 ± 0.07 b	13.25 ± 0.03 c
Arachidic acid (C20:0)	-	-	1.04 ± 0.01 c
Gondoic acid (C20:1ω9)	-	17.22 ± 0.06 b	1.00 ± 0.02 c
Erucic acid (C22:1ω9)	-	2.17 ± 0.04 b	-
SFA	21.42	8.74	10.87
MUFA	23.52	38.08	17.59
PUFA	55.06	53.19	71.54

* means ± SD followed by a letter (a–c) indicate that there were no statistically significant differences between the fatty acid compositions for sets variances with the same superscript letter according to Hartley’s Fmax test (p < 0.05) at the α = 0.05 level of significance.

. The fatty acids profile of these oils was similar to that in other studies [3,25,28]. For the selected oils, a difference between SFA (saturated fatty acids), MUFA (monounsaturated fatty acids) and PUFA compositions was observed. HSO contained a higher concentration of PUFA (71.54%), while TSO and CSO had similar PUFA concentrations (53.19% and 55.06%). MUFA were present in higher concentrations in CSO (38.08%), while their content in HSO was twice lower (17.59%). The major fatty acids found in these oils were ω6-linoleic acid (C18:2ω6) in TSO (55.06%) and HSO (55.93%), and ω3-linolenic acid (C18:3ω3) in CSO (33.90%), which was not present in TSO. Due to their composition, these oils can be used in the food industry.

The presence of carotenoids in the extracted oils was determined by the UV-VIS spectrophotometric method. The lycopene and β-carotene contents of these oils are presented in

Table 5. Carotenoid contents of vegetable oils and TS extracts (mg/100 g extract ± SD).

Extract ID	Extraction Method *	Extraction Solvent *	Lycopene Content **	β-Carotene Content **
TSO	SE	AH	3.26 ± 0.05	2.63 ± 0.07
CSO	SE	AH	0.08 ± 0.01	0.50 ± 0.03
HSO	SE	AH	0.11 ± 0.01	5.95 ± 0.08
TSO-TS	M	TSO	11.35 ± 0.06	1.42 ± 0.03
CSO-TS	M	CSO	8.94 ± 0.07	1.15 ± 0.05
HSO-TS	M	HSO	8.42 ± 0.25	1.48 ± 0.03

* SE—Soxhlet extraction, AH—acetone:hexane (1:1, v/v), M—maceration. ** there were no statistically significant differences between carotenoid contents for sets variances according to Hartley’s F_max test (p < 0.05) at the α = 0.05 level of significance.

. CSO contained traces of both carotenoids (less than 1 mg/100 g oil), while TSO had the highest content (5.89 mg carotenoids/100 g oil) due to the presence of peels and pulp residues from the tomato seed sample subjected to the SE method. β-carotene was present in a high concentration of 5.95 mg/100 g oil in HSO, being within the range of 3.15–12.54 mg/100 g oil reported by another author [35].

To assess the solubility of carotenoids in the three selected oils, the maceration method was used. Ground TS samples (A and B) were mixed with the oils (C), vortexed (D) and centrifuged in fractions (E), as shown in Figure 2. After the same extraction times, three oily extracts (TSO-TS, CSO-TS and HSO-TS) were obtained and changes in the oils’ colours were observed: TSO-TS changed from orange to red, CSO-TS changed from yellow to orange and HSO-TS changed from green to brown (Figure 3a,b). These changes in colours were associated with the carotenoid content extracted from TS, even if the vegetable oils were also pigmented.

The extracts’ contents of carotenoids (lycopene and β-carotene), without the oil contribution, are presented in Table 5. The solubility of both carotenoids in CSO and HSO was similar (~10 mg carotenoids/100 g oil), while, in TSO, the solubility was higher, at 28%. Lycopene’s solubility at 25 °C in TSO was 11.35 mg/100 g oil, close to the values of 7–8 mg/100 g oil reported in another study [36]. An amount of 7–10 mg lycopene/100 g oil was also reported in a recent study as its solubility in five vegetable oils, including sunflower, palm, soybean, olive and coconut oils [37]. The maximum solubility of pure lycopene in tomato oil is 46 mg/100 g oil, as reported by Squillace et al. [5]. These results show that the carotenoids’ solubility in oils is a very important factor for improving the extraction efficiency.

The lycopene recovery with TSO was higher, with 21% in CSO and with 26% in HSO. This behaviour may be caused by the chain length of the triacylglyceride fatty acids present in the selected oils. As also reported by Borel et al. [38] the solubility of carotenoids in oils increases with a decrease in the chain length. As can be observed from Table 4, TSO contains triacylglycerides with C16 and C18 fatty acid groups, while CSO and HSO also contain C20 and C22 fatty acid groups.

3.2. Extraction of Carotenoids with Supercritical CO₂ and Seed Oils as Modifiers

The supercritical CO₂ extraction method was used to recover carotenoids from the TS samples enriched with 20% extra seeds (TSM, CSM and HSM) as modifiers and to investigate their influence on the extraction efficiency. The final seed contents of the samples were around 26–28% (as the tomato pomace seed content). The extraction parameters of 450 bar extraction pressure, 70 °C extraction temperature, 11 kg/h CO₂ flow rate and extraction time of 10 h were chosen according to the results of a previous study [21] that established them as being favourable for carotenoid recovery. The seed content of the tomato samples, the extraction pressure and the CO₂ flow rate or extraction time influenced the extraction efficiency and the extracts’ qualities expressed by the yield, carotenoid recovery and antioxidant activity for supercritical CO₂ extraction of carotenoids from tomatoes.

Figure 4 presents the characteristics of the initial tomato sample and seeds (A), the ground sample (B) and the obtained centrifuged extracts TSM-TS, CSM-TS and HSM-TS (C) by SFE. Two extract phases were separated: red to orange oil (SFE-A) and dark-red solid oleoresin (SFE-B).

3.2.1. Modifiers’ Effects on Extraction Yield

Supercritical fluid extraction is a contact equilibrium separation process between a supercritical fluid and solid matrix, where mass transfer kinetics and solubility play important roles. The variations in the amount of the extract (g) with the amount of solvent (kg) per kg of sample (SFE extraction curves) for three samples (TSM, CSM and HSM) are presented in Figure 5a. From a mass transfer point of view, these curves resemble a typical extraction curve with three regions: a solubility-controlled phase (0–200 kg CO₂/kg sample), transition phase (200–350 kg CO₂/kg sample) and diffusion-controlled phase (350–600 kg CO₂/kg sample) for the same operating conditions (pressure, temperature and flow rate) [39].

In the first region (Reg. 1), the slope of the curve was linear and the amount of the extract was influenced by the solubility in supercritical CO₂. The compounds from the surface of the solid were solubilised in solvent. Lipids and carotenoids are compounds with high molecular weight, slight polarity and low volatility, and need higher-density conditions to be solubilised in CO₂. At higher pressures (450 bar) and moderate temperatures (70 °C), the CO₂ density is similar to that of a liquid (850–900 kg/m³). In this region, it was observed that the extraction yield decreased in the order of CSM > HSM > TSM, from 101 to 55 g extract/kg of dried sample. In this region, between 82% and 93% of the total extract is obtained. The transition between solubility-controlled to diffusion-controlled phases occurred in the second region (Reg. 2). The extract recovery was between 5% and 13% for TSM, HSM and CSM. In the third region (Reg. 3), the supercritical CO₂-extracted compounds from the inside of the solid diffused to the medium at a slow rate [9,40]. Even if more CO₂ was added, the extraction yield did not increase by more than 3–9%. Therefore, no more than 450 kg of CO₂/kg feed is needed for a significant extraction yield. This CO₂ consumption corresponds to an extraction time of about 450–480 min. During the performed SFE experiments, the CO₂ was recirculated and reused in the plant; therefore, for each SFE experiment, only around 3–4 kg of CO₂ was consumed when the plant was depressurised.

Adding ground TSM, CSM and HSM seeds to the initial samples of tomato slices, the extraction efficiency increased by 39% for HSM (91.8 g extract/kg dried sample) and by 65% for CSM seeds (108.7 g extract/kg dried sample) from that of TSM seeds (66 g extract/kg dried sample). This increase was caused by the oil content of the seed modifiers. These results are better than those reported by other authors who used other types of modifiers for different tomato samples (pulp, peels and pomace) to improve the supercritical CO₂ extraction efficiency. For solid modifiers, such as tomato seeds [5,12], hazelnut powder [17] or avocado pulp added directly to tomato samples, the same effect was observed, with the extraction yield increasing from 8.04% to 24.4% [19]. The addition of liquid modifiers and co-solvents, such as different types of edible canola oil [7], soybean oil [13,18], hazelnut oil [8,18], linseed, corn, sesame, sunflower, rapeseed and olive oils [18], had the same effect.

3.2.2. Modifiers’ Effect on Extracts’ Compositions

In the presence of oil and supercritical CO_2, the carotenoid composition of the extracts improved. Sovova and Stateva [13] studied the effect of oil as a co-solvent on the carotenoid composition of extracts and reported that β-carotene solubility in supercritical CO₂ increased by 1.3–1.5% when soybean oil was used as a co-solvent (mixed with supercritical CO₂ before entering the extractor), while, when the oil was used as a liquid modifier (being added directly to the sample), the solubility was even higher.

The presence of modifiers in solids subjected to supercritical extraction led to the extraction of oils together with carotenoids. The obtained SFE extracts were centrifuged to separate two fractions, such as the SFE-A fraction with the consistency of a pigmented oil rich in carotenoids and SFE-B fraction with the consistency of a pigmented solid oleoresin rich in lycopene. In Figure 5b, the extracts’ compositions of both fractions, SFE-A (79–86% oil) and SFE-B (14–21% solid oleoresin), are presented. It can be observed that the highest amounts of SFE-A and SFE-B were found in the TSM-TS extract (6.1 g SFE-A:1 g SFE-B) and CSM-TS extract (3.8 g SFE-A:1 g SFE-B). The carotenoid contents of the SFE-A and SFE-B fractions, determined by the UV-VIS method, are presented in

Table 6. Carotenoid contents of TSM-TS, CSM-TS and HSM-TS SFE extract fractions (mg/100 g extract ± SD).

Extract ID	Extraction Method	Extraction Solvent	Lycopene Content *	β-Carotene Content *
TSM-TS-SFE-A	SFE	scCO2	43.81 ± 0.40 a	159.73 ± 1.31 a
CSM-TS-SFE-A	SFE	scCO2	71.35 ± 0.24 a	139.70 ± 0.79 a
HSM-TS-SFE-A	SFE	scCO2	61.79 ± 0.14 a	101.35 ± 0.23 a
TSM-TS-SFE-B	SFE	scCO2	1212.68 ± 0.69 b	133.78 ± 0.76 a
CSM-TS-SFE-B	SFE	scCO2	1073.94 ± 3.45 a	93.52 ± 8.49 c
HSM-TS-SFE-B	SFE	scCO2	947.92 ± 4.36 a	123.41 ± 2.11 a

* means ± SD followed by a letter (a–c) indicate that there were no statistically significant differences between the lycopene and β-carotene contents for sets variances with the same superscript letter according to Hartley’s F_max test (p < 0.05) at the α = 0.05 level of significance.

. The SFE-A fractions contained mostly β-carotene, with values between 101.35 and 159.73 mg/100 g oil, while SFE-B fractions were enriched in lycopene, with values in the range of 947.92–1212.68 mg/100 g solid oleoresin.

This behaviour was also observed in other studies (the presence of both fractions), but with lower carotenoid contents than those reported in the present work. By the extraction of carotenoids from tomato pulp with supercritical CO₂, 80 mg β-carotene/100 g oil fraction (SFE-A) and 581.6 mg lycopene/100 g solid fraction (SFE-B) were reported by Longo et al. [17]. Valecilla-Yepez and Ciftici [41] extracted lycopene from a mixture of tomato peels and seeds (70:30), and the extract fractions contained 220–300 mg lycopene/100 g oil fraction (SFE-A) and 330–900 mg lycopene/100 g insoluble fraction (SFE-B). When hazelnut oil was used as a co-solvent in SFE, lycopene recovery from tomato pulp increased from 1.2% to 21.6% [18], while the addition of 37% tomato seeds as a modifier to tomato peels in SFE increased lycopene and β-carotene recovery from 17.5% and 37% to 46% and 68%, respectively [12].

Additionally, the modifiers directly influenced the composition of the SFE-A fraction in fatty acids. As shown in Table 4, CSO and HSO had superior quality to TSO due to the presence of ω3-linolenic acid (C18:3ω3).

Based on the presented results, samples of CSM or HSM mixed with TS could be used as raw materials to extract carotenoids with the same quality (900–1100 mg lycopene/100 g extract and 90–120 mg β-carotene/100 g extract) as those obtained from tomato pomace. The oily fraction separated from the extract was rich in carotenoids (40–70 mg lycopene/100 g extract and 100–160 mg β-carotene/100 g extract) and in essential ω3-linolenic acid (C18:3ω3), which is very rare in common oils.

3.3. Optimal Parameters for SFE Extraction with Seed Oils as Modifiers

Twelve experiments were performed to extract carotenoids by the supercritical CO₂ extraction method using tomato slices mixed with three types of seeds (TSM, CSM and HSM). Three other experiments were also performed in the central point (pressure of 400 bar, 20% CSM and 11 kg/h CO₂ flow rate). The values of the dependent variables are presented in

Table 7. Box–Behnken experimental design matrix for carotenoid SFE extraction with modifiers.

Run	Independent Variables			Dependent Variables *
	X1	X2	X3	Y1	Y2	Y3	Y4
	Extraction Pressure	Seeds Type	CO2 Flow Rate	SFE-A Yield	SFE-B Yield	SFE-A Carotenoids	SFE-B Lycopene
1	−1	−1	0	2.97	0.22	147.82	814.97
2	+1	−1	0	5.67	0.93	203.54	1212.68
3	−1	+1	0	4.76	0.75	86.17	659.17
4	+1	+1	0	7.36	1.82	163.14	947.92
5	−1	0	−1	4.32	0.73	98.57	756.22
6	+1	0	−1	8.08	1.90	198.58	1055.26
7	−1	0	+1	5.95	1.39	115.66	819.64
8	+1	0	+1	9.14	2.81	206.15	1167.62
9	0	−1	−1	3.43	0.30	172.34	956.10
10	0	+1	−1	5.27	1.09	138.35	739.99
11	0	−1	+1	4.56	0.53	189.80	1059.35
12	0	+1	+1	6.46	1.30	142.11	831.25
13	0	0	0	7.10	1.55	160.59	920.59
14	0	0	0	6.62	1.74	163.01	918.46
15	0	0	0	7.04	1.77	158.63	916.21

* SFE-A yield/(g/100 g dried sample), SFE-B yield/(g/100 g dried sample), SFE-A carotenoids/(mg/100 g extract), SFE-B lycopene/(mg/100 g extract).

. The extraction yields at different experimental runs varied between 2.97 and 9.14 g/100 g dried sample for the SFE-A fraction (oily extract) and between 0.22 and 2.81 g/100 g dried sample for the SFE-B fraction (solid extract). For the extracts’ compositions, the total carotenoid contents in the oily fractions (SFE-A) varied between 86.17 and 206.15 mg/100 g extract, while the lycopene content in the solid fractions was between 659.17 and 1212.68 mg/100 g extract.

In

Table 8. Models’ regression coefficients.

Statistical Data	Y1 = SFE-A Yield	Y2 = SFE-B Yield	Y3 = SFE-A Carotenoids	Y4 = SFE-B Lycopene
β0 (Intercept)	6.893	1.688	160.690	918.419
β1 (X1)	1.531	0.547	40.398	166.686
β2 (X2)	0.902	0.373	−22.978	−108.097
β3 (X3)	0.626	0.252	5.735	46.287
β4 (X1X2)	*	*	5.312	−27.241
β5(X2X3)	*	*	*	*
β6 (X1X3)	*	*	*	12.235
β7 (X12)	*	*	−8.237	21.640
β8 (X22)	−1.832	−0.830	*	−31.374
β9 (X32)	*	*	*	9.627
df Lack of fit	8	3	7	4
p-value Lack of fit	0.565	0.154	0.056	0.053
df Pure error	2	2	2	2
Pure Error	0.068	0.014	4.830	4.793
R2	0.983	0.961	0.968	0.999
Adjusted R2	0.976	0.889	0.950	0.998
Predicted R2	0.942	0.861	0.922	0.985

* Statistically insignificant (p > 0.05) at the α = 0.05 level of significance.

, the regression coefficients of four quadratic models that adequately described the effects of independent variables X₁–X₃ through the chosen Y₁–Y₄ responses of the SFE process are presented. The significant terms of the models from a statistical point of view (p > 0.05 at a significance level of α = 0.05) for the extraction yields of SFE-A (Y₁) and SFE-B (Y₂) were the linear terms of the pressure (X₁), seed type (X₂) and CO₂ flow rate (X₃) and the quadratic term of the seed type (X₂²). For the first response (Y₁), it can be seen that the quadratic effect of the seed type (X₂²) and the linear effects of the extraction pressure (X₁) and CO₂ flow rate (X₃) were positively correlated with the extraction yield of the SFE-A fraction (run 8 Table 7). This means that using CSM as a modifier, 450 bar as the extraction pressure and a 13 kg/h CO₂ flow rate resulted in high extraction yields. The same behaviour was found for the second response (Y₂), the extraction yield of the SFE-B fraction. Regarding the Y₃ and Y₄ responses, it can be seen that the linear terms of the extraction pressure (X₁) and seed type (X₂) significantly affected the recovery of carotenoids from the SFE-A and SFE-B fractions, as shown by the regression coefficient values. The effect of the seed type was negative, indicating that the use of the TSM modifier had the greatest effect, while the effect of the pressure was positive, indicating the use of high pressure values (450 bar). For all of the responses, it was observed that the extraction pressure (X₁) and CO₂ flow rate (X₃) had positive effects, while the seed type (X₂) had a complex effect, being positive for the Y₁ and Y₂ responses and negative for the Y₃ and Y₄ responses. The use of CSM led to high extraction yields and lower carotenoid concentrations, while the use of TSM led to higher recovery of carotenoids and lower yields. Thus, RSM analysis confirmed the importance of the type of seeds used as modifiers in carotenoid recovery from tomato slices through the SFE process.

For the statistical validation of the proposed models, the lack-of-fit values were evaluated for a confidence level of α = 0.05 (5% risk is considered significant). Coefficients with p > 0.05 were considered insignificant from a statistical point of view and were removed from the model. Additionally, the precision of the predictive models was also verified by the coefficient of determination (R²). The R² values were higher than 0.96, indicating a good fit between the experimental and predicted data, as can also be seen in Figure 6a–d for the Y₁–Y₄ responses. The predicted R² values show how well regression models make predictions. Response surface graphs between two factors keeping the third factor constant were generated to analyse the effect of them on the four responses (Figure 7). The CO₂ flow rate was kept constant at the maximum value of 13 kg/h because, as shown above, it had the lowest effect among the analysed factors.

For the extraction yields of the SFE-A (Figure 7a) and SFE-B (Figure 7b) fractions, it can be seen that an increase in pressure and the use of CSM seeds led to higher yields. Yield values of 7–8 g/100 g dried sample for SFE-A and 2–2,5 g/100 g dried sample for SFE-B could be obtained using pressures between 400 and 450 bar, X₁ = (0, 1) and 20% CSM seeds (X₂ = 0), a mixture of 10% CSM + 10%TSM seeds (X₂ = −0.5, 0) or 10% CSM + 10% HSM seeds (X₂ = 0, 0.5).

For the total carotenoid content of SFE-A (Figure 7c) and lycopene content of SFE-B (Figure 7d), an increase in pressure and the use of TSM seeds led to higher amounts of carotenoids. Total carotenoid values of 180–200 mg/100 g extract for SFE-A and 1100–1300 mg/100 g extract for SFE-B were obtained using pressures between 375 and 450 bar, X₁ = (−0.5, 1) and 20% TSM seeds (X₂ = −1), a mixture of 10% TSM + 10%CSM seeds (X₂ = −1, 0) or a mixture of 10% TSM + 5% CSM +5% HSM seeds (X₂ = −1, 0.5).

It is desirable to obtain both high extraction yields and extracts rich in carotenoids; thus a trade-off is needed. For this purpose, the desirability function was used to obtain the optimum extraction conditions for all responses optimized simultaneously. In Figure 8a–c, the optimal desirability plots for the proposed factors (X₁, X₂ and X₃) are presented. A desirability of 93.31% was obtained when 450 bar, CSM and 13 kg/h were used as the independent variables. Under these conditions, the obtained optimum response values were 8.89 g SFE-A/100 g dried sample, 2.57 g SFE-B/100 g dried sample, 198.49 mg total carotenoids/100 g SFE-A and 1174.90 mg lycopene/100 g SFE-B.

3.4. Quality of Products

By using SFE as an environmentally friendly extraction technique and vegetal samples with large productivity (tomato and camelina seeds), two natural products were obtained: a solid oleoresin (SFE-B) enriched in lycopene, which can be used as a natural colourant or additive in the food industry, and tomato and camelina oil enriched in carotenoids and ω3-linolenic acid (C18:3ω3), which can be used for consumption.

In

Table 9. Quality of the products obtained by SFE from TS enriched with a seed modifier.

Extract ID *	Extraction Yield **		Carotenoid Content **		SFE-A PUFA Content **		Antioxidant Activity **		Ref.
	SFE-A	SFE-B	SFE-A	SFE-B	C18:2ω6	C18:3ω3	SFE-A	SFE-B
CSM-TS	9.05 ± 1.02	2.61 ±0.16	203.59 ± 13.12	1172.32 ± 15.98	19.29 ± 0.10	33.90 ± 0.07	50.16 ± 4.19	71.23 ± 5.03	this study
TS	4.38 ± 0.86	0.30 ± 0.03	198.36 ± 15.45	916.50 ± 16.47	55.59 ± 0.12	-	49.65 ± 4.21	58.85 ± 4.09	[21]
TP	8.95 ± 1.07	1.85 ± 0.11	404.08 ± 26.04	1016.94 ± 12.03	55.59 ± 0.12	-	38.41 ± 3.04	67.02 ± 5.11

* CSM-TS extract from tomato slices with 6–8% tomato seed and 20% camelina seed, TS extract from tomato slices with 12–17% tomato seed, TP extract from tomato pomace with 22–28% tomato seed, ** extraction yield/(g/100 g dried sample), carotenoid content/(mg/100 g extract), PUFA content/(g/100 g oil), antioxidant activity/(% inhibition DPPH).

, a comparison between the quality of the extracts obtained by SFE from tomato slices with camelina seeds as a modifier and other extracts obtained from tomato slices and pomace is presented. The extract conditions were as follows: pressure of 450 bar, temperature of 70 °C, CO₂ flowrate of 13 kg/h and extraction time of 10 h. A quantitative analysis of the extracts showed that, when using camelina seeds, the extraction yield of the SFE-A product was higher by 107% than that when tomato slices and in the range of that when using tomato pomace. The extraction yield of the SFE-B product was 8.7 times higher than that obtained using tomato slices and 1.41 times higher than that obtained using tomato pomace. Regarding the products’ qualities, the total carotenoid content in the SFE-A product was almost half the content in the SFE-A product obtained from tomato pomace. However, the presence of ω3-linolenic acid (C18:3ω3) increased the value of this product’s antioxidant activity from 38.41 to 50.16%. For the SFE-B product, the lycopene content and the antioxidant activity were the highest from the analysed samples. In the study by Szabo et al. [1], HSO and flaxseed oil, with ω6/ω3 ratios of 3:1 and 0.3:1, respectively, were enriched with carotenoids recovered from tomato residue to increase their antioxidant activities, and their results showed that these products were suitable to be used in various food matrices.

Due to the existence of a wide range of bioactive compounds and the synergistic effects of these molecules, using camelina seeds as a modifier increases the added value of supercritical CO₂ extracts, allowing them to keep up with current trends in the field of health-related products.

4. Conclusions

The major compounds with antioxidant activities from tomato slice samples were carotenoids (lycopene and β-carotene) and ω-PUFA (ω6-linoleic acid). Three types of seeds, including tomato, camelina and hemp seeds, were chosen to be used as modifiers in the supercritical CO₂ extraction of carotenoids from tomato slices.

A quali-quantitative analysis of the seeds was performed to determine the amount of oil from the seeds and the oil quality expressed by the FAME composition and carotenoid solubility. The oil quantity of selected seeds, determined by Soxhlet extraction with AH solvent, increased from 19.17 to 41.85 g oil/100 g dried seed in the order of TSM < HSM < CSM. FAME analysis showed that camelina oil contained ω3-linolenic, gondoic and erucic acids, while hemp oil contained ω3-linolenic, ω6-linolenic, arachidic and gondoic acids and a low ω6/ω3 ratio. To determine the carotenoids’ solubility, the maceration of tomato slices in selected oils was performed and, as a result, ~10 mg carotenoids/100 g of oil was extracted from camelina and hemp oils, while ~13 mg carotenoids/100 g oil was extracted from tomato oil.

For the supercritical CO₂ extraction process at 450 bar, 70 °C and 11 kg/h CO₂ flow rate for 10 h applied to tomato slices enriched with 20 w/w% seed modifier, the extraction yield increased from 66.00 to 108.65 g extract/kg dried sample in the order of TSM < HSM < CSM. The extracts were centrifuged to separate two products, an oily fraction (79–86%) and a solid fraction (14–21%). Products analysis showed similar total carotenoid concentrations (163.14–211.05 mg/100 g extract) in the oil fraction and similar values of lycopene (947.92–1212.68 mg/100 g extract) in the solid fraction. Further, a Box–Behnken experimental design was applied to analyse the effects of three independent variables (extraction pressure, seed type and CO₂ flow rate) on four dependent variables (oil and solid yields and carotenoid composition) to identify the optimal extraction conditions for the high recovery of the two products and their quality. With the optimal SFE conditions of 450 bar, 70 °C and 13 kg/h, the CSM modifier resulted in the following yields of the two products: 90.52 g oil/kg dried sample with 203.59 mg carotenoid/100 g oil, 19.29 g ω3-linoleic acid and 33.90 ω3-linolenic, and 26.13 g solid/kg dried sample with 1172.32 mg lycopene/100 g solid. The quality of these products means that they can be used as valuable products in the pharmaceutical, cosmetic and food industries.