Optimising Supercritical Carbon Dioxide Extraction of Rosmarinic Acid from Rosmarinus officinalis L. and Enhancing Yield Through Soxhlet Coupling

Abstract

Rosmarinic acid (RA) is a bioactive phenolic compound prevalent in various medicinal plants, renowned for its significant pharmacological properties. This study aims to optimise the extraction conditions of this compound from Rosmarinus officinalis L. using the response surface methodology (RSM) with a three-variable, three-level Box–Behnken design. Optimising the parameters for supercritical CO₂ (scCO₂) extraction focused on pressure (150 to 350 bar), temperature (40 to 80 °C), and co-solvent weight percentage (5 to 15% ethanol), evaluating their impact on overall yield and RA content. The optimal conditions determined were a pressure of 150 bar, a temperature of 80 °C, and 15% ethanol, yielding a total extract of 21.86 ± 1.55%, with an RA content of 3.43 ± 0.13 mg/g dry matter (DM). Scanning electron microscopy revealed that the scCO₂ treatment induced microcracks on the surface of the rosemary powder, enhancing the fluid’s ability to penetrate the plant matrix. By employing the combined scCO₂-Soxhlet method, the RA content increased to 5.78 mg/g DM. Furthermore, the final extract obtained using the Soxhlet post-scCO₂ treatment contained only trace amounts of carnosic acid (0.38 ± 0.10 mg/g DM) and carnosol (0.38 ± 0.20 mg/g DM), compared to the crude extract obtained solely with Soxhlet, which exhibited significantly higher concentrations of 8.45 ± 2.98 mg/g DM of carnosol and 16.67 ± 0.94 mg/g DM of carnosic acid. This work highlighted an innovative extraction strategy based on the coupling of scCO₂ and Soxhlet, which significantly increased RA content while reducing concentrations of other compounds such as CA and CAR. This approach makes it possible to produce RA-enriched extracts, offering considerable potential for future large-scale applications and commercialisation.

1. Introduction

Bioactive compounds from natural products have attracted increasing interest due to their chemopreventive and chemotherapeutic potential for treating and preventing various diseases. Rosmarinic acid (RA) is an ester of caffeic acid (3,4-dihyroxcynnamic acid) and 3,4-dihdroxyphenyllactic acid (Figure 1) that exhibits antioxidant [1], anti-inflammatory [2], and antimicrobial properties [3] and is widely used in the pharmaceutical, cosmetic, and food industries [4]. This phenolic compound with the molecular formula C₁₈H₁₆O₈, was first isolated by Scarpati and Oriente in 1958 from Rosmarinus officinalis L. [5].

Plants belonging to the Boraginaceae family, subfamily Nepetoideae (Lamiaceae or Labiaceae), such as rosemary (Rosmarinus officinalis), sage (Salvia officinalis), thyme (Thymus vulgaris), oregano (Origanum vulgare), and lemon balm (Melissa officinalis), are a main source of RA. Previous studies have demonstrated the pharmacological effects of RA, including hepatoprotective activity [6], antiallergic activity [7], anti-ageing activity [8], antidepressant potential [9], cardioprotective activity [10], and anti-cancer activity [11]. Antidiabetic activity has been demonstrated [12], revealing the inhibitory effect of a rosemary extract enriched with RA on the enzyme dipeptidyl peptidase IV (DPP-IV). Inhibiting this enzyme helps reduce hyperglycaemia and haemoglobin A1c levels. The authors also showed that RA has a high affinity for DPP-IV. This interaction was confirmed by a negative binding energy of −7.8 kcal/mol.

This phenolic acid is gaining increasing attention in the pharmaceutical industry due to its cytotoxic potential against various cancer cell lines. Encalada et al. demonstrated that RA, extracted from Melissa officinalis, exhibited significant cytotoxicity against human colon cancer cells (HCT-116) at a concentration of 1 mg/mL after 24 h [13]. Additionally, rosemary extract rich in RA has shown notable cytotoxic effects against human osteosarcoma cells (MG-63) at concentrations exceeding 300 µg/mL [14]. Due to its bioactive properties, RA has also found applications in the nutraceutical industry, particularly in the formulation of solid lipid nanoparticles using Witepsol waxes [15]. Its potential as a functional food ingredient has been highlighted by its positive influence on gut microbiome growth when incorporated into solid lipid nanoparticles [16].

With the growing demand for RA and concerns about the adverse effects of synthetic compounds on human health, the food industry is increasingly being pushed to adopt natural alternatives. Plant extracts, known for their many benefits, are increasingly used in food formulations to replace, partially or totally, synthetic additives [17,18]. However, to meet this increased demand, it is essential that extraction methods be developed that are both efficient and sustainable.

Many studies have been carried out on extracting RA using conventional methods, mainly based on the use of organic solvents. These techniques include maceration [19], Soxhlet extraction [20], and extraction by reflux heating [21]. These methods are based on several key steps: the penetration of the solvent into the plant cells, the solubilisation of the phytochemicals in the plant matrix, and, finally, the diffusion of the phytochemical-enriched solvent out of the plant cells [22]. However, these methods are often associated with limited yields, thermal degradation of sensitive compounds, and excessive use of organic solvents.

In this regard, innovative extraction methods such as scCO₂ extraction, pressurised liquid extraction, and enzyme-assisted extraction have been considered as alternatives for RA extraction. In contrast, carbon dioxide in the supercritical state has proved to be a more effective alternative to conventional solvents. In the supercritical state, CO₂ has properties that are the intermediate between those of a gas and a liquid, making it possible to penetrate plant matrices with high efficiency while selectively solubilising bioactive compounds. This technique offers significant advantages, such as the preservation of temperature-sensitive compounds, the absence of toxic residues, and a reduced environmental impact, in addition to allowing precise control of extraction conditions to maximise yields.

scCO₂ is recognised for its ability to efficiently extract compounds due to its solvent properties, which can be adjusted based on pressure and temperature conditions. By mod-ifying these parameters, it is possible to control the selectivity of scCO₂, particularly for the extraction of bioactive compounds [23]. However, scCO₂ is more suitable for extracting apolar compounds. The addition of a co-solvent (modifier) enhances the solubility of polar compounds. Indeed, the dipole-dipole interactions and hydrogen bonds formed between the solute and the modifier significantly increase the solubility of the solute [24]. Among the co-solvents commonly used, ethanol is characterised by its polarity, due to the presence of a hydroxyl group. It is widely used in the food industry, as it is less harmful than other organic solvents [25]. The study conducted by Abdul Aziz et al. [23] highlighted that the extraction of Orthosiphon stamineus leaves using scCO₂ with ethanol as a co-solvent resulted in a higher concentration of RA compared to other compounds extracted.

Due to RA’s high polarity, a co-solvent is necessary to increase its concentration, thus achieving levels comparable to those of conventional methods. A previous study [26] found that rosemary extracts from scCO₂ extraction demonstrated a higher antiproliferative effect than those obtained from solvent due to their richness in bioactive compounds.

The RA content in plants rarely exceeds 1% of the dry weight [27] and varies according to plant physiology, growth and development stages, geographical and environmental conditions, and pre-and post-harvest processes. In response to the growing demand for high-volume production, it is crucial to optimise the extraction parameters according to the matrix used to ensure high yields.

One of the main challenges of the extraction process is that a significant amount of RA remains trapped in the rosemary residues, resulting in a limited overall yield. Similarly, RA yield in scCO₂ extraction can vary depending on factors such as pressure, temperature, extraction time, and use of co-solvents. Existing studies do not provide precise data on the yield of RA for scCO₂ extraction, although they offer an overview of the overall yield of bioactive compounds [28,29,30]. On the other hand, although the Soxhlet method effectively extracts bioactive compounds from Rosmarinus officinalis L., it has significant limitations for extracting RA [31]. In this context, relying solely on traditional methods, and even on innovative ones, does not allow us to exploit rosemary’s RA potential.

This study aims to optimise the conditions of RA extraction from Rosmarinus officinalis L. using supercritical scCO₂ with ethanol as a co-solvent. The key factors of pressure, temperature, and co-solvent percentage were optimised using the Box–Behnken experimental design to find the best conditions for maximising RA content. In addition, a CO₂-Soxhlet coupling method was employed to further improve extraction yield. This coupling, which combined the advantages of scCO₂ extraction with the exhaustive extraction capabilities of the Soxhlet technique, represents an original feature of this work to improve RA yield and extract purity.

2. Materials and Methods

2.1. Chemicals

Folin Ciocalteu reagent, Trolox, and Na₂CO₃ were purchased from Sigma-Aldrich (Steinheim, Germany), with rosmarinic acid (purity > 97%) from Thermo Fisher Scientific (Kandel, Germany), carnosic acid (purity ≥ 90%) from Extrasynthese (Genay, France), and carnosol from MedChemExpress (Monmouth Junction, NJ, USA). The CO₂ (purity 99.99%) was provided by Messer France. The analytical-grade solvents used were supplied by CARLO ERBA reagents (Val de Reuil, France).

2.2. Optimising scCO₂ Extraction

2.2.1. Plant Material and Extraction Methods

Rosemary was harvested in May 2022 in the Beni-Chebal region, located in the province of Taourirt (34.172031, −2.722161), in an arid bioclimatic zone in eastern Morocco. The rosemary leaves, manually separated from the stems and dried in the shade, were finely ground using an electric grinder (Silver Crest). The resulting powder was then sieved to isolate a specific particle size fraction, with a particle diameter between 0.2 and 1 mm, used in this study.

The extractions were carried out using a supercritical fluid extraction system (SFE-210057-SY10-A, EXTRATEX, France) (Figure S1 in Supplementary Materials). This device includes a 100 mL extractor placed in a temperature-controlled oven, a separate unit, and a recycling system. The system is equipped with a CO₂ pump to obtain the desired pressure and a co-solvent pump to adjust the polarity of the CO₂. A cylinder supplies the CO₂ at a pressure of over 50 bar.

A mass of 15 g of rosemary leaf powder was placed in the extractor. Extractions were carried out at temperatures ranging from 40 to 80 °C and under pressures between 150 and 350 bar. The CO₂ flow rate was set at 15 g/min, with a static state maintained until the desired pressure was reached. Ethanol, used as a co-solvent, was used to improve the solubility of the RA, with a flow rate of between 5% and 15%, based on weight. The total extraction time was 3 h.

2.2.2. Response Surface Methodology

A Box–Behnken experimental design was used to investigate the impact of three scCO₂ process factors—temperature, pressure, and co-solvent percentage—on total extraction yield and RA content. The experimental design was created using the “Design Expert” software (Version 13, Stat-Ease), and the data were analysed using analysis of variance (ANOVA), setting a significance level of 0.05. R-squared, adjusted R-squared, and p-values, alongside a lack-of-fit test, were used to assess the model’s suitability for the experimental data.

Table 1. Levels of variables used in the “Box–Behnken” experimental design.

Factors	Factors Level *
	−1	0	+1
Temperature (°C)	40	60	80
Pressure (bar)	150	250	350
% EtOH (w/w)	5	10	15

* −1 is the low value, 0 is the center value, and +1 is the high value.

shows the levels of the three factors studied: temperature (40, 60, and 80 °C), pressure (150, 250, and 350 bar), and the percentage of co-solvent (5, 10, and 15% ethanol). The results, i.e., the total extraction yield and RA content, were obtained from 15 experimental runs, including 3 central points.

The extraction conditions were chosen based on preliminary experiments and data published in the literature. The experimental design method aims to establish the relationships between factors (variables, X_i) and responses (results, Y) to define an a priori polynomial mathematical model linking the responses to the factors. The mathematical relationship between the variables and the responses can be approximated by applying the following polynomial Equation (1):

$$y={\beta }_0+{\sum }_{i=1}^k{\beta }_i{X}_i+{\sum }_{i=1}^k{\beta }_{ii}{X}_{ii}+{\sum }_{i=1}^k{\sum }_{j=i+1}^k{\beta }_{ij}{X}_{ij}+\epsilon$$

(1)

where y is the measured response, β₀ is a constant value, βi is the linear coefficient, β_ii is the quadratic coefficient, and β_ij is the interaction coefficient, while X_i and X_j are independent variables.

2.3. Soxhlet Extraction

The Soxhlet extraction method was used as a reference method. A quantity of 15 g of rosemary leaf powder was placed in a cellulose extraction cartridge and then extracted with ethanol for 6 h. The residue of the rosemary powder, previously extracted by scCO₂ under the conditions of 150 bar, 80 °C, and 10% ethanol, was then subjected to a second extraction with Soxhlet, again with ethanol and for 6 h, to maximise the RA content.

2.4. Determining the RA Content

Phenolic content was determined using ultrahigh performance liquid chromatography (UPLC) from Waters Acquity (Waters, Milford, MA, USA) equipped with an autosampler, a quaternary pump, a diode array detector, and a column furnace. Separation was performed on a C18 column (2.1 × 100 mm, 1.7 μm, Waters Acquity). The mobile phase was a mixture of (A) acetonitrile with 0.1% formic acid (v/v) and (B) water with 0.1% formic acid, at a flow rate of 0.39 mL/min under the following solvent B gradient conditions: 80% (0 min), 40% (2.55 min), 25% (4.25 min), 25% (5.10 min), and 80% (7 min). The column was maintained at 35 °C, with an injection volume of 1 μL, and the detection wavelength was 325 nm for RA and 280 nm for carnosic acid and carnosol.

The rosmarinic acid (RA, y = 12,069x + 793.9; R² = 0.999), carnosic acid (CA, y = 777.32x − 1024.8; R² = 0.98), and carnosol (CAR, 538.21x − 375; R² = 0.95) contents were expressed as the ratio of the quantified amounts of each compound by UPLC to the mass of dry rosemary, as shown in the following Equation (2):

$$R\left({\displaystyle \frac{mg}{g DW}}\right)={\displaystyle \frac{Compound weight}{Rosemary weight}}$$

(2)

The compound weight, denoted as “x”, is determined from the calibration curve, which correlates the peak area of “y” with “x”.

2.5. Evaluating Antioxidant Activity

The antioxidant power of the extracts was determined using the DPPH (1,1-diphenyl-2-picrylhydrazil) free radical scavenging method according to the protocol described by [32] with a few modifications. Absorbance was measured at 517 nm. All assays were performed in triplicate. The percentage inhibition (PI) of DPPH was determined using the following Equation (3) [33]:

$$PI (\%)={\displaystyle \frac{A_{0-}A}{A_0}}\times 100$$

(3)

where PI is the percentage of inhibition, A₀ is the optics density of the free radical (DPPH) solutions in the absence of the extract (negative control), and A is the absorbance of the free radical (DPPH) solution in the presence of the extract.

The anti-free radical power was expressed by the IC50 value calculated from the regression curves. IC50 corresponds to the inhibitory concentration required to reduce 50% of free radicals. A lower value of IC50 indicated greater anti-free radical activity.

2.6. Scanning Electron Microscopy

The surface of the rosemary powder was analysed before and after supercritical fluid extraction using an FEI Quanta-250 environmental scanning analytical electron microscope.

3. Results

3.1. Optimising Supercritical CO₂ Extraction

3.1.1. Design of the Experiments Matrix

The Box–Behnken-type experimental design was applied to optimise the extraction of rosemary using scCO₂. Temperature, pressure, and percentage of co-solvent were used as independent variables. The responses studied were total extraction yield and RA content.

Table 2. Box–Behnken experimental design matrix.

Variable Values				Responses
Run	Temperature (°C)	Pressure (bar)	% EtOH (w/w)	Yield (%)	RA Content (mg/g DM)
1	40	250	5	8.73	0.00
2	60	350	5	13.41	0.00
3	40	150	10	13.02	0.57
4	60	150	15	19.60	2.06
5	60	250	10	15.88	1.68
6	80	250	15	19.39	2.88
7	60	350	15	20.31	1.38
8	60	150	5	10.97	0.00
9	80	150	10	18.98	2.38
10	80	350	10	17.52	0.17
11	40	250	15	15.67	0.84
12	60	250	10	16.65	0.98
13	60	250	10	15.75	1.05
14	40	350	10	18.48	0.12
15	80	250	5	9.34	0.00

shows the experimental design matrix and the results obtained for each response. The total extraction yield ranged from 8.71 to 20.31%, while the RA content fluctuated between 0 and 2.88 mg/g DM.

3.1.2. Effect of Process Parameters on Extraction Yield

Table 3. Analysis of variance ANOVA and statistical data from extraction yield.

	F-Value	p-Value
Model	122.99
Temperature (A)	61.50	0.0005
Pressure (B)	36.10	0.0018
EtOH (C)	748.45	<0.0001
A*B	67.62	0.0004
A*C	13.69	0.0140
B*C	4.24	0.0946
A*A	18.50	0.0077
B*B	71.50	0.0004
C*C	73.12	0.0004
Lack of fit		0.5773
R2		0.9955
Adjusted R2		0.9874
Predicted R2		0.9612
Adeq precision		34.2945

presents the results of the ANOVA analysis of variance and the statistical data for the extraction yield model. The experimental results were fitted to a quadratic model. The model’s F-value of 122.99 implies that the model is significant. There is only a 0.01% chance that such a high F-value is due to noise. Temperature, pressure, and ethanol percentage had a significant effect on extraction yield (p < 0.05). A quadratic effect was recorded for temperature (p = 0.0077), pressure (p = 0.0004), and ethanol percentage (p = 0.0004). Note that the p-value for the interaction term between pressure and percentage of ethanol is greater than 0.05, indicating that this model term is insignificant.

The statistical data showed a very good fit for the quadratic model to the results obtained for the extraction yield. The high value of R² (0.9955), very close to 1, explains the fit of the model to the data. The difference between predicted R² and adjusted R² was 0.0262, which is below the model’s acceptance limit of 0.2 without reduction. However, the value of Adeq precision greater than 4 indicates that this model can be used to navigate the design space. The lack of a fit F-value of 0.58 implies that the lack of fit is not significant compared to the pure error, which means that the error of the model is of the same order as the pure error.

By applying the experimental design methodology, it was possible to develop an equation (or model) describing the relationship between the variables and the extraction yield. This mode, expressed as a second-order polynomial in terms of coded factors (Equation (4)), has been simplified by removing the non-significant coefficients to obtain a reduced model.

$$Yield\left(\%\right)=16.09+1.17\mathit{A}+0.8927\mathit{B}+4.07\mathit{C}-1.73\mathit{A}\mathit{B}+0.7775\mathit{A}\mathit{C}-0.9407{\mathit{A}}^2+1.85{\mathit{B}}^2-1.87{\mathit{C}}^2$$

(4)

Equation (4) illustrates the influence of the different variables, both individually and interacting, on extraction yield. For the influence of individual variables, the percentage of ethanol stands out for its major positive impact, followed by the positive effects of temperature and pressure. Regarding quadratic effects, the pressure and percentage of the co-solvent had similar coefficients, with a positive effect for pressure and a negative effect for the co-solvent. A negative quadratic effect was also observed for temperature. Regarding interactions, a marked negative interaction was noted between temperature and pressure, while a positive interaction was found between temperature and the percentage of co-solvent.

The three-dimensional graphs of the response surfaces, shown in Figure 2, show the extraction yield of rosemary as a function of two variable factors, while the third factor remained fixed at the three levels studied. The effect of the pressure–co-solvent combination on extraction yield at constant temperature is illustrated in Figure 2A. As the percentage of the co-solvent and the pressure increased, the extraction yield increased, especially at 40 and 60 °C. Increasing the pressure improved the solubility of non-volatile compounds in the solvent and promoted their extraction, especially for high-weight molecules such as flavonoids [34,35]. On the other hand, increased pressure had a negative impact on the solubility and diffusivity of volatile compounds. Indeed, complete extraction of non-volatile compounds can be achieved at lower pressures [36].

At 80 °C, increasing the pressure to 250 bar showed a negative effect on the extraction rate, although the maximum efficiency (22%) was observed at 150 bar with 15% ethanol.

Figure 2B shows the evolution in the extraction yield as a function of the percentage of co-solvent and the temperature at constant pressure. At a low percentage of co-solvent, temperature had no significant influence on the extraction yield. However, the best yields were obtained at high temperatures (80 °C) with a high percentage of co-solvent (15%), under pressures of 150 and 250 bar. On the other hand, at 350 bar, a marked increase in efficiency was recorded at low temperatures (40 °C) with a medium percentage of co-solvent. The increase in the density of the fluid under pressure improved the solvent’s ability to penetrate the plant matrix, while strengthening intermolecular interactions. This facilitated the breaking of the bonds present in the matrix and the release of the compounds it contained.

The evolution in the extraction rate as a function of pressure and temperature at the percentage of constant co-solvent is shown in Figure 2C. The lowest yields were obtained with 5% ethanol. A progressive increase in the extraction rate was observed as the percentage of co-solvent increased. The maximum efficiency was achieved at 150 bar and 80 °C with 15% ethanol. These results are in good agreement with the study by Klein et al. (2019), which showed that using 15% ethanol as a co-solvent resulted in the best extraction yield from the uvaia (Eugenia pyriformis Cambess) plant. However, the authors emphasise that extraction yield should not be the only criterion taken into account; the chemical composition of the extracts is also essential for assessing their quality [37].

3.1.3. Effect of Process Parameters on RA Content

The ANOVA results for the RA content model (

Table 4. ANOVA analysis of variance and statistical data for RA content.

	F-Value	p-Value
Model	13.41	0.0009
Temperature (A)	13.14	0.0067
Pressure (B)	9.65	0.0145
EtOH (C)	44.32	0.0002
A*B	5.35	0.0494
A*C	7.19	0.0279
B*C	0.8038	0.3961
Lack of fit	0.9644
R2	0.9096
Adjusted R2	0.8417
Predicted R2	0.7012
Adeq precision	11.6817

) indicate a good fit with a two-factor interaction (2FI) model (F-value = 13.41, p < 0.0009). Temperature, pressure, and ethanol percentage significantly influence the RA content (p < 0.05). There is an interactive effect between temperature and pressure (p = 0.0494) and temperature and ethanol percentage (p = 0.0279).

A correlation coefficient (R²) of 0.9096 confirms the model’s good fit. Equation (5), along with Figure 3, depicts the model coefficients, revealing the individual and combined effects of each factor on RA content. The ethanol percentage has the most significant impact, followed by temperature and pressure. The percentage of ethanol used as a co-solvent has the most significant effect on extraction, followed by the influence of temperature (A) and pressure (B).

$$RA\left({\displaystyle \frac{mg}{g}}DM\right)=0.9408+0.4875\mathit{A}-0.4178\mathit{B}+0.8953\mathit{C}-0.44\mathit{A}\mathit{B}+0.51\mathit{A}\mathit{C}$$

(5)

Response surface graphs (Figure 4A) illustrate the effect of temperature and pressure on the extraction of RA using scCO₂, with ethanol as a co-solvent. The response surface shows that at 40 °C, the RA content remained relatively low, less than 1 mg/g DM under all pressure and ethanol percentage conditions. At low temperatures, it appeared that neither increasing the pressure nor increasing the percentage of ethanol significantly improved RA extraction. By increasing the temperature to 60 °C, an improvement in the concentration of RA was observed compared to 40 °C. Increasing the percentage of ethanol and decreasing the pressure appeared to increase the concentration of the RA from 0 to a higher value (2.4 mg/g DM). This suggests that these two factors were beginning to influence RA extraction, although the maximum concentration reached remained intermediate. At high temperatures, a clear increase in the RA content was observed, from blue to red on the colour scale. This shows that the increase in temperature to 80 °C significantly favoured RA extraction. Temperature plays a key role in breaking the bonds between the compounds and the sample matrix, facilitating the release of bioactive substances, while the solvent extracts them by transporting them out of the matrix. The intermolecular and inter-matrix interactions between the bioactive substances and the sample matrix depend on the polarity or non-polarity of these substances [35]. A similar effect of temperature was observed during the extraction of RA using microwaves from Rosmarinus officinalis L., with an increase in its content from 1.5 mg/g at 78 °C to 2.1 mg/g at 150 °C [38]. The authors indicated that RA did not appear to be thermosensitive, thus tolerating high temperatures without significant degradation. As for the effect of ethanol percentage, it was particularly noteworthy, with higher concentrations observed at the highest ethanol levels. Pressure had an effect, but it seemed secondary to the influence of the percentage of ethanol.

Figure 4B illustrates the effect of temperature and ethanol percentage on the concentration of RA. This reached its maximum at 3.80 mg/g DM at 80 °C with 15% ethanol under a pressure of 150 bar. However, a negative effect was observed under these same conditions of temperature and percentage of ethanol, with increases in pressure to 250 bar and 350 bar, leading to a decrease in RA content, which dropped to 2.88 mg/g DM and 1.75 mg/g DM, respectively. At a low pressure of 150 bar, temperature and the percentage of co-solvent had a significant effect on the extraction of this molecule, but this effect gradually decreased with increasing pressure. According to Peev et al. [39], an increase has no significant impact on the solvent’s ability to extract RA from lemon balm.

Figure 4C shows that the completely blue response surface indicates that extraction of RA was not possible with 5% ethanol. On the other hand, by increasing the percentage of ethanol, the concentration of RA gradually improved, reaching a maximum of 15% ethanol. The study carried out by Chadni et al. [40] highlights that a water content of at least 10% in the co-solvent is necessary to optimise the extraction yield of RA from sage. However, the authors demonstrate that using pure ethanol (100%) as a co-solvent is not favourable to the extraction of RA.

Due to its high polarity, the extraction of RA using scCO₂ is impossible without the addition of a co-solvent. As the results of Hansen’s theory in our previous study showed, [41], the solubility of RA in scCO₂ is very low due to the considerable differences between the solubility parameters of scCO₂ and RA. However, as the polarity of the system increases, this solubility gap narrows, allowing for higher recovery of RA in a solvent with increased polarity. Using polar or intermediate-polarity co-solvents can increase the density of the fluid mixture, causing the plant matrix to swell. This process facilitates the breaking of chemical bonds, releasing the most polar compounds.

3.1.4. Validating Prediction Models

Predictive models establish a maximum extraction yield of 22.4% and a maximum RA content of 3.86 mg/g DM. These predictions were made under extraction conditions of 150 bar and 80 °C, using 15% ethanol as a co-solvent. To validate these models, an extraction was carried out under the same experimental conditions. The results obtained show an extraction yield of 21.86 ± 1.55% and an RA content of 3.43 ± 0.13 mg/g DM. This agreement between the experimental results and the predictions of the model testifies to the model’s predictability.

3.2. Scanning Electron Microscopy

Analysing rosemary powder before and after extraction showed a significant difference in morphology after scCO₂ treatment, as shown in Figure 5. Image 5A shows the rosemary powder before extraction. The surface is dense and compact, with tightly packed layers of material. The powder appears overall intact, with no obvious signs of structural damage. ScCO₂ extraction caused the appearance of microcracks in the powder structure (Figure 5B). This indicates significant degradation of the plant matrix, probably due to depressurisation [42] or the removal of CO₂-soluble compounds. These microcracks can facilitate access to internal compounds, thus improving extraction.

A comparable scCO₂ pretreatment effect was observed on green coco coir [42]. In this study, the authors observed cell wall damage and rupture, which resulted in the degradation of the lignocellulosic biomass. The appearance of microcracks on the surface of rosemary powder after scCO₂ treatment illustrates the mechanism of penetration of this supercritical fluid. This transformation facilitates a more efficient diffusion of solvents through the material, which increases the extraction efficiency. This phenomenon also suggests that the extraction likely removed some of the lipids and/or volatile compounds present in the trichomes [43], weakening the remaining structure.

3.3. Extracting with Soxhlet

To maximise the RA content, Soxhlet extraction was performed after scCO₂ extraction to completely deplete this compound from the plant. The results of the total extraction yield, total polyphenol content, rosmarinic acid (RA) content, carnosol (CAR) content, carnosic acid (CA) content, and IC50 value obtained by Soxhlet alone and by Soxhlet after scCO₂ extraction are shown in

Table 5. Extraction results obtained with Soxhlet alone and Soxhlet after scCO_2.

Extraction Method	Yield (%)	Total Polyphenol Content (g/100 g DM)	RA Content (mg/g DM)	CAR Content (mg/g DM)	CA Content (mg/g DM)	IC50 (μg/mL)
Soxhlet	19.42 ± 0.28	4.78 ± 0.16	4.19 ± 0.33	8.65 ± 2.98	16.67 ± 0.94	214.55 ± 6.33
Soxhlet after scCO2	7.03 ± 1.06	2.12 ± 0.15	3.7 ± 0.42	0.38 ± 0.20	0.38 ± 0.10	256.05 ± 18.69

RA: rosmarinic acid; CAR: carnosol; CA: carnosic acid; IC50: inhibitory concentration required to reduce 50% of free radicals.

. Extraction performed with Soxhlet alone and ethanol resulted in high overall efficiency of 19.42 ± 0.28%, demonstrating its significant efficiency in extracting a substantial amount of compounds. After scCO₂ extraction, the Soxhlet extraction yield dropped to 7.03 ± 1.06%, demonstrating that supercritical fluid extraction had already removed a large proportion of the extractable compounds.

Our results exceed those of Zeroual et al. [44], who observed a yield of 17.32% using ethanol as the solvent. However, this yield was improved to 21.58% when methanol was used. On the other hand, Oussaid et al. [45] achieved an even higher yield of 24.71% from Rosmarinus officinalis L. from the same region.

Total polyphenol content decreased by 56% after scCO₂ extraction, from 4.78 ± 0.16 g/100 g DM with Soxhlet alone to 2.12 ± 0.15 g/100 g DM when Soxhlet was used after scCO₂. This marked reduction in total polyphenols shows that scCO₂ extraction removed a significant fraction before the Soxhlet extraction. However, the fact that 44% of the polyphenols remained in the sample after CO₂ extraction nevertheless suggests that this method did not completely deplete these compounds, leaving a fraction still accessible for Soxhlet extraction.

For RA extraction with Soxhlet, the concentration of this molecule in raw rosemary powder without pretreatment was 4.19 ± 0.33 mg/g DM. The RA content decreased slightly to 3.7 ± 0.42 mg/g DM in the pre-extracted scCO₂ rosemary powder. This indicates that extraction with this fluid was not able to extract the majority of this compound. A small fraction was extracted with the supercritical fluid, while most of it remained accessible to the Soxhlet. Our RA contents were higher than those reported by Wellwood et al. [46], who obtained a concentration of 2.19 mg/g of fresh rosemary using a water bath extraction at 35 °C. The carnosol and carnosic acid content obtained with Soxhlet extraction in our study reached 8.65 ± 2.98 mg/g DM and 16.67 ± 0.94 mg/g DM, respectively. In comparison, the study by Hirondart et al. [47], reported concentrations of 4.39 mg/g DM for carnosol and 17.68 mg/g DM for carnosic acid using the same extraction method. The results reveal that the concentration of carnosol in our sample was higher, while carnosic acid concentrations remained comparable. These differences can be attributed to factors such as the botanical origin of the rosemary, specific extraction conditions, or the harvest season. In addition, carnosol, known for its instability and sensitivity to degradation at high temperatures [48], could also explain this variability. After scCO₂ extraction, the carnosol and carnosic acid content extracted with Soxhlet dropped to 0.38 mg/g DM, indicating that CO₂ extraction removed almost all of these two compounds. As illustrated in Figure 6 and the HPLC chromatogram (Figure 7), the extract obtained with Soxhlet after scCO₂ treatment contained very low concentrations of carnosic acid and carnosol, as well as low concentrations of phenolic compounds.

After scCO₂ extraction, the carnosol and carnosic acid content extracted with Soxhlet dropped to 0.38 mg/g DM, indicating that CO₂ extraction removed almost all of these two compounds. As illustrated in Figure 6 and the HPLC chromatogram (Figure 7), the extract obtained with Soxhlet after scCO₂ treatment contained very low concentrations of carnosic acid and carnosol, as well as low concentrations of phenolic compounds.

The concentrations of carnosic acid and carnosol in rosemary extract obtained with supercritical CO₂ and ethanol as a co-solvent reached 21.41 and 3.59 mg/g DM, respectively. ScCO₂ extraction is particularly effective for both of these compounds due to their high solubility in scCO₂ [41]. In contrast, RA is less affected by this technique, with the highest concentration of this compound being obtained with Soxhlet extraction after CO₂ extraction. This result highlights the selectivity of CO₂ extraction, which is influenced by the physicochemical properties of the compounds, thus promoting the extraction of lipophilic compounds.

According to Hirondart et al. [47], rosemary can contain up to 10.13 ± 0.02 mg/g DM of RA when extracted using a hydroalcoholic mixture of 80% ethanol using liquid extraction under pressure. However, Nguyen et al. [49] reported a maximum RA content of 13.97 mg/g DM in rosemary (Rosmarinus officinalis) with enzyme-assisted aqueous extraction. In our study, the maximum concentration of RA obtained with the combined supercritical CO₂-Soxhlet method reached 5.78 mg/g DM, compared to the results of supercritical fluid extraction alone (2.48 mg/g DM) and Soxhlet extraction alone (4.87 mg/g DM) (Figure 6).

The antioxidant activity of extracts obtained with Soxhlet extraction alone and Soxhlet after supercritical fluid extraction was assessed using the DPPH assay, based on the 50% inhibitory concentration of free radicals (IC50). A lower IC50 concentration indicates higher antioxidant power. The extract obtained with Soxhlet alone had an IC50 of 214.55 ± 6.33 μg/mL, which is lower than the IC50 of 256.05 ± 18.69 μg/mL observed for the supercritical fluid-treated extract. This indicates that the latter was less effective at neutralising free radicals. However, both values remain much higher than those of ascorbic acid, which had an IC50 of 17.26 ± 0.50 μg/mL. This finding highlights that although rosemary extracts have antioxidant properties, their effectiveness remains lower than that of the commercial antioxidant, vitamin C, which can be attributed to the use of crude extracts without purification [50,51].

The decrease observed in antioxidant activity can be explained by reduced content of polyphenols and other antioxidant compounds, such as carnosol and carnosic acid, which were largely extracted by scCO₂. According to Erkan et al. [52], carnosic acid and RA have been identified as the best free radical scavengers out of the eight different compounds found in rosemary.

The antioxidant capacity of our extracts was lower than that of ethanol rosemary extracts from different regions of Algeria, which have IC50 values ranging from 10.4 mg/L to 20.7 mg/L [53]. On the other hand, the DPPH scavenging power of our extracts exceeded that of Vietnamese rosemary leaf extract, which had an IC50 of 532.01 μg/mL [49], as well as other rosemary extracts reported in a previous study, where IC50 values ranged from 787.70 to 994.68 μg/mL [54].

4. Conclusions

This study successfully employed response surface methodology (RSM) to optimise supercritical carbon dioxide (scCO₂) extraction of rosmarinic acid (RA) from Rosmarinus officinalis L. leaves, using ethanol as a co-solvent. A higher RA concentration was achieved (3.48 mg/g DM) under optimised conditions: 150 bar pressure, 80 °C temperature, and 15% ethanol. Furthermore, the study explored the combination of scCO₂ extraction with the Soxhlet method, a technique traditionally known for its exhaustive extraction capabilities. By coupling these two methods, the RA content increased significantly to 5.78 mg/g DM. This increase demonstrates the complementary nature of the Soxhlet method, which, when combined with scCO₂ extraction, facilitates the recovery of bioactive compounds that may have been less accessible during the initial scCO₂ phase alone. The combined approach also led to a reduction in the co-extraction of carnosic acid (CA) and carnosol (CAR), allowing for a RA-rich extract. These findings indicate that this combined technique provides complementary benefits in obtaining a targeted extraction of RA, making it advantageous for applications requiring high purity of RA.

Future work will focus on the minor compounds present in AR-rich extracts. Additionally, estimating the cost of the new process through tests on industrial equipment will provide a better assessment of its effectiveness on rosemary from different regions and harvested in different seasons.