The SFE stage was performed to extract a significant portion of the low-polarity fraction of turmeric (essential oils) with supercritical CO₂, aiming to reduce its characteristic odor caused by some volatile compounds. A 3-level-2-factor central composite design (CCD) approach-based response surface methodology (RSM) analysis was proposed (22). The CCD was selected to estimate first and second-order terms of the model since it gives information about the possible curvature (by the use of the axial points) of the response variables. In this study, the selected factors were pressure (80, 140, and 200 bar) and temperature (40, 50, and 60 °C), while the response variable was oil yield (%) (Supplementary Table S1). These pressure and temperature values were obtained from previous studies (23, 24). A total of 13 experimental runs were performed, comprising five center point replicates for the assessment of experimental error, as well as four axial and four factorial design points.

In the UAE stage, 99.5% ethanol was used due to its designation as a green, safe (GRAS, generally recognized as safe), and non-toxic solvent with low cost and high recovery. A Box–Behnken design (BBD, 3-factor, 3-level) was employed to optimize UAE. The BBD was selected in this particular case since 3 factors were tested and it requires fewer runs than CCD for the same number of factors; moreover, all the design points are within the operating limits, with no axial points. The BBD included 17 randomized experiments, with five replicates at the central point. The coded and actual levels of the experimental variables, along with the response variables, are presented in Supplementary Table S3. The study variables were temperature (T, 40, 55, and 70 °C), solid/solvent ratio (Sol/Solv, 1:25, 1:35, and 1:45), and amplitude (A, 40, 60, and 80%). These values were obtained from previous studies (25). The response variables included total phenolic content (TPCmg GAE/g DE), curcuminoid content (mg/g DE), antioxidant activity (DPPH and ABTS: μmol ET/g DE), acetylcholinesterase inhibitory activity (AChE, μg/mL), and anti-inflammatory activity (lipoxygenase enzyme inhibition, LOX) (μg/mL). Ethanol is an effective, green, and safe solvent for curcuminoids extraction, its low-temperature operation prevents thermal degradation, offering economic and environmental benefits, including low cost and high recovery.

For stages 1and 2, analysis of variance (ANOVA) was applied using the significance level (p ≤ 0.05), F-statistic, and the coefficient of determination (R²) and the lack-of-fit to assess the relevance of each factor in the fitted models and to evaluate the adequacy of the model fit in each case. Among the different statistical tools, the lack-of-fit can help validating the adequacy of the model, identifying potential issues with the experimental design or data collection process and improving the model by incorporating additional terms. Design Expert® software version 13 (Stat-Ease, Inc., Minneapolis, MN, USA) was used for the experimental design, optimization and statistical analysis.

Turmeric sample (30 g) was placed in a 500 mL round-bottom flask. Then, 200 mL of distilled water was added to the flask and heated to boiling. This extraction process was carried out for a period of 3 h (26). The extracted oil was stored under refrigeration (5 °C).

Extracts were obtained with ethanol (99.5% w/w). For this, 12 g of the sample was used in the Soxhlet apparatus, with the solvent for 4 h (21, 22). After complete extraction, the solvent was removed by rotary evaporation. The solvent-free extracts were stored under refrigeration (5 °C).

The nanoemulsions were prepared according to the conditions established in previous trials using the same raw material (27). The process consisted of two phases. In the oil phase, the oil was heated to a constant temperature of 80 °C. At this temperature, the extract and SPAN 80 were added, and the mixture was stirred for one hour. The aqueous phase was prepared in parallel to the oil phase. The aqueous phase contained NaN₃ as an antimicrobial agent (0.001% w/v) (24) at 80 °C, followed by the addition of Tween 80 until complete dissolution was observed. The ratio between Tween 80/SPAN 80 surfactants was 2:1, calculated taking into account their hydrophobic balances (HLB). This specific ratio was established based on previous tests. The combination of a lipophilic (SPAN 80) and a hydrophilic (Tween 80) is a common strategy in emulsion formulation to reduce the interfacial tension between the oil and aqueous phases and achieve a stable emulsion.

A once both phases were prepared, they were mixed using a high-speed homogenizer (Ultraturrax IKA, T18) at 140,000 rpm for 15 min. Subsequently, the emulsions were treated with ultrasound at 40% amplitude for 15 min, while maintaining a water circulation bath at 12 °C. Finally, the emulsions were stored under refrigeration (5 ± 1 °C) until characterization. The same procedure was repeated to prepare the control nanoemulsion, which used an analytical standard of curcumin (>75% purity, Sigma-Aldrich, USA).

Particle size distribution (DLS), polydispersity index (PDI), Z potential and pH were measured for the nanoemulsion obtained. Z potential, DLS and PDI were measured using the technique of light dynamic (DLS) (Zetasizer Nano ZS, Malvern Instruments Ltd) at room temperature (25 °C) (28, 75). pH was recorded using a pH meter (Mettler Toledo, model FE20) calibrated with buffer solutions at pH = 4 and pH = 7, at 25 °C ± 1 °C.

The yield was determined based on the mass of the dry extract (MEs) and the amount of turmeric used (MP) (Equation 1).

A gas chromatography–mass spectrometry (GC–MS) analysis was performed to identify the compounds present in turmeric oils. The analysis was conducted using a gas chromatograph (Agilent 5,977 single quadrupole GC/MSD, USA) equipped with an HP5ms column (30 m × 0.25 mm × 0.25 μm df). The oven temperature was initially maintained at 50 °C for 2 min, then programmed to 250 °C at a rate of 6 °C/min for 15 min. The mass spectrometer operated in electron impact mode (70 eV), with the ion source temperature set at 230 °C. Helium was used as the carrier gas, and the m/z ratio range was 30 to 500 (29). Compound identification was carried out using the NIST database library, and quantification was performed based on the percentage of area.

Chromatographic separation was performed on an Agilent 1,290 UHPLC system (Agilent Technologies, Santa Clara, CA) equipped with a reversed-phase column (Zorbax Eclipse Plus C18, 2.1 × 100 mm, 1.8 μm; Agilent Technologies) maintained at 30 °C. A 5.0 μL aliquot of sample was injected, and separation was achieved using a binary solvent system consisting of (A) 0.01% (v/v) formic acid in water and (B) acetonitrile (ACN). The gradient elution program was set as follows: 0–7 min (0–30% B), 7–9 min (30–80% B), 9–11 min (80–100% B), 11–13 min (100% B), followed by re-equilibration to initial conditions (14 min, 0% B), with a constant flow rate of 0.5 mL/min.

The UHPLC system was coupled to an Agilent 6,540 qTOF mass spectrometer via an orthogonal electrospray ionization (ESI) source operating in negative-ion mode. Key MS parameters were optimized as follows: capillary voltage, 4,000 V; nebulizer pressure, 40 psi; drying gas flow rate, 10 L/min; gas temperature, 350 °C; skimmer voltage, 45 V; and fragmentor voltage, 110 V. Full-scan MS (50–1,100 m/z) and data-dependent MS/MS (50–800 m/z) modes were employed for structural elucidation.

Post-acquisition data processing was conducted using Agilent MassHunter Qualitative Analysis (v. B.08.00). Tentative compound identification was based on accurate mass measurements, isotopic distribution analysis, MS/MS fragmentation patterns, and database searches. Additional filtering strategies—including mass defect, diagnostic fragment ions, neutral loss, and background subtraction—were applied to enhance annotation reliability.

The method described by Singleton et al. (30) was used, adapted to a 96-well microplate format. The calibration curve required for this methodology was prepared using a gallic acid standard (purity >99%). A volume of 60 μL of the sample at the appropriate concentration and 60 μL of Folin–Ciocalteu reagent (FCR) were added to all wells, followed by 180 μL of sodium carbonate solution. The microplate was then incubated at 30 °C for 30 min. Subsequently, the plate was vortexed, and the samples were read at 750 nm using a microplate reader (Biotek Elx 800, USA). The total phenolic content (TPC) results were expressed as milligrams of gallic acid equivalents per gram of dry extract (mg GAE/g DE).

UAE extracts were subjected to rotary evaporation (40 °C for 15 min.) to obtain dry extracts. For the measurement, 5 mg of dry extract was dissolved in a 5 mL flask and completed with methanol (99.9% purity, Merck). The samples were then analyzed using a spectrophotometer (Genesys 10S UV–VIS, Thermo Fisher Scientific Inc., USA) at wavelength of 425 nm. The calibration curve was prepared using an analytical standard of curcumin (75% purity, Sigma-Aldrich, USA). The curcuminoid content was expressed as mg/g DE. Curcuminoid Content for UAE extracts.

The free radical scavenging capacity against 2,2-diphenyl-1-picrylhydrazyl (DPPH) was evaluated using a 96-well microplate. A volume of 60 μL of the samples and 140 μL of DPPH solution were added to the wells, followed by vortex agitation. The plate was incubated at room temperature for 30 min. Spectrophotometric readings were taken at 515 nm using a microplate reader (Biotek Elx 800, USA). The results were expressed as micromoles of Trolox equivalents per gram of dry extract (μmol TE/g DE).

The free radical scavenging capacity against ABTS [2,2′-azinobis-(3-ethylbenzothiazoline-6-sulfonic acid)] was evaluated. A volume of 60 μL of the samples was added to the microplate wells, followed by 240 μL of the ABTS radical cation solution. The microplate was vortexed before measuring the samples at 750 nm using a microplate reader (Biotek Elx 800, USA). The results were expressed as micromoles of Trolox equivalents per gram of dry extract or dry sample (μmol TE/g DE).

The inhibitory capacity of the extracts against AChE enzyme was evaluated based on the method of Ellman (31), modified by fluorescent enzymatic kinetics using ABD-F as a fluorescent probe, as described by Sánchez-Martínez et al. (32, 76). The Michaelis–Menten constant (KM value) was previously measured to determine the substrate concentration at which the reaction rate is half of the maximum velocity. A stock concentration of AChE was prepared in Tris–HCl buffer (pH 8.0, 150 mM) with the addition of 0.1% BSA to maintain the stability of the stock solution before use. Galantamine and a well without extract sample were used as positive and negative controls, respectively, for the assay reaction. Kinetic fluorescence measurements were performed in a microplate reader at an excitation wavelength (λex) of 389 nm and an emission wavelength (λem) of 513 nm, with a runtime of 15 min at 1-min intervals and a temperature set to 37 °C to determine the average velocity (Vmean) of the enzymatic reaction.

The inhibitory activity against the enzyme lipoxygenase (LOX) was measured using a fluorescence-based enzymatic kinetic method with fluorescein as a probe. The Michaelis–Menten constant (KM value) was previously determined to establish the substrate concentration at which the reaction rate reaches half of the maximum velocity (29). Quercetin (0.08 mg/mL) and a well without an extract sample were used as positive and negative controls, respectively, for the assay reaction. Kinetic fluorescence measurements were performed in a microplate reader at an excitation wavelength (λex) of 485 nm and an emission wavelength (λem) of 530 nm, with a runtime of 15 min at 1-min intervals and a temperature set to 25 °C to determine the average velocity (Vmean) of the enzymatic reaction.

The analysis was performed using a Quorum Q150R ES (USA) device with a sputter voltage of 60 mA and a sputter time of 40 s. Samples were dried and coated with 99.99% gold, and their surfaces were observed at 500 × magnification. SEM images were acquired using a Fei Quanta 200 (USA) microscope, operating at an acceleration voltage of 15 kV under high vacuum. SEM analyses were conducted on untreated turmeric (control), turmeric subjected to SFE, turmeric subjected to UAE, and turmeric subjected to sequential extraction (SFE + UAE).

Measurements were performed using a Shimadzu IRTracer-100 Fourier transform infrared (FTIR) spectrometer equipped with a DLATGS detector and a He/Ne laser, operating within a wavenumber range of 400 to 4,000 cm⁻¹. Powdered samples were analyzed using the transmission method by preparing KBr pellets (2.5 mg of sample in 150 mg of KBr), without any modifications to the sample. FTIR measurements were conducted on untreated turmeric (control), turmeric subjected to SFE, turmeric subjected to UAE, and turmeric subjected to sequential extraction (SFE + UAE).

The volatile compounds present in the oils from five different treatments were analyzed: treatments with the highest extraction yield (T4 and T6, Supplementary Table S1), treatments with the lowest yield (T1 and T2, Supplementary Table S1), treatment under optimal conditions (TO: 160 bar and 43 °C), and treatment performed using hydrodistillation (HD) as a reference for conventional extraction A total of 50 compounds were identified as predominant in the five samples; however, semi-quantification was performed only for tumerone and curlone, as they were found in higher proportions (Table 1).

Results showed that the extracts obtained through SFE contained higher amounts of tumerone (44.6–54.5%) and curlone (11.5–18.4%) compared to the HD process. The lower yield observed for HD may be attributed to the decomposition of volatile compounds due to the high temperature (~100 °C) applied during extraction (~3 h). Consequently, degradation of volatile compounds may occur leading to lower extraction efficiency compared to the SFE process (8, 35). However, the HD-extracted oil was the only one in which ar-turmerone was identified, and it was present at a high concentration (37%). Literature suggested that ar-turmerone inhibits microglial activation, a property beneficial for the treatment of neurodegenerative diseases (14). Moreover, Chowdhury et al. (38) and Devkota & Rajbhandari (14) reported ar-turmerone contents of 27.8 and 4.3%, respectively, in HD-extracted turmeric oil, which are 1.3 and 8.6 times lower than the values obtained in this study. The composition of turmeric essential oils can vary depending on the origin of the raw material, extraction methods, and process variables (14, 39).

As shown in Figure 2b, the highest tumerone percentage (54.5%) in turmeric oil was obtained under TO conditions (160 bar / 43 °C), whereas T6 had the lowest tumerone content (44.6%). T6 was characterized by a high extraction temperature (50 °C), which reduces oil solubility in supercritical CO₂. In this research results showed that TO and T6 had higher tumerone levels than those obtained through HD in the studies by Chowdhury et al. (34) and Naz et al. (35). Additionally, Figure 2b indicates that the highest curlone percentage (18.4%) in the oil was found in the TO treatment, whereas T6 had the lowest curlone percentage (16%). The curlone obtained in this study were slightly higher than those reported in previous studies (38–40), which documented curlone contents between 10 and 13%.

Phytochemical profiling of the polar fractions of C. longa, obtained using UAE in the second stage, resulted in the tentative identification of 10 compounds. These identifications were based on high-resolution mass data, MS/MS fragmentation patterns, comparisons with spectral databases (such as HMDB, Metlin, and MassBank), and existing literature. Table 2 presents the phytoconstituents detected via ESI-q-TOF-MS/MS in negative ionization mode, listing retention times (min), molecular formulas, observed [M-H]⁻ ions, mass error (ppm), and characteristic MS/MS fragment ions. Peak 1 was tentatively identified as p-hydroxybenzoic acid, displaying a deprotonated molecular ion [M − H]⁻ at m/z 137.0251. The MS/MS fragmentation produced a diagnostic product ion at m/z 93, corresponding to the loss of CO₂ [(M − H − CO2)⁻]. The decarboxylation product was also observed in the fragmentation of the precursor ions corresponding to peaks 2 and 3, generating product ions at m/z 123 and m/z 119, respectively. Based on these fragmentation patterns, the compounds were tentatively identified as vanillic acid (peak 2) and p-coumaric acid (peak 3). The product ion observed at m/z 134 in negative ionization mode for peak 4 (ferulic acid) was tentatively assigned to the fragment [M − CO − CH3 − H]⁻, resulting from consecutive losses of CO and CH3 groups. On the other hand, peaks 5–10 were tentatively identified as curcuminoid compounds based on their characteristic MS/MS fragmentation patterns, which showed agreement with previously reported data by Vardhini et al. (41), Jia et al. (42), and Singh et al. (43). Peaks 5 (tR = 8.245) and 9 (tR = 8.970) were putatively assigned as bisdemethoxycurcumin (BDMC) isomers, based on their deprotonated molecular ion [M − H]⁻ at m/z 307 and the characteristics fragments ions at m/z 145, 119, and 117. According to Jia et al. (42), m/z 145 is formed from BDMC keto form through a hydrogen transfer from benzene ring to C-4 and loss of a neutral moiety, in contrast m/z 119 is produced from BDMC enol form. Additionally, further fragmented product ion at m/z 117 is obtained from m/z 145 via the neutral loss of CO. Demethoxycurcumin isomers (DMC) (peaks 6 and 9) showed a [M − H]⁻ at m/z 337 and peak product ions at m/z 175, 173, 145, and 119. As described by Vardhini et al. (41), m/z 175 and 173 serve as diagnostic ions formed by Retro-Diels-Alder cleavage of β-diketone or keto-enol bridge in curcuminoids. Peaks 7 and 10 were proposed as curcumin isomers based on fragment ions at m/z 217, 175, and 149. The fragmentation ion at m/z 217 [M − H − 150] − represents the presence of methoxyl groups in vanillyl moieties (43).

Phytochemical profiling results were used to compare extracts obtained through UAE and conventional Soxhlet extract (used as reference). For this, a targeted profiling analysis was used based on the characterization of major phenolic compounds (Table 2). The relative abundance (peak area) of each compound (Peaks 1–10) was determined from extracted ion chromatograms (EIC) employing the deprotonated molecular ion [M − H]−. The differentially enriched compounds in the UAE-extracts are shown in the heatmap of Figure 3a. While no qualitative differences were observed between treatments, quantitative variations are evident. Curcuminoids consistently demonstrated higher abundance than phenolic acids across all extracts. Notably, extracts T7 and T15-T17 showed the most pronounced accumulation of curcumin, along with BDMC and DMC isomers. Comparative analysis revealed that the Soxhlet extraction method demonstrated significantly lower recovery efficiency for phenolic compounds relative to UAE technique. In addition, UAE-extracts were grouped according to their phenolic profile, as a result of a hierarchically cluster analysis. Figure 3b displays the t-SNE analysis results, revealing three distinct clusters. Cluster 1 comprises extracts T1, T3-T4, and T11-T13, characterized by: (i) predominant levels of bisdemethoxycurcumin I, curcumin I, and demethoxycurcumin II; and (ii) comparatively lower concentrations of curcumin II, demethoxycurcumin I, and phenolic acids. Cluster 2 (T2, T5-T6, T10, T14) exhibits a distinct phytochemical profile, characterized by: (i) higher phenolic acid content relative to other clusters, with p-coumaric acid being particularly abundant, and (ii) reduced levels of all curcuminoids except curcumin I. Cluster 3 groups extracts T7–T9 and T15–T17, occupying a central position in the t-SNE space between the other clusters. This spatial arrangement suggests a transitional role, supported by its intermediate chemical composition: while phenolic acid content is moderate compared to Clusters 1 and 2, the group is distinguished by high levels of curcumin II and demethoxycurcumin.

Results concerning bioactives extraction (TPC, curcuminoids content), antioxidant activity (DPPH, ABTS) as well as in vitro biological activities (AChE Inhibition, Anti-Inflammatory - LOX) are presented in Supplementary Table S3.

The results of ANOVA (Supplementary Table S4) showed that the solid/solvent ratio, the interaction between temperature and amplitude, and the quadratic term of the solid/solvent ratio were the variables that had a significant impact (p < 0.05) on TPC. F-tests’ p-values confirmed the statistical significance (p < 0.05) for TPC model and no significant lack of fit (p > 0.05). Experimental results (Figure 4a and Supplementary Table S3) showed that the highest total phenolic content in the extracts was 179.51 mg GAE/g DE, obtained with treatment T13 with Temperature equal to 47.5 °C (intermediate); Solid/Solvent ratio equal to 1:25 (high) and Amplitude of 30% (lowest). This TPC value was 4.3 times higher than that obtained through Soxhlet extraction (42.03 mg GAE/g DE). Similar results were found in other studies (44). The positive effect of amplitude on TPC can be attributed to higher cavitational intensity forming microbubbles, which collapse intensely, generating physical effects such as cell wall rupture. This leads to a higher extraction yield of the target compounds due to greater solvent penetration (45). Regarding the temperature effect, the literature indicates that moderately high temperatures reduce solvent viscosity while increasing solute diffusivity and solubility, thereby improving extraction. However, a substantial temperature increase during the extraction of turmeric compounds can cause degradation of the target solute (8).

The results of the analysis of variance indicate that only the quadratic term of temperature showed significant effect (p < 0.05) in curcuminoid content (Supplementary Table S5). F-tests’ p-value show significant lack of fit (p < 0.05). A significant lack of fit is an indication that the prediction of the model is not appropriate. A significant Lack of Fit F-value of 7.99 can only be caused by noise in 3.65 percent of cases. Finally, the model was considered inadequate for predicting results.

Even if it was not possible to obtain a predictive model, observation of the results presented in Supplementary Table S3 showed that curcuminoid content increased when the temperature ranged between 45 and 60 °C, the solid/solvent ratio was 1:25, and the amplitude was maintained between 30 and 45% (Supplementary Table S3). The maximum curcuminoid contents were determined to be 604.40 and 552.70 mg/g DE, obtained in treatments T1 and T13, respectively. Observations agreed with the response surface (Figure 4b) that allowed describing curcuminoids’ behavior during the UAE process as a function of the Amplitude and Temperature.

Comparing UAE with Soxhlet extraction, the curcuminoid content obtained through UAE was up to 3.3 times higher. This result can be attributed to an enhanced mass transfer caused by acoustic cavitation (44). The implosion of cavitation bubbles on the particle surface leads to cellular structure fragmentation, reducing particle size, increasing surface area, and improving mass transfer (46). The findings of this study align with previous research conducted by Landim Neves et al. (47) and Insuan et al. (44).

The results of the analysis of variance (Supplementary Table S6) indicate that for DPPH only the quadratic term of temperature showed significant differences (p < 0.05). Lack of fit showed a non-significant p-value of 0.28, which was also the evidence for the adequacy of the model. For ABTS response (Supplementary Table S6), the solid/solvent ratio was the only term that had a significant factor (p < 0.05) and non-significant Lack of fit (p < 0.05) for the model. The results indicated (Supplementary Table S3) that antioxidant activity determined by ABTS increased as the solid/solvent ratio approached 1:125. Figure 5a shows that antioxidant activity (DPPH) values were higher when temperatures ranged between 40 and 55 °C. Experimental data showed that the highest antioxidant activity by DPPH was obtained at the central point, with 1331.4 μmol TE/g DE (average of 5 central points). Nurcholis et al. (48) reported antioxidant activity (DPPH) values of 92.35 μmol TE/g DE using maceration with prior stirring (30 min at 140 rpm) followed by maceration for 48 h in a dark room. The results from Nurcholis et al. (44) were lower than those obtained in this study. Differences may be attributed to the extraction methods used and the turmeric varieties studied. High antioxidant activity values may be due to the combined effect of total phenolics, total flavonoids, and curcuminoids content (49, 50).

On the other hand, Figure 5b indicates that antioxidant activity determined by ABTS increased as the solid/solvent ratio approached 1:125 and at low temperatures, although as mentioned, the effect of temperature was not significant. The highest antioxidant activity by ABTS (231.71 μmol TE/g DE) was obtained in treatment 7.

Different studies reported that the presence of curcuminoids significantly contributes to antioxidant capacity (11, 51). The antioxidant properties of curcuminoids are attributed to their ability to reduce mitochondrial oxidative stress by enhancing the effects of superoxide dismutase, glutathione, and catalase, which collectively help neutralize free radicals (12, 13). Curcumin acts by reducing lipid peroxidation and increasing glutathione levels in the brain (52, 53).