Introduction

Environmental contamination resulting from industrial effluents is an increasingly pressing global issue, particularly due to the discharge of hazardous organic and inorganic compounds into natural water bodies. Industrial activities, including those associated with textiles, pharmaceuticals, petrochemicals, and plastics, frequently release substantial volumes of wastewater containing toxic pollutants, many of which are challenging to eliminate using conventional treatment methods. These pollutants pose significant threats to aquatic ecosystems and have harmful effects on human health and biodiversity. Consequently, the development of sustainable and cost-effective treatment technologies is of critical importance1. Over recent decades, various separation and purification techniques have been explored for their efficiency in wastewater treatment, such as adsorption, membrane filtration, ion exchange, and chemical precipitation2. Among these, membrane-based technologies have garnered considerable attention due to their high selectivity, low energy requirements, and potential for integration into existing treatment frameworks3. Notably, emulsion liquid membrane (ELM) technology is a promising technique for the removal of hazardous or toxic substances from wastewater. An ELM is a double emulsion system, typically characterized as water-in-oil-in-water (W/O/W), which facilitates the transport of solutes from an external phase (feed solution) into an internal stripping phase via a selective carrier or membrane phase4. The fundamental mechanism underlying ELM involves mass transfer through a liquid membrane, driven by concentration gradients and interfacial phenomena. ELM systems are particularly advantageous in scenarios requiring selective separation, low solute concentration, or simultaneous extraction. Nonetheless, despite these benefits, the practical application of ELM technology has been limited by stability-related challenges5. The coalescence of emulsions, leakage of the internal phase, and degradation of synthetic surfactants are among the primary obstacles hindering large-scale implementation. Traditionally, synthetic surfactants such as Span 80, Tween 80, and other non-ionic or ionic emulsifiers have been utilized to stabilize the membrane phase6. While effective in forming emulsions, these petrochemical-based surfactants often exhibit drawbacks, including limited biodegradability, potential toxicity, and environmental persistence. Moreover, their interfacial properties may deteriorate over time, resulting in emulsion breakdown and diminished separation efficiency7. In response to these limitations, there is increasing interest in developing bio-based and environmentally friendly alternatives to synthetic surfactants for membrane stabilization8.

Saponins, a diverse class of naturally occurring amphiphilic glycosides, have attracted significant attention as potential alternatives to conventional surfactants. This compound is found in a wide array of plant species, including Quillaja saponaria, Sapindus mukorossi (soapnut), and Glycyrrhiza glabra (licorice), and exhibits notable surface-active properties due to its distinctive molecular structure9,10. Saponins comprise a hydrophobic aglycone (sapogenin) moiety linked to one or more hydrophilic sugar chains, enabling them to effectively reduce surface and interfacial tension11. Their natural origin, biodegradability, and low toxicity render them ideal candidates for sustainable separation technologies. Numerous studies have underscored the potential of saponins in stabilizing emulsions within food, pharmaceutical, and cosmetic formulations8. However, their application in ELM systems for environmental remediation remains relatively underexplored. By utilizing their interfacial activity, saponins can be utilized as bio-emulsifiers in the formulation of saponin-based ELM. Beyond ELMs, several remediation approaches have harnessed saponins as natural surfactants. Aryanti et al. investigated saponin application in Micellar Enhanced Ultrafiltration (MEUF) for the removal of organic compounds and heavy metals from wastewater12. Similarly, Mohanty et al. reported the successful removal of Methyl Violet dye using saponin-based MEUF systems13. Marrucho et al. demonstrated the use of beet leaf-derived saponins for soil washing, achieving effective remediation of petroleum hydrocarbons14. Catherine Mulligan further highlighted the versatility of biosurfactants, including rhamnolipids, surfactin, sophorolipids, and saponins, in removing heavy metals and organic contaminants from soils and sediments15. More recently, Kim et al. showed that saponin concentrations above the critical micelle concentration (CMC) promoted micelle formation, which not only trapped dissolved ammonia but also facilitated the adsorption and removal of fine particulate matter (PM1)16. Collectively, these studies underscore the diverse remediation pathways enabled by saponins, ranging from micellar-based separations to soil washing and pollutant adsorption, thereby reinforcing their potential in environmental applications, including their integration into ELM systems.

In this study, a saponin-based ELM was synthesized and evaluated for catechol extraction, a phenolic compound found in industrial effluents. Catechol is a model pollutant due to its high solubility, toxicity, and extensive use in pesticides, dyes, and pharmaceuticals. Catechol is a priority pollutant, harmful to organisms at low concentrations, causing health issues such as irritability to the eyes, skin, and respiratory tract, DNA damage, vascular collapse, coma, and death17. Various technologies have been explored for catechol extraction. Tazerodi et al. utilized graphene oxide-based adsorption, while a sequencing continuous inflow reactor was applied for catechol recovery in wastewater treatment17. Similarly, D. Llorente et al. employed liquid-liquid extraction to recover antioxidant compounds, including catechol18. More recently, X. Hai et al. developed deep eutectic supramolecular solvents (DESPs), with nanoparticles for catechol extraction19. Pavón et al. utilized supported liquid membrane systems with Cyanex 923 and TBP carriers in kerosene diluent for the recovery of vanillin and catechol, a phenolic compound20. Many current methodologies exhibit significant limitations, such as elevated material costs, intricate setups and operations, complex synthesis procedures, and dependence on synthetic materials. Conversely, the Bio-based ELM process effectively mitigates these challenges by utilizing bio-based, biodegradable surfactants and oils. This strategy not only improves emulsion stability and achieves high extraction efficiency but also reduces operational costs and environmental impacts.

In this study investigates the utilization of a saponin-based ELM for stabilizing emulsions and subsequently employing it for catechol extraction from wastewater. The physicochemical properties of the saponin ELM were characterized using Interfacial tension (IFT), dynamic light scattering (DLS), Fourier-transform infrared spectroscopy (FTIR), rheometer, and microscopy. A Box-Behnken Design (BBD) was employed to optimize key parameters, while mass transfer studies evaluated the mass transfer coefficient of the saponin-based ELM. This research underscores the role of saponin as a bio-emulsifier, functioning in conjunction with bio-diluent and Span 80 to enhance emulsion stability and increase extraction efficiency in ELM systems. Its incorporation contributes to reducing the reliance on synthetic surfactants, thereby promoting a transition towards more sustainable and environmentally conscious membrane formulations.

Materials and methods

Materials

The chemicals were used in the experiments: Saponin (Saponin ex. Gypsophila Roots, 25% Sapogenin, Sisco Research Laboratories Pvt. Ltd., India), Soybean oil (doTERRA India Pvt. Ltd., density: 0.91–0.93 g/cm³, molecular weight: 885 g/mol), Span 80 (Sorbitan monooleate, Loba Chemie, India), and NaOH (Sodium hydroxide, Merck, India). Catechol (Pyrocatechol, extra pure; Sisco Research Laboratories Pvt. Ltd., India) was used as the model solute. Acetone and isopropyl alcohol (analytical grade; Merck, India) were employed for cleaning glassware and for sample preparation in FTIR, microscopy, and rheological studies.

Methods

Saponin-based elm synthesis process

The saponin-based ELM was prepared with a membrane phase (MP) consisting of saponin (3 v/v%), Span 80 (2 v/v%), and soybean oil (96 v/v%), based on optimized parameter values. The MP components were homogenized for 60 min at 1500 rpm using a magnetic stirrer. Subsequently, the internal phase (IP), consisting of 0.15 M NaOH, was added dropwise into the MP at a ratio of 3:1 (MP: IP). The resulting MP-IP mixture was stirred for 20 min at 800 rpm. A white, milky, and creamy emulsion was obtained, as shown in Fig. 1, confirming the successful synthesis of the saponin-based ELM.

Fig. 1
figure 1

The schematic representation of the synthesis of the saponin-based ELM process.

Full size image

Characterization techniques

Interfacial tension analysis of saponin-based ELM

The IFT of the saponin-based ELM was measured using a Du Nouy Tensiometer (Model K20, KRÜSS GmbH, Germany) equipped with the platinum plate method. For each experiment, 20 mL of the sample was placed in a glass beaker. The beaker was positioned on an adjustable platform controlled by a screw mechanism, allowing precise alignment of the platinum plate just above the liquid surface. The platinum plate was thoroughly cleaned with distilled water after each measurement and then flame-treated using a Bunsen burner to eliminate any residual contaminants.

Dynamic light scattering of emulsion droplets

The DLS, using a Malvern Instruments system, was employed to determine droplet size distributions in saponin-based ELM samples. Each emulsion was diluted (1:50) with deionized water to avoid multiple scattering. The diluted sample (1 mL) was transferred into a DLS cuvette. The instrument parameters were adjusted via Malvern software, providing hydrodynamic diameter data as intensity-weighted size distributions and polydispersity index (PDI).

Rheological analysis of saponin and saponin-based ELM

The rheological properties of the emulsion samples were investigated using a rotational rheometer (Modular Compact Rheometer, MCR 302e, Anton Paar) equipped with RheoCompass® software. A parallel plate geometry (diameter of 25 mm) was used for the measurements of rheological properties. During the test, the shear rate (s⁻¹) was varied from 0.1 to 100 s⁻¹ over a total duration of 180 s, and the corresponding shear stress was continuously recorded. For the saponin-based ELM, a steady shear flow test was conducted over a temperature range of 20 °C to 50 °C to examine the temperature-dependent behavior of the emulsions. Before each measurement, the emulsion sample was meticulously positioned on the lower plate, and the gap between the plates was adjusted to ensure optimal contact with the sample surface, thereby minimizing air entrapment.

Fourier transform infrared spectroscopy analysis

FTIR (Bruker ALPHA II FTIR spectrometer) analysis was performed using an ECO-ATR (Attenuated Total Reflectance) module (diamond crystal). The ATR crystal surface was cleaned with isopropanol and dried with lint-free tissue before measurements. A background spectrum was recorded before sample analysis to account for atmospheric water vapor. For analysis, 1 µL of liquid emulsion sample was deposited onto the diamond ATR crystal using a micropipette. No pressure clamp was needed, as the liquid sample formed uniform contact with the crystal surface. Spectral data were collected in 4000–500 cm⁻¹ at 4 cm⁻¹ resolution, averaging 32 scans per sample to improve signal-to-noise ratio.

Microscopy analysis of emulsions

Microscopic image analysis was performed using an optical microscope (Carson Microscope) equipped with adjustable magnification ranging from 50x to 250x. This analysis aimed to observe and compare the droplet morphology of emulsion formulations. The droplet size data obtained were used to assess emulsion stability over time by examining samples at different storage intervals. For the analysis, a small volume (~ 10 µL) of emulsion sample was carefully placed at the centre of a clean glass microscope slide, then covered with a cover slip to minimize evaporation and distortion. The slide was then positioned under the microscope for imaging. Using the ImageJ software, images of emulsions were analyzed to measure the droplet size.

Catechol extraction process and membrane stability analysis

The study aimed to evaluate the extraction efficiency of catechol using saponin-based ELM. The effect of the initial concentration of catechol in the aqueous phase was taken from 10 to 50 mg/L. A treatment ratio of 1:5 (v/v) was maintained between the ELM and catechol solution. The emulsion and catechol mixture were subjected to stirring at 800 rpm for 30 min at room temperature. Using a UV-spectrophotometer (PerkinElmer Lambda 365), after the extraction process, the residual catechol concentration was measured at a detection wavelength of 275 nm, which corresponds to the maximum absorbance (λ max) of catechol. The extraction efficiency (%) was calculated using Eq. (1), and membrane stability was studied by calculating the membrane breakage using Eq. (2).

$$\:\text{E}\text{x}\text{t}\text{r}\text{a}\text{c}\text{t}\text{i}\text{o}\text{n}\:\text{e}\text{f}\text{f}\text{i}\text{c}\text{i}\text{n}\text{e}\text{c}\text{y}\:\left(\text{\%}\right)=\frac{{C}_{i}-{C}_{f}}{{C}_{i}}\:\times\:100$$
(1)
$$\:\text{M}\text{e}\text{m}\text{b}\text{r}\text{a}\text{n}\text{e}\:\text{B}\text{r}\text{e}\text{a}\text{k}\text{a}\text{g}\text{e}\:\left(\text{\%}\right)=\frac{{V}_{i}-{V}_{f}}{{V}_{i}}\:\times\:100$$
(2)

Experimental design

In the extraction of catechol, several critical parameters were identified as independent variables: Span 80 concentration, saponin concentration, internal phase concentration (stripping agent), time, and the ratio of MP to IP. The low and high levels for each variable were designated as −1 and 1, respectively, as represented in Table 1. Factor level selection was based on prior experience and literature. Data analysis used Design-Expert® 12 software. To investigate interactions among primary variables and their impact on catechol extraction, a BBD was employed. Regression analysis estimated the response function. The permeation rate can be predicted using the quadratic model in Eq. (3). This design required 29 runs for the four variables.

$$\:Y=\:{\beta\:}_{0\:\:\:}+\:\sum\:_{i=1}^{k}{\beta\:}_{i}{X}_{i}+\:\sum\:_{i=1}^{k}{\beta\:}_{ii}{X}_{i}^{2}\:+\:\sum\:_{i=1}^{k}\:\:\sum\:_{i=1}^{k}{\beta\:}_{ij}{X}_{i}{X}_{j}+\:{e}_{0}$$
(3)

Here, Y is the predicted response, βi, βj, βij are the coefficients estimated from the regression, and Xi and Xj are the independent variables in coded levels. β0, k, and e0 are constant coefficients, a number of factors (independent variables), and model error, respectively21.

Table 1 Level and code of variable for BBD for saponin-based ELM.
Full size table

Estimation of the catechol extraction kinetics and mass transfer coefficient

The estimation of kinetic parameters was carried out using the governing equations (Eqs. 4–10), which describe solute transport through the external phase, membrane phase, and internal phase. Experimental concentration-time data were fitted into these equations to evaluate the mass transfer coefficients, namely the external coefficient (KE), membrane coefficient (K0), and internal coefficient (KI). The operating variables employed for the estimation are listed in Table 2.

$$\:ln\left(\frac{{C}_{t=t}}{{C}_{t=0}}\right)=\:-{K}_{obs}\:t$$
(4)

The (Kobs) is calculating the slope of the straight curve generated by the relationship between \(\:{ln}\left(\frac{{C}_{t=t}}{{C}_{t=0}}\right)and\:t\), using Eq. (4). The overall mass transfer coefficient (K0) of the ELM system can be calculated using Eq. (5).

$$\:\frac{1}{{K}_{0}}=\:\frac{1}{{K}_{E}}+\:\frac{1}{{K}_{I}}$$
(5)

KE of the ELM system can be calculated by Eq. (6), in which the Re value is calculated using Eq. (7).

$$\:\frac{{K}_{E}}{\sqrt{N{D}_{MB}}}=2.932\:\times\:{10}^{-7}\left(\frac{{V}_{I}+\:{V}_{M}}{{V}_{I}+\:{V}_{M}+{V}_{E}}\right){\left(\:\frac{{d}_{1}}{{d}_{2}}\right)}^{0.548}{Re}^{1.371}\:$$
(6)
$$\:{R}_{e}=\frac{N{d}_{1}^{2}{\rho\:}_{E}}{{{\upmu\:}}_{E}}$$
(7)

The diffusivity of Catechol extraction, in the MP, was estimated using Eq. (8), and Re is calculated by Eq. (7).

$$\:{D}_{MB}=\:\frac{{117.3\times\:10}^{-18}{\times\:\left({\phi\:M}_{w}\right)}^{0.5}{\times\:T}_{m}}{{{\upmu\:}}_{0}\times\:{\phi\:}_{c}^{0.6}}$$
(8)

KI can be calculated by using Eq. (9).

$$\:{K}_{I}=\:\frac{{K}_{obs}}{a}$$
(9)

Using Eq. (10), a was calculated.

$$\:a=\:\frac{{a}_{i}}{V}=\:\frac{6\alpha\:}{{d}_{32}}$$
(10)
Table 2 Experimental mass transfer and kinetic variables.
Full size table

Results and discussions

Characterization techniques

Interfacial tension analysis of saponin-based ELM

The IFT of the saponin-based ELM was evaluated over 1 to 60 min to assess interfacial dynamics and emulsifying efficiency, as shown in Fig. 2. The IFT decreased from 36.2 mN/m at 1 min to 16.0 mN/m at 60 min, indicating effective interfacial activity of saponin molecules. The IFT dropped sharply during the initial 15 min, reaching 20.1 mN/m, suggesting rapid adsorption of saponin molecules to the oil-water interface. This behavior aligns with surfactant dynamics, where surface-active molecules quickly populate the interface, reducing interfacial energy. Dziza et al. synthesized the emulsion using various surfactants such as saponin, tween 80, and citronellol glucoside. The results indicated that the interfacial properties of saponin, particularly its ability to stabilize oil-in-water emulsions, play a crucial role in enhancing the overall stability of the system22.

Fig. 2
figure 2

IFT variation over time (1 to 60 min) of saponin-based ELM (± 5% error).

Full size image

After 20 min, the IFT reduction slowed, approaching a plateau and reaching near-equilibrium by 30 min. The equilibrium IFT (16.0 mN/m) remained stable from 50 to 60 min, reflecting complete interface coverage by saponin molecules. This behavior confirms saponin’s potential as a natural emulsifier for ELM systems. A similar finding was reported by Schroen et al., using SDS (Sodium dodecylsulfate) as surfactant to study dynamic IFT at different time intervals, showing that as more time passed, the IFT decreased. This indicates the equilibrium value of IFT representing the stabilized condition23. In the present study, the initial drop in IFT enhances droplet formation, while the subsequently stabilized low IFT contributes to droplet integrity and resistance to coalescence. These findings demonstrate that saponin efficiently reduces IFT and stabilizes the emulsions.

Droplet size distribution and stability analysis

The droplet characteristics of the synthesized saponin-based ELM were evaluated using a combination of DLS and optical microscopy to provide complementary insights into size distribution and stability. The DLS analysis (Fig. 3) revealed a Z-average diameter of 0.3628 μm, with the majority of droplets distributed between 0.1 and 0.3 μm. A minor fraction of aggregates around ~ 1.296 μm was also detected, giving rise to a trimodal distribution (0.065 μm, 0.15–0.3 μm, and ~ 1.296 μm) as indicated in Fig.S1. The PDI was found to be 0.701, suggesting a broad size distribution. Although a PDI above 0.5 is typically considered highly polydisperse, such values are frequently reported in emulsion-based systems, especially in ELMs prepared with mixed surfactants. Mohammed et al. reported similar findings, where nano-emulsions synthesized using various surfactants for ice cream production exhibited PDI values greater than 0.5, indicating a broad particle size distribution24. Although the droplet size distribution shows some heterogeneity typical of natural emulsifiers, the relatively small average size (0.3628 μm) suggests that saponin, due to its amphiphilic structure, effectively reduces IFT and forms elastic interfacial films that stabilize smaller droplets and prevent coalescence25. The incorporation of saponin further enhances interfacial elasticity and droplet integrity, contributing to overall emulsion stability even in the presence of size variability.

While DLS provided an intensity-weighted snapshot of the initial droplet size distribution, microscopy was employed to monitor stability evolution over time (0–360 min) as shown in Fig. 4 (a–h). Immediately after preparation [Fig. 4 (a)], the emulsion displayed a uniform distribution of small, spherical droplets, indicating effective emulsification and strong initial stability due to saponin’s surface-active properties. After 30 min [Fig. 4 (b)], the emulsion remained stable with only a slight increase in droplet size, suggesting minimal coalescence. However, by 60 min to 120 min [Fig. 4 (c) and Fig. 4 (d)], noticeable coalescence began to occur, as evidenced by the appearance of larger droplets and a broader size distribution. These changes marked the onset of destabilization, though the emulsion still retained an overall dispersed appearance. As time progressed to 180 min [Fig. 4 (e)], more significant coalescence was observed, with increased droplet size and reduced uniformity, indicating partial breakdown of the interfacial film. Further destabilization was evident at 240 to 300 min [Fig. 4 (f) and Fig. 4 (g)], where droplets appeared larger, less spherical, and unevenly distributed. By 360 min [Fig. 4 (h)], the emulsion exhibited severe instability, characterized by large, coalesced droplets and visible phase separation, confirming the breakdown of the emulsion structure over time. Table 1 shows the emulsion droplet size range based on different times and droplet specifications.

Fig. 3
figure 3

Particle size distribution (µm) analysis of saponin-based ELM by DLS (± 5% error).

Full size image
Fig. 4
figure 4

(a-h). Microscopic images of saponin-based ELM at different time intervals (0–360 min) showing gradual droplet growth and coalescence.

Full size image

Rheological analysis of saponin and saponin-based ELM

The rheological properties of the saponin compound and saponin-based ELM were assessed by measuring viscosities over 0–100 s⁻¹ shear rates, as shown in Fig. 5 (a). Saponin and saponin-based ELM demonstrated shear-thinning property, which is a non-Newtonian behavior, with decreasing viscosity as the shear rate increased. This behavior indicates a pseudoplastic flow profile, typical for emulsified and polymeric systems. A similar finding was reported by Tsibranska et al., who synthesized emulsions using various natural saponin extracts to study their rheological properties and found that the emulsions exhibited non-Newtonian behavior and formed stable emulsions26. At low shear rates, the saponin compound showed a high initial viscosity of approximately 2200 m.Pa.s, which decreased to below 50 m.Pa.s at higher shear rates. This decline suggests strong molecular interactions among saponin molecules, which are disrupted under shear stress. The saponin-based ELM had a lower initial viscosity of approximately 1400 m.Pa.s but maintained a higher viscosity (71.67 m.Pa.s) across the shear rate range compared to pure saponin (10.59 m.Pa.s). This indicates that incorporating saponin into the ELM formulation creates a more robust emulsion network, stabilized through interfacial activity and molecular interactions with system components25,26,27.

Fig. 5
figure 5

(a) The viscosity analysis of saponin compound and saponin-based ELM by rheometer. (temperature 25 °C, shear rate 0.1–100 s[- 1) (± 5% error). (b) The effect of temperature on the viscosity of saponin-based ELM by rheometer analysis [Temperature (20 to 50 °C), and shear rate 25 s−1] (± 5% error).

Full size image

The influence of temperature on saponin-based ELM viscosity was investigated within 20–50 °C, as depicted in Fig. 5 (b). The findings showed an inverse correlation between temperature and viscosity. At 20 °C, the emulsion exhibited a viscosity of approximately 115 m.Pa.s. As the temperature increased, the viscosity gradually reduced, reaching approximately 92 m.Pa.s at 50 °C. This phenomenon can be attributed to increased thermal energy at elevated temperatures, which enhances molecular mobility and reduces intermolecular hydrogen bonding among saponin molecules28,29. Below 35 °C, the viscosity decline was moderate, indicating the compound retains structural resistance to flow. Above 40 °C, the viscosity declined due to the breakdown of micellar structures, which reduces flow resistance30. Maintaining the process temperature below 40 °C may be advantageous to preserve sufficient viscosity for stable emulsion formation. The viscosity behavior confirms that temperature was a critical parameter affecting the flow characteristics of the saponin compound in the ELM26.

Fourier transform infrared spectroscopy analysis

The pure saponin, MP, and saponin-based ELM, FTIR spectra analysis shown in Fig. 6, which were analyzed to confirm the incorporation of components and identify functional groups in ELM formation. The pure saponin shows a broad peak at 3377.63–3215.52 cm⁻¹, corresponding to O-H stretching vibrations, confirming hydroxyl groups typical of saponin molecules. A peak at 1637.51 cm⁻¹ is attributed to C = C or C = O stretch vibrations from triterpenoid or glycosidic moieties31,32. The band at 1015.75 cm⁻¹ represents C-O-C stretching, indicating glycosidic linkages33. The MP, containing Span 80, saponin, and soybean oil, shows peaks at 2924.13 cm⁻¹ and 2852.31 cm⁻¹, consistent with asymmetric and symmetric C-H stretch of methylene groups. A band at 1744.22 cm⁻¹ indicates ester C = O stretching, confirming ester functionalities from Span 80 and soybean oil. Peaks at 1461.04 cm⁻¹, 1159.39 cm⁻¹, and 720.26 cm⁻¹ relate to -CH2 bending, C-O stretching, and long-chain (CH₂) n-rocking vibrations34.

In the ELM spectrum, major characteristic peaks of saponin and MP were retained, confirming successful ELM formation. The O-H stretching band (3377.63–3215.52 cm⁻¹) from saponin remains preserved. C-H stretching bands at 2924.13 cm⁻¹ and 2852.31 cm⁻¹, and ester C = O band at 1744.22 cm⁻¹ confirm MP component incorporation. Peaks at 1637.51 cm⁻¹ and 1159.39 cm⁻¹ indicate interactions among saponin, Span 80, and soybean oil. These results confirmed the successful incorporation of saponin in ELM synthesis. The retention of functional group peaks without significant shifts indicates component interaction through physical forces rather than chemical bonding, validating the stable structure of the saponin-based ELM.

Fig. 6
figure 6

FTIR spectra analysis of pure saponin (blue), MP (red), and the saponin-based ELM (black).

Full size image

Box Behnken design

The impact of the variables and their interrelations was shown in Table S1. The model was evaluated, and the relationships between the response values and independent variables were assessed using Analysis of Variance (ANOVA). Parameter significance was determined using ANOVA, where p-values < 0.05 indicate significant model terms. Here, A, C, D, E, AB, AD, AE, BC, BD, BE, CD, CE, DE, A², B², C², D², E² were significant. A value of F, which was 118.45, showed the model’s significance. Due to noise, there was a 0.01% chance that an F-value this large. The coefficient of determination (R²) was 0.9920, indicating a strong correlation between predicted and experimental values, as shown in Fig. S2. This R² value confirms the model’s goodness of fit. The difference between the predicted R² and adjusted R² values must be below 0.2 to ensure model reliability35,36. If this condition is not met, it may indicate issues with the data or model. The predicted R² for catechol extraction efficiency is 0.9666, while the adjusted R² is 0.9837. This difference falls within the acceptable range, confirming the model’s reliability. Furthermore, Eq. (11) provides a visualization of the significant factors and their interactions.

$$\begin{array}{l} \% Extraction = 74.87--0.1633B{\rm{ }}--0.3496C{\rm{ }} + 1.43D{\rm{ }}\\ + 0.7544E{\rm{ }} + 1.68AB{\rm{ }} + 0.0160AC{\rm{ }} - 0.6650AD{\rm{ }} + 2.92AE\\ + 1.03BC{\rm{ }} + 2.56BD{\rm{ }} - 2.06BE{\rm{ }} - 2.09CD{\rm{ }} - 1.34CE{\rm{ }} - 3.57DE{\rm{ }}\\ - 2.21{A^2}--3.45{B^2} - 1.62{C^2} - 0.6899{D^2}--0.7249{E^2} \end{array}$$
(11)

The significance of the parameters, their interactions, and effects is detailed in Table S1. According to the P-values, the first-order main effects of factors A and D are statistically significant (P < 0.05). Although the second-order terms were found to be significant, the linear terms remain crucial and should not be excluded from the final model. Overall, the data presented in Table S1 corroborate the model’s effectiveness.

The fitted quadratic model exhibited statistical significance with a very high F-value and p < 0.0001, indicating that the model adequately explained the variability in the experimental data. The determination coefficient (R²) was found to be > 0.98, demonstrating an excellent correlation between predicted and experimental values. The non-significant lack of fit further confirmed the suitability of the model for describing the process behavior.

Parameter optimizations

Effect of span 80 concentration

The response surface plot model predictions indicated that an optimal extraction efficiency (~ 75%) was achieved at a Span 80 concentration of 2–2.5.5 v/v% with 30 min contact time, as shown in Fig. 7 (a); the contour plot is shown in [Fig.S2] for better visualization and understanding. Below 2 v/v%, the extraction efficiency declined due to insufficient surfactant coverage at the oil-water interface, resulting in unstable droplets, rapid coalescence, and reduced solute transport. Conversely, concentrations above 3.5 v/v% decreased efficiency owing to increased viscosity of the MP and formation of a densely packed interfacial layer, which hindered molecular diffusion37. The ANOVA results confirmed the significance of Span 80 with a strong linear effect (F = 134.40, p < 0.0001) and quadratic effect (F = 49.02, p < 0.0001), validating the observed curvature in response. Significant interactions with saponin (AB, F = 90.65, p < 0.0001) and contact time (AD, F = 14.25, p = 0.0013) further emphasized that Span 80 modulates the stability of the ELM system in synergy with other process variables.

Fig. 7
figure 7

(a). 3D plot of extraction efficiency analysis of catechol, the effect of Span 80 concentration (v/v%) vs. Time (min). (b). 3D plot of extraction efficiency analysis of catechol, the effect of Saponin concentration (v/v%) vs. Time (min). (c). 3D plot of extraction efficiency analysis of catechol, the effect of stripping agent concentration (M) vs. Time (min). (d). 3D plot of extraction efficiency analysis of catechol, the effect of MP: IP ratio vs. Time (min).

Full size image

Effect of saponin concentration

The response surface plot model predicted maximum extraction (~ 75–76%) at a saponin concentration of 2.5–3.5 v/v% with 30 min contact time, as shown in Fig. 7 (b), the contour plot is shown in [Fig.S3] for better visualization and understanding. At concentrations below 2.5 v/v%, interfacial coverage was insufficient, leading to droplet coalescence and reduced efficiency7,38. At higher concentrations (> 4 v/v%), excessive viscosity and formation of a thick interfacial layer hindered solute transport. Although the linear effect of saponin concentration was not statistically significant (F = 3.39, p = 0.0814), the quadratic effect was highly significant (F = 117.50, p < 0.0001), confirming the curvature of the response. Interaction effects with Span 80 (AB, F = 90.65, p < 0.0001), stripping agent (BC, F = 28.01, p < 0.0001), and contact time (BD, F = 211.11, p < 0.0001) were also highly significant, demonstrating that saponin acts as a synergistic bio-emulsifier in stabilizing the ELM droplets and enhancing mass transfer.

Effect of internal phase (Stripping agent) concentration

The response surface plot predicted an optimal range of stripping agent concentration was 0.1–0.15 M, providing maximum extraction efficiency (~ 76%) with 30 min contact time, as shown in Fig. 7 (c). The contour plot is shown in [Fig. S4] for better visualization and understanding. At concentrations below 0.1 M, the driving force for solute transport was insufficient, while concentrations above 0.2 M caused osmotic imbalance, droplet swelling, and rupture, leading to emulsion instability39. ANOVA confirmed the significance of stripping agent concentration, with both linear (F = 14.92, p = 0.0010) and quadratic (F = 28.83, p < 0.0001) effects contributing strongly. Interactions with saponin (BC, F = 28.01, p < 0.0001) and contact time (CD, F = 141.38, p < 0.0001) were also highly significant, reflecting the concentration-dependent and condition-sensitive role of the stripping agent in ensuring effective extraction.

The response surface plot revealed a strong synergistic influence of contact time and MP: IP ratio on extraction performance. At low contact times (1–5 min) and low MP: IP ratios (1:1), the extraction efficiency remained ~ 70% due to insufficient solute diffusion and weaker concentration gradients. As both the contact time and MP: IP ratio were increased, extraction efficiency steadily increased, reaching ~ 96% at 30 min and an MP ratio of 3, as shown in Fig. 7 (d), the contour plot is shown in [Fig. S5] for better visualization and understanding. This demonstrates that prolonged contact enhances solute transfer kinetics, while a higher MP: IP ratio provides greater stripping capacity40. Statistical analysis confirmed the strong linear significance of time (F = 261.87, p < 0.0001) and MP: IP ratio (F = 69.48, p < 0.0001). Their interaction (DE) was the most significant model term (F = 411.12, p < 0.0001), underscoring the necessity of their combined optimization for achieving maximum extraction efficiency. Quadratic terms for time and MP: IP factors were also significant (time: F = 4.77, p = 0.0416; M/P ratio: F = 5.78, p = 0.0265), validating the curvature in the response and confirming the defined optimum ranges. These results highlight that neither variable alone is sufficient, and their synergistic effect is critical for enhancing overall process performance.

Overall, the results from ANOVA, model predictions, and experimental validation confirm that Span 80 (2–2.5.5%), saponin (2.5–3.5%), stripping agent concentration (0.1–0.15 M), MP: IP ratio (3:1), and contact time (30 min) represent the optimum operating conditions. Under these optimized conditions, the model predicted an extraction efficiency of ~ 77%, which was in close agreement with the experimental findings, validating the robustness and predictive power of the RSM-based optimization approach.

Experimental investigation of catechol extraction efficiency and membrane stability

The optimized parameter values through the BBD model were used to perform the experiment for the catechol extraction to investigate the actual extraction efficiency. The membrane stability is also checked by using the membrane breakage calculation, using Eq. (2). The effect of the initial concentration of catechol on the extraction efficiency (%) and membrane breakage (%) is illustrated in Fig. 8. The extraction efficiency decreased progressively with increasing catechol concentrations, whereas membrane breakage exhibited a rising trend. At a low catechol concentration of 10 mg/L, the ELM system achieved a maximum extraction efficiency of approximately 88.9% with minimal membrane breakage (2%). This high efficiency and stability at lower solute concentration can be ascribed to the accessibility of sufficient surfactant molecules and intact interfacial film integrity, facilitating efficient transport of catechol molecules across the membrane40,41. As the catechol concentration increased to 20 and 30 mg/L, extraction efficiencies slightly declined to 80% and 75%, respectively, accompanied by an increase in membrane breakage to approximately 5% and 7%. The moderate rise in membrane rupture at these concentrations may be due to higher solute loading, which increases osmotic pressure gradients across the membrane and enhances droplet coalescence tendencies42,43. At higher catechol concentrations of 40 and 50 mg/L, extraction efficiency further decreased to 66.7% and 50%, while membrane breakage escalated sharply to 11% and 15%, respectively. The increased instability at these levels suggests that higher solute loading promotes swelling of the emulsion, leading to coalescence and rupture of the thin liquid membrane films. This phenomenon can be explained by excessive accumulation of catechol at the membrane interface, which disturbs the surfactant-stabilized interfacial layer and compromises the mechanical strength of the membrane.

Fig. 8
figure 8

The effect of catechol concentration (mg/L) on extraction efficiency (%) and membrane breakage (%) in the saponin-based ELM (± 5% error).

Full size image

Extraction mechanism of catechol through saponin-based ELM

The schematic diagram, as shown in Fig. 9, illustrates the proposed mechanism for catechol extraction using a saponin-based ELM. The emulsion droplets are composed of an IP containing the stripping agent (typically NaOH), encapsulated within an MP formed from soybean oil and stabilized by a dual-surfactant system: the synthetic non-ionic surfactant Span 80 and the natural anionic bio-emulsifier saponin. Upon introduction into the catechol-containing external aqueous phase, due to a concentration gradient, the catechol molecules diffuse into the MP 40. At the external oil-water interface, catechol partitions from the aqueous medium into the hydrophobic MP, where it continues diffusing toward the IP. Once it reaches the inner interface, catechol is captured by the stripping agent, effectively completing the extraction process. The amphiphilic nature of saponin contributes to the creation of an elastic and stable interfacial film, which enhances droplet stability and prevents coalescence during extraction26. Span 80, being non-ionic and lipophilic, complements saponin by further reducing IFT and supporting emulsion formation. In the MP, the combined action of Span 80 and saponin leads to the formation of stabilized micelle-like structures around the aqueous droplets, facilitating efficient transport of catechol via a carrier-mediated diffusion mechanism44,45. The effectiveness of catechol extraction was estimated under optimized conditions, and the results. Soybean oil was used as a green diluent, while saponin served as a bio-degradable and eco-friendly emulsifier. This combination resulted in the highest catechol extraction efficiency.

Fig. 9
figure 9

The schematic representation of the transport mechanism of catechol through the saponin-based ELM.

Full size image

Estimation of the catechol extraction kinetics and mass transfer coefficient

The mass transfer and extraction kinetics in ELM are very important because the whole process of extraction and separation depends on how efficiently a solute moves through the different phases, as discussed in Sect. 3.5. The governing equations for mass transfer and kinetics have been shown in the methods section from Eq. 4 to 10. The experimental values of the operating variables are presented in Table 2.

The kinetic analysis of the present work revealed that the external mass transfer coefficient (KE​) was higher than the membrane (K0) and internal (KI) coefficients, indicating that internal and membrane diffusion govern the overall resistance. This observation follows a similar trend to those reported in the literature for λ-cyhalothrin, chlorpyrifos, and Abamectin extraction using ELM and Pickering ELM systems, where external diffusion was found to be relatively fast, while membrane and internal phase resistances controlled the transport process4,6,46.

The present work demonstrates distinct advantages over previously reported systems. Conventional ELM studies often reported very high external mass transfer coefficients; however, these were typically accompanied by significant stability problems and rapid emulsion breakage. In contrast, the current system achieves a moderate external transfer rate (KE = 5.01 × 10⁻⁶ m/s) along with balanced membrane and internal resistances (KI = 3.48 × 10⁻⁸ m/s and K0 = 3.46 × 10⁻⁸ m/s). This controlled distribution of resistances not only agrees with reported kinetic values but also provides a more regulated diffusion process, which is essential for maintaining long-term emulsion stability. Furthermore, the optimized parameters, span 80 (2 v/v%), saponin (3 v/v%), MP: IP ratio of 3:1, stripping agent concentration of 0.15 M, and a contact time of 30 min, were found to facilitate the balance in transport resistances. Under these operating conditions, the system sustained moderate KE with controlled KI and K0, thereby minimizing breakage, enhancing stability, and ultimately achieving high extraction efficiency.

Conclusion

The present study developed and optimized a novel saponin-based ELM for efficient catechol extraction. The integration of saponin as a natural bio-emulsifier with Span80 and soybean oil resulted in a stable ELM formulation. Characterization techniques, including IFT, DLS, rheological studies, FTIR, and microscopy, confirmed the formation of a well-structured emulsion system with properties suitable for extraction. Optimization using BBD showed that saponin concentration, stripping agent concentration, and MP to IP ratio were critical factors affecting extraction efficiency. Under optimized conditions, the ELM achieved 88.9% catechol extraction efficiency at low solute concentrations, with minimal membrane breakage (2%). The extraction mechanism revealed synergistic roles of saponin and Span 80 in stabilizing the emulsion and facilitating solute transport. Kinetic analysis provided insights into mass transfer characteristics, with a mass transfer coefficient (K0) of 3.46 × 10−8 m/s. This study demonstrates saponin-based ELM potential as an environmentally friendly alternative for phenolic compound extraction. The developed saponin-based ELM system represents a promising approach for sustainable liquid-liquid extraction, combining efficiency with environmental compatibility. This research advances green separation technologies and opens avenues for further optimization and application.