Nanoemulsion design of Cynara cardunculus leaves extract rich in sesquiterpene lactones

Abstract

Background

A Cynara cardunculus leaf extract enriched in sesquiterpene lactones (CcLERSLs) exhibited strong phytotoxic activity, notably inhibiting weed root growth. However, the poor water solubility limits practical application. This study aimed to develop and optimize an oil‐in‐water nanoemulsion containing CcLERSLs, using grape seed oil as carrier, and pectin and Tween®80 as emulsifiers. Ultrasound‐assisted emulsification and response surface methodology were applied to optimize droplet size, stability and bioavailability. Phytotoxicity and storage stability were evaluated.

Results

Models for droplet size and zeta‐potential identified ultrasound amplitude and Tween®80 content as significant variables, respectively. Model validation confirmed high predictive accuracy, particularly for droplet size, and this guided the formulation of the CcLERSLs nanoemulsion (5% w/w GSO, 4% Tween® 80, 1% pectin, 80% ultrasound amplitude). Phytotoxicity against Portulaca oleracea, Plantago lanceolata and Phalaris arundinacea showed a ≥3‐fold increase in activity compared with the crude extract, particularly for root inhibition. P. oleracea was the most sensitive and the nanoemulsion outperformed the commercial herbicide at 400–800 ppm. The nanoemulsion remained physically stable at 4 and 25 °C for 90 days but degraded significantly at 45° C.

Conclusions

Nanoemulsion formulation enhances bioavailability of the CcLERSLs and significantly improves its phytotoxic activity, supporting its potential as a promising natural herbicide. © 2026 The Author(s). Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

This work introduces a novel formulation strategy that advances the potential of Cynara cardunculus leaf extract enriched in sesquiterpene lactones as a bioherbicide and contributes to the broader development of sustainable weed management tools based on natural products.

1. INTRODUCTION

Bioherbicides derived from natural or biological sources provide an eco‐friendly alternative to synthetic herbicides by mimicking natural weed control mechanisms. Consisting of organisms or plant‐based compounds, they are being increasingly adopted in sustainable agriculture in response to growing concerns over herbicide resistance and environmental impact. Advances in formulation technologies, particularly nanoencapsulation and emulsification, have enhanced the efficacy, stability and applicability of such herbicides. ¹ , ² Nanoemulsions in particular offer improved solubility, targeted delivery and reduced active ingredient requirements, although challenges in formulation stability and scalability remain. ¹

Very few studies have been carried out on plant‐derived nanoemulsions. Zainuddin et al. ³ developed a nanoemulsion that contained Parthenium hysterophorus crude extract using palm kernel oil ester (oil phase), Tensiofix and Tween® 80 (surfactants), and water (aqueous phase). A mixture design approach was employed to optimize formulation parameters to yield a stable nanoemulsion with a droplet size of 140 nm. Compared to the crude extract, the nanoemulsion exhibited enhanced phytotoxicity against Diodia ocimifolia and achieved complete inhibition at 5 versus 10 g L⁻¹ for the crude extract, thus indicating improved bioavailability. Likewise, Kaur et al. ⁴ employed ultrasonic emulsification to formulate a nanoemulsion containing Foeniculum vulgare essential oil. The formulation exhibited long‐term stability (30 days) and the droplet size was 130 nm. This formulation demonstrated superior phytotoxicity relative to the crude oil against Phalaris minor, Avena ludoviciana, Rumex dentatus and Medicago denticulata. Chen et al. ⁵ developed Artemisia argyi essential oil‐loaded nanoemulsions using Cremophor EL and polyoxyethylene lauryl ether as surfactants. These formulations (droplets <200 nm) showed enhanced phytotoxicity in greenhouse trials and enabled a two‐ to four‐fold reduction in the required concentration when compared to aqueous formulations. Stability assessments confirmed resilience under various conditions. However, temperature sensitivity and short‐term stratification upon dilution highlighted the need for immediate application after preparation.

Sesquiterpene lactones (SLs) are a family of compounds that show phytotoxic activity, making them promising candidates for sustainable weed management. ⁶ , ⁷ These compounds are mainly found in the Asteraceae family, including Cynara cardunculus L., the leaves of which provide 94.5 g SLs/kg dry weight. ⁸ Our previous research showed the phytotoxic potential of C. cardunculus leaves extract enriched in SLs (CcLERSLs) against a panel of nine weed species in the pre‐emergence stage, with a particularly marked effect on root growth inhibition, and also against Portulaca oleracea in the post‐emergence stage, with growth inhibition attributed to chlorophyll degradation and oxidative stress induction. ⁹ However, the poor water solubility of these SLs limits their practical application. In an effort to address this problem, we propose a nanoemulsion‐based delivery system to enhance bioavailability and stability under field conditions.

The aim of this study was to develop and optimize a stable oil‐in‐water nanoemulsion containing CcLERSLs, with the primary objective of enhancing the physical stability and phytotoxic potential of the system. Grape seed oil (GSO) was selected as the oil phase owing to its high content of unsaturated fatty acids and antioxidant properties, which contribute to emulsion stability and protect against oxidative degradation. ¹⁰ , ¹¹ Pectin (a natural polysaccharide) ¹² and Tween® 80 (a synthetic non‐ionic surfactant) ¹³ were used as emulsifiers owing to their effectiveness in reducing interfacial tension and preventing droplet aggregation. Moreover, the carrier components GSO, pectin and Tween® 80 are all regarded as readily biodegradable under environmental conditions, which supports the sustainability claim of the formulation. Their degradation predominantly yields low‐toxicity end products such as CO₂, water and biomass, and none of these excipients are expected to persist or bioaccumulate to a concerning extent in soil or aquatic systems. ¹⁴ , ¹⁵ , ¹⁶ Ultrasound‐assisted emulsification was employed to obtain a smaller droplet size and improve both stability and bioavailability. ¹⁷ , ¹⁸ Optimization of the formulation parameters–oil concentration, emulsifier composition and ultrasound amplitude–was conducted using the response surface methodology (RSM). Finally, the phytotoxic activity of the optimized nanoemulsion was evaluated against three weed species and its physical stability was assessed at different storage temperatures.

2. MATERIAL AND METHODS

2.1. Chemicals and plant material

Absolute ethanol (≥99.8%) supplied by BDH Prolabo (Fontenay‐sous‐Bois, France) was used as the extraction solvent. Milli‐Q water was obtained from a Millipore Ultra‐Pure Water System, Model Direct‐Q5 (Merck, Cambridge, MA, USA). Tween® 80 was purchased from Scharlau (Barcelona, Spain) and grape seed oil was obtained from Graduva (Aveiro, Portugal). Pectin was sourced from Sigma‐Aldrich (St Louis, MO, USA). SLs standards for ultra performance liquid chromatogrphy ‐ quantitative time‐of‐flight (UPLC‐QTOF) analysis were provided by the Allelopathy Group at the University of Cádiz (santamarine, cynaropicrin, aguerin B and grosheimin) and prepared according to Rial et al. ¹⁹ Deacylcynaropicrin was obtained according to Scavo et al. ²⁰ Dimethyl sulfoxide was acquired from Panreac (Barcelona, Spain).

The C. cardunculus L. (DC) leaves were collected in February 2021 and processed as described by Brás et al. ²¹ The leaves were air‐dried at room temperature to constant weight (with a moisture content below the detection limit of 1%). Dried leaves were ground using a domestic grinder (Moulinex, France) to give a powder with a particle size of <500 μm. The powdered leaves were vacuum‐sealed and stored before use.

2.2. Extraction procedure and quantification

A C. cardunculus leaf extract enriched in SLs was obtained according to the process described by Brás et al. ²² SLs were then quantified by UPLC‐QTOF under the following conditions: 120 °C source temperature, 450 °C desolvation temperature, 850 L h⁻¹ gas desolvation flow, 20 L h⁻¹ gas cone flow and 40 V cone voltage. The resolution in mass was 22 000 and the mass range was 150–800 Da. A Kinetex 1.7 μm F 5100 Å 100 × 2.1 mm (Phenomenex, Torrance, CA, USA) chromatography column was used, maintained at 65 °C. The mobile phase consisted of solvent A (water, 0.1% formic acid) and solvent B (MeOH, 0.1% formic acid) and the flow rate was set to 0.4 mL min⁻¹. The linear gradient system was as follows: 0–0.5 min, 30% B; 0.5−1 min, to 50% B; 1–2.5 min to 70% B; 2.5–3 min, to 95% B; 3–4 min to 30% B, 4–5 min, 30% B. The injection volume was 5 μL. The sesquiterpene lactone standards were dissolved in methanol at 1000 ppm and diluted with water from 10 000 to 1 ppb. Samples were filtered through a 0.22‐μm pore size membrane filter (Phenomenex, Torrance, CA, USA). Total reducing sugars were quantified using the DNS method. ²³ Total phenolic content was determined by the Folin–Ciocalteau assay, as described by Brás et al. ²¹ Chlorophylls were determined as described by Rosa et al. ²⁴

2.3. Nanoemulsion design

2.3.1. Formulation design

Oil‐in‐water CcLERSLs nanoemulsions were produced in this study. The emulsions were formulated by combining an oil phase and an aqueous phase. The CcLERSLs, GSO and Tween® 80 were used as the oil phase, whereas the aqueous phase consisted of deionized water and pectin. The oil phase was added slowly to the aqueous phase at 25 °C with magnetic stirring (ELMI, Riga, Latvia) at 800 rpm for 15 min to prepare a coarse emulsion. The CcLERSLs concentration was fixed at 0.5% w/w in each coarse emulsion. The final CcLERSLs nanoemulsions were then prepared using an ultrasonic probe. A 20 kHz ultrasonic device (Sonopuls HD3200; Bandelin, Berlin, Germany) equipped with a 13‐mm‐diameter titanium alloy (Ti‐AI6‐V4) probe (Bandelin) was used to produce disruptive forces that decreased the droplet size in the coarse emulsion to produce a nanoemulsion. Sonication was carried out in pulse mode operation with a duty cycle of 20% (30 s on/120 s off) and with 10 pulse repetitions. The formulation was cooled with ice in order to maintain a constant temperature. ¹⁷

2.3.2. Nanoemulsion optimization using response surface methodology experimental design

Box–Behnken design (BBD) was used in conjunction with the response surface methodology (RSM) to evaluate the effects of four independent factors, namely the GSO in the range of 1–5% (w/w) (X ₁), the Tween® 80 concentration range from 1 to 10% (w/w) (X ₂), the concentration of pectin from 1 to 5% (w/w) (X ₃) and the ultrasound amplitude in the range from 20 to 80% (X ₄), using a formulation with a 0.5% w/w CcLERSLs concentration. Three responses were evaluated, namely the droplet size (Y ₁), the zeta‐potential (Y ₂) and the viscosity (Y ₃). All independent variables were run at three levels in term of BBD for each of the individual coded value (−1, 0 and 1) and this resulted in 27 experimental runs including three center points (Table 1).

Table 1.

Box–Behnken design coded matrix. Values used (−1, 0, 1) for GSO (1%, 3%, 5%), for Tween® 80 (1%, 5.5%, 10%), for pectin (1%, 3%, 5%) and for ultrasound amplitude (20%, 50%, 80%)

Run	Factors
	GSO (X 1)	Tween® 80 (X 2)	Pectin (X 3)	Ultrasound amplitude (X 4)
1	0	0	1	1
2	−1	0	0	−1
3	0	−1	0	−1
4	0	0	−1	−1
5	0	1	0	1
6	0	1	1	0
7	1	−1	0	0
8	1	0	0	−1
9	0	0	0	0
10	0	0	0	0
11	0	−1	−1	0
12	0	1	−1	0
13	−1	0	−1	0
14	0	−1	0	1
15	0	0	−1	1
16	0	0	1	−1
17	−1	0	1	0
18	−1	−1	0	0
19	1	0	0	1
20	1	0	1	0
21	−1	1	0	0
22	−1	0	0	1
23	0	1	0	−1
24	1	0	–1	0
25	1	1	0	0
26	0	0	0	0
27	0	–1	1	0

minitab software (v19.0; Minitab, State College, PA, USA) was used for the experimental design and analysis of the results. Analysis of variance (ANOVA) was applied to check the statistical significance of the individual factors and their interaction effects and to evaluate the fitting quality of the polynomial model. A second‐order polynomial equation including all possible main, interaction and quadratic effects was applied as follows:

$${\mathit{Y}}_{\mathrm{i}}=\mathit{\beta }0+{\mathit{\beta }}_1{\mathit{X}}_1+{\mathit{\beta }}_2{\mathit{X}}_2+{\mathit{\beta }}_3{\mathit{X}}_3+{\mathit{\beta }}_4{\mathit{X}}_4+{\mathit{\beta }}_{12}{\mathit{X}}_1{\mathit{X}}_2+{\mathit{\beta }}_{13}{\mathit{X}}_1{\mathit{X}}_3+{\mathit{\beta }}_{14}{\mathit{X}}_1{\mathit{X}}_4+{\mathit{\beta }}_{23}{\mathit{X}}_2{\mathit{X}}_3+{\mathit{\beta }}_{24}{\mathit{X}}_2{\mathit{X}}_4+{\mathit{\beta }}_{34}{\mathit{X}}_3{\mathit{X}}_4+{\mathit{\beta }}_{11}{\mathit{X}}_{12}+{\mathit{\beta }}_{22}{\mathit{X}}_{22}+{\mathit{\beta }}_{33}{\mathit{X}}_{32}+{\mathit{\beta }}_{44}{\mathit{X}}_{42}$$ (1)

In this equation Y _i are the dependent variables, whereas X ₁, X ₂, X ₃ and X ₄ are the studied factors. β ₀ corresponds to the ordinate coefficients; β _i represents the linear coefficients; β _ij are the cross‐product coefficients between two factors; and β _ii are the quadratic coefficients.

2.3.3. Measurement of droplet size, zeta‐potential and viscosity

The droplet size and zeta‐potential of the CcLERSLs nanoemulsions were evaluated by dynamic light scattering with noninvasive back scattering using a particle size analyzer (Zetasizer Nano ZS, Malvern Panalytical, Malvern, UK), comprising a 4 mW helium‐neon gas laser beam (λ = 633 nm) with a scatter angle 173° for droplet size measurements and 13° for zeta‐potential measurements. Each sample was diluted with deionized water at a 1:10 ratio to avoid interparticle interactions and multiple scattering. Both measurements were performed at 25 °C in a folded capillary cell (DTS1070) for droplet measurements and folded capillary cell incorporated with gold‐plated electrodes (DTS1060C) in water as the dispersion medium. Three replicates were obtained for each measurement. Characterization of the CcLERSLs nanoemulsions was carried out within 48 h of preparation. The viscosities of the CcLERSLs nanoemulsions were determined on a VT 550 rotational viscometer (Haake/Thermo Fisher Scientific), using the concentric cylinder MV‐DIN at 20 °C. The tests were carried out using a controlled shear rate (ranging from 0 to 300 s⁻¹) and 10 points were collected. The resulting flow curves were evaluated according to the Newtonian model.

2.3.4. Morphological evaluation of the CcLERSLs nanoemulsions

The morphology and particle size were studied using a NovaNanoSEM 450 (FEI, Hillsboro, OR, USA) scanning electron microscope. Bright‐field (BF) and secondary electron (SE) images were recorded using the annular‐type detector installed on the microscope and recorded at voltages of 30 kV using a working distance in the range 6.5–6.6 mm. A LC300‐Cu‐150 grid with lacey carbon (C) film was used to prepare the nanoformulation samples. Both the CcLERSLs nanoemulsion and the nanoemulsion without extract were prepared by dispersing a drop of the solution onto a copper grid with lacey C. The copper grid was placed on Whatman No. 1 filter paper (47 mm diameter) inside a Petri dish. In this way, the excess solvent from the preparation could be absorbed by the paper. Additionally, the grid was kept inside the closed Petri dish for 24 h to enhance the drying process and reduce water contamination in the electron microscope.

2.4. Evaluation of phytotoxic activity

The phytotoxic activity of the nanoemulsion was evaluated against the weed species P. oleracea, Plantago lanceolata and Phalaris arundinacea (Cantueso Natural Seeds, Córdoba, Spain). Bioassays were conducted in Petri dishes as described by Rial et al., ¹⁹ with some modifications. CcLERSLs nanoemulsion was tested at different extract concentrations (100, 200, 400, 600 and 800 ppm) on 20 seeds of each species. The commercial herbicide Stomp® Aqua (BASF Agricultural Solutions, Gatersleben, Germany), the blank nanoemulsion (without extract) and the crude extract dissolved in distilled water were all tested at the same concentrations as the CcLERSLs nanoemulsion, in order to allow direct comparison of their phytotoxic effects. Four replicates were prepared for each treatment. Seeds were further incubated at 25 °C in the dark for 7 days for P. oleracea and Pl. lanceolata, and 10 days for Ph. arundinacea. The germination rate and root and shoot lengths were measured using a Fitomed system, which allowed for automatic data acquisition and statistical analysis via associated software. The data were analyzed statistically using Welch's test, with significance fixed at P < 0.01 and P < 0.05. The results are shown as a percentage difference from the control. Zero represents control, positive values correspond to stimulation, and negative values correspond to inhibition.

2.5. Storage stability tests

After identifying the optimal composition of the CcLERSLs nanoemulsion, storage stability was evaluated at different temperatures and days of storage. The formulations prepared were maintained in the dark in controlled temperature chambers at 4 °C, 25 °C and 45 °C for 0, 30 and 90 days of storage. Further measurements were carried out to assess changes in droplet size, polydispersity index (PDI) and zeta‐potential. PDI was measured by dynamic light scattering under the same conditions as described above. Experiments were carried out in triplicate. A two‐way ANOVA analysis was performed to determine whether there were any significant changes between the initial and subsequent incubation times. This analysis was performed using prism v5.0 software (GraphPad, San Diego, CA, USA).

3. RESULTS AND DISCUSSION

3.1. Chemical composition of C. cardunculus leaves extract

The extract obtained using the ethanolic pulsed ultrasound‐assisted extraction method was analyzed for its content of SLs, with particular focus on the most phytotoxic compounds: cynaropicrin, grosheimin, aguerin B and deacylcynaropicrin. The total phenolic content, chlorophyll content and sugar content also were determined. The results are presented in Table 2.

Table 2.

Chemical composition of C. cardunculus leaves extract. Values are represented as mean ± SD

Total SLs (mg SL/g extract)
Cynaropicrin	299.35 ± 6.70
Grosheimin	3.24 ± 0.64
Aguerin B	89.36 ± 3.75
Deacylcynaropicrin	15.48 ± 0.31
Total phenolic content (mg gallic acid equivalents/g extract)	28.03 ± 1.78
Total chlorophyll content (mg Chl/g extract)	1.75 ± 0.02
Sugar content (mg glucose equivalents/g extract)	27.21 ± 3.76

The SLs content in the extract was 41% and this represents a high concentration associated with strong phytotoxic activity, as previously reported by Rosa et al. ⁹ The extraction yield was 135.2 g extract per kg leaf material.

3.2. Response surface methodology experiment

The purpose of the Design of Experiment (DoE) methodology was to elucidate how formulation and process variables influence the key physicochemical properties of the system. DoE was applied to evaluate the effects of four independent variables, namely the concentrations of GSO, Tween® 80 and pectin as well as the ultrasound amplitude applied. The DoE framework allowed the systematic assessment of how these formulation factors affect critical physicochemical responses, namely droplet size, zeta‐potential and viscosity. These characteristics are key in determining the stability of nanoemulsions and their potential for interaction with biological targets. ³ , ²⁵ , ²⁶ , ²⁷ The DoE methodology provides a robust statistical and experimental basis for identifying the optimal formulation conditions that yield desirable physicochemical characteristics.

Box–Behnken design was used to optimize the levels of the formulation ingredients (GSO, Tween® 80 and pectin) and ultrasound conditions for the nanoemulsion preparation (ultrasound amplitude percentage). The ranges used for each variable were 1–5% (w/w) for GSO and pectin, 1–10% (w/w) for Tween® 80 and 20–80% for ultrasound amplitude percentage (Table 1). The results obtained for each experimental run regarding the experimental and calculated values are represented in Table S1. Relative differences between experimental and calculated values were obtained in order to evaluate the fitting properties of the resulting model for each response (Fig. 1). In terms of Y ₁, an average difference of 13% was observed, with a range from 0% to 33%, between experimental and calculated values with only four experiments having a difference >20%. In the case of Y ₂ the average difference obtained was 19%, with a range from 0% to 85%, with seven experiments with a difference >20%. Finally, for Y ₃ the average difference was 28%, with a range from 3% to 83%, with 14 experiments showing a difference greater than 20%. The three models derived from the various response evaluations indicate that viscosity had the largest discrepancies between experimental and predicted values. This finding shows that in the case of viscosity the model was not particularly suitable for explaining the results.

Figure 1.

Absolute percentage differences between experimental and model‐predicted values for each response in the Box–Behnken design.

First, it is necessary to assess the effects of the factors on the different variables. In this regard, it can be observed that the droplet size, under the conditions tested, was in the range between 222 nm and 714 nm, which means that the difference between the minimum and maximum values is approximately three times the minimum value. In the case of zeta‐potential the effect follows a similar trend, with the highest recorded negative value (−20.7 mV) being ≈5.5 times the lowest negative value (−3.7 mV). However, in the case of viscosity the experimental results indicate that the highest value (3.909 Pa.sⁿ) is approximately two million‐fold greater than the lowest value (0.002 Pa.sⁿ). Furthermore, it is clear that the viscosity results are not distributed normally (Fig. S2).

Given the experimental ranges outlined above, a reliable regression can be expected for droplet size and zeta‐potential. However, in the case of viscosity the extreme variation suggests that a reliable predictive model cannot be expected. In any case, ANOVA was used to determine the statistical significance of the model along with the impact of the variables and any potential interactions between them. A second‐order polynomial equation was used to correlate the droplet size, zeta‐potential and viscosity (i.e. the responses) with GSO, Tween® 80, pectin concentration and ultrasonic amplitude percentage (i.e. the experimental factors). The coefficients for the polynomial equation are shown in Table 3 along with their significance (P‐values).

Table 3.

Coefficients for the second‐order polynomial equation and p‐values for the linear and different interactions from the Box–Behnken design

	Y 1: droplet size		Y 2: zeta‐potential		Y 3: viscosity
Term	Coefficient	P‐value	Coefficient	P‐value	Coefficient	P‐value
Constant	−0.0027	0.0000	−0.4058	0.0000	0.5694	0.0000
Linear effects
Pectin	−0.0001	0.7750	−0.0224	0.2460	0.3960	0.0000
T80	−0.0001	0.6680	−0.1207	0.0000	0.0370	0.1590
GSO	−0.0002	0.3410	−0.0143	0.4910	0.0175	0.4920
US amplitude	−0.0008	0.0010	0.0017	0.9280	0.0104	0.6810
Quadratic effects
Pectin*Pectin	−0.0001	0.7050	−0.0171	0.5500	0.1887	0.0000
T80*T80	0.0002	0.4470	0.0339	0.2610	−0.0082	0.8280
GSO*GSO	−0.0001	0.8470	0.0178	0.5470	0.0191	0.6160
US amplitude*US amplitude	−0.0002	0.5280	0.0379	0.1990	0.0012	0.9750
Interaction effects
Pectin*T80	0.0002	0.5600	−0.0049	0.8790	0.0691	0.1310
Pectin*GSO	0.0004	0.2900	−0.0500	0.1430	0.0829	0.0760
Pectin*US amplitude	0.0001	0.7930	0.0258	0.4330	0.0260	0.5540
T80*GSO	−0.0002	0.5140	−0.0050	0.9020	0.0085	0.8450
T80*US amplitude	−0.0003	0.4030	0.0258	0.4330	0.0044	0.9190
GSO*US amplitude	−0.0003	0.3200	−0.0198	0.5450	−0.0045	0.9190
Lack‐of‐fit (P‐value)	0.826		0.152		0.019

First and foremost, it is important to highlight that both the droplet size and zeta‐potential models exhibit a nonsignificant lack of fit (P‐value >0.05). This finding indicates that in both cases the models adequately fit the experimental results under the tested factor conditions. As one would expect, in the case of viscosity the lack of fit is significant, which means that there is poor agreement between the experimental results and the predicted values generated by the model. This issue limits the applicability of the model and suggests that alternative analytical tools or modeling approaches would be necessary.

3.2.1. Droplet size ( Y ₁ )

In terms of Y ₁ it was observed that the only factor with a significant effect was X ₄ (US amplitude) in the linear terms (P = 0.0010), as shown in the Pareto chart (Fig. 2). For this factor a negative influence was observed (coefficient of −0.0008), which means that the higher the applied ultrasound amplitude applied percentage, the smaller the droplet size obtained.

Figure 2.

Pareto chart for the standardized effects on droplet size.

The main effects that the four factors have on the droplet size can be seen in Fig. 3: very large droplet size for US amplitude = −1 but a very small droplet size for US amplitude = 1. At 20% (−1 in Fig. 3) amplitude, droplets of 475 nm were formed, followed by droplets measuring 365 nm at 50% (0) and 278 nm at 80% (1). This trend demonstrates the negative effect that increasing ultrasonic amplitude has on droplet size.

Figure 3.

Main effects plot for droplet size with each factor.

The use of ultrasound enables the formation of smaller droplets and also gives a more uniform distribution, as demonstrated by the reduction in droplet size to 307 nm, from the initial 1323–1652 nm, on using dual‐frequency ultrasound. ¹⁸ The cavitation effect generated by ultrasound leads to intense mixing and shear forces, which enhance the emulsification process and reduce the need for higher concentrations of surfactants. ¹⁸ This effect is demonstrated in this experimental design as the levels of the carrier GSO and the emulsifiers Tween® 80 and pectin do not have significant effects on the droplet size.

3.2.2. Zeta‐potential ( Y ₂ )

The only factor that conditions the Y ₂ response, according to the second‐order polynomial equation model, was X ₂ (T80) in the linear terms (P = 0.0000), as shown in the corresponding Pareto chart (Fig. 4). This effect was negative (coefficient of −0.1207).

Figure 4.

Pareto chart for the standardized effects on zeta‐potential.

As the Tween® 80 concentration in the composition of the nanoemulsion increases, the zeta‐potential decreases. The zeta‐potential values are presented as positive because the RSM software only accepts positive input. This behavior can be seen from the main effect plot (Fig. 5). At a Tween® 80 concentration of 1% (−1) the zeta‐potential was −16.0 mV and this was followed by −6.5 mV at 5.5% (0) and − 4.6 mV at 10% (+1).

Figure 5.

Main effects plot for zeta‐potential with each factor. The values are inverted to avoid negative values.

The zeta‐potential of droplets in a nanoemulsion has a significant influence on the stability of the system. A higher zeta‐potential, either negative or positive, typically indicates greater stability owing to increased electrostatic repulsion between droplets, which prevents aggregation. ²⁸ The use of Tween® 80 as a surfactant results in a reduction in zeta‐potential, as demonstrated by testing different concentrations of the surfactant, although this is not immediately related to a lack of stability because Tween® 80 has the ability to manage the contact between the oil droplets and the aqueous phase, thus allowing the interactions to be stabilized. ²⁷

3.2.3. Viscosity ( Y ₃ )

The observed lack‐of‐fit (P = 0.019) suggests that the model is not adequate to describe the effects of the variables on the viscosity, as indicated previously by the large number of experimental versus predicted difference percentages that were greater than 20% (Table S1). As explained above, the results did not follow a normal distribution and therefore the response surface could not adequately describe the experimental variability across the entire range of formulations. This behavior can be mainly attributed to the nonlinear response of viscosity at higher values. As the viscosity increases it deviates from linear behavior due to increased droplet interactions and polymer entanglement, which lead to shear‐thinning effects. ²⁹ These structural changes reduce flow resistance under shear and produce behavior that cannot be accurately captured by polynomial models. This finding indicates that, beyond this viscosity threshold, the optimization of formulation conditions should focus on the viscosity range that ensures stable, homogeneous emulsions with consistent processing behavior. In an effort to improve the robustness and predictive reliability of the model, the analysis was restricted to formulations that have viscosities <0.1 Pa·sⁿ. Within this restricted range the model did not show a significant lack‐of‐fit (P > 0.05) and this confirms that the regression accurately captures the relationship between formulation variables and viscosity. Specifically, for the range between the minimum value up to 0.1 Pa.sⁿ, the results showed a distribution similar to a normal distribution (skewness = −0.4633 and kurtosis = −0.7206). As a consequence, a specific regression model was developed using experiments that covered the viscosity range below 0.1 Pa.sⁿ. The regression results gave an R ² value of 0.9351 and a regression mean square error (RMSE) of 0.0070. This model was used to compare experimental results for the optimization of both droplet size and zeta‐potential. Texture modifiers are frequently employed to improve the stability of formulations because increasing viscosity in nanoemulsions slows droplet movement. Polysaccharides, such as pectin, are among the most widely used compounds in this respect. ²⁷ Although the developed model does not accurately explain the results for viscosity, the significant increase in this parameter is compatible with the increased pectin concentration in the nanoemulsion, as reported previously by Artiga‐Artigas et al. ³⁰

3.3. Response optimization and validation

Three responses were evaluated in this study: droplet size (Y ₁), zeta‐potential (Y ₂) and viscosity (Y ₃). Given the considerable lack‐of‐fit in the viscosity model, optimization was not performed for this parameter, but the model was validated using the optimization experiments for both droplet size and zeta‐potential. Regarding the influence of the dependent variables on Y ₁ and Y ₂, it was found that for each response, a different dependent variable had negative/positive effects. As a result, the optimization of each variable was carried out individually by considering the resulting second‐order polynomial equation.

In the case of Y ₁ a study of the surface response results allowed optimization with the aim of minimizing the response in order to obtain smaller droplets. The best conditions obtained in coded form were: X ₁ = 1, X ₂ = −0.333, X ₃ = −1 and X ₄ = 1, which corresponds to a GSO concentration of 5% (w/w), a Tween® 80 concentration of 4%, a pectin concentration of 1% and an ultrasonic amplitude of 80%, to give a predicted droplet size of 222 nm. Experiments were conducted in triplicate for each response variable using the selected optimal conditions. These experiments generated an average droplet size of 244.5 ± 5.3 nm. In terms of the optimization for droplet size conditions, it was noted that the difference between the predicted (222 nm) and experimental values was statistically not significant, with a percentage divergence of 9%. This finding underlines the accuracy of the developed model and corroborates the optimal conditions identified.

For Y ₂ the optimization was performed considering a target value of −10 mV, a moderate value for formulation stability in terms of repulsive interactions between droplets. The conditions obtained in coded form were: X ₁ = −1, X ₂ = −0.386, X ₃ = −0.899 and X ₄ = −1, which corresponds to a GSO concentration of 1.2% (w/w), a Tween® 80 concentration of 3.8%, a pectin concentration of 1%, and an ultrasonic amplitude of 20%. The value predicted using the model is 10.010 mV with a confidence interval from −4.5019 to −38.6641 mV. The zeta‐potential measurement of −30.7 ± 0.6 mV is therefore within the predicted range and is in agreement with the predicted value. However, it should be noted that the confidence interval for the prediction is extremely large. In this case, an additional calculation was performed, namely a partial least square regression (PLS). The application of this regression tool gave a predicted zeta‐potential of −16.518, with a confidence range between −10.5493 and − 32.5464. This range is smaller than that obtained using the model but it is still too large.

Regarding the viscosity, the value for this experiment was 0.002 ± 0.001 Pa.sⁿ and the value obtained using the PLS model was 0.004 ± 0.001 Pa.sⁿ. Therefore, the use of the low viscosity values in the regression model will produce reliable predictive results for the viscosity only if the values are <0.1 Pa.sⁿ.

On the basis of the results outlined above, and given that droplet size is considered to be a critical parameter for the efficacy of nanoemulsion phytotoxic activity, the CcLERSLs nanoemulsion was formulated by employing the optimal conditions for droplet size (GSO 5% (w/w), Tween® 80 4% (w/w), pectin 1% (w/w) and ultrasonic amplitude of 80%), thus ensuring its suitability for subsequent studies.

3.4. CcLERSLs nanoemulsion morphological study

During the microscopy imaging, it was observed that the nanoemulsion containing CcLERSLs, despite its apparent initial density, became electron‐transparent under the initial preparation conditions. Because most of the water in the solution was removed during sample preparation–allowing the nanoemulsion to remain fully dispersed–a large portion of the matrix began to aggregate, which is what can be observed in Fig. 6. Nevertheless, most of the organic structures remained sufficiently intact to allow for a morphological and compositional nanoanalysis.

Figure 6.

SEM image of the pectin, GSO and Tween®80 matrix in which the bioactive CcLERSLs are emulsified.

By maintaining a 2.0‐nm spot size, the organic samples remained stable during imaging. In this case, the bright‐field photographs provided sufficient contrast to distinguish the C‐based samples from the lacey C of the copper grid.

The secondary electron detector clearly revealed lamellar structures distributed throughout the entire organic matrix. These structures display diverse sizes, mostly ranging from 150 to 250 nm, as observed in Figs 7 and S 2. This is consistent with the approximate measurements obtained by the dynamic light scattering analysis. Throughout all the grid sections examined, the aforementioned structures appeared exclusively within the organic matrix and were not observed independently. Figure S3 shows that these lamellar structures are predominantly spherical, possibly following the principle of minimum energy typical of nanoemulsions. ²⁶ Owing to the organic composition of all components in the CcLERSLs sample (C, O and H), it is not possible to determine the spatial arrangement of each component from the image. However, as mentioned in the previous paragraph, it is possible to confirm that all components of the emulsion are homogeneously associated.

Figure 7.

(Top left) SEM bright‐field images showing the lamellar structures observed in the CcLERSLs nanoemulsified sample. (Top right) Identification of spherical/pseudospherical structures embedded within the organic matrix formed by pectin, grape seed oil, and Tween80. (Bottom) Schematic representation of the nanoemulsion droplet structure analyzed by zeta potential and visualized by SEM.

In the case of the nanoemulsion samples without extract, Fig. 8 shows a different appearance. Initially, the sample also presented lamellar‐type structures within the organic matrix, although these structures are considerably smaller in diameter than those corresponding to CcLERSLs (Fig. 8, bottom left). This may be caused by the absence of the sesquiterpene lactone–enriched extract, because the volumes of self‐assembling polymers such as Tween® 80 tend to adapt easily to the type of content they encapsulate. ³¹ , ³² Additionally, Fig. 8 (top left and top right) shows a matrix that is much less electron‐transparent, which is associated with denser or more crystalline organic matrices.

Figure 8.

SEM bright‐field images of the nanoemulsion sample without extract, at different magnifications.

Overall, the lack of the CcLERSLs also appears to have affected the dispersion of the matrix. As observed in Fig. 8 (bottom right), despite identical preparation, the sample content is much less homogeneous.

3.5. Evaluation of phytotoxic activity

The phytotoxic activity of the CcLERSLs nanoemulsion was assessed against three common weed species: P. oleracea, Pl. lanceolata and Ph. arundinacea. In order to determine whether the nanoemulsion components themselves exhibited phytotoxic effects, a control nanoemulsion lacking the extract also was tested. Additionally, the nonformulated extract (dissolved in water) was included as a comparative treatment. All treatments were evaluated for their effects on seed germination as well as on root and shoot elongation. The results are presented in Figs 9, 10, 11.

Figure 9.

Effects of the nanoemulsion without extract (A), CcLERSLs nanoemulsion (B), CcLERSLs (C) and the herbicide Stomp® Aqua (HBC) on P. oleracea germination and root and shoot growth. Values are expressed as the percentage difference from the control, and Welch's test was used for statistical analysis. Letters a and b indicate significance for P < 0.01 and 0.01 < P < 0.05, respectively.

Figure 10.

Effects of the nanoemulsion without extract (A), CcLERSLs nanoemulsion (B), CcLERSLs (C) and the herbicide Stomp® Aqua (HBC) on Pl. lanceolata germination and root and shoot growth. Values are expressed as the percentage difference from the control, and Welch's test was used for statistical analysis. Letters a and b indicate significance for P < 0.01 and 0.01 < P < 0.05, respectively.

Figure 11.

Effects of the nanoemulsion without extract (A), CcLERSLs nanoemulsion (B), CcLERSLs (C) and the herbicide Stomp® Aqua (HBC) on Ph. arundinacea germination and root and shoot growth. Values are expressed as the percentage difference from the control, and Welch's test was used for statistical analysis. Letters a and b indicate significance for P < 0.01 and 0.01 < P < 0.05, respectively.

The CcLERSLs nanoemulsion exhibited potent phytotoxic activity against all weed species evaluated, with P. oleracea being the most sensitive (Fig. 9). Germination was inhibited by ≤40% at 800 ppm in contrast to the herbicide, which did not exhibit a significant inhibitory effect on germination. In terms of root and shoot growth were significantly inhibited, with reductions of 91% and 92% achieved, respectively and still displaying substantial inhibition at 400 ppm, with decreases of 86% in root length and 70% in shoot length. Notably, the nanoemulsion outperformed the commercial herbicide Stomp® Aqua: at concentrations of 400 to 800 ppm, it inhibited root growth, and at concentrations of 600 and 800 ppm, it inhibited shoot growth.

Germination of Pl. lanceolata (Fig. 10) was reduced by 57% at 800 ppm–a response not observed with the commercial herbicide–whereas root growth inhibition (83%) exceeded that of the commercial product (78%). In Ph. arundinacea (Fig. 11), the nanoemulsion achieved ≤80% root growth inhibition at 800 ppm, which is comparable to that of the herbicide, although germination remained unaffected. Controls with nanoemulsion without extract showed negligible phytotoxic activity, thus confirming that the observed effects are attributable to the CcLERSLs extract. Compared with the crude extract alone, the nanoemulsion showed enhanced phytotoxic activity by at least a factor of three. This finding highlights the role of nanoformulation in improving the bioavailability of SLs.

Given the high sensitivity of P. oleracea to the treatments, a morphological evaluation at 800 ppm was performed. The results are presented in Fig. 12.

Figure 12.

Morphological effects observed in P. oleracea. The upper image shows seedlings under control conditions and after treatment with the CcLERSLs nanoemulsion. The bottom image displays seedlings subjected to different treatments at 800 ppm: nanoemulsion without extract (A), CcLERSLs nanoemulsion (B), CcLERSLs crude extract (C) and the commercial herbicide Stomp® Aqua (HBC). The line in the bottom image indicates the boundary between the roots and shoots used for morphological assessment and measurements.

In the control group the roots exhibited a normal, elongated morphology without visible deformities, which reflects the typical growth patterns of untreated plants. Treatment with the nanoemulsion without extract (A) did not lead to any alterations and this indicates that the carrier itself did not exert significant phytotoxic effects. By contrast, plants exposed to the CcLERSLs nanoemulsion (B) showed severe reductions in root and shoot length, with roots appearing shorter and displaying mild axial deformations. These observations confirm the strong morphological impact of the incorporated extract. Roots treated with the crude CcLERSLs extract (C) were slightly shorter but retained an appearance relatively similar to the control, which demonstrates the less pronounced morphological effect of the crude material. Roots and shoots treated with the commercial herbicide Stomp® Aqua (HBC) were severely affected, with roots presenting extreme shortening, deformities and abnormal branching. These findings are consistent with the high phytotoxicity of the commercial herbicide. Compared with HBC, the extract‐based nanoemulsion produced roots and shoots that were noticeably shorter and thinner, while also markedly reducing seed germination–a response not observed with the commercial herbicide. These results indicate that the CcLERSLs nanoemulsion exhibited higher overall phytotoxicity than Stomp® Aqua.

The results outlined above demonstrate the high bioactive potential of the CcLERSLs nanoemulsion. This system showed at least three‐fold greater activity than the crude extract in inhibiting the root and shoot growth of P. oleracea, Pl. lanceolata and Ph. arundinacea, particularly at 800 ppm. In the context of botanical and nano‐based weed‐control agents, previous studies have shown that many plant‐extract‐ or essential‐oil‐derived bioherbicides require relatively high concentrations to produce significant phytotoxic effects. For example, a nanoemulsion of Cymbopogon nardus essential oil inhibited germination and seedling growth of Echinochloa crus‐galli in a dose‐dependent manner at concentrations between 100 and 800 ppm. ³³ Similar enhancements in biological activity also have been demonstrated for Parthenium hysterophorus L., whose nanoemulsion exhibited markedly improved inhibition of Diodia ocimifolia germination when compared with the crude extract. ³ In another study, a nanoemulsion formulated from Cananga odorata essential oil completely prevented seed germination of Amaranthus tricolor at 250 ppm, whereas a coarse emulsion required much higher concentrations for comparable effects. ³⁴ Furthermore, a bio‐nano‐herbicide based on Eucalyptus citriodora essential oil displayed enhanced pre‐ and postemergence activity against E. crus‐galli owing to superior droplet dispersion and bioavailability. ³⁵

The CcLERSLs nanoemulsion achieved >80% inhibition of root and shoot growth at 800 ppm across all three weed species, thus placing its efficacy among the highest levels reported for botanical nanoherbicides. This enhanced activity is likely to be associated with improved delivery and uptake of phytotoxic compounds. The results of our previous study, in which the putative herbicidal mechanisms were proposed for this extract, suggest that these compounds may disrupt photosynthetic efficiency, as evidenced by the pronounced decrease in total chlorophyll content, and induce oxidative stress that leads to lipid peroxidation and a decline in superoxide dismutase (SOD) antioxidant activity. These physiological and biochemical changes probably contribute to the strong inhibition of root and shoot growth observed in the present study. ⁹ Compared to conventional bioherbicide formulations and a widely used commercial herbicide (Stomp® Aqua), the CcLERSLs nanoemulsion exhibits both greater potency and lower effective concentration. These improvements are believed to be a result of its enhanced solubility, improved droplet distribution and increased cellular uptake of active constituents when compared to the crude material. These characteristics not only enhance phytotoxicity, but also indicate that the developed nanoemulsion is a competitive and sustainable alternative for weed management. Overall, the findings demonstrate that CcLERSLs nanoemulsions surpass the crude extract in efficacy while providing efficient delivery of bioactive compounds, thus highlighting their promise as an effective and environmentally friendly approach to controlling diverse weed species.

3.6. Nanoemulsion storage stability

The temperature‐response behavior of nanoemulsions in storage stability is critical for determining appropriate storage conditions, an aspect that has a direct impact on the product shelf life in real market conditions. The stability of the CcLERSLs nanoemulsion was assessed for 0, 30 and 90 days at three different temperatures (4 °C, 25 °C and 45 °C) by measuring droplet size, PDI and zeta‐potential. The three different parameters were measured immediately after production of the nanoemulsion. The results obtained for the initial time point were used as a reference for subsequent analyses after different periods of time and at different temperatures. The results are represented in Figs. 13, 14, 15.

Figure 13.

Droplet size observed in the nanoemulsion over a 90‐day incubation period at different temperatures (4 °C, 25 °C and 45 °C). Two‐way ANOVA was used for statistical analysis. Bonferroni's test was used for incubation period point differences versus initial time point at the same temperatures: *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 14.

Polydispersity index observed in the nanoemulsion over an incubation period of 90 days at different temperatures (4 °C, 25 °C and 45 °C). Two‐way ANOVA was used for statistical analysis. Bonferroni's test was used for incubation period point differences versus initial time point at the same temperatures: *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 15.

Zeta‐potential observed in the nanoemulsion over an incubation period of 90 days at different temperatures (4 °C, 25 °C and 45 °C). Two‐way ANOVA was used for statistical analysis. Bonferroni's test was used for incubation period point differences vs initial time point at the same temperatures *, P < 0.05; **, P < 0.01; ***, P < 0.001.

It can be seen from Fig. 13 that the droplet size in the nanoemulsion remained stable during 90 days in the storage stability study when kept at 4 °C and 25 °C (values between 230.0 and 252.0 nm at both temperatures). However, when the formulation was incubated at 45 °C a significant increase in droplet size was observed after 30 and 90 days of storage: the final droplet size was 1792.9 nm. The observed increase in droplet size could be attributed to sintering, the process by which two or more droplets merge together to form a single larger droplet in nanoemulsions at elevated temperatures. ²⁵

The nanoemulsion PDI values at different temperatures over the 90‐day storage period are shown in Fig. 14. It can be observed that at temperatures of 4 °C and 25 °C the PDI remained stable for the 90 days of storage (values between 0.27 and 0.30 for 4 °C and between 0.27 and 0.28 for 25 °C), whereas at 45 °C an increase in PDI was observed and this ultimately reached a PDI of 1. The increase in PDI found at 45 °C is most likely to be related to the enhancement in droplet size. Given that the PDI value is an index that reflects the size distribution of droplets, the occurrence of droplet coalescence results in a greater disparity in droplet sizes, which in turn enhances the PDI. ³⁶ The temperature‐response observed for droplet size and PDI are consistent with the results of previous studies reported in the literature. ²⁵ , ³³ , ³⁶ , ³⁷

The thermal effect on zeta‐potential after a 90‐day incubation period is shown in Fig. 15. After 30 days of incubation at 4 °C, the negative zeta‐potential increased significantly from −8.1 to −9.5 mV. However, after storage for 90 days the zeta‐potential value remained similar (−8.8 mV). After 30 days of incubation at 25 °C, the negative zeta‐potential value increased from −8.1 to −9.3 mV. However, after 90 days the zeta‐potential was different to that measured at time zero, but the value was lower than that determined after 30 days of storage (−9.2 mV). At 45 °C, zeta‐potential negative values increased significantly compared to the initial time point. After 30 days of incubation, the zeta‐potential was −16 mV and this increased slightly to −18.2 mV after 90 days.

Although the zeta‐potentials at 4 °C and 25 °C exhibited significant variation from the initial time points, the percentage of difference between values was <15%, thus suggesting that the cause was not related to the influence of temperature on the nanoemulsion. However, the observed effect at 45 °C after incubation for 90 days may be caused by an increased affinity of Tween® 80 for the oil phase as the molecules become more soluble. ³⁸ The decreased stabilizing effect of Tween® 80 leads to greater negative zeta‐potential values according to the results of the current study.

The stability of the CcLERSLs nanoemulsion is a critical factor for its practical application. The results obtained in this study demonstrate that the formulation is stable at 4 °C and 25 °C, with its physicochemical properties remaining unchanged over time. However, exposure to a temperature of 45 °C led to significant degradation, probably as a result of increased molecular motion and potential destabilization of the oil–water interface. These findings imply that, although the nanoemulsion can be stored and transported under standard laboratory and temperate field conditions, extra care is required in regions with hot climates, where temperatures may approach or exceed 40 °C. Strategies such as storage in shaded or temperature‐controlled environments, as well as rapid use following preparation, may be necessary to ensure efficacy in these settings. This temperature‐dependent stability is consistent with previous reports on plant extract nanoemulsions, which likewise show decreased shelf‐life at elevated temperatures. ³

4. CONCLUSIONS

Three predictive models were successfully developed for estimating droplet size, zeta‐potential and viscosity (in the latter case, a reduced model) for the nanoemulsion developed in this work. The models proved to have high reliability in predicting these parameters. The models were also effective for optimizing droplet size and zeta‐potential to specific application requirements and they offer a reliable means of estimating final values for all three parameters based on formulation variables before nanoemulsion formation. According to the results obtained from the model validation for droplet size, the formulation conditions used for the subsequent assays were a GSO concentration of 5% (w/w), Tween® 80 at 4% (w/w), pectin at 1% (w/w) and an ultrasonic amplitude of 80%. SEM images confirmed the formation of the nanoemulsion, showing a lamellar‐type structure consistent with the obtained zeta‐potential results. Evaluation of the phytotoxic activity of the CcLERSLs nanoemulsion showed an enhancement of growth inhibition on the tested weed species P. oleracea, Pl. lanceolata and Ph. arundinacea when compared to the crude extract. This increased activity was at least three‐fold greater than that of the crude extract and, in some cases, the nanoemulsion also outperformed the commercial herbicide Stomp® Aqua. Storage stability studies revealed that the CcLERSLs nanoemulsion remained stable at 4 °C and 25 °C over a 90‐day period, with consistent droplet sizes and PDI values, and a zeta‐potential variation of <15%. These findings show that the nanoemulsion formulation improves both the bioavailability and stability of the CcLERSLs extract, particularly enhancing its phytotoxic activity and showing it to be a promising natural herbicide candidate. From a regulatory standpoint, the nanoemulsion developed in this study could preliminarily be considered within the scope of bioherbicides under the European Union Plant Protection Products (PPP) framework.

CONFLICT OF INTEREST

The authors declare no competing of interest.

Supporting information

Table S1. Results from the experimental design regarding droplet size (Y ₁), zeta‐potential (Y ₂) and viscosity (Y ₃) for the evaluation of the variables in the experimental design.

Fig. S1. Distribution histogram of the frequency of viscosity values in the experimental design.

Fig. S2. Size distribution of the nanostructures observed in the nanoemulsions regarding their composition.

Fig. S3. Bright field images of the lamellar structures observed in a different section of the copper grid for the CcLERSLs nanoemulsified sample.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available owing to privacy or ethical restrictions.