Hydrogen-bond-driven stabilization and antioxidant synergy in Pickering emulsions stabilized by natural polyphenol-cellulose nanocrystal hybrids

Abstract

To meet the increasing demand for green and sustainable delivery systems in food, cosmetics, and pharmaceuticals and address the research gap in natural polyphenol-cellulose nanocrystal (CNC) composite stabilizers, this study developed a bio-based Pickering emulsion stabilizer via hydrogen bonding between 6’-O-caffeoylarbutin (CA) and CNCs. Results demonstrated CA/CNCs self-assembled into a dense interfacial layer, endowing the emulsion with exceptional stability that was reflected in over 85% curcumin enscapsulation efficiency after 30 days of dark storage and structural integrity at up to 80 °C, while CA-curcumin synergism enhanced antioxidant activity with 76.9% DPPH and 81.2% ABTS⁺ free radical scavenging rates. FT-IR, ¹H NMR, and computational simulations confirmed the hydrogen-bond network between CA and cellobiose as the core stabilization mechanism. This study provides a novel, eco-friendly strategy for high-performance natural stabilizers, offering the advancement of green, functional delivery systems across relevant industries.

Graphical abstract

1. Introduction

With the increasing demand for sustainability and safety in the food, cosmetics, and pharmaceuticals industries, the development of green and biocompatible delivery systems has emerged as a pivotal research focus (Wang et al., 2025). Among these emerging platforms, Pickering emulsions rely on solid particles for stabilization instead of soluble surfactants. They have attracted considerable research interest owing to excellent thermodynamic stability, low toxicity, and enhanced capacity to encapsulate bioactive compounds (Ji & Wang, 2023). Compared with soluble surfactant-stabilized emulsions, Pickering systems exhibit enhanced resilience to thermal and pH variations (Xiao, Li, & Huang, 2016; Zhang et al., 2021). They are well suited for use in food, cosmetics and pharmaceuticals fields and particularly as carriers for bioactive compounds to improve bioavailability and shelf life (Cheng et al., 2024; Jiang et al., 2023; Zhang, Fan, Liu, & Li, 2023).

Natural renewable materials, particularly cellulose nanocrystals (CNCs), have emerged as promising stabilizers for Pickering emulsions (Du Le, Loveday, Singh, & Sarkar, 2020; Wang, Wang, Zhao, Liu, & Hu, 2023). Derived from the most abundant biopolymer on Earth, CNCs possess favorable aspect ratios, excellent mechanical strength, and inherent biocompatibility, facilitating the formation of robust interfacial layers to inhibit droplet coalescence (Ma et al., 2023; Ma et al., 2023; Nie et al., 2022; Zhang et al., 2024). Recent studies have optimized CNCs as core stabilizers, with their performance significantly enhanced through modification with natural or biocompatible components (Ma, Bu, et al., 2023; Zhang et al., 2024). Bai et al. (2019) reported high-pressure microfluidization as an effective fabrication method for CNC-stabilized oil-in-water emulsions, with droplet size and creaming stability regulated by microfluidization pressure and CNC concentration (Bai et al., 2019). Chen et al. (2024) modified CNCs with cationic ε-polylysine, reducing inter-particle repulsion via electrostatic interactions and promoting denser adsorption layers at oil-water interfaces (Chen et al., 2024). Despite these advances, pure CNCs lack additional functional properties such as antioxidant activity, necessitating further modification to meet diverse application demands (Aw et al., 2022; Xie et al., 2024).

Natural polyphenols are widely used to protect active ingredients in food and pharmaceutical formulations due to their bioactivity (Shahidi & Ambigaipalan, 2015). Their incorporation into emulsions boosts antioxidant capacity, which extends food shelf life and preserves bioactive stability (Yan et al., 2022). However, most polyphenols suffer from poor aqueous solubility and insufficient stability, limiting their bioavailability and practical applications (Farooq, Abdullah, & Weiss, 2021). CNCs serve as ideal carriers for water-soluble polyphenols such as tannic acid and gallic acid via hydrogen bonding, improving emulsification performance and hydrophobicity for stable anchoring at the oil-water interfaces (Dai et al., 2021; Javed et al., 2024). However, research on interactions between water-insoluble polyphenols in solid particle form and CNC suspensions remains scarce. Notably, polyphenol crystals have been proven to adsorb at liquid-liquid interfaces and stabilize oil-in-water emulsions, supporting the feasibility of CNC-water-insoluble polyphenol nanoparticles for Pickering emulsion stabilization (Luo et al., 2011; Zembyla, Murray, & Sarkar, 2018).

6’-O-caffeoyl arbutin (CA) is a major polyphenolic component isolated from dried Vaccinium dunalianum Wight flower buds, with a dry weight content of 22% (Zhao et al., 2008). It exhibits potent antioxidant activity and melanin-inhibiting effects (Xu et al., 2014). However, CA suffers from poor aqueous solubility and insufficient stability, which requires efficient carrier technologies to address these challenges. Research gaps remain regarding CA carriers. First, the specific hydrogen bonding mechanism between CA and CNCs has not been systematically investigated, impeding precise construction and performance optimization of CA-CNC composites. Second, the synergistic enhancement of antioxidant efficacy between polyphenols and encapsulated lipophilic bioactives has been observed in curcumin-polyphenol mixtures, but has not yet been verified in CNC-CA systems, and its regulation still remain largely remains unexplored (Paciello et al., 2020). Thus, this study aims to to investigate the influence of hydrogen-bond-driven assembly of natural polyphenol CA and CNCs on the physicochemical and functional properties of curcumin-loaded Pickering emulsions, with particular focus on their stability (under dark/light, temperature, and pH conditions), microstructure, and antioxidant synergy between CA and curcumin. This work provides theoretical and practical support for green polyphenol-nanocellulose emulsion systems, advancing green delivery systems in food, comestics and pharmaceuticals fields.

2. Materials and methods

2.1. Reagents

6’-O-caffeoylarbutin (CA) (HPLC purity >90%) was isolated and purified from Quezui tea in our laboratory. Cellulose nanocrystals (CNCs) suspensions at different concentrations (0.05, 0.1, 0.2, 0.3, 0.4, 0.5 wt%) were prepared according to previously methods (Yang et al., 2024). Curcumin was obtained from Adamas-Beta Ltd. (Shanghai, China). Ultrapure water was employed for all experiments procedures. Unless otherwise specified, all other chemicals and solvents were supplied by Shanghai Titan Scientific Co., Ltd. (Shanghai, China). All reagents used were of analytical grade or higher and were applied directly in experiments without additional purification.

2.2. Preparation of curcumin-loaded CA/CNCs Pickering emulsions

The CA/CNCs mixed aqueous phase was prepared by adding specified amounts of CA to CNCs aqueous suspensions, followed by stirring at 8000 rpm for 1 min and ultrasonication (ultrasonic cell disruptor, JY92-IIDN, Ningbo Scientz Biotechnology Co., Ltd.) at 200 W for 3 min. CA/CNCs suspensions with varying CA concentrations (0, 0.025, 0.05, 0.1, 0.2 wt%) were prepared at a fixed CNCs concentration of 0.2 wt%, while suspensions with different CNCs concentrations (0, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5 wt%) were developed at a fixed CA concentration of 0.1 wt%. For the oil phase, 0.025 g curcumin was mixed with 50 g olive oil and magnetically stirred at 60 °C for 3 h until complete dissolution (final curcumin concentration is 0.05 wt%). The Pickering emulsions were formulated with an oil-to-water volume ratio of 1:4 (4 mL olive oil phase and 16 mL CA/CNCs aqueous phase) via two sequential steps. First, pre-emulsification was conducted by stirring at 12000 rpm for 1 min. This was followed by ultrasonication in an ice-water bath at 250 W for 5 min at room temperature. All samples were prepared in triplicate for subsequent measurements.

2.3. Characterization of curcumin-loaded CA/CNCs Pickering emulsions

2.3.1. Emulsions stability

Emulsion stability is closely related to ionic concentration, and zeta potential is a key indicator for evaluating emulsion stability. Zeta potential was measured using a nanoparticle size and zeta potential analyzer (Zetasizer Nano ZS90 Size, Brookhaven Instruments Corporation, Holtsville, NY, USA) under 30-cycle measuring conditions. Deionized water was used as the blank control. Zeta potentials of CA, CNCs, and CA/CNCs suspensions at different concentrations were measured in triplicate to determine the optimal CA/CNCs ratio.

2.3.2. Emulsions droplet diameter and emulsification ratio

Pickering emulsions at varying concentrations were prepared using the optimal CA/CNCs ratio. Storage stability was evaluated by photographing stationary emulsions after 7 days using a Nikon D5300 camera (Nikon Corporation, Tokyo, Japan). Droplet diameter was measured via dynamic light scattering (DLS) with a laser particle size and zeta potential analyzer (measurement range: 0.6 nm ∼ 6 μm; configured for particles >500 nm) in triplicate. The emulsification ratio (emulsion volume/total system volume) was calculated by measuring pixel distances in the images, and the emulsion volume was derived using parameters including the measured the thickness of the top emulsion phase in tube midsection images (pixel distance analysis), with deionized water as the blank control (triplicate measurements).

2.3.3. FT-IR and ¹H NMR analyses

To elucidate the hydrogen bonding interactions between CNCs and CA, Fourier-transform infrared (FT-IR) spectroscopy and proton nuclear magnetic resonance (¹H NMR) spectroscopy were employed. CA/CNCs aqueous suspensions with mass ratios of 1:3 and 3:1 (CA:CNCs) were prepared using 0.3 wt% CNCs as the base concentration, stirred at 12,000 rpm for 1 min, ultrasonication at 250 W for 3 min in an ice-water bath to prevent thermal degradation, and freeze-dried to obtained solid samples. FT-IR measurements were conducted using a IRXross spectrophotometer (Shimadzu, Japan) on pure CNCs, pure CA, and CA/CNCs composite samples over 4000–400 cm⁻¹ with 4 cm⁻¹ resolution and 32 cumulative scans. ¹H NMR spectra were acquired on a Avance II 500 MHz NMR spectrometer (Bruker, Germany) using deuterated water (D₂O) as the solvent, with 16 cumulative scans and chemical shifts referenced to the residual solvent peak.

2.3.4. Effect of temperature and pH on stability

To evaluate the impact of temperature, Pickering emulsions prepared under optimal conditions were heated in a water bath at 30, 40, 50, 60, 70, 80, and 90 °C for 1 h. Average particle size, zeta potential, and polydispersity index (PDI) were measured before and after heating. For pH stability assessment, emulsions were adjusted to pH 3.0, 5.0, 7.0, 9.0, and 11.0 using 0.1 mol/L HCl and 0.1 mol/L NaOH, homogenized at room temperature for 10 min, and evaluated for visual appearance, average particle size, zeta potential, and PDI. Deionized water was used as the blank control, and each sample was measured in triplicate.

2.3.5. Encapsulation efficiency

A curcumin standard curve was established following Marefati et al. (Marefati, Bertrand, Sjöö, Dejmek, & Rayner, 2017), (linear regression equation: y = 157.67×-0.0018,R² = 0.9990). Curcumin loading was quantified using a phase separation method. A 1.0 mL emulsion aliquot was serially diluted with 4 mL anhydrous ethanol (1:5, V/V). It was vortexed at 2000 rpm for 2 min, and subsequently centrifuged at 4000 rpm for 10 min. The supernatant was further diluted with anhydrous ethanol (1:10, V/V). Its absorbance was measured at 425 nm using UV–Vis spectrophotometer (UV-2700i, Shimadzu, Japan), with anhydrous ethanol used as blank control. All measurements were performed in triplicate. The encapsulation efficiency (EE) was calculated as:

$$\mathit{EE}\left(\%\right)=\left(C_{\mathrm{t}}-{\mathrm{C}}_{\mathrm{f}}\right)/C_{\mathrm{t}}\times 100\%.$$ (1)

Where C_t (mg/mL) denotes the initial curcumin concentration, and C_f (mg/mL) represents the free curcumin concentration in freshly prepared Pickering emulsions.

2.3.6. Microscopic observation

Optical microscopy (SK2009, Shenzhen Sikes Electric Co., Ltd., China) was used to evaluate droplet size and distribution of emulsions prepared with deionized water (stabilizer-free, blank control), 0.3 wt% CNCs, 0.1 wt% CA, and CA/CNCs composites. Cryo-scanning electron microscopy (Cryo-SEM, Regulus 8100, Hitachi, Japhan) was employed to observe interfacial morphology and spatial arrangement (cryo-fixation technique). For fluorescence imaging, Nile Blue-stained aqueous phases (excitation/emission: 365/435 nm) and Nile Red-stained oil phases (excitation/emission: 488/539 nm) were equilibrated for 12 h. Emulsions were prepared via pre-emulsification and ultrasonic homogenization, and fluorescence images were acquired at 630× magnification using a laser scanning confocal microscope (LSCMS, TELLARIS 5, Leica, Germany) (Xue, Li, & Dong, 2020).

2.3.7. Interfacial tension and rheological testing

Interfacial tension was measured using an SZ-CAMC32 universal surface tensiometer (Shanghai Sunzern Instrument Co., Ltd. China). Three aqueous phases were prepared using deionized water as the solvent. The first contained 0.1 wt% CA, the second was 0.3 wt% CNCs, and the third was a mixture of 0.1 wt% CA and 0.3 wt% CNCs. A needle containing the aqueous phase was immersed into an olive oil, and interfacial tension was calculated based on the critical droplet volume at the transition between adhesion and detachment (Meirelles, Costa, & Cunha, 2020). Rheological characterization was performed using an Anton Paar MCR 302 rotational rheometer (Anton Paar, Austria) with a 25 mm diameter stainless steel parallel plate (1 mm gap). Steady-state shear tests were conducted over a shear rate range of 0.01–100 s⁻¹ to determine apparent viscosity, and dynamic frequency sweeps within the same range were applied for linear viscoelastic analysis at a constant strain of 0.5% (Wu et al., 2021). All measurements were performed in triplicate.

2.3.8. Antioxidant activity

DPPH and ABTS⁺ free radical scavenging capacities were measured following a modified protocol (Li et al., 2021; Naksuriya & Okonogi, 2015). For the DPPH free radical scavenging assay, a 0.1 mg/mL DPPH working solution was prepared by accurately weighing 10 mg of solid DPPH, dissolving in anhydrous ethanol, and diluting the mixture to 100 mL. Emulsions were diluted 50-fold, after which 100 μL of diluted emulsion was mixed with 100 μL of DPPH working solution, then vortexed at 10,000 rpm for 30 min at room temperature. Supernatant was collected, and its absorbance was measured at 517 nm using a microplate reader (Spectra Max 190, Molecular Devices, USA) with triplicate measurements. For the ABTS⁺ free radical scavenging assay, an ABTS⁺· working solution was prepared by dissolving 0.0384 g ABTS and 0.0134 g potassium persulfate separately in water, diluting to 10 mL in volumetric flasks, mixing 10 mL of each solution, and incubating at room temperature for 12–16 h. The absorbance was adjusted to 0.70 ± 0.02 at 734 nm with anhydrous ethanol. For analysis, 40 μL of diluted emulsion was mixed with 160 μL ABTS⁺ free radical solution and incubated at 25 °C for 15 min. Supernatant was collected, and its absorbance was measured at 734 nm using a microplate reader with triplicate measurements. The scavenging rate was calculated as:

$$\text{Scavenging rate}\left(\%\right)=\left[1–\left(A_{\mathrm{T}}-A_{\mathrm{C}}\right)/A_{\mathrm{B}}\right]\times 100\%.$$ (2)

Where A_T is the absorbance of the emulsion mixed with working solution, A_C is the absorbance of the emulsion with anhydrous ethanol, and A_B is the absorbance of the working solution with anhydrous ethanol (blank control).

2.3.9. Computational simulations

Interaction Region indicator (IRI) analysis and Independent Gradient Model based on Hirshfeld partition were used to investigate intermolecular forces (Johnson et al., 2010; Lu & Chen, 2021). Correspondingly, the Independent Gradient Model based on Hirshfeld partition was employed to generate δg vs. sign(λ₂)ρ scatter plots for visualizing interaction types (Lu & Chen, 2012). Structural optimization and frequency calculations were performed using Gaussian16 at the B3LYP-D3(BJ)/def2-SVP level, with Single-point energy calculations at the B3LYP-D3(BJ)/def2-TZVP level. Wavefunction analysis and weak interaction analysis were executed with Multiwfn to explore interactions between CA and CNCs (Lu & Chen, 2022a; Lu & Chen, 2022b; Lu & Chen, 2024).

3. Results and discussion

3.1. Morphology and stability of CA/CNCs suspensions

The formation of stable suspension systems was crucial for Pickering emulsion preparation. Therefore, this study prepared a series of CA suspensions, CNCs suspensions, and CA/CNCs composite suspensions by adjusting solute concentrations. Combined with their microscopic dispersion behaviors, macroscopic characteristics, and zeta potential test results (Fig. 1), the stabilization mechanism of CA/CNCs composite suspensions was systematically investigated. From the perspective of single systems, CA suspensions (Fig. 1-A), due to their low water solubility, appeared as clear solutions at low concentrations. When the concentration increased to 0.1 wt%, turbid suspensions without obvious flocculation were formed. Their zeta potential increased with the rise in concentration to −30 mV and remained stable (Fig. 1-A1), indicating that the electrostatic repulsion was sufficient to prevent particle aggregation. CNCs suspensions (Fig. 1-C), relying on their insolubility and surface charge, formed stable colloidal systems at low concentrations through electrical double layer repulsion. As the concentration increased, the turbidity rose linearly due to enhanced light scattering by particles, with no visible flocculation. The zeta potential stabilized at −34 mV when the concentration reached 0.3 wt% (Fig. 1-C1), further confirming excellent colloidal stability.

Fig. 1.

Microscopic dispersion behaviors, macroscopic characteristics, and zeta potential test results of CA suspensions, CNCs suspensions, and CA/CNCs composite suspensions.

CA suspensions with varying concentrations (A), CA/CNCs suspensions with 0.2 wt% CNCs (constant) and varying CA concentrations (B), CNCs suspensions with varying concentrations (C), and CA/CNCs suspensions with 0.1 wt% CA (constant) and varying CNCs concentrations (D). Zeta potential of CA suspensions (A1), CNCs suspensions (B1), CA/CNCs suspensions with 0.2 wt% CNCs (constant) and varying CA concentrations (C1), and CA/CNCs suspensions with 0.1 wt% CA (constant) and varying CNCs concentrations (D1).

In composite systems, when the CNCs concentration was constant at 0.2 wt% (Fig. 1-B), pure CNCs suspensions showed significant turbidity due to Mie scattering by nanoparticles. After adding CA, turbidity decreased significantly, with the best clarity achieved at 0.05 wt%. When the concentration continued to increase to 0.1 wt%, turbidity rebounded. Correspondingly, the zeta potential remained stable at −34 mV within the range of 0–0.1 wt% CA, and dropped to −32 mV when in excess (Fig. 1-B1). When the CA concentration was constant at 0.1 wt% (Fig. 1-D), CA suspensions without CNCs showed slight turbidity. When the CNCs concentration increased from 0 to 0.5 wt% (especially 0.4–0.5 wt%), turbidity first rose and then fell. The zeta potential gradually increased from −21 mV to −41 mV with the addition of CNCs and remained stable. This phenomenon was attributed to the formation of hydrogen bonds between the water-soluble hydroxyl groups of CA and the surface hydroxyl groups of CNCs. Through steric hindrance, large aggregates were promoted to depolymerize into nanoscale aggregates, reducing light scattering to improve stability. However, excessive CA could cause excessive intermolecular association, leading to secondary agglomeration of CNCs and thus reducing stability.

As a core electrochemical parameter characterizing colloidal stability, the higher the absolute value of zeta potential, the stronger the inhibitory effect of the system on Ostwald ripening and heterocoagulation. When the absolute value exceeded 30 mV, the emulsion showed high stability (Chu, Zhang, Mei, Wang, & Xie, 2024). Based on the comprehensive results of appearance and potential tests, the composite suspension composed of 0.1 wt% CA and 0.3 wt% CNCs had the best stability.

3.2. Impact of CA/CNCs concentration on curcumin-loaded Pickering emulsions

Curcumin, a typical liposoluble natural active compound, has limited applications due to poor water dispersibility and weak chemical stability, as it is susceptible to degradation under light, heat, and oxidative stress. This study constructed a synergistic stabilization system via hydrogen bonding between CA, which possesses both biological activity and interfacial regulation functions, and CNCs. These two components co-assembled to form a dense oil-water interface shell, providing dual protection for curcumin by shielding against external degradation, reducing diffusion loss, and enabling potential synergistic effects through co-loading of CA. Pickering emulsions loaded with curcumin were prepared using a 0.1 wt% CA/0.3 wt% CNCs composite aqueous suspension as the stabilizer and olive oil containing 0.05 wt% curcumin as the oil phase, with results shown in Fig. 2. As shown in Fig. 2-A, after 15 days of storage, both the stabilizer-free group and the CA-only group exhibited significant oil-phase separation with a distinct upper oil layer. This confirmed that CA alone could not form a stable oil-water interfacial film, as it failed to inhibit oil aggregation and floating or maintain emulsion uniformity. In contrast, emulsions stabilized by CNCs or the CA/CNCs composite appeared uniformly yellow with only minor aqueous precipitate at the bottom, indicating both formed physical barriers via interfacial adsorption to effectively anchor the oil phase.

Fig. 2.

Characteristics of Pickering emulsions prepared with different stabilizers (A-C) and varying CA/CNCs concentrations (D-E) after storage.

Digital photographs (A), droplet diameters (B), and optical microscopy images (C) of the negative control (stabilizer-free) and Pickering emulsions stabilized by CA, CNCs, and CA/CNCs after 15 days of storage; digital photographs (D) and emulsification rate, zeta potential, and PDI (E) of Pickering emulsions with varying CA/CNCs stabilizer concentrations (0, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5 wt%) after 7 days of storage.

Further analysis of the particle size distribution (Fig. 2-B) and microscopic morphology (Fig. 2-C) revealed that the stabilizer-free emulsion exhibited a broad droplet size distribution ranging from 0.5 to 10 μm, with a mean diameter of 2.2 μm. Its droplets were randomly scattered with significant size differences. This phenomenon arose because the absence of a physical barrier at the oil-water interface led to uncontrolled droplet coalescence driven by Brownian motion, which was consistent with the macroscopic oil-phase separation. The CA-only stabilized emulsion exhibited a narrowed distribution ranging from 0.7 to 4 μm, with a mean diameter of 1.8 μm, and slightly improved local uniformity. This was because CA spread at the interface through hydrophilic-hydrophobic groups, reducing van der Waals forces between droplets and partially inhibiting coalescence. However, the adsorption of polyphenol particles alone at the oil-water interface was incomplete, leading to defective interfacial coverage and thus an inability to effectively resist long-term coalescence (Zembyla et al., 2018). The CNCs-only stabilized emulsion exhibited an expanded size distribution ranging from 1.2 to 11 μm, with a mean diameter of 3.8 μm. Its droplets formed ordered arrays of dense packing and showed improved monodispersity. This was attributed to the high aspect ratio of CNCs, which helps form a dense network at the oil/water interface, making them more effective than CA in stabilizing Pickering emulsions (Ji & Wang, 2025). The CA/CNCs composite stabilized emulsion exhibited a size distribution ranging from 0.85 to 13 μm, with a mean diameter of 3.4 μm, which was slightly smaller than that of the CNCs-only group. The mean diameter of the droplets decreased and they exhibited an ordered packing structure, as CA reduced the hydrophilicity of CNCs through synergistic hydrogen bonding and filled the gaps between CNCs, thereby enhancing the compactness of the interfacial film (Guo et al., 2021). Overall, both CNCs and the CA/CNCs composite stabilized Pickering emulsions. The CA/CNCs composite, with less aqueous precipitation (Fig. 2-A) and additional functional potential from the inherent activity of CA, was selected for subsequent studies, providing a basis for efficient curcumin loading and functional delivery.

To identify the optimal dosage of the CA/CNCs composite system for regulating aqueous phase retention, this study assessed emulsion stability across a 0–0.5 wt% concentration gradient. Fig. 2-D showed the stored state of corresponding Pickering emulsions, with results revealing gradual improvement in emulsion stratification as concentration increased. All low-concentration groups (≤0.2 wt%) exhibited significant oil-phase separation, attributed to insufficient stabilizer leading to incomplete interfacial films that caused droplets to coalesce easily and release the oil phase. In contrast, high-concentration groups (≥0.3 wt%) formed stable yellow homogeneous emulsions, as sufficient CA/CNCs adsorbed at the oil-water interface, effectively hindering droplet coalescence and migration to inhibit phase separation. Fig. 2-E presented three key indices, namely emulsion ratio, polydispersity index (PDI) and zeta potential of CA/CNCs-stabilized Pickering emulsions, quantifying the regulatory effect of concentration. In the low-concentration range (0–0.2 wt%), the emulsion ratio rose slowly from 50% to 70% but stayed below 80% due to insufficient stabilizers limiting oil-phase dispersion efficiency. In the high-concentration range (0.3–0.5 wt%), it increased rapidly to 92% and stabilized, indicating sufficient CA/CNCs fully covered the oil-water interface, achieving efficient oil dispersion and maximum emulsification efficiency after saturation. The PDI remained at 0.7 in the low-concentration range, reflecting non-uniform droplet sizes and severe coalescence, but dropped sharply to 0.2 in the high-concentration range with significantly improved uniformity. This occurred because high-concentration CA/CNCs inhibited coalescence through steric hindrance and electrostatic repulsion, narrowing the size distribution. Zeta potential measured −20 mV in the low range, with weak electrostatic repulsion between droplets causing high coalescence tendency, while in the high range it stabilized at −35 mV as CA/CNCs adsorption increased interfacial charge density, enhancing repulsion and inhibiting collision-induced coalescence. Therefore, optimal emulsion performance required a CA/CNCs concentration of ≥0.3 wt%, at which the emulsion was macroscopically uniform and stable without oil-phase separation. Quantitatively, the emulsion ratio reached 95% saturation, PDI hit a minimum of 0.2, and zeta potential stabilized at −35 mV. This interfacial saturation threshold enabled sufficient CA/CNCs to form a dense, uniform oil-water interface through synergistic effects. CA modified the CNCs surface to enhance interfacial film charge and steric hindrance, achieving optimal stability and uniformity. These findings provide critical parameters for industrial production of curcumin-loaded Pickering emulsions.

3.3. Hydrogen bonding interactions in CA/CNCs composites

To investigate the interfacial interactions between CNCs and CA, FT-IR and ¹H NMR analyses were conducted, with the results presented in Figs. S1–S4. Fig. S1 shows the FT-IR spectra of pure CA, pure CNCs, and CA/CNCs composites with mass ratios of 1:3 and 3:1 (CA:CNCs). All samples exhibited broad absorption bands in the 3600–3100 cm⁻¹ range, corresponding to intramolecular -OH stretching vibrations (Yang et al., 2024). The spectra of CA/CNCs composites simultaneously displayed the characteristic peaks of both CNCs and CA, with no new peaks detected compared to the individual spectra. This confirms the absence of significant chemical reactions that would form new covalent bonds between CNCs and CA. Notably, regardless of the composite ratio, whether adding a fixed amount of CA to CNCs (CA/CNCs = 1:3) or a fixed amount of CNCs to CA (CA/CNCs = 3:1), the -OH stretching absorption peak of the composites showed a slight shift and broadening relative to those of pure CNCs and CA. Such changes are typical indicators of hydrogen bonding interactions, as hydrogen bond formation modulates the vibrational frequency and intensity of -OH groups (Wang, Yang, Yang, Bai, & Li, 2021). These results suggest that CA adsorbs onto the CNCs surface primarily through hydrogen bonds formed between the abundant hydroxyl groups of both components, facilitating the construction of a polyphenol-cellulose composite system (Dai et al., 2021).

To further validate the hydrogen bonding interactions identified by FT-IR, ¹H NMR analysis was performed. Hydrogen bond formation weakens and elongates the original -OH bonds, altering the local chemical and electronic environments of the involved protons and atomic nuclei. These changes can be quantified by observing shifts in the chemical shifts of hydrogen-bonded nuclei and their adjacent protons (Zhao et al., 2020). As shown in Fig. S2, the hydroxyl hydrogens on the phenyl ring of CA are labile and typically produce indistinct signals in ¹H NMR spectra. However, when these hydrogens form hydrogen bonds with CNCs (acting as hydrogen bond acceptors via hydroxyl groups of CNCs), the electron cloud density of the CA structure is modified, leading to upfield shifts of adjacent hydrogen atoms. In the CA/CNCs composite systems, the hydroxyl hydrogens on the caffeoyl phenyl ring of CA act as hydrogen bond donors, forming interactions with hydroxyl groups of CNCs. This interaction enhances the electron cloud density of the conjugated system in the caffeoyl moiety (encompassing protons at positions 20, 21, 24, 27, and 28), strengthening the electron shielding effect on these protons. Consequently, a higher magnetic field strength is required to detect their resonance signals, manifesting as upfield shifts in the 6.25–7.55 ppm range (Fig. S3). Notably, the CA/CNCs = 1:3 system exhibited a more pronounced upfield shift compared to the CA/CNCs = 3:1 system, indicating stronger hydrogen bonding interactions and supporting its designation as the optimal formulation. For the phenol structural moiety of CA, the hydroxyl hydrogens on the phenyl ring also act as hydrogen bond donors, forming interactions with CNCs that increase the electron cloud density of the phenyl ring. This results in a modest upfield shift of the phenyl ring protons, though the magnitude is less significant than that of the caffeoyl moiety due to the relatively weak polarity of the phenol structure (compared to the highly polar conjugated system of the caffeoyl group). Furthermore, Fig. S4 shows that the hydroxyl hydrogen at position 10 of the glycoside structure of CA forms weak hydrogen bonds with CNCs, as evidenced by only slight changes in its chemical shift. This is attributed to the small polarity difference between this proton and the hydroxyl groups on the sugar rings of CNCs, which limits modifications to the electron cloud density and results in less stable hydrogen bonds.

3.4. Stability of curcumin-loaded CA/CNCs Pickering emulsions

To investigate the stability of curcumin-loaded CA/CNCs Pickering emulsions, this study systematically examined the emulsification behavior under distinct light conditions (dark vs. natural light). It also analyzed changes in particle size, PDI, and zeta potential across varying temperatures (30, 40, 50, 60, 70, 80, and 90 °C) and pH values (3.0, 5.0, 7.0, 9.0, and 11.0). Results are shown in Fig. 3.

Fig. 3.

Effect of light conditions(A), temperature (B), and pH (C) on Stability of CA/CNCs Pickering emulsions.

Fig. 3-A showed the temporal evolution of curcumin encapsulation efficiency for CA/CNCs Pickering emulsions under different light regimes. Over the 30-days observation, emulsions stored in the dark maintained a stable curcumin encapsulation efficiency above 85% with minimal fluctuations. This stability underscored the positive role of the dark environment in preserving structural integrity and curcumin-loading capacity. In contrast, emulsions exposed to natural light exhibited a gradual decrease in curcumin encapsulation efficiency after 3 days, dropping to 71.35% by day 15 and before stabilizing. This reduction stemmed from the inherent poor photostability of CA. Light exposure induces photo-oxidation of the caffeoyl and phenolic moieties in CA, leading to a reduction in the number of phenolic hydroxyl groups and alterations in its chemical structure such as the formation of quinones (Fujimoto & Masuda, 2012). These changes directly weaken its ability to form a stable hydrogen-bond network with CNCs, consequently reducing the capacity of emulsion to retain curcumin and thus lowering the encapsulation efficiency. Although the CNCs network initially provided some mitigation against CA oxidation at the oil-water interface, prolonged exposure to natural light ultimately induced significant oxidation and degradation of CA. Furthermore, the oxidized quinone derivatives are highly reactive and can undergo further covalent reactions with other CA molecules or functional groups on the CNC surface, forming irreversible aggregates (Wang et al., 2024). Such aggregation renders the interfacial film brittle and less flexible, making it more susceptible to rupture. These findings clearly confirmed that light disrupted critical CA-CNCs interfacial interactions for curcumin retention. Dark storage was thus imperative for sustaining high encapsulation efficiency and long-term emulsion stability, preserving CA-CNCs molecular synergy and ensuring consistent performance of curcumin-loaded emulsions, which held key significance for optimizing shelf life and functional efficacy in antioxidant delivery applications. Fig. 3-B showed changes in particle size, PDI, and zeta potential across temperatures (30–90 °C). Between 30–80 °C, the average droplet diameter of CA/CNCs Pickering emulsion remained stable at ∼3.5 μm, with zeta potential also remaining stable at ∼ − 36 mV. Only PDI showed a slight upward trend, rising from ∼0.22 at 30 °C to ∼0.32 at 80 °C, indicating temperature primarily affected t droplet size distribution uniformity. High temperatures weakened hydrogen bonding between CA, CNCs, and within the CNCs network while enhancing hydrophobic interactions, causing minor droplet aggregation and broadening size distribution. At 90 °C, the emulsion static stability was compromised, with the average droplet diameter increasing from ∼3.5 μm to ∼4 μm and the zeta potential decreasing from ∼−36 mV to ∼−31 mV. This destabilization is attributed to the weakening of hydrogen bonds between CA and CNCs and the enhancement of hydrophobic interactions among CNCs at elevated temperatures, leading to the relaxation of the interfacial network structure and promoting droplet coalescence (Liu et al., 2022). Thus, the 30–80 °C range better preserved emulsion structural integrity, demonstrating good thermal stability.

pH-responsive properties of Pickering emulsions are critical for applications in palatability improvement and controlled active ingredient release (Kargar, Fayazmanesh, Alavi, Spyropoulos, & Norton, 2012), with pH-tunable hydrogen bonding well-documented in the literature. Fig. 3-C examineed particle size, zeta potential, and PDI for curcumin-loaded CA/CNCs Pickering emulsions homogenized at room temperature for 10 min across varying pH values. As the pH increased from 3.0 to 7.0, emulsion droplet diameter gradually increased, and stability decreased. Notably, adjusting pH from 7.0 to 11.0 followed by homogenization reduced droplet size but failed to improve stability. In acidic environment, partially water-soluble CA underwent protonation, enhancing hydrogen bonding between CNCs surface hydroxyl groups. Transitioning from acidic to neutral conditions induced emulsion stratification, as CNCs detached from the interface, causing oil droplet coarsening and system destabilization (Mathews, Mertins, Angelov, & Angelova, 2022). At pH 11.0, phenolic compounds may hanve reacted with biopolymers in suspension and plant polyphenols were prone to oxidation and oligomerization in dissolved oxygen, forming low-solubility high-molecular-weight products (Kalashnikova, Bizot, Bertoncini, Cathala, & Capron, 2013). CA, which is naturally poorly water-soluble, exhibited further reduced solubility. Its affinity for nanocellulose via hydrogen bonding promoted CNCs surface deposits, inducing nanofiber cross-linking, constructing a continuous network around oil droplets, and stabilizing discrete oil droplets (Nanikashvili, Butenko, Deutsch, Lee, & Sloutskin, 2022). Thus, acidic conditions (pH 3.0 → 7.0) triggered oil droplet aggregation and interfacial film weakening, while alkaline conditions (pH 11.0) allowed partial droplet reconstruction. Despite pH responsiveness, CA/CNCs Pickering emulsions retained stability better in acidic environments, supporting application-specific and storage condition optimization.

In summary, curcumin-loaded CA/CNCs Pickering emulsions are most stable under dark, 30–80 °C, and acidic conditions, as light, high temperatures (≥90 °C), and alkaline environments impair stability via disrupted CA-CNCs interactions, aggregation, or incomplete droplet reconstruction. Results provided a theoretical basis for the adaptation of the emulsion to application scenarios and the optimization of storage conditions.

3.5. Microstructural characteristics of curcumin-loaded CA/CNCs Pickering emulsions

Cryo-SEM and LSCM were employed to characterize the microstructures of Pickering emulsions. These emulsions were prepared under different stabilizers and then stored for 7 days. The results were shown in Fig. 4, and the particle sizes and distributions revealed by Cryo-SEM and LSCM were consistent with those obtained from DLS and optical microscopy.

Fig. 4.

Cryo-SEM images (A) and CLSM images of Pickering emulsions prepared under different stabilizers.

For the stabilizer-free emulsions, in the cryo-SEM images (Fig. 4-A1), numerous emulsion droplets were entangled and formed a network structure through interconnection and dispersion. This directly indicated that the emulsion was existed in a thermodynamically unstable state due to the absence of effective interface regulation, with droplets lacking both directional arrangement driving forces and steric hindrance constraints. Correspondingly, in CLSM images (Fig. 4-B1), droplets appeared disordered and chaotic under bright field. Nile Red fluorescence signals were sparse, and Nile Blue showed no obvious fluorescence response. After superposition, droplet contours were barely distinguishable. These observations confirmed the intrinsic tendency of droplet coalescence and system instability from both oil-phase and water-phase perspectives.

For the CA-only stabilized emulsion, Cryo-SEM (Fig. 4-A2) revealed that although droplet aggregation was moderately alleviated, the emulsion remained dominated by irregular aggregation, accompanied by droplet coalescence and the persistence of disordered network structures. This highlighted significant limitations of CA in regulating interfacial wettability and constructing steric hindrance. CLSM observations (Fig. 4-B2) further validated this. CLSM observations (Fig. 4-B2) further validated this. Under bright field, droplets were small in size but discretely distributed. Nile Red fluorescence was dispersed with weak intensity and no blue interfacial ring was formed by Nile Blue. These results indicated that while CA possessed basic emulsifying activity, it failed to form a stable, continuous interfacial film, thus unable to provide an effective barrier for long-term droplet dispersion.

For the CNCs-only stabilized emulsion, Cryo-SEM (Fig. 4-A3) showed a trend toward discrete droplet distribution, with a marked reduction in droplet coalescence. This was attributed to electrostatic repulsion between CNCs, which drove their dispersion in the aqueous phase (Chen et al., 2024), and endowed the emulsion with a steric hindrance effect. In CLSM images (Fig. 4-B3), under bright field, droplets under bright field were larger with a more uniform distribution, as the CNCs fiber network created a stable 3D droplet-fiber network and an integrated entangled emulsion structure (Dai et al., 2021). Nile Red fluorescence highly matched droplet contours, and a weak Nile Blue fluorescence response was observed at the water phase interface. These findings verified that CNCs, leveraging their nanoparticle properties, could achieve interface adsorption and preliminary stabilization. However, the integrity and compactness of the interface film remained insufficient due to the limitations of a single component, limiting further improvements in stability.

For the CA/CNCs composite stabilized emulsion, Cryo-SEM (Fig. 4-A4) showed a high-density monodisperse droplet arrangement with significantly improved size uniformity. CLSM observations (Fig. 4-B4) more intuitively demonstrated the synergistic effect. Under bright field, the droplets were regular and ordered. The Nile Red-coated oil droplets were clear and continuous, and a complete blue interface ring was formed by Nile Blue at the oil-water interface. This phenomenon arose from CA optimizing interfacial wettability and providing anchoring sites for CNCs on the interface. CNCs and CA synergistically promoted the formation of a dense interfacial film and a stable emulsifying network through intermolecular interactions such as hydrogen bonding. This structured interface effectively inhibited droplet coalescence, correlating well with the high macroscopic stability (emulsification rate of 95%). Through the synergy of interfacial adsorption and structural regulation, the limitations of single-component stabilizers were overcome, endowing the emulsion with excellent long-term stability.

A structure-mechanism-performance correlation was validated through the dual dimensions of Cryo-SEM microdroplet arrangement and CLSM interfacial adsorption behavior, clearly illustrating the intrinsic logic by which the composite stabilizer synergistically enhances emulsion stability. Additionally, these microstructural characteristics were consistent with the corresponding macroscopic emulsion states (Fig. 2-C), thereby clearly explaining the underlying mechanism for the differential states of emulsions under the action of different stabilizers.

3.6. Interfacial tension analysis of curcumin-loaded CA/CNCs Pickering emulsions

Interfacial tension (IFT), a key parameter characterizing the difference in molecular interaction potential at the oil-water interface, is determined by the isotropic forces acting on bulk molecules with a net force of zero and the asymmetric interactions of interfacial molecules with adjacent phases which form a net force field pointing toward the interior of the phase. It directly reflects interfacial stability. To investigate the stabilization mechanism of curcumin-loaded CA/CNCs Pickering emulsions, this study measured the interfacial tension properties of different stabilizer systems using the pendant drop method, which involved dispensing the high-density aqueous phase into the low-density oil phase, results are shown in Fig. 5 and Table 1.

Fig. 5.

IFT measurement images of Pickering emulsions prepared with different stabilizer.

Table 1.

IFT of Pickering emulsions prepared with different stabilizer conditions.

Aqueous phase composition	IFT(mN/m)
No stabilizer	5.65 ± 0.17
CA	122.43 ± 0.91
CNCs	12.15 ± 0.68
CA/CNCs	29.19 ± 0.86

The stabilizer-free emulsions, with deionized water as the aqueous phase (Fig. 5-A), exhibited an interfacial tension of only 5.65 mN/m ± 0.17, which is lower than the typical value for normal olive-oil/water systems (∼20–35 mN/m). This reduction is attributed to the cavitation effect generated by sonication, which significantly decreases the oil-water interfacial tension (Abker, Wang, Chen, Ma, & Fu, 2023; Boostani et al., 2019). The extremely low interfacial constraint caused the oil phase to fail to form a stable liquid-liquid interface film, leading to spontaneous coalescence and separation driven by interfacial free energy, thus exhibiting thermodynamically unstable characteristics. Macroscopically, the droplets showed high reflectivity and irregular edges, which resulted from intense light reflection caused by exposed oil phases and discontinuous interface films. The CA-only stabilized emulsions (Fig. 5-B) had an extremely high interfacial tension of 122.43 mN/m ± 0.91. However, the rigid aggregates formed by excessive aggregation of CA molecules lacked dynamic adaptability and could not buffer external disturbances, leading to the destruction of interface film integrity and continuous exposure of the oil phase. Its highlighted features showed no significant difference from those of the stabilizer-free group, confirming that high interfacial tension was not positively correlated with actual stability and highlighting the limitation of molecular rigidity in single polyphenol stabilizers. The CNCs-only stabilized emulsions (Fig. 5-C) regulated the interfacial tension to an appropriate level of 12.15 mN/m ± 0.86 through steric hindrance and surface charge repulsion. The flexible particle adsorption layer not only reduced interfacial free energy to weaken the coalescence driving force but also inhibited oil phase aggregation through physical barriers. Macroscopically, the droplets had a smooth morphology and weakened highlighted features, reflecting basic stabilization. The CA/CNCs composite stabilized emulsions (Fig. 5-D) constructed a rigid molecule-flexible particle composite interface film through hydrogen bond synergy between CA polyphenol hydroxyl groups and CNCs surface hydroxyl groups, precisely regulating the interfacial tension to 29.19 mN/m ± 0.86. It synergistically optimized oil phase wettability and steric hindrance support higher interfacial tension are more conducive to liquid-liquid emulsification, as they allow ultrasound to generate a large number of smaller bubbles (John, Kuhn, Braeken, & Van Gerven, 2018). Macroscopically, the droplets were nearly spherical with the weakest reflectivity, confirming that the composite system could minimize oil phase exposure and achieve optimal stability. Molecular surfactants stabilize emulsions by reducing interfacial tension, a thermodynamically driven process. In contrast, solid particles stabilize emulsions by adsorbing at the interface to form a solid film, which exhibits irreversible interfacial adsorption and higher interfacial desorption energy, thereby helping to suppress emulsion instability (Binks, 2002; Chevalier & Bolzinger, 2013). In the Pickering emulsion system, the reduction of IFT is related to changes in the wettability/hydrophobicity of solid particles (Boostani et al., 2019). The IFT values of the three systems namely CA, CNCs, and CNCs/CA are all higher than those of the stabilizer-free group because both CA and CNCs are highly hydrophilic particles with a water contact angle <90°.

In summary, different stabilizers affected emulsion stability by regulating interfacial tension and interface film structure. Among them, the synergistic effect of the CA/CNCs composite system provided a key experimental basis for the stabilization design of curcumin-loaded Pickering emulsions, consistent with the research paradigm of interfacial interaction-structure construction-performance optimization in colloid and interface science.

3.7. Rheological analysis of curcumin-loaded CA/CNCs Pickering emulsions

Pickering emulsions, composed of two immiscible liquids, display unique viscoelastic behavior in rheological tests. The elastic response comes from energy storage during droplet deformation, while viscosity arises from energy dissipation at interfaces and within phases (Yuan, Zeng, Wang, Cheng, & Chen, 2021). These rheological properties are key indicators of emulsion stability and functionality. This study examined curcumin-loaded CA/CNCs Pickering emulsions with different stabilizers using steady shear and dynamic oscillatory tests to analyze their structural stability mechanisms, with results shown in Fig. 6.

Fig. 6.

Shear viscosity (A) and storage modulus (G') and loss modulus (G") (B) of Pickering emulsions prepared with different stabilizers.

The shear viscosity behavior (Fig. 6-A) revealed characteristic shear-thinning properties across all emulsions, with significant quantitative variations. The stabilizer-free emulsion displayed Newtonian fluid behavior with an initial viscosity of 105 mPa·s that decreased sharply to 1 mPa·s at a shear rate of 100 s⁻¹, representing a 99% reduction in viscosity. This behavior clearly indicated the absence of effective interfacial stabilization, with droplets interacting solely through weak van der Waals forces. In contrast, the CA-only stabilized emulsion exhibited a remarkably high initial viscosity of 1313 mPa·s due to the formation of rigid CA aggregates, yet showed poor shear stability with a viscosity reduction rate reaching 99.8%, confirming the structural brittleness of this network under shear conditions. The CNCs-only stabilized emulsion demonstrated improved performance with an initial viscosity of 605 mPa·s, maintaining better structural integrity through the formation of a nanofiber network. However, the most notable results were observed in the CA/CNCs composite stabilized emulsion, which achieved an optimal balance between rigidity and flexibility with an initial viscosity of 149 mPa·s while demonstrating exceptional shear stability (98.7% viscosity reduction). This superior performance stemmed from the synergistic stabilization mechanism where CA components provided structural framework through hydrogen bonding while the flexible CNCs network conferred dynamic adaptability.

Dynamic viscoelastic analysis (Fig. 6-B) provided deeper insights into the interfacial network dynamics. The stabilizer-free emulsion showed weak viscoelastic response with G' ranging from 0.05 to 80 Pa and G" from 0.005 to 220 Pa, undergoing a transition from elastic to viscous dominance above 10 Hz that confirmed its thermodynamic instability. The CA-only stabilized emulsion exhibited brittle network behavior with G' values between 2.0 × 10⁻⁴ and 32 Pa, displaying a gel-sol transition at 0.7 Hz that reflected its limited dynamic reorganization capability. In comparison, the CNCs-only stabilized emulsion maintained strong elastic dominance across all frequencies with G' values of 2.1–468 Pa, demonstrating the robustness of its nanofiber network. The CA/CNCs composite stabilized emulsion showed optimal viscoelastic balance with G' ranging from 0.1 to 361 Pa while maintaining elastic dominance throughout the frequency spectrum, indicating effective energy dissipation through its rigid-flexible interfacial network.

Rheological analysis revealed key structure-property relationships in Pickering emulsions. The CA/CNCs composite exhibited optimal performance through synergistic rigid-flexible network interactions. Frequency-dependent viscoelasticity provided critical design principles for functional emulsion systems, guiding advanced delivery system development.

3.8. Antioxidant activity of curcumin-loaded CA/CNCs Pickering emulsions

CA, as a natural phenolic substance, possesses excellent antioxidant potential (Wang et al., 2024). Previous studies have confirmed that the antioxidant efficacy of phenolic substances can be enhanced through interaction with cellulose. In particular, CNCs, with a higher specific surface area, have more potential binding sites with phenolic substances, thus exhibiting more outstanding antioxidant performance (Tang & Huang, 2022). Meanwhile, the inherent antioxidant effect of encapsulated curcumin cannot be ignored (Aw et al., 2022). This study analyzed the antioxidant activity of curcumin-loaded CA/CNCs Pickering emulsions, and the results were shown in Fig. 7. As can be seen from the figure, both the DPPH free radical scavenging capacity and the ABTS⁺ free radical scavenging capacity of the emulsions under different stabilizer conditions showed a consistent trend. Moreover, the Pickering emulsions loaded with curcumin exhibited better scavenging rates than those without curcumin. The DPPH free radical scavenging capacity and ABTS⁺ free radical scavenging capacity of the stabilizer-free emulsions were very weak, only 3.0% and 4.7% respectively. After adding curcumin, they increased to 25.8% and 18.2% respectively. For CA-only stabilized emulsion, the DPPH free radical scavenging capacity and ABTS⁺ free radical scavenging capacity increased to 16.6% and 17.7% respectively. After adding curcumin, they were further increased to 32.2% and 37.9%. For CNCs-only stabilized emulsion, the DPPH free radical scavenging capacity and ABTS⁺ free radical scavenging capacity were not ideal, only 10.2% and 14.2% respectively. However, after adding curcumin, they rapidly increased to 67.1% and 69.9%. This was not only related to the inherent activity of curcumin but also closely related to the uniform dispersion and stable existence of curcumin, which was consistent with the phenomena observed previously. For the CA/CNCs composite stabilized emulsion, the DPPH free radical scavenging capacity and ABTS⁺ free radical scavenging capacity were 52.6% and 62.9% respectively, showing a significant improvement compared with the single stabilizer. After adding curcumin, they rapidly increased to 76.9% and 81.2%, fully exerting the activity advantages of CA molecules and the structural advantages of CNCs. Through the regulation of dense interface and dispersion, the functions of CA molecules and curcumin were maximally released. This result was also consistent with our original purpose of selecting CA molecules as the stabilizer, that is, ingeniously utilizing the biological activity and interface regulation properties of CA.

Fig. 7.

DPPH free radical scavenging activity (A) and ABTS⁺ free radical scavenging activity (B) of Pickering emulsions stabilized with different stabilizers.

In any case, the results showed that the Pickering emulsions prepared with the CA/CNCs composite aqueous suspension as the stabilizer, regardless of the presence of curcumin, exhibited significantly higher antioxidant performance than other emulsions. This indicated that the structure-activity synergetic design was an effective strategy to break through the limitations of a single stabilizer and could achieve a performance gain exceeding the simple sum of individual components. In addition, the enhancement of the antioxidant performance of the emulsion by curcumin highly depended on the carrying capacity of the stabilizer. CA/CNCs possessed both strong basic antioxidant properties and super functional empowerment potential and were the optimal carrier for the release of the antioxidant value of curcumin. Selecting it for subsequent curcumin loading research could build a dual foundation of carrier carrying-function delivery and provide key support for the efficient application of curcumin.

3.9. Molecular interaction analysis

The Interaction Region Indicator (IRI) analysis were employed to systematically investigate hydrogen-bonding interactions in CA/CNCs composite stabilizers through high-precision quantum chemical calculations. As shown in Fig. 8-A, the δg versus sign(λ₂)ρ scatter plot revealed a distinct characteristic peak at sign(λ₂)ρ values of −0.040 with a variation of ±0.002, corresponding to δg values of approximately 0.07 a.u. This peak matched the established electron density range for hydrogen bonding interactions from 0.015 to 0.028 a.u., falling within the typical hydrogen bond range of 0.002 to 0.035 a.u. Topological analysis showed these interactions primarily originated from three sources with the caffeoyl group contributing about 60% through C=O…H-O interactions with cellobiose, phenolic hydroxyl groups contributing approximately 30% via O-H…O hydrogen bonds, and cooperative polarization effects accounting for the remaining 10% contribution. The minimal presence of sign(λ₂)ρ values greater than zero, covering only 2.7% of the area with δg values below 0.05 a.u., confirmed negligible steric hindrance effects in the system.

Fig. 8.

δg vs sign(λ₂)ρ scatter plot (a) and VMD-visualized intermolecular interactions (b) of CA/cellobiose.

Three-dimensional visualization through VMD isosurface projection provided spatial characterization of these molecular interactions. As shown in Fig. 8-B, blue regions with sign(λ₂)ρ values below −0.03 and electron densities between 0.020 and 0.025 a.u. clearly showed four directional hydrogen bonds between caffeoyl carbonyl groups and cellobiose hydroxyl groups. The dual-functional phenolic hydroxyl groups formed two distinct hydrogen-bonding networks with sign(λ₂)ρ values around −0.035. Green regions with absolute sign(λ₂)ρ values below 0.01 and electron densities under 0.01 a.u. represented van der Waals interactions in glycosidic moieties. Red regions with sign(λ₂)ρ values exceeding 0.02 accounted for less than 3% of total interactions, demonstrating excellent spatial complementarity between the molecular components.

The electronic-scale analysis revealed how CA/CNCs forms a rigid-flexible composite interface through directional hydrogen bonds with bond energies. These computational findings showed strong correlation with experimental emulsion stability results, providing fundamental theoretical insights for developing high-performance Pickering emulsion stabilizers. The study established quantitative relationships between electronic structure characteristics and macroscopic material properties while maintaining precise numerical values and proper technical terminology throughout the analysis.

4. Conclusion

Through hydrogen bond driven assembly of the natural polyphenol CA and CNCs, this study developed a bio-based Pickering emulsion stabilizer and fabricated high performance oil-in-water Pickering emulsions The CA/CNCs composite formed a dense, continuous adsorption layer at the oil-water interface, which significantly improved the static storage stability and dynamic mechanical stability of the emulsion. Additionally, the interfacially absorbed CA and curcumin in the oil phase exerted synergistic antioxidant effects, endowing the emulsion with efficient scavenging capacity for DPPH and ABTS⁺ free radicals. Combined analyses of FT-IR, ¹H NMR, and computational simulations unequivocally confirmed that the stable and extensive hydrogen-bonding network formed between CA and CNCs were the core mechanism underlying the improved emulsion performance. This work provides novel insights for developing green and sustainable food grade emulsion systems, while laying a theoretical and technical foundation for the application of natural polyphenols and nanocellulose in functional foods, cosmetics, and drug delivery, and demonstrating considerable potential for encapsulating lipophilic bioactives in nutraceutical formulations, cosmetic products, as well as in active packaging and biofilm coatings. Future research may focus on the scalability potential of this composite system and its scale-up feasibility for industrial production.

CRediT authorship contribution statement

Lu Yin: Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Kangjia Han: Writing – original draft, Visualization, Validation, Software, Investigation, Formal analysis, Data curation. Yuxiu Lv: Validation, Investigation, Data curation. Dong Xie: Software, Data curation. Qian Jiang: Data curation, Conceptualization. Ping Zhao: Writing – review & editing, Resources, Funding acquisition. Lihua Zou: Writing – review & editing, Funding acquisition, Formal analysis. Sida Xie: Writing – review & editing, Funding acquisition. Xiaoping Rao: Funding acquisition. Wei Li: Software, Resources, Conceptualization. Xiaoqin Yang: Writing – review & editing, Supervision, Project administration, Methodology, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This research was supported by the Young and Middle-Aged Academic and Technological Leaders of Yunnan Province, grant number 202205AC160049, National Natural Science Foundation of China, grant numbers 22267018 and 32060107; Yunnan Fundamental Research Projects, grant numbers 202501BD070001–027, 202401AT070295, and 202503AC100002.

Data availability

Data will be made available on request.