Commercial pea protein ingredients: a multiscale investigation of high-pressure homogenized emulsification and relation with physicochemical and interfacial properties
Yuqi Zhang, Åsmund Rinnan, Jacob J.K. Kirkensgaard, Milena Corredig, Vibeke Orlien
TL;DR
This study examines how different commercial pea proteins behave in food emulsions, revealing how their properties affect droplet size and stability.
Contribution
The study provides new insights into how pea proteins function as emulsifiers, linking their physicochemical properties to emulsion stability.
Findings
High-pressure homogenization produced emulsions with varying droplet sizes (1.1–2.7 μm, 5.8 μm, and >10 μm) depending on pea protein properties.
Oil droplet size correlated with soluble protein content and stability was mainly governed by electrostatic repulsion.
Three pea protein isolates are suitable for beverages and one for dressings based on emulsion stability.
Abstract
Utilization of commercial pea protein (PP) in food emulsions largely depends on empirical experience due to insufficient knowledge. Emulsions made of seven commercial PP ingredients by high-pressure homogenization (HPH) were studied and uncovered their relationship with physicochemical and interfacial properties of homogenized PP. Results showed that HPH produced PP supernatants of varying size and composition, leading to differences in interfacial adsorption of emulsions. Four emulsions had very small droplets (1.1–2.7 μm), in contrast to one with medium sized droplets (5.8 μm) and two with large droplets (> 10 μm). The oil droplet size was highly correlated with the protein content and size in the soluble phase, suggesting that emulsification was mainly driven by soluble proteins after homogenization. Moreover, small droplet size was not a prerequisite to flocculation and coalescence…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsProteins in Food Systems · Microencapsulation and Drying Processes · Pickering emulsions and particle stabilization
Introduction
1
The food industry is increasing its efforts to utilize consumer-friendly and sustainable plant protein ingredients. Peas (Pisum sativum L.), one of the most widely cultivated and consumed legumes worldwide, have received great attention due to their rich protein content, well-balanced amino acid profile (high lysine content), low allergenicity, and potential economic benefits (Sun et al., 2023; Zhang et al., 2022). Furthermore, pea proteins (PP) may be used due to their vast array of technological functions essential in foods, e.g., solubility, gelling, emulsifying, foaming, water, and oil absorption properties. Emulsification is one important functional property for many food applications (e.g., beverages, sausages, dressings, sauces, and soups). Because of the amphiphilic nature of protein molecules present in peas, particularly the surface activity of vicilin, PP ingredients are increasingly being used to form and stabilize emulsions (Burger & Zhang, 2019). The methods for evaluating emulsification and emulsion stability have become increasingly detailed (Burger & Zhang, 2019; McClements, 2015; Niu et al., 2023; Zhou et al., 2025). Nevertheless, the emulsifying properties of commercial PPs are not well described.
Commercial PP isolates are primarily isolated by alkali or salt extraction followed by isoelectric precipitation and spray drying, due to the limitations in cost of production and the requirement of high yields (Burger et al., 2022; Stone et al., 2015). However, large-scale processing into powder may result in commercial protein ingredients displaying different conformational structures and functional performances, especially from those of laboratory-prepared powders (Burger & Zhang, 2019; Stone et al., 2015; Taherian et al., 2011). For example, for laboratory-purified proteins the emulsifying property was reported to correlate with the solubility and interfacial properties of the proteins (Chang et al., 2023; Keivaninahr et al., 2021; Sha et al., 2021). In contrast, Burger et al. (2022) compared five commercial PP ingredients, but found no correlation between the emulsifying properties and the physicochemical (surface charge, solubility) and interfacial properties. The discrepancy between laboratory-purified proteins and commercially prepared ingredients originates from the processing-induced protein changes during the harsh industrial extraction conditions at which proteins self-assemble to form larger aggregates of various sizes. Those large protein aggregates will exhibit different emulsifying properties compared to the proteins from lab-scale extraction, wherefore the principles responsible for the oil droplet stabilization based on lab-produced isolates cannot be generalized to commercial protein ingredients. Currently, only limited number of investigations of emulsion formed with commercial PP ingredients exist, leaving a gap in the in-depth knowledge about the application of commercial PP ingredients as an emulsifier in the food industry.
This study collected seven different commercial PP ingredients to investigate their physicochemical, interfacial, and emulsifying properties. Specifically, the commercial PP ingredient suspensions were homogenized to obtain a polydisperse population of protein particles of various sizes. Interfacial adsorption was measured on the centrifugal supernatant of this dispersed phase, and the results were correlated with those of bulk emulsions, where the entire ingredient suspension was used. We hypothesize that when emulsions are processed by high-pressure homogenization (HPH), not only the droplets are refined, but the protein aggregates are also affected. The HPH-induced disintegration of PP particles may vary dependent on the initial conformation and characteristics of the proteins and aggregates. Therefore, this process can generate a new polydisperse population of PP particles, which may adsorb differently at the oil–water interface and result in different oil droplet stabilization. The goal of this study is to derive overall principles to understand the emulsification and stabilization mechanisms of commercial PP ingredients sourced from multiple suppliers. The information obtained will also provide new insights into which properties may best reflect the emulsifying properties of commercial PP, important for the development of viable future food applications. We anticipate this to be a steppingstone to connect scientists and industrial R&D to develop or optimize processing steps to fine tune the emulsification properties of PP ingredients.
Materials and methods
2
Materials and composition
2.1
Seven commercial PP ingredients, one protein concentrate (PPC) and six protein isolates (PPIs), were provided from Vestkorn A/S (Tau, Norway), Emsland group (Emlichheim, Germany), Roquette Freres (Lestrem, France), SEAH international (Wimille, France), and AGT Foods Minot (Minot, ND, US). It is noted that the processing history was not disclosed. Protein content was determined using the Dumas combustion analysis method (AOAC Method 990.03, Rapid MAX N exceed, Elementar, Langenselbold, Germany), using a nitrogen-to-protein conversion factor of 6.25. Lipid content was measured by a Soxhlet extraction unit (HT 1043, Tecator, Höganäs, Sweden) with petroleum ether as a solvent. Ash content was determined following the AOAC method 923.03 using a muffle furnace (Nabertherm, Lilienthal, Germany). Moisture content was measured by drying 1 g of the sample at 105 °C in a moisture analyzer (MA160, Sartorius AG, Gottingen, Germany). Carbohydrates content was then calculated by weight difference (100 − the sum of protein, fat, ash, and moisture).
Sunflower oil was purchased from the local supermarket. Florisil (100–200 mesh) was procured from Sigma-Aldrich (Søborg, Denmark). Other chemicals were of analytical grade and purchased from Sigma-Aldrich (Søborg, Denmark). All water used was purified using a Milli-Q water purification system (Merck, Darmstadt, Germany).
Sample and emulsion preparation
2.2
Fig. 1 shows the overview of samples and related analyses as presented in the following sections. PP ingredients were suspended in Milli-Q water under native pH and ionic strength conditions with a concentration of 1% (w/w, amount of ingredient powder dispersed in water) and stirred for 30 min at room temperature. The protein suspensions were centrifuged at 9000×g for 20 min to obtain the non-homogenized supernatants. The protein concentration in the non-homogenized supernatants was measured using a PierceTM bicinchoninic acid (BCA) assay kit (Thermo Fisher Scientific, Waltham, Massachusetts, USA).Fig. 1. The flowchart showing the different samples in bold together with the related analyses.Fig. 1
PP suspensions were pre-homogenized using an Ultra Turrax (T25 basic, IKA® Works, Wilmington, NC) at 10000 rpm for 2 min, followed by passing through a high-pressure homogenizer (Panda PLUS 1000; GEA Niro, Soavi, Italy) at 200 bars for two passes. The homogenized protein suspensions were then centrifuged at 4700×g for 30 min and the supernatants were collected.
All emulsions were prepared under the same homogenization conditions according to the method detailed by Sharan et al. (2022). In detail, coarse emulsions were prepared by pre-homogenizing the protein suspensions (1% w/w) with sunflower oil (9:1 v/v, protein suspension:sunflower, sunflower oil) using an Ultra Turrax at 10000 rpm for 2 min. The coarse emulsions were passed through a high-pressure homogenizer twice at 200 bars. The final fine emulsions were stored at 5 °C and analyzed at 0, 7 and 14 days of storage.
Properties of homogenized PP suspension
2.3
Protein concentration and particle size
2.3.1
The protein concentration in the homogenized supernatants was measured using a BCA protein assay kit (Thermo Fisher Scientific, Waltham, Massachusetts, USA). The particle size distribution of the homogenized suspensions and their supernatants was determined using a Malvern Mastersizer 3000 (Malvern Instrument Ltd., Malvern, England) with particle size measurement range 0.01–3500 μm. Samples were added to an obscuration between 4 and 10%. The refractive index (RI) values of 1.45 and 1.33 were used for the dispersed phase and deionized water, respectively. The results are reported as the volume average diameter (d_4,3_) from two independent replicates.
Interfacial properties
2.3.2
To remove the surface-active impurities of oil, the sunflower oil was treated with Florisil (Kornet et al., 2022). Florisil (15 mL) was mixed with the sunflower oil (30 mL) in centrifuge tubes and then shaken vigorously. The tubes were covered with aluminum foil to avoid light oxidation and then put on a PTR-35 multi-tube rotator (Grant Instruments Ltd., Cambridge, UK) stirring at 15 rpm at room temperature. After 24 h of rotation, samples were centrifuged twice (5000×g, 20 min, 20 °C). The final supernatant containing the stripped sunflower oil was collected for the following experiments.
Dynamic interfacial tension at the oil-water interface was performed using a pendant drop tensiometer (Surface Analyzer LSA100, LAUDA Scientific GmbH, Germany) with some modifications (Grasberger et al., 2022). The supernatants were diluted with water at a ratio of supernatant: water = 1:19. A 68 μL drop of the diluted supernatant formed at the tip of a stainless-steel needle (Ø = 1.8 mm, Krüss GmbH, Hamburg, Germany) was immersed in a cuvette containing the purified sunflower oil at room temperature (∼23 °C). The interfaces were measured for 10,800 s at room temperature. Interfacial tension values were calculated with the Young-Laplace equation by fitting the droplet contour changes. The curve inflection point, defined as the point where the curve has its largest change, was calculated by i) smoothing the data with a wide Savitzky-Golay filter (91 data points and a 1st-order polynomial), ii) normalizing the time-axis and the signal up to the first 1500 data points, and iii) calculating the lowest squared sum of the two numbers, normalized time and normalized signal. The initial adsorption rate ( , mN/m·min) was calculated using the following equation described by Grasberger et al. (2023) with slight modifications:
where γ_0_ refers to the interfacial tension at the starting point (t = 0), γ_ip_ represents the interfacial tension at the inflection point, and t_ip_ represents the time required to reach the curve inflection point.
After reaching the equilibrium, interfacial dilatational measurements were determined by an oscillation method. The drop volume was expanded and compressed at a deformation amplitude of 10% (ΔV/V) and a frequency of 0.02 Hz. A cycle consisted of three sinusoidal oscillations and a rest period of 50 s and the cycle was repeated five times. Experiments were carried out in duplicates by preparing each replicate starting from 1% protein suspension.
Properties of PP stabilized emulsions
2.4
Confocal laser scanning microscopy (CLSM)
2.4.1
The microstructure of emulsions was captured using a Leica SP5-X Confocal Laser Scanning Microscope (Leica Microsystems CMS GmbH, Wetzlar, Germany) with a 40× oil immersion objective lens as described previously with some modifications (Xiong et al., 2019; Zhang et al., 2022). Nile blue and Nile red were used to stain protein and oil, respectively. Approximately 0.5 mL of the emulsion was mixed with 10 μL of Nile red (1 mg/mL) and 20 μL of Nile blue (1 mg/mL). A drop of each stained emulsion was placed on a microscope slide and covered with a coverslip. Nile red was excited at 488 nm and emitted at the range of 520–602 nm and Nile blue was excited at 633 nm and emitted at the range of 661–749 nm. The image pixel resolution was 1024 × 1024.
Droplet size
2.4.2
The droplet size of the fresh (day 0) and stored (day 7 and 14) emulsions was determined by a Mastersizer 3000 as described before (Zhi et al., 2022). Before measurement, the emulsions were diluted ten times with Milli-Q water or 1% (w/v) sodium dodecyl sulphate (SDS) solution. An obscuration of 4–10% was used. Refractive index (RI) values of Milli-Q water and sunflower oil were set to 1.33 and 1.471, respectively, while the absorbance index was 0.01. Results are presented as particle size distribution (PSD) profiles and droplet volume average diameters (d_4,3_).
The storage stability of the PP emulsions was evaluated during storage for 14 days at 5 °C by measuring the change in oil droplet size. The flocculation and/or coalescence behavior was assessed by the addition of SDS, as SDS disrupts non-covalent interactions between proteins and dissociates flocculated droplets, allowing the measurement of individual droplet instead of flocculates. The flocculation index (FI) and the coalescence index (CI) were calculated by the Eqs. (2), (3), respectively (Zhi et al., 2022):
where d_4,3_ values were obtained from the above measurements of fresh and stored emulsions diluted with water or SDS.
Accelerated stability
2.4.3
LUMiSizer (LUM GmbH, Berlin, Germany) was used to measure accelerated emulsion stability by measuring the changes in transmission over time. 400 μL of each emulsion was centrifuged in a disposable 2 mm cuvette at a speed of 4000 rpm for 4000 s at 20 °C. The transmission profile was recorded every 10 s. The emulsions were measured the day after preparation. The first measurement was subtracted from all the remaining to remove the baseline. A one-component principal component analysis was made on the concatenated data from all measurements with the subtracted baseline, and subsequently, the score values corresponding to each measurement were evaluated separately. The first derivative of the score for each measurement was used as a parameter for instability, as this describes the rate of signal change. This gives an instability profile across the time axis per measurement, and the instability index was given as the maximum of this profile.
Viscosity
2.4.4
The viscosity of fresh emulsions was determined by a rheometer (Kinexus Pro, Malvern, Worcestershire, UK) equipped with a cylinder geometry (D = 35 mm, 4° truncation, type C25G SW1370 SS). The flow curve was recorded at 20 °C by increasing the shear rate from 0.03 to 300 s^−1^ (Christiansen et al., 2021).
Zeta potential
2.4.5
The fresh emulsion was diluted 100 times with Milli-Q water prior to measurement. The zeta potential was determined by a Malvern Zetasizer Nano (Malvern Instruments Ltd., Malvern, UK) under native ionic strength conditions, according to the method detailed by Zhang et al. (2022). The results were automatically analyzed after 10–100 runs and presented as average from independent duplicates and standard deviation.
Front-face fluorescence spectroscopy
2.4.6
Emulsions were separated following the method of Xiong et al. (2018) with slight modifications. Fresh emulsions were centrifuged at 15000×g for 30 min at 20 °C. The cream layer was collected and dried on a filter paper. The dried cream was resuspended in water, reaching the initial oil ratio (10%), and then diluted ten times with Milli-Q water to prepare the cream phase containing adsorbed proteins for measurement. The remaining water phase was collected by a syringe and centrifuged at 15000×g for 30 min at 20 °C to remove residual oil. The supernatant was collected and filtered through a 0.5 μm filter. The filtered solution containing unadsorbed protein was named the serum phase.
Fluorescence spectra of the cream and serum phases were recorded using FS920 Edinburgh Instruments fluorescence spectrophotometer (Livingston, Scotland, UK) equipped with front-face geometry (45° angle and 35.9° inclination, with an in-house built sample holder allowing for temperature controlling the sample). The temperature during measurement was controlled at 20 °C by an external temperature control unit (FL300 Recirculating Cooler, Julabo GmabH, Seelbach, Germany) connected to the sample holder. Excitation and emission matrices were recorded with excitation wavelengths from 260 to 340 nm (5 nm step) and emission wavelengths from 275 to 500 nm (1 nm step). The slit widths were set at 2 nm for excitation and emission pathways, and the dwell time was 0.15 s/nm. The obtained data are analyzed by removing the Rayleigh scatter by the method proposed by Thygesen et al. (2004) followed by parallel factor analysis (PARAFAC). This separates the signal into its underlying estimated contributions and is given in three matrices: an excitation loading matrix, an emission loading matrix, and a score matrix. The cream and serum phase were analyzed separately, and the data were both modelled using four factors. The above analyses were performed using MATLAB 2017b (Mathworks, Natick, Massachusetts, United States) with the N-way toolbox (Andersson & Bro, 2000) and in-house functions based on existing algorithms.
Small angle X-ray scattering (SAXS)
2.4.7
SAXS was measured using a Nano-inXider instrument from Xenocs SAS (Grenoble, France) having a 40 W micro-focused Cu source (Rigaku- Denki, Co., Tokyo, Japan) producing X-rays with a wavelength λ = 0.154 nm. The Nano-InXider has a Pilatus 100 k pixel-detector from Dectris (Baden, Switzerland) measuring the scattering in a q-range from 0.005 to 0.3 Å^−1^, where the scattering vector q = 4πsinθ/λ and 2θ is the scattering angle. 2D detector images were azimuthally averaged to produce 1D intensity curves as a function of q. The fresh and stored emulsions were injected into a borosilicate capillary tube and measured at room temperature for 1 h. The XSACT software was used for data reduction and logarithmic re-binning. All the SAXS data is presented in log I(q)- log q plots.
Statistical analysis
2.5
Each measurement was conducted using at least two independent biological replicates, and for each biological replicate, technical triplicates were performed. Results were shown as mean ± standard deviation. A one-way ANOVA analysis and Duncan's test were conducted using SPSS statistics software 28.0. Significant differences were indicated by p < 0.05. Pairwise correlation between the above parameters was conducted using Pearson correlation analysis, which was visualized as a heatmap using OriginPro (version 2024b; Northampton, MA, USA).
Results and discussion
3
The results of PP ingredient composition (Section 3.1), homogenized suspension (Section 3.2), and emulsion (Section 3.3) will be presented and discussed. The discussion will be supported by the Pearson correlation analysis (Fig. S1) to facilitate the identification of distinct correlation patterns within the PP ingredients and the factors affecting emulsion properties.
Composition of commercial PP ingredients
3.1
The composition of the PP ingredients is shown in Table 1. PPC had a lower protein content of 52.1% and a higher carbohydrate content of 31.3% compared to the PP isolates, in agreement with published literature (Keivaninahr et al., 2021). The protein content of the six commercial PP isolates ranged from 76.9% to 83.6%, as expected (Stone et al., 2015; Sun et al., 2023). It is important to note that the conversion factor used was 6.25, and therefore values may be overestimated. Nonetheless, in the present study the protein concentration estimation is sufficient to enable a comparison among the PP ingredients. PPC and PPI3 had considerably higher lipid contents than PPI1, PPI2, and PPI6, while PPI4 and PPI5 had very low lipid contents. It has been shown that protein isolates are often co-extracted with lipids (Keuleyan et al., 2023). The ash content of PPI5 and PPI6 was 8.2% and 7.9%, respectively, which were markedly higher than those of the other pea ingredients. This may likely be ascribed to the addition of alkalis or acids or salts remaining during extraction, thus (Sun et al., 2023). The other PP ingredients had an ash content between 3.9% and 5.9%, which was closer to reported values of 5.2%–5.4% for pea concentrates and isolates (Keivaninahr et al., 2021). The moisture content of protein isolates varied from 6.8% to 7.9%, which was lower than that of PPC. The functionality of proteins is strongly dependent on pH, hence the neutral pH should be reported, as it can significantly differ even between batches of the same commercial PP product. As seen, the pH of the PP ingredient suspensions varied only slightly between 6.7 and 7.9, which are in line with previously reported values. The pH of PP isolates commercially produced from ten different pea cultivars was in the range 6.7–8.2 (Foley et al., 2025). Othmeni et al. (2024) found that commercial PP isolate had a neutral pH of 7.5 (5% w/v in protein), while a range of 6.60–7.56 for nine commercial PP isolates was reported by Jakobson et al. (2023). A wider pH range of 5.80–7.72 (3% wt dispersion) was measured for five different commercial PP isolates (Ebert et al., 2020). Nevertheless, the native pH of the seven PP ingredients was above the reported isoelectric point (around 4.5) (Foley et al., 2025). The PP ingredients had minor compositional differences, except for PPC due to its dry fractionation process, whereas large discrepancies were seen for the particle size, d_4,3_, a consequence of the different isolation histories. Industrial wet protein extraction involves large shifts in extraction pH to facilitate protein solubility and precipitation following harsh drying conditions that result in protein denaturation and aggregation. The impact on protein changes depends on applied extraction pH as well as drying temperature and time. Despite the unknown processing conditions of the examined isolates, aggregation of different extents and variation in particle sizes are expected. Notably, d_4,3_ of PPI4, PPI5, and PPI6 were markedly higher than those of the other PP ingredients representing the presence of large particles/aggregates, while PPC contained the smallest protein size probably due to the retention of native proteins after dry fractionation. A similar noticeable variation in the particle size, d_4,3_ values ranging from 6.2 to 42.0 μm, among different commercial PP ingredients was reported by Sun et al. (2023).Table 1. Composition (wet basis) of the commercial PP ingredients. Label is the short name for the respective ingredients used throughout the paper. Values are given as the mean ± standard deviation (SD). Different letters within the same column indicate significant differences (p < 0.05).Table 1. LabelProtein content (%)Carbohydrates (%)Lipid (%)Ash (%)Moisture (%)pH of 1% ingredient suspensionParticle size (d_4,3_) of 1% ingredient suspensionPPC52.1 ± 0.2^f^31.3 ± 0.0^a^2.2 ± 0.0^b^5.8 ± 0.0^b^8.4 ± 0.0^a^6.7 ± 0.0^f^21.4 ± 0.0^g^PPI183.6 ± 0.3^a^5.3 ± 0.2^d^0.2 ± 0.0^e^3.9 ± 0.0^d^7.3 ± 0.2^e^7.8 ± 0.0^b^80.8 ± 1.8^d^PPI280.5 ± 0.0^c^8.0 ± 0.0^c^0.3 ± 0.0^d^4.4 ± 0.0^c^6.8 ± 0.0^g^7.9 ± 0.0^a^43.9 ± 0.8^e^PPI379.0 ± 0.5^d^5.8 ± 0.6^d^3.6 ± 0.0^a^4.2 ± 0.0^c^7.4 ± 0.0^d^7.0 ± 0.0^e^35.0 ± 0.3^f^PPI476.9 ± 0.1^e^9.0 ± 0.1^b^0.1 ± 0.0^f^6.0 ± 0.0^b^8.0 ± 0.1^b^7.5 ± 0.0^c^114.6 ± 1.74^a^PPI581.2 ± 0.3^b^2.9 ± 0.3^e^0.0 ± 0.0^f^8.2 ± 0.1^a^7.8 ± 0.0^c^7.5 ± 0.0^d^97.7 ± 0.6^c^PPI678.7 ± 0.1^d^5.4 ± 0.3^d^0.9 ± 0.0^c^7.9 ± 0.4^a^7.0 ± 0.1^f^6.7 ± 0.0^f^101.6 ± 0.8^b^
PP suspension without oil
3.2
To gain insight into the different PP ingredients, the effect of HPH to potentially reduce the protein aggregates particle size (without oil) was evaluated.
Protein concentration and particle size
3.2.1
Plant protein ingredients are a polydisperse mixture of native protein molecules and protein aggregates, with important consequences to particle size distribution, composition, and structural flexibility of the protein molecules. Hence, upon hydration, the PP isolates may undergo different extents of solubilization, therefore producing a suspension containing a mixture of soluble and insoluble PP particles. Assessment of protein solubility and particle size can explain their aggregation and dispersion behavior due to intermolecular, protein-associated reactions as an effect of the powder processing. Moreover, the effect of HPH on the protein is also relevant during emulsion preparation, as HPH not only leads to the formation of small oil droplets, but also contributes to the disruption of large protein aggregates. Therefore, the different homogenized PP suspensions without oil (Fig. 1) were compared for their protein solubility and particle size. The initial PP suspensions (non-homogenized supernatant, Fig. 2) differed considerably in protein solubility, ranging from 0.6 (PPI6) to 2.6 mg/mL (PPC). However, all PPIs exhibited low protein solubility, likely due to their contents of large, non-soluble aggregates. Comparing the natural mean particle size of the PP ingredients (Table 1) with the HPH treated counterparts it is clear that HPH resulted in considerably smaller particles (Table 2) due to the disruption of intermolecular interactions. Similar effect of HPH to significantly decrease the aggregate sizes of commercial PP isolates have been previously observed (Grasberger et al., 2022; Grasberger et al., 2024). Nonetheless, the mean particle size of the homogenized suspensions varied considerably and could be classified into two classes; one with PPC, PPI4, and PPI6 having the largest mean particle size (d_4,3_ > 20 μm), while the other with PPI1, PPI2, PPI3, and PPI5 having considerably smaller average diameters (d_4,3_ < 10 μm) (Table 2). Clearly, homogenization formed different protein dispersions, and this may be important for understanding the emulsifying behavior of the different protein ingredients (cf. Section 3.3). After centrifugation of the homogenized suspension, the remaining larger particles were removed, and all samples showed monomodal distributions (not shown) with d_4,3_ values much lower than 1 μm (homogenized supernatant, Table 2). The breakup of larger particles upon homogenization is expected to change the ability of the protein to remain dispersed in solution. The HPH exhibited pronounced effect on the solubility of PPI1, PPI2, PPI4, and PPI5, with PPI1 having the most substantial increase after centrifugation (Fig. 2), in agreement with other studies (Burger et al., 2022; Grasberger et al., 2022). Notably, the protein solubility of PPI3 and PPI6 did not change much between the non-homogenized suspension and after homogenization, despite the large effect on particle sizes reduction, indicating that these two homogenized isolates were dominated by poorly hydrated particles. PPI6 not only contained much less soluble proteins than all the other isolates (Fig. 2), but also markedly larger particles (516 nm) in the centrifuged homogenized supernatant. On the other hand, HPH only slightly enhanced the protein concentration of the PPC suspension (Fig. 2), as in this case, the proteins were already more soluble, compared to PPI, due to their unprocessed nature. Overall, particle size is primarily, but not solely, related to protein solubility, which was supported by the high correlation coefficient of −0.84 (d_43_-HSN and PC-H, p < 0.001, Fig. S1).Fig. 2. The solubility of non-homogenized and homogenized supernatants (cf. Fig. 1). Different uppercase (A−F) and lowercase (a−f) letters indicate significant differences (p < 0.05), and error bars represent standard deviation (SD).Fig. 2. Table 2The particle size (d_4,3_) of the homogenized suspensions and their supernatants. Values are given as the mean ± SD. Different letters within the same column indicate significant differences (p < 0.05).Table 2. Sampled_4,3_Homogenized suspension (μm)Homogenized supernatant (nm)⁎PPC23.6 ± 0.0^c^333^b^PPI16.7 ± 0.2^e^267^e^PPI25.5 ± 0.6^e^286^d^PPI38.8 ± 2.0^d^291^c^PPI426.7 ± 0.1^b^287^cd^PPI58.4 ± 0.3^d^247^f^PPI637.5 ± 0.8^a^516^a^⁎SD values are not reported as they were below the instrument's detection threshold (10 nm).
Interfacial properties
3.2.2
In the process of emulsification, proteins diffuse to the oil-water interface and adsorb, followed by conformational rearrangements and the formation of an interfacial layer. The adsorption dynamics and viscoelastic properties of the interfacial layer formed by the homogenized supernatants were measured to obtain information on the interfacial properties of the various PP ingredients.
The changes in the interfacial tension at the oil-water interface were recorded as a function of time, as shown in Fig. 3A. For all PP supernatants after centrifugation, the interfacial tension decreased very rapidly at the initial stages of adsorption, followed by a slower decrease, and finally entered a pseudo-equilibrium plateau. These changes of the interfacial tension indicate different stages of the film formation; the diffusion of the proteins from the bulk to interface, adsorption, conformational rearrangement, and a final structural consolidation (Dickinson, 2011). As seen, there was a distinguished difference among the adsorption isotherms both in levels and shapes of the interfacial tension evolution over time reflecting different interfacial properties of the proteins in the homogenized supernatants. Notably, previous studies evaluated the in-depth interfacial properties of maximum four protein ingredients by various qualitative and quantitative assessments (Choe et al., 2022; Grasberger et al., 2022; Grasberger et al., 2023; Grasberger et al., 2024). However, in this study seven different commercial PP ingredients were analyzed for their interfacial behaviors, introducing greater variability making it more challenging to derive a universal explanation. Here we attempt to evaluate the adsorption dynamics of PP at the oil-water interface by quantifying the inflection point, the interfacial tension at the inflection point (γ_ip_), the initial adsorption rate and the interfacial tension at the final time point (γ_10800_) as presented in Table 3. In general, the decrease in γ was triggered immediately as time passed, seen as different γ_0_ at the very first measurement (t = 0), in the insert Fig. 3. Moreover, the evolution of γ in the early stage was clearly different among the PP ingredients. The inflection point was taken as an estimation of the protein's time to diffuse to and adsorb on the oil-water interface. As seen in Table 3, the fastest inflection point was reached by PPI2, while PPI4 and PPI5 required significantly longer time to attain the inflection point. The interfacial tension at the inflection point (γ_ip_) varied significantly from PPI1 having the significantly lowest interfacial tension of 21.2 mN/m to PPI4 having the highest γ_ip_ value of 26.5 mN/m at the diffusion/adsorption stage. However, there was no relationship between the time to reach the inflection point and the value of γ_ip_ (see orders in low part of Table 3), highlighting different early-stage mechanisms. The interfacial adsorption rate, , accounts for the different γ patterns of the early-stage period, and varied significantly in order of increasing values: PPI3 = PPI4 < PPI5 < PPC = PPI6 < PPI2 = PPI1 (Table 3). Thus, PPI1 and PPI2 were significantly more efficient at reducing the interfacial tension at the initial adsorption stage. The diffusion and adsorption have been reported to be dependent on protein solubility (Choe et al., 2022). As seen (Fig. 2), the fast initial adsorption rate corresponded to the high content of soluble proteins for PPI1 and PPI2 in the homogenized supernatants. However, this was not the case for other isolates, especially for PPI6, which had an extremely low protein solubility, and large particle sizes (Table 2) but a medium indicating that other molecular characteristics of PPI6 promoted adsorption. Hence, was neither correlated with protein solubility (R = 0.3, and PC-H, p > 0.001, Fig. S1) nor the protein particle size (R = −0.096, and d_43_-HSN, p > 0.001, Fig. S1), indicating that they were not highly relevant, as discussed by Choe et al. (2022) and Grasberger et al. (2022). It can be hypothesized that the varied initial adsorption rates of these samples are likely governed by their differences in the protein aggregated structure and flexibility. On the other hand, the final interfacial tension γ_10800_, reflecting the molecular rearrangements at the oil-water interface, was highly correlated to PC-H with the correlation coefficient of −0.85 (p < 0.001, Fig. S1). This may suggest that a minimum level of soluble protein is needed to adsorb at the oil-water interface and to effectively reduce interfacial tension. As seen, γ_10800_ varied significantly in the increasing order: PPI1 = PPI5 < PPI2 = PPI4 < PPC ≤ PPI3 ≤ PPI6 (Table 3). However, the final interfacial tension is not only influenced by the protein solubility but also the type and size of the protein particles (a trend of positive correlation, R = 0.71, γ_10800_ and d43-HSN, p < 0.001, Fig. S1). Indeed, protein rearrangements are affected by protein conformational flexibility (Choe et al., 2022), as well as by exchanges between the adsorbed and continuous phase. Recently, Grasberger et al. (2024) argued that different adsorption rates of PP isolates with various protein aggregation states are well explained by the difference in the proteins' composition and their structure. Moreover, it has been shown that the presence of polysaccharide molecules can decrease the diffusion rate, but can also increase the rearrangement rates of soy protein isolate (Wang et al., 2020). However, in our study there were no significant correlations between or γ_10800_ and the ingredient composition (p > 0.001, Fig. S1). In agreement, PPC and PPI3, albeit containing the highest amount of carbohydrates and lipids (most likely phospholipids), respectively, did not possess improved interfacial properties compared to other PP ingredients.Fig. 3(A) Interfacial tension isotherms, insert showing the early-stage period of 0–500 s, (B) elastic modulus and viscous modulus of oil-water interface stabilized by homogenized supernatants. Different uppercase (A−E) and lowercase (a−d) letters indicate significant differences (p < 0.05), and error bars represent SD.Fig. 3. Table 3The inflection point, interfacial tension at the inflection point (γ_ip_), the initial adsorption rate and interfacial tension at pseudo-equilibrium at 10800 s (γ_10800_) measured at the oil-water interface stabilized by homogenized supernatants. Values are given as the mean ± SD. Different letters within the same column indicate significant differences (p < 0.05). The lower part of the table shows the increasing order of these different data to facilitate comparison among the PP ingredients, cf. Section 3.2.2.Table 3. SampleInflection point (s)γ_ip_ (mN/m) (mN/m.min)γ_10800_ (mN/m)PPC219.1 ± 0.0^cd^24.8 ± 0.6^b^2.36 ± 0.16^b^19.2 ± 0.5^b^PPI1198.1 ± 8.1^cd^21.2 ± 0.5^e^3.04 ± 0.03^a^16.2 ± 0.4^d^PPI2186.6 ± 2.4^d^23.7 ± 0.2^c^2.85 ± 0.02^a^18.1 ± 0.2^c^PPI3205.7 ± 8.6^cd^26.4 ± 0.2^a^1.44 ± 0.00^d^19.7 ± 0.1^ab^PPI4310.8 ± 21.4^b^26.5 ± 0.1^a^1.47 ± 0.05^d^18.1 ± 0.4^c^PPI5366.3 ± 49.1^a^22.7 ± 0.1^d^1.80 ± 0.18^c^16.6 ± 0.1^d^PPI6237.1 ± 21.1^c^24.9 ± 1.2^b^2.32 ± 0.37^b^20.3 ± 0.7^a^Inflection point: PPI2 ≤ PPI1 = PPI3 = PPC < PPI6 < PPI4 < PPI5;γ_ip_: PPI1 < PPI5 < PPI2 < PPC = PPI6 < PPI3 = PPI4; : PPI3 = PPI4 < PPI5 < PPI6 = PPC < PPI2 = PPI1;γ_10800_: PPI1 = PPI5 < PPI2 = PPI4 < PPC = PPI3 < PPI6.
Interfacial dilatational rheology reflects the intermolecular interactions occurring onto the formed film at pseudo-equilibrium, and may be related to emulsion stability (Grasberger et al., 2022; Lu et al., 2022). Oscillatory measurements were carried out after the formation of the interfacial film and the elastic and viscous modulus are presented in Fig. 3B. All the homogenized supernatants displayed a higher elastic modulus than viscous modulus, implying a solid-like nature of the interfacial film formed by the PPs. The homogenized supernatants of PPC and PPI6 both had high γ_10800_, but opposite viscoelastic modulus. PPC showed both high elastic and viscous modulus, whereas PPI6 showed low elastic and very low viscous modulus. It has been previously shown that a PP isolate prepared in the laboratory under controlled conditions, tested at concentrations comparable to the present experiments, exhibited slow adsorption at the interface and a high γ_10800_, but also showed a strong, viscoelastic interface (Grasberger et al., 2024). Large volume particles may not form a closed-packed surface area, thus a less crowded film. Hence, it is then suggested that PPI6's interfacial film was partially composed of large, stiff protein particles, not able to rearrange at the interface and incapable of forming a coherent film due to steric hindrance. While PPC's large, soluble native proteins were moderately slow and reached a high interfacial tension, they underwent rearrangements at the interface and interacted to form an elastic protein network. Of the isolates, PPI2, PPI3, and PPI4 showed the highest elastic modulus reflecting the formation of a strong viscoelastic interface surrounding the oil droplets. These homogenized supernatants exhibited moderately high values of particle size (Table 2) indicating that these sizes of soluble aggregates might not effectively reduce surface tension but form a packed film of intermolecular interactions with high elasticity. In comparison, PPI1 and PPI5 had a lower elastic modulus concomitant with the lowest γ_10800_ and smallest particles. Even though PPI1 and PPI5 seemed to exhibit superior interfacial dynamics (significantly better at reducing the final interfacial tension), the resulting film possessed weaker lateral interactions (significantly lower E′). It has been suggested that the hydrophobic amino acids have a greater affinity for the oil phase and therefore do not form inter-amino acid interactions, resulting in a weakened protein network (Kontogiorgos & Prakash, 2023). The differences in rheological properties suggest that not only size but also flexibility of particles is important, both are to some degree determined by type of particles, such as PPC primarily being native proteins retained from dry fraction and PPI6 being large protein aggregates after harsh wet extraction processing.
Emulsions
3.3
Fresh and stored emulsions were investigated by various analyses (Fig. 1) to further understand the mechanism of bulk emulsification and emulsion stability. It is important to note that the whole ingredient suspensions containing soluble and insoluble proteins and particles, and not only the soluble fractions (homogenized supernatants), were employed to prepare emulsions.
Confocal laser scanning microscopy images and droplet size
3.3.1
Fig. 4 shows the confocal laser scanning micrographs and related droplet size distribution of the freshly prepared emulsions prepared with the different PP ingredients. Most of the emulsions had spherical oil droplets but the different protein ingredients had a significant influence on the oil droplet formation and sizes. The CLMS images of the PPI1, PPI2, PPI4, and PPI5 emulsions clearly showed the presence of smaller oil droplets. Contrary, the PPI6 emulsion showed a very polydistributed system with much larger, flocculated droplets. In addition, large protein aggregates randomly attached to the surface of some oil droplet, as seen in the image of the PPI6 emulsion. Protein aggregates were observed in the PPC and PPI3 stabilized emulsions as well, together with large, coalesced oil droplets. A wide range of oil droplets size has been reported for emulsions prepared with commercial PP isolates and concentrates. For example, droplets of size 0.5–5 μm were observed by Sridharan et al. (2020), while larger droplets were reported in the range of 10–100 μm (Grasberger et al., 2022) and between 5 and 50 and 20–200 μm (Burger et al., 2022). However, this is highly dependent on the protein ingredient and emulsification method (pressure and stages of HPH). All emulsions in our study exhibited a bi-modal distribution with a fraction of small oil droplets and a larger fraction of bigger oil droplets. The first peak size did not change significantly with storage time or SDS addition in any of the samples, reflecting the presence of a fraction of small, stable oil droplets (Fig. S2). In the study by Sridharan et al. (2020), emulsions were prepared at pH 3, which resulted in self-assembled PP nanoparticles of size 0.05–0.70 μm. However, in our study, PP ingredients were used at native pH, which is considerably higher than the pI. To further investigate this, PSD profiles of homogenized PPI suspensions were measured using the RI of sunflower oil. As seen in Fig. S3, comparing the PSD profiles of homogenized PPI suspensions with the corresponding emulsions (d0) showed that few samples exhibited overlap of the submicrometer peaks. Moreover, the oil concentration in the emulsion was 10%, which is more than an order of magnitude higher than the protein concentration (approximately 0.9% PP ingredient in the emulsions). This implies that the observed size distributions are largely dominated by the volume density of oil droplets. Thus, we suggest assigning the first peak in the emulsion PSD profiles to small oil droplets. The volume average diameter (d_4,3_) of the emulsions is summarized in Table 4. PPI1, PPI2, and PPI5 showed the smallest d_4,3_ values (1.1 ± 0.0 – 1.3 ± 0.1 μm), followed by PPI4, PPC, PPI3, and the worst case PPI6 consisting of very large oil droplets (42.3 ± 1.4 μm) due to flocculated droplets (showed by the high FI of 420%). Interestingly, the droplet size order PPI1 = PPI2 = PPI5 < PPI4 < PPC < PPI3 < < PPI6 followed exactly the decreasing order of the protein concentration in the homogenized supernatant (Fig. 2), resulting in a high correlation with the R coefficient of −0.93 (p < 0.001, Fig. S1). Furthermore, the d_4,3_ value (d0) had a stronger correlation with the particle size of the homogenized supernatant (R = 0.95 for d_43_,d0 and d_43_-HSN, p < 0.001, Fig. S1), compared to that of the full homogenized suspension (R = 0.71 for d_43_,d0 and d_43_-HSS, p < 0.001, Fig. S1). The above results suggested a significant association between the size and concentration of soluble proteins after HPH and the d_4,3_ of fresh emulsions. It can be hypothesized that the soluble fractions are primarily responsible for stabilizing the oil droplets in fresh PP emulsions. As expected, the sizes of the immediately formed oil droplets (day 0) showed a positive correlation with the final interfacial tension (R = 0.73 for d_43_,d0 and γ_10800_, p < 0.001, Fig. S1).Fig. 4CLSM image and corresponding oil droplet size distribution of fresh emulsions prepared with different commercial PP ingredients. The protein was stained by Nile blue (shown as red) and the oil by Nile red (shown as green). Scale bar is 25 μm. The droplet size, d_4,3_ is shown as mean ± SD. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 4. Table 4Average droplet diameter (d_4,3_), flocculation index (FI) and coalescence index (CI) of fresh (day 0) and stored (day 7 and 14) emulsions prepared with different commercial PP ingredients. Values are given as the mean ± SD. Different letters within the same column indicate significant differences (p < 0.05).Table 4. Sampled_4,3_ (μm) with waterFI (%)CI (%)day 0day 7day 14day 0day 7day 14Day 14PPC5.8 ± 0.2^c^9.2 ± 0.4^c^10.3 ± 0.1^c^148.6 ± 3.0^c^234.5 ± 68.3^b^167.7 ± 1.2^c^65.2 ± 4.6^b^PPI11.1 ± 0.0^e^1.5 ± 0.1^f^2.2 ± 0.1^e^66.4 ± 10.5^d^114.7 ± 9.5^de^174.7 ± 24.2^c^23.3 ± 0.3^c^PPI21.3 ± 0.1^e^2.2 ± 0.2^e^5.5 ± 1.2^d^73.0 ± 5.0^d^162.6 ± 2.5^cd^475.6 ± 88.6^a^23.2 ± 5.4^c^PPI313.3 ± 0.2^b^19.2 ± 0.6^b^19.4 ± 1.1^b^193.4 ± 39.3^b^91.1 ± 0.7^e^72.6 ± 6.3^d^126.1 ± 5.2^a^PPI42.7 ± 0.1^d^2.9 ± 0.0^d^3.5 ± 0.0^de^208.1 ± 0.0^b^209.3 ± 13.1^bc^203.2 ± 10.8^c^32.8 ± 1.2^c^PPI51.1 ± 0.0^e^1.5 ± 0.0^f^3.7 ± 0.1^de^54.9 ± 2.8^d^68.0 ± 5.2^e^313.9 ± 24.1^b^28.2 ± 2.6^c^PPI642.3 ± 1.4^a^49.6 ± 0.4^a^64.4 ± 4.1^a^420.8 ± 40.8^a^357.5 ± 22.7^a^376.4 ± 9.2^b^66.2 ± 11.3^b^
All emulsions underwent an increase in droplet size during 14 days of storage (Table 4, Fig. S2) Specifically, the oil droplet size in the PPI6, PPI3, and PPC emulsions was always larger than that in the other three emulsions. Thus, the mean droplet size in the PPI6 emulsion increased from 42.3 ± 1.4 to 64.4 ± 4.1 μm, whereas the lowest d_4,3_ value of PPI1 emulsion only increased from 1.1 ± 0.0 to 2.2 ± 0.1 μm during the storage period. The instability of emulsions governed by flocculation and coalescence is quantified by FI and CI, Table 4. Flocculation is the combination of two (or more) droplets that can be induced via interdroplet protein-protein interactions between adsorbed proteins at the oil-water interface (Dickinson, 2010). Clearly, the emulsion showed flocculation immediately after preparation, as shown by the high FI values at day 0. The PPI1, PPI2, and PPI5 emulsions had the lowest level of flocculation consistent with their smaller droplets (R = 0.92 for FI,d_0_ and d_4,3_,d0, p < 0.001, Fig. S1). This might be due to their high level of soluble proteins (protein concentration in homogenized supernatants, Fig. 2) that saturate the droplet surfaces resulting in the reduction of bridging flocculation. The high FI values (d0) obtained for the PPI6 and PPI3 emulsions were aligned with their low protein solubility since insufficient proteins at the interface may induce adsorbed protein molecules or aggregates being shared between pairs of neighboring droplets. However, the PPI4 emulsion had a moderately high FI at day 0 despite containing a high amount of soluble proteins, which may be attributed to its large particle size (Table 2). These observations suggest that flocculation is not solely associated with protein solubility but influenced by other protein properties such as protein aggregate size.
The differences in FI and CI values during storage (Table 4) suggested that the emulsions had different instability mechanisms. Droplet bridging increased over time in the PPI1, PPI2, and PPI5 emulsions, but to very different extents, by 163%, 551%, and 471%, respectively, while PPI4 maintained a mediumly high FI value during storage. On the other hand, these four emulsions had significantly lower CI values than other emulsions, indicating that they were more stable to resist coalescence during storage. Coalescence is the collision of two (or more) oil droplets due to the rupture of the interfacial film, resulting in the fusion of the droplets into a bigger droplet. The rate and extent of coalescence are, therefore, also affected by characteristics of protein molecules/particles constituting the droplet interface and not only the gravitational, colloidal, hydrodynamic, and mechanical forces acting on the droplets (McClements, 2015). Apparently, PPI2 and PPI5 were resistant towards coalescence but sensitive towards flocculation, hence droplet flocculation was the main underlying mechanism of their storage instability. The strong viscoelastic interface of the PPI2 droplets may provide higher resistance to interfacial density changes, thus protecting droplets against coalescence. Despite weaker lateral interactions of the PPI5 interface, other factors apparently compensated for the weakened film to prevent coalescence. In comparison, the PPI1 and PPI4 emulsions were more stable against both flocculation and coalescence after 14 days of storage, with PPI1 exhibiting smaller oil droplets. In a word, though PPI1, PPI2, PPI4, and PPI5 all formed fresh emulsions with small oil droplets, they showed different extents of resistance to flocculation after 14 days of storage, indicating that small droplet size alone does not ensure emulsion stability against flocculation during storage. The PPI3 emulsion experienced a notable reduction in FI but the highest CI value of 126% after the two-week storage time, indicating that storage instability of the PPI3 emulsion was mainly dominated by coalescence. This observation could be partially due to the presence of phospholipids in the PPI3 system, displacing protein from the interface over time. The PPC emulsion had a low FI at day 14, but a high CI indicating a rather unstable emulsion likely due to coalescence of the encapsulated oil droplets, probably affected by the high content of carbohydrates (Table 1).
In summary, the average droplet size (d_4,3_) and FI of fresh emulsions strongly correlated with soluble protein content and size in the homogenized supernatant, whereas the droplet instability over storage time as assessed by FI and CI did not show strong correlations with them. Hence, it can be inferred that the emulsifying capacity is primarily governed by the soluble fractions in the emulsion system; however, small oil droplet size could not ensure emulsion storage stability against flocculation and coalescence. The interface composition of oil droplets during storage is likely interfered by multiple factors, such as insoluble proteins, phospholipids, or carbohydrates, inducing different destabilization mechanisms. No doubt that interfacial structure and properties are crucial for the emulsifying properties of bulk emulsion, but care must be taken upon direct and simple comparison of the results from static drop tensiometer measurement with the results from real dynamic HPH emulsification. In the following sections a detailed description is presented to obtain further understanding of the PP ingredients and bulk emulsion properties.
Accelerated stability
3.3.2
The accelerated, centrifugal stability measurement of emulsions may provide a rapid prediction of emulsion stability. The transmittance-time profiles (Fig. 5) showed striking differences among the different emulsions. Though, upon the beginning of the centrifugation, it is common for all emulsions that limited amount of transmitted light is recorded along the sample cell due to the intact emulsion. Upon increasing centrifugation time, the force promotes more light transmission caused by increasing creaming phase separation. The PPI1, PPI2, PPI4, and PPI5 emulsions exhibited less light transmitted in the final curve, which demonstrated that the boundary between the cream layer and the continuous phase was poorly detectible due to less creaming, thus stable emulsions corresponding to their low instability index (Fig. 5). These results are in full agreement with the smallest droplets in the fresh emulsion (Table 4). Moreover, the lowest instability index together with the low FI and CI suggested that PPI1 formed the most stable emulsion. The greatest changes in light transmission were observed for the PPC and PPI3 emulsions, because their oil phases moved to the top rapidly, forming a more transparent aqueous phase at the bottom with a clear boundary between the two layers. This transmission pattern resulted in high instability index. Interestingly, the PPI3 and PPI6 emulsions exhibited jagged light transmission at the end of centrifugation, which is ascribed to their less uniform oil droplet sizes (Fig. 4). When comparing the PPI3 emulsion with the PPI6 emulsion, it is hard to discriminate which emulsion is more stable based on the (ambient) instability measurements resulting in the FI and CI (Table 4). However, the results from the accelerated test indicated that the emulsion prepared with PPI6 was more stable than that prepared with PPI3 (with an instability index of 9.3 and 12.0, respectively). Apparently, a higher accelerated stability was found for the PPI6 emulsion compared to the PPC emulsion, despite that the PPI6 emulsion had a much larger droplet size and a higher FI. This observation may be attributed to the high viscosity of PPI6 emulsion (cf. next section) suppressing the movement of droplets in the accelerated test (Dapčević-Hadnađev et al., 2019; McClements, 2015). This finding supports that the accelerated instability is affected by the combination of droplet diameter, flocculation, and coalescence stability as suggested by Dapčević-Hadnađev et al. (2019), but the CI might contribute more to the accelerated stability in our case as suggested by its high correlation with the instability index (R = 0.82, p < 0.001, Fig. S1). Additionally, the emulsion stability has been reported to be associated with interfacial elasticity (Chang et al., 2023; Sha et al., 2021). However, there was no considerable correlation between the instability index and interfacial elastic modulus (R = 0.55, p > 0.001, Fig. S1).Fig. 5. Transmittance-time profiles (Lumisizer), the initial profile is red, and the final profile is green, and instability index, high instability index value corresponds to low stability, of fresh emulsions prepared with commercial PP ingredients. It is noted that the instability index is based on PCA rather than the output from the software, as the PCA gives the overall stability across the whole length of the cell, rather than a number of predefined depths. Different letters indicate significant differences (p < 0.05), and error bars represent SD. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 5
Viscosity
3.3.3
The viscosity of emulsions may also influence their stability, as high viscosity may prevent/slow down droplet mobility (Niu et al., 2023). The PP ingredients vary in composition, especially for the PPC compared to the PPIs, as well as the content of insoluble proteins/particles, which may affect the bulk properties (e.g., creaming, sedimentation) of the emulsions. The viscosities of most emulsions were quite low and nearly independent of the shear rate at low shear (0.03–10 s^−1^) (Fig. 6A). However, at this low shear rate, the viscosity of the PPI6 emulsion decreased markedly with an increasing shear rate, indicating shear thinning. This is a typical pseudo-plastic behavior of non-Newtonian materials, which is the result of the changes in spatial distribution of particles, and in particular, of the large coalesced and flocculated droplets (McClements, 2015). The high initial viscosity at a low shear rate for PPI6 emulsion is due to the highly flocculated droplets which entrap some of the continuous phase (likely protein aggregates) (Dapčević-Hadnađev et al., 2019; McClements, 2015). Generally, all emulsions produced by PP ingredients displayed increased viscosity with an increasing shear rate in the range of 100–300 s^−1^, which could be attributed to the fact that a higher shear rate leads to an increase in the efficiency and frequency of collisions between the droplets, thus promoting droplets' rearrangements.Fig. 6(A) Low-shear viscosity and (B) zeta potential of fresh emulsions prepared with different commercial PP ingredients. Different letters indicate significant differences (p < 0.05), and error bars represent SD.Fig. 6
Zeta potential
3.3.4
The charge distribution around the oil droplets is critical in imparting stability of the immediately formed oil droplets against flocculation and coalescence, thereby potentially affecting the long-term stability of emulsions (McClements, 2015). The zeta potential of the fresh emulsions was between −42 and −45 mV, except for the PPI6 emulsion, which had a significantly higher value of −38 mV (Fig. 6B). In comparison, much higher values of −23.9 and ∼ −13 mV have been reported for other commercial PPI stabilized emulsions (Sha et al., 2021; Zhang et al., 2022). On the contrary, a very low value of ∼ −60 mV was reported for emulsion stabilized by commercial PPC and PPI (Keivaninahr et al., 2021). However, their different emulsification conditions (HPH pressure, oil type and concentration, protein concentration, pH, etc.), are likely the reason for the discrepant results. Despite the slight differences in zeta potential, which were only approximately 4–7 mV between PPI6 and others, it is noted that PPI1, PPI2, and PPI5, with the most negative charge, still exhibited lower FI (day 0) partially due to their stronger charge repulsion between the formed droplets. The PPI6, having the lowest absolute value of zeta potential, seemed rather prone to flocculation, as confirmed by its markedly high FI value (day 0). This was supported by the result of correlation analysis (R = 0.71 for FI,d0 and ZP, p < 0.001, Fig. S1). Oil droplets stabilized by PPI4, on the other hand, possessed high negative surface charge, but seemingly flocculation happened emphasizing the multifaceted nature of the interfacial film and emulsion properties.
Front-face fluorescence spectroscopy
3.3.5
Front-face fluorescence spectroscopy combined with PARAFAC modeling was used to provide information of protein structure and/or particle's spatial arrangement after emulsification (Hinderink et al., 2021). Through PARAFAC, the excitation and emission loadings from the cream and the serum phases led to the assignment of different fluorescent signals from the protein (Fig. S4, Table C). Importantly, the two components, displaying an emission peak above 320 nm and an excitation peak around 290 nm, were assigned to the tryptophan residue (Trp) of the proteins (Ghisaidoobe & Chung, 2014). Based on Trp's sensitivity towards the polarity of the local environment (where an emission red-shift indicates that Trp is in a more hydrophilic environment), we suggest that the component 1 and 3 in the three systems could be assigned to Trp residues in their hydrophilic and hydrophobic environment, respectively (Fig. S4).
The scores ratio of the hydrophilic to the hydrophobic Trp were calculated (Fig. 7). In general, proteins adsorbed at the oil-water interface unfold and the hydrophobic sites are exposed during interfacial film formation. Therefore, it is expected that the Trp residues of the interfacial proteins would be located in the oil phase, hence in a locally hydrophobic environment. However, as seen in Fig. 7, the scores ratio of PP cream phase was between 1.7 (PPI1) and 4.1 (PPC), indicating that the Trp residues were in a more hydrophilic environment after adsorption. A similar seemingly counterintuitive finding was reported by Hinderink et al. (2021), who explained that the dominance of the Trp hydrophilic environment was due to the protrusion of the Trp-rich convicilin extension region into the water phase. It is, thus, plausible that the harsh ingredient processing of the PPIs resulted in aggregated “packaging defect” (thus not general hydrophobic sites buried in the native protein interior), where especially after HPH the legumin and convicilin Trp-rich regions may become protruded, contributing to the hydrophilic environment. The partitioning and distribution of hydrophobic Trp at the oil–water interface may be related to interfacial protein conformation and interactions, thereby affecting the elasticity of the interfacial film. The hydrophobicity of Trp residues could be attributed to the partitioning into the oil phase, burial within aggregates, or involving in protein-protein hydrophobic interaction. PPC emulsion was primarily stabilized with native proteins forming a highly elastic interfacial film. It could be inferred that Trp residues mainly oriented towards oil phase and participated in hydrophobic interactions at the interface, thereby strengthening the elasticity of the interfacial film (Section 3.2.2). In contrast, the PPI6 weak interface was formed by large, rigid aggregates, likely with hydrophobic Trp residues buried within the aggregated structure rather than participating in interfacial hydrophobic interactions. The relatively high contribution of hydrophilic Trp in PPI6's serum phase may be attributed to the presence of some unadsorbed, soluble particles as freely flowing aggregates with exposed Trp to water phase (Hinderink et al., 2021). PPI3 probably had different Trp partitions from PPC and PPI6 due to lacking hydrophobic peak in the serum phase. PPI2 and PPI4 showed strong, viscoelastic interfaces suggesting being composed of Trp-Trp intermolecular or inter-particle interactions contributing to the hydrophobic Trp detection. The PPI1 and PPI5 emulsions stabilized by soluble, flexible proteins had interfacial films with weakened lateral interactions, suggesting that Trp residues were mainly exposed towards the oil phase (creating a hydrophobic environment but without hydrophobic interactions).Fig. 7. The score ratio based on the PARAFAC models of cream and serum systems. The score ratio was calculated by dividing hydrophilic tryptophan score by hydrophobic tryptophan score. The scores of hydrophilic and hydrophobic tryptophan for the cream and serum systems are presented in Table S1 and Table S2, respectively. Note: The score ratio for PPI3 serum phase was not calculated since its score of hydrophobic tryptophan was too low (close to zero or negative). Different lowercase (a-e) and uppercase (A-F) letters indicate significant differences (p < 0.05), and error bars represent SD.Fig. 7
Small angle X-ray scattering
3.3.6
SAXS scattering curves of the fresh and stored emulsions in Fig. 8 show that the scattering contribution from the droplets is over the detected q-range, hence, the SAXS signal is primarily deriving from the scattering of the adsorbed proteins, as well as the protein present in the continuous phase. Changes in the scattering intensity as a function of the q vector indicate structural changes of the proteins in the emulsions (Larson-Smith et al., 2010). The SAXS data of the seven emulsions can broadly be divided into three groups, based on their initial structure and subsequent time evolution. The stable group of PPI1, PPI2, PPI4, and PPI5 emulsions started from a scattering pattern displaying a broad feature around q = 0.025 Å^−1^, indicating some nanolevel structural features around 20 nm and had similar scattering patterns with basically no or only slight changes during the 14 days. This was somehow consistent with their excellent emulsion stability (low CI and instability index). The PPI3 and PPI6 initially lacked nanolevel structural features but evolved to scattering curves similar to the starting point for the stable PPI group over the two-week storage. The structural change of PPI6 was completed within the first seven days, while PPI3 still changed between 7 and 14 days of storage. This observation indicated an unstable interfacial film prone for some structural changes during storage, supporting their high CI values. The changes in PPI6 agree with the formation of a weak interfacial film (high γ_10800_ and low elastic modulus) and as such a highly inhomogeneous emulsion (CLMS image, Fig. 4). PPI3, on the contrary, had an initial high elastic modulus and, thus, a strong interfacial film. However, the PPI3 interface might become rather unstable due to coalescence likely because of displacement of proteins by phospholipids resulting in changed SAXS curves (lipids are not detectable in this SAXS range). The PPC emulsion displayed a unique change different from other emulsions, starting with a nanostructural signature at a smaller length scale around 5 nm likely reflecting its unfolded structure. However, this feature disappeared over one week of storage with a subsequent intensity rise in low-q scattering, indicating some formation of structural growth at the interface where the replacement of adsorbed proteins might occur over time also supported by interfacial and emulsion properties.Fig. 8SAXS data of emulsions prepared with different commercial PP ingredients after storage (day 0, day 7, day 14).Fig. 8
Differences in stabilization mechanisms of oil droplets in commercial PP emulsions
3.4
To summarize the findings in this study, a schematic illustration of the differences in the interfacial composition and stabilization mechanisms of oil droplets in commercial PP's emulsions is shown in Fig. 9. Upon adsorption at the oil-water interface, PPs or PP aggregates may (re)arrange to stabilize the oil droplets against flocculation and coalescence, but the degree to which this occurs is dependent on the nature and physicochemical properties of the PP particles. As discussed, the size of oil droplet in fresh emulsion is important, but not sufficient for ensuring emulsion stability. In addition, the elasticity of the interfacial film has also been reported to be associated with emulsion stability (Chang et al., 2023; Sha et al., 2021). Therefore, it is of interest to investigate how protein structure and interactions at the oil–water interface, together with protein properties related to droplet size, influence emulsion stabilization. As shown in Fig. 9, the droplet size of fresh emulsion and the elasticity of interfacial film classified the seven emulsions into four different groups: PPC and PPI3; PPI6; PPI4 and PPI2; and PPI5 & PPI1.Fig. 9. Schematic illustration of the suggested classification based on interfacial composition and stabilization mechanisms of oil droplets in commercial PP emulsions. The Trp partitioning at the oil-water interface is illustrated based on fluorescence spectroscopy results (Section 3.3.5), oil droplet is illustrated by their relative size (Section 3.3.1), and the zeta potential on the interface (Section 3.3.4), explaining the differences in droplet flocculation and coalescence.Fig. 9
PPC: Emulsion was stabilized by native proteins forming a significantly strong, viscoelastic interfacial network, likely with Trp residues contributing to hydrophobic interactions, resulting in medium sized oil droplets. The presence of exposed hydrophobic surface regions, combined with moderate repulsive forces, led to significant flocculation and consequently an unstable emulsion.
PPI3: Despite a strong interfacial film, the insufficient PP at the interface together with a Trp-rich surface induced large droplets and immediately flocculation. Upon storage, the hydrophobic attraction concurrent with moderate repulsion rendered the droplets prone for extensive coalescence leading to emulsion instability.
PPI6: The large, inflexible PP aggregates with buried hydrophobic Trp resulted in insufficiently covered, weak interface producing large oil droplets and low net zeta-potential. Exposed Trp-rich sites on the droplet's surface may promote hydrophobic attraction and a simultaneous low repulsion in effect forming bridges between droplets contributing to the very high FI. Consequently, PPI6 produce an inadequate and unstable emulsion.
PPI4: The fresh emulsion consisted of small droplets with an elastic interface probably composed of Trp-Trp intermolecular or inter-particle interactions. However, these oil droplets formed a high flocculation at day 0 irrespective of a high negative zeta-potential, which may be attributed to PPI4's large particle size. Apparently, those flocculated droplets were stable during storage, potentially contributing to the formation of a weak droplet network. Such a structure may enhance physical stability and provide a more structured mouthfeel, which is desirable for creamy emulsions such as dressings and sauces.
PPI2: Due to the high negative surface charge, electrostatic repulsion may dominate over hydrophobic attraction retarding their flocculation upon emulsification (day 0). Furthermore, strong repulsion and an elastic interfacial film may have prevented coalescence during storage. It is noted that though resistant to coalescence, the emulsion exhibited increased flocculation during storage. This PPI2 emulsion may, thus, work for both short shelf-life beverages and light dressings.
PPI5: A weak PP interfacial film, with Trp residues oriented towards both the oil phase and the continuous phase, enveloped the small droplets. In addition, the high surface charge facilitated electrostatic repulsion, contributing to droplet stability. This emulsion is suitable for beverage applications, albeit with certain limitations such as moderate shelf life.
PPI1: This was the most stable emulsion, with small droplets resistant to both flocculation and coalescence after 14 days of storage. A compact film of small, flexible particles, with Trp residues exposed to the oil and continuous phase, and a strong charge repulsion are proposed to be the main driving forces for this stable emulsion. The strong stability makes it particularly suitable for beverage-type applications with long shelf life and low viscosity.
Overall, emulsion stability was influenced by multiple factors. Protein properties associated with droplet size, elasticity, surface charge and Trp partitioning at the oil-water interface collectively explained the diverse stabilization mechanisms, enabling their use in different food applications. The importance of developing multivariable predictive models between composition, interfacial properties, solubility, particle size, and emulsifying properties rather than pairwise correlations is, however, emphasized, and investigation and modeling including a larger set of commercial plant protein ingredients is underway.
Conclusions
4
Seven commercial pea protein ingredients (one concentrate PPC, six isolates PPI1-PPI6) showed large differences in particle size, driven by ingredient processing, which resulted in distinctly different emulsifying behaviors. The protein content and size in the soluble phases after HPH treatment were found to be significantly correlated to oil droplet size in the emulsion. Accordingly, fresh emulsions prepared by commercial PP ingredients were mainly stabilized by their soluble fractions after HPH. PPI1, PPI2, PPI4, and PPI5 facilitated effective reduction of the interfacial tension resulting in emulsions of small oil droplets (1.1–2.7 μm). Contrary, PPC was characterized by medium droplets (∼5.8 μm) and PPI3 and PPI6 had large droplets above 10 μm. However, small droplets were necessary but not sufficient for emulsion storage stability. Based on elasticity, charge, and Trp partitioning at the oil-water interface, the PP emulsions' diverse stabilization mechanisms are explained. PPI5 and PPI2 are suggested to be more suitable for short shelf-life beverage-type emulsions, whereas PPI4 exhibited properties more aligned with the requirements of dressing and sauce emulsions. In particular, PPI1 exhibited the smallest droplet size, the best accelerate stability, and the strongest long-term stability against flocculation and coalescence, making it the most effective emulsifier in this study, thereby suitable for long shelf-life protein beverages. Despite the insights provided in this study, it is acknowledged that the commercial PP ingredients were investigated at their native conditions of pH and ionic strength in the context of applied relevance, though uncontrolled conditions may limit fundamental conclusions on protein interfacial characteristics. Future research could, thus, include controlled exterior conditions and also focus on physical or enzymatic modifications of commercial PPs to improve interfacial flexibility and further optimize their emulsifying performance in food products.
CRediT authorship contribution statement
Yuqi Zhang: Writing – review & editing, Writing – original draft, Methodology, Investigation, Conceptualization. Åsmund Rinnan: Writing – review & editing, Supervision, Methodology, Conceptualization. Jacob J.K. Kirkensgaard: Writing – review & editing, Supervision, Methodology. Milena Corredig: Writing – review & editing, Methodology. Vibeke Orlien: Writing – review & editing, Supervision, Methodology, Conceptualization.
Funding
Yuqi Zhang was supported by a PhD fellowship from 10.13039/501100004543China Scholarship Council (CSC) (CSC No. 202006350037).
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.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Andersson C.A.Bro R.The N-way toolbox for MATLAB Chemometrics and Intelligent Laboratory Systems 52120001410.1016/S 0169-7439(00)00071-X · doi ↗
- 2Burger T.G.Singh I.Mayfield C.Baumert J.L.Zhang Y.Comparison of physicochemical and emulsifying properties of commercial pea protein powders Journal of the Science of Food and Agriculture 102620222506251410.1002/jsfa.1159234676547 · doi ↗ · pubmed ↗
- 3Burger T.G.Zhang Y.Recent progress in the utilization of pea protein as an emulsifier for food applications Trends in Food Science & Technology 862019253310.1016/j.tifs.2019.02.007 · doi ↗
- 4Chang L.Lan Y.Chen B.Rao J.Interfacial, and emulsifying properties nexus of green pea protein fractions: Impact of p H and salt Food Hydrocolloids 140202310865210.1016/j.foodhyd.2023.108652 · doi ↗
- 5Choe U.Chang L.Ohm J.-B.Chen B.Rao J.Structure modification, functionality and interfacial properties of kidney bean (Phaseolus vulgaris L.) protein concentrate as affected by post-extraction treatments Food Hydrocolloids 133202210800010.1016/j.foodhyd.2022.108000 · doi ↗
- 6Christiansen M.V.Skibsted L.H.AhrnéL.Control of viscosity by addition of calcium chloride and glucono-δ-lactone to heat treated skim milk concentrates produced by reverse osmosis filtration International Dairy Journal 114202110491610.1016/j.idairyj.2020.104916 · doi ↗
- 7Dapčević-Hadnađev T.Dizdar M.PojićM.KrstonošićV.Zychowski L.M.Hadnađev M.Emulsifying properties of hemp proteins: Effect of isolation technique Food Hydrocolloids 89201991292010.1016/j.foodhyd.2018.12.002 · doi ↗
- 8Dickinson E.Flocculation of protein-stabilized oil-in-water emulsions Colloids and Surfaces B: Biointerfaces 811201013014010.1016/j.colsurfb.2010.06.03320667698 · doi ↗ · pubmed ↗