Relationship between Formulation, Gelation Kinetics, Micro/Nanostructure and Rheological Properties of Sodium Caseinate Nanoemulsion-Based Acid Gels for Food Applications

Legislation and concerns about health effects of trans and saturated fatty acids have led to elimination or reduction of them in foods formulation. One of the alternatives for structuring food with healthy ingredients is using food-grade biopolymers such as proteins or polysaccharides to formulate hydrogels. The aim of the present work was to study the relationship among formulation, gelation kinetics, structure, and rheological properties of sodium caseinate (NaCas)/sunflower oil hydrogels prepared from nanoemulsions. NaCas was used as stabilizer in concentrations of 1, 2, 3, or 4 wt.%. Sucrose was also added in 2, 4, 6, or 8 wt.% to the 4-wt.% nanoemulsion. Gelation kinetics was studied by two methods: oscillatory rheometry and Turbiscan. Although gelation time values were significantly different between methods, tendencies were similar: values decreased with increasing protein and sucrose contents. However, the most influential factor on gelation time was the ratio glucono-delta-lactone (GDL)/NaCas. Structure was analyzed by confocal laser scanning microscopy and synchrotron X-ray microtomography. Low-protein content hydrogels (1 or 2 wt.%) had an inhomogeneous structure containing nano- and conventional-size droplets while the 4-wt.% hydrogel kept the initial structural characteristics: homogeneity in dispersed phase distribution and non-aggregated nanodroplets. Sucrose improved structure in terms of homogeneity. Analyses of X-ray microtomoghraphy data showed that while the porosity diminished, the wall width increased with increasing protein and sucrose contents. The hydrogel formulated with 4 wt.% NaCas and 8 wt.% sucrose showed a structure with nanodroplets evenly distributed and the highest G′∞ values of all hydrogels.


Introduction
Trans-fats have been used in margarine, snacks, confections, and a variety of foods since the 1950s.There were several reasons for the long-time use, such as high functionality for edible applications, low cost, β′ tending, high rates of crystallization, and high oxidation stability, among others.However, recent nutritional studies have claimed that trans-fat can be associated with increased risk of coronary heart disease.For this reason, reduction of trans-fat in foods has been a major concern in the last 15 years as well as the reduction of saturated fatty acid (SAFA) content, although debates about SAFA-related issues are still continuing (Sato 2018).Most fats considered healthy are liquid: vegetable and ω-3 oils.Therefore, an alternative to produce solid food with healthy lipids is using food-grade biopolymers such as proteins or polysaccharides to formulate gels.A hydrogel may be defined as a three-dimensional polymeric network with the capacity of imbibing large amounts of water (Elsayed 2019).These colloid-based systems should be prepared from food-grade ingredients, using economical and reliable processing operations (Matalanis et al. 2011).Among the sources to produce gels, starch (Meerts et al. 2017;Torres et al. 2018), vegetable proteins (Zhao et al. 2018), milk proteins (Dickinson 2006), alginate, and gelatin (López-Hortas et al. 2019) have shown a great potential for structuring food.
Sodium caseinate (NaCas) is a commercial milk product obtained by removing most of calcium phosphate from casein.Owning to its high nutritional value, health benefits, and desirable functional properties such as emulsification, water-and fat-binding, and thickening and gelation, sodium caseinate is now extensively used as an ingredient in food products.Casein micelles are composed of a complex of associated protein and calcium phosphate.The protein fraction of the casein micelles, which represents ~93% of its dry mass, is composed of four individual sub-micelles, called α s1 -, α s2 -, β-, and κ-casein, that assemble as colloidal particles having a diameter of ~200 nm (de Kruif et al. 2012).NaCas contains disaggregated sub-micelles of approximately 10-12 nm.Sunflower oil is rich in unsaturated fatty acids and is a good source for antioxidant and vitamin E (Kleingartner 2015).Therefore, a hydrogel prepared with these components will have good nutritional properties and taste, and will be a suitable replacement of trans-fats in food formulations.
In the last years, food nanotechnology is gaining more interest in areas such as food processing, packaging, nutrition, and nutraceuticals.Emulsions are considered nano when the dispersed phase contains droplets with radius sizes smaller than 100 nm (McClements 2012).Nanoemulsions have the advantages of showing good physical stability against gravitational separation and droplet aggregation, of having high optical clarity, and of enhancing bioavailability of encapsulated substances.Nanoemulsions were engineered to form viscoelastic gels even in systems with lower dispersed oil phase volume fraction compared to conventional emulsion gels (Erramreddy and Ghosh 2015;Erramreddy et al. 2019).Although droplet size is a main property that strongly affects emulsion behavior, the effects of starting from nanosystems stabilized with sodium caseinate on physical properties and structure of gels has not been described yet.Thus, it is interesting to relate initial system properties with changes in structure and rheology of the final gel and to compare nano-based hydrogels with conventional hydrogel behavior.
The knowledge of the structure properties and the gelation kinetics of hydrogel are of essential technological importance since those properties are related to rheological behavior of hydrogel and ultimately to its possible applications (Wei et al. 2016).The sol-gel transition in different systems have been investigated by disturbing and non-disturbing methods that report on different aspects of the process.For example, acidinduced gelation of enzymatically cross-linked caseinate was investigated by rheological measurements following the evolution of G′ and G″ moduli with time, to examine the impact of ionic strength on casein polymerization (Raak et al. 2019); the influence of temperature-time gelation conditions was studied to describe the effects of processing parameters on potato starch gel-forming kinetics and final gel mechanical features (Torres et al. 2018).The rheological methods reported in literature involve application of shear forces, which may modify gelation behavior since nanoemulsions are perturbed while the gel is forming.In a previous study, changes during the sol-gel transition were reported by Turbiscan (Montes de Oca-Ávalos et al. 2016).This method allows performing studies in quiescent conditions, that is, without disturbing the systems.Comparison of both methods will lead to a deeper understanding of the gelation process.
X-ray microtomography (also called X-ray imaging, IMX) is a non-destructive technique that allow investigating the three-dimensional internal structure of porous systems.In this manuscript, this technique was used to relate formulation with structure and therefore prepare the best milk protein hydrogel for applications such as desserts.It has been applied before in edible products with high amounts of voids, such as hydrogels with pore sizes in the micron range, focusing on applications in food processing.Several food systems have been investigated using this technique.A soy protein-based porous hydrogel with potential as a carrier for nutrients and flavors has been prepared by high-speed homogenization and its structural elements quantified (Guo et al. 2013).Potato starch sponges were synthetized and structurally characterized for porosity and pore network.By changing starch concentration in gels, it was possible to tailor the pore structure of these materials (Vasconcelos et al. 2012).Interior microstructure of chocolate, a multicomponent food, was described in terms of particle arrangement and structural imperfections with the aim to adjust the processing conditions appropriate for producing defectless microstructures (Reinke et al. 2016).The high energy of the synchrotron source used in this study allowed achieving a better resolution than the one obtained with the conventional X-ray techniques usually employed, being especially suitable for nanosystems.
The aim of the present work was to study the relationship among formulation, gelation kinetics, structure, and rheological properties of sodium caseinate/sunflower oil hydrogels prepared from nanoemulsions with the purpose of structuring food with healthy components.The effect of droplet size distribution of initial nanoemulsions (nanoscale of building blocks) on the properties of hydrogel was also investigated.

Nanoemulsion Preparation
Nanoemulsions were prepared using a combination of a highenergy homogenization and evaporative ripening methods previously reported for whey protein-stabilized systems, with minor changes (Lee and McClements 2010).Eight nanoemulsions were prepared, four of them with different protein concentrations, and other four with the same protein concentration but four different amount of S in the aqueous phase.NaCas was used as stabilizer in concentrations of 1, 2, 3, or 4 wt.%.The 4-wt.% NaCas-stabilized nanoemulsion may also contain S in concentrations of 2, 4, 6, or 8 wt.%.These protein and sucrose concentrations were selected because in these ranges the method was successful for preparing nanoemulsions.Higher protein or sucrose concentrations did not lead to nanosystems.Oil phase was a solution of 15 wt.%SFO in ethyl acetate.In all nanoemulsions prepared, the ratio NaCas/SFO was kept constant at 0.6.Sodium azide was used in a percentage of 0.01% to prevent microbial contamination.The preparation method has three steps.In the first step, oil and aqueous phases were mixed using an Ultra-Turrax (UT) T8 high-speed blender (S 8N-5G dispersing tool, IKA Labortechnik, Janke & Kunkel, GmbH & Co., Staufen, Germany), operated at 20,000 rpm for 1 min.Samples were kept in an ice bath (0 °C) during processing.After repeating this process three times, coarse emulsions with droplets in the microns were obtained.In the second step, the resultant coarse emulsions were further homogenized for 20 min using an ultrasonic liquid processing (US), VIBRA CELL, VC750 (power 750 W, frequency 20 kHz) model (Sonics & Materials, Inc., Newtown, CT, USA), with a 13-mmdiameter and 136-mm-length tip.Selected amplitude was 30%.The temperature of the sample cell was controlled by means of an ice bath (0 °C) with a temperature cut down control of 20 ± 1 °C during ultrasound treatment.After ultrasound treatment, fine emulsions were obtained, but droplets were still in the conventional emulsion range.In the third step, the prepared fine emulsions were placed in a rotary evaporator Buchi model R 100 (Buchi, Postfach, Switzerland) connected to a vacuum pump and a recirculating chiller to eliminate ethyl acetate.The degree of ethyl acetate evaporation was determined by carrying out a mass balance of emulsions before and after solvent evaporation (Lee and McClements 2010).The process was performed at 45 °C for 20 min.Then, the samples were cooled quiescently to room temperature (22.5 °C).The pHs of the SFO emulsions were 6.66 ± 0.05, close to 7. No buffer was added to the emulsions.Nanoemulsions were analyzed for particle size distribution, zeta potential (ζ), and polydispersity immediately after preparation.

Droplet Size Distribution, Polydispersity, and Zeta Potential
The average diameter of droplets (Z-average) from the distribution expressed in volume, the width of the droplet distribution, indicated by the polydispersity index (PDI), and the ζpotential of nanoemulsions were measured using a dynamic light scattering (DLS) device NanoBrook 90 Plus Particle Size Analyzer (Brookhaven Instruments Corporation, New York, USA) equipped with a laser and operated at room temperature.During the whole sets of experiments, the nanoemulsions were diluted until the final concentration of NaCas was 0.1 wt.%.The measurements were performed with a scattering angle of 90°.For ζ-potential experiments, KCl was added to reach a final concentration of 1 mM.Experiments were done in triplicate, and results were expressed as average and standard deviation.

Hydrogel Formation
For hydrogel formation, selected acidulant was GDL, which was used at a ratio GDL/NaCas = 0.2 or 0.4.The final pH value was 4.75 ± 0.20.The required amount of GDL was accurately weighted and added to 10 g of nanoemulsions.GDL was dissolved by stirring tubes with a vortex for 10 s and nanoemulsion was immediately placed in Turbiscan tubes or in the rheometer cell.Addition of GDL was considered zero time for the gelation kinetics.

Gelation Kinetics
Gelation kinetics was followed in two ways: in quiescent conditions and under shear.For quiescent conditions, a vertical scan analyzer Turbiscan MA 2000 (Formulaction, Toulouse, France) equipped with an infrared light source of 850 nm was used.Back scattering (BS) as a function of time was monitored in order to follow the nanoemulsion-gel transition.Nanoemulsions were evaluated at 22.5 °C, immediately after dissolving GDL and every 5 min for up to 3 h for a GDL/ NaCas ratio of 0.2 and 2 h for a GDL/NaCas ratio of 0.4.Gelation time (t gel ) was calculated from the curve average BS (BS av , the average value of BS in the 25-45-mm height of tube) vs. time as previously reported (Montes de Oca-Ávalos et al. 2016).Gelation under shear was followed by dynamic oscillatory rheology using a Paar PhysicaMCR 300 rheometer (Anton Paar Inc., Ashland, USA).After adding GDL, approximately 1.2 g of emulsion was immediately loaded into parallel plates (PP30/S) separated by 1 mm and changes in G′ and G″ moduli with time were monitored.Data were taken at 22.5 °C, every 2 min, and were recorded for up to 3 h for a GDL/NaCas ratio of 0.2 and 2 h for a GDL/NaCas ratio of 0.4.Studies were performed with a 0.01% strain at a constant frequency of 1 Hz.The time at which G′ and G″ vs. time curves crossed, that is G′ = G″, was defined as t gel .Asymptotic values of G′ (G′ ∞ ) were calculated from G′ vs. time curves using values of time higher than t gel , with the following empirical equation: where k 1 is a constant and t is the time.Values are the average of three determinations and are reported as mean and standard deviation.

Structure by Confocal Laser Scanning Microscopy (CLSM)
CLSM images were acquired with an Olympus FV300 (Olympus Ltd., London, UK) confocal laser scanning microscope equipped with an Ar gas laser (λ = 488 nm).The laser intensity was used at 10% of the maximum power.A 10× ocular was used, together with a 60× objective for a visual magnification of 600×.Acquisition was performed with the confocal assistant software Olympus FluoView version 4.1 provided with the FV300 CLSM.To stain the fat phase, 0.5 mg of Nile red was added to 5 mL of nanoemulsions.
With the aim of following microstructural changes during the sol-gel transition, images were taken immediately after GDL addition and for up to 3 h for a GDL/NaCas ratio of 0.2 and 2 h for a GDL/NaCas ratio of 0.4.

Synchrotron X-Ray Microtomography
Hydrogels were freezed-dried with a Freezone 2.5 lyophilizer (Labconco, Kansas City, MO, USA) in order to obtain aerogels.Water was removed to increase contrast and be suitable for X-ray microtomography analysis.For experimental time limitations, only gels obtained with a GDL/NaCas ratio of 0.2 were evaluated.Gels with low protein concentrations (1 or 2 wt.%NaCas) collapsed during the freeze-drying process due to the poor rigidity of the structure, and therefore, they were not analyzed.All measurements were done at the X-ray imaging (IMX) beamline of the Brazilian National Laboratory of Synchrotron Light (LNLS) that belongs to the Brazilian Centre for Research in Energy and Materials (CNPEM).Samples of 1-mm large × 0.5-mm wide were cut and mounted in the holder.After automatic alignment, microtomography was obtained by taking 1024 equally spaced images with a total rotation of 180°.3D reconstruction was performed with the software AVIZO (FEI Visualization Sciences Group, USA) using a grid of 2048 × 2048 × 2048 voxels of 0.8 μm resolution.For porosity calculation, a grid of 200 × 200 × 100 voxels was selected from the central zone of the aerogel 3D reconstruction.This region was segmented, and porosity was calculated as the percent of void space.

Statistical Analysis
Analysis of variance (ANOVA) followed by post-hoc Tukey's multiple comparison test were performed using GraphPad Prism version 6.00 for Windows (GraphPad Software, La Jolla, California, USA).An α level of 0.05 was used for significance.

Droplet Size Distribution
Figure 1 reports the droplet size distribution expressed in volume for all nanoemulsions selected in this study.The average droplet size (z-average), polydispersity index (PDI), main population maximum diameter (d max ), and the ratio main/ total population corresponding to these distributions are summarized in Table 1.Data in Fig. 1 show that all emulsions had an almost monomodal distribution with a main population comprising the majority of droplet sizes.However, as may be observed, a very small second population, corresponding to droplets with greater size than the ones of main population, was also present.The average diameter (Z-average) had values between 89 and 119 nm.The values decreased with increasing protein concentration and with sugar concentrations of 6 or 8 wt.% (Table 1).This behavior indicated the relevance of formulation.The nanoemulsion prepared with the higher protein content and the higher sucrose concentration (4-wt.%-NaCas-8-wt.%-S)showed the smaller z-average.In conventional systems, it was reported that an increase in emulsifier concentration led to better droplet coverage, and therefore, small droplet sizes were obtained (Dickinson 2006

Gelation Kinetics by Turbiscan Method
Figure 2 shows gelation kinetics of nanogel as follows by Turbiscan method for the 4-wt.%NaCas nanoemulsions, formulated without (a and c) or with 8 wt.% sucrose (b and d), gelled using two different GDL/NaCas ratio: 0.2 (part a and b) or 0.4 (part c and d).Kinetics was followed measuring changes in back scattering profiles (ΔBS) with time.The ΔBS profiles in parts a and c showed curve lines with significant variation in ΔBS values with tube height for a selected time, indicating that hydrogel was not homogeneous across the Turbiscan tube; that is the scattering units are unevenly distributed along tube length.Addition of sucrose (part b and d) improved nanoemulsion-based hydrogel structure in terms of homogeneity, and in both cases, ΔBs profiles showed straighter lines than in formulation without sugar.From these results, it might be expected that sucrose had a role in structure formation since the scattering objects were more evenly distributed along the tube for formulations containing sugar in the aqueous phase.Figure 3 was selected as an example of gelation time calculation.The starting sample is the 4-wt.%NaCas nanoemulsion without sucrose in the aqueous phase.At the time GDL was added (considered as zero time), ΔBS av started to rise very slowly.Then, as building blocks were making more connections and gelation proceed, ΔBS sharply increased showing a greater slope.At this point, gelation was notorious in bulk.As shown in Fig. 3, both behaviors had a linear tendency.Gelation time was defined as the intersection between the two lines.gels obtained from nanoemulsions using a GDL/NaCas ratio of 0.2 or 0.4.The 1-wt.% NaCas nanoemulsion-based gel was not included because the gel was very weak even after 3 h.For the protein and sucrose concentrations selected in this study, formulation had no significant effect on gelation time.The GDL/NaCas ratio, however, had a significant effect on gelation time (p < 0.001).The rate at which aggregates were formed, as detected by Turbiscan, were much faster for 0.4 than for 0.2 ratio, which is in agreement with the fact that pH diminished at a faster rate for a GDL/NaCas ratio of 0.4 than a ratio of 0.2.For example, for the nanoemulsion-based hydrogel prepared with 4 wt.%NaCas without sucrose, pH dropped from 6.8 to 5.3 in 120 min for a GDL/NaCas ratio of 0.2 while it changed from 6.8 to 4.9 in 50 min for a GDL/NaCas ratio of 0.4.

Gelation Kinetics by Oscillatory Rheology
Figure 4 reports the evolution of G′ and G″ moduli as a function of time.The nanoemulsion-based hydrogel prepared from the 4-wt.%-NaCas-8-wt.%-sucroseformulation using a GDL/ NaCas ratio of 0.4 was used as example.At the beginning of the run, the behavior of G′ and G″ with time was characteristic of a liquid system.G′ and G′′ values varied around zero.As the hydrogel formed, both moduli increased sharply.There is a time at which G′ is equal to G′′ and after this point, G′ is always significantly higher than G′′.There is a general agreement among authors considering gelation time as the time at which both moduli have the same value.Table 2 also reports gelation times as measured by oscillatory rheology.The increase in protein and sucrose content diminished gelation time, but the effect of formulation was minor compared to the effect of GDL addition (p < 0.05).The ratio GDL/NaCas was the most influential factor on gelation time.Time values obtained for a ratio GDL/NaCas of 0.2 were more than twice the ones obtained for a ratio of 0.4 for all formulations tested.Hydrogels composed of sodium caseinate and N,Ocarboxymethyl chitosan were also investigated by rheological methods, and gelation kinetics was also dependent on pH (Wei et al. 2016).Similar results were found when maize flour was used as texturizing agent in acid-unheated skim milk gels (Román et al. 2019).Interestingly, gelation times obtained by rheology were significantly higher than the ones obtained by Turbiscan method.However, tendencies were similar in both cases.Turbiscan method is a non-disturbing technique in which measurements are performed in quiescent conditions.For this reason, the method allowed sensing changes in structure in the early steps of the gelation process, when nanoemulsions were still liquid.Although at that point, texture was not the one of a hydrogel, protein particles just started to build connections to form aggregates that would make a network and retrodispersed the light.In contrast, rheological analysis are performed applying a shear force on the sample under study.Although experiments were performed in the linear viscoelastic range, in these diluted systems, the shear force was able to delay formation of aggregates and building block connections.For G′ modulus being higher than G′′ modulus, sample needed to have a more solid-like behavior, and thus, rheological method described the late state in the process when the hydrogel was formed.Therefore, rheological measurements were less sensitive to early changes in structure.
Values of G′ and G″ rose in a more advanced degree of gelation, when more and stronger bonds between protein aggregates have been formed.

Gelation Process Followed by CLSM
To investigate the effect of initial nanoemulsion NaCascontent on the gelation process, CLSM was used to follow changes in structure with time.Figure 5 shows confocal microscopy images of the gelation process starting from nanoemulsions stabilized by 2 wt.%NaCas (a, b, c, and d) or 4 wt.%NaCas (e, f, g, and h), formulated without sucrose.
Among rows, values with the same capital letter are not significantly different.Among columns, values with the same small letters are not significantly different Fig. 4 Gelation kinetics followed by the rheological method for nanogels obtained starting from 4 wt.%NaCas-stabilized nanoemulsion, with addition of 8 wt.% sucrose, using a ratio GDL/NaCas of 0.4.Black symbols: G′; Gray symbols: G″ addition, but before gelation time (b, f), differences between nanoemulsion stabilized with 2 or 4 wt.%NaCas were more evident than in the first image of the sequence.Structure in b showed droplet coalescence while structure in f was still homogenous and below microscope resolution.The images taken at gelation time (c, g) report the structure of a solid-like material.In the case of the 2-wt.%nanoemulsion-based hydrogel, many droplets with sizes of micrometers could be observed (c).The 4-wt.% nanoemulsion-based hydrogel only showed a structure formed by nanodroplets evenly distributed (g).The structure after 3 h of GDL addition (d and h) did not show significant changes with the one reported for gelation time (c and g, respectively).The low-protein-content hydrogel had an inhomogeneous structure containing nano and conventional size droplets.Increasing protein content to 4 wt.%led to a hydrogel that kept the initial structural characteristics: homogeneity in dispersed phase distribution and non-aggregated nanodroplets.

Nanoemulsion
+ GDL At gelaƟon Ɵme AŌer 3 h  and h).The images correspond to: (a,e) nanoemulsion before the addition of GDL, (b,f) nanoemulsion after the addition of GDL but before gelation, (c,g) nanoemulsion at gelation time, and (d,h) gel after 3 h at 22.5 °C Food Bioprocess Technol (2020) 13:288-299 The effect of GDL/NaCas ratio on gelation process is shown in Fig. 6.The starting system was the 4-wt.%NaCasstabilized nanoemulsion formulated with addition of 8 wt.% sucrose.Two GDL/NaCas ratios were used: 0.2 (a, b, c, and d) or 0.4 (e, f, g, and h).The first image of each sequence corresponded to the nanoemulsion structure without GDL addition (a, e).This sample had a homogeneous structure composed of nanodroplets.Once GDL was added, for the 0.2 ratio, droplets coalesced and were noticeable in the microscope (b).For the 0.4 ratio, however, there were no visible changes in structure for addition of GDL (f).At gelation time, depending on GDL/NaCas ratio, hydrogels obtained showed a completely different structure.The 0.2 hydrogel contained many droplets with micrometer size (c) while the 0.4 hydrogel only contained nanodroplets (g).The structures of hydrogel stored for 3 h depended on the GDL/NaCas ratio used.For 0.2, the obtained hydrogel was inhomogeneous and showed droplet coalescence (d), while for 0.4, it had a homogeneous structure composed of nanodroplets (h).As for the 0.4 GDL/ NaCas ratio, gelation time was shorter, the system had less time to coalesce and form larger aggregates.Thus, building blocks of final gel remained in the nanoscale.

Nanoemulsion-Based Hydrogel Long-Term Strength by Oscillatory Rheology
Table 3 summarizes the results obtained when the asymptotic values of shear storage modulus (G′ ∞ ) was calculated from curves G′ vs. time.As shown in Table 3, G′ ∞ increased with increasing contents of protein and sucrose.The hydrogel prepared from the 4-wt.%-NaCas-8-wt.%-Snanoemulsion had the highest G′ ∞ of all hydrogels.Although sucrose had a minor effect on gelation kinetics, it had a significant effect on G ′ ∞ , especially for the GDL/NaCas ratio of 0.2.Raak et al. (2019) studied acid-induced gelation of enzymatically crosslinked caseinate in different ionic milieus.Their results showed that self-assembly of casein particles in solution depended on the environment and that the degree of crosslinking and stiffness of gels correlated as may be inferred from G′ ∞ values.Tsevdou et al. (2019) reported that depending on composition of ice cream mixes, addition of phenolic compounds was able to impart weak gel-like properties to the product (G′ > G′′), demonstrating the relevance of system formulation.In agreement with these findings, the composition of aqueous phase was very much influential in long-term rheological behavior of these nano-based gels.The G′ ∞ values were also significantly affected by GDL/NaCas ratio.However, differences were smaller than the ones due to changes in formulation.The GDL/NaCas ratio was the most influential parameter on gelation kinetics, since the kinetics was determined by the rate of diminution of pH, but had a small, although significant, effect on G′ ∞ .In most cases, for the same formulation, a GDL/NaCas ratio of 0.4 led to stronger hydrogels than a ratio of 0.2.Comparing G′ ∞ values of nano-based hydrogels with the corresponding conventional gels (data not shown), it was found that they were always higher for the nano-based systems than for conventional gels.These results were expected since the building blocks in nanobased hydrogel structure should be smaller than the ones in conventional gels (gels from emulsions prepared without the third step of synthesis).

Nanoemulsion-Based Hydrogel Structure by IMX
With the aim of showing the effect of formulation on microstructure, Fig. 7 reports 3D reconstructions from images obtained by X-ray microtomography.Since the GDL/NaCas ratio had a minor effect on final hydrogel strength (G′ ∞ ), and as experimental time in the IMX line was limited, only one ratio was studied (0.2).The projections of reconstructions in a plane are shown in Fig. 8.The black regions corresponded to empty space.The hydrogel obtained from the 3-wt.%nanoemulsion without sucrose had the most open structure   4. The wall width increased with increasing protein and sucrose contents.It may be expected that the strength of the hydrogel increased with the same tendency.On the contrary, the porosity diminished with increasing protein and sucrose contents, indicating that a more dense structure was formed.In addition to the parameters corresponding to hydrogels prepared from nanoemulsions, Table 4 also reports wall width and porosity for the gels obtained from the conventional emulsions with the same formulations (without the third step of synthesis).For a selected formulation, there were significant differences between nano and conventional systems, indicating the relevance of droplet size distribution of the starting system on the physical properties of the final gel.In all cases, nanobased hydrogels had greater wall width than the corresponding conventional gel.Porosity, however, was smaller in nanosystems only for some formulations containing sucrose.G′ ∞ values were in agreement with wall width.Systems with thicker walls had greater G′ ∞ values, showing that protein/ sucrose connections in the gel network determined rheological behavior.The role of NaCas/sucrose interactions were previously reported in conventional emulsions (Belyakova et al. 2003;Dickinson 2006;Huck-Iriart et al. 2013;Huck-Iriart et al. 2016).In those systems, NaCas became a better surfactant by addition of sucrose to the aqueous phase because it diminished surface tension.Sucrose interacted with NaCas at the interface improving the solvent-protein interaction, and impeding protein/protein bonding, leading to smaller building blocks, which resulted in more homogeneous structures.

Conclusions
Nano-based hydrogels were successfully prepared from NaCas/sunflower oil nanoemulsions.The kinetics of gelation was followed by two methods: oscillatory rheology and Turbiscan.Turbiscan allowed analyzing early steps in the process, while rheology, evaluating late steps.Gelation kinetics was mainly dependent on GDL/NaCas ratio used.Asymptotic values of G′ (G′ ∞ ), however, were strongly influenced by formulation.The strength of the final hydrogel increased with increasing concentrations of protein and sucrose.The increase in wall width, as evaluated by IMX, showed that protein/ sucrose interactions were very strong since sucrose formed also part of the protein matrix.These strong interactions led to better rheological properties (greater G′ ∞ values) and more homogeneous structures (able to keep the initial nanoemulsion structural-characteristics) as analyzed by CLSM.The 4-wt.% NaCas and 8-wt.%sucrose had the most homogeneous structure and the best rheological properties of all systems studied.Usually, a dessert such as a cake may contain 15 wt.% sucrose.Therefore, a gel with this formulation would have high protein content, healthy lipids, and an amount of sucrose lower than most dessert recipes.

Fig. 8
Fig. 8 Projections of X-ray microtomography 3D reconstructions for gels prepared from nanoemulsions stabilized with (a) 3 wt.%NaCas without sucrose, (b) 4 wt.%NaCas without sucrose, (c) 4 wt.%NaCas with 2 wt.% sucrose, and (d) 4 wt.%NaCas with 8 wt.% sucrose ). Sucrose interacted with proteins and a concentration of at least 6 wt.% was necessary to see an effect on droplet size.Polydispersity values varied from 22 to 29, indicating that these systems could be considered monomodal.Values of d max were significantly smaller than z-average values (p < 0.001).As reported in Table1the second population represented less than 7.1% of total distribution.However, as diameter was calculated in volume, droplets of the small second population greatly affected z-average values, being these values twice the value of d max .In a previous study, analyzing nanoemulsion stability, we found that formulations with highprotein concentration were very stable, not showing changes in z-average with time (Montes de Oca-Ávalos et al. 2017).In addition, those systems showed a homogeneous droplet size distribution that remained unchanged with time when studied by confocal laser scanning microscopy.Therefore, it might be expected that starting from stable and homogeneous nanoemulsions, it would be possible to prepare hydrogels with good physical properties.
Table 2 reports gelation times for all

Table 3
Calculated asymptotic values of shear storage modulus (G' ∞ ) and correlation coefficients for all nanoemulsion-based gels

Table 4
Food Bioprocess Technol (2020) 13:288-299 of all, showing many dark zones and a thin protein network(Figs.7 and 8, part a).When the protein concentration was 4 wt.%, the structure showed more connections and smaller dark zones than in part a (Figs.7 and 8, part b).Addition of 2 wt.% sucrose to the 4-wt.%nanoemulsionled to a hydrogel with a similar structure than the 4-wt.%samplewithoutsucroseasobserved by the naked eye.However, a slight improvement in terms of thickness of protein network may be noticed (Figs.7 and 8, part c).The hydrogel prepared from the 4-wt.%-NaCas-8-wt.%-sucrosenanoemulsion had the most compact structure with the thicker network of all samples (Figs.7 and 8, part d).The results of quantifying images in Figs.7 and 8 are summarized in Table eAData are expressed as the mean and S.D. The number of segments considered for calculations are reported in parenthesis.Different small letter among a column indicates statistical differences (p ≤ 0.01).Different capital letter among a row indicates statistical differences (p ≤ 0.01)