Aqueous foams made from solutions of pea peptides and soybean by-products: mixture and process effects

Aqueous foams require stabilizers to be generated and to prevent their rapid destruction. Beside synthetic surfactants or polyelectrolytes, natural stabilizers can be used to produce and stabilize foams. Up to now, these are mostly of animal origins. Here, we present some new results obtained with two plant-based compounds. The first one is a powder of pea peptides. The second is a filtration permeate obtained during the industrial treatment of soybean. Together with studies of their composition, we have investigated the foamability of these solutions at different concentrations. We have used three different generation techniques to tune the foam parameters, and then studied the stability and aging of the obtained foams. These results are then compared to the ones of egg white and standard surfactant solutions. In parallel, we have performed complementary experiments at the scale of a single solution-air interface. Two different opposite behaviors are evidenced, in terms of foamability and stability, with a strong impact of the foaming process. Finally, we studied mixtures of these compounds, showing that the macroscopic features can be smoothly tuned by mixing these plant-based products, allowing us to better identify the balance between the role of the chemical composition, the concentration and the process of gas incorporation.

Aqueous foams are dispersions of gas bubbles within a liquid matrix. Such dispersions have unique properties, when compared to those of simple liquids and gas (Cantat et al., 2013; Exerowa and Platikanov, 2018; Drenckhan and Hutzler, 2005). Especially, the mechanical properties of foams makes them highly suitable for many applications (in cosmetics or in the food industry) in which the texture of the final product is crucial (Höhler and Cohen-Addad, 2005; Dollet and Raufaste, 2014). Moreover, the texture and the rheological properties can be widely tuned both by some physical parameters (bubble size and liquid fraction), and also by the chemical formulation used to produce the foam (Marze et al., 1937; Denkov et al., 2009). Indeed, foams cannot be generated from pure standard fluids: one needs to add stabilizers - surfactants, polymers, proteins, particles - to promote the initial formation of the foam, as well as to provide its stability (Tcholakova et al., 2008; Fameau and Salonen, 2014; Fameau et al., 2023). Still, despite the use of stabilizers, foams remain out-of-equilibrium materials: they age with time and eventually vanish as a consequence of different destabilization mechanisms, like drainage, coarsening and coalescence (Cantat et al., 2013; Exerowa and Platikanov, 2018; Koehler et al., 2000; Saint-Jalmes, 2006; Chieco and Durian, 2023).

In food applications, foams are mostly used because of their rheological properties - associating elasto-visco-plastic behaviors - which provide specific sensorial perception. As well, the use of aerated products can also have some advantages for health issues. Up to now, proteins (originating from milk or egg) have been widely used for stabilizing food-related foams (Boire et al., 2019), and often completed in the final formulations with various low molecular weight surfactants and stabilizers.

Today, the development of new types of stabilizer is required for multiple reasons, related to health, allergy and alimentary transitions issues. This trend is also linked to the strong demand of the consumers for products containing less additives and with simplified chemical formulations. One possible new route is to consider plant proteins (Hinderink et al., 2020; Sagis and Yang, 2022), especially those obtained from pulses (Anoop et al., 2023). Together with the goal of replacing animal-based proteins, using plant resources also allows to propose new functionalities for by-products, having low added value up to now. However, it is worth noting that - when using plant proteins - foaming and foam stability are generally not as efficient as when using animal ones. Thus, modifications through physical or enzymatic treatments are widely investigated to optimize the functionalities of these resources (Amagliani et al., 2021; García Arteaga et al., 2020; Shay et al., 2023; Yang et al., 2021; Yang et al., 2024; Perie et al., 2025).

The objectives of this study is then to characterize the ability of new plant sources to efficiently produce and stabilize aqueous foams. By efficiently, we consider here the ability to produce large volumes of homogeneous foam (not only a few bubbles sitting on top of the solution), and with these foam volumes not completely vanishing within at least 10 minutes. In the following, we report results obtained using samples originating from soybean (a solubilized filtration permeate) and yellow peas (solutions of peptides). While there are already many studies on solutions of pea proteins (Burger and Zhang, 2019; Kornet et al., 2022; Keuleyan et al., 2023; Moll et al., 2022; Qu et al., 2023; Yang et al., 2022; Shen et al., 2022), less is known regarding low molecular mass peptide solutions, and we are not aware of any studies on the functional properties of solutions of soybean filtration permeates.

Below, we first present results on the chemical composition of these resources. Then, we investigate the interfacial properties and foamability of their aqueous solutions. For the latter, it is known that the foaming process is also a crucial parameter. Consequently, we have used here three different methods of foaming: bubbling through a frit, imposing shear in a co-flow constriction, and entraining air by a open mixer (Drenckhan and Saint-Jalmes, 2015). As an advantage, this allows us to tune the initial liquid fractions and the bubble sizes of the foams, as discussed below in more details. The results of foamability and foam stability obtained for these solutions of plant resources are then compared to the ones of other standard animal protein and surfactant solutions. Lastly, we study mixtures of our two plant ingredients, eventually showing that we can widely tune the foamability and foam stability by the concentration ratio.

Products composition and properties

Two plant-based compounds are studied (provided by the OLGA company), and compared to both egg white and surfactant solutions. The first one is a Filtration Permeate (FP) of soybean. During the industrial treatment of soybeans, various co-products are obtained. The FP is the result of a process of ultra-filtration of the soluble soybean juice. It is initially recovered in a liquid state, then further processed to obtain a powder. We performed various characterisations (gel-filtration methods, ion-exchange chromatography and mass spectroscopy) allowing us to determine that it is composed of 70 % (small oligosides of 2 and 3 units), 20 % of minerals and 10 % of proteins and peptides. For the latter, 65 % have a molecular mass below 500 Da, 22 % between 500 and 5,000 Da, and 13 % bigger than 5 kDa.

The second resource tested here is a powder of pea peptides (PP). It is obtained from the hydrolysis of a pea protein isolate, and contains 90 % of peptides: 62 % have a molecular mass below 500 Da, 31 % between 500 and 1,000 Da and 7 % between 1,000 and 5,000 Da. The rest of the composition are minerals. Peas offer the advantages - once compared with proteins from other pulses - that it has a low allergenicity and quite a complete essential amino acid profile (Boukid et al., 2021). Note also that the powders are well solubilized in water, and only at the largest concentrations of PP, small micrometer-sizer aggregates are detected.

For comparisons, we also use a commercial powder of egg white proteins (ref. BP0202, purchased from IGRECA - Seiches-sur-le-Loir - FRANCE) at a concentration of 12.5 g/L (following the provider recommendations), and a standard surfactant (sodium dodecyl sulfate, SDS- from Sigma Aldrich, > 99 % purity) at a concentration 3 times its critical micellar concentration.

Liquid-air interfaces

Using a pendant drop tensiometer (Tracker - from Teclis-Scientific), the solution-air surface tension, γ, is monitored as a function of time. In practice, a solution drop (with a typical volume V of 10 μ L) is hanging at the tip of a needle, and the surface tension is derived from the shape analysis of this drop (at constant drop volume). Once equilibrium is reached, we performed sinusoidal oscillations of V, dilating and compressing the drop area, and measure if - and how much - the surface tension γ responds to these area changes. This finally allows us to measure the complex dilational modulus E * of the interfaces, which includes the elastic (E ′) and viscous (E ″) contributions. This compression/decompression protocol provides information on the interface-bulk exchange, and if the interfacial layer can, or not, reorganise itself during a compression-decompression cycle. In practice, the frequency is fixed at 0.1Hz and the typical amplitudes of the oscillations are between 5 and 10 %. In the following, we both report the complex moduli, and investigate the non-linearity of the responses by presenting Lissajous curves (surface pressure vs strain - see Section 3.1).

Foam production and stability monitoring

To test the foamability of the samples, we use three different foaming methods; schemes and pictures are given in the Supplementary Material Figure S1. Each is based on a different mechanism of gas incorporation into the continuous liquid phase (Drenckhan and Saint-Jalmes, 2015). As well, we use various optical and electrical diagnostics for characterizing the produced foams and their evolution in time (Vera et al., 2001; Feitosa et al., 2005; Höhler et al., 2014). Indeed, foams evolve in time through gravitational drainage, bubble coarsening (linked to differences in pressures) and film rupturing leading to coalescence. All of these mechanisms are coupled, with timescales depending on the liquid fraction and bubble size. It is then important to understand which of these mechanisms dominate during aging, especially to discriminate between cases where the thin films between the bubbles are strong enough to let drainage and coarsening occur without film rupturing, or are more unstable leading to an aging dominated by coalescence.

In a first device, the gas injection is made through porous glass frits immersed in the solutions. The gas injection pressure and the frit porosity are varied, in order to create foams of different liquid fractions Φ l and bubble diameters D. We used two glass porous frits: a first one with pores ranging between 160 and 200 microns (porosity 0), and the second frit with pores between 40 and 100 microns (porosity 2). Then, depending on the solution foamability and gas injection pressure, these frits respectively lead to bubble diameters D from 600 microns to a few mm (porosity #0), and from 200 microns to 800 mm (porosity 2). Note also that the time available for adsorption on a freshly created bubble, corresponding to the travel of this bubble within the solution - from one of the pores of the frit to the bottom of the foam - is between 1 and 2 s. The glass frits are located at the bottom of a Plexiglas column, with a square section (3 × 3 cm 2) and a height H = 40 cm. The gas injection is monitored by a pressure controller (Elveflow PG1113) set at 200 mbar and the injection lasts 30s; these values lead to the filling of the whole column, for a solution of high foamability. All along the column height, pairs of electrodes are disposed, to monitor the electrical conductivity of the foam and infer the liquid fraction as a function of height and time (Feitosa et al., 2005). Some examples of these liquid fractions as a function of time are given in the Supplementary Material Figure S2. We focus here on three quantities: the maximal stationary liquid fraction at a given height at the end of gas injection, and its value after 20 and 100 s of aging.

Secondly, we use the double-syringe technique (Supplementary Material Figure S1a), providing a shear-induced gas incorporation within a constriction (Gaillard et al., 2017). In this setup, one sets the initial volumes of solution V l and of gas V g to be dispersed. For all the experiments, the initial volumes are adjusted so that the solution always fills 30 % of the total volume of the setup. Therefore, in case of optimal gas incorporation, the expansion ratio r - defined as the volume of foam V F divided by the initial volume of liquid V l - must then be equal to r max = 100/30 = 3.34. In parallel, the bubble diameter D is also set by the double-syringe device (Gaillard et al., 2017): in case of high foamability, D is equal to 70 ± 20 microns. In terms of timescale, the duration of the mixing is between 15 and 20 s. Then, the generated foams are then injected in a rectangular cell, and the foam evolution is optically monitored by light transmission in the regime of multiple diffusion (Vera et al., 2001; Höhler et al., 2014). We then construct spatio-temporal diagrams summarizing the foam evolution and illustrating which destabilization mechanisms dominate (more information and examples are given in Supplementary Material Figure S3). In practice, to characterize the foaming and the aging, we focus here on the expansion ratio r and on the remaining percentage of foam after 4 min.

As a third device, we use a mechanical mixer (4.7L- KitchenAid ®) which generates the foam by a mechanism of air entrainment (Supplementary Material Figure S1c). Here the final liquid fraction and bubble diameter are set by the incorporation process, as we only impose the initial amount of solution, and the speed of the blade. The duration T of mixing is kept constant, T = 5 min. The mixing bowl is in glass, so that we can optically monitor the maximal amount of foam produced, and how it vanishes in time.

Interfacial studies

In Figure 1(top), the solution-air surface tension γ is plotted as a function of time, for different concentrations of PP. As expected, the dynamics and equilibrium values depend on the concentration. At the two lowest concentrations, only a small adsorption occurs over long timescales (hundreds of seconds). When increasing the concentration, the dynamics gets faster: the surface tension strongly decreases within the first seconds (hardly monitored with this technique), then the adsorption slows down on the timescale of a minute, while the surface tension finally reaches an equilibrium value at long times. In Supplementary Material Figure S3, we report these equilibrium values as a function of the concentration. It turns out that above a threshold concentration of about 15 g/L, the surface tension tends to saturate with the PP concentration. Figure 1(bottom) reports the results for the FP solutions. The same trends are observed on the dynamics and for the equilibrium values (Supplementary Material Figure S3). Here also, the surface tension at long times (1200s) vary less at the high concentrations. However, it still slightly decreases with the concentration, and do not show some saturation as for PP solutions.

Regarding the range of surface tensions at equilibrium, these minimum values of 40–45 mN/m mean that there are indeed amphiphilic compounds in the solutions, which both significantly adsorb at interfaces over long times, together with an efficient adsorption within the first seconds - in practice, these observations appear as good hints that such solutions could foam. After the short times dynamics, the long time evolution of γ reflects the re-organization of the different species at the interface (Fainerman et al., 1996; Leiske et al., 2016; Hinderink et al., 2022), together with a continuous adsorption of new molecules. For comparisons, the equilibrium values measured here are clearly higher than for standard surfactant or soap solutions (25–35 mN/m), but are fully consistent with the ones of milk or egg protein solutions (Tcholakova et al., 2008; Fameau et al., 2023).

In Figure 2, the complex interfacial modulus E * for the pure PP and FP solutions at a concentration of 15 g/L are reported. The results for the PP/FP mixtures are discussed below in Section 4.1. The modulus E * includes both the elastic and viscous moduli (E * = E ′ + i E ″). More precisely, we found that the elastic contribution, quantified by E ′, always dominates: for the PP solutions, one gets E ′ / E ″ = 14 ± 4, while the interfaces are a bit more viscoelastic for the FP solutions with E ′ / E ″ = 4 ± 1.

In Figure 3, we show the Lissajous curves which correspond to a more advanced representation of these experimental tests, as it allows us to directly visualize the balance between elastic and viscous contributions, as well as it reveals non-linearities in the response of the interfaces. We introduce here the surface pressure Π = γ 0 − γ, where γ 0 is the surface tension of pure water. Using Π is more straightforward, as it increases (decreases) under interfacial compression (decompression). For the pure systems, the responses are dominated by the elastic contribution: for both PP and FP interfaces, a thin and non-deformed ellipse is measured. The ellipse is wider for FP, reflecting the smaller E ′ / E ″ ratio and a more viscoelastic behavior than for PP interfaces. Concerning the effect of the bulk concentration on the results shown in Figures 2, 3, there are almost no variations for the FP solutions. For the PP-covered interfaces, more variations are observed: in fact, they are linked both to a large variability of the measurement even at a constant concentration and to the range of E * values which are surprisingly high. Indeed, such values reaching up to 100 mN/m are about twice bigger than what is often found for elastic interfacial protein layers - in that respect, Figure 2 also shows that E * reach about 60 mN/m for a standard elastic egg white interfaces. Such high moduli for PP indicate that the interfacial layer is so rigid that it can also easily break, having a low critical strain for rupture. Therefore, the measurement for PP interfaces is much tricky to perform and leads to large error bars at a given concentration, and to some variations between different concentrations. Comparatively, the value for FP-covered interfaces indicates a soft layer, less elastic than the ones covered by the proteins of egg white, but still much bigger in elasticity than what is found for surfactant-covered interfaces (for which no interfacial dilational elasticity are detected, as expected at this frequency for such small molecules).

Foaming and foam stability studies

We consecutively present the results obtained using the double-syringe setup, then the ones by bubbling, and then with the mixer: along this path, the bubble diameter D and the liquid fractions Φ l shift from values set by the experimental conditions to fully adjustable ones.

In Figure 4, the foamability with the double-syringe is illustrated by the expansion ratio r as a function of the concentration for both PP and FP solutions. Let’s recall that a low foamability means r = 1, and an optimal foamability means r max = 3.34. For both systems, it is found that the expansion ratio depends on the concentration, with a threshold value above which an efficient incorporation of air is obtained (r > 2.5). Then, at higher concentrations, r is gradually approaching r max for both chemical systems. More specifically, one can point out that the highest foamability is reached sooner for PP solutions (with a possible decrease at c = 60 g/L), while it continuously increases with concentration for FP, and always remaining smaller than for PP. The comparisons with the egg white and SDS solutions show that the ideal foamability is found for SDS and that the one of egg white solutions is similar to FP and PP at high concentrations.

Still, with this method, the foaming process appears to be sufficiently efficient so that our samples have all a foamability close to the optimal one. For this setup, the energy input and the shear deformation are strong, the duration of mixing last typically 20 s during which the foam is continuously sheared, so that the rate of bubble creation overcomes the one of bubble vanishing. As a consequence, the maximum of available gas gets incorporated into the foam, and as it limited - the expansion ratio can only saturate. This is thus not ideal to evidence differences between samples (illustrating - by the way - the importance to test the foaming with different devices). However, it brings the advantage that all the foams have the same bubble sizes and liquid fractions, allowing us to directly compare their subsequent stability and aging. Indeed, the study of the foam evolution reveals important differences between the samples. In Figure 5, we show the percentage of remaining foam after 4 min, extracted from the spatio-temporal plots, as those in Supplementary Figure S3. Here for a concentration of 15 g/L, the figure shows a clear difference between the samples. Direct observations show that strong coalescence and numerous film rupturing occur within the PP foams, while the FP foams first drain and eventually collapse. The same trend is then obtained for the other concentrations. Comparatively, foams made of egg white and surfactants solutions have a better stability, though results for pure FP foams are approaching the stability of these references.

For the bubbling method, results as those presented in Supplementary Material Figure S2 allow us to separate the effects of concentration, bubble size and composition (Figure 6). First, as expected with such a bubbling method, the smaller the bubble diameter D, the bigger the maximal liquid fraction Φ max (at a given vertical position and injection pressure): the stationary liquid fraction results from the balance between liquid rising with the bubbles and the drainage velocity, which scales with D 2 (Koehler et al., 2000; Saint-Jalmes, 2006). As well, coalescence also tends to reduce Φ max - therefore, higher values of Φ max mean a more efficient foaming. Thanks to the change of frit porosity, we have thus investigated opposite ranges of liquid fractions (i.e., wet foams with Φ max up to 0.25-0.3 and dry ones with Φ max < 0.05). Concerning the effect of concentration, at first glance, Φ max increases for both PP and FP, and reaches plateau values at high concentrations. Quantitatively, these plateau values of Φ max are fully consistent with the ones we also measured for egg white and SDS solutions, evidencing that FP and PP solutions foam (with this setup) almost as the reference ones at high concentrations. Still, differences are found at intermediate concentrations (10–30 g/L) where PP solutions appear to foam better that FP ones - this is especially visible for the small bubbles (porosity 2), with even a maximum of foamability for PP solutions and before a decay at the highest concentrations. For the large bubble experiments, the foams are in fact very unstable and the liquid fractions are low, so that it is more difficult to draw conclusions out of error bars.

The evolution with time of these foams are described in Figure 7, reporting the liquid fraction after 20s and 100s, normalized by the initial values measured at the end of bubbling. Here the concentration is fixed at 15 g/L, and the same behavior are observed at higher concentrations. Taking into account that the foams have not exactly the same initial liquid fractions and bubble diameters (oppositely to the double-syringe method), the FP foams turn out to be more stable; their decrease of liquid fraction, after 20 s and 100 s, are consistent with a foam mostly evolving under drainage, and without much coalescence (whatever the bubble size). Comparatively, the PP foams have lower ratio of liquid fractions, after 20 and 100s, which are the result of a coupling between drainage and coalescence.

In Figure 8, the results for the foaming by the mixer setup are plotted. Let’s recall that here the bubble sizes and liquid fractions are not imposed by the experimental conditions and are free to adjust. As a consequence, it is with this technique that the expansion ratio vary the most between pure PP and FP solutions. One can then point out the value of r ≈ 20 for the PP solutions, which is typically 4 times larger than for the egg white solution. In parallel, the bubble diameter D vary much less than r, and always falls between 1 and 3 mm.

Regarding the optically-monitored stability, Figure 9 shows the stability as defined by the percentage of remaining foam after 20 min. Again, despite its high foamability with this setup, the PP foams are extremely unstable and collapse rapidly by cascades of coalescence - this is even more drastic here as the foam initially produced is very dry (r ≈ 20). Thus, only about 20 % of the initial foam volume are still detected after 20 min. Oppositely, the pure FP foam (initially more wet - r ≈ 5) have stability properties close to the ones of egg white foams, with still more than 65 % of foams remaining after 20 min.

It is then clear that the same qualitative trends are obtained for the three foaming devices: in summary, the PP solutions foam well but foams are unstable, while the FP solutions foam less but foams are more stable. However, beyond such qualitative statements, we have found clear quantitative differences in terms of foamability and stability between the two chemical systems depending on the foaming methods, which - by imposing a physical mechanism of gas incorporation - impose (or let free to adjust) the initial liquid fractions and the bubble sizes.

Results on FP-PP mixtures

As pointed out before, the measurements on pure systems provide two rather opposite behaviors both for the foamability and foam stability, as well as for interfacial properties. In that respect, we have then investigated mixtures of PP and FP, to identify how the shift between these different behaviors occurs. In practice, we have focused on the two extreme foaming methods - the double-syringe and the mixer. Then, for all the mixtures, a constant total mass of 15 g/L is used (a threshold value above which the solutions of FP and PP depend less on concentration), and these mixtures are made of various ratio P of FP added to PP. Thus, by increasing P, we gradually replace PP by FP.

Interfacial studies for FP-PP mixtures

Looking back to Figure 2, one can see that increasing P provides a monotonous shift from one to the other limit: replacing PP by FP allows to smoothly tune the total interfacial elasticity E *. Note that replacing only 25 % of PP by FP already divides the modulus by a factor 2, bringing it to the more standard values found for elastic proteins layers (typically between 40 and 60 mN/m). Regarding the balance between elastic and viscous contributions, a more complex picture arises with the ratio P (Figure 3). In fact, for the intermediate values of P, wider ellipses than for the pure systems are found, meaning more viscoelastic interfacial layers. As well, the shape is no longer symmetric, with a deformed ellipse - especially for P = 50 %. Such deformation indicates non-linear behavior and differences between compression and expansion of the interface. At the interface, the mixture is thus not ideal: interactions are providing more complex viscoelastic properties, not present for the pure systems, and with competition between the species. Note that these deformed shapes can be further analysed by the GSD approach (de et al., 2023), to determine how the non linear contributions evolve with the mixing ratio - this analysis will be discussed in a future article.

Foaming and foam stability studies

Regarding the foamability with the double-syringe setup, the expansion ratio for the pure systems at 15 g/L are close, and already approaching the one of optimal foamability (Figure 4). By investigating the mixtures, we found that r smoothly vary between the values of the pure systems. In fact, we already point out that this foaming device is not discriminant between the pure systems, nor it is between their mixtures. More interestingly, the stability (monitored here on foams having the same liquid fractions and bubble sizes) depends on the formulation: back to Figure 5, we monitored a gradual and monotonous variation of the amount of remaining foam as P increases. Here, one needs to replace more than 75 % of PP to recover the stability of the pure FP.

Figure 8 shows - for the foams made with the mixer - how the expansion ratio r depends on the mixtures. A smooth decay of the expansion ratio is monitored (meaning lower volume of foam, and initially more wet ones): still, one can note here that r reaches values close to the one of pure FP with only 33 % of PP replaced. This is also the same situation for the stability (Figure 9), the stability in time is also tuned by the mixing ratio P, and 33 % of replacement of PP by FP is sufficient to provide a similar stability as pure FP.

The results show clearly that the foamability and the stability are smoothly tuned between the two opposite behaviors of the pure systems. It is important to point out that this shift is nevertheless dependant on the foaming technique used. This is well illustrated by the dependence of the stability with the mixing ratio, which has a non linear convex shape for the double-syringe (Figure 5), whereas it has a non linear concave shape for the mixer (Figure 9).

In the following, we summarize and rationalize some of the results presented before. For the PP and FP solutions, an initial concern was linked to their ability to generate foams, and about their potential as new foaming ingredients. Here, we have shown that there are indeed conditions for foaming - whatever the foaming device - and with a foamability comparable to the one of animal protein solutions. In that respect, we have determined the threshold concentrations, above which both systems reach a high foamability, as their foamabilities depends on concentration.

More precisely, between the samples tested here, the FP solutions always remain the ones with the lowest foamability, implying that large concentrations (a few tens of g/L) are required to tend to an optimal foamability. For PP solutions, the foamability is always high above a concentration of 15–20 g/L. Let’s make a few remarks here for the PP system: first, this threshold correlates well with the one for the surface tension to saturate (Figure 1, Supplementary Figure S4), evidencing analogies with standard surfactant solutions for which foamability gets optimal once the interfacial concentration reaches its highest value.

Also, the higher foamability of the PP solutions is most likely due to the presence of small molecular mass peptides, much smaller than standard proteins or polymer chains, and not present in the FP solutions. The smallest peptides are then efficient to rapidly and homogeneously cover the bubbles and to create some repulsion between bubble interfaces, allowing a foam to be dynamically build and sustained (even for the bubbling method with the shortest time of adsorption). Moreover, their intermediate size, smaller than the usual proteins and bigger than usual surfactants, is probably providing a specific organization at the interface, with molecules highly compacted and strongly adsorbed, thus inducing the highest interfacial elasticity. In fact, these small peptides clearly appear as intermediate chemical ingredients, ranging between large proteins/polymers and small surfactants, and having a specific interfacial behavior, deserving more investigations. Consequently, the decrease of foamability for the mixtures, as PP is replaced by FP, appears to be due to the decrease of the amount of these high-foaming smallest peptides.

Still on the foamability of PP solutions, we also found that the foamability decreases at the highest concentration (Figures 4, 6), evidencing a range of intermediate concentrations where the foaming is optimal (accordingly, the surface tension is also re-increasing for c = 60 g/L). We believe that this effect may be due to the limit of the solubilization, and to the impact of small aggregates occurring at the highest concentrations and destabilizing the foams.

Lastly, on foamability results, we have clearly shown how the foaming method plays an important role (Figures 4, 8). It is indeed directly linked to the fact that the method imposes - or not - the initial parameters of the foams, thus allowing differences between the samples to be attenuated, or amplified. Regarding the stability, we have monitored how the concentration and the foaming setup set the stability of FP and PP foams; as well, we have identified correlations between this stability and the interfacial elasticity. To go further on how interfacial elasticity impacts stability (Georgieva et al., 2009), one has to specify what stability means for each samples, by identifying which instability dominates and has to be related to the interfacial properties. Our results indicate that the PP foams are - at the opposite of the high foamability of the solutions - poorly stable because they are rapidly collapsing, with cascades of coalescence events: for these foams, this poor stability thus refers to a low resistance to thin film rupturing. For FP foams, the aging is consistent with the one of the reference foams, with low coalescence and only once the foam has dried by drainage.

During aging, it is known that rapid bubble re-organizations are triggered when a first film breaks: our interpretation is that the PP interfaces are too solid and brittle (consistent with the high interfacial elasticity) to smoothly adjust the skeleton of adjacent films, and to prevent the ruptures of these connected films and a subsequent cascades of ruptures. Our results thus tend to show that a too high value of interfacial dilational elasticity, meaning solid and brittle interfacial layers, is not efficient for the stability against film rupturing events. Consequently, replacing PP by FP in mixtures, both reduce this elasticity (Figure 2), making the interfacial layers more soft and viscoelastic: this then prevents film rupturing and cascades, finally increasing the stability (Figures 5, 9). In this framework, it then remains to better quantitatively understand how film rupturing and coalescence depend on the interfacial dilational elasticity, and if a possible critical value for this elasticity can be determined. As well, the elasticity considered here is the one at low frequency, and the effect of the frequency on this dependence is also to be studied.

On stability issues, it is important to also note the role of the foaming method: as pointed out before, the minimal mixing ratio to gain an optimal stability actually depends on the method used. Moreover, the concave and convex dependencies of Figures 5, 9 demonstrate again that the mechanism of gas incorporation is as much as crucial as the chemical formulation. On this point, it is worth noting that for the results of Figure 9, the foams are not initially all the same (especially in terms of initial liquid fraction - Figure 8); this has to be taken into account for the comparisons made in Figure 9 (especially, as it can explain the low stability against film rupturing of the pure PP foams by their initial lower liquid fractions).

Beyond these points revealing the importance of the foaming process and of the interfacial properties, one still has to confirm our different interpretations, as well as to tackle remaining questions (for instance, regarding the actual chemical composition and organisation at the interfaces, or the origins of the non-linear interfacial viscoelastic responses of the FP-PP mixtures).

More generally, our results confirms that foamability and time stability are not necessarily positively correlated. They also confirm that one cannot draw generic foamability or stability conclusions by focusing only on one foam generation process. In fact, we show that the foaming setup have all different advantages: the double syringe is useful to produce initially well controlled foams and to compare aging dynamics, while the mixer allows to see more differences in foamability. Regarding the link between interfacial rheology and foam stability, we believe that our study can be useful on such complex issues, by bringing a new set of data to this field.

As well, we show here that it is relevant to study by-products of plant resources, many of them having unexpected functionalities, alone or as additives to existing recipes, and this appears to be a relevant route to find alternative to animal-based ingredients for adjusting food texture.