Pea protein [ Pisum sativum ] as stabilizer for oil/water emulsions

A map of stability for various water/oil/pea protein compositions has been plotted from the numerous reported results. Two clear regions of stability were identified. High internal oil phase emulsions with 70 – 80%, v/v oil content stabilized by total pea protein concentration < 2.5%, w/v showed stability. Low oil content of 10 – 30%, v/v for a range of total pea protein concentrations > 0.5%, w/v have also been identified as stable. Intermediate oil content and pea protein concentrations > 4% w/v are unexplored regions and are likely to be areas of fruitful future research. The wide range of stability suggests that different stabilization mechanisms could be important for different compositions and careful consideration has to be taken to avoid oversimplification. Both stabilization with particles, i.e. Pickering emulsions, and protein unfolding have been suggested as mechanisms. The diverse way of describing stability makes it difficult to intercompare results in different studies. A summary of different oil types used have been presented and several properties such as dynamic viscosity, density, the dielectric constant and interfacial tension have been summarized for common vegetable oils. The type of vegetable oil and emulsion preparation techniques were seen to have rather little effect on emulsion stability. However, the different extraction methods and processing of the pea material had more effect, which could be attributed to changing composition of different proteins and to the states of aggregation and denaturing. Careful consideration has to be taken in the choice of extraction method and an increased understanding of what contributes to the stability is desirable for further progress in research and eventual product formulation.


Scope and background
Increasing interest in emulsions stabilized by vegetable proteins has arisen in recent years based on the industry shift to plant-based alternatives to the conventional meat products and the consumer interest in environmental and health issues.This gives rise to a rapidly expanding research field and increasing number of publications.Fig. 1 shows the growth in publications for pea protein stabilized emulsions published between 1992 and 2022.The research topic is relatively new and there remain many undiscovered areas to study such as development of a microstructural understanding of protein stabilized emulsions and the mechanisms of stabilization to increase shelf life.There is interest in oilin-water emulsions, water-in-oil emulsions as well as more complex systems such as 'double emulsions' where dispersed droplets consist of smaller emulsion droplets.There are several recent reviews of related fields that provide complementary information although there is only limited discussion of emulsions stabilized with pea protein.Zhang et al and Shi et al discuss protein stabilized emulsions [1,2], and there are also several reviews that describe particle stabilized emulsions particularly in the context of food [3][4][5][6].A general review of the use of pea protein in the food industry has been made by Shanthakumar et al [7].
This review gives an overview of research with pea protein as an emulsion stabilizer and what is known about the stability and stabilization mechanisms of such systems.There are some recent review articles covering other aspects of biopolymers and pea proteins as stabilizers [8][9][10] but this article summarizes particularly what has been established as regards stability with a short description of the measuring techniques and then providing information about composition and processes that give rise to 'stable' emulsions with various pea protein materials.Further papers found in literature searches covered topics such as emulsions prepared with pea protein in combination with other materials as co-emulsifiers like casein [11], xylan [12] or Tween 80 [13] and cannot be related directly to work with just pea protein.There are also articles devoted more towards the applications such as ready food products [14], 3D printing [15] or microencapsulation [16].Such topics will not be discussed in further detail.
Background information about the motivation and requirements for research in this area will be presented first.Limits of stability are tabulated and 'stability maps' drawn from the diverse results of previous studies are presented.In order to provide a uniform presentation, with composition in the same units for all studies, some physical properties of the oils have been tabulated and where necessary these values of density have been used to convert mass to volume.The various methods both to produce the emulsions and the criteria to assess their stability are described.To avoid confusion with the behaviour of other materials that have been used as co-emulsifiers, only studies with pea protein as the sole stabilizer have been included in this primary presentation of the previously published work.Even with this restriction, the scope for a range of different pea protein materials exists and is discussed further below.
Two main applications for emulsions stabilized by vegetable proteins such as pea protein are for emulsion food products and as encapsulation systems.Pea protein is a plant protein, commercially available and in contrast to soy protein, it is a non-allergenic product.It accounts for the largest fraction (36%) of the total pulse production in the world [17].It is consumed by people of different cultures and it is attractive for use and accepted globally, which can be confirmed when noting that the main pea production countries were Canada, Russia, China, USA and India [18].There have been several reports where pea proteins or pea protein fractions have been used successfully as emulsion stabilizers such as in meat analogues, cookies, milk and snack products [19].For food applications, there are several different criteria regarding emulsion stability.To substitute for usual commercial dairy and animal products, the properties such as viscosity, mouthfeel, taste and color may be important.The time requirement to keep emulsions stable vary from hours during food preparation and serving, to months or years of food storage and shelf life.Food applications include to mimic similar properties of animal products on the market, but also give rise to new applications such as low fat and high protein products.Another main use for emulsions stabilized with vegetable proteins is for encapsulation of bioactive compounds such as vitamin D and curcumin in nutraceutical formulations [9].
Peas as a plant-based protein source may have further health benefits such as antioxidant and anti-hypertensive activity, and modulate intestinal bacteria [19].As a transition process on a global scale to convert to a diet which contains more plant-based products, it is advised to offer ready meals and pre-made snacks high in plant proteins to the consumer to increase acceptance and knowledge of plant-based products [20].Well performing plant-based emulsifiers, which can replace conventional surfactants, can contribute to environmental benefits of increased shelf life of food products, less waste of degraded products and a shift to renewable and eco-friendly resources.
The stability of an emulsion is dependent on the desorption energy to detach the emulsifier from the oil-water interface.Stabilization mechanisms include steric stabilization, charge dissociation, depletion interactions and particle (Pickering) stabilization.In protein systems, steric stabilization occurs when the proteins adsorb to the droplet interface and form a surrounding diffuse layer that physically prevent droplet flocculation [21].Charge dissociation contributes to stabilization when proteins form a charged layer at the interface and the repulsive forces are enough to keep the droplets from flocculating.Pickering stabilization occurs with particles adsorbed at the interface which have a high detachment energy barrier hindering coalescence [22,23].The role of pea protein as an emulsifier is a combination of these different effects.
True thermodynamic stability would imply that the free energy change to form oil/water interface is favourable, at least when sufficient emulsifier is present.In practice many colloidal dispersions are only metastable and are found in local minimum in the free energy landscape rather than as an overall minimum.Long range interactions from charge dissociation or from polymers at interfaces create a barrier sufficient to avoid strong van der Waals attraction that occurs at low separations.If emulsion droplets grow with time, this suggests that there is not full thermodynamic stability and that the energy barrier to coalescence is low.Two further factors are important in considering emulsions with pea protein: it is often suggested that the protein is present as particles that provide stabilization according to the principles described by Ramsden and by Pickering [22,23].The large size of the emulsion droplets, many micrometres diameter, often causes creaming to occur relatively rapidly.These factors will be discussed in the light of the previous studies.
Several important ideas about Pickering stabilization of food emulsions have been described clearly by Murray [4].The criterion that free energy is reduced by undeformed particles adsorbing at an oil-water interface is clear but many stabilizers do not easily meet that definition: microgel particles and aggregates of proteins can be soft.The fluid contact angle for small stabilizer particles with an irregular shape is difficult to define.The low probability of desorption from an interface of Pickering stabilizers when compared to small amphiphilic molecules is important in providing relatively long-term stability.This slow desorption rate is a feature shared with large polymeric stabilizers that are attached to an interface in several places.Thus, sharp distinctions cannot always be drawn between methods of stabilization.Some aspects of steric stabilization may also be apparent when particles have irregular shapes or if they consist of proteins that partially unfold.The review by Murray does emphasise that for practical purposes one should distinguish between stabilization that arises from a favourable free energy change on formation of dispersed droplets and the slowing of coalescence and creaming by trapping droplets in gels or other highly viscous fluids.This exploitation of gelled phases should be distinguished from the high viscosity of high internal phase emulsions, sometimes abbreviated as HIPE, although in some respects the effects of increasing viscosity are similar.This review of the literature up to mid-2023 provides an overview of what is currently known about the stabilization mechanisms and what are the main influences on stability of emulsions made with pea protein.
In the following sections, the review will discuss the properties of the individual emulsion components, both the oils and the proteins, and then the preparation methods.Section 3 provides a summary and critical analysis of the various experimental techniques to assess physical stability.A map of stability as a novel way to present reported storage stability is provided with an extensive table of the results that have been reported in previous studies.The results that have been reported for stability as regards compositions are discussed in the context of the viscosity of the emulsions, surface potential and interfacial properties such as tension and rheology in Section 4.There is not discussion of details of the chemical degradation and chemical stability of the emulsions as this is not primarily specific to those prepared with pea proteins.Oxidative stability is described, for example, in the articles of Berton-Carabin et al and of Keramat et al [24,25].The final part, Section 5, provides conclusions and a general outlook.

Oil type
The wide use of vegetable oils in pea protein stabilized emulsions are in line with the aim to create plant-based food emulsions.The most common vegetable oils reported are canola oil, corn oil, rapeseed oil, soybean oil and sunflower oil.A few studies use less common vegetable oils such as flaxseed oil [26], hempseed oil [27] and peanut oil [28], while some others have used oils such as dodecane [29][30][31], fish oil [32], hexadecane [33], medium-chain triglycerides [34][35][36][37], silicone oil [38], trans-cinnamaldehyde [39], and the specific medium-chain triglyceride, tricaprylin [40].In some articles medium chain triglycerides are described as 'Miglyol', which is a trade name for a pharmaceutical grade oil excipient.These oils are usually predominantly formed with C 8 and C 10 alkyl chains.There are no specific studies to our knowledge where the effects of different types of oil have been investigated with pea protein in emulsions but there are some studies of interfacial properties (see Section 4.4).Fig. S1 shows the reported stability results for emulsions with pea protein and different oil types.The impact of different oils has been investigated more extensively for systems made with other proteins such as β-lactoglobulin, bovine serum albumin, lysozyme, and β-casein by Bergfreund et al [41][42][43].Although these comparisons were not of common vegetable oils but rather oils such as alkanes and chloroalkanes, Kalaydzhiev et al have studied the differences of sunflower and rapeseed emulsions with ethanol-treated rapeseed meal protein isolate [44].A general trend of the results by Bergfreund et al was that protein adsorption and denaturation decreased at interfaces with a larger difference in polarity of the oil and water phase.The more polar oil molecules interpenetrated the protein network and created weaker interfacial layers [41].The proteins experienced stronger interfacial elongational stresses with low polarity oils, which led to increased and faster unfolding [42,45].Although these studies do not make direct comparison of different oils with pea protein as the stabilizer, there are likely to be analogous effects.Several physical properties of oils are of importance when preparing emulsions.These include density and the polarizability for which the dielectric constant (relative permittivity) is a useful indicator.The dielectric constant quantity is also important in estimating Hamaker constants for calculation of van der Waals interactions.Surface tension and the oil-water interfacial tension are parameters that are also significant in selecting appropriate stabilizers.Most vegetable oils consist primarily of long-chain fatty acid triesters of glycerol (triglycerides).These oils show variation in composition according to their source that can reflect different growing conditions, varieties of individual species and processing although there are some characteristic differences.Values of physical properties for some commonly used oils are presented in Table 1.The density difference between the oil and the water is important when trying to produce stable emulsions as low-density oils may lead to increased emulsion creaming and consequent high rate of destabilization.For the listed vegetable oils, one can conclude that the peanut oil has a lower density compared to the others, whereas the flaxseed oil has a considerably higher density.The polarities of the common oils are within a similar range, the exception being dodecane on the lower end with coconut oil and octanol on the higher end.Measurements of interfacial tension for a few vegetable oils are available and particularly Gaonkar has shown that there are significant differences in oil/water interfacial tension between commercial samples and purified oils of as much as 8 to 10 mN m − 1 [46].The commercial oils show lower interfacial tension.It is also clear that ageing or time dependent effects occur notably in the commercial oils.The difference in interfacial tension with the degree of purification is more than the difference between the various vegetable oils seen in the data in Table 1.In line with the conclusions of Bergfreund et al [42], since the interfacial tensions and polarities are similar for the different vegetable oils, it is reasonable that there are no substantial differences between the emulsions made with the oils reported by Gaonkar.

Proteins and protein purification
The protein content and composition of peas have been described and reviewed by several authors.Here we provide a brief summary: there is for example, a good recent review by Grossmann [47] that relates the structures in food material to composition of pea protein.Peas consist of 13.7-38.3%protein [48] and 55-68% starch [49].The composition of pea protein differs depending on variety and growing conditions, but consists mostly of globulins (65-80%) and albumins (10-20%) [50].Example data [51,52] for the amino acid composition for pea protein are presented in Table 3.Other data that reflects the variability from growth conditions and isolation processes are available [53][54][55][56][57]. The different pea protein fractions can be classified according to the Osborne classification based on the protein solubility in water, diluted NaCl solutions, aqueous ethanol solutions and an insoluble fraction as albumins, globulins, prolamins and glutelins, respectively [58].Although the variability in different varieties are significant, the major component in pea protein is globulins followed by albumins.The globulin fraction consists mostly of 11S legumin, 7S vicilin and 7S convicilin according to the sedimentation coefficient classification.The legumin is a hexameric holoprotein with 300-400 kDa, with one bigger acidic α-subunit and one smaller basic β-subunit polypeptide linked by disulphide bridges.They eventually form the 9S trimeric prolegumin which is transformed to the 11S legumin.The vicilin is a glycosylated trimeric protein with three components and no disulphide bridges and molecular weight 146 kDa.Similarly, convicilin is trimeric protein with three components and no disulphide bridges.It has molecular weight 210-290 kDa.The albumin fraction is mainly composed of the cysteine rich fractions called PA1 (pea albumin 1) with subunits a and b and PA2 (pea albumin 2).Disulphide bridge formation is present for these fractions.
Many different ways in which the protein is extracted from the peas have been reported.This might influence the emulsion formation in several ways.Although there is reasonable similarity between reported amino acid composition between the results shown in Table 3, even with E. Olsmats and A.R. Rennie older work of e.g.Leterme et al, milling processes and particle size selection can result in different protein composition [54] and these can have different interfacial properties.To prepare pea protein isolate, alkaline extraction followed by isoelectric precipitation is the most common method, where dehulled, milled and defatted pea flour is dispersed in water, the pH is increased to alkaline pH ~8-10, the supernatant is removed after centrifugation and diluted, the pH is adjusted to the isoelectric point of the pea protein reported as pH 4 to pH 5, the sample is centrifuged again, and the protein is collected as the pellets.Other protein extraction methods include salt extraction, wet fractionation and ultrafiltration.There are also several studies of emulsions with untreated pea protein flour [28,37,[59][60][61][62][63][64].There are studies that have involved separation of various different proteins such as globulin, albumin, legumin or vicilin rich fractions [30,31,65,66].Kornet et al have identified that globulin rich fractions are more effective as emulsion stabilizers whereas albumin rich pea protein is better for stabilizing foams [65].Kimura et al investigated 7S and 11S globulins from various plants and found that 7S globulin from fava beans was better at emulsification of soy bean oil than other proteins [66].This behaviour is attributed to small differences in amino acid sequences and consequent changes in solubilisation.This study, as well as those of Baniel et al and Pedrosa et al, emphasise the importance of glycosylation in enhancement of stabilization [30,31].Other methods to modify the pea protein are extrusion, ultrasound treatment, changing the pH, cooling and heating.The change of pH is usually made by adding small amounts of acid or base to precipitate the protein at pH near the isoelectric point.
Chan et al have investigated the impact of screw extruder speed and final extrusion feed moisture content, and Sinaki et al studied the impact of die temperature during an extrusion treatment [60,67].The studies did not perform assessments of storage stability, but both emulsion capacity after centrifugation and emulsion stability after heating and centrifugation (Eq.6) decreased with extrusion treatment.Xia et al identified that emulsions with extrusion treated pea protein had a lower flocculation index and improved storage stability [68].They emphasized the importance of phosphorylation in combination with extrusion treatment to further enhance stability.These contradictory results suggest that the impact of extrusion treatment is not clear.
Generally, ultrasound treated samples had smaller droplet size than untreated samples and O'Sullivan et al reported a decrease from 5250 nm to 187 nm in droplet size but increased size distribution [69].Sha et al concluded that ultrasonication decreased droplet size but found that the droplet size of both the untreated and ultrasound treated samples increased upon storage for 14 days [70].Gao et al and Sha and Xiong investigated the water-soluble protein fraction of the legumin 11S and vicilin 7S globulin fractions, respectively, after ultrasound treatment [71,72].Gao et al reported adverse effects on storage stability, however Sha and Xiong did not perform any storage stability tests but concluded that the emulsifying activity and emulsifying capacity increased with ultrasound treatment.Zhang et al (2022) reported visually stable emulsions for samples treated with 500 W ultrasound power after 5 months of storage [73].Zhang et al (2022) and Zhang et al (2023) studied the impact of ultrasound treatment in combination with a change of pH to 12 [74,75].With ultrasound power up to 600 W, they concluded that the ultrasound and pH change had synergistic effects and improved the visual storage stability, decreased the droplet size and no change in microstructure was detected upon storage for 60 days.The benefit of concentrating on a specific protein in the pea is the possibility to identify the effects of that material and how it influences the stability, and thus draw clearer conclusions while disregarding impact from coemulsifiers or other components in the peas that might contribute to the stability.In contrast, an advantage of using plain pea flour is that it is not necessary to use expensive, time-consuming and waste producing  extraction methods and the materials can eventually be used readily on an industrial scale.In Fig. 2, only the study of Sridharan et al reported stability results for emulsions with pea flour and it reported the samples to be visually stable after 7 days with 0.5 to 0.75%, w/v pea protein in 12%, v/v oil, but there was phase separation for samples with 1.0 to 1.5%, w/v protein [63].A few further studies as mentioned above also describe studies with pea flour, but with no direct information about stable compositions.The limited amount of data for pea flour stabilized emulsions identifies an area for future studies that would include the need to map the stability.

Emulsion preparation methods
The physical process for emulsion preparation common to all studies is high shear homogenization.It is sometimes used in combination with other techniques such as microfluidization, high-pressure homogenization and ultrasonic homogenization.Fig. S2 shows the reported stability of pea protein emulsions made with different preparation techniques.While the greatest number of studies have used a high shear homogenizer as the sole process, the data points for additional techniques do not represent more stable emulsions but rather all the processes follow a similar trend.Although the literature describing emulsions with pea protein prepared using different processing methods is limited, some similar systems have been investigated.The review by Håkansson [76] and the articles by Vassaux et al [77] and O'Sullivan et al [78] as examples, give ideas as to what effects are to be expected.Håkansson compares high pressure and high shear homogenization processes and describes the droplet breakup as occurring in similar turbulent jet conditions that have been evaluated with computational fluid dynamics calculated flow patterns [76].Therefore, there is not an obvious need for high pressure homogenization in addition to high shear treatment, but as Håkansson emphasizes, the effects on high internal phase emulsions and properties other than droplet diameter are not well-studied.Vassaux et al studied the storage stability of montmorillonite particle stabilized emulsions prepared with high shear and ultrasonication processes [77].In addition to the observation of smaller droplets with increasing emulsification time for both processes, ultrasonic treatment was seen to give both considerably lower creaming and smaller droplet size.O'Sullivan et al studied ultrasound processing for emulsions made with milk protein isolate, as well as with Tween 80 [78].The droplet size decreased for longer processing times and also for higher ultrasonic amplitudes.They proposed an inverse power law relationship between the droplet size and the energy density that is dependent on the acoustic intensity, the surface area of the sonotrode, the processing time and the processing volume for emulsions with sufficient or an excess of emulsifier.Although droplet size is not the single factor determining the storage stability, reports of smaller droplet sizes, sub micrometre, using ultrasonic treatment make it promising for emulsion preparation.A complication for applications in the food industry is efficiency of pilot scale-up processing which was found to be lower than that on the laboratory scale, both for batch and continuous modes of use.

Overview
A challenge in considering stability of emulsions is the considerable variation in the measurements used to assess this.It is recognised that most samples change with time and that materials have to be regarded as metastable rather than truly in a thermodynamic minimum energy state.Measurements that are made may often reflect requirements for a particular application as there do not appear to be generally accepted criteria.The experimental methods relate primarily to assessment of droplet size, distribution of size, sample homogeneity and the changes with time, however, often empirical or semi-empirical parameters are reported.Unfortunately, it is not always easy to make inter-comparison between the different methods and parameters as often only some selected parameters are reported in each study.In the following paragraphs we mention briefly what measurements are reported frequently and the derived parameters that are used to describe the results.
In most cases it is recognised that information about droplet size and the distribution of size is key to providing information about practical aspects of stability.Usually coarsening (increase in average size) is observed with time.This leads to faster creaming of the less dense, oil phase.In this way both optical measurements of droplet size by microscopy and light scattering, as well as macroscopic visual observation of regions of different appearance may be used to make inferences about stability.Measurements of turbidity or transmission of light depends significantly on scattering and thus can also be used to assess droplet size although in practice it is often used empirically or with some secondary calibrant for turbidity.
In studies of phase behaviour, it is common to present composition diagrams that indicate stability of different phases.Although not truly stable thermodynamically, this representation is also helpful to summarize properties of emulsions.As the pea protein concentration is low, the corners of the quasi-three component phase map (Fig. 2) represent 100%, v/v water, 100%, v/v oil and 12.5%, w/w of pea protein in water.The data plotted from Table 2 represents a broad range of different aqueous phase with respect to pH and salt content that allow an overview.A further complexity in the assessment of stability arises from the difficulties in applying some measuring techniques to highly concentrated emulsions for reasons such as extensive multiple scattering of light or very high viscosity.In several studies the size of droplets has been assessed after dilution.Given the widely observed influence of oil/ water ratio on stability as seen from the data reported in Table 2 and Fig. 2, this may cause changes even if done shortly before measurement and after storage.It is also likely to be important if protocols such as those described by Pearce and Kinsella that describe dilution with sodium dodecylsulphate (SDS) solution are used [79].Although this paper describes use of about 3 mmol L − 1 SDS, not only is this surface active and will change interfacial tension, it is a denaturant for many proteins.The work of Gorinstein et al found SDS to be effective at denaturing globulins from other plant seeds within a few minutes at 30 • C at this concentration [80].Shen et al report that pea proteins denatured with SDS were improved as emulsifiers [81].
The stability index proposed by Pearce and Kinsella on the basis of a first order process that causes an increase in droplet radius with time is a measure of how long an emulsion may be stable and has units of time [79].A number of other parameters are reported in the literature that represent the size of emulsion drops at particular times or even the variation across a sample.These parameters may vary with time but often people report a dimensionless index for one particular time of observation.So that the differences between the properties that are probed by particular reported measurements may be understood, short descriptions of some more widely used parameters are provided in Table 4.

Droplet size distribution
As regards interpretation, measurements of droplet size are amongst the most straightforward.Light scattering experiments are widely used methods to study emulsion structure and droplet size.These can involve either diffraction techniques or estimates of hydrodynamic radii from dynamic light scattering [82,83].By comparing the droplet size distribution of fresh emulsions to that after storage, the technique is used to study storage stability.One frequent challenge is that samples are often highly turbid and give rise to multiple scattering.Protocols that involve dilution to avoid this difficulty need to be assessed as to whether that process may give rise to differences in measured size.Depending on the measurements that are made, droplet size is variously reported as volume-average diameter or surface-average diameter.Careful consideration has to be taken when analysing the average droplet size and E. Olsmats and A.R. Rennie

Table 2
Emulsions with pea protein that have been studied, sorted by oil type.The oil (%, v/v) and pea protein (%, w/v) concentrations are presented in common units, and the content of the continuous phase is described as buffer, salt or microbial addition.Here, pH adjustment indicates that small amounts of acid or base were added to obtain appropriate pH.The pea protein extraction methods and emulsion preparation methods are indicated.Storage temperature and experimental methods to determine emulsion stability is reported.Stability measurement techniques include flocculation index (FI) (Eq.1), coalescence index (CI) (Eq.2), emulsion stability index (ESI) (Eq.3), creaming index (Eq.4), emulsion stability (ES) (Eq.5), encapsulation efficiency (EE), emulsion capacity/activity after centrifugation (EC*/EA*) (Eq.6) and emulsion stability after centrifugation and heating (ES*) (Eq.6).Asterisk indicate centrifuged samples to mimic accelerated storage stability.additional look at the size distribution is useful.The polydispersity index, PDI, is a measure of the heterogeneity in size of a sample.While droplet size is expected to increase upon storage for physically unstable emulsions, the storage effect on the polydispersity index is more complex.The influence of pea protein concentration and treatment (ultrasound and pH shift) is reported to have a greater effect on the polydispersity index than storage effects [27,74].
Several studies quantify the storage stability by defining various different indices related to droplet size.For example, a 'flocculation index', FI, is described as a measure of the aggregated state of particles or droplets as determined by comparing flocculated droplets with original or unflocculated droplets dispersed in solutions containing 1%, w/v sodium dodecylsulphate (SDS).This has been defined as: where d is often either volume-average diameter or surface-average diameter [84].The flocculation index is reported as a percentage difference in size that might change according to the time allowed for flocculation.A related 'coalescence index', CI, is a measure of how droplet sizes in an emulsion change during storage and is determined by measuring pure emulsion samples before and after storage.This is calculated as where the droplet size is measured after various times of storage.The flocculation and coalescence indices are mostly concerned with changes of droplet size rather than the physical process that is causing the change.To distinguish between flocculation and coalescence destabilization usually requires further investigation.Light scattering to evaluate droplet size is exploited in a number of different ways, and a number of studies report an empirical measure of the homogeneity in samples that is known as the 'Turbiscan' stability index, TSI [85].This provides a scale by which either forward scattering or back-scattering of light varies from different regions of samples.The inhomogeneity is indicative in general terms of phase separation and/or creaming in samples.This type of assessment can be useful as a quality control or for comparing a range of similar samples but is difficult to relate directly to other specific physical quantities such as fraction of sample that is emulsion or droplet size.

UV-visible spectrophotometer absorbance
The turbidity of samples can be measured using a conventional ultraviolet/visible spectrophotometer.If measurements are made in a spectral region where there is low optical absorption, the light is attenuated primarily by scattering.The fraction of light that is attenuated in this way can be related to size of droplets in an emulsion.It is important that the geometry of the instrument used for the measurements is selected so that scattered light is not a significant contribution to the signal measured with the spectrophotometer.An emulsion stability index, ESI, measured by the turbidimetric method has been described as a way to quantify the rate of coalescence or aggregation and is defined as where the turbidity T = 2.303A l , A is the absorbance, l is the pathlength of the cuvette and t is the elapsed time between initial and final absorbance measurements [79].This measurement with a specific time interval for a change was based on the observations in the original work with measurements at various times that there was a simple first order process.Thus, the reported index with the ratio of an appropriate time interval to  a normalised change represented a measure of the time that the emulsion was stable.The emulsions are diluted with sodium dodecylsulphate (SDS) solution.Some variations of the method were adopted in the reported literature and the dilution factor ranged from 100 to 200 times, the absorbance wavelength between 500 and 600 nm, and the storage times were between 10 min to 24 h.However, few studies report on checks of the kinetics of growth of droplets and identify if the process can be described with this single time constant.This index is a measure of the overall ability of a protein to stabilize an emulsion by considering how resistant it is to changes such as creaming, coalescence, flocculation or sedimentation.

Microscopy and visual observation (macroscale)
Visual observation of emulsion microstructure was frequently reported in emulsion studies.For the purpose of evaluating emulsion stability, e.g.compare micrograph images after an amount of storage time, it was only used in nine of 82 studied articles (Table 4).While microscopy may give images of the emulsion microstructure that are readily interpreted qualitatively, quantitative conclusions should be drawn carefully as droplets can be distorted and so the sizes can be difficult to assess precisely.For the microscopy studies, optical microscopes were used except for one study where cryogenic scanning electron microscopy (cryoSEM) was used [69].
Macroscopic visual observation is an intuitive method to assess emulsion stability.The relative amounts of emulsion and separated phases can be measured and is often reported simply as either heights in a straight-walled container or as volume.Quantitatively, the creaming index is a measure of the cream layer separating from an emulsion during storage time.The difficulty with this type of index is the nonstandardization, where the most common way to quantify it is where H S conventionally is the volume of the droplet-depleted lower layer and H T is the total emulsion volume [86].This creaming index has also been described as creaming stability, CS [26], and emulsion stability, ES [87,88].In some other studies, the emulsion stability, ES, is described as where V A and V B are the volumes of the aqueous phases before homogenization and after storage respectively [59,89,90].

Encapsulation efficiency
An important use for emulsions is that they provide means to deliver active ingredients that may be poorly or totally insoluble in the continuous phase.Applications include nutraceuticals that can provide vitamins, antioxidants and even probiotic bacteria.Although this area has been more widely investigated using stabilizing particles from other plants [91][92][93], there is clearly emerging interest for pea proteins in this area.As with other systems, the ultimate goal for encapsulation with pea protein emulsions often relates to complex systems such as dried particles from emulsions as microcapsules or double emulsions [27,39].
Encapsulation efficiency is often used to describe the fraction or percentage of active ingredient from the source formulation that is incorporated in the emulsion or the end-product.This would usually be assessed by, for example, spectroscopic analysis to provide information about the relevant concentrations.In some cases, 'encapsulation efficiency' is used to describe the overall oil content for an oil-in-water emulsion and so as with terminology in this area, care is needed to verify the definitions used in each study.

Centrifugation stability
An alternative approach to evaluate storage stability is to use centrifugation as an accelerated substitute for long storage.The previously described methods and the defined indices for visual observation are used directly interchangeably between emulsions in the natural gravity field after storage and in accelerated gravity fields with centrifugation.While the physical effects during long time storage and after accelerated gravity are correlated, the processes that occur may not be identical and care should be taken when comparing the results from different stability experiments [79].Zhang et al have studied the centrifugation stability (stated to be at 10000 rpm for 10 min although without direct information about rotor size or effective acceleration) in comparison to the storage stability (90 days) at 4 • C [75].Visual observation and micrographs of the emulsions did not show any significant changes after storage compared to the fresh emulsions.After centrifuge treatment, the samples with lower pea protein concentration (1-2%, w/w) showed some instability with oil leakage at the top that was not present for higher concentrations due to the thicker interfacial film and a network structure formation.Consequently, the aqueous phase at the bottom after centrifugation was reduced with increasing protein concentration.What is also notable about the use of   centrifugation, is that by separating the water rich phase, it may be possible to produce ultra-high internal phase emulsions [75].Other studies that have used centrifugation stability to assess physical emulsion stability are indicated with an asterisk in Table 2.The emulsion capacity, EC, sometimes referred to as emulsion activity, EA, after centrifugation, is similar to the creaming index in Eq. 4. This index is commonly referred to as: where H E is the volume of the emulsified layer, and H T is the total volume of the sample.With additional heating and centrifugation steps, the same calculation applies to the reported emulsion stability, ES.An interesting approach to monitor emulsion stability during centrifugation similar to a time dependent stability study is by using a LUMiFuge/ LUMiSizer to analyse transmission variations [94][95][96].

Discussion of stability map
The reported studies of emulsions with pea protein were found by making a search in the Web of Science with the terms "pea protein" and "emulsion", as well as following up referenced relevant literature.These are listed in Table 2.There were 82 articles directly related to storage stability of emulsions with pea protein.Composition of studied emulsions such as oil type, oil concentration, pea protein concentration, aqueous phase content, as well as protein extraction and emulsion preparation methods, storage temperature, experimental techniques to evaluate physical stability, results and stabilization mechanisms as stated by the authors are reported.In order to provide a uniform description of the various studies, the oil content in the final emulsions is converted when necessary based on the oil density values in Table 1 to give oil content as percentage by volume.The pea protein is reported as percent, w/v of total emulsion.
The composition of the emulsions that have been studied varies from 10 to 95% v/v oil and 0.03 to 14.8%, w/v pea protein but the focus in the literature is on the lower concentration of pea protein as seen in Fig. 3. 56% of the studied samples have a concentration of ≤1% pea protein.
According to the stability results summarized in the ternary phase map shown in Fig. 2, it seems to be possible but difficult to create stable emulsions with low pea protein concentration as seen in the lower part of the diagram.The diagram is composed of points representing a composition of water, oil and pea protein.The three corner points represent samples with purely water (right corner), only oil (left corner) and the third corner represents 12%, w/v pea protein solution in water (top corner).The scaling of the third axis is chosen as it allows the range of concentrations that have been studied to be visualised more clearly.Some of the oil/water/pea protein composition ratios have been investigated independently in different studies with varying results as demonstrated by the overlapping points.While it is easy to understand that the variability is due to different measuring techniques to assess stability, different preparation methods and degrees of protein purification as well as the salt and buffer content, the main reason for different results are probably due to natural variability of different pea crops, growing conditions, varieties and ripeness.
The broad range of stability, from high internal oil phase emulsions with 70 to 80%, v/v oil and low total protein concentration, <2.5%, w/ v, to low oil concentration with 10 to 30%, v/v oil and a broad range of pea protein concentrations >0.5%, w/v, suggests that the stabilization mechanisms are different at different compositions.Sridharan et al have suggested that emulsions prepared with 10%, v/v rapeseed oil and 0.5%, w/v pea protein at pH 3 are mainly stabilized by protein molecules rather than particles [97].These conclusions are based on comparison of experimental and theoretical surface coverage, where in the experiments, protein molecules could provide surface coverage of 47% but particles would cover just 3% of the total interface.In contrast, Liang and Tang suggested that particle stabilization occurs for emulsions prepared with 20%, v/v soybean oil and 0.2-2.4%,w/v pea protein at pH 3 based on light scattering experiments of dispersed protein particles [98].Particle stabilized emulsions have also been proposed by Li et al and Sun et al for high internal oil phase emulsions with pea protein concentrations <2%, w/v [99,100].These diverse results are likely to be an indication that there are different stabilization mechanisms for different emulsion compositions and that the reported variation of stability is an effect of different pea protein compositions.The stabilization mechanisms at higher pea protein concentrations are not well explored and the argument made by Sridharan et al about surface load and oil droplet coverage may hold for the specific system, but not be a generally applicable to all the emulsion compositions with higher protein content.

Viscosity of emulsions
Although rheological properties are clearly important in processing and for end-use of emulsions, there are only a few studies that describe the properties and there are diverse results reported that perhaps largely reflect the different range of compositions.Unsurprisingly the systems prepared with low volumes of dispersed oil have fairly low viscosity: Jarzebski et al describe viscosity for pea protein stabilized emulsions with 10% oil only in the presence of lecithin and state that it is Newtonian and comparable with water [27].The pea protein used was a commercial food material.The plots indicate the viscosity is about double that of water and so is somewhat higher than the 25% increase expected by simple application of Einstein's model for increase of viscosity with concentration.In contrast, Sridharan et al show data that represent significant shear thinning up to shear rates of 1000 s − 1 for similar oil fractions with samples prepared with pea flour [63].Peng et al report less marked shear thinning in samples made with a pea protein isolate but the heat-treated protein samples are notably more viscous [101].This result suggests that the heat treatment alters the protein conformations.Zhang et al investigated emulsions with a commercial pea protein isolate and a lower oil fraction and observed shear thinning that was dependent on both pH and salt content [102].These studies suggest that some longer-range interactions between droplets may often occur and influence significantly rheology even if viscosities are not so high that they would represent gelled samples that would inhibit creaming over long periods.The long-range interactions could arise from charge dissociation of protein or other components.
A contrast is certainly observed in the rheology data presented by Li et al where a pea protein isolate was used with corn oil to prepare high internal phase emulsions (75%, v/v oil) with mixing and heat treatment [99].The viscosity, as expected, is much higher and there is clear shear thinning behaviour.The observed storage modulus was about an order of magnitude higher than the loss modulus.In such samples the gelling or viscosity can readily influence the observation of stability.It would evidently be helpful if more reports of rheological data were made as they provide useful information about properties and can lead to deeper understanding of the mechanisms of stability.

Surface potential effects in emulsions
Zeta potential is frequently used in studies of pea protein dispersions or emulsions, particularly to investigate effects of ionisation at different pH.High values of the zeta potential, either positive at low pH or negative at high pH, are generally favourable for emulsion formation and stability, as the charge effects prevents coalescence and flocculation.The isoelectric region for pea protein is reported between pH 4 to 5 [39,102,103].Protein solubility in water is also correlated with zeta potential, and commonly the lowest values are reported around the isoelectric point [87,103,104].Some processes that are reported to improve solubility are homogenization, high intensity ultrasound treatment and isoelectric precipitation of the protein [26,71,104].However, it is not clear to what extent these effects could arise from denaturation or degradation of the proteins or whether there are consequent changes of ionisation.It is also possible that dispersed particles of different size might have quite different interfaces with different distribution of hydrophobic and hydrophilic proteins and moieties.
Lam et al reported no significant difference of solubility for different pea cultivars, growing locations or season variability [3].Regarding droplet size distribution in emulsions, differences between pea protein concentrate, albumin rich fractions and globulin rich fractions in emulsions were reported by Kornet et al [65].Albumin rich fractions showed bimodal size distributions with flocculated samples, that were not present in the globulin rich and pea protein concentrate samples.Multimodal droplet size distributions are not uncommonly reported for pea protein emulsions [26,102,105].Zhang et al used pea protein microgels to prepare emulsions and proposed that the smaller population corresponds to free particles of stabilizer that are not adsorbed to the oil/water interface during the homogenization process to form the emulsions [102].Effective methods used to decrease droplet polydispersity are reported to be high intensity ultrasound and microfluidization [71,98], whereas increasing protein concentration and homogenization time are effective methods to decrease droplet size [49,62,106].Although the zeta potential and solubility are correlated with pH, no clear trend in droplet size has been reported by comparing samples of different pH.Zhang et al and Aluko et al did not report significant differences between samples [49,102], whereas Ladjal-Ettoumi et al and Chang et al reported larger droplet size at the isoelectric point [87,103].Sha et al compared zeta potential of pea protein dispersions and emulsions with 25%, v/v sunflower oil, and reported that the zeta potential had a greater negative value for emulsions as prepared [72].The same effect was observed for samples made with purified 11S and 7S proteins.These results were supported by Zhan et al, who reported the zeta potential as a function of oil fraction and saw the highest net zeta potential at 30% oil fraction and decreasing for oil fractions lower and higher [107].They suggested that the difference in zeta potential was due to difference in the droplet size and the amount of protein attached to the surface.However, in contrast, Zhang et al did not observe a significant zeta potential difference between dispersions and emulsions [74].Even though the zeta potential and droplet size measurements are frequently reported together, the effect of charge on droplet size is not clear.The divergence of results is likely to be an indication that there are multiple different mechanisms contributing to pea protein emulsion stability.

Interfacial tension and interfacial rheology
The interfacial energy, often described as an interfacial tension, of oil and water is clearly important as it determines the free energy of the interface and so several studies report the changes in this parameter when pea protein is adsorbed.However, most emulsions are not thermodynamically stable and so both the evolution of the interfacial tension as well as the elasticity and the viscosity of the interface are important.Emulsions are usually formed under conditions of high shear and even at rest, droplets would undergo fluctuations of shape that necessarily involve some changes in surface area.The interfacial rheology is therefore an important consideration in both formation and storage of emulsions.There are several reports of these parameters and it is interesting to correlate the reported results so as to understand the formation and stability of emulsions.
As regards the interfacial energy, it is widely seen that measurements such as those of the shape of a pendant drop indicate that it decreases with time.This has been seen for pea globulins as well as other vegetable  [36,[108][109][110].The changes with time have been interpreted and modelled in various ways.Most authors suggest that there is a tendency to reach an equilibrium value at long times but this may be as long as several or many hours.The changes can arise due to diffusion of protein to the interface or the rearrangement of the particles or molecules at the interface.In some cases, authors report results as surface pressure, π, that is defined as: π = γ 0 − γ p (7) where γ 0 is the interfacial energy of the 'clean' oil aqueous phase interface and γ p is the energy in the presence of protein.If the values are reported as surface pressure but a value for γ 0 is not stated, estimates can be made using values in Table 1 for the oil/water surface but there are some changes with temperature and in the presence of buffer or other salts.The values of γ 0 for many vegetable oils are similar (about 30 to 33 mN m − 1 ) but those for alkanes and silicone oils are rather different.
A survey of some reported values of interfacial energy is shown in Table 5.There is a substantial reduction of the interfacial tension of vegetable oils with water in the presence of pea protein material with typical values of 10 to 15 mN m − 1 at concentrations of 0.01 and 0.1%, w/w that can be compared with approximately 30 to 32 mN m − 1 for the 'bare' oil/water interface.It is easier to identify trends within results from individual studies as there are many different protein isolates and conditions of study.The results of Amine et al suggest that lower interfacial tensions are achieved at higher protein concentration (5%) but, as yet, there does not appear to be systematic studies with results reported for many concentrations [111].There are systematic effects of decreasing interfacial tension with increasing pH seen in Fig. 4, representing the results of Chang et al who used corn oil and observed the largest reduction of about 10 mN m − 1 for an albumin rich sample between pH 3 and pH 9 [109].There are small reductions seen by adding sodium chloride up to 0.1 mol L − 1 with the least effect for albumin and slightly larger effects for legumin-rich and globulin-rich fractions.
It is not straightforward to make direct numerical comparisons of the data that are available for interfacial rheology from different studies.Apart from the wide variety of compositions, results are reported for completely different methods that measure different quantities.Several investigations, e.g.[109][110][111], report on the results of imposing cyclic volume or pressure changes on small pendant drops.This provides measurement of elasticity and viscosity or storage modulus and loss modulus that correspond to tension in the interfacial film.Other studies describe shear of the interfacial film, e.g.[87].Apart from this substantial difference, rheological measurements often depend on amplitude of strain and the history of deformation.For these reasons, it is necessary largely to report trends in individual studies and these are not always consistent.Amine et al reported higher elasticity at pH 7 than pH 10 for low concentrations of isolate (0.1 to 0.2%) but rather similar values at high concentrations [111].Chang et al measured dilation elasticity with the largest modulus observed generally at the lowest pH, pH 3 [109].They also observed significant changes with timein the presence of 100 mmol L − 1 NaCl, there was a pronounced maximum in the modulus after about 4000 s and 7000 s for the vicilin and albumin rich fractions respectively.Grasberger et al suggest that there is less network formation and lower elasticity for films formed from a homogenized isolate [110].The polarity of the oil is seen as significant in this work with a comparison of the elasticity for pea protein isolates that was lower in 1-octanol than in n-octane.The values reported by Ducel et al for a "vaselin" were not quite as low [108].In contrast, Amine et al reported low values for high concentrations of isolate at medium chain triglyceride interface [111].
This complex pattern of variation of interfacial tension and rheological behaviour with concentration, protein composition, pH, added salt, and time suggests that further investigations will be needed for optimised formulations that may be advantageous for emulsions.Processing with high shear for short duration such as during initial emulsification may exploit different properties to those relevant for the longterm stability of emulsions.The direct investigation of the oil/water interface and mobility of the protein by Velandia et al also suggested that structure of pea protein isolate is influenced by pH [38].

Conclusions
Proteins from peas have great potential as emulsifiers to stabilize food emulsions, for encapsulation purposes and can provide protein enrichment of food products.Pea protein has been used to make stable emulsions of various compositions with an aim eventually to use in many different products with various specifications as regards physical and chemical properties that may relate to sensory perception as well as nutrition.For example, a milk substitute based on pea protein emulsion is now commercially available under the trade name 'Sproud' [112].Other products that involve emulsions could include high protein drinks, low fat spreads, stable carriers of oil soluble vitamins, or even mimics of meat patties that are processed from dispersions.The region of stability for pea protein emulsions range from high internal phase emulsions to low protein emulsions with regions of stability presented in the lower left and right corners of the ternary phase map in Fig. 2. The overlapping points describing different extent of stabilization are results of the many factors affecting the stability such as pea variety, protein purification steps, storage temperature, pH adjustment and buffer addition.The extraction and processing methods of the pea material could change composition of different proteins which affect the states of aggregation and denaturing.The wide range of stability suggests that different  stabilization mechanisms are important.A thorough understanding of the stabilization mechanisms with contributions from the different proteins or protein states, as particles, individual globular proteins or denatured and solvated chains at different compositions is desirable.This will aid careful choices to tune the protein material to optimize emulsion stability.A large unexplored region of stability was identified for intermediate oil content of 30 to 70%, v/v and this is a potential area for further studies.The aim to use vegetable proteins as protein enhancement in food emulsions opens up a new field of emulsion studies.
The review revealed diverse ways of describing emulsion stability, with different creaming and stability indices only indirectly relating to stabilization mechanisms.To simplify comparison between different studies, a uniform system to evaluate emulsion stability is preferred.As emulsions are metastable, or kinetically stable, rather than thermodynamically in a state of minimum energy, experimental techniques that evaluate the time dependence of emulsion stability are valuable.Also, methods to measure the homogeneity within samples are useful.Examples of such techniques are UV-vis spectrophotometer absorbance to derive the emulsion stability index and results based on visual observation such as the creaming and Turbiscan stability indices.From the perspective of a consumer, main criteria to assess emulsion stability are visual observation to see a homogeneous product as well as physical properties.It is, therefore, advised that this simple technique also be used as a complementary method to report results in research studies of emulsion stability.These direct measurements could be complemented by other information about physical properties.Although many studies suggest that emulsions with pea protein are stabilized by a Pickering mechanism, there is now abundant evidence, such as that from rheology, that one cannot consider simply the interfacial energy of the proteins at the oil/water interface but must consider a range of dynamic and kinetic effects both as regards emulsification and stability of the resulting emulsions.
For future research there is a need to increase the understanding of the phase behaviour of emulsions with pea protein.Phase maps can be created for different pH, various extraction methods for proteins and for a range of oil types to provide a more complete picture of the effects on stability.An important step forward is the recognition of the complexity of the pea material.With this realization comes the importance of putting the experiments into the context of stability mechanisms rather than just collecting further data that reports emulsion stability.Direct studies of the state of the stabilizing material, such as the aggregation, potential and extent of unfolding would be important.The role of the different stabilization mechanisms such as steric stabilization, charge dissociation, depletion interactions and particle or Pickering stabilization may then be possible to identify.The understanding of emulsions with pea protein can then be linked with, and applied to, the broad range of other protein and plant-product stabilized emulsion systems.

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.

Fig. 1 .
Fig. 1.Number of published papers on pea protein emulsions by year (found on Web of Science with the search term "pea protein AND emulsion".The number has an approximately exponential development trend from the first paper 1992 to the current level of >100 during one year.

Fig. 2 .
Fig. 2. Stability phase diagram of emulsions stabilized by pea protein as reported in the literature.Green circles represent the compositions with stable emulsions measured ≥14 days, blue diamonds represent stable emulsions measured <14 days, black cross represents stable emulsions measured <7 days, orange squares represent stable emulsions measured <1 day, and red triangles represent unstable emulsions.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) da Costa Cardoso et al[160].E.Olsmats and A.R. Rennie

Fig. 3 .
Fig. 3. Number of samples studied at different pea protein concentrations.

E
.Olsmats and A.R. Rennie

Table 1
Properties of oils.Density values measured at 15 • C if not otherwise stated.Oil/water and air/oil interfacial tension measured at 20 • C. * are average values from [46,116], uncertainty typically about 1 mN m − 1 .Dielectric constant values for vegetable oils measured at 25 • C and frequency range 100 Hz-500 kHz.Dynamic viscosity values measured at 22 • C if not otherwise stated.

Table 3
Amino acid composition of pea protein isolates as % of total amino acid content.

Table 4
Measurement techniques to evaluate physical storage stability and the frequency with which they are reported as percentage of the total number of articles studied.The total comes to >100% as studies may report multiple techniques and results.

Table 5
Reported values of interfacial tension for pea protein stabilized emulsions.